Introduction

Stable isotopes are powerful for analyzing trophic interactions in aquatic ecosystems. Differences in stable isotope ratios of carbon (δ 13C) and nitrogen (δ 15N) are frequently used to elucidate the diet of aquatic organisms and to describe food web structure (Peterson and Fry 1987). However, isotope analyses can produce ambiguous results when variability within food sources is large relative to differences between sources (Phillips and Gregg 2001; Fry 2006). Ratios of the stable isotopes of hydrogen relative to a standard (δ 2H) are a potentially powerful complement to other isotopes because the mass difference between protium and deuterium results in large differences in δ 2H (>100 ‰) between some sources (Doucett et al. 2007). However, the hydrogen isotope ratios of the bulk organic matter (δ 2Hom) among various aquatic primary producers including vascular plants and macro- and micro-algae has not been systematically evaluated to determine how variability in these producers might affect the use of hydrogen isotopes as food web tracers.

Hydrogen incorporated into the organic matter of aquatic photosynthetic autotrophs ultimately comes from the surrounding water. Based on published laboratory and field measurements, aquatic primary producers have δ 2Hom that is ~160–170 ‰ more negative than environmental water (Yakir and DeNiro 1990; Luo and Sternberg 1991; Luo et al. 1991; Doucett et al. 2007; Solomon et al. 2011). The very negative values of δ 2Hom are due to fractionation during photosynthesis. Plants that photosynthesize in air (terrestrial plants) experience the same theoretical fractionation during the photolysis of H2O, but a second process, transpiration is also important. Thus, the deuterium of terrestrial vegetation is enriched relative to aquatic primary producers because of differential retention of deuterium relative to protium in plant water (Roden and Ehleringer 1999; Roden et al. 2000; Barbour et al. 2004; Cuntz et al. 2007). The consistent difference between δ 2Hom of algae and terrestrial plants provides a method to identify external organic matter support of consumers in aquatic food webs (e.g., Doucett et al. 2007; Solomon et al. 2011). However, other biophysical and biochemical processes also affect the relative abundance of hydrogen isotopes in organic matter including the effects of diffusion (Roden and Ehleringer 1999; Yakir et al. 1989; Flanagan et al. 1991; Shu et al. 2008), processes associated with heterotrophic carbon metabolism (Yakir 1992; Sessions 2006; Luo and Sternberg 1992), and variations in biochemical synthesis that lead to organismal differences in molecules like lipids (Yakir 1992; Sternberg et al. 1986; Yakir and DeNiro 1990; Sessions et al. 1999). These processes of diffusion, metabolism, synthesis, and storage may lead to relative enrichment or depletion in deuterium as represented diagrammatically in Fig. 1. Thus, the δ 2Hom of producers may not be a simple function of the net effects of the best-understood processes transpiration and photosynthesis. The magnitude of the non-photosynthetic effects that influence producer hydrogen isotope composition indicated in Fig. 1 is difficult to characterize for the bulk organic matter of primary producers. A recent review by Sachse et al. (2012) evaluates hydrogen fractionation particularly in the context of using lipid biomarkers and discusses current understanding of processes affecting fractionation.

Fig. 1
figure 1

Conceptual diagram of isotope discrimination and fractionation processes affecting ΔH values between water and organic matter of aquatic primary producers. Upward arrows represent enrichment processes and downward arrows represent depletion processes. Graded shading represents uncertainty in magnitude. For example, transpiration is an enrichment process of known magnitude and heterotrophic carbon metabolism is an enrichment process of variable magnitude. See text for references that provide detail on the various processes

Full size image

Surface water δ 2H (δ 2Hwater) varies geographically due to atmospheric partitioning in vapor and precipitation as influenced by temperature, altitude, latitude, humidity, and also because of processes like evaporation of surface water and groundwater inputs (Kendall and Coplen 2001; Bowen et al. 2005). The δ 2Hom value of a primary producer is a function of both climate and hydrological conditions that influence the hydrogen isotope ratio of water. The net difference between the ratios of deuterium to protium in producer tissue and the surrounding water is the net isotope discrimination value (ΔH), which is calculated as the difference between the hydrogen stable isotope ratios of producer organic matter and water.

$$\varDelta_{\text{H}} = \left( {\delta^{ 2} {\text{H}}_{\text{om}} {-}\delta^{ 2} {\text{H}}_{\text{water}} } \right) \div \left( { 1+ \delta^{ 2} {\text{H}}_{\text{water}} \div 1000} \right)$$
(1)

Note that we refer to ΔH as an isotopic discrimination factor, not as fractionation. Fractionation is the term for differential isotope changes in discrete mass balance chemical reactions (e.g., photosynthesis) or physical processes (e.g., evaporation) (Fry 2006; Emerson and Hedges 2008). In the case of bulk organic matter, which is our interest in the context of food web studies, the net isotopic difference reflects a number of possible processes (Fig. 1) rather than a simple set of fractionations.

