- Homes on the Range - Range-wide assessment of priority areas for the conservation of prairie dog ecosystems: integrating habitat suitability, land use, and climate change Ana Davidson1, David J. Augustine3, Michael Menefee4, Dave Anderson4, Volker Radeloff5, Lindsey Sterling Krank6 1Department of Wildlife, Humboldt State University, Arcata, CA  2USDA-Agricultural Research Service, Fort Collins, CO  3Colorado Natural Heritage Program, Colorado State University, Fort Collins, CO  4Department of Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison, WI 5Prairie Dog Coalition, Boulder, CO IN A NUTSHELL: We propose to identify the top priority areas for conservation of prairie dog ecosystems across the geographic ranges of the five species of prairie dogs, from Canada to Mexico – spanning the Plains and Prairie Potholes, Great Plains, Desert, Great Basin, Great Northern, and Southern Rockies Landscape Conservation Cooperatives (LCC) regions. PROJECT SUMMARY: Prairie dogs (Cynomys spp.) are keystone species in North America’s central grasslands, but their populations have declined by over 98%, and are highly threatened by climate change and plague across this region. This has resulted in cascading losses in associated species and grassland habitat. To identify priority areas, we will first develop state-of-the-art habitat suitability models (HSMs) for all five species of prairie dogs, using ensemble forecasting and extensive spatial data that our team recently collected for all known prairie dog colonies and high resolution data on land cover, climate, elevation, slope, and soils. We will identify priority areas based on suitable habitat locations for each prairie dog species across both public and private land, as well as the presence of associated and threatened species, colony size, connectivity, and conservation potential, and use this together with expert opinion to map key areas to protect. We will identify the best areas to protect for both today’s climate and for the future, under climate change. Priority area model selection will be informed and reviewed by land managers and other regional experts, and model validation will be assessed with field data collected by experts. Resulting models and map products will inform managers on the key areas of conservation importance for prairie dogs and associated species across North America’s central grasslands. Our models will have immediate utility for land managers throughout the LCC region’s covered, and provide a valuable decision support tool for conservation and management, critically needed for mitigating the impacts of plague and climate change on prairie dog populations throughout their range. Last but not least, we will engage in community outreach to communicate our results, and work collaboratively with land managers and Non-Governmental Organizations (NGOs) to implement on-the-ground priority area conservation.

Our Team: Agency, NGO, University Partnership Dr. Ana Davidson (Humboldt State University) Dr. David Augustine (USDA-ARS)   Michael Menefee Dave Anderson (Colorado Natural Heritage Program) Dr. Volker Radeloff (University of Madison-Wisconsin) Lindsey Sterling Krank (Prairie Dog Coalition, PDC) And Coalition members, including Ghia Zalewa (PDC), Steve Forrest & Jonathan Proctor (Defenders of Wildlife), and Kristy Bly (WWF) PARTNERSHIPS AND ROLES: Dr. Ana Davidson (Research Ecologist, IWS and Adjunct Professor, UNM): has studied and published extensively on prairie dogs (Cynomys spp.) and associated species for over 20 years, and has expertise in ecological modelling, bioinformatics, and studying extinction risk in mammals. As lead PI on the project, she will be responsible for overseeing the overall project and ensuring the project’s success. Davidson’s role will be to lead the HSM modeling, priority area selection, and writing of the report for government agencies, and assist with ground-truthing.   Dr. David Augustine (Research Ecologist, USDA-ARS):  has managed and studied black-tailed prairie dogs and other wildlife habitats in the shortgrass steppe for the past 12 years. He also developed the habitat suitability model for black-tailed prairie dogs in the Great Plains LCC. He will assist and advise on the GIS modelling, priority area selection, and production of the final report for government agencies. Michael Menefee (Environmental Review Coordinator, CNHP): has headed environmental review for the Colorado Natural Heritage program for over 10 years. He has led GIS modelling efforts and regularly preforms spatial analysis for a wide variety of clients. He will be responsible for spatial data collection and preparation, assisting with GIS modeling and priority area selection, and writing of the report for government agencies. Dr. Volker Radeloff (Professor, University of Madison-Wisconsin): has published over 150 scientific papers in leading journals on spatial analysis in conservation and sustainability. He will advise on the spatial modeling and priority area selection for this project, provide spatial data, and co-author publications resulting from this work. Lindsey Sterling Krank (Environmental Scientist & Director, Prairie Dog Coalition): she has lead on-the-ground conservation and outreach of prairie dog ecosystems for over 10 years. Her role will be to lead the outreach efforts, aimed to disseminate the information from this project to government agencies and other NGO’s and help implement conservation plans based on the priority areas report.

