AN ABSTRACT OF THE THESIS OF. Michael W. Byrne for the degree of Master of Science in Wildlife Science presented on February 7, 2002.

Transcription

1 AN ABSTRACT OF THE THESIS OF Michael W. Byrne for the degree of Master of Science in Wildlife Science presented on February 7, Title: Habitat Use by Female Greater Sage Grouse in Relation to Fire at Hart Mountain National Antelope Refuge, Oregon. Abstract approved: 0 J'_' 4. The United States Fish and Wildlife Service currently uses fire as a management tool to improve Greater Sage Grouse (Centrocercus urophasianus Bonaparte) nesting and brood-rearing habitat at Hart Mountain National Antelope Refuge (HMNAR) in S.E. Oregon. Previous studies at HMNAR revealed use of burned areas by sage grouse throughout the breeding season, but quantitative evaluation of breeding-season habitat use in relation to burned areas was unavailable. Further, the effects of prescribed fire and wildfire on sage-grouse habitat use and habitat selection are poorly understood and controversial. I evaluated a comprehensive database of all prescribed fire and wildfire data ( ) and sage-grouse breeding-season habitat data ( ) as a means to evaluate and quantify the temporal and spatial effects of habitat use by female sagegrouse relative to burned areas. Nesting, brood-rearing, and broodless females generally avoided 89%, 89%, and 77%, respectively, of available burned habitats; unburned habitats were

2 typically used. All 5 nests in burns < 20 years old were unsuccessful, but nesting success in >20 year-old burns (29%, n = 6/21) and unburned areas (28%, n = 49/177) was similar. When burned habitats were used, they were typically mid- to late-successional mountain big sagebrush burns. Age of burn, and the resultant return of the shrub (i.e., sagebrush) component, was most commonly associated with use of burned areas by nesting and brood-rearing females. Fire in areas with higher biotic potential (in terms of soil productivity/structure, floral diversity, and precipitation), such as mountain big sagebrush [A. t. Nutt. var. vaseyana (Rybd.) Beetle] cover types where the shrub, herbaceous, and invertebrate components have returned, may provide favorable habitat for limited nesting, brood-rearing, and broodless female use. However, in areas with generally lower biotic potential, such as low sagebrush (A. arbuscula Nutt.) and Wyoming big sagebrush (A. tridentata Nutt. var. wyomingensis Beetle & Young) cover types, fire seemingly provided no apparent value in terms of nesting, brood-rearing, and broodless female use. Although future research may further elucidate burned habitat use trends, managers should be cautious in the use of prescribed fire in sage grouse habitats.

3 Copyright by Michael W. Byrne February 7, 2002 All Rights Reserved

4 Habitat Use by Female Greater Sage Grouse in Relation to Fire at Hart Mountain National Antelope Refuge, Oregon by Michael W. Byrne A THESIS submitted to Oregon State University in partial fulfillment of the requirements for the degree of Master of Science Presented February 7, 2002 Commencement June 2002

5 Master of Science thesis of Michael W. Byme presented on February 7, 2002 APPROVED Ma' r Professor, representing Wildlife Science Head of Department of Fisheries and Wildlife I understand that my thesis will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my thesis to any reader upon request. Michael W. Byrne, Author

6 ACKNOWLEDGEMENTS In pursuit of any goal, support and cooperation takes many forms and makes the endeavor much more productive, but most importantly, more enjoyable. I have had the fortunate opportunity to work with several agencies and individuals I would like to thank. This project would not have been possible without the longterm cooperation, financial and logistical support of the Sheldon and Hart Mountain Refuge Complex of the U.S. Fish and Wildlife Service and the financial support from the Oregon Department of Fish and Wildlife. It has been a pleasure working closely with a group I admire and respect. In particular, I would like to thank Mike Nunn, Steve Clay, Mike Dunbar, Marty Bray, Chris Farinetti, and Andy Goheen for your advice, support, and foresight; Marty Bray for your field mentorship, countless hours of help, advice, wit, and friendship; Jen Ballard for your help, advice, and good weather; Mark Giordano for your help and poker change. Gina Barr and Alicia Winters for your knack for making things run smoothly; and Jeff Jaeger and Doug Leehmann for somehow fixing everything I destroyed without a complaint, at least to me. I would like to extend an enormous amount of gratitude to Dr. John Crawford, my major professor and mentor. Your tutelage, guidance, advice, and encouragement facilitated a graduate experience that instilled an attitude of productivity, objectivity, and cooperation that will persist throughout my career. I consider myself very lucky to have worked with such a phenomenal researcher, ecologist, and individual, and I wish you the best of health.

7 I would also like to thank my field assistants: Sean Kiffe, Nolan Lewis, Larry Butcher, Marshall Lipps, and Matt Wilson. Your dedication, enthusiasm, and motivation was beyond compare and I am honored to have had the opportunity to work with you. I sincerely appreciate all of your efforts in helping me achieve my goals. Your constant entertainment value was an added benefit. I am very grateful to those at OSU that have helped me along the way. A special thanks to my committee members, Dr. Eric Forsman and Dr. Doug Johnson. I appreciate your advice, guidance, and support, and the opportunity to benefit from your wealth of knowledge and experience. Thank you Lavon Mauer and Ellen Holsberry for keeping me from being too much trouble, or at least making me feel that way; Charlotte Vickers for your guidance; Dr. Rick Miller for sharing your knowledge of sagebrush-steppe ecology, Dr. Dave Pyke for logistical support, and special thanks to Norman Swanson for your help with the Hart Mountain fire history. I sincerely appreciate the help of Dawn Davis, Mike Pope, Mike Gregg and Jay Welch for your excellent advice and assistance both in the field and in the office, and to Mike Gregg, Marty Drut, Anita DeLong, Kreg Coggins, and Norman Swanson for data collection at Hart Mountain over the years. Finally, very special thanks to my family. I appreciate your well-concealed suffering through my long-winded and overly detailed banter about sage grouse and sagebrush ecology, often devoid of any real entertainment value. You have always been a constant from which I draw strength to overcome challenges and achieve goals. This work would never have been possible without your love and support, which I consider myself incredibly fortunate to have.

