Since 2004 tissue samples have been taken from most of the individuals comprising the ex-situ population in charge of the EBTRF. Scales and muscle samples were preserved in pure ethanol and kept at -20 °C until processing. We searched EBTRF records to clarify the geographic origins, status, and current location of each crocodile. All the animals were microchipped for individual identification.

In total, we included 551 individuals in the study. The complete dataset includes 40 crocodiles that were wild in origin (Suppl. material 1: appendix S1). They were either acquired through direct captures as wildlife, from the breeding center or as confiscations. These animals included 24 seized individuals whose provenance was unknown (13 were dead in 2021), nine young individuals from Cravo Norte River (by 2021 one was dead and eight were 11 years old), and seven dead wild founders from the Cuisiana River, the Metica River, the Humea River, the Meta River, the Guachiría River, and the Vichada River. The remaining samples corresponded to 458 individuals of the captive offspring (F1 and F2) and 53 released crocodiles.

From the living individuals we evaluated, 82% belonged to the captive breeding program from five subpopulations: 316 were from the main headquarters of the program at the EBTRF in Villavicencio, Meta department; 19 were from the Parque Agroecológico Merecure in Puerto López, Meta department; five were from the Bioparque los Ocarros in Villavicencio, Meta department; four were from the Aquatic and Conservation Park Piscilago in Nilo, Cundinamarca department; and 127 were from the Parque Ecotemático Wisirare in Orocué, Casanare department. The remaining samples were from the two largest subpopulations: 44 from the EBTRF and 56 from Wisirare.

Genomic DNA was extracted from preserved tissue using the Invisorb Spin Tissue Mini Kit (Stratec) following manufacturer protocols. Seventeen microsatellite primers developed for other species of the genus and already evaluated for cross amplification by Lafferriere et al. (2016) were used. We implemented four PCRs multiplex using the Multiplex PCR kit MyTaq HS Mix (Bioline, USA) M1: CpP302, CpP305, CpP314, CpP1409, and CpP3216; M2: Cj16, Cj122, and Cu5123; M3: Cj18, Cj109, C391, CUJ131 and M4: Cj101, Cj127, Cp801, and CpDi13. Reactions were prepared in a final volume of 10 μL including: 5 μL of MyTaq HS Mix, 0.2 μL of 10X each primer (except for Cj122 and Cj109 with 0.4 μL added), a final concentration of 4 ng/ μL of DNA and the excess of ultra-pure water. Thermocycling conditions were as follows: a preliminary denaturation stage at 95 °C for 4 minutes, followed by 30 denaturation cycles at 95 °C for 30 seconds, two different annealing temperatures (57 °C for M1, M2 and M4 and 60 °C for M3) for 45 seconds and extension at 72 °C for 30 seconds, ending with a temperature of 72 °C for 5 minutes. Fragment lengths were determined using an ABI 3500 genetic analyzer. For this purpose, 1 µl of the PCR product was diluted in 99 µl water; 1 µl of this dilution was mixed with 8.5 µl Hi-Di Formamide (Applied Biosystems), 0.25 µl water and 0.25 µl GeneScan-600 LIZ Size Standard (Applied Biosystems). The Gene-Mapper 3.7 (Applied Biosystems Foster City, CA) and Osiris 2.13.1 (NCBI) software were used for scoring fragment lengths. A single person carried out the visualization and determination of allele sizes. Genetic laboratory work was conducted at the Molecular Ecology Laboratory of the Genetics Institute, National University of Colombia in Bogotá.

Our data set represents 82% of the living crocodiles of the Station and 91% of the F0 population. Of the missing founders, three corresponded to juveniles from Cravo Norte that had not reproduced to date and only one founder breeder from which no tissue sample was taken. A total of 72 alleles were revealed: 69 in F0 crocodiles and 65 in live crocodiles (89.9% of the F0 alleles, Table 1). The F0 population had eight private alleles while the live population had three, suggesting that in the living population we found wild individuals or non-genotyped parents. The average number of alleles per locus, the average allelic richness, the expected heterozygosity, and the inbreeding coefficient F_IS were higher in the F0 population than in the live population (Table 1). Fourteen loci in the live population showed significant deviation from the Hardy-Weinberg equilibrium. No significant linkage disequilibrium was found between pairs of loci. In the F0 population two loci had deviations in the F_IS coefficient for heterozygous defect and one locus for heterozygous excess. The live population showed deviations in the F_IS coefficient in seven loci for heterozygous excess (Table 1).

