An Isotopic and Morphometric Examination of Island Dogs (Canis familiaris): Comparing Dietary and Mobility Patterns in the Precolumbian Caribbean

Abstract

In precolumbian insular Caribbean archaeological sites, domestic dog (Canis familiaris) remains have been recovered from varied contexts, such as formal burials, in refuse deposits, and as modified artifacts, indicating their complex and multifaceted role within indigenous societies. In this study, isotopic and morphometric analyses provide biochemical and morphological correlations to assess this differential treatment. We examined collagen values (n = 21) of carbon (δ¹³C_co) and nitrogen (δ¹⁵N), and enamel values (n = 81) of carbon (δ¹³C_en), oxygen (δ¹⁸O_en), and strontium (⁸⁷Sr/⁸⁶Sr) of dog remains from 16 precolumbian sites. Five comparative parameters were used to assess dietary variations between different groups: buried versus nonburied, local versus nonlocal, Greater versus Lesser Antilles, chronology, and modified versus unmodified remains. The only statistically significant difference in diets was between local and nonlocal dogs. Sufficient data were available to conduct isotopic mixing models using the FRUITS statistical program on four individuals for which depositional and morphological data were available. Results of dietary modeling indicate an unexpectedly heavy reliance on plant foods consistent with intentional feeding. This approach highlights the utility of combining isotope analysis, dietary models, morphometrics, and depositional context to provide comprehensive biographic overviews of individual animals.

En del Caribe insular precolombino, los restos de perros (Canis familiaris) se recuperan de diversos contextos, como entierros, en depósitos de basura y artifactos modificados, lo que indica su papel multifacético dentro de las sociedades indígenas. En este estudio, los análisis isotópicos y morfométricos intentan proporcionar correlaciones bioquímicas y morfológicas para evaluar este tratamiento diferencial. Valores de colágeno (n = 21) de carbono (δ¹³C_co) y nitrógeno (δ¹⁵N), y valores de esmalte (n = 81) de carbono (δ¹³C_en), oxígeno (δ¹⁸O_en) y estroncio (⁸⁷Sr/⁸⁶Sr) de perros han sido examinados de 16 sitios precolombinos. Cinco parámetros evalúan las variaciones dietéticas entre los diferentes grupos: enterrados versus no enterrados, locales versus no locales, Antillas mayores versus menores, cronología, y restos modificados versus no modificados. Los resultados indican que la única diferencia significativa en las dietas fue entre locales y no locales. Los datos estaban disponibles para llevar a cabo modelos de mezcla isotópica utilizando el programa estadístico FRUITS en cuatro individuos. Los resultados del modelado dietético indican una dependencia inesperada de los alimentos vegetales, consistente con la alimentación intencional. Este enfoque destaca la utilidad de combinar análisis de isótopos, modelos dietéticos, morfometría y contexto de depósito para proporcionar descripciones biográficas de los animales.

Cross-culturally, dogs (Canis familiaris, Linnaeus 1758) are often valued members of societies, as expressed in the commonality of their inhumation, a practice rarely seen with other animals (Russell 2011). Within the precolumbian Caribbean, the role of dogs was that of a hunting aid, a companion, and potentially a food source (Grouard et al. 2013; Las Casas 1876 [1561]). Earlier studies on precolumbian Caribbean dogs assessed morphological differences (Grouard et al. 2013) and provided isotopic evidence of dietary and mobility patterns (Laffoon et al. 2015, 2019). This article provides additional morphological data and is the most expansive investigation of Caribbean C. familiaris isotopic values conducted to date.

Multi-isotopic analyses of precolumbian dogs in the insular Caribbean have primarily focused on enamel samples, with some collagen samples from El Flaco and El Carril in the Dominican Republic (Shev 2018) and from Punta Candelero in Puerto Rico (Pestle 2010). Laffoon and colleagues (2019) incorporated strontium (⁸⁷Sr/⁸⁶Sr) and carbon (δ¹³C) isotope values from El Flaco and El Cabo in the Dominican Republic and from Morel and Anse à la Gourde in Guadeloupe, demonstrating similarities between human and dog diet and mobility patterns at these sites. These human–dog linkages were previously reported globally in diverse archaeological contexts, leading to the proposition of a “canine surrogacy approach” in which dogs could be used as an isotopic surrogate for human remains (Guiry 2012).

An earlier study by Grouard and coauthors (2013) assessed the morphology of buried dogs from the region, indicating some differences in estimated withers heights (Grouard et al. 2013). These data suggest there may have been more than one variety of dog within the insular Caribbean. This notion may be supported by Las Casas, who reported the existence of two different breeds that may have received different treatment by humans (1876 [1561]). In our study, we assessed the withers heights of dogs from El Flaco and El Carril to determine whether there are correlations between morphology and differential treatment in the form of formal burials or dietary regimes.