8 TABLE OF CONTENTS Page INTRODUCTION 1 METHODS AND MATERIALS 7 Study Area 7 Trapping and Monitoring 11 Habitat Sampling 13 GIS Database 13 Data Analysis 16 RESULTS 20 Nesting 20 Brood-rearing 24 Broodless Females 29 DISCUSSION 32 LITERATURE CITED 38 APPENDICES 46

9 LIST OF FIGURES Figure Page 1. Location of Hart Mountain National Antelope Refuge, Lake County, OR Cover type classifications at Hart Mountain National Antelope Refuge, Lake County, Oregon Mean age of burned areas available to radio-marked nesting female sage grouse as a function of use response at Hart Mountain National Antelope Refuge, Lake County, Oregon, S - Selected, P - Proportional Use, L - Less than Available, A- Avoided. 4. Mean age of burned areas available to radio-marked brooding female sage grouse as a function of use response at Hart Mountain National Antelope Refuge, Lake County, Oregon, S - Selected, P - Proportional Use, L - Less than Available, A- Avoided. 5. Mean age of burned areas available to radio-marked broodless female sage grouse as a function of use response at Hart Mountain National Antelope Refuge, Lake County, Oregon, S - Selected, P - Proportional Use, L - Less than Available, A- Avoided

10 LIST OF TABLES Table 1 Potential for misclassification of locations in burned and unburned habitats; radio-marked female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Number of nest, brood, and broodless locations and number of radio-marked female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Page Sage grouse habitat characteristics (% cover) in a 1954 mountain big sagebrush burn (Degarmo Canyon Wildfire) and unburned mountain big sagebrush at Hart Mountain National Antelope Refuge, Lake County, Oregon, Use of burned areas by radio-marked nesting female sage grouse during the breeding season at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age of burns in each use category is also indicated. 5. Use of burned cover types by nesting female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age and number of bums in each category is also indicated. 6. Habitat components of sage grouse nest sites in recently burned areas (< 20 years-old), old burns (> 20 years-old), and unburned areas at Hart Mountain National Antelope Refuge, Lake County, Oregon, Use of burned areas by radio-marked brooding female sage grouse during the breeding season at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age of burns in each use category is also indicated

11 LIST OF TABLES (Continued) Table 8. Use of burned cover types by brooding female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age and number of burns in each category is also indicated. 9. Habitat components of sage grouse brood sites in recently burned areas (< 20 years-old), old burned areas (> 20 yearsold), and unburned areas at Hart Mountain National Antelope Refuge, Lake County, Oregon, Use of burned areas by radio-marked broodless female sage grouse during the breeding season at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age of burns in each use category is also indicated. 11. Use of burned cover types by broodless female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age and number of burns in each category is also indicated. Page

12 LIST OF APPENDICES Appendix Page A. Fire events at Hart Mountain National Antelope Refuge, Lake County, OR, B. Number and type of fire events and types at HMNAR, C. Percentage of burned area within utilization distribution estimates generated for pooled individual radio-marked female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Habitat selection of low sagebrush cover types by radiomarked nesting female sage grouse with Bailey simultaneous confidence intervals at HMNAR, E. Habitat selection of mountain big sagebrush cover types by radio-marked nesting female sage grouse with Bailey simultaneous confidence intervals at HMNAR, F. Habitat selection of other cover types by radio-marked nesting female sage grouse with Bailey simultaneous confidence intervals at HMNAR, G. Habitat selection of Wyoming big sagebrush cover types by radio-marked nesting female sage grouse with Bailey simultaneous confidence intervals at HMNAR, H. Habitat selection of burned cover types by radio-marked nesting female sage grouse with Bailey simultaneous confidence intervals at HMNAR, Habitat selection of low sagebrush cover types by radiomarked brooding female sage grouse with Bailey simultaneous confidence intervals at HMNAR,

13 LIST OF APPENDICES (Continued) Appendix Page J. Habitat selection of mountain big sagebrush cover types by radio-marked brooding female sage grouse with Bailey simultaneous confidence intervals at HMNAR, K. Habitat selection of other cover types by radio-marked brooding female sage grouse with Bailey simultaneous confidence intervals at HMNAR, L. Habitat selection of Wyoming big sagebrush cover types by radio-marked brooding female sage grouse with Bailey simultaneous confidence intervals at HMNAR, Habitat selection of burned cover types by radio-marked brooding female sage grouse with Bailey simultaneous confidence intervals at HMNAR, N. Habitat selection of low sagebrush cover types by radiomarked broodless female sage grouse with Bailey simultaneous confidence intervals at HMNAR, Habitat selection of mountain big sagebrush cover types by radio-marked broodless female sage grouse with Bailey simultaneous confidence intervals at HMNAR, P. Habitat selection of other cover types by radio-marked broodless female sage grouse with Bailey simultaneous confidence intervals at HMNAR, Q. Habitat selection of Wyoming big sagebrush cover types by radio-marked broodless female sage grouse with Bailey simultaneous confidence intervals at HMNAR, R. Habitat selection of burned cover types by radio-marked broodless female sage grouse with Bailey simultaneous confidence intervals at HMNAR,

