The morphological traits used to identify iguanas are described in Breuil (2013; 2016). Most consist of meristic and qualitative characteristics recorded on wild iguanas in the field on the islands of Saba and Montserrat, which are easily recorded from digital photographs taken by the authors. We also examined specimens at the Museum of Comparative Zoology (MCZ) at Harvard University, USA, based on photographs taken by Joseph Martinez and Corentin Bochaton (Muséum national Histoire naturelle, Paris).

We also reviewed photographs of Iguana found on the Internet using the Google Images search engine for the islands of St. Croix, St. Thomas (US Virgin Islands), for the coastal islets of Venezuela and for the vicinity of Cumana (Venezuela), regions known to be inhabited by melanistic iguanas, in relation to published data. Leighton et al. (2016) advocated the use of Internet images for taxonomic research to study spatial patterns in phenotypic traits that are objective, binary and easy to score, regardless of the camera angle. We used only photographs that were georeferenced to allow valid comparisons with the morphological characteristics of this new taxon (Breuil et al. 2019).

Collection and preparation of genetic material

Genomic DNA was isolated from 44 individuals from tissue, shed skin and/or blood samples using the QIAamp DNA Mini Kit (QIAGEN, Deutschland) and following the manufacturer’s recommendations (Table 1).

Fragments of the ND4 mitochondrial locus, encompassing the 3' end of the NADH dehydrogenase subunit 4 gene (ND4) and the tRNA genes histidine, serine and leucine (partial 5' end), were PCR-amplified using the primer pair and protocols of Malone et al. (2000). Sequence chromatograms were analyzed in SEQUENCHER (v5.3; Gene Codes Corp., Ann Arbor). The resulting ND4 sequences were aligned with GenBank sequences of iguanas from previous studies (Malone et al. 2000; Malone and Davis 2004; Stephen et al. 2013; Breuil et al. 2019). Sequence alignment was calculated using MAFFT (v7.18) (Katoh et al. 2005). For the ND4 analysis, the molecular data set included 21 iguanas (Table 1) from insular and continental origins.

The best nucleotide substitution model was chosen using JModelTest 2 (Darriba et al. 2012) under the Bayesian information criterion (BIC). Hierarchical relationships between individual ND4 genotypes/haplotypes were analysed using Maximum Likelihood (ML) and Maximum Parsimony (MP) tree reconstructions, and Median-Joining (MJ) haplotype networks. The evolutionary history was inferred using the ML method based on the Hasegawa-Kishino-Yano model (Hasegawa et al. 1985). The initial trees for the heuristic search were obtained automatically by applying the Neighbor-Joining and BioNJ algorithms to a matrix of paired distances estimated using the Maximum Composite Likelihood (MCL) approach, then selecting the topology with the highest log likelihood value. A discrete Gamma distribution was used to model evolutionary rate differences between sites [5 categories (+G, parameter = 0.3079)]. The tree was drawn to scale, with branch lengths measured as the number of substitutions per site. The MP tree reconstruction was obtained using the Subtree-Pruning-Regrafting (SPR) algorithm (Nei and Kumar 2000) with a search level 1, in which the initial trees were obtained by random addition of sequences (10 replicates). Node supports for tree topologies obtained by ML and MP methods were calculated as percentages of replicate trees in which the associated taxa had clustered in the bootstrap test (1000 replicates). Node supports were indicated with bootstrap support (BS ≥ 70). ML an MP analyses were conducted in MEGA7 (Kumar et al. 2016). The MJ haplotype network (Bandelt et al. 1999) was constructed using PopART (Population Analysis with Reticulate Trees) v1.7 (Leigh and Bryant 2015) with the epsilon parameter set to zero.

Analyses based on microsatellite molecular markers included 43 individuals of both insular and continental origins (Table 1). A panel of 16 microsatellite markers was used and amplified as described previously in Valette et al. (2013) and Vuillaume et al. (2015). Individuals were grouped into five populations according to their sampling localities (i.e., four islands: Montserrat, Saba, Saint Lucia and Grenadines as well as the continental population from French Guiana).

