Introduction

The US state of Florida is home to 137 introduced amphibians and reptiles, of which 56 are now established species. A vast majority of these introductions originated from the pet trade (Krysko et al. 2011). A combination of factors such as a subtropical climate, numerous ports of entry for wildlife trade, a prolific captive wildlife industry, and location in a hurricane-prone region make Florida particularly susceptible to establishment by invasive species (Engeman et al. 2011). This flood of non-native herpetofauna has had serious impacts on the native wildlife (Engeman et al. 2009; Willson 2017; Holbrook and Chesnes 2011; Mazzotti et al. 2015). Perhaps the most infamous invasive species of reptile in Florida is the Burmese python (Python bivittatus), which is thought to have been introduced to the state at some point in the 1980’s, and within only a few decades has experienced explosive population increases that have led to precipitous declines of native wildlife species (Willson et al. 2017; Dorcas et al. 2012). Other high-profile invasive reptile species that pose a substantial risk to Florida’s ecosystems include the Nile monitor lizard (Varanus niloticus; Enge et al. 2004), green iguana (Iguana iguana; Townsend et al. 2003), and Argentine black and white tegu (Salvator merianae; Mazzotti et al. 2015). Due to the negative impacts associated with invasive reptiles in Florida, risk assessment and identification of appropriate management efforts for different species are critical for ecosystem preservation.

As part of the introduced reptile fauna, a number of non-native crocodylian species have been recorded in Florida over the last few decades, including the spectacled caiman (Caiman crocodilus), Cuvier’s dwarf caiman (Paleosuchus palpebrosus), smooth-fronted caiman (P. trigonatus), Nile crocodile (Crocodylus niloticus), and West African slender-snouted crocodile (Mecistops cataphractus; Krysko et al. 2011; Rochford et al. 2016; Roberto et al. 2021). However, only one species—the spectacled caiman (C. crocodilus)—has become established in the United States (Ellis 1980). Caiman crocodilus was introduced during the 1950’s when the demand for the species for the pet trade was at its highest. This demand coincided with the prohibition of trade in American alligators (Alligator mississippiensis) to protect then-declining alligator populations. Most of the C. crocodilus individuals in Florida were thought to have originated from Colombia (King 1974; Ellis 1980). While not currently considered as serious an ecological threat as other invasive reptiles (e.g., Burmese python), the spectacled caiman is classified as a Stage 4 invasive species according to the five-stage biological invasion model outlined by Colautti and MacIsaac (2004). By definition, “the species is spatially limited in its locality, but still dominant where it occurs” (Krysko et al. 2011). However, C. crocodilus has a broad diet in its native range (Magnusson et al. 1987; Thorbjarnarson 1993) as well as its non-native range in Puerto Rico (Bontemps et al. 2016), and the species could presumably come to affect a number of sensitive species within its Florida range.

When evaluating the risks of biological invasions and the pathways by which a species has made its way into a new location, it is useful to employ a molecular ecology perspective. By analyzing phylogeographic data from an invasive species’ native range, it is possible to trace the genetic origins and assess the levels of genetic diversity in the newly established populations (Avise 2004; Allendorf et al. 2013). When a species is introduced to a new locality, it often undergoes a population bottleneck reducing genetic variation. This reduction in genetic diversity can lead to inbreeding and is often detrimental to the survival of small populations (Saccheri et al. 1998). This phenomenon can influence the invasion success of an introduced species. However, if a species is introduced from multiple source populations, then there may be an admixture effect that could lead to increased genetic diversity compared to the original source populations (Kolbe et al. 2007). The population bottleneck and admixture models are not mutually exclusive and can help predict the success of biological invasions and inform management actions. While nuclear DNA data are required to fully assess levels of genetic admixture, haploid data are useful in identifying source populations and levels of population differentiation (Allendorf et al. 2013). In crocodylians, haploid data have been used to evaluate and compare genetic diversity between populations (Amavet et al. 2021) and would be informative for addressing the molecular ecology of C. crocodilus in Florida.

