S Hamlint Kenneth L 599.7357 Mule deer in the F2mdmr Missouri River 1989 Breaks* Montana MULE DEER m THE MISSOURI RIVER BREAKS, MOM A* A A Study of Population Dynamics in a Fluctuating Environment Kenneth L. Hamlin Richard J. Mackie STATE D(*eNTS COLLECTION ^G 1 2 1990 MONTANA STATE Ll8RARy HFiciiAl5..E- 6th AVE HELENA, MONTANA 59620 piKmmtr LEASE RETURN cMoi\tariatDepartnieTit of } 'Fisfi.'WUdlife (Si *ParK$ December 1989 MONTANA STATE LIBRARY r S 599.7357 F2mdm, 1989c .1 HarrL " Y r .QQO M»e*e.«ih!l(i!sou„fl„M Breaks, mm mini j 0 I 3 0864 00068220 6 6 2006 FINAL REPORT RESEARCH PROJECT STATE: Montana PROJECT NO.: W-120-R-7-18 PROGRAM NO. : I STUDY NO. : BG-1.0 JOB NO. : 2 and 3 TITLE: Statewide Wildlife Research TITLE: Big Game Research TITLE: Statewide Mule Deer Ecology Studies TITLE: Mule deer population ecology, habitat relationships, and relations to livestock grazing management and elk in the Missouri River Breaks, Montana. Population ecology and habitat relationships of mule deer in river breaks habitat in northcentral Montana. Period Covered: July 1, 1975 - June 30, 1987 Prepared by: Kenneth L. Hamlin Richard J. Mackie Approved by: Donald A. Childress John P. Weigand Date: December 21, 1989 ACKNOWLEDGMENTS Literally hundreds of people contributed at one time or another or in one way or another to maintain the study through nearly 30 years and complete data analyses, writing, and publication. Within the Montana Department of Fish and Game/Fish, Wildlife and Parks, which provided basic support throughout, the list would include all Wildlife Division Administrators and Research Supervisors, Wildlife Managers in Administrative Regions 4 and 6, and nearly all Wildlife Management and Research Biologists that lived or worked in central and eastern Montana during the course of the study. Among all however, the contributions of several individuals stand out, for without them, the study may never have continued beyond the first few years. These include the late Wynn G. Freeman, Allan L. Lovaas, and James L. Mitchell, all of whom encouraged and provided funding within tight budgets for continuing population surveys and studies during the period 1963 through 1974. Both Freeman and Mitchell were also instrumental in developing and supporting subsequent intensive studies that began in 1975. Mitchell's unwavering support continued through his retirement in 1988. In addition, perhaps above all, we gratefully acknowledge the support and assistance of Regional Wildlife Biologist, Dr. C. Robert Watts. Bob Watts moved to Lewistown in 1969, and soon became an integral part of the study. He has since been a most outspoken advocate of continuing and expanding the work as well as an active participant in planning, in field studies, including aerial surveys, trapping and monitoring, and vegetation studies, and in review-editing our writing. He has at times been principal researcher, vital assistant, friend, outspoken supporter and critic--both our "prop" and our "conscience". Without all this, the study would not be what it became, if indeed it had continued at all. We also owe thanks to other Wildlife Division employees for special involvement or contributions. These include especially Research Biologists John G. Mundinger, David F. Pac, Duane B. Pyrah, and Gary L. Dusek, and Research Ecologist Henry E. Jorgensen, all of whom were involved in various aspects of planning, data collection, analysis, and/or review editing drafts of the manuscript. Several graduate students from Montana State University, including Eugene 0. Allen, Craig J. Knowles, Thomas J. Komberec, Robert G. Trout, Arnold R. Dood, and Shawn J. Riley, also provided assistance and/or data from thesis research studies on and adjacent to the area. Others providing field assistance included Craig Knowles, John Roseland, William Steery, and Al Rosgaard, Jr. Assistance in statistical analysis was provided by Terry N. Lonner, Ray Mule', Bill Hoskins, and Steve Martin. Margaret Morelli typed ii the final manuscript and she and her staff, with great patience, also typed the rough draft and its numerous revisions . A special thanks and acknowledgement goes to our pilots, not only for their superb flying skills, but for their knowledge of wildlife and their enthusiasm for the project, all of which added immeasurably to our accomplishments. These included: helicopter and SuperCub pilot Larry Schweitzer, Larry's Flying Service, Denton; helicopter pilots Murray and Mark Duffy, Central Helicopters, Bozeman; and SuperCub pilot Don Newton, Newton Aviation, Lewis town. We also thank the Roy area landowners and community members for their interest, assistance, friendship, and cooperation through the years. Special thanks go to landowners in the East Indian Butte Grazing District, Perry and Marge Kalal of Kalal ' s Cafe and Service Station (the site of our field station) , and Larry and Sue Kalina, who allowed us to use their field as a landing strip. We thank Martha A. Lonner, MediaWorks, for the figure graphics in this report and Diana Haker for the cover graphic. Throughout, the study was supported by the Montana Department of Fish and Game/Fish, Wildlife, and Parks under Federal Aid in Wildlife Restoration Projects W-98-R (1960-65), W-74-R(1966-68) , and W-120-R and W-130-R (1969-89). Additional support was provided by the U.S.D.I., Fish and Wildlife Service, Charles M. Russell National Wildlife Refuge. This included both direct and indirect contributions to aerial surveys, use of office and field facilities, and field assistance as needed. For this, we are especially grateful to former Refuge Managers Larry Calvert and Ralph Fries as well as many members of the Refuge staff who assisted. Staff of the U.S. D.I. Bureau of Land Management Lewistown District Office also cooperated and provided periodic assistance in deer trapping and other field activities. The cooperation and assistance of Lewistown District Biologist Larry Eichhorn was especially helpful and appreciated. Other support was provided by Montana State University, and by the Montana Agricultural Experiment Station, Montana State University. Again, we wish to thank one and all who have helped us through the years, including some whose names may have inadvertently not been mentioned. Ken Hamlin and Dick Mackie 111 TABLE OF CONTENTS CHAPTER 1. INTRODUCTION AND PERSPECTIVE 1 General Theories of Population Regulation 2 Population Regulation in Ungulates 5 The Concept of Carrying Capacity 7 Inadequacy of Existing Concepts 14 An Alternative Theoretical Framework 16 The Population-Habitat Model 17 CHAPTER 2. APPROACH AND METHODS 23 Early Studies 23 Recent Studies 24 Estimating Population Numbers 30 CHAPTER 3. CHARACTERISTICS OF THE STUDY AREA 42 Location and Physiography 42 Climate 47 Vegetation 52 Forage Plant Production, Abundance, Vigor, and Use 57 Forage Production 57 Forage Abundance 63 Changes in Abundance and Use of Forage Plants 66 Land Use 73 History of the Mule Deer Population 76 CHAPTER 4. POPULATION CHARACTERISTICS 82 Numbers and Density 1960-1987 82 Population Growth Rates and Perspective 1930-1987 85 Population Composition 91 Sex Ratio 91 Age Structure 91 Fawns as a Proportion of the Population 91 Adult Females 95 Adult Males 100 Mule Deer Condition 103 Antler Characteristics 103 Weights 103 Percent Femur Marrow Fat and Kidney Fat Index 106 Age and Condition of Coyote-killed Deer 108 IV CHAPTER 5. REPRODUCTION AND FAWN MORTALITY 110 Conception and Birth Dates 110 Pregnancy, Ovulation, and Fertilization Rates 111 Fawn Production, Mortality, and Recruitment 114 General Patterns in the Population 114 1975-1986 114 1960-1974 119 Age-specific Fawn Production and Recruitment 120 Population Effects of Age-specific Recruitment 123 Numbers 125 Ratios 125 Fawn Mortality Related to Litter Size and Sex 127 Proximate Causes of Fawn Mortality 127 Ultimate Causes of Fawn Mortality 130 Fawn Mortality in Relation to Forage Production --131 Effect of Coyote and Alternate Prey Populations on Fawn Mortality 139 Fawn Mortality in Relation to Winter Severity 146 Fawn Production and Survival in Relation to Density 149 CHAPTER 6. ADULT MORTALITY 156 Mortality Estimates 156 Female Mortality Patterns 156 Male Mortality Patterns 160 Causes of Mortality 161 Females 161 Males 162 Age-specific Mortality 162 Females 162 Males 165 Effect of Home Range Location on Mortality and Productivity of Females 167 Comparative Effects of Hunting on Females and Males 172 CHAPTER 7. SPATIAL AND SOCIAL DISTRIBUTION 176 Spatial Distribution 176 General Patterns 176 Spatial Distribution by Sex and Age 178 Temporal Changes in Spatial Distribution 180 Social Organization and Distribution 183 Group Size and Composition 185 Factors Influencing Group Size and Composition 190 Associations of Genetically Related Individuals — 194 Mother-Son and Sister-Brother Relationships -194 Relationships of Brothers 197 Relationships of Mothers and Daughters 197 Association of Sisters 202 Association of Marked Deer of Unknown Family Relationship 204 Summary and Discussion 205 CHAPTER 8. HOME RANGE AND MOVEMENTS 211 Factors Affecting Home Range Size and Use 213 Factors Affecting Summer Home Range Size of Fawns 216 Monthly Patterns of Movements 218 Emigration 218 Individual Strategies 221 Discussion 233 CHAPTER 9. HABITAT USE AND INTERSPECIFIC RELATIONS 239 Food Habits 239 Seasonal Patterns 239 Summer 239 Autumn 240 Winter 242 Spring 244 Discussion 244 General Patterns of Habitat Use 245 Use and Selection of Vegetation Cover Types 246 Influence of Vegetational Diversity on Habitat Use 251 Influence of Topographic Diversity of Habitat Use-252 Influence of Free Water on Habitat Use 253 Patterns of Habitat Selection by Sex and Age Class 256 Vegetation Cover Types 256 Vegetational Diversity and Topography 258 Distance to Free Water 260 Relationship Between Mule Deer Density and Habitat Selection 260 Seasonal Differences in Habitat Use Among Migratory Deer 261 Mule Deer Habitat Use in Relation to Elk and Cattle 263 Mule Deer and Elk 264 Mule Deer and Cattle 270 Effect of Cattle and Elk on Mule Deer Population Performance 271 Summary and Discussion 275 CHAPTER 10. POPULATION-HABITAT RELATIONSHIPS 278 Mule Deer-Forage Interactions 278 Forage Quantity 278 Forage Quality 283 Contribution of the Fixed-Stable Habitat Base to Population Dynamics 284 Discussion 289 VI CHAPTER 11. POPULATION DYNAMICS 292 Variation in Reproductive Rate 292 Variation in Mortality Rate 294 Fawn Mortality 294 Mortality of Adult Females 301 Mortality of Adult Males 307 Emigration and Immigration 308 Interaction of Reproduction and Mortality 311 Summary 316 CHAPTER 12. POPULATION REGULATION 319 CHAPTER 13. MANAGEMENT IMPLICATIONS 331 Management Philosophy and Theory 331 Management Concepts 332 General Approach to Management 338 Population Evaluation and Management 338 Harvest Management 338 Predicting Fawn Survival 344 Predator Management 345 Population Evaluation 348 Population Estimates 348 Age Structure and Sex and Age Ratios 349 Interpreting Population Trend 351 Habitat Evaluation and Management 353 Essentials of Mule Deer Habitat 354 Habitat Maintenance, Enhancement, and Development-358 APPENDICES 361 Appendix A. Arithmetically modeled population estimates for mule deer on the Missouri River Breaks study area, 1960-1987 361 Appendix B. Table Bl. Method of calculating estimated forb production on the study area using 1982 as an example 365 Table B2 . Method of calculating estimated shrub production on the study area using 1982 as an example 366 Appendix C. Common and scientific names of plants for which abbreviations are occasionally used in tables 367 Appendix D. Excerpts from quarterly narratives of CMRNWR personnel and DFWP reports, 1940-1959 368 LITERATURE CITED 378 VII CHAPTER 1 INTRODUCTION AND PERSPECTIVE Mule deer (Odocoileus hemionus) are the most widely distributed and abundant of all species of large mammals native to western North America. Important populations occur throughout a broad spectrum of habitats or environments— from relatively moist and dense coniferous forests to dry, open plains and deserts. The quality of these habitats for deer varies greatly as a result of broad differences in physical and biotic characteristics. Many are highly variable environments in which habitat quality also fluctuates seasonally and annually due to variation in weather conditions. Mule deer habitat relationships and population characteristics and dynamics, as well as management opportunities and constraints vary accordingly. However, long-term studies addressing spatial and temporal differences in population ecology and their significance in management are lacking. Our studies of mule deer in the Missouri River Breaks of northcentral Montana have provided a 28-year data set on ecology and behavior of the species in a representative, historically important and environmentally variable habitat. The studies began with investigation of range ecology and relationships of mule deer, elk, and cattle on the area during 1960-1964 (Mackie 1965, 1970). This provided basis and methodology for continuing population surveys to determine mule deer population characteristics and trends in relation to range conditions through the early 1970s (Mackie 1973). A marked decline in mule deer populations and low fawn recruitment despite seemingly favorable range conditions during 1973-1975 brought support for intensive studies of population ecology and factors regulating deer populations on the area during 1975-1987. Our first priority was to provide knowledge for management of mule deer in "breaks" habitat. However, the results also may be important from a more theoretical point of view. Few, if any, serious ecologists would deny that existing knowledge of population ecology has many shortcomings. Despite myriad theoretical, experimental, and field studies, questions and confusion about pattern and process in population organization, population-habitat relationships, and population dynamics and regulation persist. Reasonably accurate long-term data on population ecology of deer and other large mammals are rare. Thus, Caughley (1980) notes that, aside from McCullough's (1979) study of white-tailed deer in the George Reserve, Sinclair's (1977) study of African Buffalo, the long-term studies of moose and wolves on Isle Royale, and the work of Laws and his associates on elephants in east Africa, "most of the other studies that have been done coalesce into an amorphous mass of nothing much. " We believe our data set is sufficiently complete and accurate to show not only what the mule deer population has done, but why, and to address broader questions relating to issues such as carrying capacity, population regulation, yield theory, predation, and habitat selection. Perhaps somewhat more specifically, we can also address the extent to which "...the principles derived [from studies of white-tailed deer on the Edwin S. George Reserve in southeastern Michigan] are general and should apply to most large, long-lived mammals — in modern ecological parlance, K-selected species" (McCullough 1979) . We also attempted to answer age-old questions about population regulation. What determines the size of a population? Why do certain numbers occur at one time and place and other numbers at different times and places? The mule deer population we studied did not increase without limit nor did it decline to extinction. Yet, population regulation in its traditional sense is not a proper phrase for the dynamics we observed. Our results and conclusions concern population ecology and all its attendant interrelationships. The total environment of the population and its inherent heterogeneity and dynamics must be considered to understand population dynamics. We concur with Ratcliffe (1958) that animals in a population cannot be regarded simply as l/N. Individual differences in animals and the heterogeneity of their common and individual environment in time and space are all important elements of population ecology and "regulation". General Theories of Population Regulation Ecological literature of the last 50 years is replete with books and articles by proponents of various theories of population regulation. These include many syntheses, reinterpretations, and reviews (e.g. Milne 1957, Krebs 1972, and Murray 1979). Because no one theory or school of thought has been universally accepted as yet, a brief review of previous and current theories of population regulation, generally and among ungulates, is necessary to place our study and conclusions in perspective. The primary problem or question related to population dynamics and regulation that theorists and empiricists have tried to answer is composed of 2 parts. Krebs (1972: p. 269) pointed this out when he stated: "One can make two fundamental observations about populations of any plant or animal. The first is that abundance varies from place to place . . . The second observation is that no population goes on increasing without limit, and the problem is to find out what prevents unlimited increase in low-and high-density populations. This is the problem of explaining fluctuations in numbers . " The observation that no population increases without limit and that substantial destructive or debilitating forces were necessary to hold numbers in check was apparently the original impetus behind the study of population regulation. Malthus (as quoted by Davis 1950) believed that the ultimate check was food, but the immediate check was "all those customs and diseases generated by a scarcity of the means of subsistence". Darwin (1872: p. 54) observed: "There is no exception to the rule that every organic being naturally increases at so high a rate, that if not destroyed, the earth would soon be covered by the progeny of a single pair." Although the works of Malthus and Darwin inspired the search for methods of population regulation, most existing theories were framed by the work of entomologists and a few fisheries scientists from the early 1900s through the 1950s. Existing theory about pattern and process of population regulation has generally developed around 4 schools of thought. One of the earliest and, in some form, perhaps that most widely held today, is often called the biotic school. Although most commonly associated with Nicholson (1933), Howard and Fiske (1911) had earlier promoted much the same views. Nicholson believed that there was a balance in nature. In order to achieve a state of balance, he stated it was essential that a controlling factor should act more severely at high densities and less severely at low densities. He suggested that the chief "density-dependent" factor controlling population level was intraspecif ic competition for resources. Mathematical expression of this theory was accomplished (Nicholson and Bailey 1935), but its assumptions about the animals (l/N) and especially the environment (constant) were viewed as unnatural by some (Milne 1957). A second school of thought developing during the same period was that weather or climate was responsible for control of populations (Uvarov 1931). Nicholson attempted to define this school out of existence by stating that, because climate was not affected by population density, it could not be a controlling factor. He was really saying that weather couldn't be a controlling factor because it did not meet his assumption that only factors affected by population density could control populations. Apparently, the possibility that the effect of weather can vary with density was not considered important. The density-dependent school of thought already was so entrenched that the circular reasoning behind ruling weather out as a controlling factor aroused little comment. A "comprehensive" school of population regulation also developed (Thompson 1929, Andrewartha and Birch 1954). Theories developed therein held that at some time, all factors were involved in population control. In contrast to the "balance" emphasized by Nicholson, these authors emphasized the constant fluctuations of populations and indicated that populations were controlled by the total environment. The importance of weather or climate as a major regulating factor is usually apparent in the models of this "school" (Birch 1957). Milne (1957: p. 260) summarized the view of Thompson as follows: "where and when the organism meets favorable conditions its numbers increase, but these conditions never endure long enough for unlimited increase; nor do unfavorable conditions endure long enough for a decrease of numbers to zero— except in places that are in any case unsuitable for the organism." This view was essentially the same as that of Andrewartha and Birch (1954) in that the most important way in which animal populations are limited is "by shortage of time when the rate of increase, r, is positive." Milne's (1957: p. 265) own theory generally agreed with the comprehensive school, though he revived density related terms: "For the most part, control of increase is due to the combined action of (a) density-independent and (b) imperfectly density- dependent environmental factors." Milne believed that only rarely did the perfectly density-dependent factor of intraspecif ic competition play a role in population control. He also believed that a balance occurred in the environment, not between populations and the environment; that is, over time and space, favorable environmental conditions were balanced by unfavorable environmental conditions . A more recent comprehensive view (Huf faker and Messenger 1964) held that density-dependent processes control populations in favorable environments and density-independent factors control populations in unstable or marginal habitats. These earliest theories centered on the role of extrinsic factors in population regulation: food, predators, weather, disease, shelter, etc. ( Krebs 1972). Intrinsic factors and individual differences in animals were largely ignored. They came to the fore, however in a new school of thought, developed in the mid- to late- 1950s, which emphasized individual differences in animals and proposed that intrinsic changes in the population could lead to self -regulation (Chitty 1955, 1960) . Proponents of self regulation believed that uncontrolled competition for resources could ultimately lead to destruction of the food base and extinction. Because most populations do not decline to extinction, it was argued that some form of self regulation must hold animal population levels below those where resource destruction occurs. Natural selection among individuals and evolutionary changes in a population are important concepts in self regulation theories. As population numbers rise during favorable periods, more individuals of "lower quality" or viability survive. Thus, the effect of weather can be "density-dependent", being more severe at high populations with more "low quality" individuals. Because the genetic composition of a population can vary over time and with conditions, the "self-regulation" process could work through a genetic feedback mechanism (Pimentel 1961). Variations on the intrinsic (self-regulation) school of thought include the effects of: 1.) behavior, e.g. territoriality (Wynne-Edwards 1965); 2.) physiological change resulting from stress at high densities (Christian and Davis 1964); and 3.) the Chitty Hypothesis (Chitty 1967, Krebs 1978) that population regulation can result from genetic-based changes in spacing behavior. Although theories and variations of theories on population regulation have come and gone, most widely held views include the major theses of Nicholson (1933) and Smith (1935). These imply a "balance of nature" and that factors controlling populations, whether intrinsic or extrinsic, act in a density-dependent manner. Population Regulation in Ungulates Until recently, most theoretical and empirical work on population dynamics and regulation was confined to insects, fish, small mammals, and to some extent birds. However, the major "principles" derived were believed general and applied in practice to ungulates. It was only during the mid-to-late 1970s that results of long-term studies on large mammals began to appear (Laws et al . 