2550 N Diers Ave Suite L 203 W 2 nd St. second floor Grand Island, NE Grand Island, NE (308) (308) x33

Transcription

1 Rainwater Basin Joint Venture Rainwater Basin JV GIS Lab 2550 N Diers Ave Suite L 203 W 2 nd St. second floor Grand Island, NE Grand Island, NE (308) (308) x33 ASSESSING SPECIES PROBABILITY OF OCCURRENCE AND DISTRIBUTION FOR GREATER PRAIRIE-CHICKEN, SHARP-TAILED GROUSE AND LONG-BILLED CURLEW THROUGHOUT NEBRASKA November 2012 By the Rainwater Basin Joint Venture Andy Bishop 1, Roger Grosse 1, Ele Nugent 1, Christopher Jorgensen 1,2 1 Rainwater Basin Joint Venture, 2550 North Diers Ave, Grand Island, NE Nebraska Cooperative Fish and Wildlife Research Unit, 422 Hardin Hall, 3310 Holdrege Street, Lincoln, NE This work was funded in part by the Nebraska Game and Parks Commission through a Department of Energy grant to the Western Governors' Association [DOE Grant # DE-OE "Resource Assessment and Interconnection-Level Transmission Analysis and Planning (for the Western Interconnect)"] A special thanks to Mike Fritz, Joel Jorgensen, Jeff Lusk, Rick Schneider, Rachel Simpson, and Kristal Stoner for providing comments and support during the project. EXECUTIVE SUMMARY As conservation resources become increasingly limited, effective planning efforts are critical in order to prioritize species and their habitats in need of conservation. Species distribution models are one tool in the land manager s toolbox, which can facilitate an informative decision-making process as to where and how to reach management goals, given the surrounding landscape, and can be implemented in a decision support system. Here we describe a method for developing a spatial model of probability of occurrence for three species found in the Great Plains. Presence-absence data for greater prairie-chicken, sharp-tailed grouse and long-billed curlew were collected over survey routes throughout Nebraska. We utilized geographic information systems to quantify the amount of habitat surrounding each survey location at multiple spatial scales. Habitat relationships best explaining species probability of occurrence were identified using binomially distributed generalized linear models. A model selection criterion (AICc) was used to identify which model best explained species-habitat relationships and at what spatial scale habitat variables had the most influence on the species of interest. Fitted parameter estimates describing species-habitat relationships were entered into a geographic information system raster calculator to produce a spatial model of species occurrence. Predictive spatial models were validated using a subset of the dataset withheld from the analysis. Predictive spatial species occurrence models for greater prairie-chicken and long-billed curlew performed well. The sharp-tailed grouse model, however, had very poor performance, 1

2 repeatedly failing to successfully predict species occurrence. In the future, our predictive species distribution models can be refined to develop decision support tools to facilitate informative land management and energy development decisions. INTRODUCTION Effective conservation planning and development is critical for the preservation and management of species populations. Yet, developing such plans often requires an understanding of where species occur and how individuals make habitat decisions. Once it is clear what habitat characteristics are important for a species, as well as the spatial scale to which individuals respond, we can begin to address and predict species occurrence throughout a region. Furthermore, decision support tools can be developed to aid decision makers and managers during the conservation planning stages of the Strategic Habitat Conservation framework. Here we describe the methods used to develop models of species occurrence for three native grassland-dependent species in Nebraska, the greater prairie-chicken (GRPC), sharp-tailed grouse (STGR), and long-billed curlew (LBCU), using recent presence-absence data collected during roadside surveys. Habitat relationships were identified by using landscape-level variables derived from the 2010 Nebraska Landcover developed by the Rainwater Basin Joint Venture. DATA PREPARATION Regional Landscapes High land-use variability throughout the state may influence model performance; therefore we divided the state into five regions based on Major Land Resource Areas (MLRA) from the Natural Resource Conservation Service (NRCS) (Figure 1). By producing spatial models within defined land-use regions, we were able to add habitat variables to the model that are unique to a region and prevented correlated variables from driving the model fit. The southwest playas and Republican Breaks/Loess Canyon regions are considered unique landscape regions but were combined into a single region due to lack of sampling distribution across the southwest playa region (Figure 1). 2