Analysis of ecological process with hydrogen isotope ratios is not new (e.g., Macko et al. 1983), but recently a number of studies have used hydrogen isotope ratios to distinguish organic matter sources supporting aquatic food webs (e.g., Doucett et al. 2007; Jardine et al. 2009; Finlay et al. 2010; Caraco et al. 2010; Solomon et al. 2011; Cole et al. 2011; Babler et al. 2011; Karlsson et al. 2012; Batt et al. 2012; Cole and Solomon 2012; Wilkinson et al. 2013). Prior studies provide insight into the processes affecting isotopic composition of specific plant compounds (Cuntz et al. 2007; Hou et al. 2008). However, less is known about the processes affecting hydrogen isotope ratios of the bulk organic matter in the wide variety of aquatic primary producers that are studied in the context of food web analysis.

The variability in hydrogen isotopic discrimination (ΔH) among primary producers is fundamental to assessing the application of this isotope in ecological studies. Understanding of patterns should improve the interpretation of models and inform sampling and experimental designs used to answer ecological questions. For example, reviews of the variability in isotopic ratios associated with trophic transfers of carbon and nitrogen (Vander Zanden and Rasmussen 2001; Post 2002; McCutchan et al. 2003; Vanderklift and Ponsard 2003; Bunn et al. 2013) provide average and group-specific isotopic discrimination values used in mixing models in hundreds of studies. Since δ 2Hom displays a range of values even for primary producers growing in the same location (e.g., DeNiro and Epstein 1981), the net effects of isotopic discrimination may vary based on both biological and environmental factors. Here, we examine patterns in δ 2Hom for different types of aquatic autotrophs relative to water and in comparison with terrestrial material. We summarized data from prior studies on aquatic primary producers. We also measured ΔH for five categories of primary producers collected from three different aquatic ecosystems (lake, river, coastal) and their associated watersheds. We expected to find large differences in ΔH between aquatic and terrestrial primary producers but also expected that the magnitude of difference might vary among types of ecosystems and aquatic primary producers.

Materials and methods

Isotopic data collection

Water, plant, and algal samples were collected in a coastal lagoon, a river, and several lakes. Terrestrial vegetation was collected from the watersheds of these systems. We measured δ 2H of water and autotroph organic matter and categorized primary producers into five groups: terrestrial vegetation, macrophytes, macroalgae, benthic microalgae, and phytoplankton. We considered plants that do not live in water for any part of the tidal cycle as terrestrial vegetation (TV). Vascular aquatic plants were categorized as macrophtyes (MP), and algae were categorized as either macroalgae (multicellular: MA) or microalgae (unicellular). Microalgae were further classified as either phytoplankton (PHY) or benthic microalgae (BMA) according to habitat and collection method. The coastal samples were from the Virginia Coast Reserve Long Term Ecological Research site (VCR–LTER), which comprises marshes, mudflats, lagoons and barrier islands off the Eastern Shore of Virginia. The river samples were from the Hudson River between river kilometers 45 and 175 where the southern tip of Manhattan Island, New York is river kilometer 0. The lake samples were from the University of Notre Dame Environmental Research Center (UNDERC) located near Land O’Lakes, Wisconsin.