1975, Sinclair 1977, Peterson 1977, McCullough 1979, Botkin et al . 1981). Although these studies have raised many questions and stimulated discussion, they have not resulted in major new theories of population regulation. Peek (1980) summarized current thinking on natural regulation of ungulates as reflecting 1 of 2 perspectives; an ungulate-habitat interaction and a predator-ungulate interaction. The ungulate-habitat interaction model is most widely held and applied. It is essentially Nicholsonian in that density-dependent intraspecif ic competition for resources is the main regulating factor. Especially for North American deer (Davis 1950, Lack 1954, Dasmann 1971), this came to mean that they were forage-limited. It was long believed that deer were not territorial (Smith 1976, Coblentz 1977, Peek 1980), nor did they have any other intrinsic mechanisms of control. Thus, in the absence of predators, they were inherently irrupt ive, overused their forage resources, and temporarily or permanently reduced the carrying capacity (K) of their range. Caughley (1976a, 1979) modified this somewhat by indicating that generally an equilibrium population results from the dynamics of ungulates interacting with the dynamics of the plants on which they feed. Populations of both herbivores and plants fluctuate and each affects the other. Ungulates are the major or only factor affecting forage production in these models, at least in the way they are generally presented. For wildlife managers, the concept of deer irrupting, exceeding their forage base, and then abruptly declining to lower levels probably has its roots in the Kaibab deer incident (Rasmussen 1941). In textbooks and classrooms, the Kaibab story is the "classic" example used to indicate that, at least in the absence of predators, deer will "overuse" their forage and reduce carrying capacity. Interpretations have varied in their emphasis on the importance of predation and interspecific competition (livestock grazing) in the process, but the Kaibab story has generally been viewed as a typical example of intraspecif ic competition and delayed density-dependent mortality (Lack 1954). More recently, Caughley (1970) questioned the reliability of the data on the Kaibab deer herd and Burk (1973) termed the Kaibab incident "a myth. " The predator-ungulate interaction theory of ungulate population regulation (Peek 1980) was supported by indications that most irruptions of ungulates took place in the absence of predators. This theory holds that, in the presence of predators, herbivore populations seldom approach limits imposed by food. Hairston et al. (1960) and Slobodkin et al . (1967) provided a theoretical framework arguing that predation regulates the dominant members of the herbivore trophic level. Recent work (Bergerud 1971, Keith 1974, Beasom 1974, Bergerud 1978, Stout 1982, and Gasaway et al . 1983) indicated that, at least in some circumstances, predation can have a significant impact on ungulate population dynamics . According to Gasaway et al . (1983): "One of the most significant lessons to be learned from the patterns observed in Minnesota, Alaska, and on Isle Royale is that a combination of controlling factors can initiate a decline in ungulates, which may then continue to very low densities as wolf predation becomes an increasingly important controlling factor." The predator-ungulate interaction model is basically a modification of the ungulate-habitat interaction model, and most adherents of the former seem to believe that in the absence of predators, the latter is applicable. There are also those, however, who believe the predators cannot regulate populations even when they are present and abundant; thus, the ungulate-habitat interaction model is always applicable. Multiple equilibrium models (Haber 1977) combine the 2 other models and hold that predators are important at low ungulate densities but that at high densities, ungulates escape the influence of predators and are influenced by density-dependent intraspecif ic competition for resources. Because density-dependent intraspecif ic competition for resources has been assumed to be major factor regulating ungulate populations, human predation (hunting) is often neglected in studies of factors controlling populations. Management theory usually assumes that hunting losses are compensated for by decreased mortality and increased reproduction and recruitment of young among survivors. Recently, some authors (Geist 1981, Ozoga et al . 1982) have suggested that behavior may be more important than previously realized in determining deer population size, structure, and dynamics. However, these views have subsequently received little attention or comment by others. The Concept of Carrying Capacity Because most applied management theory for ungulates is based on the ungulate-habitat interaction model, a discussion of population ecology and regulation necessarily includes discussion of the carrying capacity concept. In general, "carrying capacity" refers to the number or density of animals a given habitat/environment can support. A good review of the concept, its use, and interpretation is provided by Macnab (1985). The terminology and components included in specific definitions vary widely, but animal-vegetation (forage) interactions similar to those underlying range-domestic livestock relationships and livestock production pervade most interpretations applied to deer and other wild ungulates. Thus, Macnab (1985) notes "For range management, the density of cattle providing maximum sustained production of beef is the carrying capacity of the land." More generally, Caughley (1979) states "Carrying capacity is the name we give to an equilibrium between animals and vegetation, and we index the position of that equilibrium by its characteristic density of animals." Such definitions would also indicate or at least imply that carrying capacity is a more or less finite entity, varying only in relation to changes in vegetation/forage supplies induced by animal-plant interactions . Alternatively, some theoretical definitions have recognized that "carrying capacity" may be determined by many factors and may vary considerably. Foote (1971), for example, defined carrying capacity as "The sum of the environmental factors which make a game range habitable," and further stated that "Carrying capacity changes from season to season on the same range." This complexity and the variable nature of carrying capacity was also pointed out by Edwards and Fowle (1955) when they stated: "The really important thing is to recognize that carrying capacity is not a stable property of a unit of environment but the expression of the interaction of the organisms concerned and their environment. Moreover, the carrying capacity of an ecosystem may fluctuate in response to the ebb and flow of interactions going on within it." Sinclair (1981) similarly noted "A population at carrying capacity should not be thought of as one with a stable or constant level. Rather, it is one that is fluctuating, often extensively, between certain boundaries." Hobbs et al . (1982) have empirically shown extensive annual fluctuations in energy-and nitrogen-based estimates of carrying capacity for elk (Cervus elaphus nelsoni) winter range. Following the concept that carrying capacity reflects the availability of forage resources and forage-animal interactions, different definitions and terminology have been proposed or applied, based on the management objectives for a given population. The maximum or equilibrium number of animals that can be supported under natural conditions (in a non-hunted area) has been variously termed subsistence density (Dasmann 1981), environmental carrying capacity (Clark 1976), K carrying capacity (McCullough 1979), potential carrying capacity (Riney 1982), and ecological carrying capacity (Caughley 1979). By definition, the rate of increase of both plants and herbivores at K carrying capacity is zero (Macnab 1985). "Equilibrium" is reached through a sequence of increased juvenile mortality, increased age of sexual maturity, decreased birth rate among adults, and increased adult mortality (Macnab 1985). For managed populations in which harvesting occurs, following the concept of maximum sustained yield (MSY) , other definitions and terminology have been proposed and applied to "carrying capacity. " This position on the continuum of possible ungulate/vegetation densities (Macnab 1985) has been called optimum density (Dasmann 1981), "I " carrying capacity (McCullough 1979), and economic carrying capacity (Caughley 1979). It assumes a lower standing crop of ungulates and a higher standing crop of vegetation (a lower practical or realized carrying capacity) than K carrying capacity. Also inherent is the assumption that with a higher standing crop of forage, the animals are more productive and a higher annual yield or harvest is available (MSY) . "I" carrying capacity is the definition generally recognized by range managers for domestic livestock. When applied to wild ungulate populations and combined with the concept that a high proportion of climax forage species is good, this definition has led to heavy cropping of ungulates in many locations. In commenting on this application, Sinclair (1981), citing Caughley (1976b), stated "Range managers have misguidedly used these plant criteria (decreasers, increasers, and invaders) to determine whether herbivores were overabundant or not by declaring arbitrarily that a high proportion of decreasers was good while a high proportion of invaders was bad." He (Sinclair 1981) added "Unfortunately, this led to management decisions that there were always too many animals (the reason for 30 years of elk culling in Yellowstone [National Park]). These criteria result in the ridiculous conclusion that the only good herbivore population is one vanishingly small." For practical purposes, intraspecif ic density-dependent competition for a finite forage resource is the major consideration inherent in most definitions of ungulate carrying capacity. Despite giving at least some "lip service" to other possible environmental variation, the models and graphs of most theorists display a herbivore-forage interaction where the only factor influencing forage production is the degree of use by the herbivores. Models proposed by Caughley (1976b, 1977, and 1979) and McCullough (1979) are typical of widely held views assuming a more or less determinate carrying capacity. Generally, competition for other resources (space, cover, water) has not been considered for ungulates. The typical density-dependent logistic model (Fig 1.1A) indicates that recruitment rate declines linearly as population density approaches K. Yield, in terms of number of animals that may be harvested while maintaining a stable population, typically peaks around 0 . 5K (Fig 1.1B). In theory, this occurs because as population numbers approach K, the quantity and quality of food available per capita decreases. Lag effects of a population increase may result in "overuse" of the forage, thereby lowering K. Fowler (1981) has modified typical logistic models by indicating that most density-dependent change for large mammals occurs at population levels quite close to carrying capacity (Figs. 1 . 2A and 1.2B). Nevertheless his models in this form retain a fixed carrying capacity (K) influenced only by the herbivore . In fairness to the modelers, we must point out that, to illustrate a point, they often present a simpler model than they know to be the case. Most realize that environmental CO c E o DC 3 DC O d z i- o 0) >- B. 0.5 k Population Numbers (N) 0.5 k Population Numbers (N) Figure 1 . 1 Typical density-dependent logistic models for recruitment (A) and yield (B) in relation to animal numbers as an assumed stable carrying capacity (K) is reached. CO CE C o> E o o> cr o >- B. Population 0.5 k Numbers (N) — K 0.5 k Population Numbers (N) Figure 1.2 Modified density-dependent recruitment (A) and yield (B) animal numbers as an assumed capacity (K) is reached. These that most near K. models for in relation to stable carrying models assume density dependent change takes place 10 conditions fluctuate independently of animal density and that these fluctuations invalidate simple models. Thus, Caughley (1977, p. 197) states: "The methods previously outlined for estimating MSY can cope with only a moderate degree of year-to-year fluctuation in a population's condition of life. They will be adequate for most populations. Some populations, however, have a boom and bust economy, increasing when conditions are favorable and crashing to low levels when the environment deteriorates. In this category can be placed most of the species of oceanic fish, and several vertebrates of the northern tundra. The simplifying assumption of a steady density, which at best is a dangerous abstraction, almost completely parts company with reality when applied to a boom and bust population. ...In these circumstances density and resources are seldom balanced and one cannot be predicted with confidence from the other." That caveat was further emphasized by Macnab (1985) in stating "First, in extreme environments (e.g. arid and semi-arid regions) vegetation growth and the recruitment and mortality of large herbivores are highly variable, so that the "equilibrium" is more a mathematical abstraction than an operational reality. . . .This changes none of the theory, but makes management more difficult." Similarly, McCullough (1984:p. 224) indicated that his deterministic model for the George Reserve would not work well where "environmental stochasticity is great relative to the density-dependent response of the environment." McCullough (1984) also stated that "one of the major problems confronting the wildlife biologist or resource manager is the year-to- year variation in habitat quality due to variable amounts of precipitation, severity of winter, presence or absence of an acorn crop, etc." In environmental variable areas, an ad hoc strategy is necessary to annually adjust management in response to environmental factors. However, McCullough (1984:p. 225) stated that "Fortunately, most white-tailed deer populations occur in relatively benign environments and respond well to density-dependent management." Based on findings from our study, we generally agree with the cautionary words of these authors. However, we disagree when they indicate that the assumption of relatively stable environmental conditions applies to most populations. Rather, it seems likely that simplified models may work only for absolutely stable environmental conditions and become increasingly dangerous to apply as variation increases. What 11 may seem like minor fluctuation becomes magnified upon long-term application of the principles derived from the simple models. Therefore, the simplified models may work for almost no populations. This is particularly true for ungulate populations in arid, semi-arid, or mountainous regions of North America. Underlying assumptions of fixed forage resources influenced mainly by the herbivores and density-dependent dynamics resulted in the "principles" of compensatory mortality and reproduction. Under this view, reducing the herbivore population below the level of maximal use of forage resources would result in better survival and production by those remaining. The extreme degree to which the "principles" of compensatory mortality and reproduction, density-dependence, and a fixed forage base were carried is indicated by Dasmann (1971) who stated that at the maximum level (ecological carrying capacity or subsistence density) , "one deer must die to make room for another". Often, forage resources during winter was considered to be the main limiting factor. For example, Leopold (1966), wrote "The quality and quantity of forage available to a deer population during the most critical season of the year has proven repeatedly to be the basic regulator of population level. Usually this means winter forage, but not always." The importance of the concepts of forage carrying capacity and especially winter range carrying capacity are illustrated in a typical example (Fig 1.3, Cole 1961) of the way the theory of density-dependent population dynamics and the "principles" of compensatory mortality and reproduction were applied to management of ungulates. Although this management model is simplified, it was essentially the program implemented throughout the United States . Deer needed to be reduced below the "carrying capacity" of their range by hunting. This reduction would result in greater quantity and quality of food for the survivors, thereby increasing their productivity. The reduction of deer numbers below "carrying capacity" would allow the vegetation to "recover", thus actually increasing "carrying capacity". For many years, there was no indication that this model was working. However, this was generally attributed to the belief that adequate harvests had not been achieved or that vegetation took a long time to recover following overpopulation by ungulates (lag effects). Although the model did not appear to be working on many areas, it continued to be widely applied over a broad spectrum of habitats (see McCullough 1979: p. 169), perhaps because no suitable alternative was proposed. 12 100 Fawns/ 100 Dots JUNE OCT DEC-APR JUNE OCT DEC-APR JUNE OCT POPULATION LEVEL / \ HARVEST / \ HARVEST / Rang* Carrying Capacity Deer population maintained at winter range carrying capacity by hunter harvests being equal to yearly fawn production B JUNE OCT DEC-APR JUNE OCT DEC-APR JUNE OCT POtLEVEL°N/'*~\ \ HARVEST 25 Fawns/ 100 Does 1 \ Winter Loss^ 5Q Fawns ! 199 Don \ / % '■\ vWintor Loss Or y" Reduced carrying capacity due to l 1 \ overuse during previous winter Deer population above winter range carrying capacity and declining due to inadequate hunter harvests JUNE OCT DEC-APR JUNE OCT DEC-APR JUNE OCT 25 Fama ntt DQ9S 100 Fawns 1 100 Does , ~ c 3 - - HIGH C g Q. 2 - < B > MEAN o -J + 1 - ^LOW ri - A Mule Deer Population Location Figure 1.5. Mean and range of densities for 3 Montana mule deer populations. A=Cherry Creek (Wood et al. 1989); B=Missouri River Breaks (this study); C=Northwest slope of Bridger Mountains ( Pac et al . in prep. ) density-dependent intraspecif ic competition for forage could have been important . Our data also indicated that "territoriality" of females during parturition (also see Ozoga et al . 1982 for white- tailed deer) functioned to determine home range establishment for the entire population. This "territoriality", at times, also resulted in emigration, which helped limit population growth rate. Because the environment was heterogeneous and habitat quality varied across the area, home range location (space) and use influenced recruitment and mortality rate of individual animals. During "good" conditions, deer survived and recruited fawns in "marginal" areas. During "poor" conditions, deer with home ranges in marginal areas not only failed to recruit fawns, but were subject to higher mortality 21 rates than deer in "core" areas. Recruitment and mortality were properties of individual animals in individual home ranges . Occasionally, mortality was high when density was high. This was not really the result of density per se, but the result of density-independent changes in energy gain and loss related to variable weather patterns. Seasonally and annually, variable weather modified habitat quality, structure, and its ability to support deer. When weather conditions were conducive to deer living in "marginal habitats", survival remained high despite high deer density. Also, few deer lived in marginal areas unless density was high. Thus, when density was high, more deer were vulnerable to density-independent changes in weather. "Carrying capacity" fluctuated widely and annually, usually as a result of density-independent factors (especially weather). Our model of population dynamics may be unsatisfactory to some, especially those who would prefer to "force" ecological events into mathematical formulae. Although we did not observe "classic" density-dependent population responses , they may be relatively more operational in stable environments and those lacking predators or dispersal sinks. However, there are many other types of populations and situations that must be dealt with by managers. Often, knowledge of what proximate factors are affecting populations is more important than what the ultimate regulating factor is under precisely specified and fixed conditions. We most hope to convince the reader that dogmatic, deterministic, or fixed views are often not helpful or widely applicable. A framework of general principles may apply overall, but interpretation and application of those principles must be made within the context that each population is a unique product of its total environment . 