Stable isotope analysis is increasingly used to assess diet and trophic positions of animals. Such assessments require estimates of trophic discrimination factors (TDF — offset between the isotopic composition of diet and animal tissues), with imprecise applications of TDFs leading to biased conclusions in resource use. Because TDFs are unavailable for most species, ecologists often apply values from taxonomically similar species or use trophic step increases of about 1‰ for carbon (TDF-δ¹³C) and 3‰ for nitrogen (TDF-δ¹⁵N). Such practices may be inaccurate since TDFs vary greatly, even within a species. To better understand the factors that influence TDFs, we conducted a meta-analysis of TDF-δ¹³C and TDF-δ¹⁵N for mammals and quantified variation in relation to consumer type (herbivore, omnivore, carnivore) and diet source (C3-based, C4-based, marine-based, mixture). Additionally, to guide TDF choice, we used an isotopic dataset of small mammal tissues and diet items to assess how predicted dietary contributions vary with TDFs estimated using (1) taxonomic relatedness, (2) consumer type and diet source, or (3) values derived from wild animals eating natural diets. Our meta-analysis revealed that metabolic routing and interactions between consumer class, dietary source, and the protein versus energy content of diets best explained variation in TDF-δ¹³C values (-1.5 to 7.3‰), whereas consumer class best explained variation in TDF-δ¹⁵N values (-0.5 to 7.1‰). Our test of methods to estimate TDFs indicated that ecologists should avoid relying on taxonomic relatedness when selecting TDF-δ¹³C, as mixed-diet lab studies may cause misleading results for herbivores and omnivores. Additionally, field-derived estimates could help fill TDF gaps where diets within a consumer class are absent. Overall, we suggest that using standard TDF trophic step values should be abandoned, as feeding studies are often poor proxies for natural diets, particularly for herbivores and omnivores. Instead, we provide recommendations on how to select TDFs, along with a range of TDF-δ¹³C and TDF-δ¹⁵N values depending on diet source, consumer class, and tissue type. Use of these more refined recommendations and TDF values in isotopic assessments will improve estimates of diets and trophic interactions in natural systems, leading to a better understanding of ecological interactions and communities.

See manuscript for details on data collection.