We collected larger plants from both aquatic and terrestrial environments and sampled microalgae from benthic and pelagic environments. Multicellular seaweeds such as Gracilaria vermiculaphylla and Ulva lactuca were classified as macroalgae and comprised eight species, most of which were found in the coastal marine system. These macroalgae were collected from sub-tidal habitats and not exposed to air during any part of the tidal cycle. Chara sp., found in the lake system, is a green algae (Chlorophyta) but is also considered a developmental step between macroalgae and embryophytes. There was no morphological equivalent to macroalgae for the Hudson River system. Vascular aquatic plants were categorized as macrophytes and included 13 species ranging from seagrass (Zostera marina) to pondweeds (e.g. Potamogeton pusillus) and water lilies (e.g. Nuphar variegata). The growth patterns of macrophytes influence their exposure to air, so macrophytes were further classified as emergent, floating, or submerged. Emergent plants had variable exposure to air based on plant height and tides. Floating plants had leaves on the surface of water permanently exposed to air, while submerged plants grew underwater and were not exposed to air. In the river and lake systems we collected benthic microalgae as scrapings from tiles, natural rock, or wood substrate. High levels of sediment re-suspension made this method infeasible in the coastal system, so we used a modified version of the vertical migration technique (Riera and Richard 1996) to collect phototactic benthic diatoms. Phytoplankton were sampled from incubated laboratory cultures of native planktonic assemblages grown in filtered site water as per Caraco et al. (2010) as well as from algal net tows picked clean of non-algal material. Terrestrial vegetation sampled from the lake watershed comprised broadleaf deciduous and evergreen tree species, as well as moss (Sphagnum spp.) and a shrub (Chamaedaphne calyculata). Below, we refer only to terrestrial vegetation and macrophytes as plants, and we use primary producers to describe all autotrophs including plants and algae.

Field samples were rinsed thoroughly to remove any salts and epiphytic growth. Samples were dried at 60 °C for at least 24 h prior to grinding. Subsamples of about 350 μg were weighed into silver cups for isotopic analysis. All isotopic analyses were performed at the Colorado Plateau Stable Isotope Laboratory. All samples of organic materials were pyrolized to H2 and the isotope ratio was measured on the H2 gas (Doucett et al. 2007). Organic matter samples for δ 2H were analyzed with a Thermo-Finnigan TC/EA and DeltaPLUS-XL (Thermo Electron Corporation, Bremen, Germany). Water samples were analyzed for δ 2H with cavity ring down laser spectroscopy using Los Gatos Research Off-Axis Integrated Cavity Output coupled to a CTC LC-PAL liquid autosampler. All values are reported in per mil notation (‰) and are in relation to the international standard of Vienna Standard Mean Oceanic Water (δ 2HVSMOW).

Several approaches were employed to assess the precision, accuracy, and reproducibility of the organic matter hydrogen ratio analyses. In this context, it is important to note that hydrogen in organic matter samples includes both an exchangeable and a non-exchangeable component. The analytical method measures variation in the non-exchangeable component after normalizing for the exchangeable component. A bench-top equilibration following Wassenaar and Hobson (2003) was used in all analyses wherein sample material was allowed to exchange with ambient water vapor prior to sample analysis. Variability among materials reflects differences in the non-exchangeable component as long as the proportion of exchangeable to non-exchangeable material is relatively similar among samples and standards. Relative to the primary producer material analyzed in this study, the most significant standard used in the analysis was a filamentous algae ––Cladophora sp. from the Colorado River. This material was representative of the aquatic primary producers and has a δ 2H more negative than −200 ‰. Further, the exchangeable to non-exchangeable ratio of Cladophora is likely similar at least to the other algae used in this study and possibly similar to the higher plants. The Cladophora standard provided good reproducibility in day to day sample runs and a highly linear relationship between expected and observed values (M. Caron, personal communication).

Water samples were collected at approximately the same location and time as the organic matter samples for an accurate representation of the surrounding hydrogen pool. For samples with no corresponding water value in the Hudson River data, δ 2Hwater was estimated from the location of the sampling site based on an empirical relationship consistent with the rainout effect between the measured river water δ 2H values and the distance of the sampling site from the mouth of the Hudson River (R 2 = 0.94).

We also conducted a literature search for published values of δ 2H from the total organic matter of aquatic plants and algae by using the keywords: hydrogen, deuterium, delta D, and H-2 in the Web of Science. We calculated hydrogen isotope discrimination (ΔH) from these data when environmental water isotope ratios were also reported. Note we excluded published data from Solomon et al. (2011); Cole et al. (2011); and Batt et al. (2012) because these were part of our field survey data.

Uncertainties in isotopic discrimination

Isotopic discrimination (ΔH) is calculated from Eq. 1 above and is the sum of all possible processes:

$$\varDelta_{\text{H}} = \sum \left( {\varepsilon_{\text{lw}} + \varepsilon_{\text{bio}} + \varDelta_{\text{x}} } \right)$$
(3)

where εlw is biophysical discrimination in leaf water due to uptake, transport, and transpiration; εbio is biochemical fractionation due to photosynthesis, biosynthesis of lipids, and heterotrophic carbohydrate metabolism; and Δx represents all other possible processes that influence the difference in hydrogen isotopic ratios (Fig. 1). Separate measurements of these discrimination and fractionation processes were beyond the scope of this study, but their implications are discussed below. Since organic matter samples were always more depleted in deuterium than water (δ 2Hom < δ 2Hwater < 0), ΔH values are negative following Eq. 1 (i.e. there is a negative isotope discrimination relative to source water). Relative to a constant δ2Hwater, a highly negative ΔH indicates a greater depletion in deuterium relative to protium and more negative δ 2Hom; a less negative ΔH value indicates a less negative δ 2Hom.

