Abundance and δ13C values of fatty acids in lacustrine surface sediments : Relationships with in-lake methane concentrations

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 4.0 https://creativecommons.org/licenses/by-nc-nd/4.0/ Abundance and δ13C values of fatty acids in lacustrine surface sediments : Relationships with in-lake methane concentrations © 2018 Elsevier Ltd. Accepted version (Final draft) Stötter, Tabea; Bastviken, David; Bodelier, Paul L.E.; van Hardenbroek, Maarten; Rinta, Päivi; Schilder, Johannes; Schubert, Carsten J.; Heiri, Oliver Stötter, T., Bastviken, D., Bodelier, P. L., van Hardenbroek, M., Rinta, P., Schilder, J., Schubert, C. J., & Heiri, O. (2018). Abundance and δ13C values of fatty acids in lacustrine surface sediments : Relationships with in-lake methane concentrations. Quaternary Science Reviews, 191, 337-347. https://doi.org/10.1016/j.quascirev.2018.04.029 2018 1 1 Abundance and δ13C values of fatty acids in lacustrine surface sediments:2 Relationships with in-lake methane concentrations3 4 Tabea Stöttera,*, David Bastvikenb, Paul L. E. Bodelierc, Maarten van Hardenbroeka,d, Päivi5 Rintaa, Jos Schildera, e, Carsten J. Schubertf, Oliver Heiria 6 7 aInstitute of Plant Sciences and Oeschger Centre for Climate Change Research, University of8 Bern, Switzerland9 b Department of Thematic Studies – Water and Environmental Studies, Linköping University,10 Linköping, Sweden11 cDepartment of Microbial Ecology, Netherlands Institute of Ecology (NIOO-KNAW),12 Wageningen, The Netherlands13 dSchool of Geography, Politics and Sociology, Newcastle University, Newcastle, UK14 e Department of Biological and Environmental Science, University of Jyväskylä, Jyväskylä,15 Finland16 f EAWAG, Surface Waters – Research and Management, Kastanienbaum, Switzerland17 18 19 *Corresponding author: [email protected] 21 Keywords: Methane, Fatty acids, Methane oxidizing bacteria, Stable carbon isotopes, aquatic22 invertebrates, Lakes, Sediment23 24 25 26 2 27 Highlights28 29 * Fatty acid abundances in lake sediments were compared with methane concentrations30 * Three 13C-depleted FA groups were correlated with CH4 concentration in the lakes31 * The results support earlier interpretations that MOB increase with CH4 concentrations32 * Further studies with lipids specific to MOB are needed to corroborate our results33 34 35 3 Abstract36 37 Proxy-indicators in lake sediments provide the only approach by which the dynamics of in-38 lake methane cycling can be examined on multi-decadal to centennial time scales. This39 information is necessary to constrain how lacustrine methane production, oxidation and40 emissions are expected to respond to global change drivers. Several of the available proxies41 for reconstructing methane cycle changes of lakes rely on interpreting past changes in the42 abundance or relevance of methane oxidizing bacteria (MOB), either directly (e.g. via43 analysis of bacterial lipids) or indirectly (e.g. via reconstructions of the past relevance of44 MOB in invertebrate diet). However, only limited information is available about the extent to45 which, at the ecosystem scale, variations in abundance and availability of MOB reflect past46 changes in in-lake methane concentrations. We present a study examining the abundances of47 fatty acids (FAs), particularly of 13C-depleted FAs known to be produced by MOB, relative to48 methane concentrations in 29 small European lakes. 39 surface sediment samples were49 obtained from these lakes and FA abundances were compared with methane concentrations50 measured at the lake surface, 10 cm above the sediments and 10 cm within the sediments.51 Three of the FAs in the surface sediment samples, C16:1ω7c, C16:1ω5c/t, and C18:1ω7c were52 characterized by lower δ13C values than the remaining FAs. We show that abundances of53 these FAs, relative to other short-chain FAs produced in lake ecosystems, are related with54 sedimentary MOB concentrations assessed by quantitative polymerase chain reaction (qPCR).55 We observed positive relationships between methane concentrations and relative abundances56 of C16:1ω7c, C16:1ω5c/t, and C18:1ω7c and the sum of these FAs. For the full dataset these57 relationships were relatively weak (Spearman's rank correlation (rs) of 0.34 to 0.43) and not58 significant if corrected for multiple testing. However, noticeably stronger and statistically59 significant relationships were observed when sediments from near-shore and deep-water oxic60 environments (rs = 0.57 to 0.62) and those from anoxic deep-water environment (rs = 0.55 to61 4 0.65) were examined separately. Our results confirm that robust relationships exist between62 in-lake CH4 concentrations and 13C-depleted groups of FAs in the examined sediments,63 agreeing with earlier suggestions that the availability of MOB-derived, 13C-depleted organic64 matter for aquatic invertebrates increases with increasing methane concentrations. However,65 we also show that these relationships are complex, with different relationships observed for66 oxic and anoxic sediments and highest values measured in sediments deposited in oxic67 environments overlain with relatively methane-rich water. Furthermore, allthough all three68 13C-depleted FA groups identified in our survey are known to be produced by MOB, they also69 receive contributions by other organism groups, and this will have influenced their70 distribution in our dataset.71 72 5 1. Introduction73 74 Methane (CH4) is a major greenhouse gas and lakes are an important natural source of CH4 to75 the atmosphere (Bastviken et al. 2011; Bridgham et al. 2013). CH4 