Even though the live population showed a higher Ho than the F0 population, differences between each group were not significant (H0 p = 1). Likewise, although the F0 crocodiles showed generally higher AR and F_IS, differences were statistically not significant (AR p = 0.332, FIS p = 0.332). Although there were loci where allele frequencies did not change considerably between F0 and the live populations (e.g., CpP3216, Cj127; Table 2), there were other loci that showed strong changes, and even a loss of alleles (e.g., Cj109, Cj18, Cj391, Cpp801; Table 2).

Of the 399 individuals evaluated, 325 (81.5%) had potential fathers in the subpopulation of origin registered; and for 74 (18.5%) individuals the physical registry does not correctly indicate the origin of these crocodiles (Table 3). When testing the relationships between the founding crocodiles that came from the same sites, we found that several were related (Table 4). Although the nine individuals from Cravo Norte showed unrelatedness in some cases, most were related as half and full-siblings. We found a similar result for the seized crocodiles from the Rango Rudd hatchery.

The number of alleles differed among the five subpopulations (Table 2). As expected, the largest subpopulations (EBTRF and Wisirare) showed unique alleles and the highest number of alleles (63 and 53 alleles, respectively), while the population with the lowest number of alleles was Piscilago (43 alleles). Although the other subpopulations did not show unique alleles, they did have alleles at very low frequencies (e.g., allele 203 locus Cj18 was present in five individuals or allele 193 locus Cj131 in only three individuals; Suppl. material 1: appendix S2). We identified and prioritized the management guidelines of 76 individuals that contained those rare alleles with low frequencies. The individual diversity (HL) of the living individuals that make up the entire ex-situ population varied between 0.075 and 0.947. However, 95.5% of the individuals that had an index lower than 0.6 and 74.8% were grouped between 0.2 and 0.5 (Suppl. material 1: appendix S3). No differences were found between the HL of founding crocodiles and living crocodiles (p-value = 0.292; α = 0.05).

We found that in some cases current combinations of individuals are not the most appropriate when considering their genetic profiles. For example, the EBTRF represents the largest of the C. intermedius subpopulations and contains 97% of the alleles from the entire captivity program including 55 priority crocodiles and three unique alleles; but the subpopulation has no active reproductive nucleus. Piscilago has an F0 priority male in an isolated tank only for exhibition. The three males found in Ocarros are priority crocodiles since they have scarce alleles, but two of them are related to the females located there and they have not contributed to the growth of the captive population. Since genetic parameters for the selection of reproductive individuals must be urgently considered, we proposed changes and reorganized crocodiles in the subpopulations with combinations that guarantee the recovery of rare alleles and minimize the mean kinship. All the parental combinations were assembled by the combination of the r and HL indexes with important complementary information regarding every single crocodile (i.e., size, age, sex, origin, current location, capacity of the tanks). Using this information, we considered the priority crocodiles identified with the allele frequencies, combining them with unrelated crocodiles of reproductive age that showed the lower HL. We also considered whether the selected individuals had the appropriate size and health status, as well as if they had normal growth according to the growth model estimated for the EBTRF.

We reorganized the individuals that make up the reproductive nucleus of Ocarros and the two nuclei of Piscilago, and we selected the individuals of the two new nuclei from the Universidad de los Llanos. In Wisirare we proposed not to make changes considering that transport to Wisirare is complex, and since we found an unrelated kinship level and a low HL in the individuals that made up the breeding stock. For now, we recommend considering only the six reproductive nuclei mentioned above (Table 5). Once the individuals are reorganized, we provided program officials with the necessary information to establish new pairs in EBTRF and Merecure that will depend on the capacity of the tanks. We selected six males not suitable for reproduction (with no priority and high HL) to be taken to Piscilago (one individual) and to the Hacienda Nápoles Park in Puerto Triunfo, Antioquia (five individuals). These individuals are to take part in environmental education and awareness-raising, but not for reproduction. Finally, we selected 84 juvenile individuals that have the genetic, size, health, and age requirements for release (sizes less than 2.5 m and HL < 0.6), and we moved them to a tank under semi-natural conditions in Merecure park.