Exchangeable hydrogen could potentially affect estimates of ΔH. In the bench-top equilibration conducted prior to sample analysis, the water molecules that exchange with organic hydrogen are more enriched in deuterium and thus shift the measured δ 2Hom. Direct assessments of hydrogen exchange suggest exchangeable H is on the order of 10 % for keratin, hair, and meat (Chesson et al. 2009). The key issues for our analyses are: (1) how far off are the calculated ΔH values given hydrogen exchange, and (2) how comparable are different measures of ΔH given potentially different levels of exchange among different producer bulk organic matter samples? To explore the sensitivity of ΔH to this problem, we first assumed values of δ 2H from our lake study site: −220 ‰ for algal material, −140 ‰ for vascular plant material, and −45 ‰ for water. These values resemble pre-exchange measurements if H-exchange is about 10 % (Solomon et al. 2011; Cole et al. 2011). Thus, ΔH values are −183 ‰ and −99 ‰ for the algal and vascular plant material, respectively. When the organic materials equilibrate at the analysis laboratory (δ2Hwater at CPSIL ~−90 ‰), the measured values are more positive depending on H-exchange. Assuming H-exchange in the range of 5–20 %, ΔH values would be −174 ‰ at 5 % and −147 ‰ at 20 % for algae, and −99 ‰ at 5 % and −90 ‰ at 20 % for higher plants. In this example, the worse scenario for comparison would be algae with 20 % H-exchange having a ΔH of −147 ‰ relative to terrestrial material with a 5 % H-exchange and ΔH of −90 ‰. This difference in ΔH is still large and easy to distinguish in data. Detecting differences in ΔH is more problematic when ΔH values are closer, especially if materials have different H-exchange. For the comparisons we make below, the effect of H-exchange on the ΔH values is likely small (<10 %) with the possible consequence that differential H-exchange among data categories obscure detections of some differences in statistical tests (see next section).

Statistical tests

In order to evaluate if hydrogen isotope ratios differ among primary producers, non-parametric Kruskall–Wallis tests were performed to evaluate differences in ΔH among categories of producers (test statistic reported as K). Non-parametric tests were used because our data did not meet normality or homogeneity of variances assumptions based on Kolmogorov–Smirnov tests. We used ΔH values instead of hydrogen isotope ratios (δ 2Hom) to eliminate variability associated with δ 2Hwater among the different sites. Differences were evaluated with post hoc exacted Wilcoxon Mann–Whitney rank sum tests with Bonferroni corrections (Sokal and Rohlf 2012). For the Kruskall–Wallis and Mann–Whitney tests we required that a test statistic have a probability value <0.05 to assign significance. It is possible that some significant effects were not detected (type 2 error) due to H-exchange that obscured true differences in ΔH values (see prior section). This problem is most likely where ΔH values are relatively similar for the primary producers being compared. In cases where significant differences were observed, the differences encompass any variability associated with H-exchange and are robust. Statistical tests were performed in R version 2.15.1 (R Development Core Team 2012, URL http://www.R-project.org).

Results

Aquatic primary producer hydrogen discrimination: literature survey

Variability in reported δ 2H of aquatic primary producers is substantial within and among locations (Table 1). Plotting δ 2Hwater versus δ2Hom suggests a general relationship between these two variables (Fig. 2), but not the relatively constant isotopic discrimination that would be expected if δ 2Hom were simply the result of photosynthetic fractionation. Further, the literature data suggest possible differences among major groups of aquatic producers (Fig. 2). Macrophyte ΔH values appear smaller than the other groups, but we could only find data for four species. Benthic algae had larger values of ΔH relative to macrophytes (Table 1). Macroalgal δ2Hom, for which there was the greatest number of measurements, varied substantially (Fig. 2), with ΔH values ranged from −213 to −38 ‰ (Table 1). We did not test the literature data for differences among groups because of the heterogeneity of methods used by researchers.

