Large-scale reptile extinctions following European colonization of the Guadeloupe Islands

The global biodiversity crisis is one of the most severe impacts of humans on Earth’s ecosystems (1). Assessments of present-day extinction rates by the International Union for Conservation of Nature indicate the disappearance of at least 1.07% of Earth’s vertebrate species over the past 500 years (2), a total that is probably grossly underestimated given the lack of subfossil data and poor documentation of taxa for most regions (3). Current estimates suggest, conservatively, an extinction rate increase in hundreds or even thousands of times over natural background extinction frequency (3, 4). While reconstructing extinction patterns requires fossil and archaeological data, systematically studied regional archaeological and paleobiological datasets are few and far between. When historical data are present, they sometimes demonstrate that anthropogenic impacts extend very early, as seen in the case of the human-mediated impacts on megafauna beginning c. 50,000 years ago (5). However, except in a few cases, for example, the study of prehistoric bird extinctions in the Pacific (6), most available archaeological records of extinction are based on single sites or poor regional records. Furthermore, as in the field of ecology (7), the focus in archaeology and paleontology is very often on larger or charismatic faunal species, such as ungulates or large carnivores. Systematic study of neglected fauna such as reptiles and amphibians across multiple sites within a region has been a relatively limited undertaking to date.

Squamate reptile fossil remains offer an important opportunity to examine underexplored aspects of long-term anthropogenic impacts to biodiversity. Lizards and snakes likely arose between the Early Jurassic and Late Triassic and diversified worldwide, coming to occupy a wide range of ecological niches (8). As a function of the broader “taxonomic chauvinism” that has biased against the study of reptiles, amphibians, and invertebrates (7), the extinction rates (9) and ecological functions (7) of squamates have attracted less attention than those of other taxa. However, there is an increasing recognition that reptiles have an important role to play in ecosystem function, for example, in seed dispersal and pollination, nutrient transport, and as ecosystem engineers (10). Their loss, particularly from the tropics where much of their species diversity is concentrated (11), has implications for the maintenance of ecosystem services and broader biodiversity. While modern reptile extinctions are less well documented than other taxa (12), recent efforts to study anthropogenic impacts to these taxa suggest that at least one in five reptilian species is currently under threat (10). Understanding of the longer-term history of human impacts to reptilian fauna is strongly lacking.

The Caribbean region offers an exceptional opportunity to explore long-term human impacts on snakes and lizards. The Caribbean is a biodiversity hot spot in which present-day vertebrate faunas nonetheless represent only a remnant of richer species assemblages present in the past (13–15). While the causes of Caribbean extinctions have been debated for decades (16–18), recent studies point to Late Pleistocene climate change (19) and, particularly, human colonization and activities (14–18) as leading factors. Nonetheless, systematic archaeological and paleobiological research across the region has been patchy, and with the exception of a few studies of herpetofauna, research on terrestrial nonvolant biodiversity in the Caribbean has focused overwhelmingly on mammalian biodiversity (13, 14, 17, 18, 20). However, mammalian taxa represent only a very minor component of the native terrestrial nonflying fauna of the region (14). For example, in the Lesser Antilles, native nonvolant mammals are represented by only a single taxon on most islands. Instead, most of the native terrestrial nonvolant vertebrate fauna of the Lesser Antilles is composed of squamates. To date, however, snakes and lizards in the Caribbean have seen minimal paleontological or zooarchaeological investigation (16, 21, 22), in part, because suitable deposits preserving squamate remains are either rare (in the volcanic islands) and/or situated in highly anthropized coastal areas where they are extremely susceptible to disturbance or destruction (23).

