Avian community response to a golf-course landscape unit gradient

Abstract

Although golf-course construction significantly alters natural wildlife habitat, the resulting remnant, disturbed, and introduced landscape patches may provide valuable habitat for birds. Considering the current rate of new course construction, the effects of land consumption and habitat alteration on breeding bird communities in coastal South Carolina are of immediate concern. During summer 2000 and 2001, we sampled 24 golf-course landscape units (GCLU) to assess their value to the breeding bird community. We defined a GCLU as the legally owned parcel of land where a golf course was sited, including the course and all associated development (e.g., residential housing). Sample units (n = 24) were selected to represent a gradient of GCLUs ranging from low to high landscape alteration and were subjectively classified a priori into 1 of 3 alteration groups (G1 [low], G2 [medium] or G3 [high]). We conducted Analysis of Variance procedures to determine whether estimations of species richness, Neotropical migrant richness, and degree of conservation concern differed across the gradient. We explored relative strengths of associations between landscape structure (landscape composition and spatial configuration) and avian community parameters at 2 spatial scales using stepwise multiple regression techniques. We used simple linear regression to assess the relationship between percent forested area of the GCLU and avian community parameters. Total number of species and number of Neotropical migrant species, as well as degree of conservation concern of the species present, were higher in less-altered GCLUs (F2,21 <14, P < 0.05), and were significantly influenced by percent forested area (adj. R2=0.39–0.57). The majority of birds associated with less-developed landscapes were woodland and scrub-shrub breeding species, while urban-breeding species were found primarily in the more-altered landscapes (χ2=440.3, df=6, n=4757, P < 0.001). The area of forest and disturbance patches, size variability in managed turfgrass patches, and measures of spatial complexity proved most useful in explaining variability of response variables due to landscape structure (adj. R2=0.57–0.90).