The role of aquatic and terrestrial material flows to an estuarine food web using stable isotopes

Sofia Ester e Sousa de Aguilar Dias The role of aquatic and terrestrial material flows to an estuarine food web using stable isotopes Tese de Candidatura ao grau de Doutor em Ciências do Mar e do Ambiente (Especialidade em Planeamento e Gestão Ambiental). Programa Doutoral da Universidade do Porto (Instituto de Ciências Biomédicas de Abel Salazar e Faculdade de Ciências) e da Universidade de Aveiro. Orientador– Professor Doutor Carlos Antunes Categoria– Professor auxiliar Afiliação– Escola Superior Gallaecia & Centro Interdisciplinar de Investigação Marinha e Ambiental da Universidade do Porto Co-orientador– Professor Doutor Joel C. Hoffman Categoria– Adjunct Assistant Professor Afiliação– University of Minnesota & United States Environmental Protection Agency, Mid-Continent Ecology Division, Duluth, Minnesota (USA). This thesis includes one scientific paper published in one international peer-reviewed journal, and one submitted article originated from part of the results obtained in the experimental work of this thesis and referenced as: Dias E, Morais P, Antunes C, Hoffman JC (in press) Linking terrestrial and benthic estuarine ecosystems: organic matter sources supporting the high secondary production of a non-indigenous bivalve. Biol Invasions. doi: 10.1007/s10530-014-0655-8 Dias E, Morais P, Cotter AM, Antunes C, Hoffman JC (submitted) The importance of cross-ecosystem subsidies to the lower food web of an oligotrophic estuary. “Não faças do pensamento, blocos de duro cimento.” (anónimo) i TABLE OF CONTENTS page Acknowledgments iii Figure index Vii Tables and appendix index xv Abstract xvii Resumo xix Objectives xxi Chapter 1Using stable isotopes to untangle trophic relationships in estuarine food webs 1 1.1 Introduction 3 1.2 Stable isotope ratios and estuarine food webs 4 1.3 Methods to identify which sources support consumers in food web studies 6 1.3.1 Qualitative methods 6 1.3.2 Quantitative methods 7 1.4 Characterization of end-members 11 1.5 Dominant assumptions in food web analysis 14 1.6 Conclusions 16 1.7 References 17 Chapter 2 - The importance of cross-ecosystem subsidies to the lower food web of an oligotrophic estuary 25 2.1 Introduction 27 2.2 Methods 30 2.2.1 Study area 30 2.2.2 Field sampling 31 2.2.3 Laboratory analyses 32 2.2.4 Data analyses 33 2.3 Results 35 2.3.1 Environmental data 35 2.3.2 Food web characterization 37 2.4 Discussion 47 2.4.1 Influence of river discharge 47 2.4.2 Cross-ecosystem subsidies 49 ii 2.4.3 Possible consequences of allochthony to the estuarine food web 50 2.5 References 51 Chapter 3 - Spatial and temporal variability of an estuarine benthic food web: patterns and processes in an oligotrophic ecosystem 61 3.1 Introduction 63 3.2 Methods 64 3.2.1 Study area 64 3.2.2 Field sampling 66 3.2.3 Laboratory analyses 66 3.2.4 Data analyses 67 3.3 Results 69 3.3.1 Spatial heterogeneity of estuarine food web components 69 3.3.2Temporal variability of estuarine food web components 75 3.3.3 Food web characterization 81 3.4 Discussion 88 3.4.1 Spatial heterogeneity of estuarine food web components 89 3.4.2 Temporal variability in the estuarine food web 91 3.5 References 94 Chapter 4 - Linking terrestrial and benthic estuarine ecosystems: organic matter sources supporting the high secondary production of a non-indigenous bivalve 99 4.1 Introduction 101 4.2 Methods 103 4.2.1 Study area 103 4.2.2 Field sampling 104 4.2.3 Laboratory analyses 105 4.2.4 Data analyses 106 4.3 Results 108 4.3.1 Environmental data 108 4.3.2 Organic matter sources stable isotope values 111 4.3.3 Corbicula fluminea stable isotope values 112 4.3.4 SIAR mixing model 113 4.4 Discussion 117 4.4.1. Sources of organic matter supporting C. fluminea production in the 118 iii upper estuary 4.4.2 Sources of organic matter sources supporting C. fluminea production in the lower estuary 120 4.4.3 Possible effects of C. fluminea in the estuarine food web 121 4.5 References 122 Chapter 5 - The role of aquatic and terrestrial material flows to an estuarine food web using stable isotopes: general conclusions 131 x benthic consumers (BC) and epibenthic consumers (EC). The y-axis scale varies between graphs. Fig. 3.4 Monthly average values (±SD) of particulate nitrogen (δ15NPN; A), particulate organic carbon (δ13CPOC; B), particulate organic matter C:N ratios (C:NPOM; C) and chlorophyll a (Chl a) concentration (g L-1; D) in each sampling station, along the Minho River estuary, during the sampling period. 72 Fig. 3.5 Monthly δ15N (A-C) and δ13C (D-F) average values (‰) of pelagic (A, D), benthic (B, E) and epibenthic consumers (C, F), collected along the Minho River estuary during 2011. 73 Fig. 3.6 Clustering analyses performed by station (A, B) and month (C, D) for δ13C (A, C) and δ15N (B, D) values for assemblages consisting of the main functional groups of consumers. Grey lines indicate groups of samples not separated (at < 0.05) by SIMPROF. 74 Fig. 3.7 Average values of particulate nitrogen (δ15NPN) (A), particulate organic carbon (δ13CPOC) (B), particulate organic matter C:N ratios (C:NPOM) (C) and chlorophyll a (Chl a) concentration (g L-1) (D) in each station located along the Minho river estuary, in August 2010 and during 2011. 76 Fig. 3.8 Within-seasonal and inter-annual variation of δ13C and δ15N (‰, mean ± SD) for benthic and pelagic consumers, collected in the Minho River estuary in stations (S) S1, S2, S3 and in the TFW area. Note that the y-axis scale differs between plots. 79 Fig. 3.9 Within-seasonal and inter-annual variation of δ13C and δ15N values (‰, mean ± SD) for epibenthic consumers, collected in the Minho River estuary in stations (S) S1, S2, S3 and in the TFW area. Note that the y-axis scale differs between plots. 80 Fig. 3.10 Average (±SD) δ15N and δ13C values (‰) of pelagic (white symbols), benthic (black symbols) and epibenthic consumers (grey symbols) in the brackish (S1-S3) and freshwater (TFW) areas of the Minho River estuary during winter-spring and summer 2011. The consumers shown are: zooplankton (Z), Scrobicularia plana (Sp), Hediste diversicolor (Hd), Cyathura carinata (Cy), Corophium sp. (C), Corbicula fluminea (Cf), 82-83 xi Insect larvae (In), Oligochaeta (O), Gastropoda (G), Crangon crangon (Cc), Carcinus maenas (Cm), Atyaephyra desmaresti (Ad), Pomatoschistus microps (Pm), Platichthys flesus (Pf), Gasterosteus aculeatus (Ga) and Cobitis paludica (Cp). Boxes represent the ranges for the sources collected during this study, and also the estimates values for C4 saltmarsh plants and phytoplankton (see text). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). Fig. 3.11 Proportion of each food source to Crangon crangon biomass in stations 1, 2 and 3 in August 2010 (August’), and January-April 2011, JulySeptember 2011, based on the stable isotope mixing model results. The food sources included in the model were the pelagic consumers (PC), suspension feeders and carnivorous (SF/C), collectors and carnivorous (Cl/C), and collectors (Cl). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 85 Fig. 3.12 Proportion of each food source to Carcinus maenas biomass in stations 1, 2 and 3 in August 2010 (August’), and January-April 2011, JulySeptember 2011, based on the stable isotope mixing model results. The food sources included in the model were the suspension feeders and carnivorous (SF/C), collectors and carnivorous (Cl/C), collectors (Cl), crustacean predators (Pr1), and fish predators (Pr2). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 86 Fig. 3.13 Proportion of each food source to Pomatoschistus microps biomass in stations 1, 2 and 3 in August 2010 (August’), and January-April 2011, July-September 2011, based on the stable isotope mixing model results. The food sources included in the model were the pelagic consumers (PC), suspension feeders and carnivorous (SF/C), collectors and carnivorous (Cl/C), and collectors (Cl). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 87 Fig. 3.14 Proportion of each food source to Platichthys flesus biomass in the brackish and TFW portions of the Minho River estuary in August 2010 (August’), and January-April 2011, July-September 2011, based on the stable isotope mixing model results. The food sources included in the 88 xii model were the collectors (Cl), collectors and carnivorous (Cl/C), pelagic consumers (PC), suspension feeders and carnivorous (SF/C), benthic grazers (BG), and shredders (S). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 4 and 5 refer to the results from the isotope mixing model for stations 4 and 5, respectively, due to the existence of significant differences between sessile (Corbicula fluminea) or less mobile (Atyaephyra desmaresti) stable isotope values in August 2011. Fig. 4.1 Location of the sampling stations along the Minho River estuary. 104 Fig. 4.2 Surface (open circles) and bottom (closed circles) salinity measured along the Minho River estuary, September 2010. The first and the last sampling stations were located at 8 km and 21 km away from the river mouth, respectively. 109 Fig. 4.3 Segmented linear regression for Corbicula fluminea’s δ13C (A) and δ15N (B) values in relation to salinity (relationship of type 4; Oosterbaan 1994); the 90% confidence belt and 90% confidence block of break-point are shown. Each point represents an individual Corbicula fluminea 113 Fig. 4.4 Average δ13C and δ15N values of Corbicula fluminea adjusted for one trophic level fractionation (+0.4‰ δ13C, +3.4‰ for δ15N) and potential organic matter sources for lower estuary (stations 1-4). Error bars represent one standard deviation. The sources considered relevant for Corbicula fluminea (closed square; each represents a station-specific average) in this portion of the estuary were marine phytoplankton (MP; closed diamond), estuarine phytoplankton (EP; open triangle), microphytobenthos (MPB; open circle), sediment organic matter (SOM; closed triangle) and particulate organic matter (POM; closed circle). 114 Fig. 4.5 Average δ13C values of Corbicula fluminea (closed triangle) adjusted for one trophic level fractionation (+0.4‰ δ13C) and potential particulate organic matter sources (δ13Cphytoplankton [estimated from δ13CDIC]- closed circle; δ13CPOCopen circle) along the estuarine salinity mixing. Error bars represent one standard deviation. 115 Fig. 4.6 Proportion of each food source to Corbicula fluminea’s biomass collected in the lower estuary (stations 1-4) based on the stable isotope 116 xiii mixing model. The food sources included in the model were estuarine phytoplankton (EP), particulate organic matter (POM), sediment organic matter (SOM) and microphytobenthos (MPB). Boxes indicate 50%, 75% and 95% Bayesian credibility interval. Fig. 4.7 Average δ13C and δ15N values of Corbicula fluminea adjusted for one trophic level fractionation (+0.4‰ δ13C, +3.4‰ for δ15N) and potential organic matter sources for the upper estuary (stations 5-14). Error bars represent one standard deviation. The sources considered relevant for Corbicula fluminea (closed square; each represents a station-specific average) in this portion of the estuary were freshwater phytoplankton (FP; open triangle), microphytobenthos (MPB; open circle), sediment organic matter (SOM; closed triangle) and particulate organic matter (POM; closed circle). 116 Fig. 4.8 Proportion of each food source to Corbicula fluminea’s biomass collected in the upper estuary (stations 5–14) based on the stable isotope mixing model. The food sources included in the model were freshwater phytoplankton (FP), particulate organic matter (POM), sediment organic matter (SOM) and microphytobenthos (MPB). Boxes indicate 50%, 75% and 95% Bayesian credibility intervals. 117 xiv xv Tables and appendix index Table 2.1 Stable isotope ratios of organic matter sources commonly found in estuaries. 29 Table 2.2 Characterization of the POM pool in the Minho River estuary, along the salinity mixing gradient in September 2010, March and August 2011. 37 Table 2.3 DistLM model based on the δ13C and δ15N values of zooplankton and Corbicula fluminea and fitted abiotic data. 38 Appendix A. δ15N and δ13C values of zooplankton taxa collected in September 2010, March 2011 and August 2011 in the tidal freshwater (TFW), brackish and marine portions of the Minho River estuary. 58-59 Table 3.1 Mean (±SD) δ13C and δ15N values (‰) for particulate organic matter (POM), microphytobenthos (MPB), macroalgae (mixture of Chlorophyceae and Phaeophyceae), emergent aquatic vegetation (EAV; mixture of plants from the margin), submerged aquatic vegetation (SAV; Elodea canadensis) and terrestrial plants collected in the Minho River estuary during winter, spring and summer of 2011 in stations (S) S1, S2, S3 and in the tidal freshwater (TFW) area. 78 Table 4.1 Mean water quality in the estuarine mixing zone of the Minho river (including salinity, chlorophyll a (Chl a), δ13C value of total dissolved inorganic carbon (δ13CDIC)), as well as the mean character of water column POM samples (including particulate organic carbon (POC) concentration (mg L-1), POM carbon: nitrogen molar ratios (C/N), and phytoplankton fraction (Fphyto; %) in POM samples). Average δ13C and δ15N values are reported for the organic matter sources measured, including particulate organic matter (POM: POC, PN), phytoplankton (phyto), microphytobenthos (MPB) and sediment organic matter (SOM: SOC, SN). The stations located in either the tidal freshwater area (TFW; stations 5 to 14) or brackish (stations 1 to 4) and polyhaline estuary (marine; first 3 km away from the mouth) were pooled and the means calculated. Values in parentheses represent one standard deviation. 110 xvi Table 4.2 Average δ13C (‰) and δ15N (‰) (± SD) values used in the Corbicula fluminea SIAR mixing model for the organic matter sources used in either the upper estuary (UE) or lower estuary (LE) model. These values represent average bottom values for phytoplankton and particulate organic matter (POM), and also the average values for microphytobenthos (MPB) and sediment organic matter (SOM) pooled from each group of stations. 115 xvii Abstract Estuarine food webs function within a mosaic of habitats and are supported by a mix of primary producers from both local and distant sources. Understanding the factors that govern the exchange and consumption of different organic matter (OM) sources is of paramount importance for the management and conservation of estuaries, because secondary production and food web diversity are thought to depend on the quality of the basal sources fuelling it. Therefore the studies presented in this PhD thesis aimed to improve the knowledge on the role of material flows in estuarine food webs, using the River Minho estuary as a model ecosystem, and carbon and nitrogen stable isotope analyses for estuarine biogeochemical characterization. In the first chapter, the most common analytical approaches used in the study of estuarine food webs were reviewed around three main subjects: 1) methods used to identify the pools that support consumers; 2) characterization of end-member (i.e. sources); 3) dominant assumptions in the analysis of food webs. The most commonly used approaches are qualitative approaches, combined with basic statistical analyses (e.g. ANOVA), although there has been an increase in the use of quantitative approaches such as linear mixing models (e.g. IsoError, IsoSource, SIAR). The quantitative approaches more commonly used in the study of estuarine food webs are the IsoError (determined model) and IsoSource (undetermined model) mixing models. Also, the potential of all the existing qualitative and quantitative approaches depends on the quality of the input data and on the proper acknowledgment of the assumptions/limitations of each approach. Thus, the proper characterization of the end-members and baselines is crucial for the characterization of food web dynamics. In the second chapter, a high spatial resolution study was conducted in the Minho River estuary to characterize the contribution of different OM sources to the lower food web, under naturally varying river discharge scenarios. It was possible to conclude that Minho River estuarine habitats are connected with each other and with its end-members. The degree of connectivity and the importance of allochthony varied with the river discharge and with the occurrence of a major winter flood. The contribution of terrestrialderived OM to the POM pool was higher during high river discharge conditions, and during a summer following an unusual winter flood. In the third chapter, it was verified that there was a temporal and spatial variability in the Minho river estuarine benthic food web dynamics. Spatial variability was essentially related with the characteristics of the organic matter sources available in each estuarine area, which was also influenced by the proximity to other ecosystems (i.e. higher terrestrial influence in the tidal freshwater portion and higher marine influence in the xviii brackish portion of the estuary). Temporal changes in the stable isotope ratios of consumers were driven essentially from changes occurring at the base of the food web, although there were also some evidences for a combination with changes in the resources used by consumers. River discharge was also an important factor influencing the changes at the base of the estuarine food web, both spatially and temporally. In the fourth chapter were presented evidences of how man induced activities, in particular the introduction of invasive species, can potentially change estuarine food webs. The Corbicula fluminea’s functional response was analyzed in terms of feeding behavior and food selectivity, using the natural variation in OM sources that occur in estuarine environments. It was possible to conclude that C. fluminea has the ability to adapt to environments with low food quality because it can consume terrestrial-derived OM. This can be a competitive adaptation in systems with perennial low food quality such as the Minho River estuary. Moreover, its ability to couple benthic and pelagic environments and terrestrial ecosystems demonstrates a strong potential to alter food web flows in aquatic ecosystems. In conclusion, Minho estuary food web relies on a strong connectivity between the different components, namely between the pelagic and benthic compartments, and also between the estuarine, terrestrial and marine ecosystems. Yet, further studies will be needed to disclose the impacts of disrupting ecosystems connectivity on estuarine productivity. xix Resumo As cadeias tróficas estuarinas estão inseridas em diferentes tipos de habitats, sendo suportadas por um conjunto de produtores primários com origem local ou provenientes de outras áreas. Para uma adequada gestão e conservação dos estuários, é essencial compreender quais os fatores que condicionam os movimentos da matéria orgânica nestes ecossistemas, bem como a sua utilização pelos consumidores, pois a produção secundária e a diversidade nas cadeias tróficas, depende da qualidade das fontes de matéria orgânica que as suportam. Assim, com os trabalhos científicos presentes nesta tese de Doutoramento, espera-se contribuir para uma melhor compreensão do papel que a dinâmica da matéria orgânica tem no funcionamento das cadeias tróficas estuarinas. Para realizar este trabalho, utilizou-se o estuário do rio Minho como sistema modelo e para a sua caracterização biogeoquímica, efetuaram-se análises de isótopos estáveis de carbono e de azoto. No primeiro capítulo foi feita uma revisão aos métodos mais utilizados na análise e interpretação de dados no estudo das cadeias tróficas estuarinas, tendo esta sido estruturada da seguinte forma: 1) métodos mais utilizados na identificação das fontes que suportam a produção dos consumidores; 2) caracterização das fontes de energia; 3) principais pressupostos na análise de cadeias tróficas. As abordagens mais utilizadas são as qualitativas, combinadas com análises estatíticas (e.g. ANOVA), embora a utilização de métodos quantitativos (e.g. IsoError, IsoSource e SIAR), tenha aumentado nos últimos anos. As abordagens quantitativas mais utilizadas no estudo das cadeias tróficas estuarinas são os modelos IsoError (modelo determinado) e o IsoSource (modelo indeterminado). Não obstante, o potencial de qualquer uma das abordagens existentes atualmente, depende da qualidade dos dados que são utilizados e do reconhecimento dos pressupostos e das limitações inerentes a cada uma delas. Assim, uma correta caracterização de todas as fontes possíveis é crucial, para a correta avaliação da dinâmica das cadeias tróficas estuarinas. No segundo capítulo, foi realizado um estudo com elevada resolução espacial no estuário do rio Minho, por forma a caracterizar o efeito que diferentes níveis de caudal têm na contribuição das fontes de matéria orgânica para os níveis tróficos inferiores. Concluiu-se que os habitats no estuário do rio Minho estão interligados entre si e também com os ecossistemas adjacentes. O grau de conetividade e a importância dos subsídios alóctones variaram com o caudal. A contribuição de origem terrestre para a matéria orgânica particulada na água, foi mais elevada durante períodos em que o caudal foi também mais elevado, e durante um verão de um ano em que ocorreram cheias no inverno. 2 Abstract Stable isotopes have emerged as a useful tool to understand the structure and dynamics of estuarine food webs, because they provide time-integrated information about the trophic relationships and energy flow through food webs. Two main approaches have been used to analyze and interpret stable isotope data in the context of estuarine food webs: qualitative and quantitative approaches. In the qualitative approaches, the analysis is based on the inspection of the position occupied by the stable isotope ratios of a certain consumer in relation to the stable isotope ratios of its most likely sources. In the quantitative approaches, it is calculated the proportional contribution of each source to the consumer’s biomass. Based on the analysis of studies addressing the trophic interactions in estuarine food webs between 2000 and 2011, it was possible to verify that the most commonly used approach is still the qualitative, combined with statistical analysis (e.g. ANOVA). Nonetheless there has been an increase in the use of quantitative approaches such as linear mixing models (e.g. IsoError, IsoSource, SIAR). Also, when analyzing multiple trophic levels, researchers tend to estimate the stable isotope ratios of phytoplankton, by establishing the isotopic baseline from some other biological material near the base of the food web or by using estimates from other ecosystems. Further, the interpretation of stable isotope data generally relies on several assumptions, such as that the isotopic composition of consumer’s tissues equals the weighted average of the isotopic composition of its sources, and trophic fractionation moves constantly through the food web. Since several studies contradict those assumptions, the assumptions should be acknowledge when interpreting trophic interactions based on stable isotope data in the absence of species-specific estimates. Keywords: isotope mixing model, baseline, trophic fractionation, assumptions. 