The Guadeloupe Islands in the Lesser Antilles offer an important exception to this trend and, therefore, present a prime opportunity to study the role of climatic and anthropogenic factors in the extinction of snakes and lizards in the Caribbean. Not only do the Guadeloupe Islands host cave deposits containing both Late Pleistocene and Holocene fossiliferous layers but they have also seen a series of excavations of dozens of archaeological sites across the islands over the past 30 years. Critically, these excavations all involved systematic screening of archaeological sediments using 2- to 3-mm sieves, as well as retention of all faunal material, enabling recovery and study of the bones of small taxa. These consistent practices offer a rare and exceptional opportunity for the systematic study of squamate remains across multiple sites in one Caribbean island archipelago. Beyond this, the Guadeloupe Islands also provide the opportunity to look at several phases of human activity. These islands were occupied by Indigenous fisher-forager-horticulturalists mainly originating from South America, and later by colonial populations from Europe, Africa, and Asia, whose impact on biodiversity can be compared and discussed. The suitability of Guadeloupe as a case study for exploring shifts in squamate biodiversity also, of course, stems from the fact that islands are useful model systems, offering circumscribed spaces where past evolutionary, and anthropogenic, processes can be explored (24, 25) and used to elaborate present-day biodiversity conservation strategies (26). The use of past data to set a baseline for human-induced extinction, predict future environmental impacts, and develop appropriate conservation strategies, particularly as part of the broader efforts of conservation paleobiology and applied zooarchaeology (27, 28), is especially relevant in an insular context. Islands are some of the most imperiled ecosystems on the planet (6, 29, 30), their fragility offering clues as to what the future holds in more resilient continental ecosystems.

To examine how humans have transformed the biodiversity of snakes and lizards in the Guadeloupe Islands from a deep-time perspective, we analyzed fossil squamate remains recovered from most of the known archaeological and paleontological deposits of the archipelago (Fig. 1 and table S1). Squamates represent not only the large majority but also, by far, the most diverse of the native vertebrate taxa of the Guadeloupe Islands, making them particularly suitable for studying transformations to the nonvolant vertebrate biodiversity of Guadeloupe through time. Through analysis of more than 43,000 individual squamate remains, we reconstructed the evolutionary history and diversity of these vertebrates over the past 40,000 years in relation to climate change and human-induced landscape modification.

Fossil remains attributable to 16 different taxa were identified across the 31 sites and six islands sampled in the study (table S2). Fossil data were classified into four temporal phases: (i) Late Pleistocene (32,000 to 11 650 B.P.); (ii) pre-anthropic Holocene (11,650 to 2450 B.P.); (iii) Indigenous habitation period (combining the Saladoid and Troumassoid periods) (2450 to 458 B.P.); and (iv) modern, encompassing the period after 458 B.P. until the present. The dataset obtained was used to reconstruct changes in squamate diversity on the different islands and across the Guadeloupe archipelago as a whole.

The data from Guadeloupe make it clear that major squamate extinctions followed European colonization of these islands, with at least 50 to 59% of taxa suffering extinction or extirpation in the centuries after 1492. Rates of post-European colonization squamate extinction across the individual islands of the Guadeloupe archipelago are likely many thousands of times higher than background rates, which have been modeled at 0.3 per million years for lineages of reptiles and amphibians in the West Indies (56). Lineage persistence times of millions of years through periods of major climate change, including the glacial cycles of the Pleistocene, emphasize the general resilience of taxa on the islands of the Lesser Antilles and the West Indies (56) until modern times and highlight the severity of post-European extinction in Guadeloupe. These results demonstrate that failure to consider fossil-derived data, especially regarding poorly studied lineages, probably contributes to a substantial underestimation of the magnitude of the human impact to global biodiversity.

While extinctions in the anthropic era are high, it is important to differentiate pre-Columbian and post-Columbian impacts on squamates in Guadeloupe. No squamate extinctions are recorded in the Indigenous habitation period layers of the sites investigated, and species diversity in our dataset increased in this era due to the introduction of at least two new lizard species (I. delicatissima and T. rapicauda) to the archipelago. This is in keeping with modern data suggesting that competition with introduced reptiles has little impact on native reptile species compared to other drivers (57), but this question remains difficult to investigate in the Caribbean (see Supplementary Text). The pre-Columbian anthropic era data for squamates are, however, in sharp contrast to findings for nonvolant land mammals in the broader insular Caribbean, one of the only oceanic-type island systems to have been colonized by these faunae (20). In the Late Quaternary, the West Indies were home to diverse species of endemic eulipotyphlan insectivores, megalonychid sloths, platyrrhine primates, and several families of often large-bodied rodents (14, 20), most of which experienced catastrophic collapse across the Holocene, probably in relation to human colonization and activities (14).