3 Figure 1. Regional landscape boundaries based on MLRA soils 3

4 Landcover Indices Five primary landscape level indices were created from the Nebraska Landcover (Table 1). An individual habitat index is composed of a single habitat or vegetation category, which was extracted from the Landcover. The primary index categories were: developed, cropland, woodland, grassland, and wetland. Alternative indices were created for certain regions. Conservation Reserve Program (CRP) grassland was separated from the general grassland category in the East and Loess Canyons regions where this program is widely used. Wetlands could be further subdivided. For most of the state, the term wet meadow refers to temporarily submerged or swampy grasslands associated with riverine wetlands and sloughs. These occur in low-lying land with poor drainage. However, in the Sandhills a wet meadow translates to a type of grassland that occurs in inter-dunal valleys where the water table lies near the surface and seasonally floods. Sandhill wet-meadow soils are permeable. Thus wet meadow can be lumped in with all grasses, separated out as a single index, or lumped together with wetlands. Table 1. Landcover inputs for WGA statewide habitat indices Abbreviation Habitat Category Landcover Definition crop Crops Cropland (38), Alfalfa (201), Corn (202), Fallow (203), Sorghum (206), Soybeans (207), Sunflowers (208), Wheat (209), Other (211) crp CRP Grass CRP (39), Grasses (31) dvlp Developed Other Roads (41), Rural (42), Four Lane (44), Urban/Suburban (46) dvlp0rd Developed w/o roads Rural (42), Urban/Suburban (46) gr0cr0wm Grassland w/o CRP and wet meadow Mixed Grass (71), Sandhills Grassland (73), Shortgrass (75), Tallgrass (77), Sand Sage (87) grass Grassland CRP(39), Grasses (31), Mixed Grass (71), Sandhills Grassland (73), Shortgrass (75), Tallgrass (77), Sand Sage (87), Wet Meadow (247) grs0crp Grassland w/o CRP Grasses (31), Mixed Grass (71), Sandhills Grassland (73), Shortgrass (75), Tallgrass (77), Sand Sage (87), Wet Meadow (247) 4

5 Table 1. Continued Abbreviation Habitat Category Landcover Definition ne_east Easting UTM Easting for Nebraska (in meters) wood Woodland CRP - Upland Trees (32), CRP - Riparian Trees (33), Eastern Red Cedar (59) or Juniper (66), Upland Woodland (61), Ponderosa Pine (63), Few trees w grassy understory (69), Many trees w little grass (60), Riparian canopy (241), Exotic Riparian Shrubland (242), Native Riparian Shrubland (243) wtld0wmd Wetlands w/o wet meadow Playas (12), Farmed Playa (121) or Grassland/Buffered Playa (122), Sandhills Wetlands (13), Rainwater Basin (14); Farmed Rainwater Basin (141), Early Successional RWB (142), Other Wetland (15), Emergent Marsh (152), Saline Marsh (153), CRP Wetlands (34), CRP Playa/Non-floodplain Wetland (35), Canals (48), Freshwater Lake/Sandhill Lake (101), Lagoon (102), Sand Pit/Irrigation reuse pit (103), Reservoir (104), Stock Pond (106), Emergent Marsh (152), Saline Marsh (153), River Channel (244), Un-vegetated Sandbar (245), Warm-water Slough (246) wtlnd Wetlands Playas (12), Farmed Playa (121) or Grassland/Buffered Playa (122), Sandhills Wetlands (13), Rainwater Basin (14); Farmed Rainwater Basin (141), Early Successional RWB (142), Other Wetland (15), Emergent Marsh (152), Saline Marsh (153), CRP Wetlands (34), CRP Playa/Non-floodplain Wetland (35), Canals (48), Freshwater Lake/Sandhill Lake (101), Lagoon (102), Sand Pit/Irrigation reuse pit (103), Reservoir (104), Stock Pond (106), Emergent Marsh (152), Saline Marsh (153), River Channel (244), Un-vegetated Sandbar (245), Warm-water Slough (246), Wet Meadow (247), Floodplain Marsh (248) 5

6 METHODS Survey Period for Grouse Available years of survey data varied between the geographical regions. The Rainwater Basin Joint Venture Landcover is generally current to ground condition. For this reason and to keep the models current, no route information prior to 2006 was used to produce lek models. Grouse Data Collection Survey Data were provided by Nebraska Game and Parks Commission (NGPC) and affiliated contacts. They were presented in a number of formats including: 1) Paper maps with routes, sites, and lek locations hand sketched 2) Survey forms with or without sketch of route and handwritten GPS coordinates of either stop locations or of lek locations only 3) Spreadsheet listing leks as geographical coordinates 4) ESRI map documents (.mxd) as graphics 5) ESRI shapefiles Leks were sampled using a fairly consistent route survey protocol (Walker, Meduna, and Wheeler, personal comm.). However, given the number of surveyors dispersed across the state, some variation within that framework occurred in terms of how and what data were recorded. Therefore, raw lek samples often required post-processing to create a uniform record scheme across all routes. For example, route surveyors may isolate individual leks at a single stop while others may record only the presence of leks detected nearby. Other individuals only recorded lek locations and did not provide lek absences at stop locations. We attempted to follow a simplified sampling strategy of a single occurrence or absence at each one-mile stop, which reduced artificial correlation since the landscape around each survey point is summarized at a specified scale. Additional lek samples were provided by USFWS Refuges, NFS, and other sources which ultimately were not used for modeling due to inconsistent sampling procedures. We converted all datasets to feature classes with the following information: Species, Presence/Absence listed as 1 = present, 0 = absent for GRPC and STGR, StopID if available, Route Name, Source of Data, Year(s) survey took place. We utilized a geographic information system (GIS) to convert all spatial data to a consistent geographical projection suited for Nebraska using the North American datum 1983 UTM zone 14 in ArcGIS 10.0 (ESRI, Redlands, California). Once route and lek locations were compiled into a single database, we extracted habitat index values to each survey point. Landcover indices were summarized as a percentage of the landscape at 800m, 1600m and 5000m scale radii. We randomly selected and withheld a quarter of the data for model validation, creating a training and testing dataset. Long-billed Curlew Data Collection Survey data were provided by Nebraska Game and Parks Commission (NGPC), and were completed by Cory Gregory, from Iowa State University (Gregory 2011, Gregory et al. 2012). Surveys were conducted via roadside survey and each survey route consisted of roughly 40 stops 6

