Hydrogen isotopes (δ2H) of polyunsaturated fatty acids track bioconversion by zooplankton

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 4.0 https://creativecommons.org/licenses/by/4.0/ Hydrogen isotopes (δ2H) of polyunsaturated fatty acids track bioconversion by zooplankton © 2021 The Authors. Functional Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society. Published version Pilecky, Matthias; Kämmer, Samuel K.; Mathieu‐Resuge, Margaux; Wassenaar, Leonard I.; Taipale, Sami J.; Martin‐Creuzburg, Dominik; Kainz, Martin J. Pilecky, M., Kämmer, S. K., Mathieu‐Resuge, M., Wassenaar, L. I., Taipale, S. J., Martin‐ Creuzburg, D., & Kainz, M. J. (2022). Hydrogen isotopes (δ2H) of polyunsaturated fatty acids track bioconversion by zooplankton. Functional Ecology, 36(3), 538-549. https://doi.org/10.1111/1365-2435.13981 2022 Functional Ecology. 2021;00:1–12. | 1wileyonlinelibrary.com/journal/fec Received: 14 July 2021 | Accepted: 22 November 2021 DOI: 10.1111/1365-2435.13981 RESEARCH ARTICLE Hydrogen isotopes (δ2H) of polyunsaturated fatty acids track bioconversion by zooplankton Matthias Pilecky1,2 | Samuel K. Kämmer1 | Margaux MathieuResuge1 | Leonard I. Wassenaar3 | Sami J. Taipale4 | Dominik MartinCreuzburg5 | Martin J. Kainz1,2 This is an open access article under the terms of the Creat ive Commo ns Attri bution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2021 The Authors. Functional Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society 1WasserCluster Lunz— Biologische Station, Lunz am See, Austria 2Department for Biomedical Research, Danube University Krems, Krems, Austria 3Department of Geological Sciences, University of Saskatchewan, Saskatoon, Canada 4Department of Biological and Environmental Science, University of Jyväskylä, Survontie, Finland 5Limnological Institute, University of Konstanz, Konstanz, Germany Correspondence Matthias Pilecky Email: [email protected] Funding information This work has been supported by the Austrian Science Fund (FWF; I3855) and the German Research Foundation (DFG; MA 5005/81) within the framework of the DACH collaboration (project ‘AquaTerr’). Handling Editor: Daniel Allen Abstract 1. Organisms at the base of aquatic food webs synthesize essential nutrients, such as omega3 polyunsaturated fatty acids (n3 PUFA), which are transferred to consumers at higher trophic levels. Many consumers, requiring n3 longchain (LC) PUFA, such as eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), have limited ability to biosynthesize them from the essential dietary precursor αlinolenic acid (ALA) and thus rely on dietary provision of LCPUFA. 2. We investigated LCPUFA metabolism in freshwater zooplankton using stable hydrogen isotopes (δ2H) of fatty acids as tracers. We conducted feeding experiments with the freshwater keystone grazer Daphnia to quantify changes in the δ2H value of body FA in response to the FA composition of their food and the δ2H value of the ambient water. 3. The isotopic composition of LCPUFA changed in Daphnia, depending on the integration of 2H from ambient water during de novo synthesis or bioconversion from dietary precursors, allowing us to distinguish dietary from bioconverted EPA in body tissue. We tested the applicability of these laboratory findings in a field setting by analysing δ2H values of PUFA in primary producers and consumers in eutrophic ponds to track EPA sources of zooplankton. 4. Multilinear regression models that included conversion of ALA to EPA correlated better with zooplankton δ2HEPA than seston δ2HEPA at low dietary EPA supply. 