Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Equations for Lipid Normalization of Carbon Stable Isotope Ratios in Aquatic Bird Eggs

Equations for Lipid Normalization of Carbon Stable Isotope Ratios in Aquatic Bird Eggs

  • Kyle H. Elliott, 
  • Mikaela Davis, 
  • John E. Elliott
PLOS
x

Correction

24 Apr 2014: The PLOS ONE Staff (2014) Correction: Equations for Lipid Normalization of Carbon Stable Isotope Ratios in Aquatic Bird Eggs. PLOS ONE 9(4): e96587. https://doi.org/10.1371/journal.pone.0096587 View correction

Abstract

Stable isotope ratios are biogeochemical tracers that can be used to determine the source of nutrients and contaminants in avian eggs. However, the interpretation of stable carbon ratios in lipid-rich eggs is complicated because 13C is depleted in lipids. Variation in 13C abundance can therefore be obscured by variation in percent lipids. Past attempts to establish an algebraic equation to correct carbon isotope ratios for lipid content in eggs have been unsuccessful, possibly because they relied partly on data from coastal or migratory species that may obtain egg lipids from different habitats than egg protein. We measured carbon, nitrogen and sulphur stable isotope ratios in 175 eggs from eight species of aquatic birds. Carbon, nitrogen and sulphur isotopes were enriched in lipid-extracted egg samples compared with non extracted egg samples. A logarithmic equation using the C∶N ratio and carbon isotope ratio from the non extracted egg tissue calculated 90% of the lipid-extracted carbon isotope ratios within ±0.5‰. Calculating separate equations for eggs laid by species in different habitats (pelagic, offshore and terrestrial-influenced) improved the fit. A logarithmic equation, rather than a linear equation as often used for muscle, was necessary to accurately correct for lipid content because the relatively high lipid content of eggs compared with muscle meant that a linear relationship did not accurately approximate the relationship between percent lipids and the C∶N ratio. Because lipid extraction alters sulphur and nitrogen isotope ratios (and cannot be corrected algebraically), we suggest that isotopic measurement on bulk tissue followed by algebraic lipid normalization of carbon stable isotope ratio is often a good solution for homogenated eggs, at least when it is not possible to complete separate chemical analyses for each isotope.

Introduction

Stable isotope analysis is a useful technique for tracing the origin of nutrients in tissues with applications in environmental chemistry, paleoecology, migration biology and diet reconstruction [1][10]. Stable isotope analysis applied to egg tissue is particularly useful for understanding where resources are derived for reproduction (capital vs. income breeding) and to account for variation in toxic contamination within the egg due to diet [8][14].

Agencies in a number of countries systematically collect and archive bird eggs for toxicological and chemical analyses because eggs obtained early in the season can be re-laid, collection of 10–20 eggs has little impact on bird populations numbering in the thousands or millions, and the lipid-rich matrix accumulates many of the lipophilic toxins of interest [9], [15][21]. Archived egg specimen banks allow retrospective analysis of toxic contaminants. Stable isotope analysis helps tease apart whether changes in contamination on archived tissue occur due to changes in diet or changes in toxin abundance [17], [22][24].

The most common stable isotope ratios used by ecologists are those involving carbon (13C∶12C, measured relative to the PeeDee Belemnite standard and denoted δ13C) and nitrogen (15N∶14N, measured relative to pure air and denoted δ15N) ratios. Carbon isotopes can be used to identify habitat, as δ13C varies systematically with degree of aquatic and anthropogenic input. Nitrogen isotopes are primarily used to determine trophic level, as δ15N increases predictably with trophic level. Increasingly, sulfur isotope ratios (34S∶32S, measured relative to the Vienna Cañon Diablo Troilites standard and denoted δ34S) are also used to distinguish nutrients originating from marine environments. Stable isotope ratios change in a systematic fashion as nutrients are assimilated from an animal's food (i.e. the discrimination factor). An accurate knowledge of the discrimination factor is necessary to quantitatively predict nutrient origin [5], [25].

Discrimination factors differ among tissue types, with lipids being more depleted in 13C than protein because lipid biosynthesis preferentially incorporates 12C compared with protein [26][31]. In particular, isotopic fractionation during the oxidation of pyruvate to acetyl coenzyme A, the main precursor to fatty acids, preferentially incorporates 12C [32]. Thus, despite originating from the same resource, the stable isotope ratio of consumer tissue will be more depleted in 13C if the lipid content of that tissue is high.

There are two methods for correcting for lipid content [31], [33][36]. First, lipids can be extracted chemically prior to measurement, although lipid extraction can alter δ15N by washing out nitrogenous compounds. Measurement of δ13C in lipid-extracted tissue and δ15N in non-extracted tissue overcomes that issue, but doubles cost and work load. Furthermore, the lipid extraction process can be time-consuming and can involve the use of hazardous chemicals, such as chloroform. Second, provided an index of lipid content is available, isotope ratios can be corrected for lipid content analytically without extraction. Often, the ratio of total weight of carbon to total weight of nitrogen within the sample (C∶N ratio) is used as an index of lipid content because nitrogen abundance is high in proteins and low in lipids, and C∶N ratios are readily calculated from data obtained during stable isotope analysis [31], [33][36].