22 CHAPTER 2 APPROACH AND METHODS Early Studies The data on which this report is based were derived through a series of studies of broadly different objectives and intensity, encompassing a variety of methods. In this chapter, we present a general overview of our approach and the methods employed in obtaining and analyzing the major data sets. More detailed information is provided by references cited or in presentation and analyses of specific data. Generally, the studies comprised 3 phases: (1) intensive studies emphasizing range use and relationships of mule deer, elk, and cattle during 1960-1964 (Mackie 1965, 1970); (2) extensive studies focusing primarily on mule deer population characteristics and trends in relation to range condition from 1964 through 1974 (Mackie 1973, 1976); and (3) intensive studies of mule deer population ecology and habitat relationships during 1975-1987. The early studies provided baseline data on habitat characteristics of the study area, distribution, movements, use of specific habitat/vegetation types and topographic/physiographic sites, activity patterns, food habits, and population characteristics and trends for mule deer from 1960 through 1963. These studies were primarily descriptive. Data on vegetation were obtained through classification and measurement of vegetational characteristics of habitat/vegetation types, and annual measurement of utilization and condition trends for major shrub species browsed by deer on permanent transects. Data on deer were obtained by systematic observations along vehicle routes throughout the area, recording plant species eaten at feeding sites, analyses of ruminal contents of deer collected and shot by hunters, and conducting checking stations and field check of hunters to determine distribution, numbers, sex and age composition, and condition of deer harvested (Mackie 1970) . Data on population size and characteristics were obtained primarily through counts and classifications of deer observed along the vehicle routes and from harvest information. Aerial surveys and observations were limited and provided only supplementary data on range use and population characteristics until winter 1963-64 when the entire study area was surveyed using a helicopter to count and classify deer and elk. Only 16 mule deer and 3 elk were marked, none with radio collars. 23 Comparisons of ground and aerial counts during 1960-64 indicated that helicopter surveys, flown for complete coverage following snowfall in early winter when most deer used open ridgetop habitat, provided an effective method for determining numbers and sex and age composition of mule deer on the study area. Because of a need for continuing information on population characteristics and trends in management of mule deer in breaks habitat, and termination of intensive studies, the helicopter surveys became the primary method of study during the second, extensive phase. Early winter surveys were conducted in most years from 1964 through 1975. Exceptions included 1968, 1969, when classification counts were made by the area management biologist with a fixed-wing aircraft, and 1972, when mild, dry weather precluded flying until late winter. Overall, these data provided estimates of population size and composition and trends in relation to range/environmental conditions throughout the period (Mackie 1973, 1976). Measurement of utilization and condition trends of major browse species was also continued in spring of each year on established transects. In addition, 5 permanent point-center-quarter transects measuring shrub composition, density, and plant size were established on the area during 1971 as part of a study of ecological characteristics of fragrant sumac in Montana (Martin 1972). Mule deer harvests were monitored in a limited manner through field reconnaissance during hunting seasons and the statewide harvest survey for the hunting district which includes the area. During 1972-1975, 2 graduate thesis research studies were completed on our supplementary Nichols Coulee Resource Conservation Area (NCRCA) study area (Knowles 1975, Komberec 1976). Designed to evaluate range relationships of mule deer and elk within a rest-rotation grazing system, these studies included aerial population surveys similar to those conducted on our primary study area and use of marked and radio-collared deer in studies of habitat use. Recent Studies The third phase, beginning in 1975, was precipitated by the failure of existing knowledge and theory about mule deer habitat relationships and population dynamics to explain population phenomena observed on the study area and elsewhere. Our study area and the 15 years of baseline data available provided exceptional opportunity to conduct the intensive, long-term population studies we believed necessary to develop an improved understanding of the process and mechanics of 24 population regulation in mule deer. In undertaking those studies, we also believed it necessary to closely examine and perhaps rethink current knowledge of basic mule deer population biology and ecology. Thus, we did not attempt to test specific hypotheses; nor did we employ the rigid technigues used in much scientific hypothesis testing. Instead, we approached our intensive studies more in the manner of natural historians (Bartholomew 1986, Greene 1986), with some direction but also relying on "serendipitous and unexpected" results (Greene 1986). In taking this approach, we focused heavily on obtaining as complete and detailed data as possible on population ecology. Thus, much depended on our ability to obtain detailed and reasonably accurate data on mule deer population characteristics and trends. Beginning in 1975, complete-coverage surveys employing both helicopter and fixed-wing (Piper SuperCub) aircraft were flown during autumn and spring as well as in early winter. From 1976 through 1983, fixed-wing surveys were also conducted in July. During all surveys, numbers, sex, age, group size and composition, location, and other data were recorded for mule deer and most other major species observed. As marked deer became available, observations of those deer also were recorded. During summer, autumn, and early winter surveys, deer were classified as fawns (<12 months of age), adult females, yearling males (12-24 months of age), and mature males (>24 months of age) . Yearling and mature males were distinguished by gross differences in antler size based on experience in aging and measuring antlers of hunter-killed deer. We did not attempt to distinguish yearling from mature females. Deer were classified only as adult or fawn during spring. During 1975-1986, 416 different deer were captured and marked with individually-recognizable neckbands or radio collars. Sixteen of those were recaptured and recollared once, 2 were recaptured and recollared twice. Overall, 202 were first captured as newborn fawns, 71 were 6-8 months of age, 119 were adult females, and 24 were adult (>18 month) males. Twenty seven of the newborn fawns (14 males and 13 females) and 19 (9 males and 10 females) marked at 6-8 months were offspring of marked does. Ages were assigned to all deer older than newborns on the basis of tooth replacement and wear criteria (Robinette et al. 1957). Other procedures followed in handling deer and recording data have been described by Hamlin et al. (1982) and Riley and Dood (1984). Newborn fawns were located from the air and captured by ground crews directed to the site by radio (Riley and Dood 1984). One hundred seventy-eight were fitted with 25 radio-transmitter collars and numbered metal eartags ; 24 were marked only with eartags or eartags and a vinyl earflag. Most deer over 6 months of age were captured using a drive net (Beasom et al . 1980); a few were captured with cannon nets (Hawkins et al . 1968) and a hand-held net gun (Barrett et al . 1982). We marked 89 with radio-collars, 124 with 10 cm-wide, individually recognizable neckbands, and 1 with eartags only. Seven fawns, radio collared as newborns, were later recaptured; 2 were refitted with radio collars and 5 with individually recognizable neckbands. Two adults, originally marked with neckbands, were equipped with radio collars upon recapture. Four adult females had their radio collars replaced once, 2 others had theirs replaced twice. The effective life of radio collars varied, but most fawn transmitters functioned for 1-2 years while those on adults lasted 2-5 years. During most of the study, we attempted to relocate all radio-collared deer at least twice monthly using a PA-18 Piper Super Cub with antenna mounted on the wing strut. Many were relocated more frequently. Some relocations, especially of newborn fawns which we attempted to locate at 2-3-day intervals, were made from the ground. Most relocations were made during the period one-half hour before sunrise to 2 hours after sunrise. A minor number of relocations were made during mid-day and evening hours, but no relocations were made during periods of darkness. Visual observations of the deer were made on approximately 83% of aerial radio-telemetry relocations. Relocations of radio-collared as well as neck- banded deer were also recorded from observations during the course of other fieldwork, including aerial population surveys and flights associated with efforts to locate and capture fawns in June or other deer during winter. Occasionally, special aerial searches were conducted to attempt to locate neckbanded deer that had not been observed for some time. All relocations were recorded to the nearest 3 . 2 ha from gridded aerial photographs. Social groupings and associations, numbers of f awns-at-heel with marked females, and other pertinent data, including habitat use during some years, were recorded when possible. Locations and other data were also recorded for all unmarked deer as well as for other major species observed during relocation flights. Radio-collared coyotes (Pyrah 1984, Hamlin et al . 1984) and radio-collared and neck-banded elk on the area were monitored; data were recorded in the same manner as for deer. We obtained 9,841 observations of 354 marked deer during 1 June 1976 - 31 May 1986. In addition to population estimates, relocations of marked deer provided data for estimating seasonal and annual reproduction, recruitment, and 26 mortality rates, both generally, and specific to sex and age classes of deer. They also provided information on movements and home range, social structure, and behavior. Biological materials and physical measurements were obtained from deer killed by hunters and coyotes, trapping mortalities, special collections, and deer found dead on the area. When possible, sex, age, location, and cause of death were recorded. Age was assigned on the basis of tooth replacement and wear, and later confirmed or adjusted by cementum analysis (Gilbert 1966) of an incisor extracted from deer 2 years of age and older. When available from carcasses, kidneys and attached perirenal fat were collected and a kidney fat index (KFI) was calculated following Riney (1955). Femur fat samples were taken from collected and coyote-killed deer; color and consistency was recorded and percent fat content was measured using the reagent-dry assay technique (Verme and Holland 1973). Data on animal condition and trends were also obtained by recording field-dressed weights and antler measurements (French et al . 1956, Robinette et al . 1973) for samples of deer killed by hunters. Reproductive tracts were collected from females taken in special collections. Incidence of ovulation, corpora lutea of pregnancy, and fertilization rates were determined (Cheatum 1949). Fetal ages were assigned from crown-rump measurements based on growth rates of known-age fetuses (Hudson and Browman 1959). Further data on reproductive potential were obtained during fawn capture operations and from fawn-at-heel ratios recorded for all productive females during June (Hamlin et al . 1984). Pregnancy rate also was estimated by recording the appearance and behavior of all females observed during mid June (Ozoga et al . 1982, Hamlin et al . 1984). Age-specific fawn production and survival was evaluated using fawn-at-heel ratios and percent pregnancy among individually marked females of known or assigned age (Hamlin and Mackie 1987). Seasonal and annual mortality rates were calculated from several data sets. When available, mortality rates of marked deer were used. Fawn mortality rates were estimated from deaths in a radio-collared sample and changes in fawn: female ratios recorded in seasonal population surveys. Differences between seasonal and annual population estimates also provided estimates of mortality rates. Information on relative rates and composition of hunter harvests for the hunting district including the study area were available from statewide questionnaires and telephone surveys (Cada 1985). Additionally, hunter check stations and field checks were conducted during 1960-1965 and 1976-1987 to determine relative numbers, sex, age, and condition of deer harvested on our study area each year. 27 The locations of deer and other species observed during full-coverage aerial surveys, radio-relocation flights, and other operations were recorded as the mid point of a 3.2 ha grid block. Those locations were transformed to Universal Transverse Mercator (UTM) coordinates for analysis of distribution, habitat use, and home range and movements. Seasonal and annual home ranges and movement patterns as expressed by the minimum convex polygon (Mohr 1947) and average activity radii (Hayne 1949), were calculated using TELDAY software (T.N. Lonner and D.E. Burkhalter, Users manual for the computer program TELDAY, Mont. Dept. Fish, Wildlife and Parks, unpubl . ) . Habitat use was evaluated using a block (cell) -analysis technique (Porter and Church 1987, Wood 1987) in which use was compared with availability (Byers et al. 1984) using Chi-Square tests (Everitt 1977). Analyses of distribution, movement, home range, and habitat use as well as social organization included only data obtained through 1984 when field studies were reduced to focus on data analysis and reporting. Comparative data on population characteristics and trends for adjoining breaks habitat were obtained by special aerial surveys and continuing the full-coverage, helicopter and fixed- wing surveys on the nearby Nichols Coulee Resource Conservation Area (NCRCA) study area. The NCRCA, located north of the Missouri River, 3.2 km northeast of our primary study area, served as a comparative study area for mule deer. This area, described by Knowles (1975), is also "River Breaks" habitat, but is somewhat more arid, almost totally lacks the Douglas fir (Pseudotsuga menziesii) vegetation type, and is physiographically characterized by major drainages that are steeply cut in a north-south orientation such that steep slopes adjacent to the river are primarily south-facing. River bottoms on and adjacent to the primary study area, described by Allen (1968), were used to obtain comparative population data for white-tailed deer. To assess possible effects of predation and interspecific competition on population ecology and dynamics of mule deer, we collected data on other species inhabiting the study area, including elk, white-tailed deer, pronghorn antelope, coyotes, lagomorphs, and various small rodents. Data on elk and coyote numbers, distribution, and habitat use were recorded during all complete-coverage surveys for mule deer. Special aerial surveys were flown to count and classify whitetails on the Missouri River bottoms. Aerial surveys of antelope on the area and adjoining plains were flown in late summer by the area management biologist. Siren and den area surveys of coyotes by Pyrah (1984) and project personnel provided coyote 28 population estimates. Vehicle headlight surveys to obtain estimates of lagomorph abundance were made 2-4 times each year on established routes through the area and adjoining plains (Trout 1978, Hamlin et al . 1984). Small mammal abundance was determined from 2 permanent Sherman live-trap grids (Trout 1978, Hamlin et al . 1984). Annual forage production was estimated during 1976-1986 by clipping vegetation within 2 X 5 dm and 4 X 10 dm plots in two stands in sagebrush-grassland and measuring current annual growth of shrubs on two transects in the Douglas fir-juniper type. General principles involved in establishing production estimates for herbaceous forage followed Pieper (1978). Estimates of production by shrubs generally followed Basile and Hutchings (1966) and Lyon (1968). Modifications made and more detail on techniques are discussed by Jorgensen and Mackie (1976). Browse utilization trends were measured annually in spring on transects (Cole 1958, 1959) established throughout the study area during 1959-1961. Relative changes in abundance of shrub species through the years were determined by monitoring tagged plants on browse transects, measurement of tree-shrub coverage on line-intercept transects established in 1963 (Mackie 1970), and measurement of shrub density and plant size on point-center-quarter transects established in 1971 (Martin 1972) . Food habits of mule deer were determined by examination of feeding sites and rumen analysis during 1960-1964 (Mackie 1970) and rumen analysis during 1976-1987. Data from the Roy 8NE weather station were not detailed enough to calculate several established winter severity indexes prior to 1967. To have an index of winter severity for all years of study, we constructed our own index. This constructed index gave values relative to the mildest recorded winters on the area and was calculated as follows: WSI = (Tw - Tc) + (Sc_^_Sl) is winter severity index is warmest recorded mean winter (Nov-Mar) temperature in degrees F is current mean winter temperature is total snowfall for the current winter in inches SI is lowest recorded total winter snowfall in inches where: WSI Tw Tc Sc 29 Our winter severity index (WSI) gave values highly correlated (r=0.92, n=19, P<0.01) with values from a modified Leckenby Index (Leckenby and Adams 1986), required less detailed data, and could be calculated for a longer series of years . Statistical procedures generally followed Snedecor and Cochran (1967) and Zar (1984). Analyses were conducted using the Montana State University computing service and personal computers. Statistical packages used included the Statistical Analysis System (SAS) (Ray 1982) and MSUSTAT (Lund 1983) . Estimating Population Numbers Like most mule deer habitats in the northern Great Plains, our study area was "open". Grassland - low shrub vegetation covered about half of the area and stands of coniferous vegetation were typically small or patchy, distributed along slopes, and held only scattered to low densities of trees that rarely exceeded 15 m in height. The area also comprised a relatively discrete population-habitat unit because mule deer were usually distributed on yearlong individual home ranges scattered throughout the area. During late autumn, winter, and especially early spring, the deer tended to group locally and utilize uplands and open ridgetops such that observability and counting efficiency was high. Because of this, aerial census provided a highly efficient and effective means of measuring population characteristics and trend. Our use of complete-coverage surveys eliminated possible bias resulting from sampling design. Quadrat and transect sampling require precise sampling systems that ensure random and representative effort over complex mosaics of topography and vegetation. They are also subject to the same visibility bias of any census that results in fewer animals seen than actually occur (Caughley and Goddard 1972) . Our counts always represented an absolute minimum estimate of the number of deer on the area. To develop a reasonable total population estimate, we had only to account and adjust for visibility bias. This we accomplished by developing observability indices (estimates of proportions of total deer observed) relative to season, survey conditions, aircraft, and observers. We also collected data on a variety of population characteristics for comparison and reconciliation with population estimates through arithmetic modeling (Mackie et al . 1981). A major thrust of our intensive studies during 1975-1987 was to further develop and test this approach to population estimation. 