The taxonomic affinities of the Florida caiman have been the subject of much speculation and considerable uncertainty. Since their initial introduction in the 1950’s, several studies have referred to them by various sub-specific names without any genetic data (Smith and Kohler 1977; Wilson and Porras 1983; Velasco et al. 2010). The first study to investigate the molecular systematics of this introduced population was Roberto et al. (2021). They analyzed the Cytochrome B (CytB) gene sequences of seven Florida samples collected near Everglades National Park and Homestead Air Base and found that they clustered with samples from the Caribbean coast of Colombia in the Bolivar Department. In this study, we expand on the findings of Roberto et al. (2021) to better assess how many genetic lineages of the C. crocodilus species complex are present in Florida, using a larger sample size. We add 27 samples of Florida C. crocodilus to the dataset from Roberto et al. (2021), covering much of the species’ range across South Florida. We use sequence data from two mitochondrial genes—Cytochrome B (CytB) and Cytochrome oxidase I (COI) for comparison with homologous gene fragments available on GenBank. Given the extent of the pet trade in the species and observations of variation of markings on the jaws of captured animals (S. Balaguera-Reina, pers. comm. 2023), we hypothesize that there are multiple phylogeographic origins of the spectacled caiman in South Florida, and that there is a lower amount of genetic diversity in the introduced population(s) compared to populations in the native range. Based on the relatively slow rates of molecular evolution and long generation times seen in crocodylians (Green et al. 2014), we hypothesize that the Florida caiman has not been established long enough to exhibit elevated levels of genetic diversity.

Methods

Sample collection and laboratory work

Spectacled caiman tissue samples were obtained from animals captured throughout South Florida, United States as part of a long-term control and removal effort (Godfrey et al. 2023). Animals were located and captured during crocodylian eye-shine spotlight surveys from 2012 to 2021 that targeted areas within or near two Comprehensive Everglades Restoration Plan (CERP) projects: the Biscayne Bay Coastal Wetlands (BBCW) and the C-111 Spreader Canal Western Project. Capture methods included hand capture, snake tongs, wire snares, and toggle darts. Coordinates were collected at the site of each caiman capture with a Garmin handheld global positioning system unit (Garmin International, USA). All captured animals were euthanized humanely in accordance with American Veterinary Medical Association guidelines (Underwood and Anthony 2020). Muscle tissue samples were collected during necropsies and stored in a − 20 °C walk-in freezer at the University of Florida IFAS Fort Lauderdale Research and Education Center. Finally, samples were put in 70% ethanol and shipped off to the Densmore Lab at Texas Tech University for molecular analysis.

Total genomic DNA was extracted from 27 individual samples using a PureGene isolation kit (Gentra Systems, Minneapolis, MN; Supplementary Information, SI 1). Sample quality of total DNA was then examined by gel electrophoresis with 1% agarose gels stained with ethidium bromide under UV light. DNA concentrations in ng/µL (A260/A280) were then estimated using a NanoDrop One (ThermoFisher Scientific, Waltham, MA, USA). Mitochondrial genes were targeted and amplified using primer pairs developed by Jimenez-Alonso et al. (2023) in accordance with the methods outlined in Balaguera-Reina et al. (2021). We amplified the Cytochrome B (CytB) gene using primer pairs Ccr_GluL1T (5′-GTCTGAAAAACCCCCGTTG-3′) and Ccr_ThrH1T (5′-CCCGTCTCTGTCTTACAAGG-3′), and the Cytochrome Oxidase I (COI) gene with primer pairs Ccr_TyrL1dT (5′-AACACTCGGCCACTTTACC-3′) and Ccr_SerH1T (5′-AGATGGTGGGGGTTCGATT-3′). Polymerase Chain Reaction (PCR) reactions were set up using 0.15 µL of total genomic DNA, 8.91 µL of dd H2O, and 3.44 µL of Master Mix [72.7% 5X Buffer D (Promega Corp, Madison, Wisconsin, USA), 14.5% dNTPs, 5.5% of each primer (10 µM), and 1.8% Taq DNA polymerase w/standard Taq Buffer (New England Biolabs, Ipswich, Massachusetts, USA)].

Thermocycler conditions for COI primers included an initial denaturation step of 94 °C for 3 min, followed by 30 cycles of 94 °C for 30 s, an annealing step at 49.5 °C for 45 s, an extension step at 72 °C for 1 min, and a final extension step at 72 °C for 10 min. Conditions for CytB primers were: an initial denaturation step of 94 °C for 3 min, followed by 33 cycles of 94 °C for 30 s, an annealing step at 62 °C for 45 s, an extension step at 72 °C for 1 min, and a final extension step at 72 °C for 7 min. PCR products were electrophoresed on a 1% agarose gel and then treated with ExoSAP-IT Cleanup Reagent (ThermoFisher Scientific) and incubated in a thermocycler for one cycle of 37 °C for 15 min, followed by one cycle of 80 °C for 15 min. PCR products were then cycle-sequenced using Big Dye v3.1 dye terminator (Applied Biosystems, Inc., Forster City, CA, USA) at the manufacturer’s recommended PCR conditions. Sequence products were purified by filtering the solutions through a G-50 Sephadex column (0.5 g of Sephadex/800 µL dd H2O), which was incubated at room temperature for 30 min and centrifuged at 3000 rpm for 2 min to construct the column. Approximately 10 µL of cycle-sequenced product were mailed to Eurofins Genomics LLC in Louisville, KY, USA for Sanger sequencing.