Lipid normalization models are equations that use the C∶N ratio or percent lipids to calculate the value of δ13C that would have been present in a tissue following lipid extraction from the value measured in non-extracted tissue. Such models are accurate in muscle tissue in a wide variety of animals [31], [35], [37], [38]. Furthermore, equations developed from the Post et al. dataset [31] accurately predicted 85% of the data points in a different dataset [35], suggesting that equations are robust across datasets.

Using lipid normalization models in eggs, however, has been more problematic (Table 1). For instance, δ13C in eggs collected from three species of wild birds varied widely between lipid-extracted and non-extracted tissue, and was not predicted by C∶N ratio [38], [39]. Similarly, the mean-square error for eggs was four times greater than for muscle (once three outliers and seven points with incomplete lipid extraction were excluded for muscle; no such points were apparently excluded for eggs, [38]). Waterfowl, in particular, show a different or inconsistent trend between lipid-extracted and non-extracted tissue [35], [39]. One potential reason for variation among studies is that coastal birds (e.g. eagles, waterfowl) may incorporate variable amounts of freshwater-derived rather than marine-derived lipids into their tissues, that variation may be independent of protein source, and freshwater-derived lipids are depleted in δ13C compared to marine-derived lipids [38], [39].

thumbnail
Table 1. Difference between lipid-extracted and non-extracted samples for bird egg tissue for carbon (Δδ13C), nitrogen (Δδ15N) and sulphur (Δδ34S).

https://doi.org/10.1371/journal.pone.0083597.t001

Both linear and non-linear equations have been used to describe the relationship between the C∶N ratio and the effect of lipid extraction on tissue δ13C values. For tissues that have a relatively low percent lipids, such as muscle, a linear equation appears to hold (most studies show no or only slight non-linearity [31], [35], [38]; but see [40] that found support for a non-linear equation). However, a nonlinear relationship, whereby the effect of lipid extraction is smaller at higher percent lipid content, appears to work better for eggs, which usually have higher percent lipid content than muscles [35]. The mathematical necessity of a decelerating, non-linear equation for tissues with high percent lipids is evident in the fact that a mixture of pure fatty acids (100% lipids) would have a nitrogen content of zero and an infinite C∶N ratio. A linear model would then imply that the δ13C value would be depleted infinitely, when in fact δ13C would merely be depleted by 5–6‰.

Because of the paucity of data on the effect of lipid-extraction on the eggs of non-waterfowl [35], [38], we examined the effect of lipid extraction on the eggs of six species of marine seabirds and an aquatic raptor. Those seabirds were chosen in the 1980s as part of a toxic contaminant monitoring program in Pacific Canada. The species were chosen because they feed in nearshore, continental shelf and offshore environments, and, therefore, can be used to monitor contaminants in those habitats [9], [41]. In the context of the current study, the variation in habitat (freshwater vs. nearshore vs. offshore) allows us to examine the effect of habitat on the predictability of lipid normalization equations. By including multiple individuals per species, from seven different species, we can examine both intra-specific and inter-specific variability in the relationship between lipid-extracted and non-lipid-extracted δ13C relative to the ratio of bulk C∶N. In addition, we examined the effect of a second lipid extraction process on δ13C of an eighth seabird species. The goal of the manuscript is to develop an equation for normalizing for the effect of lipids on whole egg homogenate, to examine whether that relationship is linear or nonlinear, and to determine whether the relationship is species- or habitat-specific.

Materials and Methods

We randomly selected archived egg contents from the Environment Canada's Specimen Bank [42]. We did not control for laying order (storm-petrels and auklets only lay a single egg), or days since lay, which can affect stable isotope ratios in eggs [43], [44]. As we collected eggs randomly during early incubation, we do not believe those issues created systematic bias. Eggs were collected from a nearshore marine seabird (double-crested cormorant Phalacrocorax auritus), three seabirds that forage primarily on the continental shelf during the breeding season (ancient murrelet Synthliboramphus antiquus, rhinoceros auklet Cerorhinca monocerata, pelagic cormorant Phalacrocorax pelagicus), one offshore-foraging marine seabird (Leach's storm-petrel Oceanodrama leucorhoa), a freshwater aquatic raptor (osprey Pandion haliaetus) and an aquatic bird that also uses the terrestrial environment (great blue heron Ardea herodias). For the purposes of analysis, we classified bird species into three habitat types: offshore (Leach's storm-petrel), continental shelf (rhinoceros auklet and pelagic cormorant) and those in habitats influenced by the terrestrial environment, such as lakes or estuaries (ancient murrelet, osprey, double-crested cormorant and great blue heron). Egg contents were homogenized and frozen until analysis.