5. This study provides evidence that zooplankton can compensate for low dietary EPA supply by activating LCPUFA biosynthesis and shows that herbivorous zooplankton play a crucial role in upgrading FA for higher trophic levels during low dietary EPA supply. KEYWORDS bioconversion, compoundspecific stable isotopes, deuterium, ecophysiology, eutrophication, polyunsaturated fatty acids, trophic ecology, zooplankton 2 | Functional Ecology PILECKY Et aL. 1 | INTRODUCTION Tracing and quantifying the transfer of dietary energy and nutrients within aquatic food webs is critical for understanding trophic interactions among species. For decades, stable isotopes of carbon (13C), nitrogen (15N) and hydrogen (2H) from bulk tissue samples have been routinely applied in trophic ecology as natural diet tracers (Doucett et al., 2007; Fry, 2006; Kling et al., 1992). Fatty acids (FA) are increasingly used as specific dietary source biomarkers in aquatic consumers and for disentangling autochthonous versus allochthonous diet sources (Nielsen et al., 2018). The FA content of organisms enables researchers to evaluate different diet sources (Brett et al., 2009, 2017; Galloway et al., 2014; Kainz & Mazumder, 2005; Nielsen et al., 2018), but relies on the assumption that dietary FA remain unaltered in consumers. Moreover, models from controlled experiments need to take food limitations into account, which might change the FA profiles of consumers (Taipale et al., 2015; Twining et al., 2020). Omega3 polyunsaturated fatty acids (n3 PUFA), such as αlinolenic acid (ALA; 18:3n3), eicosapentaenoic acid (EPA; 20:5n3) or docosahexaenoic acid (DHA; 22:6n3), are particularly important for trophic studies (Carlson & Neuringer, 1999), since most consumers are unable to synthesize them de novo, that is, from lowmolecularweight precursors. However, various animals have been shown to be capable of elongating and desaturating dietary C18 PUFA to C20 PUFA to adapt their n3 PUFA profile to their specific physiological needs, which is a complex process involving several enzymes and comes at higher energy costs than direct dietary acquisition (Taipale et al., 2011; Twining et al., 2016). Compoundspecific stable isotope analyses (CSIA) of FA allow natural tracing of FA sources and, compared to bulk stable isotope analysis, are more informative at disentangling food web interactions (Pilecky, Kämmer, et al., 2021; Pilecky, Winter, et al., 2021; Twining et al., 2020). The typically large δ2H differences between aquatic versus terrestrial sources (Hobson et al., 2020; Solomon et al., 2009) offer a new and promising approach for 2HCSIA to better understand the sources and fate of FA in food webs. The C– H bonds of FA are stable and their H atoms do generally not exchange with ambient water, thus distinctive dietary H isotope source signals are retained (Sessions et al., 2004). Furthermore, the linking of C and H in molecules implies that reaction kinetics, for example, during bioconversion, and their alteration due to the introduction of heavier isotopes may affect the isotopic composition of both elements. While a similar potential exists for 13C of FA, one major disadvantage of using carbon isotopes is the relatively small isotope source signal differences (Taipale, Vuorio, Brett, et al., 2016; Taipale, Vuorio, Strandberg, et al., 2016). The large stable isotope fractionations of 2H/H in nature compared to 13C/12C may provide better resolution for tracking ecophysiological processes and resolving FA sources in terrestrial and aquatic ecosystems. One key advantage is that lipid biosynthesis leads to large 2H depletion in the formed products compared to the ambient water (Chikaraishi et al., 2004; Sessions et al., 1999). During FA synthesis, H isotope fractionation is induced by variations in the H isotope composition of cellular water, acetate and the cofactor NADPH, and by different isotopologue kinetics (a) during uptake and utilization of diet, (b) via enzymes involved in the lipid biosynthetic pathways and (c) during FA allocation and metabolism (Zhang et al., 2009). Similarly, FA elongation and saturation introduces new 2H atoms into the molecule, altering its δ2H composition depending on environmental factors, such as diet, the isotopic composition of ambient water and temperature (Chikaraishi et al., 2004; Zhang & Sachs, 2007). While isotope fractionation during