Document [original]

Human impacts on aquatic ecosytems:

insights from dissolved organic

matter signatures and greenhouse gas

emissions

vorgelegt von

Clara Romero González-Quijano, M.Sc.

an der Fakultät VI – Planen Bauen Umwelt

der Technischen Universität Berlin

zur Erlangung des akademischen Grades

Doktorin der Naturwissenschaften

- Dr. rer. nat. –

genehmigte Dissertation

Promotionsausschuss:

Vorsitzender: Prof. Dr.-Ing. Reinhard Hinkelmann

Gutachter: Prof. Dr. Mark Geßner

Gutachter: Prof. Dr. Gabriel Singer (Universität Innsbruck)

Gutachterin: Prof. Dr. Sarian Kosten (Radboud Universiteit, Niederlande)

Gutachter: Jun.-Prof. Dr. Maximilian Lau (TU Bergakademie Freiberg)

Tag der wissenschaftlichen Aussprache: 19. April 2024

Berlin 2024

Preface

First of all, I would like to thank my main supervisor, Gabriel Singer: you are an amazing

scientist, but also a great person. Your positive attitude kept me going, you always had a

positive/constructive comment and a smile, even in the worst moments, when I thought nothing

was going to work out. I enjoyed a lot our “hippie” group, where the main points were to learn

and enjoy science. Thanks also for letting me be part of the Kenya experiments, it was a

wonderful experience that made me realized how lucky I am. Mark Gessner and Peter Casper,

thanks for your supervision, your input made my papers improve a lot. Tobias Goldhammer, my

life changed when you arrived to the lab, thanks a lot for your help and support and for

“adopting” me after they all moved to Innsbruck. Ruben del Campo, thanks a lot for your help y

por no dejar que mi barco se hundiera (en varias ocasiones), I really hope you will get a

permanent position in Spain soon, because you really deserve it.

To the FLEE lab, even I will never forgive you for leaving Berlin, I really appreciate your

support in the distance. You are a group of really nice people. Thomas Fuss, thanks for being

there through the years, I hope you come and visit us in Spain so we can play some beachvolley

again while our kids eat a bit of sand. Lukas Thuile, Selin Kubilay and all the others…thanks

and good luck.

To Sonia Herrero, this dissertation is also your work, we learnt so much together, thanks a lot

for being there, “every morning, in every crisis”. Our teamwork was key for this thesis, we both

grew together, thanks a lot. Thanks also to all the UWIs, for the scientific discussions, social

activities, the best broccoli ever and Christmas markets. I thank Reinhard Hinkelmann and

Gwendolin Porst for their support.

To all the students helping in the field: big thank you! Specially to Lena Meinhold, who was

always willing to help and learn, we spent hours driving and sampling around Berlin, it was a

lot of fun. Thanks also to Cleo Stratmann, for the amazing work with the permissions/paper

work.

An die CAB-Laboranten, vielen Dank, mein Leben im Labor war nicht immer einfach, aber ich

habe viel Hilfe von euch bekommen: Danke an Claudia Schmalsch, Sarah Krocker und Angela

Krüger. Danke an meine Kollegen am IGB, besonders an die Beachvolleyballer und IGB-Läufer,

unsere Spiele und Läufe um den Müggelsee haben mir viel Spaß gemacht. Danke auch an Kirsten

Pohlmann, für deine Unterstützung und Hilfe in all den Jahren. IGB-Verwaltung und -

Wartungsteam: Danke für eure Unterstützung und eure Arbeit.

To the Karlshorsters (my chosen family): Anna Jäger, Marta Alirangues and Mikael Gillefalk,

thanks for being always there, for babysitting the little monsters and for providing good wine

when it was really needed.

Preface

A las mamis de Berlín, gracias por ser mi tribu, el hombro al que llorar y la gran fuente de

conocimiento a la que acudir. Nunca compartí mucho de esta tesis con vosotras, y eso me dio

aire, para estar con vosotras tranquila, sin agobios, y así tener una parte de mi vida fuera de la

ciencia. Cruzo los dedos por muchos viajes de señoras más. Lo mismo pasa con vosotras

churripurris, gracias por llevarme a dar cabezazos por ahí, y recordarme que no ser muy friki, ni

tener mil papers, también está bien.

A mi familia, gracias por vuestra paciencia y ayuda con los nietos, por aguantar mis respuestas

en los días malos y por proveerme de jamón del rico (y croquetas, gracias Cubito) cuando más

falta me hacía. A Unai y Matías, aunque vuestra llegada hizo que el doctorado se alargara, os

doy las gracias por enseñarme a amar como nunca hubiera imaginado que amaría a alguien. Edu,

es muy difícil expresar lo que ha significado tu apoyo todos estos años. No me puedo imaginar

otro compañero de vida mejor, gracias, te quiero.

Publications of cumulative doctoral thesis

iii

Publications of cumulative doctoral thesis

1. Masese, F. O., M. J. Kiplagat, C. R. González-Quijano, A. L. Subalusky, C. L. Dutton, D. M.

Post, and G. A. Singer. 2020. Hippopotamus are distinct from domestic livestock in their

resource subsidies to and effects on aquatic ecosystems. Proceedings of the Royal Society B:

Biological Sciences 287:20193000. http://doi.org/10.1098/rspb.2019.3000

2. Romero González-Quijano, C., Herrero Ortega, S., Casper P., Gessner M, and Singer, G. 2022.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and

drivers. Biogeosciences 2022:1-34. https://doi.org/10.5194/bg-19-2841-2022

3. Romero González-Quijano, C., Herrero Ortega, S., del Campo, R., Casper, P., Gessner, M. O.,

Goldhammer, T. & Singer, G. A. Carbon dioxide emissions across an urban aquatic network.

Limnology and Oceonography (Submitted version).

4. Herrero Ortega, S., C. Romero González-Quijano, P. Casper, G. A. Singer, and M. O. Gessner.

2019. Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole-city

footprint. Global Change Biology 25:4234-4243. https://doi.org/10.1111/gcb.14799

An outline of supplementary scientific work is given in chapter 7.

This thesis was carried as part of the Research Training Group “Urban Water Interfaces (UWI)

(DFG - GEPRIS - GRK 2032: Grenzzonen in urbanen Wassersystemen), which was funded by

the German Research Foundation (DFG).

Summary

Historically, human settlements have gravitated towards water bodies to efficiently obtain

essential resources and for the sake of transport. This proximity has imposed significant stress

on aquatic ecosystems. Land cover changes, particularly urbanization and agricultural

expansion, have altered terrestrial-aquatic exchange processes and subsequently the structure

and function of these water bodies close to settlements. Urban freshwater systems are

challenged with high loads of organic carbon, nutrients, and pollutants, besides being heavily

modified morphologically and hydrologically. Using Berlin as a case study, the spectrum of

anthropogenic stressors on aquatic systems is exemplified: Here, morphologically partly heavily

altered water bodies rely on locally abstracted groundwater and face considerable wastewater

but also diffuse inputs. If urban environments are considered the rather extreme end of human

influence on water bodies, then the landcover change towards pastures associated with human

appropriation of landscape may be seen as the beginning. Today, this is exemplified by the

ongoing replacement of wildlife by livestock in the African savanna. The associated decline of

hippopotamus populations has raised concerns, as hippos play a crucial role in organic matter

and nutrient transfer to water bodies. The Mara River, hosting over 4000 hippos alongside

extensive cattle coexistence, is a river basin threatened by increasing livestock, agricultural

activities, and human settlements.

A key component of aquatic ecosystems is dissolved organic matter (DOM), a substantial carbon

reservoir vital for biogeochemical cycling. DOM comprises organic compounds from various

sources like terrestrial runoff, leaf litter, animal faeces, and algal exudates. Microbial metabolism

of DOM is central to carbon processing, releasing carbon dioxide (CO2) – a process linking

aquatic ecosystems with the global carbon cycle. Under specific conditions, primarily in oxygen-

depleted zones like sediments and wetlands, anaerobic processes produce methane (CH4), a 30-

times more potent greenhouse gas. Whether CO2 or CH4 dominates as the gaseous end product

of mineralization depends on DOM availability and lability, microbial metabolism, and

controlling environmental factors.

In this dissertation, I comprehensively explored the impact of human activities on aquatic

ecosystems, focusing on the dynamics of DOM composition, ecosystem metabolism, and the

greenhouse gases (GHG) CO2 and CH4 in urbanized Berlin and the African savanna. I delved

into the effects of human activities on DOM composition in the entire urban aquatic network of

Berlin and in the Kenyan savanna. In Kenya, I additionally investigated ecosystem metabolism

in an experimental setting. Last, I conducted an analysis of GHG emissions in highly impacted

urban aquatic ecosystems.

Firstly, I investigated spatio-temporal variations in DOM composition across urban water

bodies of Berlin, which revealed diverse DOM signatures, that were differentiated between lakes

Summary

and ponds rich in autochthonous DOM and rivers and streams dominated by allochthonous

DOM. Seasonal shifts were attributed to phenological changes and urban influences, such as

nutrient influx and point source pollution. Optical DOM properties expose the impact of

wastewater treatment plant effluents, highlighting their utility in water quality assessment, but

DOM signatures were also informative about functional ecosystem aspects like coupling to the

catchment and terrestrial systems and aquatic primary production. Similarly, in an experimental

setting in the African savanna, DOM reacted sensitively to the replacement of hippos by

livestock. In this case, I also showed an impact on aquatic ecosystem structure and function, i.e.

an increase in gross primary production (GPP). Last, I measured CO2 and CH4 fluxes in Berlin's

aquatic network, identifying their drivers and conducting an upscaling exercise. Findings

showed that urban waters emit both CO2 and CH4, the estimated total annual emissions from all

of Berlin’s surface waters were 8.5 ± 1.3 Gg CO2 and 2.6 ± 1.7 Gg CH4 (mean ± SD), estimated

from seasonal measurements of instantaneous flux across approximately 30 sites. CH4 emission

hotspots were identified in small water bodies embedded in urban green spaces. CO2 emission

rates from running waters were lower than reported in the literature, likely emphasizing the

importance of daily variability in global estimations.

In conclusion, this dissertation underscores the impact of human activities on aquatic ecosystems

in both savanna and urban environments. It is important to heed these insights in aquatic

ecosystem management.

Zusammenfassung

Seit jeher haben sich menschliche Siedlungen in der Nähe von Gewässern angesiedelt, um sich

effizient mit lebenswichtigen Ressourcen zu versorgen und um sich fortzubewegen. Diese Nähe

hat zu einer erheblichen Belastung der aquatischen Ökosysteme geführt. Veränderungen der

Bodenbedeckung, insbesondere die Verstädterung und die Ausdehnung der Landwirtschaft,

haben sich auf die terrestrisch-aquatischen Austauschprozesse und damit die Struktur und

Funktion dieser Gewässer in der Nähe von Siedlungen ausgewirkt. Städtische

Süßwassersysteme sind mit hohen Belastungen durch organischen Kohlenstoff, Nährstoffe und

Schadstoffe konfrontiert und zudem morphologisch und hydrologisch stark verändert. Am

Fallbeispiel Berlin wird das Spektrum der anthropogenen Stressoren auf aquatische Systeme

exemplarisch dargestellt: Hier sind morphologisch teilweise stark veränderte Gewässer auf

lokal entnommenes Grundwasser angewiesen und mit erheblichen abwasserbedingten, aber

auch diffusen Einträgen konfrontiert. Wenn städtische Umgebungen als das eher extreme Ende

des menschlichen Einflusses auf Gewässer betrachtet werden, dann kann die Veränderung der

Bodenbedeckung hin zu Weiden, die mit der Aneignung der Landschaft durch den Menschen

einhergeht, als Anfang angesehen werden. Ein heutiges Beispiel dafür ist die fortschreitende

Verdrängung von Wildtieren durch Nutztiere in der afrikanischen Savanne. Der damit

verbundene Rückgang der Flusspferdpopulationen gibt Anlass zur Besorgnis, da Flusspferde

eine entscheidende Rolle beim Transfer von organischem Material und Nährstoffen in die

Gewässer spielen. Der Mara-Fluss, in dem mehr als 4000 Flusspferde leben, ist ein

Flusseinzugsgebiet, das durchzunehmende Viehzucht, landwirtschaftliche Aktivitäten und

menschliche Siedlungen bedroht ist.

Ein essenzieller Bestandteil aquatischer Ökosysteme ist die gelöste organische Substanz

(DOM), ein wichtiges Kohlenstoffreservoir, das für den biogeochemischen Kreislauf unerlässlich

ist. DOM besteht aus organischen Verbindungen aus verschiedenen Quellen wie terrestrischem

Abfluss, Laubstreu, tierischen Fäkalien und Algenexsudaten. Der mikrobielle Stoffwechsel von

DOM ist von zentraler Bedeutung für die Kohlenstoffverarbeitung und setzt Kohlendioxid

(CO2) frei - ein Prozess, der aquatische Ökosysteme mit dem globalen Kohlenstoffkreislauf

verbindet. Unter bestimmten Bedingungen, vor allem in sauerstoffarmen Zonen wie Sedimenten

und Feuchtgebieten, erzeugen anaerobe Prozesse Methan (CH4), ein 30-mal stärkeres

Treibhausgas. Ob CO2 oder CH4 als gasförmiges Endprodukt der Mineralisierung überwiegt,

hängt von der Verfügbarkeit und Labilität von DOM, dem mikrobiellen Stoffwechsel und den

steuernden Umweltfaktoren ab.

In dieser Dissertation untersuchte ich umfassend die Auswirkungen menschlicher Aktivitäten

auf aquatische Ökosysteme und fokussierte mich dabei auf die Dynamik der DOM-

Zusammensetzung, den Ökosystemstoffwechsel und die Treibhausgase (THG) CO2 und CH4 im

Zusammenfassung

vii

urbanisierten Berlin und in der afrikanischen Savanne. Ich analysierte die Auswirkungen

menschlicher Aktivitäten auf die DOM-Zusammensetzung im gesamten städtischen

Gewässernetz Berlins und in der kenianischen Savanne. In Kenia untersuchte ich zusätzlich den

Stoffwechsel des Ökosystems in einem experimentellen Aufbau. Schließlich führte ich eine

Analyse der Treibhausgasemissionen in stark beeinflussten städtischen aquatischen

Ökosystemen durch.

Zunächst untersuchte ich die räumlichen und zeitlichen Schwankungen der DOM-

Zusammensetzung in den städtischen Gewässern Berlins. Dabei wurden verschiedene DOM-

Signaturen festgestellt, die sich in Seen und Teiche mit einem hohen Anteil an autochthonem

DOM und in Flüsse und Bäche mit einem hohen Anteil an allochthonem DOM unterteilen

lassen. Saisonale Verschiebungen wurden auf phänologische Veränderungen und städtische

Einflüsse wie Nährstoffzufuhr und Verschmutzung durch Punktquellen zurückgeführt.

Optische DOM-Eigenschaften zeigen die Auswirkungen von Kläranlagenabwässern auf und

unterstreichen ihren Nutzen für die Bewertung der Wasserqualität, aber DOM-Signaturen

waren auch informativ für funktionelle Ökosystemaspekte wie die Kopplung mit dem

Einzugsgebiet und terrestrischen Systemen und die aquatische Primärproduktion. In ähnlicher

Weise reagierte DOM in einem Versuchsfeld in der afrikanischen Savanne empfindlich auf die

Verdrängung von Flusspferden durch Nutztiere. In diesem Fall konnte ich auch eine

Auswirkung auf die Struktur und Funktion des aquatischen Ökosystems nachweisen, d. h. einen

Anstieg der Bruttoprimärproduktion (GPP). Schließlich habe ich die CO2- und CH4-Flüsse im

Berliner Gewässernetz gemessen, ihre Einflussfaktorenermittelt und ein Upscaling

durchgeführt. Die Ergebnisse zeigten, dass städtische Gewässer sowohl CO2 als auch CH4

emittieren. Die geschätzten jährlichen Gesamtemissionen aller Berliner Oberflächengewässer

beliefen sich auf 8,5 ± 1,3 Gg CO2 und 2,6 ± 1,7 Gg CH4 (Mittelwert ± SD), geschätzt aus

saisonalen Messungen des momentanen Flusses an etwa 30 Standorten. CH4-Emissions-

Hotspots wurden in kleinen, in städtische Grünflächen eingebetteten Gewässernidentfiziert. Die

CO2-Emissionsraten aus Fließgewässern waren niedriger als in der Literatur angegeben, was

wahrscheinlich die Bedeutung der täglichen Variabilität bei globalen Schätzungen unterstreicht.

Zusammengefasst beleuchtet diese Dissertation den Einfluss menschlicher Aktivitäten auf

aquatische Ökosysteme, sowohl in der Savanne als auch in städtischen Gebieten. Es ist von

großer Bedeutung diese Erkenntnisse für das Management aquatischer Ökosysteme zu nutzen.

  Contents 
   
viii 
 
Contents 
 
Publications of cumulative doctoral thesis ..................................................................................... iii 
Summary ................................................................................................................................................. iv 
Zusammenfassung ................................................................................................................................ vi 
Contents ............................................................................................................................................... viii 
General Introduction ............................................................................................................................. 1 
1.1 Influence of human landscape development on aquatic ecosystems .............................................. 2 
1.2 Dissolved organic matter (DOM) in aquatic ecosystems ................................................................. 5 
1.3 Ecosystem metabolism .............................................................................................................................. 8 
1.4 Greenhouse gases (GHGs) in aquatic ecosystems ............................................................................. 9 
1.6 Objectives .................................................................................................................................................. 12 
Hippopotamus are distinct from domestic livestock in their resource subsidies to and 
effects on aquatic ecosystems ........................................................................................................... 13 
2.2 Introduction .............................................................................................................................................. 14 
2.3. Material and methods ............................................................................................................................ 16 
2.3.1 Characteristics of cattle and hippo dung.................................................................................... 16 
2.3.2 Livestock versus hippopotamus loading rates of organic matter ........................................ 16 
2.3.3 Experimental mesocosms .............................................................................................................. 17 
2.3.4 Water sampling and analysis ........................................................................................................ 17 
2.3.5 Metabolism ........................................................................................................................................ 17 
2.3.6 Data analysis ..................................................................................................................................... 18 
2.4. Results ....................................................................................................................................................... 19 
2.4.1 Characteristics of cattle and hippo dung.................................................................................... 19 
2.4.2 Loading rates of organic matter by cattle and hippopotamus .............................................. 19 
2.4.3 Nutrients ............................................................................................................................................ 20 
2.4.4 Organic matter and biomass ......................................................................................................... 20 
2.4.5 DOC composition ............................................................................................................................ 20 
2.4.6 Ecosystem metabolism ................................................................................................................... 21 
2.5 Discussion ................................................................................................................................................. 25 
2.6 Conclusion ................................................................................................................................................. 27 
Supplement of Chapter 2
 ................................................................................................................... 28 
Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and 
drivers .................................................................................................................................................... 51 
3.1 Abstract ...................................................................................................................................................... 51 

Contents   
ix 
 
2 Introduction .............................................................................................................................................. 52 
3 Methods ..................................................................................................................................................... 53 
3.1 Study sites ......................................................................................................................................... 53 
3.2 Physico-chemical field measurements and water sampling .................................................. 56 
3.3 DOM characterization ................................................................................................................... 56 
3.4 Additional water-chemical analyses ............................................................................................ 58 
3.5 Data analysis ..................................................................................................................................... 59 
4. Results ....................................................................................................................................................... 60 
4.1 Physico-chemical characteristics ................................................................................................. 60 
4.2 DOM composition ........................................................................................................................... 61 
5. Discussion ................................................................................................................................................ 65 
5.1 Spatial patterns and drivers of DOM signatures ..................................................................... 65 
5.2 Seasonal patterns and drivers of DOM signatures ................................................................. 68 
5.3 DOM composition as a potential basis for urban surface water monitoring .................... 68 
6. Conclusion ................................................................................................................................................ 69 
Supplement of Chapter 3
 ................................................................................................................... 71 
Carbon dioxide emissions across an urban aquatic network ...................................................... 87 
1 Abstract ...................................................................................................................................................... 87 
2 Introduction .............................................................................................................................................. 88 
3 Methods ..................................................................................................................................................... 90 
3.1 Sampling design ............................................................................................................................... 90 
3.2 Estimation of CO2 fluxes ............................................................................................................... 90 
3.3 Theory of chamber operation and flux estimations ................................................................ 92 
3.4 Direct measurements of instantaneous CO2 flux ..................................................................... 93 
3.5 Indirect estimation of continuous flux using equilibration chambers ................................ 94 
3.6 Field sampling and laboratory analyses..................................................................................... 95 
3.7 Data analysis ..................................................................................................................................... 96 
4 Results ........................................................................................................................................................ 97 
4.1 Effects of water body type and season on CO2 fluxes ............................................................. 97 
4.2 Variability of CO2 fluxes ................................................................................................................ 99 
4.3 Comparison of Berlin CO2 fluxes with previous studies ...................................................... 102 
4.4 Annual CO2 emission from an urban aquatic network ......................................................... 102 
4.5 Drivers of CO2 fluxes and flux variability ............................................................................... 103 
5 Discussion ............................................................................................................................................... 107 
5.1 Temporal patterns ......................................................................................................................... 107 
5.2 Spatial patterns .............................................................................................................................. 108 
5.3 Potential drivers of CO2 fluxes .................................................................................................. 108 

Contents 
 
x 
 
4.5 Conclusion ............................................................................................................................................... 110 
Supplement of Chapter 4
 ................................................................................................................. 111 
Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city 
footprint .............................................................................................................................................. 123 
5.1 Abstract .................................................................................................................................................... 123 
5.2 Introduction ............................................................................................................................................ 124 
5.3 Materials and methods ......................................................................................................................... 125 
5.3.1 Study sites ....................................................................................................................................... 125 
5.3.2  CH4 emissions ................................................................................................................................ 126 
5.3.3 Extrapolation of CH4 emissions ................................................................................................. 127 
5.3.4 Water chemistry ............................................................................................................................ 128 
5.3.5 Land use ........................................................................................................................................... 128 
5.3.6 Data analysis ................................................................................................................................... 129 
5.4 Results ...................................................................................................................................................... 129 
5.5 Discussion ............................................................................................................................................... 133 
5.6 Ackowledgements ................................................................................................................................. 137 
Supplement of Chapter 5
 ................................................................................................................. 138 
Synthesis ............................................................................................................................................. 149 
6.1 DOM characteristics as indicators of landscape change .............................................................. 150 
6.2 Ecosystem functioning reveals impacts of land-use alteration .................................................. 152 
6.3 Improving global C budgets by considering spatio-temporal variability of GHG emissions 
in urban aquatic ecosystems ...................................................................................................................... 153 
6.4 Conclusions ............................................................................................................................................. 154 
6.5 Outlook .................................................................................................................................................... 155 
Complementary contributions ........................................................................................................ 156 
List of Figures .................................................................................................................................... 157 
List of Tables ..................................................................................................................................... 162 
References ........................................................................................................................................... 166 
 

CHAPTER 1

General Introduction

Throughout history, humans have decided to live close to water (Kummu et al. 2011; Wang et

al. 2022), where many ancient civilizations started. For example, Mesopotamia was located

within the Tigris–Euphrates river system and Egypt at the Nile River, because rivers provide

important resources and other services to humans. The close linkage of human activities to

aquatic ecosystems, however, entails strong pressures on water bodies. For example, widespread

land cover change to pastures, agricultural fields and urban areas perturb water bodies by

altering their morphology and adding nutrients, organic matter and pollutants (Friberg et al.

2011). For many years, human-impacted ecosystems have only been of management concern,

whereas fundamental ecological studies were dedicated to natural and seminatural aquatic

ecosystems. This focus was logical in view of the idea to first understand how natural aquatic

ecosystems function and then use them as a reference to assess the various human-impacts. The

last decades, however, have brought a growing number of studies analyzing the influence of

human activities on aquatic ecosystems (Friberg et al. 2011; Walsh et al. 2005), perhaps

motivated by the fact that there are hardly any pristine ecosystems left. Human influences on

ecosystems can indeed have devastating consequences. Alarmingly, we still do not fully

understand the manifold implications of anthropogenic activities on aquatic ecosystems,

especially if regarding, ‘land cover change’ or ‘urbanization’.

Human appropriation of a landscape often starts by usage for livestock grazing. In African

savannas, human settlements are expanding rapidly; the conversion of forest and savanna

grassland to pasture for livestock production is severely affecting aquatic ecosystems (Masese

et al. 2022). It implies the replacement of wildlife by livestock, with consequences for the

movement of terrestrial organic matter and nutrients into aquatic ecosystems (Bond et al. 2014).

At the other end of human landscape appropriation is full urbanization, with a plethora of

consequences for aquatic ecosystems worldwide (Grimm et al. 2008a).

General Introduction

1.1 Influence of human landscape development on aquatic ecosystems

Anthropogenic changes of the landscape vary considerably among regions around the world. A

widespread human influence on aquatic ecosystems is urbanization. The global urban population

is growing rapidly (Peña-Guzmán et al. 2019). While 3.9 billion people lived in urban areas in

2014 (Jensen and Wu 2018), by 2030 every third person will live in a city with at least 500.000

inhabitants (Figure 1, UN, 2018). Urban aquatic ecosystems undergo significant alterations and

face considerable challenges. They receive substantial quantities of nutrients, organic carbon,

suspended solids, and a diverse array of macro- and micropollutants, including heavy metals,

pharmaceuticals, and personal care products (Buser et al. 1999; Grimm et al. 2005; Hatt et al.

2004). Furthermore, urban aquatic ecosystems suffer from severe hydromorphological

modifications, which include the lateral and vertical disconnection from floodplains and aquifers

and longitudinal fragmentation by physical barriers such as dams (Steele and Heffernan 2013).

The degradation of the hydromorphological structure leads to the loss of habitat diversity in

urban freshwaters (Walsh et al. 2005), thereby restricting their capacity to sustain aquatic

biodiversity (Ward et al. 2002). Another notable characteristic of urban water systems is the

close integration of their natural and technical elements, resulting in numerous connections

within the urban and peri-urban water cycle (Gessner et al. 2014). The expansion of urban areas

further alters the physical characteristics of the land surface (Niemczynowicz 1999), such as

heightened sealing by paved surfaces. This impedes infiltration, accelerates surface runoff, and

affects the flow patterns and contaminant transport to the water bodies. These challenges are

encapsulated in the concept of the 'urban stream syndrome,' offering a comprehensive framework

that amalgamates the diverse impacts of watershed development and channel modification on

urban streams (Walsh et al. 2005). Global urbanization trends coincide with declining human

water supplies, coupled with a significant deterioration in water quality in certain regions.

Although innovative technologies have been developed and implemented to remove pollutants

from urban surface waters (Ramos et al. 2016), the management of urban aquatic ecosystem

often neglects crucial factors such as disruptions in connections with riparian areas and

floodplains, detachment from hyporheic zones and aquifers, and overall ecological conditions

(Gessner et al. 2014; Grimm et al. 2008b).

CHAPTER 1

Figure 1 Growth rates of urban agglomerations by size class (UN, 2018)

Berlin, with its abundant and diverse water bodies, presents an opportunity to examine

ecosystems under a spectrum of urban stressors, and ranging from minimally stressed to highly

modified systems. With a population of approximately 3.5 million inhabitants and an area

covering 889 km2, of which 6.4% consists of rivers, lakes, and artificial channels (Fromme et al.

2000), Berlin relies on the Spree and Havel Rivers for a substantial portion of its surface water

input. These rivers receive considerable wastewater inputs after tertiary treatment upstream of

the city limits (Pal et al. 2014) and receive effluents from six additional wastewater treatment

plants within the city. On average, 40% of Berlin's surface waters consist of wastewater

treatment plant effluents, reaching up to 84% locally (Heberer et al. 2002).The city operates on

a semi-closed water cycle reliant on locally abstracted groundwater. River bank filtration is

employed as a pre-treatment method to enhance the quality of drinking water distributed in

Berlin (Henzler et al. 2014). The quality of bank filtrate is susceptible to the influence of treated

wastewater discharged into upstream surface water of the filtration system (Heberer et al. 2008),

especially concerning pollutants with low removal efficiency during wastewater treatment.

While conventional wastewater treatment processes effectively remove many organic

micropollutants (OMPs), certain polar OMPs persist, entering surface water alongside treated

wastewater and manifesting in various compartments of Berlin's water cycle (Hass et al. 2012;

Richter et al. 2008).

Another important human influence on aquatic ecosystem consists on the ongoing

transformation of natural lands to pastures, and consequently, the replacement of wildlife by

livestock. This can have a particularly strong influence in freshwater ecosystems of the African

General Introduction

savanna and other ecologically similar regions (Subalusky et al. 2017). Although the link

between terrestrial and aquatic ecosystems has been traditionally considered mainly as the

transfer of leaf litter and hydrological inputs (Wallace et al. 1997), semiaquatic animals that use

terrestrial nutrients and energy may also have a significant influence on nutrient cycling and

food webs of aquatic ecosystems (Stears et al. 2018). Large mammal populations in East Africa

have decreased more than 50% over the past half-century (Craigie et al. 2010). Herbivore

communities have been extensively altered, to the extent that livestock nowadays dominates the

continent’s large mammal biomass (Hempson et al. 2017). The common hippopotamus

(Hippopotamus amphibius, from now on “hippo”) is a semiaquatic herbivore. Hippos eat large

amounts of terrestrial plants during the night (Lewison and Carter 2004) and return to the water

during the day, where they defecate, thereby transferring substantial amounts of organic matter

and nutrients from grasslands to aquatic ecosystems (Subalusky et al. 2015). Since hippo

populations have declined in many areas (Ogutu et al. 2011) and were replaced by livestock in

others, they now mainly live in conservation areas (Prins 2000). The populations continue to

decline because of continued growth of human settlements and of grassland and forest

conversion to pasture (Kinnaird et al. 2003; Prins 2000; Ripple et al. 2015). Studies assessing the

negative consequences of shifts from hippos to livestock have traditionally emphasized effects of

habitat degradation and nutrient and organic matter loading (Belsky et al. 1999; Bond et al.

2014). However, the per capita inputs of organic matter and nutrients by hippos is larger than

those by livestock because body sizes are greater and presence in or near water is longer. Their

effect on aquatic ecosystems is substantial even when hippo numbers are low (Iteba et al. 2021).

Furthermore, the composition of hippo and livestock faeces also differs, as their digestions are

different. Although livestock may sustain some ecosystem functions by maintaining the aquatic-

terrestrial linkage, a change in the composition and quantity of organic matter entering the

aquatic ecosystems in the form of faeces can affect ecosystem functions (Masese et al. 2018).

Mara River is an example of a basin threatened by the increase of livestock, agricultural activities

and human settlements (Masese et al. 2015). In the Mara River and its tributaries in the Maasai

Mara National Reserve (MMNR) there are more than 4000 hippos (Kanga et al. 2011) and over

250000 cattle contiguous to the reserve, where they coexist with wildlife (Ogutu et al. 2011).

Hippos live without cattle inside the reserve, they co-live with cattle and there are areas outside

the reserve where we can only find cattle (Veldhuis et al. 2019).

Human alterations of aquatic ecosystems can have a strong influence on aquatic biota, and

therefore on ecosystem functioning (Coscia and Kaiser 2022). To understand the character and

intensity of human impacts on aquatic ecosystem we need to gather an amalgam of

comprehensive tools which can inform from the basal alteration of carbon and nutrient subsidies,

to the ultimate consequences on the functional integrity of the system.

CHAPTER 1

1.2 Dissolved organic matter (DOM) in aquatic ecosystems

DOM plays a central role in aquatic ecosystems. It is one of the most important carbon stores,

making up more than 90% of the total organic matter in these systems, and therefore being

essential in the biogeochemical cycles of aquatic ecosystems (Song et al. 2019). DOM is

arbitrarily defined as the fraction of organic matter (OM) that can pass through a membrane

filter with a pore size of 0.45 µm. The processes behind the supply of DOM to aquatic ecosystems

are quite diverse: on the one hand, allochthonous DOM is derived from surrounding terrestrial

ecosystems (Aitkenhead-Peterson et al. 2003), its sources primarily consist of materials

exported from terrestrial environments, such as humic-rich substances with multiple aromatic

rings (Zhang et al. 2022). On the other hand, autochthonous DOM is produced within aquatic

ecosystems, by phototrophic bacteria, phytoplankton and macrophytes through photosynthesis,

as well as through excretion and secretion by a wide range of animals and by biomass decay

(Chester and Jickells 2012). Typically, these autochthonous compounds demonstrate a reduced

carbon-to-nitrogen (C:N) ratio and greater bioavailability when compared to terrestrial, humic-

like substances (Fasching et al. 2016). The exchange and transformation of DOM are intricately

intertwined with the biogeochemical cycles of biogenic elements, including carbon, nitrogen,

and phosphorus. Additionally, DOM plays a pivotal role in the processes of respiration and

photosynthesis within aquatic ecosystems. For instance, beyond the respiration of terrestrial

DOM, humic, compounds can play a non-consumptive function in metabolism by serving as

electron carriers in redox reactions (Parr et al. 2015). All these processes have a significant

impact on carbon release, burial, and nutrient cycling (Liu et al. 2019).

Our comprehension of DOM has quickly advanced beyond its sheer quantity to encompass its

molecular structure, establishing connections between the quantity and quality of aquatic DOM

and factors such as land use, climate change, and human activities (Xenopoulos et al. 2021).

Changes in land use from the cultivation of agricultural land or the introduction of livestock,

to urbanization can lead to substantial changes in DOM composition through alterations in

hydrological and biogeochemical processes (Chen et al. 2021). Indeed, recent research indicates

that land use can exert even a greater influence on DOM composition than location in the fluvial

network as expected from the traditionally accepted River Continuum Concept (RCC) (Chen et

al. 2021). Agricultural, farming and urban land uses can represent significant catchment

features, that can influence the sources, quantity and composition of aquatic DOM from local

(Wilson and Xenopoulos 2009b) to regional scales (Lambert et al. 2015). The replacement of

large mammals by livestock in savanna-like ecosystems is a clear example of drastic alteration

of DOM at local scale. Large mammal herbivores like hippos can play a substantial role as

conduits for the transfer of terrestrial organic matter and nutrients in these aquatic ecosystems

(Bond et al. 2014; Subalusky et al. 2017). Therefore, their replacement by livestock can affect

the quantity and composition of DOM reaching streams. In contrast, urbanization can influence

the composition of DOM at both local and catchment-scale through the input of anthropogenic

General Introduction

organic matter, derived from industry, sewage discharge, and household organic pollutants

(Hudson et al. 2007; Lyu et al. 2021). In comparison to non-urban aquatic environments, the

complexity and susceptibility to alterations of DOM in urban water bodies are heightened

(Graeber et al. 2012; Jaffé et al. 2008). In urban catchments, DOM typically exhibits a higher

contribution from autochthonous production rather than terrestrial input. Additionally, it tends

to have a relatively larger proportion of protein-like and/or microbial-humic DOM components

(Petrone et al. 2011; Williams et al. 2013; Yamashita et al. 2010). A comprehensive investigation

into the origin and composition of DOM can deepen our understanding of its reactivity,

environmental dynamics, and ultimate fate. This, in turn, enhances our ability to monitor and

safeguard aquatic ecosystems.

There are several methods to analyze DOM composition ranging from simple optical

measurements to complex identification of molecular compounds with high chemical resolution.

The utilization of optical measurements, specifically in assessing absorbance and fluorescence,

is prevalent for scrutinizing the composition of and discerning its origin and transformation of

DOM (Hansen et al. 2016). Parameters and indices standardized from optical data encompass

the outright absorbance or fluorescence intensity at specific wavelengths, the comparison of

different wavelength ratios, the normalization of optical properties based on carbon content, and

the evaluation of slopes within designated regions of the optical spectrum (Hansen et al. 2016).

The bioavailability of DOM is closely connected to its origins and its utilization by

microorganisms, and this aspect is not readily indicated by the concentrations of DOC (Benner

2003), but by its molecular properties, which can also inform about the source of water and

organic material (Xenopoulos et al. 2021); for instance, the degree of allochthony (Catalán et al.

2013; Lambert et al. 2015; Yamashita et al. 2010). Additionally, the degree of allochthony within

an ecosystem, and thus, the connections between inland aquatic ecosystems and their

surrounding catchments, can be determined by the ratio of allochthonous carbon inputs to

autochthonous carbon production (Carpenter et al. 2005). Thus, analyses of DOM have potential

to improve our understanding of mechanisms behind human-induced alterations of aquatic

ecosystems, as it reflects changes in land cover and affects pivotal ecological processes as

ecosystem respiration. Furthermore, absorption and fluorescence excitation-emission matrices

(EEM) analyzed by parallel factor analysis (Figure 2) (PARAFAC) identifies independently

fluorescing DOM components (Cory and McKnight 2005). In essence, PARAFAC is a technique

that decomposes the fluorescence characteristics of dissolved organic matter (DOM) into

distinct components, concurrently gauging the proportional contribution of each component to

the overall DOM fluorescence (Stedmon and Bro, 2008; Fellman et al., 2010). Subsequently, the

PARAFAC components furnish insights into the source, chemical composition, and

biogeochemical significance of aquatic DOM (Fellman et al., 2010).

CHAPTER 1

Figure 2 Example of Dissolved organic matter (DOM) components identified by PARAFAC in Berlin

surface waters. Em=emission Ex= Excitation. Component C1: humic-like and recalcitrant, C2: terrestrial

humic-like in waste water treatment impacted water, C3: humic-like, C4: terrestrial humic-like, suggested

as photo-refractory, C5: anthropogenic, microbial humic-like, C6; protein-like, linked to autochthonous

production and C7: protein-like, waste water treatment origin.

Beyond the valuable biogeochemical information provided by optical measurements, these

techniques are preferentially used in DOM monitoring because their cost-effectiveness and

rapidity, especially in comparison to analyses conducted at the molecular level (Coble 2007;

Fellman et al. 2010). When a deeper understanding of the molecular composition of DOM is

necessary, an alternative approach is the characterization of the proportions of various size

fractions of DOM through liquid chromatography–organic carbon detection (LC-OCD; Huber

et al., 2011). LC-OCD is capable of identifying distinct size classes, including biopolymers,

humic-like substances, building blocks, low molecular weight acids, humic-like substances, and

low molecular neutrals (Huber et al. 2011b). Notably, it categorizes high molecular weight

substances (biopolymers), humic-like substances (humic-like substances and building blocks),

and low molecular weight substances (low molecular weight acids and humic-like substances

and low molecular neutrals). In recent times, the utilization of advanced high-resolution mass

spectrometry techniques, such as Fourier transform ion cyclotron resonance mass spectrometry

(FT-ICR MS) and Orbitrap mass spectrometry allows the ultimate identification of specific

biomolecules conforming DOM through the assignment of molecular formulae to mass to

charge ratio (m/z) peaks derived from mass spectrometry (REF). These techniques have

contributed to unveile the intricate nature of DOM within certain environmental sample,

revealing associations between compositional turnover of DOM differing in molecular diversity

and landscape-scale environmental gradients in lakes (Kellerman et al. 2014) and rivers (Peter

et al. 2020).

General Introduction

1.3 Ecosystem metabolism

Ecosystem metabolism consists in two processes: the production and mineralization of organic

matter by photosynthesis and respiration, respectively (Odum 1971). Ecosystem metabolism is

thus considered the keystone of ecosystem functioning in aquatic systems (Venkiteswaran et al.

2015) since it is primarily linked to energy fluxes and the cycling of nutrient and organic matter

(Izagirre et al. 2007).

In aquatic ecosystems, gross primary production (GPP) is commonly estimated as the rate of O2

generation by photosynthesis (Falkowski and Raven 1997). In contrast, ecosystem respiration

(ER) is the rate of oxidation of organic carbon to inorganic compounds by both heterotrophic

and autotrophic organisms. Thus, the evolution and consumption of dissolved O2 are linked to

carbon flow in aquatic ecosystems(Carignan et al. 2000). The balance between GPP and ER is

called net ecosystem productivity (NEP) (Kemp et al. 1997). In ‘heterotrophic systems’, ER

exceeds GPP because allochthonous organic matter is mineralized in addition to organic carbon

produced by GPP. In ‘autotrophic systems,’ GPP exceeds ER, resulting in the export or burial

of organic matter produced from the conversion of carbon dioxide and inorganic nutrients

(Raymond et al. 2000). Consequently, ecosystem metabolism is considered a useful metric or

functional indicator to assess ecosystem health, responding to both natural and anthropogenic

disturbances at scales ranging from local to global (Fellows et al. 2006; Mulholland et al. 2005;

Williamson et al. 2008). Additionally, stream metabolism has the advantages as functional

indicator of involving various trophic levels and its sensitivity to a diverse array of abiotic and

biotic factors, which offer an integrated perspective of human disturbances over time and across

various habitat types (Gücker 2009; Mulholland et al., 2005). Several local conditions, such as

riparian vegetation (Figure 3) and point source pollution, affect GPP by affecting the availability

of light and nutrients, the main drivers of photosynthesis (Mulholland et al. 2001), but also ER

through changes in temperature and organic matter availability (Sinsabaugh 1997). Regional

conditions such as land use can also be important in the control of stream metabolism by the

indirect influence on hydrology and light availability (Figure 3) (Houser et al. 2005). For

instance, a recent study demonstrated that flow and light regimes are the primary controls on

the timing and magnitude of river ecosystem GPP and ER (Bernhardt et al. 2022). Likewise, the

existence of buildings in urban environments is likely to influence primary production in urban

aquatic systems through shading. Certain studies have also explored the effects of organic

matter and nutrients originating from Waste Water Treatment Plants (WWTPs) on primary

production. Furthermore, agricultural land use has been shown to affect stream ecosystem

metabolism, leading to increased GPP and ER due to greater availability of light and nutrients

(Sweeney et al. 2004; Young and Huryn 1999). The replacement of wildlife by livestock can also

have an influence on the aquatic metabolism due to cascading effects of the alteration on resource

subsidies into food webs (Dawson et al. 2016; Masese et al. 2015; Subalusky et al. 2018), but also

CHAPTER 1

by altering overall oxygen dynamics, biogeochemistry, and community composition throughout

the entire river (Dutton et al. 2018a; Masese et al. 2018).

Figure 3 Conceptual graph that illustrates the drivers of stream ecosystem metabolism across varying

spatial scales. The regional template is anticipated to influence climate, vegetation, and topography.

Factors at the watershed scale play a crucial role in determining nutrient availability and the hydrologic

regime. Concurrently, local-scale riparian canopy characteristics exert the strongest control over

terrestrial organic matter (OM) and light conditions. Seasonality is projected to impact all these factors,

potentially causing shifts in the interplay between watershed and local-scale controls on Gross Primary

Production (GPP) and Ecosystem Respiration (ER). The size of the arrows approximately reflects the

magnitude of influence. Large mammals replacement by cattle might be another local-scale control, which

might affect the OM composition. Modified from (Alberts et al. 2017).

1.4 Greenhouse gases (GHGs) in aquatic ecosystems

GHGs dynamics in aquatic ecosystems arise from the combination of aquatic production

through respiration and the import of inorganic carbon from the surrounding terrestrial

ecosystems in the catchment. The consumption (by respiration) and production (by primary

production) of organic matter result in the production and consumption of carbon dioxide. In

the absence of O2, organisms carry out anaerobic respiration, which can produce CH4 by

methanogenesis. Both CO2 and CH4 are emitted to the atmosphere through diffusion. For CH4,

besides diffusive flux, ebullition or aerenchymatic transport through aquatic plants are

important emission pathways (Bastviken et al. 2023). The total CH4 emissions to the atmosphere

depend not only on the anoxic production but also on the losses by oxidation when CH4 passes

the oxygenated water column (Borrel et al. 2011).

About 71% of the earth's surface is covered by water, but only 3% of the earth's water is

freshwater (USBR 2023). However, freshwaters can still be considerable sources of CO2 and

General Introduction

CH4, with a considerable importance in the global carbon cycle. Since the first estimations,

global numbers have changed a lot (Table 1). Compared to fossil fuel combustion, this natural

component of the global carbon cycle amounts to approximately 31% of the annual CO2

emissions (IPCC 2013) and almost half of global CH4 emissions to the atmosphere (Rosentreter

et al. 2021).

Table 1 Estimates of aquatic carbon fluxes (Pg) from Drake 2018. Black values indicate an independent

estimate was provided by the given study. Gray values were not refined by the given study but indicate

where an estimate was applied from previous or future study.

Study

Exported

to ocean

Outgassed

Stored

Photosynthesis

Input

from

land

Cole et al. 2007

0.9

0.75

0.23

0.3

1.1

Battin et al. 2009

0.9

1.2

0.6

0.3

1.9

Tranvik et al. 2009

0.9

1.4

0.6

0.3

2.1

Bastviken et al. 2011

0.9

1.48

0.6

0.3

2.2

Regnier et al. 2013

0.95

1.2

0.6

0.3

2.5

Raymond et al. 2013

0.95

2.18

0.6

0.3

3.4

Borges et al. 2015

0.95

2.78

0.6

0.3

Holgerson and Raymond

2016

0.95

3.06

0.6

0.3

4.3

Sawakuchi et al. 2017

0.95

3.88

0.6

0.3

5.1

Estimations of CO2 and CH4 fluxes in aquatic ecosystems are prone to large uncertainties, which

hampers the estimation of global GHG emissions. For CO2, this is partly due to the widespread

but rather inaccurate use of alkalinity data to derive CO2 evasion estimates. For CH4, emission

estimates are biased towards lakes, since studies on streams and rivers are scarce (Bodmer et al.

2016; Rocher-Ros et al. 2023). Additionally, seasonal variability of the fluxes has not been

determined at the global scale, limiting our ability to understand global-scale variation and

controls (Liu et al. 2022). Another source of uncertainty is the mostly ignored daily and event-

driven variability of GHG emissions, as because of practical reasons most of the measurements

have been conducted during the day. Indeed, some recent studies have shown that CO2 fluxes

from rivers are higher during night than day (Gómez-Gener et al. 2021).

There are several ways to measure CO2 and CH4 fluxes. For CO2, the most commonly used direct

methods are the boundary layer method (BLM), floating chamber (FC) and eddy covariance (EC)

measurements. For CH4, BLM and FC can also be used to determine diffusive fluxes, and gas

traps for ebullition. In the FC method, a chamber equipped with a CO2 sensor is deployed in the

field long enough to reach equilibrium between the chamber headspace and the water, this

concentration at equilibrium is used to determine the flux based on the change of concentration

CHAPTER 1

in the chamber headspace during a short period (Livingston and Hutchinson 1995), EC

measurements provide lake flux estimates over large areas (Aubinet et al. 2012), but they

require extensive post processing of the data (Erkkilä et al. 2018). To determine ebullitive CH4

flux, gas traps such as inverted funnels are placed above the sediment surface (Casper et al. 2000)

or water surface (Männistö et al. 2019) to capture the CH4 bubbles released from the sediments.

Most of the global estimates are based on the BLM (Cole and Caraco 1998), which uses the air-

water surface concentration gradient and the gas transfer velocity (generally calculated from

wind or water velocity). The concentration in water is very commonly indirectly calculated

using carbonate equilibria (Butman and Raymond 2011; Lauerwald et al. 2015; Park et al. 2018;

Raymond et al. 2013), which can lead to large errors. With this method, commonly measured

water parameters such as pH, alkalinity, dissolved inorganic carbon (DIC) and water

temperature, are used to estimate pCO2 (Lewis et al. 1998; Park 1969). The accuracy of the pCO2

calculation needs to be improved, by using corrective measures when there are large changes in

pH (Liu et al. 2020).

Human activities affect both CO2 and CH4 fluxes in various ways. For example, in the African

savannah, livestock watering points in rivers alter river ecosystems, which may affect

greenhouse gas emissions along the river continuum (Mwanake et al. 2022). Agricultural

streams present higher emission rates than forested streams (Bodmer et al. 2016). Urbanization

may affect both aquatic CO2 and CH4 fluxes due to the increases in human population densities,

wastewater effluents, change in runoff and increases in nutrient and organic carbon supplies

(Tang et al. 2021). Numerous studies have shown that urban water bodies are a significant

source of CO2 (Park et al. 2018; Yu et al. 2017) and CH4 to the atmosphere (Gonzalez-Valencia

et al. 2014; Martinez-Cruz et al. 2017). Variability in GHG emissions has been associated with

river sizes and their connectivity to terrestrial ecosystems (Hotchkiss et al. 2015; Raymond et

al. 2013; Rosamond et al. 2012). Some studies have suggested that agricultural run-off

contributes to increased GHG emissions from rivers (Smith et al. 2017), while recent findings

indicate that urban infrastructure may also play a role in elevated GHG emissions from urban

rivers (Gallo et al. 2014; Kaushal et al. 2014a). However, our understanding of environmental

controls of GHG emissions from stressed urban water bodies is still limited. A recent global

study on fluvial systems identified human population density as an important factor controlling

GHG emissions (Rocher-Ros et al. 2023). However, lakes, ponds, reservoirs and impoundments

and strongly modified systems were excluded from the study, leaving a major information gap.

General Introduction

1.6 Objectives

The present dissertation was designed to advance understanding of the impacts of human

activities on aquatic carbon dynamics. I analyzed DOM composition as a central link between

human activities and the aquatic carbon cycle. In particular, I studied how DOM is affected by

human pressures through land use change by urbanization, taking the example of a European

city, and the replacement of hippos by livestock in the African savannah. In addition, I assessed

how changes in DOM composition shape the metabolism of aquatic ecosystem and GHG fluxes.

Each of four specific objectives were addressed in a dedicated chapter (Figure 4):

1. To find out the effects of large wildlife displacement by livestock on DOM composition and

ecosystem metabolism, due to differences in the quantity and composition of faeces from

wildlife and livestock (chapter 2).

2. To discover the spatiotemporal patterns of DOM composition across a range of urban

freshwaters encompassing streams, rivers, ponds and lakes, and to identify major

environmental drivers (chapter 3).

3. To estimate carbon dioxide fluxes from an urban surface water network, determine

variability of the fluxes across the water-air interface at daily and seasonal time scale, identify

major drivers, and extrapolate the measured CO2 fluxes to the whole city of Berlin (chapter 4).

4. To quantify the net methane emissions from urban water bodies at the site and whole-city

scale and identify the underlying drivers (chapter 5).

Figure 4 Conceptual graph summarizing research included in this dissertation. Chapter 2 focuses on the

influence of human activities on African savannah rivers by analyzing the effects of replacing

hippopotamus by livestock in the catchment. Chapters 3, 4 and 5 address the influence of urban

development on carbon dynamics in contrasting aquatic ecosystems of a large European city. DOM =

Dissolved Organic Matter. Ch. = Chapter.

CHAPTER 2

Hippopotamus are distinct from domestic

livestock in their resource subsidies to and

effects on aquatic ecosystems

This study was published as:

This is the postprint version of the article.

2.1 Abstract

In many regions of the world, populations of large wildlife have been displaced by livestock, and

this may change the functioning of aquatic ecosystems owing to significant differences in the

quantity and quality of their dung. We developed a model for estimating loading rates of organic

matter (dung) by cattle for comparison with estimated rates for hippopotamus in the Mara River,

Kenya. We then conducted a replicated mesocosm experiment to measure ecosystem effects of

nutrient and carbon inputs associated with dung from livestock (cattle) versus large wildlife

(hippopotamus). Our loading model shows that per capita dung input by cattle is lower than for

hippos, but total dung inputs by cattle constitute a significant portion of loading from large

herbivores owing to the large numbers of cattle on the landscape. Cattle dung transfers higher

amounts of limiting nutrients, major ions and dissolved organic carbon to aquatic ecosystems

relative to hippo dung, and gross primary production and microbial biomass were higher in

cattle dung treatments than in hippo dung treatments. Our results demonstrate that different

forms of animal dung may influence aquatic ecosystems in fundamentally different ways when

introduced into aquatic ecosystems as a terrestrially derived resource subsidy.

Masese, F. O., M. J. Kiplagat, C. R. González-Quijano, A. L. Subalusky, C. L. Dutton, D.

M. Post, and G. A. Singer. 2020. Hippopotamus are distinct from domestic livestock in their

resource subsidies to and effects on aquatic ecosystems. Proceedings of the Royal Society B:

Biological Sciences 287:20193000

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

2.2 Introduction

The transfer of organic matter from terrestrial to aquatic environments has often been

understood to be dominated by litterfall and hydrologic transfers during storms and

precipitation events (Wallace et al. 1997; Wantzen et al. 2008). However, it is increasingly

recognized that large mammalian herbivores (LMH) can be major agents of transfer of

terrestrial organic matter and nutrients into aquatic ecosystems (Bond et al. 2014; Subalusky et

al. 2017). While rates vary widely over broad spatial and temporal scales and depend on the

characteristics of the animal vector and the recipient ecosystem (Subalusky and Post 2019; Vanni

2002), the amount can be significant, especially for low-order streams in rangelands and

pastoralist areas (Bond et al. 2014; Masese et al. 2018; Stears et al. 2018).

Terrestrial and aquatic ecosystems in many African savannah landscapes are intricately linked

via the vectoring role that LMH play in transferring large amounts of resources from terrestrial

to aquatic ecosystems (Jacobs et al. 2007; Naiman and Rogers 1997). Pathways of organic matter

and nutrient input into aquatic ecosystems by LMH include egestion and excretion during

migrations and watering (Hayward and Hayward 2012), facilitation of soil erosion (Jacobs et al.

2007) and drowning during water crossings (Subalusky et al. 2017). A prominent example is the

common hippopotamus (Hippopotamus amphibius, hereafter hippo), which migrates daily

between savannah grasslands, where it forages, and aquatic ecosystems where it rests and much

of its defaecation occurs (Subalusky et al. 2015). Resource subsidies from hippos alter primary

production and secondary production, most prominently through direct consumption by

bacteria, invertebrates and fish (Dawson et al. 2016; Masese et al. 2015; Subalusky et al. 2018),

and influence whole river oxygen dynamics, biogeochemistry and community composition

(Dutton et al. 2018b; Masese et al. 2018; Stears et al. 2018).

The expansion of human settlements, crop farming and conversion of forests and savannah

grasslands to pasture for livestock production have contributed to the loss of large populations

of wild LMH around the world (Kinnaird et al. 2003; Prins 2000; Ripple et al. 2015). In many

African savannahs, large populations of wild herbivores still dominate the biomass of

conservation areas (du Toit 2003; Young et al. 2013). However, even in these regions, wild LMH

are declining concurrently with increases in livestock such as cattle, goats and sheep (Ogutu et

al. 2016; Prins 2000). In most areas where livestock have replaced wildlife on the landscape,

their influence on aquatic systems has often been seen as negative, with research focusing on

habitat degradation, nutrient and organic matter loading and microbial contamination (Belsky

et al. 1999; Bond et al. 2014). However, livestock may take over some of the ecological roles

historically filled by wild LMH, thereby maintaining the functionally important linkage of

riverine ecosystems to their surrounding terrestrial landscapes. The degree to which ecosystem

effects of this functional linkage from livestock are similar to those from wild herbivores depends

in part on the similarity of the resource subsidies they transport.