After thawing, samples were freeze-dried. A small sample (1 mg) was removed, encapsulated in tin, and sent to the Stable Isotope Facility at University of California, Davis (http://stableisotopefacility.ucdavis.edu). On a separate sample, we extracted the lipids using a Soxhlet apparatus with petroleum ether as the solvent [45]. Specifically, a thimble filled with dried sample was placed in a Soxhlet extractor and washed with petroleum ether at 94°C for 8 hours. The solvent was then distilled off and the residue dried for 60 minutes in a drying oven. We chose petroleum ether rather than chloroform∶methanol as our solvent because several Canadian government institutions are considering the restrictions on the use of chloroform, and we wished our methods to be accessible to future researchers. After extraction, a small sample (1 mg) was removed, encapsulated, and sent to the same facility. Likewise, small samples (3 mg) of lipid-extracted and non-extracted tissue were sent to the Environmental Isotope Laboratory at University of Waterloo (http://www.uweilab.ca) for sulphur analysis.

Samples were analyzed for 13C/12C and 15N/14N isotopes using a PDZ Europa ANCA-GSL elemental analyzer (EA) interfaced to a PDZ Europa 20-20 isotope ratio mass spectrometer (IRMS; Sercon Ltd., Cheshire, UK). Samples were combusted at 1000°C in a reactor packed with chromium oxide and silvered copper oxide. Following combustion, oxides were removed in a reduction reactor (reduced copper at 650°C). The helium carrier then flowed through a water trap (magnesium perchlorate). N2 and CO2 were separated on a Carbosieve GC column (65°C, 65 mL/min) before entering the IRMS. Samples were analyzed for 34S/32S using a Europa Roboprep-20/20 EA-IRMS. During analysis, samples were interspersed with several replicates of at least two different laboratory standards. The final delta values were expressed relative to international standards Vienna PeeDee Belemnite, Vienna Cañon Diablo Troilite and air for carbon, sulphur and nitrogen, respectively.

To examine the effect of a different lipid extraction process on variation in δ13C, we also collected egg yolk from glaucous-winged gulls (Larus glaucescens), a nearshore marine seabird that also uses the terrestrial environment. We used chloroform∶methanol extraction to remove lipids from egg yolk. Lipid-extracted and non-extracted egg yolks were encapsulated and sent to the UC Davis Stable Isotope Laboratory for carbon and nitrogen analyses. All data for both projects are presented in Table S1.

We used R 2.14.2 for all statistical analyses. We used linear regression and non-linear regression based on previous published logarithmic models [35] to examine the relationship between Δδ13C (δ13Clipid extracted - δ13Cnon extracted), Δδ15N (δ15Nlipid extracted - δ15Nnon extracted), Δδ34S (δ34Slipid extracted – δ34Snon extracted) and the C∶N ratio. Linear models are appropriate for tissues with low percent lipids because as long as percent lipids is low a linear model approximates the nonlinear function; for tissues with high percent lipids (e.g., eggs), the C∶N ratio does not change linearly with percent lipids (because the denominator, nitrogen, approaches zero) necessitating a non-linear function [35],[40]. Because models with an increasing number of parameters will necessarily increase fit while including spurious relationships (i.e., a model with 175 parameters would explain 100% of the variation in our 175-sample data set), we used Akaike's information criterion to select the most parsimonious relationship. We considered both linear and non-linear functions, and functions that included species and habitat (offshore, shelf and terrestrial-influenced), and their interactions, as co-variates. We then examined whether each of the 175 values were predicted within 0.5‰ by each model, with each model recalculated to exclude that point. Because many authors will only examine these relationships within a single species, we also calculated species-specific regressions to determine the likelihood of finding a relationship given only a single species with a more limited range of lipid content. We also correlated δ13Clipid-extracted and δ13Cnon-extracted. We removed eight values, all from the same Soxhlet run, because the C∶N ratio after lipid extraction was >5.0. All raw data are available in Table S1.

Results

Lipid-extracted egg samples were more enriched in carbon, sulphur and nitrogen than non-extracted egg samples (Table 1). Across the entire dataset, there was strong support for a logarithmic relationship between Δδ13C and the C∶N ratio (Table 2, Fig. 1). Relationships between Δδ15N and the C∶N ratio, and Δδ34S and the C∶N ratio, were weaker, but still supported (Table 2). For the entire dataset of 175 eggs, 22 (12%) of the lipid-extracted egg values differed by more than 0.5‰ in δ13C when calculated using the linear model, 19 (10%) differed by more than 0.5‰ for the logarithmic model and 12 (7%) differed by more than 0.5‰ from the complete model with habitat and species interactions included. Nonetheless, the best-supported model included separate terms in the intercept (but not the slope) for each habitat (Table 2). In contrast, 39 (22%) of the lipid-extracted egg values differed by more than 0.5‰ compared with the linear “waterfowl” equation from Ehrich et al. (2011; their Table 2, “Linear”) and 173 (99%) of the values compared with the linear “non-waterfowl” equation. Lipid-extracted and algebraically-corrected δ13C values were highly correlated (R2 = 0.990; compared with R2 = 0.969 for the non-algebraically corrected values). Average repeatability (standard deviation in ‰, N = 19) for duplicates (separately homogenized) was 0.26 for δ34S, 0.12 for δ13C and 0.31 for δ15N.

thumbnail
Figure 1. Difference between lipid-extracted and non-extracted stable isotope ratios for bird egg tissue.

