Effect of varying the number of surveys on the accuracy and precision of population estimates.
True population size was 200 individuals, indicated by the horizontal line. Data were obtained using six 1 km transects and ‘incomplete visibility’, though the results for different numbers and lengths of transect did not qualitatively differ from these results in terms of medians and general characteristics (data not shown). The median estimates found by the matrix method overestimated the actual number of orangutans by about 5–10%, possibly because temporal changes in decay rate based on differences in decay rate in different stages and in different seasons overshadowed nest-specific decay rate differences. This would mean that, contrary to the expectation that long-lasting nests would be oversampled (as short-lived nests would more frequently be missed), there was not actually a tendency to find nests that would be longer-lasting in the future and thus no correction factor was necessary. The results for the matrix method for two surveys vary widely because so few (or no) nests completely decayed during the study. Note that scales on the y-axis vary among the panels.