FA synthesis in primary producers has been investigated, nothing is known about the alteration of δ2H values by consumers, including processes such as bioconversion of ALA to LCPUFA. We hypothesized that 2HCSIA tracks the integration of 2H into LCPUFA and thus provides a basis for differentiation among dietary and biosynthesized LCPUFA. To test this, we conducted controlled laboratory experiments varying the dietary quality and δ2H of ambient water to track isotope integration pathways (i.e. dietary vs. metabolic) into EPA, which is a key n3 PUFA in consumers, including Daphnia. We also evaluated the effect of introducing excess of 2H into FA, on the δ13C value of the linked C. Subsequently, we tested this hypothesis in the field by analysing water, seston and zooplankton from eutrophic ponds during the summer period to investigate, based on compoundspecific δ2H values, the origin of EPA in zooplankton by applying predictive models that tested for bioconversion or dietary allocation of FA. 2 | MATERIALS AND METHODS 2.1 | Controlled feeding experiments The unicellular green alga Acutodesmus obliquus (Kützing, 1833, formally known Scenedesmus obliquus) and the cryptophyte Cryptomonas ozolinii (Skuja, 1939) were cultured separately in 1 L Wright's cryptophyte (WC) medium at 20°C (Guillard & Lorenzen, 1972). Both algae are easily ingestible for Daphnia but differ in their nutritional quality with respect to PUFA composition (Taipale et al., 2013). A. obliquus lacks PUFA with more than 18 Catoms, whereas C. ozolinii is rich in EPA and DHA (Ahlgren et al., 1992; Masclaux et al., 2009; Table S1). WC medium was spiked above natural abundance levels by using 2H2O (99.8% D2O; SigmaAldrich) at two concentrations (c. 25, or 50 mg/L) to achieve increased δ2H values compared to standard culture water. The measured δ2H composition of the water in these three treatments was −78.9‰ ± 0.7 (reference), +92.2‰ ± 0.1 (25 mg/L) and +273.1‰ ± 10.0 (50 mg/L), relative to the VSMOW standard. The δ2H values of culture medium were monitored as described below. Laboratory feeding experiments were performed using a clone of the herbivore Daphnia magna (obtained from the University of Clermont Auvergne, France), kept in filtered (<0.7 μm) water (20°C) supplemented with 5% ADaM medium (Klüttgen et al., 1994) and fed ad libitum with the respective algae species before the start of the experiment, which kept them in a fast growing stage. For each run and treatment (in triplicate), 30 neonates (F1 generation) were | 3 Functional Ecology PILECKY Et aL. transferred into a single jar (400 ml). Before each feeding, the total organic carbon content of the algal cultures was adjusted via OD to supply ~1 mg C L−1 day−1 to D. magna in all treatments to ensure nonlimiting diet supply (Lampert, 1978). Water was replaced weekly in all jars. The F1 generation was fed algae that were precultured for at least one month with −78.9‰, +92.2‰ or +273.1‰ 2H2O as tracer. After 7 days of feeding, and before the onset of egg production, ~20 individuals (~1.5 mg dw) of Daphnia were collected from each jar, freezedried and subjected to lipid extraction. The remaining individuals were retained to produce F2 neonates, which were transferred into new jars under the same feeding and water treatments. Additionally, a subset of the F2 offspring was kept in +273.1‰ spiked lake water/ADaM while fed with nondeuterium enriched algae to investigate how the δ2H of ambient water affected the H isotope composition of FA in Daphnia (Figure 1). Comparing data from both generations revealed minor differences regarding isotopic composition of FA (Table S5; Figure S6), suggesting reported data from the second generation were at or close to dietary isotopic steady state. Experiments with small invertebrates did not require ethical approval. 