Portal de Periódicos da CAPES

Deuterium as a food source tracer: Sensitivity to environmental water, lipid content, and hydrogen exchange

2015; Wiley; Volume: 13; Issue: 5 Linguagem: Inglês

10.1002/lom3.10019

1541-5856

Grace M. Wilkinson, Jonathan J. Cole, Michael L. Pace,

Geology and Paleoclimatology Research

Limnology and Oceanography: MethodsVolume 13, Issue 5 p. 213-223 Evaluations of Existing MethodsFree Access Deuterium as a food source tracer: Sensitivity to environmental water, lipid content, and hydrogen exchange Grace M. Wilkinson, Corresponding Author Grace M. Wilkinson Department of Environmental Sciences, University of Virginia, Charlottesville, VirginiaCorrespondence: [email protected]Search for more papers by this authorJonathan J. Cole, Jonathan J. Cole Cary Institute of Ecosystem Studies, Millbrook, New YorkSearch for more papers by this authorMichael L. Pace, Michael L. Pace Department of Environmental Sciences, University of Virginia, Charlottesville, VirginiaSearch for more papers by this author Grace M. Wilkinson, Corresponding Author Grace M. Wilkinson Department of Environmental Sciences, University of Virginia, Charlottesville, VirginiaCorrespondence: [email protected]Search for more papers by this authorJonathan J. Cole, Jonathan J. Cole Cary Institute of Ecosystem Studies, Millbrook, New YorkSearch for more papers by this authorMichael L. Pace, Michael L. Pace Department of Environmental Sciences, University of Virginia, Charlottesville, VirginiaSearch for more papers by this author First published: 11 May 2015 https://doi.org/10.1002/lom3.10019Citations: 25AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Abstract Hydrogen stable isotopes (δ2H) are used for quantifying resources supporting food webs. However, application of δ2H in mixing models requires; (1) correction for environmental water (ω) in consumer tissues, (2) consideration of differential fractionation among biochemical constituents, and (3) consideration of differential H-exchange among samples and standards. We present data and sensitivity analyses addressing each of these issues and provide recommendations for future isotope food web studies. First, we determined from field data that maximum ω for aquatic consumers averaged 0.23 ± 0.03, similar to the median ω from a survey of published values (0.22 ± 0.02). Resource use estimates based solely on δ2H data were sensitive to the selected ω value. Second, to quantify the potential bias in bulk tissue analysis from differential tissue fractionation, we calculated the change in whole organism δ2H before and after lipid extraction for 61 aquatic samples. The average change in consumers' δ2H after lipid extraction was a positive shift of 11.8‰ relative to the pre-extraction value. This shift resulted in a minor change in resource use estimates when correcting for lipids. Finally, we evaluated the impact of correcting for H-exchange in samples using standards with dissimilar H-exchange portions. The impact of the correction factor for H-exchange on resource use estimates could be large if suitable standards are not used for comparison. From these analyses we conclude that despite these complicating factors, analysis of resource use is possible using whole organisms' δ2H, especially in combination with cautionary sensitivity analysis. Recently, hydrogen isotope ratios (δ2H) have been used to quantify trophic resources in marine and freshwater consumers (e.g., Bortolotti et al. 2014; Berggren et al. 2014; Hondula et al. 2014). One of the key applications of hydrogen (H) isotopes in aquatic ecosystems has been to assess the degree to which aquatic consumers use organic matter from the surrounding watershed, known as allochthony (e.g., Cole et al. 2011; Karlsson et al. 2012; Kelly et al. 2014). There is a large separation in δ2H values between allochthonous and autochthonous organic matter (material originating within the system) as well as among some aquatic primary producers such as different species of macroalgae and macrophytes (Hondula et al. 2014). This large separation in end members (organic matter sources) allows for more robust estimates of consumer resource use. However, applying H isotopes to estimate consumer resources requires: (1) correcting for environmental water δ2H in consumer tissues, (2) evaluating the influence of differential fractionation among biochemical constituents used for sources and end-members due largely to varying lipid content, and (3) properly accounting for H-exchange during laboratory analysis. Environmental water, sometimes referred to as "dietary water," is the H in a consumer's tissue that is derived directly from water in the surrounding aquatic environment and not from food resources (Hobson et al. 1999; Solomon et al. 2009). Incorporation of environmental water into consumer tissues is useful for applications such as reconstructing animal and