Abstract:
I conducted a landscape level investigation of the associations of topography, land cover, and landscape metrics with greater prairie-chicken (Tympanuchus cupido pinnatus) lek locations in Kansas. I used Kansas Department of Wildlife and Parks greater prairie-chicken lek data with 1:24000 Digital Elevation Model (DEM) grids to examine the associations of topography with lek versus random locations and traditional versus temporary lek locations. I used discriminant function analysis to develop an overall model of all lek locations combined and for each of 5 physiographic provinces. For lek versus random locations, elevation was a common predictor in all models (P < 0.001). For traditional versus temporary leks locations, topography was a poorer predictor (P < 0.05) than for lek versus random locations.
Using land cover maps created from satellite imagery, I used discriminant function analysis to model lek versus random locations and traditional versus temporary lek locations. I developed models for all lek locations combined and for each of 4 physiographic provinces within 1.6 km and 4.0 km buffers. For lek versus random locations in the 1.6 km buffer, grassland area was a common predictor (P < 0.001) to all models. For traditional versus temporary lek locations, forest area was the predictor for 3
models (P < 0.01). The 1.6 km models were better predictors than the 4.0 km models. Land cover area alone was a better predictor than landscape metrics.
I also developed combined models for the 1.6 krn buffer land cover, landscape metrics, topography, and 4.0 km buffer land cover and topography. These models showed only slight improvement in accuracy over the general land cover models. I found that lek locations in Kansas were associated with larger blocks of grassland than has been reported previously.