Unlike the extinction pattern of Guadeloupe squamates, which is short, sharp, and late, data for land mammals across the Caribbean tend to show a longer-term pattern of ongoing extinction across the Holocene [from c. 5 to 7 thousand years (ka)], albeit one that intensifies with the arrival of agriculturalists and, later, Europeans (13, 14). Caribbean bat extinctions began by 4 ka and were not climate caused (17). Cuba saw major extinctions of its land mammals in the past 2 ka, linked to the arrival of Ceramic populations (13). Archaeological data demonstrate that Indigenous populations also had a major impact on the marine ecosystems on which they were heavily reliant in the Virgin Islands and the Bahamas (58, 59). While taxonomically uneven data collection remains a problem, there is some evidence to suggest that taxonomic groups were differentially affected by early human populations. Large mammals and bats appear to have been among the first to disappear following human arrival in the Caribbean—although as more accelerator mass spectrometry radiocarbon dates are secured, it becomes clear that this was often a protracted process (14)—whereas small-bodied rodents and eulipotyphlan insectivores tended to survive to European times (14, 18).

The data from Guadeloupe, with its unique extended fossil record, allow us to further and more systematically explore extinction dynamics and the biology of resilience to human impacts. The relatively short extinction chronology (<500 B.P.) for the archipelago’s squamates points to a high degree of resilience to the kinds of impacts engendered by Indigenous populations. Pre-Columbian anthropogenic species introductions, which appear to have been so devastating in the colonial period [see also (15)], appear not to have had an impact on Guadeloupe squamates nor did pre-Columbian agricultural practices. The historical Guadeloupe dataset is also unique in permitting examination of more fine-grained extinction dynamics, enabling insights into the link between habitat preference, size, and extinction risk through time. This reveals, first, a strong link between squamate habitat preference and human impact, with terrestrial fauna showing greater susceptibility to extinction/extirpation. Our data indicate that arboreal taxa were significantly less susceptible to the environmental impacts of European colonization, with only 17% experiencing extinction/extirpation (when the very small lizards of the genus Sphaerodactylus are excluded). Our study also highlights a second factor influencing squamate extinctions/extirpation: weight/size of the species. At least 80% of the medium-sized taxa once present on Guadeloupe are now extinct or extirpated, whereas, in contrast, only 15% of the small taxa present in the Indigenous habitation era are no longer present. Model selection reveals that these two variables explain much of the observed extinction/extirpation pattern, whereas other factors (in particular, the presence of mongoose and the size of the island) appear less significant. This evidence reinforces observations previously made in Guadeloupe and the Caribbean more broadly on the basis of studies of modern lizards (45, 60) and mammals (14), linking habitat preference, body size, and extinction risk.

These trends in the data strongly point to a dominant role for recently introduced mammalian predators in Guadeloupe’s post-1492 extinction wave, coincident with findings elsewhere in the Caribbean (15). While dogs were present in Guadeloupe from the Saladoid period, along with several other introduced mammals (38), European colonial populations introduced several new mammalian predators, including cats and mongooses (H. auropunctatus) (44), that are known to be excellent squamate hunters, as well as other potential predators such as rats and raccoons (Procyon lotor). These predators are mainly terrestrial and are more likely to exploit medium-sized, ground-dwelling taxa that are easier to capture and that provide substantial protein intake. Several other authors have pointed to the important role of the mongoose, introduced originally for pest control, in the recent erosion of Antillean squamate biodiversity (61) and broader Caribbean endemic vertebrate biodiversity (15). This hypothesis would explain why, despite their large size (62), arboreal iguanas are still extant in the Caribbean today.