concentrations in lakes are76 determined by the rate of biogenic CH4 production, a process in which methanogens produce77 CH4 using the hydrogenotrophic, acetoclastic, or methylotrophic pathway (Borrel et al. 2011),78 and by the rate of CH4 diffusion and ebullition into the water column and further into the79 atmosphere. Another process that can remove CH4 from lakes is methanotrophy, a microbially80 mediated process that oxidizes CH4 with O2 or other electron acceptors, like nitrate (Hanson81 and Hanson 1996; Blees et al. 2014, Oswald et al. 2016). Most of the CH4 consumption in82 freshwaters seems to be performed by CH4-oxidizing bacteria (MOB) under oxic conditions,83 which have been estimated to oxidize 30 to 99 % of the CH4 produced in lakes (Bastviken et84 al. 2008). Biogenic CH4 has a distinctly 13C-depleted stable carbon isotopic composition, with85 δ13C values typically between -80 and -50 ‰ (vs. VPDB) (Whiticar et al. 1986; Jedrysek 2005).86 MOB, which use CH4 as their energy and carbon source, incorporate CH4-derived carbon into87 their biomass. Therefore they are characterized by very negative δ13C values as well, which can88 be even lower than those of the original CH4 due to isotopic fractionation during CH4 uptake89 (Summons et al 1994). Since CH4 is significantly more depleted in 13C than other carbon90 sources, δ13C is a good tracer for CH4-related processes in lakes. For example, modern and91 palaeoecological food web studies have used δ13C analysis to assess the importance of MOB as92 a food source for aquatic invertebrates in lakes (e.g. Sanseverino et al. 2012; Schilder et al.93 2015; Grey 2016).94 Systematic field surveys constraining CH4 production, abundance and emissions in lakes have95 only been developed very recently (e.g. Rasilo et al. 2015; Yang et al. 2015; Rinta et al. 2017).96 Since long instrumental time series are lacking it is challenging to predict how and at which97 time scales lacustrine CH4 production and emission will respond to future environmental98 6 pressures such as global warming and widespread eutrophication and reoligotrophication of99 inland waters. Proxy-based reconstructions of past changes in lacustrine carbon cycling have100 been explored as alternative approaches for constraining how the carbon cycle of lakes, and101 particularly CH4 production, oxidation, and uptake in the food web, respond to environmental102 change. Such studies are particularly relevant for assessing how these processes react on multi-103 decadal to centennial timescales to global change drivers such as increasing air and water104 temperatures and anthropogenic nutrient release, since these timescales are not covered by105 instrumental measurements.106 Proxy-based approaches used to constrain past changes in CH4 availability, production and107 oxidation in lakes have been based either on geochemical measurements on specific organic108 microfossil groups (e.g. Wooller et al. 2012; Belle et al. 2014; Rinta et al., 2016; Schilder et al.109 2017) or lipid groups (e.g. Bechtel and Schubert, 2009; Hollander and Smith, 2001; Naeher et110 al., 2014; Davies et al. 2016) that are expected to be related either in their carbon isotopic111 composition or abundance (or both) with MOB or CH4-producing microorganisms in lakes.112 Furthermore, ancient DNA (aDNA) analyses of lake sediments have recently been used to113 constrain past changes in MOB (Belle et al. 2014). If absolute or relative abundances of these114 microorganisms are in turn systematically related to CH4 concentrations in lakes, these and115 similar approaches may even allow quantitative statements about past changes in CH4 116 abundance in lakes based on lacustrine proxy records (e.g. van Hardenbroek et al. 2013;117 Schilder et al. 2015a; Elvert et al. 2016). Recent studies have demonstrated that δ13C analyses118 of chitinous remains of aquatic invertebrates may have considerable potential in this respect.119 Several aquatic invertebrate groups can feed on MOB or other microorganisms such as ciliates120 feeding on them (e.g. Kankaala et al. 2006; Deines et al., 2007; Deines and Fink, 2011; Jones121 and Grey, 2011), which leads to strongly 13C-depleted isotope signatures in their chitinous122 remains (e.g. resting egg sheaths, exoskeleton parts) preserved in lake sediments (Wooller et123 al., 2012; Belle et al. 2014; Schilder et al. 2015a). Reconstructions of changes in δ13C values of124 7 these invertebrate fossils have been used to reconstruct past changes in the relevance of CH4-125 derived carbon for different parts of lacustrine food webs. Furthermore, several studies have126 revealed quantitative relationships between δ13C values of these remains and measurements of127 CH4 abundances, suggesting that estimates of past variations in in-lake CH4 abundance may be128 possible based on δ13C analyses of some of these invertebrate groups. However, the129 mechanisms that lead from high CH4 availability in the examined lakes to a higher proportion130 of CH4-derived C in invertebrate biomass (and their microfossils) are still poorly constrained.131 One potential explanation is that CH4-rich ecosystems are characterized by higher abundances132 of MOB and therefore also a higher availability of CH4-derived carbon for filter-feeding and133 deposit feeding aquatic invertebrates.134 Here we present a survey of fatty acid (FA) concentrations and δ13C values in the surface135 sediments of 29 lakes from Finland, the Netherlands, Sweden and Switzerland. Concentrations136 of 13C-depleted FAs are compared with CH4 concentration estimates in the open water column,137 above the