Sequence chromatograms were obtained and visualized in Sequencher® version 4.7 (Gene Codes Corporation 2006). Forward and reverse sequences were trimmed so that the first 25 bases contained less than 1 base with confidence below 25%. The contigs were then assembled from the forward and reverse reads using ReAligner (Anson and Myers 1997) with a minimum match of 85% and minimum overlap of 20 bases. Ambiguous bases, contig site disagreements, and low confidence bases were either edited or replaced with an IUPAC ambiguity code. Newly collected sequence data were deposited in GenBank (accession numbers: COI = OR518300-OR518326 and CytB = OR402067-OR402093) and compared with homologous C. crocodilus genes published in the literature and available on GenBank (Farias et al. 2004; Vasconcelos et al. 2006; Venegas-Anaya et al. 2008; Roberto et al. 2020; Hernández-González and Díaz-Moreno 2020; Balaguera-Reina et al. 2020, 2021, 2022; Roberto et al. 2021; Jimenez-Alonso et al. 2023). All sequences were aligned in MAFFT version 7.511 using the iterative refinement method (FFT-NS-I; Katoh et al. 2002). Alignments were trimmed in AliView v 1.28 (Larsson 2014). We performed a parsimony-based partition homogeneity test with 100 replicates in PAUP 4.0 (Swofford 1998) on the COI and CytB alignments to check if the two genes were congruent in their phylogenetic history.

Phylogenetic analytical methods

Phylogenetic analyses were conducted using two different statistical frameworks to reconstruct the phylogenetic relationships between samples within the C. crocodilus/yacare species complex, and to examine where Florida samples clustered with previously described lineages. We used CytB-only sequences (N = 620, 1094 bp), as well as the concatenated sequences of COI + CytB (N = 235, 2639 bp). Sequences of the American alligator (A. mississippiensis Y13113.1), Chinese alligator (A. sinensis AF511507.1), Smooth fronted caiman (P. trigonatus AM493869.1), Black caiman (Melanosuchus niger KU986330—JF315312.1), and Broad snouted caiman (Caiman latirostris JN311630.1 - JF315308) were included as outgroups.

BEAST 2.6.7 software (Bouckaert et al. 2019) was used to conduct a Bayesian Inference (BI) analysis with the package, bModelTest (Bouckaert and Drummond 2017) to coestimate the site model alongside the phylogeny to avoid the use of a likelihood-based method that would a priori choose a single site model. We used a lognormal relaxed clock model with a coalescent constant population size (Roberto et al. 2021). For the CytB alignment, four independent Markov Chain Monte Carlo (MCMC) runs were conducted for 50 million generations, sampling every 5,000 generations. For the COI + CytB alignment, three independent MCMC runs were conducted with the same parameters. Trace log files were combined in the BEAST package LogCombiner and checked that all Effective Sample Size (ESS) values were greater than or close to 200 in the TRACER 1.7 software (Rambaut et al. 2018). LogCombiner was again used to resample the tree files to save on computing power, and a maximum clade credibility tree was generated after using a burnin of 25% with TreeAnnotator (Bouckaert et al. 2019). A maximum likelihood (ML) analysis was performed on both the CytB-only and concatenated COI + CytB alignments in IQTree 2.0 (Minh et al. 2020) using ModelFinder (Kalyaanamoorthy et al. 2017) to estimate the best-fit nucleotide substitution model based on Bayesian Information Criterion (BIC). Statistical support was calculated with 1,000 ultrafast bootstrap replicates. Phylogenies were visualized and plotted using FigTree v1.4.4 (Rambaut 2009). The Shimodaira–Hasegawa test (Shimodaira and Hasegawa 1999) was executed in PAUP 4.0 to test the null hypothesis of no difference between the tree topologies.

The package “ape” (Paradis and Schliep 2019) in R v4.2 (R Core Team 2022) was used to create a genetic distance matrix for the concatenated COI + CytB alignment of the molecular lineage C. c. fuscus (sensu stricto) and the Florida samples to test the null hypothesis that the Florida samples all have similar genetic distances to the trans-Andean C. c. fuscus (sensu stricto) from Colombia. For this analysis, we used the dist.dna function with the TN93 substitution model (Tamura and Nei 1993; Strimmer et al. 2009). This matrix was then used to construct an unrooted Neighbor Joining (NJ) tree of C. c. fuscus (sensu stricto) to further resolve the origins of Florida samples from this region.