7 spaced 800m apart. Survey routes were chosen randomly throughout the curlew s range in western Nebraska. Curlews were searched for by sight and sound for 5 minutes from outside the vehicle at each stop location. Surveys were conducted only on passable public roads excluding interstates and urban areas. Surveys were conducted in April prior to nest incubation, when displaying curlews are easier to detect. All surveys took place in 2008 and We used the original route data points that included GPS coordinates of all stop locations. All datasets were converted to feature classes with the following information: Species, Presence/Absence listed as 1 = present, 0 = absent, Number of curlews spotted, Route Name, Source of Data, Year(s) survey took place. All data was converted to a consistent map projection using North American Datum UTM zone 14 in ArcGIS Landcover index values were extracted to each survey point. A quarter of the data were randomly selected from the dataset and withheld for model validation, creating a training and testing dataset. Statistical Analysis We developed an a priori set of models using various habitat indices measured at various spatial extents based on the biology of the species and the region. We ran a binomial generalized linear model (GLM) for each model in the model set and used an information-theoretic criterion to determine model selection (Zuur et al. 2007). The model with the lowest AICc value was considered the best fit model (Burnham & Anderson 2002). The logistic regression equation from the top fitted model was applied to the index layers to generate spatial outputs for a final decision on which model most accurately represents reality. We validated each model using an independent testing dataset obtained by withholding a quarter of the data prior to fitting the model. Raster values from each of the spatial models were extracted to testing datasets. ROC (receiver operating characteristic) plots were created for each spatial model by assessing the ratio of true positives to false positives for all possible thresholds. AUC (area under the curve) values were calculated from the ROC plots, indicating the chance that a randomly chosen plot with an observed value of present will have a predicted probability higher than that of a randomly chosen plot with an observed value of absent. A high AUC value near 1.0 indicates good overall model performance. 7

8 RESULTS Greater Prairie-chicken: Eastern Region Table 2. Models, AICc scores and cumulative weights for Greater Prairie-chicken in eastern Nebraska. GPC ~ grass wood dvlp0rd GPC ~ crop wood dvlp0rd GPC ~ crop wood dvlp GPC ~ grass wood dvlp GPC ~ grass wood dvlp wtld0wmd GPC ~ grass wood dvlp GPC ~ grass wood GPC ~ crop wood dvlp GPC ~ grass wood dvlp wtld0wmd GPC ~ grass dvlp GPC ~ grass wood dvlp0rd GPC ~ crop wood dvlp0rd GPC ~ gr0cr0wm wood dvlp GPC ~ grass wood GPC ~ grass dvlp0rd GPC ~ grs0crp wood dvlp GPC ~ grass dvlp0rd GPC ~ grass dvlp GPC ~ crop wood GPC ~ gr0cr0wm wood dvlp0rd GPC ~ gr0cr0wm wood dvlp0rd

9 GPC ~ grs0crp wood dvlp0rd Table 2. Continued GPC ~ grs0crp wood dvlp0rd GPC ~ gr0cr0wm wood dvlp GPC ~ grs0crp wood dvlp GPC ~ gr0cr0wm wood GPC ~ grass50 + wood50 + dvlp50 + wtld0wmd GPC ~ grass50 + wood50 + dvlp GPC ~ grass50 + wood50 + dvlp0rd GPC ~ grs0crp wood GPC ~ grass50 + dvlp GPC ~ grass50 + dvlp0rd GPC ~ crop wood GPC ~ grass50 + wood GPC ~ gr0cr0wm wood GPC ~ grs0crp wood GPC ~ grs0crp50 + wood50 + dvlp0rd GPC ~ grs0crp50 + wood50 + dvlp GPC ~ gr0cr0wm50 + wood50 + dvlp0rd GPC ~ crop50 + wood50 + dvlp0rd GPC ~ crop50 + wood50 + dvlp GPC ~ gr0cr0wm50 + wood50 + dvlp GPC ~ gr0cr0wm50 + wood GPC ~ grs0crp50 + wood GPC ~ crp wood dvlp GPC ~ crop50 + wood GPC ~ crp dvlp GPC ~ crp50 + wood50 + dvlp GPC ~ crp50 + dvlp GPC ~ crp wood dvlp

10 GPC ~ crp dvlp0rd Table 2. Continued GPC ~ crp50 + dvlp0rd GPC ~ crp wood GPC ~ crp dvlp GPC ~ crp dvlp0rd GPC ~ crp50 + wood GPC ~ crp wood GPC ~ Table 3. Coefficient estimates for top-ranked greater prairiechicken model based on AICc scores. Coefficient estimates were used to create a spatial model of greater prairie-chicken probability of occurrence in eastern Nebraska. Coefficients Estimate Std. Error z value Pr(> z ) (Intercept) E-07 *** grass1600m E-12 *** wood1600m *** dvlp0rd1600m * 10

11 Figure 2. Greater prairie chicken probability of occurrence model for eastern Nebraska. 11

12 Sensitivity (true positives) GRPC Prevalence AUC: 0.93 spatial.model Specificity (false positives) Figure 3. A ROC plot and AUC calculation for the greater prairie-chicken spatial model in eastern Nebraska. We used 92 independent testing points to validate our spatial model. An AUC value of 0.93 indicated good performance for the probability of occurrence model for greater prairie-chickens in the eastern region of Nebraska. 12