Table 1 Hydrogen isotope discrimination (ΔΗ, ‰) calculated from literature data for aquatic primary producers where both producer organic matter δ 2Hom and environmental water isotope δ 2Hwater were reported (units ‰)
Full size table
Fig. 2
figure 2

δ 2Hwater and δ 2Hom of producers based on data in the literature. Symbols represent the four categories of producers: macroalgae (closed circles), macrophytes (open circles), and benthic microalgae (closed triangles)

Full size image

Aquatic primary producer hydrogen discrimination: field survey

All of the plant and algal samples in our field survey were depleted in deuterium relative to water, with δ 2Hom values ranging from 38 to 220 ‰ more negative than corresponding δ 2Hwater (Fig. 3). As with the literature data, δ 2Hwater values were somewhat related to δ 2Hom, but there was a wide variation, even among producers at the same site. For example, in the coastal system, δ 2Hwater was consistent throughout all seasons (−9.99 ‰ ± 0.9), but δ 2Hom varied from −230 to −85 ‰ among different marine producers (Fig. 3). In comparison with the coastal system, δ 2Hwater for the freshwater systems varied more in space and time. In the Hudson River, δ 2Hwater ranged from −43 to −71 ‰ among sampling locations, which was partly a function of oligohaline conditions in the most downstream sites. Samples from 18 different lakes had δ 2Hwater values that varied between −40 and −75 ‰. Temporal variation in the lakes was low based on a few of the lakes we previously sampled over the May to September period (Solomon et al. 2011; Batt et al. 2012). For example, in Peter Lake the mean (±standard deviation) of δ 2Hwater was −43.9 ‰ ± 2.4.

Fig. 3
figure 3

δ 2Hwater and δ 2Hom of producers based on data collected in this study. Symbols represent the four categories of producers: macroalgae (closed circles), macrophytes (open circles), benthic microalgae (closed triangles), and phytoplankton (open squares)

Full size image

We calculated ΔH for the 253 producer samples from our survey and classified these into five categories based on taxa, morphology, and sampling method (Table 2). The five categories were terrestrial vegetation, macrophytes, macroalgae, benthic microalgae, and phytoplankton (Table 2). Primary producers with the lowest ΔH values have the most negative δ 2Hom values relative to water. Phytoplankton and benthic algae in the lake and river system had the most negative ΔH values while terrestrial vegetation and aquatic macrophytes had the least negative values (Table 2). The category of primary producers with the most variability was macroalgae (s.d. of all observations: 55 ‰). Most species of macroalgae had low variation within ranges comparable to other groups (s.d. ~15 ‰), but as with the literature data, there was a wide variation among species within this group (Table 2). For example, ΔH values varied from −96 ‰ for Codium fragile to −201 ‰ for Ulva lactuca. Freshwater filamentous green algae, which we classified as macroalgae, also had a low (highly negative) mean ΔH that was similar to Ulva lactuca and Enteromorpha sp. (Table 2). Vascular plants including both terrestrial and aquatic macrophytes were relatively enriched in deuterium compared to algae and had less negative ΔH values. Even fully submerged species of macrophytes such as the rooted plants Zostera marina and Valisneria americana were enriched in deuterium compared to algae in the same systems.

Table 2 Mean hydrogen isotope discrimination values (ΔH, ‰) between organic matter and water grouped by primary producer types considered in this study
Full size table

Comparison of plant groups

We first contrasted the differences in ΔΗ among groups of primary producers when data for all three systems (i.e. lake, river, coastal) were combined (Fig. 4a). There were significant differences in ΔH among the five types of primary producers (Kruskal–Wallis test: K = 122, p < 0.001; n = 253). Post hoc tests revealed significant differences between phytoplankton and benthic algae (p < 0.001). Phytoplankton had the lowest average values (ΔH = −173 ‰ ± 26). Macroalgae were not significantly different from benthic algae, but were significantly different from both macrophytes (p < 0.0001) and terrestrial vegetation (p < 0.0001). Terrestrial vegetation and macrophytes had the least negative ΔH values (−85 ‰ ± 19 and −79 ‰ ± 16, respectively), and these groups were not significantly different. Emergent, floating, and submerged types of macrophytes were not significantly different from each other (K = 2.9, p > 0.2).