sediments and in deeper sediment layers (10 cm below the sediment surface). We138 focus on 13C-depleted FA groups that are produced by MOB but also receive contributions from139 other organism groups. If there is a higher contribution of MOB to particulated organic matter140 in the water column and sedimentary organic matter in surface sediments in CH4-rich systems,141 as speculated in earlier studies (e.g. van Hardenbroek et al. 2013, Schilder et al. 2015), we142 expect to see positive relationships between relative abundances of these FAs and CH4 143 concentrations in our dataset. Since the 13C-depleted FA groups in our dataset are not strictly144 limited to MOB we also assess the abundance of MOB in the examined sediments using145 quantitative polymerase chain reaction (qPCR) to support our interpretations.146 147 148 2. Study sites and study setup149 150 8 Surface sediments (0 to 2 cm) were collected from the deepest part of 29 lakes across Europe,151 including 2 Dutch (NL), 6 Finnish (FI), 11 Swedish (SE) and 10 Swiss (CH) lakes (Fig. 1). The152 lakes were sampled in a multi-year campaign and sites were selected to cover a range of small153 lake types on different bedrocks and with variable water chemistry conditions (e.g. in respect154 to transparency, pH, conductivity, deepwater oxygen concentrations) in Northern Europe155 (Finland, Sweden), Western Europe (The Netherlands) and Central Europe (Switzerland). The156 aim of the survey was to assess whether quantitative relationships existed between invertebrate157 δ13C values and variables relevant for determining the overall CH4 production, abundance and158 emissions of these lakes (e.g. Schilder et al. 2015; Rinta et al. 2015). Therefore, CH4 159 concentrations where measured in the open water column, just above the coring site, and deeper160 within the sediments (i.e. at 10 cm depth) rather than within the sediment samples that were161 analysed for invertebrate δ13C values and that are available for the FA analyses presented in162 this study. Most of the lakes were thermally stratified and developed anoxic conditions during163 the summer months. To expand the number of sediment samples accumulated under oxic164 conditions, near-shore surface sediments from the littoral zone were also analysed for 11165 Swedish lakes (lake abbreviations in small letters). All of these samples were located above the166 oxycline. The lakes are all relatively small (average surface area 0.32 km2), shallower than167 32 m, and characterized by variable nutrient concentrations and mixing conditions (Suppl.168 Table 1). The sampling of the lakes took place during late summer stratification (August-169 September) in 2010 in Sweden and 2011 in Finland, the Netherlands, and Switzerland. More170 details on the study lakes and the environmental variables measured during fieldwork are171 available in Rinta et al. (2015).172 173 3. Methods174 175 3.1 Water chemistry176 15 axis 1 scores for most of the Fennoscandian samples. C20 and C22 also followed a similar333 distribution in our dataset as the longer chain FAs. C22:1 and C18 were characterized by high334 axis 1 and 2 scores, indicating that these compounds had a different distribution in the lake335 sediments than the rest of the FAs. FAs with negative axis 1 scores, e.g. C15, C16, or C18:2 form336 another, more heterogeneous group. The observation that most of the Swiss lake sediment337 samples were also characterized by negative axis 1 scores indicates higher relative abundances338 of these compounds in the Swiss sediment samples. The littoral sample of Stora Vänstern (stv)339 was different in its FA composition from all other samples (Fig. 2). However, this sample was340 already identified as potentially contaminated by older sediments and sediment redeposition in341 the field and characterized by sandy material. We therefore excluded this sample from further342 analyses. The deep-water sample of Glimmingen (GLI), which is plotted in a similar area of the343 PCA biplot as stv, contained also sandy material, although to a lesser extent. Although GLI344 showed a different FA composition than the other lakes, the sample was not apparent as an345 outlier in the relationships between FAs and CH4 concentrations and therefore retained in346 further analyses. Since the aim of the study was to assess whether with higher CH4 347 concentrations 13C-depleted FA groups become more abundant relative to other FAs typically348 produced by aquatic organisms (see Sections 1 and 2) we eliminated longer chain FAs (C24,349 C26, C28) originating from terrestrial environments from further analyses.350 Most of the individual FAs showed a similar range of δ13C values (Fig. 3). However,351 C16:1ω7c, C16:1ω5c/t, and C18:1ω7c were more depleted in 13C values (Fig. 3), with median δ13C352 values of -45.5 ‰, -59.6 ‰, and -41.5 ‰, respectively. The rest of the analysed FAs were less353 13C-depleted with median δ13C values between -36.6 and -30.3 ‰. All of these 13C-depleted354 FAs are known to be produced by MOB, although not exclusively (see Section 5). Comparison355 of FA abundances with in-lake CH4 concentrations therefore focused on these 13C-depleted356 FAs. C16:1ω8c and C18:1ω8c, known to be produced by MOB type I and II only (Bodelier et al.357 16 2009a), were not found in the examined lake sediments or the peak areas of these specific FAs358 were too low to be quantified and analysed reliably.359 The qPCR analyses successfully quantified the number of DNA copies of total MOB in360 34 of the examined sediment samples, MOB type Ia in 34, MOB type Ib in 30 and MOB type361 II in 25 (Supplementary Table S2). However, nested PCR revealed that MOB were present in362 all but one of the 30 sediment samples. Comparison of the relative abundances of C16:1ω7c,363 C16:1ω5c/t, and C18:1ω7c with DNA copies of MOB in the sediments (Table 1) confirmed that364 C16:1ω7c was significantly correlated with the abundance of MOB Ib and C18:1ω7c with MOB Ia,365 MOB Ib, MOB II and the sum of MOB DNA copies. The sum of the 13C-depleted FAs366 correlated with MOB type Ia, MOB type Ib and the total number of DNA copies of MOB.367 C16:1ω5c/t was not correlated with the overall abundance of any MOB type.368 Over the entire dataset, relative abundances of C16:1ω7c, C18:1ω7c, and the sum of 13C-369 depleted FAs showed weak positive correlations (rs = 0.34-0.40) with surface water CH4 370 concentrations (Table 2). C16:1ω7c and the sum of 13C-depleted FAs were furthermore correlated371 with CH4 concentrations in the sediments (rs = 0.36-0.43). However, if the results were372 corrected for multiple testing none of these relationships remained significant.373 If only oxic sediments were examined, rs values were distinctly higher (0.57-0.62). Both374 surface water and sedimentary CH4 concentrations were correlated with C16:1ω7c. Furthermore,375 the sum of the 13C-depleted FAs was correlated with CH4 concentrations in the surface water,376 10 cm above and 10 cm below the sediment surface. C16:1ω5c/t was also correlated with CH4 377 concentrations in the sediments, but this relationship was no longer significant after correction378 for multiple testing (Table 2).379 For anoxic sediment samples correlations with the abundances of13C-depleted FAs were380 also stronger (rs = 0.46-0.65) than observed for the entire dataset. For this group of samples,381 C18:1ω7c was correlated with surface water CH4 concentrations (rs = 0.55) and C16:1ω7c and the382 sum of 13C-depleted FAs were correlated with CH4 concentrations in the lake water 10 cm383 17 above the sediments, although the latter two relationships were no longer significant after384 correcting for multiple testing. The strongest relationships for anoxic sediments were apparent385 between CH4 concentrations 10 cm below the sediment surface and C16:1ω7c and the sum of 13C-386 depleted FAs (rs = 0.61-0.65).387 388 5. Discussion389 390 5.1 FA composition of sediments391 392 The high content of short-chain FAs, with a classical even over odd predominance, and maxima393 at n-C16 in the FA composition of the sediment samples, indicates a predominantly394 autochthonous organic matter production (Stefanova and Disnar 2000; Woszczyk et al. 2011).395 The longer-chain FAs, with high axis 1 values in the PCA analysis (Fig. 2), are reported to be396 derived from terrestrial sources, for example C24-C30 from waxy coatings of land plants (Meyers397 2003). C22:1, which together with C18 shows a different distribution than the rest of the FAs, is398 known to be produced by zooplankton, copepods and higher plants (Pearson et al. 2007). C18 399 may originate from many sources, mainly freshwater algae (Meyers 2003). The more400 heterogeneous group of FAs with negative axis 1 values in the PCA are known to have multiple,401 mainly autochthonous sources (Pearson et al. 2007; Woszczyk et al. 2011). PCA analysis402 indicates that FA assemblages from Swiss lakes are generally more strongly influenced by FAs403 from autochthonous sources, while the Fennoscandian lakes and especially some littoral404 samples show high relative abundances of terrestrial FAs. As mentioned above (Section 3.4)405 we therefore eliminated longer-chain FAs (C24, C26, C28) originating from terrestrial406 environments from further analyses of FA abundances to reduce the effects of varying terrestrial407 influences on our comparisons between relative abundances of FAs and CH4 concentrations in408 lakes.409 18 The relatively lower δ13C values of C16:1ω7c, C16:1ω5c/t, and C18:1ω7c (Fig. 3) indicate that410 these FAs are, at least partly, produced by organisms incorporating isotopically light carbon,411 such as MOB. Literature sources confirm that they are produced by MOB, though not412 exclusively. C16:1ω7c, for example, was found to be associated with MOB type I, and also some413 MOB type II (Boschker et al. 1998; Bodelier et al. 2009a, 2012). However, this FA has also414 been reported for phytoplankton, zooplankton, fungi, mycobacteria and higher plants (Volkman415 et al. 1980; Woszczyk et al. 2011). C16:1ω5c/t was also found to be associated with MOB type I416 (Bodelier et al. 2012), and C18:1ω7c with MOB type II (Bowman et al. 1993; Deines et al. 2007;417 Bodelier et al. 2009a) and some MOB type Ia (Bodelier et al. 2009a). C18:1ω7c was also reported418 to be present in some other bacteria (Zegouagh et al. 2000). As we were mainly interested in419 relationships between MOB-derived FAs and CH4 concentrations, we focused our numerical420 analyses on these compounds, which were more depleted in 13C and were characterized by a421 δ13C signature typical for organisms incorporating CH4-derived carbon. Other chemotrophs,422 possibly abundant in stratified lakes at the chemocline, have also been reported to produce423 isotopically light lipids (Enrich-Prast et al. 2009; Zemskaya et al. 2012). However, the highest424 abundances of C16:1ω7c, C16:1ω5c/t, and C18:1ω7c were observed in oxic sediment samples and425 therefore, above the oxycline (and above any existing chemocline) in the lakes (Fig 5). Also,426 the correlations of these FAs with MOB concentrations of the sediments (Table 1), and the427 correlations of the relative abundances of these FAs with in-lake