CHAPTER 2

Ruminants such as cattle and sheep have a relatively efficient digestive system compared with

non-ruminants such as hippos and horses, and this difference in digestion produces smaller faecal

particle sizes in ruminants (Fritz et al. 2009; Thomas and Campling 1977). Non-ruminants, such

as hippos, have longer mean retention times than ruminants, which enhances nutrient extraction

from ingesta, leading to a greater ratio of C to nutrients, reflecting relatively lower quality of

their dung (electronic supplementary material, Table S1). Ruminants also forage on a broader

selection of plant species compared with nonruminants (Field 1970; Noirard et al. 2004), and by

so doing they ingest a wider variety of metabolites and chemicals (Webster et al. 1999), which

may result in differences in the chemical composition of dung and its leachate, and consequently

its effect on ecosystem processes. Differences in particle size and composition are likely to

influence the way in which dung inputs from ruminants and non-ruminants influence aquatic

ecosystems. Dung comprising large particles with a high ratio of C to nutrients, as expected

from non-ruminants such as hippos (electronic supplementary material, Table S2), is

qualitatively similar to the seasonal input of leaf litter to aquatic ecosystems in temperate forests

(Subalusky et al. 2018; Webster et al. 1999). These inputs are expected to deposit in the benthos

as relatively refractory material that increases ecosystem respiration and is incorporated into

the detrital food web. Dung comprising small particles that are relatively high in nutrients, as

expected from ruminants such as cattle, is expected to remain suspended in the water column,

which could decrease light penetration, but also be more likely to increase both water column

and benthic primary production (Garg and Bhatnagar 1999). Furthermore, the addition of

nutrient-rich ruminant dung from cattle to aquatic ecosystems already receiving large inputs of

carbon-rich non-ruminant dung from hippos may lead to interactions between the two subsidies

in decomposition rates and ecosystem effects (Kominoski et al. 2015). The incremental

displacement of wild herbivore populations by livestock raises the question of how this change

may impact the transfer of nutrients and organic matter into inland waters and the ensuing

ecosystem responses.

The Mara River and its seasonal tributaries in the Maasai Mara National Reserve (MMNR) in

Kenya host more than 4000 hippos (Kanga et al. 2011). There are also over 250 000 cattle in

communal lands adjoining the MMNR, where livestock coexist with wildlife (Ogutu et al. 2011).

This distribution results in a displacement pattern with hippo areas inside the reserve, mixed

hippo and livestock areas outside the reserve and only livestock grazing areas further away from

the reserve (Veldhuis et al. 2019). This overlapping distribution of livestock and wildlife raises

the question of how aquatic ecosystems will respond to the displacement of wild LMH by

livestock as agents of resource transfer from terrestrial to aquatic environments. Here, we

characterize the particle size and stoichiometry of cattle and hippo dung, estimate the loading

of organic matter by cattle and hippos into the Mara River and conduct an experiment in

recirculating experimental stream mesocosms to test the impacts of these different inputs on the

function of aquatic ecosystems. Previous research has already shown that the quantity of inputs

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

by LMH has substantial impacts on the aquatic ecosystem (Dutton et al. 2018b; Subalusky et al.

2018). For our experiment, we used a replacement design to compare ecosystem effects of cattle

dung and hippo dung both independently and in combination with one another. We measured

the effects of both hippo and cattle dung on nutrients, dissolved organic carbon (DOC) quantity

and quality, gross primary production (GPP) and ecosystem respiration (ER). We hypothesized

that cattle dung inputs would lead to higher nutrient concentrations and increased GPP, while

hippo dung inputs would lead to higher concentration of DOC and increased ER. Furthermore,

we hypothesized that these parameters may change nonlinearly along a gradient of low to high

subsidy quality––that is, from a system dominated by hippo dung to one dominated by cattle

dung––depending on the strength of the interaction between the high C and high nutrient

inputs. We further hypothesized that cattle dung would lead to a more diverse DOC composition

given the broad foraging strategy of cattle compared with hippos.

2.3. Material and methods

2.3.1 Characteristics of cattle and hippo dung

Macro- and micronutrient composition of cattle and hippopotamus faecal samples were analysed

at the Leibniz Institute for Zoo and Wildlife Research, Berlin, Germany. Before analysis, dried

samples were ground to a particle size of about 1 mm. For C and N, samples were weighed and

analysed on an elemental analyser (Hekatech, Thermo Finnigan). For P, samples were weighed,

ashed in a muffle furnace at 550°C and then digested before analysis on an inductively coupled

plasma optical emission spectrometer (ICP-OES) (PerkinElmer, Ueberlingen, Germany). Crude

protein was calculated as 6.25 ×N. Carbohydrates (sucrose, D-glucose, D-fructose and starch)

were analysed using commercial enzymatic test kits from R-Biopharm (Darmstadt, Germany).

For mineral analysis (Ca, Mg, Fe, K), samples were microwave digested and analysed by a

PerkinElmer ICP-OES.

2.3.2 Livestock versus hippopotamus loading rates of organic matter

We developed a model to estimate cattle loading rates of organic matter (dung) into the Mara

River (electronic supplementary material S1) and compared results with existing estimates of

loading rates for hippos in the river extracted from Subalusky et al. 2015. We used literature to

estimate daily dry matter intake (DMI) of zebu cattle, and the proportion of organic matter

(OM) egested or excreted. We estimated cattle loading rates of OM as a fraction of time spent

in the river, and we multiplied the per capita loading rate by the cattle population to get the

total loading rates for all cattle. We then compared the loading of cattle and hippopotamus dung

in two areas of the Mara River where their distributions overlap.

CHAPTER 2

2.3.3 Experimental mesocosms

We used experimental stream mesocosms constructed out of PVC canvas measuring 4.2 m long

and 19 cm wide (Subalusky et al., 2018). Water was recirculated in each mesocosm by

paddlewheels affixed to a shaft that was powered by a motor, with each of three shafts handling

six streams (electronic supplementary material S2). We had three replicates for each of six dung

treatments in a replacement design ranging from 100% hippo dung to 100% cattle dung with

20% increments of replacement (electronic supplementary material, Figure S1). This approach

allowed us to test for potential interactive effects between dung types, recognizable by nonlinear

responses to the dung treatment gradient. Treatments were randomly distributed among

mesocosms, with a replicate of each treatment in each of the three blocks. A total of 120 g (wet

weight, 1.7 g l−1) of dung was distributed in each mesocosm once at the beginning of the

experiment in order to study ecosystem responses arising from differences in dung quality due

to nutrient leaching and mineralization rates. To accelerate biofilm growth, mesocosms were

inoculated with periphyton scraped off rocks from the Amala River, a tributary of the Mara

upstream of wildlife. Each mesocosm was lined with six unglazed ceramic tiles that were used

for weekly sampling of biofilms. Each week, one tile from each mesocosm was destructively

sampled without replacement for analysis of ash-free dry mass (AFDM).

2.3.4 Water sampling and analysis

We collected water samples weekly, including day 1, for analysis of ammonium (NH4+), soluble

reactive phosphorus (SRP), total phosphorus (TP), nitrite (NO2-), nitrate (NO3-), total suspended

solids (TSS), particulate organic matter (POM), dissolved organic carbon (DOC) concentration

and composition, and chlorophyll a (Chl-a). Further details on analysis of nutrients and DOC,

Chl-a, TSS and POM concentrations are available in electronic supplementary material S3. We

characterized DOC by absorbance and fluorescence analyses, which provide proxies for

DOCsource and/or biological availability (Fellman et al. 2010; Jaffé et al. 2008). To characterize

DOC, we used parallel factor analysis (PARAFAC) to decompose 349 excitation–emission

matrices (EEMs) into fluorescent components (Stedmon and Bro 2008), and size-exclusion

chromatography (SEC) (Huber et al. 2011b), which separates three size fractions: humic

substances (HS), high molecular weight non-humic substances (HMWS) and low molecular

weight substances (LMWS). Further details on collection and analysis of DOC composition data

are available in the electronic supplementary material S4.

2.3.5 Metabolism

In each mesocosm, we recorded dissolved oxygen (DO) and water temperature every 1 min for

six weeks with MiniDOT loggers (PME, Vista, CA, USA). Light intensity (0 to 320 000 lux)

was recorded with HOBO Pendant Temperature/Light Data Loggers (UA-002-64; Onset,

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Bourne, MA, USA). We then estimated GPP and ER from diel changes of DO, temperature and

irradiance using an inverse modelling procedure that included temperature- dependent ER,

light-dependent GPP and reaeration (Fuss et al. 2017); details are provided in electronic

supplementary material S5.

2.3.6 Data analysis

We used linear mixed effect models (LMEMs) to test the effect of dung treatment on response

variables DOC, Chl-a, AFDM, TSS, POM, SRP, NH4+, NO2- and NO3- with the lme function in

the ‘nlme’ package (Pinheiro et al., 2016) in R (R Core Team, 2017). LMEMs were used after

residuals displayed linear responses to dung treatment. LMEMs included dung treatment (six

levels) and time (week 0 to 6) as fixed effects and block as a random effect. We also included an

interaction of dung treatment with time to test for differences in the temporal dynamics of

responses. Response variables were log-transformed when appropriate to meet normality

assumptions. We ran a separate model for each variable and computed marginal R2 (R2 m,

variance explained by fixed factors) and conditional R2 (R2c , variance explained by the entire

model, i.e. by fixed and random factors) coefficients with the r.squaredGLMM function in the

MUMIN package (R Core Team, 2017). To test the effect of cattle dung on ecosystem

respiration, or production, we fitted a three-parameter sigmoid Gompertz model (Gompertz

1825), given by

𝑌(𝑡)=𝐾 ∙𝑒−𝑙𝑎𝑔∙𝑒−𝑟𝑎𝑡𝑒∙𝑡,

to daily GPP and ER for each dung treatment (i.e. with data pooled across three streams) and

separately for each replicate stream; this yielded estimates for the upper asymptote (K), growth

rate (rate) and a dimensionless parameter for location along the time axis (lag), which shifts the

graph to the left or right and is related to the time taken to reach the upper asymptote (maximum

GPP or ER), with high values indicating faster progression towards maximum GPP or ER. Y(t)

is the expected value (GPP or ER) as a function of time (days since start of the experiment) and

t is time in days. These parameter estimates were then regressed against dung treatment (%

cattle dung) using general additive mixed modelling (GAMM) to avoid strong assumptions

about potential relationships. GAMMs were built using penalized cubic regression splines with

degrees of freedom automatically identified based on the generalized cross-validation score

(GCV). Further, to investigate weekly changes in ecosystem metabolism (GPP, ER, GPP/ER

and net ecosystem production (NEP)), weekly means for each stream were computed (total six

weeks). We then tested for differences among dung treatments using GAMMs (Zuur et al.,

2007), and included dung treatment as a fixed effect and block as a random effect. GAMMs were

fitted using the R package mgcv (Wood and Wood 2015). Principal component analysis (PCA)

was used for dimension reduction of DOC quality data (absorbance- and fluorescencederived

indices FIX, β/α, humification index (HIX); PARAFAC components C1 to C7; SEC results

CHAPTER 2

HMWS (in %), HS (in %), LMWS (in %)). While optical indices and SEC results are expressed

as ratios or percentages and thus describe composition with little or no influence of DOC

quantity, PARAFAC components were used in the form of quantitatively reliable absolute

fluorescence intensities. All variables were scaled to zero mean and unit s.d. prior to use in PCA.

Statistical analyses were performed with R version 3.3.1 (R Core Team, 2017) using the

packages vegan (Oksanen et al., 2013), sem (Fox 2006) and deSolve (Soetaert 2010).

2.4. Results

2.4.1 Characteristics of cattle and hippo dung

Cattle dung had lower C:N: P ratio than hippo dung (electronic supplementary material, Table

S3). C:N: P was 155.2 : 5.1 : 1.0 for cattle dung and 261.4 : 7.6 : 1.0 for hippo dung (electronic

supplementary material, Table S3); per unit C, cattle dung was thus richer in N and P than hippo

dung by 13 and 69%, respectively. Further, cattle dung was enriched in the micronutrients Ca,

Fe, K and Mg by 3.6–31% (electronic supplementary material, Table S3).

2.4.2 Loading rates of organic matter by cattle and hippopotamus

We estimated that cattle spend 10 min in the Mara River or tributaries per day, and each on

average loads 22.3 g DM (86.6 g wet mass) (kg body mass)−1day−1 into the river (electronic

supplementary material S1). Thus, an average animal (265 kg) defaecates 12.5 kg faeces (wet

mass) every day, and 0.0866 kg (0.70% of daily defaecation) goes into the Mara River. In

comparison, an average hippo (1500 kg) defaecates 17.4 kg faeces (wet mass) every day, and 8.7

kg (50%) goes into the Mara River (Subalusky et al. 2015). Using cattle population estimates

(Lamprey and Reid 2004; Ogutu et al. 2011; Ogutu et al. 2016), we estimated the total daily

loading was 1157 kg faeces into the Mara River inside the MMNR, 2599 kg outside the MMNR

and 7364 kg along the Talek River. Within the MMNR, livestock loading is only around 6% of

loading due to cattle and hippopotamus because cattle are not supposed to be grazed in the

reserve, but illegal grazing of a small number of cattle nevertheless occurs. Outside the MMNR,

where the Maasai pastoralists keep large numbers of cattle, loading by cattle along the Mara

and Talek rivers increases to nearly 16 and 57%, respectively, of the total organic matter loading

due to cattle and hippopotamus. These loading rates are based on the assumption that all cattle

within either the Mara or Talek sub-catchment visited the river for watering or crossing at least

once per day. However, some cattle may use water pans for their water needs during certain

portions of the year.

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

2.4.3 Nutrients

Dung treatment had a significant effect on nutrient concentrations, with a significant increase

in SRP, ammonium, nitrite and nitrate with increasing proportion of cattle dung (Table 1;

electronic supplementary material, Figure S2). There was also a significant effect of time on

nutrient concentrations, reflecting different rates of leaching, uptake and retention in biomass

(Table 1). Notably, there was a greater than 90 and 56% reduction in SRP and ammonia,

respectively, within the first two weeks across all treatments (electronic supplementary material,

Figure S3a,b). Similarly, nitrite significantly declined after the second week, while nitrate

increased (electronic supplementary material, Figure S3c,d; Table 1).

2.4.4 Organic matter and biomass

Dung treatment had a significant effect on organic matter (DOC and POM), TSS, water column

Chl-a and biomass of biofilms (AFDM) (Table 1). There was a linear increase in these variables

from low to high proportion of cattle dung (electronic supplementary material, Figure S4; Table

1). DOC concentration was considerably higher with the presence of cattle dung through the

entire experiment (electronic supplementary material, Figure S4a). There was also a significant

effect of time on these variables (Table 1). DOC concentrations increased by greater than 50%

over the experimental period (electronic supplementary material, Figure S4a), and Chl-a,

AFDM, TSS and POM increased by greater than 100% (electronic supplementary material,

Figure S4b–e). Further, we observed that the smaller particles of cattle dung (Fritz et al. 2009;

Thomas and Campling 1977) remained suspended in water while those from hippo dung sank

to the bottom, which was reflected by higher TSS and POM in the water column with higher

proportions of cattle dung (electronic supplementary material, Figure S4d,e).

2.4.5 DOC composition

The PARAFAC model consisted of seven components (electronic supplementary material S4,

Table S4 and Figures S5 and S6): four humic-like, one reduced humic-like and two protein-like

fluorescence components. The first two PCA axes explained 32.5 and 27.0% of the total variance,

respectively, and efficiently depicted treatment differences and development of DOC

composition throughout the experimental time (Figure 3). The 100% cattle treatment was

clearly separated from all other treatments and furthest from the 100% hippo treatment, in

particular along PC1. By contrast, PC2 was more important for capturing temporal changes,

but also contributed to definition of distinct treatment specific DOC composition at the

experiment start. At the start of the experiment, all dung treatments produced DOC with the

highest share of low molecular weight substances and rich in aromatic structures and humic

substances indicative of leaching from plant material (Figure 3a,b). DOC from cattle dung was

more fluorescent and humic compared with hippo dung, which, by contrast, had seemingly

CHAPTER 2

fresher DOC with relatively N-deficient high molecular weight substances (Figure 3a,b). Over

the duration of the experiment DOC composition in all dung treatments changed in parallel

towards a common endpoint of higher concentrations of less fluorescent, less humic and less

aromatic DOC (Figure 3a,b). High molecular weight substances with high C:N, likely

carbohydrates from primary production, became more important towards the end of the

experiment. Notably, DOC in the 100% cattle dung treatment experienced a strong and long-

lasting phase of humic fluorescence buildup before rejoining the other treatments0 trend. In an

effort to quantify overall compositional dynamics, we summed Euclidean path lengths from the

start to the end of the experiment in the multivariate space described by all PCs. While the 0

and 20% cattle dung treatments here resulted in DOC with minimal turnover, the 60% cattle

treatment had the highest compositional turnover of DOC (Figure 3c). This suggests the sum

of two processes––leaching from the dung and autochthonous production––cause very dynamic

DOC in treatments with a higher proportion of cattle dung.

2.4.6 Ecosystem metabolism

Dung treatment strongly influenced temporal trends in GPP and ER (electronic supplementary

material S5). There was a significant effect of dung treatment on the maximum production value

(K) and the rate of increase in GPP; as the proportion of cattle dung increased, GPP increased

slower but reached a higher maximum (Figure 1a,b). GPP increase also started later with less

cattle dung (Figure 1d), but this lag effect was insignificant owing to excessively high parameter

estimates for two streams (0% cattle dung) that had poor data coverage in the first two weeks.

Notably, while maximum GPP increased linearly with dung treatment, the rate of increase and

the lag changed nonlinearly with dung treatment and suggested stronger changes when even

only small fraction of hippo dung was replaced by cattle dung. Because of the use of different

theta values for temperature dependence of ER in metabolism models (Demars et al. 2015;

Perkins et al. 2012; Sand-Jensen et al. 2007), we used a higher modelled value of 1.1085 since

our attempts to use a common value of 1.045 (Riley and Dodds 2013) were unsuccessful. We

subsequently performed a sensitivity analysis to compare results of using both theta values on

the findings, and concluded that the response in GPP and ER to dung treatment remained

unchanged (electronic supplementary material, Figures S7–S9).

Table 1 Results of mixed-effects models for loge(Χ)-transformed dissolved organic carbon (DOC, mg l−1), chlorophyll a (Chl-a, mg l−1), ash-free dry mass (AFDM, mg

cm−2), total suspended solids (TSS, mg l−1), particulate organic matter (POM, mg l−1), soluble reactive phosphorus (SRP, μg l−1), total phosphorus (TP, mg l−1),

ammonium (mg l−1), nitrite (mg l−1) and nitrate (mg l−1). The marginal R2 (GLMM(m); fixed effects only) and the conditional R2 (GLMM(c); fixed and random

effects) represent the proportion of variance explained by each model; s.e. = standard error; s.d. = standard deviation; *p < 0.05, **p < 0.01, ***p < 0.001

fixed effects

SRP

ammonium

nitrite

nitrate

DOC

Chl-a

AFDM

TSS

POM

(s.e.)

intercept

0.87

(0.08)***

0.1

(0.02)***

0.14

(0.01)***

0.39

(0.05)***

1.17

(0.03)***

1.11

(0.05)***

0.79

(0.07)***

1.08

(0.04)***

0.32

(0.10)***

dung

treatment

0.004

(0.001)**

0.001

(0.0002)*

0.001

(0.0002)***

0.002

(0.001)*

0.002

(0.001)**

0.003

(0.001)***

0.003

(0.001)*

0.002

(0.001)*

0.003

(0.002)*

time

−0.19

(0.02)***

−0.02

(0.05)***

−0.02

(0.003)***

0.03

(0.01)***

0.04

(0.004)***

0.22

(0.02)***

0.12

(0.02)***

0.11

(0.01)***

0.27

(0.03)***

dung

treatment x

time

0.001

(0.0003)*

0.0002

(0.0001)*

0.0002

(0.00004)***

—

−0.0004

(0.0002)*

—

ANOVA for

fixed effects

F-value

intercept

58.1***

19.6***

100.0***

630.7***

7986.4***

4265.4***

1938.7***

6273.1***

857.7***

dung

treatment

4.6*

6.5**

27.9***

11.6**

21.3***

30.4***

40.0***

42.9***

33.1***

time

161.5***

17.4***

76.1***

43.3***

103.2***

33.7***

89.3***

285.3***

263.4***

dung

treatment x

time

6.9*

6.3*

20.9***

1.3

0.4

1.2

1.3

4.8*

1.7

random

effects

s.d.

block

(intercept)

0.03

<0.001

residuals

0.26

0.06

0.038

0.13

0.09

0.24

0.25

0.12

0.34

GLMM(m)

0.58

0.2

0.51

0.34

0.51

0.74

0.47

0.72

0.71

GLMM(c)

0.58

0.2

0.55

0.46

0.61

0.75

0.47

0.76

0.71

CHAPTER 2

Figure 1 Dynamics of GPP (a) and ER (e) over 44 days as fitted with a three-parameter sigmoid Gompertz

model. To test relationship with dung treatment, we plotted mean and s.d. of upper asymptote K,

maximum rate of increase and lag for Gompertz models for GPP (b,c,d ) and ER ( f,g,h) as a function of

dung treatment, respectively, and fitted a smoothing model (grey line with shaded area represents

smoother mean and s.e.; smoother significance, R2 and GCV are supplied in the figures). Note that

parameter estimates for the smoothing models (n = 3 per treatment) were based on fits to data of

individual flumes. Note also log-scale for lag in (d ) owing to excessive lag in two flumes with 0% cattle

dung.

Figure 2 Weekly measures of flume-scale GPP (a), flume-scale ER (b), GPP : ER (c) and NEP (d ). The

dashed line indicates NEP = 0, and most of the mesocosms were net heterotrophic until day 7 and then

switched.

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Dung treatment did not significantly affect K, rate and lag of ER (Figure 1e–h), though there

was the suggestion of some nonlinear trends in response to dung treatment. Replacing 20–60%

of hippo dung with cattle dung slightly increased the maximum rate of respiration, but the

streams with 80% cattle dung had exceptionally low maximum ER (Figure 1e,f ). The rate of

increase in respiration and the lag parameter showed distinctly nonlinear but insignificant

trends, with lowest values at intermediate dung replacement (Figure 1g,h). Our analysis of

Gompertz parameters did not account for the blocking factor in our experimental design, but

separately computed general additive mixed models did not identify a significant effect of block

on any Gompertz parameter of GPP or ER (data not shown). Using weekly averages of GPP

and ER in GAMMs with a smoother for dung treatment interacting with time, and accounting

for blocks as a random factor, we found a significant weekly increase in GPP with increasing

proportion of cattle dung (Figure 2). There was also a significant interaction between the dung

treatment smoother and time (electronic supplementary material S6, Table S5), further

indicating that the positive effect of cattle dung on GPP increased with time. We also found a

significant main effect of dung treatment on weekly means of ER and a significant interaction

between dung treatment and time (electronic supplementary material, Table S5). GAMMs did

not identify a significant effect of block on GPP or ER (electronic supplementary material, Table

S5). Most streams were heterotrophic during the first week, after which they were all

autotrophic, with NEP peaking at week 4 (28 days; Figure 2a,d).

Figure 3 PCA based on descriptors of DOC. DOC composition changed over time towards a common

endpoint composition when plotting scores (mean ± s.d. per treatment and time) (a). The PCA was based

on PARAFAC components C1 to C7, high and low molecular weight substances (HMWS, LMWS), ratio

of HMWS : LMWS and C : N of HWMS, humic-like substances (HS), aromaticity via specific ultraviolet

absorption at 254 nm (SUVA), humification index (HIX), fluorescence index (FIX), freshness index β : α

(FreshIndex) and an absorbance-based indicator of molecular size (E2 : E3) (b). Stream-specific changes

of DOC composition were quantified as cumulative Euclidean distance in PCA space considering all its

dimensions and progress along a path of consecutive time points; the graph shows average total path

length per treatment (c). Note that the arrows in (a) designate the time series from early to late in

experiment for each treatment, and ‘total DOC dynamics’ in (c) describes the approximate ‘length’ of the

temporal arrows in (a).

CHAPTER 2

2.5 Discussion

Our results show that replacing hippo dung with cattle dung produces different responses in

aquatic ecosystems. The smaller particle sizes and higher quality (lower C :N: P ratio) of cattle

dung compared with hippo dung appeared to promote increased leaching of nutrients and

increased assimilation (Mathuriau and Chauvet 2002; Stohlgren 1988). Indeed, cattle dung

stimulated higher primary production in both the benthos and water column compared with

hippo dung, although the rate of increase in GPP was nonlinear (Figure 1c). Cattle dung also

increased biofilm biomass in the water column, which is a cumulative measure of both microbial

and algal production. However, there were no dung treatment effects on the fitted sigmoid

parameters of ER. Hippo dung, which was composed of larger particles, tended to sink to the

bottom of the streams and reduce benthic production, suggesting it may do the same in aquatic

systems, especially during low flows (Dawson et al. 2016; Dutton et al. 2018b; Subalusky et al.

2018). By contrast, cattle dung tended to remain suspended or dissolve, and in aquatic systems,

it may become dispersed by river discharge into a larger area, creating potentially more

widespread and diffuse effects. Because light is one of the key determinants controlling

production and composition of periphyton or algae in aquatic ecosystems (Mosisch et al. 2001),

it is likely that cattle dung more strongly stimulated the autotrophic component (algae) of

periphyton while hippos stimulated the heterotrophic component (bacteria/fungi), which led to

higher GPP per unit biomass of periphyton among cattle dung treatments. Although we used a

theta value (1.1085) which is different from the typical value (1.045) that is commonly used in

modelling GPP and ER (electronic supplementary material S4), the conclusion reached that

cattle dung stimulated higher GGP values than hippo dung remains unchanged. This is because

the trends and trajectory of change in both GPP and ER are the same for both theta values

(electronic supplementary material, Figures S7–S9). Moreover, ER, which is more temperature

sensitive than GPP (Demars et al. 2015; Perkins et al. 2012), did not respond to dung treatment,

irrespective of the theta value used. This is intriguing and suggests different drivers for GPP

and ER in the experiment. This can be explained by a lack of coupling between GPP and ER,

which explains the increasing concentration of DOC and microbial biomass in cattle dung-

dominated treatments over time. Thus, it is likely that increased ER from heterotrophs in the

hippo dung treatment was offset by the increased autotrophic respiration in the cattle

treatments. If there was any coupling of GPP and ER, some variation in ER would have

occurred, because a proportion of ER is autotrophic respiration (Griffiths et al. 2013; Robert O.

Hall and Beaulieu 2013). There were also effects of dung treatment and treatment by time

interactions in the composition of DOC. In streams that received higher proportions of cattle

dung relative to hippo dung, DOC displayed a strong increase in diversity over time, moving

from a dominance of allochthonous DOC, through a dominance of microbially produced DOC,

to finally a dominance of autochthonously produced DOC from primary production (Figure 3).

DOC in cattle dung treatments also showed a higher contribution of humic-like components

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

associated with microbial activity and high fractions of a fulvic acid-like component of higher

plant material origin (electronic supplementary material S6). The difference in DOC

composition between hippo dung and cattle dung could be due to differences in digestion

efficiency between cattle and hippos (Fritz et al. 2009); it may also result from cattle foraging

on a wider selection of plants and thus encountering a wider variety of metabolites and chemicals

than hippos (Field 1970; Noirard et al. 2004). The strong response of GPP to dung treatment

left a strong imprint on DOC (Fuss et al. 2017). For instance, as GPP peaked over time, DOC

concentration increased in concert, and composition shifted from predominantly allochthonous

towards increasingly autochthonous.

While we recognize that inputs by hippos and cattle likely vary in quantity across time and

space, which impacts how they affect river ecosystem function, this study specifically focused on

comparing the impact of input quality from these two large herbivores. Although there are no

data for African savannahs showing rates of OM and nutrient loading by livestock into rivers,

preliminary findings from the Mara River show that 10–15% of cattle that visit watering points

defaecate and/or urinate in the river (J. Iteba 2019, unpublished data). Our model for estimating

dung inputs by cattle show that only a small fraction of daily dung production by cattle is

deposited directly into the river, compared with 50% of hippo dung. However, owing to variation

in cattle and hippo numbers across the landscape, cattle inputs can range from around 6 to 57%

of total organic matter loading due to cattle and hippos (electronic supplementary material,

Table S1). Our estimates of both hippo and livestock loading have some uncertainty around

them that could be improved with more detailed knowledge of animal time budgets and

population sizes. For example, cattle numbers in the basin can more than double in the dry

season, when livestock are herded in for increased forage (Lamprey and Reid 2004), suggesting

our estimates for cattle loading are conservative. Although the majority of cattle dung is

deposited outside the river, some proportion of it likely enters aquatic systems during large

rainfall events. Furthermore, it is likely that the trend towards increasing populations of cattle

and other livestock, such as goats and sheep, is likely to continue.

This research increases our knowledge about how resource subsidies from cattle may influence

aquatic ecosystem function and highlights similarities and differences between subsidies

transported by cattle versus hippos. Similar to other LMH such as hippos and ungulates (Jacobs

et al. 2007; Subalusky et al. 2015), cattle can create biogeochemical hotspots through

congregation and egestion. However, cattle subsidies are more likely to increase nutrient

concentrations and stimulate primary production in recipient aquatic ecosystems, which could

have pronounced bottom-up effects on food webs. In addition, other aspects of cattle behavior

may have pronounced ecosystem effects. The development of cattle footpaths can channel water

and nutrients from terrestrial to aquatic environments. Large herds of cattle visit watering

points during the dry season and concentrate in riparian areas, where they deposit dung and

CHAPTER 2

urine, which may contribute substantially to nutrient flux at a time when low water runoff limits

fluxes by hydrological vectors. Dung and urine deposited around waterholes also enrich the soil

and vegetation, and this enrichment could increase fluxes of organic matter for aquatic

consumers during inundation and/or litterfall.

2.6 Conclusion

Here we show that cattle and hippo dung have contrasting effects on aquatic ecosystem function,

likely caused by differences in faecal particle size and stoichiometry of major elements (C:N:P

ratio). Increasing inputs of cattle dung led to higher GPP and a more complex and diverse DOC

composition. By contrast, hippo dung reduced benthic primary production and led to a delayed

response in GPP, which is consistent with whole-river observations (Dutton et al. 2018a; Dutton

et al. 2018b). In landscapes where livestock are displacing hippos, these differences may lead to

substantial changes in aquatic ecosystem structure and function. Taken collectively, our results

expand the current understanding of the role played by large mammalian herbivores in the

functioning of aquatic ecosystems in African savannahs. However, they also emphasize the

species-specific nature of many of these ecological roles and suggest that species introductions

and/or rewilding efforts seeking to replace extinct species with modern analogues may have

unintended outcomes (Bar-On et al. 2018; Schweiger and Svenning 2020). Our results highlight

the need for more research on the ecological consequences of introduced large herbivores and

replacement of native populations by anthropogenic change.

Data accessibility. The data supporting this article have been deposited with the Dryad Digital

Repository at https://doi.org/10.5061/ dryad.jh9w0vt79

Authors’ contributions. F.O.M. conceived of the study, designed the study, collected field data,

performed data analysis and drafted the manuscript. M.J.K. and C.R.G.-Q. collected field data

and performed laboratory sample analysis; A.L.S., C.L.D. and D.M.P. designed the study and

critically revised the manuscript; G.A.S. designed the study, performed data analysis and

critically revised the manuscript. All authors gave final approval for publication and agree to be

held accountable for the work performed herein.

Competing interests. We declare we have no competing interests. Funding. This work was

supported by the International Foundation for Science (Research grant no. A/5810-1), an

Alexander von Humboldt Postdoc fellowship to F.O.M.; the German research foundation

(within the Research Training Group on Urban Water Interfaces, GRK 2032) to C.R.G.-Q. and

the U.S. National Science Foundation to D.M.P. (DEB 1354053 and DEB 1753727).

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Supplement of Chapter 2

Hippopotamus are distinct from domestic

livestock in their resource subsidies to and

effects on aquatic ecosystems

Masese, F.O., Kiplagat, M.J., González-Quijano, C.R., Subalusky, A.L., Dutton, C.L., Post, D.M.

& Singer, G.A.

Journal: Proceedings of the Royal Society B: Biological Sciences

Article URL: http://dx.doi.org/10.1098/rspb.2019.3000

This file includes:

Electronic supplementary material S1: Characteristics of dung in African savannah and loading

rates of organic matter (dung) by cattle and hippopotamus in the Mara River, Kenya

Electronic supplementary material S2: Experimental mesocosms and design and characteristics

of hippo dung and cattle dung used in experiment

Electronic supplementary material S3: Dung treatment effects on nutrients and organic matter

Electronic supplementary material S4: DOM composition

Electronic supplementary material S5: Modeling metabolism

Electronic supplementary material S6: Weekly measures of Ecosystem Metabolism

1. Electronic supplementary material S1:

(a) Characteristics of dung for large mammalian herbivores of the African savannah

Dung from herbivore species varies considerably in C:N:P stoichiometry because of differences in body size, quality of their diet (foraging

strategy) and digestive physiology (e.g. foregut and hindgut fermenters, or for comparison ruminants and non-ruminants (Edwards 1991;

Sitters et al. 2014). For example, the dung of browsers has higher N concentrations than that of grazers (Codron et al. 2007), which results

in differences in dung C:N:P ratios (Sitters et al. 2014).

Table S1 Mean dung C, N, P concentrations and stoichiometry for some large mammalian herbivores of the African savannah

Herbivore

species

Digestive

physiology and

feeding strategy*

(mg

g -1)

N (mg

g -1)

(mg

g -1)

C:N

C:P

N:P

C:N:P

References

Bushbuck

Ruminant browser

417.0

18.9

3.3

25.0

164.0

6.1

164:6.1:1

(Sitters et al. 2014)

Giraffe

Ruminant browser

499.0

23.9

3.3

16.1

142.0

9.1

142:9.1:1

(Sitters et al. 2014)

Goat

Ruminant browser

29.7#

1.7#

0.3#

19.2

102.3

6.4

102:6.4:1

(Sileshi et al. 2017)

Cattle

Ruminant grazer

29.1#

1.3#

0.5#

23.3

79.2

3.6

79:3.6:1

(Sileshi et al. 2017)

Zebu cattle

Ruminant grazer

28.4#

11.3

2.2

30.4

155.2

5.1

127:5.1:1

this study

Buffalo

Ruminant grazer

348.0

10.9

2.4

30.5

153.0

5.3

153:5.3:1

(Sitters et al. 2014)

Hartebeest

Ruminant grazer

403.0

8.6

3.0

52.9

153.0

3.6

153:3.6:1

(Sitters et al. 2014)

Reedbuck

Ruminant grazer

389.0

17.1

3.1

21.9

103.0

4.8

103:4.8:1

(Sitters et al. 2014)

Waterbuck

Ruminant grazer

379.0

14.5

2.9

29.1

145.0

5.3

145:5.3:1

(Sitters et al. 2014)

Wildebeest

Ruminant grazer

358.0

13.5

3.5

27.3

119.0

4.8

119:4.8:1

(Sitters et al. 2014)

Zebra

Non-ruminant

grazer

414.0

12.2

3.5

45.7

213.0

4.0

213:4.0:1

(Sitters et al. 2014)

Hippopotamus

Non-ruminant

grazer

33.7#

9.8

1.3

34.4

261.4

7.6

261:7.6:1

this study

Hippopotamus

Non-ruminant

mixed feeder

34.9#

1.0#

0.2#

34.9

222.8

6.3

223:6.3:1

(Subalusky et al. 2015)

Elephant

Non-ruminant

mixed feeder

447.0

13.5

1.8

34.4

221.0

7.9

221:7.9:1

(Sitters et al. 2014)

*Herbivore species were grouped per digestive physiology and feeding strategy based on (Cerling et al. 2003; Codron et al. 2007; Kingdon and Largen

1997).

#These numbers are % per weight of dry matter

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

(b) Loading of organic matter (dung) by cattle and hippopotamus in the

Mara River, Kenya

Methods: We estimated cattle loading rates of organic matter (dung) into the Mara River, Kenya

inside and outside of the Maasai Mara National Reserve (MMNR), where there is an overlapping

distribution of livestock and hippopotamus. We used literature estimates of the daily dry matter

intake (DMI) of Zebu cattle, which were determined as a fraction of their body mass (BM) (Elliott

and Fokkema 1961; Oyenuga and Olubajo 1975). It has been established that DMI scales with BM

in herbivores (Clauss et al. 2007; Müller et al. 2011). We used literature values to determine DMI,

which has been estimated to range from 1.9 % - 2.5 % BM for cattle on African pasture (Elliott and

Fokkema 1961; Oyenuga and Olubajo 1975). Two values have been provided in the literature for

BM of cattle in the Mara Basin; 180 kg (Lamprey and Reid 2004) and 350 kg (Hoffman 2007), so

we used an average weight of 265 kg. Similarly to estimates of hippopotamus loading rates that

have been determined for the Mara River (Subalusky et al. 2015), we used the low intake value of

1.9 % BM (equivalent to 5035 g DM) for the wet season (lasting 6 months) and the highest value

of 2.5 % BM (6625 g DM) for the dry season. We assumed that the cattle population in the Mara

were in metabolic equilibrium and used the following equation to estimate the mass of organic

matter (OM) excreted or egested (ex/eg):

Mass of ex/eg OM = Mass Food Consumed (DMI) x % ex/eg OM

To determine the amount of OM (dry matter) excreted or egested, we used literature estimates

showing cattle on a grass diet excrete or egest 57% (Zhu et al. 2018). Cattle preferentially defecate

in rivers during watering or crossings (Bond et al. 2012), so we used time budgets to estimate the

per cent of excretion and egestion occurring in the river. Both cattle and hippos have long mean

gut retention times for particles (71 h for hippopotami and 66 h for cattle) and fluids (26 h for

hippopotami and 32 h for cattle) (Clauss et al. 2004; Clauss et al. 2007; Müller et al. 2011). Thus,

we assumed constant excretion and egestion rates by cattle throughout the day, as was done in the

hippopotamus study (Subalusky et al. 2015). Using behavioural data collected at livestock watering

points across the Mara River basin (Iteba J, unpublished data), we determined that cattle spend an

average of 10 minutes in or near the river during watering and/ or crossings. To obtain the daily

per-cattle rate of loading into the Mara River, we multiplied total per-cattle daily excretion and

egestion rates by the fraction of time spent in the river per day.

We estimated the total loading rates to the Mara River by multiplying the per-cattle loading rate

by cattle population estimates in the Mara River in 2002 (Reid et al. 2003), and compared these with

CHAPTER 2. SUPPLEMENT

the hippopotamus population in 2006 (Kanga et al. 2011). We then compared the loading of cattle

and hippopotami in two areas of the Mara River where their distribution overlaps: the Mara River

outside the Maasai Mara National Reserve and along the Talek River. We assumed that all cattle

within either the Mara or Talek sub-catchment visited the river for watering or crossing at least

once per day. Some cattle may use water pans for their water needs during certain portions of the

year, which may make our loading estimates on the upper end of potential inputs.

Results: We estimate cattle in the Mara basin have a daily dry matter intake of 25 g DM kg-1 in the

dry season and 19 g DM kg-1 in the wet season. This is in comparison to the daily dry matter intake

of 4.5 g DM kg-1 in the wet season and 6.8 g DM kg-1 in the dry season for hippopotamus (Subalusky

et al. 2015). We estimate that an average cattle excretes or egests 10.5 g DM kg cattle-1day-1 in the

wet season, and 13.8 g DM kg cattle-1day-1 in the dry season. Assuming that cattle consumption is

averaged over 6 months of wet season and 6 months of dry season (Subalusky et al. 2015), and that

they spend 10 minutes in the river per day, we estimate an average cattle loads 22.3 g DM kg cattle-

1day-1 to the river. Using % dry mass estimates from cattle faeces in the field (25.7% dry mass), we

calculated that 22.3 g DM equals 86.6 g faeces (wet mass), thus an average cattle (265 kg) defecates

12.5 kg faeces (wet mass) every day, and 0.0866 kg (0.69% of daily defecation) of that goes into the

Mara River. In comparison, an average hippopotamus (1500 kg) defecates 17.4 kg faeces (wet mass)

every day, and 8.7 kg (50%) of that goes into the Mara River (Subalusky et al. 2015). Using

population estimates from 2000 (Lamprey and Reid 2004; Reid et al. 2003), we estimated total daily

loading for the cattle population in the Mara River outside the reserve (MMNR) and along the

Talek River to be 2599 kg and 7364 kg faeces (wet mass), respectively (Table S1). Although the

cattle population estimates of 2000 are old, a study shows that by 2016 the numbers had not changed

significantly, although the numbers were higher between 2005 and 2010 (Ogutu et al. 2016). In

comparison, the total daily loading from excretion and egestion of hippopotamus population in the

Mara River outside the reserve (MMNR) and along the Talek River (1,571 and 648 individuals,

respectively) is estimated to be 13,668 kg and 5,638 kg faeces (wet mass), respectively, which is

equivalent to a total of 4,586 kg day-1 DM (Table S1). Of the total organic matter loading due to

cattle and hippos, cattle contribute 6-57% of inputs.

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Table S2 Estimated loading rates of organic matter (dung) by cattle and hippopotamus in the Mara River,

Kenya

Hippopotamus and cattle populations and

loading numbers

Inside

Reserve

Outside

Reserve

Talek

River

Hippoptamus numbersx

1,924

1,571

648

Cattle numbers*

13,350

30,000

85,000

Total loading by hippopotamus population (kg day-

1, wet wt)y

16,739

13,668

5,638

Total loading by cattle population (kg day-1, wet

wt)

1,157

2,599

7,364

*Data sources- (Lamprey and Reid 2004; Reid et al. 2003). Cattle numbers outside the reserve are for the

Koyake Group Ranch, while numbers for the Talek represent all other Group Ranches, estimated from the

conservative number of 100,000 cattle in the group ranches outside the MMNR.

xHippopotamus numbers and density are from (Kanga et al. 2011; Subalusky et al. 2015), respectively.

yEstimates of hippopotamus loading rates are from (Subalusky et al. 2015).

2. Electronic supplementary material S2: Characteristics of hippo dung and

cattle dung and experimental mesocosms

(a) Characteristics of cattle and hippo dung

Macro- and micro-nutrient composition of of cattle and hippopotamus faecal samples used for the

mesocosm experiment were analysed at the Leibniz Institute for Zoo and Wildlife Research, Berlin,

Germany (Table S3). Before the analysis, all dried samples (60⁰C for 48h) were grounded with an

IKA A 11 Basic mill (IKA-Werke GmbH & Co. KG, 79219 Staufen, Germany) to a particle size of

about 1mm. For C and N, samples were weighed and loaded into tin cups and analysed on a

elemental analyser (Hekatech-Elemental analyser, Thermo Finnigan). For P, samples were

weighed, ashed in a muffle furnace at 550 °C, then digested before analysis on a Perkin-Elmer ICP-

OES (Perkin Elmer, Ueberlingen, Germany). Crude protein was calculated as 6.25*N (Dijkslag et

al. 2019). For analyses of carbohydrates (sucrose, d-glucose, d-fructose, starch) we used enzymatic

tests, commercial kits from r-biopharm (R-Biopharm AG, 64297 Darmstadt, Germany) in which

standard solutions were included. Additionally, a lab standard always was run in all nutrient

analyses to check for reproducibility and accuracy of the tests. For mineral analysis (Ca, Mg, Fe, K)

samples were microwave digested and analyzed by AAS (Atom-Absorption-Spectroscopy).

CHAPTER 2. SUPPLEMENT

Table S3 Characteristics of hippo dung and cattle dung used in the mesocosms in this study

Parameter

Hippo dung

Cattle dung

Mean particle sizes in mm*

17.8

0.4

Carbon (% of dry matter)

33.71

28.36

Nitrogen (% dry matter)

0.98

1.13

Protein (% dry matter)

6.13

6.55

Fructose (mg g-1)

0.00

Glucose (mg g-1)

0.51

0.52

Sucrose (mg g-1)

0.26

0.47

Starch (mg g-1)

2.59

1.87

Ca (mg g-1)

6.15

8.07

Fe (mg g-1)

3.79

4.14

K (mg g-1)

9.62

10.29

Mg (mg g-1)

1.63

1.69

P (mg g-1)

1.29

2.23

N (mg g-1)

9.81

11.32

C:N:P

261.4:7.6:1.0

127.2:5.1:1.0

*From (Fritz et al. 2009; Thomas and Campling 1977).

(b) Experimental set-up of mesocosms

Mesocosms were constructed out of PVC canvas measuring 4.2 m long and 19 cm wide (Subalusky

et al. 2018). Water was recirculated in each mesocosm by paddlewheels affixed to a shaft that was

streams were located in an open field, and the entire array was covered with a shade cloth to yield

even light distribution. Mesocosms were lined with washed gravel and filled with river water from

a region upstream of most herbivore inputs. Mean (±SD) velocity and depth across channels were

0.078 ± 0.013 m s-1 and 7.8 ± 0.7 cm, respectively. Water levels were maintained by rainfall and

additions of rainwater. The river water had the following physicochemical characteristics: total

suspended materials = 1.11±0.1 mg L-1, temperature = 19.4±0.7 ⁰C, dissolved organic carbon =

1.62±0.5 mg L-1, nitrate = 1.48±0.4 mg L-1, soluble reactive phosphorus = 0.06±0.06 mg L-1, and a

concentration of ammonia below detection limits (10 μg L-1). Background nutrient and DOC

concentrations were lower than treatment level concentrations in all treatments.We had three

replicates for each of 6 dung treatments in a replacement design ranging from 100% hippo dung to

100% cattle dung: H100 = 100% hippo, 0% cattle; H80 = 80% hippo, 20% cattle; H60 = 60% hippo,

40% cattle; H40 = 40% hippo, 60% cattle; H20 = 20% hippo, 80% cattle; and H0 = 0% hippo, 100%

cattle.

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Fresh hippo dung and cattle dung were collected from hippo paths and Maasai livestock pens,

respectively. Dung from 5 different hippo paths and 3 cattle pens was thoroughly homogenized in

buckets before use. We had three replicates for each of 6 dung treatments in a replacement design

ranging from 100% hippo dung to 100% cattle dung with 20% increments of replacement (Figure

S1). In contrast to a simpler pure hippo vs. pure cattle design, this approach allowed us to test for

potential interactive effects between dung types, recognizable by non-linear responses to the dung

treatment gradient. Treatments were randomly distributed among mesocosms, with a replicate of

each treatment in each of the three blocks. A total of 120 g (wet weight, 1.7 g L-1) of dung was

distributed in each mesocosm once at the beginning of the experiment in order to study ecosystem

responses arising from differences in dung quality due to nutrient leaching and mineralization rates.

This concentration of dung is lower than field estimates for hippo sites in the Mara river (4 g L-1,

Subalusky et al. 2015), but it provided a sufficient quantity to elicit ecosystem responses without

creating hypoxic conditions.

To accelerate biofilm growth, mesocosms were inoculated with periphyton scraped off rocks from

the Amala River. Each mesocosm was lined with 6 unglazed ceramic tiles that were used for weekly

sampling of biofilms. Each week, one tile from each mesocosm was destructively sampled without

replacement, and biofilm was scrubbed off into a known volume of water and filtered through pre-

weighed and pre-combusted GF/F filters (Whatman International Ltd., Maidstone, England) for

analysis of ash-free dry mass (AFDM).

CHAPTER 2. SUPPLEMENT

Figure S1 Experimental set-up and dung used in mesocosms: (a) allocation of dung treatments in three blocks

driven independently by paddle wheels, (b and c) layout and details of mesocosms, (d) hippo dung, and (e)

cattle dung.

Block A

Block B

Block C

H100

mesocosms

Experimental design

H100

H0H20

H60 H40

H60

H80

H20

H0H40

H40 H80

H80

(a) (b)

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

3. Electronic supplementary material S3: Dung treatment effects on

nutrients and organic matter

Water samples for ammonium, nitrate, nitrite, and SRP were filtered on site through pre-combusted

(450°C for 4 h) and pre-washed Whatman GF/F filters into acid-washed HDPE bottles, and stored

at 4°C until analysis within 48 hr. For TSS and POM, water was filtered on site through pre-

combusted and pre-weighed GF/F filters. Water samples for DOC concentration and composition

were filtered on site through a double layer of pre-combusted Whatman GF/F filters (pore size 45

µm) followed by GF/75 filters (pore size 3 µm) into acid-washed and pre-combusted glass vials.

DOC samples were then acidified with 2 N hydrochloric acid (HCl) to pH 2 and refrigerated at 4°C

until analysis. TP and TN was determined following the persulfate digestion method (APHA 1998).

We measured SRP, TN, TP, NO3-2 and NH4+ in water samples using standard colorimetric methods

(APHA 1998). We measured DOC concentration using a Shimadzu TOC-V-CPN fitted with an

inorganic C removal unit. We extracted Chl-a in 90% ethanol and determined concentrations

spectrophotometrically (APHA 1998). We measured TSS concentration (g L− 1) by drying filters

with the adhered sediments and subtracting the filter weight. POM in TSS was further determined

gravimetrically after ashing filters at 450 °C for 4 h, re-weighing them, and subtracting the ashed

weight from TSS. Biofilm biomass (AFDM) was measured similarly to POM using the filtered

slurry from the scraped tiles and expressed per unit area.

CHAPTER 2. SUPPLEMENT

(a) Dung treatment effects on nutrient concentrations

Figure S2 Influence of dung treatment on (a) soluble reactive phosphorus (SRP), (b) nitrite, (c) ammonium,

and (d) nitrate concentrations. Asterisks are displayed for significant linear relationships across low-high

proportions of cattle dung for each sampling occasion (α ≤ 0.05). *P < 0.05, **P < 0.01, ***P < 0.001.

Dung treatment (% cattle dung)

020 40 60 80 100

Nitrate (mg L-1)

8Day 1

Day 7

Day 14, R2 = 0.41**

Day 21

Day 28

Day 35, R2 = 0.35*

Day 42

Dung treatment (% cattle dung)

020 40 60 80 100

Nitrite (mg L-1)

0.00

0.05

0.10

0.15

0.20

0.25

Day 1, R2 = 0.53***

Day 7, R2 = 0.69***

Day 14, R2 = 0.55***

Day 21, R2 = 0.35*

Day 28, R2 = 0.28*

Day 35, R2 = 0.42**

Day 42

Dung treatment (% cattle dung)

Ammonium (mg L-1)

0.0

0.1

0.2

0.3

0.4

Day 1, R2 = 0.86***

Day 7

Day 14

Day 21

Day 28

Day 35

Day 42

Dung treatment (% cattle dung)

SRP (mg L-1)

0.0

0.5

1.0

1.5

2.0 Day 1, R2 = 0.54***

Day 7, R2 = 0.71***

Day 14

Day 21

Day 28

Day 35

Day 42

(a)

(d)

(c)

(b)

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Figure S3 Influence of time on (a) soluble reactive phosphorus (SRP), (b) nitrite, (c) ammonium, and (d)

nitrate concentrations among dung treatments

Time in days

010 20 30 40

Nitrite (mg L-1)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

H20

H40

H60

H80

H100

Time in days

010 20 30 40

Ammonium (mg L-1)

0.0

0.1

0.2

0.3

0.4

0.5

H20

H40

H60

H80

H100

Time in days

010 20 30 40

SRP (mg L-1)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

H20

H40

H60

H80

H100

Time in days

010 20 30 40

Nitrate (mg L-1)

H20

H40

H60

H80

H100

(a)

(d)

(c)

(b)

CHAPTER 2. SUPPLEMENT

(b) Dung treatment effects on organic matter

Figure S4. Influence of dung treatment on a) DOC, b) chlorophyll-a (Chl-a), c) ash-free dry mass (AFDM),

d) total suspended solids (TSS), and e) particulate organic matter (POM) concentrations. Asterisks and model

fits are displayed for significant linear relationships (α ≤ 0.05). *P < 0.05, **P < 0.01, ***P < 0.001.

Dung treatment (% cattle dung)

020 40 60 80 100

POM (mg L-1)

60 Day 1

Day 7

Day 14, R2 = 0.82***

Day 21, R2 = 0.69***

Day 28, R2 = 0.88***

Day 35, R2 = 0.86***

Day 42, R2 = 0.90***

Dung treatment (% cattle dung)

DOC (mg L-1)

40 Day 1, R2= 0.36**

Day 7, R2= 0.70***

Day 14, R2 = 0.57***

Day 21, R2 = 0.31*

Day 28, R2 = 0.36**

Day 35, R2 = 0.19

Day 42, R2 = 0.62**

Dung treatment (% cattle dung)

020 40 60 80 100

TSS (mg L-1)

Day 1, R2 = 0.30*

Day 7, R2 = 0.30*

Day 14, R2 = 0.37**

Day 21, R2 = 0.57***

Day 28, R2 = 0.40**

Day 35, R2 = 0.53***

Day 42, R2 = 0.77***

Dung treatment (% cattle dung)

AFDM (mg cm-2)

40 Day 1

Day 7

Day 14, R2 = 63***

Day 21, R2 = 67***

Day 28, R2 = 81***

Day 35, R2 = 52***

Day 42, R2 = 49**

Dung treatment (% cattle dung)

020 40 60 80 100

Chl-a ( g L-1)

0.0

50.0

100.0

150.0

200.0

250.0

Day 1

Day 7

Day 14, R2 = 0.60***

Day 21, R2 = 0.50***

Day 28, R2 = 0.63***

Day 35, R2 = 0.47**

Day 42, R2 = 0.40**

(a)

(c)

(b)

(d)

(e)

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

4. Electronic supplementary material S4: DOM composition

Animal dung releases dissolved organic matter of carbon (DOC) into water with nutrients. DOC in

water contains thousands of molecules (Coble 2007) that influence ecosystem processes through

light attenuation and nutrient availability (Ishii and Boyer 2012). For instance, the optically active

part of DOC, which is known as coloured dissolved organic carbon (CDOC), is several times that of

chlorophyll in coastal areas (Coble 2007). Determining the composition of DOC allows the study of

factors such as land use and land management practices, as they affect spatial and temporal

variations in stream biogeochemistry and ecosystem functioning (Fellman et al. 2010; Ishii and

Boyer 2012; Masese et al. 2017; Mwanake et al. 2019). Here, we studied DOC released by leachates

of cattle dung and hippo dung in order to understand and compare their properties and influence

on ecosystem processes.

We characterized the optically active DOC fraction by absorbance and fluorescence analyses, which

provide proxies for DOC source and/or biological availability (Fellman et al. 2010; Jaffé et al. 2008).

DOC absorbance spectra (250–600 nm, every 5 nm) and fluorescence excitation–emission matrices

(EEMs, excitation wavelength from 250 to 600 nm, in 5 nm increments and emission range of 250–

550 nm in 1.77 nm increments) were measured simultaneously on a Horiba Aqualog (Horiba Ltd,

Kyoto, Japan) spectrophotometer using a 1 cm quartz cuvette and a scan speed of 12,000 nm min-1

with a response time of 0.01 s. MilliQ water was used as an optical blank. Naperian absorption

coefficients were calculated from absorbance scans (Green and Blough 1994) and used to calculate

a number of indices. A ratio of absorption coefficients E2:E3 (a250:a365), which declines with increasing

molecular size, was used to provide further information on DOC aromaticity and molecular weight

(Helms et al. 2008). The spectra slope ratio (SR), which is a ratio of the short wavelength slope (S275-

295) and the long wavelength slope (S350-400), served as an indicator of molecular weight and

photodegradation-induced shifts (Helms et al. 2008). The DOC-standardized specific UV

absorption at 254 nm (SUVA254), which is commonly used as an indicator of aromaticity (Weishaar

et al. 2003), was computed by dividing decadal absorbance by cuvette path length (in m) and by

DOC concentration (in mg C L-1).

