https://orcid.org/0000-0003-1050-5849
Abstract
Montane cloud forests are highly threatened ecosystems that are vulnerable to climate change. These complex habitats harbor many species that suffer the negative consequences of this global phenomenon, such as shifts in their distribution and habitat use. The Central American clouded oncilla (Leopardus pardinoides oncilla) is the smallest and most endangered wild cat in Mesoamerica and is primarily reported in cloud forests throughout its distribution. The species is poorly understood, with no studies conducted in Central America assessing its habitat preferences. To bridge this knowledge gap, we sampled two mountain ranges in Costa Rica with camera traps and conducted an occupancy analysis to understand the anthropogenic and environmental features that influence oncilla habitat use within them. Additionally, we conducted spatial predictions of habitat use across its northern and southern range in Costa Rica to identify priority conservation areas for the species. We found that Central American clouded oncilla habitat use is driven primarily by environmental factors. Our results showed that oncillas select habitats with denser tree cover at high elevations, closer to permanent water sources, which may provide them with high prey density and a favorable habitat structure for their survival. Spatial predictions identified two main regions as conservation priority areas where threat mitigation efforts and monitoring should be implemented: the Caribbean slope of the Talamanca mountains, and the Arenal-Monteverde forest complex. The occupancy modeling approach turned out to be very useful to assess the spatial associations of the species with the environment and mapping the conservation priority areas. Future research and mitigation actions should focus on potential threats that could negatively impact Central American clouded oncilla populations and habitat use, including the role of mesopredators and feral species.
Figures
Citation: Ramírez-Fernández JD, Fox-Rosales LA, Mooring MS, Delgado-Carazo JC, Blankenship SR, Powell JR, et al. (2024) Distribution and habitat use patterns of the endangered Central American clouded oncilla (Leopardus pardinoides oncilla) in Costa Rica. PLoS ONE 19(9): e0310562. https://doi.org/10.1371/journal.pone.0310562
Editor: Shijo Joseph, Kerala University of Fisheries and Ocean Studies, INDIA
Received: April 16, 2024; Accepted: September 3, 2024; Published: September 17, 2024
Copyright: © 2024 Ramírez-Fernández et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The Oncilla Conservation project was funded by the Small Wild Cat Conservation Foundation, the Mohamed bin Zayed Species Conservation Fund (# 190520765), the Rufford Foundation (#29081-1), and the Shared Earth Foundation. MSM was funded by Point Loma Nazarene University and San Diego Zoo Global in collaboration with the Talamanca Elusive Mammal project and the Quetzal Education and Research Center (QERC). TGO received a productivity fellowship from the National Council of Technological and Scientific Development (CNPq) of Brazil. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The Central American clouded tiger-cat or oncilla (Leopardus pardinoides oncilla, Figs 1 and 2) is part of the Leopardus “tigrina” species complex. It was recently suggested that the Central American clouded oncilla is a subspecies of the clouded tiger-cat (L. pardinoides), a former subspecies of the complex which was recently elevated to species level [1]. The clouded tiger-cat ranges discontinuously from the southern Central American montane forests of Costa Rica and Panama, through the Andean chain of South America, to northwestern Argentina. The Central American subspecies is the smallest (1.5–3.5 kg), rarest, and most endangered wild cat in all Mesoamerica [1–4]. Categorized as “Vulnerable” in the IUCN Red List under L. tigrinus [5], and as “Endangered with extinction” by Costa Rican environmental authorities [6, 7]. Across its distribution it has been associated with different habitats, although Central American populations have been reported to occur mainly in montane cloud forest [2, 5].
These tropical montane ecosystems are known to be highly threatened by climatic change [8–10]. Increases in temperature, dry season length, and number of dry days are just a few examples of the consequences produced by climate change in tropical highlands [11]. In Costa Rica, the negative effects of this global phenomenon have been modeled for highland amphibians, reptiles, and birds [11–13]. This situation forces many middle elevation species to change their distribution to higher elevations, thereby changing species composition and interactions in the altitudinal gradient [14, 15].
Cloud forests are particularly vulnerable to climate change because (1) their existence requires specific environmental conditions [16–18], (2) they have a natural patchy distribution [19, 20], and (3) much of their original distribution has already disappeared [21, 22]. It has been suggested that the ecological complexity of cloud forests makes them an “archipelago in the highlands” [23]. Considering that these ecosystems are under strong pressure from human activities, have great hydrological value [24, 25], high biodiversity and endemism [20, 23], montane cloud forests must be prioritized for conservation.
The Central American clouded oncilla is not only the most endangered wild cat in Mesoamerica, but also an endemic subspecies restricted to a highly susceptible and fragile habitat in Costa Rica and Panama. Because felid predators play a key role in regulating prey populations and structuring animal communities, establishing their current distribution and abundance is a priority issue for conservation action. The application of modelling techniques for estimating species distributions is a crucial tool for assessing the environmental factors driving the occurrence of rare species such as the oncilla. There are several approaches to model species distributions, with some techniques using presence only data and others using presence absence data. These models have been used to delineate species distributions at a global scale [1] and make predictions of future distribution under climate change scenarios [26]. A major issue in modeling the distribution of elusive species is accounting for detectability during the sampling stage. Failure to account for lack of detection when a species is present, may result in less precise inferences of species distribution and habitat associations [27]. Occupancy models estimate the distribution of a species while accounting for imperfect detection, by modeling the detectability of the species at a given site and thereby reducing the risk of false absences [28]. This makes occupancy analysis useful for assessing the drivers of species occurrence while distinguishing between ecological and observational processes.
Regarding the “tigrina” species complex split after de Oliveira et al. [1], two clouded tiger-cat subspecies are recognized, the Andean clouded tiger-cat L. pardinoides pardinoides in South America and the clouded oncilla L. pardinoides oncilla in Central America. Research on the clouded tiger-cat (under L. tigrinus) has been performed on South American populations in order to understand its ecological requirements and monitor its populations [29–32], allowing to identify areas where conservation efforts should be enforced for this species to thrive. However, for the Central American populations no research has been implement on these subjects.