13 Greater Prairie-chicken: Loess Canyons Table 4. Models, AICc scores and cumulative weights for Greater Prairie-chicken in the Loess Canyons, Nebraska. GPC ~ grass800m + wood800m + dvlp800m GPC ~ crop800m + wood800m + dvlp800m GPC ~ grass800m + wood800m + dvlp800m + wtld0wmd800m GPC ~ grs0crp800m + wood800m + dvlp800m GPC ~ gr0cr0wm800m + wood800m + dvlp800m GPC ~ grass800m + dvlp800m GPC ~ crp800m + wood800m + dvlp800m GPC ~ crp800m + dvlp800m GPC ~ grass800m + wood800m + dvlp0rd800m GPC ~ gr0cr0wm800m + wood800m + dvlp0rd800m GPC ~ grs0crp800m + wood800m + dvlp0rd800m GPC ~ grass800m + wood800m GPC ~ grass1600m + wood1600m + dvlp1600m GPC ~ grass1600m + wood1600m + dvlp1600m + wtld0wmd1600m GPC ~ gr0cr0wm800m + wood800m GPC ~ grs0crp1600m + wood1600m + dvlp1600m GPC ~ crop1600m + wood1600m + dvlp1600m GPC ~ gr0cr0wm1600m + wood1600m + dvlp1600m GPC ~ grs0crp800m + wood800m GPC ~ crop800m + wood800m + dvlp0rd800m GPC ~ grass800m + dvlp0rd800m GPC ~ grass1600m + wood1600m + dvlp0rd1600m GPC ~ grass1600m + wood1600m

14 Table 4. Continued GPC ~ gr0cr0wm1600m + wood1600m GPC ~ grs0crp1600m + wood1600m + dvlp0rd1600m GPC ~ gr0cr0wm1600m + wood1600m + dvlp0rd1600m GPC ~ grs0crp1600m + wood1600m GPC ~ crop1600m + wood1600m + dvlp0rd1600m GPC ~ crop800m + wood800m GPC ~ crop1600m + wood1600m GPC ~ grass1600m + dvlp1600m GPC ~ grass1600m + dvlp0rd1600m GPC ~ crp1600m + wood1600m + dvlp1600m GPC ~ grass5000m + wood5000m + dvlp5000m GPC ~ grass5000m + wood5000m + dvlp5000m + wtld0wmd5000m GPC ~ crop5000m + wood5000m + dvlp5000m GPC ~ grass5000m + wood5000m GPC ~ grass5000m + wood5000m + dvlp0rd5000m GPC ~ grs0crp5000m + wood5000m + dvlp5000m GPC ~ crop5000m + wood5000m GPC ~ gr0cr0wm5000m + wood5000m + dvlp5000m GPC ~ crop5000m + wood5000m + dvlp0rd5000m GPC ~ grs0crp5000m + wood5000m GPC ~ grs0crp5000m + wood5000m + dvlp0rd5000m GPC ~ gr0cr0wm5000m + wood5000m GPC ~ gr0cr0wm5000m + wood5000m + dvlp0rd5000m GPC ~ crp5000m + wood5000m + dvlp5000m GPC ~ crp1600m + wood1600m GPC ~ crp1600m + dvlp1600m GPC ~ grass5000m + dvlp5000m GPC ~ crp5000m + wood5000m

15 Table 4. Continued GPC ~ grass5000m + dvlp0rd5000m GPC ~ crp800m + wood800m GPC ~ crp800m + dvlp0rd800m GPC ~ crp5000m + dvlp5000m GPC ~ crp1600m + dvlp0rd1600m GPC ~ crp5000m + dvlp0rd5000m GPC ~ Table 5. Coefficient estimates for top-ranked greater prairiechicken model in the Loess Canyon region based on AICc scores. Coefficient estimates were used to create a spatial model of greater prairie-chicken probability of occurrence in the Loess Canyons, Nebraska. Coefficients Estimate Std. Error z value Pr(> z ) (Intercept) grass800m E-07 *** wood800m E-03 ** dvlp800m E-16 *** 15

16 Figure 4. Greater Prairie chicken probability of occurrence model for the Loess Canyons region, Nebraska. 16

17 Sensitivity (true positives) GRPC Prevalence AUC: 0.92 spatial.model Specificity (false positives) Figure 5. ROC plot and AUC calculation for the greater prairie-chicken Loess Canyons spatial model. We used 204 independent testing points to validate the spatial model. An AUC value of 0.92 indicated good performance for the probability of occurrence model for greater prairiechickens in the Loess Canyons region of Nebraska. 17