The expected heterozygosity obtained in the currently living crocodiles of the EBTRF is similar and even higher than that reported for wild populations of other species of the genus Crocodylus, evaluated with the same loci (e.g., Isberg et al. 2004; McVay et al. 2008; Mauger et al. 2017). This shows that, although the captive population of the Orinoco crocodile experienced an allele loss compared with the founder population, it maintains an important part of the variability in terms of heterozygosity, with genetically viable individuals to be reproduced and used for conservation and management. This is because few breeding pairs comprised unrelated and genetically diverse individuals and there has not been a generational turnover that might result in reproduction between relatives.

Our results showed no statistical difference between observed heterozygosity and the allelic richness between the live and the founder populations. However, a decrease in variability was detected by the loss of alleles (Table 2). This phenomenon has already been reported in ex-situ population of other reptiles: in a captive population of the northern river terrapin Batagur baska, the first generation maintained only 73% of the founder alleles, while there were no significant differences in the heterozygosities between founders and the first generation (Spitzweg et al. 2018). Furthermore, in captive populations of the Jamaican yellow boa Chilabothrus subflavus, a loss of genetic diversity due to the differential reproduction of individuals was detected in the first generations by the allelic richness and not by the heterozygosities (Tzika et al. 2008). This reflects the limited efficiency of tests based on heterozygosity variations to detect recent inbreeding (Luikart et al. 1998; Tzika et al. 2008). A similar situation was detected in the EBTRF, where variations in allele frequencies showed that only a few reproducing founders segregated alleles to the next generation. Consequently, knowing the genetic profile of individuals is crucial for developing strategies that prevent genetic loss.

One of the objectives of captive breeding programs is to guarantee the survival of the offspring, which can be compromised by phenomena such as inbreeding and captive adaptation (Farquharson et al. 2021). We found that the inbreeding coefficient is higher in F0 than in the live population and in the living population we have deviations in seven loci due to excess heterozygotes while in the F0 population there are no loci with this deviation. This may be because, initially, many confiscated individuals came from the same breeding farm and thus may be closely related. However, most of these crocodiles did not reproduce; and, if they did, they were combined with wild or seized crocodiles, decreasing the F_IS of the living ones. For the future management of the program, generational change must be considered since species’ response to captivity adaptation may differ in the first generation in relation to subsequent generations (Farquharson et al. 2021). Captive adaptation can improve fitness at the population level in the captive environment, but when animals are returned to the wild, captive adaptations can be maladaptive and contribute to the poor success of reintroduction programs (Frankham 2008; Jule et al. 2008).

The captive breeding program of the EBTRF plays a key role in Orinoco crocodile conservation. Nonetheless, management of these captive populations was not guided by the standards necessary to conserve and maximize genetic diversity, despite the previous recommendation for genetic monitoring (see Williams and Osentoski 2007). Furthermore, genetic management has been recommended within the actions framed in PROCAIMAN (MAM 2002). The only genetic characterization of the EBTRF ex-situ population was carried out by Cuervo-Alarcón and Burbano-Montenegro (2012). In this study, the sampling coverage of the captive population was limited; of the seven wild founders, only one was used in the analysis. We genotyped the same crocodiles and obtained different numbers of alleles for the same loci (e.g., 16 alleles at locus Cj16 versus five in our study). Furthermore, their division of the samples in two groups based on age is inadequate, returning inconsistent results with an admixture of F0, F1, and F2 generations in the established groups, as well as an erroneous determination of crocodiles with wild origin.