30 Normally, to obtain the most accurate population estimates, multiple aerial surveys are recommended for each sampling period. The number of replicates required depends upon the count as a proportion of the total population and the number of marked deer as a proportion of the total population (Rice and Harder 1977). As with most studies, our funding was limited. Because of this, our earlier experience, and our objective of concurrently estimating deer numbers and collecting other types of data, we directed our efforts toward 3-4 population surveys at different times of the year rather than 3-4 replicates during one period. Although this approach initially reduced our confidence in any one estimate, over time it had the advantage of providing estimates at shorter time intervals that could be reconciled with sex and age composition and mortality data from other sources. Also, because the survey data were collected over many years, data sets were developed that could be analyzed similar to replicate counts made near the same time in one year. For example, the surveys flown with the same pilot and observers during late December or early January each year from 1980 through 1986 could be used as replicates to determine the average percentage of deer observed at that time of the year. Although the percentage of the population marked (generally 4-9% during winter and spring) was lower than desirable (Robson and Regier 1964), the total count as a percentage of the total population (60-83%) was much higher than usual . The 3-4 population estimates made each year during 1976-87 generally supported one another and, when correlated with total counts, provided support or basis for improving estimates for earlier years when population size was estimated from a single complete coverage survey in winter or spring. Any estimate, before being finalized, had to include reasonable reconciliation of numbers in the various sex and age categories. These accounted for proportion observed in concurrent classifications as well as known or estimated mortality and natality based on hunting statistics, survival of marked deer, dispersal of marked deer, and population numbers and composition from prior and subsequent estimates. Our estimates based on the Lincoln index for years when adequate numbers and distribution of marked deer were present were seldom far from numbers estimated from known or estimated natality and mortality, especially during early winter and spring. The proportion of marked deer observed during aerial surveys (observability) was generally consistent within a season from year to year when the same pilot and observer were used. Observability was most consistent and precise and population estimates were most accurate during early winter 31 and spring. Population estimates for July, a time when yearlings were dispersing and males were more observable than females, were considered reliable only as a general index to population size. Considering their potential shortcomings, the July estimates were relatively accurate (Table 2.1). Observability of deer during July appeared to increase with population size up to a density of about 3.9 deer/km2. Observability of deer during autumn surveys also appeared to increase with density to a population of at least 3.9 deer /km2. Once population level was accounted for, observability was lower during surveys flown with pilot D than with pilot B (Table 2.2). Observability was generally consistent, however, for each pilot. The greatest deviations between Lincoln and modeled population estimates for autumn (Table 2.3) occurred in 1983 and 1985. Sex and age composition in modeled estimates (Appendix A) had to be consistent with estimates for the previous spring and the subsequent early winter, given natality and mortality during the period. We considered the Lincoln estimate for autumn 1983 inaccurate because 1) bad weather caused a 5-day break in flying mid way through the survey, allowing deer movement between areas, and 2) apparent drought-related movement of some deer from adjacent prairies onto the study area occurred in late summer. The Lincoln estimate for autumn 1985 was considered inaccurate because an unusually early snowstorm resulted in some deer moving from summer to winter ranges during the survey. Overall, the average deviation between the Lincoln and modeled estimates for autumn was 13.1%. When 1983 and 1985 were excluded, the average deviation was 10.0%. Generally, precision for the proportion of marked deer observed during autumn was less than that for winter and spring, even when the same pilot was used. This probably reflected more variable conditions during autumn and the influence of deer density on observability. Observability of deer during early winter helicopter surveys was very consistent from year to year (Table 2.4). An average of 66% of all marked deer were observed in surveys flown with pilot A during 1977-1979 and this estimate was precise (CV = 1.8%). During 1980-1986, pilot B was employed, the surveys were flown at an altitude 20-30 m higher than in previous years, and an average 78.8% of the marked deer were observed. Again, the coefficient of variation was low (6.6%). Flying at slightly higher altitude apparently enabled us to observe and record deer which previously had moved before the helicopter over ridges and into areas we had covered earlier. Observability was as high or higher during surveys in years of no snow cover (1981) as in years of complete snow cover 32 01 u ■a c. Q. O o I- o 3 «- a. x -o feU Is "! I .a 1 o a. Q c "S o o oo o> o* «-o ^ 00 •* fO S IA m t- n Kt ro o*- 00 o *t *- O Is- "~> Q O- «- -* <- O O O O N M s* -^ >f «- O o o d o o O Q O m *- nO in 'O r- r- r^ \r\ QJ Q S -t ^ IO N •* lA m lA r- 00 00 r W M ■* lAf Oj J) Oj 1.1 Qj OJ T3 T3 T3 T3 T3 "O L. L- t- C_ t_ U o o o o o o (J U O U U L L. L L. t_ L ID 0J !_ o u u 1) fc t/> o i 33 00 •* O *- C CJ ° i TR 13 s r*- co m r>- oo -o «- «- «M 00 •- S8 S 12 L. > h. «- «- f\i »0 IT> ^ t> 5 "o 1 •a "8 O 00 "8"S X O a .c u a> o ■•- o -t V 4) 4-1 I > « ° 8 8 & - at 3 X J3 a -O n £1 £ £ 3 3 U 3 3 3 U (-) u <_> o O u o i_ t_ L. L L l_ c g S a 8 n a a 01 a. 0) a 3 D ^ 3 D U 3 in l/> LO c/v en 00 l/l h- 00 O Q *- o r^ r*- t^ oo ao *-> i i i i i s 2 g ? g g s g 2 34 *J u o e QO c •H T3 3 r-l O a •H » r^ 60 00 c ON ■H rH .— 1 1 0) r- -a r-- o Ol B H B » o CO u CD H-4 u CO en CD t» 4-1 TJ id 3 a 4-) •H (0 4-> c/> C/> 0) X cfl ^H CD to U c CQ •H t+H i-t 0) X) > C •H cO a; IB •h 0) 1-4 X 3 0) O X! c/l c C/l M •H S c rH CD o -C u 4-1 c •H 1-1 J o IH a o C! u o lm ■H 4-1 C/l CO 0) E 4-1 M tfl o a <+4 -H c «J •rH cn 0) ^H CO c > o ■H •H > 4-1 u cfl 3 H u> 3 D.T3 O C Ph to # en • CM HI rH .a co e- o H •H J 4J CO B ■H 0 > In CD C4-I a CD -a 4-J 0) CO .— i F CD ■H XI 4-1 o cn s W O M ■H J 4-1 a) ^ o m c ■-H w o ai u X) c c o ■H M -^ J cfl > O a> ^ Q 1+4 tl 0) u 4-1 .-1 CO u E Tl ■H o 4_> r; 09 W o X U OJ C xJ •h a ►J M X) oj o S -H •H V4 E-i 0) IX CO CM CN o I o o in o- •« oo o» O >H CM CO E o\ m CM CO o in t^ CO CO O C o CO -H rH rH rH rH ^ rH rH 3 en 03 aw >> CU o CL, > U 3 en C K O CDx) r^ r» I"* r^ CO rH CO CO CT. CO ■H CU QJ l-l l-l vo vo vo ao CD CD r~ 00 r^ vO (d O'O 4) o O o O O o o O O 0 •H U CO,!) en Or*.Q CU l-i l-i O id *3 a o ■O 4-1 u cu , c (4h •X n, « r- CO in CM 33° n 01 a, X) cu ■a rH X) CU 3 0) M> B r* £ H l-l QJ CU <-t CM o t*l CN CO CTv o O 00 rH CM co » ^H U > eg o CD vn vO m CM -3- VO r^ CO CU l-l CO m vo CD O av CM r->° ,- * cu 1 & * o u 1 CJ=0 s * > go S 4J cu co UjSi ai ox) & ^o • « - o o^o ►»o 4-1 4-1 . en S 00 OHxj U C° [>> wo - Co -I4J (Mfl O 2 a 3 i S E i XI CO . XI cm c co<^ CU 4-1 O co , C S 1 CO 4-) > T3 co 0 » » CO r">0 CO o Bu >vO cd co >>o •■ * r^3 XI C 873X1 4J o u >, rH^j 4-1 u r^i! a u „-,*J B IH r^ rH a 3 O U CU 3 CM CU C CU 4J CU o o U iH B -P u cu C 4-1 CU S O 4-> O X Oo i > C l-l coo ►HO en CM 4->0 C CO l-l 0 •H o) UTH i-H CN m o 3 a) O rH CO r-{ • rH CO -* "-> > O cn 003 O co C 4-1 CM 0 UH rH CJ CO pa a i m i CM 1 Oh 1 rH O oo B i CM u en S3.Q co cd en d-* 3 efl 4-1 a. cu o O l-i i-H < » >^ 3 > o o o •P-H rH i-H .-h apt: o o o cd ClO CO CO CJ-H Qj l-l CM CM CM CM CM CM CM CM cj 4-> O 4-1 CJ 4-> O 4-) O 4-1 O 4-1 0 4-1 O l-i en 14-1 ar^ at^ a^ ar^ ai~~ ai~- ar~- ar^ ars ar- C0-H CO o<» o-j- o-t rH rH rH 36 (1983 and 1985). Patchy snow cover (1982 and 1986) may result in more variable observability. The Lincoln and modeled population estimates were very close for early winter (Table 2.3), deviating by an average of 3.7%. Spring surveys were conducted with a helicopter during 1978-1980 and a Piper SuperCub from 1981 through 1987 (Table 2.5). The SuperCub surveys were generally flown a week or two later than the helicopter surveys, shortly after "green-up" when most deer concentrated on open ridgetops to feed on new growth. Observability from the helicopter was high and similar to early winter except during 1978 when survey conditions were poor, the marked sample was low, and deer were actively moving to their normal ranges from areas along the river where they had concentrated during the severe winter. Average observability from the SuperCub was also high (68.3%), and the coefficient of variation between years was low (8.0%). Both cloud cover and stage of vegetation "green-up" may influence observability from a fixed-wing aircraft during spring. For example, slightly more than 60% of the marked deer were observed in 1983 and 1986 when skies were generally overcast, and the 1986 survey was conducted before "green-up" was sufficiently advanced for maximum deer use of open areas. Excluding those years, observability averaged 71%. The average deviation between Lincoln and modeled population estimates for spring was 6.5% (Table 2.3). The accuracy of our counts and percentages of marked deer observed during winter and spring were at least as high and usually higher than reported by others for mule deer (Bartmann et al. 1986), white-tailed deer (Rice and Harder 1977, Floyd et al.1979, Beasom et al . 1986), and moose (LeResche and Rausch 1974). Variation in percentages of marked deer observed during winter and spring among years was much less than reported elsewhere for deer (Bartmann et al . 1986, Beasom et al. 1986), but similar to that for aerial surveys of moose conducted by experienced observers under excellent conditions (LeResche and Rausch 1974). The high precision of observability indices for winter and spring provided confidence in the accuracy of annual population estimates for those seasons. For example, during 6 of the 7 winter surveys flown with pilot B, the proportion of marked deer observed varied only from 0.77 to 0.83 (Table 2.4). During the seventh year, when survey conditions were poor, observability was 0.68. If we assume a total count of 1,000 deer and apply a mean observability index of 0.80 (52 of 65 marked deer seen), our Lincoln estimate would be 1,250 deer. Considering the possible range in observability recorded for most years, we might expect the total population to lie between 1,200 (1,000/0.833) and 1,300 (1,000/0.769). 37 CD 3 cu Xi o 4-> 4-1 -rH cd ro ro rH -f o CO I1* B CO r- rH 00 r— rH CO CO CO CXI 3 o cd ■rH en co a> o CM CM rH CO CTi CM rH «J rH rH rH rH rH 3 U) CO 3.W OJ o eu > lH 3 (0 C l-l o cu Tj 00 o ro o\ <* CO CD •H CD 0) CO rH UO CO 00 O CO •* rH o rH 4-1 1-1 Q I a -M cu U r- CO co rH CO -* rH CM CO en ro r- Ol t-H IT) CO CD rH 4-1 0) 1-1 CO o co St VO CO m ^t !•» CD o ai 11 CM ex CO 4J 73 05 CO rS "C rC rC 4J 3 •a a> 4-1 0) 3 4J 4-> • •> •rH • • -H c 4-1 Cfi u X O l-l l-l T3 ■3 rH • •a -o > cd 2 m C o ■H rH o 0 C C c c 3 e a, o s E U s c 3 a - ex 3 3 • 3 'H, u •H O o 4-1 3 O O 4-1 3 » °> 4-> c * •a 3 . s ■ U lH CO l-i u CO CO • CO ■H o en s o CO o CO oo • oo . co 3 00 • oo • CO en °> >> X) 0) CO 1) 01 >. >, U 01 >% >, o ■a T3 - "° 3 & a > cu 3 a. oj 3 0) c lH 01 0) 3 0) 3 lH oj 3 £ 3 o o o 3 o o o o r4 3 rH C Ol In In 3 u 3 u X O O 4-> o a rH 3 s i-H s rH CO 3 CO 3 > 00 CO 3 CO 3 > •H rH •H o) CO CD CO CO CO CO CO PQ CO CQ co o = PQ co PQ co o x: u 4-) Q CU 01 o o u rH < < o co CO. P3 to CQ u> CQ (Xpa ■H CU « 01 >-l u o> ro 3 > o 4-1 -H rH ,a rQ rO XJ rd JO rO a. «; O 3 3 3 3 3 3 3 cd CO O O O O O O O U -H CU l-l CM CM CM IH u u lH u u lH U 3 l-l pg l-l CO l-l CO 01 1) 0) u 01 Ol 01 1 O 01 i 01 1 1) 1 D. a Cu cu CU CU Cu X CO 4-1 4-1 o 4-> O 4-1 o 3 3 3 3 3 3 3 U CO 4-1 Cur- Cur- Cur- CO CO CO CO CO CO CO cd -H cd o 03 e C 03 C o ■H U o en c ■H O n 03 0) M 03 >i ■P 1/3 (0 Q) U Cn •H (X4 43 5 c 5 o CO in < 5 c E CD o o CO o ^r CO CD CD T~ 1 CO CO TO 3 c O O 10 CD (/> B j*: 3 CO O , - CD 10 CO .O GO CD S h CO t_ CD JZ CD CD > (A DC 3 O CO CD 'C o Lm 3 CO ■u O o 0) > CO is o 2 o ^- # (1) J3 +J 4-1 0 c o •H ■P CO u o H Q) J=. P •h P CO 4J •H • XI (JJ rn M x: 0) > CO ■H M M ITI (1) ■a M c Xi ra M U) ■H >1 U en ■s CD .£ X! en P •H X 4-1 0 ^ If) ■p r (-; •* OJ n p P X QJ M 01 •P 0 •r-\ m e Di C •» •H 0J ■* 03 n (JJ r, u m r0 Cu >i m -a B 3 P < W (N n 0) U 3 Cn •H 44 V*' } - ' ■ -A < ^.■"•^-Ja". N< € t ' h \V aK" ft* • '*i ~ JtU L*>\ i.-^t* 3& ~£>$&* V-1 v*. b,y .'•Mr,- •V- ■V-; ¥j£& *SSr< 1 S?^! % .«y j* 7&«? ^ffef ,* ^Wi (V^- *#5 sit.' < *. V •«# «« . id CD (0 >i T3 3 +J to CD U ■H 03 (a e o ■p o A •H 1-1 CD < ■ m a; n D> ■H En 45 B Figure 3.4. Aerial views of the study area showing (A) drainage head areas, (B) mid-level drainage areas, and (C) steeper terrain near the Missouri River. 46 additional permanent sources of stockwater, on the area. Nearly all of the reservoirs were small (< 1 ha) and more than one-third were dry by late summer in average water years . During the driest years only 28 contained water, mostly in small amounts and of low quality. Three of the 4 wells were developed during the last 2 years of the study to supply watering tanks at various sites on the south-central portion of the area. During years or periods of above-average precipitation, water is also available in natural depressions throughout the area . Access is provided by a paved highway (U.S. 191) along the western boundary, several unpaved "roads" extending through the area from the highway, and numerous vehicle "trails" along ridgetops. All unpaved roads and trails are impassable when wet. Climate The climate is semiarid, characterized by moderately low and variable precipitation, low relative humidity, moderate to strong winds, and great extremes in temperature. Records for the U.S. Department of Commerce weather station, Roy 8 NE, located approximately 25 km southwest of the center of the study area show a 42-year (1943-1985) mean annual temperature of 6.5 C (range 4.3 - 8.3 C, CV = 4%). January is the coldest month (Fig. 3.5), with an average temperature of -8.4 C, while July is the warmest (20.8 C) . Mean annual (1939-1985) precipitation is 35.4 cm (range 17.9 - 63.9 cm, CV= 27%), most of which occurs as rain during spring and summer (Fig. 3.5). Precipitation is normally highest during June (7.5 cm) and lowest during February (1.0 cm). Total snowfall ranged from 0.2 m to 2.2 m. Snow depths exceeding 0.3m are rare, though occasionally accumulations up to 0.9 m persist for significant periods of time (e.g. 1977-78 and 1978-79). Warm, southwesterly "chinook" winds occur periodically during many winters to moderate temperatures and reduce snow accumulations. The average frost-free season is 128 days, but the effective growing season is shorter, usually extending only from mid-April through the end of June. Trends in annual temperature, growing season and annual precipitation, and winter snowfall and severity (Figs. 3.6 and 3.7) illustrate the marked seasonal, annual, and periodic fluctuations that characterize the climate and weather conditions on the area. Long-term (1580-1980) trends in moisture were calculated as a relative moisture index from tree-ring data for Douglas fir (Pseudotsuga menziesii) on the study area by the University of Arizona Tree Ring Laboratory (D. Meko and C. Stockton, pers . commun., Fig. 3.8). These 47 ■14 -* — 1 1 1 1 1 1 1 1 1 1 r Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month E c o o o -\ 1 1 1 1 1 1 1 1 1 1 r Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Figure 3.5 A. Forty-two year (1943-1985) mean and standard deviation of monthly temperatures at Roy 8NE weather station. B. Forty-seven year (1939-1985) mean and standard deviation of monthly precipitation at Roy 8NE weather station. 48 Roy 8 NE Weather Station Adjacent Weather Stations O o j_ o Q. £ c c < c ■ m •H 00 - ■ u ■ s 0 U-l -a " CD - o p - °? 3 U - -H - u - in r^ ^■v — i <* CN ■ i^ u " Q) _ P r ' -70 a, en u ■ (D - - CO . m lm "— » CO CO 4 ■ to 0) >■ -a . c _ •H ■ o >i - CD to - in "> p. . 4 m Q) • P t- C a) •H 1 - > CO OO . o in > -1 d> •H ■ <* P X - (0 cn - CD O . m p p <* £ _ i <* < p - *r . O) o CO o CM xapui A;u9Aes JeiuiM n P Di •H fa 50 o • ■» CO U (0 en C § o N CD .c u \H ^ &«H ^ s < 8 c P 0 4-t 5 *H 0 8 O -H >i O) oo g +J T- m )_, -h HuM 0*fc! W 0) o oo CO year with Univ T— ^— ^ n ,° i ^■2 H (0 0 -j cu o ;H n U^4"1 Eh 8 i— index ouglas Meko, 0) Q § 3 EQ CO +J o T" mois d fr and (0 0 -~ • H -H -P C co +J 3 ,* 0 fl O U M ■H r-H 0 U (IJ (fl -P 3 Pi U CO En in xapui 9jnis;o|/j eAjjeiay CO ro ■H cw X CM 3 m ■h O £ 13 CO m <* r-~ i I oo CM in i i o\ CN 00 •J- oo ro ON ro CO CM CO c^ iH CM ~* iH 00 CD r^ m O CM o\ rH -3- rH iH o 00 CO e'- en r-. CT> CO co in cm CO r>» XI 0) D D .— 1 en In w cd Ed X! -C 4-> id CO CO o U t-> o CO (0 OJ (j 1-1 oj x; » u cd « OJ X5 3 u ^ o x: in U) (0 * OJ OJ j_) C.-H >N 05 4J - CM X) C OJ ™j xi en » u M 0 u m QO « ca 3 u ~ a> •? > « cd 00 "J OJ w jC _ 4-> c •H ^ ^ I-1 05 a , - 2 « 3 u OJ +J XJ OJ l_l 3 o (X C 0< . en •H „, -i s T3 . u £■ CI >. >% cd t-l OJ ^ u u 1. » CU *> m CD X! » £ CO o P, O OJ w "S U a OJ g en 3 o •> X! en "~i u CJ CO ' CO u ^ E 73 00 14-1 3 en C cd ■O ,„ c * 4-1 0) CO « a > rH ki in ^ .. tc, < 58 <0.05 1.35 354 6.70 760 0.06 0.4 0.06 12.5 0.34 78 0.66 70 0.04 10.4 0.02 0.07 0.71 218 Table 3.5. Abundance (number of plants/2 frequency) of yellow sweetclover in 310 4x10 dm plots on 31 sites, 1977-1986. First Second Year Year Avg. Weight3 Year Plants Plants (g) per plant kg/ha 1977 11/1.9 0/0 1978 3227/44.2 21/3.5 1979 0/0 1407/38.1 1980 90/3.5 0/0 1981 2582/54.2 9/2.9 1982 2563/53.9 291/26.5 1983 544/22.9 769/46.8 1984 3205/35.5 21/3.2 1985 46/3.2 0/0 1986 3695/62.6 108/11.9 a Average weight (g) per plant taken from separate clipped plots, same period. Although this often was the case, cool temperatures also occurred in the absence of precipitation. Thus, mean temperature for April and May was significantly negatively correlated with forb production, but precipitation during those months was not (Table 3.6). The negative relationship between mean growing season temperature and forage production appeared to be independent of a correlation with precipitation. Multiple regression models incorporating both temperature and precipitation variables were computed for the period 1976- 1982, when production of all forage classes was measured (Table 3.7). Mean temperature during May and precipitation during the previous July-April period together explained 91.5% of the annual variation in forb yield (R2 = 0.915, F = 21.55, P = 0.009). Stepwise procedures indicated that both variables were significantly related to forb yield; mean May temperature explained 73% of the variation and an additional 18.5% was explained by July-April precipitation. Production of 3 shrub species was best predicted by precipitation during July-May 59 en 3 o •H > en 3 en u CD > cd oo ■a H CD en U) IT) u OO X) c d .a . 3 CM U oo en i & o\ p ^ ■-< 0) o Q 4-1 CO S " •H <" hi 4-1 J? c > oj .. •H G (0 « ■-H at C u o t-. -H o +J u cd ■p ■H cu a. .-1 -H P. U B OJ •H u en a. t CO • CO OJ .-1 .a cd E-« to id U -i to .G -* t/3 ^ J3 3 -a i-i .-< J3 Xi OJ l-l CO -H O >H en 0) -a ■rH .—i u -a OJ 01 3 ■H a. kl >H en x: 00 en •a XJ rH >-l OJ O -H *J d a> co X) OJ C rH OJ JD p. cd 0) -H •O l-i (3 cd M > 0> co * xi C O td 4-1 U QJ u P. je o I-I cd 3 -x m in CD * * CM co * 0> CO CM CD a o •H a •H O OJ l-l p. s-> cd S i i-H 3 •-> -x to CO * -X eD co CM O 00 c o •H 4J cd 4-1 ■H p. •H U ai V4 P. >v cd £ 4J P. co -x CO CO a. cr* CO co en •X ■X en r- CO * CM en CO •X in oo C O P, •H U OJ l-l P. l-l a. 4-J a> CO -X CM o CO •X o a o 4-1 cd u OJ l-l p. 3 "J co vd in CO CO O in r-~ r^ o o o i i co en m cr> s cd •H l-i P. < •X oo CM CO CM eo O O^ O l OJ u 3 4-> cd l-l CD p. a OJ c cd OJ B cd l-i P. < + CN CO r- * en CO o i l-l 3 4J nj l-l CI) p, B Q) *-> C cd OJ e cd CM oo o^ en l-i cd 1) OO C ■H u 3 •a c o •a 0) i-i 3 en cd OJ E XJ 3 l-i x: en U QJ O. to l-i o ■o OJ ■H c o cd QJ c ■H 3 o l-l HO CD X\ 4-1 OO c ■H l-i 3 T3 T3 C cd C o en ca QJ CO u o op p. 3 O to U J2 aO 4-1 c a; o ^: e ^ e o ij CM QJ l-l cd OO c •H l-i 3 X) QJ l-i cd cd u « cd ™ cd c ^ n QJ l-l 3 4-1 cd P. ^ •H OJ U g< OJ s u OJ Dl. H cd 4-1 o o V a, 4-1 a cd cj •H C n oo Ml •H •iH c/> CO •K ■X * 60 CO 0) ro ■H X X) u 3 i-H o> l-l 01 ex -C •H U) CO >H 0) X X) i-H XI 3 i-H XI u u 0) n) 0 -3 ■iH ■H [14 CO 5h u 03 > 4-1 C 0) 13 C r-> 0) co a 01 0) •H a 13 U ,0. .-l Cb 3 01 a. l-l ■H in x: >h X) r4 o C en 01 01 •a .-1 c x> ai m P.-H 01 Lj •a m c > o cm CO 00 CO o CM CO 10 * * co CO CO 00 o 00 o * CO 00 00 CO CO * CM 00 o CO CO * -X CM O * CO CO ■X CO 00 CO 10 * * LT1 rH o a) 0) a. n t-l r4 c 3 a 3 0 4-1 +j a 4-1 •H id cd 0 cd 4J 13 u u •H r-l cd a 01 a. 4-1 0) 4-1 01 c OJ 0 p. ex ra a. ■r4 u 0 1-1 •h e c E 4-> E P, 3 ■H 3 4-1 a> 0 0) •H <11 •H 4-J 4-1 4-1 ra 4-1 •H 4-1 CX 4-1 U cd cd ra 4-J 4-1 •H 01 U 4-1 rJ ■H c rrj c U a u 0) •H 01 a rfl 4-J cd 0) cd ex ex 0, ex ■H 0) ■H 0) rJ QJ E ■H E u B ex E CX E rH 0) U 0 01 •H ■H 4-> 0) 4^ VJ >-> U ^ !*i >"N V-i r-l p, cd 0J cd cd cd P. C cx C s l-l 2 S s < cd ra jG 1 p. 1 1 1 1 0) rH 01 u rH i-H u rH 1-1 a ■H E 1-1 •H ^» •H 0) •H 0) r-l ra M cd kl X> 1-1 rQ >, P. (*, _e ex JE a B ex e cd <«! ra 1 < 1 < 01 < 01 s 1 2: ^ >% 4-1 4-1 r^ rH X! rH ■a a. t3 ex -a rH -a 3 c 3 c 01 c o> c 3 C •-> id •-> r0 CO ra CO cd "-) ra N CO Ol rH I r- r- en U m 0) C •H r-l 3 13 G o 13 01 1-1 3 CO ra 0) E X 3 rJ rC CO u 01 ex rJ o 01 ■H X on c •H r-l 3 13 01 1-1 ra ra 4-J ra 13 aj u 3 *J ra l-l 01 ex e cu t-i c o ra OJ GO a •H 3 o rJ aj x 4-1 00 c ■H l-l 3 13 13 C ra l-l o rJ ex o E E O Vj QJ U ra ra 13 c o c o en cd 01 cd CO 4-1 ■H DO ex c •H -H U 3 Ol o V4 l-l rX| 00 m o o • • o o V V a, Pi 4-J c ra u •H C C 00 QO -H ■H c/i CO * * 61 (R2 = 0.795, F = 19.4, P = 0.008). Temperature variables did not significantly (P > 0.25) add to the predictive power of regressions on shrub yield. However, a combined index of forb and shrub production was most significantly related to a combination of precipitation from July through April and mean temperature for May (R2 = 0.87 8, F - 14.43, P = 0.017). Stepwise procedures indicated that the precipitation factor contributed most to the regression (R2 = 0.643, F = 9.02, P = 0.03), but mean May temperature also contributed significantly (R2 = 0.235, F = 7.72, P = 0.05). Production measurements were limited to forbs and grasses after 1982. Addition of those data to the regression model did not significantly alter results. Precipitation during April and May continued to be significantly positively correlated with grass production (r = 0.863), explaining 74% of the variation in yield between years. Production of forbs also continued to be significantly correlated with mean temperature for May and precipitation from July-April, although the predictive capability of the resulting regression was diminished (R2 = 0.720, F = 10.28, P = 0.007) compared to that derived from data for 1976-1982. An examination of the residuals of this regression indicated that production during 1984 and 1986 deviated most. A regression of mean temperature for May and precipitation during July-April on forb production for the 9 years that excluded 1984 and 1986 had very high predictive power (R2 = 0.941, F = 47.62, P = 0.0006). Data from the residuals of the full regression indicated it overestimated forb production when precipitation increased in years following drought. This apparently occurred because the regression could not account for moisture necessary to recharge soil moisture above the wilting point, when additional moisture becomes effective for plant growth. As noted earlier, soil moisture measurements probably would be best correlated with forb yield because they encompass temperature, precipitation, and past conditions. In the absence of soil moisture measurements, the regression of mean temperature for May and precipitation during July-April gave adequate predictive power, providing it was recognized that yield was overestimated in the year of recovery following drought. Despite those overestimates, the regression accurately predicted trend in production (stable, increase, or decrease) in production from previous years. Based on these findings, an index of relative summer forage production (Fig. 3.9) was computed for each year from 1959 through 1987 as forage yield estimates using the regression equation: [ Y ( forb and shrub yield ) = 1,704 - 30.04 X1 (mean temperature for May) + 24.61 X2 ( precipitation 62 CO sz X E E c/) 600 500 1960 1965 1970 1975 Year 1980 1985 Figure 3 . 