Habitat Selection Attributes of an Avian Community: A Discriminant Analysis Investigation

Abstract

Habitat use characteristics of all bird species occurring in the lower Colorado River valley were studied. Seventy—two transects, each through homogeneous stands of vegetation and together representing all riparian habitat types occurring in the area, were censused three times monthly for 2 yr (1977, 1978) to provide distributional records for all species. From the census data we divided the transects into three groups for each bird species each season: transects where the species was recorded in both 1977 and 1978, transects where the species was absent both years, and transects where the species occurred in only one of the years. We used discriminant analyses to classify the habitat on each transect as suitable or unsuitable for each species, using a set of eight species composition measures, four foliage density measures, and foliage height diversity (FHD). Mean habitat discriminability of all species varied from a low of 86% in summer to a high of 91% in late summer. Most species showed extensive variability among seasons in both the degree of habitat selectivity and the criteria used in habitat selection. This variability reflected both changes in population structure of the species and responses to changing resource and climatic pressures. Ecological attributes of the species were examined to determine which factors shaped the habitat use patterns of each. Feeding guild played only a small role in both the degree and criteria of habitat selectivity, except for highly specialized feeders such as nectivores and frugivores. Species with restricted distributions were usually habitat specialists, but species with widespread distributions included both habitat generalists and specialists. During spring and summer nonresident, short—stay species showed greater habitat selectivity than did extended—stay species, but during nonbreeding seasons both short—stay and extended—stay species showed equivalent habitat selectivity. That both the mix of habitat generalists and specialists and the habitat selectivity of individual species varied seasonally implies that the organization of habitat use of the community is the dynamic product of many interacting factors. The importance of individual habitat attributes was also examined. Species showed preferences for particular habitat attributes significantly more frequently than they avoided habitat attributes. FHD was the most frequently important variable in differentiating areas used from areas not used. Proportions of honey mesquite, cottonwood, and willow in the vegetation were also frequently important variables in habitat selection by avian species, but usually in conjunction with FHD. Foliage density characteristics were less frequently significant, but for species where these measures were important, they tended to be the primary habitat selection criterion. In summer and late summer, habitat selection criteria of individual species seemed independent of the selection criteria of the other species in the community. In other seasons there was evidence of species interactions in these habitat selection patterns. These empirical findings for an entire avian community over all seasons are an advance in our understanding of how habitat use patterns of communities are organized. The extensive seasonal variability in habitat use by avian species and variability of distributions of individual species between years imply that studies of community organization can best be done with large—scale, long—term studies.