We also compared C. familiaris isotope values from Puerto Rico, the Dominican Republic, Cuba, Saint-Martin, Guadeloupe, Barbados, and Grenada (Figure 1; Supplemental Table 1) according to five criteria: local versus nonlocal, Early Ceramic (500 BC–AD 600) versus Late Ceramic Ages (AD 600–1500/1600), buried versus nonburied individuals, Greater versus Lesser Antillean, and modified (e.g., pendants) versus unmodified remains. Where applicable, published morphometric data permitted comparisons between, diet, localness, and morphology. Additionally, a dietary mixing model was applied to four buried dogs, providing a detailed biography of these individuals. Estimated withers heights provide morphological evidence for the existence of two native “breeds” of dogs; details regarding FRUITS mixing models are further addressed in the online supplement (Supplemental Text 1).

Figure 1. Map of precolumbian sites from which samples were acquired. (1) Cueva Belica, Cuba; (2) El Flaco, Dominican Republic (D.R.); (3) El Carril, D.R.; (4) Playa Grande, D.R.; (5) Cueva de Berna, D.R.; (6) El Cabo, D.R.; (7) Manantial de Cabo san Rafael, D.R.; (8) Punta Candelero, Puerto Rico; (9) Sorcé, Vieques; (10) Hope Estate, Saint-Martin; (11) Cathédrale de Basse-Terre, Guadeloupe; (12) Morel, Guadeloupe; (13) Anse à la Gourde, Guadeloupe; (14) Silver Sands, Barbados; (15) La Poterie, Grenada; (16) Pearls, Grenada.

A total of 81 teeth and 24 bone collagen samples of C. familiaris were sourced from precolumbian sites throughout the Caribbean, including from a dog burial at El Flaco (Supplemental Table 1), and were analyzed by Shev (2018). Methods of stable isotope analysis, morphometric analysis, and FRUITS dietary mixing models are published extensively elsewhere and are presented online (Supplemental Text 1).

Results

Withers Height

Withers height was estimated for dogs from El Flaco (n = 2) and was added to Grouard and coauthors’ (2013) data (Supplemental Table 2).

Collagen Isotope Data

In total, 21 samples yielded high-quality collagen. An overlap in isotope values can be observed in all parameters of examination (Figure 2a). Further details on sample conditions and results are provided in the online supplement (Supplemental Text 1).

Figure 2. (a) Bivariate plot of dog δ¹⁵N and δ¹³C values from El Flaco and El Carril, Dominican Republic; Morel and Cathédrale de Basse-Terre, Guadeloupe; and Hope Estate, Saint-Martin (Shev 2018); (b) chart showing the estimated bioavailable ⁸⁷Sr/⁸⁶Sr local ranges (represented by boxes) of each site/island and whether each sample fits into or outside these ranges.

Enamel Isotope Data

Local versus Nonlocal

Thirteen of 50 samples possess strontium isotope ratios interpreted as nonlocal (Figure 2b) and showed relative enrichment in δ¹³C_en values (local, μ = −11.26‰, Mdn = −11.28‰; nonlocal, μ = −10.66‰, Mdn = −10.99‰). The Greater Antillean nonlocal δ¹³C_en mean value was 0.61‰ higher than in locals; in the Lesser Antilles the nonlocal δ¹³C_en value was 0.7‰ higher (Figure 3a). A t-test confirmed a significant difference in means (t = 2.2638, p = 0.029079). Excluding outliers larger than one standard deviation (critical t = 2.0484, p = 0.021275) also demonstrated significant differences. Details regarding limitations in Caribbean isoscapes are available online (Supplemental Text 1).

Figure 3. (a) Box plot showing median δ¹³C_en values (horizontal line), inner quartiles (box), value ranges (whiskers), and outliers (dots) for all comparative parameters; (b) FRUITS dietary mixing model results for El Flaco (FND 2270), Morel (FND 2727 and FND2729), and Cathédrale de Basse-Terre (CBT5002).

Burials versus Nonburials

δ¹³C_en values were available for 16 buried and 28 nonburied dogs, showing broad similarities between groups (burials, μ = −10.86‰, Mdn = −10.99‰; nonburials, μ = −11.07‰, Mdn = −11.17‰) and demonstrating no significant dietary differences between buried individuals and those that were not. There was no correlation between localness and the propensity for burial: 21.4% of buried and 26.7% of nonburied individuals were nonlocal.