14 I dedicate this thesis to my mother, Beverly C. Byrne, and to my grandmother, Elizabeth H. Kunkel

15 Habitat Use by Female Greater Sage Grouse in Relation to Fire at Hart Mountain National Antelope Refuge, Oregon. INTRODUCTION The Greater sage grouse (Centrocercus urophasianus Bonaparte), a once abundant sagebrush-steppe obligate, has decreased in abundance throughout its range in the western United States. Crawford and Lutz 1985 estimated the number of sage grouse in Oregon declined 60% since These changes resulted primarily from the loss of habitat through sagebrush (Artemesia L. spp.) control programs, agricultural activities, and altered fire regimes (Dalke et al. 1963, Braun et al. 1977, Johnsgard 1983, Crawford and Lutz 1985, Drut et al. 1994a). Crawford and Lutz (1985) found that estimates of sage grouse productivity in Oregon declined and survival has varied substantially since the 1950's. Because of sage grouse population declines, the species has been petitioned for Threatened or Endangered status under the Endangered Species Act of 1973 in Colorado, Utah, and Washington. Crawford (1982) determined that the population decline was unrelated to harvest, and Crawford and Lutz (1985) suggested that the population decline was a consequence of poor nesting and brood-rearing habitat. Consequently, management of sage grouse habitats will be critical for the recovery of sage grouse populations.

16 2 During the early 19th century, increased settlement of the Great Basin by American pioneers brought about the introduction of exotic herbivores and plants, cultivation of extensive areas, and suppression of natural wildfires (Miller 1989, Miller and Eddleman 2000). These changes, in conjunction with climatic fluctuations, resulted in alterations in plant composition and community structure (Willis et al. 1993, Gruell 1995, Miller and Eddleman 2000). Sage grouse declines coincided with increased utilization of rangelands for grazing and agricultural activities from 1900 to 1915 (Patterson 1952). As a result, many rangeland habitats became dominated by a dense sagebrush overstory with corresponding decreases in grass and forb components (Vale 1974, Miller 1989, Pyle and Crawford 1996). Although high densities of sagebrush (>25%) may improve wintering habitat for sage grouse by increasing winter food, spring and summer habitats have deteriorated by a reduction in the herbaceous component that provides nesting cover and key food forbs (Sime 1991). Gruell (1995) indicated that the exclusion of fire in sagebrush habitats has affected plant succession and composition, resulting in shrub-dominated stands at the expense of the herbaceous component. Furthermore, altered fire regimes have resulted in the expansion of western juniper (Juniper occidentalis Hook.) woodlands, also depleting the shrub and herbaceous understory (Miller and Rose 1995). A fundamental concept in sage grouse management is to provide a balance of sagebrush, grasses, and forbs necessary for successful fulfillment of all lifehistory needs throughout the year (Connelly et al. 2000b). Prescribed fire has

17 3 been suggested as a means to rehabilitate shrub-dominated sagebrush habitats (Frandsen 1985, Bunting et al. 1987). The importance of the herbaceous component during the prenesting, nesting, and brood-rearing periods has been well documented (Barnett and Crawford 1994, Drut et al 1994a, Drut et al 1994b, Gregg et al. 1994, DeLong et al. 1995). In addition, prescribed fire may improve habitat quality for sage grouse (Klebenow 1985, Bunting et al. 1987) and create successionally variable habitats (Laycock 1991). Klebenow (1972) and Gates (1983) indicated that prescribed fire patterns that resulted in a mosaic would, ideally, create spatial diversity and habitat juxtapositions that could enhance sage grouse habitat use. The use of prescribed fire as a management tool for sage grouse habitat is controversial (Dalke et al. 1963, Braun et al. 1977). Specific concerns relate to removal of sagebrush to manage for sagebrush obligate species, and the unknown reestablishment rates of sagebrush following fire. In presettlement times, fire return intervals were typically years in low sagebrush (A. arbuscula Nutt.) communities, years in Wyoming big sagebrush (A. tridentata Nutt. var. wyomingensis Beetle & Young) communities, and years in mountain big sagebrush [A. t. Nutt. var. vaseyana (Rybd.) Beetle] communities (Wright and Bailey 1982, Miller and Rose 1999). The amount of time for sagebrush to return to a burned area can be quite substantial, site specific, and variable (Watts and Wambolt 1996, Miller and Eddleman 2000, Nelle et al 2000). A broad or over-