We tested the departures from Hardy-Weinberg expectations and linkage disequilibria using exact tests (1200 permutations) as implemented in the fstat software ver. 2.9.3.2 (Goudet 2001). Significance levels were adjusted for multiple tests using Bonferroni’s standard correction (Rice 1989). We calculated the genetic polymorphism at all loci for each population by computing the allelic richness (Ar), the expected heterozygosity (He) and Fis (Weir and Cockerham 1984) using fstat ver. 2.9.3.2 (Goudet 2001) with 1,200 permutations.

We calculated paired Fst values between populations (Weir and Cockerham 1984) using fstat ver. 2.9.3.2 (Goudet 2001). Their associated significance was computed and tested using global tests implemented in fstat ver. 2.9.3.2 (Goudet 2001). Significance levels were adjusted for multiple tests using Bonferroni’s standard correction (Rice 1989). In addition, we assessed relationships between populations with a Factorial Correspondence Analyses (FCA) based on individual genotypes and using the FCA procedure in genetix v. 4.05.2 (Belkhir et al. 2004). Finally, the population structure was also investigated using the individual-based approach implemented in the structure software (Pritchard et al. 2000). Based on the Bayesian cluster approach, this method allowed the inference of both the number, K, of genetic clusters and the admixture coefficient of individuals to be assigned to the inferred clusters. Initially, we replicated 15 independent runs for each value of K (with K varying from 1 to 9) with a total of 1,000,000 iterations and a burn-in of 100,000. To determine the number of genetic clusters from Structure analyses, we used the Structure Harvester program (Earl and Vonholdt 2012) to compare the mean likelihood and variance per K values calculated from the 15 independent runs (Evanno et al. 2005). The results gave a Delta K value per tested K value allowing the determination of the most likely number of inferred clusters. Subsequent runs were performed to test for the presence of genetic sub-structure within each cluster when K > 1 was initially inferred. These subsequent runs used the same conditions as above (Fig. 2).

Figure 2. 

Hierarchical genetic structure of Iguana inferred by Structure and Structure harvester. Bar plots show admixture coefficient of each analyzed individuals (represented by each vertical bar) for the inferred genetic clusters K (represented by a different color). The graphs show Delta K values (Evanno et al. 2005) as a function of K (number of clusters) and calculated from posterior probabilities of the data (i.e., ln [P(D|K)]). The results inferred K = 2 genetic clusters when they were initially based on the overall sampling. A subsequent run, including individuals from Montserrat-Saba and French Guiana, revealed significant genetic substructure (i.e., K = 2) separating individuals from Montserrat-Saba and those from French Guiana. Independently, another subsequent run including only individuals from Saint Lucia and the Grenadines revealed no genetic substructure. The bar plots were produced using Distruct 1.1 program (Rosenberg 2004) from the average of the 15 replicates. The names of the taxa are given according to Breuil et al. (2019) and to the conclusion of this work for Saba and Montserrat.

Based on microsatellite variation, genetic diversity (He) ranged from 0 to 0.82 (Table 2) and allelic richness (Ar) ranged from 1 to 3.86. No linkage disequilibrium was detected (adjusted p-value threshold = 0.0004) and no loci deviated from the Hardy-Weinberg expectations (adjusted p-value threshold = 0.0006). Analyses of microsatellite loci for population structure resulted in the identification of three groups (Fig. 2). Pairwise Fst values ranged from 0.104 to 0.909 (overall Fst = 0.595) and many were significant after Bonferroni correction (Indicative adjusted nominal level (5%) for multiple comparisons was 0.005) (Table 3). These results were corroborated by a FCA procedure which identified the same three groups: one including individuals from St. Lucia and Grenadines, a second including individuals from French Guiana, and the last including individuals from both Montserrat and Saba. Bayesian clustering using Structure and the Structure Harvester programs, also supported population subdivision, giving the highest posterior probability for K = 2. Specifically, individuals from Montserrat-Saba-French Guiana were mainly allocated to the first genetic cluster (i.e., mean ± sd of individual admixture coefficients: 0.997 ± 0.002; 0.998 ± 0.001 and 0.996 ± 0.0001, respectively), whereas individuals from Saint Lucia-Grenadines were allocated mainly to the second genetic cluster (i.e., mean ± sd of individual admixture coefficients: 0.997 ± 0.005 and 0.976 ± 0.029, respectively). Subsequent analyses within each of these genetic clusters revealed further sub-structure separating Montserrat-Saba and French Guiana individuals. In contrast, no genetic sub-structure was detected when considering individuals from Saint Lucia and the Grenadines (whatever the individuals, they showed similar inferences for genetic clusters) (Fig. 2).

