Accurate estimates of density and population size are fundamental for appropriate monitoring of populations. Forest dwelling large mammals that live in herds are particularly difficult for estimating population density and size because of the heterogeneity of their spatial distribution at local scales and the difficulty of counting groups. Previous work has dealt with heterogeneity of spatial-temporal distribution by the independent use of gas movement theory and spatially explicit distribution models. Here we propose to combine both approaches, and a novel probabilistic method to minimize inflation of group size estimates, to estimate density and population size of a Neotropical social mammal, the white-lipped peccary (Tayassu pecari). From February to August 2013, we placed eight camera-traps at distinct locations at least 2 km apart from one another, at Cardoso Island State Park, a 110-km2 protected area in the Atlantic forest of Brazil. We estimated groups/area using a gas movement function, and individuals/group from a pathway probability function, and used both to derive a final density estimate (individuals/area). We then generated a habitat suitability model based on seven environmental layers and 32 location points, and extracted the area with high probability of occurrence to estimate population size in the park.
Results/Conclusions
In 1037 camera trap*days, we recorded 309 10-second videos of white-lipped peccary herds. Group density estimates were sensitive to the time interval used to separate independent events (10 seconds: 7 groups/km2; 1 hour: 1.35 groups/km2; 1 day: 1.15 groups/km2); and so were group size estimates (1.15 individuals/10 seconds; 5.92 individuals/hour: 6.96 individuals/day). Nonetheless, one estimate compensated the other resulting in a similar final density estimate across time intervals (8 individuals/km2). Population density applied to the entire area of the park (110km2) reveals a population of 880 individuals. Habitat suitability analyses revealed that four environmental variables (distance from continental edge, altitude, vegetation, and slope) collectively explained 84% of the variation in white-lipped occurrence. Using a model threshold of 0.27 (maximum test sensitivity and specificity), we determined that only 21% of the park’s area was in fact suitable for the species, with a final population estimate of 185 individuals. The present study reveals that a combination of analytical tools allows for more robust estimates of population density and size of forest-dwelling social mammals. Our approach also permits disentangling distinct factors affecting population size (group size, group density, and suitable area), and facilitates pinpointing sources of variation in population size over time.