Ecological effects of winter road grooming on bison in Yellowstone National Park by Daniel David Bjornlie A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Fish and Wildlife Management Montana State University © Copyright by Daniel David Bjornlie (2000) Abstract: The. effects of winter recreation on wildlife in Yellowstone National Park (YNP) have become high-profile issues. Snowmobiling is perhaps the most contentious of these issues. The road grooming needed to support snowmobile travel in YNP has also come under examination for its effects on bison {Bison bison) ecology. Data were collected from November 1997 through May 1998 and again from December 1998 through May 1999 on the effects of road grooming on bison in Madison-Gibbon-Firehole (MGF) area of YNP. Synoptic bison surveys of the entire study area were conducted 33 times during the study. Peak bison numbers in the study area occurred in late March/early April and were correlated with snow water equivalent measurements in the Hayden Valley area (1997-98: r2 = 0.62, P < 0.001; 1998-99: r2 = 0.64, P < 0.001). Data from an infrared trail monitor set up on the Mary Mountain trail between the Hayden Valley and the Firehole Valley suggest that this trail is the sole corridor for major bison distributional shifts between these locations. Of the 28,293 individual bison observations made during the study, 8% of the activities were traveling, while 69% were foraging. These percentages were nearly identical during the period of winter road grooming (7% and 68%, respectively). The majority of foraging activities during this period (77%) involved displacing snow, while 12% of the traveling activities involved displacing snow. The majority of travel took place off roads (P < 0.001). Bison utilized geothermal features, stream and riverbanks, and a network of established trails to travel in the study area. Peak road use by bison occurred in April, with the lowest values during the road-grooming period. Groomed road use by bison in the MGF area of YNP seems to be an activity that is neither sought out nor avoided. The minimal use of roads compared to off-road areas, the short distances traveled on the roads, the decreased use of roads during the road grooming period, and the increased costs of negative interactions with over-snow vehicles suggest that roads are not the major influence on bison ecology that has been proposed.  ECOLOGICAL EFFECTS OF WINTER ROAD GROOMING ON BISON ' IN YELLOWSTONE NATIONAL PARK by Daniel David Bjomlie A thesis submitted in partial fulfillment o f the requirements for the degree o f Master o f Science in Fish and Wildlife Management MONTANA STATE UNIVERSITY Bozeman, Montana May 2000 ii APPROVAL o f a thesis submitted by Daniel David Bjomlie This thesis has been read by each member o f the thesis committee and has been found to be satisfactory regarding content, EngUsh usage, format, citations, bibliographic style, and consistency, and is ready for submission to the CoUege o f Graduate Studies. Ernest R. Vyse Approved for the Department o f Biology (Signature) / 'ZOOTt (Date) Bruce R. McLeod Approved for the College o f Graduate Studies (Signature) (Date) iii STATEMENT OF PERMISSION TO USE In presenting this thesis in partial fulfillment o f the requirements for a master’s degree at Montana State University, I agree that the Library shall make it available to borrowers under rules o f the Library. I fI have indicated my intent to copyright this thesis by including a copyright notice page, copying is allowed only for scholarly purposes, consistent with “fair use” as prescribed in the U.S. Copyright Law. Requests for perthission for extended quotation from or reproduction o f this thesis in whole or in parts may be granted only by the copyright holder. ACKNOWLEDGEMENTS . This research was funded by the United States Geological Survey - Biological Resources Division. Special thanks go to P J . Gogan for organizing and administering the bison research projects. I would also like to thank the United States Department o f the Interior - National Park Service - Yellowstone National Park for their logistical support and advice. I would like to thank D.M. Fagone5 M.J. Ferrari5 A.R. Hardy5 S.C. Hess5 R. Jaffe5 J. McDonald, and AC . Pils for their assistance in the collection o f field data; R. Abegglen5 C.A. Van De Polder, W.W. Wimberly, and the rest o f the Madison Junction maintenance workers for their mechanical help and logistical support; L.T. Inafuku5 M.P. Keator5 R.R. Siebert5 D. Young5 and the rest o f the West District rangers for their cooperation and logistical support; W. Clark, G. Kurz5 and J.A. Mack o f the Yellowstone bison office for equipment, information, and reviewing the study plan; S. Cherry for help with the regression model; P.J. Gogan5 S.C. Hess5 L.R. Irby5 and A.V . Zale for reviewing the manuscript; and RA . Garrott for advice, criticism, and support throughout the study. VTABLE OF CONTENTS Page 1. INTRODUCTION............................................................ I 2. STUDY A R E A ....................................................................... 6 3. M ETHODS..................................................................... 10 Spatial and Temporal Snowpack Variatidn....................................................................10 Synoptic Bison Surveys.......................................... 10 Bison Travel Monitoring..................................................... 13 Behavioral Observations.....................................................................................................17 4. RESULTS ......................................................................................*........................................... 19 Spatial and Temporal Snowpack Variation . . . ............................................................. 19 Bison Population and Distribution Dynamics.......................... 19 Bison Activity B udgets ................... ......................... ,.....................................................25 Bison Road U s e ..................... ........................... : ........................... ..................................28 5. D ISCUSSION ................................ 35 LITERATURE CITED 44 vi LIST OF TABLES Table Page I. Distribution o f bison among the 3 major river drainages in Madison-Gibbon-Firehole area o f Yellqwstpne National Park.................................. *.....................................................................................23 . 2. Percentage o f activities o f individual bison observed during complete and partial synoptic bison surveys in the Madison- Gibbon-Firehole area o f Yellowstone National P a rk ....................................26 3. The amount o f road use by bison for each o f 7 road sections in the Madison and Firehole Valleys o f Yellowstone National Park....................................... 31 ■ 4. Results o f the regression model o f the factors influencing bison road use in the Madison-Gibbon-Firehole area o f Yellowstone National Park................................................................... 32 5. The percentages o f reactions o f bison groups traveling on roads to interactions with over-snow and wheeled vehicles in the Madison-Gibbon-Firehole area o f Yellowstone National Park............................................................... ............. .................... 