The ML tree based on ND4 sequences had the highest log likelihood of – 2861.13 (Fig. 3). The MP analysis, using the same data, generated 2 trees of length 302 (not shown). The consistency index was 0.659, the retention index was 0.892 and the composite index was 0.644 (0.588) for all sites and parsimony-informative sites (in parentheses). ND4 sequences from Montserrat and Saba specimens were placed in a well-supported monophyletic group (Fig. 3; ML BS = 98, MP BS = 97). The Montserrat and Saba sequences were closely related to GenBank sequence HM352501 from an Iguana iguana specimen from Venezuela (Sucre: Cumana). In the MJ network (Fig. 4, Table 4), four mutational steps separate HM352501 and “Hap_4” (= HM352505 and MN590163 to 68) encompassed most of samples from Montserrat and Saba (nine in the present study and 13 from published studies). Only two ND4 haplotypes were found to characterise Montserrat and Saba iguanas represented by 24 specimens (9 from this study, 13 from GenBank consisting of 13 from Saba and 11 from Montserrat).

The Saba-Montserrat-Venezuela populations formed a monophyletic group that is the sister group of the populations from French Guiana and Brazil, this group is, in turn, the sister group of the group of St. Lucia-Grenadines that was previously described as two endemic subspecies Iguana iguana sanctaluciae for the endemic population of St. Lucia and Iguana iguana insularis for the population of the Grenadines (Breuil et al. 2019).

From these genetic data and the following morphological data, it is clear that the endemic iguanas from Saba and Montserrat form a distinct evolutionary entity that we recognize as a new species which is formally describe below accompanied by a taxonomic analysis.

Figure 3. 

The Maximum Likelihood (ML) tree of the ND4 sequences of iguanas. The percentage of trees in which the associated taxa clustered is shown next to the branches, as bootstrap support (BS ≥ 70) for ML and Maximum Parsimony (MP) topologies respectively. Sequences amplified by the authors of this study were indicated by colored diamond-shaped marks. Marks are colored based on the sampling sites. Sequences from published studies were labelled by their NCBI GenBank access numbers (Fig. 4). The names of the taxa are classified according to the conclusions of this work. Roman numerals refer to clades identified by Stephen et al. (2013).

Figure 4. 

The Median-Joining (MJ) network of the ND4 sequences of iguanas. The sequences amplified by the authors of this study and the corresponding haplotypes from published studies were labelled from “Hap_1” to “Hap_10” (details in Table 4). Other sequences from published studies were labelled by their NCBI GenBank access number. Black circles are median vectors representing extinct or unsampled haplotypes. The remaining colored circles represent haplotypes as nodes in the networks. The circles are colored based on the sampling sites. The size of circles corresponds to the number of specimens with identical sequence. The number of mutational steps is indicated by hatch marks. The names of the taxa are classified according to the conclusions of this work.

In the US Virgin Islands (St. Thomas, St. Croix, St. John) and the British Virgin Islands (BVI: Tortola) as well as on the coastal islands of Venezuela, and on the coast in the vicinity of Cumana, there are also melanistic iguanas (Lazell 1973; MacLean 1982; Buurt 2005; Falcón et al. 2018), but we found no morphological and color differences from the melanistic iguanas of Saba-Montserrat. Thus, as a hypothesis, we consider here that these melanistic iguanas belong to the same lineage as I. melanoderma (see below).

Such a discontinuous distribution of this new species is puzzling. It can be explained by: [1] the natural dispersal from northern Venezuela (mainland and coastal islands) where iguanas with the same melanistic phenotype are found, as in Margarita, Los Roques and La Blanquilla, [2] human dispersion by pre-Columbian Indians, [3] recent dispersal by Modern Man, [4] convergent evolution, [5] the regression of a wider distribution area, or [6] a combination of the previous hypotheses.

