Browsing by Subject "Maxent"
Now showing 1 - 10 of 10
- Results Per Page
- Sort Options
Item Open Access American chestnut (Castanea dentata) habitat modeling: identifying suitable sites for restoration in Shenandoah National Park, Virginia(2013-12-06) Santoro, JenniferSince 2008, The American Chestnut Foundation’s (TACF) Appalachian Trail MEGA-Transect Project has engaged citizen scientists to collect American chestnut occurrence data over the length of the Appalachian Trail. This data helps TACF to locate surviving trees for use in their breeding program and expand their knowledge of chestnuts across the East Coast. However, this dataset is limiting in that it considers only the ridge-top habitat of the trail. To remedy this, we conducted an extensive sampling of side-trails in Shenandoah National Park in order to study more diverse elevation and habitat gradients. Expanding the dataset allows us to draw more informed conclusions about habitat for surviving American chestnuts. To achieve this, I developed a series of species distribution models, including GLM, CART, and Maxent models, based on field observations and spatial data of environmental variables. These predictive distribution models were then combined to generate a comprehensive map of the most likely surviving American chestnut occurrence locations across Shenandoah National Park. Additionally, projections based on future climate were made for the Maxent model to 2050 and 2070 in order to see if habitat for surviving trees might shift in the face of climate warming. Overall, the three species distribution modeling techniques tended to agree on location, but not quantity, of suitable habitat for surviving chestnuts. All models found elevation, sand, and slope to be the most significant habitat predictors in Shenandoah. Climate change models produced only subtle range shifts; as a generalist species, American chestnuts may not face adverse effects of future climate warming. Mapping these results provides valuable information to both Shenandoah National Park and TACF as they continue to search for, study, and restore American chestnuts in the Appalachian forest.Item Open Access Brown Hyena (Hyaena brunnea) Distribution: Nuances in Modeling a Generalist Species(2020-04-24) Killea, AllisonThe brown hyena plays a critical role as a top carnivore across southern Africa alongside other megafauna including lions and cheetahs. However, because it is perceived to face fewer threats to its population, there are fewer research efforts aimed at understanding its distribution and spatial ecology. The Hyaena Distribution Mapping Project in partnership with the IUCN Hyaena Specialist Group is currently working to update the global range map of the brown hyena. A critical follow up to this work is the creation of a species distribution model to better understand the environmental factors that affect their spatial ecology. However, given the wide variation of habitat types within the hyena’s range, a single model may not be sufficient to capture the nuances within their global distribution. I created a global model of the brown hyena as well as two biome specific models, one for Deserts and Xeric Shrublands and one for Tropical and Subtropical Grasslands, Savannas, and Shrublands as defined by the World Wildlife Fund. I propose that brown hyena ecology varies based on the specific ecosystem of an individual and that by modeling their distribution at a biome level we can better predict potential habitat. However, a general model may still be informative as hyenas do range widely, and one individual can take advantage of a variety of habitats. By comparing these models and their performance, we can create a more robust description of how brown hyena ecology varies throughout southern Africa.Item Open Access Identification and prioritization of lands for restoration of Piedmont prairie in North Carolina(2007-12-07T15:59:32Z) Taecker, EricaIn the central Piedmont of North Carolina, prairies and savannas were noted by European settlers to have covered a significant portion of the landscape. Piedmont prairie is valued for its extraordinary biodiversity; at least 277 plant species, some endemic, are associated with this unique area. Rich prairie ecosystems in the Piedmont were maintained by both naturally-occurring and human-ignited fires, which created open fields or patches of prairie within oak-pine-hickory or Piedmont longleaf pine forests. Anthropogenic changes to fire regimes and land use have fragmented the Piedmont prairie ecosystem, such that several of its plant species are now federally endangered. Effective conservation of this native ecosystem in our rapidly developing state depends on a solid understanding of its science. Just as importantly, it necessitates the ability for conservation agencies to act efficiently to protect and maintain areas of intact prairie, while quickly identifying and protecting other areas with restoration potential. This masters project compares the suitability of two multivariate modeling tools, CART (Classification and Regression Tree) and Maxent (Maximum entropy), for predicting the potential geographic distribution of the