EEMs were corrected for the water Raman scatter, Rayleigh–Tyndall effect and the inner filter

effect (McKnight et al. 2001; Parlanti et al. 2000), and used to calculate three fluorescence indices:

fluorescence index (FIX) (McKnight et al. 2001), freshness index (β/α) (Wilson and Xenopoulos

2009a), and humification index (HIX; unitless) (Ohno 2002). The FIX provides information on DOC

origin, distinguishing terrestrially derived DOC (FIX~1.2) from microbially derived DOC

(FIX~1.9), and was calculated as the ratio of emission intensity at 450–500 nm for an excitation of

CHAPTER 2. SUPPLEMENT

370 nm (McKnight et al. 2001). β/α indicates the proportion of recently produced DOC relative to

more decomposed DOC (Parlanti et al. 2000; Wilson and Xenopoulos 2009a). β/α values >1

indicate that DOC is primarily of autochthonous origin and values 0.6-0.8 indicate primarily

allochthonous origin (Huguet et al. 2009a). HIX is directly proportional to the humic content of

DOC, where HIX values around 1–2 are associated with non-humified plant material and values >

10 are commonly reported for fulvic acid extracts (Ohno 2002; Zsolnay et al. 1999).

We used parallel factor analysis (PARAFAC) to decompose 349 EEMs into fluorescent components

of DOC (Stedmon and Bro 2008). PARAFAC was conducted using DOMFluor toolbox 1.7

following Stedmon & Bro (2008) in Matlab 7.11.0 (MathWorks, Massachusetts, USA). The number

of components was determined by using split half validation and assessed with random initialization

fits and residual analysis (Stedmon and Bro 2008). We further characterized DOC using size-

exclusion chromatography (SEC) (Huber et al. 2011a), which separates three size fractions: humic

substances (HS), high-molecular weight non-humic substances (HMWS) and low molecular weight

substances (LMWS).

Fluorescence EEMs were very dissimilar and occurred over a wide range of excitation (ca. 250−450

nm) and emission (ca. 270−600 nm) wavelengths (Figures S4 and S5). The PARAFAC model

consisted of seven components (referred as C1–C7) whose fluorophores were compared with

literature (Table S2). The position and spectral shape for the seven components are shown in

Figures S4 and S5. Four humic-like (C1, C4, C5 & C6), one reduced humic-like (C2) and two

protein-like (C3 and C7) fluorescence components were identified across our dataset, with C1, C3,

C5, and C6 being among the most commonly observed components in aquatic ecosystems (Murphy

et al. 2014). C1 and C4 are located in the fluorescence region that usually define the ubiquitous

humic-like Peaks C and A, respectively (Coble 1996), and are related to high molecular weight

humic substances of terrestrial origin (Fellman et al. 2010). In addition, component 4 has been

shown to be resistant to photodegradation (Stedmon and Markager 2005a). C5 was similar to peak

M, and resembled components of high molecular weight, humic-like, terrestrial material (Fellman

et al. 2010) with increased aromatic carbon content, indicating higher plant material as a likely

source (Cory and McKnight 2005). C6 had both a primary excitation peak (ca 250−270 nm) and a

secondary excitation peak (340−420 nm), which have been associated with large molecular size,

hydrophobic compounds (Wu et al. 2003). Protein-like C3 and C7 spectra resemble those of

tryptophan and tyrosin free amino acids, respectively, and have been classified as originating from

microbial DOM sources (Cory and McKnight 2005; Fellman et al. 2010). C7 was also the most

redshifted component in our study, resembling peak T (Fellman et al. 2010).

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Figure S5 Observed excitation and emission wavelengths for maximum fluorescence of the 7 PARAFAC

components identified in our dataset.

Figure S6 Emission and excitation loadings of the 7 PARAFAC components

Component 1

Component 5

Component 4Component 3

Component 2

Component 7Component 6

Excitation

Emission

Table S4 Fluorescent components of DOM as identified by parallel factor analysis (PARAFAC). Given are observed excitation and emission

wavelengths for maximum fluorescence, alignment with distinct fluorescence peaks and PARAFAC components identified in previous studies, probable

sources of DOC and a literature-based component descriptiona.

PARAFAC

component

(this study)

Excitation

maximum

(nm)

Emission

maximum

(nm)

Peak name and

PARAFAC

component s

(previous studies)

Probable

sources*

Description

<250, 250

428-444

CCa,Cb,Cd, MCd, βP, 1Sma,

4SMb, 1Ma, 11CMK

T, A, M

UVA humic-like component. Low

molecular weight, biological activity,

widespread

<250, 250

516-530

(500-550)

4CMK

T, M

Hydroquinone-like component. reduced

humic-like component

270-276

320-332

BCa, δP, 8CMK, 6SMa,

7SMb, 5SMB, 7Ma, 6Mb,

4CK

T, A, M

Protein- and tryptophan-like component,

microbial-produced, widespread

<250, 250

436-456

ACa,Cb, ACb, αP

UVC humic-like, fulvic acid component.

<250, 250

378-382

ACb, MCd, βP, 1Sma,

4SMb, 1Ma

T, A, M

UVA humic-like component. Polycyclic

aromatic, increased aromatic carbon

content.

256-262

(366-378)

446-472

ACa,Cd, CCa,Cd, αP

UVC humic-like + UVA humic-like

component. reduced humics, widespread.

254

302

BCb,Cd, TCd, γP, 13CMK,

4SMa, 8SMb, 1Ma, 7Mb

T, A, M

Protein- and tyrosine-like component. may

indicate more degraded peptide material

aValue in parentheses is secondary maximum. See text for discussion of probable origins. * T, terrestrial plant or soil organic matter; A, autochthonous

production; M, microbial processing.

ca Coble et al. (1990); cbCoble (1996); cdCoble et al. (1998); PParlanti et al. (2000); SMaStedmon and Markager (2005b); SMbStedmon and Markager (2005a);

MaMurphy et al. (2008); MbMurphy et al. (2011); CMKCory and McKnight (2005); SMBStedmon et al. (2003); CKCory and Kaplan (2012).

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

5. Electronic supplementary material S5: Modeling metabolism

We estimated flume-scale GPP and ER following Fuss et al. (2017) by fitting a differential

equation model (Hotchkiss and Hall 2014; Van de Bogert et al. 2007) to diel DO concentration

measured at a single site (Marzolf et al. 1998; Odum 1956). The model simulates temporal

changes in DO concentration (dDO/dt) as the result of parameterized GPP, ER and reaeration

(RF, eqn 1):

𝑑DO

𝑑𝑡 =(GPP−ER+RF)×1

𝑧 (1)

where GPP adds DO to the water by photosynthesis; ER consumes DO and RF is the gas

exchange at the water–air interface. GPP (g O2 m-2 min-1) was modelled with light saturation

(Ratkowsky, 1986; Uehlinger, König & Reichert, 2000) as:

GPP=PAR

P1+P2+PAR (2)

where PAR (W m-2) is the observed, instantaneous PAR. P1 (W min g-1 O2) is the inverse of the

slope of a photosynthesis–irradiance curve at low light intensity and P2 (m2 min g-1 O2) is the

inverse maximum photosynthesis rate. Daily GPP (GPP24, g O2 m-2 day-1) was integrated from

P1, P2, the light record and the time step ∆t between light measurements:

GPP24=∑𝑃𝐴𝑅𝑡

𝑃1+𝑃2+𝑃𝐴𝑅𝑡

𝑡 𝑒𝑛𝑑

𝑡=𝑡0 × ∆𝑡 (3)

Since ER (g O2 m-2 min-1) is a strongly temperature-dependent process (Kirschbaum 1995), it

was modelled with the van’t Hoff–Arrhenius equation (Parkhill and Gulliver 1999):

ER=ER20

(24 ×60) × 𝜃(T−20) (4)

where ER2420 (g O2 m-2 day-1) is the daily rate of ER standardized to 20 °C and T (°C) is the

observed, time specific ambient stream temperature, and 𝜃 (theta) is the temperature dependance

on respiration. Because different authors have used different values of 𝜃 (e.g., Demars et al.

2015), and our modeling efforts were not successful with the commonly used value of 1.045, we

decided to model this value and obtained a value of 1.1085 that we used in our model. Since

diurnal variations in temperature in the mesocosms was high (mean daily range 14 °C - 26 °C),

using a higher value for theta was more relevant for our analysis. Moreover, our model outputs

were greatly improved. In order to investigate ER at in situ temperature, we translated ER2420

to ER24insitu (g O2 m-2 day-1) using recorded in situ temperature measurements T (°C) for every

time interval ∆t:

CHAPTER 2. SUPPLEMENT

ER24𝑖𝑛𝑠𝑖𝑡𝑢 =∑𝐸𝑅20

(24×60)

𝑡 𝑒𝑛𝑑

𝑡=𝑡0 × 1.1085(𝑇𝑡−20)×∆𝑡 (5)

The reaeration flux RF (g O2 m-2 min-1) was computed as

RF =𝑘 × DOdeficit (6)

where k is the temperature-dependent vertical gas exchange velocity (m min-1) and DOdeficit (g

m-3) is the difference of the observed DO concentration (DO) to DO at 100% saturation (DOSat):

DOdeficit =DOSat − DO (7)

DOSat was calculated from observed, time-specific ambient stream temperature and atmospheric

pressure (Benson and Krause Jr 1984). The vertical gas exchange velocity k (m min-1) is related

to the reaeration coefficient K (min-1) by multiplication with depth (m) (Marzolf et al. 1998;

Raymond et al. 2012). We used a reaeration coefficient measured in 6 mesocosms (2 each for

each block) by degassing water by boiling and then cooling in air-tight containers before

carefully filling the mesocosms with minimal bubbling. The slope of the linear increase in DO

concentration was used as an estimate of re-aeration. Temperature dependence of gas exchange

was calculated according to Elmore (1961) and Bott (Bott 1996):

𝐾𝑇=K20 × 1.024𝑇−20 (8)

where KT and K20 are reaeration coefficients at ambient stream temperature T and at 20 °C,

respectively. For model fitting, the time derivative dDO/dt of eqn (1) was approximated by

differences in ∆DO/∆t across the observed time intervals, and a discretized time series of DO

was predicted using observed, time-specific temperature and light conditions, barometric

pressure and a chosen parameter set P1, P2, ER2420 and K20 (Fuss et al. 2017; Hotchkiss and

Hall 2014; Van de Bogert et al. 2007):

DO𝑡+1 =DO𝑡+(GPP𝑡− ER𝑡+RF𝑡)×∆𝑡×1

𝑧 (9)

DOt+1 (g O2 m-2) was computed from DO𝑡 and GPP, ER and RF were computed from

temperature and light conditions at the previous time point t. ∆𝑡, the time interval between t

and t + 1, is needed to scale up the minute-specific rates accordingly and is chosen in agreement

with the observed time series. Equation (9) was obtained by forward differencing or Eulerian

integration of eqn (1) (Soetaert & Herman, 2009). A first observed DO measurement is used as

a starting value (DO𝑡0 ), from which all subsequent DO𝑡 values are computed. To fit P1, P2,

ER2420 and K20 to empirical data, we used eqn (9) in an inverse modelling approach that

repeatedly models a DO time series with updated parameter values and minimizes the sum of

squared residuals of the modelled to the observed DO time series. We estimated a reaeration

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

coefficient (k) by filling 6 clean mesocosms (2 for each block) with degassed (boiled and cooled)

water and then used recorded DO and temperature to model reaeration (K20) without GPP and

ER. K20 was then used as a starting value to reliably model P1, P2, ER2420 and K20.

Over the last decade, temperature depencency of ER is an active topic of discussion and different

authors have used different values (theta) for this dependence (Demars et al. 2015; Perkins et al.

2012; Sand-Jensen et al. 2007). Our attempts to use a value of 1.045 (Riley and Dodds 2013)

were unsuccessful, so we decided to use a higher modeled value of 1.1085. To arrive at this theta

value (1.1085), we selected 50 days from different dung treatments and different days from

among the 44 days experimental period (6 weeks) and modeled theta along with P1, P2 and ER

(4-parameter model) and used a fixed reaeration coefficient (k) that we measured in our

mesocosms. An average theta value was then obtained from successfully modeled days (see

below), which we then fixed for a 3-parameter model (P1, P2 and ER modeled and k and theta

fixed) we subsequently used to model all days of the experiment. Since ER increases with

temperature, using a higher value for theta was more relevant for our data, which displayed a

wide range in water temperature (mean daily range 14 °C - 26 °C).

A number of checks were done to pick the number of days that were successfully modeled and

whose results were used for subsequent analyses. First, we used nlm in the metabolism FIT

function to minimize the negative log-likelihood between measured and modeled DO values.

Low values (< -100) of sum of squared residuals for each model were considered indicative of a

successful and constrained fit. Secondly, model fits (graphs) were inspected to confirm that the

modeled DO values perfectly or closely matched measured DO values (Figure S5). Finally, the

modeled outputs for GPP and ER were inspected to make sure that they made sense. For

instance, cases where GPP values were negative or ER values were zero or positive were

discarded.

Sensitivity analysis

To determine the effect of using different values of theta on our estimates of GPP and ER, we

performed a sensitivity analysis and re-run the model using a value of theta (1.045) that is

common in the literature. By using a higher value of theta (1.1085), our model outputs were

better constrained, i.e., the sum of squared residuals of the modeled to the measured DO time

series obtained using a theta value of 1.1085 were lower for most streams compared with when

a theta value of 1.045 was used (Figure S5). Better performance of the higher value of theta was

also confirmed by the higher number of days that were successfully modeled: Of the 567 days

out of 774 days (18 streams x 43 days) that had complete data, 400 days (70.5%) were

successfully modeled by a theta value of 1.1085, while only 311 days (54.9%) were successfully

modeled by the common theta value of 1.045.

CHAPTER 2. SUPPLEMENT

The metabolism results obtained using the two theta were different, but trends in GPP, ER,

GPP:ER ad NEP in response to dung treatment were generally similar (Figures S6 and S7). In

both cases (Figures S6 and S7), GPP, GPP:ER and NEP increased with increasing proportions

of cattle dung. However, in all cases the ranges of values were much reduced for the lower theta

value (1.045). For instance, for GPP, the highest value obtained using a theta of 1.045 was

around 4 O2 m-2 day-1 with most of the values below 3 O2 m-2 day-1, while for the higher theta

value (1.1085), the highest value was twice as high (around 8 O2 m-2 day-1). Similar trends in low

ranges for the lower theta value were observed for ER, GPP:ER and NEP. Moreover, the range

in ER was very low (0.2-0.5 O2 m-2 day-1) (Figure S7b). This lack of variation in ER, which is

very sensitive to temperature variation, to a low theta value (1.045) gave more credence to our

use of a higher theta value (1.1085). Moreover, the higher theta value enabled us to successfully

model more days, which enabled us to more effectively evaluate the effect of dung treatment on

ecosystem metabolism in our mesocosms.

Figure S7 Performance of different values of theta in modeling metabolism in our experimental

mesocosms. A higher value of theta (1.1085, upper panel) performed better for most streams when

compared with a common literature value of 1.045 (lower panel). a and b are model fits for day 1, and c

and d are model fits for day 10 in the hippo dung treatment (100 % hippo dung). The black bold line is for

measured dissolved oxygen concentration (mg/L) while the red line is for the modeled dissolved oxygen

concentration. The red dotted line is measured temperature and the blue dotted line is light intensity. The

green dotted line is for oxygen saturation. Modeling for each day was performed from mid-night (0

minutes, 24:00 hrs) to mid-night, 1440 minutes, 23:59 hrs).

Time in minutes

Dissolved oxygen concentration (mg/L)

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

Figure S8 Model outputs using a theta value of 1.1085. Weekly measures of flume-scale gross primary

production (GPP), (a) flume-scale ecosystem respiration (ER; b), GPP:ER (c) and net ecosystem

production (NEP; d) using a theta value of 1.1085. The dotted line indicate NEP = 0, and most of the

mesocosms were net heterotrophic on until day 7 and then switched.

Dung treatment (% cattle dung)

020 40 60 80 100

NEP (O2 m-2 day-1)

-2

Week 1

Week 2

Week 3

Week 4

Week 6

Week 7

Dung treatment (% hippo dung)

GPP (O2 m-2 day-1)

Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Dung treatment (% cattle dung)

020 40 60 80 100

GPP : ER

15 Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Dung treatment (% hippo dung)

ER (O2 m-2 day-1)

0.0

0.5

1.0

1.5

2.0 Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

(a)

(d)

(c)

(b)

CHAPTER 2. SUPPLEMENT

9o99

Figure S9 Model outputs using a theta value of 1.045. Weekly measures of flume-scale gross primary production

(GPP), (a) flume-scale ecosystem respiration (ER; b), GPP:ER (c) and net ecosystem production (NEP; d) using a

theta value of 1.045. The dotted line indicate NEP = 0, and most of the mesocosms were net heterotrophic on until

day 7 and then switched.

Dung treatment (% cattle dung)

020 40 60 80 100

NEP (O2 m-2 day-1)

6Week 1

Week 2

Week 3

Week 4

Week 6

Week 7

Dung treatment (% cattle dung)

020 40 60 80 100

GPP : ER

14 Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Dung treatment (% cattle dung)

ER (O2 m-2 day-1)

0.0

0.2

0.4

0.6

0.8

1.0

Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Dung treatment (% cattle dung)

GPP (O2 m-2 day-1)

6Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

(a)

(d)

(c)

(b)

Hippopotamus are distinct from domestic livestock in their resource subsidies to and effects on aquatic ecosystems

6. Electronic supplementary material S6: Weekly measures of Ecosystem

Metabolism

To investigate weekly changes in ecosystem metabolism (GPP, ER, GPP/ER and net ecosystem

production [NEP]), weekly means (1 value per stream per week, total 6 weeks) were used.

Significant differences among dung treatments were tested using generalized additive mixed

models (GAMMs, Zuur et al. 2007) after residuals in GLMM displayed non-linear responses

to dung treatment. GAMM models included dung treatment as a fixed effect, and block and

stream as random effects. Models were fitted using the the mgcv-package (Wood and Wood

2015) in the R platform (R Core Team 2017).

Table S5: Summary of generalized additive mixed modeling (GAMM) analyses to determine the effect of

dung treatment on ecosystem metabolism - gross primary production (GPP, mg O2 m-2 day-1), ecosystem

respiration (ER, ER, mg O2 m-2 day-1), GPP:ER and net ecosystem production (NEP, mg O2 m-2 day-1),

which displayed nonlinear responses to dung treatments.

SE= standard error; EDF = estimated degrees of freedom; F = ANOVA F-test value between the fitted

and a null model. Significance: *P < 0.05, **P < 0.01, ***P < 0.001

Measures of ecosystem metabolism

Variables

GPP

GPP:ER

NEP

Intercept (estimate(SE); t

value

4.05(0.36);

11.10***

0.77(0.04);

20.90***

4.50(0.36);

12.59***

3.41(0.34);

10.08***

Dung Treatment

(estimate(SE); t value

-0.07(0.01); -

4.85***

-0.01(<0.01);

-5.17***

-0.07(0.01); -

4.67***

-0.07(0.01); -

4.67***

Dung Treatment x Time

(estimate(SE); t value

0.01(<0.01); -

4.81***

<0.01(<0.01);

-7.65***

0.01(<0.01); -

4.65***

0.01(<0.01); -

4.29***

Block (EDF(F))

<0.01 (0)

0.81(1.81)

<0.01 (0)

Adj. R2

0.41

0.40

0.44

Explained deviance (%)

48.7

43.8

51.7

CHAPTER 3

Dissolved organic matter signatures in

urban surface waters: spatio-temporal

patterns and drivers

This study was published as:

This is the postprint version of the article.

3.1 Abstract

Advances in analytical chemistry have facilitated the characterization of dissolved organic

matter (DOM), which has improved understanding of DOM sources and transformations in

surface waters. For urban waters, however, where DOM diversity is likely to be high, the

interpretation of DOM signatures is hampered by a lack of information on the influence of land

cover and anthropogenic factors such as nutrient enrichment and release of organic

contaminants. Here we explored the spatiotemporal variation of DOM composition in

contrasting urban water bodies, based on spectrophotometry and fluorometry, size-exclusion

chromatography and ultrahigh-resolution mass spectrometry, to identify linkages between

DOM signatures and potential drivers. The highly diverse DOM we observed distinguished

lakes and ponds, which are characterized by a high proportion of autochthonous DOM, from

rivers and streams where allochthonous DOM is more prevalent. Seasonal variation in DOM

composition was apparent in all types of water bodies, apparently due to interactions between

phenology and urban influences, such as nutrient supply, the percentage of green space

surrounding to the water bodies and point source pollution. Optical DOM properties also

revealed the influence of effluents from wastewater treatment plants, suggesting that simple

optical measurements can be useful in water-quality assessment and monitoring, informing

about processes both within water bodies and their catchments.

Romero González-Quijano, C., S. Herrero Ortega, P. Casper, M. Gessner, and G. Singer.

2022. Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns

and drivers. Biogeosciences 2022:1-34.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

3.2 Introduction

Urban freshwaters typically receive high loads of organic carbon, nutrients and micropollutants,

ranging from pharmaceuticals and personal care products to industrial chemicals and more

(Schwarzenbach et al. 2006). Although routine wastewater treatment is increasingly effective,

chemical stressors in urban freshwaters remain widespread. Prominent reasons are pollution

legacies (Baume and Marcinek 1993 ; Ladwig et al. 2017) and continued uncontrolled inputs,

particularly by stormwater runoff (Council 2009). In addition, urban surface waters tend to

suffer from severe hydromorphological modifications. This includes the lateral and vertical

disconnection from floodplains and aquifers and results in large impacts on the extent and

complexity of riverine habitat (White and Walsh 2020). Furthermore, the disruption of

connectivity limits the self-purification capacity of urban surface waters (D'Arcy et al. 2007),

which can lead to turbid water and visually unpleasant and potentially harmful algal

blooms(Carpenter et al. 1998). This and the limited recognition of urban freshwaters as

providers of ecosystem services (Huser et al. 2016) calls for improved water management

strategies that consider ecological in addition to hygienic and chemical criteria (Gessner et al.

2014).

The concentration and chemical composition of dissolved organic matter (DOM), generally

quantified as dissolved organic carbon (DOC), are key characteristics of aquatic ecosystems.

Both concentration and composition are governed by allochthonous inputs and internal

biological production and transformation processes (Williams et al. 2016). Typically, however,

water quality monitoring only considers concentration and bulk quality properties (e.g.

biological oxygen demand, BOD) as measures of DOM availability to, and degradation by,

heterotrophic microbes (Jouanneau et al. 2014). This focus is at odds with the extreme diversity

of DOM observed in freshwaters, where thousands of compounds can be chemically

distinguished (Kellerman et al. 2014; Peter et al. 2020; Stanley et al. 2012). This high diversity

and the strong spatio-temporal variation of DOM composition suggests much potential for

DOM characteristics to provide insights into the state of freshwater ecosystems in water quality

assessment and monitoring. In fact, additional insights into freshwater ecosystems may be

gained if the very high diversity of DOM can be used to inform about water quality for

ecosystem assessment and monitoring purposes.

Recent progress in analytical methods has increasingly enabled the detailed characterization of

DOM to elucidate the sources and fates in surface waters (Xenopoulos et al. 2021). Optical

properties can inform not only about the chemical characteristics of DOM but also, for example,

about large-scale gradients in aquatic networks (Creed et al. 2015) or the degree of aquatic-

terrestrial ecosystem coupling (Catalán et al. 2013; Lambert et al. 2015; Sankar et al. 2020;

Yamashita et al. 2010). Fluorescence excitation-emission matrices (EEM) can be processed by

parallel factor analysis (PARAFAC) to identify independently fluorescing DOM components

CHAPTER 3

(Cory and McKnight 2005). Size-exclusion chromatography partitions bulk DOM into

molecular size fractions, which also tend to differ in origin and bioavailability (Huber et al.

2011b). Finally, the advent of ultrahigh-resolution mass spectrometry (FT-ICR-MS or

Orbitrap-MS) has greatly refined the characterization of DOM, revealing associations between

compositional turnover of DOM differing in molecular diversity and landscape-scale

environmental gradients in lakes (Kellerman et al. 2014) and rivers (Peter et al. 2020).

In the present study we explored variation in the chemical composition of DOM over time and

space in contrasting urban surface waters, hypothesizing that a detailed chemical

characterization of DOM yields signatures of various human influences. To this end, we

explored linkages between chemical composition of DOM and potential drivers determining

DOM signatures, including land cover, eutrophication and chemical pollution, which we

captured by using a suite of proxies. Our specific goals were to: (i) describe spatio-temporal

patterns of DOM composition across a range of urban freshwaters encompassing streams, rivers,

ponds and lakes; (ii) identify environmental factors accounting for the observed patterns; and

thereby (iii) explore how information on DOM composition could be included in urban

freshwater assessment and monitoring, complementing approaches and metrics currently used.

3.3 Methods

3.3.1 Study sites

The study was conducted in 32 freshwater sites located in the city of Berlin, Germany. Nearly

6.5% of the municipal area (889 km2) is covered by freshwaters. These comprise 60 lakes (>1

ha), about 500 ponds, the two slow-flowing lowland rivers Spree and Havel, and numerous

streams, ditches and canals. Selection of the 32 study sites followed a stratified random sampling

design (Figure 1a, Table S1). Based on geographical information for Berlin`s water bodies, we

randomly selected 7 sites in each of 4 strata: lakes, ponds, rivers and streams. Rivers and streams

were classified according to a width cutoff of 5 m. Monitoring data on water chemistry (Berlin

city administration, SenUVK 2009-2014) were used in a cluster analysis to identify highly

polluted sites. These were excluded from the pool used for randomly selecting study sites.

Instead, two organically polluted rivers (H1 and H2) and streams (H3 and H4) were deliberately

added to lengthen the environmental gradient. Sites H1 and H2 received WWTP effluents

(Figure 1a, Table S1) and sites H3 and H4 presented high levels of diffuse pollution. Other sites

for, for some of which monitoring data were unavailable (streams and ponds), were also affected

by pollution (Table S1): Pond P4 was formerly connected to an old waste water treatment plant

and still receives stormwater inflow during heavy rain events; S5 is located immediately

downstream of a WWTP; and R7 became a receiving stream in 2015 (Nega et al., 2019), which

was too recent for the site to become classified as polluted based on the monitoring data. Land

use data obtained from the Berlin city administration (Senate Administration for Environment,

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

2017) were used to calculate the proportion of paved areas and green spaces within a 50-m

perimeter around the selected water bodies using open-source geoinformation software (QGIS

Development Team 2017). The 50-m perimeter was chosen to capture influences in the

immediate vicinity of the sites, such as of the riparian zone and slightly beyond, but not of the

whole catchments, which are highly variable in size and tend to be difficult to define in urban

areas. Delineation of the 50-m perimeter enabled us to distinguish particularly between urban

sites adjacent to paved surfaces vs. green spaces. Tufekcioglu (2020) and Johnson (2005) used

buffer zones of similar size and a study on ponds using perimeters of up to 3200 m found 50 and

100 m to be most appropriate to assess land-cover effects (Declerck et al., 2006). All samples

were taken during base flow conditions (Figure S5).

CHAPTER 3

Figure 1 Map of 32 sampling sites in the city of Berlin, including 7 lakes (dark green), 7 ponds (light

green), 9 streams (light blue), and 9 rivers (dark blue), including two heavily polluted stream sites and

two heavily polluted river sites. Wastewater Treatment Plants (WWTP) are shown in orange, arrows

point to locations where the effluents are discharged (a). Scores of a Principal Component Analysis (PCA)

of DOM characteristics are shown as color gradients for all sites sampled in four seasons (b,c). The PCA

is based on DOC concentrations, all absorbance and fluorescence data, absolute component-specific

fluorescence intensities from PARAFAC, and data from size-exclusion chromatography. Different colours

indicate differences in DOM composition. Site codes are given in Table S1. Sites marked by asterisks (*)

were restricted to 3 seasons and hence excluded from the PCA.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

3.3.2 Physico-chemical field measurements and water sampling

We repeatedly sampled all 32 sites in each of four campaigns conducted over an annual cycle,

first in spring (April-May 2016), then in summer (July-August 2016), autumn (September-

October 2016) and winter (February-March 2017). All field visits occurred during base flow

conditions (Figure S5), We measured water temperature, pH, the dissolved oxygen (DO)

concentration and electrical conductivity using a hand-held WTW Multiprobe 3320 (pH320,

OxiCal-SL, Cond340i, Weilheim, Germany) or a smarTROLL probe (In-Situ, Fort Collins, CO,

USA). We also collected integrative water samples (2 L) from the upper 0.5 m water layer for

chlorophyll-a and DOM analyses. The water was kept cool in acid-washed polycarbonate

Nalgene bottles placed in a cooling box pending filtration in the laboratory (GF75, 0.3 μm

average pore size; Advantec, Tokyo, Japan) within 6 hours after sampling. Additional volumes

of surface water were filtered through pre-combusted glass fiber filters (GF75) directly in the

field. These filters were placed in acid-washed, pre-combusted (450 °C, 4 h) glass vials (15-20

mL) sealed with a PTFE septum in a screw-cap for later measurements of dissolved organic

carbon (DOC) concentrations, DOM fluorescence and absorbance, and DOM molecular size

distribution. The water passed through the filter was collected in acid-washed polyethylene

tubes for analyses of soluble reactive phosphorus (SRP), nitrate (NO3-), nitrite (NO2-),

ammonium (NH4+) and trace organic compounds (TrOCs). We also took unfiltered water

samples for total phosphorus (TP) analysis. For each variable, we collected three replicate

samples at each site in each season. We stored all samples in the dark in a cooling box during

transport. To preserve samples and remove all inorganic carbon, we acidified (pH 2) the water

for DOC, NO3-, NO2- and NH4+ analyses with 2 M HCl within 6 hours after sample collection.

DOC concentrations and DOM fluorescence and absorbance were measured within 24 h.

Filtered water for analyses of SRP, NO3-, NO2-, NH4+ and TrOCs was frozen at -20 °C.

3.3.3 DOM characterization

We determined total DOC concentrations by high-temperature catalytic combustion and

infrared spectrometry on a TOC-V Analyzer (Shimadzu, Kyoto, Japan), with a 0.5 mg L-1 limit

of quantification and a typical analytical precision of 3%. DOM absorbance and fluorescence

were simultaneously determined on an Aqualog instrument (Horiba Ltd, Kyoto, Japan) using

ultra-pure water as a blank. From each site and season, we measured three analytical replicates

of each of the three independent samples. We generated nine measurements (each of three

process replicates was measured 3 times) immediately after 3 blanks. The high level of

replication allowed identification of artefactual measurements and outlier removal following a

visual check of absorbance spectra and fluorescence excitation-emission matrices (EEMs). The

fluorescence data was expressed in Raman units, removing the need for an external

quantification standard.

CHAPTER 3

Iron can form stable complexes with DOC and interfere with optical DOC measurements, so

that the two variables are not independent (Maranger et al., 2003). We found that the quotient

of light absorbance at 420 nm (a420) and DOC concentration, a measure of the optical signal

returned per unit DOC (Weyhenmeyer et al., 2014), was indeed significantly correlated (p =

0.002) with the Fe concentration measured in the monitoring program of the Senate of Berlin

in two of our lakes (L5 and L7: 0.06 ± 0.03 mg/L), four of the rivers (H1, R1, R6 and R7: 0.30

± 0.14) and two of the streams (H3 and H4: 0.31 ± 0.14). The relationship explained 31% of the

overall variation (Figure S6). Consequently, Fe could have influenced our optical estimates of

DOC concentration. However, because our analysis rests on differences in DOM composition as

opposed to concentration (see below), it is unlikely that the presence of Fe notably influenced

the spatial and temporal patterns observed in our study.

We calculated several indices from the absorbance spectra (Table S3): the specific UV absorption

(SUVA254) as a proxy for DOM aromaticity (Weishaar et al. 2003), the ratio of absorbance at

250 and 365 nm (E2:E3) as an (inverse) indicator of molecular size (Peuravuori and Pihlaja

1997), the ratio of E4:E6 as an indicator of humification (Chen et al. 1977), and the ratio of slopes

(SR) computed from short and long wavelength regions (Loiselle et al. 2009) as another negative

correlate with DOM molecular weight. We used the fluorescence data to compute the freshness

index β/α (Table S3) (Wilson and Xenopoulos 2009b), which indicates the relative importance

of recently produced DOM (Parlanti et al. 2000). Furthermore, we calculated the fluorescence

index (FI) as the ratio of fluorescence intensities at the emission wavelengths of 470 and 520 nm

(obtained at an excitation wavelength of 370 nm), which has proved useful to distinguish the

relative contributions of terrestrial plants (FI~1.2) and microbes or algae (FI~1.4) as sources of

DOM (Cory and McKnight 2005; Cory et al. 2010; Fellman et al. 2010; Jaffé et al. 2008). Finally,

we computed the humification index (HIX) as a proxy for humic substances (Ohno 2002). EEMs

were used for PARAFAC, a multivariate three-way modeling approach decomposing EEMs into

individual fluorophores (Bro 1997; Stedmon and Bro 2008). We derived 7 components from a

total of 116 EEMs and compared their loading spectra with the OpenChrom/OpenFluor

database (http://www.openfluor.org) (Murphy et al. 2014). Prior to PARAFAC, we interpolated

missing data in Rayleigh scatter regions to expedite the modeling process (Bro 1997). The

calculations for PARAFAC were performed using Matlab (version 7.11.0, MathWorks) and the

DOMFluor Toolbox (1.7) following Stedmon & Bro (2008). We limited the number of

components to 10, rigorously checked residual EEM plots, and assessed the final models by

split-half validation (Figure S1) as recommended by Stedmon and Bro (2008).

The molecular size distribution of DOM was analyzed by liquid size-exclusion chromatography

in combination with UV and IR detection of organic carbon and UV detection of organic

nitrogen (LC-OCD-OND) (Huber et al. 2011b). The instrument was calibrated with IHSS

Suwannee River I Humic Acid and Fulvic Acid standards (International Humic Substance

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Society, St Paul, MN, USA). Carbon and nitrogen detectors were calibrated with potassium

hydrogen phthalate (C) and sodium nitrate (N). Limits of quantification were 0.1 mg C L-1, and

0.01 mg N L-1, analytical precision based on repeated standard measurements was better than

3%. We determined concentrations of three molecular size fractions: humic-like substances (HS-

C and HS-N reported in mg C L-1 and mg N L-1, respectively), high-molecular weight non-humic

substances (reported as HMWS-C and HMWS-N, in mg C L-1 and mg N L-1) and low-molecular

weight substances (LMWS, in mg C L-1).

To examine the molecular composition of DOM, we used ultrahigh-resolution Fourier-

Transform Ion Cyclotron Mass Spectrometry (FT-ICR-MS). We extracted DOM on Agilent

Bond Elut PPL solid-phase columns (Dittmar et al., 2008) from 1 L of filtered water acidified to

pH 2. We then diluted extracts to 10 µg L-1 C in 1/1 ultrapure water/methanol before broadband

mass spectrometry on a 15 Tesla Solarix FT-ICR-MS (Bruker Daltonics, Bremen, Germany) in

electrospray ionization negative mode (300 accumulated scans, ion accumulation time of 0.1 s,

flow rate of 240 µL/h). We performed internal mass calibration and exported the raw mass lists

from 150 to 1000 Da for further data processing using previously established R code (del Campo

et al. 2019). Briefly, we first applied a method detection limit similar to Riedel & Dittmar (2014)

before aligning m/z values across samples (Del Campo et al., 2019). Subsequently, we assigned

chemical formulas to mean m/z values assuming single-charged deprotonated molecular ions

and Cl-adducts for a maximum elemental combination of C100H250O80N4P2S2, respecting

chemical constraints. To eliminate doubtful formula assignments, we performed (i) an accurate

assessment of mass error including its partitioning into random and systematic components

(Savory et al. 2011): (ii) an exploration for stable isotope validation by daughter peaks (Koch et

al. 2007), and (iii) a homologous series assessment based on CH2, CO2 and H2O as chemical

building blocks for aliphatic, acid-based and alcohol-based elongation (Koch et al., 2007). To

condense the mass-spectrometric data, we grouped formulas into 12 non-overlapping molecular

groups (Lesaulnier et al. 2017) based on elemental composition and calculated the average

molecular mass, number of formulas (molecular richness) and total intensity for each of them.

In addition, we computed the double-bond equivalents (DBE) and the aromaticity index (AI) as

indicators of unsaturated compounds, and the molecular lability boundary (MLB) as a measure

of lability. Finally, we used van Krevelen plots to present the sum formulas derived from the

FT-ICR-MS data in a space defined by O:C (oxygen richness) and H:C (saturation) ratios. We

used random order of plotting to avoid bias due to systematic overplotting of thousands of

compounds with identical O:C and H:C ratios.

3.3.4 Additional water-chemical analyses

NO3-, NO2- and NH4+ were analyzed on a FIAcompact (MLE GmbH, Dresden, Germany). TP

was measured using the same technique with unfiltered water samples that were digested with

K2S2O8 (30 min at 134 °C). We measured chlorophyll-a concentrations spectrophotometrically

CHAPTER 3

(HITACHI U2900, Tokyo, Japan) following hot ethanol extraction (Jespersen and

Christoffersen, 1987) of three GF75 filters from each individual water sample. Concentrations

of 18 trace organic compounds (TrOCs) were determined by HPLC-MS/MS (Shimadzu, Kyoto,

Japan) (Zietzschmann et al., 2016). These included chemicals such as acesulfame (a sweetener),

benzotriazole (a corrosion inhibitor), and drug residues like carbamazepine and gabapentin

(Table S7).

3.3.5 Data analysis

We used repeated-measures ANOVA to test for differences among types of water bodies and

sampling periods (referred to as seasons hereafter) for a variety of response variables; as the

interaction between water body type and season was not significant we recomputed models

including main effects only. Furthermore, we assessed the importance of seasonal variation in

each water body type by computing a respective variance component using a type-II ANOVA

(aka variance component analysis) for data from each water body type with season and site ID

as random factors; this approach facilitates the assessment of temporal variation as a fraction of

total variation within each water body type. Normal distribution was assessed graphically by

quantile plots and histograms. For ANOVA, data were log(x) or √x -transformed to achieve

conditions of normality and variance homogeneity of the residuals.

For constrained multivariate analyses we considered land cover adjacent to the water bodies,

trophic state and micropollutant load as drivers of variation in DOM chemical composition. We

used the percentages of urban green space and paved areas as proxies for land cover,

concentrations of TP, NH4+, NO3- and chlorophyll a as a measure of trophic state; and the mean

TrOC concentration as a proxy for micropollutant load. We also performed a principal

component analysis with all the TrOCs.

We followed a three-step approach to analyze the spatio-temporal patterns of DOM

composition: First we identified major axes of variation in DOM composition by a PCA based

on quantitative indicators of DOM, analytically accessible fractions thereof or quantitative

proxies: DOC concentration, all absorbance and fluorescence indices, component-specific

fluorescence intensities from PARAFAC normalized to DOC, and the size-exclusion

chromatography data. Only the 27 sites sampled in all four seasons were included in this

analysis. All variables were standardized to a mean of zero with a variance of 1 to ensure equal

weighting, and projected onto the ordination space using Pearson correlations of the variables

with PCA axes in a distance biplot (sensu Legendre and Legendre, 2012). To explore spatial

patterns, we mapped PC1 and PC2 scores onto Berlin´s landscape using QGIS (QGIS

Development Team, 2017).

Second, we used the same dataset as the dependent matrix in a redundancy analysis (RDA) with

the set of potential drivers described above used as predictor variables. The goal of the RDA

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

was to identify potential drivers of DOM composition and thereby assess, reciprocally, whether

various DOM descriptors are ecologically informative. We started with the full RDA model and

forward-selected drivers (Legendre and Legendre 2012). For hypothesis tests in the RDA,

permutations were restricted to account for repeated measurements at the same sites across

seasons by first permuting sets of four seasonal measurements across sites and then permuting

across seasons within each site. To check our ability to identify drivers behind major variation

observed in DOM composition, we used Procrustes analysis to assess the similarity of PCA and

RDA ordinations, including a permutation-based test of the non-randomness of the achieved

superimposition (Mardia 1979; Peres-Neto and Jackson 2001).

Third, we exploited results of the FT-ICR-MS to facilitate interpretation of the two major axes

of variation in DOM chemical composition resulting from the PCA. The FT-ICR-MS data were

only available for three seasons and were purely compositional (relative intensities), as the many

thousands of compounds contained in the spectra cannot be calibrated to yield concentrations.

To link the quantitative and compositional datasets, we correlated PCA scores with compound-

specific relative intensities of the mass spectra. The compound-specific correlation coefficients

were then used as colour codes in van Krevelen plots. FT-ICR-MS-derived information such as

the richness or average weight of specific molecular groups was also projected onto the PCA

ordination space as arrows, provided correlation coefficients were >0.2. All statistical analyses

and graphs were made with R 3.2.4 (R Core Team 2016).

3.4. Results

3.4.1 Physico-chemical characteristics

Among all physico-chemical variables, only DOC concentration and temperature differed

significantly among types of water bodies (p<0.05 and p<0.001, respectively). Temperature

varied strongly across seasons, but still proved significantly different among water body types,

with lakes and rivers being warmer than ponds and streams. DOC concentrations did not vary

across seasons, but were significantly higher in ponds and streams than in lakes and rivers.

Ponds also showed the highest chlorophyll-a concentrations and rivers the lowest, but these

differences were not significant.

Separate ANOVAs for each water body type showed that seasonal variation in TP and NH4+

concentrations was highest in rivers and streams (Table S2). Seasonal variation in NO3-

concentrations was generally high, but systematic differences were neither detected among

seasons nor sites (Table S2). Seasonal variation of chlorophyll-a concentrations was also high

and similar across types of water bodies.

The analysis of TrOCs identified acesulfame, a widely used artificial sweetener (Buerge et al.,

2009), in 72 out of a total of 120 samples taken at 32 sites across all seasons (Table S8). Similarly,

CHAPTER 3

two corrosion inhibitors, benzotriazole and methylbenzotriazole (Cotton and Scholes 1967;

Tamil Selvi et al. 2003), occurred in 68 and 63 samples, respectively. Fifteen other TrOCs were

detected in at least 2 and up to 62 samples (Table S9). Rivers showed the highest concentrations

throughout the year. The first principal component of the PCA considering all TrOCs explained

61% of the total variance (Figure S3) and separated streams and rivers with higher

concentrations from ponds and lakes where concentrations of TrOCs were lower and often

undetectable, particularly in ponds (Table S9). The strong positive correlations between most

of the TrOCs suggested the applicability of a simple average TrOC concentration as a proxy for

micropollutant load in further analysis; this mean was computed across all TrOCs after z-

standardization of each TrOC for equal weighting.

3.4.2 DOM composition

PARAFAC modeling resulted in 7 components referred to as C1-C7 (Table S4, Figure S1).

Components C6 and C7 were previously found to be protein-like, whereas all other components

have been reported as humic-like (Table S4). In contrast to the standard physico-chemical

variables and results from size-exclusion chromatography (Table S7), the PARAFAC

components and absorbance and fluorescence indices generally showed significant differences

among water body types (Table S5 and S6).

The first axis of the PCA analyzing spatio-temporal patterns of DOM chemical composition

explained 34% of the total variance (Figure 2). PC1 was largely defined by the negative loadings

for C1 and C2 (representing humic substances originating from wastewater treatment),

SUVA254 and LMWS (Figure 2b). Furthermore, PC1 correlated positively with the absorption

slope ratio, E2:E3 (molecular size), β/α and HMWS-C. This axis separated water body types,

from lakes on the right to ponds, rivers, and finally streams on the left. The optical proxies

identified PC1 as a gradient spanning from lakes, where DOM had lower aromaticity and

contained more freshly produced material, to streams, which showed high aromaticity and low

proportions of fresh DOM. Pond P4, which was identified as an outlier because of particularly

high NH4+ concentrations, also showed a rather distinct DOM composition.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Figure 2 Ordination of sites (a) by PCA based on DOM characteristics (b): (i) indices derived from

measurements of absorbance (E2:E3 indicating molecular size, E4:E6 representing the humification ratio,

the slope ratio SR, and SUVA254) and fluorescence (freshness index β:α, fluorescence index FI, and

humification index HIX), (ii) PARAFAC components C1 to C7, and (iii) data from size exclusion

chromatography (humic-like substances HS, high-molecular weight non-humic substances HMWS, low-

molecular weight substances LMWS). (c) Potential drivers of DOM composition, that were used as

constraints in the RDA, were mapped onto the PCA ordination, with the significant constraints marked

by an asterisk (*). (d) FT-ICR-MS-derived indices and molecular groups mapped onto the PCA ordination

representing only groups correlated with PC1 or PC2 (r>0.2; oxygen richness O:C, saturation level

indicated by H:C, double-bond equivalents DBE, aromaticity index AI, molecular lability boundary MLB,

molecular groups g1 and g2 indicating black carbon without and with heteroatoms, g5 consisting of

unsaturated aliphatics, g7 representing saturated fatty acids, g8 and g9 denoting carbohydrates without

and with heteroatoms N, S or P, and g10 comprising peptides). The molecular group measures are either

average masses (marked by ‘_a’) or counts of molecules (marked by ‘_c’).

PC2 explained an additional 21% of the total variance and correlated positively with HMWS

(mg N/L) and β/α, and negatively with HIX. An exploration of spatio-temporal variation by

plotting site-specific PC scores (Figure 3) identified PC2 as the axis capturing temporal

variation, with the four seasons aligning vertically at most sites. Winter and summer had the

lowest and highest PC2 scores, respectively, with transitional seasons located in between. Thus,

higher proportions of humic substances in winter contrast with more labile DOM in summer.

In agreement with the variable-specific seasonal variance components, the degree of seasonal

differentiation differed among water body types also in multivariate space, being higher in

CHAPTER 3

streams and ponds than in the larger lakes and rivers (Figure 3b). Except for site H3, seasonal

variability was poorly reflected by PC1, which largely captured variation among individual

water bodies or water body types, separating flowing from standing waters. Visual inspection

of PCA scores mapped across Berlin (Figure 1b,c) did not reveal a spatial signature transcending

types of water bodies. RDA identified the areal percentage of green space adjacent to the water

bodies, TP, NH4+, NO3- and the mean TrOC concentration as significant predictors of DOM

composition (Figure S4). The resulting PCA and RDA ordinations for DOM were strongly

correlated (Procrustes rotation 0.73, p<0.001), suggesting that the considered predictors were

indeed major drivers of variation in DOM chemical composition.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Figure 3 (a) PCA biplot based on DOM absorbance and fluorescence indices, PARAFAC components and

size exclusion chromatography data from 4 contrasting types of urban freshwater bodies, including lakes,

ponds, rivers, and streams, in addition to two streams and two rivers specifically selected as highly

polluted sites. Each of the ellipses represents one sampling site that was visited 4 times, once in each

season. Site codes are given in Table S1. (b) Visual comparison of site-specific seasonal variation based on

the size, shape and orientation of ellipses, plotted separately per site. (c) Seasonal variation across sites

illustrated by ranking the sampling dates at each site according to the PC2 scores, as shown in the inset.

The stacked histograms show frequencies of the seasons across the four ranks. Summer samples tend to

produce high scores at most sampling sites, whereas winter samples tend to score low.

CHAPTER 3

High-resolution mass spectrometric analyses of samples from three seasons provided additional

insights into the chemical composition of DOM. Overall, we detected 6446 molecular formulas,

most of them representing molecular groups typical of humic material derived from soils. This

includes highly unsaturated O-rich compounds, polyphenols and other aromatic structures,

followed by unsaturated aliphatic polyphenols, and polycyclic aromatic compounds with

aliphatic chains. The van Krevelen plots revealed a positive correlation of lignin-like molecules

and carbohydrates with PC1 of DOM and identified these molecules as abundant in lakes (Figure

S2). In contrast, the negative association of proteins with PC1 was typical of streams.

Information on the molecular groups identified by FT-ICR-MS and projected on the PCA space

(Figure 2d) showed carbohydrates and sugars containing N, S or P to be positively related to

PC1. Furthermore, PC1 was negatively related to black carbon, polyphenols and polycyclic

aromatic compounds with aliphatic chains, which are all typical of soil-derived humic material,

as well as with unsaturated aliphatics, saturated fatty acids and peptides, indicating that all of

these molecular groups were more important in streams. Lastly, the computed molecular lability

boundary (MLB), carbohydrates, sugars without heteroatoms (N, S or P) and unsaturated

aliphatics were positively related to PC2, while AI, DBE, black carbon and polyphenols were

negatively related to PC2.

3.5. Discussion

3.5.1 Spatial patterns and drivers of DOM signatures

Our results show that the chemical composition of DOM in contrasting surface waters of the

metropolitan area of Berlin, Germany, is highly diverse. This reflects both aquatic-terrestrial

linkages and DOM transformations within the aquatic systems (Fonvielle et al. 2021). Clear

differences among the four types of water bodies we investigated were due to distinct signatures

of streams and rivers vs. ponds and lakes. This was revealed especially by the first principal

component (PC1) of a PCA (Figure 2), which reflects the dominant gradient defined by variation

in DOM composition across the 32 urban sites included in the study. Since optical measurements

play an important role in our analysis of DOM, it is important to consider potential interference

by iron. Elevated iron concentrations lead to brownification, similar to effects of allochthonous

DOM, and Fe and DOC also form stable complexes, so that the two variables are not

independent (Maranger et al., 2003). However, Fe data available from the Senate of Berlin for

two of our lakes (L5 and L7: 0.06 ± 0.03 mg/L), four of the rivers (H1, R1, R6 and R7: 0.30 ±

0.14) and two of the streams (H3 and H4: 0.31 ± 0.14), all concentrations were below the

threshold of 1 mg/L, suggesting that Fe increases may result in a420/DOC increases

(Weyhenmeyer et al., 2014).

Stream DOM exhibited higher aromaticity (as indicated by SUVA254) and lower amounts of

recently produced, low-molecular DOM (as indicated by the freshness index or the slope ratio)

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

than lakes at the opposite end of the gradient. This pattern matches results from agricultural

streams near Berlin, where SUVA254 values up to 3 L m-1 mg-1 were reported (Graeber et al.

2012) and from an urban river in southwestern Korea (SUVA254 values of 2.5 L m-1 mg-1) (Park

2009). The distinct signature is also reflected in other DOM components, such as the

fluorophore C2, which was more important in streams and identified as terrestrial humic

material (Murphy et al. 2011). Streams also showed higher levels of humic-like (C1) and protein-

like (C7) compounds, whereas higher values of the freshness index characterized lakes. These

patterns consistently indicate that the arrangement of sites along PC1 reflects a gradient of

allochthonous vs autochthonous sources of DOM. A corollary of this finding is that despite the

potentially pervasive influence of the urbanized surroundings, urban streams in particular are

more tightly linked to the terrestrial environment than urban lakes, just as is the case for flowing

and standing waters in natural landscapes (Larson et al. 2014).

In contrast to natural landscapes, however, the linkage of urban waters with their terrestrial

surroundings is mediated by paved surfaces and engineered flow paths, including roof run-off

into rain gutters, extensive (partially leaky) sanitation networks and sewage overflows in

WWTPs that are activated following heavy rainfall or snowmelt. The urban gradient from

allochthonous to autochthonous DOM sources we document could thus be driven by surface

run-off rather than soil seepage and subsequent delivery of DOM to surface waters via

groundwater Although we did not sample after major storms (Figure S5), we would expect

legacy effects of past runoff events to differ among sites, depending on the extent of green space

and impervious surface area in the surroundings of the sites. This interpretation is supported

by higher levels of proteins (Figure 2) characterizing the urban streams and rivers, as opposed

to soil-derived humic DOM signatures typical of unimpacted streams and rivers (Hutchins et al.

2017). The proteins could originate from surface runoff integrating various sources of urban

pollution but they might also derive from WWTPs, as implied by the nature of some of the

PARAFAC components (Table S4). For instance, the humic fluorophore C2 has been reported

in WWTP effluents that may be discharged into urban surface waters (Murphy et al. 2011).

Point-source inputs were also identified as drivers of DOM composition by the influence of

TrOCs in our RDA and their correlations with C2 and C7, all of which are components of

WWTP effluents.

Lakes differ from streams by a typically greater importance of autochthonous production. Since

this production is fostered by abundant nutrient supply (given sufficient light), elevated nutrient

concentrations should coincide with DOM signatures indicative of autochthonous carbon

sources. This pattern has been found in agricultural streams, where the freshness index β:α

indicating autotrophic activity was related to high nitrogen concentrations (Wilson and

Xenopoulos 2009b). However, it contrasts with the negative relation between nitrogen

concentration and the proportion of fresh DOM found across our study sites, where high

CHAPTER 3

nutrient concentrations were instead strongly related to DOM components of WWTP effluents.

This typically resulted in an allochthonous DOM character at high-nitrogen sites. Notably,

signatures like lower β/α in WWTP-impacted sites may also be a consequence of the highly

processed nature of DOM that underwent degradation in a WWTP.

Similarly, the TP concentration was significantly related to DOM composition in our RDA,

where phosphorus-rich water bodies also proved to have more allochthonous than

autochthonous DOM. This points to inputs from urban surface runoff rather than groundwater

inflow where long flow paths and residence times provide ample opportunities for phosphorus

immobilization. As with N, additional phosphorus may derive from WWTP effluents, as

suggested by the positive relationship between TP concentration and the fluorophore C2 as a

putative tracer of WWTP effluents (Murphy et al. 2011). Overall, the negative relationships

between nutrient availability and the importance of autochthonous components in the DOM

pool suggests that while streams and rivers may efficiently collect N and P from the urban

environment; lakes are more efficient at channeling nutrients into autochthonous production.

Thus, the autochthonous DOM signature in urban lakes appears to be largely independent of

nutrient supply and rather be facilitated by longer water residence times, higher water

temperature and favorable light conditions.

Our results on urban surfaces driving urban allochthonous DOM composition meet our

expectation that land cover notably influences the composition of DOM in urban surface waters

(Sankar et al. 2020; Williams et al. 2016). This conclusion is supported by results of our RDA,

which identified the presence of green spaces in the perimeter of the water bodies as a significant

influence. However, the relationship between land cover and DOM composition must be

interpreted with caution because all lakes were situated in areas with green spaces in their

surroundings, whereas streams ran through areas dominated by buildings and paved surfaces.

The urban running waters, more than lakes and ponds, thus received high surface runoff during

rain events, including high inputs of pollutants and allochthonous DOM.

Except for ponds and some lakes, all investigated water bodies had direct surface water

connections, which could result in spatial autocorrelation. In addition, spatial patterns may arise

from the prominent land cover gradients in Berlin, ranging from forested areas to densely

populated urban centers. Since the sampling design of our study does not lend itself to a formal

analysis of spatial autocorrelation, we explored spatial patterns with DOM proxies in maps

(Figure 1b,c) but found no obvious relationships. Instead, type-specific characteristics of the

water bodies were pronounced, largely independent of hydrological connections. Factors

potentially contributing to the resulting heterogeneity across the surface waters in the city

include specific local stressors such as point-source inputs of pollutants, spatially variable urban

surface runoff delivering allochthonous DOM, and hydraulic-engineering structures such as

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

sluices. Thus, our map of DOM composition (Figure 1b,c) could be interpreted as visualizing

heterogeneity in the conditions of urban surface freshwaters.