To successfully colonize an island, iguanas must first arrive on that island by natural or human means, occupy an available ecological niche, or if not, be able to successfully compete with the local species. Based on what we know about the competition between Iguana delicatissima and Iguana iguana (Breuil 2013, 2016; Vuillaume et al. 2015), it is likely that Iguana melanoderma first became established in Montserrat-Saba and prevented the subsequent establishment of Iguana delicatissima. We found no evidence of mitochondrial introgression with I. delicatissima in this new taxon, whereas we discovered mitochondrial introgression with I. delicatissima in Iguana iguana sanctaluciae (Breuil et al. 2019). Vuillaume et al. (2005) found no evidence of introgression with microsatellites in the Saba population. The possibility of nuclear introgression should be investigated in Montserrat to test the hypothesis of an ancient presence of I. delicatissima.

The first iguanas were collected in Saba by F. Laglois in 1879 (Garman 1887) and were housed at the MCZ (Harvard), but they no longer appear to be present according to the database of this museum. The first mention of iguanas in Montserrat seems to be recent, as this species is not mentioned by Dunn (1934) or by Barbour (1937), although herpetologists, such as Garman (1887), visited the island from where he described the endemic species Anolis lividus. This does not mean, however, that there were no iguanas on this island at that time. The first mention of iguanas in Montserrat was probably published by Underwood (1962).

Saba and Montserrat are separated by 150 km. Between the two is the Christopher Bank or St. Kitts Bank (St. Eustatius, St. Kitts, Nevis) which was inhabited in historical times by Iguana delicatissima perhaps with the exception of St. Kitts (Lazell 1973). Nowadays, only a small population of Iguana delicatissima is present in St. Eustatius (Van den Burg et al. 2018b) and St. Kitts and Nevis have no Iguana delicatissima. In this context, the presence of Iguana melanoderma in Saba and Montserrat appears to be an anomaly in respect to the historical distribution of Iguana delicatissima. This species has inhabited in historical times, with the exception of these two islands, all islands from Martinique to Anguilla. This can be explained if Iguana delicatissima extended its geographical range with the help of Amerindians after the arrival of I. melanoderma in Saba and Montserrat.

Steadman et al. (1984), Reitz (1994), and Pregill et al. (1994) reported the presence in two ceramic assemblages in Montserrat (± 2500 years ago) of Iguana cf. iguana bones that were used with other vertebrates as food. In Montserrat, the iguana is believed to have been introduced by Amerindians (Young and Hilton 2008) while Underwood (1962) thought it was a recent introduction but at that time no archeo-zoological iguana remains were known. Edgar (2009) suggested that the population may be native to Montserrat but that further studies are needed.

In Saba, a pre-ceramic occupation dating from 3300 BP with dense midden deposits, mainly land crabs and birds, located at 400 m a.s.l., in the tropical forest, was studied by Hofman and Hoogland (2003), but did not reveal any iguana remains. For these authors, the use of these limited food resources and the nature of the tools suggest a temporary seasonal occupation. Other Amerindian sites on Saba are dated between 400 and 1400 AD, with a major occupation period between (800–1400 AD) (Hofman and Hoogland 2003). These latter coastal archeological sites contain Iguana iguana bones (Hofman and Hoogland 1991), but it appears that this attribution is based only on the fact that Iguana iguana is the species present today on this island.

The similarity in morphology and coloration of these two populations (Breuil 2013, 2016) and their genetic homogeneity with respect to the ND4-Leu sequence and microsatellites suggest that an introduction from Saba to Montserrat or vice versa during the Amerindian period is a viable hypothesis. Our results clearly show that this Montserrat iguana belongs to the same group as Saba. The iguana bones in the Saba and Montserrat refuse middens strongly indicate that, while the attribution of bones to Iguana iguana by Hofman and Hoogland (1991) is correct with respect to delicatissima, Iguana cf. melanoderma is associated with Amerindians artifacts. However, this does not exclude the possibility of natural and/or Amerindian introductions from the eastern coastal region of Venezuela.