CH4 concentrations in our428 dataset (Table 2) support that MOB are a relevant source of these FAs in our study lakes.429 430 431 5.2 Relationships between FA and MOB abundances432 433 The distribution of MOB types I and II across the entire set of analysed lakes differs from analyses434 of terrestrial soils where type II usually dominates (Pan et al., 2010; Bodelier et al., 2013).435 19 This difference has been observed in freshwater sediments before (Borrel et al., 2011), where436 MOB type II are less dominant and sometimes even absent (Sundh et al. 2005; Schubert et al.,437 2010). Some of the genetic clusters belonging to the abundant type Ib MOB have been438 designated as typical freshwater MOB, only being found in these habitats (Lüke and Frenzel,439 2011; Knief 2015).440 Relatively strong correlations (rs = 0.42-0.69) are apparent between relative abundances of 13C- 441 depleted FAs and all three MOB groups (Ia, Ib, II), as well as with the sum of MOB. These 442 relationships are apparent even though different confounding processes potentially affect the 443 relative abundance data of FAs (e.g. in-lake production rates of short chain FAs by organisms 444 other than MOB) and the absolute abundances of MOB DNA expressed relative to the organic 445 content of the sediments (e.g. sediment homogeneity, variable accumulation rates, dilution by 446 terrestrial organic matter). The observed relationships confirm that in our dataset abundances of 447 13C-depleted FAs are positively related with abundances of MOB as quantified by qPCR. 448 The strongest and most consistent correlations were observed between MOB abundances and 449 C18:1ω7c. However, the abundances of the different MOB types were strongly inter-correlated (rs450 = 0.67-0.83). Therefore, this finding does not necessarily indicate that C18:1ω7c is produced by all 451 three MOB types in our study lakes, since, in principle, the apparent correlation of C18:1ω7c to all 452 MOB types could also result from production by one MOB type only. C16:1ω7c is only related 453 significantly to MOB type Ib. This suggests that MOB type Ib may be one of the main producers 454 of C16:1ω7c in the studied ecosystems. However, significant amounts of this FA have been 455 demonstrated to be produced also by MOB type Ia as well as type II (Bowman et al., 1993; 456 Bodelier et al., 2009a). 457 The relative abundance of C16:1ω5c/t was not clearly related to MOB abundance, although the 458 depletion in 13C supports that the compound contains CH4-derived carbon. In general, C16:1ω5c/t 459 was not very abundant in the analysed sediments. C16:1ω5 is not abundant in Methylobacter, 460 20 which is the type Ia genus mostly dominating freshwater sediments (Borrel et al., 2011) and this 461 could explain the lack of relationship of C16:1ω5c/t with MOB abundance. 462 The sum of C16:1ω7c,C16:1ω5c/t, and C18:1ω7c correlated most closely with MOB type Ib. This 463 relationship is possibly driven by C16:1ω7c, since in most sediments this was the most abundant 464 of the three more 13C-depleted FAs. However, the sum of C16:1ω7c,C16:1ω5c/t, and C18:1ω7c is more 465 robustly correlated with MOB type Ib than C16:1ω7c, suggesting that the less abundant 13C- 466 depleted FAs, particularly C18:1ω7c, reinforced this relationship. 467 468 469 5.3 Relationships between FA abundances and CH4 concentrations470 471 As expected we observed positive relationships between the abundance of 13C-depleted FAs472 and CH4 concentrations in our dataset. However, only relatively weak relationships were473 observed when all sediment samples were examined together, with C16:1ω7c and the sum of 13C-474 depleted FAs correlating with CH4 concentrations in the surface water and the sediments and475 C18:1ω7c with CH4 concentrations in the surface water. However, none of these relationships476 remained significant after correction for multiple testing (Table 2). Relationships between FA477 abundances and CH4 concentrations were noticeably stronger and statistically significant when478 oxic and anoxic sediments were examined separately. The highest correlations between FA479 concentrations (for C16:1ω7c and the sum of 13C-depleted FAs) and lake-water CH4 480 concentrations were apparent for oxic sediments (rs = 0.56-0.62), whereas observed rs values481 were slightly lower (for C18:1ω7c, C16:1ω7c and the sum of13C-depleted FAs) for anoxic sediments482 (rs = 0.46-0.55). These findings agree with the interpretation that the proportion of organic483 carbon originating from MOB increases in lakes and in lake sediments with higher CH4 484 concentrations, as has been suggested to explain observed relationships between estimates of485 in-lake CH4 abundance and δ13C values of aquatic invertebrates that can feed on MOB (van486 21 Hardenbroek et al. 2013; Schilder et al. 2015). However, our results also suggest that this487 proportion increases to a different extent in sediments deposited in oxic and anoxic488 environments. The highest abundances of 13C-depleted FAs were observed in oxic deep-water489 and littoral sediments (Fig 4). CH4 can be oxidized in both oxic and anoxic conditions, but490 aerobic CH4 oxidation is considered to be the dominant process, oxidizing up to 99 % of CH4 491 produced in lakes (Bastviken et al. 2008; Blees et al. 2014). The growth of aerobic MOB is not492 only limited by CH4 but also by the availability of electron acceptors, mainly oxygen (Amaral493 and Knowles 1995; Hanson and Hanson 1996). In oxic deep-water and littoral environments494 oxygen is generally abundant and MOB can profit most from increasing CH4 availability,495 