3 1.1 Introduction Estuaries are among the most productive ecosystems in the world, and are characterized by complex trophic dynamics associated with their diverse habitats and organic matter (OM) sources. Earlier studies of trophic dynamics in estuaries were usually based on consumers’ stomach content analysis (Whitfield 1988). Although this can reveal consumers’ diets and trophic relationships with higher taxonomic resolution, they do not provide information about assimilation. Stable isotopes emerged as a useful tool to understand the structure and dynamics of estuarine food webs, because they provide time-integrated information about the trophic relationships and energy flow through food webs (Pasquaud et al. 2007). The first applications of stable isotopes in food web studies were qualitative, providing inferences about the relationships between consumers and their resources based on inspections of species’ position in bi-plots (e.g. Haines 1976, Peterson et al. 1985). During the last 20 years, quantitative approaches have emerged to quantify the differences in the stable isotopes ratios of individuals, species or others (e.g. Hsieh et al. 2002, Richoux and Froneman 2007, Vinagre et al. 2011), and more complex mixing models to determine dietary contributions of different sources to a certain consumer (e.g. Phillips 2001, Lubetkin and Simenstad 2004, Parnell et al. 2010). These approaches have dramatically improved our understanding on the complexity of estuarine food webs, providing new insights on its spatial and temporal variability (e.g. Chanton and Lewis 2002, Baeta et al. 2009, Atwood et al. 2011, Antonio et al. 2012), food chain length (e.g. Akin and Winemiller 2008), and changes in the food web subsidies and structure induced by human activities (e.g. Cole et al. 2004, Wu et al. 2009). Nonetheless the use of more quantitative approaches has raised a prolific debate due to the common underlying assumptions of each approach, especially how they incorporate sources’ and consumers’ variability or uncertainty (Phillips and Gregg 2001, 2003, Parnell et al. 2010, Semmens et al. 2013). In this chapter we provide an overview on the most common analytical approaches used to interpret stable isotope data on food webs, structured around three main subjects: 1) methods used to identify the pools that support consumers’ biomass; 2) characterization of end-members (i.e. sources); 3) dominant assumptions in food web analysis. To identify the most common analytical approaches used to interpret stable isotope data in estuarine food web studies we analyzed the papers published between 2000 and 2011 and cataloged in Scopus. The papers were identified using the keywords “food web”, “estuary” and “stable isotopes”. 4 1.2 Stable isotope ratios and estuarine food webs Stable isotopes are atoms of the same element with different masses, because of differences in the number of neutrons. Atomic mass is essentially the sum of the masses of the protons and neutrons (each of a weight equal to about 1 atomic mass unit), because electrons have a relatively insignificant mass. For instance, the common isotope of carbon has six protons and six neutrons. It thus has an atomic mass of about 12, designated as 12C. It has a rare isotope with seven neutrons, designated as 13C. Stable isotopes, unlike radioisotopes, do not decay (Criss 1999). The most commonly used elements in food web studies are nitrogen (N), carbon (C) and sulphur (S). The C isotope ratios (δ13C: 13C/12C) are useful to discriminate between primary producers with different photosynthetic pathways (e.g. C3 and C4 plants), and since they exhibit little or no differences between trophic levels, they are useful to determine the original sources of dietary C (Peterson and Fry 1987). Similarly, the S isotope ratios (δ34S: 34S/33S) vary substantially among primary producers, but it changes little between trophic levels and therefore are also useful tracers to identify organic matter sources (Peterson et al. 1985, Peterson and Howarth 1987). These stable isotope ratios are particularly useful in the identification of organic matter (OM) sources in estuaries, because estuaries receive inputs from multiple sources, including riparian vegetation, marsh vegetation, submerged and emergent aquatic vegetation and phytoplankton (Cloern et al. 2002, Hoffman and Bronk 2006). Riparian plants that utilize the C3 pathway have a δ13C of about -28‰ because there is an uptake fractionation of about -21‰ over atmospheric CO2 (δ13C: -7‰) (Smith and Epstein 1971). In contrast, plants associated with the C4 pathway (e.g. Spartina spp.) are more 13Cenriched (δ13C: -13‰) owing to reduced fractionation (Smith and Epstein 1970, Fry and Sherr 1984). Freshwater and estuarine phytoplankton will also differ, because they utilize isotopically distinct pools of dissolved inorganic carbon (DIC). In general, microphytobenthos (MPB) are more 13Cenriched than phytoplankton, due to the existence of a diffusive boundary layer at the sediment-water interface that reduces isotopic fractionation (France 1995). Upland plants, phytoplankton and marsh plants also derive their sulfate from different origins, thus producing distinguishable δ34S values. Upland plants use the sulfate available from the rain water (δ34S: 2‰ to 8‰), while estuarine phytoplankton derived their sulfate from seawater (δ34S ~ +18‰). Marsh plants derive sulfite from a more variable and usually light δ34S inorganic sulfur pool in the sediments (δ34S typically -10‰ to +5‰). For MPB the values are intermediate (δ34S: 4‰ to 14‰), depending upon the proportions of reduced sulfur and seawater at the sediment/water interface (Peterson and Howarth 1987, Sullivan and Moncreiff 1990, Currin et al. 1995, Newell et al. 1995). Thus, stable C and S isotopes 5 can help to differentiate between benthic and pelagic environments and between vascular plant and phytoplankton sources. The N isotope ratios (δ15N: 15N/14N) become enriched in 3-4‰ between two consecutive trophic levels and therefore are a powerful tool to estimate consumers’ trophic position (Minagawa and Wada 1984, Peterson and Fry 1987, Vander Zanden et al. 1999). They can also help to separate 15N-depleted terrestrial OM (δ15N: -4‰ to 4‰) from 15N-enriched aquatic OM sources (δ15N: 6‰ to 10‰) (Peterson and Fry 1987, Cloern et al. 2002). However, δ15N can be altered by changes in the isotope value of N substrates due to preferential uptake of isotopically light N (Cifuentes et al. 1988), nitrification, denitrification (Mariotti et al. 1981) or by microbially mediated degradation (Miyake and Wada 1971; Altabet 1988). Also the form and amount of anthropogenic nitrogen addition can influence the baseline δ15N. Nitrate from atmospheric sources typically have δ15N values varying between 0‰ and 8‰ (Kreitler and Jones 1975, Kellman and HillaireMarcel 2003). The δ15N values of nitrate from synthetic fertilizers range from -3‰ to 3‰ (Macko and Ostrom 1994, Kellman and Hillaire-Marcel 2003) and nitrate from human and animal waste typically have δ15N values varying from 9‰ to 25‰ (Kreitler and Browning 1983, Rolston et al. 1996). We conducted a literature survey, on Scopus, to assess the number of studies published that investigated the estuarine food webs dynamics using stable isotopes, from 2000 to 2011. In this period, a total of 113 studies were published. The number of publications increased through time, with the most numbers of papers published in 2011 (n= 17) (Fig. 1.1). Overall, North America (United States and Canada), Europe, and Australia produced the majority of the publications with 33%, 13% and 12% of the total, respectively. The main topics studied were the identification of OM sources supporting the estuarine food web and their structure (89% of total), although other subjects were evaluated, such as the effects of anthropogenic inputs to the food web and consumers’ movements. The stable isotope ratios most commonly used were δ13C and δ15N combined (52% of total), though in some studies one of those, or both, were used in combination with other elemental tracers, such as δ34S or fatty-acids. Details on the main approaches used to characterize estuarine food webs are discussed below. 6 Year 200020012002200320042005200620072008200920102011 Number of publications 0 2 4 6 8 10 12 14 16 18 Fig 1.1 Number of studies published between 2000 and 2011 and indexed at Scopus, using the keywords “food web”, “estuary” and “stable isotopes”. 1.3 Methods to identify which sources support consumers in food web studies There were two main approaches used to identify the resources supporting the production of consumers: qualitative and quantitative approaches. In a qualitative approach, the analysis is based on the inspection of the position occupied by the stable isotope ratios of a certain consumer, usually in a bi-plot (2 tracers), in relation to the stable isotope ratios of its most likely sources. In a quantitative approach, the proportional contribution of each source is estimated using a simple calculation or a mixture of more complex equations. 1.3.1 Qualitative methods The most used method to identify trophic relationships in estuarine food webs, was the qualitative approach combined with statistical analyses (i.e. t-test, ANOVA, ANOSIM), which accounted for 61% of the papers published. Though we recognize that a visual analysis should always be done to constrain the possible sources to a consumer, this methodology is particularly limiting when studying complex food webs (i.e. multiple sources, several trophic levels and feeding strategies). Also, statistical analyses can be very informative in quantifying the differences in the stable isotope ratios of producers and/or consumers, or in assessing spatial and temporal variability in the stable isotope 7 ratios of a certain consumer or group of consumers. Yet, the strength of statistical analyses depends on the sample size and on within-group variability. As the number of potential food sources increases, the ability to identify dietary contributions becomes more difficult. So, in the last decades, several models have been developed to identify and quantify the importance of various food sources to the consumers’ biomass. 1.3.2 Quantitative methods Several approaches have been proposed to quantify the contribution of multiple sources to consumers’ biomass: geometric (Ben-David et al. 1997), linear mixing models (Phillips 2001, Lubtekin and Simenstad 2004, Hall-Aspland et al. 2005, Parnell et al. 2010), and spatially-based approaches (Rasmussen 2010). The geometric approaches use δ values to determine Euclidean distances between consumers and sources, and an inverse relationship is assumed between those distances and the contribution of each source to consumers’ diet, i.e. the shorter the distance, the greater the contribution (Ben-David et al. 1997). However, this method overestimate rare food sources (Ben-David et al. 1997), and the equations provided are inaccurate to identify dietary contributions (Phillips 2001). For this reason, geometric approaches have been replaced by more recent quantitative approaches, such as the linear mixing models. In fact, the most commonly used quantitative approaches, in the surveyed studies of estuarine food webs, were the linear mixing models74% of those that used quantitative approaches. Since linear mixing models are the most widely used models, they will be discussed in more detail. The linear mixing models have become a fundamental method to quantify the contribution of various data sources to a consumer based on their respective stable isotope ratios. These models use a set of mass-balance equations to estimate the proportional contribution of each food source to a consumer, assuming that all sources sum to 1 (Phillips 2001). For example, in a system with n tracers and n+1 sources, solving the mass-balance equations will give the exact proportional contribution of each source (Phillips 2001). Assuming that we have two stable isotope ratios (δ15N, δ13C) and three sources (A, B, C), the equations would be represented as follows: δ13CM= fA.δ13CA + fB.δ13CB + fc.δ13CC (1) δ15NM= fA.δ15NA + fB.δ15NB + fc.δ15N (2) fA + fB + fC =1 (3) 8 where δM is the isotopic composition of a consumer’s tissue and fA, fB and fC are the proportional contributions of sources A, B and C. These are the main principles underlying the linear mixing model IsoError (Phillips and Gregg 2001), which takes into account the source’s and consumer’s stable isotope ratios variability. An alternative formulation, the IsoConc linear mixing model (Phillips and Koch 2002) also considers the elemental concentration (e.g. %C,% N) of each source to estimate its proportional contribution to the consumer. However, most food webs are very complex, with the number of sources frequently exceeding the number of tracers plus one. In this situation, researchers can constrain the possible sources using prior information to allow a determinate model (e.g. Schwamborn et al. 2002), or they will have to use an indeterminate model. The latter implies that there is not a unique solution to the model. The most commonly used mixing model is IsoSource40% (Phillips and Gregg 2003). This model estimates all multivariate solutions that fit the model bounding conditions (all contributors sum to 1) within a specific tolerance, and then outputs the probability distribution of the all univariate estimates under the domain of acceptable multivariate solutions. However, IsoSource does not explicitly incorporate uncertainty in the variables. The formula only requires the mean values of consumers, diet sources and trophic fractionation; the tolerance term implicitly recognizes variability in the diet sources. Alternative modelling approaches were developed to estimate proportional source contributions including SOURCE and STEP (Lubtekin and Simenstad 2004) or the Moore Penrose pseudoinverse model (Hall-Aspland et al 2005). SOURCE is used to estimate consumers´ direct and indirect uptake of autotrophic sources and consumer trophic levels (Lubtekin and Simenstad 2004). STEP estimates a consumer’s diet, which may include autotrophs or heterotrophs, or both (Lubtekin and Simenstad 2004). Both models require that the diet source isotope ratios included do not overlap, so they use a nearest neighbor distance measurement (NND2) to determine whether two sources are sufficiently distinct to be considered individually in model calculations (Lubtekin and Simenstad 2004). These models identify the outer bounds of possible mixtures instead of examining every possible biological solution as with IsoSource, making it computationally less demanding (Lubtekin and Simenstad 2004). The Moore Penrose pseudo-inverse model attempts to provide a unique solution of source contribution to a consumer using a single isotopic tracer and matrix algebra (HallAspland et al 2005). However, the model yields a similar solution to the mean generated by IsoSource (Hall-Aspland et al 2005), but fails to acknowledge other feasible solutions as with IsoSource. 9 None of these indeterminate models allows for uncertainty in the variables because they all require mean values without distribution information for the stable isotope ratios of dietary sources and consumers, as well as for the estimates of trophic fractionation and trophic level. Alternative approaches, such as the Bayesian mixing models MixSIR and SIAR, have emerged to cope with these limitations. Bayesian inference incorporates variability in all the inputs, while allowing for multiple dietary sources and generates potential dietary solutions as assigned probability distributions (Parnell et al. 2010). SIAR (Parnell et al. 2010) and MixSIR (Moore and Semmens 2008) basically use the same approach, aside from minor differences in the fitting algorithms implemented (SIAR uses Markov Chain Monte Carlo sampling methods and MixSIR uses Sample Importance Resampling) (Moore and Semmens 2008, Parnell et al. 2010). The main difference between these methods is that SIAR includes a residual error term (Parnell et al. 2010). Possible unknown sources of error can be related to the physiology or ecology of the consumers studied. For instance, if the consumer under study goes through ontogenic shifts, the residual error likely will be high and thus, further analysis (e.g. literature review and/or basic statistics such as segmented linear regressions analysis) and modelling are required. An extension of the Bayesian approach has been developed to incorporate intrapopulation variability, where the variance in the diet of individual consumers across multiple levels of population structure can be acknowledged (Semmens et al. 2009). There are some limitations that should be considered before applying Bayesian mixing models, namely: 1) dietary source variability and trophic fractionation may not be normally distributed, 2) Bayesian mixing models do not assume isotope routing (i.e. isotopes can be assimilated differently within the body of the consumer (which is a caveat in all the food web approaches), 3) the model will always attempt to fit a model, even if the sources lie outside the isotopic mixing polygon (Parnel et al. 2010). Another problem, raised by Fry (2013), is that likelihood densities can be multimodal in underdetermined mixing models, so the credibility intervals estimated by the Bayesian mixing models (usually 95%) can be misleading. Several strategies have been proposed to reduce multimodality: incorporate more consumer data or prior information, change source geometry by re-grouping sources and include residual error (Semmens et al. 2013 and references therein). One common problem that researchers have to deal with is the interpretation of the output of indeterminate mixing models. The most intuitive approach is to interpret measures of central tendency such as means, median or modes (most likely values), but this is not justified by the structure of the most common mixing models and when used alone they can be misleading. Although some unique solutions may be more likely than 10 others, it is advocated that the outputs should be characterized by the distribution of possible solutions (Phillips and Gregg 2003, Semmens et al. 2013). Several approaches have been suggested to solve indeterminate mixing models: a) statistically subsample the range of feasible solutions and then calculate averages from the subsample (approach used in IsoSource); b) geometric sector approach, where the contribution of each source is inversely proportional to sector areas in a two dimensional space; c) extension of the previous approach, where a secondary polygon (danger zone) is included and defined based on halfway points between the exterior polygon and a centroid. The mixture data points falling within this danger zone are dominated by the assumed centroid contributions and little constrained by the measured data. All these approaches are detailed by Fry (2013). However the graphical analysis proposed by Fry (2013) has no statistical inferences, ignores uncertainty in both sources and mixture data, and do not provide probabilistic estimates of source contributions (Semmens et al. 2013). Nonetheless, other approaches can be used to try constraining mixing models, such as adding more tracers or conduct a prior source selection for aggregating and reducing the number of sources (Philips and Gregg 2003) based, for instance, on consumer’s feeding behavior. Although some models like SOURCE and STEP may help to decide which sources can be aggregated, there are some examples for which the existing models do not seem to have a satisfactory solution. For example, in a study aiming to identify the main dietary sources of the invasive bivalve Corbicula fluminea (Dias et al. in press), the authors identified five possible dietary sources based on the bivalves’ feeding modes: estuarine phytoplankton (EP), marine phytoplankton (MP), particulate organic matter (POM; mixture of marine and freshwater POM), sediment organic matter (SOM) and microphytobenthos (MPB) (Fig. 1.2). The MP and MPB have similar isotope ratios, similar C/N values (~7) and both are possible sources, based on a first inspection of the bi-plot (i.e. some C. fluminea present high δ15N values). All the existing mixing models would have difficulties to solve these two sources, because they are highly correlated, i.e. the increasing proportional contribution of MP would implicate the decrease in the proportional contribution of MPB, and vice-versa. The aggregation of these sources would not be realistic because they represent completely different compartments of the system (pelagic vs. benthic), and the use of one or the other would imply differences in feeding strategies (filter feeding vs. pedal feeding). In this case, the existing knowledge on the species’ physiology was crucial to discard MP as a likely source (Dias et al.in press). 11 Fig. 1.2 Average δ13C and δ15N values of Corbicula fluminea adjusted for one trophic level fractionation (+0.4‰ δ13C, +3.4‰ for δ15N) and potential organic matter sources in the brackish estuary. Error bars represent one standard deviation. The sources considered relevant for Corbicula fluminea (closed square; each represents a station-specific average), in this portion of the estuary, were marine phytoplankton (MP; closed diamond), estuarine phytoplankton (EP; open triangle), microphytobenthos (MPB; open circle), sediment organic matter (SOM; closed triangle) and particulate organic matter (POM; closed circle). Figure in Dias et al. (in press). Other models use spatial variability to estimate sources’ proportional contribution when their isotope ratios are not necessarily distinct (Rasmussen 2010). This approach may be useful in systems where sources’ variability is predictable across some spatial axis (e.g. altitude, river distance), which use the slopes of change along the spatial axis to estimate sources’ contribution to the consumer by assuming that the consumer’s stable isotope ratios is a weighted mixture of the sources along the linear gradients (Rasmussen 2010, Gray et al. 2011). The main disadvantages of this approach are that the proportions of the sources in a consumer’s diet must be constant along the gradient, and the knowledge about the isotope gradients may be difficult to acquire (Layman et al. 2012). 