2.2 | Field sampling Water, dietary plankton (seston) and zooplankton were collected monthly from nine shallow (1– 2 m depth) fishponds located in northern Austria (48°49′N, 15°17′E, 510 m) from June to September 2020. Permission for sampling was granted by the private owners. Pond water samples were collected in triplicate, filtered through a 30µm mesh and stored in plastic vials for δ2H analysis of the water. The most edible seston size fraction for zooplankton (<30 µm; Brooks & Dodson, 1965; Vanderploeg & Paffenhöfer, 1985) was sampled in triplicate from each pond using a Schindler trap, prefiltered through a 30µm mesh and retained on preweighed and precombusted GF/C filters (0.7 µm), and subsequently stored at −80°C until further analysis. Zooplankton samples were collected in triplicate from each pond by gently pulling an Apstein's zooplanktonnet (500 µm; HydroBios GmbH) below the surface (~0.5 m). Zooplankton size fractions (>500, 500– 250 and <250 µm particle size) were collected in sterile Falcon® tubes (50 ml). Both the filters and zooplankton were frozen (−80°C), freezedried (Virtis Genesis Freeze dryer, for minimum 24 hr) and subsequently stored frozen (−20°C) until further processing. Zooplankton samples (>500 µm) from one sampling (July) were separated into cladocerans, copepods and Chaoborus larvae for taxaspecific isotope analysis. Because no significant differences in δ2H values of FA were detected, all other samples were processed as homogenized bulk samples. 2.3 | Gas chromatography (GC) and isotope ratio mass spectrometry (IRMS) Lipids were extracted from all samples according to Heissenberger et al. (2010). Briefly, freezedried samples were homogenized and mixed with chloroform:methanol (2:1 V/V) following the addition of 0.9% NaCl, sonication, vortexing and centrifuging 3× to remove nonlipid materials. Extracted lipids were evaporated to a final volume of 1.5 ml under N2 gas flow. For FAME formation, samples were incubated with sulphuric acid:methanol (1:100 V/V) for 16 hr at 50°C, following the addition of KHCO3 and hexane. Samples were shaken, vortexed and centrifuged and the upper organic layers collected, pooled and concentrated under a N2 gas flow. FIGURE 1 Outline of the controlled laboratory experiment. The two algae Scenedesmus obliquus and Cryptomonas ozolinii were grown in medium spiked with different 2H2O concentrations (−78.9‰, +92.2‰ or +273.1‰, VSMOW) and subsequently fed to the first generation of Daphnia. Neonates of the second generation were raised under the same conditions as their mothers, except for those receiving the nondeuterium enriched diets. This group was split with one half kept in 2H enriched ambient water while still receiving nonenriched diet (blue square), while the other half was kept in a nonenriched environment and served as control. Data of first and second generation were compared and suggested dietary isotopic steady state for generation 2 (see Supporting Information) 4 | Functional Ecology PILECKY Et aL. Fatty acid methyl esters (FAME) were quantified using a gas chromatograph (TRACE GC ThermoFisher Scientific, Detector: FID 260°C, Carrier gas: He: 1 ml/min, Detector gases: H2: 40 ml/min, N2: 45 ml/min, air: 450 ml/min, temperature ramp: 140°C (5 min)— 4°C/min– 240°C (20 min) = 50 min) equipped with a temperatureprogrammable injector and an autosampler. A Supelco SP2560 column (100 m, 25 mm i.d., 0.2 µm film thickness) was used for FAME separation and quantification. Chromeleon 7 software (ThermoFisher) was used for peak integration. FAME were identified by comparison of retention times to known reference standards (37component FAME mix, 47885U, Supelco; SigmaAldrich). Fatty acid concentrations were quantified using calibration curves based on different known standard concentrations. FA are reported as µg/ mg dry weight after applying a conversion factor for each individual FAME accounting for the mass fraction of the methyl group. 