33 vii LIST OF FIGURES Figure Page 1. Madison-Gibbon-Firehole study area o f Yellowstone National ' Park................................................... 7 2. Temporal trends in snowpack as indexed by snow water equivalent (SWE) measurements recdrded at the Natural Resources Conservation Service automated SNOTEL sites in the West Yellowstone (2,042 m) and Canyon (2,466 m) areas o f Yellowstone National Park during the 1997-98 and 1998-99 field seasons............................................................................................. 20 3. Temporal trends in the number o f bison enumerated in the Madison- Gibbon-Firehole study area o f Yellowstone National Park.................. 20 4. The correlation between snow water equivalent and measured at the Canyon SNOTEL site and the number o f bison enumerated in the Madison-Gibbon-Firehole area o f Yellowstone National Park.................. 21 5. The number o f events recorded as bison by an infrared trail monitor placed on the Mary Mountain trail between the Hayden Valley and the Firehole River Valley in Yellowstone National Park.............................22 6. The number o f events recorded as bison by an infrared trail monitor placed, on the Oneiss Creek trail between the Madison Riyer and Cougar Meadows in Yellowstone National Park.......................................... 24 7. The locations o f traveling bison groups in = 383) in the Madison- Gibbon-Firehole study area o f Yellowstone National Park........................ 27 8. The temporal trends in the proportion o f all bison groups observed traveling on roads, established trails, and o ff roads and established trails in the Madison-Gibbon-Firehole area o f Yellowstone National Park..................................................................................28 9. Bison use o f the Madison-Gibbon-Fifehole road system in Yellowstone National Park............................................... 10. Frequency o f distances traveled by 90 groups o f bison observed during behavioral observations in the Madison- Gibbon-Firehole area o f Yellowstone National Park . . . . IX ABSTRACT The. effects o f winter recreation on wildlife in Y ello wstone National Park (YNP) have become high-profile issues. Snowmobiling is perhaps the most contentious o f these issues. The road grooming needed to support snowmobile travel in YNP has also come under examination for its effects on bison {Bison bison) ecology. Data were collected from November 1997 through May 1998 and again from December 1998 through May 1999 on the effects o f road grooming on bison in Madison-Gibbon-Firehole (MGF) area o f YNP. Synoptic bison surveys o f the entire study area were conducted 33 times during the study. Peak bison numbers in the study area occurred in late March/early April and were correlated with snow water equivalent measurements in the Hayden Valley area (1997-98: / = 0.62, P < 0.001; 1998-99: r2 = 0.64, P < 0.001). Data from an infrared trail monitor set up on the Mary Mountain trail between the Hayden Valley and the Firehole Valley suggest that this trail is the sole corridor for major bison distributional shifts between these locations. O f the 28,293 individual bison observations made during the study, 8% o f the activities were traveling, while 69% were foraging. These percentages were nearly identical during the period o f winter road grooming (7% and 68%, respectively). The majority o f foraging activities during this period (77%) involved displacing snow, while 12% o f the traveling activities involved displacing Snow. The majority o f travel took place o ff roads (P < 0.001). Bison utilized geothermal features, stream and riverbanks, and a network o f established trails to travel in the study area. Peak road use by bison occurred in April, with the lowest values during the road­ grooming period. Groomed road use by bison in the MGF area o f YNP seems to be an activity that is neither sought out nor avoided. The minimal use o f roads compared to off-road areas, the short distances traveled on the roads, the decreased use o f roads during the road grooming period, and the increased costs o f negative interactions with over­ snow vehicles suggest that roads are not the major influence on bison ecology that has been proposed. iINTRODUCTION Over the past several decades, there has been a change in the focus o f human use o f public lands. The number o f outdoor recreationists has increased rapidly with the increase in human population, disposable income, and leisure time (Knight and Gutzwiller 1995). This emphasis on recreation creates a new form o f effects on the resources o f public lands. In areas that were once used mainly for extractive industries like mining and logging, the predominant activity is now shifting to outdoor recreation (Knight and Gutzwiller 1995). Instead o f localized effects, resource managers must now deal with a much broader range o f effects over much larger areas (Cole and Knight 1991). It has been assumed that because recreation effects are dispersed over a wider area, they are diluted and cause less damage than extractive uses (Youmans 1999). This dispersion, however, is a primary reason why recreation effects can be so extensive (Flather and Cordell 1995). For many animals in high latitudes, the winter season is the most physiologically taxing. Therefore, the additional effects o f winter recreation can have major effects. Goodrich and Berger (1994) found that “since the quiet approach o f researchers sometimes caused black bears (Ursus americanus) to abandon their dens and cubs, skiing and other recreational activities could have the same or more heightened effects.” Snowmobiling has received the vast majority o f attention dealing with winter recreation effects (Canfield et al. 1999). The reports on the effects o f snowmobile activity show a broad range o f results. In Maine, Lavigne (1976) found that snowmobile trails enhanced white-tailed deer (Odocoileus virginianus) mobility during periods o f deep snow. 2Dorrance et al. (1975) reported that whereas some white-tailed deer avoided areas where snowmobiles were present, there were no significant changes in home range or daily movement patterns. Other research found more negative effects, such as using snowmobiles for illegal hunting (Malaher 1967) and reduced home ranges o f white-tailed deer in high-use snowmobile areas forcing them into less preferred habitat (Huff and Savage 1972). Snowmobiles in Canada created well-used trails through ungulate winter ranges that wolves (Cams lupus) had not previously been able to access, thus increasing predation (Claar et al. 1999). There has also been concern about air, water, and noise pollution created by snowmobiles (Anne 1981, National Park Service 1999, Youmans 1999). It has been reported in several other studies, however, that snowmobiles confined to designated trails or roads are less disturbing to wildlife than cross-country skiers or persons on foot. Average flight distances for elk (Cervus elaphus) and mule deer (Odocoileus hemionus) in Yellowstone National Park were less for disturbances attributed to snowmobiles than for those attributed to skier interactions (Anne 1981). Freddy et al. (1986) found that responses by mule deer to persons on foot were longer in duration, more often involved running, and involved greater energy costs than reactions to snowmobiles on trails. These examples support the observation made by Parker et al. (1984:484) stating, “Greater flight distances occur in response to skiers or individuals on foot than to snowmobiles, suggesting that the most detrimental disturbance to the wintering animals is that which is unanticipated.” 