Preliminary Report: Mapping Black-tailed Prairie Dog (Cynomys ludovicianus) Colonies Across Montana Using the 2005 National Agriculture Imagery Program (NAIP) Imagery Prepared for: Bureau of Land Management Miles City Field Office and Montana Department of Fish, Wildlife, and Parks Nongame Program Prepared by: Bryce A. Maxell, Scott Blum, and Karen Walker Montana Natural Heritage Program a cooperative program of the Montana State Library and the University of Montana The document should be cited as: Maxell, B.A., S. Blum, and K.V. Walker. 2010. Preliminary Report: Mapping Black-tailed Prairie Dog (Cynomys ludovicianus) colonies across Montana using the National Agriculture Imagery Program (NAIP) 2005 imagery. Report to the Miles City Field Office of the Bureau of Land Management and the Nongame Program of the Montana Department of Fish, Wildlife, and Parks. Helena, MT: Montana Natural Heritage Program. 27 pp. plus an appendix. EXECUTIVE SUMMARY Numerous animal and plant species are dependent on, or closely associated with, the burrowing and foraging activities of Black-tailed Prairie Dogs (Cynomys ludovicianus). In Montana, these species include the federally endangered Black-footed Ferret (Mustela nigripes) and numerous state Species of Concern such as the Ferruginous Hawk (Buteo regalis), Mountain Plover (Charadrius montanus), and Burrowing Owl (Athene cunicularia) (MPDWG 2002, MTNHP and MFWP 2010). To manage and protect Black-tailed Prairie Dogs and associated species, the Conservation Plan for Black-tailed and White-tailed Prairie Dogs in Montana calls for statewide prairie dog abundance and distribution standards, including complexes defined by a 7km distance to nearest neighbor rule that are greater than 5,000 acres (Category 1), between 1,000 and 5,000 acres (Category 2), and less than 1,000 acres (Category 3) (MPDWG 2002). The conservation plan also calls for inventorying and monitoring prairie dog distribution and status and identifying isolated colonies in need of special consideration for conservation or possible use in restoration of colonies depopulated by plague. Achievement of these and other goals of the conservation plan has been hindered by inaccuracies in the most recent statewide polygonal GIS coverage for prairie dog colonies and incomplete mapping of prairie dog colonies across their range in Montana due to the associated logistical difficulties and costs of acquiring and maintaining this information. In hopes of addressing these deficiencies and fulfilling the goals of the Conservation Plan for Black-tailed and White-tailed Prairie Dogs in Montana, we completed a statewide pilot mapping of Black-tailed Prairie Dog colonies using natural color and color infrared images from the National Agriculture Imagery Program’s (NAIP) 2005 imagery (USDA FSA 2010). We did this by coding a gridded network of one hectare (ha) cells as to whether they showed any evidence of vegetation or soil disturbance associated with the presence of recent prairie dog activity. We then buffered the resulting potential colony polygons to identify 1.5, 3.0, and 7.0 km complexes and summarized colony and complex numbers and acres by ownership (federal, state, tribal, private) and administrative unit (FWP region, BLM Field Office, Tribal Reservation, USFS District, and county). We generally felt that the 2005 NAIP imagery worked well for identifying areas with evidence of recent activity for Black-tailed Prairie Dogs. Alterations to vegetation and mounds associated with their burrows seemed to stand out well against the background vegetation and soils in many areas at map scales of between 1:5,000 and 1:30,000; scales around 1:10,000 often seemed to work best. However, White- tailed Prairie Dog (Cynomys leucurus) colonies were not easily detected on the 2005 NAIP imagery, potentially as a result of extirpation of colonies and lower densities and less obvious burrow structures relative to Black-tailed Prairie Dogs. In addition, badland areas were problematic to review in general because the barren soils in these regions are similar in appearance to areas where prairie dogs have removed vegetation and created spoil piles from burrow diggings; areas in southern Phillips County were particularly problematic to review on this front. The area between Terry and Melstone was also a problematic area to review because, while there were apparent burrow entrances, the typical vegetation alterations associated with prairie dog activity were not always evident. Areas with previous evidence for Black-tailed Prairie Dog activity were usually independently identified as having evidence for recent prairie dog activity on the 2005 NAIP imagery. Furthermore, it seems promising that potential colonies identified under this effort have size class and spatial distribution patterns that are similar to previously gathered information for Montana. However, it is important to note that estimates of acreage for areas with recent evidence of prairie dog activity are biased high by an unknown magnitude because other ground features, such as ant mounds and Richardson’s Ground Squirrel (Spermophilus richardsonii) burrows, have likely been misinterpreted as evidence of recent prairie dog activity. Similarly, only a portion of each individual grid cell had to show evidence of recent prairie dog activity to be coded as such. Ground truthing is needed to correct for these biases. 