Coiba National Park (CNP) comprises an offshore archipelago of around 35 islands and islets of different areas off the coast of the Panama Pacific. This archipelago was once attached to the Mainland and separated from it around 12,000 ya, creating a barrier for further species colonization. In this study, the herpetological richness, and community assemblages were assessed along four different islands of CNP. Amphibian richness was lower in the islands (maximum five spp.) compared to the mainland (11 spp.); meanwhile, reptile richness was similar between the mainland and Coiba island (21 spp.). The area of the islands was positively correlated with amphibian species richness but not with reptile richness and no effect of distance was found. The herpetological communities were less similar between the Mainland and the islands of CNP, possibly reflecting not only dispersal limitations but also specific adaptation to the more pristine habitats of the latest. Based on these results, CNP may harbor a unique herpetological fauna with local adaptations to the habitat, influenced by other elements and past anthropogenic factors that need to be investigated in more detail.

Herpetological Surveys

Surveys were conducted between July 2019 and August 2023 along transects in 18 sites from the Mainland in front of CNP and on four islands of CNP archipelago (Figure 1). The islands surveyed included Canales de Afuera, Rancheria, Coiba, and Jicaron (Table 1). Several field trips occurred each year except for 2020 when only one site was visited (Table 1_SuppInfo).

Visual encounter surveys (VES) (Crump and Norman, 1994; Rodda et al., 2007) were used to identify metamorphic amphibians and terrestrial reptiles. VES were conducted along one to six randomly deployed 2 m wide and 100 m long transects at each site during day and night without time constraints. Each transect was separated at least 20 m from each other, and two people constantly surveyed the transects. Search effort (in person-h) was calculated by multiplying the total search time at each site by the number of observers (Maritz and Alexander, 2012), excluding time allocated to manipulate captured individuals. This was necessary to ensure a clearer identification of the species, moreover at night, as it has been occurred during other surveys using VES (Vences et al., 2008; Metcalf et al., 2020; Quilumbaquin et al., 2023). The capture rate at each site was calculated by dividing the number of identified individuals by the total search time in each site (Linares and Eterovick, 2013). Assuming that some species cannot be detected, pitfall traps were used (Todd et al., 2007; Kenneth Dodd Jr., 2010). When possible, one or two arrays of pitfall traps were deployed at each site and were separated at least 100 m from transects to provide independent sampling units. Individuals were identified in the field at the species level and compared with dichotomic keys (Köhler, 2008, 2011).

Data Analysis

To estimate α diversity, species richness (Sobs) was estimated (Magurran, 2021). In addition, individual-based rarefaction curves with 95% confidence intervals based on the Mao Tau function (Colwell, 2006) were produced using the software Estimates v9.1.0 (Colwell, 2006). To determine β diversity the Jaccard dissimilarity index with the presence–absence of species was calculated using the vegan package (Oksanen et al., 2022). Clusters were produced using the hierarchical clustering algorithm using the hclust function and the unweighted pair group method with arithmetic mean to create dendrograms using the package BiodiversityR (Kindt and Coe, 2005). Counts of individual specimens were averaged across all visits and each transect was visited between one to five times (Marsh and Haywood, 2010).  Data in all transects from sites (1) to (6) were assigned to Mainland, and those from (9) to (17) were assigned to Coiba island (Figure 1). An updated biodiversity repository for CNP does not exist and specimen collections are difficult to access. Hence, the surveys were further supplemented with occurrences of amphibians and terrestrial reptiles from 2018 to 2023 downloaded from the Global Biodiversity Information Facility (GBIF) website (www.gbif.org). These occurrences were added for the α and β diversity analyses. GBIF is an open access platform that integrates data resulting from specimen surveys and collections worldwide (Heberling et al., 2021) and is being used in herpetological biodiversity assessments (Sillero et al., 2014; Vasconcelos and Nascimento, 2014). The point-occurrence records were obtained using the draw area tool encircling the study islands in CNP and a section along the mainland coast where the study sites were located. Data curation before analysis included the deletion of those: without geographic coordinate, without specific year, and without a clear description of location. Scientific names were also validated to check for synonyms and to abide to actual scientific nomenclature. From 121 occurrences in total, 51 individuals (45 reptiles and 6 amphibians) belonging to 14 reptile and 3 amphibian species, were included and assigned to the closer study site. Presence–absence data of additional species of amphibians and terrestrial reptiles reported by De la Riva (1997a, 1997b), and Pérez-Santos and Martínez (1997) for some islands of CNP and by Martinez and Rodríguez (2003) for Mainland, were also included in the β diversity analyses assuming these species are extant species (Table 2_SuppInfo; Table 3_SuppInfo). Pearson product–moment correlations were conducted to test the effect of island area and distance from the Mainland on species richness. Statistical analyses were conducted using the software R v4.2.2 (R Core Team, 2023). Data presented here are mean ± SD (range) unless otherwise indicated.