Date of Award
Doctor of Philosophy (PhD)
Wildlife and Fisheries Biology
Robert F. Baldwin, Committee Chair
Robert B. Powell
David L. White
Hoke S. Hill Jr
Systematic conservation planning is a fertile scientific discipline capable of examining global natural resource threats including land cover conversion, habitat fragmentation, biodiversity loss, and climate change. Many of these threats are acting in concert to accelerate the rate of change which increases the urgency of action. The scale of these changes creates unique spatial analysis and computational challenges. This study was designed to address several of these challenges pertaining to the analyses of publically protected areas, habitat connectivity, and private lands conservation. I examine how using high-throughput and high-performance computing helps conservation practitioners ask and solve larger ecological questions. I describe when practitioners might consider using these technologies based on the types of questions that are asked. I improve upon existing circuit-theory based habitat connectivity modeling by using parallel processing and develop new methodologies which facilitate fine-grained analyses over broad geographic extents. I explore how species-level habitat connectivity can inform broad patterns of current and future landscape connectivity in the southeastern U.S. by incorporating species with a diverse group of dispersal abilities and including forecasted threats to connectivity. Since most of the conservation lands in the eastern U.S. are privately-owned, I explore the spatial patterns, distribution, and social covariates around private lands conservation (e.g., easements) throughout the Appalachian region. Harnessing the power of high-throughput and high-performance computing facilitated a 72-125 times speedup for several large-landscape spatial analyses. These improvements arm researches with the ability to ask questions which more closely capture the complexity seen in nature. Circuit-theory based habitat connectivity is notoriously computationally expensive and the results presented herein suggest computational bottlenecks can be overcome with a 160 times speedup. Moreover, the results illustrated circuit-theory based connectivity maps converge on a near solution much faster than previously thought and thus the description of this â€˜convergence factor' will save researchers many hours of computation and make broad-extent analyses practicable. Species-derived connectivity maps indicate far fewer large contiguous areas will be available to facilitate animal movements across ecoregions in 2100. Overall species potential habitat cores were reduced by 36% reaching as high as 72% for one species. I observed a more simplified landscape connectivity in 2100 where values were dissected and depleted for the highest quantiles. Despite these discouraging findings, private lands conservation may provide a â€˜stop-gap' measure to slow these changes. I found conservation easements throughput Appalachia nearer to urban developments, major road networks, on arable lands, and surrounded by areas of higher diversity than random. These results differ from that of global protected areas but the drivers of easement location may be more tightly linked with social processes, rather than environmental ones. Collectively, this research can help guide practitioners about what science and technology is needed and at the appropriate scale to match their conservation questions. Although the increasing threats to biodiversity conservation are well documented, strategies such as those presented here may help facilitate positive conservation outcomes.
Leonard, Paul, "SOLUTIONS TO SPATIAL AND COMPUTATIONAL CHALLENGES FOR LARGE-LANDSCAPE CONSERVATION PLANNING" (2016). All Dissertations. 1835.