1 Our digitization effort identified 8,852 potential prairie dog colonies on the 2005 NAIP imagery that range in size from 2.5 acres to 2,945 acres; 2,598 (29%) of these had previous confirmation of prairie dog activity in the immediate area. Potential colonies were identified in 8 counties that do not have previously confirmed observations of prairie dogs. For complexes defined by the 1.5 km distance rule, our effort identified 2,474 complexes that contain 2.5 to 38,766 acres of potential colonies. Ten of these complexes contain greater than 5,000 acres of potential prairie dog colonies, 73 complexes contain between 1,000 and 5,000 acres, and 2,391 complexes contain less than 1,000 acres. For complexes defined by the 3.0 km distance rule, our effort identified 1,032 complexes that contain 2.5 to 109,883 acres of potential colonies. Sixteen of these complexes contain greater than 5,000 acres of potential prairie dog colonies, 42 complexes contain between 1,000 and 5,000 acres, and 974 complexes contain less than 1,000 acres. For complexes defined by the 7.0 km distance rule, our effort identified 289 complexes that contain 2.5 to 397,086 acres of potential colonies. Six of these complexes contain greater than 5,000 acres of potential prairie dog colonies, 16 complexes contain between 1,000 and 5,000 acres, and 267 complexes contain less than 1,000 acres. The colony boundaries defined by this mapping effort have not been verified with a ground truthing effort. Therefore, we suggest the following uses and limitations to this information: (1) The 2,598 colonies identified on the 2005 NAIP imagery that are corroborated with other evidence of recent prairie dog activity in the statewide Point Observation Database should be immediately used in environmental reviews and other colonies should be added to this coverage as they are verified. (2) Given the time delay involved, exact boundaries of colonies mapped from the 2005 NAIP imagery should be evaluated in the field for environmental reviews and other purposes for which an exact colony boundary is needed. (3) Whether they have been corroborated with other evidence of recent activity or not, all potential colonies mapped in this effort should be used in conjunction with predicted distribution models to examine conservation priorities. However, finalization of conservation priorities in statewide or regional management plans should not proceed without evaluation with ground truthing. (4) We strongly caution that acreage estimates resulting from this mapping effort should not be used in conservation plans or for other purposes until magnitudes of biases are identified with ground truthing and the appropriate corrections can be determined. We currently know that these estimates are likely to be biased high because other ground features such as ant mounds and Richardson’s Ground Squirrel burrows have likely been misinterpreted as evidence of recent prairie dog activity and because many of the one ha grid cells were not fully occupied with evidence of recent prairie dog activity. On the other hand, we also have reason to believe that the estimates are biased low because many colonies smaller than an acre in size were likely missed by this mapping effort. Finally, we strongly encourage ground truthing and would suggest the following approach: (1) Ground truth potential colonies for evidence of recent prairie dog activity, current occupancy, and the presence of other species associated with praire dog colonies whenever possible in the course of other fieldwork. A coverage of colonies and one ha grid cells loaded onto GPS units or PDAs can greatly assist with this. (2) Systematically ground truth all potential colonies identified outside of the current documented range of Black-tailed Prairie Dogs in order to expand knowledge of their known distribution, document isolated colonies, and identify ground features that are being misinterpreted as prairie dog burrows on the NAIP imagery. (3) Ground truth a spatially balanced sample of all identified potential colonies and evaluate them for evidence of recent prairie dog activity, current occupancy, and the presence of other species associated with praire dog colonies. (4) Intensively examine a spatially balanced sample of those colonies that are ground truthed by evaluating evidence of recent prairie dog activity and occupancy in each of the individual 1 ha grid cells that compose the potential colony as identified on the 2005 NAIP imagery. 