3Yellowstone National Park (YNP) is a microcosm o f the effects felt by growing recreational activities (Anne 1981, Gunther 1991, Cassirer et al. 1992, National Park Service 1999). Annual visitation to YNP has increased from 2,404,862 people in 1982 to 3,131,381 in 1999 (Tammy Wert, YNP, unpubl. data). Total winter visitation has increased at an even greater rate, from 77,679 in 1984-85 to 124,275 in 1998-99 (National Park Service 1999). The first permits were granted for snowcoaches in 1955 to bring tourists into YNP, with snowmobile tourism beginning in 1963-64 when the first private snowmobiles entered the Park (Aune 1981). In order to restrict the use o f snowmobiles to the roads and facilitate better access to the Park, the National Park Service (NPS) began road-grooming operations in 1970. The number o f snowmobile and snowcoach (over-snow vehicle) visitors entering YNP increased from 12,239 in 1970-71 to 42,597 in 1977-78 (Tammy Wert, YNP, unpubl. data). This increase in winter visitation was large enough to warrant a study o f the impacts o f winter recreation on wildlife (Aune. 1981). The conclusion was that although minor impacts occurred, recreation activity was not a major factor influencing wildlife distribution and cover use. Since that time, over-snow vehicle (OSV) visitation has increased to 86,977 in 1998-99 (National Park Service 1999). Concurrent with increasing Park visitation was a steady increase in the YNP bison {Bison bison) population (National Park Service 1998). The winter o f 1996-97 brought above-average snow depths and thick ice layers that impeded foraging (Cheville et al. 1998). In search o f forage, bison left YNP and over 1,000 were subsequently shot in a controversial management action to protect livestock from potential exposure to 4brucellosis (Keiter 1997, National Park Service 1998). Prior to the killing o f bison in 1996-97, an internal unpublished Park Service report (Meagher 1993) suggested that road grooming facilitated bison migration out o f the Park. The contention raised was that groomed roads permit bison to move more freely and efficiently about the Park in winter, thus allowing them to save energy by avoiding snow displacement and to access more and higher quality winter forage than would be possible without the road system. Enhanced nutrition and overall postwinter health are purported consequences, leading to excessive numbers, habitat deterioration, range expansion, and movements o f animals outside the Park. Prompted by the assertions in this report and the killing o f the bison in the winter o f 1996-97, several advocacy groups filed suit against the National Park Servile, leading to an out-of-court agreement to write an environmental impact statement (EIS) (Baskin 1998, National Park Service 1998). To provide information for this EIS, more research was needed on the effects o f road grooming on bison movements, distribution, and behavior. The goal o f this study, therefore, was to quantify bison distribution, movement, and activity patterns and evaluate the ecological consequences o f winter road grooming. Four testable hypotheses were formulated to address the purported effects o f road grooming on bison. The first hypothesis was that road grooming facilitates major bison distributional shifts. Second was that displacing snow while traveling is the major energetic cost to bison in the winter. The third hypothesis was that the majority o f bison travel in winter would take place on groomed roads within the Madison-Gibbon-Firehole study area, facilitating long distance movements by bison seeking to escape the costs o f 5traveling in snowpack. The fourth hypothesis stated that traveling on groomed roads results in less energy expenditure than traveling off-roads. STUDY AREA The study area consists o f the drainages o f the upper MadisonRiver east from the Park boundary at West Yellowstone to Madison Junction, the Gibbon River upstream to Norris Geyser Basin, and the Firehole River upstream from Madison Junctibn to Old Faithful (Fig. I). The meadows and geothermal features in this area are considered the primary winter range for the Mary Mountain bison subpopulation, which is the largest in YNP and comprises 60% to 75% o f the entire YNP bison population (Thome et al. 1991, Meagher 1993). The study area is approximately 7,200 hectares with elevations ranging from 2,000 m to 2,250 m. Winter visitation in this area is the heaviest in the Park, with 60% o f winter visitors to YNP stopping at Old Faithful (National Park Service 1998). Although not within the formal study area, the Hayden Valley in the central region o f the Park to the east o f the study area and the meadows above and along Cougar and Duck creeks to the northwest o f the study area are also important to bison movements within the study area. The Hayden Valley is higher in elevation (2,466 m) than the Madison-GibbOn-Firehole (MGF) study area. Bison move into the study area from this region in winter (Meagher 1993). The Cougqr arid Duck Creek area is lower in elevation (2,042 m) than most o f the study area, and along with the Madison River, is utilized as an egress corridor for bison in the study area (National Park Service 1998). These egress points are o f major concern due to the threat o f brucellosis transmission to cattle outside o f the Park (Cheville et al. 1998). 7uck Cr. Gneiss Cr. Trail Gibbon RCougar Cfc- \Cougar Meadows Madison R. Madison Junction West Yellowstone /X v/ Waterways / X / Groomed Roads Mary Mountain Trail Nez Perce Cr. HaydenValley Ungroomed Roads Bison Study Area Trail Monitor Locations irehole R. Old x Faithful Kilometers Figure I. Madison-Gibbon-Firehole study area o f Yellowstone National Park. The shaded area represents the bison range, as delineated by Ferrari (1999). Groomed and ungroomed roads within the study area are denoted. The locations o f infrared trail monitors on bison off-road trails are identified. 