In this study, we identified the environmental features that had the greatest influence on habitat use by Central American oncilla in Costa Rica; we also predicted oncilla habitat use in two regions of the country, enabling conservation efforts to be allocated appropriately. We tested the following hypotheses, oncilla habitat use is: (1) mainly driven by vegetation structure and topography; (2) negatively influenced by anthropogenic activity; and (3) driven by a combination of environmental and anthropogenic features.
Materials and methods
Study site
The study was conducted in two mountain ranges in Costa Rica spanning protected and unprotected areas (Fig 3). In the Arenal-Monteverde Protected Zone, within the Tilarán mountain range, in the northwest part of the country, we surveyed the Monteverde Cloud Forest Preserve. In central and southern Costa Rica, we sampled several protected areas of the Talamanca mountain range: Chirripó National Park (CNP), Los Quetzales National Park, Tapantí Macizo de la Muerte National Park (TMMNP), La Amistad International Park, and Las Tablas Protected Zone, and a mix of private reserves and unprotected areas within the CNP and TMMNP buffer zones. Vegetation sampled at most sites consisted of montane cloud forest, with some areas of elfin forest and paramo habitat.
Base imagery derived from the United States Geological Survey, Digital Elevation–Shuttle Radar Topography Mission (SRTM) (https://www.usgs.gov/).
Climate at all sites consisted of a dry (December-May) and wet season (June-November), and differences in rainfall between sites, with the Talamanca region being wetter than Arenal-Monteverde (S1 Table).
Because camera traps are a non-invasive tool with no animal captures involved, this study did not required clearance from an animal ethics board. This study is framed under MINAE and CONAGEBIO research permits (resolutions SINAC-ACC-PI-R-044-2020, R-SINAC-PNI-ACLAP-028-2020, and R-001-2021-OT-CONAGEBIO).
Occurrence records
We compiled all the published data and available unpublished records of oncillas in Costa Rica between 2011 and 2022, including camera trap data, live records with photographic evidence, and carcasses. The records of oncilla were compiled with location details and GPS coordinates of each sighting. For this study, we used our own data (Oncilla Conservation–OC) and gathered additional camera trap data from four collaborators: (1) Tropical Science Center Monteverde (MV); (2) Nai Conservation (NAI); (3) Cloudbridge Nature Reserve (CNR); and (4) Quetzal Education & Research Center (QERC). We compiled a total of 712 independent records of clouded oncilla. Records were checked for accurate species identification to avoid common mistaken identities with margays (Leopardus wiedii) and ocelots (Leopardus pardalis).
Sampling design
We set 65 camera trap stations in specific areas of the Talamanca mountain range where we did not have data from collaborators and where habitat characteristics matched reports for oncilla in Costa Rica. The exact location of the camera trap stations was based on accessibility. Excluding stolen and damaged cameras, we collected data from 53 camera trap stations from 2019 to 2022 for the analysis. Prior to installation of cameras, we removed vegetation in the camera detection field that could interfere with the motion sensor to reduce the number of "ghost photos" caused by windblown vegetation. At each station, a single camera trap was placed at a height of ca. 40 cm on a natural (game) trail created by recent wildlife activity (e.g., tapir trails). We used several brands and models of passive infrared camera traps for each sampling station (BUSHNELL ® Trophy Cam HD Essential and Trophy Cam HD Aggressor, and BROWNING ® Strike Force HD Max and Dark Ops HD Max). We set the cameras to take a sequence of three photos with the minimum delay possible (0–5 s).
A similar sampling design from MV, NAI, CNR, and QERC databases is detailed in [33–36], respectively. Because camera traps are a non-invasive tool with no animal captures involved, this study did not require clearance from an animal ethics board.
Covariates
We assessed a total of eight covariates to model oncilla habitat preferences (Table 1). This included six anthropogenic and environmental covariates to model the occupancy parameter: a) percent tree cover; b) distance to permanent water bodies; c) distance to nearest paved road; d) elevation; e) terrain ruggedness index; f) distance to nearest settlement (Table 1). Based on prior studies of oncilla and other tropical felids [32, 37–39], we hypothesized that oncillas would prefer areas with denser vegetation cover, higher elevation, more rugged terrain, and areas further away from roads and settlements. To model the detection parameter, we measured two additional covariates: g) camera trapping effort; and h) survey (Table 1). We hypothesized that a greater trapping effort would result in a higher detection probability. The “Survey” covariate was included to account for unmodeled heterogeneity between survey methods and study sites in the different databases (S1 Dataset).
All covariates were z-transformed prior to model construction to allow direct comparisons and facilitate model convergence. We conducted Spearman correlations to examine collinearity between covariate pairs and considered covariates to be correlated whenever ρ ≥ |0.70|. All spatial covariate data were extracted using QGIS [40]. Elevation and terrain ruggedness were obtained from a Shuttle Radar Topography Mission digital elevation model [41]; tree cover data were derived from the Global Forest Change dataset [42]; distance to settlements came from the OpenStreetMap database and includes cities, towns, villages, and hamlets [43]; distance to paved roads and permanent water bodies were derived from the Costa Rican National Geography Institute [44].
Occupancy modeling
We modeled oncilla occupancy using single-season, single-species occupancy models [28]. These models estimate two parameters: the probability that the species is present at a sampled site (occupancy, ψ), and the probability that the species is detected at a site given that it is present (detection, p). Because oncillas are a highly mobile species and our sampling was done with camera traps, we interpret oncilla occupancy as the probability of habitat use [45]. We first built a detection history matrix, using an occasion length of 15 days, with the package ‘camtrapR’ [46] in the R programming language ver. 4.2.3 [47].