Genetic analysis of haploid diversity

We created subsets of both CytB-only and concatenated COI + CytB sequences based on phylogenetic results and in accordance with geographic locations at both country and regional scales. Sequences from Florida were grouped according to their phylogenetic affinities and analyzed alongside sequences from the native range to compare genetic diversity between the native and introduced populations. In all but the C. c. fuscus (sensu stricto) lineage and Florida populations, CytB-only alignments were used for analytical comparisons due to lack of COI sequence data. Population genetic analyses were conducted on the following groups: trans-Andean lineages (176 sequences); C. c. fuscus (sensu stricto) from Colombia, Panama, and Costa Rica (150 sequences); cis-Andean lineages + C. yacare (439 sequences); C. c. crocodilus Upper Branco lineage (25 sequences); Florida caiman (34 sequences).

We performed an analysis of DNA polymorphism in DnaSP v6.12 (Rozas et al. 2017) within groups to estimate the number of polymorphic sites, nucleotide diversity, number of haplotypes, and haplotype diversity (Nei 1987). Fixation indices (FST) were estimated between select groups using 5000 permutation replicates of the equation outlined in Hudson et al. (1992), implemented in DnaSP v6.12. Genetic differentiation was defined as low for FST < 0.05, moderate for FST between 0.05 and 0.15, high for FST between 0.15 and 0.25, and very high for FST > 0.25 (Men et al. 2017; Balaguera-Reina et al. 2021). Additionally, we calculated Tajima’s D (Tajima 1989) and Fu and Li’s F and D (Fu and Li 1993) statistics to detect selective neutrality. Haplotype data were then used to construct TCS haplotype networks in PopART (Clement et al. 2002; Leigh and Bryant 2015).

Results

Phylogenetic analysis

We obtained a total of 27 new sequences of both CytB and COI for the Florida caimans. The assembled sequences ranged from 1097–1225 bp for CytB, and 1463–1521 bp for COI. These sequences were incorporated into the CytB and COI alignments. The parsimony-based partition homogeneity test found no significant incongruence (p > 0.05) between the two genes and so we used the concatenated COI + CytB alignment as well as the CytB-only alignment for subsequent analyses. For the ML analysis the TIM + F + G4 substitution model was used on the CytB-only alignment, and the TVM + F + G4 model on the concatenated COI + CytB alignment, as chosen by ModelFinder based on Bayesian Information Criterion (BIC). Both the BI analysis and ML analyses yielded very similar topologies for both alignments, and the Shimodaira-Hasegawa test found no significant differences between tree topologies (p > 0.05).

All major mitochondrial molecular lineages recognized by Roberto et al. (2020) were recovered in both the BI and ML analyses of the CytB gene (Fig. 1A, B). These same analyses for the concatenated COI + CytB alignment recovered all lineages that were present in the reduced dataset (Fig. 2). We found very high statistical support (Bayesian posterior values > 0.98, Bootstrap support values > 90) for three distinct lineages of C. crocodilus sensu stricto, three lineages of C. c. fuscus/chiapasius, and two lineages of Caiman yacare. Deeper nodes indicating phylogenetic relationships between these clades had lower statistical support and differed from other studies that used fossil calibrations (Roberto et al. 2020; Amavet et al. 2023). The Upper Branco and Southwestern Amazon clades were basal to all other C. crocodilus and C. yacare lineages. In both the BI and ML analyses of the CytB gene, there were 17 sequences from Florida that clustered within the C. c. fuscus (sensu stricto) lineage and 10 sequences that clustered within the Upper Branco clade of C. c. crocodilus in the state of Roraima, Brazil. The Florida samples nested within the C. c. fuscus clade were polyphyletic, while the samples nested in the Upper Branco clade of C. c. crocodilus were monophyletic, with 100% posterior probability support and 96% bootstrap support.