2 ACKNOWLEDGEMENTS We thank Allison Begley, Nongame Biologist with the Montana Department of Fish, Wildlife, and Parks in Billings, Lauri Hanauska-Brown, Nongame and Threatened and Endangered Species Programs Section Manager at the Montana Department of Fish, Wildlife, and Parks, and Bobby Baker, Wildlife Biologist with the Miles City Field Office of the Bureau of Land Management, for recognizing the importance of this project and working to provide the funds to make it possible. We also thank Scott Story and Adam Messer in the Strategic Planning and Data Services Bureau of Montana Fish, Wildlife, and Parks for their advice on the methodology used to map Black-tailed Prairie Dog colonies using NAIP imagery. Finally, we thank members of the Montana Prairie Dog Working Group for providing constructive comments on the analyses presented in this report. This project was supported by agreements with the Montana Department of Fish, Wildlife, and Parks (FWP Purchase Order #090196) and the Bureau of Land Management (BLM AFA – ESA010009, Task order #16). 3 TABLE OF CONTENTS Introduction.................................................................................................................................. 6 Project Goals................................................................................................................................ 7 Methods....................................................................................................................................... 7 Digitization of Prairie Dog Colonies........................................................................................ 7 Analyses of Colonies and Complexes.......................................................................................8 Results..............................................................................................................................……… 9 Summaries of Potential Colonies and Complexes.................................................................... 9 Discussion / Recommendations................................................................................................... 11 Strengths and Weaknesses of Digitization from NAIP Imagery.............................................. 11 Suggested Uses and Limitations of NAIP Mapped Colonies and Complexes………………. 12 Suggestions for Ground Truthing……………………………………………………………. 12 Suggestions for Future Digitizing and Management of Statewide Database of Colony Boundaries…………………………………………………………………………………… 13 Literature Cited ........................................................................................................................... 14 Figures Figure 1. Point observations and 20 km x 20 km tiles coded as high density (red), low density (green), and no previously documented prairie dog activity (blue)……... 15 Figure 2. Predicted distribution model (blue is predicted low habitat suitability and read is predicted high habitat suitability) overlaid by 20 km x 20 km tiles coded as high density (red), low density (green), and no previously documented prairie dog activity (blue) ………............................................................................................. 15 Figure 3. Example of one hectare grid cell network overlying the 2005 NAIP imagery with areas coded for evidence of recent prairie dog activity (orange cells)……... 16 Figure 4. Example of one hectare grid cells coded as having evidence of recent prairie dog activity (cells with darker shading) and aerial flight intercept lines from 2008 FWP aerial surveys (red lines)……………………………………………... 17 Figure 5. Example of 20 km x 20 km tiles (yellow lines) with one hectare grid cells coded for evidence of recent prairie dog activity (lighter small grid cells), previous observations of Black-tailed Prairie Dogs in the statewide point observation database (yellow points) and flight intercepts of Black-tailed Prairie Dog colonies recorded during FWP’s 2008 aerial surveys (red lines)........................... 18 Figure 6. Potential prairie dog colonies digitized from the 2005 NAIP imagery (black) along with previously confirmed areas with Black-tailed Prairie Dog activity in the statewide point observation database (purple) and flight intercepts of Black- tailed Prairie Dog colonies recorded during FWP’s 2008 aerial surveys (red)………………………………………………………………………………. 19 4 Figure 7. Size class distribution of potential prairie dog colonies digitized from the 2005 NAIP imagery (black) (A), and total acres of potential prairie dog colonies within complexes defined by the 1.5 km (B), 3.0 km (C), and 7.0 km (D) distance rules……………………………………………………………………... 