Differentiating wild from captive animals: an isotopic approach

Data source and compilation

We searched peer-reviewed publications in the Web of Science database (https://clarivate.com/webofsciencegroup/solutions/web-of-science/) from 1945 to 2021 using the terms “isotop*” AND “wild OR free-rang*” AND “captiv* or farm*” as a topic. We added the search terms “NOT ‘chicken OR hen* OR cattle OR pig’” to exclude domestic animals from the results. The search returned 295 hits, initially sorted based on the title, keywords, and abstract. First, we considered studies of stable isotopes involving any nondomestic vertebrate species. In a second instance, we selected only research related to carbon, nitrogen, hydrogen, oxygen, or sulfur isotopes that met one of the following criteria: (1) used stable isotope to differentiate wild from captive animals; (2) conducted isotopic analysis in wild and captive individuals of the same species in the same study. Paleontological publications and those studies exclusively using compound-specific isotope analysis (CSIA) techniques were excluded. We also excluded studies that did not show basic isotopic statistical information, such as average, standard deviation, or error. Data were extracted only from original research papers rather than those found in reviews or meta-analysis studies to avoid duplicates. After these two filtering steps, 47 studies remained to be analyzed in this review (Table S1). We also checked the references list from the main studies and reviews to ensure that all relevant papers were included.

Data were initially collected from the texts and tables of articles. When they were not or were only partially available, we contacted the authors asking for the missing or raw data. As a last resort, we estimated isotopic values from the figures, when available, using PlotDigitizer software, version 2.1.1 (PlotDigitizer, 2022).

Data and metadata structure

We selected variables related to the taxon classification, biology and morphology of animals, samples data (tissue, geographic location, year, and period), rearing system, isotopic records (values of mean, standard deviation, minimum and maximum, range), methodological records from the isotope analyses (analytical error, lipid extraction, and international reference material), and identification of publication were registered (Table 1 and Table S1). Regarding the rearing system, we classified the animals as “wild” or “captive.” On some occasions, the origin of the samples was uncertain; thus, we followed the authors’ conclusions about their wild or captive origin. All stable isotope results are expressed in the conventional delta (δ) notation, in units per mil (‰).

We used the systematized metadata of our database to present an overall picture of the studies analyzing stable isotopes in wild and captive animals. Additionally, we used the isotopic ratios’ mean, standard deviation, and range of wild and captive animals from the database to evaluate general differences and similarities. For this, we performed Student’s t-test for differences in δ¹³C, δ¹⁵N, δ¹⁸O, δ²H, and δ³⁴S considering the whole database. We also tested for differences between wild and captive animals for δ¹³C and δ¹⁵N per continent, taxon group, and/or dietary category. Finally, we performed Student’s t-test or one-way ANOVA to compare δ¹³C and δ¹⁵N in wild and captive animals considering the terrestrial taxon together (amphibian, reptile, bird, and mammal) and fish per continent. Due to the lack of suitable studies, we did not perform more detailed analyses for δ²H, δ¹⁸O, and δ³⁴S. In addition, despite the possible effects of different types of captivity on the isotopic signature of animals, we could not access details of captivity conditions in most studies, making it impossible to categorize them as suggested by Hudson (1989) or CITES. Thus, in the database, they were called “captivity” only.

To gain a deeper understanding, we also assessed each study individually. For studies that analyzed isotopic differences between the two environments, we relied on the results reported by the authors. For studies that measured but did not compare isotopic ratios in the wild and captivity, we tested for such differences (t-test, ANOVA, Wilcox-test, Kruskal-Wallis, and linear mixed model) when the original data were available. In this approach, we considered the different categories of wild or captivity when explicitly presented by the authors. Inferential tests were preceded by analyses of normality (Shapiro-Wilk test) and equality of variances. We performed statistical tests in the R platform, version 4.1.0 (R Development Core Team), with a significance level of 5% in all hypothesis testing.

Quality assurance and control

As part of the quality assurance, we carefully checked the data in different steps of the database building, trying to keep the information as close as possible to the original one. For example, we double-checked for mean, standard deviation, and range of isotopic ratios extracted by PlotDigitizer. We also double-checked the original source outliers detected by boxplots for the same variables.

When not provided, the geographic coordinates of the samples were estimated based on the authors’ most detailed geographic information (e.g., city, region, fishing area zones). Species names were kept as in the original publications, and as “fishes” are a paraphyletic group, we checked the taxonomic class of each species using Eschmeyer’s Catalog of Fishes (van der Laan & Fricke, 2023). Species diet classifications were checked using the R package Sider (Healy et al., 2018) or searched for in peer-reviewed papers.

An overall picture of the use of stable isotopes to differentiate wild and captive animals

The first studies using stable isotopes to distinguish wild and captive animals dated around two decades ago and aimed to evaluate the potential of SIA as a tool to differentiate wild from recent farm-escaped salmon (Salar salar, Dempson & Power, 2004) and mink (Mustela vison, Hammershøj, Asferg & Kristensen, 2004). Since then, the number of studies using this tool and its applications has been growing.

From the 47 selected publications, 33 used stable isotopes to distinguish wild from captive animals (Table S2), and 14 analyzed stable isotopes in wild and captive vertebrates for different purposes (Table S3). We found studies distributed in 37 countries (Fig. 1A), with the United States and Italy having more publications (n = 6 each). Regarding the continents, Europe (n = 21) and Africa (n = 3) had the largest and smallest number of studies, respectively.