Fig. 1
figure 1

A Phylogenetic tree showing major lineages of the Caiman crocodilus/yacare species complex based on the mitochondrial CytB gene sequences (N = 620, 1094 bp). For each branch on the tree, statistical support values are indicated by Bayesian posterior support to the left of the slash, and bootstrap values to the right of the slash. Yellow stars indicate clades where the Florida caiman sequences were nested. B Geographic locations of CytB sequences collected from C. crocodilus and C. yacare across their ranges. The light blue polygon indicates the range of C. yacare (Campos et al. 2020) and the light red polygon indicates the range of C. crocodilus (Balaguera-Reina and Velasco 2019). Yellow stars indicate where samples were collected from Florida

Full size image
Fig. 2
figure 2

Phylogeny of Caiman crocodilus based on the concatenated COI + CytB alignment. Statistical support is indicated at major nodes with posterior support values before the slash and bootstrap support after the slash. Colored triangles indicate major lineages as defined by Roberto et al. (2020). The light blue box denotes the C. c. fuscus (sensu stricto) lineage. Tree tips labeled in red indicate the 27 Florida samples sequenced for this study

Full size image

The genetic pairwise difference matrix revealed substantial differentiation of the 10 Florida sequences of cis-Andean origin from the other 17 Florida sequences, as indicated by the NJ tree (Supplementary Information, SI 2). Generally, genetic distances were shorter between samples within Colombia, and longer between samples from Colombia and Costa Rica, and Colombia and Panama. Within the 17 Florida sequences of trans-Andean origin there were two distinct clusters. These sequences were genetically closest to sequences from various locations throughout the trans-Andean regions of Colombia including the Upper Magdalena River Basin and Paramillo region near the Colombia/Panama border.

Genetic diversity analysis of trans-Andean Caiman crocodilus populations

The trans-Andean group consisted of 176 sequences from Colombia, Panama, Costa Rica, El Salvador, Mexico, and the US (Florida). For subset groups based on country and molecular lineage, the haplotype network analysis of the CytB gene yielded a total of 22 haplotypes based on 42 polymorphic sites, with an overall nucleotide diversity (π) of 0.0053. Within this group, the analysis revealed 3 haplotypes of C. c. chiapasius, 6 of C. c. fuscus each in Panama and Costa Rica, 13 of C. c. fuscus in Colombia, and 2 in Florida, USA. The most frequent haplotype (haplotype 2 as defined in Supplementary Information, SI 3) consisted of 68 sequences primarily from the C. c. fuscus (sensu stricto) lineage in Colombia, with several closely related haplotypes of much lower frequency occurring in geographically similar localities (Fig. 3A). Haplotype diversity ranged from 0.159 in Florida to 0.802 in Panama. We found reduced nucleotide diversity in the C. c. fuscus Florida population (π = 0.0002) compared to all other trans-Andean subsets by country (Table 1). Pairwise FST values for this group were significant (χ2 = 448.12, df = 84, p < 0.01), and indicated very high genetic differentiation between all but one (C. c. fuscus in Colombia and Panama) of ten pairwise comparisons. Neutrality tests based on Tajima’s D, Fu and Li’s D, and Fu and Li’s F indicated no significant deviations from the neutrality model for any groups based on country and molecular lineages (Table 1).

Fig. 3
figure 3

A Haplotype map of the CytB-only alignment for trans-Andean lineages of Caiman crocodilus, and B cis-Andean lineages of C. crocodilus most closely related to the C. c. crocodilus haplotype from Florida. Numbers indicate haplotype number, as defined in Supplementary Information, SI 3. Each hash mark indicates a single nucleotide mutation

Full size image
Table 1 Genetic diversity of Caiman crocodilus across trans-Andean Subspecies in different countries based on CytB sequences (1094 bp)
Full size table

Genetic analysis of the concatenated COI + CytB sequences of the C. c. fuscus (sensu stricto) population in Colombia found 23 haplotypes, based on 30 polymorphic sites, with an overall haplotype diversity of 0.881 and an overall nucleotide diversity of 0.0024 (Table 2). We found large amounts of variability in genetic diversity between the populations defined by Díaz-Moreno et al. (2021). Our results corroborated those of Díaz-Moreno et al. (2021) and found the same population genetic indices indicating elevated levels of diversity in the Upper Magdalena River Basin (UMRB). We also found very low levels of nucleotide diversity in the Catatumbo and Guajira regions of northeastern Colombia (π = 0.0 and 0.0002, respectively) as well as in the Pacifico region in the southwest (π = 0.0). The Florida population of trans-Andean origin had the next lowest nucleotide diversity in this group, and shared haplotypes with samples from the UMRB, Lower Magdalena River Basin (LMRB), Middle Magdalena River Basin (MMRB), Catatumbo, Paramillo, and Guajira regions. The Florida population had the lowest genetic divergence from the LMRB (FST = 0.253, p < 0.001), followed by the UMRB (FST = 0.384, p < 0.01), and the highest divergence from the Pacifico region in southwestern Colombia (FST = 0.972, p < 0.001; Table 3). The neutrality tests found no significant deviations from neutrality at the regional level, but the C. c. fuscus (sensu stricto) population in Colombia as a whole showed significant negative deviations (Fu and Li’s D = − 2.706, Fu and Li’s F = − 2.334). The Florida population had a non-significant negative deviation for Tajima’s D (− 0.720), and non-significant positive deviations for Fu and Li’s D and F (1.033 and 0.592, respectively; Table 2).