We used up to 120 days from each camera trap station to meet the population closure assumption [48]. Because our surveys were carried out over several years and some of the sites were sampled more than once, we adopted a stacked approach whereby each unique camera-year combination was treated as a sampling site [49].
Occupancy models were implemented in a Bayesian modelling framework using the R package ‘ubms’ [50]. We ran several models, each representing a biologically relevant hypothesis. Initially we ran a null model, representing the hypothesis that oncilla occupancy and detectability do not vary along any of the 8 covariates. We also ran a null model with Year in the occupancy parameter to explore possible changes in occupancy with year. Then we ran two versions of the most parameterized model, including all 6 environmental predictors of the occupancy parameter and both the Effort and Survey measures of the detection parameter. One of these models included a random effect of camera trap station on the occupancy parameter to account for possible pseudoreplication from our stacked approach. The other model did not include this random intercept. Lastly, we ran a model using only environmental variables (testing hypothesis 1) and another model with only anthropogenic variables (testing hypothesis 2). To obtain the most parsimonious model, we ran versions deleting one parameter at a time (the parameter with the smallest effect size) to select the best model. We ran models using four Markov chains of 40,000 iterations each, with half discarded for burn-in. We used default non-informative priors. We ranked models using expected log pointwise predictive density (elpd). We calculated elpd using leave-one-out cross-validation for pairwise model comparisons. We assessed model fit by running the MacKenzie and Bailey goodness of fit test on the global model [51]. Additionally, we assessed model convergence by observing Rhat statistics > 1.05 and by visually examining the trace plots.
Spatial predictions
Using the results of our occupancy modeling, we predicted oncilla occupancy across our study sites and the protected areas within them. We divided the area into 500x500 m grids to match the scale of the modelling and extracted covariate information for each grid cell. We then projected our most parsimonious occupancy model to predict oncilla habitat use probability in each grid cell.
Results
After a total sampling effort of 59,687 trap-nights, we obtained a total of 253 independent detection events of oncilla.
The models with the highest predictive accuracy included the random intercept effect for Station and did not include Year, suggesting static habitat use for the species during the sampling period (Table 2). The best model suggests that oncilla habitat use is most strongly influenced by tree cover (β = 0.69 ± 0.32, 95% CI = 0.06–1.32), elevation (β = 0.68 ± 0.26, 95% CI = 0.17–1.20), and distance to nearest water source (β = -0.42 ± 0.18, 95% CI = -0.76 –-0.07) (Fig 4), and to a lesser extent by terrain ruggedness (β = 0. 23 ± 0.20, 95% CI = -0.18–0.63). Although the coefficient suggests a positive effect on oncilla occupancy, terrain ruggedness had the smallest effect size and its 95% confidence interval included zero. Detectability was higher with increased survey effort (β = 0.34 ± 0.14, 95% CI = 0.06–0.62). Derived parameter estimates were 0.30 ± 0.05 (95% CI: 0.21–0.41) for habitat use and 0.09 ± 0.02 (95% CI: 0.06–0.13) for detection. Our global model performed well under the Mackenzie and Bailey goodness of fit test (Chi square = 1130.91, P = 0.3). Spatial predictions suggest high habitat suitability for oncilla in the Talamanca mountains and to a lesser extent the Tilarán mountain range, with the highest suitability in Tilarán in the core zone of the Children’s Eternal Rainforest, which we did not survey (Fig 5). In the Talamanca range, suitability was high throughout the Caribbean slope and lower in the Pacific slope (Fig 5).
Habitat use probability in response to (a) tree cover percentage; (b) elevation above sea level; and (c) distance to permanent water sources.
Habitat suitability for the Central American clouded oncilla in (a) the Arenal-Monteverde forest complex and (b) the Talamanca mountain range in northern and southern Costa Rica.
Models are ranked by expected predictive accuracy (elpd); P denotes the detection parameter and ψ the habitat-use parameter.
Discussion
We have conducted the first extensive analysis of the Central American clouded oncilla habitat use, thus shedding light on its ecological preferences at the northern portion of its range. Our analysis revealed that environmental features influence oncilla habitat use, supporting hypothesis 1. Our results indicate that oncillas prefer areas of dense tree cover at higher elevations, close to permanent water sources; there was little to no influence of terrain ruggedness, distance to roads, and distance to settlements on habitat use.
Tree cover emerged as the main environmental driver of oncilla occupancy, with the species occupancy increasing in denser areas. This pattern has also been seen for tiger-cat populations in the Brazilian drylands [37, 52]. Within the mountainous landscapes of the Talamanca range, vegetation cover varies from lowland rainforest to cloud forest at middle elevations, and elfin forest and paramo at high elevations. The latter habitat would be unsuitable for oncilla according to our results. However, we did record oncillas in the paramo at areas surrounded by oak forest or nearby oak trees which provide enough cover but were not tall enough to be categorized as a forest in the satellite imagery used in our analysis [42]. A preference for denser vegetation cover has also been documented for Andean clouded tiger-cats and Atlantic Forest tiger-cats [1]. Even the savanna tiger-cat shows a preference for denser vegetation within the savannas and semi-arid dry shrub-woodlands [38, 52, 53].
Small mammals, especially rodents, are reported as the Leopardus “tigrina” species complex’s main prey [5, 54–56], and denser areas are likely to provide both cover and a prey base. Particularly in the Talamanca mountain forests, mice are significantly more abundant in areas of denser vegetation and greater environmental complexity (i.e., oak forest) than in paramo [57]. Thus, the habitat use pattern of oncilla appears to be concentrated in areas of higher rodent abundance. In terms of prey diversity, the Talamancan deermouse (Peromyscus nudipes) (an endemic species included in the oncilla diet [54]) is the most abundant small mammal in this region [57]. Felid populations elsewhere have been shown to increase their habitat use in areas of higher prey density [58, 59]. However, we note the need to establish long-term standardized surveys of rodent abundance and diversity along with camera trap monitoring of small wild cats to evaluate the effect of rodent abundance on the habitat use of oncilla.