The only other study of C. intermedius population genetics considering wild individuals was carried out in Hato El Frío in Venezuela by Lafferriere et al. (2016) and aimed at reporting multiple paternity in the species. When comparing the EBTRF population with El Frío Biological Station population, the Venezuelan individuals have a greater diversity in terms of allele composition with 90 alleles in the 17 loci. It is remarkable that the locus CpP1610 was monomorphic in our study while in Venezuela it was polymorphic with two alleles, but with one allele more frequent than the other (Lafferriere et al. 2016). However, the overall He / Ho level in the EBTRF was a little higher than in Venezuela (0.622 / 0.575 vs 0.524 / 0.544). This difference was probably due to different ways in which individuals reproduced since even though the Venezuelan individuals were born from reintroduced individuals, they follow the principles of a natural population; while, in the EBTRF they have been dependent on arbitrary human management that has reproduced the same individuals without a generational change. Finally, unlike the Venezuelan population that did not show significant deviations from Hardy-Weinberg expectations at any locus (Lafferriere et al. 2016), in the EBTRF we found deviations in 14 loci for live crocodiles. These deviations were expected since the animals originated from a few breeding pairs crossed without scientific basis or management.

The breeding program for Crocodylus intermedius in Colombia aims to preserve and increase as much as possible the current genetic diversity and to produce neonates with the highest genetic diversity possible to support management actions (MAM 2002). Based on our data, we suggest that selective breeding should be implemented, and some mating combinations should be avoided. To achieve this goal, we proposed a robust system of 16 polymorphic microsatellite loci for estimating the relationship, the individual diversity and rarity of the living crocodiles. This, combined with information of age, size, sex, and location, allowed us to design combinations for planning breeding groups in each subpopulation. The tool enables a simultaneous maximization of genetic diversity, combining non-related diverse individuals and individuals containing rare alleles to achieve a genetic gain by minimizing the relationships between the individuals combined and guaranteeing no loss of alleles in the following generations.

Our work is necessary and complements the previous data, since most captive breeding projects are not monitored genetically, and only recently attention has been paid to the pedigree or relatedness of breeders using conservation genetic approaches (e.g. Spitzweg et al. 2018). Furthermore, this is one of the few studies combining relatedness information with the homozygosity by loci, which can be very useful when the number of individuals involved is large, allowing discrimination only when the r index may not be enough.

Despite the living crocodiles of our sample retaining approximately 90% of the genetic diversity of the wild-caught founder individuals with the presence of three unique and 13 rare alleles, the difference in the number of alleles and the allele frequencies among the five subpopulations revealed that the diversity is unevenly distributed between groups. If no action is taken to balance this, the loss of rare genetic diversity in the next few generations could be drastic, jeopardizing the viability of the program (Groombridge et al. 2012). To solve this, we explicitly recommend using the combination of genetic data with the information from the archive records so as not to rely solely on the latter; since, as we found when performing the parent pairs analysis, a large part of the archive files is wrong in determining the origin of the individuals. Finally, we explicitly recommend implementing conservation genetic assessments for other captive breeding projects to preserve maximum genetic diversity and to avoid inbreeding depression (recommended by Xu et al. 2005; Tzika et al. 2008; Spitzweg et al. 2018).

The EBTRF conservation program covers a very restricted range of the historical natural distribution of the species in Colombia, and key individuals (e.g., from Vichada department) had rare alleles, suggesting that the genetic diversity of the Station does not cover the unknown threatened possible diversity available in the wild. It is necessary and urgent to evaluate wild populations, as well as to enrich the diversity of the Station ́s population by including wild individuals from unsampled sites (e.g., Guayabero / Duda / Lozada Rivers). These individuals must be genotyped to determine the presence of rare alleles, individual genetic diversity, and degree of relationship. As we demonstrated here, the basic assumption of unrelated founders may be incorrect, particularly given the often-imprecise nature of information on their origin (Gautschi et al. 2003). We recommended completing the dataset with the missing crocodile samples and including them in the management guidelines. In turn, it is necessary to genotype the crocodiles that are going to be born to have a complete genetic profile of the program, to evaluate future trends in allele frequencies and to restructure combinations if necessary.