Table 2 Genetic diversity of Caiman crocodilus fuscus (sensu stricto) across trans-Andean populations in Colombia, based on concatenated CytB and COI sequences (2639 bp)
Full size table
Table 3 Caiman crocodilus FST values between regional populations in Colombia compared to the Florida population of trans-Andean origin
Full size table

Genetic diversity analysis of cis-Andean Caiman crocodilus/yacare populations

The cis-Andean group consisted of 439 sequences from Brazil, Peru, French Guiana, Colombia, Venezuela, Trinidad and Tobago, Paraguay, Bolivia, and the US (Florida). This included sequences of both C. crocodilus and C. yacare, which was found to be nested within the species complex. The haplotype network analysis of the CytB gene for this group yielded a total of 79 different haplotypes for the group based on 156 polymorphic sites, with an overall nucleotide diversity (π) of 0.0206. The haplotype with the largest number of sequences (H = 69) was haplotype 1 in the Brazilian Amazon Basin, which corroborates previous studies (Balaguera-Reina et al. 2021). Haplotype diversity ranged from 0.0 in Florida to 0.885 in Brazil (Table 4).

Table 4 Genetic diversity of Caiman crocodilus across cis-Andean Subspecies in different countries based on CytB sequences (1094 bp)
Full size table

We found extremely low nucleotide diversity in the C. c. crocodilus Florida population (π = 0.0) compared to all other cis-Andean subsets by country. The single C. c. crocodilus Florida haplotype had only 2–3 nucleotide mutations differentiating it from the two haplotypes of the Upper Branco clade (Fig. 3B). FST values for these groups were not significant (χ2 = 2779.14, d.f. = 702, p > 0.05) overall, but individual pairwise comparisons indicated very high levels of significant differentiation (> 0.25) between nearly all groups. When phylogeographic lineages were compared below the sub-specific level, the Upper Branco population of C. c. crocodilus had much higher genetic diversity (π = 0.001, hd = 0.343) compared to the C. c. crocodilus samples from Florida. The C. c. crocodilus Florida samples were most differentiated from C. yacare samples from Paraguay (FST = 0.98, p < 0.01) and Bolivia (FST = 0.98, p < 0.01), while they were least differentiated from C. c. crocodilus samples from Peru (FST = 0.63, p < 0.01) and Brazil (FST = 0.80, p < 0.01). When compared with the Upper Branco population, the C. c. crocodilus samples in Florida had very high levels of differentiation (FST = 0.82, p < 0.01). C. c. crocodilus subsets in both Peru and Trinidad and Tobago had significant deviations from neutrality (2.585 and − 1.831 for Tajima’s D, respectively, and 1.927 and − 2.196 for Fu and Li’s F, respectively). The C. c. crocodilus subset from Trinidad and Tobago also had a significant negative value for Fu and Li’s D (− 2.229). There were no values calculated for the Florida population of this lineage, as the genetic diversity was too low.

Genetic diversity analysis of Florida Caiman crocodilus populations

There were 3 haplotypes found for both the CytB and the COI genes in the Florida caimans. For CytB based on 34 sequences, there were two haplotypes separated by only a single nucleotide mutation, and a third haplotype separated by 37 mutations (Fig. 4A). The overall nucleotide diversity was 0.0158, and haplotype diversity was 0.506. The highly differentiated haplotype consisted of samples collected from the C-111 Spreader Canal northwest of Homestead, FL as well as the 8.5 Square Mile Area community (Fig. 4B). This haplotype clustered within the Upper Branco lineage from the cis-Andean group, while the other two haplotypes clustered within the C. c. fuscus (sensu stricto) lineage. There was a very large amount of genetic differentiation between these two groups (FST = 0.998, p < 0.001), consistent with the phylogenetic results reported above. Additionally, there was no genetic differentiation between sample localities within the same subspecies (average FST = 0.0), but extremely large levels between sample localities across the two subspecies (average FST = 0.99).