8The study area consists o f extensive, flat meadows associated with the major river systems. Forested regions and the canyons o f the major rivers break up these areas. There is also significant geothermal activity in the study area. Warm sites near these and other geothermal areas are snow-free earlier in spring and have a longer growing season than surrounding areas. They also produce areas o f reduced snow cover or bare ground in the winter (Meagher 1973, Despain 1990). In 1988 several major fires burned through the study area (Despain 1990). Prior to these fires, approximately 61% o f the study area was dominated by climax lodgepole pine (Firms contorta) or serai stages (Aune 1981). The 1988 fires resulted in large areas o f open canopy with charred snags and downed wood. Many o f these areas have been rapidly recolonized by lodgepole pine (Knight 1996). The climate in the study area is characterized by long, cold winters and short, cool summers. Records from the National Oceanic and Atmospheric Administration (NOAA) - for West Yellowstone, Montana, indicate the mean monthly temperature ranges from 15.3°C in July to -10.6°C in January (Aune 1981). Mean annual snowfall is 418 cm and mean snow depth exceeds 46 cm for an average o f 126 days per year. The beginning o f snow accumulation at 2,100 m occurs in late October and the end o f the snowpack at this elevation is normally in late May (Despain 1990). The road system within the study area consists o f paved, 2-lane roads along the major rivers (Fig. I). From West Yellowstone the road extends 22 km to Madison Junction, passing through meadows, forests and the Madison Canyon to the junction o f the Firehole and Gibbon Rivers. From Madison Junction the road along the Gibbon 6 9River passes through the Gibbon Canyon and Gibbon Meadows and on to Norris for a distance o f 22 kilometers. The road from Madison Junction to Old Faithful follows the Firehole River for 26 kilometers through Firehole Canyon, Lower and Midway geyser basins. The roads in the study area are open to visitors in wheeled vehicles (WV) from April 15 until October 31. From November I until the third week in December the roads in the study area are closed to all vehicles driven by visitors. During this time, WV traffic is restricted to YNP personnel only. However, the roads are not plowed and snow is allowed to accumulate on the roads. Road grooming begins the evening prior to the opening date o f the OSV season, which is approximately the third week o f December. Roads are groomed nightly until the end o f the OSV season, in early to mid March. The study area roads are then closed to all visitor vehicles and the roads are plowed to pavement. Travel is restricted to YNP personnel until approximately April 15. METHODS I began data collection for the first field season o f this study in mid November 1997 and continued through May 1998. Data collection began again in early December 1998 and continued through May 1999. The OSV season extended from December 19 to March 9 during the winter o f 1997-98, and from December 16 to March 14 during the winter o f 1998-99. Spatial and Temporal Saowpack Variation Snow water equivalent (SWE) was used as an index o f snowpack in the study area. This index calculates the equivalent amount o f water contained in a column o f snow (Fames 1996). These data were gathered from remotely operated Natural Resources Conservation Service (NRCS) SNOTEL sites near West Yellowstone and Canyon Village (Fig. I). The West Yellowstone site, at an elevation o f .2,042m, reflects SWE in the lower-elevation valley bottoms o f the study area, while the Canyon SNOTEL site, at 2,466m, reflects SWE in the Hayden Valley (Fig. I, inset), which is the major summer range for bison in YNP (Meagher 1973). Synoptic Bison Surveys The spatial and temporal patterns o f bison population, distribution, and activities were characterized by conducting synoptic surveys o f the entire study area at 10-day intervals using methodology developed by Ferrari (1999). The study area was divided into 72 units ranging in size from 6 to 716 ha, with 6 survey routes designated through the units. Each member o f a 3-person crew traversed I o f these routes using 10 11 snowmobiles and/or snowshoes each day, attempting to locate all bison within each unit. Therefore, using a 3-person crew, it was possible to survey the entire study area in 2 days. Surveys were conducted in a manner that minimized the possibility o f missing or double counting bison. This was accomplished by crew members starting each o f the survey routes simultaneously and by surveying the'3 Madison and Gibbon routes on I day and the 3 Firehole routes on the other. The long distance between these 2 areas decreased the probability ofbisqn moving from I area to another between survey days. For all groups o f bison detected during surveys we recorded the location o f each group in Universal Transverse Mercator (UTM) coordinates (using USGS 7.5 minute maps) and the age and sex composition o f the group divided into cows, bulls, calves o f the year, arid unknown adults. In addition, the activity o f each observed bison was classified as traveling, foraging, or resting. Traveling bison were defined as any individual that was engaged in directed, sustained travel. This would exclude animals slowly wandering and stopping frequently to forage. Bison travel was categorized as either on roads, on established trails, or o ff o f roads and established trails. Foraging animals were defined as any individual that was feeding or slowly moving around in search o f forage. Resting animals were defined as stationary individuals, either lying down or standing, that were obviously not engaged in either foraging or traveling. If bison were displacing snow in any o f these activities, we noted snow depth in relation to bison anatomy using a reference diagram from Carbyn et aL (1993) and placed it into I o f (j categories (1-20 cm, 21-40 cm, 41-60 cm, 61-80 cm, 81-1.00 cm, >100 cm). To augment data collected during the complete synoptic surveys, I also performed 12 partial surveys o f randomly selected blocks o f survey units during the 8-day interval between each complete survey. For these partial surveys, the 6 routes were combined into 3 geographic strata consisting o f 2 adjacent travel routes. These were the Madison River/lower Firehole River, the Gibbon River drainage, and the middle/upper Firehole drainage. Each morning o f a partial survey, a stratum was randomly selected. Then I o f the 2 travel routes within that stratum was randomly selected and followed that day. The survey schedule was tailored so that 3 partial surveys were conducted between each complete survey, with I route in each o f the 3 strata. Approximately half o f the survey units in each o f the 3 strata were then covered once in the 10-day rotation, and the other half o f the,units were covered in the next rotation. No units were resurveyed until all Units in the study area had been surveyed once. These partial surveys employed the same data collection techniques used in the complete synoptic surveys. The total number o f bison enumerated during each complete synoptic survey was calculated and considered a census o f the bison population in the study area. Linear regression (Minitab Inc. 1998) was used tq examine correlations between changes in the number o f bison detected within the study area each winter and snowpack in the Hayden Valley, as indexed by the mean SWE measurements at the Canyon SNOTEL site for the 10-day period centered on the date o f each bison population estimate. Linear regression was also used to examine the correlations between changes in the distribution o f bison in the Madison and Firehole River valleys o f the study area. Bison activity budget data from all complete and partial synoptic surveys front both years were aggregated to obtain information on bison activity patterns by month through the winter/spring field season. The locations o f all traveling bison groups observed during complete and partial synoptic surveys were plotted to determine the spatial extent o f traveling bison in the study area using ArcView GIS software (ESRI 1998). The differences in the mean number o f bison groups observed traveling per 2-week time interval on roads, trails, and off-roads and off- trails were tested using I -way analysis o f variance, and orthogonal linear contrasts (PROC GLM; SAS 1998). Bison Travel Monitoring Data on