18 4 exuberant use of fire may negatively affect sagebrush ecosystems (Fischer et al. 1996, Nelle et al 2000) Prescribed fire has the potential to increase grasses and forbs in sagebrush habitats (Pyle and Crawford 1996), but Miller and Eddleman (2000) found that the outcome of fire is related to sagebrush species/subspecies, site potential, site condition, plant groups, and bum pattern and size. Prescribed fire in mountain big sagebrush stands in Wyoming produced twice the amount of perennial forbs in burned areas than in unburned areas for 3 years post-burn in a relatively mesic environment (Cook et al. 1994). Martin (1990) observed an increase in forbs after fire in a mesic Idaho site, and Pyle and Crawford (1996) found that prescribed fire in mountain big sagebrush-bitterbrush [Purshia tridentata (Pursh.) DC.] stands increased several key sage grouse forbs and did not adversely affect the abundance of key invertebrates used by sage grouse chicks. Winter (1984) found a greater abundance of insects in burned big sagebrush stands. In contrast, fire in mountain big sagebrush stands in Idaho had a negative impact on sage grouse nesting and brood-rearing habitats because of the long period of time necessary for a substantial sagebrush component to return (Nelle et al. 2000). Fischer et al. (1996) found that fire in Wyoming big sagebrush stands did not increase key forbs and decreased beetle abundance. Prescribed fire in a Wyoming big sagebrush stand in Montana resulted in less herbaceous growth than other forms of sagebrush control, which included spraying and plowing, but no differences were observed among the 3 treatments after 30 years (Watts and Wambolt 1996)

19 5 Several studies have addressed the relationship between fire and sage grouse habitat use with differing conclusions. Gates (1983) reported a greater use of burned areas than unburned areas by sage grouse, whereas Benson et al. (1991), Sime (1991) and Fischer et al. (1996) did not. Prescribed fire did not influence seasonal movement patterns of female sage grouse in southeastern Idaho (Fischer et al. 1997), but Connelly et al. (2000a) observed that fire negatively affected male sage grouse lek attendance. Benson et al. (1991) observed that sage grouse use of burned areas was proximate to unburned stands of sagebrush, and recommended that prescribed fire be designed to burn in a mosaic and be conducted on small areas. Openings in the sagebrush canopy may increase forbs and improve brood habitat (Sime 1991), but could be a detriment to winter and nesting habitat (Gates 1983, Nelle et al. 2000). Fire has become an increasingly popular tool to manage sage grouse habitats with a degraded herbaceous component to reduce shrub cover and promote herbaceous growth. However, information that relates fire histories to proportional use of habitats by sage grouse during the breeding season is unavailable. Although some researchers have investigated the short-term effects of fire on sage grouse habitat use, none have used long-term breeding season data in conjunction with detailed long-term fire histories to determine temporal or spatial effects of fire on sage grouse habitat use and selection. Based on preliminary observations at HMNAR, I hypothesized that female sage grouse used burned areas in greater proportion than their availability (i.e., selected)

20 6 during the reproductive period. The goal of my study was to provide a better understanding of the long- and short-term influence of fire on female sage grouse habitat use and selection by nesting, brood-rearing, and broodless females. My objectives were to: 1) measure proportional use of burned areas by female sage grouse as a function of burned area availability and, 2) determine potential factors, such as habitat components and burned area characteristics, which may affect habitat use and selection.

21 7 METHODS AND MATERIALS STUDY AREA Hart Mountain National Antelope Refuge (HMNAR) consists of approximately 114,375 ha of sagebrush-steppe habitat 70 km northeast of Lakeview, Lake County, Oregon (Fig. 1). Elevation ranged from 1500 in to 2450 in. The topography was characterized by flat sagebrush plains, interrupted by rolling hills, ridges, and draws (Gregg et al. 1994). Since 1940, mean annual precipitation was 29 cm, annual temperature averaged 6 C, and temperature ranged from -22 C in winter to 36 C in summer (Gregg 1992). HMNAR cover types were defined by Soil Conservation criteria and cover type boundaries were delineated by interpretation of color-infrared photographs (Gregg 1992). I subdivided cover types on the refuge into 5 types: low sagebrush, mountain big sagebrush, Wyoming big sagebrush, burned areas, and "other" cover types (Fig. 2). "Other" included quaking aspen (Populus tremuloides Michx.), mountain mahogany (Cercocarpus ledifolius Nutt.), western juniper, basin big sagebrush (A. t. Nutt. var. tridentata Nutt.), silver sagebrush (A. cana Pursh.), and meadows (e.g., Carex L. spp., Juncus L. spp). Low sagebrush covered approximately 38% of the refuge and was most common on ridges and in areas with rocky soils. Graminoid species frequently associated with low sagebrush stands were bluegrass (Poa L. spp.), bluebunch wheatgrass [Pseudoroegneria spicata

22 S Area of Detail N Kilometers Figure 1. Location of Hart Mountain National Antelope Refuge, Lake County, Oregon.

23 N Wyoming Big sagebrush Low Sagebrush Mountain Big sagebrush Other Cover Types Bunted Areas Kilometers Figure 2. Cover type classifications at Hart Mountain National Antelope Refuge, Lake County, Oregon.