human migrations using δ2H precipitation gradients (Hobson et al. 2004; O'Grady et al. 2012). For studies of aquatic consumer resource use dietary water is problematic. The environmental water correction (ω), formalized by Solomon et al. (2009), varies by consumer and increases with increasing trophic level (Birchall et al. 2005). The influence of environmental water on consumer δ2H is calculated as a mixing model (1)where ωtot is the trophically compounded value of ω (ω = ωtot for primary consumers), the environmental water correction parameter (bounded between 0 and 1), and δ2HConsumer, δ2H2O, and δ2HResources are the H isotope ratio values of the consumer, water, and end members, respectively (Solomon et al. 2009). Surface water δ2H2O is considerably less negative than organic matter resources (Stiller and Nissenbaum 1980; Kendall and Coplen 2001), particularly in the case of aquatic primary producers (Hondula et al. 2014; Yang et al. 2014). As such, not correcting for the influence of environmental water on consumer tissue δ2H underestimates consumer use of autochthonous sources. Conversely, over correcting for the influence of environmental water can cause corrected consumer δ2H to become so negative that the value falls outside the bounds of the end members. Biosynthesis of different tissue types (e.g., lipids) discriminates against the incorporation of deuterium, the heavier H isotope (Soto et al. 2013). Thus, the bulk value of δ2H can be influenced by the tissue composition of the consumer in ways that are unrelated to food sources. In particular, lipids are highly depleted in 2H making them substantially more negative (in del units) than other tissue constituents such as protein (Smith and Epstein 1970; Sessions et al. 1999). This fractionation in lipid synthesis also occurs for carbon isotopes (DeNiro and Epstein 1977; Bodin et al. 2007). In order to correct for this potential bias, lipids can be extracted from the samples before analysis (Folch et al. 1957; Arrington et al. 2006), or in the case of carbon isotopes, a correction can be applied to the whole organism (hereafter bulk) isotope values based on the carbon to nitrogen (C : N) ratio of the sample (Post et al. 2007), a proxy for lipid content. Few studies have evaluated the influence of lipids on bulk consumer δ2H (Jardine et al. 2009; Soto et al. 2013). Those few that have indicate that lipids can influence bulk tissue δ2H values for aquatic consumers, however, the magnitude of the influence of lipids on bulk tissue δ2H has not been thoroughly evaluated and a correction has not been formulated. The lipid bias is further complicated for bulk δ2H measurements as the H in lipids is nonexchangeable while some other constituents undergo H-exchange with ambient water vapor (Soto et al. 2013). For example, approximately 20% of the H in proteins exchanges (Chesson et al. 2009). As such, tissue composition (e.g., percent lipid, protein, carbohydrate) influences the overall bulk tissue H-exchange value for the sample and can lead to differential exchange values among sample types. This differential exchange can be accounted for by correcting measured values after laboratory analyses. A common method for accounting for H-exchange is the comparative equilibration method (CEM; Wassenaar and Hobson 2003). CEM analyzes calibrated organic standards along with similar, but unknown samples after allowing exchange of both standards and samples with ambient water vapor to occur. The unknown sample values are then corrected based on the regression between measured vs. expected values for the standard (Doucett et al. 2007; Kelly et al. 2009). The common assumption is that the sample tissue composition is similar enough to the standards in the lab that assuming a single value of H-exchange for the material (based on the standards) is sufficient. If the standard and unknown samples have different H-exchange values, this assumption may lead to erroneous corrections which would in turn impact mixing model calculations of consumer resource use. The goal of this article is to assess the impact of ω value selection, potential lipid bias, and H-exchange assumptions during analysis on estimates of aquatic consumer resource use. Recent publication of δ2H data from a variety of aquatic ecosystems and consumers provided a unique opportunity to empirically evaluate the maximum likely value of ω and the impact that ω value selection has on consumer resource use estimates. Additionally, we evaluated the effect of lipids on bulk consumer δ2H values by measuring both lipid-extracted and bulk samples for numerous aquatic consumers and end members. Finally, we performed a sensitivity analysis of the influence on mixing model estimates of resource use by applying realistic H-exchange fractions to known δ2H values and then correcting those values for nonexchangeable H based on standards. From these analyses