Specifically (A) carbon (Δδ13C), (B) nitrogen (Δδ15N) and (C) sulphur (Δδ34S) increases with ratio of carbon to nitrogen by weight (C∶N ratio) across seven aquatic bird species: ancient murrelet (ANMU), double-crested cormorant (DCCO), great blue heron (GBHE), Leach's storm-petrel (LESP), osprey (OSPR), pelagic cormorant (PECO) and rhinoceros auklet (RHAU). Also shown are results from studies listed in Table 1 (eider average with SD bars shown, Arctic birds) and best-fit habitat- and species-dependent regression models listed in Table 2. (D) Δδ13C for groups within our study compared with arithmetic lipid-correction models proposed by Post et al. [29], Ehrich et al. [33] (filled lines) and within our own study (dashed lines).

https://doi.org/10.1371/journal.pone.0083597.g001

thumbnail
Table 2. Ranking of models used to describe the difference between lipid-extracted and non-extracted bird egg tissue.

https://doi.org/10.1371/journal.pone.0083597.t002

There was a statistically-significant relationship between Δδ13C and the C∶N ratio for all of the species with sample sizes greater than two (ancient murrelet: t5 = 4.34, P = 0.007; double-crested cormorant: t8 = 3.73, P = 0.006; Leach's storm-petrel: t66 = 6.43, P<0.0001; pelagic cormorant: t24 = 7.32, P<0.0001; rhinoceros auklet: t49 = 6.56, P<0.0001; gull yolk: t17 = 4.44, P = 0.0004), except ospreys (t10 = −0.13, P = 0.90).

Discussion

Lipid extraction enriched egg δ13C values by ∼2‰, δ15N values by ∼1‰ and δ34S by ∼0.5‰. The degree of enrichment in egg δ13C values (Δ13C) was correlated with the percent lipids, as inferred by the C∶N ratio. Indeed, across the entire dataset, and within all marine species, the C∶N ratio strongly predicted Δ13C. As more than 90% of the samples were estimated within ±0.5‰ by the best-fit algebraic equation (logarithm model), our algebraic equation is a robust method of calculating lipid-extracted δ13C values given δ13C values measured on non-extracted egg tissue from marine birds.

Our models for egg tissue were as accurate as those for muscle tissue [31], [35]. In contrast, most past attempts at providing a method for calculating lipid-extracted values for egg tissues have been unsuccessful (Table 1). Those methods focused on migratory or coastal species, such as bald eagles, eiders, snow geese and other waterfowl (Table 1). For migratory birds—at least those that are capital breeders—egg lipids can be derived from energy reserves obtained on non-breeding grounds with different δ13C signatures; egg protein, in contrast, may be derived from the breeding grounds [35], [39], [46]. Differences in δ13C may be particularly large for coastal birds switching between freshwater and marine prey bases [17], or for other species showing dietary switches during egg-laying [47], [48], even if they are non-migratory. For instance, the ospreys whose eggs were included in this study were tracked via satellites to wintering grounds in both marine and freshwater environments [49]. If a portion of their egg lipid δ13C values were derived from wintering grounds and their protein δ13C values were mainly derived from breeding grounds, that would explain why the lipid-extracted and non-extracted samples were consistently different [35], [39]. We therefore concur with Oppel et al. [39] that algebraic correction for the effect of lipids on δ13C in eggs is accurate for birds relying on resources acquired concurrently with reproduction for egg synthesis, such as income breeders, but not for coastal or migratory birds that may bring lipid reserves from a habitat with a different δ13C value. Specifically, algebraic corrections are applicable where lipids and proteins would be expected to be acquired in isotopically similar landscapes, with capital and income breeding being extremes of a continuum with actual reproductive strategies ranging from high to low reliance on stored nutrients [50], [51]. It is also important to emphasize that our results were primarily for egg homogenate, which has a higher proportion of protein (mostly albumin) than yolk. Arithmetic correction for yolk may be more problematic because of the lower proportion of protein in yolk and may account for problems encountered by previous authors working with yolk [39].