Introduced predators may also help to explain another trend in the Guadeloupe squamate data. Unexpectedly, the study did not reveal any major influence of island size on the survival of Guadeloupe snakes and lizards, although biogeographic models and fossil data (see Supplementary Text) predict that larger islands should be more resilient and have more areas of refuge from human impacts. In Guadeloupe, however, the least affected islands, Les Saintes and Petite-Terre Islets, are also the smallest. Tellingly, these islands were the least attractive for human settlers, and, critically, both are mongoose free (44), while Petite-Terre was never subjected to large-scale agriculture or notable habitat destruction and has been uninhabited since 1972 (31), as reflected in the low proportion (20%) of exogenous introduced taxa in its current herpetofauna.

While mongoose presence is important, it is, as highlighted by our models, clearly not the sole driver of squamate extinctions in Guadeloupe. A role for other introduced predators (e.g., cats and rats) should not be overlooked. Furthermore, introduced taxa are strong anthropization benchmarks (63) that are correlated with historic and modern land use for intensive agriculture. Impacts of intensive agriculture include habitat fragmentation and destruction, soil quality erosion, and extirpation of the primary insects consumed by terrestrial squamates (64). The widespread dry forest that likely hosted some of Guadeloupe’s extinct lizards (Leiocephalus and Pholidoscelis) is the most sensitive of all the island biomes to soil degradation (65) and was strongly affected in the Lesser-Antilles (66). These landscape transformations, similar to alien predators, may have had an important impact on biodiversity. Marie-Galante, which is the most heavily affected island in Guadeloupe, with an extinction/extirpation rate of squamate taxa above 70%, differs from the other islands by its early and intensive cultivation since the 17th century. In light of its small size, this likely left few areas of natural refuge. In contrast, the archipelago’s other islands remained mostly undisturbed until the 18th century (31).

Our study strongly highlights the need to prioritize study of squamates, as well as other species affected by taxonomic chauvinism (7), in both modern and fossil/archaeological contexts. Research on defaunation has highlighted the important loss of large-bodied mammals through body-size downgrading (30) and keystone species through trophic downgrading (67), but we know much less about size-differential loss and broader ecosystem impacts in reptiles, amphibians, and invertebrates. Increasing evidence nonetheless points to a tendency for larger squamates to fall prey more quickly than smaller ones to extinction processes in a number of island contexts (60, 62), and scientists are beginning to document the impact to ecosystems of such body size–biased species removal. For example, extirpation and extinction of larger-bodied lizards have been shown to affect seed dispersal and plant recruitment in recent studies in the Canary Islands (68). Preservation of larger species that ensure effective seed dispersal is crucial to the maintenance of functional seed dispersal services and associated genetic diversity, meaning that they should be an important conservation target (68). In Guadeloupe as elsewhere, smaller squamates appear to have been less susceptible to extinction and extirpation, but medium-sized lizards were actually more heavily affected than the largest species. This size-based extinction pattern may reflect the hunting behaviors of introduced mammalian predators, which selected prey size suitable to their own size and needs. Our findings highlight the need to target medium-sized species in conservation efforts in the Guadeloupe Islands.

The relatively late extinctions of snakes and lizards in Guadeloupe also have important broader conservation implications. Our dataset suggests, minimally, several thousand years of human occupation without impact to squamates in Guadeloupe [and as much as almost 5000 years if pollen and charcoal data are correct in indicating earlier human occupation (34)]. The persistence of Guadeloupe squamates through major periods of Pleistocene climate oscillation and Indigenous island occupation that, elsewhere in the Caribbean, saw substantial faunal extinctions highlights the resilience of these animals to disturbance. On the other hand, Guadeloupe’s only nonvolant native mammal, the medium-sized rodent, A. rayi (20), also persisted through to colonial times, only going extinct in the historical era, and marine ecosystems in the Lesser Antilles (69), including Guadeloupe (48, 70), also appear to have been relatively resilient to pre-Columbian exploitation. Whatever the nature of pre-Columbian environmental impacts, however, they appear to have been massively ramped up following European colonization, when squamate assemblages on many islands in the archipelago saw marked decline in terms of both taxonomic and morphological diversity (71). These declines should cause serious concern. Squamates do not appear to be a sensitive group of fauna in Guadeloupe, and we should accordingly see their disappearance as a warning to humanity about the perils of continued landscape destruction and disturbance that threatens to undermine not only global biodiversity but also our planet’s ability to support human societies.