18 Greater Prairie-Chicken: Loess Hills Table 6. Binomial GLM models, AICc scores and cumulative weights based on 2010 data for greater prairie-chicken in the Loess Hills region, Nebraska. GPC ~ gr0cr0wm1600m + wood1600m + dvlp0rd1600m GPC ~ crop1600m + wood1600m + dvlp0rd1600m GPC ~ grass1600m + wood1600m + dvlp0rd1600m GPC ~ grs0crp1600m + wood1600m + dvlp0rd1600m GPC ~ gr0cr0wm1600m + wood1600m GPC ~ grass1600m + wood1600m GPC ~ grs0crp1600m + wood1600m GPC ~ crop1600m + wood1600m + dvlp1600m GPC ~ gr0cr0wm1600m + wood1600m + dvlp1600m GPC ~ crop1600m + wood1600m GPC ~ grass1600m + wood1600m + dvlp1600m GPC ~ grs0crp1600m + wood1600m + dvlp1600m GPC ~ grass1600m + wood1600m + dvlp1600m + wtld0wmd1600m GPC ~ grs0crp5000m + wood5000m GPC ~ grass5000m + wood5000m GPC ~ crp1600m + wood1600m + dvlp1600m GPC ~ gr0cr0wm5000m + wood5000m GPC ~ grs0crp5000m + wood5000m + dvlp0rd5000m GPC ~ grass5000m + wood5000m + dvlp0rd5000m GPC ~ grs0crp5000m + wood5000m + dvlp5000m GPC ~ grass5000m + wood5000m + dvlp5000m GPC ~ crp5000m + wood5000m GPC ~ crp5000m + wood5000m + dvlp5000m GPC ~ gr0cr0wm5000m + wood5000m + dvlp0rd5000m

19 Table 6. Continued GPC ~ gr0cr0wm5000m + wood5000m + dvlp5000m GPC ~ crop5000m + wood5000m GPC ~ grass5000m + wood5000m + dvlp5000m + wtld0wmd5000m GPC ~ crop5000m + wood5000m + dvlp0rd5000m GPC ~ crop5000m + wood5000m + dvlp5000m GPC ~ crp1600m + wood1600m GPC ~ grass800m + wood800m GPC ~ grs0crp800m + wood800m GPC ~ crop800m + wood800m + dvlp0rd800m GPC ~ grass800m + wood800m + dvlp0rd800m GPC ~ grs0crp800m + wood800m + dvlp0rd800m GPC ~ gr0cr0wm800m + wood800m GPC ~ crop800m + wood800m GPC ~ crop800m + wood800m + dvlp800m GPC ~ grass800m + wood800m + dvlp800m GPC ~ grs0crp800m + wood800m + dvlp800m GPC ~ gr0cr0wm800m + wood800m + dvlp0rd800m GPC ~ gr0cr0wm800m + wood800m + dvlp800m GPC ~ grass800m + wood800m + dvlp800m + wtld0wmd800m GPC ~ crp800m + wood800m + dvlp800m GPC ~ crp800m + wood800m GPC ~ grass800m + dvlp0rd800m GPC ~ grass5000m + dvlp5000m GPC ~ grass800m + dvlp800m GPC ~ grass5000m + dvlp0rd5000m GPC ~ grass1600m + dvlp0rd1600m GPC ~ grass1600m + dvlp1600m GPC ~ crp800m + dvlp0rd800m

20 Table 6. Continued GPC ~ crp800m + dvlp800m GPC ~ crp1600m + dvlp1600m GPC ~ GPC ~ crp1600m + dvlp0rd1600m GPC ~ crp5000m + dvlp0rd5000m GPC ~ crp5000m + dvlp5000m Table 7. Coefficient estimates for top-ranked greater prairiechicken model based on the 2010 data collected in the Loess Hills region based on AICc scores. Coefficient estimates were used to create a spatial model of greater prairie-chicken probability of occurrence in the Loess Hills, Nebraska. Coefficients Estimate Std. Error Z Value Pr(> z ) (Intercept) gr0cr0wm1600m * wood1600m E-06 *** dvlp0rd1600m

21 Figure 6. Greater prairie chicken probability of occurrence model for the Loess Hills region, Nebraska. Model is based on 2010 data. 21

22 Sensitivity (true positives) GRPC Prevalence AUC: 0.77 spatial.model Specificity (false positives) Figure 7. ROC plot and AUC calculation for the greater prairie-chicken Loess Hills spatial model. We used 53 independent testing points from the 2010 data to validate the spatial model. An AUC value of 0.77 indicated the performance for the probability of occurrence model for greater prairie-chickens in the Loess Hills region of Nebraska. 22

23 Greater Prairie-chicken: Sandhills Table 8. Binomial GLM models, AICc scores and cumulative weights for greater prairie-chicken in the Sandhills region, Nebraska. GPC ~ gr0cr0wm5000m + wood5000m + dvlp0rd5000m + ne_east + ne_east GPC ~ gr0cr0wm1600m + wood1600m + ne_east + ne_east GPC ~ gr0cr0wm5000m + wood5000m + ne_east + ne_east GPC ~ crop5000m + wood5000m + ne_east + ne_east GPC ~ gr0cr0wm1600m + wood1600m + dvlp1600m + ne_east + ne_east GPC ~ gr0cr0wm1600m + wood1600m + dvlp0rd1600m + ne_east + ne_east GPC ~ crop5000m + wood5000m + dvlp0rd5000m + ne_east + ne_east GPC ~ crop5000m + wood5000m + dvlp5000m + ne_east + ne_east GPC ~ crp5000m + wood5000m + ne_east + ne_east GPC ~ grass1600m + wood1600m + ne_east + ne_east GPC ~ gr0cr0wm800m + wood800m + dvlp800m + ne_east + ne_east GPC ~ grs0crp800m + wood800m + dvlp800m + ne_east + ne_east GPC ~ grs0crp1600m + wood1600m + ne_east + ne_east GPC ~ crp5000m + wood5000m + dvlp5000m + ne_east + ne_east GPC ~ grass1600m + wood1600m + dvlp1600m + ne_east + ne_east GPC ~ grass1600m + wood1600m + dvlp0rd1600m + ne_east + ne_east