1.4 Characterization of end-members It is critical to properly characterize the end-members for stable isotope-based food web analyses. End-members are the source materials used in the mixing models to determine the proportional contribution of the two sources, or more, comprising the mixture being evaluated (Peterson 1999). For example, when studying the diet of a suspension feeding organism, all the sources available in the water column must be sampled. The number of end-members may increase with the complexity of the 18 Boyd CE, Goodyear CP (1971) Nutritive quality of food in ecological systems. Arch Hydrobiologia: 256-279. Burkhardt S, Riebesell U, Zondervan I (1999) Effects of growth rate, CO2 concentration, and cell size on the stable carbon isotope fractionation in marine phytoplankton. Geochim Cosmochim Acta 63: 3729-3741. Cabana G, Rasmussen JB (1996) Comparison of aquatic food chains using nitrogen isotopes. P Natl Acad Sci 93: 10844-10847. Caut S, Angulo E, Courchamp F (2009) Variation in discrimination factors (δ15N and δ13C): the effect of diet isotopic values and applications for diet reconstruction. J Appl Ecol 46: 443-453. Chanton J, Lewis FG (2002) Examination of coupling between primary and secondary production in a river-dominated estuary: Apalachicola Bay, Florida, USA. Limnol Oceanogr 47: 683–697. Cifuentes LA, Sharp JH, Fogel ML (1988) Stable carbon and nitrogen isotope biogeochemistry in the Delaware estuary. Limnol Oceanogr 33: 1102-1115. Cloern JE, Canuel EA, Harris D (2002) Stable carbon and nitrogen isotope composition of aquatic and terrestrial plants of the San Francisco Bay estuarine system. Limnol Ocenogr 47: 713-729. Cole ML, Valiela I, Kroeger KD, Tomasky GL, Cebrian J, Wigand C, McKinney RA, Grady SP, Carvalho Da Silva MH (2004) Assessment of a δ15N Isotopic Method to Indicate Anthropogenic Eutrophication in Aquatic Ecosystems. J Environ Qual 33: 124-132. Criss RE (1999) Principles of stable isotope distribution. Oxford University Press, pp.254. Currin CA, Newell SY, Paerl HW (1995) The role of standing dead Spartina alterniflora and benthic macroalgae in salt marsh food webs: considerations based on multiple stable isotope analysis. Mar Ecol Prog Ser 121: 99-116. DeNiro MJ, Epstein S (1981) Influence of diet on the distribution of nitrogen isotopes in animals. Geochim Cosmochim Acta 45: 341-351. Dias E, Morais P, Antunes C, Hoffman JC (in press) Linking terrestrial and benthic estuarine ecosystems: organic matter sources supporting the high secondary production of a non-indigenous bivalve. Biol Invasions. DOI: 10.1007/s10530-014-0655-8. Dierking J, Morat F, Letourneur Y, Harmelin-Vivien M (2012) Fingerprints of lagoonal life: migration of the marine flatfish Solea solea assessed by stable isotopes and otolith microchemistry. Est Coast Shelf Sci 104-105: 23-32. Eddins SGN (2001) Calculation of terrestrial inputs to particulate organic carbon in an anthropogenically impacted estuary in the southeastern United States. Water Air Soil Poll 127: 227-241. 19 Finlay JC, Doucett RR, McNeely C (2010) Tracing energy flow in stream food webs using stable isotopes of hydrogen. Freshwater Biol 55: 941-951. França S, Vasconcelos RP, Tanner S, Máguas C, Costa MJ, Cabral HN (2011) Assessing food web dynamics and relative importance of organic matter sources for fish species in two Portuguese estuaries: a stable isotope approach. Mar Environ Res 72: 204-215. France RL (1995) Carbon-13 enrichment in benthic compared to planktonic algae: foodweb implications. Mar Ecol Prog Ser 124: 307-312. Fry B (2013) Alternative approaches for solving underdetermined isotope mixing problems. Mar Ecol Prog Ser 472: 1-13. Fry B, Allen YC (2003) Stable isotopes in zebra mussels as bioindicators of riverwatershed linkages. River Res. Applic. 19: 683-696. Fry B, Arnold C (1982) Rapid 13C/12C turnover during growth of brown shrimp (Penaeus aztecus). Oecologia 54: 200-204. Fry B, Sherr EB (1984) δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contr mar Sci 27: 15-47. Gannes LZ, O’Brien DM, Martínez del Rio C (1997) Stable isotopes in animal ecology: assumptions, caveats, and a call for more laboratory experiments. Limnol Oceanogr 78: 1271-1276. Gray DP, Harding JS, Elberling B, Horton T, Clough TJ, Winterbourn MJ (2011) Carbon cycling in floodplain ecosystems: out-gassing and photosynthesis transmit soil δ13C gradient through stream food webs. Ecosystems 14: 583-597. Haines EB (1976) Stable carbon isotope ratios in the biota, soils and tidal water of a Georgia salt marsh. Estuarine and Coastal Marine Science 4: 609-616. Hall-Aspland SA, Hall AP, Rogers TL (2005) A new approach to the solution of the linear mixing model for a single isotope: application to the case of an opportunistic predator. Oecologia 143: 143-147. Hobson KA, Sease JL (1998) Stable isotope analyses of tooth annuli reveal temporal diet records: an example using Steller sea lions. Mar Mammal Sci 14: 116-129. Hoeinghaus DJ, Vieira JP, Costa CS, Bemvenuti CE, Winemiller KO, Garcia AM (2011) Estuary hydrogeomorphology affects carbon sources supporting aquatic consumers within and among ecological guilds. Hydrobiologia 673: 79-92. Hoffman JC, Bronk DA (2006) Interannual variation in stable carbon and nitrogen isotope biogeochemistry of the Mattaponi River, Virginia. Limnol Oceanogr 51: 2319-2332. Hsieh H-L, Chen C-P, Chen Y-G, Yang H-H (2002) Diversity of benthic organic matter flow through polychaetes and crabs in a mangrove estuary: δ13C and δ34S signals. Mar Ecol Prog Ser 227: 145-155. 20 Jim S, Jones V, Ambrose SH, Evershed RP (2006) Quantifying dietary macronutrient sources of carbon for bone collagen biosynthesis using natural abundance stable carbon isotope analysis. Brit J Nutr 95: 1055-1062. Kellman L, Hillaire-Marcel C (2003) Evaluation of nitrogen isotopes as indicators of contamination sources in agricultural watershed. Agric Ecosyst Environ 95: 87-102. Kelly LJ, Martinez del Rio C (2010) The fate of carbon in growing fish: an experimental study of isotopic routing. Physiol Biochem Zool 83: 473-480. Koch PL (2007) Isotopic study of the biology of modern and fossil vertebrates. In: Michener R, Lajtha K (eds) Stable Isotopes in Ecology and Environmental Science. Blackwell Publishing, Boston, pp. 99-154. Kreitler CW, Browning LA (1983) Nitrogen-isotope analysis of groundwater nitrate in carbonate aquifers: natural sources versus human pollution.J Hydrol (Amst) 61. 285-301. Kreitler CW, Jones DC (1975) Natural soil nitrate: the cause of the nitrate contamination of ground water in Runnels County, Texas. Ground Water 13: 53-61. Layman CA, Araujo MS, Boucek R, Hammerschlag-Peyer CM, Harrison E, Jud ZR, Matich P, Rosenblatt AE, Vaudo JJ, Yeager LA, Post DM, Bearhop S (2012) Applying stable isotopes to examine food-web structure: an overview of analytical tools. Biol Rev 87: 545-562. Laws EA, Bidigare RR, Popp BN (1997) Effect of growth rate and CO2 concentration on carbon isotopic fractionation by marine diatom Phaeodactylum tricornutum. Limnol Oceanogr 42: 1552-1560. Liu K-K, Kao S-J, Chiang K-P, Gong G-C, Chang J, Cheng J-S, Lan C-Y (2013) Concentration dependent nitrogen isotope fractionation during ammonium uptake by phytoplankton under an algal bloom condition in the Danshuei esuary, northern Taiwan. Mar Chem 157: 242-252. Lubetkin SC, Simenstad CA (2004) Multi-source mixing models to quantify food web sources and pathways. J Appl Ecol 41: 996-1008. Macko SA, Ostrom NE (1994) Pollution studies using stable isotopes. In: Lajtha K, Michener RH (eds) Stable isotopes in ecology and environmental science. Blackwell Scientific Publications, Oxford, p. 45-62. Maier GO, Simenstad CA (2009) The role of marsh-derived macrodetritus to the food web of juvenile chinook salmon in large altered estuary. Estuar Coasts 32: 984-998. Mariotti A, Germon JC, Hubert P, Kaiser P, Letolle R, Tardieux A, Tardieux P (1981) Experimental determination of nitrogen kinetic isotope fractionation: some principles; ilustration for the denitrification and nitrification processes. Plant Soil Sci 62: 413-430. Marty J, Planas D (2008) Comparison of methods to determine algal δ13C in freshwater. Limnol Oceanogr: Methods 6: 51-63. 21 McCarthy JJ (1980) Nitrogen In: Morris I (ed) The physiological ecology of phytoplankton. Blackwell Scientific p. 191-233. McCutchan JH, Lewis Jr WM, Kendall C, McGrath CC (2003) Variation in trophic shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos 102: 378-390. Minagawa M, Wada E (1984) Stepwise enrichment of 15N along food chains: further evidence and the relation between δ15N and animal age. Geochim Cosmochim Acta 48: 1135-1140. Miyake Y, Wada E (1971) The isotope effect on the nitrogen in biochemical, oxidation-reduction reactions. Rec Oceanogr Works Jpn 11: 1-6. Mook WG, Tan FC (1991) Stable carbon isotopes in rivers and estuaries. In: Degens ET, Kemp S, Richey JE (eds.) Biogeochemistry of major world rivers scope 42, John Wiley, Chichester. Moore JW, Semmens BX (2008) Incorporating uncertainty and prior information into stable isotope mixing models. Ecology Letters 11: 470-480. Mulkins LM, Jelinski DE, Karagatzides JD, Carr A (2002) Carbon isotope composition of mysids at a terrestrial-marine ecotone, Clayoquot Sound, British Columbia, Canada. Est Coast Shelf Sci 54: 669-675. Needoba J A, Waser NA, Harrison PJ, Calvert SE (2003) Nitrogen isotope fractionation in 12 species of marine phytoplankton during growth on nitrate. Mar Ecol Prog Ser 255: 81-91. Newsome SD, Fogel ML, Kelly L, Martinez del Rio C (2011) Contributions of direct incorporation from diet and microbial amino acids to protein synthesis in Nile tilapia. Funct Ecol 25: 1051-1062. Newsome SD, Tinker MT, Monson DH, Oftedal OT, Ralls K, Staedler MM, Fogel ML, Estes JA (2009) Using stable isotopes to investigate individual diet specialization in California sea otters (Enhydra lutris nereis). Ecology 90: 961-974. Parnell AC, Inger R, Bearhop S, Jackson AL (2010) Source partitioning using stable isotopes: coping with too much variation. PlosOne 5: e9672. Pasquaud S, Lobry J, Elie P (2007) Facing the necessity of describing estuarine ecosystems: a review of food web ecology study techniques. Hydrobiologia 588: 159-172. Pasquaud S, Pillet M, David V, Sautour B, Elie P (2010) Determination of fish trophic levels in an estuarine system. Est Coast Shelf Sci 86: 237-246. Pennock JR, Velinsky DJ, Ludlam JM, Sharp JH, Fogel ML (1996) Isotopic fractionation of ammonium and nitrate during uptake by Skeletonema costatumimplications for δ15N dynamics under bloom conditions. Limnol Oceanogr 41: 451–459. Peterson BJ (1999) Stable isotopes as tracers of organic matter input and transfer in benthic food webs: a review. Acta Oecol 20: 479-487. 22 Peterson BJ, Fry B (1987) Stable isotopes in ecosystem studies. Ann Rev Ecol Syst 18: 293-320. Peterson BJ, Howarth RW (1987) Sulfur, carbon and nitrogen isotopes used to trace organic matter flow in the salt-marsh estuaries of Sapelo Island, Georgia. Limnol Oceanogr 32: 1195-1213. Peterson BJ, Howarth RW, Garritt RH (1985) Multiple stable isotopes to trace the flow of organic matter in esstuarine food webs. Science 227: 1361-1363. Phillips DL (2001) Mixing models in analyses of diet using multiple stable isotopes: a critique. Oecologia 127: 166-170. Phillips DL, Gregg JW (2001) Uncertainty in source partitioning using stable isotopes. Oecologia 127:171-179. Phillips DL, Gregg JW (2003) Source partitioning using stable isotopes. Coping with too many sources. Oecologia 136: 261-269. Phillips DL, Koch PL (2002) Incorporating concentration dependence in stable isotope mixing models. Oecologia 130: 114-125. Popp BN, Laws EA, Bidigare RR, Dore JE, Hanson KL, Wakeham SG (1998) Effect of phytoplankton cell geometry on carbon isotopic fractionation. Geochim Cosmochim Acta 62: 69-77. Post DM (2002) Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83: 703-718. Rasmussen JB (2010) Estimating terrestrial contribution to stream invertebrates and periphyton using a gradient-based mixing model for δ13C. J Anim Ecol 79: 393-402. Richoux NB, Froneman PW (2007) Assessment of spatial variation in carbon utilization by benthic and pelagic invertebrates in a temperate South African estuary using stable isotope signatures. Est Coast Shelf Sci 71: 545-558. Rolston DE, Fogg GE, Decker DL, Louie DJ, Grimser ME (1996) Nitrogen isotope ratios identify nitrate contamination sources. Calif Agric 50: 32-36. Ruesink JL, Roegner GC, Dumbauld BR, Newton JA, Armstrong DA (2003) Contributions of coastal and watershed energy sources to secondary production in a Northeastern Pacific estuary. Estuaries 48: 1079-1093. Sachs JP, Repeta DJ, Goericke R (1999) Nitrogen and carbon isotopic ratios of chlorophyll from marine phytoplankton. Geochim Cosmochim Acta 63: 1431-1441. Schwamborn R, Ekau W, Voss M, Saint-Paul U (2002) How important are mangroves as a carbon source for decapod crustacean larvae in a tropical estuary? Mar Ecol Prog Ser 229: 195-205. 23 Semmens BX, Ward EJ, Moore JW, Darimont CT (2009) Quantifying interand intra-population niche variability using hierarchical Bayesian stable isotope mixing models. PlosOne 4: e6187. Semmens BX, Ward EJ, Parnell AC, Phillips DL, Bearhop S, Inger R, Jackson A, Moore JW (2013) Statistical basis and outputs of stable isotope mixing models: comment on Fry (2013). Mar Ecol Prog Ser 490: 285-289. Smith BN, Epstein S (1970) Biogeochemistry of the stable isotopes of hydrogen and carbon in salt marsh biota. Plant Physiol 46: 738-742. Smith BN, Epstein S (1971) Two categories of 13C/12C ratios for higher plants. Plant Physiol 47: 380-384. Solomon CT, Carpenter SR, Clayton MK, Cole JJ, Coloso JJ, Pace ML, Vander Zanden MJ, Weidel BC (2011) Terrestrial, benthic, and pelagic resource use in lakes: results from a three-isotope Bayesian mixing model. Ecology 92: 1115-1125. Vander Zanden MJ, Shuter BJ, Lester N, Rasmussen JB (1999) Patterns of food chain length in lakes: a stable isotope study. Am Nat 154: 406-416. Vander Zanden MJ, Rasmussen JB (2001) Variation in δ15N and δ13C trophic fractionation: implications for aquatic food web studies. Limnol Oceanogr 46: 2061-2066. Varo N, Amat JA (2008) Differences in food assimilation between two coot species assessed with stable isotopes and particle size in faeces: linking physiology and conservation. Comp Biochem Phys A 149: 217-223. Vinagre C, Salgado J, Cabral HN, Costa MJ (2011) Food web structure and habitat connectivity in fish estuarine nurseriesimpact of river flow. Estuar Coast 34: 663-674. Waser NAD, Harrison PJ, Nielsen B, Calvert SE, Turpin DH (1999) Nitrogen isotope fractionation during the uptake and assimilation of nitrate, nitrite, ammonium, and urea by marine diatom. Limnol Oceanogr 43: 215-224. Wilson RM, Chanton J, Lewis FG, Nowacek D (2010) Concentration-dependent stable isotope analysis of consumers in the upper reaches of a freshwater-dominated estuary: Apalachicola Bay, FL, USA. Estuar Coasts 33: 1406-1419. Whitfield AK (1988) The fish community of the Swartvlei estuary and the influence of food availability on resource utilization. Estuaries 11: 160-170. Wu Y-T, Wang C-H, Zhang X-D, Zhao B, Jiang L-F, Chen J-K, Li B (2009) Effects of saltmarsh invasion by Spartina alterniflora on arthropod community structure and diet. Biol Invasions 11: 635-649. York JK, Tomasky G, Valiela I, Repeta DJ (2007) Stable isotopic detection of ammonium and nitrate assimilation by phytoplankton in the Waquoit Bay estuarine system. Limnol Oceanogr 52: 144-155. 24 Zeng Q-F, Kong F-X, Zhang E-L, Tan X (2007) Organisms from microbial food chain in Taihu lake. Environmental Science 28: 1670-1674. 25 Chapter 2 The importance of cross-ecosystem subsidies to the lower food web of an oligotrophic estuary Ester Dias1,2, Pedro Morais2,3, Anne M. Cotter4, Carlos Antunes2,5,6, Joel C. Hoffman4 1 ICBAS - Instituto de Ciências Biomédicas Abel Salazar, Universidade do Porto, Rua de Jorge Viterbo Ferreira 228, 4050-313 Porto, Portugal. 2 CIMAR/CIIMAR - Centro Interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto, Rua dos Bragas 289, 4050-123 Porto, Portugal. 3 CIMA - Centro de Investigação Marinha e Ambiental, Campus de Gambelas, Universidade do Algarve, 8005-139 Faro, Portugal. 4 Mid-Continent Ecology Division, National Health and Environmental Effects Research Lab, US Environmental Protection Agency, 6201 Congdon Blvd, Duluth, MN 55804, USA. 5 Aquamuseu do Rio Minho, Parque do Castelinho s/n, 4920-290 Vila Nova de Cerveira, Portugal. 6 ESG - Escola Superior Gallaecia, Largo das Oliveiras s/n, 4920 Vila Nova de Cerveira, Portugal. 26 Abstract Understanding the factors that govern the exchange and consumption of different organic matter (OM) sources is of paramount importance for management and conservation of estuaries. Our objective was to determine the importance of autochthonous and allochthonous OM in supporting the production of the Minho River estuary lower food web, under different river discharge conditions. Using stable isotopes of carbon (C) and nitrogen (N) we observed that pelagic (zooplankton) and benthic (Corbicula fluminea) primary consumers rely on a mixture of autochthonous and allochthonous OM, which included terrestrial-derived OM. Owing to an increase in the phytoplankton fraction in the POM pool during typical low river discharge conditions, autochthonous OM was the most important contributor to consumers’ biomass, declining with increasing river discharge. Based on SIAR mixing model, reliance on phytoplankton ranged from 11% to 60% during low flow discharge and from 2% to 30% (mode values) during high flow discharge. Allochthony increased with increasing river discharge, especially for consumers associated with benthic pathways. However, the highest contribution of terrestrial-derived OM (up to 70%) was observed during low river discharge conditions of an atypical hydrological year, preceded by winter high floods. We hypothesized that during high flow pulses, C loads into the estuary increased and phytoplankton biomass decreased, thereby decreasing phytoplankton availability to the food web. Although the majority of food sources were being filtered from the water column, primary consumers’ stable isotope data reveal that both pelagic and benthic consumers rely on sediment OM and microphytobenthos, providing evidence for reliance on benthic C. Its ability to consume terrestrial-derived OM and to access both pelagic and benthic food sources couple the terrestrial ecosystems with the estuarine pelagic and benthic environments. Keywords: allochthony, connectivity, river discharge, stable isotopes, pelagic, benthic. 27 2.1 Introduction Estuarine food webs function within a mosaic of habitats, which include intertidal sand or mud flats, tidal creeks and channels, saltmarshes, mangroves, and habitats defined by the salinity and depth gradient (Hagy III and Kemp, 2013). Connectivity between habitats is facilitated by physical exchanges of water (with associated nutrients, plankton and other particulate organic matter [OM] sources) and movements or migrations of animals for feeding, spawning or other ontogenetic changes (Sheaves, 2009; Hagy III and Kemp, 2013). Understanding the factors that govern the exchange and consumption of different OM sources is of paramount importance for management and conservation of estuaries, because secondary production and food web diversity depend on the quality of the basal OM sources (Rooney and McCann 2011). Three influential conceptual models have tried to explain the processes that govern the exchange and consumption of different OM sources in riverine food webs: the River Continuum Concept (RCC), the Flood Pulse Concept (FPC), and the Riverine Productivity Model (RPM). The RCC (Vannote et al. 1980) was the first conceptual model to link the physical characteristics of stream reaches with changes in consumers’ composition. According to this model, large rivers receive the majority of their OM from upstream processing of dead leaves and woody debris. Downstream transport of recalcitrant material influences trophic dynamics of large rivers because autochthonous primary production is limited by depth and turbidity. The RCC generally relied on inferences based on data collected in headwater streams; later, it was recognized that the potential to extrapolate to downstream areas was limited (Sedell et al. 1989). The FPC (Junk et al. 1989) was based on large river-floodplain systems, such as the Amazon and Mississippi Rivers, and states that the pulsing of river discharge into the floodplain influences primary and secondary production in large, lowland rivers. Thus, most animal biomass would be derived from production within the floodplain (e.g. live plant tissue and detritus) and not from downstream transport. In contrast, the RPM (Thorp and Delong 1994) hypothesized that metazoan production in large rivers is primarily fuelled by autochthonous production (phytoplankton, benthic phytoplankton, and aquatic vascular plants and mosses), and also by direct inputs from the riparian zone. The RPM states that labile autochthonous OM and moderately labile OM subsidies from the riparian zone will compensate for the typically greater abundance of recalcitrant OM from upstream leakage or floodplain inputs. The current consensus for river-estuary complexes is that phytoplankton predominantly supports upper trophic levels, although consumers may assimilate plant (C3) material in rivers with high sediment loads and low transparency during high-flow 34 Multivariate analyses were performed to reveal natural groupings in the data according to sampling period and portion of the estuary. A cluster analyses was performed using Euclidean similarity distance based on δ13C and δ15N stable isotope values of all the consumers collected. A similarity profile routine (SIMPROF) was used to test for the presence of sample groups (Clarke et al. 2008). DistLM and cluster analyses were performed using PRIMER v.6.1.11 (Clarke and Gorley 2006) with PERMANOVA+1.0.1 add-on package (Anderson et al. 2008). One-way ANOVA (factors “sampling period” or “portion of the estuary”) were performed to determine bulk zooplankton and C. fluminea changes in stable isotope values through time and space. When data did not meet the assumptions, the nonparametric Kruskal-Wallis was used. These analyses were performed using STATISTICA 7, Statsoft. To quantify OM source contributions to the most frequently sampled zooplankton groups (Calanoids and Caridea larvae) and C. fluminea tissue, we used a dual-stable isotope mixing model. The mixing model estimates a proportional contribution for each food source, assuming all sources sum to 1. However, the model equations are indeterminate when there are more than n+1 sources relative to stable isotopes (i.e. more than three sources for two stable isotopes). Because there were more than three potential food sources in the estuary, we used a stable isotope mixing model that uses Bayesian inference to solve the indeterminate equations and produce a probability distribution that represents the likelihood a given source contributes to the consumer (Parnell et al. 2010). The model, Stable Isotope Analysis in R (SIAR), allows each of the sources and the trophic enrichment factor (TEF; or trophic fractionation) to be assigned as a normal distribution, rather than a single datum (Parnell et al. 2010). We used the SIAR package (Stable Isotope Analysis in R; Parnell et al. 2010), which is part of the open source statistical language R (R Development Core Team 2007). SIAR will produce a range of feasible mixing problem solutions to which are assigned credibility intervals (CI), analogous to the confidence intervals used in frequentist statistics (in this study, 95% CI; Parnell et al. 2010). SIAR also includes a residual error term. For modelling purposes, the estuary was divided into three portions based on salinity, as a result fro the DistLM routine: tidal freshwater (TFW), brackish and marine portions. For the mixing model, we estimated the average (± SD) δ13C and δ15N values for the various OM sources. At each station, the phytoplankton 13C (δ13Cphytoplankton) value was estimated from the DIC δ13C (δ13CDIC) value, assuming an uptake fractionation of -21‰ (i.e. δ13Cphytoplankton= δ13CDIC - 21‰; Peterson and Fry 1987). For each portion of the estuary, the average δ13Cphytoplankton value used in the SIAR mixing model was the mean of station-specific average surface and bottom δ13Cphytoplankton values (based on station 35 replicates) for those stations within each portion. We obtained δ15Nphytoplankton values for the SIAR mixing model from POM sampled during August 2011 at all stations where the POM samples were comprised of nearly 100% phytoplankton based on phytoplankton C:Chl a and Chl a and POC concentrations (Canuel et al 1995, Marty and Planas 2008). For the model, the value used in the TFW and brackish portions of the estuary was the mean station-specific average δ15Nphytoplankton values for those stations within each portion. For the marine stations, we used the values described by Bode et al. (2007) for the Iberian Peninsula Coast (6 ± 1.5 ‰). The average MPB, POM, and SOM δ13C and δ15N values were calculated from all the values for those stations within each portion. The zooplankton and C. fluminea δ13C values were corrected for lipid content because lipids are depleted in 13C compared to protein and carbohydrates (DeNiro and Epstein 1977). Variability in lipid content can bias bulk tissue δ13C values and thereby cause dietary or habitat shifts to be incorrectly interpreted. We corrected zooplankton tissue data for lipid content using the mass balance correction model proposed for zooplankton by Smyntek et al. (2007; Eq. 5), and for C. fluminea muscle tissue data the mass balance correction for fish muscle tissue proposed by Hoffman and Sutton (2010; Eq. 6), which uses estimates of C:Nprotein and Δδ13Clipid that are similar to those from the muscle tissue found for other fish (e.g. Sweeting et al. 2006) and taxonomic groups (e.g. shrimp and zooplankton; Fry et al. 2003, Smyntek et al. 2007). Zooplankton were also corrected for ethanol preservation (+0.4 ‰ δ13C, +0.6 ‰ δ15N; Feuchtmayer and Grey 2003). For the SIAR mixing model, we adjusted the δ13C and δ15N values for one trophic level using the TEF estimates from Post (2002; +0.4 ± 1.3 ‰ δ13C, +3.4 ± 1.0 ‰ δ15N). 