explaining the highest abundances of 13C-depleted FA groups observed for sediments deposited496 under oxic condition. In the lakes where anoxic conditions develop, the CH4 concentrations in497 the deep water layers are high during summer stratification, when our samples were taken.498 However, the absence of oxygen or other electron acceptors can limit CH4 oxidation (Rudd et499 al. 1976; Amaral and Knowles 1995) at the sediment-water interface and in stratified, anoxic500 lakes CH4 oxidation is usually most extensive at the oxycline in the open water (Bastviken et501 al. 2008, Schubert et al. 2010, Milucka et al. 2015). Our results could be explained if MOB502 occur at higher concentrations where the sediment-water interface coincides with the oxycline503 than in the open water, since here the sediments would provide a stable substrate for these504 microorganisms to grow on, and/or if FAs originating from MOB in the open water would be505 partly decomposed during sedimentation and contribute to a lower extent to FAs measured at506 the sediment-water interface than MOB growing in the surfical sediment layers. In these507 situations, a high contribution of MOB-derived FAs would be expected for sediment samples508 located at or just below the oxycline, as confirmed by the highest abundance of 13C-depleted509 FAs observed for our dataset in oxic samples overlain by relatively CH4-rich water.510 The abundances of C16:1ω7c and of the sum of 13C-depleted FAs in the surficial sediment layers511 were robustly correlated with CH4 concentrations of the deeper sediment layers 10 cm below512 22 the sediment surface, both for the anoxic and oxic sediment samples (rs = 0.57-0.59 and 0.61-513 0.65, respectively; Table 2; Fig. 4). This suggests that the overall CH4 richness of sediments,514 and supply of CH4 to the uppermost sediment layers, also promote higher abundances of CH4-515 derived carbon in sedimentary organic carbon in the surface sediment samples.516 517 6. Conclusions and implications for palaeoenvironmental reconstructions518 519 We show that in 29 small lakes across Europe the abundance of13C-depleted FA groups520 relative to other FAs produced in freshwater ecosystems increases with increasing CH4 521 concentrations, at least when relationships are examined separately for sediments deposited in522 oxic and anoxic environments. This is expected if the relative contribution of CH4-derived523 carbon in lacustrine sedimentary organic matter, originating from MOB, increases with524 increasing in-lake CH4 concentrations. This interpretation is supported by the analysis of the525 number of DNA copies of MOB in the examined sediments, which in our dataset is clearly526 correlated with the abundance of 13C-depleted FAs. However, our analyses also indicate that527 the proportion of 13C-depleted FA groups was highest in oxic sediment samples deposited in528 environments with relatively high CH4 concentrations, suggesting that the proportion of CH4-529 derived sedimentary organic carbon may also be elevated in these sediments. In contrast, lower530 proportions of these 13C-depleted FAs were observed in sediments deposited in anoxic sections531 of the study lakes. Our analyses also show that different relationships between the relative532 abundance of 13C-depleted FA groups and in-lake CH4 concentrations are observed for533 sediments deposited in oxic and anoxic sediments. This implies that relationships between the534 proportion of CH4-derived carbon in aquatic organic matter and CH4 concentrations may also535 differ between these two depositional environments, at least for sedimentary organic matter536 deposited in small temperate lakes such as the ones we examined in our study.537 23 The robust correlations observed between CH4 concentrations of the sediments and the538 abundance of 13C-depleted FAs suggest that these relationships are not just with CH4 539 abundances at or above the sediments. Instead, CH4-rich lakes, characterized by high CH4 540 production and concentrations in the sediments, seem to be generally characterized by higher541 abundances of 13C-depleted FA groups in the surface sediment, although we again observed542 different relationships between the abundance of these FAs and CH4 concentrations in the543 sediments for oxic and anoxic environments.544 Our findings have two potential implications for approaches reconstructing past changes545 in CH4 availability in lakes based on geochemical analyses of lake sediments. First, they546 confirm that higher abundances of 13C-depleted FAs, from FA groups that are known to be547 produced by MOB, can be found in the sediments of small lakes under higher CH4 548 concentrations. This agrees with earlier interpretations that the observed relationships between549 d 13C values of unspecific deposit- or filter-feeding invertebrate groups and estimates of in-lake550 CH4 abundance (van Hardenbroek et al. 2013; Schilder et al. 2015) can be explained by higher551 MOB abundances under higher in-lake CH4 concentrations. The results therefore support that552 reconstructions of d 13C values of CH4-sensitive invertebrate groups, such as Chironomini and553 Daphnia, may provide information on past changes in in-lake CH4 concentrations in small554 temperate lakes, as has been suggested in earlier studies (e.g. van Hardenbroek et al. 2013;555 Schilder et al. 2015), since the d 13C values of these organism groups can be expected to be556 strongly driven by the availability of 13C-depleted lacustrine organic matter, which, as our557 results suggest, is related to CH4 concentrations. Second, our results