We found a linear relationship in which the probability of oncilla habitat use increased with elevation. In the tropical Andes, clouded tiger-cat populations have been shown to exhibit greater occupancy at elevations around 2500 meters above sea level [32]. The mountain ranges in which Central American clouded oncillas occur do not reach the high elevations of the Andes (> 4000 m), with the highest summits in Talamanca averaging 3600–3800 m above sea level, which is within the maximum known limit recorded for clouded tiger-cats (3960 m) [1], and those in the Tilarán all below 1900 m above sea level. The bulk of natural habitat in both mountain ranges is represented by upland cloud forest, which is the oncilla’s most important habitat. Hence, the positive effect of elevation on the species’ habitat use may be related to favorable habitat structure at mid and high elevations. At higher elevations there is also less human disturbance, which is especially true in the Talamanca mountain range. We note that some oncilla detections occurred at elevations of more than 3200 m, not far from Costa Rica’s highest summit (3820 m).
Distance to nearest water source was an important determinant of oncilla habitat use, with occupancy increasing in areas near water sources. This pattern contrasts with that found in savanna tiger-cats [37, 53]. The effect size of distance to water sources was smaller than both tree cover and elevation, suggesting that this predictor is the least important of the three. Due to high rainfall, small creeks and other water sources not visible in satellite imagery are likely to be present in several areas of the oncilla’s range. Additionally, the cloud forest environment is very humid. This suggests that water availability per se is not a limiting factor for the species. The perceived effect of this variable might be an artifact of the sampling design, given that forested remnants in non-protected areas are mainly concentrated around riparian vegetation, which is protected by Costa Rican law. As a result, camera trap stations in non-protected areas were mainly located around these riparian zones. Nevertheless, riparian areas could enhance resource availability for small mammals like rodents, which in turn would provide a prey base for oncillas.
We found no evidence of terrain ruggedness influencing oncilla habitat use, although the coefficient suggests a positive effect. Since oncillas are cloud forest specialists, we anticipated a positive relationship between habitat use and both elevation and topographic ruggedness. Although topographic ruggedness has not been tested on savanna tiger-cat populations, it has been shown to positively affect occupancy of other felids elsewhere [60, 61]. Further sampling in less rugged terrain may show evidence of terrain ruggedness positively influencing oncilla occupancy. Alternatively, it could be possible that the species favors high elevation environments with lower ruggedness. Rugged terrain presents challenges for locomotion and is energetically more expensive compared with movement over trails and natural walkways. Felid species generally prefer moving along trails, and oncillas specifically have been suggested to use trails made by tapirs [34]. Such trails tend to be located in less rugged areas, which may explain the apparent lack of response of oncillas to terrain ruggedness.
Oncilla habitat use did not change in relation to distance to roads, with the species neither avoiding nor selecting areas near roads. This suggests that the species may be at risk of roadkill mortality on roads that cross the forests. For instance, Central America’s main road, the Pan-American highway, passes close to some of our sampling sites and serves as a border between two of Costa Rica´s major highland protected areas: Tapantí Macizo de la Muerte, and Los Quetzales National Parks. Oncillas are known to be killed on this and nearby roads [62]. Roads have been documented to affect the habitat use of felids elsewhere (e.g., [32, 62–65]). Alternatively, in the Caatinga shrubland and thorn forest, savanna tiger-cat occupancy did not vary in response to distance to the nearest road [37]. Roads facilitate access of poachers to remote areas, and even in otherwise pristine forests, roads often lead to widespread habitat loss and poaching. Although deforestation was not a major threat at our study sites, roads still pose a serious problem for oncillas due to the risk of collision mortality. In addition, poachers who are otherwise hunting for meat could use roads to kill oncillas opportunistically. Because the bulk of the oncilla range in the Talamanca mountains is not crossed by roads, roadkill mortality is perhaps not a range-wide threat. Still, as many as three oncillas per year are being killed along the Pan-American highway in Costa Rica [62].
Although we had expected a negative effect of human settlements on oncilla habitat use, this variable was poorly supported in the models. This result is in contrast to studies conducted in the Brazilian Caatinga [37] and Colombian Andes [32]. Some characteristics of our sampling procedures could potentially explain this. All of our cameras were located in fairly pristine areas, and the vast majority were inside protected areas. Furthermore, a large portion of our cameras were in difficult-to-access terrain. Roughly 93% of our camera trap stations were more than 1 km away from the nearest settlement, with some as far away as 20 km. For tiger-cat populations in the Brazilian drylands, 800-m from the nearest household has been documented as a critical distance affecting the species’ habitat use [52]. If this pattern also holds true for oncilla, we expect a marked negative effect on habitat use in areas less than 800-m away from the nearest settlement. Alternatively, oncillas are known to occasionally prey on poultry by attacking chicken coops in local communities [66], thus oncillas in human-dominated landscapes might select habitats closer to settlements if these provide a foraging resource. Retaliatory killing is known to be a threat for some oncilla populations, as happens with other small wild cat species [67]; hence, we would expect a negative effect of distance to settlements on the species’ abundance more than on habitat use per se. For some areas, management schemes and the tolerance of local residents could explain a lack of response to settlements. At a private reserve in the Brazilian Caatinga, for example, tiger-cat habitat use did not change with distance to households, likely because the tiger-cats are not persecuted there [53].
Our spatial predictions suggest that the stronghold for the species in Costa Rica lies in the Talamanca mountains, with high suitability throughout Chirripó, Los Quetzales, and Tapantí National Parks and La Amistad International Park. High suitability was also found in the Arenal-Monteverde forest complex, but this area is smaller (48,273 ha.). Our maps suggest high suitability on the Caribbean slope of the Talamanca mountains, a relatively unexplored area that warrants further field surveys. On the Pacific slope of the mountains, suitability is lower as a result of past deforestation. Suitability was also found to be low in the highest area of Chirripó National Park, coinciding with the open paramo. In this area, oncillas could still be present, finding cover and prey in shrubby vegetation that does not qualify as proper tree cover. In-situ vegetation metrics such as litter depth and basal area could further elucidate oncilla habitat use in the paramo. In the Andes, litter depth was found to strongly influence clouded tiger-cat occupancy [32].