20 Figure 8. Dominant ownership of colonies; private (gray), federal (yellow), tribal (brown), state (blue)……………………………………………………………… 22 Figure 9. Ten complexes containing greater than 5,000 acres of potential prairie dog colonies as defined by the 1.5 km distance rule (A). Most of these are dominated by private land ownership (gray), but three are dominated by tribal ownership (brown), one is dominated by federal ownership (yellow), and state ownership (blue) is a component of most (B)……………………………………. 23 Figure 10. Sixteen complexes containing greater than 5,000 acres of potential prairie dog colonies as defined by the 3.0 km distance rule (A). Most of these are dominated by private land ownership (gray), but one is dominated by tribal ownership (brown), one is dominated by federal ownership (yellow), and state ownership (blue) is a component of most (B)……………………………………. 24 Figure 11. Six complexes containing greater than 5,000 acres of potential prairie dog colonies as defined by the 7.0 km distance rule (A). Most of these are dominated by private land ownership (gray), but one is dominated by tribal ownership (brown), one is dominated by federal ownership (yellow), and state ownership (blue) is a component of most (B)……………………………………. 25 Tables Table 1. Numbers of potential colonies and complexes in various acreage classes digitized from the 2005 NAIP imagery ……………………………………......... 26 Table 2. Numbers of potential colonies and total acres by FWP Region…………………. 26 Table 3. Numbers of potential colonies and total acres by BLM Field Office……………. 26 Table 4. Numbers of potential colonies and total acres by Tribal Reservation…………… 26 Table 5. Numbers of potential colonies and total acres by U.S. Forest Service District...... 26 Table 6. Numbers of potential colonies and total acres by County……………………...... 27 Appendix A Summary of Mapping of Potential Prairie Dog Colonies for each 20 km x 20 km Map Tile………………………………………………………………………... 28 5 INTRODUCTION Prior to widespread European settlement, Black-tailed Prairie Dogs (Cynomys ludovicianus) were widely distributed across the Great Plains of the United States with an overall estimated population of more than 5 billion animals (Hoogland 1996). In Montana, the species was abundant and widely distributed throughout grassland and shrub/grassland habitats east of the Continental Divide during the 1800s, but declined in abundance during the 20th century as a result of conversion of native rangelands to agricultural production, poisoning campaigns, sylvatic plague, urbanization, and recreational shooting (MPDWG 2002). The burrowing and foraging activities of this species have structured plant and animal communities across the Great Plains for the past 3 million years and numerous animal species are dependent on, or closely associated with, their colonies (Hoogland 1996, Roelle et al. 2006). In Montana, these species include the federally endangered Black-footed Ferret (Mustela nigripes) and numerous state Species of Concern such as the Ferruginous Hawk (Buteo regalis), Mountain Plover (Charadrius montanus), and Burrowing Owl (Athene cunicularia) (MPDWG 2002, MTNHP and MFWP 2010). To manage and protect Black-tailed Prairie Dogs and species associated with prairie dog colonies, the Conservation Plan for Black-tailed and White-tailed Prairie Dogs in Montana (MPDWG 2002) calls for statewide prairie dog abundance and distribution standards that: achieve a mix of prairie dog colonies and complexes (groups of colonies) capable of accommodating Black-footed Ferret recovery, supporting viable, well-distributed populations of other wildlife species associated with prairie dogs, and capable of sustaining a viable population of black-tailed prairie dogs distributed over 90% of the historic range of the species (page 15) The conservation plan has specific goals for conservation of prairie dogs and associated species according to three categories of prairie dog colony size using a 7 km rule for distance to nearest neighbor as follows. Category 1: A minimum of two black-tailed prairie dog complexes sufficient to maintain viable populations of black-footed ferrets. These should be at least 100 km apart, with each encompassing at least 5,000 acres of prairie dogs Category 2: A total of 36,000 acres occupied by black-tailed prairie dogs, composed of at least 20 complexes of at least 1,000 acres Category 3: Complexes less than 1,000 acres in size…plus scattered isolated colonies of any acreage (pages 15-16) The conservation plan also calls for inventorying and monitoring prairie dog distribution and status and identifying isolated colonies in need of special consideration for conservation or possible use in