Fifty-five species from different vertebrate taxonomic groups (mammals, birds, reptiles, amphibians, and fishes) were studied. The most studied groups varied on different continents: while studies on reptiles and amphibians were concentrated in Asia and Oceania, studies with fish were distributed worldwide (Fig. S1). All studies focused on one species (n = 36) or a group of species from the same taxonomic class (n = 11). Amphibia was the least representative group, recording in only one study, while fish accounted for over 46% of the publications (Fig. 1B).

Carbon and nitrogen stable isotopes in combination accounted for 80.8% of the studies, and 88.2% used at least one of those elements (Fig. 1C). Hydrogen, oxygen, and sulfur stable isotopes together account for less than 15% of the publications (Fig. 1C). Muscle and inert organic structures were the most analyzed tissues presenting in 46.81% and 42.55% of the studies, respectively. Muscle was used in 83.33% of the works with fishes, while inert tissues were used in 86.96% of studies involving the other taxonomic groups.

The experimental design, captivity condition, and description of the publications varied widely (Table S4). While some studies measured stable isotopes for only two categories (wild and captive) of the same species at the same moment and in the same geographic regions, others had a complex design, considering several types of wild or captivity various species, countries, period, and samples origin.

Patterns and potential of SIA in distinguishing between wild and captive animals

Analysis of the publications individually

To identify differences between wild and captive individuals, the studies performed graphical analyses of overlapping, discriminant tests, frequentist statistics (such as t-tests and ANOVA), or the combination of the last two statistical methods. Overall, 83.9% of the studies found significant differences or no overlapping among categories of wild and captive animals analyzed, while 16.1% distinguished among some modalities where authors considered different wild or captive conditions. In addition, no publication failed to differentiate between all categories of wild and captivity (Table S2). Accurately identifying the group to which individuals belonged (wild or captive) ranged from 58% to 100% when using discriminant tests.

From the studies using δ¹³C and δ¹⁵N, 90.3% and 89.3% reported significant differences between wild and captive animals, respectively. The δ¹³C was usually higher in the wild than in captive fishes. The opposite was found for the other taxonomic groups (Fig. 2). δ¹⁵N was consistently higher in wild birds and did not show clear trends in the other taxonomic groups (amphibians could not be evaluated since there was only one study) (Fig. 2). The studies used δ²H and δ¹⁸O when geographic variation was also involved (δ²H in birds and reptiles and the δ¹⁸O in fishes and amphibians). The δ³⁴S was used in two studies with birds or fishes, involving expected differences in the proportion of marine/terrestrial diet (Table S2).

Regarding the publications that measured but did not compare stable isotopes in wild and captive animals, we accessed the original data for 10 of 14 studies. We could distinguish between all (64.3%) or some (21.4%) categories in which authors used different wild or captive conditions. We found no differences between treatments in 14.3% of the publications (Table S5).

An overall picture of the use of stable isotopes to differentiate wild and captive animals

The development and understanding of methods to accurately identify the origin of an animal are crucial to ensure that wildlife farming fulfills its role as a conservation strategy (Phelps, Carrasco & Webb, 2014; Tensen, 2016; Lyons, Natusch & Jenkins, 2017; Natusch, 2018). Stable isotope analyses are an important biomarker of animal provenance. Here we summarize the main aspects of the use of stable isotopes as a tool to differentiate wild and captive animals.

Fish was the most studied group involving taxa of a single class (Actinopteri) and mainly species used for human consumption (e.g., European seabass, meagre, and different species of salmon). Only two fish species were considered globally threatened by IUCN and included in the CITES appendix (Tables S3). These findings suggest that beyond environmental impacts, identifying seafood origin involves concerns related to food safety, leading to labeling regulation about the rearing system in European Union and the United States (Commission Regulation (EC), 2001; Agricultural Marketing Service (AMS), 2009). However, few countries have clear requirements for the origin of the breeding system yet (El Sheikha & Xu, 2017).

Mammals, birds, and reptiles studies were mainly associated with ecological or forensic purposes, such as identifying the origin of a wolf population (Canis lupus) in an area where they were previously extinct (Kays & Feranec, 2011), the use of stable isotopes to identify the provenance of invasive alien species (Trachemis scripta) (Hill et al., 2020), or to detect crocodile lizard (Shinisaurus crocodilurus), short-beak echidnas (Tachyglossus aculeatus), and yellow-crested cockatoos (Cacatua sulphurea) laundering (van Schingen et al., 2016; Brandis et al., 2018; Andersson et al., 2021). Most studies in mammals, birds, or reptiles involved living species listed in the CITES appendices and relied on inert keratinous tissues, demonstrating SIA as a noninvasive biomarker of the rearing system.

Around 90% of the studies relied on δ¹³C and δ¹⁵N in animal tissues, which was not a surprise, considering the assumption that wild and captive animals have different diets, which is reflected in their isotopic ratios (e.g., Dempson & Power, 2004; Natusch et al., 2017; Hill et al., 2020). The δ²H and δ¹⁸O have been largely used to infer animals’ geographic origin and movement due to their pattern of variation in response to hydrological processes and the linkage with those in animal tissues (Hobson & Wassenaar, 2019). However, the tissue-environment relationship of δ²H and δ¹⁸O may also be affected by local factors, such as food-web relationships (Vander Zanden et al., 2016), physiology, and the proportion of water in the animals’ diet (Magozzi et al., 2019). The lack of dietary and trophic studies using δ²H and δ¹⁸O is also reflected in the wild-versus-captive studies. No research used δ³⁴S exclusively to distinguish wild from captive animals.

Patterns and potential of SIA in distinguishing between wild and captive animals