24 10 (Pursh.) A. Love], Idaho fescue (Festuca idahoensis Elmer), and squirreltail [Elymus elymoides (Raf.) Swezey]. Wyoming big sagebrush covered approximately 33% of the refuge and was most common in xeric areas with less productive soils. Commonly associated graminoid species in Wyoming big sagebrush stands were bluegrass, squirreltail, and cheatgrass (Bromus tectorum L.). Mountain big sagebrush covered approximately 13% of the refuge and was most common at higher elevations in mesic areas with deeper and more productive soils. Mountain big sagebrush areas with a significant component of bitterbrush and snowberry (Symphoricarpos Duham. spp.) were common in highelevation areas. Graminoid species frequently associated with mountain big sagebrush stands were Idaho fescue, rough fescue (F. campestris Rydb.), and needlegrass (Achnatherum Piper spp. and Stipa L. spp.). Other cover types covered approximately 16% of the refuge. Livestock grazing was eliminated from HMNAR in December Before 1991, grazing averaged approximately 12,000 animal unit months (AVMs), allocated from mid-april to mid-december in a rest rotation, deferred grazing system (Gregg 1992). Livestock grazing was the primary management tool with some prescribed fire and herbicide use during this time. When livestock were removed in 1990, prescribed fire became the principal land management practice at the refuge (USDI 1994). I defined areas as burned if they had visible signs of fire or a known history of being burned. Since 1947, 70 fires burned approximately 14% of the refuge. Twenty-nine percent of

25 11 mountain big sagebrush, 14% of low sagebrush, 12% of other cover types, and 5% of Wyoming big sagebrush communities at HMNAR are known to have burned since 1947 (Fig. 2, Appendix A). Prescribed fire and escaped prescribed fires composed 73% of total fire events (Appendix B). Seventy-one percent of fires occurred from and of these, 78% were prescribed fires. TRAPPING AND MONITORING Female sage grouse were trapped between 15 March and 15 April annually from 1989 to 2000, except Habitat use data for this study were compiled from research conducted at HMNAR by M. Gregg , A. Delong 1992, M. Drut 1993, K. Coggins , N. Swanson 1997, and M. Byrne Grouse were captured with spotlighting techniques (Giesen et al. 1982, Wakkinen et al. 1992). Sex and age of captured birds were determined from wing molt and plumage characteristics (Crunden 1963, Beck et al. 1975, Ottomeier and Crawford 1996). Females were fitted with numbered aluminum leg bands and 20-g necklace-mounted radiotransmitters with a battery life of approximately 350 days [Advanced Telemetry Systems (ATS), Isanti, MN]. Each radio-marked female was monitored for 1 breeding season only to maintain independence of samples (e.g., nest sites, brooding areas) among years.

26 12 Grouse were visually located 2-4 times weekly from the ground with a hand-held Yagi antenna and a portable ATS receiver from 1 April through 31 July. Approximately equal effort was expended to gather location data for all radio-marked females and visual locations of the same individuals were not gathered on consecutive days. Location and cover-type assessment data were gathered in a manner that minimized disturbance to radio-marked grouse. Emphasis was placed on avoiding flushing radio-marked individuals or influencing their movements. Visual locations during the breeding-season were estimated with a Global Positioning System (GPS) unit or by visual reference to terrain features on a 1:24,000 topographic map. Visual location data were entered in a GIS database. When monitoring indicated a female had initiated a nest, the nest was visually located, marked on a map, and monitored remotely. A nest was classified as successful if > 1 egg hatched, as determined by detachment of the eggshell membrane (Gregg et al. 1994). Females with broods were monitored to determine habitat use and brood success from hatching date through July. Broods were considered successful if > I chick was recruited into the population by 1 August. Females with unsuccessful nests were monitored to determine habitat use.

27 13 HABITAT SAMPLING Habitat characteristics at nest and brood locations were sampled with 2 perpendicular transects 10-m in length centered at the nest, brood, or randomly selected site. Shrub canopy cover was measured with the line-intercept method with a gap criterion of 10cm (Canfield 1941). I classified height of shrubs 3 categories: short (< 40 cm), medium (40-80 cm), and tall ( 80 cm). Species composition and forb and grass cover were measured in 5 plots (2-x 5-dm) placed equidistantly along each transect (Daubenmire 1959). Tallest droop height of graminoids, excluding flower stalks, was measured and classified into 2 classes: short (< 20 cm) and tall (> 20 cm). Plants were identified to species according to available vegetative and phenological characteristics; otherwise, vegetation was identified to genus. All plant nomenclature follows Natural Resource Conservation Service authority (USDA 2002). GIS DATABASE I used Erdas Imagine 8.4 (Erdas, Inc.), ArcView 3.2, and ArcInfo 8 (Environmental Systems Research Institute, Inc.) to construct a GIS database that included fire histories ( ), cover types, and locations of radio-marked female sage grouse ( ). Fire history layers were created with on-screen

28 14 digitizing from aerial photography and fire-history maps. Polygons developed for old burns in which the shrub component had returned were typically more difficult to map accurately and had correspondingly more error potential than fire polygons created from aerial photography. Ground estimation was used to delineate fire patterns and boundaries of older fires. Based on comparison of fire polygons generated with ground estimation with photographic records available for older burns, accuracy of plotted boundaries of older fires ranged from 0.5 m to 100 in. In burned areas where the shrub component had not returned, fire boundaries generated from orthorectified aerial photos were delineated within a m error. Locations that were not gathered with GPS units and differentially corrected also had higher error rates because of the reduced accuracy inherent in less sophisticated GPS systems or location coordinates obtained from UTM grids placed on 7.5-minute topographic quads. A buffer analysis of all locations included in the final analysis was used to address the potential classification error of sage grouse locations as burned or unburned. Error buffers were created for all fire polygons and grouse locations. The width/radius of each buffer was the maximum possible error, which was determined to be 100m for fire polygon boundaries and grouse locations. Grouse locations with buffers that overlapped polygon buffers were defined as potential misclassifications of locations in burned or unburned habitats. Based on this worst-case analysis, the mean potential for misclassification of burned vs. unburned locations was 9% (Table 1).