According to Schmidt and Inger (1957): “The common iguana, Iguana iguana, is apparently a recent immigrant to the southern Lesser Antilles and the Virgin Islands from northern South America, and has driven out the native rock iguana (genus Cyclura…) from these islands”. Platenberg and Boulon (2006) suggested that this species was introduced to the latter islands by the pre-Colombian Indians, possibly to replace the native rock iguana as a food source, although it may also have floated or rafted there. According to Mayer (2012), all Iguana iguana present in the Virgin Islands were introduced. Recent introductions of iguanas from the pet trade into the USVI may have occurred from different locations (Falcón et al. 2013). According to Falcón et al. (2018), haplotypes of Caribbean iguanas were found on the islands of Vieques (Puerto Rico Bank) and St. Croix (USVI). These authors also found that all individuals from St. Thomas (USVI) with the melanistic phenotype shared haplotypes found in iguanas native to two Caribbean Islands not mentioned in their study. These preliminary genetic data concerning the melanistic iguanas of the Virgin Islands and Puerto Rico Bank and their morphology, which is indistinguishable from the iguanas of Saba-Montserrat, suggest that these melanistic iguanas belong to I. melanoderma. Genetic studies with microsatellites for these populations are urgently needed to answer this question in view of the presence of invasive iguanas (Iguana iguana iguana, Iguana rhinolopha) on the Virgin Islands and Puerto Rico Bank.

The Pleistocene natural deposits of Marie-Galante do not contain the remains of Iguana delicatissima nor Iguana iguana while many other reptiles, now extinct, are present. Iguana delicatissima appears on this island with Amerindians artifacts that suggest human introduction (Grouard 2001; Bochaton et al. 2015). The current distribution of I. delicatissima is thus a mixture of natural colonization and Amerindian introductions as it is now demonstrated.

Based on our data and those of Stephen et al. (2013), we propose that the origin of these insular populations with the black phenotype results from a first colonization from coastal Venezuela (Cumana area) and the Venezuelan continental islands such as Margarita, Los Roques or La Blanquilla where melanistic iguanas are present to Saba or Montserrat. A pre-Columbian colonization from one of these two islands to the other may have followed, and a third colonization towards the Puerto Rico Bank and St. Croix Bank. The original differentiation leading to this black phenotype and genetic differentiation may have taken place in the Cumana region. The colonization of Saba-Montserrat and the Puerto Rico Bank, Virgin Islands, and St. Croix Bank, with little or no divergence, may be the consequence of multiple, diachronic, Amerindians or recent colonization events.

Stephen et al. (2013) suggested that there were two natural dispersal events in the Lesser Antilles. The older one gave birth to the population of St. Lucia which is the most differentiated and is now considered as a subspecies I. iguana sanctaluciae (Breuil et al. 2019), and the most recent event to the Saba-Montserrat population, less differentiated from the Venezuela population, considered here as I. melanoderma. The phylogeny of Stephen et al. (2013) proposed the existence of three lineages in the Common Green Iguana. The deepest and therefore the oldest is represented by the Curaçao population which, in this context, warrants a specific status. The intermediate lineage is represented by the Central American clade and the most recent is the South American lineage to which the insular populations of St. Lucia-Grenadines and Saba-Montserrat-Venezuela belong. Breuil (2013, 2016) proposed to consider the Central American iguana clade as Iguana iguana rhinolopha and the South America clade as Iguana iguana iguana. Breuil et al. (2019) upgraded these two subspecies to the species level.

The five differences (5/818 = 0,6 %) between the haplotype (HM352505) of Saba-Montserrat and the haplotype of Venezuela (HM352501) suggest differentiation, with a genetic divergence, approximated by a molecular clock of 1.29 million years for every 1% sequence divergence, at the ND4-Leu Locus (Malone et al. 2000), of ka. However, if the dark coloration of I. melanoderma is an adaptation to the cold environment of Saba, we have no parsimonious explanation for the presence of melanistic iguanas in northeastern Venezuela, and on the St. Croix and Puerto Rico Banks, except a very recent diaspora in a context where this black coloration does not show any loss of fitness in a very safe and protective environment.