and a review of reported ω values, we synthesized recommendations for future research using δ2H to estimate consumer resource use. Materials and Procedures Literature review of aquatic consumer ω values We searched published papers to expand on the summary of ωtot values reported in Solomon et al. (2009). Using Web of Science and the key terms hydrogen, deuterium, aquatic, and consumer, we included studies that either directly estimated the environmental water parameter through controlled laboratory experiments or inferred ωtot from field sampling and a well constrained, known diet. Only studies that measured the nonexchangeable H fraction were considered as this is currently the most common analysis that is widely available. Additional information such as the trophic position of the consumer, habitat, tissue type sampled, and the environmental water estimation method were also recorded. If trophic position was not stated explicitly, it was inferred from the information provided in the study. These variables were then used to discern if there were any patterns in ω values among consumers, trophic levels, or study types. Empirically deriving the maximum value of ωtot We also searched the literature for data on aquatic consumers, resources, and water δ2H values (δ2H2O) for multiple systems. We corrected the reported raw values of consumer δ2H for environmental water using the system specific δ2H2O and values of ωtot ranging between 0 and 0.5, by increments of 0.01. We used ωtot, the final ω value after trophic compounding was accounted for (Solomon et al. 2009), to make the analysis comparable across consumers. Consumers were characterized by the lowest taxonomic level provided in the study and were analyzed by taxonomic group. The number of corrected consumer δ2H values that fell outside the bounds of the study-specific end members after correcting for each ωtot value between 0 and 0.5 was tabulated and used to assess the empirical maximum of ωtot. The most autochthonous consumers were the most likely to be pushed outside the bounds of the mixing model, thereby minimizing any confusion between terrestrial and water isotopic values in highly allochthonous consumers. Impact of ω on estimates of consumer resource use We evaluated the impact of the environmental water parameter on estimates of consumer resource use using two different datasets. First, we examined the impact of environmental water value selection on estimates of Daphnia allochthony from a δ2H mixing model. As the δ2H of environmental water (δ2H2O) is more similar to the δ2H of allochthonous material than autochthonous material, the dietary water correction can potentially have a large impact on the estimate of consumer allochthonous resource use. Using the following two-source, δ2H Bayesian mixing model (2)described in Wilkinson et al. (2013), where δ2HT and δ2HA are the isotope end members for terrestrial and aquatic organic matter, respectively, and δ2HCons is the isotope value of the aquatic consumer. We used three different ω values (0.1, 0.2, and 0.3) in the model and estimated allochthonous resource use. The isotope data used in the model are presented in Table 1. Table 1. δ2H isotope mean values and standard deviations (s.d.) for the three sensitivity analyses. Daphnia were assumed to be primary consumers for this analysis. Parameter Notation Value (‰) s.d. (‰) Algae δ2HA −220 17.0 Terrestrial δ2HT −120 15.2 Daphnia A δ2HCons −140 0.5 Daphnia B δ2HCons −160 0.5 Daphnia C δ2HCons −180 0.5 Daphnia D δ2HCons −200 0.5 Environmental water correction (a) ω 0.22a 2.0 Lake water δ2H2O −45 1.0 a Unless otherwise noted for the analysis Second, we evaluated the impact of environmental water value selection on macroalgal resource use by cultured hard clams (Mercenaria mercenaria) in a Virginian coastal bay using a multi-isotope mixing model with three potential clam resources. Specifically, we used the δ2H, δ13C, and δ15N data from Hondula and Pace (2014). The potential resources for clams included in the model were macrophytes (MP; Zostera marina and Spartina alterniflora), macroalgae (MA; Codium fragile, Gracillaria vermicuphylla, Agardhiella subulata, and Ulva letuca), and microalgae (MI; phytoplankton and benthic algae) based on the analysis in Hondula and Pace (2014). Distributions for the clams and end member data comprised all available data from all sampling periods. Using the following 3-source, 3-isotope Bayesian mixing model (3)in which ΔC and ΔN are the trophic enrichment values (1.05‰ ± 0.75‰ and 3.42‰ ± 0.83‰, respectively), ω was varied between 0.1 and 0.3 and the resulting source portions (φMP, φMA, and φMI) were recorded. Additionally, the same analysis was performed using only the δ2H and δ15N data and the δ2H and δ13C data in order to determine if the removal of an isotope tracer would increase the influence of environmental water value selection. The