3.5.2 Seasonal patterns and drivers of DOM signatures

Seasonal variation in DOM signatures occurred in all types of water bodies mostly independent

from variation among the four water body types. With a few exceptions, H3 being the most

prominent example, seasonal variation of DOM composition was consistent across all water

body types. (Figure 3a,b), Assessed separately at each site (Figure 3b), DOM was generally

fresher in summer and autumn than in winter and spring, as indicated by higher ratios of β:α

and more HMWS-N as indicators of polysaccharides and proteins (Thurman 1985), whereas

humic matter was more abundant in winter, and the pattern in spring was not clear-cut. Our

rank-based analysis of PC2 scores (Figure 3c) suggests a consistent seasonal pattern of changes

in DOM composition across sites, which emerged even though the variation within individual

sites was limited along PC2.

At least four potential processes could account for the observed seasonal turnover in DOM

composition: exudates of aquatic primary producers, microbial and sunlight-induced

transformation of DOM, and terrestrial inputs from riparian vegetation (Cory et al. 2015;

Spencer et al. 2009), all of which could be influenced by the urban environment. Seasonal

variation in light conditions could be important in influencing DOM composition by primary

producers, independent of nutrient supply (see above), and temperature changes might also play

a role, especially in determining rates of microbial DOM transformations. Pulses of leaf litter

falling or swept or blown into urban water bodies could be an additional source of DOM varying

with season (Gessner et al. 1999). This holds particularly for urban green spaces and water

courses lined by woody riparian vegetation. However, quantification of the relative importance

of different drivers of seasonal patterns remains difficult based on the data currently available

for urban settings.

The ponds and streams included in our study showed higher and less predictable seasonal

changes in DOM composition than the lakes and rivers, as revealed by the pattern along PC2

(Figure 3). This indicates that the nature and degree of aquatic-terrestrial coupling in urban

settings leaves an imprint on seasonal changes in DOM composition. Therefore, the more

extensive the time series data from surveys of DOM dynamics, the better can they inform about

ecosystem conditions, complementing established procedures in water quality assessment and

monitoring.

3.5.3 DOM composition as a potential basis for urban surface water

monitoring

The fact that our analysis of DOM composition revealed specific characteristics of individual

water bodies underlines the potential value of DOM descriptors as indicators that could be

CHAPTER 3

included in water-quality assessment and monitoring. Some sites deviated from the general

pattern observed for water bodies of the same type. P4, for example, was formerly connected to

an old waste water treatment plant and appeared to be influenced by previously unrecognized

stormwater runoff. This legacy matches the particularly high levels of nutrients characterizing

this site, especially NH4+, combined with a distinct DOM composition. Similarly, S5, located

immediately downstream of a WWTP, although not specifically selected as a highly polluted

site, also showed a distinct DOM composition as reflected by its highly negative PC1 score

(Figure 2a), indicating that the allochthonous influence was likely the strongest among all sites.

Site R7 showed the same pattern as S5, and although not initially recognized as being affected

by a WWTP, its DOM composition revealed that it had received WWTP effluents, which has

actually happened since the end of 2015 (Nega et al., 2019). The distinct signatures at these

individual sites are thus a promising starting point for incorporating information on DOM

composition in water-quality assessment and monitoring. DOM optical indices would be highly

cost-effective to apply and yield information that is not easily obtained by classic approaches.

Robustness of such assessments would further increase when they are based on continuous time

series. This could strengthen the implementation of current legal frameworks such as the EU

Water Framework Directive aiming at an integrative water-quality assessment, including of

urban water bodies.

3.6. Conclusion

The composition of DOM analyzed in a suite of contrasting water bodies of a large metropolitan

area, the city of Berlin in Germany, is diverse, varying widely in molecular size and other

features related to the degree of allochthonous inputs and conveying a distinct urban character.

DOM features clearly differentiated water body types, from lakes with highly abundant

autochthonous DOM to streams with more allochthonous DOM. Seasonal variation of DOM

was prevalent in all water body types but likely to be driven not only by phenology but also by

urban influences such as nutrient supply, WWTP effluents, reduced leaf litter input or flashy

runoff resulting from sealed surfaces. Nutrient supply, the percentage of green space and

concentrations of trace organic pollutants (as proxies for point source influences) were identified

as drivers of DOM composition. In particular, simple optical measurements of DOM

characteristics were sufficient to detect WWTP effluents, a result that was corroborated by our

data on TrOCs. This suggests that optical analysis of DOM could be a useful approach to

complement current water-quality assessments and monitoring. Such analyses are fast,

inexpensive and easily implemented, and could be further supported by more sophisticated,

potentially automated analyses such as the mass-spectrometric quantification of TrOCs. DOM

composition can inform about processes both within water bodies and in the terrestrial

surroundings; therefore, water-quality assessments could benefit from integrating information

on DOM composition. Robustness of the approach would increase if the DOM assessments were

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

based on time series or even continuous monitoring, for which knowledge and technology are

already available; this could indeed strengthen assessments as implemented in legal frameworks

such as the EU Water Framework Directive.

Author contributions. All authors contributed to designing the study. CR and SH collected the

data. CR did the optical analysis and the PARAFAC modeling, GS carried out the FT-ICR-MS

analysis. CR and GS conducted the statistical analysis. CR led the manuscript writing, jointly

with GS. All authors discussed results and edited the manuscript.

Competing interests. The authors declare not to have a conflict of interest.

Supplement. The supplement related to this article is available online.

Data availability. The data is available at https://doi.org/10.5281/zenodo.6563595

Acknowledgments. We thank A. Köhler at the Senate Berlin (SenUVK) for water quality data,

authorities and private land owners for providing access to the study sites, C.N. Stratmann for

obtaining permissions, U. Mallok for nutrient analyses, C. Schmalsch for the LCOCD analysis,

S. Krocker and T. Fuss for the DOC analysis and T. Goldhammer for advice with chemical

analyses. C.N. Stratmann, Meinhold, I. Ajamil, G. Idoate, L. Thuile-Bistarelli, A. Sultan, R.

Schulte, E. Tupper, T. Fuss, R. del Campo, A. Wieland, and M. Bethke for field assistance. G.

Aschermann and A. Putschew kindly enabled TrOC analyses. Access to FT-ICR-MS and

associated expertise was generously provided by T. Dittmar during a stay of G. Singer at the

University of Oldenburg that was funded by the Hanse-Wissenschaftskolleg Delmenhorst.

Thank you also to K. Pypkins for support with GIS and to B. Kleinschmit for thoughts on the

sampling strategy and data analysis. This project was funded by the German Research

Foundation (DFG) through the Research Training Group ‘Urban Water Interfaces’ (UWI;

GRK 2032).

CHAPTER 3. SUPPLEMENT

Supplement of Chapter 3

Dissolved organic matter signatures in

urban surface waters: spatio-temporal

patterns and drivers

Table S1 Coordinates, land cover, origin and special features. Longitude is given in decimal degrees East

and latitude in decimal degrees North.

Site

Site name

Water

body

type

Latitude

Longitud

Agri-

culture

(%)

Fo-

rest

(%)

Urba

pave-

ment

(%)

Urban

green

space

(%)

Origin

Special

features

Teltowkanal

River

52.44239

13.32454

Artificial

Channelized,

WWTP

Teltowkanal

River

52.42642

13.52039

100

Artificial

Channelized,

WWTP

Wuhle

Stream

52.52562

13.57913

Natural

Tegeler Fliess

Stream

52.63442

13.38013

Natural

Biesdorfer See

Lake

52.50331

13.5497

Artificial

Obersee

Lake

52.54856

13.48972

Artificial

Ploetzensee

Lake

52.5438

13.33049

100

Natural

Gross

Glienicker

Lake

52.46417

13.11489

Natural

Havel

Lake

52.4431

13.14453

100

Natural

Schlachtensee

Lake

52.44066

13.21183

Natural

Müggelsee

Lake

52.43837

13.6451

Natural

Hoheheideteic

Pond

52.57694

13.16428

100

Natural

Protected

area

Hamburger

Teich

Pond

52.56738

13.44549

Artificial

Ruhwaldteich

Pond

52.52573

13.25998

Artificial

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Kienhorstbec

ken

Pond

52.57724

13.34556

100

Artificial

Former

WWTP

runoff input

Mittelfeldteich

Pond

52.61208

13.23045

100

Artificial

Protected area

Neurandteich

Pond

52.63883

13.27377

Artificial

Möwensee

Pond

52.55282

13.33545

Artificial

Müggelspree

River

52.42985

13.68912

100

Natural

Channelized

Landwehrkan

River

52.51935

13.31959

Artificial

Channelized

Spree

River

52.53613

13.21622

100

Natural

Channelized

Kuhlake

River

52.57817

13.16509

100

Natural

Protected

area

Neukölln

Canal

River

52.48936

13.43949

Artificial

Channelized

Spree

River

52.47137

13.49683

100

Natural

Channelized

Panke

River

52.5369

13.36759

Natural

Channelized,

WWTP

(from 2015)

Zingergraben

Stream

52.58209

13.38594

Artificial

Channelized

Schwarzer

Graben

Stream

52.56488

13.34918

Natural

Channelized

Graben 1

Buch

Stream

52.62384

13.46883

100

Artificial

Graben 73

Buchholz

Stream

52.62881

13.45315

100

Artificial

Erpe

Stream

52.45888

13.61245

Natural

WWTP

Koppelgraben

Stream

52.62065

13.41089

Unknown

Plumpengrabe

Stream

52.41513

13.5628

100

Natural

Channelized

CHAPTER 3. SUPPLEMENT

Table S2 Physico-chemical characteristics (mean ± SD and % variance explained) of four contrasting

types of water bodies in the city of Berlin. Means and standard deviations were computed across all

seasons and sites. The percentages of variance explained (% Var) refer to the effect of season within each

water body type, calculated by type-II ANOVA (aka variance component analysis), with season treated as

a random factor. F-values refer to results of repeated-measures ANOVAs testing for differences among

water body types (*** p<0.001, * p<0.05, ns = not significant).

Water

body

Temperature

DOC

NH4+

NO3-

Chlorophyll

Type

(°C)

Var

(mg/

Var

(mg/L)

Var

(mg/L)

Var

(mg/L)

Var

(μg/L

)

Var

Lakes

14.6 ±

6.9

7.5 ±

2.7

0.05 ±

0.05

0.07 ±

0.07

0.22 ±

0.36

6.2 ±

14.2

Ponds

13.7 ±

5.3

10.3

± 3.0

0.09 ±

0.07

0.27 ±

0.67

0.03 ±

0.06

7.3 ±

7.7

Rivers

15.2 ±

6.2

8.0 ±

1.6

0.10 ±

0.08

0.15 ±

0.14

1.12 ±

1.66

2.1

±2.9

Streams

11.3 ±

4.9

11.7

± 5.5

0.26 ±

0.31

0.36 ±

0.65

0.91 ±

1.53

5.3 ±

9.7

Fwater body

9.4***

3.8*

2.8ns

1.3ns

2.5ns

1.0ns

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Table S3 Description of absorbance and fluorescence indices.

Variable

Description

SUVA254

Proxy for DOM aromaticity (Weishaar et al. 2003)

E2:E3

Ratio of absorbance at 250 and 365 nm, as an (inverse) indicator of molecular size

(Peuravuori and Pihlaja 1997) (Chen et al. 1977)

E4:E6

Indicator of humification (Chen et al. 1977)

Ratio of slopes (SR) computed from short and long wavelength regions as another

negative correlate with DOM molecular weight (Loiselle et al. 2009)

Fluorescence index (FI) Ratio of the fluorescence intensities at the emissions 470 and

520 (obtained at excitation wavelength of 370nm). Indicator of DOM derived from

terrestrial plants (FI around 1.2) or from microbes or algae (FI around 1.4) (Cory

and McKnight 2005; Cory et al. 2010; Fellman et al. 2010; Jaffé et al. 2008)

HIX

Humification index (HIX) as a proxy for humic substances (Ohno 2002)

β/α

Freshness index β/α (Wilson and Xenopoulos 2009b), which indicates the relative

importance of recently produced DOM (Parlanti et al. 2000)

CHAPTER 3. SUPPLEMENT

Table S4 Designation, excitation (Ex) and emission (Em) wavelengths of PARAFAC components, and

the number of studies with matching components reported in OpenFluor (checked on the 28th March

2022) (Murphy et al. 2014).

PARAFAC

component

OpenFluor

reference

matches (0.95)

Explanation and selected

references

250

446

Humic-like, peak A (Coble 1996);

humic-like and recalcitrant (C1)

(Hansen et al. 2016)

250

500

Terrestial humic-like in waste water

treatment impacted water, (G1)

(Murphy et al. 2011); ubiquitous and

recalcitrant humic (C2) (Chen et al.

2017)

306

408

Humic-like, peak M (Coble, 1996);

humic-like (C3) (Stedmon and

Markager 2005b)

256

444

Terrestrial humic-like, suggested as

photo-refractory (Yamashita et al.

2010); (C2) terrestrial humic-like (C3)

(Williams et al. 2010)

250

382

Anthropogenic, microbial humic-like

(C6) (Williams et al. 2016)

294

352

Similar to tryptophan (C3) (Catalán et

al. 2015); protein-like, linked to

autochthonous production (C3)

(Amaral et al. 2016)

276

326

Protein-like, peak B (Coble, 1996);

waste water treatment protein (C2)

(Teymouri 2007)

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Table S5 Variables of absorbance and fluorescence analyses (mean ± SD and % variance explained) in

contrasting types of urban surface waters. Means and standard deviations were computed across all

seasons and sites. The percentages of variance explained (% Var) refer to the effect of season within each

water body type, calculated by a type-II ANOVA (aka variance component analysis), with season treated

as a random factor. F-values refer to results of repeated-measures ANOVA testing for differences among

water body types (***p<0.001, **p<0.01, *p<0.05, ns = not significant). Abbreviations explained in

Table S3.

Water

body

type

SUVA254

E2:E3

E4:E6

HIX

β/α

Var

Lakes

1.55

0.39

8.99

2.14

3.02±

1.34

1.38 ±

0.27

1.61 ±

0.08

0.77 ±

0.08

0.86

0.09

Ponds

2.14

0.51

6.65

1.24

3.12±

0.74

1.20 ±

0.19

1.52 ±

0.06

0.83 ±

0.04

0.70

0.04

Rivers

2.25

0.15

7.04

0.98

4.40±

14.27

0.017 ±

0.002

1.68

±0.11

0.85 ±

0.03

0.79

0.09

Streams

2.50

±0.52

6.32

8 ±

0.87

3.53±

2.31

0.97 ±

0.13

1.63 ±

0.14

0.86 ±

0.05

0.73

0.10

Fwater

body

11.8*

5.8*

2.6 ns

9.2***

3.5*

4.9**

5.5**

CHAPTER 3. SUPPLEMENT

Table S6 PARAFAC components (mean ± SD and % variance explained) in contrasting types of urban

surface waters. Means and standard deviations were computed across all seasons and sites. The

percentages of variance explained (% Var) refer to the effect of season within each water body type,

calculated by a type-II ANOVA (aka variance component analysis), with season treated as a random factor.

F-values refer to results of repeated-measures ANOVA testing for differences among water body types

(***p<0.001, **p<0.01, *p<0.05, ns = not significant).

Water

body

type

Var

Lakes

0.17 ±

0.10

0.13 ±

0.06

0.24 ±

0.12

0.28

0.12

0.31 ±

0.18

0.19 ±

0.10

0.18

0.11

Ponds

0.24 ±

0.14

0.24 ±

0.11

0.40 ±

0.22

0.46

0.21

0.52 ±

0.38

0.16 ±

0.11

0.25

0.14

Rivers

0.51 ±

0.34

0.34 ±

0.17

0.66 ±

0.37

0.44

0.12

0.61 ±

0.26

0.32 ±

0.22

0.25

0.14

Streams

0.76 ±

0.47

0.57 ±

0.30

1.00 ±

0.58

0.85

0.63

1.08 ±

0.73

0.37 ±

0.29

0.35

0.18

Fwater

body

6.3**

10.4**

7.6***

7.2*

8.4***

2.2n.s

2.6n.s

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Table S7 Results of size exclusion chromatography (mean ± SD and % variance explained) of samples

from contrasting types of urban surface waters. Means and standard deviations were computed across all

seasons and sites. The percentages of variance explained (% Var) refer to the effect of season within each

water body type, calculated by a type-II ANOVA (aka variance component analysis), with season treated

as a random factor. F-values refer to results of repeated-measures ANOVA testing for differences among

water body types (***p<0.001, **p<0.01, *p<0.05, ns = not significant). HS, humic-like substances;

HMWS, high-molecular weight non-humic substances; and LMWS, low-molecular weight substances.

Water

body type

HMSW

LMWS

(mg C/L)

Var

(mg N/L)

Var

(mg C/L)

Var

(mg N/L)

Var

(mg C/L)

Var

Lakes

0.96 ± 0.74

0.11 ± 0.07

4.14 ± 1.49

0.25 ± 0.11

0.83 ± 0.26

Ponds

1.32 ± 0.60

0.16 ± 0.06

6.29 ± 2.42

0.33 ± 0.13

1.19 ± 0.46

Rivers

0.59 ± 0.20

0.09 ± 0.03

5.21 ± 0.97

0.31 ± 0.09

1.10 ± 0.41

Streams

0.73 ± 0.45

0.10 ± 0.05

7.21 ± 3.75

0.41 ± 0.27

1.48 ± 0.62

Fwater body

4.2*

2.9ns

1.3ns.

3.7*

CHAPTER 3. SUPPLEMENT

Figure S1 Emission and excitation wavelengths of PARAFAC components. Solid lines represent

emission spectra, dashed lines excitation spectra. Lines in different shades of grey refer to models using

different sample sub-sets of a split-half validation analysis.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Figure S2 Van Krevelen plots showing all molecules (sum formulas) identified by FT-ICR-MS analysis

of DOM samples collected at 32 urban sites over three seasons (summer, autumn and winter). Colour

indicates molecule-specific Spearman correlation coefficients of the relative intensities of each compound

with the first (a) and second (b) axis of the PCA shown in Figures 2 and 3. The data points were plotted

in random order to avoid bias resulting from identical O:C and H:C ratios for many sum formulas.

CHAPTER 3. SUPPLEMENT

Table S8 Trace organic compounds (TrOCs) analyzed in samples collected in urban surface waters. LLoQ

= Limit of Quantification. Frequency refers to the number of occasions where concentrations exceeded

the LLoQ.

Acrony

LLo

(µg

/L)

Frequency

Name

Description

ACS

0.1

Acesulfame

Sweetener

ATS

0.05

Amidrotrizoic

Radiocontrast agent

BTA

0.1

Benzotriazole

Corrosion inhibitor

BZF

0.1

Benzafibrate

Lipid-lowering agent

CBZ

0.05

Carbamazepine

Anticonvulsant

DCF

0.05

Diclofenac

Analgesic/anti-inflammatory agent

FAA

0.1

4-formylamin

metabolite of

metamizol

Analgesic

GAB

0.1

Gabapentin

Drug for epilepsy treatment/pain

killer

GPL

0.05

Gabapentin-lactam

Derivate of gabapentin

IOM

0.1

Iomeprol

Radiocontrast agent

IOP

0.01

Iopromide

Radiocontrast agent

MBT

0.1

Methylbenzotriazole

Corrosion inhibitor

MTP

0.1

Metoprolol

Beta blocker

PRI

0.05

Primidone

Anticonvulsant

SMX

0.1

Sulfamethoxazole

Antibiotic

VAL

0.1

Valsartan

At1-receptor antagonist

VLX

0.1

Venlafaxine

Antidepressant

VSA

0.1

Valsartan acid

Antihypertensive agent

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Table S9 Mean concentrations and standard deviations of Trace Organic Compound (TrOC) per water

body type. See Table S8 for full names. BZF, SMX and VLX were always below the limit of quantification

(LLoQ) and are hence omitted from the table.

TrOC concentration (µg/L)

Acronym

Lakes

Ponds

Rivers

Streams

ACS

0.23 ± 0.17

0.15 ± 0.16

0.28 ± 0.17

0.78 ± 1.35

ATS

0.08 ± 0.12

<LLoQ

0.74 ± 1.12

0.43 ± 0.88

BTA

0.34 ± 0.51

0.38 ± 0.91

2.37 ± 3.31

2.16 ± 3.61

CBZ

0.07 ± 0.08

<LLoQ

0.37 ± 0.48

0.41 ± 0.66

DCF

<LLoQ

0.97 ± 1.38

0.88 ± 2.14

FAA

0.15 ± 0.23

<LLoQ

1.25 ± 1.66

2.10 ± 4.25

GAB

0.27 ± 0.36

<LLoQ

0.42 ± 0.43

0.74 ± 1.12

GPL

0.10 ± 0.17

<LLoQ

0.19 ± 0.42

0.13 ± 0.18

IOM

0.18 ± 0.28

<LLoQ

1.18 ± 2.36

1.44 ± 2.91

IOP

0.09 ± 0.19

0.01 ± 0.02

0.27 ± 0.32

1.46 ± 3.79

MBT

0.27 ± 0.39

0.11 ± 0.24

0.91 ± 1.01

0.69 ± 1.27

MTP

<LLoQ

0.47 ± 0.58

0.63 ± 1.55

PRI

0.03 ± 0.02

<LLoQ

0.16 ± 0.22

0.26 ± 0.52

VAL

<LLoQ

0.39 ± 0.44

0.97 ± 3.65

VSA

0.70 ± 1.02

<LLoQ

3.22 ± 3.84

3.33 ± 5.72

CHAPTER 3. SUPPLEMENT

Figure S3 Principal Component Analysis (PCA) of 32 urban sites in the city of Berlin over four seasons

(a) and Trace Organic Compounds (TrOCs) (b). Site S5 had extreme PC1 and PC2 scores; the site was

included in the analysis but is not presented in the biplot to better visualize variability among the other

sites. Abbreviations of the TrOCs (B) are explained in Table S7.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Figure S4 Redundancy Analysis (RDA) of urban sampling sites (a) visited 4 times over one year, the

DOM characteristics included in the analysis (b) and the predictor variables (c), the last marked by an

asterisk (*) when significant. DOM characteristics include (i) absorbance and fluorescence indexes (E2:E3,

molecular size, E4:E6, indicator of humification, SR, slope ratio, β:α, freshness index, SUVA254 and HIX,

humification index), (ii) PARAFAC components (C1 to C7), and (iii) fractions derived from size exclusion

chromatography (HS, humic-like substances; HMWS, high-molecular weight non-humic substances; and

LMWS, low-molecular weight substances).

CHAPTER 3. SUPPLEMENT

Figure S5 Precipitation and flow at a site within the city of Berlin during the study period, with the grey

boxes indicating the four sampling periods.

Dissolved organic matter signatures in urban surface waters: spatio-temporal patterns and drivers

Figure S6 Relationship between iron (Fe) and the absorbance at 420nm relative to DOC(a420/DOC)

CHAPTER 4

Carbon dioxide emissions across an urban

aquatic network

This study was submitted to Limnology and Oceanography as:

This is the preprint version of the article.

4.1 Abstract

Freshwater carbon dioxide (CO2) emissions represent a globally important carbon flux.

However, despite rapid urbanization, the CO2 fluxes from urban waters remain poorly

constrained and challenges remain for reliable upscaling, particularly due to the diversity of

urban aquatic ecosystems. Using floating chambers to measure instantaneous fluxes as well as

monitor them over multiple days, we estimated emissions seasonally at 32 sites across Berlin,

encompassing ponds, lakes, streams, and rivers. As potential drivers of CO2 emissions, we

evaluated land cover, nutrients, dissolved organic matter (DOM) composition, chlorophyll-a,

and micropollutant concentrations. CO2 fluxes from ponds and lakes ranged from -22 to 585

gC-CO2m-2y-1, aligning with previous studies. Rivers and streams exhibited lower fluxes (22 to

809 gC-CO2 m-2y-1), likely attributable to low gas exchange in the channelized lowland running

waters. Fluxes were higher with more aromatic DOM, pointing to respiration of labile

allochthonous DOM, likely sourced from wastewater. Contrary, fluxes were lower at abundant

fresh DOM and higher phosphorus concentration, suggesting drawdown of CO2 by primary

production. This coincided with a higher percentage of urban paved area. Extrapolation to

Berlin's aquatic network yielded an annual emission estimate of 8.5 Gg of C-CO2 from urban

streams. An implication of this study is that urban rivers, lakes, and ponds should be considered

when establishing global budgets of CO2 emissions from surface waters.

Romero González-Quijano, C., Herrero Ortega, S., del Campo, R., Casper, P., Gessner,

M. O., Goldhammer, T. & Singer, G. A. Carbon dioxide emissions across an urban aquatic

Network. Submitted to Limnology and Oceanography

Carbon dioxide emissions across an urban aquatic network

4.2 Introduction

Accurate quantification of the global carbon (C) cycle remains a priority in the current era of

climate change (IPCC 2023). Although covering only 3% of the global land surface (Vachon et

al. 2020a), inland waters contribute a sizeable share to Earth’s C budget (Butman et al. 2016;

Cole et al. 2007; Le Quéré et al. 2018; Tranvik et al. 2009). Lakes, ponds, streams and rivers act

as intermediaries between the continents to the oceans, transporting an estimated 0.95 Pg C

yr−1 of terrestrially derived C to the sea (Borges et al. 2015; Drake et al. 2018; Holgerson and

Raymond 2016; Sawakuchi et al. 2017), but at the same time assuming important roles in the

sequestration, transformation and outgassing of various C species. This includes the emission

of gaseous carbon species such as carbon dioxide (CO2) received by inland waters from soils and

groundwater influx, a portion of which is subsequently vented to the atmosphere (Raymond et

al. 2013). In addition, inland waters receive sizeable amounts of dissolved and particulate

organic carbon from their catchments, part of which is eventually converted to CO2 before being

emitted to the atmosphere as well (Butman and Raymond 2011; McDonald et al. 2013; Raymond

et al. 2013; Striegl et al. 2012).

An important consequence of inland waters receiving and transforming inorganic and organic

carbon is that most are oversaturated with dissolved CO2, which results in diffusive CO2 fluxes

to the atmosphere due to concentration gradients at the water-air interface. The first global

estimates of CO2 evasions from inland waters with a focus on rivers (Cole and Caraco 1998;

Richey et al. 1980) have regularly been updated (Aufdenkampe et al. 2011; Battin et al. 2023;

Battin et al. 2009; DelSontro et al. 2018; Raymond et al. 2013; Tranvik et al. 2009). As a result,

the latest report of the Intergovernmental Panel of Climate Change (IPCC 2023) recognizes

inland waters as a significant component of the global C cycle, functioning as a net source of

CO2 to the atmosphere. However, all estimates of global CO2 emissions from inland waters

suffer from large uncertainties. This is partly due to methodological reasons but also because of

biases towards natural and rural environments especially in Europe and North America (Pickard

et al. 2021), whereas data on urban environments are scarce.

Another important source of uncertainty is the notoriously large temporal variability of CO2

fluxes on different time scales. This variability hampers the consolidation of datasets, including

measurements made at various temporal and spatial scales, which are needed to deliver robust

long-term estimates over large areas. Importantly, dissolved CO2 concentrations in inland

waters are influenced by various event-driven changes that superimpose variation at the daily

and seasonal time scale (Clow et al. 2021; Ulseth et al. 2018). However, studies capturing

seasonal and annual variability of CO2 emissions from freshwaters are limited (Finlay et al. 2019;

Huotari et al. 2011; Shao et al. 2015). Furthermore, nighttime emissions have been found to

exceed emissions during the day in both lakes (Shao et al. 2015) and rivers (Gómez-Gener et al.

CHAPTER 4

2021), although the data available to date are insufficient to consider incorporating diel variation

of CO2 fluxes in large-scale estimates (Gómez-Gener et al. 2021). Improving the temporal

resolution of gas flux measurements during day and night time would thus deliver better models

to map current and project future emissions (Martinsen et al. 2018).

A strategy to improve upscaling estimates of CO2 fluxes is through the analysis of drivers and

the use of mechanistic models. Identifying the potential drivers of CO2 fluxes in inland water

bodies is critical for accurate inclusion in global CO2 budgets, understanding the drivers of

spatial and temporal variability will deliver better sampling strategies and will help improve

upscaling models (Ray et al. 2023).

More than half of the world's population currently lives in cities, and this number is projected

to increase in the future, with 68% of the world population expected to be urban by 2050 (UN,

2018). Stressors such as inputs of pollutants and nutrients or land-cover changes involving soil

imperviousness and a range of other catchment characteristics potentially make urban waters

important hotspots of CO2 emissions (Gallo et al. 2014; Kaushal et al. 2014b). Multiple evidence

indicates indeed that rapidly urbanizing areas may enhance CO2 emissions from water bodies

(Park et al. 2018; Wang et al. 2017; Yu et al. 2017). Quantifications of their contribution to CO2

emissions (Gu et al. 2022; Wang et al. 2021) have been linked to trophic state (Sepulveda-

Jauregui et al. 2018a; Vachon et al. 2020b). However, multiple other factors and their interaction

characterizing urban environments may play an additional role in controlling emission rates

from urban lakes, ponds, rivers and streams. Identifying these drivers and sources of CO2 in

urban aquatic ecosystems will help to better understand the underlying biogeochemical

processes and improve total emission estimates (Smith et al. 2017).

In the present study, we aim to estimate CO2 fluxes from an urban aquatic network comprising

lakes, ponds, rivers and streams in a major metropolitan area, taking into account variation at

both the daily and seasonal scale. We hypothesize flowing water bodies will have higher

emission fluxes because of their greater supply of C and nutrient resources from the city. On the

contrary, we hypothesize higher temporal variability will occur in lentic ecosystems due to

stronger changes in functioning at daily scale (e.g. changes of primary production by light).

Fluxes might be influenced by urban stressors as nutrients or organic matter. Finally, we want

to derive an upscaled estimate of the aquatic CO2 fluxes from a whole city.

Carbon dioxide emissions across an urban aquatic network

4.3 Methods

4.3.1 Sampling design

We sampled 32 sites within the city boundaries of Berlin, Germany (Figure 1, Table S4,

Supporting Information). The site selection is detailed in Herrero Ortega et al. (2019) and

Romero González-Quijano (2022). Briefly, we sampled seven randomly chosen lakes, ponds,

rivers and streams each and also included four additional sites identified as polluted based on

monitoring data. Two of them were classified as streams (H3 and H4) and two as rivers (H1 and

H2). To characterize the sites, we calculated the percentage of urban paved areas, green areas,

forested areas and agricultural land for a 50 m buffer zone around each site using QGIS (QGIS

Development Team, 2017). The spatial data were kindly provided by the Senate of Berlin.

Figure 1 Map of Berlin, Germany, showing the city’s aquatic network, land use, and the 32 sampling

sites selected in the present study.

4.3.2 Estimation of CO2 fluxes

We estimated CO2 fluxes with two complementary approaches to cover two different time scales

(Table 1), as detailed further below. Briefly, in the first approach, we deployed a floating flux

chamber connected to a Los Gatos ultra‐portable greenhouse gas analyzer (UGGA 24P and

30P; Los Gatos Research, Inc., Mountain View, CA, USA) to measure instantaneous CO2 flux.

We sampled at each site once per season over one year from April 2016 to March 2017. We

continuously recorded the CO2 concentration in the chamber headspace for at least 15 min at

each site and repeated this procedure three times to obtain triplicate direct flux estimates (Table

1). In the second approach, we used inexpensive CO2 sensors mounted inside custom-built

floating equilibration chambers that were exposed for approximately one week, measuring every

CHAPTER 4

5 minutes, at each site and in each season, except in winter when the sensors failed at the low

temperatures during this time (Table 1). We used these continuous measurements to indirectly

estimate continuous time series of CO2 flux over one week, which we then used to estimate the

CO2 flux representative for an average day in each season. These measurements also allowed us

to determine diel variation of CO2 flux by calculating the average daily variance of the flux over

one week. We explored the influence of water body type and season on the CO2 fluxes

determined with both approaches and extrapolated the measured fluxes to the whole city of

Berlin.

Table 1 Comparison of CO2 flux measurement methodologies using two types of floating chambers

deployed at 32 sites in the city of Berlin.

Flux

measurement

approach

Instantaneous, direct

Continuous, indirect

Chamber type

Floating flux chamber

Floating equilibration chamber

Sensor

Los Gatos portable gas analyzer

connected to flux chamber

Senseair CO2 logger mounted in

equilibration chamber headspace

Method

Direct flux measurement using the

floating-chamber method

Computation of flux by the boundary

layer method based on reconstructed

time series of CO2 concentrations in

water and air concentrations

Time scale

Three flux estimates per site and

season within one hour between 8 and

12 am,1 measurement per second over

a duration of 15 min

Flux estimates derived from CO2

concentrations measured at 5 min

intervals to capture diel and daily

variation over periods of 1 week

Advantages

Rapid instrument deployment and

data collection, no risk of sensor loss,

accurate data

More representative estimate of mean

fluxes, capture of flux variation at diel

and daily time scale, cheap sensors

Disadvantages

Low temporal and spatial resolution,

lack of replication, expensive

instrument

Limited data accuracy, sensor failure at

low temperature, risk of sensor loss or

damage, need of many sensors and

chambers to cover spatial and temporal

variability

Carbon dioxide emissions across an urban aquatic network

4.3.3 Theory of chamber operation and flux estimations

The temporal change of the CO2 partial pressure inside the headspace of flux chambers results

from a dynamic equilibration process. Short-term measurements taken immediately after

deployment of a flux chamber on a water body oversaturated with CO2 typically shows an

approximately linear increase in CO2 concentration followed by a plateau when CO2

concentration reaches equilibrium between water-air interface. The pseudo-linear increase at

the beginning is used to measure emission rates with flux chambers, whereas the equilibrium

level reached in the equilibration chambers deployed for a prolonged period serves to compute

the CO2 concentration in the water. However, if gas exchange between water and air is slow

and CO2 concentrations vary greatly over time, for instance because of light-dependent

fluctuations of photosynthetic carbon fixation, then equilibrium conditions (i.e. periods of zero

net exchange) may only be transient, because the CO2 partial pressure measured in the

headspace constantly lags behind the concentration in the water. Furthermore, the resulting

incomplete equilibration leads to a dampened amplitude in the time series of CO2 concentration

in the water.

The gas flux F (mol m-2 s-1) across the water-air interface is given by:

𝐹 =(𝐶𝑂2(𝐴𝑄)−𝐶𝑂2(𝐻𝑆)) ∙ 𝑘 Eq (1),

where k is the gas exchange velocity (m s-1), CO2(AQ) the concentration in the water, and CO2(HS)

the equilibrium concentration in the water at a given concentration in the air (mol m-3). Positive

and negative fluxes indicate whether the water is a net source or sink of CO2 to the atmosphere.

CO2(HS) can be computed from the relative concentration in air (ppm) commonly used to report

trace gas concentrations and atmospheric pressure:

𝐶𝑂2(𝐻𝑆)=𝑘𝐻∙𝑥𝐶𝑂2(𝐻𝑆)∙10−6 ∙𝑃 Eq (2),

where kH is Henry's constant (mol m−3 atm−1), xCO2(HS) is the measured molar (or volumetric)

fraction (ppm) in the chamber headspace, and P is the atmospheric pressure (atm).

In the confined chamber headspace, the flux F (mol m-2 s-1) across the water surface leads to a

dynamic change of CO2 concentration:

𝑑𝑥𝐶𝑂2(𝐻𝑆)

𝑑𝑡 =𝐹∙𝐴∙𝑉𝑚∙106

𝑉 Eq (3),

where V (m3) and A (m2) are the volume and area of the chamber, respectively, Vm is the molar

gas volume (m3 mol-1) computed from the ideal gas law, as R*T/P. The change of xCO2 in the

headspace is thus directly proportional to the flux F, which in turn depends on the concentration

CHAPTER 4

gradient between water and headspace. This leads to non-linear behavior of xCO2HS over time

that is best described by combining Eq 1 and 3:

𝑑𝑥𝐶𝑂2(𝐻𝑆)

𝑑𝑡 ∙𝑉

𝐴∙𝑉𝑚∙106=(𝐶𝑂2(𝐴𝑄)− 𝑘𝐻∙𝑥𝐶𝑂2(𝐻𝑆)∙10−6 ∙𝑃)∙𝑘 Eq (4).

Given an increasing (or decreasing) trend of xCO2(HS) in the headspace of a floating chamber,

this equation can be rearranged,

𝑘𝐻∙𝑥𝐶𝑂2(𝐻𝑆)∙10−6 ∙ 𝑃 = −𝑑𝑥𝐶𝑂2(𝐻𝑆)

𝑑𝑡 ∙𝑉

𝐴∙𝑉𝑚∙106∙1

𝑘+𝐶𝑂2(𝐴𝑄) Eq (5),

to determine k and CO2(AQ) by linear regression with Y=ki×X+b, where Y=kH×xCO2(HS)×10-

6×P, X=-dxCO2(HS)/dt×V/(A×Vm×106), ki=1/k, and d=CO2(AQ). The differential quotient

dxCO2(HS)/dt is approximated by ∆xCO2(HS)/ ∆t computed as the difference between consecutive

data points in the time series or as a local slope estimate spanning a larger time window.

As most time series show a prolonged phase of linear increase of xCO2HS directly after chamber

deployment (i.e., dxCO2HS/dt remains quasi-constant), a simplified approach can be used that

consists of directly applying Eq (3) (Cole et al. 2010; Lorke et al. 2015):

𝐹 =𝑑𝑥𝐶𝑂2(𝐻𝑆)

𝑑𝑡 ∙𝑉

𝐴∙𝑉𝑚∙106=𝑑𝑥𝐶𝑂2(𝐻𝑆)

𝑑𝑡 ∙𝑃∙𝑉

𝐴∙𝑅∙𝑇∙106 Eq (6),

where dxCO2HS/dt is the slope of the time series during its pseudo-linear initial phase. R is the

universal gas constant (0.082 L atm K−1 mol−1), T is the air temperature (K), and A is the area

covered by the chamber (m2).

Eq (4) also allows dynamic modeling of CO2(AQ) when extended time series and an estimate for

k are available. Eq (4) is rearranged for this purpose to:

𝐶𝑂2(𝐴𝑄)=𝑑𝑥𝐶𝑂2(𝐻𝑆)

𝑑𝑡 ∙𝑉

𝐴∙𝑉𝑚∙106∙𝑘 +𝑘𝐻∙𝑥𝐶𝑂2(𝐻𝑆)∙10−6 ∙ 𝑝 Eq (7),

where dxCO2(HS)/dt can again be approximated as ∆xCO2(HS)/ ∆t, xCO2(HS) is the average of the

two considered time points, and k is the temperature-corrected (see Eq (8) below) gas-exchange

velocity. The resulting CO2(AQ) is valid for the midpoint between the consecutive time points.

This approach allows translating the xCO2(HS) time series into two time series for xCO2(AQ) and

CO2(AQ)

4.3.4 Direct measurements of instantaneous CO2 flux

To determine instantaneous fluxes, we connected a flux chamber to an ultra‐portable

greenhouse gas analyzer (UGGA 24P and 30P; Los Gatos Research, Inc., Mountain View, CA,

Carbon dioxide emissions across an urban aquatic network

USA) in a closed loop and measured the CO2 concentration every second during three 15-min

periods on a single day per site and season. We lifted the chamber between measurements to

assure atmospheric background concentration at the start of each run. All measurements were

made in the morning between 8 and 12 a.m. to minimize the risk of any diel changes confounding

comparisons among sites. The resulting data supported application of Eq (6) to estimate flux.

We then used these flux estimates to calculate the gas exchange velocity (k) according to Eq (1),

using actual measurements of CO2(HS) and CO2(AQ) made with the Los Gatos analyzer. CO2(AQ)

was measured by equilibrating a volume of water with air in the headspace of a closed vial by

vigorous shaking before injecting a gas sample from the headspace into a closed loop connected

to the Los Gatos analyzer (Wilkinson et al. 2018) (see Supporting Information, heading one).

To enable comparisons of gas transfer velocities among sites, we standardized k to k600 using

Schmidt number scaling:

𝑘600 =𝑘(600

𝑆𝑐)−0.5 Eq (8),

where k is the temperature-specific gas exchange velocity and Sc is the temperature-adjusted

Schmidt number (Jähne et al. 1987).

4.3.5 Indirect estimation of continuous flux using equilibration chambers

To determine flux rates indirectly over an extended period, we deployed equilibration chambers

equipped with ELG CO2 loggers (SenseAir AB, Delsbo; Sweden) (Bastviken et al. 2015) for

about one week at each site per season and recorded the CO2 molar fraction in the chamber

headspace (xCO2(HS)) at 5 min intervals. The ELG CO2 loggers use a non-dispersive infrared

CO2 sensor covering a concentration range of 0−5000 ppm. Due to losses of chambers or sensors

during deployment in the field, the final dataset comprised data from 37 weeks. The one-week

time series of xCO2(HS) per site and season were used to derive a continuous time series of CO2

flux, using Eq (7) to compute CO2AQ and Eq (1) to Eq (2) to compute fluxes.

To estimate the gas exchange velocity k, we employed different approaches to account for

differences between standing and flowing waters. We considered various empirical models using

combinations of fetch (Vachon and Prairie 2013) and wind speed (Cole and Caraco 1998) for

ponds and lakes, and slope, flow velocity, water depth and channel width for streams and rivers

(Raymond and Cole 2001) (Table S6, Supporting Information). Slopes of streams and rivers

were derived from a topographic map, assuming a minimum slope of 0.01 m/m for very flat

terrain. Selection of the most appropriate empirical model was guided by comparing the k600

values (Table S6, Supporting Information) with k600 calculated from the instantaneous flux

CHAPTER 4

measurements. For lakes and ponds, we followed Vachon (2013), which resulted in estimates of

k600 according to:

𝑘600 =2.51+1.48 ∙ 𝑈10 +0.39∙𝑈10 ∙𝑙𝑜𝑔10 𝐴𝐿 Eq (9),

where U10 is the wind speed at 10 m above surface (m s-1) and AL refers to the lake area (m2).

These k600 values were positively correlated (r=0.43 p< 0.05) with the estimates of k600 obtained

with the flux-chamber method. Wind speed at 10 m above surface (U10) was extrapolated from

wind speeds measured at 5 m above ground (see Supporting Information, heading two) at a

monitoring station close to Lake Müggelsee (site L7, Table S4, Supporting Information). This

resulted in a time series of k600 for each lake and pond.

Since none of the calculated k600 values for streams and rivers (Table S6, Supporting

Information) was correlated to the k600 values measured with the flux chamber, we used the k600

obtained with the flux chamber method. Wind speed appears not to influence k in streams and

rivers, where water velocity and thus discharge-related characteristics are most influential,

which we assumed not to change during one-week deployment periods. This assumption was

supported by our field observations of constant discharge and is further justified by (i) the rather

low influence of evapotranspiration by riparian vegetation in urban streams and (ii) lack of major

precipitation events during our measurements. Thus, we obtained a weekly time series of k for

each stream and river for each season, where temporal variability was only due to temperature

fluctuations.

Finally, we assessed variability of the fluxes from two different points of view: Firstly, we

explored the temporal variability of CO2 fluxes at different time intervals ranging from season

to ours: seasonal variability, day to day variability and diel variability.

We calculated seasonal standard deviation (SD), day-to-day SD and diel SD. Additionally, we

assessed the percentages of variance explained (% Var), that refer to the effect of season, day to

day and time of the day within each water body type, computed by type-II ANOVA (variance

component analysis), with season, day and time of the day treated as random factors. Secondly,

we explored whether variability of the fluxes resulted from the dynamics of the CO2

concentration gradients, or from k. Finally, we calculated the percentage of variability

explained by k and ΔCO2.

4.3.6 Field sampling and laboratory analyses

We used a WTW Multiprobe 3320 (pH320, OxiCal-SL, Cond340i, Weilheim, Germany) or a

smarTROLL probe (In-Situ, Inc., Fort Collins, CO, USA) for field measurements of pH,

temperature, electrical conductivity and dissolved oxygen concentration, once per season. We

Carbon dioxide emissions across an urban aquatic network

also measured water depth and wind speed at each occasion. Furthermore, we collected triplicate

water samples that were filtered (MFS GF75, 0.3 μm nominal pore-size, Advantec Co., Ltd.,

Tokyo, Japan, pre-combusted at 450°C for 4 hours) and stored in a dark cool box until analysis.

Additional unfiltered water samples were collected for analysis of total phosphorus (TP) and

chlorophyll a. Filtered samples for the analysis of dissolved organic carbon (DOC) concentration

and the fluorescence, absorbance and molecule size distribution of dissolved organic matter

(DOM) were stored in acid-washed and pre-combusted (450 °C for 4 hours) glass vials sealed

with a PTFE septum cap. Filtered water for analyses of nitrate (NO3-), nitrite (NO2-), ammonium

(NH4+) and trace organic compounds (TrOCs) was collected in Falcon tubes and frozen until

analysis. We acidified samples for DOC, NO3-, NO2- and NH4+ with 2M HCl to pH 2 within 6

hours after collection. Samples were analyzed for NO3-, NO2-, NH4+ with an analytical module

FIAcompact (MLE GmbH, Dresden, Germany). TP was analyzed in the same way, after

digestion of unfiltered water samples with K2S2O8 (30 min at 134 °C).

Chlorophyll-a concentration was determined spectrophotometrically (Hitachi U2900; Tokyo,

Japan) following filtration of 1-L water samples in the laboratory within 6 hours after collection

and extraction with ethanol (Jepersen 1987). DOC concentration and DOM absorbance and

fluorescence were measured within 24 hours after collection on a a Shimadzu TOC-V Analyzer

and a Horiba Aqualog, respectively. The fluorescence data were used for PARAFAC modelling

which was carried out in a previous study (Table S1, Supporting information) (Romero

González-Quijano et al. 2022). Furthermore, we analyzed the molecular size distribution of

DOM by liquid size-exclusion chromatography in combination with UV combustion and IR

detection of organic carbon and UV detection of organic nitrogen (LC-OCD-OND) (Huber et

al. 2011b; Romero González-Quijano et al. 2022). Finally, concentrations of a total of 18 trace

organic compounds (TrOCs) were determined by high performance liquid chromatography-

tandem mass spectrometry (HPLC-MS/MS, Shimadzu, Kyoto, Japan) following Zietzschmann

et al. (2016).

4.3.7 Data analysis

We used linear mixed models to test for differences in log-transformed CO2 fluxes among

seasons and water body types, and the interaction of both, considering site as a random effect.

We analyzed the importance of seasonal and daily variability in each water body type by

calculating a variance component using a Type-II ANOVA (variance component analysis) for

data from each water body type with day, season and site ID as random factors; this approach

facilitated the assessment of seasonal and daily variability as a fraction of total variation within

each water body type. We then calculated the variance component from gas exchange velocity

(k) and the concentration gradient (ΔCO2). We graphically evaluated whether the assumptions

of normal distribution and variance homogeneity of the residuals were met by inspecting

CHAPTER 4

histograms and quantile plots of log(x)- or √x-transformed data. To investigate the controls on

the CO2 flux, we condensed all potential predictor variables by three complementary principal

component analyses (PCA). The first PCA used variables describing DOM composition,

including DOM indices derived from absorbance and fluorescence analyses, size-exclusion

chromatography and PARAFAC. The second PCA used physical and water-chemical variables

(NH4+, NO3-, NO2-, chl a, TP), as well as land cover, water temperature and water depth. Finally,

the third PCA focused on Trace Organic Compounds (TrOC). We z-standardized all variables

to ensure equal weighting before running the PCAs. We then used the principal components

(PC) with eigenvalues >1 from the three PCAs as predictors in three Random Forest (RF)

analyses, a flexible, non-parametric regression belonging to the Classification and Regression

Tree analysis (CART) family (Feld et al. 2016). We carried out a separate analysis for each of

the three response variables: instantaneous CO2 flux, average of the one-week continuous flux

measurements, and daily variance of the one-week continuous flux measurements.

We also extrapolated the CO2 fluxes to the total surface water area of Berlin to obtain the total

CO2 emission footprint for one year. This was achieved by calculating the total surface area of

each type of water body in the city and multiplying it by the estimated total seasonal emission

from each water body type the number of days in each season, excluding 55 days in winter when

the sites were ice-covered, and finally summing up the seasonal estimates across the four types

of water bodies. We then estimated the total annual emissions from each type of water body by

multiplying the seasonal total emission from each water body type (g C-CO2 m-2) with its

respective total surface area. We finally calculated the total CO2 emission footprint of Berlin's

surface waters as the sum of the annual emissions from each of the four types of water body.

The uncertainties were calculated using error propagation rules at each step of the calculations.

4.4 Results

4.4.1 Effects of water body type and season on CO2 fluxes

Fluxes varied greatly over the seasons and between types of water bodies (Figure 2). Average

fluxes from lakes and ponds exceeded those for rivers and streams showing much higher rates

but also a much higher variation for instantaneous than for continuous measurements (Table

2). Instantaneous and continuous fluxes were positively correlated (R = 0.69, p < 0.001; Figure

S1, Supporting Information), but differed markedly in magnitude: Instantaneous fluxes were on

average 2.5 times higher than the average continuous fluxes. However, differences depended on

the type of water body. Estimates were strongly correlated with a slope near 1 for streams and

rivers (Figure S1, Supporting Information, Table 2), whereas the correlation for ponds and lakes

was not significant (Figure S1, Supporting Information).

Carbon dioxide emissions across an urban aquatic network

Figure 2 Map of instantaneous CO2 fluxes (g C-CO2 m-2 d-1) at 32 sites distributed across the city of

Berlin, Germany, in four seasons.

Linear mixed models did not reveal significant differences among seasons or types of water

bodies for any of the fluxes, although average CO2 fluxes from streams in summer and autumn

tended to be higher than from the three other types of water bodies. However, fluxes

significantly differed when we merged flowing and standing waters (p < 0.05 for instantaneous

fluxes, p < 0.01 for continuous fluxes), with higher average fluxes determined by continuous

measurements than standing waters throughout the year (Table 2, Figure 3). CO2 fluxes

determined by instantaneous measurements ranged from -0.45 (lake L2 in summer) to 7.32 g C-

CO2 m-2 d-1 (stream H4 in spring) with an average of 0.65 ± 0.91 g C-CO2 m-2 d-1 across the four

seasons and 32 sites. Continuously measured fluxes ranged from -0.18 (lake L4 in spring) to

1.52 g C-CO2 m-2 d-1 (stream H3 in spring) with an average across sites and seasons of 0.41 ±

0.07 g C-CO2 m-2 d-1 (Figure 3). Most of the sites were net sources of CO2; however, a few lakes,

ponds and streams acted as CO2 sinks during summer (Figure 3) when fluxes were assessed by

instantaneous measurements. On the contrary, only one lake acted as a CO2 sink, and that only

in spring when fluxes were continuously measured for a week.

CHAPTER 4

Figure 3 Summary of instantaneous and continuous measurements of CO2 fluxes from four different types

of water bodies in the city of Berlin, Germany, determined in four seasons.

4.4.2 Variability of CO2 fluxes

The collection of continuous measurements of CO2 fluxes also allowed us to compare the

variability of the measured xCO2HS with the computed xCO2AQ. The average variance of xCO2AQ

was 4.2 times higher than that of xCO2HS (Figure 4 C,D). Differences in variance for xCO2AQ and

xCO2HS depended on water body type. When k at a given site was high, the amplitude of xCO2AQ

exceeded that at sites where k was low.

In general, continuous flux measurements showed much higher day-to-day and diel variability

for lakes and ponds than for streams and rivers, where diel variability was near 0 (Figure S2,

Supporting Information, Table 2). In contrast, seasonal variability of continuous measurements

was very similar across all water bodies. Instantaneous measurements also gave very similar

estimates of seasonal variability among water body types. Only rivers showed lower seasonal

variability than the other three types of water bodies.

Carbon dioxide emissions across an urban aquatic network

100

Figure 4 Selected high-resolution time courses of water temperature and wind speed (A,B), xCO2(AQ) and

xCO2(HS) (C,D), and CO2 and k from an urban stream (S3) in spring (A,C,E) and an urban pond (P5) in

summer (B,D,F), illustrating variability at different scales.

For standing waters, k generally accounted for a greater portion of the variability in CO2 flux

than ΔCO2, especially for the continuous flux measurements (Table 2). This contrasts with

flowing waters, where most of the variability was caused by ΔCO2 for both types of flux

measurements (Table 2).

Table 2 Mean and SD of CO2 fluxes from surface waters in the city of Berlin, Germany, made by instantaneous and continuous measurements. Seasonal SD, day-to-

day SD and diel SD. The percentages of variance explained (% Var) refer to the effect of season, day to day and time of the day within each water body type, calculated

by type-II ANOVA (variance component analysis), with season, day and time of the day treated as random factors. The percentage of variability explained by k and

ΔCO2. NA* Data insufficient for calculations

Instantaneous flux measurements

Continuous flux measurements

Water

body type

Mean flux

(g C-CO2

m-2 d-1)

SD of

seasonal

flux (g C-

CO2 m-2 d-

k600

(%)

ΔCO2 (%)

Mean flux (g

C-CO2 m-2 d-1)

SD of

seasonal

flux

(g C-CO2

m-2 d-1)

Seasonal

variance

(%)

SD of day-to-day flux

(g C-CO2 m-2 d-1)

Day-to-

day

variance

(%)

SD of daily

flux (g C-CO2

m-2 d-1)

Diel

variance

(%)

k600 (%)

ΔCO2 (%)

Lake

0.41 ±

0.80

0.49 ±

0.41

48 ± 5

52 ± 5

0.27 ± 0.48

NA*

0.033 ± 0.032

NA*

0.037 ± 0.042

NA*

92.0 ± 0.5

8.0 ± 0.5

Pond

0.56 ±

0.56

0.44 ±

0.25

56 ± 17

43 ± 17

0.24 ± 0.13

0.46 ± 0.25

72 ± 40

0.075 ± 0.065

28 ± 40

0.083 ± 0.046

0.11 ±

0.16

15.5 ± 0.5

River

0.46 ±

0.38

0.48 ±

0.39

37 ± 1

63 ± 1

0.51 ± 0.44

1.54 ± 0.50

100 ± 0

0.001 ± 0.001

0 ± 0

0.006 ± 0.006

0 ± 0

30.2 ± 2.7

69.8 ± 2.7

Stream

0.79 ±

0.59

0.81 ±

0.95

42 ± 5

58 ± 5

0.68 ± 0.51

1.93

100

0.002 ± 0.001

0.007 ± 0.007

26.0 ± 5.4

74.1 ± 5.4

121

Carbon dioxide emissions across an urban aquatic network

102

4.4.3 Comparison of Berlin CO2 fluxes with previous studies

Flux estimates from Berlin´s standing waters based on instantaneous and continuous

measurements ranged from -0.45 to 3.38 g C-CO2 m-2 d-1, similar to the range of rates

determined in other standing waters distributed across 150 sites around the world (-0.18 to 2.90

g C-CO2 m-2 d-1) (Figure 5, Table S8, Supporting Information). Fluxes from Berlin´s flowing

waters covered ranged from -0.13 to 7.32 g C-CO2 m-2 d-1, however, values reported from other

flowing water ranged 0.34 to 19.2 g C-CO2 m-2 d-1. (Figure 5).