29 15 Table 1. Potential for misclassification of locations in burned and unburned habitats: radio-marked female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, % Potential Misclassification Year Nesting Brood-rearing Broodless Mean Mean Prescribed fires and wildfires in sagebrush typically burn more than one cover type. Therefore, I generated 50 random points within each burned area, determined the cover type classification of each point, and used the cover type of the majority (i.e., > 50%) of random points in the polygon as the classification of the burn. The same method was used to determine mean elevation of burned areas. GIS layers of sage grouse locations during the breeding were generated from visual locations of radio-marked females. Breeding-season locations were categorized into 3 classes: nesting, brood-rearing, and broodless females. Use and availability layers were generated for all location categories.

30 16 DATA ANALYSIS Because of small sample size for individual birds, locations were pooled across individuals for each year-specific category of nesting, brooding, and broodless female locations (e.g., 1989 nests, 1989 broods, 1989 broodless, 1990 nests, 1990 broods, 1990 broodless, etc.) to compare habitat use with habitat availability. Use was defined as observed presence in a specific cover type. I analyzed habitat use for 203 nest locations from 203 individuals, 503 brood locations from 58 individuals, and 694 broodless female locations from 158 individuals (Table 2).

31 17 Table 2. Number of nest, brood, and broodless locations of radio-marked female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Nest Brood Broodless Year Ind. Loc. Ind. Loc. Ind. Loc Total I used 2 methods to estimate the amount of each cover type that was available to sage grouse during each year-specific portion of the breeding season. In cases where the combined sample from all radio-marked grouse was >30, I used a 95% fixed kernel estimator, with least square cross validation (LSCV). If the combined sample of locations from all grouse was <30, I used a 100% Minimum Convex Polygon utilization estimator (Mohr 1947, Seaman and Powell 1996). I defined cover types that were available to sage grouse as the area within the utilization polygon boundary. Utilization distributions were generated with the Animal Movement extension in ArcView 3.2 (Hooge et al. 1999). 1 used Chi-square goodness-of-fit tests with Bailey confidence intervals to compare observed and expected proportions of sage grouse locations in each

32 18 cover type (Neu et al. 1974, Manly et al. 1993, Cherry 1996). Although other methods could have been used (e.g., Johnson 1980, Aebischer et al. 1993), I chose the Neu et al. (1974) method because McLean et al. (1998: 793) found that it better identified "habitat selection patterns consistent with known requirements" compared with other methods. I used P < 0.05 as the cutoff for determination of significance in all statistical tests. Use of burned and unburned habitats was defined as: 1) Selected - cover types used in greater proportion than their availability; 2) Cover types used in proportion to their availability; 3) Cover types used less than their availability; and 4) Avoided - no measured use. All assumptions outlined by Alldredge et al. (1998) for the use of Chisquare tests in resource selection studies were addressed. I used the Schoener (1981) estimator, as described by Swihart and Slade (1985), to test for autocorrelation of locations. Five locations that did not meet independence criteria were removed from the data set. Each year-specific category was analyzed separately to control for annual variation and changes in burned area availability (Schooley 1994). Birds with a large deviation in number of locations from the mean number within each category were excluded from the analysis to avoid the possible anomalous influence of those individuals in the overall analysis (Alldredge and Ratti 1992). Differences in breeding-season movement patterns of radio-marked individuals at HMNAR were investigated by comparing the sizes of individual 100% MCP utilization distributions. Selection analysis results for each category were pooled across years to control for the influence of 4th order habitat

33 19 changes that may have been influenced by climatic variation and could possibly affect year-to year habitat use (Johnson 1980, Schooley 1994). All types of fires (e.g., prescribed, escaped prescribed, and wildfires) were pooled and defined as I cover type to reduce potential Type I errors associated with an increased number of habitat categories included in the analysis (Alldredge and Ratti 1986). I used logistic regression models to evaluate use patterns of burned areas by nesting, brood-rearing, and broodless females. Explanatory variables included in the model were: cover type burned, age and size of burn, annual precipitation, year, and elevation. In the logistic model, cover type use was categorized as: used (selected, used in proportion, or used less than available) or not used (avoided or no measurable use). AMANOVA (Multivariate Analysis of Variance) was used to compare habitat components (e.g., shrub, forb, and grass cover) of utilized areas by nesting and brood-rearing females in burned and unburned habitats. If significant differences were found (P < 0.05), Tukey-Kramer multiple comparisons were used to determine what habitat components contributed to the differences (Ramsey and Schafer 1997).

