| This dissertation is a treatise on spatially-explicit resource selection on two very different vertebrate species. The North Atlantic right whale ( Eubalaena glacialis) is the most endangered large whales in the world. Ship strikes and fishing gear entanglement are impediments to recovery. The gopher tortoise (Gopherus polyphemus) is an imperiled species whose upland habitats are shrinking from urban and agricultural development. Determining spatial distribution of important resources is important for conservation strategies. Historical and modern thinking of habitat selection theory and analytical techniques are reviewed and applied to these species. Fine-scale resource selection of sea surface temperature (SST), derived from AVHRR imagery, is evaluated for right whales in the southeastern U.S. calving grounds. Aerial survey data (December-March, 1991-1998) including survey tracklines and right whale locations were entered into a Geographic Information Systems (GIS) for comparing whale use of SST to availability based on search effort. Using Monte Carlo techniques, mean and standard deviation for SSTs and latitudes of whale-sightings were compared to sampling distributions derived from available SSTs and latitudes. From these data, it was concluded that the North Atlantic right whale uses SSTs and latitudes non-randomly. Broad-scale habitat selection for gopher tortoises was evaluated from the 2003 Land Cover/Land Use map (Florida Fish and Wildlife Conservation Commission). Based on land cover and ancillary data, potential gopher tortoise habitat was developed for northeast Florida. Data collected at random field locations, including actual land cover type, vegetation structure, and gopher tortoise presence/absence were used to develop a spatially-explicit probabilistic (logistic regression) model, and associated uncertainty estimates, of gopher tortoise occupancy. Logistic regression was also used to develop a local model of habitat suitability based on vegetation structure. Exploratory analysis, based on AIC criterion, implied gopher tortoise presence was inhibited by understory vegetation and leaf litter, but probability of presence increased with additional ground cover. The potential for combining vegetation structure data with land cover data for refining the probabilistic model was explored. Following future model refinement and validation, a simple protocol is proposed for evaluating remaining gopher tortoise habitat for the state of Florida. |