2Hand 13CCSIA were performed according to Pilecky, Winter, et al. (2021), Pilecky, Kämmer, et al. (2021), using a Thermo Trace 1310 GC (ThermoFisher), coupled by a ConFlo IV interface (ThermoFisher) to a continuousflow isotoperatio mass spectrometer (DELTA V Advantage, ThermoFisher). FAME were separated using either a VFWAXms 60 m column, 0.25 mm ID, film thickness 0.25 µm; or a VFWAXms 30 m column, 0.32 mm ID, film thickness 1 µm (both Agilent). The temperature programme for the 60 m GC column started at 80°C, held for 2 min, after which the temperature was ramped by 30°C/min to 175°C, by 5°C/min to 200°C and finally by 2.4°C/min to 250°C, which was then maintained for 30 min. The temperature programme for the 30 m GC column started at 80°C, held for 2 min, after which the temperature was ramped by 30°C/min to 175°C, and then by 5°C/min to 240°C, which was held for 35 min. For δ13C analysis, FAME were oxidized to CO2 in an online combustion reactor filled with Ni, Pt and CuO wires, at a temperature of 1,000°C. For δ2H analysis, FAME were pyrolysed to H2 gas by passing them through a high temperature graphite conversion reactor at 1,200°C. Samples were run against certified MeC20:0 stable isotope reference material (USGS70: δ13C = −30.53‰, δ2H = −183.9‰, USGS71: δ13C = −10.5‰, δ2H = −4.9‰ and USGS72: δ13C = −1.54‰, δ2H = +348.3‰). Weightedaverage combined δ2H values of monounsaturated fatty acids (MUFA = C14:1, C16:1, C18:1) were obtained by integrating over all isoforms (e.g. C18:1n7 + C18:1n9) because no clear baseline separation could be achieved. FA δ13C/δ2H values ( 𝛿IFA ) were corrected for the methyl group addition during methylation according to Pilecky, Winter, et al. (2021) and Pilecky, Kämmer, et al. (2021). Values for δ13C are referenced to Vienna PeeDee Belemite (13C:12C = 0.01118). Values for δ2H are standardized against Vienna Standard Mean Ocean Water (2H:1H = 155.76 ppm). For δ2H analysis of pond and experimental waters used, three replicates each sample were filtered before isotope analysis using a L2130I (Picarro Inc.) and using the techniques described elsewhere (Coplen & Wassenaar, 2015). For waters with D2O added, we used IAEA604 (+799.0‰) and IAEA VSMOW2 (0.0‰) as our bracketing standards. 2.4 | Data analysis Data analysis was conducted and plots were produced in R (Version 4.0.2, R Core Team, 2020) using the packages rstatix, ggplot2, ggpubr, lme4, relaimpo v2.2 and corrplot. Isotope data are summarized as the mean carbon or hydrogen isotope δ value ± SD. Unless indicated, values reported from the controlled experiment are from the second generation of Daphnia at dietary steady state. Paired t test and ANOVA were used for comparison of dietary and consumer compoundspecific δ values. The Pearson method was used for linear correlations. Multiple linear regression models were developed by a stepwise backward elimination process (Ghani & Ahmad, 2010). Explanatory variables were accepted or eliminated based on the combination of pvalues and Feldman's relative importance (Feldman, 2005; Grömping, 2006). 3 | RESULTS 3.1 | Fatty acid variation in δ2H values In the controlled laboratory feeding experiments, Daphnia fed C. ozolinii had significantly higher EPA contents than those fed A. obliquus (6.0 ± 1.9 and 1.1 ± 1.1 µg/mg, respectively, ANOVA, F1 = 110.5, p < 0.001), but significantly lower ALA contents (18.0 ± 11.5 and 34.4 ± 5.7 µg/mg, respectively, ANOVA, F1 = 3 8 . 2, p < 0.001; Table S1). The controlled feeding experiments using algae grown under various 2H water treatments revealed a strictly dietary allocation pattern with no significant differences between diet and consumers in the δ2H values of saturated FA (SFA = 14:0, 16:0, 18:0; ANOVA, F6 = 1.9, p = 0.172), and the PUFA ALA (ANOVA, F3 = 2.4, p = 0.139), stearidonic acid (SDA, 18:4n3; ANOVA, F3 = 2.7, p = 0.117) and gammalinolenic acid (GLA, 18:3n6; ANOVA, F3 = 1.42, p = 0.248). Linoleic acid (LIN, 18:2n6) was isotopically slightly 2Hdepleted in Daphnia (ANOVA, F3 = 5.5, p = 0.031). Daphnia δ2H values of MUFA showed dietdependent 2H enrichment; however, the δ2H values of diet and consumer FA were different. EPA in Daphnia that were feeding on A. obliquus, which does not contain EPA, was significantly enriched in 2H (−77.8‰ ± 43.5) compared to the δ2H values of the dietary precursors ALA and SDA (−202.8‰ ± 3.07 and −219.7‰ ± 10.2; ANOVA, F5 = 7.8, p = 0.0017). In contrast, Daphnia fed C. ozolinii showed similar δ2H values compared to diet for 14:0 (ANOVA, F3 = 2.6, p = 0.130), 16:0 (ANOVA, F3 = 5.2, p = 0.038), 18:1 (ANOVA, F3 = 0.3, p = 0.586), SDA (ANOVA, F3 = 1.9, p = 0.188) and EPA (ANOVA, F3 = 0.4, p = 0.516), whereas significant 2H isotope 𝛿 13CFA = ( 13C∕12CSample 13C∕12CVPDB −1 ) × 1, 000. 