Hap_4 (GB HM352505), the most common haplotype, is shared by these two insular populations, but Saba also has its own haplotype (Hap_5) (Fig. 4). This most common haplotype is, according to Falcón et al. (2018), also present in the Puerto Rico Bank. This is a strong argument in favor of a recent diaspora of this melanistic iguana throughout the Caribbean Region. A similar situation was found in Leptodactylus fallax where no genetic differences were found between the Montserrat population and the Dominican population (Hedges and Heinicke 2007). These authors suggested that Amerindians were responsible for this distribution pattern even though it had been present in the past in St. Kitts, Martinique, and St. Lucia (Lescure 2000).

Montserrat consists of three major volcanic centers: to the north are the heavily eroded Silver Hills (ca. 2.6–1.2 Ma); in the center are the Centre Hills (ca. 950–550 ka), also extinct and crossed by deep erosive canyons; and to the south is the massif comprising the South Soufrière Hills (ca. 135–125 ka) and the Soufrière Hills (ca. 170 ky to present) (Harford et al. 2002). Based on geographical distance and age, Montserrat is a better candidate for first natural colonization by propagules from Venezuela. Montserrat is located 150 km southeast of Saba.

Saba’s oldest rocks are about 400 ky years but most of the volcanic deposits were produced in the last 70,000 years and have increased the size of the island on the edge of Mt Scenery. The last eruption is dated to 280 years BP and covers Amerindian artifacts but underlies those of the European settlers who colonized the island in 1640. European settlers may have been attracted to the island because of the presence of grassland instead of tropical rainforest caused by an eruption shortly before European settlement (Roobol and Smith 2004). Under these conditions, a recent major bottleneck may have occurred in the iguana population.

A bottleneck hypothesis and/or founder effect to explain the low genetic diversity of these insular iguanas is congruent with the fact that of the 16 common microsatellite loci used in this study, 9 are monomorphic in Saba while 8 are monomorphic in Montserrat with twice as many individuals. For example, the Iguana delicatissima population of Petite Terre (Guadeloupian Archipelago) has 8/15 monomorphic microsatellite loci and the Chancel Islet population 6/15. In both populations, serious bottlenecks have recently occurred (Breuil 2002; 2009). Indeed, this panel of microsatellite molecular markers revealed high polymorphism in all the other common iguana populations studied by Valette et al. (2013) and Vuillaume et al. (2015).

Results based on microsatellite markers revealed a higher level of genetic diversity in Montserrat iguanas than in Saba iguanas. Montserrat iguanas show a genetic signature close to that of iguanas from French Guiana (Fig. 2). However, this did not appear to be the result of hybridization between two populations as several alleles are also present in the Central and South American clade populations (unpublished data). This should be investigated in a further study of the Montserrat population to check for the presence of exotic iguanas.

According to Miralles et al. (2017), in their review of the taxonomic revision of Amblyrhynchus cristatus, there is no objective definition of subspecies, but for these authors, the subspecies rank is useful to refer to distinct population-level units that have not attained independent evolutionary lineage status, and so do not meet the species criteria. These authors thus described five new subspecies of Amblyrhynchus cristatus based on morphology, mitochondrial DNA, and microsatellite clusters as we did in this study. Most of their subspecies live on a single island, while two of them live on two islands and one island has two subspecies.

Breuil et al. (2019) discussed the arguments for naming the St. Lucia and Grenadines populations as one subspecies or a new species with two subspecies. The choice was made to adopt the most conservative solution for this emblematic species. The difficulty lies mainly in the different concepts used to distinguish between the species and subspecies categories (Torstrom et al. 2014). Hybridization is known between the endemic iguanas of the Lesser Antilles and the two invasive lineages i.e., Iguana iguana and Iguana rhinolopha (Breuil 2013, 2016; Vuillaume et al. 2015) and now with the Curacao lineage in Saint Maarten (Van den Burg et al. 2018a, 2018b). In addition, hybridization has recently occurred between endemic Cyclura nubila and Iguana iguana (Moss et al. 2018). In this context, it is clear that there are no intrinsic barriers to the reproduction among all the clades of Iguana identified first by Malone et al. (2000) and their subsequent work (Stephen et al. 2013). We chose to recognize these melanistic iguanas at the species level, integrating the source population of Venezuela and making it fit the criterion of an independent evolutionary unit.