bison road use were collected opportunistically by a 4-person field crew traveling independently on the road system daily in trucks and on snowmobiles. This extensive travel resulted in coverage o f a majority o f the study area road system on a daily basis. Each crew member recorded the sections o f the road system traveled daily and attributes o f any bison group observed using the road. Daily survey effort was determined by summing the total number o f kilometers traveled by crew members excluding road sections that were traveled by more than I crew member within one half hour o f each other. Bison traveling less than 50 meters on the roads were not recorded as road use. I f possible, the direction o f travel, location o f the group at the point where they were first observed, and the locations o f where the group accessed and exited the road were identified, th e group age and sex composition were recorded using the same categories employed in the synoptic surveys. In some instances, access and exit points could not be determined. To acquire data on nocturnal road use by bison, .bison tracks on the road were recorded in the early morning prior to the onset o f daily travel by Park visitors, as crew members were traveling to their respective survey areas. Because the road system was groomed every evening, any bison tracks on the roads in the early morning were made by bison traveling the roads the previous night. The number o f bison and the age and sex composition o f bison groups could not be determined from tracks. However, all other data collected were the same as those collected for diurnal road use. Regression models were constructed in an effort to examine potential factors that may have contributed to the propensity for bison to travel on the groomed road system. The time intervals for the models were centered on the date o f each complete synoptic survey. The response variable in these models was the number o f bison groups observed traveling on the roads per 100 km traveled by observers in each interval. Potential explanatory variables included the number o f bison counted in the study area during the synoptic survey (Bison), the mean SWE at West Yellowstone for that time interval (SWE), an indicator variable specifying whether the roads were groomed (Groom; 0 - ungroomed, I - groomed), and ah indicator variable for year (Year; 0 - 1997-98 ,1 - 1998 99). The statistical software package Minitab (Minitab Inc. 1998) was utilized in conducting these regression analyses. Because o f limited sample , sizes, a corrected Akaike’s information criterion (AICc) value was calculated for each o f the models (Burnham and Anderson 1998). Models were then ranked based on AAIC0 values, with the most parsimonious model having the Idwest value. After inspecting the plots o f the residuals, an outlier was noticed. This outlier corresponded to a time interval during which crew members could not travel the study area roads for much o f the time due to road plowing operations. Thus, it was discarded due to less than 50% normal survey 15 effort, providing one legs observation for the models. The recommended number o f observations per each parameter in a model has been suggested as 6 to 10 (Neter et al. 1996). Therefore, inference from models at or below this range should be viewed with some caution. Residualplots and Durbin-Watson test statistics were utilized to test for the presence o f autocorrelation (Neter et al 1996). Models were run using all possible combinations o f the 4 parameters. In addition, terms addressing the possible 2-way interactions between Year and Bison, and Year and SWE were included with each model that contained those parameters. The total candidate list o f models was then 24. The road system in the study area was divided into 10 sections (3 from West Yellowstone to Madison Junction, 4 from Madisoh. Junction to Old Faithful, and 3 from Madison Junction to Nonris) based upon topography o f the study area. A chi-squared goodness-of-fit test was used to illustrate spatial and temporal trends in road use. The chi-squafed expected values for each road section were weighted based upon the proportion o f km traveled by observers in that section o f the total km traveled in that time interval. To obtain data on bison use o f established trails for distributional shifts, I placed a Trailmaster 1500 infrared trail monitor on the 35-km-long Mary Mountain trail between the Hayden Valley and the Lower Geyser Basin o f the Firehole Valley (Fig. I). A second monitor was placed on the Gneiss Creek trail leading from the Madison River at Seven Mile Bridge to the Cougar Meadows area (Fig. I). The Mary Mountain trail was chosen because it is considered to be the major migration route for bison moving between the Hayden Valley summer range and the Madison-Firehole winter range (Meagher 1973, 16 1993). The Gneiss Creek trail was chosen based upon observations o f heavy bison travel on the trail made by researchers in previous winters. OnCe well-defined trails were established in the snowpack by traveling bison early in the season, the monitors were placed on the trails to quantify the number o f bison using the trails and the times, dates, and direction o f use. Monitors were set up with infrared beams crossing the trail in areas where bison were restricted to traveling in single file. The monitors recorded the date and time o f any “events” that broke the infrared beam. The sensitivity o f the monitors was set to ensure that only large animals were recorded. Cameras were set up in conjunction with each o f the monitors. In order to reduce the possibility o f exposing the entire roll o f film on I bison group, the monitors were programmed with a I O-minute camera delay so that once the beam was broken and a picture was taken, the camera did not take another picture for IO minutes. The photographs provided the species identity o f the lead animals o f each group that broke the beam as well as recording the direction in which they were traveling. By analyzing the timing o f the events immediately following the photographs, the species identity o f the entire group could then be determined. For instance, 25 events taking place from 12:42 to 12:47 was considered a single group. A photograph o f the lead animal identifying it as a bison was then interpreted to mean that the remaining 24 events were also bison. Data were then downloaded and film was replaced at 7-10 day intervals. The monitor on the Gneiss Creek trail was initially set up in late November o f 1997 and was taken down in late May o f 1998. It was placed back on the trail in mid October o f 1998 and maintained until early June o f 1999. The monitor on the Mary Mountain trail was initially set up in early December o f 1997 and taken down in early July o f 1998. It was set up again in mid September o f 1998 and left in place until early July o f 1999. Behavioral Observations BeMviofal observations p f traveling bison groups Were conducted regularly through each field season to gain additional insights into this activity and the potential influence o f groomed roads and human interactions. Two sampling regimes were used to collect these beMvioral observations. During the partial synoptic surveys, the first bison group observed traveling on the road system and the first group observed traveling o ff o f the road were each