𝛿 2HFA = (2 H∕ 1 HSample 2H∕1HVSMOW −1 ) × 1, 000. | 5 Functional Ecology PILECKY Et aL. enrichment was detected in 18:0 (ANOVA, F3 = 21.7, p < 0.001), 14:1 (ANOVA, F3 = 20.1, p < 0.001), 16:1 (ANOVA, F3 = 9.0, p = 0.009) and a slight but significant decrease in the δ2H of LIN (ANOVA, F3 = 10.7, p = 0.006) and ALA (ANOVA, F3 = 8.1, p = 0.013; Figure 2). The δ2H composition of FA in Daphnia raised in 2Hspiked water (+351.9‰ vs. reference treatment) showed no 2H enrichment compared to the control for LIN (−85.1‰ ± 52.3 vs. −48.5‰ ± 55.8, t test, p = 0.516), GLA (−49.2‰ ± 42.8 vs. −66.5‰ ± 54.5, t test, p = 0.736) and ALA (−221.8‰ ± 8.4 vs. −219.5‰ ± 18.4, t test, p = 0.886), while MUFA were significantly 2H enriched (−53.9‰ ± 8.0 vs. −110.7‰ ± 13.9, ANOVA, F3 = 54.3, p < 0.001) with δ values close to those fed with 25 mg/L 2H2Osupplemented diet (−25.4‰ ± 16.7). ARA (−2.3‰ ± 11.1) and EPA (10.4‰ ± 10.7) of Daphnia fed A. obliquus were significantly enriched in 2H compared to the control algae (−73.5‰ ± 18.8 and −77.8‰ ± 10.7, respectively; ANOVA, F2 = 21.7, p = 0.002), while SDA was not significantly enriched in 2H (−229.1‰ ± 40.0 vs. −246.9 ± 20.7, t test, p = 0.641). We found no effect of environmental water on δ2H values of FA of interest for Daphnia fed with C. ozolinii raised in 2Henriched water compared to the control treatment (Figure 2). 3.2 | Fatty acid variation in δ13C values The introduction of 2H via diet or ambient water did not change the δ13C values for SFA (ANOVA, F3 = 0.1, p = 0.723), ALA (ANOVA, F3 = 0.2, p = 0.628) or LIN (ANOVA, F3 = 0.4, p = 0.527). Daphnia raised in 2Henriched water were more depleted in 13C compared to the control group with δ13C values of 14:1 (−31.74‰ ± 0.98 vs. −27.10 ‰ ± 1.12; ANOVA, F3 = 26.0, p < 0.001), 16:1 (−31.87‰ ± 2.88 vs. −28.89‰ ± 0.98; ANOVA, F3 = 6.5, p = 0.022), GLA (−34.28‰ ± 2.04 vs. −30.57‰ ± 2.91; ANOVA, F3 = 4.5, p = 0.053), SDA (−31.15‰ ± 1.94 vs. −29.22‰ ± 1.28; ANOVA, F3 = 4.1, p = 0.061), ARA (−30.63‰ ± 1.68 vs. −27.89‰ ± 1.37; ANOVA, F3 = 8.05, p = 0.013) and EPA (−32.73‰ ± 3.17 vs. −28.07‰ ± 0.53; ANOVA, F3 = 10.6, p = 0.006; Figure 3). 3.3 | δ 2H and δ13C analyses of LCPUFA in pond zooplankton The δ2H values of water in the ponds varied only slightly over the sampling timeframe, although each pond was distinctive, which is FIGURE 2 Compoundspecific deuterium allocation into fatty acids of Daphnia via diet (green, orange and blue) or environmental water (purple). MUFA experienced significant changes in δ2H values, as well as a significant enrichment in deuterium when Daphnia were raised in 2H2O, indicating FA synthesis. Strict dietary controlled allocation of fatty acids can be seen for LIN and ALA. Deuterium enrichment in ARA and EPA can be seen in Daphnia when raised in deuteriumenriched environmental water with poor diet quality (Acutodesmus obliquus) but not with high diet quality (Cryptomonas ozolinii) in which case EPA is also strictly dietary controlled 6 | Functional Ecology PILECKY Et aL. relevant for establishing comparative H isotope baseline conditions (−48.9‰ ± 6.0; min: −59.1‰, max: −37.1‰). None of the ponds deviated from the local meteoric evaporation line, indicating a lack of progressive evaporation over the time frame of sampling. These eutrophic ponds had generally high chlorophylla concentrations (89 ± 38 µg/L), with the exception of two ponds (< 20 µg/L at