For example, Cyclura nubila nubila differs by less than 1% from the ND4 sequence of C. nubila caymanensis . The same level of divergence is also found between C. cychlura cychlura and C. cychlura inornata which are considered valid subspecies (Malone and Davis 2004). With 1.47 % (12/818) difference between the two haplotypes (JQ340913, JQ340914) from Brazil and the two haplotypes from Saba-Montserrat (HM352505) and Venezuela (HM352501), the differences are greater than in these Cyclura subspecies. Powell and Henderson (2005), Powell et al. (2005), Powell (2006), and Henderson and Powell (2009) suggested that the Saba and Montserrat populations (and possibly the historical population on St. Croix) may warrant designation as a separate species from the Common Green Iguana found elsewhere. We follow their recommendations here.

According to Stephen et al. (2013), the Central American clade (Mexico to northern Panama) is distinct from the South American clade and suggests the presence of cryptic species. The difficulty with naming these two clades is that many authors, such as Lazell (1973) or Stephen et al. (2013) considered that the only morphological distinction between these two clades is “enlarged tubercle scales on the snout” (i.e., horns). Breuil (2013; 2016) demonstrated that there are many morphological differences between these clades and that, in addition, the horns of iguanas of the southern group of Lesser Antilles are different from the horns of rhinolopha from Central America. Even if some populations or some individuals in some Central American populations have only small or no horns at all, this does not change the taxonomic issue. The first Central American iguanas in Mexico were described as a full species, Iguana rhinolopha, which clearly belongs to the Central American clade. This name should therefore be used for the Central American clade from Mexico to northern Panama, called number III by Stephen et al. (2013) (Fig. 3).

Clade II of (Stephen et al. 2013) comprises subclade IIa, called north west of Andes, which contains iguanas from Columbia, Peru and Ecuador, whereas subclade IIb, called south east of Andes, includes iguanas from Venezuela, Suriname, Brazil, and the Lesser Antilles (Saba, Montserrat, St. Lucia). According to our results, in our working region, we have the choice to consider that we have one species (Iguana iguana) with three subspecies (iguana, sanctaluciae, insularis), and Iguana melanoderma as another species, with the following distribution:

• Iguana iguana iguana (French Guiana and Brazil);

• Iguana iguana sanctaluciae (St. Lucia);

• Iguana iguana insularis (St. Vincent and Grenadines);

• Iguana melanoderma (north-eastern Venezuela, Venezuelan coastal islands, Saba-Montserrat Puerto Rico Bank, Virgin Islands, St. Croix Bank).

According to Stephen et al. (2013), populations in Brazil, Suriname and Venezuela have the same unique PAC haplotype (JN811116) as the Saba-Montserrat iguanas (23 iguanas), which is different from the haplotypes found in clades IIb and III. The single PAC haplotype identified in Saint Lucia (also present in the Grenadines populations, unpublished results) is a synapomorphy of the iguanas of this southern group. So, we have now recognized one new species, Iguana insularis, with two subspecies:

• Iguana insularis insularis comb. nov. of the Grenada Bank (including the Grenadines);

• and Iguana insularis sanctaluciae comb. nov. from Saint Lucia.

Naming species and subspecies in the South American clade has not resolved phylogenetic relationships with continental iguanas. For example, the formation of the Isthmus of Panama which was previously thought to be 2.5 My old (Iturralde-Vinent 2006) may be more than 10 My older (Montes et al. 2015). This new dating challenges the age of the first colonization of Central America by ancestral iguanas. Our taxonomic proposal simply reflects the morphological and genetic originalities of the endemic insular horned iguanas of the southern Antilles and the melanistic iguanas of the northern Antilles, regardless of changes in the original natural distribution area. The characterization of all these insular endemic taxa provides a solid basis for conservation.

The description of Iguana melanoderma, with its morphological and genetic diagnostic characteristics, will enable conservationists to differentiate between endemic and exotic iguanas. For example, the IUCN Red List (Bock et al. 2018) considered the Common Green Iguana, Iguana iguana, to be of “Least Concern” but failed to differentiate between populations that do not have the same threat levels. With our taxonomic proposal, these endemic insular populations will be considered as a conservation unit with their own assessments.