model was also run with just the δ13C and δ15N data to determine if the δ2H, and therefore, potentially ω value, were substantially influential on the resource use posterior distributions. The Daphnia and clam mixing models were written in R (R Core Development Team) using JAGS (Just Another Gibbs Sampler). The distributions of the source portions (φA and φT or φMP, φMA, and φMI) were center log transformed as these distributions were bounded between zero and one (Semmens et al. 2009). The median value of Daphnia percent allochthony (φT) and clam percent macroalgal use (φMA) are reported and discussed throughout, however, the quantiles of the full posterior distributions are provided in the figures. Changes in the median values between model runs are reported as changes in percentage points (i.e., absolute changes in the median values of resource use). Change in δ2H value from lipid extraction of bulk tissues During the ice-free season, three consumer groups (zoobenthos, zooplankton, and fishes) and two end members (periphyton and terrestrial vegetation) were collected from four lakes on the University of Notre Dame's Environmental Research Center's property. Zooplankton were collected at night via oblique tows of a conical net (80 μm for zooplankton, 153 μm for Chaoborus) and separated by taxa under a dissecting microscope. Zoobenthos were collected using an Eckman dredge and separated from the sediments. A sample of dorsal muscle tissue was excised from fishes collected via minnow trapping, electrofishing, and angling. Periphyton was scraped from tiles suspended above the sediments of the lake in the littoral zone and terrestrial leaf samples were collected from trees within the watershed. Consumers and organic matter samples were also collected from nine sampling sites along the Hudson River estuary. Organic matter samples included cyanobacteria, filamentous algae, various submerged and floating aquatic vegetation, and terrestrial leaf material. Dorsal muscle tissue was collected from 10 fish species. Zoobenthos from the soft sediments were sampled using a Ponar sampler at midchannel. Zebra mussels (Dreissena polymorpha) were collected midchannel at depths of 0.5–4 m. All of the materials for isotope analysis were dried at 60°C and ground to a fine powder. Individual consumer samples from the Hudson River and the lakes were divided and half of the sample underwent lipid extraction and the other portion remained unchanged. Lipids were extracted with a 50 : 50 methanol–chloroform solution following the procedure of Folch et al. (1957). Isotope values were measured at the Colorado Plateau Stable Isotope Laboratory (CPSIL) using the benchtop equilibration method for exchangeable H described in Doucett et al. (2007). CPSIL uses a number of standards, one of which is Cladophora spp., a chlorophyte alga. All values are expressed in relation to the international standard of Vienna Standard Mean Oceanic Water (VSMOW). The change in δ2H between lipid extracted and bulk samples (Δδ2H) is reported here using the per mil (‰) notation and as a percent of the pre-extraction bulk value. We compared Δδ2H to consumer C : N values as an indicator of consumer lipid content where a higher C : N implies greater relative lipid composition. Impact of lipid correction on estimates of consumer resource use Average lipid correction values (Δδ2H) calculated from the samples described above were applied to end member and Daphnia δ2H values (Table 1) in a two-source (terrestrial material and algae) Bayesian mixing model (4)where the sample specific corrections (subscripts the same as Eq. 2) are applied before the environmental water correction (ωtot = 0.2 for this analysis). The model was run with sample-specific lipid corrections and without lipid corrections (Δδ2H = 0) to compute the difference in the terrestrial fraction (φT) between the two model runs. Estimating realistic H-exchange values An H-exchange value for Daphnia pulicaria of 0.074 was calculated based on reported tissue composition (Ventura and Catalan 2005) and the H-exchange values for those tissue types (Culebras and Moore 1977; Chesson et al. 2009). The likely H-exchange value for algae of 0.20 was calculated based on the average tissue composition (Becker 1994) of 17 algal species and the H-exchange values for those tissue types. The likely H-exchange for terrestrial material was assumed to be 0.23 which is the H-exchange value for cellulose (Chesson et al. 2009). For the H-exchange sensitivity analysis, the isotope data in Table 1 was considered the true but unmeasured δ2H values of the consumer and end members (δ2HTrue). The realistic H-exchange values, reported above, were used in this simple linear mixing model (5)to calculate what the measured δ2H (δ2HMeasured) values of the samples would be if ambient water vapor (δ2HVapor) was −90‰ (the approximate value at CPSIL). A keratin standard