Figure 5 Instantaneously (A) continuously measured CO2 fluxes (B) from water bodies in the city of

Berlin, Germany, in comparison to fluxes reported in published studies (C).

4.4.4 Annual CO2 emission from an urban aquatic network

Extrapolation of the fluxes measured continuously over one-week periods per season to a whole

year resulted in emission estimates that differed greatly among types of water bodies, (Table 3).

Extrapolation of the instantaneous flux measurements led to broadly similar results (Table 3).

These differences were mainly driven by the large differences in the surface area covered by the

different types of water bodies. Extrapolation to the entire aquatic network of the city results in

an annual emission estimate of 8.5 ± 1.3 Gg C-CO2 for the instantaneous and 4.9 ± 2.0 Gg C-

CO2 for the continuous flux measurements (Table 3).

CHAPTER 4

103

Table 3 Estimates of annual CO2 emissions (± 95% CL) from four types of water bodies in the city of

Berlin based on fluxes measured at four occasions either during 15-min deployments of flux chambers

(instantaneous measurements) or during continuous one-week measurements with equilibration

chambers.

Water

Annual CO2 emissions (Gg C-CO2)

Total surface

body type

Instantaneous

measurements

Continuous

measurements

area (km2)

Lakes

3.6 ± 0.7

0.8 ± 1.2

29.7

Ponds

0.4 ± 0.1

0.2 ± 0.2

2.1

Rivers

4.3 ±0.3

3.7 ± 0.3

21.4

Streams

0.2 ± 0.2

0.2 ± 0.3

0.8

Total

8.5 ± 1.3

4.9 ± 2.0

54.0

4.4.5 Drivers of CO2 fluxes and flux variability

The PCAs performed to summarize DOM data (Tables S1-S2, Supporting Information) and

physico-chemical (Table S3, Supporting Information) as potential drivers of CO2 (Figure 6,

Table 4) resulted in up to four principal components (PC) with eigenvalues >1. The PCA on the

DOM data explained 36% (PC1), 19% (PC2), 15% (PC3) and 12% (PC4) of the total variance,

and the four PCs from the PCA on the physico-chemical data explained 21, 14, 11 and 10% of

the variance, respectively. In the PCA on the TrOC data, only PC1 and PC2 had an eigenvalue

>1, explaining 35% (PC1) and 15% (PC2) of the total variance.

Carbon dioxide emissions across an urban aquatic network

104

Figure 6 Principal component analyses (PCA) summarizing potential drivers of CO2 fluxes from four

types of urban water bodies: A) Analysis based on dissolved organic matter (DOM) characteristics derived

from absorbance/fluorescence measurements, size-exclusion chromatography and PARAFAC (Table S1-

S2 Supporting Information; B) analysis based on water depth and temperature as well as land-cover and

physico-chemical variables such as NH4+, NO3-, NO2-, chl a, and TP (Table S3, Supporting Information;

and C) analysis based trace organic compounds as summarized in Table S5, Supporting Information.

CHAPTER 4

105

Table 4 Description and interpretation of PCA axes identified in the final random forest (RF) models as

potential drivers of CO2 fluxes. DOM: dissolved organic matter, CHEM: physico-chemical variables.

PC axis

PCA

Variables

PC description

Interpretation

PC1

DOM

Negative: PARAFAC

components C1, C2,

C3, SUVA254

Highly aromatic, humic-

like DOM of terrestrial

origin (including

wastewater treatment

plant effluents)

Gradient from allochthonous

aromatic DOM in streams and

rivers (-PC1) to

autochthonous labile DOM in

lakes (+PC1)

Positive: slope ratio,

E2:E3, β/α, HMWS-

Low molecular weight

DOM of aquatic microbial

origin

PC1

CHEM

Negative: urban paved

and TP

Gradients of depth and

land-use.

Depth-land cover gradient,

ranging from deep standing

waters in green spaces (+PC1)

to shallow water bodies, with

large surrounding paved areas

and high TP concentrations (-

PC1)

Positive: depth and

percentage of natural

areas in the

surroundings

PC2

CHEM

Negative: agricultural

land use

Chlorophyll-a gradient

Water temperature and

primary production gradient

Positive: water

temperature,

chlorophyll-a

concentration

PC4

CHEM

Negative: Wind

Separation of water bodies

by main abiotic factors

explaining k in lakes (wind)

or streams (temperature)

Separation of lakes with high

wind influence to warm

streams

Positive: water

temperature

The RF analysis of instantaneous CO2 fluxes identified PC1 from the DOM PCA and PC4 from

the PCA based on physico-chemical data as the most important drivers (Table S7, Supporting

Information). Partial plots from RF showed that instantaneous fluxes decreased with negative

PC1 scores of the DOM PCA and remained relatively constant when the scores were positive,

indicating that fluxes decreased along the aromaticity gradient. Elevated fluxes were associated

with low proportions of fresh DOM low and high DOM aromaticity (Figure 7). Partial plots

also showed that instantaneous fluxes were relatively constant with negative PC4 scores of the

physico-chemical PCA, but to increase with positive PC4 scores (Figure 7). This result indicates

that instantaneous fluxes were positively related to temperature (Table 4, Table S7, Supporting

Information).

Carbon dioxide emissions across an urban aquatic network

106

Similarly, the RF model for the continuous flux measurements was mainly explained by PC1 of

the DOM and physico-chemical PCAs. Continuous fluxes were also negatively related to PC1

of the DOM PCA and positively related to PC1 from the physico-chemical PCA (Figure 7),

indicating higher fluxes at deeper sites.

Figure 7 Partial plots of the main drivers identified in random forest models for instantaneousCO2 fluxes,

continuous fluxes, and daily variability of the continuous fluxes.

Daily variability of the continuously measured CO2 fluxes was mainly explained by PC2 and

PC1 of the physico-chemical PCA (Figure 7 E, F). Daily variability was positively related to

PC2; however, it mainly increased with positive PC2 scores, indicating higher daily variability

at elevated temperature and chlorophyll a concentration (Figure 7E). Daily variability decreased

with PC1 scores, indicating higher daily variability at sites with high proportion of paved areas

and low TP concentrations (Figure 7). None of the response variables (CO2 fluxes and daily flux

variability) showed a relationship with PC1 of the TrOC PCA, PC2, PC3 or PC4 from the DOM

PCA, nor PC3 from the physico-chemical PCA.

CHAPTER 4

107

4.5 Discussion

4.5.1 Temporal patterns

The current global estimate of aquatic CO2 emissions is poorly constrained at present (Pilla et

al. 2022). Large-scale CO2 flux estimates have traditionally been derived from CO2

concentrations in water (DelSontro et al. 2018; Raymond et al. 2013), generally calculated from

pH or alkalinity, an approach that typically overestimates fluxes (Abril et al. 2015) and

underrates flux variability (Attermeyer et al. 2021). Improving the accuracy of estimates thus

requires both improved methods and a consideration of temporal and spatial variability. The

two complementary methods used in the present study address this gap by assessing temporal

variation of CO2 fluxes at the seasonal, daily and diel scale across multiple sites in a large city.

We found that lotic and lentic water bodies responded differently to temporal variations on CO2

fluxes. Streams responded strongly to seasonal effects by increasing CO2 fluxes during summer

and autumn. Warmer temperatures during the summer and autumn can increase the metabolic

activity of aquatic organisms and microbial communities, leading to higher rates of organic

matter decomposition and CO2 production (Findlay 2021). Furthermore, the higher input of

organic material, such as plant litter, into the streams in autumn (Lidman et al. 2017; Pozo et al.

1997) can also boost microbial decomposition, contributing to higher CO2 fluxes (Amaral et al.

2021). On the other hand, lakes and ponds reacted stronger to diel and day-to-day variation.

The longer water residence times in lentic ecosystems compared to rivers and streams, allows

for more time for biogeochemical processes to occur (Brooks et al. 2014). This can lead to

increased diel and day-to-day variation in CO2 fluxes in lakes and ponds. Additionally, lakes and

ponds often have a higher density of aquatic plants and algae. These organisms can undergo

photosynthesis during the day, leading to CO2 uptake, and then release CO2 through respiration

at night, resulting in diel variation in CO2 fluxes. Rivers and streams, with faster flow rates, may

have lower densities of these organisms and, as a result, exhibit relatively stable CO2 fluxes on

a daily basis. Therefore, for accurate estimates, it is essential to sample both day and night in

lakes and ponds on multiple days or use an integrative method that averages across diel and

daily time scales, as demonstrated in our study.

Water bodies did not only differ on temporal variability but also on the mechanisms driving

GHG emissions. Wind speed exerted a strong control on gas exchange velocity in lakes and

ponds of Berlin, thus regulating CO2 daily dynamics in lentic waters (Prytherch and Yelland

2021). On the other hand, in rivers and streams, the variability related to the concentration

gradient (ΔCO2) had a greater impact on CO2 fluxes than the variability associated with gas

exchange velocity (k). This result contrasts with some previous studies on streams, such as

(Kokic et al. 2015) and (Wallin et al. 2011) which found gas exchange velocity to be more

influential. The difference in our findings could be attributed to the location of our research sites

Carbon dioxide emissions across an urban aquatic network

108

in an urban setting within a flat lowland terrain. In this type of environments, river

channelization, along with very gentle channel slopes, can significantly reduce the gas exchange

velocity by limiting the natural turbulence needed for efficient gas exchange. Our study

highlights the nature of CO2 flux regulation in aquatic ecosystems. Understanding these nuances

is essential for accurate modeling and management of carbon dynamics in these diverse aquatic

environments.

4.5.2 Spatial patterns

Rivers and streams are often subject to a higher input of allochthonous organic matter but also

dissolved inorganic carbon, which can explain the increased CO2 emissions observed here in

comparison to lakes and ponds. The higher spatial connectivity of streams and rivers with the

landscape make them great conduits for terrestrial CO2 emissions (Hotchkiss et al 2015). In

addition, in these flowing water bodies, the decomposition of externally derived organic material

plays a significant role in CO2 production (Saarela et al. 2022) , as opposed to lakes and ponds,

where CO2 production is predominantly reliant on autochthonous sources of organic matter,

such as aquatic plants and algae (Tranvik et al. 2009). Interestingly, our observations revealed

distinct patterns of CO2 dynamics across various aquatic environments. Some lakes, ponds, and

streams only acted as carbon sinks during the spring season, while rivers emitted carbon dioxide

consistently throughout the year. This difference in carbon flux behavior underlines the need

for a comprehensive approach to capture the intricacies of CO2 dynamics.

While we could compare instantaneous carbon fluxes across all sites, it is important to note that

our ability to obtain continuous flux data was limited due to the loss of several flux chambers,

particularly at the lake sites. The choice between utilizing instantaneous flux measurements or

continuous measurements depends on the research objectives and priorities. If achieving broad

spatial coverage is of utmost importance, instantaneous flux measurements may be appropriate.

On the other hand, if capturing daily and day-to-day variations in CO2 flux is the primary goal,

continuous flux measurements, are crucial for a more comprehensive understanding of carbon

dynamics in these aquatic ecosystems.

4.5.3 Potential drivers of CO2 fluxes

DOC quantity has been previously related to the concentration of CO2 in streams (Kaushal et

al. 2014a; Lapierre et al. 2013; Lennon 2004; Zhu et al. 2012); however, is still unclear how DOM

composition may affect CO2 fluxes from freshwaters to the atmosphere, especially in urban

settings. We found DOM composition to have a stronger influence on CO2 fluxes than simply

DOC concentration. In this study, fluxes were higher when aromatic compounds were

prominent in the DOM pool, and lower when fresh DOM was abundant, possibly driven by the

respiration of labile allochthonous DOM supplied, particularly with effluents from wastewater

CHAPTER 4

109

treatment plants. Indeed, we found C2, a humic-like component related to WWTP effluents,

was positively related to the CO2 fluxes. This result reinforces other studies indicating that

labile organic matter of anthropogenic origin may increase CO2 production and emissions from

urban surface waters (Griffith and Raymond 2011; Zhao et al. 2016) or that high concentrations

of CO2 in effluents from WWTPs lead to increased CO2 outgassing (Alshboul et al. 2016b). In

contrast to our results, respiration of relatively fresh DOM, less than 5 years old, was found to

be the dominant source of CO2 outgassing from rivers (Mayorga et al. 2005).

The percentage of urban paved area and TP concentration were negatively related to CO2 fluxes

in our study, in contrast with the positive relationship of CO2 emissions and TP concentration

previously found in lakes (Allesson et al. 2021). Physical characteristics of the water bodies we

examined also appeared to be important determinants of CO2 fluxes, especially water depth,

which was positively related to CO2 fluxes, although it must be acknowledged that the deepest

water bodies were larger lakes located mainly in forested areas, where the extent of paved

surfaces was small and nutrient concentrations lower than in the other urban water bodies we

investigated. Depth has been previously recognized as a factor influencing CO2 fluxes for

standing waters (Kankaala et al. 2013; Macklin et al. 2018). Percentage of paved surface was

negatively related to CO2 fluxes in our study, on the contrary, percentage of urban land has been

previously positively related to CO2 emissions from rivers (Gu et al. 2022). Temperature was

another influencing factor of CO2 fluxes in our study, fluxes were higher with higher

temperatures, as has been previously reported (Audet et al. 2020; Peacock et al. 2021). Other

environmental factors such as flow velocity, discharge or channel slope, which all affect water

exchange velocity, proved not to play important roles in our study. This lack of influence is in

contrast to more natural rivers and streams and could be due to the extensive channelization

and lowland character of the running waters we investigated. Regarding the drivers of the

variability of the fluxes, elevated temperatures generally stimulate microbial activity, which, in

turn, leads to more dynamic daily fluctuations in CO2 production and release, as we found in this

study. Additionally, higher chlorophyll-a levels were associated with increased daily fluctuations

in CO2 fluxes. Chlorophyll-a is an indicator of phytoplankton abundance (Boyer et al. 2009), and

its elevated presence often signifies greater biological activity, which can contribute to higher

CO2 variability. Interestingly, the location of water bodies played a significant role in daily

variability. Areas with a low percentage of paved surfaces, buildings, and other impervious

structures exhibited higher daily variability in CO2 fluxes. Paved surfaces and impervious

structures tend to increase runoff and disrupt natural processes that can influence the

concentration variability of CO2. In more natural or less urbanized areas, the ability of water

bodies to regulate carbon dioxide dynamics was more pronounced, leading to greater daily

variability. In summary, our findings indicate that physical factors, such as water temperature,

chlorophyll-a concentration, and the urban environment's influence on runoff, play a crucial role

Carbon dioxide emissions across an urban aquatic network

110

in shaping the daily variability of CO2 fluxes. Understanding the interplay between these factors

is essential for a comprehensive grasp of carbon dynamics in aquatic ecosystems.

Our study attempted to cover the aquatic network of a whole city, including all types of water

bodies, which is rarely reported in the literature. The streams and rivers showed lower emission

rates than found in other studies (Table S8, Supporting Information). Low discharge, flow

velocities and inflow of groundwater supersaturated with CO2 resulting from respiration during

soil and groundwater passage likely contributed to this outcome. Our results are similar to a

recent study carried out in the municipality of Beijing, where rivers from the urban area showed

lower emissions than global average (Wang et al. 2023). These results highlight the importance

of considering urban water bodies as a distinct group when estimating large-scale GHG

emissions from freshwaters. Traditionally, urban surface waters have not been considered when

estimating global CO2 fluxes, our estimation showed that freshwaters account for 0.17% of the

total estimated emissions from the city of Berlin in 2017 (UBA 2017).

4.5 Conclusion

The data we collected in four seasons across the aquatic network of the city of Berlin suggest

that the magnitude and variability of CO2 emissions from urban freshwaters may differ from

emissions typical of water bodies in natural and agricultural landscapes. An implication of this

finding is that urban streams, rivers, lakes and ponds should be considered distinct categories of

water bodies when establishing global budgets of CO2 emissions from surface waters. Rapid

growth of urban sprawl is a global phenomenon and urban waters in Berlin and elsewhere are a

net source of CO2 to the atmosphere, even though the total emissions we estimated were lower

than in other studies.

Acknowledgments. We thank A. Köhler at the Senate Berlin (SenUVK) for water quality data,

authorities and private land owners for providing access to the study sites, C.N. Stratmann for

obtaining permissions, U. Mallok for nutrient analyses, C. Schmalsch for the LCOCD analysis.

C.N. Stratmann, Meinhold, I. Ajamil, G. Idoate, L. Thuile-Bistarelli, A. Sultan, R. Schulte, E.

Tupper, T. Fuss, R. del Campo, A. Wieland, and M. Bethke for field assistance. G. Aschermann

and A. Putschew kindly enabled TrOC analyses. This project was funded by the German

Research Foundation (DFG) through the Research Training Group ‘Urban Water Interfaces’

(UWI; GRK 2032).

Author contributions. CR, SH, MOG, PC and GS designed the study. CR adapted the CO2 sensors

and together with SH, collected the data. CR did the flux calculations, the optical analysis and

the PARAFAC modeling. CR and GS conducted the statistical analysis. CR led the manuscript

writing, with guidance by GS and RdC. All authors discussed the results and edited the

manuscript. Data accessibility statement. Data will be available at ZENODO.

CHAPTER 4. SUPPLEMENT

111

Supplement of Chapter 4

Carbon dioxide emissions across an urban

aquatic network

CO2 calculation: loop method

We measured pCO2 using the loop method according to (Wilkinson et al. 2018), where the gas

concentration (XCO2 in ppm) in the injected gas sample (Xsample) is calculated from the

difference between the average equilibrium and baseline values:

𝑋𝑠𝑎𝑚𝑝𝑙𝑒 =∆X𝑉𝑙𝑜𝑜𝑝 +𝑉𝑠𝑎𝑚𝑝𝑙𝑒

𝑉𝑠𝑎𝑚𝑝𝑙𝑒

Where, Vloop is the sum of the internal loop volume of the instrument and the volume of the

external loop connection and the volume of the external loop connection.

2. Wind speed calculation

To calculate wind speed at 10m (U10 ms-1) we used the empirical power law (Elliot, 1979):

U10 = Uz* (10 / z) 1/7

Uz is the measured wind speed (ms-1), z is the height it was measured.

Carbon dioxide emissions across an urban aquatic network

112

3. Tables

Table S1 Designation, excitation (Ex) and emission (Em) wavelengths of PARAFAC components, and

the number of studies with matching components reported in OpenFluor (checked on the 28th March

2022) (Murphy et al. 2014). From Romero González-Quijano 2022

PARAFA

componen

OpenFluor

reference

matches

(0.95)

Explanation and selected references

Humic-like, peak A (Coble 1996) humic-like and

recalcitrant (C1) (Hansen et al. 2016)

Terrestial humic-like in waste water treatment

impacted water, (G1) (Murphy et al. 2011);

ubiquitous and recalcitrant humic (C2) (Chen et al.

2017)

Humic-like, peak M (Coble, 1996); humic-like (C3)

(Stedmon and Markager 2005b)

Terrestrial humic-like, suggested as photo-refractory

(Yamashita et al. 2010); (C2) terrestrial humic-like

(C3) (Williams et al. 2010)

Anthropogenic, microbial humic-like (C6) (Williams

et al. 2016)

Similar to tryptophan (C3) (Catalán et al. 2015);

protein-like, linked to autochthonous production (C3)

(Amaral et al. 2016)

Protein-like, peak B (Coble, 1996); waste water

treatment protein (C2) (Teymouri 2007)

Table S2 Description of absorbance and fluorescence indices (from Romero González-Quijano 2022)

Variable

Description

SUVA254

Proxy for DOM aromaticity (Weishaar et al. 2003)

E2:E3

Ratio of absorbance at 250 and 365 nm, as an (inverse) indicator of molecular size

(Chen et al. 1977; Peuravuori and Pihlaja 1997)

E4:E6

Indicator of humification (Chen et al. 1977)

Ratio of slopes (SR) computed from short and long wavelength regions as another

negative correlate with DOM molecular weight (Loiselle et al. 2009)

Fluorescence index (FI) Ratio of the fluorescence intensities at the emissions 470 and

520 (obtained at excitation wavelength of 370nm). Indicator of DOM derived from

terrestrial plants (FI around 1.2) or from microbes or algae (FI around 1.4) (Cory and

McKnight 2005; Cory et al. 2010; Fellman et al. 2010; Jaffé et al. 2008)

HIX

Humification index (HIX) as a proxy for humic substances (Ohno 2002)

β/α

Freshness index β/α (Wilson and Xenopoulos 2009b) , which indicates the relative

importance of recently produced DOM (Parlanti et al. 2000)

CHAPTER 4. SUPPLEMENT

113

Table S3 Physico-chemical variables across water body types.

Variable

Lakes

Ponds

Rivers

Streams

Temperature

14.6 ± 6.9

13.7 ± 5.3

15.2 ± 6.2

11.3 ± 4.9

DOC (mg/L)

7.47 ± 2.70

10.34 ± 3.00

7.96 ± 1.62

11.67 ± 5.45

TP (mg/L)

0.05 ± 0.05

0.09 ± 0.07

0.10 ± 0.08

0.26 ± 0.31

NH4+ (mg/L)

0.07 ± 0.07

0.27 ± 0.67

0.15 ± 0.14

0.36 ± 0.65

NO3- (mg/L)

0.22 ± 0.36

0.03 ± 0.06

1.12 ± 1.66

0.91 ±1.53

Chlorophyll a

(μg /L)

6.24 ± 14.18

7.33 ± 7.7

2.09 ±2.93

5.28 ± 9.74

Carbon dioxide emissions across an urban aquatic network

114

Table S4 Sites coordinates and land uses. Longitude is given in decimal degrees East and latitude in

decimal degrees North. Land cover was calculated with QGIS- layer, for a 50 m buffer around water

bodies using the software QGIS (QGIS Development Team, 2017) and data provided by the Senate

Administration for Environment, Transport and Climate Protection of Berlin.

Name

Type

Latitude

Longitude

Agricul-

tural(%)

Fores-

ted(%)

Urban

Pavement

(%)

Urban

Green(%)

Teltowkanal

river

52.44239

13.32454

Teltowkanal

river

52.42642

13.52039

100

Wuhle

stream

52.52562

13.57913

Tegeler Fliess

stream

52.63442

13.38013

Biesdorfer See

lake

52.50331

13.5497

Obersee

lake

52.54856

13.48972

Ploetzensee

lake

52.5438

13.33049

100

Gross Glienicker

lake

52.46417

13.11489

Havel

lake

52.4431

13.14453

100

Schlachtensee

lake

52.44066

13.21183

Müggelsee

lake

52.43837

13.6451

Hoheheideteich

pond

52.57694

13.16428

100

Hamburger

Teich

pond

52.56738

13.44549

Ruhwaldteich

pond

52.52573

13.25998

Kienhorstbecken

pond

52.57724

13.34556

100

Mittelfeldteich

pond

52.61208

13.23045

100

Neurandteich

pond

52.63883

13.27377

Möwensee

pond

52.55282

13.33545

Müggelspree

river

52.42985

13.68912

100

Landwehrkanal

river

52.51935

13.31959

Spree

river

52.53613

13.21622

100

Kuhlake

river

52.57817

13.16509

100

Neukölln Ship

Canal

river

52.48936

13.43949

Spree

river

52.47137

13.49683

100

Panke

river

52.5369

13.36759

Zingergraben

stream

52.58209

13.38594

95.35

4.65

Schwarzer

Graben

stream

52.56488

13.34918

Graben 1 Buch

stream

52.62384

13.46883

100

Graben 73

Buchholz

stream

52.62881

13.45315

100

Erpe

stream

52.45888

13.61245

Koppelgraben

stream

52.62065

13.41089

Plumpengraben

stream

52.41513

13.5628

100

CHAPTER 4. SUPPLEMENT

115

Table S5 List of trace organic compounds analyzed

TrOC

abreviat

ion

Lower

Limit of

Quantific

a-tion

(LLoQ)

number of

sites with

conc >

LLoQ

Name

Description

ACS

0.1

acesulfame

sweetener

ATS

0.05

amidrotrizoic

radiocontrast agent

BTA

0.1

benzotriazole

corrosion inhibitor

BZF

0.1

benzafibrate

lipid-lowering agent

CBZ

0.05

carbamazepine

anticonvulsant

DCF

0.05

diclofenac

analgesic/anti

inflammatory

FAA

0.1

4-formylamin

metabolite of

metamizol

analgesic

GAB

0.1

gabapentin

epilepsy treatment

/pain killer

GPL

0.05

gabapentin-

lactam

IOM

0.1

iomeprol

radiocontrast agent

IOP

0.01

iopromide

radiocontrast agent

MBT

0.1

methylbenzotri

azole

corrosion inhibitor

MTP

0.1

metoprolol

beta blocker

PRI

0.05

primidone

anticonvulsant

SMX

0.1

sufamethoxazol

antibiotic

VAL

0.1

valsartan

AT1- receptor

antagonist

VLX

0.1

venlafaxine

antidepressant

VSA

0.1

valsartan acid

high blood preasure

Carbon dioxide emissions across an urban aquatic network

116

Table S6 Summary of all the k600 calculated for this study.

Name_k600

Formula

Comments

k600_chamber

𝑘 = 𝐹

𝑘𝐻(𝐶𝑂2_𝑤𝑎𝑡𝑒𝑟 – 𝐶𝑂2_𝑎𝑖𝑟)

𝑘600 =𝑘(600

𝑆𝑐)−0.5

Floating chamber method.

(Duc et al.

2013)

k600_wind_field

𝑘600=2.07+0.215𝑈101.7 ·( 𝑆𝑐

600)−0.5

U10 calculated from wind

measured in the field at 0.5m

(Cole and

Caraco 1998)

k600_wind_mug

𝑘600=2.07+0.215𝑈101.7 ·( 𝑆𝑐

600)−0.5

U10 calculated from wind

measured at Müggelsee station

at 5m

(Cole and

Caraco 1998)

k600_fetch_field

𝑘600=2.13+2.18·𝑈10+0.82·𝑈10

·𝑙𝑜𝑔10𝑓𝑒𝑡𝑐ℎ

Fetch in km. U10 calculated

from wind measured in the

field at 0.5m

(Vachon and

Prairie 2013)

k600_fetch_mug

𝑘600=2.13+2.18·𝑈10+0.82·𝑈10

·𝑙𝑜𝑔10𝑓𝑒𝑡𝑐ℎ

Fetch in km. U10 calculated

from wind measured at

Müggelsee station at 5m

(Vachon and

Prairie 2013)

k600_area_field

𝑘600=2.51+1.48·𝑈10+0.39·𝑈10

·𝑙𝑜𝑔10𝐿𝐴

U10 calculated from wind

measured in the field at 0.5m

(Vachon and

Prairie 2013)

k600_area_mug

𝑘600=2.51+1.48·𝑈10+0.39·𝑈10

·𝑙𝑜𝑔10𝐿𝐴

U10 calculated from wind

measured at Müggelsee station

at 5m

(Vachon and

Prairie 2013)

k600_ray_1

𝑘600=(𝑉𝑆)0.89 ·𝐷0.54 ·5037

Velocity (V) as an average of

V estimated from depth (D)

and width (W).

(Raymond et

al. 2012)

k600_ray_3

𝑘600=𝑉0.85 ·𝑆0.77 ·1162

V as an average of V estimated

from depth (D) and width (W).

(Raymond et

al. 2012)

k600_ray_4

𝑘600=(𝑉𝑆)0.76 ·951.5

V as an average of V estimated

from depth (D) and width (W).

(Raymond et

al. 2012)

k600_ray_5

𝑘600=𝑉𝑆·2841+2.02

V as an average of V estimated

from depth (D) and width (W).

(Raymond et

al. 2012)

CHAPTER 4. SUPPLEMENT

117

Table S7 Random Forest (RF) models. Variables used as predictors were PC axes with Eigenvalues >1

from three separate PCAs based on DOM characteristics; water depth, land cover and physico-chemical

variables (CHEM); and Trace Organic Compounds. Only variables with a relative influence in the RF

>10% are shown.

Measurement method

Variance

explained (%)

Predictor

Variable

Deviance explained

by model (%)

Instantaneous CO2 flux

in flux chambers

PC1 DOM

PC4 CHEM

Continuous CO2 flux in

equilibration chambers

PC1 DOM

PC1 CHEM

Daily CO2 flux

variability in

equilibration chambers

PC2 CHEM

PC1 CHEM

Carbon dioxide emissions across an urban aquatic network

118

Table S8 Literature values of CO2 fluxes for the comparison with our fluxes

Study

Yea

lenti

c/lo

tic

in gC-CO2

m2y-1

(Kling et al. 1992) in (Abnizova et al. 2012)

1992

lenti

91.25

(Hamilton et al. 1994) in (Abnizova et al.

2012)

1994

lenti

368.65

(Hamilton et al. 1994) in (Abnizova et al.

2012)

1994

lenti

1095

(Roulet et al. 1997) in (Abnizova et al. 2012)

1997

lenti

616.85

(Cole and Caraco 1998) in (Abnizova et al.

2012)

1998

lenti

29.2

(Cole and Caraco 1998) in (Abnizova et al.

2012)

1998

lenti

51.1

(Striegl and Michmerhuizen 1998) in

(Abnizova et al. 2012)

1998

lenti

3.65

(Duchemin et al. 1999) in (Abnizova et al.

2012)

1999

lenti

248.2

(Riera et al. 1999) in (Abnizova et al. 2012)

1999

lenti

135.05

(Riera et al. 1999) in (Abnizova et al. 2012)

1999

lenti

200.75

(Riera et al. 1999) in (Abnizova et al. 2012)

1999

lenti

(Riera et al. 1999) in (Abnizova et al. 2012)

1999

lenti

21.9

(Casper et al. 2000) in (Abnizova et al. 2012)

2000

lenti

175.2

(Cole et al. 2000) in (Abnizova et al. 2012)

2000

lenti

-65.7

(Cole et al. 2000) in (Abnizova et al. 2012)

2000

lenti

219

(Hope et al. 2001)

2001

lotic

662.26

(Huttunen et al. 2002a) in (Abnizova et al.

2012)

2002

lenti

51.1

(Huttunen et al. 2002b) in (Abnizova et al.

2012)

2002

lenti

149.65

(Huttunen et al. 2002b) in (Abnizova et al.

2012)

2002

lenti

153.3

(Richey et al. 2002) in (Butman and

Raymond 2011)

2002

lotic

830

(Huttunen et al. 2003) in (Abnizova et al.

2012)

2003

lenti

76.65

(Huttunen et al. 2003) in (Abnizova et al.

2012)

2003

lenti

58.4

(Huttunen et al. 2003) in (Abnizova et al.

2012)

2003

lenti

62.05

(Huttunen et al. 2003) in (Abnizova et al.

2012)

2003

lenti

83.95

(Huttunen et al. 2003) in (Abnizova et al.

2012)

2003

lenti

40.15

CHAPTER 4. SUPPLEMENT

119

Study

Year

lentic

/lotic

in gC-CO2

m2y-1

(Åberg et al. 2004) in (Abnizova et al.

2012)

2004

lentic

105.85

(Åberg et al. 2004) in (Abnizova et al.

2012)

2004

lentic

94.9

(Repo et al. 2007) in (Abnizova et al.

2012)

2007

lentic

51.1

(Repo et al. 2007) in (Abnizova et al.

2012)

2007

lentic

160.6

(Repo et al. 2007) in (Abnizova et al.

2012)

2007

lentic

149.65

(Yao et al. 2007) in (Butman and

Raymond 2011)

2007

lotic

830-1560

(Blodau et al. 2008) in (Abnizova et al.

2012)

2008

lentic

7.3

(Shirokova et al. 2009) in (Abnizova et al.

2012)

2009

lentic

109.5

(Humborg et al. 2010) in (Butman and

Raymond 2011)

2010

lotic

1850

(Dubois et al. 2010) in (Butman and

Raymond 2011)

2010

lotic

1182

(Aufdenkampe et al. 2011)

2011

lentic

240

(Aufdenkampe et al. 2011)

2011

lentic

(Aufdenkampe et al. 2011)

2011

lentic

130

(Butman and Raymond 2011)

2011

lotic

-882 to 4008

(Aufdenkampe et al. 2011) in (Butman and

Raymond 2011)

2011

lotic

1675

(Butman and Raymond 2011)

2011

lotic

2370

(Aufdenkampe et al. 2011)

2011

lotic

1600

(Aufdenkampe et al. 2011)

2011

lotic

2720

(Aufdenkampe et al. 2011)

2011

lotic

720

(Aufdenkampe et al. 2011)

2011

lotic

2630

(Aufdenkampe et al. 2011)

2011

lotic

260

(Aufdenkampe et al. 2011)

2011

lotic

560

(Alin et al. 2011) in (Lauerwald et al.

2015)

2011

lotic

2065

(Alin et al. 2011) in (Lauerwald et al.

2015)

2011

lotic

1478

(Striegl et al. 2012) in (Lauerwald et al.

2015)

2012

lotic

750

(Tian et al. 2012) in (Gu et al. 2021)

2012

lotic

1898

(Crawford et al. 2013)

2013

lotic

1930.03

Carbon dioxide emissions across an urban aquatic network

120

Study

Year

lentic

/lotic

in gC-CO2

m2y-1

(de Fátima F. L. Rasera et al. 2013) in

(Lauerwald et al. 2015)

2013

lotic

1880

(de Fátima F. L. Rasera et al. 2013) in

(Lauerwald et al. 2015)

2013

lotic

2079

(Wallin et al. 2013) in (Gu et al. 2021)

2013

lotic

1423.5

(Natchimuthu et al. 2014)

2014

lentic

4.818

(Khadka et al. 2014) in (Gu et al. 2021)

2014

lotic

91.25

(Giesler et al. 2014) in (Gu et al. 2021)

2014

lotic

292

(Sepulveda-Jauregui et al. 2015)

2015

lentic

43.34

(Lauerwald et al. 2015)

2015

lotic

2004

(Lauerwald et al. 2015)

2015

lotic

1075

(Lauerwald et al. 2015)

2015

lotic

543

(Lauerwald et al. 2015)

2015

lotic

305

(Lauerwald et al. 2015)

2015

lotic

1351

(Lauerwald et al. 2015)

2015

lotic

648

(Lauerwald et al. 2015)

2015

lotic

2829

(Lauerwald et al. 2015)

2015

lotic

1403

(Lauerwald et al. 2015)

2015

lotic

193

(Lauerwald et al. 2015)

2015

lotic

758

(Lauerwald et al. 2015)

2015

lotic

2035

(Lauerwald et al. 2015)

2015

lotic

1849

(Lauerwald et al. 2015)

2015

lotic

2215

(Lauerwald et al. 2015)

2015

lotic

1946

(Lauerwald et al. 2015)

2015

lotic

459

(Lauerwald et al. 2015)

2015

lotic

831

(Smith and Kaushal 2015) in (Gu et al.

2021)

2015

lotic

83.95

(Jones et al. 2016)

2016

lentic

(Holgerson and Raymond 2016)

2016

lentic

154.08

CHAPTER 4. SUPPLEMENT

121

Study

Year

lentic

/lotic

in gC-CO2 m2y-

(Holgerson and Raymond 2016)

2016

lentic

92.89

(Holgerson and Raymond 2016)

2016

lentic

104.56

(Holgerson and Raymond 2016)

2016

lentic

98.19

(Holgerson and Raymond 2016)

2016

lentic

91.54

(Holgerson and Raymond 2016)

2016

lentic

50.32

(Gerardo-Nieto et al. 2017)

2017

lentic

92.59

(Gerardo-Nieto et al. 2017)

2017

lentic

55.10

(Natchimuthu et al. 2017)

2017

lotic

7008

(Smith et al. 2017)

2017

lotic

4201.15

(Smith et al. 2017)

2017

lotic

49110.75

(Smith et al. 2017)

2017

lotic

3759.5

(Smith et al. 2017)

2017

lotic

930.75

(Jiang et al. 2017) in (Gu et al. 2021)

2017

lotic

1835.95

(Lee et al. 2017) in (Gu et al. 2021)

2017

lotic

343.1

(Peacock et al. 2019)

2019

lentic

74.85

(Peacock et al. 2019)

2019

lentic

-18 to 343

(Ni et al. 2019) in (Gu et al. 2021)

2019

lotic

1445.4

(Reiman and Xu 2019) in (Gu et al. 2021)

2019

lotic

47.45

(Li et al. 2020) in (Gu et al. 2022)

2020

lotic

2018.45

(Gu et al. 2022)

2021

lotic

1799.45

(Bargrizan et al. 2022)

2022

lotic

218

(Zhang et al. 2022)

2022

lentic

-106.40

Carbon dioxide emissions across an urban aquatic network

122

4. Figures

Figure S1 Relationship between CO2 fluxes determined with two different approaches,

instantaneous measurements on single days in the morning vs continuous measurements over a

week. Dashed line indicates 1:1 line. Lentic: lakes and ponds, lotic: rivers and streams.

Figure S2 Daily (A), day to day (B) (from the Continuous fluxes) and seasonal standard

deviation (SD) (C) (from the Instantaneous fluxes)

CHAPTER 5

123

Methane emissions from contrasting urban

freshwaters: Rates, drivers, and a whole‐

city footprint

This study was published as:

This is the postprint version of the article.

5.1 Abstract

Global urbanization trends impose major alterations on surface waters. This includes impacts

on ecosystem functioning that can involve feedbacks on climate through changes in rates of

greenhouse gas emissions. The combination of high nutrient supply and shallow depth typical

of urban freshwaters is particularly conducive to high rates of methane (CH4) production and

emission, suggesting a potentially important role in the global CH4 cycle. However, there is a

lack of comprehensive flux data from diverse urban water bodies, of information on the

underlying drivers, and of estimates for whole cities. Based on measurements over four seasons

in a total of 32 water bodies in the city of Berlin, Germany, we calculate the total CH4 emission

from various types of surface waters of a large city in temperate climate at 2.6 ± 1.7 Gg

CH4/year. The average total emission was 219 ± 490 mg CH4 m−2 day−1. Water chemical

variables were surprisingly poor predictors of total CH4 emissions, and proxies of productivity

and oxygen conditions had low explanatory power as well, suggesting a complex combination

of factors governing CH4 fluxes from urban surface waters. However, small water bodies (area

<1 ha) typically located in urban green spaces were identified as emission hotspots. These results

help constrain assessments of CH4 emissions from freshwaters in the world's growing cities,

facilitating extrapolation of urban emissions to large areas, including at the global scale.

Herrero Ortega, S., C. Romero González-Quijano, P. Casper, G. A. Singer, and M. O.

Gessner. 2019. Methane emissions from contrasting urban freshwaters: Rates, drivers, and a

whole-city footprint. Global Change Biology 25:4234-4243.

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

124

5.2 Introduction

More than half of the world's population currently lives in cities and this fraction is projected to

rise to two‐thirds by the year 2050 (UN 2016). This global urbanization trend leads to heavy

modifications of freshwaters worldwide, as encapsulated in the “urban stream syndrome” for

running waters (Walsh et al. 2005). Symptoms characterizing this syndrome include strong

nutrient and pollutant loading, even when effective sanitation is in place, and disruptive changes

in the hydromorphology of urban freshwaters resulting from altered connectivity, surface

sealing in the catchment, bank hardening, and channel modification by canalization and a

multitude of other engineering measures (Gessner et al. 2014; Grimm et al. 2008b; Roy et al.

2016). As a result, strong impacts on urban surface waters have been documented on biological

communities and ecosystem properties such as oxygen regimes and organic matter dynamics

(Birch and McCaskie 1999; Paul and Meyer 2001; Waajen et al. 2014). A particularly important

consequence of enhanced oxygen depletion and organic matter loading in freshwaters is the

stimulation of methanogenesis in sediments and, thus, increased emission of methane (CH4)

across the water–atmosphere interface (Grinham et al. 2018b). This suggests that urban

freshwaters could act as an important source of CH4 to the atmosphere (Gonzalez-Valencia et

al. 2014; Martinez-Cruz et al. 2017; Wang et al. 2018). Empirical data on CH4 emissions from

urban freshwaters are scarce, however, and have not been included in global emission

estimates(Bastviken et al. 2011; IPCC 2013), nor in systematic assessments of CH4 evasion from

all potential sources in cities. In fact, most studies on freshwaters assessing urban CH4 emissions

were limited to a single type of water body and a single season (López Bellido et al. 2011; Wang

et al. 2018; Zhang et al. 2014; Zhang et al. 2016) with only one recent investigation in a tropical

megacity considering multiple surface waters and temporal patterns (Martinez-Cruz et al. 2017).

Equivalent information is lacking from urban freshwaters in temperate climates, where

seasonality is more pronounced than in the tropics. Information available on individual urban

water bodies suggests that the drivers behind CH4 emissions are similar to those in rural, forest,

and other natural areas (Martinez-Cruz et al. 2017; Yu et al. 2017). All else being equal, shallow

waters, which are typical of urban areas (McEnroe et al. 2013), are likely to emit more CH4 per

surface area, because the travel times of CH4 bubbles generated by ebullition events and rising

from the sediment to the water surface are likely to be shorter, limiting CH4 oxidation by

methanotrophy in the oxic water column (Bastviken et al. 2004; Holgerson 2015). The small

size of most urban water bodies also suggests that land use in the surroundings and associated

inputs of organic matter, nutrients, and contaminants can strongly influence water quality and

ecosystem properties. Large supplies of labile organic matter, whether from the catchment or

through intense primary production boosted by nutrient availability, coupled with subsequent

oxygen depletion are both conducive to methanogenesis (Segers 1998). This points to a high

potential of urban freshwaters to produce and emit CH4 to the atmosphere, unless toxic

substances curb biological activity. In view of the importance and large gaps in information on

CHAPTER 5

125

rates and drivers of CH4 emissions from urban freshwaters, the aims of this study were to (a)

determine CH4 fluxes at different times of the year from a range of contrasting urban

freshwaters; (b) identify drivers of CH4 emissions from the different types of water bodies; and

potentially important component of global urban CH4 emissions from freshwaters. Based on the

limited information available to date, we predicted rates to be particularly high in small, shallow,

and nutrient‐rich standing waters with sediments rich in organic matter.

Figure 1 Map of the metropolitan area of Berlin, Germany, showing land use and freshwater sampling

locations in lakes (L1–7), ponds (P1–7), rivers (R1–7), streams (S1–7), and four additional running water

sites characterized by high nutrient concentrations (H1–4)

5.3 Materials and methods

5.3.1 Study sites

The study was conducted in the city of Berlin, Germany, an urban area with 3.5 million

inhabitants on 892 km2 (Heberer 2002), of which 54 km2 (6%) are freshwaters (Figure 1).

Freshwaters in Berlin include two mid‐sized rivers feeding and draining several larger shallow

lakes (Knappe et al. 2005), about 60 smaller lakes (>1 ha), and more than 500 ponds (Heberer

2002). When canals for transportation and ditches for sewage and rainwater collection are

added, the surface river network reaches a total length of about 560 km (SenUVK, 2018). River

flow is slow because of the low terrain slope (0.01%; Knappe et al., 2005), locks, and weirs.

Multiple wastewater treatment plants (WWTP) within the city discharge treated effluents into

the urban freshwater network (Heberer 2002). Four categories of surface waters were

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

126

distinguished: lakes, ponds, rivers (including canals), and streams (including ditches). Lakes

were classified as water bodies ≥1 ha according to a lake inventory for Berlin (SenUVK, 2005).

Rivers and streams were differentiated by width (rivers >5 m). Seven locations were randomly

selected from each of the four categories. Four additional running water sites were also included

because of particularly high nutrient (NO3−, NH4+, total phosphorus [TP]) and dissolved

organic carbon (DOC) concentrations recorded in a monitoring program over the five previous

years (SenUVK 2009–2014). However, CH4 emissions at these sites were found not to differ

significantly from those of the randomly selected sites and were thus treated as rivers (H1–2) or

streams (H3–4), depending on size. Thus, a total of 32 sites (Figure 1; Table S1) were each

sampled four times, in spring (April–May), summer (July–August), and fall (September–

October) 2016, and in winter (February–March) 2017 just after ice out because of unusually cold

weather late in the season.

5.3.2 CH4 emissions

Floating chambers were deployed at one selected point in each water body to estimate rates of

total, diffusive, and ebullitive CH4 fluxes to the atmosphere. The chambers were anchored but

several meters of rope and tubing allowed for some free movement. The position in lakes was

randomly chosen along the contour line of average water depth to avoid potential bias caused

by taking measurements at the deepest point (Schilder et al. 2013). Since the bathymetry of

ponds was unknown, the central point (not necessarily the deepest) was used in those cases; this

was less critical than for lakes because water depth in ponds varied much less. In running waters,

chambers were deployed within 2 m from the shore (Grasset et al. 2016). Cylindrical floating

chambers (area: 0.071 m2; headspace volume 5.4 L) were used in lakes and ponds to determine

CH4 emission rates. Slightly wider and shorter but otherwise similar chambers (0.126 m2;

headspace volume 16.8 L) were used in streams and rivers. The chamber headspace was

connected in a closed loop to an ultra‐portable greenhouse gas analyzer (UGGA 24P and 30P;

Los Gatos Research) before deploying a single chamber three times at each location to measure

CH4 headspace concentrations every second for 15 min (Pirk et al. 2016). Chambers were opened

between series of measurements and equilibrated with the surrounding air. All fluxes were

measured between 8 and 12 a.m. to minimize any possible influence of systematic diel variations.

Atmospheric pressure and wind speed 1 m above the water surface were simultaneously

determined using a portable weather station (Kestrel 4000; Nielsen‐Kellerman). Total CH4 flux

(F) to the atmosphere was calculated as:

𝐹 = ∆𝐶

∆𝑡 ×𝑉×𝑃

𝐴×𝑅𝑇 ×10−6 ×8,640×103×16 (mg day−1 m−2)

where ΔC is the concentration change in the headspace of the static chamber (ppmv), Δt is the

chamber deployment time (s), V is the volume (m3) of the chamber headspace, A is the area of

the static chamber (m2), R is the universal gas constant (8.3143 m3 Pa mol−1 K−1), P is

CHAPTER 5

127

atmospheric pressure (Pa), and T is air temperature (K) during the measurement. All

concentration data were plotted to visually identify whether any sampling errors or ebullition

events occurred. When initial values deviated from the atmospheric concentration measured

before deploying a chamber, the first data points were removed and the fluxes calculated based

only on the time span where the concentration increased linearly. Total fluxes were calculated

as the difference between initial and final concentrations during the considered deployment time.

Diffusion fluxes were computed for the first period of linear concentration increases after the

deployments. This was usually during the first 30 s when ebullition was observed. If no

ebullition occurred, the period was extended to up to 15 min. When no ebullition event was

observed, we calculated diffusive fluxes based on the entire exposure period of 15 min (Gerardo-

Nieto et al. 2017). Ebullition events were recognized by sudden steep concentration increases,

which were occasionally followed by a decline. Only concentration increases with an r2 > .7

were taken into account to compute diffusive fluxes (Martinez-Cruz et al. 2017; Sepulveda-

Jauregui et al. 2018b). Ebullition flux was calculated as the difference between the total and

diffusive flux. To assess the reliability of the calculated fluxes from the chamber technique, other

commonly adopted methodologies were used in tandem with the flux measurements by the

chamber technique. Specifically, CH4 concentrations of surface waters were used to calculate

diffusive fluxes following the thin boundary layer (TBL) methodology (see Supporting

information), and inverted funnels deployed above the sediment for a week were used to calculate

ebullition fluxes (see Supporting information).

5.3.3 Extrapolation of CH4 emissions

Total CH4 fluxes measured with the chamber technique were first averaged for each type of

water body and season and then extrapolated to the duration of each season (mg CH4/m2)

(Panneer Selvam et al. 2014). Seasons were defined following the solar calendar: spring (March

20, 2016–June 21, 2016), summer (June 21, 2016–September 21, 2016), autumn (September 22,

2017–December 21, 2017), and winter (December 22, 2017–March 19, 2018). Fifty‐five days of

ice cover were excluded for the winter estimate, with the period of ice cover being established

based on regular visits of a reference lake in Berlin (L7). To standardize the ice‐cover period

among the different water bodies, we defined the start as the date where the minimum daily

temperature dropped below 0°C for 3 days in a row and the end as the date when mean daily

temperature rose above the freezing point for at least 1 week. Total annual emissions from each

type of water body were estimated by multiplying the seasonal total emission from each type of

water body (mg CH4/m2) by the respective surface area of all water bodies in the city of Berlin

assigned to that water body type. The total CH4 emission footprint of Berlin's surface waters

was then calculated as the sum of the annual emissions by each of the four types of water bodies.

Estimates of variation (i.e., uncertainties) were obtained by applying error propagation rules at

each step.

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

128

5.3.4 Water chemistry

Dissolved oxygen (DO), pH, electrical conductivity, and temperature were measured at 0.5 m

depth with an in situ multiprobe (smarTROLL) or a WTW Multiprobe 3320 (pH320, OxiCal‐

SL, Cond340i). Integrative water samples were collected from the upper 0.5 m water layer.

Alkalinity was measured by titrating (888 Titrando, Metrohm) unfiltered water in the

laboratory. To determine particulate organic carbon (POC), known volumes (0.2–2 L) were

filtered through precombusted (5 hr, 450°C) and preweighed GF75 glass fiber filters (average

pore size 0.3 μm; Advantec). The filters were dried and weighed, and a weighed portion was

subsequently used for elemental analysis (Vario EL; Elementar Analysensysteme GmbH) to

determine POC. The filtrate was stored in acid‐washed and precombusted glass vials with a

polytetrafluoroethylene‐ lined screw cap for later measurements of DOC and dissolved

inorganic carbon (DIC) on a TOC analyzer (TOC‐V; Shimadzu). A second GF75 filter produced

in the same way was used for spectrophotometric analysis of chlorophyll a (chl a) after hot

ethanol extraction (Jespersen and Christoffersen 1987). Soluble reactive phosphorus, NO − 3,

NO−2, and NH4+ in the filtrate were analyzed spectrophotometrically on a flow injection

analyzer (FIA compact; MLE GmbH), and TP was determined in the same way after digesting

unfiltered water samples with K2S2O8 (30 min at 134°C). The concentrations of SO2− 4 and Cl−

were measured by ion chromatography (Dionex ICS 1000; Thermo Scientific). We further

characterized dissolved organic matter (DOM) by absorbance and fluorescence

spectrophotometry (Aqualog). Fluorescence spectra were recorded in a 1 cm quartz cuvette at

excitation wavelengths ranging from 250 to 600 nm at 5 nm increments and emission

wavelengths of 250–650 nm measured at 1.77 nm increments. These optical measurements were

performed within 48 hr after sampling. The resulting data yielded the following indicators of

DOM quality (Table S2): humification index (HIX), fluorescence index (FIX), biological activity

index (β:α), specific UV absorbance (SUVA), spectral slope between 275 and 295 nm (S275–295),

spectral slope between 350 and 400 (S350–400), and the spectral slope ratio (SR).

5.3.5 Land use

The total area of each type of water body and of four categories of land use (forest and natural

areas, green space, agricultural land, paved areas) within a 50 m wide strip along the shores of

each site were calculated using Quantum GIS (Development Team), based on land‐use data

freely available from the Senate Department for the Environment, Transport and Climate

Protection of Berlin. Historical reviews and personal communication with citizens and

authorities complemented the database to determine whether a given water body was natural or

man‐made and whether it had any other distinct anthropogenic features.

CHAPTER 5

129

5.3.6 Data analysis

All statistical analyses were performed with R version 3.2.2 (R Development Core Team, 2010).

Linear mixed models were used on log‐transformed data to test for differences in total CH4

emissions among seasons, types of water bodies, and the interaction of both, taking into account

the repeated‐measures nature of the data. Tukey post hoc tests were used for pairwise

comparisons. Wilcoxon signed rank test was used to compare estimates of diffusive flux by the

TBL and chamber method, as well as to compare ebullitive flux assessed with the funnel traps

and the chamber method. To explore possible controls of total CH4 emissions, the large number

of variables recorded to characterize the water bodies was first condensed by a principal

component analysis (PCA). The analysis was based on water temperature, a range of water

chemical variables (conductivity, pH, alkalinity, DO, TP, NH4+, NO3-, Cl−, DOC, DIC, chl a),

including DOM properties (SUVA, S275–295, S350–400, SR, β:α, FIX, and HIX), and land use (relative

coverage by forest, agriculture, paved areas, and green space). All variables were z‐standardized

prior to the PCA. Subsequently, all principal components with eigenvalues >1 were used as

predictors in a multiple linear regression (MLR) model with total CH4 emission as the response

variable. MLR models were built stepwise in both directions and compared by means of Akaike's

information criterion to identify the most parsimonious model. Last, CH4 emission was

individually regressed against all variables contributing most to the PCA axes included as

responses in the final MLR.

5.4 Results

Total CH4 emissions determined with the chamber technique from surface waters in the city of

Berlin averaged 219 ± 490 (SD) mg CH4 m−2 day−1 across all 32 locations and seasons. These

fluxes averaged across all sites were higher in summer (p < .05) than in all other seasons,

coinciding with the highest water temperatures (Figure 2; Table S4). No significant differences

were found among the other seasons. Ponds showed the highest emission (503 ± 699 mg CH4

m−2 day−1) in all seasons (Figure 2), with fluxes significantly exceeding (p < .05) those from

rivers (123 ± 285 mg CH4 m−2 day−1) and streams (118 ± 348 mg CH4 m−2 day−1) but not from

lakes (159 ± 473 mg CH4 m−2 day−1). Within each of the four types of water bodies, seasonal

differences were only significant between summer and winter in lakes, ponds, and rivers (p <

.05), whereas streams never showed any significant difference among seasons.

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

130

Figure 2 Seasonal changes in (a) daily mean air temperature in Berlin Tempelhof recorded by the German

Meteorological Office, with the light gray area representing a period of ice cover on the larger lakes and

the dark gray areas representing the sampling periods, and (b) Total methane (CH4) emissions from four

types of urban water bodies. Box plots show the median (horizontal line), interquartile range (box limits),

highest and lowest values within 1.5 times the box size from the median (whiskers) and outliers (points)

Total CH4 emission derived from all chamber measurements indicated a higher contribution of

ebullition (80%). Although the relative contribution of ebullition varied among types of water

bodies (Table 1; Figure S2). Estimates of ebullition and diffusive fluxes derived from different

methodologies also showed some differences. Ebullition fluxes estimated by 1 week deployments

of funnels accounted for an average of 62% of the emissions at those sites where ebullition was

observed (N = 12), compared to 51% based on measurements at the same sites made with the

chamber technique (Table S3). Ebullition fluxes determined with the two techniques were

positive correlated (Spearman's ρ = 0.73; p < .01). There were no significant differences in

ebullition fluxes among individual water bodies within each type. In contrast, diffusive fluxes

estimated by the two methods were significantly different for lakes (p < .001), ponds (p = .024),

rivers (p < .01), and streams (p < .001). However, despite these differences, the values obtained

with the different methods were in a broadly similar range for most of the observations. Taking

into account the calculated areas of the different types of surface waters in the city of Berlin

(Table 1), the annual total CH4 emission estimated by the chamber method was 2.6 ± 1.7 Gg

CHAPTER 5

131

CH4. Lakes alone contributed almost two‐thirds to the total emissions, due to the large total

lake area, while streams contributed the least (Table 1).