support that lipids558 originating from MOB may be more abundant in environments with high CH4 concentrations559 in the water or the sediments, and that lipid records may provide insights into past changes in560 CH4 concentrations in lakes. Although FAs have been reported to decay rapidly in lake561 sediments (e.g. Muri and Wakeham 2006) other lipid groups produced by MOB or562 methanogens, such as bacteriohopanepolyols, and glycerol dialkyl glycerol tetraethers563 24 (GDGTs) (Coolen et al., 2008; Naeher et al., 2012; Sinninghe Damsté et al., 2012) have been564 analysed in downcore records and related to past variations in CH4cycling of lakes. However,565 our results also suggest that the relationship between the abundance of lipids originating from566 MOB and CH4 concentrations may be complex and strongly influenced by oxygen availability567 at the sediment water-interface and by a different preservation of lipids produced in the open568 water column compared with those produced in sediments.569 A major constraint of our study is that other organisms than MOB can contribute to the570 three 13C-depleted FA groups detected at our study sites and that FA groups strictly limited to571 MOB were not detected in our survey. Although the different lines of evidence (FA analyses,572 qPCR analyses, correlations with CH4 concentrations) are all consistent with the interpretation573 that the contribution of MOB-derived organic matter in lake sediments increases with574 increasing in-lake CH4 concentrations, it is unclear how the observed relationships were575 influenced by the production of C16:1ω7c,C16:1ω5c/t, and C18:1ω7c FAs by other organisms than576 MOB. Our results should therefore be corroborated by similar surveys focusing on lipid groups577 specific to MOB such as more specific FAs or specific bacteriohopanepolyols (e.g.578 methylcarbamate lipids related to C- 35 amino-bacteriohopanepolyols, Rush et al. 2016), or579 alternative approaches which, e.g., assess the abundance of MOB-derived DNA relative to other580 bacterial and algal DNA in lake sediments. An additional uncertainty associated with our581 approach is the extent to which catchment carbon cycling reinforced or masked the relationship582 between FAs and CH4 concentrations. For example, relationships are noticeably stronger in our583 dataset when analyses focus on shorter chain FAs only, since longer chain FAs originating from584 terrestrial plants apparently influence the relative concentrations of FAs in the analysed585 sediments. Similar effects may affect the relative abundances of short-chain FAs in our data as586 well. Future studies would therefore ideally aim to more fully quantify varying terrestrial587 organic matter input between and across lake basins to assess the extent that this organic matter588 source influences FA abundance analyses in surficial lake sediment samples.589 31 Volkman, J.K., Johns, R.B., Gillan, F.T., Perry, G.J., Bavor Jr, H.J., 1980. Microbial lipids of826 an intertidal sediment—I. Fatty acids and hydrocarbons. Geochim. Cosmochim. Acta827 44, 1133–1143. doi:10.1016/0016-7037(80)90067-8828 Whiticar, M.J., Faber, E., Schoell, M., 1986. Biogenic methane formation in marine and829 freshwater environments: CO2 reduction vs. acetate fermentation—Isotope evidence.830 Geochim. Cosmochim. Acta 50, 693–709. doi:10.1016/0016-7037(86)90346-7831 Wooller, M.J., Pohlman, J.W., Gaglioti, B. V., Langdon, P., Jones, M., Walter Anthony,832 K.M., Becker, K.W., Hinrichs, K.-U., Elvert, M., 2012. Reconstruction of past833 methane availability in an Arctic Alaska wetland indicates climate influenced methane834 release during the past ~12,000 years. J. Paleolimnol. 48, 27–42. doi:10.1007/s10933-835 012-9591-8836 Woszczyk, M., Bechtel, A., Gratzer, R., Kotarba, M.J., Kokociński, M., Fiebig, J., Cieśliński,837 R., 2011. Composition and origin of organic matter in surface sediments of Lake838 Sarbsko: A highly eutrophic and shallow coastal lake (northern Poland). Org.839 Geochem. 42, 1025–1038. doi:10.1016/j.orggeochem.2011.07.002840 Yang, H., Andersen, T., Dörsch, P., Tominaga, K., Thrane, J.-E., Hessen, D.O., 2015.841 Greenhouse gas metabolism in Nordic boreal lakes. Biogeochemistry 126, 211-225.842 Zegouagh, Y., Derenne, S., Largeau, C., Saliot, A., 2000. A geochemical investigation of843 carboxylic acids released via sequential treatments of two surficial sediments from the844 Changjiang delta and East China Sea. Org. Geochem. 31, 375–388.845 doi:10.1016/S0146-6380(00)00007-3846 Zemskaya, T.I., Sitnikova, T.Y., Kiyashko, S.I., Kalmychkov, G. V., Pogodaeva, T. V.,847 Mekhanikova, I. V., Naumova, T. V., Shubenkova, O. V., Chernitsina, S.M., Kotsar,848 O. V., Chernyaev, E.S., Khlystov, O.M., 2012. Faunal communities at sites of gas- and849 oil-bearing fluids in Lake Baikal. Geo-Marine Lett. 32, 437–451. doi:10.1007/s00367-850 012-0297-8851 32 Figures and Tables Fig. 1. Study sites located in Finland (6 lakes), the Netherlands (2), Sweden (11) and Switzerland (10). 