Fig. 4
figure 4

A Haplotype map of the Florida Caiman crocodilus samples based on the CytB gene, and B corresponding geographic localities of individual samples from South Florida. Note that the C-111 Spreader and 8.5 Square Mile Area localities cluster within a highly differentiated haplotype group. Each hash mark on the haplotype map (A) indicates a single nucleotide mutation

Full size image

Discussion

Our results support at least two distinct origins of the Florida caiman, one origin being the Magdalena River Basin and Caribbean coast of Colombia, and the second being the Upper Branco River Basin in the state of Roraima in northern Brazil. These two origins seemingly place the Florida caimans into two different currently recognized molecular lineages, with the Colombian-sourced individuals belonging to C. c. fuscus (sensu stricto) and the Brazilian-sourced individuals belonging to C. c. crocodilus (Upper Branco River basin). The population of the C. c. crocodilus lineage was composed of samples from two localities associated with the C-111 Spreader Canal Western Project between western Miami and the Everglades, while the C. c. fuscus population was found on the outskirts of the Everglades, near the Homestead Air Base, and the Biscayne Bay Coastal Wetlands (BBCW) Project to the southeast of Miami. These two populations exhibit very strong genetic differentiation (FST = 0.998), and do not share haplotypes with one another. With only three haplotypes and 38 variable sites, the Florida caimans have a relatively increased level of nucleotide diversity, and a slightly reduced haplotype diversity when compared to other countries’ populations in the native range. Thus, the Florida population follows the trend that the most genetically diverse populations are those that consist of multiple phylogenetically distinct clades. It should be noted however that some of the populations classified by country have a smaller sample size (e.g., Venezuela, N = 8) and this could affect the amount of detected polymorphism.

Neither the Brazilian nor Colombian-origin lineages in Florida have high levels of variation within their populations. The Brazilian C. c. crocodilus lineage in Florida has no nucleotide or haplotype diversity when compared to other populations in the cis-Andean range, and the Colombian C. c. fuscus lineage in Florida has reduced levels compared to native trans-Andean populations. The lack of diversity between the ten Florida C. c. crocodilus samples might be indicative of a single isolated introduction, with the current population descended from only a few founders. The founder effect is a common phenomenon in non-native species, and the introduced individuals have a lower level of genetic diversity compared with the source population as a whole (Mayr and Huxley 1954; Abdelkrim et al. 2005). The single haplotype of the C. c. crocodilus samples is further support for a very limited number of animals in the original source population. As such, neutrality tests for this population yielded values of zero.

A pairwise FST value of 0.82 was found between the C. c. crocodilus Florida population and the native Upper Branco source population. High FST values between source and introduced populations can be an indication of a founder effect where a population bottleneck drastically decreases genetic similarity between the populations (Abdelkrim et al. 2005). Alternatively, this could be a result of limited sampling in either the native or non-native ranges. This latter possibility should be seriously considered, as the samples from the Upper Branco region in Roraima, Brazil were collected by Roberto et al. (2020), who reported that the sampling of this cryptic lineage was limited to a relatively small area. They speculated that the range for this lineage could extend all the way northeast through the Guiana Shield to coastal Guyana and Suriname. Furthermore, Roberto et al. (2021) traced the origins of both introduced Paleosuchus species in Florida to populations in the Guiana Shield region. Guyana is a country known to export live wild-sourced specimens of Paleosuchus spp. and C. crocodilus in recent years (Sinovas et al. 2017). A Guyana or Suriname origin could explain why all the Florida C. c. crocodilus samples sourced from this molecular lineage consisted of a single unique haplotype not shared with any samples from northern Brazil. More extensive sampling in that region may better elucidate the genetic relatedness between the C. c. crocodilus population in Florida and the native population in the Upper Branco region.

The Colombian lineage has greater diversity than the Upper Branco lineage in Florida. It is possible that this lineage was the first population to be established, as the first reports on the Florida caiman suggested that Colombia was the source country, with the Caribbean port city of Barranquilla as the primary export hub for the species (Medem 1971; Busack 1974; Ellis 1980; Roberto et al. 2021). The higher level of diversity for this population is also likely the result of either a larger founder population than that from Brazil, or a founder population from a source with heightened levels of genetic diversity. We examined the genetic diversity of populations of C. c. fuscus (sensu stricto) in Colombia, as defined by Díaz-Moreno et al. (2021) and found variable levels of both nucleotide and haplotype diversity. The neutrality tests corroborated the findings of Díaz-Moreno et al. (2021) that regional populations had no significant genetic disequilibrium, but taken as a whole, C. c. fuscus (sensu stricto) was found to have significant negative deviations from neutrality for Fu and Li’s D and F, which could be caused by historic population expansion. We found no significant deviations from neutrality in either of the C. c. crocodilus or C. c. fuscus populations in Florida.