The use of occupancy modeling allowed us to identify areas of suitable habitat for the species, and hence delineate conservation priority areas for the species in the country. By leveraging a large camera trap dataset, we were able to account for unequal sampling effort and imperfect detection in an occupancy modeling framework. Lack of accounting for sampling effort and detectability can lead to biased estimates of habitat suitability using other modeling approaches [68]. A limitation of our approach is that we did not sample the full range of covariates, as cameras were often placed in areas where we expected a priori to find clouded oncillas. Nevertheless, the large number of sampling units and the geographic spread of sample sites allowed us to account for most of the species’ proposed range in the country. Our spatial prediction of habitat suitability matches that of de Oliveira et al. [1], which identified the Talamanca mountain range as the species’ stronghold in Central America.
Conclusion and future directions
We have shown that Central American clouded oncilla habitat use patterns in the northern portion of its range are driven primarily by environmental factors. Future research should address the influence of mesopredators, such as the ocelot, on the habitat use patterns of the oncilla. Ocelots have been documented to negatively affect the populations of small Neotropical felids [69]. However, as the ocelot is mostly a lowland species, its numbers in clouded tiger-cat ranges are typically too small to cause an effect, as has been noted in Colombia [32, 70]. We also do not know if ocelot numbers could be limiting clouded tiger-cats to higher altitudes as it seems to do for savanna tiger-cats in Amazonia [69]. Additionally, assessing and mitigating other threats such as the presence of feral dogs and domestic cats is imperative due to the potential threat of disease transmission [52, 67]. Although most of the current suitable habitat for the unique Central American clouded oncilla falls under some form of protection, threats such as long-term climate change and disease transmission could wreak havoc on oncilla populations in the future. Density estimates are also much needed, as so far none have been published for L. pardinoides. Lastly, field surveys on the Panamanian side of the Talamanca mountains and in Costa Rica’s central mountain range are needed to fully delineate the scope of the conservation needs/requirements, status, and strategic conservation planning for this unique felid population of Central America.
Supporting information
S1 Table. Sampled areas to assess occupancy and habitat use patterns by the Central American clouded oncilla (Leopardus pardinoides oncilla) in the mountains of Costa Rica.
https://doi.org/10.1371/journal.pone.0310562.s001
(DOCX)
S2 Table. Covariate correlation matrix.
There were no correlations, hence all covariates were used in the modeling.
https://doi.org/10.1371/journal.pone.0310562.s002
(DOCX)
S1 Dataset. Site covariates data used for running the analysis of this manuscript.
https://doi.org/10.1371/journal.pone.0310562.s003
(XLSX)
S2 Dataset. Observation covariates data used for running the analysis of this manuscript.
https://doi.org/10.1371/journal.pone.0310562.s004
(XLSX)
S3 Dataset. Detection history matrix used for running the analysis of this manuscript.
https://doi.org/10.1371/journal.pone.0310562.s005
(XLSX)
Acknowledgments
We would like to thank the many private-land owners for allowing us to conduct part of this study in their properties, particularly, to Rossy Estrada, Gisella Fernández, Roger Fernández, Bedrich Sruta, César Vindas, Shankar Syed, Virgita Mora, and the Serrano and Seelye families. We also are grateful with our many collaborators from SINAC–Área de Conservación Central and Área de Conservación La Amistad–Pacífico, and Greilin Fallas, Madelyn Peterson, Maximilian King, and Casey McConnell from Cloudbridge Nature Reserve for logistic support in the field. Paula Ledezma from Tropical Science Center provided data on Monteverde Cloud Forest Preserve climate and size. Benjamin Luke and Nick Hawkins kindly shared pictures on the Central American clouded oncilla for the Oncilla Conservation project. Lastly, we would like to thank the comments of the assigned reviewer Nabaz R. Khwarahm, which improved the quality of the manuscript.
References
- 1. de Oliveira TG, Fox-Rosales LA, Ramírez-Fernández JD, Cepeda-Duque JC, Zug R, Sanchez-Lalinde C, et al. Ecological modeling, biogeography, and phenotypic analyses setting the tiger cats’ hyperdimensional niches reveal a new species. Nature. 2024;14: 2395–2413. pmid:38287072
- 2.
Sunquist ME, Sunquist FC. Family Felidae (cats). In: Wilson DE, Mittermeier RA, editors. Handbook of the mammals of the world vol. 1. Carnivores. Barcelona: Lynx Edicions; 2009. p. 54–168.
- 3. Trigo TC, Freitas TRO, Kunzler G, Cardoso L, Silva JCR, Johnson WE, et al. Inter-species hybridization among Neotropical cats of the genus Leopardus, and evidence for an introgressive hybrid zone between L. geofffroyi and L. tigrinus in Southern Brazil. Mol Ecol. 2008;17: 4317–4333. https://doi.org/10.1111/j.1365-294X.2008.03919.x.
- 4. Trindade FJ, Rodrigues MR, Figueiró HV, Li G, Murphy WJ, Eizirik E. Genome-wide SNPs clarify a complex radiation and support recognition of an additional cat species. Mol Biol Evol. 2021;38: 4987–4991. pmid:34320647
- 5. Payán E, de Oliveira T. Leopardus tigrinus. The IUCN Red List of Threatened Species. 2016; e.T54012637A50653881. https://dx.doi.org./10.2305/IUCN.UK.2016-2.RLT.T54012637A50653881.en
- 6. Sistema Nacional de Áreas de Conservación (SINAC). Resolution n° R-SINAC-CONAC-092-2017, stating Costa Rican mammal species red list. La Gaceta. San José, Costa Rica. 2017.