Fig. 4
figure 4

Mean ΔΗ values based on the differences between hydrogen stable isotope ratios of water (δ 2Hwater) and organic matter (δ 2Hom) for 5 categories of primary producers from three different watershed types. Panels are a data combined from all three systems, b river only, c lake only, d coastal system only. Categories are abbreviated as follows: MA macroalgae, MP macrophytes, BMA benthic microalgae, PHY phytoplankton, TV terrestrial vegetation. Error bars are standard errors; letters reflect significant differences between categories based on post hoc exacted Wilcoxon Mann–Whitney rank sum tests with Bonferroni corrections. Categorization of primary producers is discussed in the text

Full size image

When each system was considered independently, ΔH values of the groups displayed similar patterns (Fig. 4b–d). In the Hudson River, terrestrial vegetation and macrophytes had the least negative values of ΔH and were not significantly different (Fig. 4b). Phytoplankton ΔH values were most negative (−187 ‰ ± 30) and significantly different than both groups of higher plants (p < 0.001). Benthic algae collected from the river had ΔH values (−195 ‰ ± 9) similar to phytoplankton.

In lakes, both groups of higher plants (terrestrial and aquatic macrophytes) had less negative values of ΔΗ than either type of microalgae (phytoplankton and benthic microalgae). However, macroalgae were not significantly different from any of the other groups (Fig. 4c). Macroalgae in the lakes included both Chara sp. and filamentous epilithic green algae. Chara was highly enriched in deuterium resulting in a much less negative ΔH relative to filamentous green algae (Chara ΔH = −50 ‰ ± 14; filamentous algae ΔH = −216 ‰ ± 32).

Values of ΔH for primary producers in the coastal system were also significantly different among groups (K = 49, p < 0.001). The aquatic primary producer groups were significantly different from each other but all overlapped with terrestrial vegetation (Fig. 4d). Pairwise comparisons between genera revealed significant differences between several types of macroalgae. For example, Ulva lactuca had a much more negative ΔH value than all the other macroalgae (p < 0.05, note Enteromorpha sp. was excluded because n = 1 for this taxa). Within the macrophtye group, the submerged seagrass, Zostera marina, was not significantly different from the emergent marsh grass Spartina alterniflora.

The ordering of ΔH between groups was relatively consistent across all three systems with algae having the most negative values and terrestrial vegetation and macrophytes having the least negative values (Fig. 4). However, when individual groups were tested for differences among systems (i.e., differences of a group among the lake, river, and coastal sites), all groups differed significantly except for macroalgae. For example, macrophytes exhibited differences between sites (K = 12, p < 0.01). The coastal macrophytes (Spartina alternaflora and Zostera marina) had significantly less negative ΔH values than the river or lakes based on post hoc comparisons (p < 0.05). Values of ΔΗ for benthic algae were significantly different among the three sites (K = 17, p < 0.001). The most negative ΔH values were in the river and the least negative were observed in the coastal system. Benthic algal ΔH values in the river were significantly different from those in both the lake (p < 0.01) and coastal (p < 0.01) system. Phytoplankton ΔH values were also different among systems (K = 7, p < 0.01) with the river having a more negative value than the lakes (p < 0.01).

Terrestrial vegetation ΔH values were also significantly different (K = 20, p < 0.001) as would be expected given differences in conditions (e.g., transpiration) among the sites. Plants from the coastal system had the most negative ΔH values while plants from the river system had the least negative values. Plants sampled in the lake watersheds had ΔΗ values that were significantly different from those in the coastal (p < 0.05) or river (p < 0.001) system.

Discussion

We found a consistent pattern of hydrogen isotope discrimination for groups of primary producers in three types of aquatic systems, supporting the use of hydrogen isotope ratios to distinguish energy sources in aquatic food webs. Our results are consistent with other observations (Epstein et al. 1976; Doucett et al. 2007; Finlay et al. 2010; Caraco et al. 2010) that algae are strongly depleted in deuterium compared to terrestrial plants. However, some of the other variability we observed cannot yet be explained with current understanding of stable hydrogen isotopes. Our results indicate that ΔH can be used to differentiate among some, but not all groups, of aquatic primary producers. The organic matter of primary producers is not simply a function of δ 2Hwater and a fixed ΔH but tends to vary among sites and producer types. In freshwater systems, terrestrial vegetation can be distinguished from algal primary producers but not from vascular aquatic plants. Application of hydrogen isotopes may, therefore, be complicated in lakes or rivers where macrophytes are a significant part of the food web.