with an H-exchange value of 0.16 (Chesson et al. 2009) was used to correct (δ2HCorrected) the measured values (δ2HMeasured) using the following equation (6)the coefficients of which have been empirically determined from a regression analysis of measured δ2H of the keratin standard exchanging with the same water vapor (δ2Hvapor) as the samples (independent variable) vs. the expected δ2H of the nonexchangeable portion of the keratin standard based on an assumed del value. The δ2HCorrected values and the true δ2H values were then used in the two-source δ2H mixing model (Eq. 2) and the posterior distributions of φT were compared. Assessment Values of ω from literature and field data In total, we found or calculated ωtot values for 24 consumers from both freshwater and marine habitats (Table 2). There is not a significant difference in mean ωtot values among trophic positions, between habitats, between muscle and bulk tissues, between aquatic habitats, or between assumed and known diet. Excluding the snail which has a reported ω of 0.0, ωtot ranged from 0.06 to 0.39. The mean and standard deviation are 0.22 and 0.02, respectively (Fig. 1). Table 2. Literature review of dietary water (ω) values ("muscle" = protein, "bulk" = whole organism). Assumed isotope diet composition is inferred for studies where the diet was not directly observed yet reasonably well known, whereas known isotope diet composition is from lab reared animals fed a known diet. Reference Consumer Aquatic habitat Aquatic zone ω s.d. Trophic position Tissue type Diet Estep and Dabrowski (1980) Snail Marine Benthic 0.00 NA 2 Muscle Assumed Macko et al. (1983) Amphipod Marine Pelagic 0.12 NA 2 Bulk Known Hondula and Pace (2014) Clam Marine Benthic 0.15 0.09 2 Muscle Known Malej et al. (1993) Jellyfish Marine Pelagic 0.34 0.12 3 Bulk Assumed Finlay et al. (2010) Shredder insects Freshwater Benthic 0.12 0.12 2 Bulk Assumed Finlay et al. (2010) Scraper insects Freshwater Benthic 0.06 0.06 2 Bulk Assumed Solomon et al. (2009) Mosquito Freshwater Pelagic 0.39 0.04 2 Bulk Known Solomon et al. (2009) Zooplankton Freshwater Pelagic 0.20 0.04 2 Bulk Known Wilkinson et al. (2013)a Zooplankton Freshwater Pelagic 0.29 0.04 2 Bulk Assumed Berggren et al. (2014)b Cyclopoids Freshwater Pelagic 0.16 0.03 3 Bulk Assumed Berggren et al. (2014)b Calanoids Freshwater Pelagic 0.16 0.05 3 Bulk Assumed Berggren et al. (2014)b Cladocerans Freshwater Pelagic 0.29 NA 2 Bulk Assumed Wang et al. (2009) Chironomids Freshwater Benthic 0.31 0.03 2 Bulk Known Bortolotti et al. (2013) Snails Freshwater Benthic 0.21 0.03 2 Bulk Assumed Nielson and Bowen (2010) Brine shrimp Saline lake Pelagic 0.38 NA 2 Chitin Known Babler et al. (2011) Gizzard Shad Freshwater Benthic 0.08 0.04 2 Muscle Assumed Solomon et al. (2009) Chaoborus Freshwater Pelagic 0.14 0.06 3 Bulk Assumed Wilkinson et al. (2013) Chaoborus Freshwater Pelagic 0.13 0.03 3 Bulk Assumed Jardine et al. (2009) Water Strider Freshwater Neustic 0.25 NA 3 Bulk Known Solomon et al. (2009) Fish Freshwater Pelagic 0.12 0.02 3 Muscle Assumed Jardine et al. (2009) Trout Freshwater Pelagic 0.30 NA 3 Muscle Known Graham et al. (2014) Atlantic Salmon Mixed Pelagic 0.36 0.05 3 Muscle Known Graham et al. (2014) Arctic Char Mixed Pelagic 0.35 0.05 3 Muscle Known Solomon et al. (2009) Trout Freshwater Pelagic 0.23 0.03 3–4 Muscle Known a Calculated from the difference in zooplankton and phytoplankton δ2H from the three systems with the most autochthonous consumers b Calculated from zooplankton samples that were assumed to be autochthonous based on a δ13C mixing model Figure 1Open in figure viewerPowerPoint Rank plot of ω values presented in Table 2. The error bars for each point are standard deviation, when reported. The inset is the distribution of the mean ω values reported in the literature. For the field data of aquatic consumer δ2H, all consumers remained within the bounds of the study and system specific mean end members at an ωtot value of 0.15 and below (Fig. 2). All consumers across all studies fell outside the bounds of the end members when ωtot was 0.50. The consumers in some groups such as calanoid copepods (Figs. 2A,G) and Mesocyclops (Fig. 2C) fell outside the bounds of the end members when ωtot was below 0.50. Using the criteria that 90% of the consumers must fall within the end members, the empirical maximum for ωtot is 0.23 ± 0.03 averaged across all consumers and studies in Figure 2. Figure 2Open in figure viewerPowerPoint The percent of consumers that fall outside the bounds of the end members after correcting for dietary water (ωtot) using different values (increments of 0.05 shown though analysis was more refined). (A–D) Zooplankton from Wilkinson et al. (2013), E) fish from Babler et al. (2011), F–H) zooplankton from Berggren et al. (2014). While no clear pattern emerged from the ωtot values and auxiliary data collected from the literature, there