Piedmont prairie ecosystem in nine Piedmont counties of North Carolina. Natural Heritage “Element Occurrence” point location data of four prairie species were the basis for the models, which considered environmental variables such as elevation, topographic relative moisture index (TRMI), slope, relative aspect, soil clay content, and soil effective cation exchange capacity (ECEC) in the prediction of potential prairie extent. Further, a basic prioritization of the resulting prairie “habitat” patches mapped in GIS highlights areas adjacent to existing protected areas in which to focus conservation and restoration efforts. The results indicated that the habitat model of prairie created by Maxent reasonably predicts known prairie species occurrences without over-generalizing the possible distribution of prairie in the study area. Maxent also highlights that ECEC is the most important predictor variable of prairie distribution, followed by clay content. The CART technique resulted in similar accuracy and explanatory variables, but when mapped, “habitat” covered a large proportion of the study area, less useful for targeting regions for further study. The preliminary prioritization suggests that several zones around Charlotte, NC and in Davidson County warrant further investigation for prairie remnants. With sufficient additional information about current land use and cover, the prioritization can be further refined to reduce the effort needed to find suitable sites for the restoration and conservation of Piedmont prairie and its associated forest cover types.Item Open Access Invasive Exotic Plants of the Eno River Watershed(2010-04-30T20:02:44Z) Starke, LesleyInvasive exotic species are an international threat to biodiversity. Management of invasive species is divided into three approaches: prevention of introduction outside of native range; eradication of invasions; and containment and control strategies. Prevention is unfortunately limited by accurate predictions and border control measures which are difficult to implement. Similarly, eradication is made difficult due to the fast acting and aggressive behavior of many invasive species, some of which are naturalized for many years before control measures are implemented. This leaves containment and control as management strategies for many managers today. Land protection groups in the United States including non-profit land trusts and governmental agencies – local to national -- address invasive species on nearly all protected lands. I have consulted with the Eno River Association of Durham and Orange counties in North Carolina to address the management of three invasive plant species of concern: tree of heaven (Ailanthus altissima), Chinese privet (Ligustrum sinense), and multiflora rose (Rosa multiflora). After assembling an observational data set of these three species, I used Maxent, a maximum entropy based machine-learning software, to model the potential distribution of each species within the Eno River watershed. Distributions of all three species were best predicted by soil type and distance to rivers. Properties of the Eno River State Park master plan -- a land protection priority list for the Eno River Association and the Eno River State Park -- were analyzed and ranked for the total area and the percent coverage of invasive plants from the modeled distributions.Item Open Access Invasive Plant Management Plan for the Duke Forest, Durham, NC(2009-04-24T19:36:52Z) Wright, LornaRecently, populations of invasive plants are increasing in the Duke Forest and detrimentally affecting the growth of Pinus taeda (loblolly pine) and natural forest communities. To effectively control the spread of invasives, the Duke Forest Resource Manager needs to know what invasive plants are located throughout the Forest, where they are located and what factors are associated with their presence. Therefore, a sample of the invasive plant population was recorded using a GPS unit. This data was then used to model the distribution of each of the invasive plants throughout the Forest. I used Maxent to create these predicted distributions. Ailanthus altissima, Lonicera japonica, and Microstegium vimineum are the dominant invasive species present. Of the 15 invasive plants recorded, most species were found along roads and streams. The management activity that was most correlated with presence of invasive plants was harvesting, though no specific harvesting technique (i.e. seed-tree, salvage, selective or clear cut) predicts invasive plant presence than any other. The predicted distribution maps will be used to complete a targeted inventory of invasive plants throughout the Duke Forest. The inventory process should begin in Natural Heritage areas predicted to have high priority species and multiple invasive species. When feasible, control treatments should be applied at the same time as inventorying particularly on small, peripheral populations. In addition, the Duke Forest Manager can prevent future invasions through monitoring and early removal of plants in areas where soil disturbing management activities have taken place.Item Open Access Maximum Entropy Modeling for