History Update Homes on the Range 2005  Original Data Request in 2013 After discussion and research  Scientific Credibility  Better strategy- Larger landscape approach, collaborative  Evolution RANGEWISE ASSESSMENT OF PRIORITY AREAS FOR THE CONSERVATION OF PRAIRIE DOG ECOSYSTEMS: INTEGRATING HABITAT SUITABILITY, LAND USE PATTERNS AND CLIMATE CHANGE Follow up to Bob’s Luce’s 2004 Report. We recognize the states helped build that report and ultimately we’d have a much better result if we could get your input and participation on this report too. Same approach as Luce; but with updated methodology.

Homes on the Range addresses management needs Eco-regional landscape approach Uses keystone species Input from land & wildlife managers at the planning, review, and implementation phases Goes beyond SWAP planning Direct conservation actions that you want to implement Pro-active planning tool to get ahead of species listing Ex: Arctic Grayling in MT - Success story because Conservation planning and implementation got ahead of regulations. Current summaries on conservation actions from the Luce Report State of the art ensemble forecast modeling (landuse and climate change) Long-term planning & decision support tool

Prairie dogs are central to Homes on the Range because they are keystone species of North America’s central grasslands Because prairie dogs function as ecosystem engineers and keystone species in North America’s grasslands, their conservation and management lies at the core of many landscape-level conservation efforts Bergstrom et al., 2013, Conservation Letters

Conservation of prairie dogs has implications for efforts to conserve species that use alternative habitats -Loss of prairie dogs has resulted in declines in species associated with the habitats they create (burrowing owl and mountain plover) and those dependent or heavily reliant on prairie dogs as prey (black-footed ferrets and ferruginous hawks) Spatial optimization of conservation efforts will be essential where range of prairie dog dependent species overlap with their associate species

Overall Goals: 1) Generate models and map products that inform managers on key areas of conservation importance for all 5 prairie dog species and associated biodiversity, across North America’s central and western grasslands. Models will provide land managers throughout the 5 species range with a valuable decision support tool for conservation and management. 2) This effort will be followed by community outreach to communicate results, and working collaboratively with land managers and NGOs to implement the on-the-ground conservation actions PROJECT GOALS & OBJECTIVES: We propose to identify the top priority areas for prairie dog ecosystem conservation across the ranges of all 5 prairie dog species, from Canada to Mexico (Fig. 3). Our project will be focused on identifying priority areas on public lands, and consists of three phases; the first of which already has been completed. The focus of the first phase was to collect data on all known current and historic prairie dog occurrences and colonies on public lands. The second phase will focus on spatial modelling to identify priority areas for conservation. We will do this under today’s climate and project priority areas into the future, under a warming climate. We will inform and validate our model with ground-truthing and expert opinion, and publish the priority areas maps in a report titled, “Homes on the Range: top priority areas for prairie dog ecosystem conservation”, for NGOs and agencies to use for guiding their conservation and management efforts. We will also provide a two-page report summary for wildlife and natural resource managers and convey the results of the project in LCC-hosted webinars. The third phase of our project will focus on community outreach to communicate our results, and working collaboratively with land managers and NGO’s to implement on-the-ground priority area conservation. Here, we are seeking funds to support the second and third phases of our project. Geographic distribution of the five prairie dog species

Methods: Identifying suitable landscapes Our process for identifying priority areas will involve 3 steps: Identify a large pool of suitable landscapes that could contribute to future prairie dog conservation Compile existing data from state, federal and other land managers on current status/distribution of prairie dogs and associated species. Compile GIS layers on land ownership, land use/land cover Use (a) and (b) combined with analyses of existing distribution of habitat patch sizes and connectivity to identify landscapes for more detailed analyses of habitat suitability. First, at the scale of the geographic range of each species, we will identify suitable landscapes based on landcover and landscape composition. Specifically, we will use the National Landcover Database to identify extensive areas of native grassland (i.e., to mask out areas of cropland, forested land, shrubland, urban, and open water). In addition to measuring the amount and percentage of these habitats in a given area (landscape composition), we will use the program GUIDOS (http://forest.jrc.ec.europa.eu/download/software/guidos/) to quantify landscape structure and especially the amount of grassland that is core-habitat in large patches, the amount that is in small, isolated patches, and the amount that is in corridors connecting core areas. Landscapes will also be evaluated in terms of the amount of public versus private landownership within them, and the type of public land. We will select landscapes within the geographic ranges of prairie dogs based on their size, ownership, and connectivity of grassland patches at a broad spatial scale, recognizing that only landscapes > 25,000 acres in size are likely to make major contributions to the conservation of prairie-dog-associated species such as black-footed ferrets and mountain plovers. This first step of identifying large, suitable landscapes from which we can conduct habitat suitability models and select priority areas is an important strategy for making this project logistically feasible, considering the extensive amount of data acquisition, preparation, and analysis required for such a large area that spans across much of the central US.