24 Table 8. Continued GPC ~ grs0crp1600m + wood1600m + dvlp1600m + ne_east + ne_east GPC ~ gr0cr0wm800m + wood800m + dvlp0rd800m + ne_east + ne_east GPC ~ grs0crp1600m + wood1600m + dvlp0rd1600m + ne_east + ne_east GPC ~ grass800m + wood800m + dvlp0rd800m + ne_east + ne_east GPC ~ grass800m + dvlp0rd800m + ne_east + ne_east GPC ~ crop1600m + wood1600m + dvlp1600m + ne_east + ne_east GPC ~ grass1600m + dvlp1600m + ne_east + ne_east GPC ~ grass1600m + dvlp0rd1600m + ne_east + ne_east GPC ~ crop800m + wood800m + dvlp800m + ne_east + ne_east GPC ~ grass1600m + wood1600m + dvlp0rd1600m + wtld0wmd1600m + ne_east + ne_east GPC ~ crop1600m + wood1600m + dvlp0rd1600m + ne_east + ne_east GPC ~ grass800m + wood800m + dvlp0rd800m + wtld0wmd800m + ne_east + ne_east GPC ~ grass800m + wood800m + ne_east + ne_east GPC ~ crop800m + wood800m + dvlp0rd800m + ne_east + ne_east GPC ~ crop1600m + wood1600m + ne_east + ne_east GPC ~ crp800m + wood800m + dvlp800m + ne_east + ne_east GPC ~ gr0cr0wm800m + wood800m + ne_east + ne_east GPC ~ grass5000m + dvlp0rd5000m + ne_east + ne_east GPC ~ grass5000m + dvlp0rd5000m + wtld0wmd5000m + ne_east + ne_east

25 Table 8. Continued GPC ~ grs0crp800m + wood800m + ne_east + ne_east GPC ~ crp1600m + wood1600m + dvlp1600m + ne_east + ne_east GPC ~ crop800m + wood800m + ne_east + ne_east GPC ~ crp1600m + wood1600m + ne_east + ne_east GPC ~ crp800m + dvlp800m + ne_east + ne_east GPC ~ crp800m + dvlp0rd800m + ne_east + ne_east GPC ~ crp800m + wood800m + ne_east + ne_east GPC ~ crp1600m + dvlp1600m + ne_east + ne_east GPC ~ crp1600m + dvlp0rd1600m + ne_east + ne_east GPC ~ crp5000m + dvlp5000m + ne_east + ne_east GPC ~ crp5000m + dvlp0rd5000m + ne_east + ne_east GPC ~ grass5000m + dvlp0rd5000m + wtld0wmd5000m GPC ~ gr0cr0wm5000m + wood5000m + dvlp0rd5000m GPC ~ gr0cr0wm1600m + wood1600m GPC ~ gr0cr0wm1600m + wood1600m + dvlp0rd1600m GPC ~ gr0cr0wm1600m + wood1600m + dvlp1600m GPC ~ grass1600m + wood1600m + dvlp0rd1600m + wtld0wmd1600m GPC ~ gr0cr0wm800m + wood800m + dvlp800m GPC ~ gr0cr0wm800m + wood800m GPC ~ gr0cr0wm5000m + wood5000m GPC ~ grass800m + wood800m + dvlp0rd800m + wtld0wmd800m GPC ~ gr0cr0wm800m + wood800m + dvlp0rd800m GPC ~ grs0crp800m + wood800m + dvlp800m GPC ~ grass800m + dvlp0rd800m GPC ~ grass1600m + dvlp0rd1600m GPC ~ grass800m + wood800m

26 Table 8. Continued GPC ~ grass1600m + dvlp1600m GPC ~ grass1600m + wood1600m GPC ~ grass1600m + wood1600m + dvlp0rd1600m GPC ~ grass800m + wood800m + dvlp0rd800m GPC ~ grass1600m + wood1600m + dvlp1600m GPC ~ grs0crp800m + wood800m GPC ~ grs0crp1600m + wood1600m GPC ~ grass5000m + dvlp0rd5000m GPC ~ grs0crp1600m + wood1600m + dvlp1600m GPC ~ grs0crp1600m + wood1600m + dvlp0rd1600m GPC ~ crop800m + wood800m + dvlp800m GPC ~ crp800m + wood800m + dvlp800m GPC ~ crop1600m + wood1600m + dvlp1600m GPC ~ crop1600m + wood1600m GPC ~ crop1600m + wood1600m + dvlp0rd1600m GPC ~ crp800m + dvlp800m GPC ~ crop800m + wood800m + dvlp0rd800m GPC ~ crp1600m + wood1600m + dvlp1600m GPC ~ crop5000m + wood5000m + dvlp0rd5000m GPC ~ crop800m + wood800m GPC ~ crop5000m + wood5000m GPC ~ crp5000m + wood5000m GPC ~ crop5000m + wood5000m + dvlp5000m GPC ~ crp5000m + wood5000m + dvlp5000m GPC ~ crp1600m + wood1600m GPC ~ crp1600m + dvlp1600m GPC ~ crp800m + wood800m GPC ~ crp800m + dvlp0rd800m GPC ~ crp5000m + dvlp5000m

