Estimation of parrot and hornbill densities using a point count distance sampling method

Abstract

The suitability of point count distance methods for estimating densities of tropical parrots and hornbills was assessed during surveys in Indonesia. The methods will perform well, so long as the following are considered. (1) Enough bird records must be accumulated to model species' detection curves precisely. For some species, around 2000 point counts may be needed and, in very rare species, the method may not be appropriate. Pooling data across habitats, species or years may increase precision in cases of small sample size. (2) Point counts are likely to be less biased than line transects because bird detection rates close to the recorder may be higher and there may be less chance of double-counting birds. Distances greater than 400 m between census points are unnecessary. (3) Count periods lasting ten minutes may be needed to ensure that most birds close to the recorder are detected. Controlled flushing of concealed birds after the main count period may also be appropriate. (4) The best time of day for census is the period when bird detectability is high but bird mobility low. For many large avian frugivores, this corresponds to the period between one hour after dawn and 10.30 h. (5) Records of flying birds must be excluded from density calculations. In the species studied, between 2% and 20% should be added to density estimates to compensate for the exclusion of flying birds.