Photovoltaic Optimization: A Spatial Analysis of California(2012-04-26) Schrager, SamuelAdoption of renewable energy for electricity generation holds significant potential to produce emission reduced power, in addition to other benefits. Great promise has been expected for solar energy, specifically photovoltaic (PV) solar panels, which harness the photoelectric effect to produce electricity. Over the past 20 years, the average number of PV facilities in the United States have increased nearly 40% per year. Even with this growth, as of 2011, PV provided only .2% of all national electric generating capacity. However, if historical growth trends can continue, solar energy may stand on the threshold of much larger adoption rates. Within this context, a holistic understanding of the social, environmental and economic elements that play a role in aiding PV growth may prove fruitful. Questions, which have remained unanswered, include what parties adopt, what kind of sites are most common, where development most frequent and what are the drivers of PV adoption? In this paper, a literature review of quantitative and social studies, related to PV adoption is conducted. The output of the literature review is used to select environmental, economic and social variables, which guide a spatial model building process. The model goal is to predict PV adoption hotspots. The spatial boundary of this study is limited to California, chosen due to a national dominance of the PV market. A maximum entropy based model, Maxent, was selected due to its high regard within the field of species distribution modeling, its ability to predict unoccupied habitat, the complex relationships it fits between indicator variables, and the ease at which it integrates and visualizes spatial data. Verification is conducted by comparing model output with historical adoption trends and remote sensing. Model results are analyzed for potential PV utilization market segment and policy implications. The study concludes with suggestions for further research.Item Open Access Modeling Pronghorn Migration Corridors in the Northern Great Plains(2010-04-29T17:33:50Z) Poor, Erin E.While terrestrial seasonal migrations worldwide continue to decline, the migrations of pronghorn antelope (Antilocapra americana) in the Northern Great Plains remain poorly studied. Development due to energy exploration and extraction within this region has recently increased, possibly placing restrictions on pronghorn movement. In this study, I used two habitat modeling methods, maximum entropy and Analytical Hierarchy Process (AHP), to identify suitable pronghorn habitat, and two corridor modeling methods, circuit theory and least-cost path, to identify seasonal migration corridors. I used pronghorn locations during the 2008 spring and fall migration seasons to identify which habitat and corridor models perform the best based on pronghorn occupancy and area included in the corridor. The maximum entropy model performed better than AHP, resulting in corridors that included more pronghorn locations than those created using AHP. Additionally, corridors derived from circuit theory included more pronghorn locations within a smaller area than those created using the least-cost path method. I recommend using the maximum entropy cost surface and the area of overlap between the two corridors at 15% habitat suitability level for future management actions. Without further study and conservation efforts built on this new knowledge, pronghorn populations may eventually decline and the functioning and biodiversity of the Northern Great Plains may be greatly impaired.Item Open Access Species Distribution Modeling for Bog Turtles (Glyptemys muhlenbergii) in North Carolina(2013-12-06) Dick, KevinThe bog turtle (Glyptemys muhlenbergii) is the smallest turtle species in North America and is listed as a threatened species under the federal Endangered Species Act. Accurate detection of its specialized wetland habitat and subsequent tagging of individuals for monitoring purposes is critical for improving conservation efforts with this species. Parts of the Piedmont region in North Carolina have historically served as habitat for bog turtles, but few populations are now known to occur there. Increases in residential development, agricultural land use, and the draining of wetland areas over the past several decades have likely contributed to their current extirpation from this part of the state. Most wildlife managers no longer survey for bog turtles in most of the Piedmont as efforts are both time and cost prohibitive, and funding generally all allocated for work in counties where they have a better chance of locating bog turtles during a given survey event. Several managers acknowledge that there may still be bog turtles living in the Piedmont, but because of present limitations, there is currently no conservation plan for them. GIS and predictive modeling were used as a low-cost method for locating potential sites within four North Carolina counties that exhibit suitable habitat characteristics for bog turtles. Such predictions may prove useful in documenting new occurrences of bog turtles in both the