- 7. Ramírez-Fernández JD, Sánchez R, May-Collado LJ, González-Maya JF, Rodríguez-Herrera B. Revised checklist and conservation status of the mammals of Costa Rica. Therya. 2023;14: 233–244. https://doi.org/10.12933/therya-23-2142.
- 8. Hilbert DW, Bradford M, Parker T, Westcott DA. Golden bowerbird (Prionodura newtonia) habitat in past, present and future climates: predicted extinction of a vertebrate in tropical highlands due to global warming. Biol Conserv. 2004;116: 367–377. https://doi.org/10.1016/S0006-3207(03)00230-1.
- 9. Laurance WF, Useche DC, Shoo LP, Herzog SK, Kessler M, Escobar F, et al. Global warming, elevational ranges and the vulnerability of tropical biota. Biol Conserv. 2011;144: 548–557. https://doi.org/10.1016/j.biocon.2010.10.010.
- 10. González-Zeas D, Erazo B, Lloret P, De Bièvre B, Steinschneider S, Dangles O. Linking global climate change to local water availability: limitations and prospects for a tropical mountain watershed. Sci Total Environ. 2019;650: 2577–2586. pmid:30293009
- 11. Pounds JA, Fogden MPL, Campbell JH. Biological response to climate change on a tropical mountain. Nature. 1999;398: 611–615. https://doi.org/10.1038/19297.
- 12. Pounds JA, Bustamante MR, Coloma LA, Consuegra JA, Fogden MPL, Foster PN, et al. Widespread amphibian extinctions from epidemic disease driven by global warming. Nature. 2006;439: 161–167. pmid:16407945
- 13. Karmalkar AV, Bradley RS, Diaz HF. Climate change scenario for Costa Rican montane forests. Geophys Res Lett. 2008;35: 1–5. https://doi.org/10.1029/2008GL033940.
- 14. Dirnböck T, Essl F, Rabitsch W. Disproportional risk for habitat loss of high-altitude endemic species under climate change. Glob Chang Biol. 2010;17: 990–996. https://doi.org/10.1111/j.1365-2486.2010.02266.x.
- 15. Ripple WJ, Estes JA, Beschta RL, Wilmers CC, Ritchie EG, Hebblewhite M, et al. Status and ecological effects of the world’s largest carnivores. Science. 2014;343: 1241484. pmid:24408439
- 16.
Bruijnzeel LA, Scatena FN, Hamilton LS. Tropical montane cloud forests: science for conservation and management. Cambridge: Cambridge University Press; 2010.
- 17. Berry ZC, White JC, Smith WK. Foliar uptake, carbon fluxes and water status are affected by the timing of daily fog in saplings from a threatened cloud forest. Tree Physiol. 2014;34: 459–470. pmid:24835239
- 18. Oliveira RS, Eller CB, Bittencourt PRL, Mulligan M. The hydroclimatic and ecophysiological basis of cloud forest distributions under current and projected climates. Ann Bot. 2014;113: 909–920. pmid:24759267
- 19. Foster D, Swanson F, Aber J, Burke I, Brokaw N, Tilman D, et al. The importance of land-use legacies to ecology and conservation. Bioscience. 2003;53: 77–88. https://doi.org/10.1641/0006-3568(2003)053[0077:TIOLUL]2.0.CO;2.
- 20. Ponce-Reyes R, Nicholson E, Baxter PWJ, Fuller RA, Possingham H. Extinction risk in cloud forest fragments under climate change and habitat loss. Divers Distrib. 2013;19: 518–529. https://doi.org/10.1111/ddi.12064.
- 21. Cayuela L, Golicher DJ, Rey-Benayas JM. The extent, distribution, and fragmentation of vanishing montane cloud forest in the highlands of Chiapas, Mexico. Biotropica. 2006;38: 544–554. https://doi.org/10.1111/j.1744-7429.2006.00160.x.
- 22. Hu J, Riveros-Iregui DA. Life in the clouds: are tropical montane cloud forests responding to changes in climate? Oecologia. 2016;180: 1061–1073. pmid:26739003
- 23. Jiménez-García D, Peterson AT. Climate change impact on endangered cloud forest tree species in Mexico. Rev Mex Biodiv. 2019;90: e902781. https://doi.org/10.22201/ib.20078706e.2019.90.2781.
- 24. Bruijnzeel LA. Hydrology of tropical montane cloud forests: a reassessment. Land Use Water Resour Res. 2001;1: 1–18. https://doi.org/10.22004/ag.econ.47849.
- 25. Hostettler S. Tropical montane cloud forests: a challenge for conservation. Bois et forets des Trop. 2002;274: 19–31.
- 26. Hama AA, Khwarahm NR. Predictive mapping of two endemic oak tree species under climate change scenarios in a semiarid region: range overlap and implications for conservation. Ecol Inform. 2023;73: 101930. https://doi.org/10.1016/j.ecoinf.2022.101930.
- 27. Guillera-Arroita G, Lahoz-Monfort JJ, MacKenzie DI, Wintle BA, McCarthy MA. Ignoring imperfect detection in biological surveys is dangerous: a response to ‘fitting and interpreting occupancy models’. PLoS ONE. 2014;9: e99571. https://doi.org/10.1371/journal.pone.0099571.
- 28. Mackenzie DI, Nichols JD, Lachman GB, Droege S, Royle JA, Langtimm CA. Estimating site occupancy rates when detection probabilities are less than one. Ecology. 2002;83: 2248–2255. https://doi.org/10.1890/0012-9658(2002)083[2248:ESORWD]2.0.CO;2.
- 29. Cuyckens GAE, Perovic PG. Distribución y conservación del tirica (Leopardus tigrinus) en Las Yungas, Argentina. Agraria. 2013;7: 17–21.