chosen for an intensive beMvioral observation. In addition, between each complete synoptic survey I set aside a day solely for intensive behavioral observations o f all traveling bison groups detected. One o f the 3 river dr ainages in the study area was randomly chosen as the starting point for each o f the daylong surveys. Each subsequent intensive beMvioral survey day started with a different randomly chosen river drainage until all 3 had been used as a starting point. Once a traveling bison group was detected on these surveys, continuous observations were conducted until the group could no longer be seen or until it stopped traveling for a period o f over 5 minutes. The path traveled by the group during the observation period was recorded on 7.5-minute USGS topographical maps in the field. Data collected included distance traveled on and o ff roads, travel time, mode o f travel (single file vs. multiple animals abreast), whether snpw was being displaced, and interactions with Park visitors. The data recorded for interactions with Park visitors included the number o f bison/visitor interactions during the observation time and the type 18 o f visitor interaction, categorized as snowmobile, snpwcoach, or skier. Interactions were defined as any snowmobile, snowcoach, or individual approaching within 100 meters o f any traveling bison. Bison reactions to these interactions were recorded and categorized as either ran, pushed off road, changed direction o f travel, or none. I f a reaction included more than I o f these categories (i.e. changed direction and ran), then the reaction was classified in the category assumed to cost the bisOn the most energy. It was assumed that running cost more energy than being pushed o ff o f the road and that changing direction cost the least o f the 3 negative reactions. 19 RESULTS Spatial and Temporal Sndwpack Variation Snowpack began to accumulate in late October/early November and continued to build throughout the winter (Fig. 2). Peak snowpack occurred in early April and was followed by a rapid meltoff period lasting about 6 weeks. Canyon SWE was generally about twice that o f West Yellowstone. Peak SWE during the winter o f 1998-99 was 44% higher than 1997-98 at Canyon and 71% higher at West Yellowstone. West Yellowstone peak SWE during the winter o f 1997-98 was 37% lower than the historic average SWE (1967 -1 9 9 7 ) , whereas 1998-99 peak SWE was 24% higher than the historic average. Peak SWE at Canyon was only 2% lower than the historic average (1981 - 1997) during 1997-98, whereas it was 43% above the historic average during 1998-99. Bison Population and Distribution Dynamics Complete synoptic bison surveys were conducted 33 times during this study, 17 during 1997-98 and 16 during 1998-99. Three surveys were conducted each month except for December 1997 and 1998 and in May 1999, when only 2 were conducted due to logistic limitations. The bison population in the study area ranged from 299 to 888 in 1997-98 and 464 to 921 in 1998-99, with peaks near April I for both years (Fig. 3). There was a positive correlation between SWE at the Canyon SNOTEL site and the number o f bison counted in the study area for both the 1997-98 and 1998-99 field seasons (1997-98: r2 = 0.62, P < 0.001; 1998-99: r2 = 0.64, P < 0.001) (Fig. 4), suggesting that snowpack in the Hayden Valley influences the number o f bison in the study area. 60 i 20 — West Yellowstone — Canyon 1997-98 1998-99 1997-98 12/1 12/21 1/10 1/30 2/19 3/11 3/31 4/20 5/10 5/30 Date Figure 2. Temporal trends in snowpack as indexed by snow water equivalent (SWE) measurements recorded at the Natural Resources Conservation Service automated SNOTEL sites in the West Yellowstone (2,042 m) and Canyon (2,466 m) areas o f Yellowstone National Park during the 1997-98 and 1998-99 field seasons. 1000 -I 1997- 98 1998- 99 800 - 600 -o 400 200 12/1 12/21 1/10 1/30 2/19 3/11 3/31 4/20 5/10 5/30 Date Figure 3. Temporal trends in the number o f bison enumerated in the Madison-Gibbon- Firehole study area o f Yellowstone National Park. Data are from complete synoptic surveys o f the study area conducted at approximate 10-day intervals during the 1997-98 ( » = 1 7 ) and 1998-99 (« = 16) field seasons. 1000 -I 21 800 c in 600 - O a 400 - E3 Z 200 ■ 1997-98 o 1998-99 O O O O OO OO 0 H------------------ 1 i i i i 0 10 20 30 40 50 SWE (cm) Figure 4. The correlation between snow water equivalent (SWE) measured at the Canyon SNOTEL site and the number o f bison enumerated in the Madison-Gibbon-Firehole study area o f Yellowstone National Park. Data are from the 1997-98 (r2 = 0.62, P < 0.001, M= 17) and 1998-99 (r2 = 0.64, P < 0.001, « = 16) field seasons. The infrared trail monitor placed on the Mary Mountain trail between the Hayden Valley and the Firehole recorded 6,256 events identified as bison during the study, 2,473 in 1997-98 and 3,783 in 1998-99. These events included travel in both directions. The majority o f these events (74%) were diurnal travel, occurring between 6:00 and 17:59. During the OSV season, 81% o f the bison events were diurnal. The number o f events in a 2-week period recorded as bison ranged from 0 to 437 in 1997-98, and 4 to 676 in 1998-99 (Fig. 5). Because the monitor was not set up until December 1997, bison moving into the study area earlier in that year were not recorded. There were several mechanical failures with the trail monitors during December 1997 and November 1998 causing missing data. 22 □ 1997-98 □ 1998-99 600 - 400 200 - Figure 5. The number o f events recorded as bison by an infrared trail monitor placed on the Mary Mountain trail between the Hayden Valley and the Firehole River Valley in Yellowstone National Park. Data are from the 1997-98 (n = 2,473) and 1998-99 (n = 3,783) field seasons. Problems with missing data due to monitor failures occurred in December 1997 and November 1998. Bison were not evenly distributed throughout the 3 major river valleys o f the study area (Table I). The Gibbon River Valley contained the lowest percentage o f bison observed during complete synoptic surveys in the study area during both field seasons with an overall average o f 8%. The Madison River Valley contained an average o f 19% o f the bison during the study, while the Firehole River Valley consistently contained the largest percentage with an average o f 73%. The largest fluctuations in bison distribution within the study area took place in the Madison and Firehole valleys. There was a strong inverse relationship between the percentages o f bison in these 2 valleys in both seasons (1997-98: coefficient = -1.00, r2 = 0.94, P = 0.001; 1998-99: coefficient = -0.99, r2 = 23 0.85, P < 0.009). The percentage o f bison was lowest in the Madison River Valley in the mid-winter months o f February and March in both seasons (Table I). Table I . The distribution o f bison among the 3 major river valleys o f the Madison- Gibbon-Firehole study area o f Yellowstone National Park. Data are from complete synoptic bison surveys conducted at 10-day intervals over the entire study area during the 1997-98 and 1998-99 field seasons. Numbers for each area are reported as percents o f the total number o f bison detected for that month. The mean number o f bison enumerated per survey and the total observations for each month are given by year. Year Area. Dec Jan Feb Mar Apr May 1997-98 Gibbon 5 7 4 4 3 5 Firehole 81 73 79 82 80 65 Madison 14 20 17 14 18 31 Mean/Surv 309 . 