9 An index of summer forage production (kg/ha) for forbs and shrubs at selected sites on the Missouri River Breaks study area as determined by clipping and regression analysis. July-April ) ]. This index indicated that 1961, 1980, and 1985 were years of extremely poor forage production on the study area, while 1959, 1965, 1975, 1978, and 1979 were characterized by forage production more than 1 standard deviation above the mean. Although production in other years, such as 1963, 1967, 1968, 1973, 1974, 1982, and 1986, was indicated to be less than 1 standard deviation above the mean, all stand out in field notes, general observations, or other data as years of above-average forage production. Forage Abundance in 2 g. Forage production was directly measured on only 4 sites vegetation types (Table 3.4), but a variety of data , Tables 3.2 and 3.3) were available to estimate the 63 relative abundance of the same species in other types. Those data were used to calculate estimates of total forage produced on the study area during summer and autumn. The estimates were developed with very conservative parameters that included the following: 1) Distribution cells in which deer were not observed during aerial surveys (17% of area) were not included in forage calculations although they undoubtedly received at least minor use by deer. 2) Minor vegetation-cover types in which relative forage abundance was not measured (an additional 9% of area) were either assigned no value for forage, or a level equivalent to the value in the lowest type measured. Several of these types (river riparian, pine- juniper- Douglas fir, and grassy bottoms) were probably above average in forage production. 3) Only those forbs known to be preferred by deer during summer and, autumn were used to calculate forb production. 4) The production of 2 very important shrubs (rubber rabbitbrush and green rabbitbrush) was not estimated or included, although they often averaged up to one-third of the autumn diet. 5) Only one-half of the forage in the big sagebrush habitat in which deer were observed was considered usable. 6) After the previous 5 factors had been included in calculations, only one-half of the estimated forage yield was considered usable by deer. 7) Regrowth of forage plants was not considered. An example of the data and calculations involved in these estimates is presented in Appendix B. The results (Table 3.8), together with data presented in Figure 3.9 indicated that the quantity of forage during summer and autumn was adequate to support existing deer populations throughout the period 1959-1986. Only in 1985, did the available quantity of forage during summer and autumn appear to approach a shortage. However, even then, given the conservative assumptions of the calculations and the fact that an autumn "green-up" of grasses and forbs occurred in 1985, it is unlikely that the quantity of forage was insufficient. Similar conclusions about the quantity of summer-autumn forage were made for a Colorado mule deer range (Wallmo et al . 1977). 64 Table 3.8. Estimated forage production on the study area and estimated number of deer it would support during summer and autumn, 1976-1986. Year Total kg of Forbs on Study Area Total kg Forbs Available 464,213 353,075 3 ,009,784 4 ,543,311 327,102 705,630 1 ,214,169 565,691 227,705 84,569 2 ,232,415 Number of Deer Supported Summer- Autumn3 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1,375,136 1,137,342 7,375,912 10,609,783 1,037,409 2,016,061 3,226,553 1,351,978 616,985 248,510 3,900,078 1,719 1,308 11,147 16,827 1,211 2,613 4,497 2,095 843 313 8,268 Number of Deer Total kg Total Number Supported Summer-Autumn kg Shrubs of Deer by Forbs Year Shrub Species Available Supported & Shrubs 1976 1,282,604 641,302 2,375 4,094 1977 486,148 243,074 900 2,208 1978 2,450,196 1,225,098 4,537 15,684 1979 2,151,563 1,075,781 3,984 20,808 1980 489,131 244,565 906 2,117 1981 988,475 494,237 1,830 4,444 1982 1,081,635 540,817 2,003 6,500 1983 347,645 173,822 644 2,739 1984 742,405 371,202 1,375 2,218 1985 242,030 121,015 448 761 1986 2,018,451 1,009,225 3,738 12,006 a Number of deer supported = Estimated total kg forbs and/or shrubs available on study area / 1.8 kg per deer day / 150 days, summer-autumn period. 65 The quantity of forage available on the study area during winter and spring was not directly measured, however some observations indicated that it was unnecessary. Forage quantity or quality probably was not limiting during spring (April- June) because more green forbs were available then than during July when plots were clipped (Table 3.3). Also, deer made considerable use of ubiquitous new green grasses during spring. The primary forage species used during winter (big sagebrush and Rocky Mountain juniper) were essentially unlimited in abundance. No hedging, highlining, or other indications of overuse of those species was noted. As pointed out by Wallmo et al. (1977), the total use of those species may be limited because of the bacteriocidal effects of their essential oils. Substantial use must be made of other species to dilute big sagebrush and Rocky Mountain juniper. Although most other forage available during winter, with the possible exception of rabbitbrush, were nutritionally below maintenance requirements for deer, adequate quantities were available to mix with big sagebrush and Rocky Mountain juniper. Rabbitbrush was essentially used entirely every year, even at the lowest observed deer densities. At low deer densities, however, it may be available for a longer portion of the winter period, raising the average nutritional plane of deer. Regardless of deer density, forage quantity appeared to be adequate, but forage quality, by itself, was probably inadequate, to sustain deer during any winter. Although winter forage supplied a portion of maintenance requirements, winter survival appeared to depend on fat accumulation and storage during summer and autumn, the length and severity of winter, and energy conservation behavior (Wood et al. 1962, Loveless 1967, Silver et al . 1969, Thompson et al . 1973, Bucsis 1974, Short et al. 1974, Mackie et al. 1976, Mautz et al. 1976, Walmo et al. 1977, Mautz 1978, and Youmans 1979). Changes in Abundance and Use of Forage Plants The percentage of current annual growth twigs or "leaders" on shrubs used by mule deer was recorded each year, beginning in 1959 for rubber rabbitbrush and 1960 for fragrant sumac (Fig. 3.10). These 25 plant transects were established according to the criteria of Cole (1958 and 1959) and originally included information collected on age, condition, and form class of the plants, as well as percent leader use. Transects were measured in spring of each year and, thus, recorded use for the previous summer, autumn, and winter. Although Mackie (1975) determined that these transects did not 66 100-1 90 - 80 - o 70 - (0 => 60- o T3 (0 O 50- 40 - 30 - 20 - 10- 0 Rubber Rabbitbrush Fragrant Sumac r"V*"*rI I /■*■» : \ 1Kr 1960 1965 1970 1975 Year 1980 1985 Figure 3.10 Trend in percent leaders used by spring each year for two important browse species, 1959- 1976 and 1980-1987. Bars represent standard eror of the mean. reliably measure what they were originally intended to (especially condition), some useful information was obtained. One criticism of the Cole transects was that too few were established for each deer population to determine whether changes in use were statistically significant. Transect density on our study area was at about the highest level in the state. From 3 to 7 (X = 5.3) rubber rabbitbrush transects and from 2 (1960 only) to 10 (X = 7.2) fragrant sumac transects were measured each year. The relative number of transects was high for our study area, but few statistical differences were observed in percent leader use among years (Fig. 3.10). Because utilization was measured by categories, 90% leader use was the highest possible value. The upper confidence level of leader use was never below 88% for rubber rabbitbrush during any of the 25 years of measurement. The lower confidence level was 30% during 1965 and 49% during 1963, but was usually always above 60% leader use. Annual mean percent leader use for rubber rabbitbrush was consistently high (X = 83.6+1.9 SE, range 69- 67 89, CV=0.055) throughout the entire period 1959-1986 despite wide fluctuations in deer numbers. Annual mean percentage of leader used on fragrant sumac varied considerably (X = 33.2+9.5 SE, range 3-81, CV=0.68). Lowest use was recorded during both low (1973-1976) and high (1968-1970 and 1984-1986) deer population levels (Fig. 3.10). From transects established during 1959-1963, data on longevity of tagged plants were available for 6 transects on fragrant sumac and 2 transects on rubber rabbitbrush. Because these transects of 25 plants were established using nearest neighbor (of the same species) techniques, new plants establishing between the original tagged plants could be detected. Deaths of tagged plants were recorded as they occurred through the years. Existing records were summarized and, during measurements in spring 1985, particular attention was given to dead tagged plants and new plants that had established over the years. A total of 4 of 150 fragrant sumac plants had died on the 6 transects since 1963 and 19 new plants had established, for a net increase of 15 plants. A total of 4 of 50 rubber rabbitbrush plants had died on 2 transects, and 29 new plants had established, for a net increase of 25 plants. Although a few of the new shrubs may have been overlooked when the transects were first established, the overall conclusion was that the number of fragrant sumac and rubber rabbitbrush plants on those transect areas increased after 1963. Because we had increasing difficulty in finding 25 plants to measure within one rubber rabbitbrush transect area without tagged plants, we believe the number of rabbitbrush plants declined through the years on that area. Line-intercept transects were established within mature stands (100+ years) in the pine- juniper and Douglas fir-juniper vegetation types in 1963 and remeasured during 1983. Within the pine- juniper type (Table 3.9), both ponderosa pine and Rocky Mountain juniper increased in coverage over the 20 year period, further closing the overstory. Significant increases in canopy coverage occurred for rubber rabbitbrush and green rabbitbrush, but canopy coverage of current, gooseberry, and snowberry declined significantly. Within the Douglas fir-juniper type (Table 3.10), canopy coverage of both Douglas fir and Rocky Mountain juniper increased significantly, but canopy coverage of Ponderosa pine declined. Among understory species, only canopy coverage of fragrant sumac increased significantly. Canopy coverage of rose, snowberry, gooseberry, and currant decreased significantly. 68 Table 3.9. Proportion of line intercept transects in Pine-Juniper habitat types covered by various shrub species, 1963 and 1983. a Shrub Change Speciesb 1963 1983 From 1963 Z value Artr 0.0008 0.0010 0 0.91 Chna 0.0007 0.0022 + 7 . 12**c Chvi 0.0000 0.0004 + 11.11** Jusc 0.0866 0.1446 + 23.51** Rhar 0.0098 0.0088 0 1.33 Ribes spp. 0.0015 0.0011 - 2.00* Pipo 0.2620 0.3073 + 13.00** Rosa spp. 0.0096 0.0107 0 1.39 Symp spp. 0.0155 0.0070 - 10.38** Fourteen 100-foot lines at 4 permanently marked transects, See plant species list, Appendix C. * Significantly different at P < 0.05. ** Significantly different at P < 0.01. Table 3.10. Proportion of line intercept transects in Douglas fir habitat type covered by various shrub species, 1963 and 1983. a Shrub Species' 1963 1983 Change From 1963 Z value Jusc 0.3116 0.3567 Rhar 0.0029 0.0094 Ribes spp. 0.0046 0.0019 Pipo 0.1931 0.1810 Rosa spp. 0.0458 0.0177 Prvi 0.0304 0.0343 Psme 0.5607 0.6290 Symp spp. 0.0561 0.0495 + + 0 + 6.62**c 5.87** 3.18** 2.13* 11.07** 00 61** 10* a Two 100-foot lines at each of 2 permanently marked transects b See plant species list, Appendix C. c * Significantly different at P < 0.05. ** Significantly different at P < 0.01. 69 Overall, these data indicated that, in unburned timbered habitats, the trend during the last 20 years has been toward increasing overhead canopy, with a decline in coverage by snowberry, currant, gooseberry, and probably rose. Shrub species showing an increase, or at least no significant decrease, included fragrant sumac, rubber rabbitbrush, and green rabbitbrush. Eichorn and Watts (1984) reported that burning of mature stands in the Douglas fir-juniper type on this area increased canopy coverage of snowberry, rose, and chokecherry. They also noted a similar increase in shrubs on burned pine-juniper types, though the ultimate increase and canopy coverage was less than occurred on the Douglas fir- juniper type. As part of a study of the ecology of fragrant sumac in Montana, Martin (1972) established 5 point-center-quarter transect plots on the study area, each of which included 20 tagged fragrant sumac plants . These plants and transects were first measured in 1971 and subsequently remeasured in 1982 and 1986 (Table 3.11). None of the 100 fragrant sumac plants, tagged in 1971, were dead in either 1982 or 1986. Plant diameter, area, volume, corrected plant area, and corrected plant volume were all significantly greater in 1982 and 1986 than in 1971. The percent crown dead remained the same in 1982 as 1971, but there was significantly less percent crown dead in 1986 than in 1971 or 1982. Height was significantly lower in 1986 than 1982. Despite considerable girdling by microtine rodents in winter 1978-79, all fragrant sumac plants present in 1971 continued to live, and in aggregate, increased in area and volume. Although, as noted above, all 100 fragrant sumac plants tagged in 1971 were still alive in 1986, point-centered- quarter measurements for 1972 and 1982 (Table 3.12) indicated a relative decline in the number, density, and frequency of plants. That resulted because the point-centered-quarter data gives relative values, not absolute values. The 2 sets of data (Tables 3.11 and 3.12) indicated that, although numbers of fragrant sumac and rubber rabbitbrush plants probably were similar in 1972 and 1982, the numbers of other shrubs, especially snowberry, had increased. Four of 5 of these sites had been burned more than 25 years previously and, as predicted by Eichorn and Watts (1984), some overstory sensitive shrubs may have continued to increase following initial measurements in 1972. Collectively, the measurements indicated that increasing or at least stable shrub populations occurred on most transects despite 1 to 2 "irruptions" of the mule deer population over the intervening years, consistently heavy use of rubber rabbitbrush plants by deer, and at least occasional 70 8 i ■ -^ E 0 o. O — o > t_ a. < ^ o ai *J E -» C 3^ ID — • E _* O -^ a. > II 3 "O o 0 ra l. t_ a> 41 o o o o + 1 +1 o- o -o On o o o- o r- nO o I1 nO o + 1 b d no in On r- t O 00 CM «- «- + 1 +1 +1 00 00 n- u-> -r- o ■ o» O 0) 3 V|^ IS a. > L. u> 11 ui x: a; *j -> o >. c — i 01 ■m i> C 3 10 *J O III ■i- .a h- .,- a; c o oi c ■ ^ 0) t/» l_ 01 (A <+_ •f- H- ■o -o (0 o *j o c re c u 3 i- O *- [_ ■!- u c Ol 4-1 •(- C l/l 01 u o 1- 2 «- c 10 c c ji n 10 <- -C .C 4-. O 00 L l- o 11 11 «- ai O) l I- c .•o CO -i- VI *-. 1/1 u^ ITN 11 o o -* o d '- m v| VIO Q. O- o — ' — ' 0_ •i- E U w o o o ?' I' *' in o ■** r^ 00 00 odd s— **- c c c u Ol 01 •1- l/l 1/1 c 1) u Ol o + 1 o 41 o + 1 oo Q 0 II 10 a ■D "O vO o o co1 o f\J NO 00 00 On On c ■a "O 01 rg 10 a u n L. a a 11 a c c 3 3 TD O O a a It 01 u a u u X X T3 11 11 C ro r\j -o 00 00 o o- .c *~ *~ ai l_ 1) • o o X no ■+- 00 o l/l l/l 11 u 3 3 fN- "O O- C rO <0 «- 10 > > C C\J 10 00 —^ — 1 X o- 71 Tl tu 1-1 CD 4-1 c OJ u o N p, oo o> m H c -a 0 C CO en J=> CXI 3 r^ Ul a\ .C H l/> - CO cfl 3 OJ O 1-4 •H cfl l-l d >>*. > -n 3 <4-4 4-1 o 01 l-l CO a: M ,a CO H OJ 3 14 c « ■a 1-4 G 0) (0 > •H • Pi p^ u •H c 1-4 0) 3 3 O tr CO CD C/J U •H 4-4 32 » aj OJ X CJ 4-> C co c C •rH •H J3 H 4J o •H ■a 3 » to >^ 4-1 4-1 U •H OJ CO •Si C a 1) CO •o u 4-1 OJ > V4 -H OJ +J 4-J co 1-4 .—I m a; 3 prf cr co OJ •-t ./O (0 E-" OJ ,a B 3 2 en 4-1 C CO a, o OJ f ) c OJ 3 4J .— 1 1-1 to o > O. B u c 0) 3 cr 01 fc4 OJ ^ .3 v o ni > >, -H 4-J 4J •H cfl U) c CD o pi o l-l CO OJ i-H CO oo H o no r-~ in -j- o\ co t-t o\ 00 CO CM r«« o in o co or 00 CJ\ r^ -3- CO <4D lO r-. co CM i-l rH rH a\ cm ■H CM o •H c CD a CM CM CM CM CM CM CM CM CM CM CM CM f"» 00 r» 00 r- oo r-- oo r^ oo r» oo CTi CTi Oi C* ON Oi o\ o\ a. o\ CT> ON en OJ •H u l-l u 3 « u u CD CO o a c CD 4J a. X! iV, o ■C d l-l CO Bi 00 k: o ►-> < o OJ (X en a cd i-H P. ai OJ c/j 72 heavy use of fragrant sumac plants. Significant declines in shrub coverage was noted only in mature forested types where overstory coverage had increased. The decline in shrub coverage in these types probably resulted from successional trends rather than overuse by deer or other animals. The considerable annual fluctuation in forage production was attributable to annual climatic variation and did not appear to be influenced by deer populations. Long term trends in forage abundance, in addition to being influenced by climate, appeared to be primarily affected by forest maturation and periodic wild fires. Land Use Prehistorically, the "breaks" apparently were a common hunting area for several Indian tribes; none maintained exclusive use of the area, especially over long periods of time. Overall human use of and impact on the land and wildlife was probably relatively low and intermittent. The journals of Lewis and Clark indicate an abundance of wildlife, including the newly described mule deer, in the vicinity of the study area in 1805. Major influent species, which occurred at that time but not at present, include bison (Bison bison), Audubon bighorn sheep (Ovis canadensis auduboni) , grizzly bear (Ursus arctos) , and wolf (Canis lupus). Minor numbers of trappers and traders traversed the area from the early 1800s to the 1860s, when the use of steamboats on the Missouri River led to the first major influx of people. "Woodhawks" became established along the river to supply steamboats with wood, and a small "town" existed at a steamboat landing at Rocky Point on the north-central edge of the study area from the 1860s to the turn of the century. Rocky Point served as a rendezvous for a wide variety of people inhabiting the surrounding area. A short-lived attempt to establish a settlement (Carroll) in the northeastern corner of the study area occurred during 1874-1877. Domestic livestock grazing, which persists as the major land use, became widespread during the early 1880s when large livestock companies moved cattle, sheep, and horses onto the study area and adjacent breaks and plains. The loss of at least half of all livestock in the area during the severe 1886-87 winter reduced total livestock numbers through the 1890s. At the same time, large livestock companies were replaced by an increased number of individual ranchers grazing smaller herds. The Homestead Acts of 1905 and 1909 led to settlement and attempts to cultivate upland sites over much of the area. Settlement peaked during 1908-1914 and included development of 73 a school, post office, and store (Little Crooked) on the south-central edge of the study area and a post office (Wilder) in the north-central portion. It is notable perhaps, that the influx of humans and early agricultural efforts coincided with the longest continuous period of above-normal moisture (1907-1918) in the area since about 1850 (Fig. 3.11). Years of low and high moisture were interspersed during 1919- 1929, resulting in abandonment of the most marginal homestead sites. Extreme drought during 1930-1940 and other factors led to abandonment of most remaining homesteads and general depopulation of the area. By 1960, there were only 3 headquarter ranches and 2 seasonal (haying and winter livestock feeding) operations located on the study area; there were none by the 1970s. The only current inhabitants are employees of the Charles M. Russell National Wildlife Refuge (CMRNWR), Sand Creek Field Station, developed along Highway 191 in the northwest corner of the area during the early 1980s. 2.8 1 Homestead Era (1905-1920) Study Period 1850 1870 1890 1910 1930 Year 1950 1970 1990 Figure 3.11 Relative moisture index for the study area during years 1850 through 1986 calculated from Douglas fir growth ring data. Estimates for 1981-1986 based on our regressions. See Fig. 3.8 for source acknowledgement . Abandonment of homesteads resulted in reversion of nearly all formerly cultivated tracts to grassland and other native vegetation along with seeding of some fields to introduced crested wheatgrass (Agropyron desertorum) . Alfalfa (Medicago 74 sativa) and barley (Hordeum spp.) were grown for hay and wildlife food on several river bottoms until the 1970s when all cropping of bottomlands was phased out and fields were allowed to revert to natural vegetation. The only remaining cultivation occurred as a small area of dryland grains on private lands on the extreme western edge of the study area. Numerous tracts of land privatized during homesteading were repurchased by the Federal government following abandonment or to secure lands adjacent to the Missouri River for construction of the Fort Peck Dam which was completed in 1939. Publicly owned lands now comprise about 70% of the study area. About 45% lies within the CMRNWR, established in 1936 and administered by the U.S. Fish and Wildlife Service. Refuge lands are open to hunting of most Montana game species as provided for by State regulations for hunting districts which encompass the area. Except for about 5% owned by the State of Montana, remaining public lands are administered by the Bureau of Land Management. For grazing purposes, most of the area has been administered as part of the East Indian Buttes State Grazing District, a large common allotment that also includes rolling plains to the southwest. The remainder consists of small units or pastures along the river bottom and two privately-owned upland tracts on which livestock were wintered until the 1970s. About 1600 animal units (AUs) are authorized; however, actual use on the study area has varied greatly from year to year and through time. Grazing is primarily by cattle, but a few horses were included through the years. Domestic sheep were grazed on adjacent plains and into the breaks along the southern edge of the study area until the 1960s. Mule deer are