2.3 Results 2.3.1 Environmental data The average monthly river discharge during the sampling period was similar to the average monthly estimates for the historical data set (1991-2005) (Fig. 2.2). Thus, September 2010 and August 2011 average river discharge values were close to the typical low river discharge conditions of Minho River estuary, and March 2011 average river discharge values were typical for the end of the wet season (Fig. 2.2). In September 2010, saltwater intrusion was detected up to 12 km from the river mouth (stations 1-13 were brackish, stations 14-21 were within the tidal freshwater (TFW) portion of the estuary; Fig. 2.3 A). In March 2011, owing to high river discharge, stations 36 1-6 were brackish and 7-21 were freshwater (Fig. 2.3B). In August 2011, saltwater intrusion resulted in measurable salinity up to 16 km from the mouth. Stations 1-6 were marine, 7-16 were brackish and 17-21 were within the TFW (Fig. 2.3C). Fig. 2.3 Bottom (closed circles) and surface (open circles) salinity (A-C), chlorophyll a (Chl a) concentration (g L-1; D-F), ammonia (NH4+; G-I) and nitrates (NO3-; J-L) concentrations (M) determined for the estuarine salinity mixing for the river Minho estuary in September 2010 (low discharge), March 2011 (high discharge), and August 2011 (low discharge). The quality and quantity of POM varied between sampling periods and among portions of the estuary. The C/NPOM above 10 in September 2010 indicates a substantial contribution of terrestrial-derived OM to the POM pool in the TFW (Hedges et al. 1986, 1997; Table 2.2). In the brackish estuarine area, C/NPOM (7-9) varied between the C:N of terrestrial-derived OM (> 10) and of marine phytoplankton (~7; Hedges et al. 1986, 1997), indicating that brackish POM was a mixture of riverine and marine POM (Table 2.2). In March 2011, the C:NPOM varied between 6 and 12 indicating that the POM pool was a mixture of phytoplankton and terrestrial-derived OM (Table 2.1). In August 2011, the C:NPOM averaged above 7, the estimates of the phytoplankton proportion in the POM pool (> 50%) and high chlorophyll a (Chl a) concentrations indicate that phytoplankton was a major contributor to the POM pool (Table 2.2, Fig. 2.3 D-F). 37 Table 2.2 Characterization of the POM pool in the Minho River estuary, along the salinity mixing gradient in September 2010, March and August 2011. Month Location C:NPOM average (SD) POC (µM) average (SD) Phytoplankton’s proportion % POC (SD) September 2010 TFW 10.2 (0.9) 29.0 (11.9) 17.2 (18.8) Brackish 8.3 (1.1) 25.8 (13.1) 18.7 (14.0) March 2011 TFW 8.4(1.6) 17.9 (6.3) 30.0 (21.4) Brackish 9.2 (1.9) 11.9 (4.9) 36.8 (26.2) August 2011 TFW 7.0 (0.5) 27.0 (4.8) 74.4 (19.0) Brackish 7.0 (1.1) 27.5 (7.8) 56.6 (25.6) Marine 8.7 (2.6) 16.3 (4.6) 7.7 (5.1) The POC concentrations in September 2010 and August 2011 were almost the double that measured in March 2011, and concentration values were similar along the estuarine salinity gradient (Table 2.2). The average NH4+ concentration (M; bottom and surface samples combined) was the highest in March 2011 (5.7 ± 6.6 M) and decreased during in the low flow months (September 2010: 4.0 ± 1.2 M; August 2011: 2.3 ± 1.3 M; Figs. 2.3 G-I). The average NO3concentration (M; bottom and surface samples combined) were higher than NH4+ for all the sampling periods with values increasing towards TFW (Figs. 2.3, J-L). 2.3.2 Food web characterization The results from the DistLM indicate that the environmental variables that accounted for the most variability in the δ13C and δ15N values of both zooplankton and C. fluminea were salinity and Chl a concentration (50% of the total variance; Table 2.3). The δ13C and δ15N values of zooplankton and C. fluminea did not demonstrate a significant association with river discharge (Table 2.3). However, significant differences were observed between sampling periods for both zooplankton and C. fluminea. Bulk zooplankton samples collected in the TFW were significantly depleted in 13C in March 2011 (H(2,25)= 9.70, p= 0.008) and 13C-enriched in the brackish stations (H(2,23)= 6.79, p= 0.03) compared to the other sampling periods. The δ15N values were significantly higher during August 2011 (TFW: H(2,25)= 18.45, p= 0.0001; brackish: H(2,23)= 8.87, p= 0.012). The C. fluminea δ13C values were not significantly different (p> 0.05). However, those collected in August had the highest δ15N values (TFW: H(2,25)= 14.15, p= 0.0008; brackish: t= -2.54, df= 13, p= 0.025). 38 Table 2.3 DistLM model based on the δ13C and δ15N values of zooplankton and Corbicula fluminea and fitted abiotic data. Marginal tests Variable SS (trace) pseudo-F p proportion (%) Salinity 58.26 41.76 < 0.05 31.0 River discharge 5.17 2.63 0.075 2.75 Chl a 39.41 24.66 < 0.05 21.0 Sequential tests Variable AIC SS (trace) pseudo-F p proportion (%) Salinity 33.61 58.26 41.76 < 0.05 31.0 Chl a 5.61 35.12 34.15 < 0.05 19.0 The cluster analysis results corroborate the finding that salinity was the main factor structuring consumers’ δ13C and δ15N values (Fig. 2.4), owing to either differences in the OM sources available or differences in the stable isotope values of the OM sources, or both. The first branch separates consumers enriched in 13C and depleted in 15N (A), from those showing the opposite trend (B) (Figs. 2.4, 2.5). Group A was essentially comprised of zooplankton collected in marine and brackish stations during August 2011 and of zooplankton collected in the brackish stations during March 2011 (Fig. 2.4). Group B included all the organism collected in the TFW (benthic and pelagic), and some 15Nenriched consumers, such as zooplankton predators (e.g. fish larvae collected in the marine and brackish stations– Fl; Fig. 2.4). Group A clusters into group AA, which includes Caridea larvae collected in September 2010, and group AB, which includes pelagic zooplankton. Caridea larvae from group AA had the lowest δ15N values, suggesting a great contribution of the 15N depleted SOM to their biomass (Fig. 2.5). Group AB cluster into group ABA and ABB, essentially separating the zooplankton collected in August from those collected in March, which were more 13C-depleted and 15N-enriched, owing to the existence of differences in the diet (Figs. 2.6, 2.7). For instance, assuming typical trophic fractionation (+0.4 δ13C, +3.4 δ15N), it is likely that zooplankton collected in the marine stations in August 2011 were using a 15Ndepleted and 13Cenriched source such as MPOM or MPB; in the brackish stations in March 2011 they were using a more 15Nenriched source such as macroalgae and/or a more 13Cdepleted source such as brackish phytoplankton (Fig. 2.6). 39 Fig. 2.4 Dendograms from clustering analyses of primary consumers δ13C and δ15N values collected in September 2010 (black triangle), March 2011 (inverted grey triangle) and August 2011 (grey square) in the marine (M), brackish (B) and TFW (F) portions of the Minho River estuary. The first letters (one or two) represent the species codes (listed in Appendix A) and the last letter represents the portion of the estuary where they were collected (M, B or F). Grey lines indicate groups of samples not separated (at < 0.05) by SIMPROF. The SIAR mixing model indicates that OM source nutritional contribution to the primary consumers varied by sampling period, portion of the estuary, and taxonomic group (Figs. 2.8, 2.9 and 2.10). The main difference among sampling periods was the substantial phytoplankton contribution to all the consumers modelled during August 2011 (Figs. 2.8, 2.9 and 2.10). In September 2010 and March 2011, calanoid copepods (calanoid copepods not identified in TFW and Acartia spp. in the brackish and marine portions of the estuary) and Caridea larvae were generally supported by terrestrial-derived OM (September 2010), MPB (September 2010) and SOM (September 2010 and March 2011). The proportion of phytoplankton in the tissues was always lower than 40% (Figs. 2.8 and 2.9). Calanoid copepods collected in the TFW portion of the estuary in March 2011 were not modelled because they had δ13C values outside the range of δ13C values from the OM sources sampled (Fig. 2.6). Also, cladocerans and Chironomidae larvae had low δ13C values, suggesting that they were feeding on a 13C-depleted source that may have not been sampled (Fig. 2.6). Differences were observed between portions of the estuary, with terrestrial-derived OM contribution increasing in the TFW stations (Figs. 2.8, 2.9). 40 FPOM BP 13C (‰) -35 -30 -25 -20 -15 -10 15N (‰) 0 2 4 6 8 10 12 Macroalgae MPB SOM BPOM Cf C Ca M B FPOM BP Fig. 2.5 Average stable isotope ratios (±SD) of common Minho River estuary zooplankton species (codes in Appendix A) and Corbicula fluminea collected in the tidal freshwater (A), brackish (B)portions of the estuary in September 2010. Organic matter (OM) sources include freshwater (FP) and brackish (BP) phytoplankton, freshwater (FPOM) and brackish (BPOM) particulate OM, microphytobenthos (MPB), submerged aquatic vegetation (SAV), terrestrial plants (Terr), macroalgae, and sediment OM (SOM). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). -40 -35 -30 -25 -20 -15 15N (‰) -4 -2 0 2 4 6 8 10 12 SAV MPB FP SOM FPOM Terr. Cf M Ca Cy A 41 -40 -35 -30 -25 -20 -15 15N (‰) -4 -2 0 2 4 6 8 10 12 14 FPOM FP MPB SOM Terr. SAV EAV Fl Cf Ca Ch Cd Hy A 13C (‰) -35 -30 -25 -20 -15 -10 15N (‰) 0 2 4 6 8 10 12 Macroalgae SOM BPOM FPOM BP MPB Cl A Pl Fl O H B Fig. 2.6 Average stable isotope ratios (±SD) of common Minho River estuary zooplankton species (codes in Appendix A) and Corbicula fluminea collected in the tidal freshwater (A), brackish (B) portions of the estuary in March 2011. Organic matter (OM) sources include freshwater (FP) and brackish (BP) phytoplankton, freshwater (FPOM) and brackish (BPOM) particulate OM, microphytobenthos (MPB), submerged (SAV) and emergent (EAV) aquatic vegetation (SAV), terrestrial plants (Terr), macroalgae, and sediment OM (SOM). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). 42 EAV -35 -30 -25 -20 -15 15N (‰) -4 -2 0 2 4 6 8 10 12 14 16 SAV MPB SOM FPOM Terr. FP Ca Hy Cf MCh A EAV 13C (‰) -35 -30 -25 -20 -15 -10 15N (‰) -4 -2 0 2 4 6 8 10 12 14 A Macroalgae Ok FPOM BPOM Terr. SOM MPOM BP MPB C O D Fl Cf M B Fig. 2.7 Average stable isotope ratios (±SD) of common Minho River estuary zooplankton species (codes in Appendix A) and Corbicula fluminea collected in the tidal freshwater (A), brackish (B) and marine (C) portions of the estuary in August 2011. Organic matter (OM) sources include freshwater (FP), brackish (BP) and marine (MP) phytoplankton, freshwater (FPOM), brackish (BPOM) and marine (MPOM) particulate OM, microphytobenthos (MPB), submerged (SAV) and emergent (EAV) aquatic vegetation (SAV), terrestrial plants (Terr), macroalgae, and sediment OM (SOM). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). 43 13C (‰) -35 -30 -25 -20 -15 -10 15N (‰) -4 -2 0 2 4 6 8 10 12 14 MPOM MPB MP Macroalgae BPOM Fl P C A D Pl Cl Ok C Fig. 2.7 (continuation) Average stable isotope ratios (±SD) of common Minho River estuary zooplankton species (codes in Appendix A) and Corbicula fluminea collected in the tidal freshwater (A), brackish (B) and marine (C) portions of the estuary in August 2011. Organic matter (OM) sources include freshwater (FP), brackish (BP) and marine (MP) phytoplankton, freshwater (FPOM), brackish (BPOM) and marine (MPOM) particulate OM, microphytobenthos (MPB), submerged (SAV) and emergent (EAV) aquatic vegetation (SAV), terrestrial plants (Terr), macroalgae, and sediment OM (SOM). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). The contribution of terrestrial-derived OM and OM in the sediment (SOM and MPB) to C. fluminea production was higher in September 2010 and March 2011 than August 2011 (Fphytoplankton ≤ 50%; Fig. 2.10). As with zooplankton, the proportion of phytoplankton in the tissue muscle was higher in August 2011 (Fphytoplankton: 50-100%; Fig. 2.10). Differences were also observed between portions of the estuary. In the TFW stations, the terrestrial-derived OM (FPOM) was a relevant alternative to phytoplankton (especially in September 2010 and March 2011), whereas in the brackish stations, (brackish and freshwater), phytoplankton contributed the most to C. fluminea biomass (September 2010: 30-55%; August 2011: 67-100%; Fig. 2.10). 50 up to 60% of terrestrial-derived OM to primary consumers’ biomass in the brackish estuary, thus suggesting that OM from the upper portion of the estuary was being exported and consumed in the lower portion of the estuary. In August 2011, the contribution of phytoplankton to consumer biomass in the TFW was ≤80%, and ≤100% in the brackish portion of the estuary. The δ13C and δ15N values from calanoid copepods, Caridea larvae and C. fluminea, suggest that they were using a more 13C-depleted source than brackish phytoplankton, which could be terrestrial plants or freshwater POM. Their high δ15N values, however, indicate that the most likely source was freshwater POM, which was largely comprised of phytoplankton. Thus, as in September 2010, the freshwater OM was exported from the TFW portion of the estuary, and functioned as an important subsidy to organisms in the lower portion of the estuary. Accordingly we estimate that up to 80% of consumer biomass was derived from FPOM. Also in the brackish portion of the estuary, there was another group of consumers that consumed a more 13C-enriched mixture of OM sources (Centropages, Oikopleura sp. and Ostracoda), indicating consumption of macroalgae and marine POM. The importance of marine material (phytoplankton and POM), increased with the proximity to the mouth of the river, where it comprised 26-88% of calanoid copepod biomass. Therefore, marine ecosystem can also subsidize the estuarine food web during low flow conditions. In March 2011, the consumers collected in the brackish area of the estuary relied on 13C-enriched sources, such as macroalgae and brackish phytoplankton, and 15N-depleted sources (e.g. SOM). We estimated that SOM was the most important source contributing to calanoid copepod biomass (27-61%). Because the δ13C and δ15N values of calanoid copepods were similar to those from polychaeta larvae and ostracods, they could be also using the same OM sources. Because marine intrusion was confined to the first 6 km of the estuary, it is possible that during high discharge, the marine subsidies to the estuary are minor. If consumers were using marine phytoplankton, which has a typical δ15N of 6 ± 1.5‰ (Bode et al. 2007) and δ13C of -20.5 ± 1.3‰ (McMahon et al. 2013), and assuming a trophic fractionation of +3.4 ‰ for N and +0.4‰ for C, their δ13C and δ15N values would have to be close to -21‰ for δ13C and higher than 8‰ for δ15N, which was not the case. 2.4.3 Possible consequences of allochthony to the estuarine food web Calanoid copepods, Caridea larvae and C. fluminea obtained the majority of their food sources from the water column. The quality of these food sources varied with river discharge and along the estuarine mixing zone, revealing that primary consumers have the ability to use alternative and less labile OM sources (terrestrial-derived OM and SOM) when phytoplankton availability is low. 51 Although the mechanisms underlying the use of refractory material are poorly understood, previous studies proposed that terrestrial OM could become available to pelagic consumers via several mechanisms: microbial uptake of terrestrial dissolved organic carbon followed by direct or indirect consumption of the microbes (Berggren et al. 2010), direct consumption of dissolved terrestrial organic carbon (Speas and Duffy 1998) or direct consumption of terrestrial-derived particles (Cole et al. 2006). Corbicula fluminea has a preference for small-sized living and suspended POM (Atkinson et al. 2011), and detritus have been identified in gut content analysis (Hill 1985 cited in Foe and Knight 1985). Therefore, it is likely that they are able to consume directly terrestrial-derived particles. If so, it is possible that C. fluminea may also act as initiators to break down plant cell wall structural polysaccharides, thus providing more labile OM for benthic organisms through the production of faeces and pseudofaeces. However, further studies are needed to clarify the presence of an enzymatic mechanism that enables C. fluminea to perform this function (Dias et al in press). Thus, primary consumers in the Minho River estuary have the ability to adapt to low food quality conditions, especially during high-flow pulses, by using less labile OM sources such as terrestrial-derived OM or SOM. Its ability to consume terrestrial-derived OM and to access both pelagic and benthic food sources couple the terrestrial ecosystems with the estuarine pelagic and benthic environments. Nonetheless, further studies are needed to understand how the use of less labile OM sources affects the productivity of primary consumers and upper trophic levels. Acknowledgements We thank Eduardo Martins and the staff at Aquamuseu do Rio Minho for their help while conducting the field work; Katie Bentley for preparing POM samples for stable isotope analysis; Jacinto Cunha, Martina Ilarri and Luis Vieira for helping with the sorting and identification of zooplankton samples; Allan T. Souza and Martina Ilarri for helping with data analysis; Rute Pinto for providing the map of the study area. E. Dias was supported by a Ph.D. grant from Iceland, Liechtenstein and Norway through the EEA Financial Mechanism and the Norwegian Financial Mechanism (PT0010). P. Morais was supported by a post-doc scholarship financed by Fundação para a Ciência e Tecnologia, Portugal (SFRH/BPD/40832/2007). This work was partially supported by the project MIGRANET SOE 2/P2E288. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA. 2.5 References Altabet MA (1988) Variations in nitrogen isotopic composition between sinking and suspended particles: implication for nitrogen cycling and particle transformation in the open ocean. Deep-Sea Res 35: 535-554. 52 Alves AM (1996) Causas e processos da dinâmica sedimentar na evolução actual do litoral do Alto Minho. PhD Dissertation, Universidade do Minho. Anderson MJ, Gorley RN, Clarke KR (2008) PERMANOVA + for PRIMER: guide to software and statistical methods. PRIMER-E, Plymouth. Antunes C, Araújo MJ, Braga C, Roleira A, Carvalho R, Mota M (2011) Valorização dos recursos naturais da bacia hidrográfica do rio Minho. Final report from the project Natura Miño-Minho, Centro interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto. Atkinson CL, First MR, Covich AP, Opsahl SP, Golladay SW (2011) Suspended material availability and filtration-biodeposition processes performed by a native and invasive bivalve species in streams. Hydrobiologia 667: 191-204. Berggren M, Ström L, Laudon H, Karlsson J, Jonsson A, Giesler R, Bergström AK, Jansson M (2010) Lake secondary production fueled by rapid transfer of low molecular weight organic carbon from terrestrial sources to aquatic consumers. Ecol Lett 870-880. BirdLife International (2012) Important Bird Areas factsheet: Minho and Coura estuaries. Available at http://www.birdlife.org. Accessed 5 November 2012. Bode A, Alavarez-Ossorio MT, Cunha ME, Garrido S, Peleteiro JB, Porteiro C, Valdés L, Varela M (2007) Stable nitrogen isotope studies of the pelagic food web on the Atlantic shelf of the Iberian Peninsula. Prog Oceanogr 74: 115-131. Boltovskoy D, Izaguirre I, Correa N (1995) Feeding selectivity of Corbicula fluminea (Bivalvia) on natural phytoplankton. Hydrobiologia 312: 171-182. Canuel EA, Cloern JE, Ringelberg DB, Guckert JB (1995) Molecular and isotopic tracers used to examine sources of organic matter and its incorporation into the food webs of San Francisco Bay. Limnol Oceanogr 40: 67-81. Carabel S, Godínez-Domínguez E, Verísimo P, Fernández L, Freire J (2006) An assessment of sample processing methods for stable isotope analyses of marine food webs. J Exp Mar Biol Ecol 336: 254-261. Caraco N, Bauer JE, Cole JJ, Petsch S, Raymond P (2010) Milennial-aged organic carbon subsidies to a modern river food web. Ecology 91: 2385-2393. Carpenter SR, Cole JJ, Pace ML, Van de Bogert M, Bade DL, Bastviken D, Gille CM, Hodgson JR, Kitchell JF, Kritzberg ES (2005) Ecosystem subsidies: terrestrial support of aquatic food web from 13C addition to contrasting lakes. Ecology 86: 27372750. Chanton J, Lewis FG (2002) Examination of coupling between primary and secondary production in a river-dominated estuary: Apalachicola Bay, Florida, USA. Limnol Oceanogr 47: 683–697. 