357 587 ^53 793 485 Total 618 1,070 1,762 1,959 2,379 1,860 1998-99 Gibbon 16 13 13 12 10 4 Firehole 49 65 76 79 67 67 Madison . 35 22 11 10 23 29 Mean/Surv 488 . 576 842 829 7Q3 593 - Total 975 1,727 2,525 2,487 2,136 1,353 Data from the infrared trail monitor placed on the Gneiss Creek trail between the Madison River and Cougar Meadows indicated that bison leaving the Madison River area in mid-winter did not move west into the Cougar MeadbwsZDuck Creek area. Overall, 24 7,325 events identified as bison were recorded by this trail monitor, 3,168 in 1997-98 and 4,157 in 1998-99. Over both years, and during the OSV seasons, 76% o f the events recorded were diurnal travel, occurring between 6:00 and 17:59. The number o f bison events recorded by this monitor peaked moderately in early January o f the first season and early December o f the second season (Fig. 6). The mean SWE values for these 2- week time intervals were 7.3 and 7.5 cm, respectively, suggesting a possible threshold effect causing bison to vacate the Cougar Meadows area. From late January through March, there was little or no bison use o f this trail (Fig. 6). This period corresponds with the maximum snowpack in the study area. Large peaks occurred during the spring in both years, corresponding with the start o f snowpack melt at West Yellowstone (Fig. 6). 900 i □ 1997-98 □ 1998-99 97-98 SWE 98-99 SWE 600 - Figure 6. The number o f events recorded as bison by an infrared trail monitor placed on the Gneiss Creek trail between the Madison River and Cougar Meadows in Yellowstone National Park. Data are from the 1997-98 (n = 3,168) and 1998-99 (n = 4,157). Mean snow water equivalent (SWE) values recorded at the Natural Resources Conservation Service automated SNOTEL site near West Yellowstone are also plotted. 25 Bison Activity Budgets During the 33 complete and 99 partial synoptic bison surveys conducted over both field seasons, a total o f 29 ,184 individual bison observations were recorded. O f these observations, bison activity data were obtained from 28,293 (1997-98: n = 12,871; 1998- 99: n = 15,512) (Table 2). Bison spent 69% o f their time foraging, 23% resting, and 8% traveling. During the OSV season, the percentage o f foraging (68%) and traveling (7%) remained nearly the same. Bison were observed displacing snow in 42% o f the foraging observations and 6% o f the traveling observations. During the OSV season, 77% o f the foraging qbservations and 12% o f the traveling observations involved displacing snow. Plotting the locations o f all traveling bison groups observed during complete and partial synoptic surveys demonstrates that in areas where bison were not constricted by topography to travel along road corridors, they were found traveling widely over the study area (Fig. 7). The locations o f traveling bison observed during the OSV season are also distributed widely over the study area, suggesting that groomed roads are not a major attractor for traveling bison. During complete and partial synoptic surveys, 383 traveling bison groups, representing 2,323 individual bison, were observed. Travel was not equal among the 3 categories (ANOVA; P < 0.001). Overall, the number o f bison groups observed traveling off-road and off-trail (x = 17.2) was higher than road (x = 6.0; P < 0.001) and trail (x = 8.8; P = 0.004) travel. During the OSV season, the level o f off-road and off-trail travel (x = 15.3) was higher than travel on roads (x = 5.0; P - 0.012). Overall, travel on roads accounted for 19% o f all bison travel observed during these surveys (Fig. 8). The use o f 26 established trails accounted for 27%, and peaked during February, March, and April. Off-road and off-trail travel was the most common category o f travel throughout the study, accounting for 54% o f all observed travel. During the OSV period, road use remained low, accounting for only 17% o f bison travel observed during that period. Off­ road and off-trail travel accounted for 52% o f all travel, while use o f established trails increased to 31% o f observed travel .during the OSV period. Table 2. Percentage o f activities o f individual bison observed during complete and partial synoptic bison surveys in the .Madison-Gibbon-Firehole study area o f Yellowstone National Park. Data are reported as percentages o f total observations for each month. Data are ffomthe 1997-98 (n = 12,871) and 1998-99 (n = 15,512) field seasons.________ Displacing Snow Not Displacing Snow Month Total No. Obs. Traveling Foraging Traveling Foraging Resting aNovember 326 0.0 26.4 . 0.0 48.2 25.5 December 2,089 0.5 61.3 4.5 6.9 26.8 January 4,568 0.9 63.3 4.0 3.7 28.1 February 5,805 0.6 44.3 7.9 22.6 24.7 March 6,224 0.6 19.3 8.7 51.4 20.0 April 5,825 0.1 3.3 12.0 69.4 .. 15.2 May 3,456 0.0 0.0 6.2 64.8 29.0 Total Obs. 28,293 130 8,223 2,193 11,256 6,491 Percent 0.5 29.1 7.8 39.8 22.9 aNovember data are from 1997-98 partial synoptic surveys only. Duck Cr. Kilometers Figure 7. The locations o f traveling bison groups (n - 383) in the Madison-Gibbon-Firehole study area o f Yellowstone National Park. The shaded area denotes the bison range, as delineated by Ferrari (1999). Dark circles represent locations from the over-snow vehicle (OSV) season (n = 166), when roads are groomed. Open circles represent data from the wheeled vehicle (WV) season (n = 217). The number o f traveling bison events recorded by the trail Mary Mountain and Gneiss Creek trial monitors are also given. 28 Figure 8. The temporal trends in the proportion o f all bison groups observed traveling on roads, established trails, and o ff roads and established trails in the Madison-Gibbon- Firehole area o f Yellowstone National Park. Data are percentages o f total bison travel for each time period. Numbers at the top o f the figure are the number o f traveling groups observed during each 2-week period. Bison Road Use While there was a difference in the magnitude o f road use by bison between years, the general seasonal pattern o f bison road use was similar for both years (Fig. 9). During 42,576 km o f travel on the study area roads (1997-98: n = 22,113 km; 1998-99: n = 20,463 km), 812 bison groups were observed traveling on the roads, 277 in 1997-98 and 535 in 1998-99. As seen in the data from synoptic bison surveys (Fig. 8), the rate o f bison use o f the road system during the OSV season declined and was lower than the peaks in fall and spring. Road use decreased from a fall peak to a minimum in early winter, increased and leveled off in mid-winter, and peaked in April. This peak in road use coincided with the beginning o f spring snow meltoff at the lower elevations o f the study area (Fig. 2) as well as with peak numbers o f bison in the study area (Fig. 3). O f all bison road use events that could be recorded as either nocturnal or diurnal during the OSV season, 51 o f 335 (15%) were either bison groups observed traveling at night or tracks found in the morning indicating nocturnal use. These data suggest that nocturnal use o f the road system by bison was not a major factor in bison travel. E O 0 1 3 O □ 1997-98 □ 1998-99 [ OSV season I N' N$> Figure 9. Bison use o f the Madison-Gibbon-Firehole road system in Yellowstone National Park. Data are presented as the number o f groups observed traveling on roads per IOO km o f roads traveled by crew