Using Poisson and Negative Binom Regression Models on Birds Population İn Dönemeç Delta

Abstract

The aim of this study to make a statistical estimate to bird populations in Dönemeç Delta using Poisson and negative binom regression models. According to Poisson regression model a deviance statistic greater than one (1) indicates that there is an over-dispersion in the bird’s population. The over-dispersion value in the Poisson regression was much greater than one (156.615). In contrast, the over-dispersion value in negative binom regression was close to one (1.277). Therefore, parameter estimations were interpreted according to negative binom regression. The effects of seasons, habitats and order (ordo) were found to be statistically significant on population density (p<0.05). Summer season when taken as a reference in all seasons only winter season was statistically significant (p<0.05). The population in sandy, farmland and stream edge habitats seen different according to other habitats (meadow, reeds and marshy) (p<0.05). When the Anseriformes ordo is taken as reference only two bird order were insignificant (p>0.05) according to 13 orders. © 2018, Centenary University. All rights reserved.