53 Cifuentes LA, Sharp JH, Fogel ML (1988) Stable carbon and nitrogen isotope biogeochemistry in the Delaware estuary. Limnol Oceanogr 33: 1102-1115. Clarke KR, Gorley RN (2006) PRIMER v6: user manual/tutori-al. PRIMER-E, Plymouth. Clarke KR, Somerfield PJ, Gorley RN (2008) Testing of null hypotheses in exploratory community analyses: similarity profiles and biota-environment linkage. J Exp Mar Biol Ecol 366: 56-69. Cloern JE, Canuel EA, Harris D (2002) Stable carbon and nitrogen isotope composition of aquatic and terrestrial plants of the San Francisco Bay estuarine system. Limnol Ocenogr 47: 713-729. Cole JJ, Carpenter SR, Pace ML, Van de Bogert MC, Kitchell JL, Hodgson JR (2006) Differential support of lake food webs by three types of terrestrial organic carbon. Ecol Lett 9: 558-568. Cole JJ, Carpenter SR, Kitchell J, Pace ML, Solomon CT, Weidel B (2011) Strong evidence for terrestrial support of zooplankton in small lakes based on stable isotopes of carbon, nitrogen, and hydrogen. P Natl Acad Sci 108: 1975-1980. Cole JJ, Solomon CT (2012) Terrestrial support of zebra mussels and the Hudson River food web: a multi-isotope, Bayesian analysis. Limnol Oceanogr 56: 1802-1815. Deegan LA, Garritt RH (1997) Evidence for spatial variability in estuarine food webs. Mar Ecol Prog Ser 147: 31-47. DeNiro MJ, Epstein S (1977) Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197: 261–263. Dias E, Morais P, Antunes C, Hoffman JC (in press) Linking terrestrial and benthic estuarine ecosystems: organic matter sources supporting the high secondary production of a non-indigenous bivalve. Doi: 10.1007/s10530-014-0655-8. EIONET (2012) Available at http://bd.eionet.europa.eu/activities/Natura_2000. Accessed 5 November 2012. Ferreira JG, Simas T, Nobre A, Silva MC, Schifferegger K, Lencart-Silva J (2003) Identification of sensitive areas and vulnerable zones in transitional and coastal Portuguese systems. Application of the United States National Estuarine Eutrophication Assessment to the Minho, Lima, Douro, Ria de Aveiro, Mondego, Tagus, Sado, Mira, Ria Formosa and Guadiana systems. INAG/IMAR Technical Report. Feuchtmayer H, Grey J (2003) Effect of preparation and preservation procedures on carbon and nitrogen stable isotope determinations from zooplankton. Rapid Commun Mass Sp 17: 2605-2610. Foe C, Knight A (1985) The effect of phytoplankton and suspended sediment on the growth of Corbicula fluminea (Bivalvia). Hydrobiologia 127: 105-115. 54 France RL (1995) Carbon-13 enrichment in benthic compared to planktonic algae: foodweb implications. Mar Ecol Prog Ser 124: 307–312. Fry B (1991) Stable isotope diagrams of freshwater food webs. Ecology 72: 22932297. Fry B (2002) Conservative mixing of stable isotopes across estuarine salinity gradients: a conceptual framework for monitoring watershed influences on downstream fisheries production. Estuaries 25: 264-271. Fry B, Allen YC (2003) Stable isotopes in zebra mussels as bioindicators of riverwatershed linkages. River Res. Applic. 19: 683-696. Fry B, Sherr EB (1984) δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contr mar Sci 27: 15-47. Goericke R, Montoya, JP, Fry B (1994) Physiology of isotopic fractionation in algae and cyanobacteria. In: Lajtha K, Michener RM (eds.), Stable isotopes in ecology and environmental science, Blackwell Scientific Publications, pp 187-221. Hagy III JD, Kemp WM (2013) Estuarine food webs. In: Day Jr. JW, Crump BC, Kemp WM, Yáñez-Arancibia A (eds), Estuarine Ecology, Wiley-Blackwell, pp 417-441. Hedges JI, Clark WA, Quay PD, Richey JE, Devol AH, Santos UM (1986) Compositions and fluxes of particulate organic material in the Amazon River. Limnol Ocenogr 31: 717-738. Hedges JI, Keil RG, Benner R (1997) What happens to terrestrial organic matter in the ocean? Org Geochem 27: 195-212. Hoffman JC, Bronk DA (2006) Interannual variation in stable carbon and nitrogen isotope biogeochemistry of the Mattaponi River, Virginia. Limnol Oceanogr 51: 2319-2332. Hoffman JC, Bronk DA, Olney JE (2007) Tracking nursery habitat use by young American shad in the York River estuary, Virginia using stable isotopes. T Am Soc 136: 1285-1297. Hoffman JC, Bronk DA, Olney JE (2008) Organic matter sources supporting lower food web production in the tidal freshwater portion of the York River estuary, Virginia. Estuar Coast 31: 898-911. Hoffman JC, Sutton TT (2010) Lipid correction for carbon stable isotope analysis of deep-sea fishes. Deep-Sea Res I 57: 956-964. Howarth RW, Schneider R, Swaney D (1996) Metabolism and organic carbon fluxes in the tidal freshwater Hudson River. Estuar Res Fed 19: 848-865. Huryn AD, Riley RH, Young RG, Arbuckle CJ, Peacock K, Lyon G (2001) Temporal shift in contribution of terrestrial organic matter to consumer production in a grassland river. Freshwater Biol 46: 213-226. 55 Jones RI, Grey J, Sleep D, Quarmby C (1998) An assessment, using stable isotopes, of the importance of allochthonous organic carbon source to the pelagic food web in Loch Ness. Proc Biol Sci 265: 105-111. Junk WJ, Bayley PB, Sparks RE (1989) The flood pulse concept in river-floodplain systems. In: Dodge DP (ed) Proceedings of the International Large River Symposium. Can Spec Publ Fish Aquat Sci 106, p. 110-127. Kasai A, Nakata A (2005) Utilization of terrestrial organic matter by the bivalve Corbicula japonica estimated from stable isotope analysis. Fisheries Sci 71: 151-158. Kendall C, Silva SR, Kelly VJ (2001) Carbon and nitrogen isotopic compositions of particulate organic matter in four large river systems across the United States. Hydrol Process 15: 1301-1346. Kendall C, Elliott EM, Wankel SD (2007) Tracing anthropogenic inputs of nitrogen to ecosystems. In: Michener RH, Lajtha K (eds) Stable isotopes in Ecology and Environmental Science. Blackwell Publishing, p. 375-449. Lewis WM, Jr, Hamilton SK, Lasi MA, Rodríguez M, Saunders JF (2000) Ecological determinism on the Orinoco floodplain. BioScience 50: 681-692. Lorenzen CJ (1967) Determination of chlorophyll and pheo-pigments: Spectrophotometric equations . Limnol Oceanogr 12: 343-346. Lorrain A, Savoye N, Chauvaud L, Paulet Y-M, Naulet N (2003) Decarbonation and preservation method for the analysis of organic C and N contents and stable isotope ratios of low-carbonated suspended particulate material. Analytica Chimica Acta 491: 125-133. McMahon RF (2002) Evolutionary and physiological adaptations of aquatic invasive animals: r selection versus resistance. Can J Fish Aquat Sci 59, 1235-1244. Mariotti A, Germon JC, Hubert P, Kaiser P, Letolle R, Tardieux A, Tardieux P (1981) Experimental determination of nitrogen kinetic isotope fractionation: some principles; ilustration for the denitrification and nitrification processes. Plant Soil Sci 62: 413-430. Marty J, Planas D (2008) Comparison of methods to determine algal δ13C in freshwater. Limnol Oceanogr: Methods 6: 51-63. McMahon K, Hamady LL, Thorrold SR (2013) A review of ecogeochemistry approaches to estimating movements of marine animals. Limnol Oceanogr 58: 697-714. Meili M, Kling GW, Fry B, Bell RT, Ahlgren I (1996) Sources and partitioning of organic matter in a pelagic microbial food web inferred from the isotopic composition δ13C and δ15N of zooplankton species. Arch Hydrobiol Spec Issues Advanc Limnol 48: 53-61. Miyake Y, Wada E (1971) The isotope effect on the nitrogen in biochemical, oxidation-reduction reactions. Rec. Oceanogr 11: 1-6. Parnell AC, Inger R, Bearhop S, Jackson AL (2010) Source partitioning using stable isotopes: coping with too much variation. PlosOne 5: e9672. 56 Pasquaud S, Lobry J, Elie P (2007) Facing the necessity of describing estuarine ecosystems: a review of food web ecology study techniques. Hydrobiologia 588: 159-172. Peterson BJ, Fry B (1987) Stable isotopes in ecosystem studies. Ann Rev Ecol Syst 18: 293-320. Post DM (2002) Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83: 703-718. Roach KA (2013) Environmental factors affecting incorporation of terrestrial material into large river food webs. Freshwater Science 32: 283-298. Rooney N, McCann KS (2011) Integrating food web diversity, structure and stability. Trends Ecol Evol 27: 40-46. del Rio CM, Wolf N, Carleton SA, Gannes LZ (2009) Isotopic ecology ten years after a call for more laboratory experiments. Biol Rev 84: 91-111. Sedell JR, Richey JE, Swanson FJ (1989) The river continuum concept: A basis for the expected ecosystem behavior of very large rivers? In: Dodge DP (ed) Proceedings of the International Large River Symposium. Can Spec Publ Fish Aquat Sci 106: 49-55. Sheaves M (2009) The consequences of ecological connectivity: the example of the coastal ecosystem mosaic. Mar Ecol Progress Ser 391: 107-115. Sin Y, Wetzel RL, Anderson IC (1999) Spatial and temporal characteristics of nutrient and phytoplankton dynamics in the York River estuary, Virginia: analyses of longterm data. Estuaries 22: 260-275. Smith BN, Epstein S (1970) Biogeochemistry of the stable isotopes of hydrogen and carbon in salt marsh biota. Plant Physiol 46: 738-742. Smith BN, Epstein S (1971) Two categories of 13C/12C ratios for higher plants. Plant Physiol 47: 380-384. Smyntek PM, Teece MA, Schulz KL, Thackeray SJ (2007) A standard protocol for stable isotope analysis of zooplankton in aquatic food web research using mass balance correction models. Limnol Oceanogr 52: 2135-2146. SNIRH (2012) Sistema nacional de informação de recursos hídricos. Available at: http://snirh.pt. Accessed 31 October 2012. Sousa R, Nogueira AJA, Gaspar M, Antunes C, Guilhermino L (2008) Growth and extremely high production of the non-indigenous invasive species Corbicula fluminea (Muller, 1774): possible implications for ecosystem functioning. Est Coast Shelf Sci 80: 289-295. Speas DW, Duffy WC (1998) Uptake of dissolved carbon (DOC) by Daphnia pulex. J Freshwat Ecol 13: 457-463. 57 Sweeting CJ, Polunin NVC, Jennings S (2006) Effects of chemical lipid extraction and arithmetic lipid correction on stable isotope ratios of fish tissues. Rapid Communications in Mass Spectrometry 20: 595-601. Thompson JK, Parchaso F (2010) Corbula amurensis conceptual model. U.S. Geological Survey. Thorp JH, Delong MD (1994) The riverine producivity model: an heuristic view of carbon sources and organic processing in large river ecosystems. Oikos 70: 305-308. Thorp JH, Delong MD (2002) Dominance of autotrophic carbon in food webs of heterotrophic rivers. Oikos 96: 543-550. Vannote RL, Minshall G, Cummins KW, Sedell JR, Cushing CE (1980) The river continuum concept. Can J Fish Aquat Sci 37: 130-137. Zeug SC, Winemiller KO (2008) Evidence supporting the importance of terrestrial carbon in a large-river food web. Ecology 89: 1733-1743. 58 Appendix A. δ15N and δ13C values of zooplankton taxa collected in September 2010, March 2011 and August 2011 in the tidal freshwater (TFW), brackish and marine portions of the Minho River estuary. Zooplankton taxa Sampling month Location 15N (±SD) ‰ δ13C (±SD) ‰ Code Copepoda Calanoida Calanus sp. March 2011 Brackish 8.5 (0.1) -18.9 (0.3) Cl Acartia sp. March 2011 Brackish 8.1 (0.2) -18.8 (0.04) A August 2011 Brackish 10.7 (1.8) -24.9 (4.5) August 2011 Marine 7.2 (0.2) -17.5 (0.2) Centropages sp. September 2010 Brackish 7.0 (0.6) -20.1 (0.3) C March 2011 Brackish 7.9 -19.1 August 2011 Brackish 7.3 -17.0 August 2011 Marine 7.3 (0.1) -17.3 (0.1) Calanoida n.i. September 2010 TFW 9.9 -27.3 Ca September 2010 Brackish 9.5 (0.5) -24.2 (3.1) March 2011 TFW 11.2 (1.0) -33.9 (1.7) August 2011 TFW 12.5 (0.1) -30.3 (0.4) August 2011 Brackish 12.3 (0.2) -30.5 (0.1) Cyclopoida n.i September 2010 TFW 10.0 -27.3 Cy August 2011 Brackish 17.7 (0.1) -21.0 (0.4) Harpaticoida n.i. March 2011 Brackish 8.6 (0.2) -21.2 (3.9) H Branchiopoda Cladocera n.i. March 2011 TFW 10.7 -32.0 Cd Ostracoda n.i. August 2011 Brackish 6.9 (0.03) -16.9 (0.1) O August 2011 Marine 6.5 (0.2) -16.9 (0.1) Malacostraca Caridea n.i. September 2010 TFW 8.1 -27.4 M September 2010 Brackish 5.9 (1.4) -26.3 (1.5) August 2011 TFW 11.1 (0.4) -26.4 (0.3) August 2011 Brackish 11.6 (0.1) -26.1 (0.5) 59 Appendix A (continuation)  Zooplankton taxa Sampling month Location 15N (±SD) ‰ δ13C (±SD) ‰ Code Podon sp. August 2011 Marine 6.8 (0.3) -16.3 (0.1) P Hydracaridae n.i. August 2011 TFW 12.8 -26.4 Hy Polychaeta Polychaeta larvae n.i. March 2011 Brackish 8.0 -18.3 Pl August 2011 Marine 6.8 -16.7 Ascidiacea Oikopleura n.i. August 2011 Brackish 7.0 -18.1 Ok August 2011 Marine 7.3 (0.1) -17.1 (0.1) Osteichthyes Fish larvae n.i. March 2011 Brackish 10.3 (0.2) -18.1 (0.0) Fl August 2011 Brackish 13.4 (0.6) -23.0 (2.8) August 2011 Marine 12.2 -18.8 66 desmaresti and by the epibenthic fishes Cobitis paludica and Platichthys flesus (CostaDias et al. 2010). 3.2.2 Field sampling Sampling was conducted in August 2010 and from January to September 2011, at five stations (S) along the salinity gradient (Fig. 3.1): S1near to the Minho river mouth, S2at the mouth of Coura river, an estuarine tributary of Minho river; ca. 4 km away from the river mouth, S3located at the salinity transition zone, and at 8 km upstream from the river mouth, S4 and S5located at 15 km and 21 km, respectively, upstream the river mouth (tidal freshwater area; TFW). All the stations were sampled during full-moon spring tides. At each station, surface (50-100 cm below the surface) and bottom water samples (0.5 m off the bottom) were collected using a 2 L Ruttner bottle. From these samples, we determined the concentration of chlorophyll a (Chl a: g L-1), concentration and isotopic composition of POM ([POM]: mg L-1, particulate organic carbon (POC) δ13C, particulate nitrogen (PN) δ15N, molar C:N). Salinity was measured with an YSI model 6820 QS probe and reported using the Practical Salinity Scale. The POM and Chl a water samples (POM: 1L, Chl a: 0.5L) were pre-filtered with a 150 µm sieve and filtered onto a pre-combusted (500 ºC for 2 h) Whatman GF/F and Whatman GF/C filters, respectively, and kept frozen (-20 ºC) until analysis. The microphytobenthos (MPB) samples were collected in each station, and at every sampling period, by scraping artificial substrates that were fixed in the sediment and left in the estuary during the period of this study. Macroalgae were collected in S1 and S2, and vascular plants in S3, S4 and S5. Zooplankton samples were collected using a plankton net (200 m mesh) towed near the surface. Samples were immediately preserved in 70% ethanol. Macrozoobenthos were sampled using a van Veen grab. Epibenthic organisms, such as C. crangon, C. maenas, A. desmaresti and fish species, were sampled in August 2010 and January, March, April, July-September 2011 using a 1 m beam trawl (5 mm mesh size) towed at 2 km h-1. All the consumers, except zooplankton, were kept frozen (-20 ºC) until analyses. 3.2.3 Laboratory analyses Filters collected for Chl a analysis were extracted in 90% acetone and analyzed on a Spectronic 20 Genesys spectrophotometer. Chl a concentration was calculated following the equations proposed by Lorenzen (1967). 67 Filters for POM and MPB analysis were fumigated with concentrated HCl to remove inorganic carbonates, rinsed with deionized water, placed in a sterile Petri dish, and dried at 60 ºC for 24 h (Lorrain et al. 2003). Sediment samples were rinsed with 10% HCl to remove carbonates, rinsed with deionized water, and dried at 60 ºC for 48h. Both acidification procedures are expected to produce only slight changes, ca. 0.4‰, in sample 15N values (Lorrain et al. 2003, Carabel et al. 2006). Macroalgae and vascular plants were cleaned with deionized water to remove epiphytes, dried at 60 ºC, and ground to a fine powder with a mixer mill for posterior stable isotopic analyses. All the consumers were sorted, identified, measured when applicable, dried in an oven at 60 ºC and ground to a fine homogeneous powder using a mortar and pestle. The macrozoobenthos samples consisted of whole individuals, except for bivalves where we used the foot muscle for stable isotopes analyses, while for epibentic crustaceans and fish we used muscle tissue. Stable isotope ratios were measured using a Thermo Scientific Delta V Advantage IRMS via a Conflo IV interface (Centro Interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto). Stable isotope ratios are reported in  notation, X: X = (Rsample/Rstandard -1) x 103, where X is the C or N stable isotope, R is the ratio of heavy:light stable isotopes. Pee Dee Belemnite and air are standards for 13C and 15N, respectively. The analytical error, the mean standard deviation of replicate reference material, was ±0.1‰ for δ13C and δ15N. 3.2.4 Data analyses Multivariate analyes were performed to reveal natural groupings in the data according to time and space. Multiple cluster analyses were performed using Euclidean similarity distance based on δ13C and δ15N values of consumers. Organisms were grouped according to their functional feeding group, which included pelagic feeders (e.g. zooplankton; Kleppel 1993), suspension feeders (e.g. clams and poliquetes; Verdelhos et al. 2005, Nordström et al. 2009, Atkinson et al. 2011), benthic grazers (e.g. gastropods; Bode et al. 2006), collectors (e.g. Diptera larvae, amphipods, isopods; Gerdollf and Hughes 1994, Henriques-Oliveira et al. 2003, Cardoso et al. 2004), shredders (e.g. A. desmaresti; Pestana et al. 2007) and predators (e.g. fish, Carcinus maenas, Crangon crangon; Phil 1985). The polychaete H. diversicolor and the isopod C. carinata can also be carnivorous, so they were considered to belong to two distinct functional groups (Wägele et al 1981, Nordström et al. 2009). Cluster analyses were then performed pooling functional groups by station or by month. For these analyses, the months were only used when all the benthic and epibenthic functional groups where available. A similarity profile 68 routine (SIMPROF) was used to test for the presence of sample groups (Clarke et al. 2008). These analyses were performed using PRIMER v.6.1.11® (Clarke and Gorley 2006). One-way ANOVA analyses (factor “station” or “season”) were performed to determine species-specific (epibenthic organisms) or functional group (zooplankton and benthic organisms) changes in stable isotope ratios through time and space. When data did not meet the assumptions (i.e. normality and homogeneity of variances), the nonparametric Kruskal-Wallis test was used. These analyses were performed using STATISTICA 7, Statsoft. Carbon (C) and nitrogen (N) stable isotope ratios bi-plots were used to examine the most likely OM sources for zooplankton and benthic macroinvertebrates. The isotopic ranges used for POM, MPB, EAV, SAV and terrestrial plants were determined in this study. The isotopic ranges for phytoplankton, sediment organic matter (SOM) and C4 saltmarsh plants were compiled from previous studies conducted in this ecosystem (freshwater and estuarine phytoplankton and SOM; Dias et al. in press, Dias et al. unpublished data), and compiled from the literature (marine phytoplankton and C4 saltmarsh plants: Fry and Sherr 1984, Deegan and Garritt 1997, Bode et al. 2007, Vinagre et al. 2008, MacMahon et al 2013). These bi-plots were interpreted in combination with ANOVA, and considering the foraging behavior of each species or functional feeding groups, to infer food web relationships. We used a dual-stable isotope mixing model, that uses Bayesian inference to solve indeterminate linear mixing equations (i.e for two stable isotopes and more than three diet sources), to quantify the importance of pelagic and benthic pathways supporting epibenthic production. Indeterminate linear mixing equations produce a probability distribution that represents the likelihood of a given source (e.g. primary consumer) to contribute to the secondary consumer (Parnell et al. 2010). We used the model Stable Isotope Analysis in R (SIAR), which allows each of the sources and the trophic enrichment factor (TEF; or trophic fractionation) to be assigned a normal distribution (Parnell et al. 2010). The SIAR package (Parnell et al. 2010) is part of the open source statistical language R (R Development Core Team 2007). SIAR produces the distribution of feasible solutions to the mixing problem and estimates credibility intervals (95% CI in this study), which is analogous to the confidence intervals used in frequentist statistics. SIAR also includes a residual error term. In the SIAR mixing model, we adjusted the δ13C and δ15N values for one trophic level using the TEF estimates from Post (2002; +0.4 ± 1.3‰ δ13C, +3.4 ± 1.0‰ δ15N). The consumers δ13C values were corrected for lipid content, because lipids are depleted in 13C compared to protein and carbohydrates (DeNiro and Epstein 1977). 69 Variability in lipid content can bias bulk tissue δ13C values and thereby cause dietary or habitat shifts to be incorrectly interpreted. We corrected zooplankton tissue data for lipid content using the mass balance correction model proposed for zooplankton by Smyntek et al. (2007; Eq. 5). In the case of benthic and epibenthic consumers, we used the mass balance correction for fish muscle tissue, as proposed by Hoffman and Sutton (2010; Eq. 6), which uses estimates of C:Nprotein and Δδ13Clipid that are similar to those from the muscle tissue found for other fish (e.g. Sweeting et al. 2006) and taxonomic groups (e.g. shrimp and zooplankton; Fry et al. 2003, Smyntek et al. 2007). Zooplankton was also corrected for ethanol preservation (+0.4‰ δ13C, +0.6‰ δ15N) (Feuchtmayer and Grey 2003). 