Predicting Future Range Expansion of Whooping Crane (Grus americana) Winter Habitat Using Long-Term Census and Remotely Sensed Data
The whooping crane (Grus americana) is one of the most threatened crane species in the world and has been identified as an endangered species since 1967. The last wild-population of the whooping crane, the Aransas-Wood Buffalo Population (AWBP), breeds in Wood Buffalo National Park and surrounding areas in the Northwest Territories of Canada and migrates 2,500 km through the Central Flyway to their wintering grounds within the Aransas National Wildlife Refuge and surrounding areas on the Texas Coast. The Endangered Species Act (ESA) recovery plan for this species outlines objectives to down-list the species from endangered to threatened status, including an objective for a self-sustaining wild population size of at least 1,000 individuals, including 250 breeding pairs. However, the feasibility of meeting this objective requires an assessment of space-use by wintering whooping cranes and of the amount of available habitat within the winter range to support the recovery goal population size. Historic location data for 42 color-banded whooping crane individuals were used to analyze space-use strategies by three identified classes of cranes during the winter season; i.e., subadult (immature), associated (non-mating pair), and paired/family (mating pair). Space-use was analyzed by estimating winter whooping crane home range and core area extents from kernel density estimators. The resulting home range estimates supported previous descriptions based on observation data of the distribution of the three identified classes on the wintering grounds. The resulting core area estimates yielded a similar spatial distribution as winter territories identified by U.S. Fish and Wildlife Service annual winter monitoring, and identified a positive relationship between core area size and land cover diversity. To further evaluate the suitability of habitat for winter whooping cranes, Maxent software was used to create a winter whooping crane distribution model. The species distribution model was developed from over 5,000 whooping crane locations from six winter seasons and 35 environmental and ecological data layers that were considered to be potentially significant factors relating to the ecology of wintering whooping cranes. Environmental layers included salinity levels, hydric soils, and mesohabitats based on land cover characteristics for the extent of the study area. Ecological data included GIS-derived spatial layers such as distance from road and development, percent vegetative cover, and landscape metrics (patch density, patch diversity, and edge density). Correlation and principle component analyses, and a stepwise variable reduction method were used to determine the most parsimonious model with the minimum number of discriminatory variables. The final winter whooping crane distribution model included 11 data layers that matched observed whooping crane winter locations (AUC=0.824 ±0.005). Based on the final model, winter whooping crane distribution is negatively influenced by distance from seagrass, distance from estuarine vegetated marsh, patch contiguity, average density of road, and distance to woody habitat, and probability of occurrence increased with distance from anthropogenic development, percent estuarine vegetated marsh, and distance from road. Furthermore, results from the model suggest more suitable whooping crane winter habitat is located northeast of their current range and that more than 181,000 ha of highly suitable habitat is located within our study area, which exceeds the 125,000 ha of habitat targeted for conservation by the U.S. Fish and Wildlife Service to have enough winter habitat for 1,000 individual whooping cranes (USFWS 2012). In the final chapter, a grid-based simulation model was developed to evaluate carrying capacity for five distinct landforms that comprise the assumed current winter whooping crane range. Fixed minimum average territory sizes and randomized territory sizes following a Monte Carlo method were used to determine how many territories could be established by paired/family whooping cranes based on the spatial mosaic of available suitable habitat. Suitable habitat was further refined by fitting a probability density function to whooping crane observations and computed suitability values based on long-term winter whooping crane location data. Paired/family whooping cranes were determined to occupy habitat with suitability values >0.426. This threshold suitability value was selected based on the probability density function relationship. The grid-based model includes three preliminary rules to create territories, including analysis of a raster stack comprised of an occupancy layer, suitability values, and unique cell numbers. The model allowed a pair to start at one randomly chosen raster cell from the raster stack that was unoccupied and had a suitability value greater than the threshold. The model then used a spatially coherent adjacent suitable cell search function to expand territories into surrounding suitable cells. Once the territory size was met for the existing pair, the next pair randomly selected a suitable starting cell to establish a territory, and this process continued until no more territories could be established on each of the five landforms. Simulation using the Monte Carlo methods consistently estimated a greater number of potential territories on each landform compared to the fixed-territory size simulation. In total, the Monte Carlo method concluded a total of 159 territories could be established within the current range of winter whooping cranes, which equates to 318 reproductive pairs. However, only 47% of the total amount of suitable habitat for paired/family whooping cranes was included in the simulated territories. Lastly, approximately 186.3 km2 of suitable habitat for paired/family whooping cranes is located within protected areas, which leaves approximately 296.9 km2 of suitable habitat for territorial whooping cranes that are not protected on the winter grounds. These simulations are the first to use habitat suitability and space-use strategies in the prediction of carrying capacity of whooping crane wintering grounds. With the addition of more space-use strategies that will come from improved data collection methods, these simulation models can be modified to further evaluate habitat use and carrying capacity within the winter range of the endangered AWBP and assist in predicting future range expansion as the population increases towards down-listing goals.
Whooping crane, Species distribution model, Rule-based model, Habitat suitability, Maxent
Davis, N. (2019). <i>Predicting future range expansion of whooping crane (Grus americana) winter habitat using long-term census and remotely sensed data</i> (Unpublished dissertation). Texas State University, San Marcos, Texas.