Methods: Identifying suitable landscapes Our process for identifying priority areas will involve 4 steps: Identify a large pool of suitable landscapes that could contribute to future prairie dog conservation Compile existing data from state, federal and other agencies on current status/distribution of prairie dogs and associated species Compile GIS layers on land ownership, land use/cover Use (a) and (b) combined with analyses of existing distribution of habitat patch sizes and connectivity to identify landscapes for more detailed analyses of habitat suitability. First, at the scale of the geographic range of each species, we will identify suitable landscapes based on landcover and landscape composition. Specifically, we will use the National Landcover Database to identify extensive areas of native grassland (i.e., to mask out areas of cropland, forested land, shrubland, urban, and open water). In addition to measuring the amount and percentage of these habitats in a given area (landscape composition), we will use the program GUIDOS (http://forest.jrc.ec.europa.eu/download/software/guidos/) to quantify landscape structure and especially the amount of grassland that is core-habitat in large patches, the amount that is in small, isolated patches, and the amount that is in corridors connecting core areas. Landscapes will also be evaluated in terms of the amount of public versus private landownership within them, and the type of public land. We will select landscapes within the geographic ranges of prairie dogs based on their size, ownership, and connectivity of grassland patches at a broad spatial scale, recognizing that only landscapes > 25,000 acres in size are likely to make major contributions to the conservation of prairie-dog-associated species such as black-footed ferrets and mountain plovers. This first step of identifying large, suitable landscapes from which we can conduct habitat suitability models and select priority areas is an important strategy for making this project logistically feasible, considering the extensive amount of data acquisition, preparation, and analysis required for such a large area that spans across much of the central US. Need Agency/Expert input on how to weight these factors when developing the pool of landscapes to analyze

Methods: Habitat Suitability Modelling Our process for identifying priority areas will involve 4 steps: Identify a large pool of suitable landscapes that could contribute to future prairie dog conservation Generating habitat suitability models within these landscapes using state-of-the-art ensemble forecasting prairie dog occurrence data and environmental variables Once the models are parameterized with current climate, we will generate HSMs based on projected climate in 2080 Habitat Suitability Modelling: Within the identified suitable landscapes, we will develop habitat suitability models using state-of-the-art ensemble forecasting (Araújo and New 2007). We will use ‘biomod2’ package in R to build our habitat suitability models, which allows for combining predictions across multiple (an “ensemble” of) modelling techniques, including selected maximum entropy modeling (MaxEnt), generalized linear models (GLM), random forest models (RF) generalized additive models (GAM), gradient boosting models (GBM), to depict the relationships of species with their environment (Thuiller et al. 2009, 2012). Ensemble modeling is a powerful approach for simulating species distributions across different initial conditions, model classes, model parameters, and boundary conditions in order to improve prediction accuracy (Araújo and New 2007). To build our model, we will use the known prairie dog occurrence data that our team has collected for all prairie dog species on public lands, and use similar environmental variables to those used by Augustine et al. (2012) for developing black-tailed prairie dog HSMs within the Great Plains LCC (Table 1) (Guisan and Thuiller 2005, Augustine et al. 2012, Stanton et al. 2012). We will parameterize the HSMs with our data of prairie dog occurrences, most of which are located on public lands due to data availability, but our habitat suitability maps will cover both public and private lands throughout each landscape. A key component of our work is that the influence of topography and soil attributes on prairie dog habitat suitability will be modelled at the highest spatial resolution that input datasets (i.e., National Elevation Datasets and the Soil Survey Geographic database) will currently permit, thus allowing us to evaluate and compare habitat suitability across priority areas at a spatial resolution that is relevant to land managers (i.e., at scales of individual pastures and properties). We will use the range boundaries of each species to limit the area from which random pseudo-absence points can be selected during the model fitting process. Once the models are parameterized with current climate, we will generate HSMs based on projected climate in 2080 (Guisan and Thuiller 2005, Stanton et al. 2012).