3.3. Results 3.3.1 Spatial heterogeneity of estuarine food web components The stations sampled presented distinct salinity patterns (Fig. 3.2). S1 and S2 were overall marine to brackish, with average (±SD) salinity varying between of 23.1 ± 11.3 (S1) and 20.5 ± 12.4 (S2). Salinity in S4 and S5 was always lower than 0.5 (Fig. 3.2), and so they are considered to be freshwater stations. In S3, the salinity values were intermediate, 5.4 ± 8.2 (Fig. 3.2), owing to daily marine water intrusion and its central position in the estuary. The isotopic composition of primary producers varied markedly along the salinity gradient in the Minho River estuary, with δ15N values ranging from -3.0‰ to 14.0‰, and δ13C values ranging from -31.5‰ to -12.8‰ (Fig. 3.3A). These ranges encompass the spatial and temporal variability of all the primary producers sampled. The highest δ15N values were observed for SAV collected in S5, which varied between 7.6‰ and 14.0‰, and the lowest δ15N values were observed for terrestrial plants in the TFW, which varied between -3.1‰ and 9.0‰ (Fig. 3.3A). The highest δ13C values were observed for macroalgae (-19.0‰ to -12.8‰ in S1) and MPB (-20.3‰ to -16.4‰ in S1), and the lowest were observed for freshwater phytoplankton (-32.2‰ to -29.2‰ in the TFW) (Fig. 3.3A). The δ13C values of primary producers (phytoplankton, MPB) collected along the entire gradient increased towards the river mouth (Fig. 3.3A). The POM pool, which is a mixture of different sources, including phytoplankton and vascular material, also varied along the salinity gradient. The δ15N values of POM inceased ca. 2‰ towards the TFW (Fig. 3.4A), while the δ13C values followed an inverse 70 pattern. On average (± SD), POM δ13C values in stations closer to the river mouth (S1 and S2) were higher than those in the TFW: -24.0 ± 2.1 ‰ and -28.0 ± 1.0 ‰, respectively (Fig. 3.4B). The average (± SD) POM δ13C values in S3 were intermediate between S1-S2 and S4-S5: -26.0 ± 1.9 ‰ (Fig. 3.4B). The quality of the POM pool, as indicated by C:NPOM, was similar between stations and varied between 8.6 ± 2.3 (S3) and 10 ± 1.7 (S4) (Fig. 3.4C). The C:NPOM close to 10 in TFW indicates that terrestrial-derived OM contributed substantially to the POM pool in this portion of the estuary (Hedges et al. 1986, 1997). In the brackish portion of the estuary (S1-S3), C:NPOM varied between 8 and 9 (Fig. 3.4C), which is within the range of terrestrial-derived OM (>10) and marine phytoplankton (~7) (Hedges et al. 1986, 1997), thus suggesting that brackish POM was a mixture of riverine and marine POM. Fig. 3.2 Salinity variation along the Minho River estuary, in the 5 sampling stations (S) sampled along the estuarine mixing gradient during the studied period (2010-2011). 71 Fig. 3.3 δ15N and δ13C values of primary producers (A) and primary and secondary consumers (B) collected in five stations (S1-S5) along the salinity gradient of the Minho River estuary, between 2010 and 2011. The primary producers include particulate organic matter (POM), microphytobenthos (MPB), phytoplankton, macroalgae, Terrestrial plants, emergent (non identified) and submerged (Elodea canadensis) aquatic vegetation. Consumers include the pelagic primary consumers (PC), benthic consumers (BC) and epibenthic consumers (EC). The y-axis scale varies between graphs. 72 Fig. 3.4 Monthly average values (±SD) of particulate nitrogen (δ15NPN; A), particulate organic carbon (δ13CPOC; B), particulate organic matter C:N ratios (C:NPOM; C) and chlorophyll a (Chl a) concentration (g L-1; D) in each sampling station, along the Minho River estuary, during the sampling period. Overall, consumers were 15N-depleted and 13C-enriched towards the river mouth. Pelagic consumers were 15Ndepleted towards the river mouth (H(4,77)= 40.46, p< 0.001), with average (±SD) values varying between 8.9 ± 0.9 ‰ (S1) and 11.5 ± 0.5 ‰ (TFW) (Fig. 3.5A). Although some benthic consumers (C. carinata and H. diversicolor) were 15Ndepleted towards the river mouth (Fig. 3.5B), the overall differences were not statistical significant (F(4,115)= 1.34, p= 0.25). In the brackish portion of the estuary, C. carinata was the benthic consumer with the highest δ15N values (close to 12‰), and Scrobicularia plana was the benthic consumer with the lowest δ15N values (7.3 ± 1.8 ‰) (Fig. 3.5B). In the TFW, insect larvae had the highest δ15N values (11.1 ± 1.5 ‰), while C. fluminea (9.0 ± 0.7 ‰), Gastropoda (9.0 ± 1.9 ‰) and Oligochaeta (9.1 ± 1.1 ‰), had the lowest (Fig. 3.5B). Overall, epibenthic consumers were 15N-depleted towards the river mouth (F(3,26)= 5.99, p= 0.003), with the exception of P. flesus. Nonetheless, the highest δ15N values were observed in S2, varying between 12.6 ± 0.5 ‰ (C. crangon) and 14.6 ± 0.4 ‰ (P. microps) (Fig. 3.5C). In stations S1, S2 and S3, P. microps had the highest δ15N values and C. crangon had the lowest δ15N values (Fig. 3.5C). In the TFW, A. desmaresti had the lowest δ15N values (8.0 ± 0.4 ‰) and Gasterosteus aculeatus had the highest δ15N values (14.8 ± 0.5 ‰) (Fig. 3.5C). 73 Fig. 3.5 Monthly δ15N (A-C) and δ13C (D-F) average values (‰) of pelagic (A, D), benthic (B, E) and epibenthic consumers (C, F), collected along the Minho River estuary during 2011. Pelagic consumers were 13C-enriched towards the river mouth (H(4,77)= 44.0, p< 0.001) with values varying between -17.5 ± 1.1 ‰ (S2) and -28.5 ± 3.4 ‰ (TFW) (Fig. 3.5D). The δ13C values of benthic consumers were significantly enriched towards the river mouth (H(4, N=120)= 86.33, p< 0.001), ranging from -27.2 ± 1.0‰ (C.fluminea) to -24.1 ± 1.6‰ (insect larvae) at the TFW, and from -22.2 ± 2.5 ‰ (Corophium sp.) to -14.7 ± 0.9 ‰ (C. carinata) at S1 and S2 (Fig. 3.5E). Overall, the δ13C values of C. carinata were the highest among benthic consumers in the stations S1-S3 (-21.9‰ to -13.2‰), and the δ13C values of H. diversicolor (-31.2‰ to -17.5‰) in S1-S2, and C. fluminea (-27.5‰ to - 24.4‰) in S3, were the lowest (Fig. 3.5E). 74 Fig. 3.6 Clustering analyses performed by station (A, B) and month (C, D) for δ13C (A, C) and δ15N (B, D) values for assemblages consisting of the main functional groups of consumers. Grey lines indicate groups of samples not separated (at < 0.05) by SIMPROF. 75 In the TFW the benthic consumers with higher δ13C values were the insect larvae (δ13C: - 25.2 ± 2.1 ‰) and the oligochaeta (δ13C: -24.1 ± 1.6 ‰) (Fig. 3.5E). The benthic consumer with the lowest δ13C values was C. fluminea (δ13C: -27.2 ± 0.9 ‰) (Fig. 3.5E). The δ13C values of epibenthic consumers were enriched towards the river mouth (F(3,26)= 53.05, p< 0.001), and ranged from -25.0 ± 0.2 ‰ (Cobitis paludica) in the TFW to -16.5 ± 1.1 ‰ (C. crangon) in S2 (Fig. 3.5F). Overall, P. flesus was the epibenthic predator with the lowest δ13C values (S1, S2, S3), contrarily to C. crangon (S1, S2, S3) and G. aculeatus (TFW) which presented the highest δ13C values (Fig. 3.5F). The cluster analyses grouped consumers collected in S4 and S5 (Fig. 3.6). Consumers collected in S1 and S2 were also clustered, but consumers collected in S3 were grouped differently according to the stable isotope ratio analyzed. 3.3.2 Temporal variability of estuarine food web components The δ15N and δ13C values of the majority of the estuarine food web components varied seasonally. For POM samples (Fig. 3.7), the δ15NPN values increased in the summer, although the lowest values were observed in all stations in August 2011, except in S4 (Fig. 3.7A). The δ13CPOC values increased in the summer in brackish stations, but decreased in the summer in the TFW area (Fig. 3.7B). The period when the largest differences between stations S1 and S2 versus the TFW stations in δ15NPN and δ13CPOC values were observed corresponded to the occurrence of a Chl a maximum in the estuary (August 2011: 2.3 ± 2.2 µg L-1) (Fig. 3.7D). Inter-annual differences were observed in the POM values in all stations (Fig. 3.7). In August 2010, the POM pool was 13Cenriched in relation to August 2011 in S2 and in TFW stations, and 13Cdepleted in S3. The POM δ15N values were similar in all stations in August 2010 (between 4.8‰ and 4.9‰), but were 15N-enriched in relation to samples collected in S1 in August 2011 (Fig. 3.7). 82 Z -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 15N (‰) 0 2 4 6 8 10 12 14 16 MA MPB MP EP POM SOM Cy Hd Sp Pm Pf Cm Cc S1 - Winter/Spring Z MA S1 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 0 2 4 6 8 10 12 14 16 SOM EP POM MP MBP Cy Hd Sp Z Pm Pf Cm Cc MA S1 - Summer 13C (‰) -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 15N (‰) 0 2 4 6 8 10 12 14 16 SOM POM MPB MA C4 plants C3 plants MP EP Cy Hd CZ Pm Pf Cc Cm S2 - Winter/Spring 13C (‰) -28 -26 -24 -22 -20 -18 -16 -14 -12 0 2 4 6 8 10 12 14 16 SOM C3 plants POM MPB MA C4 plants MP EP Z Cy Hd Pm Cm Pf Cc S2 - Summer Fig. 3.10 Average (±SD) δ15N and δ13C values (‰) of pelagic (white symbols), benthic (black symbols) and epibenthic consumers (grey symbols) in the brackish (S1-S3) and freshwater (TFW) areas of the Minho River estuary during winter-spring and summer 2011. The consumers shown are: zooplankton (Z), Scrobicularia plana (Sp), Hediste diversicolor (Hd), Cyathura carinata (Cy), Corophium sp. (C), Corbicula fluminea (Cf), Insect larvae (In), Oligochaeta (O), Gastropoda (G), Crangon crangon (Cc), Carcinus maenas (Cm), Atyaephyra desmaresti (Ad), Pomatoschistus microps (Pm), Platichthys flesus (Pf), Gasterosteus aculeatus (Ga) and Cobitis paludica (Cp). Boxes represent the ranges for the sources collected during this study, and also the estimates values for C4 saltmarsh plants and phytoplankton (see text). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). 83 -35 -30 -25 -20 -15 15N (‰) 0 2 4 6 8 10 12 14 16 SOM FP POM MPB MP EP Cy Hd Cf C Z Pm Pf Cc Cm S3 - Winter/Spring -35 -30 -25 -20 -15 0 2 4 6 8 10 12 14 16 SOM POM MPB EP MP FP C3 plants Cy Hd Cf C Z Pm Pf Cm Cc S3 - Summer 13C (‰) -35 -30 -25 -20 -15 15N (‰) -5 0 5 10 15 SOM Terr POM MPB EAV SAV EP FP ZPf O Cf C Ad In Ga TFW - Winter/Spring G 13C (‰) -35 -30 -25 -20 -15 -2 0 2 4 6 8 10 12 14 16 SAV SOM POM Terr EAV EP FP MPB C Cf In O Ad Pf Ga Cp G Z TFW - Summer Fig. 3.10 (continuation) Average (±SD) δ15N and δ13C values (‰) of pelagic (white symbols), benthic (black symbols) and epibenthic consumers (grey symbols) in the brackish (S1-S3) and freshwater (TFW) areas of the Minho River estuary during winter-spring and summer 2011. The consumers shown are: zooplankton (Z), Scrobicularia plana (Sp), Hediste diversicolor (Hd), Cyathura carinata (Cy), Corophium sp. (C), Corbicula fluminea (Cf), Insect larvae (In), Oligochaeta (O), Gastropoda (G), Crangon crangon (Cc), Carcinus maenas (Cm), Atyaephyra desmaresti (Ad), Pomatoschistus microps (Pm), Platichthys flesus (Pf), Gasterosteus aculeatus (Ga) and Cobitis paludica (Cp). Boxes represent the ranges for the sources collected during this study, and also the estimates values for C4 saltmarsh plants and phytoplankton (see text). The dotted lines indicate the expected stable isotope ratio of consumers utilizing phytoplankton, assuming typical trophic fractionation (+0.4‰ δ13C and +3.4‰ δ15N per trophic level). 84 largely comprised of phytoplankton (C:NPOM ~7). The δ13C values of these consumers were similar to δ13CPOC values and also followed the same trend as POC (Figs. 3.7B, 3.8). The benthic primary consumers were the most relevant prey for the epibenthic secondary consumers analyzed, except for C. crangon, according to the SIAR mixing model results Overall, pelagic consumers (PC) were the most important prey for C. crangon in S1 and S2, with the highest values (95% CI) in April (S2: 39.5-87.2%) and August 2011 (S1: 54.3-93.4%; S2: 43.3-61.2%) (Fig. 3.11). Inter-annual variability was observed in S2. In August 2010, PC were the source with the lowest contribution (0.027.3%) to C. crangon biomass, whereas the contribution of the suspension feeder (SF/C) H. diversicolor (12.3-60.5%) and of the deposit feeder (Cl/C) C. carinata (34.8-64.6%) increased, when compared to August 2011 estimates (Fig. 3.11). In S3, collectors (Cl) were the most important prey for C. crangon, followed by PC, and their importance decreased throughout summer (Fig. 3.11), in opposition to SF/C and Cl/C. The most important prey for C. maenas and P. microps in S1 and S2 were the SF/C, but the proportion of this source varied monthly (Figs. 3.12, 3.13). In S3, the most important prey identified for C. maenas were the Cl (20.3-65.5%) and SF/C (1.3-58.2%) (Fig. 3.12). For P. microps, the Cl/C were the most important prey during winter, decreasing towards summer (Fig. 3.13). We ran two models for flounder, one for the brackish estuary (S1-S3) and another for the TFW area (S4-S5), because we noticed that some flounders δ13C and δ15N values did not represent the expected stable isotope ratios for the station where they were collected. For instance, some flounders collected in S5, which is the uppermost station, had δ13C values between -24‰ and -22‰, which are similar to those shown by P. microps collected in S3 (13 km downstream). Therefore, we selected the flounders that had stable isotope ratios (especially δ13C) similar to those from other less mobile species, such as P. microps for the brackish water stations and C. paludica (A. desmarestii in the absence of C. paludica) for the TFW area. Thus, in the brackish estuary, the main prey of P. flesus were the benthic invertebrates, especially collectors, with its importance decreasing during summer. The opposite trend occurred for PC and SF/C (Fig. 3.14). Overall, the most important prey for P. flesus were the shredders (S) in the TFW area, with the highest contributions in March (68.1-90.9%) and August 2011 (46.7-89.6%), followed by the Cl and benthic grazers (BG). The PC were only relevant in July 2011 (1.7-49.8%) (Fig. 3.14). 85 0 10 20 30 40 50 60 70 80 90 100 Proportion Sources A - STATION 1 C/Cl PCCl SF/C 0 10 20 30 40 50 60 70 80 90 100 Proportion Sources B - STATION 2 C/Cl PCCl SF/C 0 10 20 30 40 50 60 70 80 90 100 Proportion Sources C - STATION 3 C/Cl PCCl SF/C Fig. 3.11 Proportion of each food source to Crangon crangon biomass in stations 1, 2 and 3 in August 2010 (August’), and January-April 2011, July-September 2011, based on the stable isotope mixing model results. The food sources included in the model were the pelagic consumers (PC), suspension feeders and carnivorous (SF/C), collectors and carnivorous (Cl/C), and collectors (Cl). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 86 Fig. 3.12 Proportion of each food source to Carcinus maenas biomass in stations 1, 2 and 3 in August 2010 (August’), and January-April 2011, July-September 2011, based on the stable isotope mixing model results. The food sources included in the model were the suspension feeders and carnivorous (SF/C), collectors and carnivorous (Cl/C), collectors (Cl), crustacean predators (Pr1), and fish predators (Pr2). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 87 Fig. 3.13 Proportion of each food source to Pomatoschistus microps biomass in stations 1, 2 and 3 in August 2010 (August’), and January-April 2011, July-September 2011, based on the stable isotope mixing model results. The food sources included in the model were the pelagic consumers (PC), suspension feeders and carnivorous (SF/C), collectors and carnivorous (Cl/C), and collectors (Cl). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 88 Fig. 3.14 Proportion of each food source to Platichthys flesus biomass in the brackish and TFW portions of the Minho River estuary in August 2010 (August’), and January-April 2011, JulySeptember 2011, based on the stable isotope mixing model results. The food sources included in the model were the collectors (Cl), collectors and carnivorous (Cl/C), pelagic consumers (PC), suspension feeders and carnivorous (SF/C), benthic grazers (BG), and shredders (S). Data presented include the most likely solution (mode) and the 95% Bayesian credibility intervals. 4 and 5 refer to the results from the isotope mixing model for stations 4 and 5, respectively, due to the existence of significant differences between sessile (Corbicula fluminea) or less mobile (Atyaephyra desmaresti) stable isotope values in August 2011. 3.4 Discussion This study provides evidence for spatial and temporal variability in the River Minho estuarine food web. Spatial variability was essentially related with the characteristics of the organic matter sources available in each area, which was also influenced by the proximity to other ecosystems, namely higher terrestrial influence in the tidal freshwater portion and higher marine influence in the brackish portion of the estuary. Temporal changes in the stable isotope ratios of consumers were driven essentially from changes 89 occurring at the base of the food web, although there were also some evidences for a combination with changes in the use of resources by the consumers. The possible mechanisms behind the spatial and temporal variability are discussed below. 3.4.1 Spatial heterogeneity of estuarine food web components During this study, a large variability in the isotopic composition of primary producers was observed along the salinity gradient in the Minho River estuary, with δ15N values ranging from -3.0‰, to 14.0‰, and δ13C values ranging from -31.5‰ to -12.8‰. The variability in the δ13C values of primary producers were essentially related with differences in the source of carbon dioxide used. Commonly, C uptake by C3 plants, on land, involves a net fractionation of about 21‰ between the atmosphere (-7‰) and plant biomass resulting in δ13C values of ca. -28‰ (Peterson and Fry 1987). Primary producers uptaking C from solution will display more variable δ13C values owing to the higher variability in sources of dissolved CO2, which include carbonate rock weathering, atmosphere or respired OM (Peterson and Fry 1987). In consequence of that variability, several studies showed a direct relationship between dissolved inorganic carbon (DIC) δ13C values and salinity, where the mixing between riverine DIC 13C-depleted with the marine DIC 13Cenriched, creates a gradient in the δ13CDIC values along the estuarine mixing gradient (Chanton and Lewis 2002, Fry 2002). Thus, the primary producers, that obtain inorganic carbon from solution, will be influenced by the portion of the estuary where they were collected. However, the composition of the OM pool available in each portion of the estuary was also different, and contributed to the observed variability. Marine phytoplankton and macroalgae (13C-enriched) were only available in the brackish portion of the estuary, while in the TFW there was an increasing contribution of vascular material, which included aquatic and terrestrial vegetation (in general, 13C-depleted). River discharge also influenced the quality of the POM available in each portion of the estuary. During high river discharge periods, the contribution of terrestrial-derived OM increased in the TFW, and during low river discharge, marine intrusion was detected up to station S3, which likely explains some of the highest δ13CPOC values observed (δ13C: -24‰ to -22‰) as 13Cenriched marine POM (δ13C ca. -20‰, typical of marine phytoplankton; MacMahon 2013) mixed with 13C-depleted estuarine POM. Primary consumers were 13C-enriched towards the mouth of the estuary, reflecting the variability in the OM sources available in each station. In the TFW, terrestrial-derived OM was likely an important contributor to pelagic and benthic consumers, especially 90 during winter. In the brackish estuary, marine phytoplankton (or MPOM) was an important energy source, especially during summer owing to high marine intrusion. Another factor influencing both δ15N and δ13C values was the habitat (pelagic vs. benthic) from where primary consumers obtained their energy. Primary consumers relying on benthic or detrital C, such as C. carinata or insect larvae, were generally more 15Nand 13Cenriched than those that relied mainly on OM sources in the pelagic pathway, such as zooplankton or H. diversicolor. One plausible explanation is related with the existence of a diffusive boundary layer at the sediment-water interface that reduces isotopic fractionation (France 1995). The between-habitat differences in the stable isotope ratios of primary consumers appear to influence the isotopic composition of both crustacean and fish epibenthic consumers, which were also 13C-enriched towards the river mouth. Despite these differences, SIAR mixing results indicate that P. microps and C. maenas relied essentially on benthic filter feeding organisms (i.e. polychaetes), regardless of the station where they were collected. However, C. crangon and P. flesus presented some differences according to the station where they were collected. Based on stable isotope ratios, C. crangon was feeding on 13C-enriched and 15N-depleted sources in S1 and S2 (zooplankton and/or H.diversicolor), and on 13Cand 15N-enriched sources in S3 (collectors and/or carnivorous benthic consumers). SIAR mixing model results suggest that C. crangon was supported essentially by pelagic primary consumers in S1 and S2, and by benthic primary consumers (collectors) in S3, shifting from a general pelagic pathway to an increase reliance on benthic pathway upriver. Prey abundance may explain the differences observed between stations, because gut content analyses indicated that epibenthic predators, like P. microps, C. crangon or C. maenas, are opportunistic carnivore that select prey based on relative availability (Phil 1985). However, to the best of our knowledge, thorough studies are missing on the relative abundance of zooplankton and benthic macroinvertebrates along the salinity mixing gradient in the Minho River estuary. Platichthys flesus relied mostly on the detrital pathway, by feeding on collectors in the brackish portion of the estuary, and on shredders in the TFW portion of the estuary. Collectors essentially feed on fine POM (0.5µm-1 mm) from a great variety of resources including coarse POM (> 1 mm) fragmentation, periphyton, phytoplankton and microorganisms (Henriques-Oliveira et al. 2003). Shredders feed on coarse POM, which includes leaves from the riverine vegetation and macrophytes (Henriques-Oliveira et al. 2003). 91 3.4.2 Temporal variability in the estuarine food web Seasonal changes were observed in the stable isotope ratios of POM, MPB and SAV. Although cluster analysis did not reveal any clear seasonal pattern in the stable isotope ratios of consumers, there was an overall trend for 13C-enrichment during summer. Temporal variation in C and N isotope ratios of OM sources were expected to occur due to hydrologic variability and to annual senescence of vascular plants. Cloern et al. (2002) found high variability (approximately 5‰ to 10‰) in monthly δ13C and δ15N values within some plant groups, associated with annual cycles of growth and senescence in the San Francisco Bay estuary. We cannot describe well the temporal variability in isotopic composition of aquatic plants because they were sampled according to habitat (i.e. aquatic, emergent or submerged), and thus a species-specific analysis it is not possible. However, seasonal variability was observed in the stable isotope ratios of MPB and POM. MPB were 13C-enriched during summer (except in S2), suggesting that benthic phytoplankton was using a 13C-enriched marine dissolved inorganic pool, which can be associated with a higher marine intrusion occurring under low river discharge periods. The δ13CPOC values increased towards summer in brackish portion of the estuary, while the opposite pattern was observed in stations located at the TFW portion of the estuary. During summer, river discharge declined, which decreases terrestrial inputs and increases residence time, favoring the accumulation of living and detrital phytoplankton (Hoffman and Bronk 2006). Our results suggest that during summer, and especially in August 2011, there was an increase in the contribution of phytoplankton to the POM pool. The concentration of Chl a increased during summer, peaking in August 2011 in the TFW stations (up to 8 times) and in the middle of the estuary (up to 6 times), suggesting the occurrence of a phytoplankton bloom during this month. Moreover, the δ13CPOC values in the TFW (δ13CPOC: -30‰) were similar to those estimated for freshwater phytoplankton in this estuary (δ13C: -31‰; unpublished) and the C:NPOM lower than 10, which suggests a decrease in the contribution of terrestrial-derived OM to the POM pool. The opposite occurred during winter, when the river discharge increased. The δ13CPOC values (δ13C: - 28‰ to -24‰) and C:NPOM ratio (> 10) suggest a substantial contribution of terrestrialderived OM to the POM pool, advected from upland or riparian habitats or both (Hedges et al. 1997, Hoffman et al. 2008). The low Chl a concentrations (0.8 ± 0.8 µg L-1) during this period also suggests that the contribution of phytoplankton to the POM pool decreased owing to the suppression of primary production as a consequence of increased turbidity and rapid flushing rates (Sin et al. 1999, Hoffman and Bronk 2006). 