27 Table 8. Continued GPC ~ GPC ~ crp1600m + dvlp0rd1600m GPC ~ crp5000m + dvlp0rd5000m Table 9. Coefficient estimates for top-ranked greater prairie-chicken model based on the 2010 data collected in the Sandhills region based on AICc scores. Coefficient estimates were used to create a spatial model of greater prairie-chicken probability of occurrence in the Sandhills, Nebraska. Coefficients Estimate Std. Error z value Pr(> z ) (Intercept) *** gr0cr0wm5000m ** wood5000m ** dvlp0rd5000m * easting E-07 *** easting E-05 *** 27

28 Figure 8. Greater prairie chicken probability of occurrence model for the Sandhills region, Nebraska. 28

29 Sensitivity (true positives) GRPC Prevalence AUC: 0.83 spatial.model Specificity (false positives) Figure 9. A ROC plot and AUC calculation for the greater prairie-chicken spatial model in the Sandhills region. We used 79 independent testing points to validate the spatial model. An AUC value of 0.83 indicated the performance for the probability of occurrence model for greater prairie-chickens in the Sandhills region of Nebraska. 29

30 Sharp-tailed Grouse: Sandhills Table 10. Binomial GLM models, AICc scores and cumulative weights for sharp-tailed grouse in the Sandhills region, Nebraska. STGR ~ grass5000m + dvlp5000m STGR ~ crp5000m + dvlp5000m STGR ~ grs0crp5000m + wood5000m + dvlp5000m STGR ~ grass5000m + wood5000m + dvlp5000m STGR ~ gr0cr0wm5000m + wood5000m + dvlp5000m STGR ~ crp5000m + wood5000m + dvlp5000m STGR ~ crop5000m + wood5000m + dvlp5000m STGR ~ grass800m + dvlp800m STGR ~ grass800m + wood800m + dvlp800m STGR ~ grs0crp800m + wood800m + dvlp800m STGR ~ grs0crp1600m + wood1600m + dvlp1600m STGR ~ grass1600m + dvlp1600m STGR ~ grass1600m + wood1600m + dvlp1600m STGR ~ gr0cr0wm800m + wood800m + dvlp800m STGR ~ gr0cr0wm1600m + wood1600m + dvlp1600m STGR ~ crop1600m + wood1600m STGR ~ STGR ~ grs0crp1600m + wood1600m STGR ~ grass1600m + wood1600m STGR ~ gr0cr0wm1600m + wood1600m STGR ~ crp1600m + wood1600m STGR ~ crp800m + dvlp800m STGR ~ crop1600m + wood1600m + dvlp1600m STGR ~ crop800m + wood800m + dvlp800m

31 Table 10. Continued STGR ~ crop5000m + wood5000m STGR ~ crp800m + wood800m + dvlp800m STGR ~ crp1600m + wood1600m + dvlp1600m STGR ~ crop800m + wood800m STGR ~ crp1600m + dvlp1600m STGR ~ grass800m + wood800m STGR ~ grs0crp800m + wood800m STGR ~ gr0cr0wm800m + wood800m STGR ~ crp800m + wood800m STGR ~ crp5000m + wood5000m STGR ~ grs0crp5000m + wood5000m STGR ~ grass5000m + wood5000m STGR ~ gr0cr0wm5000m + wood5000m Table 11. Coefficient estimates for top-ranked sharp-tailed grouse model in the Sandhills region based on AICc scores. Coefficient estimates were used to create a spatial model of sharp-tailed grouse probability of occurrence in the Sandhills, Nebraska. Coefficients Estimate Std. Error z value Pr(> z ) (Intercept) E-16 *** grass5000m dvlp5000m ** 31

32 Figure 10. Sharp-tailed grouse probability of occurrence model for the Sandhills region, Nebraska. 32

33 Sensitivity (true positives) STGR Prevalence AUC: 0.53 spatial.model Specificity (false positives) Figure 11. A ROC plot and AUC calculation for the sharp-tailed grouse spatial model in the Sandhills region. We used 79 independent testing points to validate the spatial model. An AUC value of 0.53 indicated poor performance for the probability of occurrence model for sharp-tailed grouse in the Sandhills region of Nebraska. 33

34 Long-billed Curlew: Sandhills and Panhandle Table 12. Binomial GLM models, AICc scores and cumulative weights for long-billed curlews in the Sandhills and Panhandle region, Nebraska. LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grass800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grass800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crop800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp0rd800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ grass800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + crop800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grass5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grs0wmd800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grass1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ dvlp800m + crop800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grass5000m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grass5000m + rugged5000m + rugged5000m_2 + easting + easting

35 Table 12. Continued LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crop5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grass5000m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grass1600m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grs0wmd5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grass1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grs0wmd5000m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grass800m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crp800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp0rd800m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + crop5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ grass5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ dvlp5000m + crop5000m + rugged5000m + rugged5000m_2 + easting + easting