Variability of δ2Hwater

Climate and hydrologic conditions influence the isotopic composition of meteoric, surface, and ground- water (Schiegl and Vogel 1970; Kendall and Coplen 2001; Bowen et al. 2005); therefore hydrogen isotope values vary across systems. Water in the lake and river systems we considered was predictably more depleted in deuterium than the coastal marine system (Kendall and Coplen 2001). Intense mixing and exchange with the ocean as well as low freshwater input maintains more constant values in the coastal lagoons we sampled, for which the δ 2Hwater was near mean ocean water values of 0 ‰. Variability in δ 2Hwater among lakes and along the Hudson River (upriver to downriver) is consistent with seasonal, continental, precipitation, and evaporation effects that influence these freshwater systems (Kendall and Coplen 2001). The range of these differences (lake to lake, upriver to downriver) for our study sites was on the order of 30 ‰. For lakes, evaporation will vary with surface area to volume ratios, thus differences in lake morphometry are likely related to differences in δ 2Hwater. Also, the relative inputs of groundwater versus surface water to lakes may differentially influence their δ 2Hwater (Krabbenhoft and Webster 1995). The gradient of δ 2Hwater observed in the Hudson River is likely related to inputs of freshwater sources, evaporation, mixing within the tidal freshwater estuary, and mixing of oligohaline water at the most downstream end of the system.

Variability in δ 2Hwater among systems or within a system influences the δ 2H of plant organic matter (Figs. 2, 3). This water-driven variability in δ 2Hom can transfer up the food web and complicate interpretation of mixing model results for organisms with long tissue turnover times. Comparisons of hydrogen isotopes ratios from different systems should account for isotopic differences in the δ 2H of water, especially between water bodies that vary in size, salinity, latitude, elevation, and precipitation. Since environmental parameters can affect δ 2H of organic matter by changing δ 2Hwater, environmental variation would most strongly affect δ 2H of organic matter in primary producers with high turnover rates such as algae. However, temporal variability in δ 2Hwater was low in the lake, river, and lagoon systems we studied.

In the context of food web analyses, published hydrogen ratios should only be used from a separate location or study if differences in initial water hydrogen isotope ratios are taken into account, especially for freshwater ecosystems. The development of a global database of water isotope values (GNIP, administered by the International Atomic Energy Association and the World Meteorological Organization, http://www-naweb.iaea.org/napc/ih/IHS_resources_gnip.html) is making hydrogen isotope ratio data more available. As the geographic coverage of this database grows, it should be possible to pair δ 2Hwater values with primary producer measurements from similar sites to constrain estimates of hydrogen isotope discrimination (ΔH).

Biophysical influences on ΔH

We observed smaller ΔH values in terrestrial plants relative to those of most aquatic primary producers, consistent with transpiration effects (e.g. differential evaporation of protium relative to deuterium in leaf water). However, aquatic macrophytes including completely submerged species had ΔH values similar to terrestrial plants. This was an unexpected result. These data suggest that deuterium enrichment in leaf water of terrestrial plants due to transpiration is not the only explanation for differences between δ 2H in aquatic and terrestrial plants since functional stomata are not a sole predictor of deuterium enrichment. While some emergent and floating-leafed macrophytes do transpire, stomata on most submerged aquatics are considered non-functional because of wax occlusions (Sculthorpe 1967), which prevent transpiration even if exposed to air. The observed similarities in ΔΗ for macrophytes and terrestrial vegetation in our data may reflect different enrichment and depletion processes for aquatic versus terrestrial plants. In aquatic plants for which there is no transpiration, deuterium enrichment in leaf water (especially for submerged plants) should be negligible and plant tissue hydrogen stable isotopes might be expected to simply reflect the fractionation associated with photosynthesis. For example, Yakir and DeNiro (1990) found this expected photosynthetic fractionation values in Lemna plants of about −170 ‰ relative to water. However, they also measured post-photosynthetic fractionation (about +150 ‰) related to metabolism (Yakir and DeNiro 1990). Hence, for Lemna the δ2Hom was not simply the result of photosynthesis but reflected the consequences of other processes. For our data, the similarity of ΔH in comparing macrophytes with terrestrial plants is unresolved and requires further research.

In addition to transpiration, other biophysical influences on leaf water are unlikely to explain the enrichment we found in aquatic macrophytes. Aquatic macrophtyes (Rascio 2002) and unicellular green algae (Yakir 1992) both have high levels of exchange between photosynthetic cells and environmental water. A direct pathway between water uptake and photosynthesis minimizes leaf/cell water heterogeneity. The differences we observed in ΔΗ values between the three systems may be related to differences in environmental conditions like relative humidity or physiological differences between the autotrophs found at each location.