- 30. Bonilla-Sánchez A, Gómez-Ruíz DA, Botero-Cañola S, Rendón-Jaramillo U, Ledesma-Castañeda E, Solari S. Riqueza y monitoreo de mamíferos en áreas protegidas privadas en Antioquia, Colombia. Mastozool Neotrop. 2020;27: 266–281. https://doi.org/10.31687/saremMN.20.27.2.0.11.
- 31. González-Maya JF, Zárrate-Charry DA, Arias-Alzate A, Lemus-Mejía L, Hurtado-Moreno AP, Vargas-Gómez MG, et al. Spotting what’s important: priority areas, connectivity, and conservation of the northern tiger cat (Leopardus tigrinus) in Colombia. PLoS ONE. 2022;17: e0273750. pmid:36099258
- 32. Cepeda-Duque JC, Andrade-Ponce G, Montes-Rojas A, Rendón-Jaramillo U, López-Velasco V, Arango-Correa E, et al. Assessing microhabitat, landscape features and intraguild relationships in the occupancy of the enigmatic and threatened Andean tiger cat (Leopardus tigrinus pardinoides) in the cloud forests of northwestern Colombia. PLoS ONE. 2023;18: e0288247. pmid:37428730
- 33. Rogan JE. First locality record of melanistic oncilla (Leopardus tigrinus) in Monteverde, Costa Rica. Neotrop Biol Conserv. 2021;16: 427–434. https://doi.org/10.3897/neotropical.16.e65464.
- 34. Brenes-Mora E. Patrones de actividad, selección de hábitat y atropellos de danta (Tapirus bairdii) en bosque atravesado por una carretera en la Cordillera de Talamanca, Costa Rica. M. Sc. Thesis, Universidad Nacional de Costa Rica. 2018. Available from: https://repositorio.una.ac.cr/bitstream/handle/11056/14856/TFG_Esteban%20Brenes-Mora.pdf?sequence=1&isAllowed=y
- 35. Smith G. Estimating the diversity of bird and mammal communities using camera traps in a cloud forest environment, Costa Rica. M. Sc. Thesis. University of Plymouth. 2021. Available from: https://www.cloudbridge.org/wp-content/uploads/2021/09/Estimating-the-diversity-of-bird-and-mammal-communities-using-camera-traps-in-a-cloud-forest-environment-Costa-Rica.pdf
- 36. Mooring MS, Eppert AA, Botts RT. Natural selection of melanism in Costa Rican jaguar and oncilla: a test of Gloger’s rule and the temporal segregation hypothesis. Trop Conserv Sci. 2020;13: 1–15. https://doi.org/10.1177/1940082920910364.
- 37. Marinho PH, Bezerra D, Antongiovanni M, Fonseca CR, Venticinque EM. Estimating occupancy of the vulnerable northern tiger cat Leopardus tigrinus in Caatinga drylands. Mammal Res. 2017;63: 33–42. https://doi.org/10.1007/s13364-017-0330-4.
- 38. Fox-Rosales LA, de Oliveira TG. Habitat use patterns and conservation of small carnivores in a human-dominated landscape of the semiarid Caatinga in Brazil. Mamm Biol. 2022;102: 465–475. https://doi.org/10.1007/s42991-022-00245-3.
- 39. Wang B, Rocha DG, Abrahams MI, Antunes AP, Costa HCM, Gonçalves ALS, et al. Habitat use of the ocelot (Leopardus pardalis) in Brazilian Amazon. Ecol Evol. 2019;9: 5049–5062. pmid:31110661
- 40. QGIS Development Team. QGIS: A Free and Open Source Geographic Information System. 2019. Available from: https://www.qgis.org/en/site/
- 41. USGS. USGS EROS Archive—Digital Elevation—Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global. 2018. Available from: https://www.usgs.gov/centers/eros/science/usgs-eros-archive-digital-elevation-shuttle-radar-topography-mission-srtm-1-arc?qt-science_center_objects=0#qt-science_center_objects
- 42. Hansen MC, Potapov P V., Moore R, Hancher M, Turubanova SA, Tyukavina A, et al. High-resolution global maps of 21st-century forest cover change. Science. 2013;342: 850–853. pmid:24233722
- 43. OpenStreetMap. Planet OSM. 2021. Available from: https://planet.openstreetmap.org/
- 44. Instituto Geográfico Nacional. IGN Cartografía 1:5mil. Sistema Nacional de Información Territorial. 2017. Available from: https://www.snitcr.go.cr/ico_servicios_ogc_info?k=bm9kbzo6MjY=&nombre=IGN
- 45. Mackenzie DI, Royle JA. Designing occupancy studies: general advice and allocating survey effort. J Appl Ecol. 2005;42: 1105–1114. https://doi.org/10.1111/j.1365-2664.2005.01098.x.
- 46. Niedballa J, Sollmann R, Courtiol A, Wilting A. camtrapR: an R package for efficient camera trap data management. Methods Ecol Evol. 2016;7: 1457–1462. https://doi.org/10.1111/2041-210X.12600.
- 47.
R Development Core Team R. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. 2011. https://doi.org/10.1007/978-3-540-74686-7.
- 48. MacKenzie DI, Nichols JD, Hines JE, Knutson MG, Franklin AB. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology. 2003;84: 2200–2207. https://doi.org/10.1890/02-3090.
- 49.
Kéry M, Royle JA. Modeling population dynamics with count data. In: Kéry M, Royle JA, editors. Applied hierarchical modeling in ecology: analysis of distribution, abundance and species richness in R and BUGS. Vol. 2: Dynamics and advanced models. Cambridge: Academic Press; 2021. p. 65–156. https://doi.org/10.1016/B978-0-12-809585-0.00002-8.
- 50. Kellner KF, Fowler NL, Petroelje TR, Kautz TM, Beyer DE, Belant JL. ubms: an R package for fitting hierarchical occupancy and N‐mixture abundance models in a Bayesian framework. Methods Ecol Evol. 2022;13: 577–584. https://doi.org/10.1111/2041-210X.13777.