The fossil record offers an opportunity to understand and learn from past human practices, to examine which human activities offered sustainable lifeways and which were particularly destructive, and to draw on this information to create realistic and data-driven strategies for mediating anthropogenic impacts in a diverse variety of cultural and ecological contexts (5, 27, 28, 72). The archaeological record of extinctions from Guadeloupe gives us a clear baseline carrying disturbing news and reason to be deeply concerned, but at the same time, it offers opportunities to explore solutions and to examine the long-term impact of diverse human lifeways. Agriculture per se does not seem to be harmful to Guadeloupe’s squamates, but the shift from diversified agriculture to agricultural specialization and monoculture crops after French colonization in 1635 CE likely negatively affected biodiversity through losses in plant and animal diversity, reduction in pollination services, and changing hydrology, for example (73). Perhaps the greatest human impact to squamates, however, derives from the introduction of a variety of “Old World” human commensal species (15, 46). This points to a strong need to prioritize research and conservation measures relating to invasive predators to mitigate global biodiversity loss (74).

To examine changes to squamate biodiversity over the two last millennia, the squamate osteological remains from 31 Guadeloupe archaeological and paleontological deposits spanning the Late Pleistocene through to modern times were sorted and studied (by C.B. and S.B.) as part of the present study (Fig. 1, table S1, and figs. S1 and S2). Pleistocene and Holocene layers predating the arrival of the first human populations in Guadeloupe around 5000 years ago (34) were only available in the paleontological deposits of Marie-Galante Island (Fig. 1, 27 to 29). Pre-Columbian cultural period remains were available for all sites studied. Full details regarding dates, references, and repository locations for each studied deposit are indicated in the Supplementary Materials (table S1). A total of 43,790 bone remains of squamates were isolated and taxonomically identified from the selected assemblages (table S2). These faunal remains had been previously collected from archaeological deposits using 0.5- to 3-mm mesh size sieves and were all retained for further study. Taxonomic identifications were obtained by consulting osteological criteria of taxonomic relevance and by comparison with large sets of modern specimens from the Comparative Anatomy, and Reptiles and Amphibians collections of the Muséum National d’Histoire Naturelle (Paris, France), the University of Bordeaux (Bordeaux, France), the Edgar Clerc Museum (Le Moule, Guadeloupe), the Museum of Comparative Zoology (Harvard, USA), and the Florida Museum of Natural History (Gainesville, USA). Details of these identifications for most taxa were previously discussed in a series of papers (see Supplementary Text for details). The taxonomic attributes discussed in these papers provided the basis for the taxonomic identifications made in the present study. The chronological context of identified bone remains was established on the basis of archaeological data to reconstruct the evolutionary and anthropogenically altered history of Guadeloupe’s squamate diversity during the past millennia. Because of the diversity of archaeological contexts studied and the lack of chronological resolution for most archaeological deposits, we grouped pre-Columbian cultural era finds (combining the Saladoid and Troumassoid periods) into a single chronological bin. Given that we found that most Guadeloupe squamate species went extinct after these periods, this did not significantly affect our study findings. These data were supplemented with published historical and modern data (see Supplementary Text for details).