Black-Tailed Prairie Dog Habitat Suitability Model GPLCC

Methods: Identifying priority areas Our process for identifying priority areas will involve 4 steps: Identify a large pool of suitable landscapes that could contribute to future prairie dog conservation Generating habitat suitability models within these landscapes using state-of-the-art ensemble forecasting prairie dog occurrence data and environmental variables Once the models are parameterized with current climate, we will generate HSMs based on projected climate in 2080

Online Portal for HOTR Products: Making Reports, Data and Maps Easily Accessible for Researchers and Land Managers CNHP has extensive experience hosting online data portals, with its Colorado Wetland Information Center serving as an excellent example of a rich online data source that researchers and land managers rely on across the state. CNHP partners with CSU’s research library system to ensure all public reports, data and maps are accessible through the library. Finally we would reach out to the USGS to host key products on Science Base as they did for the SGP LCC study led by Dave Augustine.

Using Prairie Dog Data in Management NRCS Black-footed ferret Initiative http://www.nrcs.usda.gov/wps/portal/nrcs/detail/mt/programs/financial/eqip/?cid=stelprdb1253844 The NRCS Black-footed ferret Initiative is using the results of the GP LCC study to help select reintroduction sites.

Methods: Identify priority areas under landuse & climate change Our process for identifying priority areas will involve 4 steps: Identifying suitable landscapes Generating habitat suitability models within these landscapes Incorporate landuse and climate change predictions into habitat suitability models (climate envelope approaches) Habitat Suitability Modelling: Within the identified suitable landscapes, we will develop habitat suitability models using state-of-the-art ensemble forecasting (Araújo and New 2007). We will use ‘biomod2’ package in R to build our habitat suitability models, which allows for combining predictions across multiple (an “ensemble” of) modelling techniques, including selected maximum entropy modeling (MaxEnt), generalized linear models (GLM), random forest models (RF) generalized additive models (GAM), gradient boosting models (GBM), to depict the relationships of species with their environment (Thuiller et al. 2009, 2012). Ensemble modeling is a powerful approach for simulating species distributions across different initial conditions, model classes, model parameters, and boundary conditions in order to improve prediction accuracy (Araújo and New 2007). To build our model, we will use the known prairie dog occurrence data that our team has collected for all prairie dog species on public lands, and use similar environmental variables to those used by Augustine et al. (2012) for developing black-tailed prairie dog HSMs within the Great Plains LCC (Table 1) (Guisan and Thuiller 2005, Augustine et al. 2012, Stanton et al. 2012). We will parameterize the HSMs with our data of prairie dog occurrences, most of which are located on public lands due to data availability, but our habitat suitability maps will cover both public and private lands throughout each landscape. A key component of our work is that the influence of topography and soil attributes on prairie dog habitat suitability will be modelled at the highest spatial resolution that input datasets (i.e., National Elevation Datasets and the Soil Survey Geographic database) will currently permit, thus allowing us to evaluate and compare habitat suitability across priority areas at a spatial resolution that is relevant to land managers (i.e., at scales of individual pastures and properties). We will use the range boundaries of each species to limit the area from which random pseudo-absence points can be selected during the model fitting process. Once the models are parameterized with current climate, we will generate HSMs based on projected climate in 2080 (Guisan and Thuiller 2005, Stanton et al. 2012).

Climate Change Prairie dogs in the lower elevation and southern part of their range are threatened by drought Consistent with long-term prairie dog population declines in the American Southwest and northern Mexico, due to drought Stresses by climate will increase with continued climate warming Base our models on projected climate in 2080 Davidson et al., 2014, JWM