- 51. MacKenzie DI, Bailey LL. Assessing the fit of site-occupancy models. J Agric Biol Environ Stat. 2004;9: 300–318. https://doi.org/10.1198/108571104X3361.
- 52. de Oliveira TG, Lima BC, Fox-Rosales L, Pereira RS, Pontes-Araújo E, de Sousa AL. A refined population and conservation assessment of the elusive and endangered northern tiger cat (Leopardus tigrinus) in its key worldwide conservation area in Brazil. Glob Ecol Conserv. 2020;22. https://doi.org/10.1016/j.gecco.2020.e00927.
- 53. Fox-Rosales LA, de Oliveira TG. Interspecific patterns of small cats in an intraguild-killer free area of the threatened Caatinga drylands, Brazil. PLoS ONE. 2023;18:e0284850. pmid:37083683
- 54. Gardner AL. Notes on the little spotted cat, Felis tigrina oncilla Thomas, in Costa Rica. J Mammal. 1971;52: 464–465.
- 55. Trigo TC, Tirelli FP, Machado LF, Peters FB, Indrusiak CB, Mazim FD, et al. Geographic distribution and food habits of Leopardus tigrinus and L. geoffroyi (Carnivora, Felidae) at their geographic contact zone in southern Brazil. Stud Neotrop Fauna Environ. 2013;48: 56–67. https://dx.doi.org/10.1080/01650521.2013.774789.
- 56. Marinho PH, Bezerra D, Antongiovanni M, Fonseca CR, Venticinque EM. Activity patterns of the threatened northern tiger cat Leopardus tigrinus and its potential prey in a Brazilian dry tropical forest. Mamm Biol. 2018;89: 30–36. https://doi.org/10.1016/j.mambio.2017.12.004.
- 57. Ramírez-Fernández JD, Barrantes G, Sánchez-Quirós C, Rodríguez-Herrera B. Habitat use, richness, and abundance of native mice in the highlands of the Talamanca mountain range, Costa Rica. Therya. 2023;14: 49–54. https://doi.org/10.12933/therya-23-2227.
- 58. Tortato MA, Oliveira-santos LGR, Moura MO, de Oliveira TG. Small prey for small cats: the importance of prey- size in the diet of southern tiger cat Leopardus guttulus in a competitor-free environment. Stud Neotrop Fauna Environ. 2023;58: 75–86. https://doi.org/10.1080/01650521.2021.1902202.
- 59. Santos F, Carbone C, Wearn OR, Rowcliffe JM, Espinosa S, Moreira MG, et al. Prey availability and temporal partitioning modulate felid coexistence in Neotropical forests. PLoS ONE. 2019;14: 1–23. pmid:30861045
- 60. Lamichhane BR, Lamichhane S, Regmi R, Dhungana M, Thapa SK, Prasai A, et al. Leopard (Panthera pardus) occupancy in the Chure range of Nepal. Ecol Evol. 2021;11: 13641–13660. pmid:34707806
- 61. Dyck MA, Iosif R, Promberger–Fürpass B, Popescu VD. Dracula’s ménagerie: A multispecies occupancy analysis of lynx, wildcat, and wolf in the Romanian Carpathians. Ecol Evol. 2022;12: 1–13. https://doi.org/10.1002/ece3.8921.
- 62. Arellano A. Counting dead cats on Costa Rica’s highways: Q&A with Daniela Araya Gamboa. Mongabay. 2023 Apr 5 [Cited 2024 February 26]. Available from: https://news.mongabay.com/2023/04/counting-dead-cats-on-costa-ricas-highways-qa-with-daniela-araya-gamboa/.
- 63. Klar N, Herrmann M, Kramer-Schadt S. Effects and mitigation of road impacts on individual movement behavior of wildcats. J Wildl Manage. 2009;73: 631–638. https://doi.org/10.2193/2007-574.
- 64. Blackburn A, Heffelfinger LJ, Veals AM, Tewes ME, Young JH Jr. Cats, cars, and crossings: the consequences of road networks for the conservation of an endangered felid. Glob Ecol Conserv. 2021;27. https://doi.org/10.1016/j.gecco.2021.e01582.
- 65. Grilo C, Borda-de-Água L, Beja P, Goolsby E, Soanes K, le Roux A, et al. Conservation threats from roadkill in the global road network. Glob Ecol Conserv. 2021;30: 2200–2210. https://doi.org/10.1111/geb.13375.
- 66. Ramírez-Fernández JD, Blanco Murillo P, Araya-Chavarría F, Delgado-Carazo JC, Gallardo V, Pastor-Parajeles S. Integrated actions for the conservation of the Central American oncilla. CATnews. 2021;74: 40–42.
- 67. Mugerwa B, Adhya T, Ratnayaka A, Thudugala A, Napolitano C, Sanderson J, et al. Are we doing enough to protect the World’s small wild cats? CATnews. 2020;71: 42–47.
- 68. MacKenzie DI. Modeling the probability of resource use: the effect of, and dealing with, detecting a species imperfectly. J Wildl Manag. 2010;70: 367–374. https://doi.org/10.2193/0022-541X(2006)70[367:MTPORU]2.0.CO;2.
- 69. de Oliveira TG, Fox-Rosales LA, Paemelaere EAD, Ferraz KMPMB. The dominant mesopredator and savanna formations shape the distribution of the rare northern tiger cat (Leopardus tigrinus) in the Amazon. Nature. 2022;12: 18697. pmid:36333380
- 70.
Valderrama-Vásquez CA. Densidad de ocelote y abundancias relativas de ocelote y margay, usando datos de cámaras trampa en la cordillera oriental de los Andes colombianos. In: Payán E, Castano-Uribe C, editors. Grandes felinos de Colombia. Panthera, Conservación Internacional, Fundación Herencia Ambiental Caribe; 2013. p. 132–144.