98 senegalensis in the Tagus estuary (Portugal): A stable isotope approach. Estuar Coast Shelf S 76: 255-264. Wägele J-W, Welsch U, Müller W (1981) Fine structure and function of the digestive tract of Cyathura carinata (Kroyer) (Crustacea, Isopoda). Zoomorphology 98: 69-88. Yokoyama H, Ishihi Y (2007) Variation in food sources of macrobenthos along a land-sea transect: a stable isotope study. Mar Ecol Progr Ser 346: 127-141. Yokoyama H, Tamaki A, Harada K, Shimoda K, Koyama K, Ishihi Y (2005) Variability of diet-tissue isotopic fractionation in estuarine macrobenthos. Mar Ecol Progr Ser 296: 115-128. Zeug SC, Winemiller KO (2008) Evidence supporting the importance of terrestrial carbon in a large-river food web. Ecology 89: 1733-1743. 99 Chapter 4 Linking terrestrial and benthic estuarine ecosystems: organic matter sources supporting the high secondary production of a non-indigenous bivalve Ester Dias1,2, Pedro Morais2,3,4, Carlos Antunes2,5,6, Joel C. Hoffman7 1 ICBAS - Instituto de Ciências Biomédicas Abel Salazar, Universidade do Porto, Rua de Jorge Viterbo Ferreira 228, 4050-313 Porto, Portugal. 2 CIIMAR/CIMAR - Centro Interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto, Rua dos Bragas 289, 4050-123 Porto, Portugal. 3 CIMA - Centro de Investigação Marinha e Ambiental, Campus de Gambelas, Universidade do Algarve, 8005-139 Faro, Portugal. 4 ICCE - International Center for Coastal Ecohydrology, Solar do Capitão-Mor, Horta das Figuras, EN 125, 8005-518 Faro, Portugal. 5 Aquamuseu do Rio Minho, Parque do Castelinho s/n, 4920-290 Vila Nova de Cerveira, Portugal. 6 ESGEscola Superior Gallaecia, Largo das Oliveiras s/n, 4920 Vila Nova de Cerveira, Portugal. 7 Mid-Continent Ecology Division, National Health and Environmental Effects Research Lab, US Environmental Protection Agency, 6201 Congdon Blvd, Duluth, Minnesota, 55804, USA. Dias E, Morais P, Antunes C, Hoffman JC (in press) Linking terrestrial and benthic estuarine ecosystems: organic matter sources supporting the high secondary production of a non-indigenous bivalve. Biol Invasions. doi: 10.1007/s10530-014-0655-8 100 Abstract The Asian clam, Corbicula fluminea, is among the most pervasive invasive species in freshwater ecosystems worldwide. Our objective was to study C. fluminea’s functional response in terms of feeding behavior and food selectivity, using the natural variation in organic matter (OM) sources that occur in estuarine environments. Using C and N stable isotopes, we identified and quantified the contribution of different OM sources supporting the production of C. fluminea along the salinity gradient occupied in the Minho River estuary (NW-Iberian Peninsula, Europe), where this species presently dominates the benthic macrofauna biomass. We observed a pronounced shift in the quality of OM available for C. fluminea along the estuarine mixing zone. Stable isotope analysis, POM C/N, and phytoplankton contribution estimates based on C: Chl a revealed that POM was largely comprised of terrestrial-derived OM in freshwater stations (TFW) and was increasingly comprised of phytoplankton, a more palatable food source, towards the polyhaline estuary. A similar shift in the isotopic composition along the estuarine mixing zone was observed in C. fluminea, suggesting a shift in food resources. Accordingly, based on a Bayesian stable isotope mixing model, there was an upstream-downstream counter gradient in the contribution to C. fluminea biomass from terrestrial-derived OM (41-64% in TFW) and phytoplankton (29-55% in the brackish estuary). Although the majority of the food sources identified were filtered from the water column (70-80%), reliance on sediment OM and microphytobenthos provided evidence for deposit feeding by C. fluminea. We conclude that C. fluminea has the ability to adapt to environments with low food quality because it can consume terrestrial-derived OM. This can be a competitive adaptation in systems with perennial low food quality such as the Minho River estuary. Moreover, its ability to couple benthic and pelagic environments and terrestrial ecosystems demonstrates a strong potential to alter food web flows in aquatic ecosystems. Keywords: Corbicula fluminea, stable isotopes, benthic food web, Minho estuary. 101 4.1 Introduction The Asian clam, Corbicula fluminea, is among the most widespread invasive species in freshwater ecosystems (Sousa et al. 2008a). Native to Southeast Asia, this species has dispersed worldwide. It was reported for the first time in Europe in the early 1980s (Mouthon 1981) and presently is spread throughout the continent (e.g. Araujo et al. 1993, Morais et al. 2009, Munjiu and Shubernetski 2010, Caffrey et al. 2011). Corbicula fluminea can cause several ecological and economic impacts: it can compete with and displace native mollusc populations (Hakenkamp and Palmer 1999, Vaugh and Hakenkamp 2001), influence biogeochemical cycles (Phelps 1994, Cherry et al. 2005, Cooper et al. 2005) and biofoul human water systems (Darrigran 2002). For these reasons, it has been listed as one of the 100 worst invasive species in Europe (DAISIE 2008). However, C. fluminea can also benefit benthic species; shell production can increase substrate availability for attachment and refugia (Gutiérrez et al. 2003), and C. fluminea faeces and pseudo-faeces may constitute a food resource for other organisms (Howard and Cuffey 2006, Gergs et al. 2011). Its high salinity tolerance (up to salinities of 10-14) enables the clam to colonize both freshwater and brackish habitats (Morton and Tong 1985) within estuaries. High densities have been documented in numerous estuaries: 4,185 clams m-2 in the Minho River estuary (Sousa et al. 2005), >5,000 clams m-2 in the Paraná River delta (Boltovskoy et al. 1995), and 11,142 clams m -2 in the Mondego River estuary (Franco et al. 2012). The invasion success of this species has been attributed to their life history characteristics, including rapid individual growth, early sexual maturity, high fecundity, broad environmental tolerance, wide genetic variability and phenotypic plasticity (Morton 1997). Corbicula fluminea has numerous feeding strategies and this allows it to compete with native species through multiple trophic pathways. This species has high filtering capacity (0.3-10 m3.m-2.d-1; Strayer et al. 1999), can non-selectively suspension feed (Way et al. 1990, Boltovskoy et al. 1995), and has the ability to deposit feed when little planktonic food is available (Hakenkamp and Palmer 1999; Vaugh and Hakenkamp 2001). Previous studies revealed that C. fluminea can select components present within seston but has a low trophic fidelity (Atkinson et al. 2010), which probably favours its successful invasion. In estuaries, characterizing the food sources of C. fluminea is complicated by the spatial and temporal complexity of estuarine food webs (Canuel et al. 1995, Deegan and Garritt 1997). Estuarine food webs can incorporate organic matter (OM) inputs from many sources, including riparian vegetation, submerged and emergent aquatic vegetation (and associated epiphytic algae), as well as phytoplankton and microphytobenthos produced in 102 situ (Cloern et al. 2002, Hoffman and Bronk 2006). Carbon (δ13C) and nitrogen (δ15N) stable isotope analysis is a powerful tool to characterize energy flow through estuarine food webs (Pasquaud et al. 2007, Hoffman et al. 2008). For consumers, the stable isotope composition (13C/12C and 15N/14N) of tissues is a time-integrated signal of the food sources in the ecosystem that were incorporated into an organism’s structural components and energy reserves (Peterson and Fry 1987). Thus, the stable isotope ratio of a consumer reflects its diet, demonstrating an average trophic fractionation (i.e., the difference between the consumer and its diet) of +0.4‰ δ13C and +3.4‰ δ15N per trophic level (Post 2002). The C and N stable isotope ratios of upland plants, marsh vegetation, and freshwater and estuarine algae differ with respect to C and N source and method of C fixation (Goerike et al. 1994, Lajtha and Marshall 1994). The riparian plants that use the C3 pathway have a δ13C of about -28‰ because there is an uptake fractionation of about -21‰ over atmospheric CO2 (δ13C -7‰) (Smith and Epstein 1971). In contrast, C4 plants (e.g. Spartina spp.) are more enriched in the heavy isotope (δ13C -13‰) owing to reduced fractionation (Smith and Epstein 1970, Fry and Sherr 1984). Freshwater and estuarine phytoplankton will also differ where they utilize isotopically distinct pools of dissolved inorganic carbon (DIC). As with C3 vascular plants, their 13C fractionation is about -21‰, but can shift with DIC concentration, phytoplankton growth rate, and nutrient availability (Goericke et al. 1994). In freshwater ecosystems, where the DIC δ13C is highly depleted (lower than -10‰), phytoplankton may also be distinguished from riparian vegetation (Hoffman and Bronk 2006). In general, microphytobenthos (MPB) are more 13C-enriched than phytoplankton due to the existence of a diffusive boundary layer at the sedimentwater interface that reduces isotopic fractionation (France 1995). The N stable isotope composition can also help to separate 15N-depleted terrestrial OM (δ15N -4‰ to 4‰) from 15N-enriched aquatic OM sources (δ15N 6‰ to 10‰) (Peterson and Fry 1987, Cloern et al. 2002). The C/N ratios can aid source discrimination. For example, higher plant tissues are characteristically carbon-rich (C/N: 20-500) compared to algae (C/N ~ 7) owing to the predominance of N-free biomacromolecules (e.g. lignin, tannin, hemicellulose, cellulose, cutin, and suberin) over proteins (C/N: 3-4) in higher plants (Hedges et al. 1986). However, this distinction can be blurred as vascular plant tissues preferentially gain N during microbial decay, and as plankton preferentially lose N versus C during decay (Hedges et al.1997 and references therein). The objective of this research was to study C. fluminea’s functional response in terms of both feeding behavior and food selectivity by using the natural variation in available OM sources that occurs in an estuarine environment over the salinity gradient occupied by C. fluminea. We studied this species in the Minho River estuary (NW-Iberian Peninsula, Europe), where it represents more than 90% of the total benthic macrofauna 103 biomass and is present at high densities (up to 4,185 clams m-2). In 2005, its production was estimated to be about 465 g ash free dry weight (AFDW) m-2 year-1 (Sousa et al. 2005, 2008b). Because C. fluminea has high filtering capacity (Strayer et al. 1999), we hypothesized that where phytoplankton contribution to the suspended OM pool was high, C. fluminea would respond by preferentially deriving energy from this high-quality food source; however, where the available suspended OM quality was low (e.g. due to high contributions of detritus derived from terrestrial OM) or quantity limited, C. fluminea would increase the relative proportion of deposit feeding. To test our hypothesis, we identified potential OM sources to the food web and quantified their respective contributions to the production of C. fluminea along the estuarine mixing zone in the Minho River estuary using C and N stable isotope analysis. The impact that an invasive species may have in an ecosystem is related to its abundance or biomass (or both; Parker et al. 1999, Ricciardi 2003). Thus, knowledge about the feeding strategies that are utilized by this highly abundant population can further our understanding of the underlying mechanisms by which C. fluminea can alter energy flows in an invaded ecosystem. 4.2 Methods 4.2.1 Study area The Minho River is located in the NW-Iberian Peninsula (SW Europe; Fig. 4.1). The annual average discharge is 300 m3 s-1 (Ferreira et al. 2003). Its watershed is 17,080 km2, of which 95% is located in Spain and 5% in Portugal. The river is 343 km long; 76 km serves as the northwestern Portuguese-Spanish border (Antunes et al. 2011). The limit of tidal influence is about 40 km inland, and the uppermost 30 km are a tidal freshwater wetland (TFW; Sousa et al. 2008b). Its estuary has an area of 23 km2, of which only 9% is intertidal. The estuary is mesotidal with tides ranging between 0.7 m and 3.7 m (Alves, 1996). The mean depth of the estuary is 2.6 m and the maximum depth is ca. 26 m (Antunes et al. 2011). Due to its ecological importance, the Minho River estuary and the international section of the River Minho were designated as a Natura 2000 site (EIONET 2012) and as an Important Bird Area (BirdLife International 2012). 104 Fig. 4.1 Location of the sampling stations along the Minho River estuary. 4.2.2 Field sampling In September 2010, 14 stations located at 1 kilometer intervals along the main river channel were sampled during full-moon spring tide. The first and the last sampling stations were located at 8 km and 21 km away from the river mouth, respectively (Fig. 4.1). These stations encompass the distribution range of C. fluminea along the brackish and tidal freshwater areas. Sampling was conducted at a small spatial scale to guarantee a proper characterization of the estuarine mixing zone. We chose to sample in September, near the end of the annual low flow period (typically, from July through October; SNIRH 2012), for a number of reasons. First, salt water intrusion is maximized, which favors the connectivity between the estuarine and marine environments. Second, residence time is elongated during low flow, which potentially allows phytoplankton biomass to accumulate. Third, without pronounced and varying riverine inputs, biogeochemical variability is reduced, which results in reduced variability in the OM sources and isotopic character of suspended particles (Hoffman and Bronk 2006). Based on the feeding modes of C. fluminea (e.g. Boltovskoy et al. 1995, Hakenkamp and Palmer 1999), the OM sources of interest were phytoplankton (freshwater, estuarine and marine), microphytobenthos (MPB), sediment organic matter (SOM), and particulate organic matter (POM; a mixture of OM inputs to the estuary). We directly sampled MPB, SOM and POM, but used a proxy for phytoplankton (using DIC δ13C; δ15N values were obtained from POM in a similar cruise done in August 2011) since it is difficult to isolate algae from the complex bulk POM present in estuarine environments. 105 At each station, surface (50-100 cm below the surface) and bottom water samples (0.5 m off the bottom) were collected using a 2-L Niskin bottle. From these samples, we measured the concentration of chlorophyll a (Chl a: g L-1), concentration and isotopic composition of POM (mg L-1, δ13CPOC, δ15NPN, molar C/N), and isotopic composition of total dissolved inorganic carbon (CO2: δ13CDIC). Salinity was measured with an YSI model 6820 QS probe and reported using the Practical Salinity Scale. Replicate POM and Chl a water samples (POM: 1L, Chl a: 0.5L) were pre-filtered with a 150-µm sieve and filtered onto a pre-combusted (500 ºC for 2 h) Whatman GF/F and Whatman GF/C filters, respectively , and kept frozen (-20 ºC) until analysis. To measure δ13CDIC values, replicate water samples were injected (6 mL) into a 10 mL exetainer (Labco Limited), containing 0.5 ml of phosphoric acid 85% (v/v), with a sterile syringe filter coupled with a 0.2 m cellulose acetate membrane acrodisc. Samples were kept cooled during field sampling and then stored at 4 ºC until analysis. MPB samples were collected at three stations situated across the salinity gradient (stations 1, 8 and 14). At each station, one PVC pipe (5 cm diameter; 50 cm length) was fixed in the sediment and left in the estuary for two weeks. At the end of the period, algae colonizing at the sediment interface were scraped from each pipe into a separate vial, stored on ice, and returned to the laboratory where we applied the same procedure used for POM samples. At each station, we used a van Veen grab to collect C. fluminea specimens and sediment. Ten clams per station were randomly selected for analysis (140 specimens total). Both C. fluminea specimens and SOM samples were kept frozen (-20 ºC) until analysis of sediment organic carbon (SOC) δ13C values and nitrogen (SN) δ15N values. 4.2.3 Laboratory analyses Filters for POM and MPB analysis were fumigated with concentrated HCl to remove inorganic carbonates, rinsed with deionized water, placed in a sterile Petri dish, and dried at 60 oC for 24 h (Lorrain et al. 2003). Sediment samples were rinsed with 10% HCl (also to remove carbonates), rinsed with deionized water, and dried at 60 ºC for 48h. Both acidification methods are expected to produce only slight changes in sample 15N values (ca. 0.4‰; Lorrain et al. 2003, Carabel et al. 2006). The shell length of each C. fluminea specimen was measured (± 0.01 mm). Then, the foot was excised, dried in an oven at 60 ºC and ground to a fine powder with a mortar and pestle for stable isotope analysis. 106 Stable isotope ratios were measured using a Costec 4010 EA and Thermo Delta Plus XP isotope ratio mass spectrometer (IRMS) (United States Environmental Protection Agency, Mid-Continent Ecology Divison, Duluth, Minnesota). Stable isotope ratios are reported in  notation, X: X = (Rsample/Rstandard - 1) x 103, where X is the C or N stable isotope, R is the ratio of heavy: light stable isotopes, and Pee Dee Belemnite and air are standards for 13C and 15N, respectively. The analytical error, the mean SD of replicate reference material, was ±0.1‰ for δ13C and δ15N. To control for sample processing quality, we did not include those C. fluminea samples with a SD between replicates (i.e. two sub-samples of the same sample) >0.2‰ δ13C or δ15N in subsequent data analyses. DIC δ13C was measured using a GasBench II system interfaced to a Delta V Plus IRMS (Davis Stable Isotope Facility, University of California). The analytical error was ±0.2‰ δ13C. Filters processed for Chl a analysis were extracted in 90% acetone and analyzed on a Spectronic 20 Genesys spectrophotometer. Chl a concentration was calculated following Lorenzen (1967). 4.2.4 Data analyses We used a segmented linear regression to investigate if there was a continuum of C. fluminea δ13C and δ15N values with respect to salinity or clam size. Segmented linear regression analysis performs distinct linear regressions to determine values in the x-axis that are smaller or greater than a given breakpoint. The selection of the breakpoint and function type (among 6 options) is based on maximizing the statistical coefficient of explanation, and performing tests of significance (Oosterbaan 1994). A range of breakpoint values are analyzed. The breakpoint with the highest coefficient of explanation is selected. We conducted the segmented regression analyses using SegReg software (Waterlog 2012). If a breakpoint was identified between either stable isotope ratio (δ13C and δ15N values were analyzed separately) and salinity or shell length, two (or more) groups were defined based on the independent variable (i.e., salinity, shell length). The segmented linear regression analysis did not identify a breakpoint with respect to clam size; however, a positive relationship was found between clam size and the stable isotope values. We then tested the differences between C. fluminea’s stable isotope values collected in areas subjected to different salinity regimes (the lower and upper estuary, as identified by the SegReg analysis) using an analysis of covariance (ANCOVA). The test was performed using salinity as the grouping variable (upper and lower estuary groups) and shell length as a covariate. Statistical analyses were performed using STATISTICA 7 software. 