restoration of colonies depopulated by plague (MPDWG 2002). Achievement of these and other goals of the conservation plan has been hindered by incomplete mapping of prairie dog colonies across their range in Montana due to the associated logistical difficulties and costs of acquiring and maintaining this information. The most recent statewide polygonal GIS coverage for prairie dog colonies has a large number of inaccuracies in colony locations and boundaries and lacks mapping for large numbers of colonies on private lands. Mapping prairie dog colonies using National Agriculture Imagery Program (NAIP) imagery from the U.S. Department of Agriculture’s Farm Service Agency may be a cost effective means of acquiring and maintaining a statewide polygonal coverage that can be used to meet the goals of Montana’s Conservation Plan for Black-tailed and White-tailed Prairie Dogs (MPDWG 2002). NAIP imagery is high resolution digital ortho photography gathered on a 3 to 5 year cycle during the growing season and is freely 6 available to government agencies and the general public within a year of its acquisition. NAIP imagery contains four bands of information (red, green, blue, and near infrared) to yield both natural color and color infrared images, is acquired at a one-meter ground sample distance (i.e., one-meter per pixel), and has a horizontal accuracy of six meters (USDA FSA 2010). Four-band NAIP imagery was acquired statewide with one-meter resolution for the first time in 2005 and was made available in the spring of 2006. Statewide four-band imagery was acquired again in 2009 and this was made available in early March of 2010. Cursory reviews of 2005 NAIP imagery in 2006 and 2007 showed promise for identification of prairie dog colonies and the Montana Department of Fish, Wildlife, and Parks (FWP), subsequently funded a pilot project to investigate its systematic use for this purpose (Maxell 2009a). After successful completion of the pilot project, the Miles City Field Office of the Bureau of Land Management (BLM) then funded this effort to systematically map prairie dog colonies across the remainder of the state. PROJECT GOALS The goals of this effort were to: (1) map areas with evidence of recent prairie dog activity on the 2005 NAIP imagery across the possible range of Black-tailed Prairie Dogs in Montana; (2) examine the use of NAIP imagery in detection of recent evidence of activity of White-tailed Prairie Dogs (Cynomys leucurus); (3) identify complexes of colonies using the 7 km rule in the Conservation Plan for Black- tailed and White-tailed Prairie Dogs in Montana (MPDWG 2002), a 1.5 km rule proposed for use in complexes for recovery of Black-footed Ferrets (Biggins et al. 2006), and an intermediate 3 km rule that may be useful in identifying linkage areas between complexes; (4) identify acres and the size class distribution of colonies in each complex relative to the Category 1 (>5,000 acres), Category 2 (1,000- 5,000 acres), and Category 3 (<1,000 acres) criteria in the conservation plan (MPDWG 2002); (5) summarize numbers and acres of colonies and complexes by ownership (federal, state, tribal, private) and administrative unit (FWP region, BLM Field Office, Tribal Reservation, USFS District, and county); (6) begin to evaluate this mapping effort by identifying the number of polygons mapped with previous evidence of prairie dog activity from ground or flight surveys; and (7) develop a draft plan for ground truthing and managing the polygons that have been mapped. METHODS Digitization of Prairie Dog Colonies We used ArcMap 9.3 © ESRI software and file geodatabase for digitization and subsequent GIS and tabular analyses. In order to provide for faster processing times and more manageable datasets during the digitization effort, all areas east of the Continental Divide were broken up into 717 tiles each covering an area of 400 square kilometers (20 km on a side) using the Marine Geospatial Ecology Fishnet Tool. We then classified tiles as high density, low density, or no documented presence of prairie dog colonies based on the density of prairie dog colonies recorded in the statewide Point Observation Database (POD) and predicted distribution models created by the Montana Natural Heritage Program using the methodology outlined in Maxell; tiles that were heavily forested mountainous areas were excluded (2009b) (Figures 1 and 2). Within each of the 20 km x 20 km tiles, we documented areas with recent evidence of prairie dog activity using a network of 40,000 one ha grid cells (100 meters on each side) rather than trying to digitize colony boundaries in order to enhance repeatability of classifications across different colonies and observers. We chose this grid cell size because it allowed patterns associated with mounds and burrow entrances to