the most common and abundant large wildlife species on the area. Others, in approximate order of abundance are Rocky Mountain elk (Cervus elaphus) , white-tailed deer (O. virginianus) , pronghorn antelope (Antilocapra americana) , and coyote (Canis latrans) . Mule deer, elk, and coyotes range throughout the study area, while whitetails are restricted largely to Missouri River bottoms and antelope range into the area along major ridges. Elk, bison, and grizzly bear were essentially extirpated from the study area by the mid 1880s; wolves were gone by the 1920s. The present elk population developed as a result of release of 31 animals, transplanted from Yellowstone National Park, on the study area in 1951. Since then, numbers and distribution have increased consistently. Coyotes have always been present, but numbers apparently were low during the late 1940s and early 1950s when the toxicants strychnine and compound 1080 were first used in massive predator control 75 campaigns. Numbers gradually increased after the mid-1950s despite some continued control efforts with toxicants through 1971. History of the Mule Deer Population Mule deer apparently were common in the vicinity of our study area during the early 19th Century. As the Lewis and Clark expedition moved through the area in 1805, their journals note that on May 19 the party killed a minimum of 10 deer, on May 20, Clark killed two deer and "the hunters killed . . . several deer merely for their skins", and on May 23, Clark killed four deer in the morning and saw a number of deer in the afternoon" (DeVoto 1953, Burroughs 1961). Later records suggest that deer also were common during the mid-to-late 1800s. Koch (1941), citing journals of his father and others wintering along the Missouri River immediately below our study area in 1869-70, states "[they] ...were able to go out every two or three days, about as one goes to the market now, and bring in a deer or antelope..." Similarly, Hornaday (1908) recounted a successful hunting trip for mule deer bucks in "bad-lands" along the Missouri east of our study area during October 1901. Up to that time deer apparently were so common that "...certain ranchmen of the North Side [of the river] had slaughtered great numbers of... deer to feed their dogs." These sketchy early records indicate only that mule deer were widely distributed and occurred in some abundance during recent historical times. They tell us little about absolute or relative numbers or fluctuations through time. Of this we can only speculate based on more recent population characteristics and habitat relationships of mule deer in the area. It is unlikely, however, that deer were more widely distributed or generally more abundant in historical times than today. They may also have been subject to equal or greater temporal fluctuations in occurrence. Relative moisture indexes calculated from tree-ring data (Fig. 3.8) indicate consistently wide variation in environmental conditions through the years. Periods of drought were interspersed with years of ample moisture, and mild with severe winters. Relatively dry to average conditions prevailed during the time of the Lewis and Clark passage through the area and during the 1860s and 1870s. The 1880s were relatively moist and included the exceptionally severe winter of 1886-87 during which more than half of the cattle herds in the area died. Deer must have suffered substantially as well. All of this would suggest that mule deer numbers and distribution probably fluctuated greatly throughout the 1800s and earlier as they do today. 76 Recollections of long-time residents of the area indicate that deer numbers were relatively high during 1890s and early 1900s but declined to scarcity during the 1920s and early 1930s. This decline, which coincided with great increase in human populations on the area during 1905-1920 and drought from the mid 1920s through the mid 1930s, was also documented in increasingly restricted hunting regulations. A 4-month hunting season with a bag limit of 8 deer established in 1895 was progressively reduced to 3 months and 3 deer in 1905, 2 months and 3 deer (of which 2 had to be bucks) in 1913, 2 months and 1 deer of either sex in 1919, and 2 months and "one male deer with visible horns" in 1921. From 1921 to 1930, the eastern sixty percent of the study area was closed to hunting and, during 1931 and 1932, the entire area was closed. Hunting resumed for bucks-only in 1933, but was prohibited on the portion of the area within the CMRNWR from 1937 through 194 7 and during 1949. An August 1935 survey of the area that later became the CMRNWR by O.J. Murie indicated that "the mule deer has become very scarce on some parts of this range, but is still present and widely distributed." The period from the 1920s to 1935-1937 marked the historical low in mule deer populations on the study area. Some deer persisted, however, and by the time the drought ended in 19 38, most of the human population had left the area, formerly cultivated tracts had reverted to rangelands, and the overall intensity of human usage had been greatly reduced. All of this apparently set the stage for increasing mule deer numbers during the late 1930s into the 1940s (Mackie 1970). The newly established Fort Peck Game Range brought resident observers with biological interest to the area during the late 1930s. Quarterly narratives by Game Range personnel, beginning in 1940, provided qualitative information on wildlife abundance and forage and cover conditions through the 1940s and 1950s. Passage of the Pittman-Robertson Act and its adoption by the Montana Legislature in 1941 enabled the Montana Fish and Game Department to employ biologists and begin collecting quantitative data on and near the study area during 1947. Selected statements from quarterly narratives and other reports during 1940-59 that relate to mule deer populations and factors that we have found to influence mule deer populations are presented in Appendix D. The reader is encouraged to read Appendix D to obtain more specific information as well as "the flavor of the times". Although such qualitative information does not enable us to precisely estimate numbers, it was recorded every year and provided a continuous record of impressions and events. When combined 77 with the more scattered and diverse quantitative data available (e.g. early censuses), and both are interpreted in light of findings from our studies during 1960-87, we believe that reasonable estimates of mule deer numbers and trends during the period can be derived. Deer were still considered scarce in 1940, but immediately thereafter and through 1947, the narratives (Appendix D) indicated that mule deer populations increased rapidly. Concurrently, range and forage conditions were considered "excellent" through 1947. A deer drive count on 4 mi2 in autumn 1944 produced an observed density of 8.5 mule deer/mi2. More recent information from that same area indicates that the actual density could have been as high as 11-17 mule deer/mi2. In 1948, deer were considered sufficiently abundant to permit hunting of bucks within the Game Range for the first time since it's establishment in 1935. Also, Game Range managers and state personnel were starting to discuss the "unmentionable proposition of opening the season on does . " The first substantial quantitative data on mule deer populations were collected by aerial strip census during September 1947 (Brown 1947, Appendix D, Table 3.13). Aerial strip censuses were also conducted during February 1948, 1950, and 1951. These censuses recorded observed densities ranging from 3.4 to 5.7 mule deer/mi2. Our adjustments of those figures (Appendix D, Table 3.13), based on the areas flown and our more recent comparative information, indicate actual densities of 7-12.5 mule deer/mi2. The lowest estimated density (5-8/mi2) was observed during mid-late winter 1950, the second of 2 consecutive severe winters. Pre-hunting season density may have been as high as 15-18 mule deer/mi2 in autumn 1951 (Appendix D) . Good forage conditions prevailed from 1951-1954, but drought and poor forage conditions occurred from mid-summer 1955 through 1958. Fawn survival was good during the early period, but had declined by 1958-59. An attempt to remove all deer from a 3.13 mi2 fenced pasture adjacent to our study area in February 1958 indicated a minimum density of 7.7 mule deer/mi2 (Appendix D, Table 3.13) . The accuracy of early population estimates is subject to question, especially when the individuals reporting and interpreting them did not do the work. We believe, however, that the estimates, and especially the ranges presented (Table 3.13, Appendix D) , are reasonable. 78 Table 3.13. Density estimates for mule deer in northeast Fergus County, Montana, 1944-1958. Date 1943 20 Oct. 1944 Sept. 1947 Feb. 1948 Feb. 1950 Feb. 1951 Feb. 1958 Type of Census Density Estimate (Range) Remarks Unknown -- probably minimal ground counts of selected areas or "a guess . " Drive count of 4 mi2 by 2 people walking. Aerial strip census. Aerial strip census. 1.3/mi2 0.5 /km2 Probably substantial underestimate . 8.5/mi2 Small area sampled. 3. 3 /km2 12.6/mi2 Reasonable. 4. 9 /km2 (10-15/mi2) +202 Aerial strip census. Aerial strip census Attempted total kill within 3.13 mi2 pasture, 8.0/mi2 3.1 /km2 (6-10/mi2) 6. 7 /mi2 2. 6 /km2 (5-8/mi2) 10.0/mi2 3. 9 /km2 (8-12/mi2) 7.7/mi2 2.95/km2 Reasonable. +202 Probably low. +202 Reasonable. +202 Reasonable, small area sampled. Collectively, information and data on historical deer numbers and trend indicate mule deer were at least moderately abundant in the vicinity of our study area throughout the 1800s and early 1900s. Thereafter, mule deer populations on our area declined to historic low levels by the early 19 30s, but started to increase by the late 19 30s and especially early 1940s. The increase in deer numbers after the 1930s has been attributed by various people to the following factors, both singly and in combination: 1) human depopulation of the area - both as a harvest effect and a habitat recovery effect; 2) increased precipitation - the ending of the drought; 3) predator control; and 4) increasingly effective law enforcement. Most of the historically highest human population on the area during the teens and 1920s had abandoned their homesteads 79 and left the area by 1940. These people had to a large extent depended on deer for subsistence and hunted yearlong. The onset of World War II further reduced the population of young men living near the area and war-time rationing of ammunition, tires, and gas further reduced hunting pressure by people from the surrounding area. As Holibaugh (1944) stated, "Perhaps the tire and gas situation has eliminated to a certain extent the hunting, both illegally and legally." Many factors, including illegal hunting; habitat disruption by homesteading, plowing, and timber harvest; severe drought; and vegetation depletion by domestic livestock grazing and grasshoppers (Murie 1935) probably contributed to the decline in deer numbers. The increase in mule deer probably resulted from a decrease in the influence of the above factors, and especially from improved forage conditions. The increase in mule deer numbers coincided with the improved forage conditions that resulted from increased precipitation during 1938, 1941-44, and 1946. Predator control is the most often cited reason for the increase in deer and antelope during the 1940s, both locally and throughout the western United States. Information from this area indicated that mule deer on this area had reached historically high levels by autumn 1947 and possibly as early as 1944. However, the narratives indicate that no substantial, effective coyote control program began until December 1946. Sporadic trapping and very limited poisoning in local areas took place prior to that. Indeed, the narratives and coyote damage complaints indicate that coyote populations and depredation increased from 1940-1946! Mule deer populations had apparently reached such high levels prior to 1947 that Federal biologists already considered the deer range to be approaching "overuse". The evidence did indicate that intensive predator control during the late 1940s and early 1950s resulted in increased mule deer fawn survival, at least through 1956. This increase in fawn survival, however, apparently did not result in much, if any, increase in mule deer populations within the riverbreaks habitat. Our recent information on mule deer dispersal does substantiate, however, local suspicion that predator control may have helped create an "excess" that successfully recolonized surrounding areas of less secure habitat. Although mule deer population recovery occurred within the riverbreaks habitat prior to effective predator control, it may have occurred even faster had predator control been implemented earlier. Yearlong subsistence hunting undoubtedly contributed to low deer populations in the 1920s and 1930s. It is unlikely that the restrictions on hunting effectively protected deer to 80 any great extent. Enforcement was largely lacking, and during the homestead era as well as through subsequent periods of drought and depression, residents shot deer as need and opportunity presented. Depopulation of the area, rather than increased law enforcement, probably contributed most to the decline in illegal harvest of deer. Mule deer populations increased throughout the improved range conditions of the 1940s and declined somewhat during the severe winters of 1948-49 and 1949-50. Populations may then have increased during good conditions in the early 1950s, but declined following drought during 1955 and 1956. Regardless of population level, and in spite of consistently "overused" browse species after 1948, deer apparently remained "in excellent condition", "very fat", and "one of healthiest and most productive [herds] in the country". This seemed to be uniformly true except during periods of drought (mid-late 1950s), when deer were in "bad condition", regardless of numbers. We believe the available data are sufficient to indicate that from 1947-1959, and possibly back to 1944, the mule deer population on our study area fluctuated within the same range of densities as it did during the period 1960-1987. 81 CHAPTER 4 POPULATION CHARACTERISTICS Numbers and Density 1960-1987 Estimated total numbers of mule deer on the study area during autumn, early winter, and spring from 1960 through 1987 are shown in Figure 4.1. Annual and seasonal numerical population estimates by sex and age class are presented in Appendix A. 2000 -l ** c 1600- 1400- CO 1200- LU ■ 1000- a 800- Q) s 600- o 400- a> E 3 200- 0- 1930 Figure 4 .7. Mean 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1940 1950 r 1960 Year 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1970 1980 1987 Estimated mule deer population levels during early winter, 1930-1987. See text for derivation of estimates prior to 1960 and for explanation of curves A (stable growth rate) and B (semi-logistic growth). Mule deer populations during the early to mid 19 30s were extremely low. Old timers talk about the rarity of seeing deer and that seeing a fresh deer track during this period was cause for excitement. When we compare these recollections to our observations during low deer numbers in the mid 1970s (400 deer, 1.5/km2), we believe that fewer than 100 deer (0.4/km2) probably occurred on the study area during the mid-1950s. For purposes of calculation, we chose reasonable estimate of the low. 50 deer (0.2 /km2) as a 89 We also know that population growth began during 1938 or 1939. Deer numbers increased progressively through 1947 or 1948, apparently reaching a population high that couldn't have been much higher than 1500 deer based on both a reasonable aerial strip census estimate plus 20% and practically achievable population growth rates. The population increase was probably halted by emigration of yearlings and mortality during the severe winters of 1948-49 and 1949-50. Between 1938 and 1948, the population grew in a manner similar to that illustrated (Fig. 4.7), regardless of the degree of certainty about the number of years or deer involved. Depending on the number of years involved (8-10) and initial (25-100) and ultimate deer numbers (1200-1500), the average instantaneous rate of growth for the deer population was in a range of r=0 . 248-0 . 484 . Based on growth rates observed during 1960-1987, growth rates in the projected range were achievable during 1938-1948, although a sustained growth rate of 0.484 over 8 years was unlikely. There was at least some hunting loss of adult males during those years and at least a minimal amount of illegal kill of all deer and natural mortality of adult females and fawns must have occurred. Thus, we believe it is reasonable to assume that the average instantaneous growth rate during the period 1938-1947 was less than 0.484. The use of 50 deer in 1938 for a starting point and 1500 deer in 1947 as the population peak (Fig. 4.7) yields an average annual instantaneous growth rate of r = 0.378, near the midpoint of the potential range. Growth curve A (Fig. 4.7) represents an annual growth rate of 0.378 between 1938 and 1947. It is unlikely that the growth rate was precisely the same each year, so the shape of the actual growth curve probably varied somewhat from that plotted. An alternative growth curve (B, Fig. 4.7) assumed a growth rate that decreased steadily with density (semi- logistic) as observed by McCullough (1983) for white-tailed deer on the George Reserve. That curve assumed a maximal growth rate of 0.76 for the initial year, a decline to 0.50 the second year, and then progressive declines with density to about 0.20 during 1946-47. Other alternative growth patterns exist, but none deviate significantly from the general pattern displayed in Figure 4.7. During the period 1938-1948, the population probably grew in a manner similar to irruptive or introduced populations (McCullough 1979 and 1983, Caughley 1970). Subsequent to 1948, the population fluctuated widely around what some might call an "equilibrium density" of 3.8 deer/km2. The major body of theoretical population work centers around the irruptive 90 growth phase of populations. Our results are mainly concerned with the fluctuations of an established population. Population Composition Sex Ratio Three separate data sets, each collected over periods of 7-12 years, provided estimates of sex ratio among fawns from birth through 6-9 months of age. A ratio of 110 males: 100 females was indicated by a combined sample of 218 newborn fawns captured and marked during mid-June. Ratios of 93 and 106 males: 100 females were indicated by samples of 145 fawns killed by hunters and 74 fawns captured and marked during winter, respectively. Those ratios were not significantly different from each other or from a 1:1 ratio (chi square tests, all P > 0.50). The overall average (103 males: 100 females, n=437) indicated that sex ratios were nearly equal or slightly favored males during the first year of life as generally reported in the literature. Sample sizes were inadequate to detect possible differences among years. Among adults, females always outnumbered males (Tables 4.1 and 4.2). Classification in autumn, prior to the hunting season, averaged 42 males:100 females (range, 20-64:100). Early winter (post-season) ratios averaged 31 males: 100 females and ranged from 13 to 50:100. The decline in proportion of males to females after the first year is not surprising in a population where males are harvested more heavily than females; though other factors may also have been involved. Martinka (1978) found adult females to outnumber adult males by more than 2:1 in an unhunted mule deer population. Gavin et al . (1984) reported 3 females per male and Kie and White (1985) 2.6 females per male in unhunted populations of white-tailed deer. Sex ratios fluctuated widely among years as a result of variable and selective hunting pressure and variable fawn recruitment. Because of the relatively smaller numbers of adult males compared to females, the annual cohort of yearlings made up a larger portion of the male population than the female population. Population estimates indicated that yearlings comprised an average of 18% (3-30%) of the female population and 43% (10-66%) of the male population. Age Structure Fawns As A Proportion Of The Population At birth pulse, newborn fawns comprised up to 55% of the population. Because fawns experienced higher mortality rates than adults, the proportion of fawns declined from June 91 cn CD > U 3 en u O) m 71 ti eo •n c 3 o In 00 G H 3 +J 3 CO a o m 4-1 TJ (1) C •rH e (0 CD U CD m 4_) 01 3 * 4-1 C in o •H en 4-) •H en en CD u l-i u. m y o i-i u CD > U •H o m •H J-J •H CO l-l rH 3 3 O u. en o en u. •H n 1-1 CD » CD vn 73 oo CD o .-1 vo 3 o x: rH 01 rO CO o o 3 73 < o o rH cn .. 0) cn rH r] CO [* a 01 CD L>_, u. o o rH e/> .. o> tn rH (II to rH a 10 CD rE u. CO rlH 3 < en CD rH CO G CD ■M tn rH ft) 3 rH 73 (0 <: S 73 0) U •H CD C4H rO -H a en 3 tn 2 eo o c o <4H -H O 4J CO OJ > Q. >H 5»% 01 Eh CO rQ o u co 01 orococ\itnr~in-3'iniri-3'rH-*co<ooooi/ivoi/i vo avOcooo-ooor^csii/ieNOi^ m oom-s-oir^vocoocNcorvr^f^^ovo i^ OrOlO^ONNCOO-IOairHcIlcOCD ncoioocooioiNN-fHOiniiincji ■jNincotototo-jineoin-j^eocNH ")ooo tOfflHlO-JOlONfl-t-JtO-JteiaeO coNCMroconm-3p-*cocco>oor^ -3-r^CT>^-OlOrHr^r~.-3-r--OrH000\0l on cm i-ii-irHrHrHcMiovomr^-J-rod CNl CN CO c 3 o fO c0 tO eoeoeOeoeOeOeOeO o o o o OHNrliniONCOOlOHNrl^intO vovovovor^r^r^r^r^cooooooooocooo ejio^o*(no>a\c^o\cjia\CTic^o\cTioo\ C efl CD 00 I o VO c OJ CD e CD B o <-W T3 CD -o 3 rH U X 01 01 in 0) > ■a c CO CO iH 3 O cj CO c CD l-l OJ •a •H cn C O u CO eo e CD O O CO E •O c cO cn 0! i-H CO e o> 00 cfl c ai u u 01 0-. 92 •o c si T3 c 3 o on a o u X) 0) c •H s l-l 0) w CD T3 M en 3 C co • ►o CO CD t»i M .— 1 m u cd >, CD X) 3 H 4-> o CO u CO 0) .* X (0 B CD o> M CJ PQ 01 P lJ CD 00 > a •H •H tti u T> ■H X) l-i 3 c O o CO •H U) ■M ■H •H S CO o ■ a.1^- S CD O 0\ u H c O O VO •H ai 4-> H CO rH » 3 co CX >-» o CD O. > l-l Vi 3 CD en CD XlrH CO CD •H H Vj 3 CD x: 10 r-J ■* CD rH 43 al E-. co dl CD CO .