36 Table 12. Continued LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grass1600m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp0rd5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crop5000m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grass800m + easting + easting LBCU ~ grass5000m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp0rd1600m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grs0wmd1600m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crop800m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crop1600m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + crop5000m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crop1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp0rd1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ dvlp5000m + crop5000m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crp5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grs0wmd1600m + rugged1600m + rugged1600m_2 + easting + easting

37 Table 12. Continued LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + rugged5000m + rugged5000m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grs0wmd800m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + crop800m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + crop1600m + easting + easting LBCU ~ crp800m + rugged800m + rugged800m_2 + easting + easting LBCU ~ grass1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + crop1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ grass1600m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp0rd5000m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crp800m + easting + easting LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ grass800m + easting + easting LBCU ~ crp5000m + rugged5000m + rugged5000m_2 + easting + easting

38 Table 12. Continued LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + easting + easting LBCU ~ dvlp1600m + crop1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ dvlp1600m + crop1600m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crp1600m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crp5000m + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + easting + easting LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crp1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ dvlp800m + crop800m + easting + easting LBCU ~ crp5000m + easting + easting LBCU ~ crp1600m + easting + easting LBCU ~ crp800m + easting + easting LBCU ~ crp1600m + rugged1600m + rugged1600m_2 + easting + easting LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grs0wmd5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grass5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grs0wmd5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grass5000m + rugged5000m + rugged5000m_

39 Table 12. Continued LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp0rd5000m + rugged5000m + rugged5000m_ LBCU ~ crop5000m + wtld0wmd5000m + wtld0wmd5000m_2 + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grass800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + crop5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grass800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crp5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp0rd800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grs0wmd800m + rugged800m + rugged800m_ LBCU ~ rugged5000m + rugged5000m_2 + wtld0wmd5000m + wtld0wmd5000m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grs0wmd5000m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grass1600m + rugged1600m + rugged1600m_ LBCU ~ rugged800m + rugged800m_2 + wtld0wmd800m + wtld0wmd800m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grs0wmd5000m

40 Table 12. Continued LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grs0wmd800m + rugged800m + rugged800m_ LBCU ~ crop800m + wtld0wmd800m + wtld0wmd800m_2 + rugged800m + rugged800m_ LBCU ~ grass5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grs0wmd1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grass1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crp800m + rugged800m + rugged800m_ LBCU ~ grass800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grs0wmd1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + crop800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp0rd1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + grass5000m LBCU ~ dvlp5000m + crop5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + grass5000m LBCU ~ rugged1600m + rugged1600m_2 + wtld0wmd1600m + wtld0wmd1600m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grs0wmd1600m LBCU ~ crop1600m + wtld0wmd1600m + wtld0wmd1600m_2 + rugged1600m + rugged1600m_

41 Table 12. Continued LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + grass1600m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grs0wmd1600m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + rugged1600m + rugged1600m_ LBCU ~ dvlp800m + crop800m + rugged800m + rugged800m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crp1600m + rugged1600m + rugged1600m_ LBCU ~ crp800m + rugged800m + rugged800m_ LBCU ~ crp5000m + rugged5000m + rugged5000m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + grass1600m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + crop1600m + rugged1600m + rugged1600m_ LBCU ~ grass1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grass800m LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crop5000m LBCU ~ grass5000m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grass800m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp0rd1600m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp0rd800m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + grs0wmd800m LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m + crop5000m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crop1600m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + grs0wmd800m LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp0rd5000m

42 Table 12. Continued LBCU ~ wtld0wmd1600m + wtld0wmd1600m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crop800m LBCU ~ dvlp1600m + crop1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd800m + wtld0wmd800m_ LBCU ~ dvlp5000m + crop5000m LBCU ~ crp1600m + rugged1600m + rugged1600m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m + crop1600m LBCU ~ wtld0wmd5000m + wtld0wmd5000m_ LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + crp1600m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m + crop800m LBCU ~ grass1600m LBCU ~ wtld0wmd1600m + wtld0wmd1600m_2 + dvlp1600m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + crp800m LBCU ~ wtld0wmd800m + wtld0wmd800m_2 + dvlp800m LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + dvlp5000m LBCU ~ wtld0wmd5000m + wtld0wmd5000m_2 + crp5000m LBCU ~ grass800m LBCU ~ dvlp1600m + crop1600m LBCU ~ LBCU ~ dvlp800m + crop800m LBCU ~ crp1600m LBCU ~ crp800m LBCU ~ crp5000m

43 Table 13. Coefficient estimates for top-ranked long-billed curlew model in the Sandhills and Panhandle region based on AICc scores. Coefficient estimates were used to create a spatial model of long-billed curlew probability of occurrence in the Sandhills and Panhandle region, Nebraska. Coefficients Estimate Std. Error z value Pr(> z ) (Intercept) E-16 *** wtld0wmd800m ** wtld0wmd800m * grass800m ** rugged800m *** rugged E-05 *** easting E-06 *** easting E-07 *** 43

44 Figure 12. Long-billed curlew probability of occurrence model for the Sandhills and Panhandle region, Nebraska. 44