Stage of embryonic development also may have impacted the relationship between C∶N ratio and Δδ13C. As the embryo develops, there may be differential uptake of varying nutrients and, therefore, variation in 13CO2 expiration. For instance, chicken yolk δ15N was depleted by 1.0‰ after 15 days of development and chicken albumin δ13C was depleted by 0.2‰ after 3 days of development [43]. In principle, if 13C was excreted differentially to 12C in yolk relative to plasma, that could increase the variance in Δδ13C relative to lipid content. As we did not know the age of the embryo, this issue could play a role in our dataset, although collectors attempted to obtain eggs shortly after laying.

At the early stages of development (prior to significant growth of the embryo), the egg interior primarily consists of the cysteine-rich protein albumin (egg white) and lipid-rich yolk. Many authors have separated the two components and analyzed stable isotopes separately on each component [39], [43]. As we were interested in establishing lipid normalization equations for retrospective analyses of specimen banks for toxicological assays, where specimens have usually been homogenized, we did not examine each component separately, except for the gull eggs. Interestingly, our lipid-extracted C∶N ratio, 4.33 (SD = 0.50), is almost identical to that of egg albumin (4.28 based on the albumin amino-acid sequence), suggesting following lipid extraction on egg homogenate we were left almost entirely with albumin or with a matrix that was very similar in makeup. In particular, simple statements that pure protein should have a C∶N ratio of ∼3.0 [30] seem erroneous as the C∶N ratio will depend on the particular amino acid composition of the protein.

The influence of habitat and breeding strategy on Δδ13C

Pelagic foragers (Leach's storm-petrels) had higher Δδ13C than nearshore foragers (rhinoceros auklets and pelagic cormorants), which had higher Δδ13C than birds foraging in terrestrial-influenced habitats (ancient murrelets, double-crested cormorants, great blue herons and ospreys). Ancient murrelets may forage farther offshore than pelagic cormorants, but for the current classification scheme they appeared to group with those species feeding in terrestrial-influenced habitats. Perhaps there is a larger difference in the carbon isoscape in winter (where lipids were derived) relative to breeding (where non-lipids were derived) habitat for birds feeding offshore than for those feeding in the terrestrial environment. Alternatively, within the capital-income breeding strategy continuum, Leach's storm-petrels may rely more on endogenous stores for egg production and the terrestrial-influenced species may rely more on exogenous stores for egg production [48], [50][52]. Ehrich et al. [35] found higher Δδ13C for terrestrial birds (along with one seabird and several shorebirds) than waterfowl. Either variation in habitat or breeding strategy (endogenous or capital vs. exogenous or income) may explain why Δδ13C differed between terrestrial birds and waterfowl in that study.

The effect of solvent

The solvent used during lipid extraction is known to effect measurement of δ13C and δ15N, at least in fish and invertebrates [45], [52]. The ideal solvent would extract all lipids, and only lipids, but in practice both lipids and non-lipids vary in polarity; non-polar lipids, such as triglycerides, are more soluble in non-polar solvents (e.g. petroleum ether, polarity index ∼0.1) and polar lipids, such as phospholipids, are more soluble in polar solvents (e.g. 2∶1 chloroform∶methanol, polarity index ∼4.4). Thus, no solvent will extract all lipids and all solvents will extract some non-lipids, such as hydrophobic amino acid or carbohydrate derivatives. In one study, δ13C in lipid-extracted tissue varied among three different solvents, with chloroform-methanol (Bligh-Dyer method, [54]) causing systematic errors on one type of invertebrate tissue [45]. In contrast, another study on fish tissue found that chloroform-methanol extraction was somewhat more effective than other techniques [53].

We believe a solvent effect is present for bird eggs. The arithmetic correction for Δδ13C proposed by Ehrich et al. [35] is quite effective for non-waterfowl (gulls in our study) and waterfowl [39] in other studies that used chloroform∶methanol as the solvent (Fig. 1d). The Ehrich et al. [36] equation calculates higher Δδ13C values than our equations, which would be consistent with chloroform-methanol being a stronger solvent (extracting a greater overall portion of lipids, but also extracting some non-lipids). Furthermore, pure gull albumin had a C∶N ratio of 3.72±0.06 and lipid-extracted whole egg had a C∶N ratio of 3.72±0.04 in the gull dataset using the chloroform-methanol extraction process, whereas lipid-extracted whole egg had a C∶N ratio of 4.33±0.50 in the seabird dataset using the petroleum ether extraction process. Thus, we believe that chloroform-methanol extractions remove more, or at least different, lipids (and other molecules) than petroleum ether, and therefore cause a larger Δδ13C.

Why use lipid normalization equations?