The EBTRF contains the largest subpopulation (about 370 individuals), the largest number of tanks available, and a high genetic diversity involving three unique alleles. More than 150 crocodiles have passed through the EBTRF and have died from recent hatchlings to the first clutches of 1991 and the F0. After 2005, fewer eggs from the EBTRF were incubated since eggs from Wisirare, Piscilago and later Occarros began to be carried to the Station for incubation. Considering that the EBTRF subpopulation has the highest number of adult crocodiles with unique diversity, it is necessary to re-implement the breeding stock with these individuals. It is urgent to maintain a balance in the proportion of eggs incubated according to their origin and the number of parents that produce them. In the EBTRF we found juvenile individuals that we considered as priority because they contained alleles at low frequencies (Suppl. material 1: appendix S2). However, these individuals have not attained reproductive age, so we recommend keeping them until they can be included in the reproductive nuclei.

Through the implementation of the crosses proposed here, the program will ensure highly genetically variable offspring that preserve the available genetic diversity. By combining the offspring produced by different reproductive pairs, we will be able to form groups of unrelated and highly diverse individuals that, according to the requirements of natural populations, could be released into the wild.

This research supports the actions defined in PROCAIMAN to advance the recovery of populations of the Orinoco crocodile in Colombia. This is of urgent application since, even though management actions were established 20 years ago, the natural situation of the species has apparently not changed or even has deteriorated (see Medem 1981; Lugo 1996; Seijas et al. 2010; Espinosa-Blanco and Seijas 2012; Babarro 2014; Parra-Torres et al. 2020). By using the genetic system developed here, it is urgent to genetically characterize wild populations to define whether they need genetic management. The evaluation of the captive populations together with natural populations, as well as demographic and ecological studies, should guide the recovery of eggs and hatchlings both ex-situ and in-situ and define a reintroduction protocol, including the monitoring of introduced and wild populations.

However, more support and research are needed to comply with what has been established in PROCAIMAN (MAM 2002). Currently, neither the population sizes nor important ecological characteristics are known, such as whether there is reproduction in wild populations, the magnitude of reproduction, physiological parameters associated with reproductive events, nesting times, demographic structure of populations, etc. As a crucial component for the success in the recovery of the species, focused environmental education and awareness initiatives must be developed. These efforts should aim to foster actions and establish an extensive dialogue with human communities regarding the coexistence of this Colombian crocodile and its recovery.

The author has declared that no competing interests exist.

No ethical statement was reported.

The development of this project was possible thanks to the financial support of the agreement N°PE.GDE. 1.4.8.1.20.009 between the Corporación para el Desarrollo Sostenible del Área de Manejo Especial La Macarena (CORMACARENA) and the Science Faculty of the Universidad Nacional de Colombia (UNAL): “Aunar esfuerzos técnicos, científicos y financieros para realizar la caracterización genética por medio de marcadores moleculares nucleares variables (microsatélites) de la mayor población cautiva del críticamente amenazado caimán del Orinoco (Crocodylus intermedius), presente en la Estación de Biología Tropical Roberto Franco (EBTRF)” and the Convocatoria para el Fortalecimiento de Alianzas Interdisciplinarias de Investigación y Creación Artística de la Sede Bogotá de la Universidad Nacional de Colombia (UNAL)-2018: “Genética de la conservación de la mayor población cautiva del críticamente amenazado caimán del Orinoco (Crocodylus intermedius): una contribución para su supervivencia”.

Conceptualization: MVR, AMSG. Data curation: AMSG, MVR. Formal analysis: AMSG, MVR. Funding acquisition: AMSG, MVR. Investigation: MVR, AMSG. Methodology: AMSG, MVR. Project administration: AMSG, MVR. Resources: AMSG, MVR. Software: AMSG, MVR. Supervision: MVR, AMSG. Validation: AMSG, MVR. Visualization: AMSG, MVR. Writing – original draft: MVR, AMSG. Writing – review and editing: AMSG, MVR.

Ana M. Saldarriaga-Gómez https://orcid.org/0000-0003-3466-6313

Mario Vargas-Ramírez https://orcid.org/0000-0001-8974-3430

All of the data that support the findings of this study are available in the main text or Supplementary Information.