33 Fig. 2. Principal components analysis (PCA) summarizing variations in relative abundances of FAs in the surface sediment samples. Site names in capital and lowercase letters refer to samples taken in the deepest parts of lakes and littoral samples, respectively. -1.0 1.0 -1.0 1.0 C14 C15 aC15 iC15 C16:1ω5c/t C18:1ω9 C18:1ω7c C16:1ω7c C16 C18:2 C18 C20 C22:1 C22 C24 C26 C28 GRI HAR LIL SKO KIS ILR GLI ERS MRN STV SGL GER HIN ROT SEE HIJK WAY MEK LOV NIM VAL SYR JYV BUR HAS HUT LAU NUS SCW gri har lil sko kis ilr gli ers mrn stv sgl Swedish lakes, littoral sediments Swiss lakes, deep sediments Swedish, Finnish and Dutch lakes, deep sediments PCA axis 1 PCA axis 2 34 Fig. 3. Boxplots indicating the distribution of FA δ13C values in deep water sediment samples from the 29 examined lakes. Central horizontal lines indicate median values and the bottom and top of boxes the 25th and 75th percentiles, respectively. Circles indicate samples more than 1.5 times the interquartile range above the third and below the first quartile, respectively. Boxplots for the 13C- depleted FAs that are discussed in detail in the text are marked in grey. C16:1ω7c C14 iC15 aC15 C15 C16:15c/t ω C18:1ω9 C18:2 C17 C16 C24 C26 C18 C20:1 C20 C22:1 C18:17c ω C22 C28 -80 -70 -60 -50 -40 -30 -20 13 δC (‰) 35 Fig. 4. Total abundance of 13C-depleted FAs (C16:1ω7c,C16:1ω5c/t, and C18:1ω7c) in the sediment samples compared with CH4 concentrations ([CH4]) in the surface waters (A- B), 10 cm above the sediment (C-D), and 10 cm below the sediment surface (E-F) (note deep sediments Anoxic conditions A FE DC B 0.1 1 10 1 10 0.1 10 1000 0.1 0.1 10 1000 10 100 1000 10 100 1000 0.0 0.50.40.30.20.1 0.0 0.50.40.30.20.1 0.0 0.50.40.30.20.1 13C-depleted FAs / short-chain FAs 13C-depleted FAs / short-chain FAs 13C-depleted FAs / short-chain FAs Oxic conditions deep sediments littoral sediments [CH4] 10 cm below sediment surface (mmol l-1) [CH4] 10 cm above the sediment (mmol l-1) [CH4] surface water (mmol l-1)[CH4] surface water (mmol l-1) [CH4] 10 cm above the sediment (mmol l-1) [CH4] 10 cm below sediment surface (mmol l-1) 36 the logarithmic scale). FAs are expressed as abundances relative to the sum of shortchain FAs (C14 to C22). Samples obtained from the deepest point of lakes with anoxic bottom waters (A, C, E) are plotted separately from samples obtained in the centre of lakes with oxic bottom waters (filled circles) and in shallower sections of lakes (open circles; B, D, F). Concentrations are expressed per volume of water (A-D) or wet sediment (E-F). 37 Table 1 Spearman rank correlation (rs) values for relationships between relative abundances of FAs in the sediments and DNA copies (pmoA gene) of MOB Ia, MOB Ib, MOB II and all MOB. *,** and *** mark p-values of below 0.05, 0.005 and 0.0005, respectively, and values in brackets indicate relationships that are no longer significant after False Discovery Rate (FDR) correction for multiple testing (Garcia 2004). The rs values are provided for C16:1ω7c, C16:1ω5c/t, C18:1ω7c, and the sum of these 13C-depleted FAs. C16:1ω7c C16:1ω5c/t C18:1ω7c Sum MOB Ia - - 0.50** 0.42* MOB Ib 0.42* -0.63*** 0.53** MOB II - - 0.62** - Total MOB --0.69*** 0.45* 38 Table 2 Spearman rank correlation (rs) values for relationships between relative abundances of FAs in the sediments and CH4 concentrations. Symbols *, ** and *** mark p-values below 0.05, 0.005 and 0.0005, respectively, and values in brackets indicate relationships that are no longer significant after FDR correction for multiple testing (within each group; Garcia 2004). Relationships with CH4 concentrations are calculated for relative abundances of C16:1ω7c, C16:1ω5c/t, and C18:1ω7c, the sum of these 13C-depleted FAs. All lakes C16:1ω7c C16:1ω5c/t C18:1ω7c Sum CH4 surface water (0.34*) - (0.34*) (0.40*) CH4 +10cm - - - - CH4-10cm (0.43*) - (0.36*) Oxic sediments C16:1ω7c C16:1ω5c/t C18:1ω7c Sum CH4 surface water 0.57* - - 0.62* CH4 +10cm - - - 0.56* CH4 -10cm 0.57 * (0.51*) - 0.59* Anoxic sediments C16:1ω7c C16:1ω5c/t C18:1ω7c Sum CH4 surface water - - 0.55* - CH4 +10cm (0.46*) - - (0.46*) CH4-10cm 0.65** - - 0.61** 39 Supplementary material Table S1 Geographical location and limnological characteristics of the 29 examined lakes. See Rinta et al. (2015) for more details Lake name Abbreviation Country Lon decimal °E Lat decimal °N Altitude m a.s.l. Lake area (km2) Max. depth (m) O2 bottom (mg l-1) Core depth (m) Organic matter C/N CH4 -10 cm (μmol l-1) Erssjön ERS SE 12.16 58.37 75 0.063 5.0 7.0 5.0 13.6 459.7 Glimmingen GLI SE 15.57 57.93 145 1.672 31.5 7.4 14.0 11.6 36.9 Grissjön GRI SE 15.14 58.77 139 0.227 16.0 1.9 12.2 20.1 204.6 Hargsjön HAR SE 15.24 58.27 108 0.994 6.2 5.9 6.2 9.4 1142.7 Illersjön ILR SE 14.99 58.58 96 0.069 12.1 0.2 12.1 11.4 1469.6 Kisasjön north KIS SE 15.65 58.01 99 0.958 8.5 0.2 8.5 10.2 575.1 Lillsjön LIL SE 16.14 58.66 84 0.026 7.7 0.1 7.7 16.4 432.8 Mårn MRN SE 15.87 58.59 27 0.617 15.3 0.2 15.3 10.2 1322.5 Skärgölen SGL SE 16.23 58.76 72 0.156 13.0 0.1 12.6 13.3 215.4 Skottenesjön SKO SE 12.14 58.35 51 0.263 6.0 7.5 2.5 12.9 276.7 Stora Vänstern STV SE 15.15 58.62 102 1.135 21.4 0.7 21.4 12.8 516.2 Hijkermeer HIJK NL 6.49 52.89 14 0.023 2.0 5.6 1.2 14.3 502.0 40 De Waay WAY NL 5.15 51.93 0 0.041 14.5 0.0 14.5 10.5 1443.2 Jyväsjärvi JYV FI 25.77 62.24 78 3.032 25.0 0.0 24.1 11.1 518.8 Lovojärvi LOV FI 25.03 61.08 108 0.050 18.1 0.0 17.7 12.2 1529.8 Mekkojärvi MEK FI 25.14 61.23 136 0.003 4.0 0.1 4.0 18.8 198.8 Nimetön NIM FI 25.19 61.23 152 0.004 12.7 0.0 12.4 14.3 392.0 Syrjänalunen SYR FI 25.14 61.19 138 0.009 9.2 0.1 9.1 14.2 408.5 Valkea-Kotinen VAL FI 25.06 61.24 156 0.041 8.3 0.0 5.8 14.3 198.5 Burgäschisee BUR CH 7.67 47.17 434 0.204 30.0 0.1 30.0 9.0 1482.0 Gerzensee GER CH 7.55 46.83 603 0.240 10.7 0.1 9.3 10.5 1257.9 Hasensee east HAS CH 8.83 47.61 434 0.067 5.8 0.3 5.5 9.8 543.0 Hinterburgsee HIN CH 8.07 46.72 1516 0.046 11.4 0.1 11.0 8.9 1421.6 Hüttwilersee HUT CH 8.84 47.61 434 0.344 15.5 0.1 15.4 8.9 980.1 Lauenensee LAU CH 7.33 46.40 1381 0.087 3.5 6.8 3.0 8.2 321.7 Nussbaumersee NUS CH 8.82 47.62 434 0.253 8.0 0.1 8.0 9.3 492.9 Rotsee ROT CH 8.31 47.07 403 0.443 16.0 0.1 14.6 9.4 1791.6 Schwarzsee SCW CH 7.28 46.67 1046 0.446 9.1 0.1 8.8 10.0 495.5