any time; Table S4). Cladocerans were the predominant zooplankton taxa in all ponds (>50%). Six ponds also contained high numbers of calanoid copepods and Chaoborus larvae. Mass fractions of ALA, EPA and DHA varied in seston during the sampling period (Table S4; Figure 4c). EPA contents of zooplankton were independent of seston EPA (Pearson, r = −0.039, p = 0.83) and remained similar from June to August (7.6 mg/g ± 3.3; 6.5 mg/g ± 3.2; 6.7 mg/g ± 2.2) with a slight increase in September (10.5 mg/g ± 3.3; ANOVA, F11 = 6.0, p < 0.001). A significant correlation was found between δ2H composition of seston and zooplankton for all FA (Pearson; p < 0.01), except C14:1 (p = 0.67), C18:0 (p = 0.05), SDA (p = 0.09) and ARA (p = 0.04). Except for 14:0, all SFA were enriched in 2H in zooplankton compared to seston (16:0: t(31) = −9.44, p < 0.0001; 18:0: t(31) = −10.88, p < 0.0001), while LCPUFA were more depleted (ARA: t(27) = 9.93, p < 0.0001; EPA: t(31) = 10.59, p < 0.0001; DHA: t(23) = 3.50, p = 0.0046; Figure 4a). The 2H depletion of EPA in zooplankton compared to seston was significantly correlated with a decrease in the EPA mass fraction in seston (Pearson, r = 0.82, p < 0.001). Similar correlations of ∆δ2HZooplanktonSeston and the respective FA mass fraction in seston were detected for ALA (R = 0.61, p < 0.001) and SDA (R = 0.4, p = 0.025; Figure 4b). Changes in δ2H of the individual FA were independent of zooplankton taxa, and no significant differences between Chaoborus larvae, copepods or cladocerans were found (Figure S6; ANOVA, df = 18, p = 0.053). The δ13C composition did not significantly differ among n3 PUFA (ALA, SDA, EPA; ANOVA, F3 = 0.4, p = 0.532) or n6 PUFA (LIN, ARA; ANOVA, F2 = 0.8, p = 0.380) in both seston and zooplankton in all ponds (Figure S5). 3.4 | Conversion models for predicting EPA δ2H values of zooplankton To predict the δ2H composition of zooplankton EPA, a multiple linear regression model was developed using stepwise backwards iteration to correlate δ2HEPA values of zooplankton to a combination of δ2H values of other zooplankton and seston FA, as well as the δ2H values of the environmental water. After iteration and convergence, the best fit model, using δ2H of zooplankton FA and environmental water, was: Overall, predictions of δ2HEPA values using this model (F4,27 = 16.42, p < 0.001, R2 = 0.67) were better than predictions based on correlations with dietary (i.e. seston) δ2HEPA (F1,30 = 13.97, p < 0.001, R2 = 0.32). Other potential covariates, such as SFA or dietary and zooplankton SDA were insignificant and thus removed from the model iteration process. However, when considering only samples with high seston EPA (>0.6 mg/g), seston δ2HEPA values were a better predictor of zooplankton δ2HEPA than from the conversion model. The calculations of Feldman's relative importance revealed a shift in the explanatory value from somatic ALA, 16:1, and environmental water, associated with conversion to dietary EPA at values of approximately 0.3– 0.5 mg/g dry weight of seston EPA (Figure 5a). 4 | DISCUSSION The aim of this study was to assess the potential of using δ2H to differentiate between dietary and bioconverted LCPUFA, in particular EPA, of Daphnia. EPA of Daphnia was significantly enriched in 2H when raised in 2Hsupplemented water on an EPAfree diet, showing 𝛿2 H EPA =0.65𝛿 2 H ALA +0.40𝛿 2 H 16:1 +1.08𝛿 2 H H2O − 38.26. FIGURE 3 δ13C values of FA detected in Daphnia fed Acutodesmus obliquus. No significant differences for FA could be seen for SFA, LIN and ALA. FA that are suspected to be de novo synthesized (14:1, 16:1), or had to be converted from a precursor (EPA, ARA) were significantly depleted in 13C. A nonsignificant trend can be seen for the conversion intermediates GLA and SDA | 7 Functional Ecology PILECKY Et aL. integration of H from ambient water during FA conversion and biosynthesis. Applying 2HCSIA to investigate FA bioconversion in consumers under field conditions revealed that