107 To quantify OM source contributions to C. fluminea tissue, we used a dual-stable isotope mixing model. The mixing model estimates a proportional contribution for each food source, assuming all sources sum to 1. However, the model equations are indeterminate when there are more than n + 1 sources relative to stable isotopes (i.e., more than three sources for two stable isotopes). Because there were more than three potential food sources in the estuary, we used a stable isotope mixing model that uses Bayesian inference to solve the indeterminate equations and produce a probability distribution that represents the likelihood a given source contributes to the consumer (Parnell et al. 2010). The model, Stable Isotope Analysis in R (SIAR) allows each of the sources and the trophic enrichment factor (TEF; or trophic fractionation) to be assigned as a normal distribution, rather than a single datum (Parnell et al. 2010). We used the SIAR package (Stable Isotope Analysis in R; Parnell et al. 2010) which is part of the open source statistical language R (R Development Core Team 2007). SIAR will produce a range of feasible solutions to the mixing problem to which are assigned credibility intervals (CI), analogous to the confidence intervals used in frequentist statistics (in this study, 95% CI; Parnell et al. 2010). SIAR also includes a residual error term. For modelling purposes, the estuary was divided into two portions based on evidence from the segmented regression analysis: upper and lower estuary. We estimated the average (± SD) δ13C and δ15N values for the various OM sources differently for each OM source. At each station, the phytoplankton 13C (δ13Cphytoplankton) value was estimated from the DIC δ13C (δ13CDIC) value, assuming an uptake fractionation of -21‰ (i.e., δ13Cphytoplankton= δ13CDIC-21‰; Peterson and Fry 1987 and references there in). However, isotopic fractionation can vary with DIC concentration, phytoplankton growth rate, and nutrient availability (Mook and Tan 1991). To corroborate the DIC-based estimates, we used an independent method to estimate δ13Cphytoplankton values; this alternative method was shown to yield values similar to isolated algae samples (Marty and Planas 2008). A two source, single stable isotope mixing model was used to estimate δ13Cphytoplankton by correcting particulate organic carbon δ13C (δ13CPOC) for algal biomass based on Chl a concentration (C:Chl a= 80; Marty and Planas 2008): δ13CPOM= x . (δ13Cphytoplankton) + (1 - x) . (δ13Ca) where x represents the proportion of algal carbon in the POM pool and δ13Ca = -27.0‰, which is the typical value for C3 plants in the TFW portion of the estuary, for the TFW portion and δ13Ca = -20‰ for marine phytoplankton in the polyhaline estuary (McMahon et al 2013). The proportion of algal carbon in the POC pool was calculated according to the modified relationship from Wienke and Cloern (1987) in Canuel et al. (1995). For this 114 Fig. 4.4 Average δ13C and δ15N values of Corbicula fluminea adjusted for one trophic level fractionation (+0.4‰ δ13C, +3.4‰ for δ15N) and potential organic matter sources for lower estuary (stations 1-4). Error bars represent one standard deviation. The sources considered relevant for Corbicula fluminea (closed square; each represents a station-specific average) in this portion of the estuary were marine phytoplankton (MP; closed diamond), estuarine phytoplankton (EP; open triangle), microphytobenthos (MPB; open circle), sediment organic matter (SOM; closed triangle) and particulate organic matter (POM; closed circle). In the stations closest to the river mouth (stations 1 and 2), salinity values can be higher than 20 during summer, depending on the tidal cycle. Salinities above 8 are lethal for adult clams (Mackie and Claudi 2010), although they can tolerate higher salinities for short periods (Ilarri and Sousa 2012b). Thus, it is still possible to find live clams in the stations closest to the river (1 and 2) during high tide. However, repeated fluctuations in salinity levels likely stress the clams. One common response to environmental stress in bivalves is valve closure (Shumway 1977, Kramer et al. 1989), and thus C. fluminea would not be able to consume marine phytoplankton during high tide, when it is available. For that reason, we excluded marine phytoplankton from the model. Based on the SIAR mixing model (95% CI), phytoplankton contributed between 29.1% and 54.7%, and POM contributed between 19.1% and 56.1% to C. fluminea in the lower estuary (Fig. 4.6). 115 Fig. 4.5 Average δ13C values of Corbicula fluminea (closed triangle) adjusted for one trophic level fractionation (+0.4‰ δ13C) and potential particulate organic matter sources (δ13Cphytoplankton [estimated from δ13CDIC]- closed circle; δ13CPOCopen circle) along the estuarine salinity mixing. Error bars represent one standard deviation. Table 4.2 Average δ13C (‰) and δ15N (‰) (± SD) values used in the Corbicula fluminea SIAR mixing model for the organic matter sources used in either the upper estuary (UE) or lower estuary (LE) model. These values represent average bottom values for phytoplankton and particulate organic matter (POM), and also the average values for microphytobenthos (MPB) and sediment organic matter (SOM) pooled from each group of stations. Source δ 13C δ 15N UE phytoplankton -38.5 (±0.7) 5.0 (±1.0) LE phytoplankton -32.0 (±2.4) 6.0 (±1.0) UE POM -27.8 (±1.3) 4.9 (±0.9) LE POM -27.4 (±1.3) 5.3 (±0.7) UE MPB -24.0 (±2.7) 7.3 (±2.2) LE MPB -18.7 (±5.2) 7.8 (±1.1) UE SOM -28.0 (±0.8) 0.2 (±2.5) LE SOM -26.1 (±1.5) 1.7(±0.3) 116 Fig. 4.6 Proportion of each food source to Corbicula fluminea’s biomass collected in the lower estuary (stations 1-4) based on the stable isotope mixing model. The food sources included in the model were estuarine phytoplankton (EP), particulate organic matter (POM), sediment organic matter (SOM) and microphytobenthos (MPB). Boxes indicate 50%, 75% and 95% Bayesian credibility interval Fig. 4.7 Average δ13C and δ15N values of Corbicula fluminea adjusted for one trophic level fractionation (+0.4‰ δ13C, +3.4‰ for δ15N) and potential organic matter sources for the upper estuary (stations 5-14). Error bars represent one standard deviation. The sources considered relevant for Corbicula fluminea (closed square; each represents a station-specific average) in this portion of the estuary were freshwater phytoplankton (FP; open triangle), microphytobenthos (MPB; open circle), sediment organic matter (SOM; closed triangle) and particulate organic matter (POM; closed circle). Corbicula fluminea δ13C values in the upper estuary were similar to 13CPOC values, as well (Fig. 4.4 and 4.7, Table 4.2). Based on the results from the SIAR mixing model (95% CI), POM contributed between 40.9% and 64.4%, SOM contributed between 14.1 and 23.7%, and MPB contributed between 8.5 and 21.4% to C. fluminea biomass (Fig. 117 4.8). The contribution of phytoplankton to C. fluminea biomass in the TFW (10.6-16.4%) decreased ca. 20-40% in relation to the lower estuary (Fig. 4.6 and 4.8). Fig. 4.8 Proportion of each food source to Corbicula fluminea’s biomass collected in the upper estuary (stations 5–14) based on the stable isotope mixing model. The food sources included in the model were freshwater phytoplankton (FP), particulate organic matter (POM), sediment organic matter (SOM) and microphytobenthos (MPB). Boxes indicate 50%, 75% and 95% Bayesian credibility intervals. 4.4 Discussion A pronounced shift in the quality of OM available for C. fluminea in the Minho River estuary occurred along the estuarine mixing zone. In concert, stable isotope analysis, POM C/N ratios, and phytoplankton contribution estimates based on C: Chl a indicated that POM in freshwater stations was largely comprised of terrestrial-derived OM, a refractory food source, and was increasingly comprised of phytoplankton, a highly palatable food source, towards the polyhaline estuary. A similar shift in isotopic composition along the estuarine mixing zone was observed in C. fluminea, indicating a shift in the food sources used in response to the change in food quality. Accordingly, the SIAR mixing model attributed most C. fluminea biomass to terrestrial-derived OM in the TFW, with an increasing contribution from estuarine phytoplankton seaward. Notably, the stable isotope data indicates C. fluminea has the ability to consume and assimilate terrestrial-derived OM. Therefore, C. fluminea appears to link the estuary with the terrestrial ecosystem in the Minho River estuary benthic food web. The shifts in the 118 sources of OM consumed by C. fluminea along the estuarine mixing gradient and the possible effects of its feeding modes to the estuarine food web are discussed below. 4.4.1. Sources of organic matter supporting C. fluminea production in the upper estuary Terrestrial-derived OM was estimated as the major contributor to C. fluminea biomass (50% to 64%) in the upper portion of the Minho River estuary. Terrestrial-derived OM is a prominent C source to rivers; terrestrial inputs are often equal to or larger than the autochthonous primary production (Caraco and Cole 2004). Yet, numerous studies have demonstrated that estuarine secondary production is primarily supported by autochthonous OM, especially phytoplankton (Deegan and Garritt 1997, Chanton and Lewis 2002). Nevertheless, there is growing evidence that terrestrial inputs can subsidize estuarine food webs, supporting up to 80% of invertebrate and fish biomass (Kasai and Nakata 2005, Hoffman et al. 2008). In the Minho River TFW portion, the POM pool was comprised mostly of terrestrialderived OM (C/N > 10) and the phytoplankton contribution to this pool was ca. 20% (surface). C. fluminea stable isotope values were similar to POM, which suggests that growth of this bivalve in the upper portion of the estuary was supported by terrestrialderived OM, which was directly consumed by filter feeding. If phytoplankton were contributing greatly to C. fluminea biomass, we would expect to have observed both C. fluminea 13C and δ13Cphytoplankton values follow a similar trend along the estuarine mixing zone (Fig. 4.5) and δ13C values of C. fluminea would have been lower in the TFW portion, between -35‰ and -38‰ (Fig. 4.7), which was not the case. Moreover, the available environmental data indicate that phytoplankton standing stock was low and that the system was net-heterotrophic (i.e. low δ13CDIC values). The Chl a concentration was lower than 1 g L-1 (0.74 g L-1, this study; also, 0.70 g L-1 in 2009-2010, Brito et al. 2012) and based on C:Chl a, phytoplankton fraction in the POM was ≤ 20%. The latter estimate should be treated cautiously. Chl a concentration is often used to estimate phytoplankton biomass; the C:Chl a ratio typically ranges from 30 to 50 (e.g. Legendre et al. 1999). This ratio, however, varies within and among algal species (e.g. Chan 1980, Kruskopf and Flynn 2005, Putland and Iverson 2007), ranging from 5 to 345 mg C mg Chl a -1 (Putland and Iverson 2007). In a recent study, the effect of variation in the C:Chl a ratio on the proportion of algal C in POC was examined for different levels of Chl a concentrations (Marty and Planas 2008). The authors concluded that there is a positive relationship between the phytoplankton percentage in POC and the C:Chl a ratio, yet under low Chl a concentrations (0.6 µg L-1), the contribution of phytoplankton would be low regardless of the C:Chl a ratio (Marty and Planas 2008). Therefore, even if the C:Chl a ratio used was 119 as high as 100, the contribution of phytoplankton to POC pool would remain close to 10%. Thus, C. fluminea appears to have been assimilating a relatively low quality, high flux food source (terrestrial inputs) in response to low phytoplankton availability (though rates of primary production were not measured). In laboratory (18 clams per 18 L funnel) and field growth experiments with C. fluminea, clams were food limited at Chl a concentrations between 16 µg L-1 and 19 µg L-1 (Foe and Knight 1985), which are 22-26 times higher than the average Chl a concentration estimated, and 4-5 times higher than the peak concentration measured (3.6 µg L-1, station 7). While it is difficult to compare clam densities and filtration rates measured in the laboratory with those measured in the field, these results suggest that C. fluminea may have been using alternative OM sources where phytoplankton availability was limited to maintain its high abundance in the Minho River estuary. Although organic detritus has been identified in C. fluminea gut contents (Hill 1985 cited in Foe and Knight 1985), to our knowledge, this is the first study that quantifies the importance of terrestrial-derived OM as a food source to this species. Plant residues contain structural polysaccharides, such as cellulose and hemicellulose that are thought to be digested only by microorganisms (Sakamoto and Toyohara 2009). However, cellulase and hemicellulase activities have been found in the digestive organs of Corbicula japonica (Sakamoto et al. 2009, Niiyama and Toyohara 2011). We hypothesize that C. fluminea is using a similar biochemical process to digest a less palatable OM source (terrestrialderived OM), and simultaneously catalyze vascular plant-derived refractory substances into assimilable OM. Though terrestrial-derived OM was the main food source for C. fluminea in the upper portion of the estuary, sediment OM and MPB also contributed to their biomass. Filter feeding alone cannot balance C. fluminea’s energy budgets (Aldridge and McMahon, 1978), so pedal feeding is an important source of nutrition, especially in food-limited environments (Boltovskoy et al. 1995). A critical assumption in the SIAR mixing model is that the OM source values measured were similar over the prior 2-3 months because the data collected during this study were obtained during a single sampling event. Although isotopic turnover rate studies of C. fluminea have not been conducted, it is likely that the isotopic turnover period of the C. fluminea population is about three months. For individuals, the isotopic turnover period depends on both somatic growth and metabolic turnover rates; in organisms that grow rapidly, somatic growth rates essentially determine the isotopic turnover period (del Rio et al. 2009). C. fluminea have rapid growth throughout their life and, owing to their rapid sexual maturity, have a population-level turnover time (in biomass) of 73-91 days (McMahon 2002). Moreover, previous studies have demonstrated 120 that stable isotope values of other bivalves Dreissena polymorpha (zebra mussel) and Corbula amurensis (Asian clam; formerly known as Potamocorbula) are spatially and temporally well aligned with environmental processes (e.g. watershed inputs, chemistry of river water; Fry 2002, Fry and Allen 2003). The same is likely true for C. fluminea owing to its rapid growth rate and similarlyshort life span (McMahon 2002, Thompson and Parchaso 2010). There are several features that we considered relevant when interpreting this data over the turnover period of C. fluminea. First, the sampling occurred at the end of the summer, during low flow conditions (below 25% percentile of the monthly averages along a hydrologic year; Confederación Hidrográfica del Miño-Sil, personal communication) that persisted over the summer months (July-September), a timeframe similar to the turnover period of C. fluminea. Second, the stable isotope values of terrestrial inputs are relatively constant (Cloern et al. 2002). Third, the low inflow variation associated with low flow conditions results in δ13CDIC values that are more stable (Striegl et al. 2001, Bade et al. 2004, Lehmann et al. 2004) and, therefore, δ13Cphytoplankton values are likely similar throughout the low flow period. Thus, it is likely that the biogeochemical conditions in September were similar to the summer low flow biogeochemical conditions that preceded sampling. 4.4.2 Sources of organic matter sources supporting C. fluminea production in the lower estuary In the brackish estuary, phytoplankton and POM were estimated to compose the majority of C. fluminea’s biomass, contributing between 29-55% and 19-56%, respectively. Although Chl a was higher in the brackish estuary than in the TFW portion, it was still within the range identified as limiting for C. fluminea growth (Foe and Knight 1985). Our estimates of the phytoplankton fraction in estuarine POM were around 20%. Therefore, positive selection for phytoplankton must have occurred prior to ingestion or during digestion, as the phytoplankton contribution to C. fluminea biomass was proportionally higher than the phytoplankton contribution to the bulk POM pool. Although C. fluminea is a generalist consumer, these clams prefer small-sized living and suspended particulate OM (Atkinson et al. 2011). The mechanisms involved in this selection, however, are not understood. Corbicula fluminea also filtered and assimilated POM, which was a mixture of terrestrial-derived OM and marine POM, transported by surface freshwater outflow and bottom marine inflow, respectively. 121 4.4.3 Possible effects of C. fluminea in the estuarine food web Corbicula fluminea filtered from the water column the majority (70-80%) of its food sources. The quality of these filtered food sources used by C. fluminea changed along the estuarine mixing zone, revealing that this invasive bivalve has the ability to use alternative and less labile OM sources (terrestrial and sediment OM) when phytoplankton is less abundant. If C. fluminea can utilize the refractory components of terrestrial detritus, it is possible that they may also act as initiators to break down plant cell wall structural polysaccharides. However, further studies are needed to clarify the presence of an enzymatic mechanism that enables C. fluminea to perform this function. Corbicula fluminea can influence seston concentration (Leff et al. 1990) and particle size (Atkinson et al. 2011), as well as decrease Chl a concentrations by 20-70% (Cohen et al. 1984). Thus, the high filtration rates of C. fluminea (Strayer et al. 1999) and preference for small-sized living and suspended POM (Atkinson et al. 2011) may condition the availability of edible particles to other native suspension feeders. By removing suspended material from the water column at high rates, this species increases OM deposition into the sediments (Hakenkamp and Palmer 1999). This may increase sediment OM available to C. fluminea, as well as to other benthic organisms. In a recent study conducted in the Minho River estuary, a positive relationship was found between both the density and biomass of benthic invertebrates (chiromidae larvae, amphipods, and gastropods) and increasing densities of C. fluminea (Ilarri et al. 2012a). The authors concluded that the increase in habitat structure provided by shells and the increase in waste products (faeces and pseudofaeces) may play essential roles in this positive relationship (Ilarri et al. 2012a). Additionally, the stoichiometry of C. fluminea biodeposits from Ichawaynochaway Creek (USA) revealed that they represent a higher quality food source when compared to the biodeposits of a native bivalves (Elliptio crassidens) or seston (10-45 µm) (Atkinson et al. 2010) and thereby constitute a higher quality food source for benthic organisms. Although C. fluminea filtered from the water column the majority of the food sources identified in this study, sediment OM and benthic periphyton did contribute between ca. 10-20% to their biomass, indicating C. fluminea was deposit feeding. This adaptation to access benthic carbon when water column food resources were apparently limiting is likely an additional factor explaining C. fluminea’s success in the Minho River estuary. Furthermore, this species has a broader diet breadth than is known for native unionids species (Vaughn and Hakenkamp 2001, Atkinson et al. 2010), which can be an advantage when competing with co-occurring native species for resources and habitat. 122 In conclusion, this study reveals that C. fluminea has the ability to adapt to environments with low food quality, not only because it deposit feeds, but also because it can filter and assimilate terrestrial-derived OM. This can be a competitive adaptation in systems with perennial low food quality such as the Minho River estuary. Moreover, its ability to directly (by consuming terrestrial inputs) and indirectly (by conditioning terrestrial OM for consumption by other benthic consumers in the form of biodeposits) couple the terrestrial ecosystem with estuarine benthic and pelagic environments constitutes a new identified process by which C. fluminea may alter food web flows in aquatic ecosystems. As an invader, the implication of this adaptive and flexible feeding behavior is that it likely facilitates its widespread success at establishing in new aquatic environments. Acknowledgements We thank Eduardo Martins, Joana Campos and the staff at Aquamuseu do Rio Minho for their help while conducting the field work; Anne M. Cotter for preparing POM samples for stable isotope analysis, for running the samples and for all the technical support with the IRMS; Rute Pinto for providing the map of the study area; and three anonymous reviewers for their helpful comments. E. Dias was supported by a Ph.D. grant from Iceland, Liechtenstein and Norway through the EEA Financial Mechanism and the Norwegian Financial Mechanism. P. Morais was supported by a post-doc scholarship financed by Fundação para a Ciência e Tecnologia, Portugal (SFRH/BPD/40832/2007). This work was partially supported by the project MIGRANET SOE 2/P2E288. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA. 4.5 References Aldridge DW, McMahon RF (1978) Growth, fecundity, and bioenergetics in a natural population of the freshwater clam, Corbicula fluminea Philippi, from north central Texas. J Mollus Stud 44: 49-70. Alves AM (1996) Causas e processos da dinâmica sedimentar na evolução actual do litoral do Alto Minho. PhD Dissertation, Universidade do Minho. Antunes C, Araújo MJ, Braga C, Roleira A, Carvalho R, Mota M (2011) Valorização dos recursos naturais da bacia hidrográfica do rio Minho. Final report from the project Natura Miño-Minho, Centro interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto. Araujo R, Moreno D, Ramos MA (1993) The Asiatic clam Corbicula fluminea (Müller, 1774) (Bivalvia: Corbiculidae) in Europe. Amer Malac Bull 10: 39-49. Atkinson CL, First MR, Covich AP, Opsahl SP, Golladay SW (2011) Suspended material availability and filtration-biodeposition processes performed by a native and invasive bivalve species in streams. Hydrobiologia 667: 191-204. 123 Atkinson CL, Opsahl SP, Covich AP, Golladay SW, Conner LM (2010) Stable isotopic signatures, tissue stoichiometry, and nutrient cycling (C and N) of a native and invasive freshwater bivalve. J N Am Benthol Soc 29: 496-505. Bade DL, Pace ML, Cole JJ, Carpenter SR (2006) Can algal photosynthetic inorganic carbon isotope fractionation be predicted in lakes using existing models? Aquat Sci 68: 142-153. BirdLife International (2012) Important Bird Areas factsheet: Minho and Coura estuaries. Available at http://www.birdlife.org. Accessed 5 November 2012. Boltovskoy D, Izaguirre I, Correa N (1995) Feeding selectivity of Corbicula fluminea (Bivalvia) on natural phytoplankton. Hydrobiologia 312: 171-182. Brito AC, Brotas V, Caetano M, Coutinho TP, Bordalo AA, Icely J, Neto JM, Serôdio J, Moita T (2012) Defining phytoplankton class boundaries in Portuguese transitional waters: An evaluation of the ecological quality status according to the Water Framework Directive. Ecol Ind 19: 5-14. Caffrey JM, Evers S, Millane M, Moran H (2011) Current status of Ireland’s newest invasive speciesthe Asian clam Corbicula fluminea (Müller, 1774). Aquatic Invasions 6: 291-299. Canuel EA, Cloern JE, Ringelberg DB, Guckert JB (1995) Molecular and isotopic tracers used to examine sources of organic matter and its incorporation into the food webs of San Francisco Bay. Limnol Oceanogr 40: 67-81. Carabel S, Godínez-Domínguez E, Verísimo P, Fernández L, Freire J (2006) An assessment of sample processing methods for stable isotope analyses of marine food webs. J Exp Mar Biol Ecol 336: 254-261. Caraco NF, Cole, JJ (2004) When terrestrial organic matter is sent down the river: Importance of allochthonous C inputs to the metabolism in lakes and rivers. In Polis A, Power ME, Huxell G (eds.). Food Webs at the Landscape Level, University of Chicago Press, Chicago, Illinois, pp. 301-316. Chan AT (1980) Comparative physiological study of marine diatoms and dinoflagellates in relation to irradiance and cell size. II. Relationship between photosynthesis, growth, and carbon/chlorophyll a ratio. J Phycol 16:428-32. Chanton J, Lewis FG (2002) Examination of coupling between primary and secondary production in a river-dominated estuary: Apalachicola Bay, Florida, USA. Limnol Oceanogr 47: 683-697. Cherry DS, Scheller JL, Cooper NL, Bidwell JR (2005) Potential effects of Asian clam (Corbicula fluminea) die-offs on native freshwater mussels (Unionidae) I: watercolumn ammonia levels and ammonia toxicity. J N Am Benthol Soc 24, 369-380.