The Florida C. c. fuscus population did not have the lowest nucleotide diversity among native populations, and those with the greatest diversity were populations in the Magdalena River Basin. There were three shared haplotypes, with one haplotype consisting of individuals from five different geographically distinct populations from the Magdalena River and northeastern Colombia. The Florida caiman samples belonged to two shared haplotypes, one of which was the geographically diverse haplotype described above, while the other was a haplotype shared between the Upper Magdalena River Basin (UMRB) and the Paramillo region near the Panama border. Thus, phylogenetically, the Florida samples clustered closely with samples from multiple geographic localities.

The UMRB animals had the highest amount of mitochondrial genetic variation in the trans-Andean range of C. c. fuscus (sensu stricto). Díaz-Moreno et al. (2021) suggested that the high diversity within the UMRB was probably due to the ongoing translocations and restocking of caimans at localities within this region. The Florida samples had the second lowest level of genetic differentiation from the UMRB samples (FST = 0.38), and the lowest amount from the Lower Magdalena River Basin samples (LMRB; FST = 0.25). This lends support to Medem’s (1971) claim that a primary source for the population in Florida was likely in the coastal area near the city of Barranquilla, where the Magdalena River meets the Caribbean. However, the lack of strong geographical structure and the presence of gene flow between Colombian populations make it difficult to pinpoint the exact localities from which the Florida caimans descend.

While the genetic origins of the Florida caiman appear to be Colombia and Brazil, some caution should be exercised when concluding that these countries were the actual export sources. Trade records from the late 1970’s indicate that in several cases subspecies non-native to the exporting country were sent to the US. For example, 1979 CITES trade records indicate that almost 5,000 live individuals of C. c. crocodilus were imported to the US from Guatemala, a country that has no native populations of that subspecies (CITES Secretariat and UNEP-WCMC 2022). Between 1950 and 1990 C. crocodilus were imported to the US in the tens of thousands largely for the booming pet trade that coincided with the new regulations surrounding trade in American alligators (Busack 1974; Ellis 1980). Reliable early records of imports prior to the mid-1970’s are scarce and insufficient to definitively pinpoint the export sources that led to the established populations we see today.

Obtaining information on the genetic origins serves as an important baseline in monitoring future changes in genetic diversity within the Florida caiman populations. One important management objective for these animals should be to examine and monitor the extent to which the two differentiated populations in Florida come into genetic contact with one another. We recommend that the two distinct lineages in Florida be managed as separate populations, and that measures should be taken to prevent gene flow between them. Investigation into waterway connectivity between the groups should also be a high priority to see whether there are barriers to gene flow. The population of C. c. crocodilus from Brazil appears to be concentrated in the area around the C-111 Spreader Canal Western Project of Florida between the suburbs of western Miami and the Everglades. Given the low genetic diversity and geographically limited range in Florida, this population may be more easily contained than the more widespread and diverse population of C. c. fuscus. Contact between the Florida populations could lead to genetic admixture, which in some species has been shown to increase the success of biological invasions (Kolbe et al. 2004; Wolfe et al. 2007; Kraus 2015; Lawson Handley et al. 2011; Rius and Darling 2014). Any admixture between the C. c. crocodilus and C. c. fuscus lineages in Florida could increase both the success and severity of the Florida caiman invasion, and further genetic analysis may be necessary to detect admixture early on before it becomes widespread.

We propose that future studies of the Florida caiman populations examine nuclear loci to corroborate the mitochondrial data, and to more closely examine genetic relationships within and between populations throughout Florida. Microsatellite genotyping has proven useful in elucidating population structure and detecting gene flow under a range of demographic parameters in crocodylians (Dever and Densmore 2001; Dever et al. 2002; Ray et al. 2004). One study by Rodriguez et al. (2011) used microsatellite data to detect admixture of American crocodile (C. acutus) populations in Florida due to human-mediated translocation, and it can be assumed that similar effects could occur as Florida caiman populations of different origins come in contact with each other. Microsatellite data can also be used to estimate the number of functional populations and introduction pathways of invasive species through approximate Bayesian computation (Guillemaud et al. 2010; Benazzo et al. 2015). These methods were used by Wood et al. (2016) to study gene flow between introduced populations of the Nile Monitor (V. niloticus) and assess likely introduction scenarios. A similar analysis could be useful in resolving population dynamics of Florida caimans. By controlling the populations and preventing gene flow, it may be possible to more effectively manage C. crocodilus in Florida and reduce its impact on local ecosystems.