Experimental validation confirms a carbon stable isotope lipid normalization procedure for Pacific salmon

Abstract

Carbon stable isotope analysis is an important tool in studies of fish ecology. Studies using this tool must account for the effect of lipids present in fish tissue on carbon stable isotope (δ13C) values. Simple correction equations have been developed to correct for this effect. For taxa with high fat content, choosing an accurate correction equation can have a significant impact on results. Pacific salmon (Oncorhynchus spp.) are a lipid-rich genus. Their broad marine distribution and ecological, cultural, and commercial importance make them prime candidates for stable isotope analysis. To determine both an accurate lipid correction equation for Pacific salmon δ13C values, and the effect of lipid extraction on bulk nitrogen isotope (δ15N) values, we performed pairwise isotope analysis on lipid-extracted and untreated muscle samples from 68 Chinook salmon O. tshawytscha spanning a size range of 165-750 mm and C:N values of 2.96-14.25. We compared the fit of existing δ13C lipid correction equations from 3 previously published models to our data, and optimized the top performing model using a leave-one-out cross validation. The model of Kiljunen et al. (2006; https://doi.org/10.1111/j.1365-2664.2006.01224.x) performed the best (mean squared error: 0.22, r2: 0.91), while the optimized model only slightly improved on it (MSE: 0.20, r2: 0.93). For δ15N, we determined that Chinook δ15N values significantly increased by 0.6‰ following lipid extraction. Our results confirm a lipid normalization procedure that is broadly applicable to Pacific salmon, and supports streamlined analysis of δ13C and δ15N from a single untreated muscle tissue sample.

Dealing with lipid effects and lipid-extraction biases in δ13C and δ15N isotopic studies: a solution based on 28 marine invertebrate, fish and mammal species

“…However, our data showed, as expected based on published studies (e.g., Logan et al 2008, Lesage et al 2010, Groß et al 2021, Lerner & Hunt 2022, a linear relationship between  15 Nbulk and  15 Nlipid-extracted (Fig. 1d).…”

“…Paired t -tests were used to examine, for each species, the effect of lipid extraction on δ 13 C, δ 15 N and C:N. The difference between lipid-extracted and bulk aliquots was expressed as Δδ 13 C (δ 13 C lipid-extracted −δ 13 C bulk ) and Δδ 15 N (δ 15 N lipid-extracted - δ 15 N bulk ). 13 C discrimination is known to be related to C:N bulk in a log-linear relationship (Logan et al 2008, Lesage et al 2010, Fischer-Rush et al 2021, Lerner & Hunt 2022), and this relationship was tested here.…”

“…Because  15 Nbulk is expected to be linearly related to  15 Nlipid-extracted (Logan et al 2008, Lesage et al 2010, Groß et al 2021, Lerner & Hunt 2022, this relationship was tested here (see Results) to accurately retroestimate  15 Nbulk from  15 Nlipid-extracted.…”

“…This procedure was based on the assumption that an absence of nitrogenous components in chemical solvents would prevent alteration of δ 15 N values. However, this assumption was refuted by empirical assessments, with multiple studies reporting significant and uneven discrimination, resulting in 15 N enrichment, across species following lipid extraction (Choy et al 2016, Clark et al 2019, Lerner & Hunt 2022, but see Cloyed et al 2020). An alternative to lipid extraction has been to analyze carbon and nitrogen isotopes using separate aliquots (Logan et al 2008, Lesage et al 2010).…”

“…But thus far, little guidance exists for investigators in making that choice. Also, equivalent models for reverting lipid-extracted δ 15 N values back to their bulk un-extracted values have been published (Logan et al 2008, Lesage et al 2010, Groß et al 2021, Lerner & Hunt 2022), making older datasets, where lipid-extraction was the norm, suitable for long-term monitoring of environmental or trophic change.…”