Biochemical influences on ΔH

We found ΔΗ values in phytoplankton, some macroalgae, and some benthic algae in the range of −160 to −170 ‰ that were approximately consistent with a large depletion in deuterium due to photosynthesis. The δ 2Hom of these primary producers relative to water suggests that most of the organic matter is produced with little depletion or enrichment of deuterium beyond what occurs as a consequence of photosynthesis. The most deuterium-depleted primary producers we observed were some microalgae and filamentous macroalgae. In a few cases, these ΔH values were even more negative than the range of −160 to −170 ‰ expected from photosynthesis. Since lipid biosynthesis strongly fractionates against deuterium (e.g., Sessions et al. 1999), ΔH < −170 ‰ may be an indication of species where lipids are a substantial component of producer biomass. However, the highest lipid contents we found reported in the literature for the most deuterium-depleted algae (relative to water) in our data were only 6.4 and 9 % for species of Ulva and Enteromorpha, respectively (Wahbeh 1997). If lipids in these macroalgae are 100 ‰ depleted relative to the value for the whole alga (Smith and Epstein 1970), a simple mixing model where 90 % of the mass is −170 ‰ and 10 % of the mass is −270 ‰ for the lipid fraction results in a final tissue δ 2Hom of −180 ‰. Lipid content would need to be near 30 % for these algae to achieve a tissue isotopic value similar to the values we measured for Ulva and Enteromorpha. Thus, these macroalgae as well as some filamentous algae and cyanobacteria are more depleted in deuterium than expected based on net isotopic discrimination due to photosynthesis and reasonable assumptions about lipid content (Fig. 2). This conclusion, however, must be tempered by uncertainty about expected fractionations related to photosynthesis and lipid synthesis as well as the δ 2H of non-lipid organic matter in these and other algae.

Variability in ΔH for benthic microalgae across our systems could be related to differences in community composition of the organisms we sampled. Although microscopic inspection suggested algae dominated our samples, other chemosynthetic and heterotrophic microbes were present. These organisms could affect measured values of δ 2H especially in cases where microbes (e.g. methane oxidizers) rely on sources of hydrogen that are highly depleted in deuterium (Deines et al. 2009).

Terrestrial succulent plants use Crassulacean Acid Metabolism (CAM) as an adaption to water stress—a condition not expected in the aquatic environment. However, observations of substantial diel hydrogen ion changes (5–290 mmol H+ kg−1) in aquatic species indicate the operation of an acid metabolism (Keeley 1998). Separation of carbon uptake and reduction during “aquatic” acid metabolism (Rascio 2002) may provide a competitive advantage for carbon acquisition in the aquatic environment where diffusion of carbon is many times lower than in air. These aquatic producers use CO2 as a carbon source. In the absence of aquatic acid metabolism, most aquatic primary producers use bicarbonate. Bicarbonate uptake in aquatic plants and algae is widely observed including in many of the genera included in this study (e.g., Potamogeton: Sand-Jensen et al. 1992). However, we also measured the isotopic composition of genera for which there is evidence of CO2 use and acid metabolism (e.g., Pontedaria: Pagano and Titus 2007). These macrophytes and macroalgae did not have an H-isotopic composition distinct from those that use HCO3 −.

The high variability we found for ΔH in macroalgae may be a promising area to investigate for a more complete understanding of drivers of aquatic producer δ 2H. Chara, a submerged macroalgae that grows unattached in open water, was the most deuterium-enriched macroalgae in the lake system. Chara has traits that make it intermediate between macroalgae and embryophytes, and is considered the closest algal relative of higher plants. Other branched and highly structured macroalgae like the coastal seaweeds Agardhiella subulata and Codium fragile were also enriched in deuterium relative to more simply structured Ulva lactuca, which grows as sheets only two cells thick. These differences in structure may be related to differences in composition that influence δ 2H.

Conclusions

We found large and consistent differences in hydrogen isotope discrimination among major groups of aquatic primary producers across freshwater and marine ecosystems. Relative to water deuterium was depleted in microalgae, variable in macroalgae, and relatively enriched in both terrestrial vegetation and aquatic macrophytes. These differences can aid in partitioning energy sources in many aquatic ecosystems, particularly in distinguishing algal and terrestrial sources. Based on our results, hydrogen isotopes are less useful in distinguishing sources in aquatic food webs where both macrophytes and terrestrial vegetation are important, but δ 13C and δ 15N are often distinct for these sources (e.g., Batt et al. 2012). Food web studies using hydrogen isotopes should either measure the δ 2H of primary producers directly, or where this is infeasible (e.g. phytoplankton), account for the variability in isotope discrimination relative to water thereby providing a means to estimate δ 2H of producer biomass. Finally, while ΔΗ was relatively constant within most producers groups for a given system, the causes of the large variability of ΔH in macroalgae require further investigation.