SPATIAL VARIATION OF INFLUENTS ON AVIAN DIVERSITY: A MULTIVARIATE APPROACH |
| |
| Authors: | Kathleen C Parker |
| |
| Institution: | University of Georgia , Athens, Georgia 30602 |
| |
| Abstract: | Breeding bird censuses from 132 homogeneous sites representing a variety of habitats throughout North America are analyzed and a bird species diversity model based on environmental characteristics and sampling procedures is presented. Bivariate regression models relate the number of bird species detected in a site (NSPECIES) to plot size, latitude, and the number of bird pairs encountered during censusing. These models consistently overestimate NSPECIES in desert shrublands, grasslands, tundra, and scrub and underestimate NSPECIES in forested habitats, which indicates that the effects of the environmental and methodological variables on NSPECIES vary with respect to habitat. The expansion method of regression analysis is used to generate a multivariate model that accounts for this spatial variation in the influence of the independent variables on NSPECIES. With the exception of temperate wetlands, which have the third highest mean value for NSPECIES of nine habitat groups, forested sites and woodlands have higher mean NSPECIES values than more open habitats. Use of the expansion method to account for spatial variation in the effects exerted by independent variables introduces a geographically realistic element often lacking in broad-scale models. (Key words: avian diversity, sampling intensity, multiple regression analysis, vegetation structure, habitat types.) |
| |
| Keywords: | radar precipitation estimates southeast United States water resources |
|
|