The first four axes of the PCA to characterize the 32 investigated water bodies in terms of water

chemistry and land use accounted for 58% of the total variability. PC1 and PC2 clearly separated

the four types of water bodies (Figure 3a,c), with PC1 separating running from standing waters

mainly based on differences in land use (green space, paved, or agricultural) and the DOM

spectral ratio (SR), and PC2 separating larger from smaller water bodies based on conductivity

and solute concentrations (e.g., NH4+, Cl-), DOM descriptors (SUVA, β:α), and chl‐a

concentration. PC3 captured smaller scale water chemical differences based on DOM descriptors

(e.g., S350–400) and proxies of productivity (e.g., NH4+, chl a, and DO), and indicates a slight

tendency of lakes to differ from other water bodies (Figure 3b,d). Finally, PC4 tended to separate

autumn samples from all others, mainly based on high DOC concentrations.

Table 1 Annual methane (CH4) emission footprint of the metropolitan area of Berlin, Germany, separated

by type of water body (mean ± SD)

Type

water

body

Area

(km2)

Emission footprint

(mg CH4/year)

CH4 emission (mg CH4 m-2 day-1)

Ebullition

Diffusion

Lakes

29.7

1,712

1,498

100

342

Ponds

2.11

385

598

300

564

120

166

Rivers

21.4

461

552

109

275

Streams

0.79

218

317

Total

2,594

1,718

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

132

Figure 3 Principal component analysis of 32 water bodies sampled over four seasons, based on potential

explanatory variables for CH4 emissions. (a) Water body types differed mainly along the first two principal

components, (b) PC3 indicates a slight tendency of lakes to differ from all other water bodies, and PC4

tended to distinguish autumn from all other seasons. (c, d) Dominant variables creating the ordination

space relate to land use, water chemistry, and optical properties of dissolved organic matter. Black lines

are scaled structure coefficients (scaling factor of 8), that is, correlations with the principal components.

Gray lines show analogous correlations with particulate organic carbon, which were added a posteriori

because data from only three seasons were available. Only variables with a structure coefficient >0.15 in

Linear regression analyses using the PC scores showed that the most parsimonious model

explaining total CH4 emissions involved PC1 and PC3 (r2 = .30; p < .001). The most important

variables contributing to these two PCA axes were POC, chl a, S350–440, and DO: Total CH4

emissions were related to elevated concentrations of POC and chl a, lower DOM molecule size,

and DO depletion. These patterns appear to be largely driven by differences among lakes (Figure

S1), which produced similar relationships with emission data when lakes were analyzed alone.

No such patterns emerged for the three other types of water bodies analyzed alone. Ponds were

the only exception in that low DO concentrations in surface water were weakly related to CH4

emission (r2 = .19; p = .04).

CHAPTER 5

133

5.5 Discussion

Global estimates of CH4 emissions from freshwaters and other sources are still plagued by large

uncertainties (Bastviken et al. 2011; Deemer et al. 2016; Stanley et al. 2016) with one of the big

unknowns being emissions from surface waters of urban areas. Our estimate of the freshwater

CH4 footprint of a large metropolitan area in a western industrialized region is an important

step toward reducing these uncertainties. The estimated annual emissions of Berlin's surface

waters (2.6 ± 1.7 Gg CH4, mean ± SD) are similar to the CH4 footprint of freshwaters in a

tropical megacity, Mexico City (3.7 ± 4.4 Gg CH4year-1; Martinez‐Cruz et al., 2017), the only

other urban area where a range of surface waters was investigated to obtain an emission estimate

for an entire metropolitan area.

The similar annual values for the two cities mask an important difference, however, namely a

six times larger total surface area of Berlin's freshwaters compared to Mexico City, although the

total land area covered by Berlin is 40% smaller. As a result, the estimated annual CH4 footprint

expressed per surface area of Berlin's freshwaters is eight times lower than in Mexico City (49

vs. 411 Mg CH4 km−2 year−1); this number changes only marginally (i.e., by 2%) when potential

emissions during the nearly 2 month period of ice cover are added to the annual estimate for

Berlin. The discrepancy between the two cities points to several non‐mutually exclusive factors

driving emissions from urban freshwaters.

Temperature could be one of those factors, as suggested by a trend of increasing emission fluxes

toward the tropics identified in a comparison of urban surface waters distributed across the globe

(Table 2). However, this relationship with latitude based on data from 17 cities is rather weak

(Spearman's ρ = 0.29) and not significant (p = .16). Furthermore, although the annual mean

temperatures in Berlin and Mexico City reflect the location of the two cities in distinct climates,

the difference of <10°C (9.0 and 15.9°C, respectively) cannot account for much more than a

twofold, or possibly threefold, difference in microbial metabolic rates (Davidson and Janssens

2006), even when Berlin's greater temperature variability is taken into account (Bernhardt et al.

2018). Ebullition fluxes can show stronger responses to small temperature changes than

diffusive fluxes (Aben et al. 2017) but are still unlikely to fully account for the observed

difference in CH4 emissions between Berlin and Mexico City. This suggests that additional

features of urban surface may have to be considered. Such features include resource availability

related to human population density (10 times higher in Mexico City than in Berlin), pollution

control policies (Grimm et al. 2008c), and stormwater and sanitary infrastructure (Smith et al.

2017). This conclusion is supported by the hypereutrophic conditions reported for all water

bodies analyzed by Martinez‐Cruz et al. (2017). Our budget calculation is based on

measurements of total flux including both diffusion and ebullition made with floating chambers.

This enabled a first approximation of total annual emissions, for a large metropolitan area

encompassing a wide range of different water bodies. Expanding the coverage of these

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

134

measurements at different scales, both spatial (within and among water bodies) and temporal

(diel to interannual), would reduce the uncertainties associated with the estimates available at

present. In addition, a comparison with alternative methods can help constrain and validate

these estimates. Therefore, we also computed diffusive fluxes by the commonly employed TBL

approach and determined ebullitive fluxes at selected sites by deploying funnel traps for 1 week.

The TBL approach makes several assumptions, particularly on piston velocities (k) depending

on wind speed, which makes this method vulnerable to biases, especially in aerodynamically

rough and heterogeneous urban environments. This could be one reason for several

discrepancies observed between the two methods used to derive diffusive fluxes in our study

(Table S3). The use of anchored rather than freely drifting chambers could also have contributed

to the observed differences in running waters, mainly because unnatural water turbulence

created by the chambers could unnaturally increase fluxes (Lorke et al. 2015). However, the

typically slow flow of the lowland streams and ditches in Berlin makes it unlikely that this error

was large. Ebullition fluxes assessed with inverted funnels deployed for 1 week produced

remarkably similar results as our short‐term measurements of ebullition, despite the

documented high stochasticity and spatial heterogeneity of ebullition (Wik et al. 2013). This

suggests that the results of our short‐term chamber measurements were broadly realistic across

sites.

Although lower than in Mexico City, the calculated total annual emission per km2 from Berlin's

freshwaters (49 Mg CH4 km−2 year−1) is more than twice that of the global average (22 Mg CH4

km−2 year−1) reported by Bastviken et al. (2011) for 4.6 million km2 of global freshwater surfaces.

The fraction of urban areas contributing to freshwater surfaces globally is unknown, but our

rates for Berlin, like those for other urban freshwaters (Table 2), were higher than both the

average calculated for lakes and ponds at northern latitudes (Wik et al. 2016) and values for

streams and rivers globally (Stanley et al. 2016). This could suggest that urban areas in general

contribute disproportionally to CH4 emissions from freshwaters. Given that there are >500

urban centers worldwide with >1 million inhabitants each and that urbanization trends continue

(UN, 2016), emissions of CH4 from urban areas may be sufficiently considered in large‐scale

estimates. An extremely rough estimate assuming 3 Gg of CH4 emitted annually by each of the

500 most densely populated cities in the world results in a total annual emission of 1.5 Tg CH4,

but emissions from the total urbanized area globally are evidently much larger. A related

question is whether surface waters also contribute significantly to the total CH4 footprint of

metropolitan areas. Currently, the answer to this question is speculative, too, because other

sources of CH4 have not been quantified. However, a recent estimate of 20,000 Tg of CO2 emitted

by the city of Berlin in 2012 (Reusswig et al. 2014) suggests that even the high total CH4 fluxes

from Berlin's surface waters would contribute little to the total greenhouse gas emissions from

the city, equivalent to 0.004% in CO2 equivalents.

CHAPTER 5

135

Table 2 Methane (CH4) emission flues from urban freshwaters.

Elevation

(masl)

CH4 flux (mg CH4 m-2 day-1)

Reference

Climatic zone and location

Total

Diffusive

Boreal

Lake Vesijärvi in Enonselkä,

Finland

3.8

(López Bellido et al.

2011)

Pond in Linköping, Sweden

128

(Natchimuthu et al.

2014)

Temperate

Lakes in Berlin, Germany

159

This study

Lake Rotsee, Lucerne,

Switzerland

419

(Schubert et al. 2010)

Ponds in Berlin, Germany

503

117

This study

Open water in a wetland in

Florida, USA

123

(Morin et al. 2017)

Rivers in Berlin, Germany

123

This study

Streams in Berlin, Germany

118

This study

Small modified streams in

Baltimore, USA

11.5

(Smith et al. 2017)

Streams receiving WWTP

effluents, Germany

13.3

(Alshboul et al. 2016b)

Modified section of the Jian

River in Shunyi, Beijing,

China

374

(He et al. 2018)

Dammed section of the

Chaobai River in Shunyi,

Beijing, China

2,134

(He et al. 2018)

Ponds in Queensland,

Australia

276

129

(Grinham et al. 2018a)

Subtropical

Nambol Turel stream in

Nambol, State of Manipur,

India

777

134

(Khoiyangbam and

Basanta Kumar 2014)

Shanghai River network,

Shanghai, China

3.1-296

(Yu et al. 2017)

Pond in Yichang, Hubei

Province, Central China

595

(Xiao et al. 2014)

Yangtze River network in

Chongqing, South-west

China

259

22.4

(Wang et al. 2018)

Lake Donghu, Wuhan, China

23.3

(Xing et al. 2005)

Lakes in the urban areas of

States of Mexico and

Michoacán, Mexico

2,080-

2,840

277

(Gonzalez-Valencia et

al. 2014)

Lakes in Mexico City,

Mexico

2,230

500

(Martinez-Cruz et al.

2017)

Ponds in Mexico City,

Mexico

2,230

(Martinez-Cruz et al.

2017)

Rivers in Mexico City,

Mexico

2,230

2,400

(Martinez-Cruz et al.

2017)

Tropical

Lakes in the urban areas of

State of Veracruz, Mexico

464

2,819

(Gonzalez-Valencia et

al. 2014)

High variability of CH4 emissions rates in space and time is common (Deemer et al. 2016;

DelSontro et al. 2010) and also apparent in our dataset on surface waters in Berlin. Despite this

high variability both within and across water bodies, PCA could differentiate between standing

and flowing waters, and subsequent regression analyses identified water chemistry and the

predominant land use near each site as factors influencing CH4 emissions (Figure 3). Ponds, in

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

136

particular, were identified as hotspots of CH4 emissions in Berlin, with the annual average

emissions four times higher than from lakes, streams, and rivers (Table S3). This information is

important, not least because anthropogenic ponds are neglected water bodies in terms of CH4

emissions both in cities and other landscapes (Grinham et al. 2018a). For example, Berlin has a

detailed inventory of all lakes and their water quality is regularly assessed in monitoring

programs. In contrast, no systematic information is available on ponds, despite the fact that

these small water bodies are increasingly recognized as important urban habitats (Hassall

2014). Although emissions did not significantly differ when lakes and ponds were statistically

treated as categories, a significant negative relationship emerged between log‐transformed CH4

emission flux and lake and pond surface area (r2 = .46; p = .01), corroborating a previously

observed pattern of increasing CH4 flux to the atmosphere with decreasing size of water bodies

(Bastviken et al. 2004; Grinham et al. 2018a; Holgerson and Raymond 2016; Wik et al. 2016).

Nevertheless, even though ponds had high emissions, their contribution to the overall emission

budget is low in comparison to lakes, which account for more than 50% of the total emission

from freshwaters in Berlin (Table 1), owing to the 14 times larger total water surface area of

lakes. In addition to differences in depth and shoreline development, land use adjacent to the

ponds and lakes (Figure 3d) could play a role in producing this relationship, since half of the

investigated lakes in Berlin are surrounded by forests. In contrast, urban ponds are mostly

associated with green spaces throughout the city where they are likely to receive anthropogenic

inputs resulting, for example, from feeding of waterfowl, fertilizer application, or pet waste

(Hobbie et al. 2017).

The particularly high variability in emissions rates that we observed from running waters was

not clearly related to riparian land cover or other characteristics. High emission rates

characterized some stream sites experiencing diffuse nutrient inputs from agriculture (S3 and

S6) or some highly engineered streams (paved riparian areas, channelization; S7, S2), but this

was not universally true for other water bodies showing similar characteristics (S1 and S4). This

inconsistency is not readily explained by toxic effects, because concentrations of a range of heavy

metals and synthetic chemicals that we analyzed were mostly below detection limits in both

water and sediments (S. Herrero Ortega, M.O. Gessner, G.A. Singer, and P. Casper, unpublished

data). Likewise, a strong influence of WWTP was not apparent. While emissions at some sites

receiving WWTP discharge (H1 and R7) were higher than at other sites, those at S5, which was

also influenced by WWTP effluents, were among the lowest. This variability differs from other

observations where a contribution of WWTP to CH4 concentrations was significant (Alshboul

(Alshboul et al. 2016a; Garnier et al. 2013), and may be due to the fact that our study sites were

not located directly downstream of WWTP outlets.

The relation between oxygen concentration and total CH4 emission was also weak (r2 ≤ .12),

although oxygen concentrations varied widely across sites (Table S4). When interpreting these

CHAPTER 5

137

data, it must be borne in mind, however, that our measurements in surface water are not

necessarily good proxies of conditions conducive to methanogenesis in sediments. Furthermore,

differences in chemical characteristics and land use had little explanatory power; only their

combination produced a clear relationship while substantial scatter still remained. Clearly, a

multitude of factors create complex environmental conditions in urban freshwaters that make it

a challenge to tease apart individual drivers of CH4 emissions from these systems. Overall,

however, the variables with the highest explanatory power in our study (i.e., POC, chl a, and

DO) all point to trophic state as a determinant of CH4 emissions from urban freshwaters. This

is in line with results of Martinez‐Cruz et al. (2017) and DelSontro et al. (2018) and is also

reflected by the conspicuous peaks in DOC and chl a in autumn (Figure 3; Table S4) when the

emissions from ponds were highest. This result and our finding that ponds act as hotspots of

CH4 fluxes to the atmosphere are important contributions toward robust assessments of CH4

emissions from whole cities and extrapolation to large areas, including global estimates.

5.6 Ackowledgements

We are grateful to the many students and technicians for their invaluable assistance during

extensive fieldwork, especially to C.N. Stratmann, L. Meinhold, I. Ajamil, G. Idoate, L.T.

Bistarelli, A. Sultan, R. Schulte, E. Tupper, T. Fuß, A. Wieland, and M. Bethke. We also thank

M. Sachtleben for help and advice with material preparation and construction, A. Sepúlveda‐

Jauregui for advice on calculating gas fluxes, K. Pypkins for support with GIS, B. Kleinschmit

for advice in sampling strategy and data analyses, A. Köhler at the Senate Administration for

Environment, Transport and Climate Protection Berlin (SenUVK) for providing data on water

quality, and the numerous administrative bodies and private pond owners for granting sampling

permissions. This study was funded by the German Research Foundation (DFG) as part of the

Research Training Group on Urban Water Interfaces (GRK 2032).

Author contributions. All authors contributed to designing the study. CR and SH collected the

data. SH carried the statistical analysis, jointly with GS. SH led the manuscript writing. All

authors discussed results and edited the manuscript.

Methane emissions from contrasting urban freshwaters: rates, drivers and a whole-city footprint

138

Supplement of Chapter 5

Methane emissions from contrasting urban

freshwaters: rates, drivers and a whole-city

footprint

Sonia Herrero Ortega, Clara Romero González-Quijano, Peter Casper, Gabriel A.

Singer and Mark O. Gessner

Supplementary Material and Methods

CH4 concentration in water and diffusive flux

A 30-mL syringe fitted with a stop-cock was used to equilibrate 20 mL of water with 10

mL of air collected at the site by vigorous shaking for one minute (Bastviken et al. 2008;

Sobek et al. 2003). A subsample of the headspace (8-10 mL) was injected into a closed

20-mL vial (silicon-PTFE septum; Macherey-Nagel, Düren, Germany) filled with a

saturated salt solution (Daelman et al. 2012) with the injected gas replacing salt solution

escaping through a second needle. Three replicates were taken at each site. The vials

were kept upside-down at 4 °C pending analysis of the headspace gas composition on a

gas chromatograph (GC2014, Shimadzu, Kyoto, Japan). The chromatograph was

equipped with a Shimadzu autosampler (HS20; 1-mL injection loop), three columns

(each 1/8”; packed with Haysep N, 80/100 mesh, 1 m; Haysep D, 80/100 mesh, 4 m; and

Haysep N, 80/100 mesh, 1.5 m), a Shimadzu flame-ionization detector (FID; FID-2014)

for CH4 analysis, and two other detectors for CO2 and N2O analysis, data of which are

not reported here.

CH4 concentrations (mol m-3) in the water (Caq) were calculated as follows:

𝐶𝑎𝑞 =1

𝑉𝑤×(𝐶𝑤×𝑉𝑤+𝑝𝐶𝐻4×𝑉𝐻𝑆

𝑅×𝑇 −1.8×10−6 ×𝑃𝑎𝑡𝑚 ×𝑉𝐻𝑠

𝑅×𝑇 )

where Cw is the concentration of CH4 in the water phase of the syringe (mol m-3); Vw is

the volume of water in the syringe (m-3); pCH4 is the partial pressure (Pa) of CH4 in the

head space at the sampling water temperature; VHS is the volume of gas in the syringe

(m-3); R is the gas constant (8.3143 m3 Pa mol-1 K-1); T is the water temperature during

CHAPTER 5. SUPPLEMENT

139

sampling (K); 1.8×10-6 is the molar fraction (dimensionless) of CH4 in the atmosphere

assuming a global average partial pressure of 1.8 ppm (IPCC, 2014), and Patm is the

atmospheric pressure at the sampling site (Pa).

The partial pressure of CH4 in the headspace at the sampling water temperature (Pa)

was calculated from CGC, the concentration of CH4 in the headspace reported by the gas

chromatograph (mol m-3), as follows:

𝑝𝐶𝐻4= 𝐶𝐺𝑐 ×𝑅×𝑇.

Cw (mol m-3) was calculated according to Henry´s law:

𝐶𝑤=𝑝𝐶𝐻4×𝐾𝐻,

where Henry’s solubility constant KH (mol m-3 Pa-1) was calculated from temperature

according to Weiss (1970) and Sander (2015) using:

𝐾𝐻=𝛽×1 (𝑅×𝑇𝑠𝑡𝑝)

⁄

and

lnβ = A1 + A2 × (100/T) + A3 × ln(

/100),

where β is the Bunsen solubility coefficient for CH4 (dimensionless); T is the water

temperature measured on the sampling day (K); A1 (-67.1962), A2 (99.1624) and A3

(27.9015) are constants given by Yamamoto et al. (1976); R is the gas constant (8.3143

m3 Pa mol-1 K-1); and Tstp is the freezing-point temperature (273.15 K).

Diffusive CH4 flux was calculated from measured concentrations in the water using the

thin boundary equation (MacIntyre et al. 1995):

F = (K600 × (Sc /600)-n)× (Caq - Ceq) × 24 × 1000 ×16,

where Sc is the dimensionless Schmidt number for CH4 at the ambient water

temperature (Wanninkhof 1992), n = 2/3 for wind speeds <3.7 m s−1 and n = 1/2 for

wind speeds >3.7 m s−1 for lakes, ponds and most of the streams (except S5, H3 and

H4). One river (R4) had no or very little flow and we assumed a smooth water surface.

At all other sites, fluxes were calculated based on the assumption that n = 1/2 (Guérin

et al. 2007). F is the diffusive flux (mg d-1 m-2), and Ceq was calculated as follows:

𝐶𝑒𝑞 =1.8×10−6×𝑃𝑎𝑡𝑚×𝐾𝐻,

Methane emissions from contrasting urban freshwaters: rates, drivers and a whole-city footprint

140

assuming a global atmospheric CH4 molar fraction of 1.8×10-6 given a global average

partial pressure of 1.8 ppm (IPCC, 2014), Caq is the concentration at the water surface

(mol m-3), K600 (m h-1) is the gas transfer velocity for a Schmidt number of 600, based

on a frictionless wind speed at 10 m above the ground in m s-1 (U10), calculated according

to Cole and Caraco (1998):

K600

= (2.07 + 0.215 × U101.7) × 0.01.

CH4 ebullition with inverted funnels

Ebullition traps were used to collect gas released from sediments: Limnos traps 0.34 m

in diameter (Limnos, Turku, Finland) and self-made inverted funnels (0.2 m diameter)

with a graduated flask screwed on top. Duplicate traps were deployed in each water body

0.3 m above the sediment surface (Casper et al. 2000) and left in place for a week. Upon

retrieval, the flasks were closed under water with a butyl stopper, the gas volume was

measured and a subsample of 1-2 mL was taken with a gas-tight syringe and injected

into crimped (silicone-PTFE septum; Macherey-Nagel, Düren, Germany) pre-

evacuated vials (20 mL) that had been flushed with N2. Gas analyses were conducted by

gas chromatography (GC 2014, Shimadzu, Kyoto, Japan) immediately upon return to

the laboratory. All lakes, six of the seven ponds and three rivers where water depth

exceeded 50 cm were suitable for these measurements.

Table S1 Characteristics of 32 freshwater sites studied in the metropolitan area of Berlin, Germany. Land use refers to a strip extending 50 m away from the shore. n.a.:

no data available. *Areas for rivers and streams calculated for arbitrary stretches 1 km long to enable intuitive comparisons of relative size with lakes and ponds. †Mean

value of four sampling campaigns. WWTP = wastewater treatment plant

Site

Type

Name

Latitude

Longitude

Area

Max.

depth*

Width

Origign

Land Use (%)

Special features

(ha)

(m)

Agriculture

Forest

Paved area

Green

space

Lake

Biesdorfer See

52°30'11.9"

13°32'58.9"

7.56

5.3

Artificial

Lake

Obersee

52°32'54.8"

13°29'23.0"

3.79

3.1

Artificial

Lake

Plötzensee

52°32'37.7"

13°19'49.8"

7.73

7.5

Natural

100

Lake

Groß Glienicker

See

52°27'51.0"

13°06'53.6"

66.7

11.3

Natural

Lake

Untere Havel

52°26'35.2"

13°08'40.3"

1000

Natural

100

Lake

Schlachtensee

52°26'26.4"

13°12'42.6"

41.6

8.9

Natural

Lake

Müggelsee

52°26'18.1"

13°38'42.4"

765

8.9

Natural

Pond

Hoheheideteich

52°34'34.2"

13°09'52.5"

0.84

1.6

Natural

100

Protected area

Pond

Hamburger

Teich

52°34'02.6"

13°26'43.8"

0.17

1.6

Artificial

Pond

Ruhwaldteich

52°31'32.6"

13°15'35.9"

0.14

2.2

Artificial

Pond

Kienhorstbecken

52°34'38.1"

13°20'44.0"

0.37

1.3

Artificial

100

Pond

Mittelfeldteich

52°36'43.5"

13°13'49.6"

0.41

4.8

Artificial

100

Protected area

Pond

Neurandteich

52°38'19.8"

13°16'25.6"

0.11

0.3

Artificial

Pond

Möwensee

52°33'13.2"

13°20'01.9"

0.6

2.1

Artificial

River

Müggelspree

52°25'47.3"

13°41'19.0"

n.a.

Natural

100

Channelized

River

Landwehr Canal

52°31'09.7"

13°19'10.5"

Artificial

Channelized

River

Spree River 1

52°32'10.3"

13°12'58.4"

n.a.

Natural

100

Channelized

River

Kuhlake

52°34'41.4"

13°09'54.3"

n.a.

Natural

100

Protected area

River

Neukölln Ship

Canal

52°29'22.3"

13°26'22.5"

Artificial

Channelized

River

Spree River 2

52°28'13.7"

13°29'49.2"

131

Natural

100

Channelized

River

Panke River

52°32'11.6"

13°22'00.1"

n.a.

Natural

Channelized,

WWTP

River

Teltow Canal 1

52°26'35.7"

13°19'35.0"

2.5

Artificial

Channelized,

WWTP

River

Teltow Canal 2

52°25'35.1"

13°31'13.4"

n.a.

Artificial

100

Channelized

Site

Type

Name

Latitude

Longitude

Area

Max.

depth*

Widt

Origign

Land Use (%)

Special

features

(ha)

(m)

Agriculture

Forest

Paved

area

Green space

Strea

Zingergraben

52°34'55.5

13°23'09.4

0.18

0.8

Artificial

Channelize

Strea

Schwarzergrabe

52°33'53.6

13°20'57.1

0.11

Natural

Channelize

Strea

Ditch 1

52°37'25.8

13°28'07.8

0.35

Artificial

100

Strea

River Erpe

52°27'32.0

13°36'44.8

0.54

Natural

WWTP

Strea

Koppelgraben

52°37'14.4

13°24'39.2

0.25

Unknow

Strea

Plumpengraben

52°24'54.5

13°33'46.1

0.26

1.6

Natural

100

Channelize

Strea

River Wuhle

52°31'33.2

13°34'46.5

0.42

Natural

Strea

Tegeler Fließ

52°37'53.7

13°22'31.5

0.62

Natural

CHAPTER 5. SUPPLEMENT

143

Table S2 Summary and description of DOM optical properties, modified from Catalán et al.(2013) and

Fasching et al. (2014).

Variable

Definition

Interpretation

References

SUVA (L mg-

1 m-1 )

Ratio of absorbance

coefficient at 254 nm and

DOC concentration (mg L-1

)

Informs on

aromaticity of DOM,

with values generally

between 1 and 6 L

mg-1 m-1

(Weishaar et al.

2003)

S350-400

Ratio of absorption at 350

and 400 nm

Inversely correlated

to molecular weight

(Helms et al. 2008)

S275-295

Ratio of absorption at 275

and 295 nm

Inversely correlated

to molecular weight

(Helms et al. 2008)

Slope ratio of S275-295 to

S350-400

Inversely correlated

to molecular weight

(Helms et al. 2008)

Biological

activity index

(β/α)

Ratio of emission intensities

at 380 and the maximum

between 420 and 435 nm at

an excitation wavelength of

310 nm

Indicator of recent

biological activity or

recently produced

DOM

(Huguet et al. 2009b;

Wilson and

Xenopoulos 2009b)

Humification

index (HIX)

Area under the emission

spectrum between 435 and

480 nm divided by that

between 300 and 345 nm,

given an excitation at 254

Indicator of

humification degree

(Fellman et al. 2010;

Huguet et al. 2009b;

Ohno 2002; Zsolnay

et al. 1999))

Fluorescence

index (FIX)

Ratio of the emission

intensities at 470 and 520

nm at an excitation

wavelength of 370 nm

Indicator of DOM

derived from

terrestrial plants

(low FI ×1.2) or

from microbes or

algae (high FI × 1.4)

(Cory and McKnight

2005; Fellman et al.

2010; Jaffé et al.

2008)

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

144

Table S3 Average annual CH4 emissions (total, diffusive and ebullition flux) to the atmosphere measured

in situ with a chamber connected to an ultraportable gas analyser; diffusive flux calculated from CH4, and

ebullition flux calculated from data collected with inverted funnels (IF) placed on the sediment surface.

n.a.: no data available

Water

body

Site

Total flux

Diffusive

flux

Ebullitive

flux

Diffusive

flux

(TBL)

Ebullitive

flux (IF)

(mg CH4

m-2 d-1)

(mg CH4

m-2 d-1)

(mg CH4

m-2 d-1)

(mg CH4

m-2 d-1)

(mg CH4

m-2 d-1)

Lake

39 ± 27

29 ± 31

10 ± 13

12 ± 21

0 ± 0

753 ± 1123

50 ± 91

459 ± 737

8 ± 8

781 ± 416

128 ± 263

17 ± 15

112 ± 254

35 ± 55

376 ± 333

37 ± 37

35 ± 39

2 ± 4

8 ± 10

12 ± 6

37 ± 41

0 ± 0

9 ± 12

n.a.

32 ± 27

30 ± 29

2 ± 5

33 ± 44

15 ± 17

31 ± 37

29 ± 38

1 ± 2

2 ± 1

0 ± 0

Pond

99 ± 190

15 ± 26

84 ± 170

14 ± 26

512 ± 679

1215 ±

1026

131 ± 131

1070 ±

1108

31 ± 48

1178 ±

903

413 ± 449

187 ± 176

271 ± 333

172 ±

162

289 ± 249

883 ± 1155

590 ± 837

178 ± 242

880 ±

1513

927 ± 747

299 ± 297

157 ± 255

145 ± 103

49 ± 77

294 ± 211

77 ± 75

51 ± 62

20 ± 29

12 ± 0

n.a.

454 ± 753

30 ± 37

466 ± 772

8 ± 10

347 ± 373

River

6 ± 7

0 ± 0

12 ± 16

n.a.

199 ± 381

52 ± 74

147 ± 354

62 ± 100

50 ± 40

14 ± 6

0 ± 1

57 ± 84

n.a.

1 ± 1

0 ± 0

7 ± 11

n.a.

7 ± 5

0 ± 0

19 ± 33

n.a.

5 ± 3

0 ± 0

32 ± 45

n.a.

226 ± 442

20 ± 35

218 ± 439

17 ± 23

214 ± 85

551 ± 943

45 ± 36

506 ± 943

106 ±

180

n.a.

4 ± 3

0 ± 0

8 ± 10

Stream

9 ± 6

0 ± 0

5 ± 5

n.a.

138 ± 175

0 ± 0

12 ± 11

n.a.

3 ± 2

0 ± 0

9 ± 11

n.a.

1 ± 0

n.a.

44 ± 58

31 ± 30

13 46

5 ± 1

n.a.

363 ± 624

3 ± 1

360 ± 624

28 ± 37

n.a.

362 ± 710

61 ± 52

301 ± 669

116 ±

171

n.a.

23 ± 24

0 ± 0

10 ± 15

n.a.

28 ± 40

0 ± 0

6 ± 8

n.a.

CHAPTER 5. SUPPLEMENT

145

Figure S1. Methane emission rates in relation to selected explanatory variables measured in four seasons,

except for POC in spring, where data were unavailable.

Methane emissions from contrasting urban freshwaters: Rates, drivers, and a whole‐city footprint

146

Figure S2. Seasonal changes in diffusive and ebullitive CH4 fluxes from four types of urban water bodies.

Box plots show the median (horizontal line), interquartile range (box limits), highest and lowest values

within 1.5 times the box size from the median (whiskers) and outliers (points). Isolated horizontal lines

are singular values.

Table S4 Physical and chemical surface water variables of 32 freshwater bodies in the city of Berlin, Germany, averaged per water body type and season. Values represent

means ± standard deviations. DO = dissolved oxygen, TP = total phosphorus, DOC = dissolved organic carbon

Water

Season

Temperature

Conductivity

Alkalinity

SRP

NH4+

NO3-

NO2-

DOC

SO42-

Chl a

body

(mg L-1)

(°C)

(μS cm-1)

(mmol L-

(mg L-1)

(μg L-1)

Lakes

Spring

11.8 ± 1.3

11.7 ± 1.5

624 ± 205

7.8 ± 0.3

2.7 ± 1.6

0.07 ±

0.02

0.01 ±

0.01

0.42 ±

0.41

0.42 ± 0.41

0.01 ± 0.01

6.6 ± 2.0

128 ± 75

3.9 ± 4.7

Summer

10.0 ± 1.4

23.9 ± 2.8

638. ± 213

7.8 ± 0.2

3.5 ± 1.2

0.02 ±

0.02

0.01 ±

0.01

0.22 ±

0.38

0.21 ± 0.38

<0.01 ± 0.01

7.4 ± 2.6

130 ± 80

5.7 ± 7.2

Autumn

8.1 ± 1.6

17.5 ± 2.6

658 ± 144

7.9 ± 0.3

3.7 ± 1.4

0.06 ±

0.05

0.05 ±

0.08

0.19 ±

0.39

0.19 ± 0.39

0.01 ± 0.02

28.8 ±

11.4

131 ± 82

15.5 ±

26.2

Winter

10.9 ± 1.5

6.2 ± 1.9

454 ± 130

8.1 ± 0.2

2.6 ± 1.1

0.12 ±

0.10

0.02 ±

0.01

0.48 ±

0.45

0.48 ± 0.45

0.02 ± 0.01

8.0 ± 3.8

130 ± 79

2.7 ± 2.5

Ponds

Spring

8.1 ± 2.4

11.11 ± 1.5

334 ± 42

8.0 ± 0.4

3.0 ± 1.0

0.15 ± 0.2

0.04 ±

0.04

0.06 ±

0.09

0.06 ± 0.09

0.01 ± 0.01

10.2 ± 3.8

22 ± 23

7.7 ± 4.6

Summer

5.1 ± 3.9

20.0 ± 1.1

347 ± 116

7.8 ± 0.3

3.9 ± 1.3

0.37 ± 0.9

0.04 ±

0.03

0.00 ±

0.01

<0.01 ± 0.01

<0.01 ±

<0.01

11.4 ± 2.1

22 ± 20

6.7 ± 5.9

Autumn

4.9 ± 3.8

15.0 ± 1.5

347 ± 94

7.2 ± 1.3

2.7 ± 1.4

0.47 ±

1.09

0.05 ±

0.06

0.04 ±

0.09

0.04 ± 0.09

<0.01 ±

<0.01

39.1 ±

16.7

18 ± 19

13.0 ±

12.5

Winter

7.2 ± 2.9

6.3 ± 1.7

318 ± 104

7.8 ± 0.4

2.6 ± 1.3

0.15 ±

0.28

0.04 ±

0.05

0.09 ±

0.13

0.09 ± 0.13

0.02 ± 0.01

8.8 ± 2.2

36 ± 41

5.3 ± 5.2

Rivers

Spring

10.1 ± 0.9

13.9 ± 3.5

685 ± 141

8.1 ± 0.2

3.0 ± 0.6

0.11 ±

0.03

0.04 ±

0.03

1.07 ±

1.35

1.07 ± 1.35

0.04 ± 0.05

7.5 ± 1.6

177 ± 62

4.3 ± 4.4

Summer

5.3 ± 2.1

21.4 ± 2.0

985 ± 381

8.0. ± 0.4

3.0 ± 1.8

0.20 ±

0.17

0.11 ±

0.14

1.73 ±

2.32

1.73 ± 2.32

0.02 ± 0.02

8.42 ± 2.2

170 ± 55

1.8 ± 2.0

Autumn

5.8 ± 2.3

17.9 ± 1.7

829 ± 266

8.0 ± 0.3

2.2 ± 0.6

0.10 ±

0.05

0.11 ±

0.06

1.56 ±

1.61

1.56 ± 1.61

0.02 ± 0.01

36.56 ±

9.3

190 ± 57

2.0 ± 2.9

Winter

11.6 ± 1.2

6.2 ± 1.8

688 ± 275

8.0 ± 0.1

2.9 ± 0.7

0.17 ±

0.21

0.03 ±

0.02

1.26 ±

1.04

1.26 ± 1.04

0.05 ± 0.07

7.79 ±

1.39

179 ± 54

3.0 ± 2.5

Streams

Spring

8.9 ± 2.3

8.8 ± 1.6

557 ± 327

8.0 ± 0.3

3.8 ± 1.6

0.20 ±

0.20

0.13 ±

0.23

1.16 ±

1.33

1.16 ± 1.33

0.05 ± 0.07

11.9 ± 6.0

101 ± 49

13.2 ±

14.8

Summer

4.5 ± 2.5

17.7 ± 2.7

762 ± 357

7.9 ± 0.5

2.4 ± 0.8

0.41 ±

0.48

0.39 ±

0.33

1.26 ±

2.24

1.26 ± 2.24

0.05 ± 0.04

10.8 ± 2.8

82 ± 49

2.8 ± 2.0

Autumn

6.0 ± 2.4

15.1 ± 1.1

985 ± 279

7.9 ± 0.2

2.9 ± 1.1

0.16 ±

0.14

0.17 ±

0.11

1.21 ±

1.53

1.21 ± 1.53

0.07 ± 0.09

56.8 ±

22.2

123 ± 48

2.6 ± 4.4

Winter

8.4 ± 2.9

6.4 ± 1.0

593 ± 410

7.7 ± 0.3

3.4 ± 1.9

0.61 ±

1.12

0.11 ±

0.17

1.27 ±

1.35

1.27 ± 1.35

0.04 ± 0.03

12.9 ± 7.6

89 ± 70

4.3 ± 7.7

CHAPTER 6

149

Synthesis

This thesis aimed to enhance our understanding of human impacts on carbon cycling in

freshwater ecosystems, particularly focusing on dissolved organic matter (DOM) composition,

ecosystem metabolism, and greenhouse gas (GHG) dynamics. I conducted a comprehensive

investigation including two distinct stages of human landscape development: transitioning from

a near-natural ecosystem impacted by cattle herding, the Mara River in the African savannah,

to a highly urbanized ecosystem, the aquatic network of one of Europe's largest cities, Berlin.

In the following chapter, I will outline the key contributions of this doctoral dissertation to the

current knowledge regarding the role of dissolved organic matter in mediating human impacts

on aquatic ecosystem functions with implications for the carbon cycle. I delved into the effects

of human activities on DOM composition in the Kenyan savannah (Chapter 2) and in the entire

urban aquatic network of Berlin, Germany (Chapter 3). Additionally, I examined the influence

of human activities on ecosystem metabolism (Chapter 2). Finally, I conducted an analysis of

GHG emissions and their drivers in highly impacted aquatic ecosystems (Chapters 4 and 5).

Only a limited number of studies have considered both urban lentic and lotic waters, whereas in

this dissertation, I investigated rivers, streams, ponds, and lakes within the aquatic network of

an entire city (Chapters 3, 4, and 5).

Synthesis

150

6.1 DOM characteristics as indicators of landscape change

Human development has important effects on DOM in aquatic ecosystems (Xenopoulos et al.

2021). The use of simple optical descriptors of DOM (absorbance and fluorescence) allowed me

to identify the effect of human-induced changes on DOM composition, independent of the origin

of the impact on the system. This has implications for future water management strategies. As

optical measurements are inexpensive and DOM may provide useful information on

anthropogenic impacts, water authorities could consider including DOM composition analyses

in their regular monitoring. I found that optical results were in accordance with more

sophisticated and expensive methodologies such as ultrahigh-resolution Fourier-transform ion

cyclotron mass spectrometry (FT-ICR-MS). For instance, as revealed by our mesocosm

experiments (Chapter 2), the replacement of hippo by cattle dung in the riparian area of the

Mara River results in an increase in humic fractions but also diversity of DOM: When cattle

dung was more abundant than hippo dung, the diversity of DOM increased significantly over

time. This shift progressed from an initial dominance of allochthonous DOM, transitioning to a

prevalence of microbially produced DOM, and eventually culminating in the dominance of

autochthonous DOM generated by primary production. In the mesocosms affected by cattle

dung, DOM had more humic-like components associated with microbial activity and a

significant proportion of a fulvic acid-like component originating from higher plant material.

The disparities in DOM composition between DOM deriving from hippo and cattle dung could

stem from variations in digestion efficiency between cattle and hippos, as noted by Fritz et al.

(2009), and also from differences related to gross primary production, where DOM became an

indicator of ecosystem functioning.

As Chapter 3 showed, DOM composition in an urban aquatic network also showed clear

differences among water body types exposed to strong human influences. DOM in streams was

characterized by higher aromaticity and lower amounts of recently produced DOM, similar to

previous studies in human-impacted inland waters (Graeber et al. 2012; Park 2009), whereas

DOM in lakes showed the opposite. Importantly, physical connections between aquatic and

terrestrial environments in urban systems are often paved surfaces and constructed drainage

systems, including sewage overflows. The “urban allochthonous signal”-gradient showed in

Chapter 3 is thus suggested to be affected by runoff events than seepage through soils and

subsequent delivery to lakes and rivers via groundwater. In contrast to (near-)natural

allochthony, which can normally be recognized by signatures showing high proportions of soil-

derived humic DOM (Hutchins et al. 2017), I found high levels of proteins to be characteristic

of “urban allochthony signal” which may specifically originate from WWTPs, as also implied by

the nature of some of the PARAFAC components. Furthermore, I found seasonal variation in

DOM characteristics to be prevalent in lakes, ponds, rivers and streams but independent from

variation in levels of allochthony across all water bodies. In summer and autumn, DOM

signatures suggested a higher proportion of more recent origin constituents than in winter and

CHAPTER 6

151

spring. Ponds and streams showed higher and less predictable seasonal turnover of DOM

constituents than lakes and rivers.

Figure 1 Variation in DOM composition along urban aquatic ecosystems as indicated by PARAFAC

components C2 and C1, SUVA254 as a proxy for DOM aromaticity (Weishaar et al. 2003), E2:E3 as the

ratio of absorbance at 250 and 365 nm, which serves as an (inverse) indicator of molecular size (Chen et

al. 1977; Peuravuori and Pihlaja 1997), and the Freshness index β/α (Wilson and Xenopoulos 2009b)

indicating the relative importance of recently produced DOM (Parlanti et al. 2000). Pictures by C. Romero

The analysis of DOM composition provides insight into the extent of human impact on

freshwater ecosystems across spatial scales, from individual river reaches to entire catchment

areas. In particular, the studies included in this dissertation suggest that DOM composition can

be useful as an indicator of land use changes, both in an African savannah and an urban

ecosystem. Moreover, DOM composition can serve as an indicator to assess human influences,

distinguishing between aquatic ecosystems ranging from near-natural to highly urbanized. This

highlights the potential to use DOM characteristics as indicators that could be integrated into

routine water-quality assessments and monitoring programs conducted at different spatial and

temporal scales. Assessments based on DOM optical indexes are cost-effective and provide

information that traditional methods, as the measure of the dissolved organic concentration,

often struggle to capture. However, to enhance the robustness of such DOM-based assessments,

continuous monitoring of DOM composition, rather than snapshots as in the present study, is

likely to be beneficial. This could strengthen water-quality assessments in line with the EU

Water Framework Directive and other legal frameworks.

Lakes- E2:E3 & β:α

fresh DOM

Higher autochthony signal

Streams – C2, C1 & SUVA254

More recalcitrant DOM

Higher allochthony signal

Synthesis

152

6.2 Ecosystem functioning reveals impacts of land-use alteration

Aquatic ecosystem metabolism has also been suggested as an indicator of human impacts such

as land-use alteration (Bernot et al. 2010; Jankowski et al. 2021). The change of African savannah

to pasture, which is accompanied by the substitution of cattle for hippos, produces various

responses in aquatic ecosystem functioning. Foremost, cattle dung increases nutrient

concentrations and stimulates primary production in aquatic ecosystems. However, in the

experiment we carried out in Kenya, where cattle and hippo dung were added to mesocosms in

different proportions, ecosystem respiration was unaffected, suggesting that changes in

resources caused by supplying a different type of dung were too small to trigger an answer of

periphyton. The results presented in Chapter 2 improve knowledge on the consequences of land-

use change for aquatic ecosystem functioning in African savannah, but also point to a need for

more research on the ecological significance of replacing populations of large native herbivores

by livestock.

These effects of the changes of land use in the savannah on ecosystem functions could be

compared to the effects of urbanization on freshwater ecosystems studied in Chapter 3, 4 and 5.

In some parts of the world, waste water is directly discharged into receiving surface waters

without or with very little treatment. In other areas, like Berlin, waste water treatment is

effective, unless sewage overflow results in raw wastewater ending up in surface waters during

heavy rain events. I could identify WWTP inputs in the aquatic network in Berlin, where I

found some protein-like components indicative of wastewater inputs, associated with high CO2

fluxes. In other studies, WWTP effluents have also been found to promote respiration (Aristi et

al. 2015) and to decrease Gross Primary Production (GPP) (Rodríguez-Castillo et al. 2017). In

another study, photosynthetic rates immediately downstream of WWTPs did not increase

because nutrients were already high because of agricultural nutrient loading upstream of the

WWTP (Venkiteswaran et al. 2015).

CHAPTER 6

153

6.3 Improving global C budgets by considering spatio-temporal variability

of GHG emissions in urban aquatic ecosystems

At present, urban aquatic ecosystems are not specifically included in global estimates of carbon

dioxide and methane emissions from freshwaters, which have large uncertainties (Aufdenkampe

et al. 2011; Bastviken et al. 2011; Deemer et al. 2016; Stanley et al. 2016; Tranvik et al. 2009).

The data presented in Chapters 4 and 5 of this dissertation contribute towards reducing these

uncertainties. The estimated annual emissions of all of Berlin’s surface waters amount to 8.5±

1.3 Gg CO2 (Chapter 4) and 2.6 ± 1.7 Gg CH4 (72.8 ± 47.6 Gg CO2-equivalents, mean ± SD,

Chapter5). CO2 emissions from Berlin´s flowing waters (Chapter 4) were lower than rivers and

streams reported in the literature (Gu et al. 2022; Holgerson and Raymond 2016). This may be

partly due to the lowland character of the streams and rivers in Berlin, with low flow velocities,

and partly to the channelized nature of these water bodies, which translates to low gas transfer

velocities. CH4 emissions, in contrast, were higher than those from lentic and lotic natural water

bodies (Chapter 5), especially from ponds, where CH4 emissions were probably linked to high

productivity. As a result, urban ponds emerged in this dissertation as hotspots for CH4 emissions

(Chapter 5). This should be considered in urban water management, as ponds have been shown

to have positive environmental effects in urban areas, for instance as nutrient retention systems

or biodiversity hotspots (Hassall 2014), but they also contribute to greenhouse gas emissions

(Chapter 5). Nevertheless, further research is needed to better understand the mechanisms and

relationships between trophic state, nutrient levels, microbial activity, and CH4 and CO2

emissions in urban aquatic ecosystems to inform strategies towards emission reductions.

DOM composition was related to CO2 flux (Chapter 4) and influenced by urban factors, such as

trace organic compounds or paved surfaces, suggesting that understanding the sources and

composition of DOM in urban water bodies is crucial for assessing the environmental impact of

urban surface waters, including their contribution to CO2 fluxes across the water-air interface.

This may have implications for urban planning and environmental management of urban water

interfaces in the context of carbon cycling and climate change mitigation.

A major challenge in our effort to estimate Berlin-wide CO2 emissions from various water bodies

was choosing the right method to calculate the gas exchange velocity, k. Different methods have

been previously used to calculate k (Raymond and Cole 2001; Schelker et al. 2016), but there are

no comparative studies for urban aquatic ecosystems. Several factors potentially affecting gas

exchange are likely to differ in importance between flowing and standing urban waters: water

velocity might be crucial for rivers and streams, whereas wind speed might be more important

for lakes and ponds. More studies should focus on improving methodologies before continuing

attempts to improve global estimates. Additionally, diel variability of fluxes has not yet been

considered in global estimates. For CO2 (Chapter 4), I studied this variability from different

angles: variability associated with concentration changes, k, and seasonal and daily fluctuations.

Synthesis

154

Variability differed among urban lakes, ponds, rivers and streams, with higher daily variability

in ponds and lakes than in rivers and streams. Additionally, for flowing waters, concentration

changes were more important than gas exchange velocity in determining flux variability. It is

important to consider this variability when designing sampling strategies and upscaling from

local to larger scales, considering important differences in the characteristics of standing and

flowing waters.

6.4 Conclusions

● Cattle dung increases GPP and diversifies dissolved organic matter DOM composition, while

hippo dung reduces primary production and delays GPP responses. These differences, driven by

variation in dung particle size could result in significant changes in aquatic ecosystem structure

and function when livestock replaces hippos in the African savannah. The findings underscore

the species-specific nature of ecological roles and suggest that introducing or rewilding species

as replacements for extinct ones may lead to unintended consequences.

● DOM composition from the Berlin aquatic network differed significantly between water body

types, with lakes having more natural DOM and streams containing more DOM from external

sources. Seasonal variation was observed in all water bodies, potentially influenced by factors

such as urbanization, nutrient supplies, wastewater treatment plant effluents, and changes in

leaf litter inputs.

● Key drivers of DOM composition included nutrient supply, the importance of green space, and

trace organic pollutants, which could be detected using simple optical measurements. This study

suggests that analyses of optical DOM properties can complement existing water-quality

assessments, providing a fast and cost-effective method. Continuous monitoring of DOM

composition could enhance water-quality assessments in line with the EU Water Framework

Directive.

● Urban waters emit both CO2 and CH4. The estimated total annual emissions from all of Berlin’s

surface waters are 8.5 ± 1.3 Gg CO2 and 2.6 ± 1.7 Gg CH4 (mean ± SD).

● Carbon dioxide emission rates from Berlin’s running waters were lower than reported in the

literature for urban streams and rivers. CO2 variability played an important role in our study.

Daily variability should be considered when carrying out global estimations, especially for ponds

and lakes.

● Trophic state variables like particulate organic carbon (POC), chlorophyll a (chl a), and

dissolved oxygen (DO) concentration had the highest explanatory power for CH4, suggesting

that trophic state plays a crucial role in CH4 emissions from urban freshwaters. Ponds were

identified as CH4 emission hotspots. These findings contribute to a better understanding of CH4

emissions from urban areas, aiding in constraining city-level and global estimates.

CHAPTER 6

155

6.5 Outlook

● Explore the effects of changes from hippos to cattle in the African savannah on GHG emissions.

In order to be able to capture GHG, developing cheaper sensors would be a first step.

● Collaborate with stakeholders in the water sector to establish DOM analyses as an approach in

freshwater monitoring. Develop knowledge-transfer strategies to incorporate DOM optical

properties in global regular monitoring.

● Implement new methodologies and develop guidelines for measurements and calculations of

GHG for urban aquatic ecosystems – temporal variation of GHG from freshwaters at a global

scale.

Figure 2 Conceptual illustration of the four chapters and the suggested future studies (boxes in dash

lines)

Complementary contributions

156

Complementary contributions

(not included in this dissertation)

1. Attermeyer, K., J. P. Casas-Ruiz, T. Fuss, A. Pastor, S. Cauvy-Fraunié, D. Sheath, A. C. Nydahl,

A. Doretto, A. P. Portela, B. C. Doyle, N. Simov, C. Gutmann Roberts, G. H. Niedrist, X.

Timoner, V. Evtimova, L. Barral-Fraga, T. Bašić, J. Audet, A. Deininger, G. Busst, S. Fenoglio,

N. Catalán, E. de Eyto, F. Pilotto, J.-R. Mor, J. Monteiro, D. Fletcher, C. Noss, M. Colls, M.

Nagler, L. Liu, C. Romero González-Quijano, F. Romero, N. Pansch, J. L. J. Ledesma, J. Pegg,

M. Klaus, A. Freixa, S. Herrero Ortega, C. Mendoza-Lera, A. Bednařík, J. A. Fonvielle, P. J.

Gilbert, L. A. Kenderov, M. Rulík, and P. Bodmer. 2021. Carbon dioxide fluxes increase from

day to night across European streams. Communications Earth & Environment 2:118.

2. Bravo, A. G., D. N. Kothawala, K. Attermeyer, E. Tessier, P. Bodmer, J. L. J. Ledesma, J. Audet,

J. P. Casas-Ruiz, N. Catalán, S. Cauvy-Fraunié, M. Colls, A. Deininger, V. V. Evtimova, J. A.

Fonvielle, T. Fuß, P. Gilbert, S. Herrero Ortega, L. Liu, C. Mendoza-Lera, J. Monteiro, J.-R.

Mor, M. Nagler, G. H. Niedrist, A. C. Nydahl, A. Pastor, J. Pegg, C. Gutmann Roberts, F.

Pilotto, A. P. Portela, C. R. González-Quijano, F. Romero, M. Rulík, and D. Amouroux. 2018.

The interplay between total mercury, methylmercury and dissolved organic matter in fluvial

systems: A latitudinal study across Europe. Water Research 144:172-182.

3. Goldsmith, G. R., S. T. Allen, S. Braun, N. Engbersen, C. R. González-Quijano, J. W. Kirchner,

and R. T. W. Siegwolf. 2019. Spatial variation in throughfall, soil, and plant water isotopes in a

temperate forest. Ecohydrology 12:e2059.

List of Figures

157

List of Figures

Chapter 1 General Introduction

Figure 1 Growth rates of urban agglomerations by size class (UN, 2018)

Figure 2 Example of Dissolved organic matter (DOM) components identified by

PARAFAC in Chapter 3. Em=emission Ex= Excitation. Component C1: humic-like and

recalcitrant, C2: terrestrial humic-like in waste water treatment impacted water, C3:

humic-like, C4: terrestrial humic-like, suggested as photo-refractory, C5: anthropogenic,

microbial humic-like, C6; protein-like, linked to autochthonous production and C7:

protein-like, waste water treatment origin.

Figure 3 Conceptual graph that illustrates the drivers of stream ecosystem metabolism

across varying spatial scales. The regional template is anticipated to influence climate,

vegetation, and topography. Factors at the watershed scale play a crucial role in

determining nutrient availability and the hydrologic regime. Concurrently, local-scale

riparian canopy characteristics exert the strongest control over terrestrial organic matter

(OM) and light conditions. Seasonality is projected to impact all these factors, potentially

causing shifts in the interplay between watershed and local-scale controls on Gross

Primary Production (GPP) and Ecosystem Respiration (ER). The dashed line signifies

the expected influence of hydrology on the retention of terrestrial organic matter within

stream channels. The size of the arrows approximately reflects the magnitude of

influence. Large mammals replacement by cattle might be another local-scale control,

which might affect the OM composition. Modified from (Alberts et al. 2017).

Figure 4 Conceptual graph summarizing research included in this dissertation. Chapter

2 focuses on the influence of human activities on African savannah rivers by analyzing

the effects of replacing hippopotamus by livestock in the catchment. Chapters 3, 4 and 5

address the influence of urban development on carbon dynamics in contrasting aquatic

ecosystems of a large European city. DOM = Dissolved Organic Matter. Ch. = Chapter.

Chapter 2 Hippopotamus are distinct from domestic livestock in their resource

subsidies to and effects on aquatic ecosystems

Figure 1. Dynamics of GPP (a) and ER (e) over 44 days as fitted with a three-parameter

sigmoid Gompertz model. To test relationship with dung treatment, we plotted mean

and s.d. of upper asymptote K, maximum rate of increase and lag for Gompertz models

for GPP (b,c,d ) and ER ( f,g,h) as a function of dung treatment, respectively, and fitted a

smoothing model (grey line with shaded area represents smoother mean and s.e.;

smoother significance, R2 and GCV are supplied in the figures). Note that parameter

estimates for the smoothing models (n = 3 per treatment) were based on fits to data of

individual flumes. Note also log-scale for lag in (d ) owing to excessive lag in two flumes

with 0% cattle dung.

Figure 2. Weekly measures of flume-scale GPP (a), flume-scale ER (b), GPP : ER (c) and

NEP (d ). The dashed line indicates NEP = 0, and most of the mesocosms were net

heterotrophic until day 7 and then switched.