biosynthesis of EPA in zooplankton increased with decreasing dietary EPA supply, whereby a critical threshold appeared to be at 0.5 mg EPA/g dry weight of seston. According to the environmental water δ2H baseline values, bioconversion of ALA to EPA in Daphnia resulted in significant changes in δ2HEPA values. This study demonstrated that stable H isotopes of FA in aquatic organisms can be used in trophic ecology as a highly informative tracer to better understand how consumers use and/or convert dietary resources. Integration of 2H from environmental water into FA occurs during redoxreaction steps via NADH/NADPH, which is a common cofactor in many catabolic/anabolic processes (Yang & Sauve, 2016), leading to 2Henriched alkyl chains during FA synthesis and elongation (Baillif et al., 2009; Figure 5b). In general, the involvement of heavier isotopes decelerates reaction kinetics. Generally, if a reaction directly involves breaking a 12C– 2H bond, it is seven times slower compared to a reaction breaking a 12C– 1H bond (primary isotope effect; Westheimer, 1961). If the heavier isotope is not directly involved in a chemical reaction but is present in the nonreacting part of a heavier molecule (M + 1), the reaction rate constant (k) is also lower than the lighter isotopologue with mass (M). This results in reaction products with a lower 2H isotopic content than their educts, that is, the originally used substrates (secondary isotope effect; Cleland, 2005; Maggi & Riley, 2010). In our case, we assume that kM+1 during FA elongation was constant for secondary isotope effects, regardless of whether the additional mass is introduced by substitution of 12C with 13C or 1H with 2H. Thus, if an FIGURE 4 δ2H and δ13C in FA of seston (<30 μm size fraction) and zooplankton in nine different ponds sampled from June to September 2020. (a) Significant enrichment in 2H of the saturated fatty acids C16 and C18 in zooplankton compared to seston, and significant depletion in 2H of ARA and EPA. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (b) Relationships between fatty acid mass fractions in seston and fractionation of hydrogen isotopes in fatty acids of zooplankton relative to seston; the isotopic fractionation of 2H in EPA decreased strongly with increasing dietary EPA mass fractions (Pearson correlation values were calculated for differences in zooplankton to seston plotted vs. the FA concentration in the seston). (c) Principal component analysis of the relative FA composition of seston and zooplankton. Only FA contributing to more than 0.1% were considered 8 | Functional Ecology PILECKY Et aL. increasing amount of 2H is introduced into the acyl chains, a correspondingly lower amount of 13Ccontaining isotopomers is processed compared to the isotopically lighter 12C - a l l - 1Halkyl molecules, leading to an overall 13C depleted isotopic composition of the FA. This means that, in isotope studies, 13Cand 2Hvalues of individual FA should not be considered as independent variables. In our controlled feeding experiments, the integration of 2H into FA from ambient water was clearly seen in case of MUFA, as the FIGURE 5 Multiple linear regression models (MLRM) based on δ2H values of FA suggest significant conversion activity in zooplankton in eutrophic ponds. (a) In a MLRM using environmental water, zooplankton ALA (precursor), C16:1 (proxy for FA synthesis) and seston EPA δ2H values, a switch in Feldman's relative importance, calculating their contribution to the model at the respective level, occur at mass fractions between 0.3 and 0.5 mg/L dry weight in seston indicating transition from conversion to dietary allocation at this threshold. (b) Tracking H atoms during biochemical conversion of ALA to EPA by fatty acid synthase and other involved enzymes displays the involvement of water H atoms (red) various energy sources (lilac) and H atoms subjected to isotopic fractionation during the potentially reversible desaturation processes (light green)