—I u at CO CD c I- 3 -H 1— 1 ct-c 4-> W O CO CO J* CD «■* 4J >H en •< go CO o co Pl, go a) o &4 CO -cf CO NC^ COCO in r^ co ■* -J co J MOCOO •a- -j- lOrHcO-3-r^cOOOOcOrHOCJirHr~- COONcnNOONOH j-com-3- (X H >-. CD E-" co X o cd jd u T) m X)XlX)rHrHrHrHrH i— IHHrHH CCCccJcOcOcOcO<3c0c0c0cOc0 3 3 3 -H -H -H -H -H O -H -H -H -H -H OOOMMMUMPdMUUV-IM MWMCDCDCDCDCDCJCDCDCDCDCD cOcOcOcOcOcOcOcOcOcOcOcO •H -H -H -H -H -H -H -H -H -H -H uuumUUV-Immmmu CDOJCDCDCDCDCDCDCDCDCDCD HNcOOr-CDOlOHCN-JcOlOM»OlOHNcO-JcOcOr- lOcOiOlOCOiOcOcOcOr^M^tNMvM^rNOOOOCOCOCOOOCOCO I I 1 I I I I I I I I I I I I I I I I I I I I I I OHMcO-CTlOcONOOOlOHcO'tincOrscOOlOHNcO-cflO'O covocococovo^ococo^or^i^r^rvtsi^Nr^r^cocococDCDooco CJ\CTcC^O\<^CJcc^CJ\C^CJ\CJ\C^O>CJcC^C^CJ\CJcOlO\CTiC^CJiCJcCTc CD CO CD u CO -a 3 4-> o CTc • CO u a\ CD rH X e 4J 0) P. 01 U a x CD 00 CO -H CD rH u CD CO X. c r^ o xi CO CO o o 4-1 c 0) u CO •r-1 ■v CO 00 II CO I < r- O CO <*! o\ o rH 2 c - ■H CO CD X) l-l CD CO CD >-, >T3 U 3 3 4-1 CO CO a) CO > > M M CD 01 CO CO XI XI o o 0) rH > CO •H -H 4-> M cd CD rH CO 3 B rH 3 rH c_> < CO 0) cd x: U 4J cd 4-1 i-l CO 3 C u 4-1 O X O CD CO (Xj c o CO -H X 4-1' CO 4J > CO Vj X) 0) 0) 3 CO X >■> o DO G •H M 3 CD > w C O 3 O 2 en IB £ U O 41 93 through May during all years. The variable recruitment rates to autumn, early winter, and spring for fawns (Tables 4.1, 4.2, and 4.3) resulted in considerable variation in the proportion of fawns in the population during any season or year. Fawn: female ratios at birth pulse averaged 127:100 and ranged from 99 to 148 fawns: 100 females during 1975-1986. There was an average of 74 fawns: 100 females during autumn, but ratios varied considerably from 31 to 120 fawns: 100 females. During early winter, the average ratio dropped to 65 fawns: 100 females, within a range of 29-116 fawns: 100 females (Fig. 4.8). In spring, the average was 39 fawns: 100 adult deer and the range was 5 to 82 fawns: 100 adults. Conversion of these data to fawn: 100 female ratios, based on estimated numbers of males in spring populations, yielded an average of 50 fawns: 100 females with a range of 6 to 103 fawns: 100 females . CO ' E 0) 0) +-> LL L. O o £ >. . , i_ CO c CD LU 5 ■ CO 120 -i 110 - 100 - 90 - 80 - 70 - 60 - 50 - 40 - 30 - 20 - 10 - 1960 1965 1970 1975 Year 1980 1985 Figure 4.8. Fawns: 100 females during early winter, 1960-1987 94 Table 4.3. Mule deer population composition during spring, as determined from ground and aerial surveys, 1961-1987, Missouri River Breaks study area. Type of Observation Number Classified Percent aee Fawns: 100 Year Adults Fawns Adults 1961 Ground 378 58.7 41.3 70.3 1962 Ground 289 74.7 25.3 33.8 1963 Ground 177 72.3 27.7 38.3 1964 Ground 57 68.4 31.6 46.2 1972 Ground 66 89.4 10.6 11.9 1973 Aerial 235 80.9 19.1 23.7 1976 Aerial 192 90.1 9.9 11.0 1977 Aerial 293 76.8 23.2 30.2 1978 Aerial 246 79.3 20.7 26.2 1979 Aerial 409 60.6 39.4 64.9 1980 Aerial 683 54.9 45.1 82.1 1981 Aerial 738 62.1 37.9 61.1 1982 Aerial3 869 65.6 34.4 52.5 Aerialb 295 69.5 30.5 43.9 1983 Aerial3 694 62.1 37.9 61.0 Aerialb 488 62.9 37.1 59.0 1984 Aerial3 846 80.3 19.7 19.7 Aerial0 122 88.5 11.5 13.0 1985 Aerial 643 95.0 5.0 5.2 1986 Aerial 566 85.0 15.0 17.7 1987 Aerial 861 60.7 39.3 64.6 1961-1987 Mean 73.9 26.1 39.1 a March b April c May Adult Females Data on the age structure of adult female populations were available from deer killed by hunters for 1960-64 and from hunter-kills and winter trapping during 1978-1985 (Table 4.4). Age distribution among deer killed by hunters may not have been representative of the population age structure because some age-classes of females, especially yearlings and the oldest females, appeared to be harvested at higher rates than others (see Chapter 6). Nevertheless, these data, if not mathematically reconcilable from age-class to age-class, appeared to represent the general changes in age structure. 95 Table 4.4. Adult female age structure from hunter-killed and captured mule deer, expressed as percentages in each age class. Age in Years 1960 46 22 13 1961 63 21 8 1962 35 11 31 1963 10 10 20 1964 19 32 21 1960-64 173 20 16 1978 33 21 24 1979 10 40 0 1980 17 29 6 1981 55 20 29 1978-81 115 23 22 1982 55 33 29 1983 49 31 16 1984 47 13 34 1985 27 4 0 Mean 13 years 22 18 Year N 1 2 3-6 7+ 28 37 54 17 37 20 40 30 37 11 41 23 36 18 50 10 47 18 35 17 38 17 33 6 42 10 40 12 85 11 43 17 For example, yearlings were relatively poorly represented in the population during autumn 1962 and 1963 (Table 4.4) as expected based on relatively poor fawn recruitment during 1961 and 1962 as compared with 1960 and 1963. Poor recruitment of the 1983 and 1984 cohorts was also indicated by the hunter- killed samples (Table 4.4). There was little difference in age structure for a period of high harvest rates for females (1960-1964) and a period of no harvest of females (1978-1981, Table 4.4). Females had not been legally harvested for 6 years prior to autumn 1981. The period of 1978-1981 was a time of population growth and young deer were well represented in the population. Modeled estimates of age structure each year from 1968 through 1986 are shown in Fig. 4.9. The model added the number of fawns recruited from each cohort starting with 1960, and annually reduced those numbers based on estimates of hunting and winter losses and dispersal, before advancing it to the next age class. Age-specific differences in mortality (Chapter 6) were also considered. For years when data were 96 1 2345678 9+ Age Class 1 2345678 9+ Age Class 1 2345678 9+ Age Class Figure 4.9. Modeled age structure of the female population during spring, 1968-1986. Age class 1 is 12 month-old females. 97 also available from trapping and hunter-killed samples, we considered the modeled estimates more accurate than those derived from small samples of trapped and hunter-killed deer. Given those small samples, randomness, hunter selection, and misaging could significantly influence proportions assigned to each age class. The general form and structure of the 2 estimates was similar, though age-specific differences appear (Fig. 4.10). Both generally detected cohorts with poor survival (1971, 1975, 1983, and 1984). The general age structure of the population was the same during 2 different periods of population increase, 1968-1971 and 1979-1983 (Fig. 4.9). Age structure was also similar for 2 periods of decline, 1972-1974 and 1984-1986. The major difference was that the population decline during 1972-1974 was more severe and population size was lower than during 1984-1986. Continued low fawn recruitment during 1975-1977 further reduced population size and perpetuated the small base in female age structure during the earlier decline. It is apparent from age structural estimates (Fig. 4.9 and Table 4.4) that age structure alone is not a useful criterion for predicting future population trends. Rather, it was most useful for determining or verifying what had happened in the past. Used alone, or with invalid assumptions, age structural data may be potentially misleading in determining population trend. Population declines during both 1971-72 and 1983-84 were preceded by several years in which the pyramidal age structure represented the classic, "healthy and growing" population. Both declines were represented by sharp, rather than gradual deviations from this age structure. A stable age structure cannot be expected for this population. Persistence of a "normal" or "stable" age structure requires relatively constant natality and mortality. On our study area, natality and mortality of fawns occurred as periods of several years of both boom and bust. Hunting mortality was variable. Although natural mortality of adults appeared relatively stable during most years, it occasionally occurred as a pulse of catastrophic mortality. The greatest age structural stability occurred from spring 1975 through spring 1978 (Fig. 4.9), when fawn survival was low, but relatively stable, and adult female mortality was low and stable. Only 16.5% of 466 adult females aged during 1960-1985 were 7 years or older; 40% were yearlings or 2-years-old. The maximum age recorded for females was 16 1/2 years. 98 ■ Age Structure from Model 0 Age Structure based on trapped and hunter-killed sample 1981 3 4 5 6 7 Age Class 3 4 5 6 7 Age Class Figure 4.10. A comparison during six years of female age structure based on the modeled estimate with female age structure based on a sample of trapped and hunter-killed deer. 99 Adult Males Information on the age structure of adult male populations was provided by deer killed by hunters during 1960-1964 and 1976-1986 (Table 4.5). Male age structure and changes were also estimated using the model that added in each recruited cohort, starting with 1960, and reduced its numbers on the basis of annual and average estimates of hunting and winter losses (Fig. 4.11). Age-specific differences in mortality were applied to 3 age classes: yearlings, 2-year- olds, and 3-year-old and older males, based on mortality data from 1976-1986 (Chapter 6). Data were also available to age males as either yearlings or older adults from aerial classifications during 1976-1986 (Table 4.6). The 3 estimates of male age structure generally followed the same trend, though some differences were apparent (Table 4.6). Aerial classifications usually over-estimated yearling males as compared to estimates based on fawn recruitment. That probably occurred because, from the air, some small- antlered, 2-year-old males were misclassif ied as yearlings. The proportion of yearling males in the harvested sample was lower than estimates based on fawn recruitment. That was also expected because age-specific mortality data indicated that yearling males were harvested at a lower rate than older and larger males. Because those 2 estimates deviated as expected from the estimate based on fawn recruitment, we considered the estimates based on fawn recruitment to be the most accurate. Although age structures for harvested males, 1960-1964 and 1976-1986 (Table 4.5), probably underestimated proportions of yearling males and overestimated older males in the population, they did indicate the numerical importance of yearling males to the total harvest, even during years following poor fawn recruitment. The modeled age structure (Fig. 4.11) also indicated the importance of yearling males to the surviving male population. Because the total male population after the hunting season was usually one-third or less that of the female population, the size of each annual cohort had a proportionately greater effect on numbers of males in the population than on female numbers (Figs. 4.9 and 4.11). As a result of hunter selection for older, larger- antlered males, the male age structure generally retained a more pyramidal shape than that for females, even at low population levels following poor fawn recruitment (Figs. 4.9 and 4.11). Harvest rates of males that averaged over 40% reduced the influence of strong year classes much sooner than occurred among the female population. Only 3.4% of 739 adult males aged during 1960-1986 were 7 years or older. Seventy percent (70%) were yearlings or 2 years old. The oldest male was 11 1/2 years-old. 100 1 234 5678 9* VtOinrirtrtrT 200 2345678 9+ 12345678 9+ 1984 200 150 100 234 5678 9+ 1234 5678 9+ 1982 200M 1983 150 100 50H 234 5678 9+ Age Class 1985 200 150 100 50 ««- 1234 5678 9+ u 1 1986 lav? *aa 234 5678 9+ Age Class 12345678 9+12345678 9+ Age Class Age Class Figure 4.11 Modeled age structure of the male population during spring, 1968-1986. 101 Table A. 5. Adult male age structure from hunter-killed deer, expressed as percentages in each age class. 1960 1961 1962 1963 1964 Year N 73 66 31 16 43 Age in Years 53 59 42 50 47 18 14 48 19 16 - 6 21 20 10 31 30 7 + 8 8 0 0 7 1976 1977 1978 1979 16 21 11 29 50 38 18 52 25 14 31 27 25 39 51 21 1980 1981 1982 1983 1984 1985 1986 71 102 54 95 49 16 46 54 49 41 55 22 0 20 21 36 36 17 39 19 20 22 11 23 27 39 75 58 3 4 0 1 0 6 2 Mean 16 years 41 25 31 Table 4.6. Percentage of yearling mule deer in the male segment of the population during autumn as determined by three methods, 1976-1986. Year Methc )d Spring Autumn Population Aerial Hunter Estimates Classifications Harvest 20 39 (31)a 50 (16)a 39 41 (20) 38 (21) 44 52 (33) 18 (11) 58 65 (57) 52 (29) 66 62 (139) 54 (71) 59 55 (151) 49 (102) 53 63 (96) 41 (54) 63 67 (146) 55 (95) 21 32 (107) 22 (49) 10 10 (61) 0 (16) 38 29 (42) 20 (46) 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 Mean 43 47 36 a Number in parenthesis is the sample size. 102 Mule Deer Condition Antler Characteristics Antler characteristics, especially of yearling males, have been related to nutritional qualities of the diet and physical condition in deer (French et al . 1956, McEwen et al . 1957, Robinette et al . 1973, and Rasmussen 1985). Generally, antler size as expressed by total weight, beam diameter, number of points, and beam length increases with nutritional plane. Antler characteristics of mule deer on our study area (Table 4.7) indicated that yearling males were in relatively poor condition during 1961, 1970, 1984, and 1985. They also may have been in relatively poor condition during 1964. Better than average condition was apparent for 1963, 1974, 1978, 1979, and 1987. Data for other years either indicated average condition or conflicting conclusions were given by beam diameter and antler point measurements. Although antler measurements did not exactly reflect overall deer herd performance in all years, they accurately reflected forage conditions and fawn survival during years when antler measurements were at extremes in either direction. Because antler measurements may also be affected by conditions prevailing at birth, genetic factors, hunter selection, and stocastity, "noise" may mask changes resulting from nutritional status during years of less than extreme condition. Weights Relative deer weights by sex and age class have also been considered an indicator of nutritional status (McEwen et al . 1957 and Klein 1970). Limited data on deer weights recorded during some years of the study (Table 4.8) verified that all deer were in relatively poor condition during 1961 and 1984. During 1962, the weights of yearling and 2-year-old males as well as 2-year-old-and-older females were above average despite poor fawn survival. Weights of adult females were probably above average that year because few females recruited fawns to autumn during either 1961 or 1962 and, relieved of lactation stress and with good forage conditions after June, most females recovered body condition by autumn 1962. Weights of adult females were usually below or near average during years of drought and/or during years following several years of above average fawn recruitment. For example, antler beam diameter (Table 4.7) and weights of yearling males (Table 4.8) were above average during 1981, but weights of adult females were slightly below average after having 103 CO Oi o vO 0> 3 en d U co l-i 1) > In 3 O CD x: c o u CD CD T3 3 E l-i O en u •H M m 4-1 U « co U ^-- l-i e 3 B X) CD IH > a> o 4-1 XI OJ cd a a s ■H U X) S m cd • 0) CM ca en i-i oo o .H CD CM CO O CO > • • • 1-1 co cm co CM rd as 4-1 cn C O O o. »i i ct) > m In co r^mocor-~i-i-3- -* o N Ol ul CM N H CO CO CM CO CO CM CO <3- o o cm rH -3- r^ .-I CM rH rH rH CM 00 VO -* CM vo <- CO CO vo rHcocNCMCMintocomcovococMOcDin ^J-vor^oovOvor^aocor^Lnr^r^^cMin NOCO<0(MIOOOOlOMn( -t n in m co CM CM CM CM CM riooi com ~3- ■» io -j m co i/i m CM CMCMCMCMCMCMCM -* O r^ O rH -3- f-^ rH CM rH rH en CO CO CO rH r^ CD t-H CM CM rH CO r~l CO 00 o\ VO rH co VO -* vo 01 VO iOs oiom vo vo oo ct> r- CD ^ o 4J ■H a 4-1 3 CO O ■H O > 01 en T3 a. •H •a ■ rH cd T3 -H c 4-1 •H •H en x: xi *■"' Cu id 0 U A u cd 4-1 4-1 3 a* CD O l*> B 00 ON CO rH •H H T3 O -3 u C E •+-I to CO ID m co XI 4-1 r- CO cr- X T3 rH 104 CO O Ol m OJ (tl id OJ u « u > ■H Pi o in OJ a o u 14-1 CD a; -a 3 e o DO |X a co P CO IX a CO I* X x: 60 •H (V a & M CO c tu s .—1 3 id 4-1 s IX 3 < 2 00 -3- OJ ■-H u XI a) 01 0) 6-i >H oi m io h oi ■!• <r -I -* -* -d- -* - Is". vo CO ro CO CO CO VO OJ CO CO CO VO m CO CO H N H N m K) H N rH O O CM CM CM CM CM m n co vo rH m 00 CM Is-- vo -a rH as r*. oo oi m H co in rH o\ rH CD X -H rd 3 • 00 r- CT> co CM CM X 4-1 01 a xt i~* Oi rH 4-1 X 01 00 rH ■o QJ *J 4H in <* - — 4-1 CO o •-• ro cd X 3 ^ 4J •H 1 CO m x r-H 4_» in c QJ X) U c* rH rH rH CO jQ 01 -r| QJ td ^ s QJ femal pring, CO 00 O CO to TJ 1 — 1 lO ou r- rH 4-1 c cd C 00 * 2. £ o 4J u 0J • 3 cd CJ QJ o " o\ t-i rH 43 QJ U X 3-2 rH T> cu 1/5 >> CO C 0 ■H 4-1 .. c cd CM o-i CO CM CM CM (1) l-l 3 a - ^ rH 4-1 u 00 en CM r- CM I--. ^1 3 cd 0) CTl rH rH co i_i ^ W> rH ;timated mn, wint rH QJ X) rH O 1J QJ S h o S a u i— i CD U rH r— > Oi rH rH o> cd Cd QJ 4-1 S3 iH i"» IH •H •a CO QJ rH 3 rH cd 3 CO ratios , uring a o • 3 4-J U CO IH O CM lO CO co 00 CO Cd O 0) o 00 co CN 00 00 r» u qj •r) rH a CTi r-\ rH cd :►, 4-J cd rH QJ Cd a u >, >-. 14 Ql QJ female atios d U • 4H Xj CM 01 c U a ™ > o QJ i— 1 QJ o> m in o ^o o £} .. r~- 00 m o CM rH cn * QJ °"d" rH rt rl" rH *§ 4H Q> co 5 o) cd O 4h X) a ai en -3- rH CM rH 00 in CO QJ ^ id CO a u o 4-4 rH u u = CO JO 3 > vO in rH CM H QJ 2 p. +-> cd r- in • ■j- t-i cd o •H ■^ ° cd OI rH rH 4-1 X) rH ^ n a - rH C ' QJ U S a ■H 0) 3 C CO > Cd a> CO 4-> C cd 3 QJ Vh oi m -H en 3 4h cd 4-J OI Cd T-, w •-> HH rH O C CO CO X cd X) rH cd xi 4-1 O 3 C QJ • o (0 • ■H ■rl QJ o ■H 4J CM •H rH .o *j • U 4J > o rH 4-1 CO • VH CO 0) O, u o, c u 4H cd X O QJ in QJ •H C QJ QJ < QJ QJ 3 00 ■H > P-i •u 3 CO Q 4J CO QJ 4J -rl 4-1 IH 0) c >-> >-, O r2 QJ U rH Cd Cd rH QJ QJ I 0) 0J t-i P-, o t/1 < 4H Dm X XI a 4J T3 4-1 4-> u cd •H O •H cd cd cd E-" IH P-. E t-i rH 0) (a JD u TJ 116 4-1 o in u o> u t_ OJ a Kl «- O a) t- r- o o o o o o o h- CO o o «- o o o o o o m a U 0) <1) +J on ( ) r i id 0) 4) 4-) c C +-) O J =1 u 00 b O o o m oo UO -0 *- -J- 00 O o- a a oo 117 number of fawns lost annually. Recruitment of fawns to 1 year of age averaged 35% of maximum potential and ranged from 83% in 1979 to 4% in 1984. Generally, the loss of potential production prior to mid-June was low, but was exceptionally high during 1984 (38%) and 1985 (42%). Losses from the estimated maximum production to mid-June included non- breeding, non-pregnancy, less than maximum ovulation and fertilization rate, abortion, resorption, stillbirth, and immediate post-partum abandonment. Of fawns surviving to mid-June, an average of 42% (range 4-69%, Table 5.3) died by late September. The lowest rate of mortality during summer was for the 1979 cohort; the highest for the 1985 cohort. The rate of fawn mortality during summer was less variable than during autumn and winter (Table 5.3). Mortality was exceptionally low during summer 1979, but exceeded 30% during all other summers except 1978 (28%). Of all fawns dying between mid-June and the following May, 65% died by the end of September. Mortality rate of fawns during autumn averaged 12% (range 0-35%), was lower than during any other season (Table 5.3), was more variable (CV=113%) than during any other season, and was generally either very low or very high. Highest mortality was observed during 1975, 1976, 1977, and 1984. Overall, however, only 9% of the fawns that died during their first year of life died during autumn. Over-winter mortality rates averaged 33% (range 5-82%) and accounted for 26% of all deaths of fawns. Although winter was longer than other periods, almost all mortality occurred by the end of March, so mortality rates were generally comparable among seasons. Mortality rates during winter were high during 1975-76, 1983-84, and 1984-85; moderate during 1977-78, 1981-82, and 1985-86; and lower during other years. Total annual mortality for fawns that were alive in mid- June averaged 62% and ranged from 17 to 94%. Annual mortality was less than 50% during only 4 of 12 years. Despite annual mortality of fawns that usually exceeded 50%, population growth occurred in some of these years. With no hunting mortality, adult female numbers increased slightly from spring 1976 to spring 1978 despite annual fawn mortality of 72% for the 1976 cohort and 75% for the 1977 cohort. Adult female numbers increased substantially from 1978 to 1979, when annual fawn mortality was 43%. Despite annual fawn mortality of 56% and 42% and moderate hunting mortality of adult females, the number of adult females also increased somewhat from spring 1981 through spring 1983. Seasonal estimates of fawn mortality were further refined by calculating instantaneous monthly rates of mortality for 118 radio-collared fawns during 1976-1986 (Fig. 5.2). Mortality rates were highest ( r>0 . 10/month) during June, July, August, and February. Mortality was moderate during September, October, January, and March, and very light during November, December, April, and May. o *- CO DC V) c >. 5 CO CO u_ r o 1— n 2» LL in -C -j •*-* n i- 0) o c ^ CO ■*- I— c Q) CO n CO c 0.25 -i 0.00 i 1 1 1 1 1 1 1 1 1 r Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Month N 168 151 129 106 95 86 82 116 110 109 107 106 Figure 5.2 Instantaneous monthly mortality rate for radio- collared mule deer fawns, 1976-1986. 1960-1974 Early winter classifications were made during most years of this period (Table 4.2), but autumn (Table 4.1) and spring (Table 4.3) classifications were made during few years prior to 1975. Assuming average fawn production each year, annual first-year mortality rates were 43%, 69%, 62%, and 53% for the 1960-1963 cohorts. Annual mortality was 89% for the 1971 cohort and 77% for the 1972 cohort. Because spring classifications were not made during other years, annual mortality could not be calculated directly. Observed fawn mortality rates were 12%, 24%, <1%, and 14% during autumn; and 4%, <1%, <1%, and 17% during winter for the 1960-1963 cohorts, respectively. Observed winter fawn mortality rate was approximately 77% for the 1971 cohort. 119 Mortality during other years was calculated from population estimates and/or average mortality rate for other years with similar weather patterns. Calculated fawn mortality rates for 1960-1974 were similar to those observed during 1975-1986. Autumn mortality was generally low, as during 1975-1986, but was somewhat higher during 1960-1963 because of higher harvests of antlerless deer. Autumn mortality was highest during the severe drought year of 1961. Winter mortality was generally low during both periods, except for occasional catastrophic mortality. Age-specific Fawn Production and Recruitment Fawn production and survival records were obtained for 114 marked females during June 1976-May 1984 (Hamlin and Mackie 1987). Twenty-seven provided information only on breeding success as fawns. The remaining 87 females included 34 (39%) known-aged deer captured as newborn fawns or as 1/2 or 1-1/2 year olds. Another 16 (18%) were aged by eruption- wear techniques (Robinette et al . 