Our lipid normalization equations will be of use in accounting for the effect of lipids in toxicological studies of purely marine birds, although we urge the establishment of separate equations for each species. There are many reasons why lipid extraction is sub-optimal: chloroform is not allowed at many laboratories for health reasons, the Soxhlet apparatus used in the current studies cost over $10 000, the apparatus created a bottleneck of only eight samples processed each day, δ13C (and δ15N/δ34S) will vary among extraction techniques and, finally, to avoid overestimation of δ15N or δ34S (which cannot be corrected algebraically) it would be necessary to complete two different analyses, doubling expenses. Lastly, lipid extraction removes at least some amino acids, which can insert some (unwanted) variability into diet estimation. For studies of nutrient allocation or prey use, very accurate and unbiased measurements of δ13C may be necessary, but for toxicological studies that primarily use stable isotope ratios as correlative, predictive variables, the algebraic approach may be sufficient; across all species, lipid-extracted and algebraically-corrected values were highly correlated (R2 = 0.99) so that in correlative studies essentially no information would be lost by using the algebraic approach. Furthermore, where the algebraic equations do not provide an accurate index of lipid extracted δ13C because lipids and proteins are derived from different environments, the algebraic equations will give the value more associated with lipophilic contaminants; the stable isotope ratios measured on lipid-extracted tissue (albumin) will not reflect the diet associated with the prey from which the lipophilic contaminants are derived. We, therefore, urge the use of algebraic corrections, rather than lipid extraction, in toxicological studies.

What equation to use?

The most-widely used equation is the Post et al. [31] equation, which is accurate for muscle from many different organisms [31], [35]. Muscle is relatively low in lipids, so the non-linearity apparent in our dataset is not a significant issue (i.e. the “Post” equation is similar to the “Ehrich” equation at low C∶N ratio, Fig. 1d). Because egg tissue can have much higher levels of lipids, a non-linear equation is necessary (Fig. 1a,d, Table 2). To estimate lipid-extracted egg tissue values (using a petroleum ether extraction process), we therefore suggest using the equationwhich can be increased or decreased depending on the habitat of the study species (following Table 1). The C∶N Ratio is the ratio of the weight of carbon in the sample to the weight of nitrogen in the sample, often reported in µg. The equation is not species-specific, as we did not find that species was an important term in our models.

Supporting Information

Table S1.

Stable isotope data analyzed in the manuscript.

https://doi.org/10.1371/journal.pone.0083597.s001

(XLSX)

Acknowledgments

We thank J. Roth for access to the Soxhlet apparatus. M. Guigueno and D., J. and L. Jamieson completed the lipid extraction and sample encapsulation. R. McDonald and three anonymous reviewers provided useful comments on a previous version of this manuscript.

Author Contributions

Conceived and designed the experiments: KHE JEE MD. Performed the experiments: KHE MD. Analyzed the data: KHE. Contributed reagents/materials/analysis tools: JEE. Wrote the paper: KHE.