does appear to be an upper threshold for bulk tissue ωtot of approximately 0.40 for aquatic consumers. The analysis of zooplankton consumer δ2H corroborated this, indicating that ωtot could not exceed 0.50 as all consumers fell outside the bounds of the end members when ωtot was this high. Together, these results suggest an upper limit for bulk tissue ωtot of 0.40–0.50 which is in agreement with the maximum value of approximately 0.40 put forth by Solomon et al. (2009). Using more stringent criteria that 90% of consumers remain within the bounds of the end members, we determined that the maximum ωtot value for consumers using the δ2H field data was 0.23, which is very similar to the median of the literature values (0.22). Although the field data are limited to freshwater consumers, they encompass more than 60 temperate ecosystems and are surprisingly consistent between studies. The literature median and survey-derived maximum are also very similar to previous compilations (Solomon et al. 2009) and modeled ω values (Cole and Solomon 2012). Considering the multiple lines of evidence, a reasonable assumed ω value is approximately 0.20 for aquatic consumers in the absence of other data. Influence of ω value selection on resource use estimates The Daphnia data from Table 1 was used to assess the impact of ω on estimates of consumer allochthonous resource use (φT) in a δ2H-based model. The median of Daphnia φT ranged from 7% to 87% allochthonous when ω was 0.10 (Fig. 3A). In general, as ω increased, Daphnia φT decreased. There was a decelerating non-linear decrease in the median φT with increasing ω. The range in median φT also decreased at higher ω. For example, the range in median φT for ω of 0.30 was only 7–69% allochthonous (Fig. 3C). Figure 3Open in figure viewerPowerPoint The sensitivity of resource use estimates (φT or φMA, φMI, φMP) for Daphnia and clams (Mercenaria mercenaria) assuming different values of ω in the analysis. (A–C) The posterior distributions of φT for the four Daphnia in Table 1 using different values of ω in a Bayesian mixing model (Eq. 2), and (D–F) The posterior distributions of φMA, φMI and φMP for clams for the three-source, three-isotope mixing model (Eq. 3). The data on cultured hard clams from Hondula and Pace (2014) was used to assess the impact of ω on estimates of consumer resource use in a multi-isotope modeling approach with more than two potential resources. In all three versions of the model analysis that included δ2H, the median values of clam resource use (φMP, φMA, and φMI) did not change by more than two percentage points for each value of ωtot evaluated (Fig. 3D–F). Instead, the resource use posterior distributions were much more sensitive to the isotopes used in the model. The model runs using δ2H + δ15N, δ15N + δ13C, and δ2H + δ13C + δ15N all had median φMA values between 63 and 69%. The exclusion of δ15N in the δ2H + δ13C model run yielded substantially different results (data not shown), with a median φMA of 31%. However, even in the δ2H + δ13C analysis, changing ω between 0.1 and 0.3 did not change the posterior distribution of φMA. From these analyses, we conclude that there can be a substantial influence of ω value selection on resource use estimates when δ2H is the only tracer in the mixing model. With the addition of another isotope tracer and another resource, the influence of ω value selection was greatly diminished. It is difficult to tease apart the confounding effect of the addition of isotope tracers and additional resources in the model as the inclusion of either necessarily further constrains the influence of environmental water in the mixing model compared to the one-isotope, two-resource model. Influence of lipid correction on resource use estimates In total, 61 samples of lipid extracted consumer and end member tissue were analyzed (Table 3). The values reported are the shift, in per mil units (‰), from the bulk sample. A positive shift is a less negative number; that is the lipid-free sample is less negative than the bulk sample. The range in the difference in δ2H before and after extraction (Δδ2H) was from −0.82‰ (algae) to 18.43‰ (zooplankton) and the average Δδ2H across all samples was 11.8‰ (Table 3). Only the algal samples were more negative than the bulk samples after extraction. All other consumer and end member samples were less negative after extraction consistent with expectation relative to the effect of removing lipids. Variation among groups was significant (α = 0.05) in Δδ2H (ANOVA, F6,54 = 2.42, p-value = 0.03). A post hoc Tukey test revealed that algae and zooplankton Δδ2H were significantly different. There were no significant relationships between consumer C : N and Δδ2H or bulk δ2H when all consumers were pooled or for individual consumer groups and end members. Table 3. The change in δ2H after lipid extraction (Δδ2H = lipid free δ2H − bulk ti