Using linear regression to measure bird abundance

This study investigated methods for identifying daily incidence rates for bird species. It focused on relationships between incidence rates, site and season. We used sightings of 23 common resident species routinely reported every month from January 2004 to December 2007 at seven wetland locations in the Thale Noi non-hunting area of southern Thailand. Our findings revealed that the log-linear model gives a quite satisfactory fit, so it appears a suitable type of model for bird abundance. On taking logarithms of the incidence rates though, the zero counts must be replaced by an appropriate constant. Our model suggests that Cattle Egret (Bubulcus ibis) was found at the Thale Noi non-hunting area with the highest incidence rate. In contrast, we found a low mean of model outputs for Lesser Whistling-Duck (Dendrocygna javanica) relative to the mean in the data, and this species was not observed on at least 25 % or 3 days per year. These data had a low number of zeros and a large number of various species. Therefore, we recognize a remark on “what is being counted” that it is important to reasonably explain the species abundance in terms of statistical and ecological approaches.