Page 43 - GIS for Science, Volume 3 Preview

DISTRIBUTION DATA FOR BIODIVERSITY CONSERVATION
Transportation and energy infrastructure, forest harvests, grazing, pesticide use, and residential and commercial development (among many others) take place in specific locations. A major goal of conservation is to ensure that these activities are carried out in areas away from the habitats where imperiled species occur. The more precise the species distribution is known, the better we can avoid conflicts between human activities and species’ needs. However, the complete distribution of many species is often challenging to determine with precision, especially for rare species. They can be hard to detect or identify, occur in remote areas, or undergo seasonal movements, creating challenges to documenting their distributions.
Traditionally, conservation practitioners have used two approaches to compiling distribution information. Field surveys conducted by qualified botanists and zoologists that produce confirmed records of species presence (and, sometimes, confirmed absence) offer the highest degree of spatial accuracy and confidence. But comprehensive biodiversity inventory is an immense challenge that has never been fully realized. The time and resources required for field surveys, the numbers of species involved, the inaccessibility of most privately owned land, and the vastness of natural landscapes combine to limit the numbers of records available for most species. As a result, field-based observations will almost always significantly underestimate species distributions.
The alternatives most often employed are coarse estimates of species ranges. For example, many US Fish and Wildlife Service (FWS) range maps of threatened and endangered species include the counties where species are thought to occur. Another example is broad-range maps developed by experts who depict boundaries around known localities. These types of maps almost always overestimate species distributions, encompassing much area of unoccupied habitat (Hurlbert and Jetz 2007). A potential consequence is ill-informed land use decisions and unnecessary conflicts for areas where a species may not exist.
Today, NatureServe integrates GIS with cloud-based computational analyses, machine learning algorithms, and web-based tools for collaborative spatial
science to transform distribution information for our nation’s most vulnerable species. The foundation of this approach is habitat suitability modeling, also called species distribution modeling, an academically mature tool that combines verified records of species occurrences with environmental data layers to identify a suite of conditions that describe a species’ habitat requirements (Guisan and Thuiller 2005). Machine learning algorithms then identify where else that suite of conditions occurs on the landscape. The output map is a probability distribution, with higher values indicating the closest match to the conditions at localities where the species is verified to occur.
NatureServe applies these tools to produce a dramatic refinement in our understanding of imperiled species distributions. For decades, NatureServe’s network of botanists, zoologists, and ecologists have conducted field surveys using standardized approaches for documenting species occurrences and identifying the places and conditions most important for their persistence. Shared data management systems support the aggregation of these data into a national database of verified locations of a wide range of species, from ferns to frogs, mussels to mammals, and pollinators to vascular plants. This same network of experts also reviews model outputs, evaluating where they under- or overpredict, allowing models to be adjusted to produce more precise estimates of imperiled species distributions.
An example of the power of habitat models can be seen in the golden-cheeked warbler (Setophaga chrysoparia). This declining neotropical migratory bird is designated as endangered under the US Endangered Species Act (ESA) and is considered globally imperiled (G2) by NatureServe’s conservation status assessment. Distribution data provided by the FWS, which administers the ESA for terrestrial and freshwater plants and animals, shows that the species occurs across a swath of counties in central Texas. The Texas Department of Transportation (DOT) must consider this listed species for potential impacts from any planned road construction projects in these counties. But the habitat for this beautiful little songbird is not likely to occur throughout all these counties. A habitat suitability model, based on verified
ABC
Varying depictions of the distribution of the golden-cheeked warbler in Texas. Map A, county distribution provided by the FWS. Map B, modeled habitat probability. Map C, habitat model of thresholds, showing areas where the warbler is most likely to occur.
Preventing Species Extinctions 31