Testing the efficacy of camera-trap sampling designs for monitoring giant pandas in a heterogeneous landscape.

Abstract

The use of camera traps is prevalent in the ecological study of giant pandas (Ailuropoda melanoleuca). The reliability of camera-trap surveying results greatly depends on sampling designs that significantly influence the detection probability of the target species. Few studies have tested the efficacy of sampling designs on camera-trap surveys for monitoring giant pandas in a heterogeneous landscape. In this study, we conducted camera trapping of giant pandas based on two different sampling schemes in Changqing National Nature Reserve of China, and evaluated their outcomes based on three aspects: occupancy analysis, photographic rate, and activity pattern. The results demonstrated that both climate heterogeneity and distance to the nearest road had a strong positive influence on site occupancy, and slope and forest cover had a significant negative impact on site occupancy. Significant differences in the direction or magnitude of variables' influences indicated that there were apparently spatial-temporal dynamics of giant panda distribution between two sampling schemes. The low detection probabilities indicated that both sampling schemes were not robust to accurately monitor giant pandas in the whole study area. We recommended that more suitable sampling designs with local covariates be developed for camera-trap surveys monitoring giant pandas to account for temporal variability and small-scale variation in heterogeneous landscapes.