How well do bird atlas reporting rates reflect bird densities? Correlates of detection from the Fynbos biome, South Africa, with applications for population estimation
Relationships between true population densities and reporting indices from atlas data are important for the calculation of population sizes, though these relationships are remarkably little-known and likely confounded by issues of detection. We examine issues of detection for a single-observer point-count survey across the Fynbos biome in South Africa. We created an index of relative abundance comparable to atlas reporting rate and calculated detection coefficients for each species. We explore various models that explain relative abundance as a function of detection covariates, e.g. mass, colour, group size, vocal behaviour and density. Density was consistently included across models and the best predictor when used alone in a validation exercise. We then calculated mean reporting rates for the citizen science South African Bird Atlas Project2 (SABAP2). A model of species’ reporting rates as determined by density for this set of birds suggested the correlation between atlas reporting rates and density estimates was weak, with species capable of using modified habitats exhibiting higher reporting rates than expected from the density estimates. There was a positive correlation between bird density and reporting rates corrected for by species mass. We used this relationship to calculate densities for the Fynbos bird species and compare results to published data.