As the range of invasive iguanas increases worldwide (Falcón et al. 2013), and mainly in the Caribbean (Powell et al. 2011; van den Burg et al. 2018a, 2018b), the probability of invasive iguanas arriving on Saba increases with trade, travel, and tourism on St. Maarten (Yokoyama 2012), another Dutch island 45 km north of Saba. It is thus necessary to conserve these original populations and to be able to differentiate them from invasive iguanas of different lineages.

The Saba Conservation Foundation (SCF) is responsible for nature management on the island for the local government and runs programs for the Saban Black Iguana by reducing defoliation by goats through the restriction of the number of free-ranging domestic animals (Powell et al. 2005). As a seed dispersing animal, this taxon contributes to the recolonization of landscapes after human, animal or natural destruction.

We found no evidence of genetic introgression in the Saban Black Iguana population; the samples studied for genetic analysis, collected in 2011, have all the characteristics of this new taxon. However, according to a photograph published in the book by Powell et al. (2005) (fig. 92), but not taken by these authors, invasive iguanas could be (or could have been) present in Saba, unless the locality of the photograph is in error.

Moreover, it is clear from the Google photographs of iguanas taken in the Virgin Islands that the invasive iguana (Iguana rhinolopha) is also present and hybridization is occurring as shown by the presence of intermediate phenotypes in these islands and by genetic data (Falcón et al. 2018). The same holds true for Margarita Island (Venezuela).

Iguana iguana is listed in CITES Appendix II, but export quotas exist for many countries for pet trade and products (leather, meat). No distinction is made between native and introduced populations, or between continental and insular populations (Powell and Henderson 2007). With regard to its endemicity [1], the small size of these islands (Saba: 13 km², Montserrat: 100 km²) [2], that this new taxon occupies less than half of Saba and occupied less than 10 % of Montserrat before the eruption of the Soufrière Hills in 1995 [3], that the total effective population of this species in Saba and Montserrat could be around 400 adults [4], that there is a pressure on the biotope [5], that hunting is rare but present [6], that dogs and cats, as well as motor vehicles, kill iguanas [7], and that the invasive iguanas could arrive soon [8], this new taxon is, based on IUCN criteria, critically endangered on the islands of Saba and Montserrat.

Priority actions for the conservation of the species Iguana melanoderma are biosecurity [1], minimization of hunting [2], and habitat conservation [3]. The maritime and airport authorities of both islands must be vigilant about the movements of iguanas, or their sub-products, in either direction, even if the animals remain within the same nation’s territory. Capacity-building and awareness-raising should strengthen the islands’ biosecurity system and could enhance pride in this flagship species.

Key stakeholders in conservation efforts are the Dutch Caribbean Nature Alliance (DCNA), the Saba Conservation Foundation (SCF), the Montserrat National Trust (MNT) and the UK Overseas Territories Conservation Forum (UKOTCF).

The DCNA is already involved in the conservation of Iguana delicatissima on St Eustatius¹. The MNT is implementing a project funded by the Darwin Initiative entitled “Adopt a Home for Wildlife”, which can also be harnessed for iguana conservation. MNT and UKOTCF suggested the following priority actions for the native iguana: preparation of an inventory, including a survey of the geographical distribution and ecology of the taxon, assessment of conservation status and regular monitoring, as well as habitat mapping and assessment of conservation purposes. The listing of Fox’s Bay Bird Sanctuary as a Ramsar Site was also recommended².

In conclusion, the pioneering work of Malone et al. (2000) and her subsequent works (Malone and Davis 2004; Stephen et al. 2013), followed by those of Breuil (2013, 2016), Vuillaume et al. (2015), and Breuil et al. (2019) clearly demonstrated that there are morphological, mitochondrial, and nuclear divergences between what was considered to be a single species, Iguana iguana. Thus, different clades have been identified genetically and morphologically without naming them. These diagnosable insular entities are here considered to form two new species: Iguana melanoderma and Iguana insularis. Southeastern Andes populations are considered to be Iguana iguana and populations in the northern Isthmus of Panama are considered to be Iguana rhinolopha. Morphological, genetic, ecological, and ethological studies are needed to characterize and name all lineages not included in these four recognized species. Our work provides the tools for a better protection of these insular flagship species, mainly against alien iguanas.