Piedmont counties of Iredell, Davie, and Davidson, as well as in the higher quality bog turtle habitat regions of Wilkes County. The Maxent distribution model was used as it is capable of producing accurate habitat predictions for species with small sample sizes. A total of 28 areas with species presence and 16 different environmental variables were used in the analysis. The model returned several sites within Wilkes County exhibiting higher levels of predicted suitability, and a smaller number of sites within Iredell County with moderate levels of suitability. The predicted sites in Iredell County were previously unknown to wildlife managers, and may help to direct future survey work in those locations. If these model predictions can be translated to positive detection of turtles in the field, spatial modeling work of this kind may begin to play a larger role in the conservation efforts for the species.Item Open Access Use of Site Occupancy Modeling to Delineate a Jaguar Corridor in Southern Belize(2010-04-28T20:08:08Z) Petracca, Lisanne S.This analysis, a component of Panthera's Jaguar Corridor Initiative, incorporated 184 interviews with local hunters and farmers in a site occupancy framework to determine the likelihood of jaguar and jaguar prey occupancy in 90 16-km2 grid cells in the Toledo District of southern Belize. This output was compared with that of Maxent, a presence-only species distribution modeling technique, to determine if both approaches led to similar conclusions. Site occupancy analysis revealed that jaguar occupancy was associated with percent daily chance of seeing armadillo, higher elevation, and proximity to protected areas and forest cover. Prey species analysis revealed that likelihood of white-lipped peccary (Pecari tajacu) occupancy was associated with greater forest cover and proximity to protected areas and water/wetlands; collared peccary (Tayassu pecari) occupancy with greater forest cover and proximity to agriculture and settlements; red brocket deer (Mazama americana) occupancy with greater forest cover, proximity to agriculture, and higher elevation; and armadillo (Dasypus novemcinctus) occupancy with greater agricultural area, lower elevation, and greater distance from water/wetland. Site occupancy models were unable to be fitted for the paca (Agouti paca) and white-tailed deer (Odocoileus virginianus) populations. Following the weighting of jaguar output to 2.0 and prey species output to 1.0, total Psi (probability of occupancy) was calculated for each of the 90 grid cells. Cells with the maximum possible Psi value (meaning that the jaguar and all modeled prey species were present) were identified, and the final cells were chosen from this subset based on having relatively low future threat. These cells were cross-checked with Maxent output to ensure that the corridor contained areas of high habitat suitability. The proposed corridor extends along the eastern flank of the Toledo District and connects Sarstoon-Temash National Park with the protected areas of local NGOs TIDE (Toledo Institute for Development and Environment) and YCT (Ya’axche Conservation Trust). The analysis concludes with recommendations and concerns specific to the communities that fall within the corridor.Item Open Access Wolves in lower Michigan: habitat modeling and capacity estimation(2010-12-07) Claeys, GinaGray wolf populations have been rebounding in the Great Lakes region, after being nearly extirpated from the conterminous United States. Breeding populations of wolves have established in the Upper Peninsula of Michigan but have not yet become widely established in the Lower Peninsula (LP) of Michigan, despite several sightings of wolves in the LP since 2004. The objective of this analysis was to determine the quantity of potential habitat for wolves in the LP using wolf occurrence data obtained by radio telemetry, Maxent software, and ArcGIS. The habitat model was evaluated to determine if the potential habitat could support a viable population. A species distribution model (SDM) was created in Maxent and analyzed in ArcGIS to estimate the amount of potential habitat for wolves in the LP. Connectivity of potential habitat was evaluated by looking at least cost paths, corridors, and potential dispersal land. I found that there is 2,674-3,246 km2 of potential habitat in the LP, which could support 52-63 wolves. This is below the minimum of 100 necessary to sustain a viable population. All habitat patches at least 50 km2 in size are in the northern LP, but are not connected by land suitable for dispersal. Thus, although there are habitat patches of sufficient size in lower Michigan, it is likely that packs of wolves that may inhabit these patches will remain relatively isolated and have difficulties dispersing throughout lower Michigan. It is unlikely that they will be able to inhabit or disperse through the southern LP without land management geared towards creating larger unfragmented habitat patches and suitable dispersal corridors. This analysis highlights the need for land management geared towards wolves in order to allow them to continue to recover their former range in the United States.