34 20 RESULTS NESTING Unburned mountain big sagebrush was consistently selected in all years (x2= 21.53, 4 df, P < 0.001). A high-elevation late-successional (39-42 years old) mountain big sagebrush burned area was selected in 1991 and However, in 1999, there was no difference in shrub, forb, or grass cover in this burn compared with unburned mountain big sagebrush habitats (Table 3). With the exception of mountain big sagebrush, other unburned cover types were used proportionally or used less than available (Appendices D, E, F, G, H). Table 3. Sage grouse habitat characteristics (% cover) in a 1954 mountain big sagebrush burn (Degarmo Canyon Wildfire) and unburned mountain big sagebrush at Hart Mountain National Antelope Refuge, Lake County, Oregon, Burned (1954) Unburned (1999) Resource Mean SE Mean SE Forb Short Grass Tall Grass Low shrub Medium shrub Tall shrub

35 21 Burned areas composed 15% of cover types available to nesting females (Appendix Q. Burned areas were generally avoided for nesting, with only 11% of burned areas used (Table 4, Table 5). Use tended to increase with increasing age of the bum (Table 4, Fig. 3). Table 4. Use of burned areas by radio-marked nesting female sage grouse during the breeding season at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age of burns in each use category is also indicated. Use Response % Mean Age (yr) Select Proportional Less than available Avoid

36 22 Table 5. Use of burned cover types by nesting female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age and number of bums in each category is also indicated. Burned Cover Type Mountain big sagebrush Low sagebrush Response % Use Mean Age N % Use Mean Age N Select Proportional Less than available Avoid Table 5. (Continued). Burned Cover Type Wyoming big sagebrush Other Response % Use Mean Age N % Use Mean Age N Select Proportional Less than available Avoid

37 23 N 50 -r 40t CI Mean Age A Use Response 10-f- 0 A {ir- -I---i- 0 S P L A Use Response Figure 3. Mean age of burned areas available to radio-marked nesting female sage grouse as a function of use response at Hart Mountain National Antelope Refuge, Lake County, Oregon, S - Selected, P - Proportional Use, L - Less than Available, A- Avoided. Lack of data-point convergence in burned areas used by nesting females precluded logistic modeling, because 90% of burned areas used by nesting females were in mountain big sagebrush (Table 3). Mountain big sagebrush burns were used more than other burned cover types (X2 = 12.09, 3 df, P = 0.008), and burned areas > 20 years old were used more than burned areas < 20 years old (z = 11.63, 1 df, P < 0.001). Ally nests in burns < 20 years old were unsuccessful. Nesting success in >20 year burns (29%, n = 6/2 1) was similar to nesting success in unburned areas (28%, n = 49/177). Habitat components for nests in burns > 20 years old and nests in unburned areas were similar (Table 6).

38 24 Table 6. Habitat components of sage grouse nest sites in recently burned areas (< 20 years-old), old burns (> 20 years-old), and unburned areas at Hart Mountain National Antelope Refuge, Lake County, Oregon, < 20-year-old burns > 20-year-old burns Unburned Resource Mean SE Mean SE Mean SE Forb Short Grass Tall Grass Short shrub Medium shrub Tall shrub BROOD-REARING Estimated sizes of areas utilized by individual brood-rearing females (100% MCP) were highly variable (Mean = 1347 ha, SE = 435, 95% Cl = ). Unburned low sagebrush and mountain big sagebrush cover types were consistently selected or used in proportion to their availability by brooding females. Unburned Wyoming big sagebrush and other unburned habitats were used less than available or avoided (,, = 9.52, 4 df, P = 0.048). Other unburned cover types were used in proportion to their availability or used less than available (Appendices I, J, K, L, M). Burned areas composed 14% of habitats available to brooding females (Appendix Q. Burned areas were generally avoided by brooding females, with only 11% of burned areas used (Table 7, Table 8). Exceptions included a 44 year-

39 25 old high-elevation late-successional mountain big sagebrush burned area that was selected in A 7-year-old mountain big sagebrush burn, a 30-year-old highelevation mid-successional mountain big sagebrush burn, and a 36-year-old highelevation late-successional mountain big sagebrush burn were used in proportion to their availability in 1990 and Use tended to increase with increasing age of the burn (Table 7, Fig. 4). Table 7. Use of burned areas by radio-marked brooding female sage grouse during the breeding season at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age of burns in each use category is also indicated. Response % Mean Age (yr) Select Proportional Less than available Avoid

40 26 Table 8. Use of burned cover types by brooding female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age and number of burns in each category is also indicated. Burned Cover Type Mountain big sagebrush Low sagebrush Response % Use Mean Age N % Use Mean Age N Select Proportional Less than available Avoid Table 8. (Continued). Wyoming big sagebrush Other % Use Mean Age N % Use Mean Age N Response Select Proportional Less than available Avoid

41 27 40-r % V Mean Age A Use Response 0 A S P L A Use Response 20 Figure 4. Mean age of burned areas available to radio-marked brooding female sage grouse as a function of use response at Hart Mountain National Antelope Refuge, Lake County, Oregon, S - Selected, P - Proportional Use, L - Less than Available, A- Avoided. The logistic model that best fit the data from brooding females suggested that: 1) burned areas > 20 years-old were used more than burned areas < 20 yearsold (,j = 9.39, 1 df, P = 0.002); 2) burns in mountain big sagebrush were used more than burns in other cover types (, = 11.52, 3 df, P = 0.009); and 3) there was a year-effect on use of burns in 1990 and 1996 where brooding females tended to exhibit more use of burned areas(, = 20.96, 9 df, P = 0.017). However, I found no differences in any characteristics of the annual samples, such