members during the 1997-98 (n = 277 groups, 22,113 km) and 1998-99 (n = 535 groups, 20,463 km) field seasons. The over-snow vehicle (OSV) season is delineated by dashed lines and is the period when the roads are groomed. Bison did not travel consistently on the same sections o f the study area road system throughout the season (Table 3). Bison road use seemed to follow the distribution shifts o f the bison population within the study area. The intensity o f use was not in contiguous road sections, suggesting that road use events were not long distance movements, but rather movements within certain sections. The 4-km-long Madison Canyon section had the highest level o f road travel in the fall and spring when bison were more concentrated in the Madison area (Table I). This road section, about 5 km west o f Madison Junction, is constricted from the north by high canyon walls and from the south by the Madison River, necessitating travel on the road for that stretch. During midwinter, road use was highest in the Firehole sections from Fountain Flats to Old Faithful. Bison were concentrated in this area during this time interval (Table I). The road sections in the Gibbon River Valley were excluded from the analysis due to the low percentage o f bison and amount o f movement between other portions o f the study area. The regression model analyses to explore potential mechanisms that influence bison road travel yielded 3 essentially equivalent models differing by a AAICc value o f only 0.63 (Burnham and Anderson 1998) (Table 4). Models I and 2 differ by only I parameter and model 3 contains, all 4 parameters, indicating that all 4 parameters influence bison road travel. The coefficients and 95% confidence intervals for the Groom parameter are negative, indicating that bison road use decreased during road grooming. The Bison and SWE parameters were positively correlated with road use indicating that as the snowpack and number o f bison in the study area increased, road use also increased. The Year parameter was found in all o f the top 4 models, indicating that there was a difference in road use between years that was not captured by the other parameters in the model. Ofall 24 models, none that included the 2-way interaction terms fell within 3 AAICc values o f the most parsimonious model, suggesting interactions between SWE and 31 Table 3. The amount o f road use by bison for 7 road sections in the Madison and Firehole Valleys o f Yellowstone National Park. Data are presented as the percentage o f chi-squared value contributed by each road section for each 2-week time interval. Dashes represent time intervals with non-significant chi-squared values at a = 0.05. Bold type in . boxes indicates the road section or sections that contributed the largest percentage o f chi- squared value for that time interval. The Upper Madison and Fountain Flats sections in the May 1-15 interval had lower than expected values o f road use. The largest values in all other time intervals were higher than expected values. Data are from the 1997-98 and 1998-99 field seasons. Sections start with the lower Madison section at West Yellowstone and progress upstream up the Firehole to Old Faithful.____________________ Road Sections Time Interval n km Lower Mad. Mad Cany Upper Mad Firehole Ftn Cany Flats Old Midway Faithful Nov 16-30 17 664 - - 'T - - - Dec 1-15 34. 1747 - - - - - ■ - - Dec 16-31 11 2780 <1 5 7 2 8 . 27 BI Jan 1-15 29 3955 11 g <1 18 I 10 <1 Jan I6r3I 49 3592 - - - . - - - - Feb 1-15 50 3429 17 3 2 5 <1 2 Feb 16-28 39 2254 16 3 10 13 m M 8 Mar 1-15 45 2640 - - - - - - - Mar 16-31 63 2577 - - - - - - - Apr 1-15 122 3013 4 14 <1 ' 6 2 <1 Apr 16-30 96 3138 2 16l| 21 I 7 <1 7 May.1-15 37 1768 2 19 S 11 mi 5 <1 May 16-29 10 1001 9 13 I Yellowstone Ecosystem. International Association o f Wildland Fire, Fairfield, Washington, USA. Knight, RU ., and KU Gutzwiller. 1995. Wildlife and recreationists: coexistence through management and research. Island Press, Washington, D.C., USA. Lavigne, G.R. 1976. Winter response o f deer to snowmobiles and selected natural factors. Thesis, University o f Maine, Orono, Maine, USA. Lemke, T.O., LA. Mack, and D.B. Houston. 1998. Winter range expansion by the northern YeUowstone elk herd. Intermountain Journal o f Sciences 4:1-9. Malaher, G.W. 1967. Improper use o f snow vehicles for hunting. Transactions o f the North American WildUfe and Natural Resources Conference 32:429-433- Meagher, M. 1973. The bison o f YeUowstone National Park; National Park Service Scientific Monograph Series Number. I, National Park Service, Washington D.C., USA. Al —--------. 1993. Winter recreation-induced changes in bison numbers and distribution in Yellowstone National Park. Unpublished National Park Service report. Yellowstone National Park, Wyoming, USA. Messier, F., J. Huot, D. Le Henaff, and S. Luttich. 1988. Demography o f the George River caribou herd: evidence o f population regulation by forage exploitation and range expansion. Arctic 41:279-287. Mintab Inc. 1998. Minitab user’s guide. Release 12.1. Minitab Inc, State College, Pennsylvania, USA. National Park Service. 1998. Draft environmental impact statement for the interagency bison management plan for the state o f Montana and Yellowstone National Park. National Park Service, Denver, Colorado, USA. ---------- . 1999. Winter use plan draft environmental impact statement for the Yellowstone and Grand Teton National Parks and John D. Rockefeller, Jr., Memorial Parkway. National Park Service, Denver, Colorado, USA. Neter, J., M.H. Kutner, C.J. Nachtsheim, and W. Wasserman. 1996. Apphed linear statistical models. Fourth edition. Richard D , Irwin Inc,, Chicago, Illinois, USA. Parker, K.L., C.T. Robbins, and T.A. Hanley. 1984. Energy expenditures for locomotion by mtile deer and elk. Journal o f Wildlife Management 48:474-488. Phillips, G.E. 1998. Effects o f human-induced disturbance during calving season on reproductive success o f elk in the upper Eagle River Valley. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Reynolds, P.E. 1998. Dynamics and range expansion o f a reestablished muskox population. Journal o f Wildlife Management 62:734-744. SAS Institute. 1998. SAS/STAT user’s guide. Version 7. SAS Institute, Cary, North Carolina, USA. Smith, B.L. and R.L. Robbins. 1994. Migrations and management o f the Jackson elk herd. United States Department o f the Interior, National Biological Survey resource publication 199, Washington, D.C., USA. Thome, E.T., M. Meagher, and R. Hillman. 1991. BmceUosis in free-ranging bison: three perspectives. Pages 275-287, in R.B. Keiter and M-S. Boyce, eds., The Greater YeUowstone ecosystem: redefining America's wilderness heritage. Yale University Press, New Haven, Connecticut, USA. 48 Van Camp, J. 1975. Snow conditions and the winter feeding behavior o f Bison bison in EUc Island National Park. Canadian Wildlife Service report, Calgary, Alberta, Canada. White, DJr., K.C. KendaU, and H.D. Picton. 1999. Potential energetic effects o f mountain climbers on foraging grizzly bears. Wildlife Society Bulletin 27:146- 151. Youmans, H. 1999. Project overview. Pages 1.1-1.18 in G. Joslin and H. Youmans, coordinators. Effects o f recreation on Rocky Mountain wildlife: a review for Montana. Committee on Effects o f Recreation on Wildlife, Montana Chapter o f The Wildlife Society, Helena, Montana, USA.