Figure 3 PCA based on descriptors of DOC. DOC composition changed over time

towards a common endpoint composition when plotting scores (mean ± s.d. per treatment

and time) (a). The PCA was based on PARAFAC components C1 to C7, high and low

molecular weight substances (HMWS, LMWS), ratio of HMWS : LMWS and C : N of

HWMS, humic-like substances (HS), aromaticity via specific ultraviolet absorption at 254

nm (SUVA), humification index (HIX), fluorescence index (FIX), freshness index β : α

(FreshIndex) and an absorbance-based indicator of molecular size (E2 : E3) (b). Stream-

List of Figures

158

specific changes of DOC composition were quantified as cumulative Euclidean distance

in PCA space considering all its dimensions and progress along a path of consecutive time

points; the graph shows average total path length per treatment (c). Note that the arrows

in (a) designate the time series from early to late in experiment for each treatment, and

‘total DOC dynamics’ in (c) describes the approximate ‘length’ of the temporal arrows in

(a).

Supplement of Chapter 2 Hippopotamus are distinct from domestic livestock in their

resource subsidies to and effects on aquatic ecosystems

Figure S1: Experimental set-up and dung used in mesocosms: (a) allocation of dung

treatments in three blocks driven independently by paddle wheels, (b and c) layout and

details of mesocosms, (d) hippo dung, and (e) cattle dung.

Figure S2. Influence of dung treatment on (a) soluble reactive phosphorus (SRP), (b)

nitrite, (c) ammonium, and (d) nitrate concentrations. Asterisks are displayed for

significant linear relationships across low-high proportions of cattle dung for each

sampling occasion (α ≤ 0.05). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure S3. Influence of time on (a) soluble reactive phosphorus (SRP), (b) nitrite, (c)

ammonium, and (d) nitrate concentrations among dung treatments.

Figure S4. Influence of dung treatment on a) DOC, b) chlorophyll-a (Chl-a), c) ash-free

dry mass (AFDM), d) total suspended solids (TSS), and e) particulate organic matter

(POM) concentrations. Asterisks and model fits are displayed for significant linear

relationships (α ≤ 0.05). *P < 0.05, **P < 0.01, ***P < 0.001

Figure S5. Observed excitation and emission wavelengths for maximum fluorescence of

the 7 PARAFAC components identified in our dataset.

Figure S6. Emission and excitation loadings of the 7 PARAFAC components

Figure S7. Performance of different values of theta in modeling metabolism in our

experimental mesocosms. A higher value of theta (1.1085, upper panel) performed better

for most streams when compared with a common literature value of 1.045 (lower panel).

a and b are model fits for day 1, and c and d are model fits for day 10 in the hippo dung

treatment (100 % hippo dung). The black bold line is for measured dissolved oxygen

concentration (mg/L) while the red line is for the modeled dissolved oxygen

concentration. The red dotted line is measured temperature and the blue dotted line is

light intensity. The green dotted line is for oxygen saturation. Modeling for each day

was performed from mid-night (0 minutes, 24:00 hrs) to mid-night, 1440 minutes, 23:59

hrs).

Figure S8. Model outputs using a theta value of 1.1085. Weekly measures of flume-scale

gross primary production (GPP), (a) flume-scale ecosystem respiration (ER; b), GPP:ER

indicate NEP = 0, and most of the mesocosms were net heterotrophic on until day 7 and

then switched.

Figure S9. Model outputs using a theta value of 1.045. Weekly measures of flume-scale

gross primary production (GPP), (a) flume-scale ecosystem respiration (ER; b), GPP:ER

indicate NEP = 0, and most of the mesocosms were net heterotrophic on until day 7 and

then switched.

Chapter 3 Dissolved organic matter signatures in urban surface waters: spatio-

temporal patterns

List of Figures

159

Figure 1 Map of 32 sampling sites in the city of Berlin, including 7 lakes (dark green), 7

ponds (light green), 9 streams (light blue), and 9 rivers (dark blue), including two heavily

polluted stream sites and two heavily polluted river sites. Wastewater Treatment Plants

(WWTP) are shown in orange, arrows point to locations where the effluents are

discharged (a). Scores of a Principal Component Analysis (PCA) of DOM characteristics

are shown as color gradients for all sites sampled in four seasons (b,c). The PCA is based

on DOC concentrations, all absorbance and fluorescence data, absolute component-

specific fluorescence intensities from PARAFAC, and data from size-exclusion

chromatography. Different colours indicate differences in DOM composition. Site codes

are given in Table S1. Sites marked by asterisks (*) were restricted to 3 seasons and hence

excluded from the PCA.

Figure 2 Ordination of sites (a) by PCA based on DOM characteristics (b): (i) indices

derived from measurements of absorbance (E2:E3 indicating molecular size, E4:E6

representing the humification ratio, the slope ratio SR, and SUVA254) and fluorescence

(freshness index β:α, fluorescence index FI, and humification index HIX), (ii) PARAFAC

components C1 to C7, and (iii) data from size exclusion chromatography (humic-like

substances HS, high-molecular weight non-humic substances HMWS, low-molecular

weight substances LMWS). (c) Potential drivers of DOM composition, that were used as

constraints in the RDA, were mapped onto the PCA ordination, with the significant

constraints marked by an asterisk (*). (d) FT-ICR-MS-derived indices and molecular

groups mapped onto the PCA ordination representing only groups correlated with PC1

or PC2 (r>0.2; oxygen richness O:C, saturation level indicated by H:C, double-bond

equivalents DBE, aromaticity index AI, molecular lability boundary MLB, molecular

groups g1 and g2 indicating black carbon without and with heteroatoms, g5 consisting

of unsaturated aliphatics, g7 representing saturated fatty acids, g8 and g9 denoting

carbohydrates without and with heteroatoms N, S or P, and g10 comprising peptides).

The molecular group measures are either average masses (marked by ‘_a’) or counts of

molecules (marked by ‘_c’).

Figure 3 (a) PCA biplot based on DOM absorbance and fluorescence indices, PARAFAC

components and size exclusion chromatography data from 4 contrasting types of urban

freshwater bodies, including lakes, ponds, rivers, and streams, in addition to two streams

and two rivers specifically selected as highly polluted sites. Each of the ellipses represents

one sampling site that was visited 4 times, once in each season. Site codes are given in

Table S1. (b) Visual comparison of site-specific seasonal variation based on the size, shape

and orientation of ellipses, plotted separately per site. (c) Seasonal variation across sites

illustrated by ranking the sampling dates at each site according to the PC2 scores, as

shown in the inset. The stacked histograms show frequencies of the seasons across the

four ranks. Summer samples tend to produce high scores at most sampling sites, whereas

winter samples tend to score low.

Supplement of Chapter 3 Dissolved organic matter signatures in urban surface waters:

spatio-temporal patterns

Figure S1 Emission and excitation wavelengths of PARAFAC components. Solid lines

represent emission spectra, dashed lines excitation spectra. Lines in different shades of

grey refer to models using different sample sub-sets of a split-half validation analysis.

Figure S2 Van Krevelen plots showing all molecules (sum formulas) identified by FT-

ICR-MS analysis of DOM samples collected at 32 urban sites over three seasons

(summer, autumn and winter). Colour indicates molecule-specific Spearman correlation

coefficients of the relative intensities of each compound with the first (a) and second (b)

axis of the PCA shown in Figures 2 and 3. The data points were plotted in random order

to avoid bias resulting from identical O:C and H:C ratios for many sum formulas.

List of Figures

160

Figure S3 Principal Component Analysis (PCA) of 32 urban sites in the city of Berlin

over four seasons (a) and Trace Organic Compounds (TrOCs) (b). Site S5 had extreme

PC1 and PC2 scores; the site was included in the analysis but is not presented in the

biplot to better visualize variability among the other sites. Abbreviations of the TrOCs

(B) are explained in Table S7.

Figure S4 Redundancy Analysis (RDA) of urban sampling sites (a) visited 4 times over

one year, the DOM characteristics included in the analysis (b) and the predictor variables

(c), the last marked by an asterisk (*) when significant. DOM characteristics include (i)

absorbance and fluorescence indexes (E2:E3, molecular size, E4:E6, indicator of

humification, SR, slope ratio, β:α, freshness index, SUVA254 and HIX, humification

index), (ii) PARAFAC components (C1 to C7), and (iii) fractions derived from size

exclusion chromatography (HS, humic-like substances; HMWS, high-molecular weight

non-humic substances; and LMWS, low-molecular weight substances).

Figure S5 Precipitation and flow at a site within the city of Berlin during the study

period, with the grey boxes indicating the four sampling periods.

Figure S6 Relationship between iron (Fe) and the absorbance at 420nm relative to

DOC (a420/DOC)

Chapter 4 Carbon dioxide emissions across an urban aquatic network

Figure 1 Map of Berlin, Germany, showing the city’s aquatic network, land use, and the

32 sampling sites selected in the present study.

Figure 2 Map of instantaneous CO2 fluxes (g C-CO2 m-2 d-1) at 32 sites distributed across

the city of Berlin, Germany, in four seasons.

Figure 3 Summary of instantaneous and continuous measurements of CO2 fluxes from

four different types of water bodies in the city of Berlin, Germany, determined in four

seasons.

Figure 4 Selected high-resolution time courses of water temperature and wind speed

(A,B), xCO2(AQ) and xCO2(HS) (C,D), and CO2 and k from an urban stream (S3) in spring

(A,C,E) and an urban pond (P5) in summer (B,D,F), illustrating variability at different

scales.

100

Figure 5 Instantaneously (A) continuously measured CO2 fluxes (B) from water bodies

in the city of Berlin, Germany, in comparison to fluxes reported in published studies (C).

102

Figure 6 Principal component analyses (PCA) summarizing potential drivers of CO2

fluxes from four types of urban water bodies: A) Analysis based on dissolved organic

matter (DOM) characteristics derived from absorbance/fluorescence measurements,

size-exclusion chromatography and PARAFAC (Table S1-S2 Supporting Information; B)

analysis based on water depth and temperature as well as land-cover and physico-

chemical variables such as NH4+, NO3-, NO2-, chl a, and TP (Table S3, Supporting

Information; and C) analysis based trace organic compounds as summarized in Table S5,

Supporting Information.

104

Figure 7 Partial plots of the main drivers identified in random forest models for

instantaneousCO2 fluxes, continuous fluxes, and daily variability of the continuous fluxes.

106

Supplement of Chapter 4 Carbon dioxide emissions across an urban network

111

Figure S1 Relationship between CO2 fluxes determined with two different approaches,

instantaneous measurements on single days in the morning vs continuous measurements

122

List of Figures

161

over a week. Dashed line indicates 1:1 line. Lentic: lakes and ponds, lotic: rivers and

streams.

Figure S2 Daily (A), day to day (B) (from the Continuous fluxes) and seasonal standard

deviation (SD) (C) (from the Instantaneous fluxes)

122

Chapter 5 Methane emissions from contrasting urban freshwaters: Rates, drivers,

and a whole‐city footprint

123

Figure 1 Map of the metropolitan area of Berlin, Germany, showing land use and

freshwater sampling locations in lakes (L1–7), ponds (P1–7), rivers (R1–7), streams (S1–

7), and four additional running water sites characterized by high nutrient concentrations

(H1–4)

125

Figure 2 Seasonal changes in (a) daily mean air temperature in Berlin Tempelhof

recorded by the German Meteorological Office, with the light gray area representing a

period of ice cover on the larger lakes and the dark gray areas representing the sampling

periods, and (b) Total methane (CH4) emissions from four types of urban water bodies.

Box plots show the median (horizontal line), interquartile range (box limits), highest and

lowest values within 1.5 times the box size from the median (whiskers) and outliers

(points)

130

Figure 3 Principal component analysis of 32 water bodies sampled over four seasons,

based on potential explanatory variables for CH4 emissions. (a) Water body types differed

mainly along the first two principal components, (b) PC3 indicates a slight tendency of

lakes to differ from all other water bodies, and PC4 tended to distinguish autumn from

all other seasons. (c, d) Dominant variables creating the ordination space relate to land

use, water chemistry, and optical properties of dissolved organic matter. Black lines are

scaled structure coefficients (scaling factor of 8), that is, correlations with the principal

components. Gray lines show analogous correlations with particulate organic carbon,

which were added a posteriori because data from only three seasons were available. Only

variables with a structure coefficient >0.15 in (c) or (d) were plotted

132

Supplement of Chapter 5 Methane emissions from contrasting urban freshwaters: Rates,

drivers, and a whole‐city footprint

138

Figure S1. Methane emission rates in relation to selected explanatory variables measured

in four seasons, except for POC in spring, where data were unavailable

145

Figure S2. Seasonal changes in diffusive and ebullitive CH4 fluxes from four types of

urban water bodies. Box plots show the median (horizontal line), interquartile range (box

limits), highest and lowest values within 1.5 times the box size from the median

(whiskers) and outliers (points). Isolated horizontal lines are singular values

146

Chapter 6 Synthesis

149

Figure 1. Variation in DOM composition along urban aquatic ecosystems as indicated

by PARAFAC components C2 and C1, SUVA254 as a proxy for DOM aromaticity

(Weishaar et al. 2003), E2:E3 as the ratio of absorbance at 250 and 365 nm, which serves

as an (inverse) indicator of molecular size (Chen et al. 1977; Peuravuori and Pihlaja 1997),

and the Freshness index β/α (Wilson and Xenopoulos 2009b) indicating the relative

importance of recently produced DOM (Parlanti et al. 2000). Pictures by C. Romero

151

Figure 2 Conceptual illustration of the four chapters and the suggested future studies

(boxes in dash lines)

155

List of Tables

162

List of Tables

Chapter 1 General Introduction

Table 1 Estimates of aquatic carbon fluxes (Pg) from Drake 2018. Black values indicate

an independent estimate was provided by the given study. Gray values were not refined

by the given study but indicate where an estimate was applied from previous or future

study.

Chapter 2 Hippopotamus are distinct from domestic livestock in their resource

subsidies to and effects on aquatic ecosystems

Table 1. Results of mixed-effects models for loge(Χ)-transformed dissolved organic

carbon (DOC, mg l−1), chlorophyll a (Chl-a, mg l−1), ash-free dry mass (AFDM, mg

cm−2), total suspended solids (TSS, mg l−1), particulate organic matter (POM, mg l−1),

soluble reactive phosphorus (SRP, μg l−1), total phosphorus (TP, mg l−1), ammonium

(mg l−1), nitrite (mg l−1) and nitrate (mg l−1). The marginal R2 (GLMM(m); fixed

effects only) and the conditional R2 (GLMM(c); fixed and random effects) represent the

proportion of variance explained by each model; s.e. = standard error; s.d. = standard

deviation; *p < 0.05, **p < 0.01, ***p < 0.001

Supplement of Chapter 2 Hippopotamus are distinct from domestic livestock in their

resource subsidies to and effects on aquatic ecosystems

Table S1: Mean dung C, N, P concentrations and stoichiometry for some large

mammalian herbivores of the African savannah

Table S2. Estimated loading rates of organic matter (dung) by cattle and hippopotamus

in the Mara River, Kenya.

Table S3 Characteristics of hippo dung and cattle dung used in the mesocosms in this

study.

Table S4: Fluorescent components of DOM as identified by parallel factor analysis

(PARAFAC). Given are observed excitation and emission wavelengths for maximum

fluorescence, alignment with distinct fluorescence peaks and PARAFAC components

identified in previous studies, probable sources of DOC and a literature-based component

descriptiona.

Table S5: Summary of generalized additive mixed modeling (GAMM) analyses to

determine the effect of dung treatment on ecosystem metabolism - gross primary

production (GPP, mg O2 m-2 day-1), ecosystem respiration (ER, ER, mg O2 m-2 day-1),

GPP:ER and net ecosystem production (NEP, mg O2 m-2 day-1), which displayed

nonlinear responses to dung treatments.

Chapter 3: Dissolved organic matter signatures in urban surface waters: spatio-

temporal patterns

Supplement of Chapter 3 Dissolved organic matter signatures in urban surface waters:

spatio-temporal patterns

Table S1: Coordinates, land cover, origin and special features. Longitude is given in

decimal degrees East and latitude in decimal degrees North.

Table S2: Physico-chemical characteristics (mean ± SD and % variance explained) of

four contrasting types of water bodies in the city of Berlin. Means and standard deviations

were computed across all seasons and sites. The percentages of variance explained (%

List of Tables

163

Var) refer to the effect of season within each water body type, calculated by type-II

ANOVA (aka variance component analysis), with season treated as a random factor. F-

values refer to results of repeated-measures ANOVAs testing for differences among

water body types (*** p<0.001, * p<0.05, ns = not significant).

Table S3: Description of absorbance and fluorescence indices.

Table S4: Designation, excitation (Ex) and emission (Em) wavelengths of PARAFAC

components, and the number of studies with matching components reported in

OpenFluor (checked on the 28th March 2022) (Murphy et al., 2014).

Table S5: Variables of absorbance and fluorescence analyses (mean ± SD and % variance

explained) in contrasting types of urban surface waters. Means and standard deviations

were computed across all seasons and sites. The percentages of variance explained (%

Var) refer to the effect of season within each water body type, calculated by a type-II

ANOVA (aka variance component analysis), with season treated as a random factor. F-

values refer to results of repeated-measures ANOVA testing for differences among water

body types (***p<0.001, **p<0.01, *p<0.05, ns = not significant). Abbreviations

explained in Table B2.

Table S6: PARAFAC components (mean ± SD and % variance explained) in contrasting

types of urban surface waters. Means and standard deviations were computed across all

seasons and sites. The percentages of variance explained (% Var) refer to the effect of

season within each water body type, calculated by a type-II ANOVA (aka variance

component analysis), with season treated as a random factor. F-values refer to results of

repeated-measures ANOVA testing for differences among water body types (***p<0.001,

**p<0.01, *p<0.05, ns = not significant).

Table S7: Results of size exclusion chromatography (mean ± SD and % variance

explained) of samples from contrasting types of urban surface waters. Means and

standard deviations were computed across all seasons and sites. The percentages of

variance explained (% Var) refer to the effect of season within each water body type,

calculated by a type-II ANOVA (aka variance component analysis), with season treated

as a random factor. F-values refer to results of repeated-measures ANOVA testing for

differences among water body types (***p<0.001, **p<0.01, *p<0.05, ns = not

significant). HS, humic-like substances; HMWS, high-molecular weight non-humic

substances; and LMWS, low-molecular weight substances.

Table S8: Trace organic compounds (TrOCs) analyzed in samples collected in urban

surface waters. LLoQ = Limit of Quantification. Frequency refers to the number of

occasions where concentrations exceeded the LLoQ.

Table S9: Mean concentrations and standard deviations of Trace Organic Compound

(TrOC) per water body type. See Table S8 for full names. BZF, SMX and VLX were

always below the limit of quantification (LLoQ) and are hence omitted from the table.

Chapter 4 Carbon dioxide emissions across an urban aquatic network

Table 1 Comparison of CO2 flux measurement methodologies using two types of floating

chambers deployed at 32 sites in the city of Berlin.

Table 2 Mean and SD of CO2 fluxes from surface waters in the city of Berlin, Germany,

made by instantaneous and continuous measurements. Seasonal SD, day-to-day SD and

diel SD. The percentages of variance explained (% Var) refer to the effect of season, day

to day and time of the day within each water body type, calculated by type-II ANOVA

(variance component analysis), with season, day and time of the day treated as random

101

List of Tables

164

factors. The percentage of variability explained by k and ΔCO2. NA* Data insufficient

for calculations

Table 3 Estimates of annual CO2 emissions (± 95% CL) from four types of water bodies

in the city of Berlin based on fluxes measured at four occasions either during 15-min

deployments of flux chambers (instantaneous measurements) or during continuous one-

week measurements with equilibration chambers.

103

Table 4 Description and interpretation of PCA axes identified in the final random forest

(RF) models as potential drivers of CO2 fluxes. DOM: dissolved organic matter, CHEM:

physico-chemical variables

105

Supplement of Chapter 4 Carbon dioxide emissions across an urban aquatic network

111

Table S1 Designation, excitation (Ex) and emission (Em) wavelengths of PARAFAC

components, and the number of studies with matching components reported in

OpenFluor (checked on the 28th March 2022) (Murphy et al. 2014). From Romero

González-Quijano 2022

112

Table S2 Description of absorbance and fluorescence indices (from Romero González-

Quijano 2022)

112

Table S3 Physico-chemical variables across water body types.

113

Table S4 Sites coordinates and land uses. Longitude is given in decimal degrees East

and latitude in decimal degrees North. Land cover was calculated with QGIS- layer, for

a 50 m buffer around water bodies using the software QGIS (QGIS Development Team,

2017) and data provided by the Senate Administration for Environment, Transport and

Climate Protection of Berlin.

114

Table S5 List of trace organic compounds analyzed

115

Table S6 Summary of all the k600 calculated for this study.

116

Table S7 Random Forest (RF) models. Variables used as predictors were PC axes with

Eigenvalues >1 from three separate PCAs based on DOM characteristics; water depth,

land cover and physico-chemical variables (CHEM); and Trace Organic Compounds.

Only variables with a relative influence in the RF >10% are shown.

117

Table S8 Literature values of CO2 fluxes for the comparison with our fluxes

118

Chapter 5 Methane emissions from contrasting urban freshwaters: Rates, drivers,

and a whole‐city footprint

123

Table 1 Annual methane (CH4) emission footprint of the metropolitan area of Berlin,

Germany, separated by type of water body (mean ± SD)

131

Table 2 Methane (CH4) emission flues from urban freshwaters.

135

Supplement of Chapter 5 Methane emissions from contrasting urban freshwaters: Rates,

drivers, and a whole‐city footprint

138

Table S1. Characteristics of 32 freshwater sites studied in the metropolitan area of

Berlin, Germany. Land use refers to a strip extending 50 m away from the shore. n.a.: no

data available. *Areas for rivers and streams calculated for arbitrary stretches 1 km long

to enable intuitive comparisons of relative size with lakes and ponds. †Mean value of four

sampling campaigns. WWTP = wastewater treatment plant

141

List of Tables

165

Table S2. Summary and description of DOM optical properties, modified from Catalán

et al.(2013) and Fasching et al. (2014).

143

Table S3. Average annual CH4 emissions (total, diffusive and ebullition flux) to the

atmosphere measured in situ with a chamber connected to an ultraportable gas analyser;

diffusive flux calculated from CH4, and ebullition flux calculated from data collected with

inverted funnels (IF) placed on the sediment surface. n.a.: no data available

144

Table S4. Physical and chemical surface water variables of 32 freshwater bodies in the

city of Berlin, Germany, averaged per water body type and season. Values represent

means ± standard deviations. DO = dissolved oxygen, TP = total phosphorus, DOC =

dissolved organic carbon

147

References

166

References

Aben, R. C. H. and others 2017. Cross continental increase in methane ebullition under

climate change. Nature Communications 8: 1682.

Åberg, J., A.-K. Bergström, G. Algesten, K. Söderback, and M. Jansson. 2004. A comparison

of the carbon balances of a natural lake (L. Örträsket) and a hydroelectric reservoir (L.

Skinnmuddselet) in northern Sweden. Water Research 38: 531-538.

Abnizova, A., J. Siemens, M. Langer, and J. Boike. 2012. Small ponds with major impact: The

relevance of ponds and lakes in permafrost landscapes to carbon dioxide emissions.

Global Biogeochemical Cycles 26.

Abril, G. and others 2015. Technical Note: Large overestimation of pCO2 calculated from

pH and alkalinity in acidic, organic-rich freshwaters. Biogeosciences 12: 67-78.

Aitkenhead-Peterson, J. A., W. H. McDowell, and J. C. Neff. 2003. 2 - Sources, Production,

and Regulation of Allochthonous Dissolved Organic Matter Inputs to Surface Waters,

p. 25-70. In S. E. G. Findlay and R. L. Sinsabaugh [eds.], Aquatic Ecosystems.

Academic Press.

Alberts, J. M., J. J. Beaulieu, and I. Buffam. 2017. Watershed Land Use and Seasonal Variation

Constrain the Influence of Riparian Canopy Cover on Stream Ecosystem Metabolism.

Ecosystems 20: 553-567.

Alin, S. R. and others 2011. Physical controls on carbon dioxide transfer velocity and flux in

low‐gradient river systems and implications for regional carbon budgets. Journal of

Geophysical Research: Biogeosciences 116.

Allesson, L., B. Koehler, J.-E. Thrane, T. Andersen, and D. O. Hessen. 2021. The role of

photomineralization for CO2 emissions in boreal lakes along a gradient of dissolved

organic matter. Limnology and Oceanography 66: 158-170.

Alshboul, Z., J. Encinas-Fernández, H. Hofmann, and A. Lorke. 2016a. Export of Dissolved

Methane and Carbon Dioxide with Effluents from Municipal Wastewater Treatment

Plants. Environmental Science & Technology 50: 5555-5563.

Alshboul, Z., J. Encinas-Fernández, H. Hofmann, and A. Lorke. 2016b. Export of Dissolved

Methane and Carbon Dioxide with Effluents from Municipal Wastewater Treatment

Plants. Environ Sci Technol 50: 5555-5563.

Amaral, V., D. Graeber, D. Calliari, and C. Alonso. 2016. Strong linkages between DOM

optical properties and main clades of aquatic bacteria. Limnology and Oceanography

61: 906-918.

Amaral, V., T. Ortega, C. Romera-Castillo, and J. Forja. 2021. Linkages between greenhouse

gases (CO2, CH4, and N2O) and dissolved organic matter composition in a shallow

estuary. Science of The Total Environment 788: 147863.

References

167

APHA. 1998. Standard Methods for the Examination of Water and Wastewater. APHA-

AWWA-WEF.

Aristi, I. and others 2015. Mixed effects of effluents from a wastewater treatment plant on

river ecosystem metabolism: subsidy or stress? Freshwater Biology 60: 1398-1410.

Attermeyer, K. and others 2021. Carbon dioxide fluxes increase from day to night across

European streams. Communications Earth & Environment 2: 118.

Aubinet, M., T. Vesala, and D. Papale. 2012. Eddy covariance: A practical guide to

measurements and data analysis. Springer.

Audet, J., M. V. Carstensen, C. C. Hoffmann, L. Lavaux, K. Thiemer, and T. A. Davidson.

2020. Greenhouse gas emissions from urban ponds in Denmark. Inland Waters 10:

373-385.

Aufdenkampe, A. K. and others 2011. Riverine coupling of biogeochemical cycles between

land, oceans, and atmosphere. Frontiers in Ecology and the Environment 9: 53-60.

Bar-On, Y. M., R. Phillips, and R. Milo. 2018. The biomass distribution on Earth. Proc Natl

Acad Sci U S A 115: 6506-6511.

Bargrizan, S., T. K. Biswas, K. D. Joehnk, and L. M. Mosley. 2022. Sustained high CO2

concentrations and fluxes from Australia’s largest river system. Marine and

Freshwater Research 73: 540-551.

Bastviken, D., J. Cole, M. Pace, and L. Tranvik. 2004. Methane emissions from lakes:

Dependence of lake characteristics, two regional assessments, and a global estimate.

Global Biogeochemical Cycles 18.

Bastviken, D., J. J. Cole, M. L. Pace, and M. C. Van de Bogert. 2008. Fates of methane from

different lake habitats: Connecting whole-lake budgets and CH4 emissions. Journal of

Geophysical Research: Biogeosciences 113.

Bastviken, D., I. Sundgren, S. Natchimuthu, H. Reyier, and M. Gålfalk. 2015. Technical Note:

Cost-efficient approaches to measure carbon dioxide (CO2) fluxes and concentrations

in terrestrial and aquatic environments using mini loggers. Biogeosciences 12: 3849-

3859.

Bastviken, D., L. J. Tranvik, J. A. Downing, P. M. Crill, and A. Enrich-Prast. 2011.

Freshwater Methane Emissions Offset the Continental Carbon Sink. Science 331: 50-

50.

Bastviken, D. and others 2023. The importance of plants for methane emission at the

ecosystem scale. Aquatic Botany 184: 103596.

Battin, T. J. and others 2023. River ecosystem metabolism and carbon biogeochemistry in a

changing world. Nature 613: 449-459.

Battin, T. J., S. Luyssaert, L. A. Kaplan, A. K. Aufdenkampe, A. Richter, and L. J. Tranvik.

2009. The boundless carbon cycle. Nature Geosci 2: 598-600.

References

168

Baume, O., and J. Marcinek. 1993 Das Berliner Gewässernetz und seine Nährstoffbelastung.

Wasserwirtschaft Wassertechnik, 6: 40-45 (In German).

Belsky, A. J., A. Matzke, and S. Uselman. 1999. Survey of livestock influences on stream and

riparian ecosystems in the western United States. Journal of Soil and Water

Conservation 54: 419-431.

Benner, R. 2003. 5 - Molecular Indicators of the Bioavailability of Dissolved Organic Matter,

p. 121-137. In S. E. G. Findlay and R. L. Sinsabaugh [eds.], Aquatic Ecosystems.

Academic Press.

Benson, B. B., and D. Krause Jr. 1984. The concentration and isotopic fractionation of oxygen

dissolved in freshwater and seawater in equilibrium with the atmosphere 1. Limnol

Oceanogr 29: 620-632.

Bernhardt, E. S. and others 2022. Light and flow regimes regulate the metabolism of rivers.

Proceedings of the National Academy of Sciences 119: e2121976119.

Bernhardt, J. R., J. M. Sunday, P. L. Thompson, and M. I. O'Connor. 2018. Nonlinear

averaging of thermal experience predicts population growth rates in a thermally

variable environment. Proceedings of the Royal Society B: Biological Sciences 285:

20181076.

Bernot, M. J. and others 2010. Inter-regional comparison of land-use effects on stream

metabolism. Freshwater Biology 55: 1874-1890.

Birch, S., and J. McCaskie. 1999. Shallow urban lakes: a challenge for lake management, p.

365-377. In D. M. Harper, B. Brierley, A. J. D. Ferguson and G. Phillips [eds.], The

Ecological Bases for Lake and Reservoir Management. Springer Netherlands.

Blodau, C. and others 2008. A snapshot of CO2 and CH4 evolution in a thermokarst pond

near Igarka, northern Siberia. Journal of Geophysical Research: Biogeosciences 113.

Bodmer, P., M. Heinz, M. Pusch, G. Singer, and K. Premke. 2016. Carbon dynamics and their

link to dissolved organic matter quality across contrasting stream ecosystems. Science

of The Total Environment 553: 574-586.

Bond, T., D. Sear, and T. Sykes. 2014. Estimating the contribution of in-stream cattle faeces

deposits to nutrient loading in an English Chalk stream. Agricultural Water

Management 131: 156-162.

Bond, T. A., D. Sear, and M. Edwards. 2012. Temperature-driven river utilisation and

preferential defecation by cattle in an English chalk stream. Livestock Science 146: 59-

66.

Borges, A. V. and others 2015. Globally significant greenhouse-gas emissions from African

inland waters. Nature Geoscience 8: 637-+.

Borrel, G. and others 2011. Production and consumption of methane in freshwater lake

ecosystems. Research in Microbiology 162: 832-847.

References

169

Bott, T. L. 1996. Primary productivity and community respiration. Methods in stream

ecology: 533-556.

Boyer, J. N., C. R. Kelble, P. B. Ortner, and D. T. Rudnick. 2009. Phytoplankton bloom status:

Chlorophyll a biomass as an indicator of water quality condition in the southern

estuaries of Florida, USA. Ecological Indicators 9: S56-S67.

Bro, R. 1997. PARAFAC. Tutorial and applications. Chemometrics and Intelligent

Laboratory Systems 38: 149-171.

Brooks, J. R., J. J. Gibson, S. J. Birks, M. H. Weber, K. D. Rodecap, and J. L. Stoddard. 2014.

Stable isotope estimates of evaporation : inflow and water residence time for lakes

across the United States as a tool for national lake water quality assessments.

Limnology and Oceanography 59: 2150-2165.

Buser, H.-R., T. Poiger, and M. D. Müller. 1999. Occurrence and Environmental Behavior of

the Chiral Pharmaceutical Drug Ibuprofen in Surface Waters and in Wastewater.

Environmental Science & Technology 33: 2529-2535.

Butman, D., and P. A. Raymond. 2011. Significant efflux of carbon dioxide from streams and

rivers in the United States. Nature Geoscience 4: 839-842.

Butman, D., S. Stackpoole, E. Stets, C. P. McDonald, D. W. Clow, and R. G. Striegl. 2016.

Aquatic carbon cycling in the conterminous United States and implications for

terrestrial carbon accounting. Proceedings of the National Academy of Sciences 113:

58-63.

Carignan, R., D. Planas, and C. Vis. 2000. Planktonic production and respiration in

oligotrophic Shield lakes. Limnology and Oceanography 45: 189-199.

Carpenter, S. R., N. F. Caraco, D. L. Correll, R. W. Howarth, A. N. Sharpley, and V. H. Smith.

1998. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecological

Applications 8: 559-568.

Carpenter, S. R. and others 2005. Ecosystem subsidies: terrestrial support of aquatic food

webs from 13c addition to contrasting lakes. Ecology 86: 2737-2750.

Casper, P., S. C. Maberly, G. H. Hall, and B. J. Finlay. 2000. Fluxes of methane and carbon

dioxide from a small productive lake to the atmosphere. Biogeochemistry 49: 1-19.

Catalán, N., A. M. Kellerman, H. Peter, F. Carmona, and L. J. Tranvik. 2015. Absence of a

priming effect on dissolved organic carbon degradation in lake water. Limnology and

Oceanography 60: 159-168.

Catalán, N., B. Obrador, C. Alomar, and J. L. Pretus. 2013. Seasonality and landscape factors

drive dissolved organic matter properties in Mediterranean ephemeral washes.

Biogeochemistry 112: 261-274.

Cerling, T. E., J. M. Harris, and B. H. Passey. 2003. Diets of East African Bovidae based on

stable isotope analysis. Journal of Mammalogy 84: 456-470.

References

170

Chen, H., Z.-l. Liao, X.-y. Gu, J.-q. Xie, H.-z. Li, and J. Zhang. 2017. Anthropogenic

Influences of Paved Runoff and Sanitary Sewage on the Dissolved Organic Matter

Quality of Wet Weather Overflows: An Excitation–Emission Matrix Parallel Factor

Analysis Assessment. Environmental Science & Technology 51: 1157-1167.

Chen, S. and others 2021. Agricultural land use changes stream dissolved organic matter via

altering soil inputs to streams. Science of The Total Environment 796: 148968.

Chen, Y., N. Senesi, and M. Schnitzer. 1977. Information provided on humic substances by

E4/E6 ratios. Soil Sci. Soc. Am. J. 41, 352–358.

Chester, R., and T. D. Jickells. 2012. Marine geochemistry. John Wiley & Sons.

Clauss, M. and others 2004. Intake, ingesta retention, particle size distribution and

digestibility in the hippopotamidae. Comparative Biochemistry and Physiology Part

A: Molecular & Integrative Physiology 139: 449-459.

Clauss, M., A. Schwarm, S. Ortmann, W. J. Streich, and J. Hummel. 2007. A case of non-

scaling in mammalian physiology? Body size, digestive capacity, food intake, and

ingesta passage in mammalian herbivores. Comparative Biochemistry and Physiology

Part A: Molecular & Integrative Physiology 148: 249-265.

Clow, D. W., R. G. Striegl, and M. M. Dornblaser. 2021. Spatiotemporal Dynamics of CO2

Gas Exchange From Headwater Mountain Streams. Journal of Geophysical Research:

Biogeosciences 126: e2021JG006509.

Coble, P. G. 1996. Characterization of marine and terrestrial DOM in seawater using

excitation-emission matrix spectroscopy. Marine Chemistry 51: 325-346.

Coble, P. G. 2007. Marine optical biogeochemistry: the chemistry of ocean color. Chemical

reviews 107: 402-418.

Coble, P. G., C. E. Del Castillo, and B. Avril. 1998. Distribution and optical properties of

CDOM in the Arabian Sea during the 1995 Southwest Monsoon. Deep Sea Research

Part II: Topical Studies in Oceanography 45: 2195-2223.

Coble, P. G., S. A. Green, N. V. Blough, and R. B. Gagosian. 1990. Characterization of

dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature 348:

432.

Codron, D., J. A. LEE‐THORP, M. Sponheimer, J. Codron, D. De Ruiter, and J. S. Brink.

2007. Significance of diet type and diet quality for ecological diversity of African

ungulates. Journal of Animal Ecology 76: 526-537.

Cole, J. J., D. L. Bade, D. Bastviken, M. L. Pace, and M. Van de Bogert. 2010. Multiple

approaches to estimating air-water gas exchange in small lakes. Limnol. Oceanogr.

Meth 8, 285–293,.

Cole, J. J., and N. F. Caraco. 1998. Atmospheric exchange of carbon dioxide in a low-wind

oligotrophic lake measured by the addition of SF6. Limnology and Oceanography 43:

647-656.

References

171

Cole, J. J., M. L. Pace, S. R. Carpenter, and J. F. Kitchell. 2000. Persistence of net heterotrophy

in lakes during nutrient addition and food web manipulations. Limnology and

Oceanography 45: 1718-1730.

Cole, J. J. and others 2007. Plumbing the Global Carbon Cycle: Integrating Inland Waters

into the Terrestrial Carbon Budget. Ecosystems 10: 172-185.

Cory, R. M., K. H. Harrold, B. T. Neilson, and G. W. Kling. 2015. Controls on dissolved

organic matter (DOM) degradation in a headwater stream: the influence of

photochemical and hydrological conditions in determining light-limitation or

substrate-limitation of photo-degradation. Biogeosciences 12: 6669-6685.

Cory, R. M., and L. A. Kaplan. 2012. Biological lability of streamwater fluorescent dissolved

organic matter. Limnol. Oceanogr. 57: 1347-1360.

Cory, R. M., and D. M. McKnight. 2005. Fluorescence spectroscopy reveals ubiquitous

presence of oxidized and reduced quinones in dissolved organic matter. Environmental

science & technology 39: 8142-8149.

Cory, R. M., M. P. Miller, D. M. McKnight, J. J. Guerard, and P. L. Miller. 2010. Effect of

instrument-specific response on the analysis of fulvic acid fluorescence spectra.

Limnology and Oceanography: Methods 8: 67-78.

Coscia, I., and M. J. Kaiser. 2022. The impact of the human activities on aquatic ecosystems.

Journal of Fish Biology 101: 331-332.

Cotton, J. B., and I. R. Scholes. 1967. Benzotriazole and Related Compounds as Corrosion

Inhibitors For Copper. British Corrosion Journal 2: 1-5.

Council, N. R. 2009. Urban Stormwater Management in the United States. Washington, DC:

The National Academies Press. https://doi.org/10.17226/12465.

Craigie, I. D. and others 2010. Large mammal population declines in Africa’s protected areas.

Biological Conservation 143: 2221-2228.

Crawford, J. T., R. G. Striegl, K. P. Wickland, M. M. Dornblaser, and E. H. Stanley. 2013.

Emissions of carbon dioxide and methane from a headwater stream network of interior

Alaska. Journal of Geophysical Research: Biogeosciences 118: 482-494.

Creed, I. F. and others 2015. The river as a chemostat: fresh perspectives on dissolved organic

matter flowing down the river continuum. Canadian Journal of Fisheries and Aquatic

Sciences 72: 1272-1285.

D'Arcy, B. J., T. Rosenqvist, G. Mitchell, R. Kellagher, and S. Billett. 2007. Restoration

challenges for urban rivers. Water Science and Technology 55: 1-7.

Daelman, M. R. J., E. M. van Voorthuizen, U. G. J. M. van Dongen, E. I. P. Volcke, and M.

C. M. van Loosdrecht. 2012. Methane emission during municipal wastewater

treatment. Water Research 46: 3657-3670.

Davidson, E. A., and I. A. Janssens. 2006. Temperature sensitivity of soil carbon

decomposition and feedbacks to climate change. Nature 440: 165-173.

References

172

Dawson, J., D. Pillay, P. J. Roberts, and R. Perissinotto. 2016. Declines in benthic

macroinvertebrate community metrics and microphytobenthic biomass in an estuarine

lake following enrichment by hippo dung. Scientific Reports 6: 37359.

de Fátima F. L. Rasera, M., A. V. Krusche, J. E. Richey, M. V. R. Ballester, and R. L. Victória.

2013. Spatial and temporal variability of pCO2 and CO2 efflux in seven Amazonian

Rivers. Biogeochemistry 116: 241-259.

Deemer, B. R. and others 2016. Greenhouse Gas Emissions from Reservoir Water Surfaces:

A New Global Synthesis. BioScience 66: 949-964.

del Campo, R., R. Gómez, and G. Singer. 2019. Dry phase conditions prime wet-phase

dissolved organic matter dynamics in intermittent rivers. Limnology and

Oceanography 0.

DelSontro, T., J. J. Beaulieu, and J. A. Downing. 2018. Greenhouse gas emissions from lakes

and impoundments: Upscaling in the face of global change. Limnology and

Oceanography Letters 3: 64-75.

DelSontro, T., D. F. McGinnis, S. Sobek, I. Ostrovsky, and B. Wehrli. 2010. Extreme

Methane Emissions from a Swiss Hydropower Reservoir: Contribution from Bubbling

Sediments. Environmental Science & Technology 44: 2419-2425.

Demars, B. O., J. Thompson, and J. R. Manson. 2015. Stream metabolism and the open diel

oxygen method: Principles, practice, and perspectives. Limnology and Oceanography:

Methods 13: 356-374.

Dijkslag, M., M. Elling-Staats, Y. Yen, L. Marchal, and R. Kwakkel. 2019. The effects of

coarse and wet feeding on performance parameters, gastrointestinal tract and tibia

traits, and digesta phytase activity in egg-type pullets, either fed a low or moderate

phosphorus diet. Poultry science 98: 4729-4744.

Drake, T. W., P. A. Raymond, and R. G. M. Spencer. 2018. Terrestrial carbon inputs to

inland waters: A current synthesis of estimates and uncertainty. Limnology and

Oceanography Letters 3: 132-142.

du Toit, J. T. 2003. Large herbivores and savannah heterogeneity. Pages 292-309 in: J.T. du

Toit, K.H. Rogers and H.C. Biggs (eds.) The Kruger Experience: Ecology and

Management of Savannah Heterogeneity. Island Press, Washington, DC, USA.

Dubois, K. D., D. Lee, and J. Veizer. 2010. Isotopic constraints on alkalinity, dissolved

organic carbon, and atmospheric carbon dioxide fluxes in the Mississippi River.

Journal of Geophysical Research: Biogeosciences 115.

Duc, N. T., S. Silverstein, L. Lundmark, H. Reyier, P. Crill, and D. Bastviken. 2013.

Automated Flux Chamber for Investigating Gas Flux at Water–Air Interfaces.

Environmental Science & Technology 47: 968-975.

References

173

Duchemin, E., M. Lucotte, and R. Canuel. 1999. Comparison of static chamber and thin

boundary layer equation methods for measuring greenhouse gas emissions from large

water bodies §. Environmental Science & Technology 33: 350-357.

Dutton, C. L., A. L. Subalusky, S. C. Anisfeld, L. Njoroge, E. J. Rosi, and D. M. Post. 2018a.

The influence of a semi-arid sub-catchment on suspended sediments in the Mara River,

Kenya. PLOS ONE 13: e0192828.

Dutton, C. L., A. L. Subalusky, S. K. Hamilton, E. J. Rosi, and D. M. Post. 2018b. Organic

matter loading by hippopotami causes subsidy overload resulting in downstream

hypoxia and fish kills. Nature Communications 9: 1951.

Edwards, P. 1991. Seasonal variation in the dung of African grazing mammals, and its

consequences for coprophagous insects. Functional Ecology: 617-628.

Elliott, R., and K. Fokkema. 1961. Herbage consumption studies on beef cattle. 1.

Intakestudies on Afrikander and Mashonacows onveldgrazing-1958/9. Rhodesia

Agricultural Journal 58: 49-57.

Elmore, H. 1961. Effects of water temperature on stream sanitation. J. Sanitary Eng. 87: 59-

71.

Erkkilä, K. M. and others 2018. Methane and carbon dioxide fluxes over a lake: comparison

between eddy covariance, floating chambers and boundary layer method.

Biogeosciences 15: 429-445.

Falkowski, P. G., and J. A. Raven. 1997. Aquatic photosynthesis. Blackwell Science, Malden,

UK.

Fasching, C., A. J. Ulseth, J. Schelker, G. Steniczka, and T. J. Battin. 2016. Hydrology

controls dissolved organic matter export and composition in an Alpine stream and its

hyporheic zone. Limnology and Oceanography 61: 558-571.

Feld, C. K., P. Segurado, and C. Gutiérrez-Cánovas. 2016. Analysing the impact of multiple

stressors in aquatic biomonitoring data: A 'cookbook' with applications in R. Sci Total

Environ 573: 1320-1339.

Fellman, J. B., E. Hood, and R. G. Spencer. 2010. Fluorescence spectroscopy opens new

windows into dissolved organic matter dynamics in freshwater ecosystems: A review.

Limnol Oceanogr 55: 2452-2462.

Fellows, C. S. and others 2006. Benthic Metabolism as an Indicator of Stream Ecosystem

Health. Hydrobiologia 572: 71-87.

Field, C. R. 1970. A Study of the Feeding Habits of the Hippopotamus (Hippopotamus

Amphibius Linn.) in the Queen Elizabeth National Park, Uganda, With Some

Management Implications. Zoologica Africana 5: 71-86.

Findlay, S. E. G. 2021. Chapter 4 - Organic Matter Decomposition, p. 81-102. In K. C.

Weathers, D. L. Strayer and G. E. Likens [eds.], Fundamentals of Ecosystem Science

(Second Edition). Academic Press.

References

174

Finlay, K., R. J. Vogt, G. L. Simpson, and P. R. Leavitt. 2019. Seasonality of pCO2 in a hard-

water lake of the northern Great Plains: The legacy effects of climate and limnological

conditions over 36 years. Limnology and Oceanography 64: S118-S129.

Fonvielle, J. A. and others 2021. Exploring the Suitability of Ecosystem Metabolomes to

Assess Imprints of Brownification and Nutrient Enrichment on Lakes. Journal of

Geophysical Research: Biogeosciences 126: e2020JG005903.

Friberg, N. and others 2011. Biomonitoring of Human Impacts in Freshwater Ecosystems:

The Good, the Bad and the Ugly, p. 1-68. In G. Woodward [ed.], Advances in

Ecological Research. Academic Press.

Fritz, J., J. Hummel, E. Kienzle, C. Arnold, C. Nunn, and M. Clauss. 2009. Comparative

chewing efficiency in mammalian herbivores. Oikos 118: 1623-1632.

Fromme, H., A. Köhler, R. Krause, and D. Führling. 2000. Occurrence of cyanobacterial

toxins—microcystins and anatoxin-a—in Berlin water bodies with implications to

human health and regulations. Environmental Toxicology 15: 120-130.

Fuss, T., B. Behounek, A. J. Ulseth, and G. A. Singer. 2017. Land use controls stream

ecosystem metabolism by shifting dissolved organic matter and nutrient regimes.

Freshwater Biol 62: 582-599.

Gallo, E. L., K. A. Lohse, C. M. Ferlin, T. Meixner, and P. D. Brooks. 2014. Physical and

biological controls on trace gas fluxes in semi-arid urban ephemeral waterways.

Biogeochemistry 121: 189-207.

Garg, S. K., and A. Bhatnagar. 1999. Effect of different doses of organic fertilizer (cow dung)

on pond productivity and fish biomass in stillwater ponds. Journal of Applied

Ichthyology 15: 10-18.

Garnier, J. and others 2013. Budget of methane emissions from soils, livestock and the river

network at the regional scale of the Seine basin (France). Biogeochemistry 116: 199-

214.

Gerardo-Nieto, O., M. S. Astorga-España, A. Mansilla, and F. Thalasso. 2017. Initial report

on methane and carbon dioxide emission dynamics from sub-Antarctic freshwater

ecosystems: A seasonal study of a lake and a reservoir. Science of The Total

Environment 593-594: 144-154.

Gessner, M. O., E. Chauvet, and M. Dobson. 1999. A Perspective on Leaf Litter Breakdown

in Streams. Oikos 85: 377-384.

Gessner, M. O. and others 2014. Urban water interfaces. Journal of Hydrology 514: 226-232.

Giesler, R. and others 2014. Catchment-scale dissolved carbon concentrations and export

estimates across six subarctic streams in northern Sweden. Biogeosciences 11: 525-

537.

Gómez-Gener, L. and others 2021. Global carbon dioxide efflux from rivers enhanced by

high nocturnal emissions. Nature Geoscience 14: 289-294.

References

175

Gompertz, B. 1825. XXIV. On the nature of the function expressive of the law of human

mortality, and on a new mode of determining the value of life contingencies. In a letter

to Francis Baily, Esq. F. R. S. &c. Philosophical Transactions of the Royal Society

of London 115: 513-583.

Gonzalez-Valencia, R., A. Sepulveda-Jauregui, K. Martinez-Cruz, J. Hoyos-Santillan, L.

Dendooven, and F. Thalasso. 2014. Methane emissions from Mexican freshwater

bodies: correlations with water pollution. Hydrobiologia 721: 9-22.

Graeber, D., J. Gelbrecht, M. T. Pusch, C. Anlanger, and D. von Schiller. 2012. Agriculture

has changed the amount and composition of dissolved organic matter in Central

European headwater streams. Science of The Total Environment 438: 435-446.

Grasset, C., G. Abril, L. Guillard, C. Delolme, and G. Bornette. 2016. Carbon emission along

a eutrophication gradient in temperate riverine wetlands: effect of primary

productivity and plant community composition. Freshwater Biology 61: 1405-1420.

Green, S. A., and N. V. Blough. 1994. Optical absorption and fluorescence properties of

chromophoric dissolved organic matter in natural waters. Limnol Oceanogr 39: 1903-

1916.

Griffith, D. R., and P. A. Raymond. 2011. Multiple-source heterotrophy fueled by aged

organic carbon in an urbanized estuary. Marine Chemistry 124: 14-22.

Griffiths, N. A. and others 2013. Agricultural land use alters the seasonality and magnitude

of stream metabolism. Limnology and Oceanography 58: 1513-1529.

Grimm, N., J. M. Grove, S. A. Pickett, and C. Redman. 2008a. Integrated Approaches to

Long-Term Studies of Urban Ecological Systems, p. 123-141. In J. Marzluff et al.

[eds.], Urban Ecology. Springer US.

Grimm, N. B. and others 2008b. Global Change and the Ecology of Cities. Science 319: 756-

760.

Grimm, N. B. and others 2008c. The changing landscape: ecosystem responses to

urbanization and pollution across climatic and societal gradients. Frontiers in Ecology

and the Environment 6: 264-272.

Grimm, N. B., R. W. Sheibley, C. L. Crenshaw, C. N. Dahm, W. J. Roach, and L. H. Zeglin.

2005. N retention and transformation in urban streams. Journal of the North American

Benthological Society 24: 626-642.

Grinham, A. and others 2018a. The importance of small artificial water bodies as sources of

methane emissions in Queensland, Australia. Hydrol. Earth Syst. Sci. 22: 5281-5298.

Grinham, A., M. Dunbabin, and S. Albert. 2018b. Importance of sediment organic matter to

methane ebullition in a sub-tropical freshwater reservoir. Science of The Total

Environment 621: 1199-1207.

Gu, C., S. Waldron, and A. M. Bass. 2021. Carbon dioxide, methane, and dissolved carbon

dynamics in an urbanized river system. Hydrological Processes 35: e14360.

References

176

Gu, S., S. Li, and I. R. Santos. 2022. Anthropogenic land use substantially increases riverine

CO2 emissions. Journal of Environmental Sciences 118: 158-170.

Guérin, F. and others 2007. Gas transfer velocities of CO2 and CH4 in a tropical reservoir

and its river downstream. Journal of Marine Systems 66: 161-172.

Hamilton, J. D., C. A. Kelly, J. W. M. Rudd, R. H. Hesslein, and N. T. Roulet. 1994. Flux to

the atmosphre of CH4 and CO2 from wetland ponds on the Hudson-bay Lowlands

(HBLS) ). JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES 99: 1495-

1510.

Hansen, A. M., T. E. C. Kraus, B. A. Pellerin, J. A. Fleck, B. D. Downing, and B. A.

Bergamaschi. 2016. Optical properties of dissolved organic matter (DOM): Effects of

biological and photolytic degradation. Limnology and Oceanography 61: 1015-1032.

Hass, U., U. Duennbier, and G. Massmann. 2012. Occurrence and distribution of

psychoactive compounds and their metabolites in the urban water cycle of Berlin

(Germany). Water Research 46: 6013-6022.

Hassall, C. 2014. The ecology and biodiversity of urban ponds. WIREs Water 1: 187-206.

Hatt, B. E., T. D. Fletcher, C. J. Walsh, and S. L. Taylor. 2004. The Influence of Urban

Density and Drainage Infrastructure on the Concentrations and Loads of Pollutants

in Small Streams. Environmental Management 34: 112-124.

Hayward, M. W., and M. D. Hayward. 2012. Waterhole use by African Fauna. South African

Journal of Wildlife Research 42: 117-127, 111.

He, B., J. He, J. Wang, J. Li, and F. Wang. 2018. Characteristics of GHG flux from water-air

interface along a reclaimed water intake area of the Chaobai River in Shunyi, Beijing.

Atmospheric Environment 172: 102-108.

Heberer, T. 2002. Tracking persistent pharmaceutical residues from municipal sewage to

drinking water. Journal of Hydrology 266: 175-189.

Heberer, T., D. Feldmann, K. Reddersen, H.-J. Altmann, and T. Zimmermann. 2002.

Production of Drinking Water from Highly Contaminated Surface Waters: Removal

of Organic, Inorganic, and Microbial Contaminants Applying Mobile Membrane

Filtration Units. Acta hydrochimica et hydrobiologica 30: 24-33.

Heberer, T., G. Massmann, B. Fanck, T. Taute, and U. Dünnbier. 2008. Behaviour and redox

sensitivity of antimicrobial residues during bank filtration. Chemosphere 73: 451-460.

Helms, J. R., A. Stubbins, J. D. Ritchie, E. C. Minor, D. J. Kieber, and K. Mopper. 2008.

Absorption spectral slopes and slope ratios as indicators of molecular weight, source,

and photobleaching of chromophoric dissolved organic matter. Limnol Oceanogr 53:

955-969.

Hempson, G. P., S. Archibald, and W. J. Bond. 2017. The consequences of replacing wildlife

with livestock in Africa. Scientific Reports 7: 17196.

References

177

Henzler, A. F., J. Greskowiak, and G. Massmann. 2014. Modeling the fate of organic

micropollutants during river bank filtration (Berlin, Germany). Journal of

Contaminant Hydrology 156: 78-92.

Herrero Ortega, S., C. Romero González-Quijano, P. Casper, G. A. Singer, and M. O.

Gessner. 2019. Methane emissions from contrasting urban freshwaters: Rates, drivers,

and a whole-city footprint. Global Change Biology 25: 4234-4243.

Hobbie, S. E., J. C. Finlay, B. D. Janke, D. A. Nidzgorski, D. B. Millet, and L. A. Baker. 2017.

Contrasting nitrogen and phosphorus budgets in urban watersheds and implications

for managing urban water pollution. Proceedings of the National Academy of Sciences.