Greater versus Lesser Antilles

Diets of dogs from the Greater (n = 48) and Lesser Antilles (n = 25) were similar (G. Antilles, δ¹³C_en μ = −11.11‰, L. Antilles, δ¹³C_en μ = −10.88‰; G. Antilles, Mdn = −11.17‰, L. Antilles, Mdn = −11.07‰). Given the disparity in sample numbers between the two regions, we ran a t-test (critical t = 1.9939, p = 0.33867), which confirmed there was no significant difference in mean values.

Early (500 BC–AD 600) versus Late Ceramic Age (AD 500/600–1500)

There were no significant differences in means (Early, δ¹³C_en μ = −10.83‰; Late, δ¹³C_en μ = −11.12‰). Both groups demonstrated considerable overlap.

Modified versus Unmodified

No dietary differences were observed between modified (n = 22) and unmodified remains (n = 44). Both groups exhibited similar values (modified, δ¹³C_en μ = −11.12‰, Mdn = −11.07‰; unmodified, δ¹³C_en μ = −10.98‰, Mdn = −11.08‰). The vast majority of modified remains (n = 22, 95.7%) were nonlocal, with the exception of a perforated canine from Manantial del Cabo de San Rafael (⁸⁷Sr/⁸⁶Sr = 0.7092, local range = 0.70914–0.70924).

Dietary Mixing Models

The results of the FRUITS modeling are presented in Figure 3b and Table 1 (see also Supplemental Text 1). They suggest that these dogs mainly consumed plants foods with modest amounts of C₄ plants (likely maize), possibly indicating intentional feeding. Of these, the two nonlocal dogs (FL FND2270 and MO FND2729) exhibited relatively higher proportions of marine protein consumption.

Table 1. Dietary Estimates (%) for Four Caribbean Dog Specimens Derived from the FRUITS Dietary Mixing Model.

Discussion

We found overlapping mean values in four of the five comparative parameters. The values are similar to those of humans from the region, likely reflecting broad-spectrum diets comprised of C₃ plants supplemented with C₄ crops, as well as terrestrial and marine proteins (Laffoon et al. 2019; Pestle and Laffoon 2018). These data suggest that dogs are an effective isotopic surrogate in the precolumbian Caribbean.

The only significant difference in mean δ¹³C values that we found is between nonlocals and locals. Of all samples (n = 49), 22.4% were deemed nonlocal. They likely either migrated alongside humans to new locations or were exchanged between different human groups. Similar mobility patterns have been suggested for Anse à la Gourde and Morel, where 30% of studied dogs were nonlocal (Laffoon et al. 2015, 2019; Plomp 2013).

Ethnographies of indigenous peoples from the South American lowlands provide insight into how dogs may have been treated in the past (see Koster 2009). The exchange of hunting dogs by renowned dog breeders such as the Waiwái of Guyana and Brazil (Howard 2001:248) may serve as a useful analogy. Ethnohistoric sources describe the use of dogs in Hispaniola as valued hunting aids, although it is possible that dogs were also a food source (Las Casas 1876 [1561]:341). However, given the sparse evidence of butchery or cooking, the consumption of dogs may have been restricted to times of food scarcity (Wing 2008). It is unlikely that dogs were traded as food; it is more probable that they were migratory companions or prized dogs exchanged between communities.

One nonlocal specimen from El Carril (FND 30) had an ⁸⁷Sr/⁸⁶Sr ratio (0.7090) denoting a likely coastal origin. This sample also had the most enriched δ¹³C_en (−8.8‰) and δ¹⁸O_en (−1.8‰) values of any dog from Hispaniola (Shev 2018). This individual likely subsisted on a diet rich in marine proteins, with higher oxygen values indicating natal origins in an arid, low-altitude, or coastal region (Wang et al. 2016). Two nonlocal dogs from Silver Sands in Barbados demonstrated similarly high δ¹³C_en values (−8.8‰ and -8.6‰), whereas one exhibited an ⁸⁷Sr/⁸⁶Sr ratio (0.7075) suggesting possible natal origins from a nearby island. According to the FRUITS modeling, two nonlocals (FL FND2270 and MO FND2729) consumed higher proportions of marine proteins, indicating that some individuals that were exchanged or migrated alongside humans were possibly “sea dogs” consuming foods such as pelagic fish.

Conclusion

Broad similarities and trends can be seen in the paleodietary and mobility signatures of dogs analyzed throughout the insular Caribbean, regardless of time period, burial context, and location. At a regional scale, geographic patterning of isotopic values of dogs appears to mirror that of humans for the most part. The most significant disparity in diet occurs between local and nonlocal dogs, which is statistically significant enough to merit consideration. Nonlocal dogs likely received different foods than did locals, perhaps reflecting differences in social value or spatially structured foodways between certain individuals. The findings from the dietary mixing model suggest that the two nonlocals had diets higher in marine protein than the two local dogs; however, more data are needed to accurately assess whether there is a broader correlation between mobility and enriched isotopic values.