References

  1. 1. Roth JD (2003) Variability in marine resources affects arctic fox population dynamics. J Anim Ecol 72: 668–676.
  2. 2. West JB, Bowen GJ, Cerling TE, Ehleringer JR (2006) Stable isotopes as one of nature's ecological recorders. Trends Ecol Evol 21: 408–414.
  3. 3. Chiaradia A, Forero MG, Hobson K, Swearer S, Hume F, et al. (2012) Diet segregation between two colonies of little penguins Eudyptula minor in southeast Australia. Austr Ecol 37: 610–619.
  4. 4. Choy ES, Gauthier M, Mallory ML, Smol JP, Lean D, et al. (2010) An isotopic investigation of mercury accumulation in terrestrial food webs adjacent to an Arctic seabird colony. Sci Tot Environ 408: 1858–1867.
  5. 5. Jardine TD, Kidd KA, Fisk AT (2006) Applications, considerations, and sources of uncertainty when using stable isotope analysis in ecotoxicology. Environ Sci Toxicol 40: 7501–7511.
  6. 6. Elliott KH, Crump D, Gaston AJ (2010) Sex-specific behavior by a monomorphic seabird represents risk partitioning. Behav Ecol 21: 1024–1032.
  7. 7. Wolf N, Carleton SA, Martinez del Rio CM (2009) Ten years of experimental animal isotopic ecology. Funct Ecol 23: 17–26.
  8. 8. Inger R, Bearhop S (2008) Applications of stable isotope analyses to avian ecology. Ibis 150: 447–461.
  9. 9. Elliott JE, Elliott KH (2013) Tracking marine pollution. Science 340: 556–558.
  10. 10. Elliott KH, Cesh LS, Dooley JA, Letcher RJ, Elliott JE (2009) PCBs and DDE, but not PBDEs, increase with trophic level and marine input in nestling bald eagles. Sci Tot Environ 407: 3867–3875.
  11. 11. Guigueno MF, Elliott KH, Levac J, Wayland M, Elliott JE (2012) Differential exposure of alpine ospreys to mercury: melting glaciers, hydrology or deposition patterns? Environ Int 40: 24–32.
  12. 12. Elliott JE, Levac J, Guigueno M, Shaw P, Wayland ME, et al. (2012) Factors influencing legacy pollutant accumulation in alpine osprey: biology, topography or melting glaciers?. Environ Sci Tech 46: 9681–9689.
  13. 13. Day RD, Roseneau DG, Vander Pol SS, Hobson KA, Donard OF, et al. (2012) Regional, temporal, and species patterns of mercury in Alaskan seabird eggs: Mercury sources and cycling or food web effects? Environ Poll 166: 226–232.
  14. 14. Ito M, Kazama K, Niizuma Y, Minami H, Tanaka Y, et al. (2012) Prey resources used for producing egg yolks in four species of seabirds: insights from stable-isotope ratios. Ornithol Sci 11: 113–119.
  15. 15. Braune BM, Hobson KA, Malone BJ (2005) Regional differences in collagen stable isotope and trace element profiles in populations of long-tailed duck breeding in the Canadian Arctic. Sci Tot Environ 346: 156–168.
  16. 16. Braune BM, Trudeau S, Jeffrey DA, Mallory ML (2011) Biomarker responses associated with halogenated organic contaminants in northern fulmars (Fulmarus glacialis) breeding in the Canadian Arctic. Environ Poll 159: 2891–2898.
  17. 17. Hebert CE, Weseloh DVC, Idrissi A, Arts MT, O'Gorman R, et al. (2008) Restoring piscivorous fish populations in the Laurentian Great Lakes causes seabird dietary change. Ecology 89: 891–897.
  18. 18. Holmström KE, Järnberg U, Bignert A (2005) Temporal trends of PFOS and PFOA in guillemot eggs from the Baltic Sea, 1968–2003. Environ Sci Tech 39: 80–84.
  19. 19. Mallory ML, Braune BM (2012) Tracking contaminants in seabirds of Arctic Canada: Temporal and spatial insights. Mar Poll Bull 64: 1475–1484.
  20. 20. Vander Pol SS, Becker PR, Kucklick JR, Pugh RS, Roseneau DG, et al. (2004) Persistent organic pollutants in Alaskan murre (Uria spp.) eggs: Geographical, species, and temporal comparisons. Environ Sci Tech 38: 1305–1312.
  21. 21. Crosse JD, Shore RF, Jones KC, Pereira MG (2012) Long-term trends of PBDEs in gannet (Morus bassanus) eggs from two UK colonies. Environ Poll 161: 93–100.
  22. 22. Elliott JE (2005) Chlorinated hydrocarbon contaminants and stable isotope ratios in pelagic seabirds from the North Pacific Ocean. Arch Environ Contamin Toxicol 49: 89–96.
  23. 23. Elliott JE (2005) Trace metals, stable isotope ratios, and trophic relations in seabirds from the North Pacific Ocean. Environ Toxicol Chem 24: 3099–3105.
  24. 24. Vo ATE, Bank MS, Shine JP, Edwards SV (2011) Temporal increase in organic mercury in an endangered pelagic seabird assessed by century-old museum specimens. Proc Natl Acad Sci U S A 108: 7466–7471.
  25. 25. Bond AL, Daimond AW (2011) Recent Bayesian stable-isotope mixing models are highly sensitive to variation in discrimination factors. Ecol Appl 21: 1017–1023.
  26. 26. Sotiropoulos MA, Tonn WM, Wassenaar LI (2004) Effects of lipid extraction on stable carbon and nitrogen isotope analyses of fish tissues: potential consequences for food web studies. Ecology of Freshwater Fish 13: 155–160.
  27. 27. Sweeting CJ, Polunin NVC, Jennings S (2004) Tissue and fixative dependent shifts of delta C-13 and delta N-15 in preserved ecological material. Rapid Comm Mass Spectr 18: 2587–2592.
  28. 28. Sweeting CJ, Polunin NVC, Jennings S (2006) Effects of chemical lipid extraction and arithmetic lipid correction on stable isotope ratios of fish tissues. Rapid Comm in Mass Spectr 20: 595–601.