Hoffman, C. M. 2007. Geospatial mapping and analysis of water availability-demand-use

within the Mara River Basin. Florida International University, Miami.

Holgerson, M. A. 2015. Drivers of carbon dioxide and methane supersaturation in small,

temporary ponds. Biogeochemistry 124: 305-318.

Holgerson, M. A., and P. A. Raymond. 2016. Large contribution to inland water CO2 and

CH4 emissions from very small ponds. Nature Geosci advance online publication.

Hope, D., S. M. Palmer, M. F. Billett, and J. J. C. Dawson. 2001. Carbon dioxide and methane

evasion from a temperate peatland stream. Limnology and Oceanography 46: 847-857.

Hotchkiss, E. R. and others 2015. Sources of and processes controlling CO2 emissions change

with the size of streams and rivers. Nature Geosci 8: 696-699.

Hotchkiss, E. R., and R. O. Hall. 2014. High rates of daytime respiration in three streams:

Use of delta O-18(O2) and O-2 to model diel ecosystem metabolism. Limnol Oceanogr

59: 798-810.

Houser, J. N., P. J. Mulholland, and K. O. Maloney. 2005. Catchment disturbance and stream

metabolism: patterns in ecosystem respiration and gross primary production along a

gradient of upland soil and vegetation disturbance. Journal of the North American

Benthological Society, 24, 538–541.

Huber, S. A., A. Balz, M. Abert, and W. Pronk. 2011a. Characterisation of aquatic humic and

non-humic matter with size-exclusion chromatography–organic carbon detection–

organic nitrogen detection (LC-OCD-OND). Water research 45: 879-885.

Hudson, N., A. Baker, and D. Reynolds. 2007. Fluorescence analysis of dissolved organic

matter in natural, waste and polluted waters—a review. River Research and

Applications 23: 631-649.

Huguet, A., L. Vacher, S. Relexans, S. Saubusse, J.-M. Froidefond, and E. Parlanti. 2009a.

Properties of fluorescent dissolved organic matter in the Gironde Estuary. Organic

Geochemistry 40: 706-719.

Huguet, A., L. Vacher, S. Relexans, S. Saubusse, J. M. Froidefond, and E. Parlanti. 2009b.

Properties of fluorescent dissolved organic matter in the Gironde Estuary. Organic

Geochemistry 40: 706-719.

References

178

Humborg, C. and others 2010. CO2 supersaturation along the aquatic conduit in Swedish

watersheds as constrained by terrestrial respiration, aquatic respiration and

weathering. Global Change Biology 16: 1966-1978.

Huotari, J. and others 2011. Long-term direct CO2 flux measurements over a boreal lake:

Five years of eddy covariance data. Geophysical Research Letters 38.

Huser, B. J., M. Futter, J. T. Lee, and M. Perniel. 2016. In-lake measures for phosphorus

control: The most feasible and cost-effective solution for long-term management of

water quality in urban lakes. Water Research 97: 142-152.

Hutchins, R. H. S., P. Aukes, S. L. Schiff, T. Dittmar, Y. T. Prairie, and P. A. del Giorgio.

2017. The Optical, Chemical, and Molecular Dissolved Organic Matter Succession

Along a Boreal Soil-Stream-River Continuum. Journal of Geophysical Research:

Biogeosciences 122: 2892-2908.

Huttunen, J. T. and others 2003. Fluxes of methane, carbon dioxide and nitrous oxide in

boreal lakes and potential anthropogenic effects on the aquatic greenhouse gas

emissions. Chemosphere 52: 609-621.

Huttunen, J. T. and others 2002a. Exchange of CO2, CH4 and N2O between the atmosphere

and two northern boreal ponds with catchments dominated by peatlands or forests.

Plant and soil 242: 137-146.

Huttunen, J. T. and others 2002b. Fluxes of CH4, CO2, and N2O in hydroelectric reservoirs

Lokka and Porttipahta in the northern boreal zone in Finland. Global Biogeochemical

Cycles 16: 3-1-3-17.

IPCC, 2013. Climate change 2013: The physical science basis. In T. F. Stocker, D. Qin, G. K.

Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex & P. M.

Midgley (Eds.), Contribution of working group I to the fifth assessment report of the

Intergovernmental Panel on Climate Change (pp. 465–570). Cambridge, UK:

Cambridge University Press.

IPCC, 2014. Climate change 2014. Synthesis report: Contribution of Working Groups I, II

and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate

Change [R.K. Pachauri and L.A. Meyer (eds.)], IPCC. Geneva, Switzerland.

IPCC, 2023. Global Carbon and Other Biogeochemical Cycles and Feedbacks, p. 673-816. In

C. Intergovernmental Panel on Climate [ed.], Climate Change 2021 – The Physical

Science Basis: Working Group I Contribution to the Sixth Assessment Report of the

Intergovernmental Panel on Climate Change. Cambridge University Press.

Ishii, S. K., and T. H. Boyer. 2012. Behavior of reoccurring PARAFAC components in

fluorescent dissolved organic matter in natural and engineered systems: a critical

review. Environmental science & technology 46: 2006-2017.

References

179

Iteba, J. O., T. Hein, G. A. Singer, and F. O. Masese. 2021. Livestock as vectors of organic

matter and nutrient loading in aquatic ecosystems in African savannahs. PLOS ONE

16: e0257076.

to calculate river metabolism from diel oxygen–concentration curves. Environmental

Modelling & Software 22: 24-32.

Jacobs, S. M. and others 2007. Nutrient Vectors and Riparian Processing: A Review with

Special Reference to African Semiarid Savannah Ecosystems. Ecosystems 10: 1231-

1249.

Jaffé, R., D. McKnight, N. Maie, R. Cory, W. H. McDowell, and J. L. Campbell. 2008a. Spatial

and temporal variations in DOM composition in ecosystems: The importance of long-

term monitoring of optical properties. Journal of Geophysical Research:

Biogeosciences 113.

Jähne, B., K. O. Münnich, R. Bösinger, A. Dutzi, W. Huber, and P. Libner. 1987. On the

parameters influencing air-water gas exchange. Journal of Geophysical Research:

Oceans 92: 1937-1949.

Jankowski, K. J., F. H. Mejia, J. R. Blaszczak, and G. W. Holtgrieve. 2021. Aquatic ecosystem

metabolism as a tool in environmental management. WIREs Water 8: e1521.

Jensen, O., and H. Wu. 2018. Urban water security indicators: Development and pilot.

Environmental Science & Policy 83: 33-45.

Jespersen, A. M., and K. Christoffersen. 1987. Measurments of chlorophyll-a from

phytoplankton using ethanol as extraction solvent. . Archiv für Hydrobiologie, 109,

445-454. .

Jiang, T. and others 2017. Characteristics of dissolved organic matter (DOM) and

relationship with dissolved mercury in Xiaoqing River-Laizhou Bay estuary, Bohai

Sea, China. Environmental Pollution 223: 19-30.

Jones, J. R., D. V. Obrecht, J. L. Graham, M. B. Balmer, C. T. Filstrup, and J. A. Downing.

2016. Seasonal patterns in carbon dioxide in 15 mid-continent (USA) reservoirs.

Inland Waters 6: 265-272.

Jouanneau, S. and others 2014. Methods for assessing biochemical oxygen demand (BOD): A

review. Water Research 49: 62-82.

Kanga, E. M., J. O. Ogutu, H. Olff, and P. Santema. 2011. Population trend and distribution

of the Vulnerable common hippopotamus Hippopotamus amphibius in the Mara

Region of Kenya. Oryx 45: 20-27.

Kankaala, P., J. Huotari, T. Tulonen, and A. Ojala. 2013. Lake-size dependent physical

forcing drives carbon dioxide and methane effluxes from lakes in a boreal landscape.

Limnology and Oceanography 58: 1915-1930.

References

180

Kaushal, S. S. and others 2014a. Land Use and Climate Variability Amplify Carbon, Nutrient,

and Contaminant Pulses: A Review with Management Implications. JAWRA Journal

of the American Water Resources Association 50: 585-614.

Kaushal, S. S., W. H. McDowell, and W. M. Wollheim. 2014b. Tracking evolution of urban

biogeochemical cycles: past, present, and future. Biogeochemistry 121: 1-21.

Kellerman, A. M., T. Dittmar, D. N. Kothawala, and L. J. Tranvik. 2014. Chemodiversity of

dissolved organic matter in lakes driven by climate and hydrology. Nature

Communications 5: 3804.

Kemp, W. M., E. Smith, M. Marvin-Dipasquale, and W. R. Boynton. 1997. Organic carbon

balance and net ecosystem metabolism in Chesapeake Bay Mar. Ecol. Prog. Ser. 150:

229–248.

Khadka, M. B., J. B. Martin, and J. Jin. 2014. Transport of dissolved carbon and CO2

degassing from a river system in a mixed silicate and carbonate catchment. Journal of

Hydrology 513: 391-402.

Khoiyangbam, R. S., and W. Basanta Kumar. 2014. Influence of waste water disposal on water

quality and methane emission from Nambol Turel: Feeder stream of Loktak Lake in

Manipur, India. International Journal of Recent Scientific Research, 5, 974–979. .

Kingdon, J., and M. Largen. 1997. The kingdom field guide to African mammals. Zoological

Journal of the Linnean Society 120: 479.

Kinnaird, M. F., E. W. Sanderson, T. G. O’Brien, H. T. Wibisono, and G. Woolmer. 2003.

Deforestation Trends in a Tropical Landscape and Implications for Endangered Large

Mammals. Conservation Biology 17(1), 245–257.

Kirschbaum, M. U. 1995. The temperature dependence of soil organic matter decomposition,

and the effect of global warming on soil organic C storage. Soil Biology and

biochemistry 27: 753-760.

Kling, G. W., G. W. Kipphut, and M. C. Miller. 1992. The flux of CO2 and CH4 from lakes

and rivers in the artic Alaska. Hydrobiologia 240: 23-36.

Knappe, A., P. Möller, P. Dulski, and A. Pekdeger. 2005. Positive gadolinium anomaly in

surface water and ground water of the urban area Berlin, Germany. Geochemistry 65:

167-189.

Koch, B. P., T. Dittmar, M. Witt, and G. Kattner. 2007. Fundamentals of Molecular Formula

Assignment to Ultrahigh Resolution Mass Data of Natural Organic Matter. Analytical

Chemistry 79: 1758-1763.

Kokic, J., M. B. Wallin, H. E. Chmiel, B. A. Denfeld, and S. Sobek. 2015. Carbon dioxide

evasion from headwater systems strongly contributes to the total export of carbon

from a small boreal lake catchment. Journal of Geophysical Research: Biogeosciences

120: 13-28.

References

181

Kominoski, J. S., A. D. Rosemond, J. P. Benstead, V. Gulis, J. C. Maerz, and D. W. P.

Manning. 2015. Low-to-moderate nitrogen and phosphorus concentrations accelerate

microbially driven litter breakdown rates. Ecological Applications 25: 856-865.

Kummu, M., H. de Moel, P. J. Ward, and O. Varis. 2011. How close do we live to water? A

global analysis of population distance to freshwater bodies. PLoS One 6: e20578.

Ladwig, R., L. Heinrich, G. Singer, and M. Hupfer. 2017. Sediment core data reconstruct the

management history and usage of a heavily modified urban lake in Berlin, Germany.

Environmental Science and Pollution Research 24: 25166-25178.

Lambert, T. and others 2015. Landscape Control on the Spatial and Temporal Variability of

Chromophoric Dissolved Organic Matter and Dissolved Organic Carbon in Large

African Rivers. Ecosystems 18: 1224-1239.

Lamprey, R. H., and R. S. Reid. 2004. Expansion of human settlement in Kenya's Maasai

Mara: what future for pastoralism and wildlife? Journal of Biogeography 31: 997-1032.

Lapierre, J.-F., F. Guillemette, M. Berggren, and P. A. del Giorgio. 2013. Increases in

terrestrially derived carbon stimulate organic carbon processing and CO2 emissions

in boreal aquatic ecosystems. Nature Communications 4: 2972.

Larson, J. H. and others 2014. Relationships between land cover and dissolved organic matter

change along the river to lake transition. Ecosystems 17: 1413-1425.

Lauerwald, R., G. G. Laruelle, J. Hartmann, P. Ciais, and P. A. G. Regnier. 2015. Spatial

patterns in CO2 evasion from the global river network. Global Biogeochemical Cycles

29: 534-554.

Le Quéré, C. and others 2018. Global Carbon Budget 2018. Earth Syst. Sci. Data 10: 2141-

2194.

Lee, K. Y., R. van Geldern, and J. A. C. Barth. 2017. A high-resolution carbon balance in a

small temperate catchment: Insights from the Schwabach River, Germany. Applied

Geochemistry 85: 86-96.

Legendre, P., and L. Legendre. 2012. Numerical ecology.

Lennon, J. T. 2004. Experimental evidence that terrestrial carbon subsidies increase CO2

flux from lake ecosystems. Oecologia 138: 584-591.

Lesaulnier, C. C. and others 2017. Bottled aqua incognita: microbiota assembly and dissolved

organic matter diversity in natural mineral waters. Microbiome 5: 126.

Lewis, E., D. Wallace, and L. J. Allison. 1998. Program developed for CO{sub 2} system

calculations, p. Medium: ED; Size: 40 p.

Lewison, R. L., and J. Carter. 2004. Exploring behavior of an unusual megaherbivore: a

spatially explicit foraging model of the hippopotamus. Ecological Modelling 171: 127-

138.

References

182

Li, S., J. Luo, D. Wu, and Y. Jun Xu. 2020. Carbon and nutrients as indictors of daily

fluctuations of pCO2 and CO2 flux in a river draining a rapidly urbanizing area.

Ecological Indicators 109: 105821.

Lidman, J., M. Jonsson, R. M. Burrows, M. Bundschuh, and R. A. Sponseller. 2017.

Composition of riparian litter input regulates organic matter decomposition:

Implications for headwater stream functioning in a managed forest landscape. Ecology

and Evolution 7: 1068-1077.

Liu, Q. and others 2019. Impact of land use on the DOM composition in different seasons in

a subtropical river flowing through a region undergoing rapid urbanization. Journal

of Cleaner Production 212: 1224-1231.

Liu, S., D. E. Butman, and P. A. Raymond. 2020. Evaluating CO2 calculation error from

organic alkalinity and pH measurement error in low ionic strength freshwaters.

Limnology and Oceanography: Methods 18: 606-622.

Liu, S. and others 2022. The importance of hydrology in routing terrestrial carbon to the

atmosphere via global streams and rivers. Proceedings of the National Academy of

Sciences 119: e2106322119.

Livingston, G. P., and G. L. Hutchinson. 1995. Enclosure-based measurement of trace gas

exchange: applications and sources of error, in: Biogenic trace gases: measuring

emissions from soil and water, edited by: Matson, P.A. and Harriss, R.C.. Blackwell

Science Ltd.

Loiselle, S. A. and others 2009. The optical characterization of chromophoric dissolved

organic matter using wavelength distribution of absorption spectral slopes. Limnology

and Oceanography 54: 590-597.

López Bellido, J., E. Peltomaa, and A. Ojala. 2011. An urban boreal lake basin as a source of

CO2 and CH4. Environmental Pollution 159: 1649-1659.

Lorke, A. and others 2015. Technical Note: Drifting vs. anchored flux chambers for

measuring greenhouse gas emissions from running waters. Biogeosciences Discuss.

12: 14619-14645.

Lyu, L., G. Liu, Y. Shang, Z. Wen, J. Hou, and K. Song. 2021. Characterization of dissolved

organic matter (DOM) in an urbanized watershed using spectroscopic analysis.

Chemosphere 277: 130210.

MacIntyre, S., R. Wanninkhof, and J. Chanton. 1995. Trace gas exchange across the air-

water interface in freshwater and coastal marine environments. In: Matson, PA,

Harriss, RC (Eds.), Freshwater and Coastal Marine Environments. Blackwell Science,

Oxford, UK, pp. 52–97.

Macklin, P. A., I. G. N. A. Suryaputra, D. T. Maher, and I. R. Santos. 2018. Carbon dioxide

dynamics in a lake and a reservoir on a tropical island (Bali, Indonesia). PLOS ONE

13: e0198678.

References

183

Männistö, E. and others 2019. Multi-year methane ebullition measurements from water and

bare peat surfaces of a patterned boreal bog. Biogeosciences 16: 2409-2421.

Mardia, K. V., Kent, J.T. and Bibby, J.M. . 1979. Multivariate Analysis.

Martinez-Cruz, K., R. Gonzalez-Valencia, A. Sepulveda-Jauregui, F. Plascencia-Hernandez,

Y. Belmonte-Izquierdo, and F. Thalasso. 2017. Methane emission from aquatic

ecosystems of Mexico City. Aquatic Sciences 79: 159-169.

Martinsen, K. T., T. Kragh, and K. Sand-Jensen. 2018. Technical note: A simple and cost-

efficient automated floating chamber for continuous measurements of carbon dioxide

gas flux on lakes. Biogeosciences 15: 5565-5573.

Marzolf, E. R., P. J. Mulholland, and A. D. Steinman. 1998. Reply: Improvements to the

diurnal upstream-downstream dissolved oxygen change technique for determining

whole-stream metabolism in small streams. Can J Fish Aquat Sci 55: 1786-1787.

Masese, F. O., K. G. Abrantes, G. M. Gettel, S. Bouillon, K. Irvine, and M. E. McClain. 2015.

Are Large Herbivores Vectors of Terrestrial Subsidies for Riverine Food Webs?

Ecosystems 18: 686-706.

Masese, F. O., K. G. Abrantes, G. M. Gettel, K. Irvine, S. Bouillon, and M. E. McClain. 2018.

Trophic structure of an African savanna river and organic matter inputs by large

terrestrial herbivores: A stable isotope approach. Freshwater Biology 63: 1365-1380.

Masese, F. O., T. Fuss, L. T. Bistarelli, C. Buchen-Tschiskale, and G. Singer. 2022. Large

herbivorous wildlife and livestock differentially influence the relative importance of

different sources of energy for riverine food webs. Science of The Total Environment

828: 154452.

Masese, F. O., J. S. Salcedo-Borda, G. M. Gettel, K. Irvine, and M. E. McClain. 2017.

Influence of catchment land use and seasonality on dissolved organic matter

composition and ecosystem metabolism in headwater streams of a Kenyan river.

Biogeochemistry 132: 1-22.

Mathuriau, C., and E. Chauvet. 2002. Breakdown of leaf litter in a neotropical stream. Journal

of the North American Benthological Society 21: 384-396.

Mayorga, E. and others 2005. Young organic matter as a source of carbon dioxide outgassing

from Amazonian rivers. Nature 436: 538-541.

McDonald, C. P., E. G. Stets, R. G. Striegl, and D. Butman. 2013. Inorganic carbon loading

as a primary driver of dissolved carbon dioxide concentrations in the lakes and

reservoirs of the contiguous United States. Global Biogeochemical Cycles 27: 285-295.

McEnroe, N. A., C. J. Williams, M. A. Xenopoulos, P. Porcal, and P. C. Frost. 2013. Distinct

Optical Chemistry of Dissolved Organic Matter in Urban Pond Ecosystems. PLOS

ONE 8: e80334.

References

184

McKnight, D. M., E. W. Boyer, P. K. Westerhoff, P. T. Doran, T. Kulbe, and D. T. Andersen.

2001. Spectrofluorometric characterization of dissolved organic matter for indication

of precursor organic material and aromaticity. Limnol Oceanogr 46: 38-48.

Morin, T. H., G. Bohrer, K. C. Stefanik, A. C. Rey-Sanchez, A. M. Matheny, and W. J. Mitsch.

2017. Combining eddy-covariance and chamber measurements to determine the

methane budget from a small, heterogeneous urban floodplain wetland park.

Agricultural and Forest Meteorology 237-238: 160-170.

Mosisch, T. D., S. E. Bunn, and P. M. Davies. 2001. The relative importance of shading and

nutrients on algal production in subtropical streams. Freshwater Biology 46: 1269-

1278.

Mulholland, P. J. and others 2001. Inter-biome comparison of factors controlling stream

metabolism. Freshwater Biology 46: 1503-1517.

Mulholland, P. J., J. N. Houser, and K. O. Maloney. 2005. Stream diurnal dissolved oxygen

profiles as indicators of in-stream metabolism and disturbance effects: Fort Benning as

a case study. Ecological Indicators 5: 243-252.

Müller, D. W. and others 2011. Phylogenetic constraints on digesta separation: variation in

fluid throughput in the digestive tract in mammalian herbivores. Comparative

Biochemistry and Physiology Part A: Molecular & Integrative Physiology 160: 207-

220.

Murphy, K. M., K. R. and others 2011. Organic Matter Fluorescence in Municipal Water

Recycling Schemes: Toward a Unified PARAFAC Model. Environ. Sci. Technol. 45:

2909-2916.

Murphy, K. R., C. A. Stedmon, T. D. Waite, and G. M. Ruiz. 2008. Distinguishing between

terrestrial and autochthonous organic matter sources in marine environments using

fluorescence spectroscopy. Mar. Chem. 108: 40-58.

Murphy, K. R., C. A. Stedmon, P. Wenig, and R. Bro. 2014. OpenFluor– an online spectral

library of auto-fluorescence by organic compounds in the environment. Analytical

Methods 6: 658-661.

Mwanake, R. and others 2019. Land use, not stream order, controls N2O concentration and

flux in the upper Mara River basin, Kenya. Journal of Geophysical Research:

Biogeosciences.

Mwanake, R. M., G. M. Gettel, C. Ishimwe, E. G. Wangari, K. Butterbach-Bahl, and R. Kiese.

2022. Basin-scale estimates of greenhouse gas emissions from the Mara River, Kenya:

Importance of discharge, stream size, and land use/land cover. Limnology and

Oceanography 67: 1776-1793.

Naiman, R. J., and K. H. Rogers. 1997. Large Animals and System-Level Characteristics in

River Corridors: Implications for river management. BioScience 47: 521-529.

References

185

Natchimuthu, S., B. Panneer Selvam, and D. Bastviken. 2014. Influence of weather variables

on methane and carbon dioxide flux from a shallow pond. Biogeochemistry 119: 403-

413.

Natchimuthu, S., I. Sundgren, M. Gålfalk, L. Klemedtsson, and D. Bastviken. 2017.

Spatiotemporal variability of lake pCO2 and CO2 fluxes in a hemiboreal catchment.

Journal of Geophysical Research: Biogeosciences 122: 30-49.

Ni, M., S. Li, J. Luo, and X. Lu. 2019. CO2 partial pressure and CO2 degassing in the Daning

River of the upper Yangtze River, China. Journal of Hydrology 569: 483-494.

Niemczynowicz, J. 1999. Urban hydrology and water management – present and future

challenges. Urban Water 1: 1-14.

Noirard, C., M. Berre, R. Ramousse, C. Sépulcre, and P. Joly. 2004. Diets of sympatric

hippopotamus (Hippopotamus amphibius) and zebus (Bos indicus) during the dry

season in the ‘W’ National Park (Niger Republic). Game Wildl. Sci. 21, 423–431.

Odum, E. 1971. Fundamental of ecology. W.B. Saunders, Philadelphia.

Odum, H. T. 1956. Primary Production in Flowing Waters. Limnol Oceanogr 1: 102-117.

Ogutu, J. O., N. Owen-Smith, H.-P. Piepho, and M. Y. Said. 2011. Continuing wildlife

population declines and range contraction in the Mara region of Kenya during 1977–

2009. Journal of Zoology 285: 99-109.

Ogutu, J. O. and others 2016. Extreme Wildlife Declines and Concurrent Increase in

Livestock Numbers in Kenya: What Are the Causes? PLOS ONE 11: e0163249.

Ohno, T. 2002. Fluorescence Inner-Filtering Correction for Determining the Humification

Index of Dissolved Organic Matter. Environmental Science & Technology 36: 742-

746.

Oyenuga, V., and F. Olubajo. 1975. Pasture productivity in Nigeria: II. Voluntary intake and

herbage digestibility. The Journal of Agricultural Science 85: 337-343.

Pal, A., Y. He, M. Jekel, M. Reinhard, and K. Y.-H. Gin. 2014. Emerging contaminants of

public health significance as water quality indicator compounds in the urban water

cycle. Environment International 71: 46-62.

Panneer Selvam, B., S. Natchimuthu, L. Arunachalam, and D. Bastviken. 2014. Methane and

carbon dioxide emissions from inland waters in India – implications for large scale

greenhouse gas balances. Global Change Biology 20: 3397-3407.

Park, J.-H. 2009. Spectroscopic characterization of dissolved organic matter and its

interactions with metals in surface waters using size exclusion chromatography.

Chemosphere 77: 485-494.

Park, P. K. 1969. Oceanic CO2 system: an evaluation of ten methods of investigation.

Limnology and Oceanography 14(2), 179-186.

References

186

Park, Y., F. Meng, and M. A. Baloch. 2018. The effect of ICT, financial development, growth,

and trade openness on CO2 emissions: an empirical analysis. Environmental Science

and Pollution Research 25: 30708-30719.

Parkhill, K. L., and J. S. Gulliver. 1999. Modeling the effect of light on whole-stream

respiration. Ecological Modelling 117: 333-342.

Parlanti, E., K. Wörz, L. Geoffroy, and M. Lamotte. 2000. Dissolved organic matter

fluorescence spectroscopy as a tool to estimate biological activity in a coastal zone

submitted to anthropogenic inputs. Organic Geochemistry 31: 1765-1781.

Parr, T. B., C. S. Cronan, T. Ohno, S. E. G. Findlay, S. M. C. Smith, and K. S. Simon. 2015.

Urbanization changes the composition and bioavailability of dissolved organic matter

in headwater streams. Limnology and Oceanography 60: 885-900.

Paul, M. J., and J. L. Meyer. 2001. Streams in the Urban Landscape. Annual Review of

Ecology and Systematics 32: 333-365.

Peacock, M. and others 2021. Small artificial waterbodies are widespread and persistent

emitters of methane and carbon dioxide. Global Change Biology 27: 5109-5123.

Peacock, M., J. Audet, S. Jordan, J. Smeds, and M. B. Wallin. 2019. Greenhouse gas emissions

from urban ponds are driven by nutrient status and hydrology. Ecosphere 10: e02643.

Peña-Guzmán, C. and others 2019. Emerging pollutants in the urban water cycle in Latin

America: A review of the current literature. Journal of Environmental Management

237: 408-423.

Peres-Neto, P. R., and D. A. Jackson. 2001. How well do multivariate data sets match? The

advantages of a Procrustean superimposition approach over the Mantel test. Oecologia

129: 169-178.

Perkins, D. M. and others 2012. Consistent temperature dependence of respiration across

ecosystems contrasting in thermal history. Global Change Biology 18: 1300-1311.

Peter, H., G. Singer, A. J. Ulseth, T. Dittmar, Y. T. Prairie, and T. J. Battin. 2020. Travel

Time and Source Variation Explain the Molecular Transformation of Dissolved

Organic Matter in an Alpine Stream Network. Journal of Geophysical Research:

Biogeosciences 125: e2019JG005616.

Petrone, K. C., J. B. Fellman, E. Hood, M. J. Donn, and P. F. Grierson. 2011. The origin and

function of dissolved organic matter in agro-urban coastal streams. Journal of

Geophysical Research: Biogeosciences 116: n/a-n/a.

Peuravuori, J., and K. Pihlaja. 1997. Molecular size distribution and spectroscopic properties

of aquatic humic substances. Analytica Chimica Acta 337: 133-149.

Pickard, A. and others 2021. Greenhouse gas budgets of severely polluted urban lakes in

India. Science of The Total Environment 798: 149019.

References

187

Pilla, R. M. and others 2022. Anthropogenically driven climate and landscape change effects

on inland water carbon dynamics: What have we learned and where are we going?

Glob Chang Biol 28: 5601-5629.

Pirk, N., M. Mastepanov, F. J. W. Parmentier, M. Lund, P. Crill, and T. R. Christensen. 2016.

Calculations of automatic chamber flux measurements of methane and carbon dioxide

using short time series of concentrations. Biogeosciences 13: 903-912.

Pozo, J., E. González, J. R. Díez, J. Molinero, and A. Elósegui. 1997. Inputs of Particulate

Organic Matter to Streams with Different Riparian Vegetation. Journal of the North

American Benthological Society 16: 602-611.

Prins, H. H. T. 2000. Competition Between Wildlife and Livestock in Africa, p. 51-80. In H.

H. T. Prins, J. G. Grootenhuis and T. T. Dolan [eds.], Wildlife Conservation by

Sustainable Use. Springer Netherlands.

Prytherch, J., and M. J. Yelland. 2021. Wind, Convection and Fetch Dependence of Gas

Transfer Velocity in an Arctic Sea-Ice Lead Determined From Eddy Covariance CO2

Flux Measurements. Global Biogeochemical Cycles 35: e2020GB006633.

QGIS Development Team. 2017. QGIS Geographic Information System. Open Source

Geospatial Foundation Project. http://qgis.osgeo.org.

R Core Team. 2016. R: A language and environment for statistical computing. R Foundation

for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

---. 2017. R: A language and environment for statistical computing. R Foundation for

Statistical Computing.

Ramos, S., V. Homem, A. Alves, and L. Santos. 2016. A review of organic UV-filters in

wastewater treatment plants. Environment International 86: 24-44.

Ray, N. E. and others 2023. Spatial and temporal variability in summertime dissolved carbon

dioxide and methane in temperate ponds and shallow lakes. Limnology and

Oceanography 68: 1530-1545.

Raymond, P. A., J. E. Bauer, and J. J. Cole. 2000. Atmospheric CO2 evasion, dissolved

inorganic carbon production, and net heterotrophy in the York River estuary.

Limnology and Oceanography 45: 1707-1717.

Raymond, P. A., and J. J. Cole. 2001. Gas exchange in rivers and estuaries: Choosing a gas

transfer velocity. Estuaries 24: 312-317.

Raymond, P. A. and others 2013. Global carbon dioxide emissions from inland waters. Nature

503: 355-359.

Raymond, P. A. and others 2012. Scaling the gas transfer velocity and hydraulic geometry

in streams and small rivers. Limnology and Oceanography: Fluids and Environments

2: 41-53.

Regnier, P. and others 2013. Anthropogenic perturbation of the carbon fluxes from land to

ocean. Nature Geosci 6: 597-607.

References

188

Reid, R. S. and others 2003. People, wildlife and livestock in the Mara ecosystem: The Mara

count 2002. International Livestock Research Institute, Nairobi, Kenya.

Reiman, J. H., and Y. J. Xu. 2019. Dissolved carbon export and CO2 outgassing from the

lower Mississippi River – Implications of future river carbon fluxes. Journal of

Hydrology 578: 124093.

Repo, E. and others 2007. Release of CO2 and CH4 from small wetland lakes in western

Siberia. Tellus B: Chemical and Physical Meteorology 59: 788-796.

Reusswig, F., B. Hirschl, and W. Lass. 2014. Climate‐neutral Berlin 2050. Results of a

feasibility study. Berlin, Germany: Senate Department for Urban Development and

the Environment. Retrieved from

Richey, J., JT Brock, RJ Naiman, RC Wissmar, and R. Stallard. 1980. Organic carbon:

oxidation and transport in the Amazon River. Science 207 (4437), 1348-1351.

Richey, J. E., J. M. Melack, A. K. Aufdenkampe, V. M. Ballester, and L. L. Hess. 2002.

Outgassing from Amazonian rivers and wetlands as a large tropical source of

atmospheric CO2. Nature 416: 617-620.

Richter, D., G. Massmann, and U. Dünnbier. 2008. Identification and significance of

sulphonamides (p-TSA, o-TSA, BSA) in an urban water cycle (Berlin, Germany).

Water Research 42: 1369-1378.

Riera, J. L., J. E. Schindler, and T. K. Kratz. 1999. Seasonal dynamics of carbon dioxide and

methane in two clear-water lakes and two bog lakes in northern Wisconsin, USA.

Canadian Journal of Fisheries and Aquatic Sciences 56: 265-274.

Riley, A. J., and W. K. Dodds. 2013. Whole-stream metabolism: strategies for measuring and

modeling diel trends of dissolved oxygen. Freshwater Science 32: 56-69.

Ripple, W. J. and others 2015. Collapse of the world’s largest herbivores. Science

Advances 1: e1400103.

Robert O. Hall, J., and J. J. Beaulieu. 2013. Estimating autotrophic respiration in streams

using daily metabolism data. Freshwater Science 32: 507-516.

Rocher-Ros, G. and others 2023. Global methane emissions from rivers and streams. Nature

621: 530-535.

Rodríguez-Castillo, T., J. Barquín, M. Álvarez-Cabria, F. J. Peñas, and C. Álvarez. 2017.

Effects of sewage effluents and seasonal changes on the metabolism of three Atlantic

rivers. Science of The Total Environment 599-600: 1108-1118.

Romero González-Quijano, C., S. Herrero Ortega, P. Casper, M. O. Gessner, and G. A.

Singer. 2022. Dissolved organic matter signatures in urban surface waters: spatio-

temporal patterns and drivers. Biogeosciences 19: 2841-2853.

Rosamond, M. S., S. J. Thuss, and S. L. Schiff. 2012. Dependence of riverine nitrous oxide

emissions on dissolved oxygen levels. Nature Geoscience 5: 715-718.

References

189

Rosentreter, J. A. and others 2021. Half of global methane emissions come from highly

variable aquatic ecosystem sources. Nature Geoscience 14: 225-230.

Roulet, N., P. Crill, N. Comer, A. Dove, and R. Boubonniere. 1997. CO sub (2) and CH sub

(4) flux between a boreal beaver pond and the atmosphere. Journal of Geophysical

Research. D. Atmospheres 102: 29.

Roy, A. H. and others 2016. Urbanization and stream ecology: diverse mechanisms of change.

Freshwater Science 35: 272-277.

Saarela, T. and others 2022. Dissolved organic matter composition regulates microbial

degradation and carbon dioxide production in pristine subarctic rivers. Biogeosciences

Discuss. 2022: 1-28.

Sand-Jensen, K., N. L. Pedersen, and M. Sondergaard. 2007. Bacterial metabolism in small

temperate streams under contemporary and future climates. Freshwater Biol 52: 2340-

2353.

Sankar, M. S. and others 2020. Land use and land cover control on the spatial variation of

dissolved organic matter across 41 lakes in Mississippi, USA. Hydrobiologia 847:

1159-1176.

Savory, J. J. and others 2011. Parts-Per-Billion Fourier Transform Ion Cyclotron Resonance

Mass Measurement Accuracy with a “Walking” Calibration Equation. Analytical

Chemistry 83: 1732-1736.

Sawakuchi, H. O. and others 2017. Carbon Dioxide Emissions along the Lower Amazon

River. Frontiers in Marine Science 4.

Schelker, J., G. A. Singer, A. J. Ulseth, S. Hengsberger, and T. J. Battin. 2016. CO2 evasion

from a steep, high gradient stream network: importance of seasonal and diurnal

variation in aquatic pCO2 and gas transfer. Limnology and Oceanography 61: 1826-

1838.

Schilder, J. and others 2013. Spatial heterogeneity and lake morphology affect diffusive

greenhouse gas emission estimates of lakes. Geophysical Research Letters 40: 5752-

5756.

Schubert, C. J. and others 2010. Oxidation and emission of methane in a monomictic lake

(Rotsee, Switzerland). Aquatic Sciences 72: 455-466.

Schwarzenbach, R. P. and others 2006. The Challenge of Micropollutants in Aquatic

Systems. Science 313: 1072-1077.

Schweiger, A. H., and J.-C. Svenning. 2020. Analogous losses of large animals and trees,

socio-ecological consequences, and an integrative framework for rewilding-based

megabiota restoration. People and Nature 2: 29-41.

Segers, R. 1998. Methane production and methane consumption: a review of processes

underlying wetland methane fluxes. Biogeochemistry 41: 23-51.

References

190

SenUVK. (2005). Gewässeratlas von Berlin. Berlin, Germany: Senatsverwaltung für

Stadtentwicklung.

SenUVK. (2018). Senate Department for the Environment, Transport and Climate

Protection. Retrieved from https ://www.berlin.de/senuv k/

umwelt/wasser/ogewaesser

Sepulveda-Jauregui, A. and others 2018a. Eutrophication exacerbates the impact of climate

warming on lake methane emission. Science of The Total Environment 636: 411-419.

Sepulveda-Jauregui, A., K. Martinez-Cruz, M. Lau, and P. Casper. 2018b. Assessment of

methane and carbon dioxide emissions in two sub-basins of a small acidic bog lake

artificially divided 30 years ago. Freshwater Biology 63: 1534-1549.

Sepulveda-Jauregui, A., K. M. Walter Anthony, K. Martinez-Cruz, S. Greene, and F.

Thalasso. 2015. Methane and carbon dioxide emissions from 40 lakes along a north–

south latitudinal transect in Alaska. Biogeosciences 12: 3197-3223.

Shao, C. and others 2015. Diurnal to annual changes in latent, sensible heat, and CO2 fluxes

over a Laurentian Great Lake: A case study in Western Lake Erie. Journal of

Geophysical Research: Biogeosciences 120: 1587-1604.

Shirokova, L., O. Pokrovsky, S. Kirpotin, and B. Dupré. 2009. Heterotrophic bacterio‐

plankton in thawed lakes of the northern part of Western Siberia controls the CO2

flux to the atmosphere. International Journal of Environmental Studies 66: 433-445.

Sileshi, G. W., N. Nhamo, P. L. Mafongoya, and J. Tanimu. 2017. Stoichiometry of animal

manure and implications for nutrient cycling and agriculture in sub-Saharan Africa.

Nutrient cycling in agroecosystems 107: 91-105.

Sinsabaugh, R. L. 1997. Large-Scale Trends for Stream Benthic Respiration. Journal of the

North American Benthological Society 16: 119-122.

Sitters, J., M. J. Maechler, P. J. Edwards, W. Suter, and H. Olde Venterink. 2014. Interactions

between C: N: P stoichiometry and soil macrofauna control dung decomposition of

savanna herbivores. Functional Ecology 28: 776-786.

Smith, R. M., and S. S. Kaushal. 2015. Carbon cycle of an urban watershed: exports, sources,

and metabolism. Biogeochemistry 126: 173-195.

Smith, R. M., S. S. Kaushal, J. J. Beaulieu, M. J. Pennino, and C. Welty. 2017. Influence of

infrastructure on water quality and greenhouse gas dynamics in urban streams.

Biogeosciences 14: 2831-2849.

Sobek, S., G. Algesen, A. Bergström, M. Jansson, and L. Tranvik. 2003. The catchment and

climate regulation of pCO2 in boreal lakes. Global Change Biology 9: 630-641.

Song, K. and others 2019. Characterization of CDOM in saline and freshwater lakes across

China using spectroscopic analysis. Water Research 150: 403-417.

References

191

Spencer, R. G. M. and others 2009. Photochemical degradation of dissolved organic matter

and dissolved lignin phenols from the Congo River. Journal of Geophysical Research:

Biogeosciences 114.

Stanley, E. H., N. J. Casson, S. T. Christel, J. T. Crawford, L. C. Loken, and S. K. Oliver. 2016.

The ecology of methane in streams and rivers: patterns, controls, and global

significance. Ecological Monographs 86: 146-171.

Stanley, E. H., S. M. Powers, N. R. Lottig, I. Buffam, and J. T. Crawford. 2012. Contemporary

changes in dissolved organic carbon (DOC) in human-dominated rivers: is there a role

for DOC management? Freshwater Biology 57: 26-42.

Stears, K. and others 2018. Effects of the hippopotamus on the chemistry and ecology of a

changing watershed. Proceedings of the National Academy of Sciences 115: E5028-

E5037.

Stedmon, C. A., and R. Bro. 2008. Characterizing dissolved organic matter fluorescence with

parallel factor analysis: a tutorial. Limnology and Oceanography: Methods 6: 572-579.

Stedmon, C. A., and S. Markager. 2005a. Resolving the variability in dissolved organic matter

fluorescence in a temperate estuary and its catchment using PARAFAC analysis.

Limnol Oceanogr 50: 686-697.

---. 2005b. Tracing the production and degradation of autochthonous fractions of dissolved

organic matter by fluorescence analysis. Limnology and Oceanography 50: 1415-1426.

Stedmon, C. A., S. Markager, and R. Bro. 2003. Tracing dissolved organic matter in aquatic

environments using a new approach to fluorescence spectroscopy. Mar. Chem. 82: 239-

254.

Steele, M. K., and J. B. Heffernan. 2013. Morphological characteristics of urban water bodies:

mechanisms of change and implications for ecosystem function. Ecological

Applications 24: 1070-1084.

Stohlgren, T. J. 1988. Litter dynamics in two Sierran mixed conifer forests. II. Nutrient

release in decomposing leaf litter. Canadian Journal of Forest Research 18: 1136-1144.

Striegl, R. G., M. M. Dornblaser, C. P. McDonald, J. R. Rover, and E. G. Stets. 2012. Carbon

dioxide and methane emissions from the Yukon River system. Global Biogeochemical

Cycles 26: n/a-n/a.

Striegl, R. G., and C. M. Michmerhuizen. 1998. Hydrologic influence on methane and carbon

dioxide dynamics at two north‐central Minnesota lakes. Limnology and Oceanography

43: 1519-1529.

Subalusky, A. L., C. L. Dutton, L. Njoroge, E. J. Rosi, and D. M. Post. 2018. Organic matter

and nutrient inputs from large wildlife influence ecosystem function in the Mara River,

Africa. Ecology 99: 2558-2574.

References

192

Subalusky, A. L., C. L. Dutton, E. J. Rosi-Marshall, and D. M. Post. 2015. The hippopotamus

conveyor belt: vectors of carbon and nutrients from terrestrial grasslands to aquatic

systems in sub-Saharan Africa. Freshwater Biology 60: 512-525.

Subalusky, A. L., C. L. Dutton, E. J. Rosi, and D. M. Post. 2017. Annual mass drownings of

the Serengeti wildebeest migration influence nutrient cycling and storage in the Mara

River. Proceedings of the National Academy of Sciences 114: 7647-7652.

Subalusky, A. L., and D. M. Post. 2019. Context dependency of animal resource subsidies.

Biological Reviews 94: 517-538.

Sweeney, B. W. and others 2004. Riparian deforestation, stream narrowing, and loss of

stream ecosystem services. Proc Natl Acad Sci U S A 101: 14132-14137.

Tamil Selvi, S., V. Raman, and N. Rajendran. 2003. Corrosion inhibition of mild steel by

benzotriazole derivatives in acidic medium. Journal of Applied Electrochemistry 33:

1175-1182.

Tang, W., Y. Jun Xu, and S. Li. 2021. Rapid urbanization effects on partial pressure and

emission of CO2 in three rivers with different urban intensities. Ecological Indicators

125: 107515.

Teymouri, B. 2007. Fluorescence spectroscopy and parallel factor analysis of waters from

municipal waste sources.

Thomas, S., and R. C. Campling. 1977. Comparisons of some factors affecting digestibility in

sheep and cows. Grass and Forage Science 32: 33-41.

Thurman, E. M. 1985. Classification of Dissolved Organic Carbon, p. 103-110. In E. M.

Thurman [ed.], Organic Geochemistry of Natural Waters. Springer Netherlands.

Tian, Y. Q., D. Wang, R. F. Chen, and W. Huang. 2012. Using modeled runoff to study DOC

dynamics in stream and river flow: A case study of an urban watershed southeast of

Boston, Massachusetts. Ecological Engineering 42: 212-222.

Tranvik, L. J. and others 2009. Lakes and reservoirs as regulators of carbon cycling and

climate. Limnology and Oceanography 54: 2298-2314.

UBA. 2017. https://www.umweltbundesamt.de/daten/klima/treibhausgas-emissionen-in-

deutschland/kohlendioxid-emissionen accesed on 24th March 2023.

Ulseth, A. J., E. Bertuzzo, G. A. Singer, J. Schelker, and T. J. Battin. 2018. Climate-Induced

Changes in Spring Snowmelt Impact Ecosystem Metabolism and Carbon Fluxes in an

Alpine Stream Network. Ecosystems 21: 373-390.

UN. 2016. The world's cities in 2016. Data booklet (ST/ESA/ SER.A/392). New York, NY:

United Nations Department of Economic and Social Affairs

UN, 2018. The World’s Cities in 2018. New York, NY: United Nations Department of

Economic and Social Affairs

USBR. 2023. https://www.usbr.gov/.

References

193

Vachon, D., T. Langenegger, D. Donis, S. E. Beaubien, and D. F. McGinnis. 2020a. Methane

emission offsets carbon dioxide uptake in a small productive lake. Limnology and

Oceanography Letters 5: 384-392.

Vachon, D., and Y. T. Prairie. 2013. The ecosystem size and shape dependence of gas transfer

velocity versus wind speed relationships in lakes. Canadian Journal of Fisheries and

Aquatic Sciences 70: 1757-1764.

Vachon, D. and others 2020b. Paired O2–CO2 measurements provide emergent insights into

aquatic ecosystem function. Limnology and Oceanography Letters 5: 287-294.

Van de Bogert, M. C., S. R. Carpenter, J. J. Cole, and M. L. Pace. 2007. Assessing pelagic and

benthic metabolism using free water measurements. Limnol Oceanogr-Meth 5: 145-

155.

Vanni, M. J. 2002. Nutrient Cycling by Animals in Freshwater Ecosystems. Annual Review

of Ecology and Systematics 33: 341-370.

Veldhuis, M. P. and others 2019. Cross-boundary human impacts compromise the Serengeti-

Mara ecosystem. Science 363: 1424-1428.

Venkiteswaran, J. J., S. L. Schiff, and W. D. Taylor. 2015. Linking aquatic metabolism, gas

exchange, and hypoxia to impacts along the 300-km Grand River, Canada. Freshwater

Science 34: 1216-1232.

Waajen, G. W. A. M., E. J. Faassen, and M. Lürling. 2014. Eutrophic urban ponds suffer

from cyanobacterial blooms: Dutch examples. Environmental Science and Pollution

Research 21: 9983-9994.

Wallace, J. B., S. L. Eggert, J. L. Meyer, and J. R. Webster. 1997. Multiple Trophic Levels of

a Forest Stream Linked to Terrestrial Litter Inputs. Science 277: 102-104.

Wallin, M. B. and others 2013. Evasion of CO2 from streams – The dominant component of

the carbon export through the aquatic conduit in a boreal landscape. Global Change

Biology 19: 785-797.

Wallin, M. B., M. G. Öquist, I. Buffam, M. F. Billett, J. Nisell, and K. H. Bishop. 2011.

Spatiotemporal variability of the gas transfer coefficient (KCO2) in boreal streams:

Implications for large scale estimates of CO2 evasion. Global Biogeochemical Cycles

25.

Walsh, C. J., A. H. Roy, J. W. Feminella, P. D. Cottingham, P. M. Groffman, and R. P.

Morgan. 2005. The urban stream syndrome: current knowledge and the search for a

cure. Journal of the North American Benthological Society 24: 706-723.

Wang, G., S. Liu, S. Sun, and X. Xia. 2023. Unexpected low CO2 emission from highly

disturbed urban inland waters. Environmental Research 235: 116689.

Wang, G. and others 2021. Intense methane ebullition from urban inland waters and its

significant contribution to greenhouse gas emissions. Water Research 189: 116654.

References

194

Wang, X. and others 2018. CH4 concentrations and fluxes in a subtropical metropolitan river

network: Watershed urbanization impacts and environmental controls. Science of The

Total Environment 622-623: 1079-1089.

Wang, X. and others 2017. pCO2 and CO2 fluxes of the metropolitan river network in

relation to the urbanization of Chongqing, China. Journal of Geophysical Research:

Biogeosciences 122: 470-486.

Wang, Y., A. G. L. Borthwick, and J. Ni. 2022. Human affinity for rivers. River 1: 4-14.

Wanninkhof, R. 1992. Relationship between wind speed and gas exchange over the ocean.

Journal of Geophysical Research: Oceans 97: 7373-7382.

Wantzen, K. M., C. M. Yule, J. M. Mathooko, and C. M. Pringle. 2008. 3 - Organic Matter

Processing in Tropical Streams, p. 43-64. In D. Dudgeon [ed.], Tropical Stream

Ecology. Academic Press.

Ward, J. V., K. Tockner, D. B. Arscott, and C. Claret. 2002. Riverine landscape diversity.

Freshwater Biology 47: 517-539.

Webster, J. R. and others 1999. What happens to allochthonous material that falls into

streams? A synthesis of new and published information from Coweeta. Freshwater

Biology 41: 687-705.

Weishaar, J. L., G. R. Aiken, B. A. Bergamaschi, M. S. Fram, R. Fujii, and K. Mopper. 2003.

Evaluation of Specific Ultraviolet Absorbance as an Indicator of the Chemical

Composition and Reactivity of Dissolved Organic Carbon. Environmental Science &

Technology 37: 4702-4708.

Weyhenmeyer, G. A., Y. T. Prairie, and L. J. Tranvik. 2014. Browning of Boreal Freshwaters

Coupled to Carbon-Iron Interactions along the Aquatic Continuum. PLOS ONE 9:

e88104.

White, J. Y., and C. J. Walsh. 2020. Catchment-scale urbanization diminishes effects of

habitat complexity on instream macroinvertebrate assemblages. Ecological

Applications 30: e2199.

Wik, M., P. M. Crill, R. K. Varner, and D. Bastviken. 2013. Multiyear measurements of

ebullitive methane flux from three subarctic lakes. Journal of Geophysical Research:

Biogeosciences 118: 1307-1321.

Wik, M., R. K. Varner, K. W. Anthony, S. MacIntyre, and D. Bastviken. 2016. Climate-

sensitive northern lakes and ponds are critical components of methane release. Nature

Geoscience 9: 99-105.

Wilkinson, J., C. Bors, F. Burgis, A. Lorke, and P. Bodmer. 2018. Measuring CO2 and CH4

with a portable gas analyzer: Closed-loop operation, optimization and assessment.

PLOS ONE 13: e0193973.

References

195

Williams, C. J. and others 2016. Human activities cause distinct dissolved organic

matter composition across freshwater ecosystems. Global Change Biology 22: 613-

626.

Williams, C. J., P. C. Frost, and M. A. Xenopoulos. 2013. Beyond best management practices:

pelagic biogeochemical dynamics in urban stormwater ponds. Ecological Applications

23: 1384-1395.

Williams, C. J., Y. Yamashita, H. F. Wilson, R. Jaffé, and M. A. Xenopoulos. 2010. Unraveling

the role of land use and microbial activity in shaping dissolved organic matter

characteristics in stream ecosystems. Limnology and Oceanography 55: 1159-1171.

Williamson, C. E., W. Dodds, T. K. Kratz, and M. A. Palmer. 2008. Lakes and streams as

sentinels of environmental change in terrestrial and atmospheric processes. Frontiers

in Ecology and the Environment 6: 247-254.

Wilson, H. F., and M. A. Xenopoulos. 2009a. Effects of agricultural land use on the

composition of fluvial dissolved organic matter. Nature Geoscience 2: 37.

Wilson, H. F., and M. A. Xenopoulos. 2009b. Effects of agricultural land use on the

composition of fluvial dissolved organic matter. Nature Geoscience 2: 37-41.

Wood, S., and M. S. Wood. 2015. Package ‘mgcv’. R package version: 1-7.

Wu, F., R. Evans, and P. Dillon. 2003. Separation and characterization of NOM by high-

performance liquid chromatography and on-line three-dimensional excitation emission

matrix fluorescence detection. Environmental science & technology 37: 3687-3693.

Xenopoulos, M. A. and others 2021. How humans alter dissolved organic matter composition

in freshwater: relevance for the Earth’s biogeochemistry. Biogeochemistry 154: 323-

348.

Xiao, S. and others 2014. Gas transfer velocities of methane and carbon dioxide in a

subtropical shallow pond. Tellus B: Chemical and Physical Meteorology 66: 23795.

Xing, Y., P. Xie, H. Yang, L. Ni, Y. Wang, and K. Rong. 2005. Methane and carbon dioxide

fluxes from a shallow hypereutrophic subtropical Lake in China. Atmospheric

Environment 39: 5532-5540.

Yamamoto, S., Alcauskas, J.B., Crozier, T.E., 1976. Solubility of methane in distilled water

and seawater. J. Chem. Eng. Data 21, 78–80.

Yamashita, Y., L. J. Scinto, N. Maie, and R. Jaffé. 2010. Dissolved Organic Matter

Characteristics Across a Subtropical Wetland’s Landscape: Application of Optical

Properties in the Assessment of Environmental Dynamics. Ecosystems 13: 1006-1019.

Yao, G. and others 2007. Dynamics of CO2 partial pressure and CO2 outgassing in the lower

reaches of the Xijiang River, a subtropical monsoon river in China. Science of the Total

Environment 376: 255-266.

References

196

Young, H. S. and others 2013. Effects of mammalian herbivore declines on plant

communities: observations and experiments in an African savanna. J Ecol 101: 1030-

1041.

Young, R. G., and A. D. Huryn. 1999. Effects of land use on stream metabolism and organic

matter turnover. Ecological Applications 9: 1359-1376.

Yu, Z. and others 2017. Carbon dioxide and methane dynamics in a human-dominated

lowland coastal river network (Shanghai, China). Journal of Geophysical Research:

Biogeosciences 122: 2017JG003798.

Zhang, L., Y. J. Xu, and S. Li. 2022. Riverine dissolved organic matter (DOM) as affected by

urbanization gradient. Environmental Research 212: 113457.

Zhang, S. and others 2014. Detection of methane biogenesis in a shallow urban lake in

summer. Journal of Soils and Sediments 14: 1004-1012.

Zhang, T., X. Huang, Y. Yang, Y. Li, and R. A. Dahlgren. 2016. Spatial and temporal

variability in nitrous oxide and methane emissions in urban riparian zones of the Pearl

River Delta. Environmental Science and Pollution Research 23: 1552-1564.

Zhao, Y., K. Song, S. Li, J. Ma, and Z. Wen. 2016. Characterization of CDOM from urban

waters in Northern-Northeastern China using excitation-emission matrix fluorescence

and parallel factor analysis. Environmental Science and Pollution Research 23: 15381-

15394.

Zhu, D., H. Chen, Q. a. Zhu, Y. Wu, and N. Wu. 2012. High Carbon Dioxide Evasion from

an Alpine Peatland Lake: The Central Role of Terrestrial Dissolved Organic Carbon

Input. Water, Air, & Soil Pollution 223: 2563-2569.

Zhu, Y., L. Merbold, D. Pelster, E. Diaz‐Pines, G. N. Wanyama, and K. Butterbach‐Bahl.

2018. Effect of dung quantity and quality on greenhouse gas fluxes from tropical

pastures in Kenya. Global Biogeochemical Cycles 32: 1589-1604.

Zietzschmann, F., G. Aschermann, and M. Jekel. 2016. Comparing and modeling organic

micro-pollutant adsorption onto powdered activated carbon in different drinking

waters and WWTP effluents. Water Research 102: 190-201.

Zsolnay, A., E. Baigar, M. Jimenez, B. Steinweg, and F. Saccomandi. 1999. Differentiating

with fluorescence spectroscopy the sources of dissolved organic matter in soils

subjected to drying. Chemosphere 38: 45-50.

Zuur, A., E. N. Ieno, and G. M. Smith. 2007. Analyzing ecological data. Springer Science &

Business Media.