Methods: Priority area rankings Our process for identifying priority areas will involve 4 steps: Identifying suitable landscapes Generating habitat suitability models within these landscapes Incorporate climate change predictions into habitat suitability models (climate envelope approaches) 4. Identifying current & future priority areas within predicted suitable habitat for each prairie dog species Quantification of top priority areas Expert input Habitat connectivity Habitat Suitability Modelling: Within the identified suitable landscapes, we will develop habitat suitability models using state-of-the-art ensemble forecasting (Araújo and New 2007). We will use ‘biomod2’ package in R to build our habitat suitability models, which allows for combining predictions across multiple (an “ensemble” of) modelling techniques, including selected maximum entropy modeling (MaxEnt), generalized linear models (GLM), random forest models (RF) generalized additive models (GAM), gradient boosting models (GBM), to depict the relationships of species with their environment (Thuiller et al. 2009, 2012). Ensemble modeling is a powerful approach for simulating species distributions across different initial conditions, model classes, model parameters, and boundary conditions in order to improve prediction accuracy (Araújo and New 2007). To build our model, we will use the known prairie dog occurrence data that our team has collected for all prairie dog species on public lands, and use similar environmental variables to those used by Augustine et al. (2012) for developing black-tailed prairie dog HSMs within the Great Plains LCC (Table 1) (Guisan and Thuiller 2005, Augustine et al. 2012, Stanton et al. 2012). We will parameterize the HSMs with our data of prairie dog occurrences, most of which are located on public lands due to data availability, but our habitat suitability maps will cover both public and private lands throughout each landscape. A key component of our work is that the influence of topography and soil attributes on prairie dog habitat suitability will be modelled at the highest spatial resolution that input datasets (i.e., National Elevation Datasets and the Soil Survey Geographic database) will currently permit, thus allowing us to evaluate and compare habitat suitability across priority areas at a spatial resolution that is relevant to land managers (i.e., at scales of individual pastures and properties). We will use the range boundaries of each species to limit the area from which random pseudo-absence points can be selected during the model fitting process. Once the models are parameterized with current climate, we will generate HSMs based on projected climate in 2080 (Guisan and Thuiller 2005, Stanton et al. 2012).

Methods: Ground-truthing, expert review, and priority area selection Priority area model selection will be informed and reviewed by land managers and other regional experts Attend State agency prairie dog working groups Contact key individuals directly Provide template maps using LandScope (http://www.landscope.org/) Model validation will be assessed with ground-truthing by experts Hold workshops on modeling results. Request feedback on draft results via meetings or workshops with land managers, agency representatives and other stakeholders Ground-truthing & expert review: Priority area model selection will be informed and reviewed by land managers and other regional experts, and model validation will be assessed with ground-truthing by experts. To engage land managers and regional experts, we will attend scheduled State agency prairie dog working groups and contact key individuals directly. We will provide template maps using LandScope (http://www.landscope.org/), the NatureServe software program, whereby a person can directly draw key areas onto a map, which can then be exported as a spatial layer. We will ground-truth the best areas selected by our model, and informed by experts, to confirm the presence of prairie dogs and associated species and conservation potential. Our ground-truthing efforts will involve site visits with local land managers to assess and confirm the presence of prairie dogs and associated species, colony health and activity, and conservation potential.

Outreach & Implementation We will partner with land managers and landowners to implement priority area conservation. Fundraising for on-the-ground conservation implementation Engaging with local communities through field projects, education, and community outreach Outreach: In addition to conducting outreach to inform and review our modeling efforts, we will partner with land managers and landowners to implement priority area conservation. These efforts will involve fundraising for on-the-ground conservation implementation, and engaging with local communities through field projects, education, and community outreach. The US Fish and Wildlife Service and Natural Resources Conservation Service have recently implemented a prairie dog incentive package in certain western states, which also provides an exciting new opportunity for facilitating such on-the-ground conservation. Outreach and support from land managers is key to creating net conservation gains for the prairie dog ecosystem. Our team has extensive experience in community outreach and leading on-the-ground conservation throughout NA’s the central grasslands, and the partnerships we already have in place from our decades of experience working in this system are fundamental to seeing priority area conservation realized. For this project, specifically, we have worked successfully with twelve state agencies, Natural Heritage Programs, federal land managers and the Western Association of Fish & Wildlife Agencies (WAFWA) to obtain prairie dog occurrence data from all States and in Mexico and Canada. We will continue to work closely with these agencies and organizations during the next phases of our project.

LCC Steering Committees LCC Focal Areas Desert Geographic Area LCC Steering Committees

Products Maybe remove this slide? A report titled, “Homes on the range: top priority areas for prairie dog ecosystem conservation”, for NGO’s and agencies to use for guiding their conservation and management efforts. And… HSMs and GIS data layers of colonies and priority areas for NGO’s and agencies Private land areas that have the greatest leverage for land exchanges and purchase Implementing the priority areas conservation on-the-ground Publishing a peer-reviewed scientific paper on the work.

Timetable Keep this slide?

Budget BUDGET: $250,000 GIS specialist Roles of Menefee, Sterling-Krank, and Davidson Travel

Questions?

Variables