  29. 29. Bodin N, Le Loc'h F, Hily C (2007) Effect of lipid removal on carbon and nitrogen stable isotope ratios in crustacean tissues. J Exp Mar Biol Ecol 341: 168–175.
  30. 30. Hussey NE, Olin JA, Kinney MJ, McMeans BC, Fisk AT (2012) Lipid extraction effects on stable isotope values (δ13C and δ15N) of elasmobranch muscle tissue. J Exp Mar Biol Ecol 434: 7–15.
  31. 31. Post DM, Layman CA, Arrington DA, Takimoto G, Quattrochi J, et al. (2007) Getting to the fat of the matter: models, methods and assumptions for dealing with lipids in stable isotope analyses. Oecologia 152: 179–189.
  32. 32. De Niro MJ, Epstein S (1977) Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197: 261–263.
  33. 33. Kiljunen M, Grey J, Sinisalo T, Harrod C, Immonen H, et al. (2006) A revised model for lipid-normalizing delta C-13 values from aquatic organisms, with implications for isotope mixing models. J Appl Ecol 43: 1213–1222.
  34. 34. Mintenbeck K, Brey T, Jacob U, Knust R, Struck U (2008) How to account for the lipid effect on carbon stable-isotope ratio (delta C-13): sample treatment effects and model bias. J Fish Biol 72: 815–830.
  35. 35. Ehrich D, Tarroux A, Stien J, Lecomte N, Killengreen S, et al. (2010) Stable isotope analysis: modeling lipid normalization for muscle and eggs from arctic mammals and birds. Meth Ecol Evol 2: 66–76.
  36. 36. Tarroux A, Ehrich D, Lecomte N, Jardine TD, Bêty J, et al. (2010) Sensitivity of stable isotope mixing models to variation in isotopic ratios: evaluating consequences of lipid extraction. Meth Ecol Evol 1: 231–241.
  37. 37. Lesage V, Morin Y, Rioux È, Pomerleau C, Ferguson SH, et al. (2010) Stable isotopes and trace elements as indicators of diet and habitat use in cetaceans: predicting errors related to preservation, lipid extraction, and lipid normalization. Mar Ecol Prog Ser 419: 249–265.
  38. 38. Ricca MA, Miles AK, Anthony RG, Deng X, Hung SSO (2007) Effect of lipid extraction on analyses of stable carbon and stable nitrogen isotopes in coastal organisms of the Aleutian archipelago. Can J Zool 85: 40–48.
  39. 39. Oppel S, Federer RN, O'Brien DM, Powell AN, Hollmen TE (2010) Effects of lipid extraction on stable isotope ratios in avian egg yolk: is arithmetic correction a reliable alternative? Auk 127: 1–7.
  40. 40. Logan JM, Jardine TD, Miller TJ, Bunn SE, Cunjak RA, et al. (2008) Lipid corrections in carbon and nitrogen stable isotope analyses: comparison of chemical extraction and modelling methods. J Anim Ecol 77: 838–846.
  41. 41. Elliott JE, Noble DG, Norstrom RJ, Whitehead PE, Simon M, et al.. (1992) Patterns and trends of organic contaminants in Canadian seabirds, 1968–1990. In Persistent Pollutants in the Marine Environment, C.H. Walker and D.R. Livingston (Eds.) Pergamon Press. Oxford, pp. 181–194.
  42. 42. Elliott JE (1984) Collecting and archiving wildlife specimens in Canada. In: R.A. Lewis, N. Stein, and C.W. Lewis (eds.) Environmental Specimen Banking and Monitoring as Related to Banking, Martinus Nijhoff, The Hague, pp. 45–64.
  43. 43. Sharp CM, Abraham KF, Burness G (2009) Embyro development influences the isotopic signatures of egg components in incubated eggs. Condor 111: 361–365.
  44. 44. Hobson KA, Thompson JE, Evans MR, Boyd S (2005) Tracing nutrient allocation to reproduction in Barrow's goldeneye. J Wildl Manage 69: 1221–1228.
  45. 45. Schlechtriem C, Focken U, Becker K (2003) Effect of different lipid extraction methods on δ13C of lipid and lipid-free fractions of fish and different fish feeds. Isotopes in Environ Health Stud 39: 135–140.
  46. 46. Yohannes E, Valcu M, Lee RW, Kempenaers B (2010) Resource use for reproduction depends on spring arrival time and wintering area in an arctic breeding shorebird. J Avian Biol 41: 580–590.
  47. 47. Jacobs SR, Elliott KH, Gaston AJ, Weber JM (2009) Fatty acid signatures of female Brünnich's guillemot Uria lomvia suggests reliance on local prey for replacement egg production. J Avian Biol 40: 327–336.
  48. 48. Morrissey CA, Elliott JE, Ormerod SJ (2010) Diet shifts during egg laying: implications for measuring contaminants in bird eggs. Environ Poll 158: 447–454.
  49. 49. Elliott JE, Morrissey CA, Henny CJ, Inzunza ER, Shaw P (2007) Satellite telemetry and prey sampling reveal contaminant sources to Pacific Northwest ospreys. Ecol Appl 17: 1223–1233.
  50. 50. Thomas VG (1989) Body condition, ovarian hierarchies, and their relation to egg formation in anseriform and galliform species. Proc Int Ornithol Congr XIX: 353–363.
  51. 51. Jönsson KI (1997) Capital and income breeding as alternative tactics of resource use in reproduction. Oikos 57–66.
  52. 52. Hobson KA, Jehl JR (2010) Arctic waders and the capital-income continuum: further tests using isotopic contrasts of egg components. J Avian Biol 41: 465–572.
  53. 53. Logan JM, Lutcavage ME (2008) A comparison of carbon and nitrogen stable isotope ratios of fish tissues following lipid extractions with non-polar and traditional chloroform/methanol solvent systems. Rapid Comm Mass Spectr 22: 1081–1086.
  54. 54. Bligh EG, Dyer WJ (1959) A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37: 911–917.