To better understand the causes of the extinction/extirpation events observed in the fossil archaeological record, we compared alternative models following an approach similar to that previously used to investigate the extinction risks faced by New Zealand lizards (75). Briefly, as a response variable, we considered the status of reptile taxa on the different islands defined as extinct (i) or extant (ii) (see table S3). We only included taxa for which past occurrence was clearly demonstrated (see Results) and excluded putative occurrences hypothesized solely on the basis of biogeographic reconstruction. We considered the following explanatory variables: (i) adult body mass of the species in grams (ln transformed), (ii) habitat preferences (terrestrial or arboreal), (iii) activity phase (nocturnal or diurnal), (iv) size of the island in square kilometers (ln transformed), (v) maximum altitude of the island in meters (ln transformed), (vi) whether the island is occupied by mongooses (76), and (vii) number of human inhabitants per square kilometer in 1950 (77, 78). We did not consider the body size (length) of the reptile species, as such a variable is not comparable in snake and lizard species and is strongly correlated with their weight. The presence of mongoose on the islands is associated with intensive past and present cultivation of sugar cane, as this animal was originally introduced to eliminate pests from these fields (76). We did not consider among the explanatory variables other predators known to strongly affect reptile faunas such as cats and rats since these are present on all Guadeloupe islands. The analysis of the data is challenging because it contains several levels of nestedness and interdependence. To take the phylogenetic relationship between species into account, we used the phylogeny from Pyron et al. (79) to test for phylogenetic signals. Since we had several observations per species (i.e., the different islands), we added short terminal branches so that the correlation within a species was high. We then built a relevant phylogenetic correlation structure that was used to fit models that consider several variables. Since we considered a non-Gaussian response, we used GLMMs. We opted for the glmmPQL function in the package MASS because it enables the inclusion of a correlation structure [similar to phylogenetic generalized least squares (PGLS)] while being able to model non-Gaussian responses (unlike PGLS). It also enables the inclusion of some predictors as random effects, which is relevant here since the effects of the islands and species variables can be considered random. This framework uses a penalized quasi-likelihood, which does not enable the calculation of information criteria such as Akaike information criterion (AIC) or Bayesian information criterion (BIC). Therefore, we tested the significance of the effects with standard tests on the coefficients. For comparison, we also fitted the same models with a GLM with a binomial response. As there is strong multicollinearity in our dataset, the results obtained should be considered with caution but are still relevant for a comparison with our GLMMs. The R package MuMIN v.1.43.17 executed within the R v. 3.6.3 software environment was used for automated GLMM selection using, as criterion, the BIC adjusted for small sample sizes. All possible combinations of the six explanatory variable retains were considered. Nagelkerke’s R² was used as a measure of the explanatory power of the best models.

This paper is dedicated to our friend and colleague A.T. who passed away in 2019 before seeing the completion of this work in which she was involved. We thank the Guadeloupe Department of Archaeology (Regional Office of Cultural Affairs, French Ministry of Culture), C. Stouvenot, and D. Bonnissent for allowing access to the archaeological material investigated in this study. We also wish to thank all the excavation directors of the sites investigated here: P. Bodu, D. Bonnissent, M. Boudadi-Maligne, D. Cochard, F. Casagrande, A. Chancerel, E. Clerc, C. Colas, P. Courtaud, A. Delpuech, P. Fouéré, J. Gagnepain, C. Hofman, M. Hoogland, M.-A. Paulet-Locart, T. Romon, N. Sellier-Segard, N. Serrand, C. Stouvenot, and M. de Waal. We are grateful to A. Crowther for discussion of botanical evidence from the Caribbean and to R. Fernandez who provided useful feedback concerning our modeling approaches. We are also grateful to C. Giovas and R. Fernandes for valuable comments that helped us to improve this manuscript. This is publication ISEM 2021-052. Funding: Some of the results presented in this study were acquired within the framework of the CNRS-INEE SEEG and the FEDER program 2016–2020 “Ecosystéme insulaire tropical, réponse de la faune vertebrée terrestre à 6 000 ans d’anthropisation de la Guadeloupe” (ECSIT), with support from a European PO-FEDER grant, the Guadeloupe Regional Council, the DAC of Guadeloupe, and the CNRS. This work was mainly funded a Ph.D. Grant granted to C.B. by the French Ministére de la recherche et de l’enseignement supérieur (2013-2016). C.B. was funded by a grant provided by the FYSSEN Foundation, and N.B. and C.B. are supported by the Max Planck Society. S.G. has received funding from the Muséum National d’Histoire naturelle “ATM Biodiversité actuelle et fossile 2005–2010: Quaternary Vertebrate peopling of the Guadeloupe archipelago, Lesser Antilles” for the initial data acquisition and research. Author contributions: C.B. designed the study, collected the data, identified all discussed bone remains with the assistance of S.B., and performed data analysis with the assistance of E.P. C.B. and N.B. wrote the paper. E.P., S.G., O.L., A.L., I.I., A.T., and S.B. took part in the analysis of the data and added editorial input. Competing interests: The authors declare they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.