be detected on the NAIP imagery within individual grid cells so that they could be coded somewhat independently of one another. Similarly, we felt this grid cell size would be an appropriate scale for on- 7 the-ground evaluations of grid cells. However, this approach does bias acreage estimates high because only a portion of each individual grid cell had to show evidence of recent prairie dog activity to be coded as such. Each grid cell was examined for evidence of recent prairie dog activity on high resolution true color and color infrared 2005 NAIP imagery at scales varying between 1:3,000 and 1:30,000 on a 19 inch computer screen. Smaller map scales were more useful in detecting vegetation contrasts resulting from prairie dog activity and larger map scales were more useful in detecting mounds associated with burrow entrances; scanning tiles at a scale of 1:10,000 often seemed to work best. At the beginning of our efforts we discussed coding of grid cells in detail as a group, but only one of us (SAB) coded all of the grid cells for purposes of consistency. Models identifying slopes less than 10 percent, digital topographic maps, black and white digital ortho quadrangle maps, and predicted distribution models of prairie dogs were also used at these scales to assist the digitizing process by locating flat areas and areas with vegetation contrasts indicative of recent evidence of prairie dog activity. However, 1 ha grid cells were only classified as having evidence of recent prairie dog activity if they contained evidence of mounds at the mouths of prairie dog burrows somewhere within the grid cell on the 2005 NAIP imagery (Figure 3). All initial examination and coding of grid cells as having “evidence for recent activity = 1” or having “no evidence for recent activity = 0” were made blind to any documentation of the known locations of prairie dog towns. However, after all grid cells on each 20 km x 20 km tile were initially coded, a second pass was made through each tile using prairie dog town locations documented in the statewide observation database and flight intercepts from FWP’s 2008 aerial surveys (Figures 4 and 5). Any additional colonies with obvious mounds at the mouths of burrows detected on the 2005 NAIP imagery on this second pass were then coded as “assisted evidence for recent activity = 2”. After the second pass through each 20 km x 20 km tile the following statistics were recorded: (1) total number of colonies digitized; (2) total number of unique colonies mapped in statewide point observation database; (3) total number of flight intercepts of colonies during FWP’s 2008 aerial surveys; (4) total number of colonies initially missed, but found with the assistance of observations in the statewide point observation database; (5) total number of colonies initially missed, but found with the assistance of FWP’s 2008 aerial surveys; (6) total number of previously undocumented colonies prior to this effort; (7) total number of observations in the statewide point observation database with no colony evident on the 2005 NAIP imagery; and (8) total number of flight intercepts during FWP’s 2008 aerial surveys with no colony evident on the 2005 NAIP imagery (Appendix A). Analyses of Colonies and Complexes After completing the digitizing effort, we merged grid cells in all 717 of the 20 km x 20 km tiles into a single vector coverage and dissolved adjacent individual grid cells into colonies for additional analyses; grid cells that only touched another grid cell on a corner were assigned a separate colony identification number. Colonies were buffered by 750, 1,500, and 3,500 meters using the Analysis/Proximity/Buffer tool to create the 1.5, 3.0, and 7.0 km complex boundaries. For each complex distance rule, the “Dissolve Type = All” option was used to dissolve overlapping buffers and create a single multipart polygon. We then exploded this multipart polygon into separate polygons with unique identification numbers for each complex using the Data Management/Features/Multipart to Singlepart tool. We used spatial joins to identify colonies with previous corroborating evidence of prairie dog activity and to assign colonies to the complexes they fell within under each of the 1.5, 3.0, and 7.0 km complex rules. We assigned colonies and complexes to various administrative boundaries (BLM Field Offices, FWP Regions, Tribal Reservations, U.S. Forest Service Districts, Counties) and land ownership classes (federal, state, tribal, and private) using the Analysis/Overlay/Identity tool. Finally, we summarized acres for each complex under the three complex rules for various administrative boundaries and land ownership types using group by queries in a file geodatabase. Figures and charts were created in ArcMap 9.3 © ESRI software and Microsoft Excel. 8