Considerations for using occupancy surveys to monitor forest primates: A case study with Sclater's monkey (Cercopithecus sclateri)

Count-based indices and distance sampling are widely used to monitor primate populations. Indices are often confounded by variation in detectability, whereas distance sampling is generally ineffective with species that flee or hide from observers and where it is difficult to accurately measure detection distances. We tested occupancy modeling as a means to monitor Sclater's monkey (Cercopithecus sclateri), an endemic of Nigeria. We evaluated effects of survey methodology, habitat, and human disturbance on detection probability and site occupancy. Average detectability was high (p = 0.81), but varied substantially between two observers. Occupancy was highest in areas with intermediate levels (20-40%) of farmland and secondary forest, and was unaffected by human disturbance. Sampling plots (4 and 6.25 ha) did not concurrently contain >1 monkey group, were likely closed to monkey movements during the replicate surveys of each plot, and were spatially separated so that it was unlikely the same group was observed in >1 plot. These conditions enabled the conversion of occupancy to group density. Scaled to 6.25 ha, model-weighted occupancy averaged 0.230 (SE 0.103), yielding an estimate of 3.7 groups/km ² (95% CI 1.4-7.7 groups/km ²). Because some groups straddled plot boundaries, we assumed that half of these groups were inside the plots, resulting in an adjusted estimate of 3.1 groups/km ². Our results illustrate that occupancy can be suitable for monitoring vigilant forest primates where detection distances are difficult to measure. However, special attention is required to choose spatial and temporal scales that accommodate the method's closure and independent-detection assumptions.