42 28 as sample size, age distribution, or capture locations that helped to explain the year-effect. Plant phenology information was unavailable, but could be an influence. No other variables (i.e., size of burn, precipitation, or elevation) contributed to model performance. Brood sites in > 20 year-old burned areas had greater cover of short shrubs than brood sites in burned areas that were < 20 years old (t44 = 3.369, P < 0.001) (Table 9). Brood sites in > 20 year-old burned areas had more cover of medium shrubs than brood sites that were in burned areas < 20 years old (t44 = 2.96, P = 0.003). Forb cover was similar in < 20 year-old burned areas and > 20 year burned areas. There were no differences in brood-site habitat variables for > 20 year-old and brood sites in unburned areas. Table 9. Habitat components of sage grouse brood sites in recently burned areas (< 20 years-old), old burned areas (> 20 years-old), and unburned areas at Hart Mountain National Antelope Refuge, Lake County, Oregon, < 20 year-old burns > 20 year-old burns Unburned Resource Mean SE Mean SE Mean SE Forb a Short Grass 34.2a a 09 Tall Grass 8.3a b b 0.6 Short shrub b b 0.7 Medium shrub 0. l a 0.1 0a 0 0.9a 0.2 Tall shrub a a Within-row means followed by a different letter are significantly different at (P < 0.05).

43 29 BROODLESS FEMALES Estimated sizes of areas utilized by broodless females (100% MCP) were highly variable (Mean = 2096 ha, SE = 354, 95% Cl = ). Unburned low sagebrush was consistently selected in all years (x2 = 11.33, 4 df, P = 0.023). Compared to brooding and nesting females, broodless females exhibited more plasticity in terms of proportional use of unburned habitats (Appendices N, 0, P, Q, R). Burned areas composed 21% of habitats available to broodless females (Appendix Q. Burned areas were selected during 1989, 1990, and 1992; otherwise available burned areas were avoided (77%), with only 23% of burned areas used (Table 10, Table 11). A year-old high-elevation latesuccessional mountain big sagebrush burn was selected during 1989, 1990, and 1992, a 27 year-old mountain big sagebrush burn was selected during 1989, and an year-old low sagebrush burn was selected during 1989 and 1992.

44 30 Table 10. Use ofburned areas by radio-marked broodless female sage grouse during the breeding season at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age of burns in each use category is also indicated. Response % Mean Age (yr) Select Proportional Less than available Avoid Table 11. Use of burned cover types by broodless female sage grouse at Hart Mountain National Antelope Refuge, Lake County, Oregon, Mean age and number of burns in each category is also indicated. Burned Cover Type Wyoming big sagebrush Other Response % Use Mean Age N % Use Mean Age N Select Proportional Less than available Avoid Table 11. (Continued). Burned Cover Type Wyoming big sagebrush Other Response % Use Mean Age_ N % Use Mean Age N Select Proportional Less than available Avoid

45 31 Use increased with increasing age of burn (Table 10, Fig. 5). No variables (i.e., cover type burned, age of burn, size of burn, precipitation, year, or elevation) contributed to logistic-regression model performance. 30, S P L A Use Response - 50 Mean Age 40 A Use Response Figure 5. Mean age of burned areas available to radio-marked broodless female sage grouse as a function of use response at Hart Mountain National Antelope Refuge, Lake County, Oregon, S - Selected, P - Proportional Use, L - Less than Available, A- Avoided.

46 32 DISCUSSION My analysis indicated that unburned areas were generally selected and burned areas were generally avoided by female sage grouse at HMNAR during the breeding season. Although variables such as size of burn, annual precipitation, and elevation did not significantly contribute to the logisticregression model, age of burn and the cover type burned affected sage grouse use of burned areas. Age of burn, and the resultant return of the shrub (i.e., sagebrush) component, was most positively associated with use of burned areas by nesting and brood-rearing female sage grouse. The cover type burned also had a strong influence on selection of nest and brood sites. When burned habitats were used, they were typically > 20 year-old burns that occurred in mountain big sagebrush. Use of these particular burned areas may be related to the higher biotic potential (in terms of soil productivity/structure, floral diversity, and precipitation) typically associated with mountain big sagebrush habitats and, subsequently, more rapid recovery of these cover types after fire (Miller and Eddleman 2000). However, there was a much higher proportion of mountain big sagebrush burns at HMNAR than bums in any other cover type. Consequently, analysis results may have been biased towards use of mountain big sagebrush burns because of the limited number of burns in the other 3 cover types available to nesting, brood-rearing and broodless females.

47 33 Dunn and Braun (1986) observed that sage grouse avoided areas without sagebrush cover as they migrated towards wintering grounds. My study demonstrated a similar trend for nesting, brood-rearing, and broodless females. Sime (1991) observed sage grouse use of a 12-year-old burned area in spring but saw no overall difference in use of burned and unburned areas. I observed little or no use of burned areas < 20 years old by breeding females in cover types other than mountain big sagebrush. Several studies have indicated a post-fire increase in cover, abundance, and nutritive content of forbs (Harniss and Murray 1973, Bailey and Anderson 1978, Cook et al. 1994, Pyle and Crawford 1996, McDowell 2000), which theoretically should benefit sage grouse. However, the loss of the shrub component may have countered any benefits from increases in forb cover, abundance, or nutritive content, at least during the nesting, brood-rearing, and broodless stages. Due to the significant amount of time necessary for the sagebrush component to return in and environments, the role that fire may play in long-term (> 100 years) rehabilitation of degraded sagebrush ecosystems remains poorly understood. Foundations for evaluating these possible effects, such as sitespecific long-term monitoring of burned areas, should be implemented to assist future researchers. Mapping error or location error is always a concern in studies of habitat use. In my analysis, I estimated that the maximum misclassification rate of grouse locations due to mapping error and location error was 9%. This worst-case