1957) at capture and later by dental cementum techniques at death. Thirty-seven females (43%) were aged only by eruption-wear. Data on fawn production and survival were not complete for all females during each period of the year, thus data presented here are a composite, by age class, over the 8 year period. None of 41 female fawns sampled had bred (Fig. 5.3). Among older females, production increased steadily with age, reaching a peak for females breeding at 5-1/2-years and giving birth at 6 years. Production declined for females giving birth at 7 years and remained lower than for 3 to 6 year old females thereafter. Production for 7-year-old females was 12%, 16%, and 21% lower than for 4-, 5-, and 6-year-old females, respectively. Production was similar for 3- and 7- year-old females, but 7 year old and older females produced 22% more fawns than first year breeders. Females of ages 4, 5, and 6 years had the highest recruitment rate of fawns to 6 months. Recruitment to 6 months declined substantially for 7-year-old females and increased somewhat thereafter, but never returned to rates sustained by 4-, 5-, and 6-year-old females. Recruitment of fawns to 12 months was highest for females giving birth at 4 and 5 years and declined considerably after 5 years, generally remaining higher only than recruitment rates of first year breeders. The highest rate of fawn loss during summer and autumn occurred among females giving birth at 3 and 7 years; the lowest was for 4- and 5-year-old females. The highest rates of over-winter fawn loss were for 2-, 6-, 8-, and 9-year-old 120 2.0 -| 1.8- 1.6- 1.4- 1 1-2 H ^ 1.0 J (A | 0.8- LL 0.6 -| 0.4 0.2 -| 0.0 Fawns : Pregnant Female Initial Production Recruitment to 6 Mo. Recruitment to 12 Mo. ^100 Pregnancy Rate -75 -50 0) 3 O -25 0) tt> T 1 1 7 8&9 10+ Female age at Parturition N 41 28 27 33 23 23 19 23 18 Initial Production N 27 29 29 24 19 17 20 17 Recruitment to 6 Months N 22 25 21 17 13 15 17 12 Recruitment to 12 Months Figure 5.3. Age-specific production and recruitment of fawns by female mule deer in the Missouri River Breaks, Montana, 1976-1984. females, and the lowest were for 4-, 7-, and 10+ year olds. The highest rates of fawn loss over the entire 12 month period occurred among 2-, 6-, 8-, and 9-year-old females. Recruitment of fawns to 12 months was the first parameter to decline with age, decreasing slightly from 4- to 5-year-old females and substantially from 5- to 6-year-olds. Recruitment to 6 months declined slightly from age 5 to 6 and substantially from age 6 to 7 . Most of this decline was related to a decline in pregnancy rate, because the number of fawns per pregnant female did not decline until between 7 and 8 years of age. 121 Speculations about the effects of previous reproductive status on age-specific recruitment rates were based on 87 pairs of years . We used recruitment to 6 months as the reproductive parameter because it indicated that the female had undergone full lactation stress. Sample size by age class ranged from 10 to 15 cases. Over all age classes, 55% of females recruited fawns to 6 months in successive years, 38% in alternate years, and 7% did not recruit fawns for 2 successive years. Females of age 4, 5, and 6 years were most likely to recruit fawns when they had also recruited fawns the previous year (71%). Eight and 9-year-old females were least likely to recruit fawns in successive years (27%). Females of ages 3, 8, and 9 years were most likely to recruit fawns in only 1 of 2 years (57%). Trends in age-specific fawn production and recruitment (Fig. 5.3) were generally typical of mammals (Caughley 1977); climbing from puberty and forming a convex plateau. However, relative declines after middle age were more severe than typically seen. Data presented by Clutton-Brock et al . (1982) also indicated a similar abrupt decline in fecundity for middle-aged red deer hinds on the island of Rhum. Their data, however, indicated a subsequent recovery in fecundity that we did not observe. Some additional data were collected on reproduction and recruitment by age class during 1984 and 1985, both years of drought and nutritional stress. Data for recruitment to late autumn for small samples of females observed during 1984 and 1985 (Table 5.4) indicated that all age classes of females produced and recruited fewer fawns than during 1976-1983. Very few fawns of the 1984 and 1985 cohorts were recruited to 1 year of age (6 fawns:100 females in 1984 and 21:100 in 1985). Females giving birth at 3 years and at 7 years and older, declined the least in fawn recruitment relative to the average for 1976-1983. First year breeders and females giving birth at 4 , 5, and 6 years underwent the greatest decline in recruitment . Nutritional stress appeared to have greatest influence on first time breeders, as commonly reported, but females normally producing the most fawns (4-, 5-, and 6-year-olds) also were greatly affected. Except for first year breeders, females giving birth at 3 years and 7 years of age and older had recruited fewer fawns per capita the prior year than other deer (Table 5.4). Thus, at least some members of those age classes were most likely to be in the best condition during 1984 and 1985 and consequently produced most of the recruited fawns. Four- to 6-year-old females, that usually recruited the highest number of fawns per capita, may have been in the poorest condition going into 1984 and 1985 because of cumulative reproductive stress, and thus experienced the 122 Table 5.4. Age-specific fawn recruitment to late autumn for female mule deer during 1984 and 1985, compared to the average for 1976-1983. 198^ i and 198! 1976-1983 Fawns 198 Female Number Number Fawns i4 and 1985 Age of of Per Pei as a . Percentage Class Females Fawns Female Feme tie of 1976-1983 1 -2a 8 1 0.13 0.81 (27)b 16 2 -3 11 5 0.45 0.93 (29) 48 3 -4 13 4 0.31 1.34 (29) 23 4 -5 12 4 0.33 1.46 (24) 23 5 -6 10 2 0.20 1.37 (19) 15 6 -7 5 2 0.40 0.88 (17) 45 7 -9 9 4 0.44 1.00 (20) 44 9 + 7 4 0,57 1.06 (17) 54 TOTAL 75 26 0.35 0.90 (182) 39 a Age at conception -- age at parturition. b Female sample size. greatest decline in recruitment of fawns. First year breeders experienced an equally great decline in recruitment, probably because more than for other age groups, nutritional demands of body growth competed with reproductive demands. Based upon estimated age structure, 3-year-old and 7- year-old-and-older females produced 54% of fawns surviving through late autumn in 1984 and 1985. If recruitment had been equivalent to the average for 1976-1983, those age classes would have produced 32% of surviving fawns. However, the difference in survival by female age class had no measurable impact on the population because few fawns of the 1984 and 1985 cohorts survived to 1 year of age regardless of the age of their mother. Population Effects of Age-Specific Recruitment We hypothesized that age-specific variation in recruitment of fawns, in conjunction with changes in age structure of the female population, might explain declines in fawn survival on the area during the early-mid 1970s. However, analyses using several different models did not support that hypothesis. Numbers of fawns produced and fawn: 100 female ratios generated by the models are presented in Table 5.5. Calculations were based on female age structure (Fig. 4.9) and the recruitment by age class observed during 123 c - O en en cd -H Pi t^ •O-H B m P -H "H S 3 i-i u ^r u « -H .-I rt 2 > CO j_i oi en *-" 3 o ^ •73 u QJ ** In » id p, a> e oo o m u 01 4-> rH C flj QJ B B QJ 4-1 n_, •H 3 l-i U a> u XI ai 4-1 m gi id 14H 3 l-i U a> i-i U1 u d QJ 3 P. td en 4-4 I 01 14_, 00^ o -a & £ ■H 4-1 cd o in td •P l-i C o oi id oi .a B 3 O m IT) 01 rH 43 id t- o in o OJ rH H .. id in E cj Oi 3 U, m [j-, + CM in QJ rH in m C e 5 a, id It, U-, o o .-i ■a 0) 4-> UH -rH 0 3 u u o 01 01 B 3 in 2 id o o CQ 4-1 O in |J 0) QJ H- rH XI cm id B S 3 QJ 2 bu 0) V4 01 01 43 14-1 cd BOB 3 0J 2 tX4 CM m DC -H ID ai » r^ oo r^ id uicoooooinr-^^vj-cj-iu-iCTvincMiOrHrHcococMr-* ininininr-^r^r~-r^r».r~.iDu->ininioiDr^ooooiD uioooininininoinoooo OOO^r^rHCMr^rOOCTilOiH^HO CM -3 CO rH rH rH CM -J ~j ci mOOOOi-HrHOO HN-t^r-Ninio 3--3-inininr CM u-i u-i o in m m r-~ cm oo r~- cm co CO -$■ ~3 in -3- CO m m m Oi N N CM CM CM u-l O u-l OI CM CM o o o o CO -3 o o co m u-l u-i o o 00 o m r» m o u-l lO Id ID iH r- -3- -3 <3 t-\ i/iuiomo oinoominmo O O o O u-l o i-h r- CO ID 00 -3 ID oimH^omHin 00 O a\ id r^ co •j- r~ inmiON'j cOCOlOCMcOcO-3in id r-~ f»- r^ id r~ U-| rH u 00 Ol o rH CM co -3- u-l ID id ID ION r^ r» r^ r-« r*. r~- OJ 0\ (J\ Ch ai o\ Ui o% c^ e^ >' rH rH rH rH rH rH rH i-H rH a\ o r» oo rH CM CO 00 co -3- 00 00 Oi CM u-l ID oo oo CM Oi a cd QJ Q 1H id > u •H 14H -H U 0) D. in 0) OO cd co oo CTi ID a> rH 00 e •H 10 3 T3* QJ 4-1 •H 3 In U 01 iH ro 00 OJ 00 cd IH 0J > m CO 00 o\ ID en 0) 4-1 ■o 0) u 3 73 O u OJ DO cd x: u 4-1 cd 4-1 00 c •r| B 3 in in CO . -3 id O) u 3 "" 4-1 14H U O 3 u o 4-1 "H u 01 VH QO Vh 10 u O o V4 0) 43 01 rH id E 0) 14H u 0) 43 E E 3 C 3 II XI II 3 < cd P3 x) 01 4-> •H 3 u u OJ u cd 14H 14H o 4-1 cfl IH 1-1 O u 01 • J3 cd B •P 3 •rl C Q. cd x) U QJ > U U 01 QJ a. in en o g. cd 14H O 124 1976-1983. Alternative estimates of fawn numbers and ratios were generated by assuming that all females older than yearlings recruited fawns at the same average rate. Numbers Given the estimated female age structure each year from 1968 through 1986, there was little difference in the average number of fawns recruited whether recruitment varied by age for females (Column Al, Table 5.5) or whether recruitment rate was the same for all 2+ females (Column Bl, Table 5.5). Fawn production did not vary significantly more than numbers of females producing those fawns in either case. Number of females producing fawns was more important to total fawn production than age-specific differences in fawn recruitment among females. The greatest difference in fawn recruitment for any individual year was during 1969, when variation in recruitment by age class produced about 15% fewer fawns than would have been produced by egual recruitment by all age classes (Table 5.5). Potential differences in fawn recruitment between the 2 models are shown in Figure 5.4. When the population was expanding and had relatively more females in younger age classes, fawn recruitment would have been somewhat depressed by the age-specific variation in recruitment observed during 1976-83. The reverse was true in declining populations with relatively more older animals. In neither case, however, would general population trend be greatly affected. The observed number of fawns recruited (Column CI, Table 5.5) varied considerably more than any variation owing to age- specific differences in recruitment. Age-specific differences in fawn recruitment, operating as an intrinsic mechanism of population control, probably contributed a relatively minor amount of the observed variation in fawn recruitment. Ratios We observed little variation in average fawn: female ratios between the model assuming age-specific differences in recruitment (Column A2, Table 5.5) and the model assuming equal recruitment across all age classes (Column B2, Table 5.5). The same was true for fawn:2+ female ratios (Columns A3 and B3, Table 5.5). Observed fawn:female and fawn:2+ female ratios (Columns C2 and C3, Table 5.5) varied considerably more than ratios assuming constant recruitment. Given the female age structures estimated for each year 1968-1986, and assuming that annual fawn recruitment was stable for the population but varied among age classes as observed for 1976-1983, fawn: female ratios could have varied from 52 to 83:100 with a mean of 67:100 and a coefficient of 125 15 10 5 - 0 -- o c P C X 10 O W C -H -H •H+J 1^ w CO (0 1-1 T3 CD C cp^ i y w o 3 x ^ e . " CD o (0 to 2 CD u T3 T3 (0 n •H P c o •H to (0 Ifl CO CD ST *■** O Cn-H 0) „, H ^ P> CD p S-l -C TS T3 P d) CD C > .C O M P •H QJ -P m +J to <0 X3 -H P o t-l 4-1 o -a U0UBIA8Q o/o CX CD P H X! O UPC (0 CD CD U 4-1 O in CD U 3 CJi ■H 135 available information indicated that deer were in average or better condition during those years. We conclude that in most years, fawn survival is directly related to forage quality as determined by the length of time plants remain green and succulent, providing quality forage for lactation. Four of the 6 years excepted may be explained by cycles of fat storage and depletion. If fat reserves are high, females may be able to produce and nurse fawns successfully, despite poor quality current forage. On the other hand, a female coming out of winter in extremely poor condition, with a loss of muscle tissue as well as fat, may lose her fawn in utero and/or not have recovered enough to successfully rear a fawn, despite good forage conditions during summer. The assumption inherent in regressions of fawn survival against estimates of forage production is that improved forage conditions result in improved physical condition of deer and, in turn, higher fawn production and survival. Studies of captive deer document that improved nutrition results in improved deer condition and improved fawn production and survival (Verme 1962, 1965, 1969; and Robinette et al . 1973). French et al . (1956), Robinette et al . (1973), and Rasmussen (1985) further suggested that antler characteristics of yearling males were sensitive to nutritional conditions. Because we collected data on antler characteristics of yearling males relatively consistently, we were able to regress those data against annual forage production estimates. All antler characteristics and measurements of yearling male mule deer were positively related to forb production (Table 5.7), which should indicate that deer condition improved with increased forage production and quality. As with previous regressions of fawn survival on forage production, antler size (if it was related to forage quantity) should be better-related to forage per capita than just relative annual forage production. Antler measurements regressed against an index of forage per capita also indicated significant relationships, but rather than improving the relationship, the inclusion of a deer density factor weakened the relationship (Table 5.7). Because of that, we believe the relationship of antler size with forage production estimates was not with forage quantity, but with coincident forage quality. Trends in fawn survival and antler characteristics of yearling males coincided during most years (Fig. 5.7). This relationship was especially apparent during 1960-1964 and 1977-1986, underscoring the general relationship between fawn survival, deer condition, and forage conditions. Even the exceptions tended to verify the expectation that antler size and deer condition were related to forage conditions . Fawn 136 Table 5.7. Correlation coefficients for regressions of antler characteristics of yearling male mule deer against forb production and per capita forb production, 1976-1986, Missouri River Breaks, Montana. Dependent Variables Independent Variables Forb Production (kg/ha) Per Capita Forb Production (kg/ha/deer) Number of antler 0.793 <0.01 11 points per side. Percentage antlers 0.789 <0.01 11 2x2 points per side or more. 0.703 0.751 <0.05 <0.01 11 11 Beam diameter (mm) 0.799 <0.01 10 0.603 0.06 10 Product of beam diameter and percentage antlers 2x2 points per side or greater. 0.825 <0.01 10 0.792 <0.01 10 survival during 1965 was much below expectation based on current forage conditions . Both number of antler points per side and % 2X2 point antlers of yearling males, however, increased in 1965 over that of 1964 (Fig. 5.7), as expected, based on forage conditions. That information verified our projections that forage conditions were good during summer 1965, and that fawn survival was probably poor because of lag effects in female condition following the preceding severe winter. Forage conditions were apparently sufficient to result in average or better antler size of yearling males during 1973-1977, when fawn survival was lower than average (Fig. 5.7). Although forage conditions were excellent during 1975, the 2 measurements of antler size displayed in Figure 5.7 declined. However, another measure of antler size (beam diameter) was near the highest recorded (Table 4.7). Antler size that year, regardless of the measure used, was at least near average, and certainly greater than in 1961, 1984, and 1985, when deer were in extremely poor condition and antlers were extremely small. 137 Antler Points Per Side o oi i_ ID E a> LL o o c (0 u. o o CM t- O o o O) o CO o o o 8 in CO o> o CO m T- CO 0) >- o o> IO -8 o o CM jejeaiB 9 sjuiod 3x3 seieyy 6u;|jeeA % jaqiuaoea - seieiuaj o(H :sumbj .c c ■P 0) u a -t-) 0 x: 4-1 -a E " 0 tn U Q) oo E I 'H g>S Hi)'1 -Q (1) I | 0.1 3 0.0 Missouri River Breaks Study Area Yellow Water Triangle Study Area Projected from data for the Yellow Water Triangle Study Area 1972 1977 Year 1982 1986 Figure 5.8 Coyote density, as determined by the siren- answer survey, for the Missouri River Breaks and Yellow Water Traingle study areas. 139 Numbers of coyotes responding to the siren during summer declined from 1972-1973 to a low in 1977 (Fig. 5.8). Estimated coyote numbers increased from 1977 to 1984 and then declined through 1986. Notably, and contrary to general public perception, coyote populations on both areas declined following the 1972 ban on the use of Compound 1080 for predator control. Data on numbers of coyotes and the minimum number of litters on the study area during summer was provided by a combination of siren-answer and den area surveys (Pyrah 1984) . Estimates for 1984-1986 may be low relative to prior years because den area searches were less intensive. Thus, numbers may not have declined to the degree indicated in Figure 5.9. It is possible peak numbers were actually reached during 1984, as indicated by siren-answers, and declined thereafter. Total numbers of coyotes on the area during 1977-1985 (Fig. 5.9) varied by 1.6 times from low to high, a difference of 52 coyotes. That difference was mostly the result of variation in pup production and survival . The number of adults varied by 1.5 times from low to high, or 17 adults. Because pups comprise about one-half to two-thirds of summer coyote populations, it was not surprising that the minimum number of litters each year varied in a similar manner as total coyotes. 150 i Total Number of Coyotes Number of Litters Number of Adult Coyotes Figure 5 . 9 Number of coyotes and minimum number of coyote litters estimated for the Missouri River Breaks study area, 1977-1986, using a combination of siren-answer and den area surveys. 140 Populations of deer mice (Peromyscus maniculatus) increased irregularly from 1976 to 1983, declined abruptly in 1984, and slowly increased through 1986 (Figure 5.10). Microtine rodent (Lemmiscus curtatus and Microtus spp . ) populations experienced 3 irruptions during 1976-1987. Populations were at peak levels from summer 1978 through 1982 through autumn 1983. Another summer 1986. No trapping was field observations indicated that in 1986 continued through autumn autumn 1979 and from summer irruption started during conducted during 1987, but the irruption that started 1987 and was at least as great as previous highs 180 16(H 140 120- g 100- 80 -I 60 40- 20- 0 o Deer Mice -• Microtines 0 A / \ ft / \ y / \ / \ / . i / - S 0.2 Q Q> ** O >• 0.1 o 0.0 J Coyote Litters Coyote Density 24 20 C 3 16 ^ 12 O O «< o 8 0.05). The small variation in the number of adult coyotes among years also made it unlikely that fawn mortality should vary with numbers of coyotes. Although there was a tendency for fawn mortality to be lower when alternate prey populations were highest (Fig. 5.13), the relationship was not statistically significant. Fawn survival to autumn and winter was highest during microtine irruptions and lowest during the periods between microtine irruptions (Fig. 5.13). Use of average numbers of microtines during summer as an index may have obscured the relationship. Trapping in mid-summer and early autumn and general observations during 1978 indicated that microtine populations did not reach high levels until late summer, after almost all mortality of fawns that summer had occurred (Hamlin et al . 1984). Although more microtines than usual were present during 1982, highest numbers occurred from autumn 1982 through early summer 1983. Thus, most mortality of the 1983 cohort occurred after microtine populations had declined. Similarly, white-tailed jackrabbit populations peaked during summer 1983 and had "crashed" by November. Population levels of microtine rodents were probably influenced by vegetation production. All 3 irruptions were concurrent with much greater than average production of grasses and f orbs . A certain vegetation cover threshold may need to be met (Birney et al . 1976) before microtine population growth can take place. Alternate prey populations, especially microtine rodents, influenced the degree of predation by coyotes on mule deer fawns. Annual fawn survival was lowest for the 1976, 1977, 1983, 1984, and 1985 cohorts (Fig. 5.13). All prey populations were at lows during those years, except 1983 when survival to early winter was relatively good for the 1983 fawn cohort, but overwinter survival was poor. Severe declines in both microtine rodent and jackrabbit populations had begun by autumn 1983. Consequently, much of the mortality of the 1983 fawn cohort occurred after alternate prey populations had declined. Fawn survival increased substantially for the 1978 and 1979 cohorts, over that for 1976 and 1977 cohorts and coincided with both excellent forage conditions and a microtine rodent irruption. Populations of other alternate prey species were relatively low during 1978 and 1979. 144 CO > 3 (/) C CO Autumn Early Winter Spring 0 J Prey index Microtines 1976 1978 1980 1982 Year 1984 1986 Figure 5.13 Relationships among fawn survival to autumn, early winter, and spring; an index to alternate prey numbers; and microtine numbers. Although microtine populations had reached lows in 1980 and 1981, fawn survival did not decline to the degree of previous and subsequent declines (Fig. 5.13). Increasing numbers of deer mice and jackrabbits during 1980 and 1981 (Figs. 5.10 and 5.11) may have helped lessen decline in fawn survival during those years the degree of though fawn mortality 1981-82. was relatively high during the severe winter of 145 Survival of fawns was not greater than average during summer 1982, but little mortality occurred after microtine populations began to increase during late summer. Fawn survival was high during summer 1983 but declined substantially after microtine and jackrabbit populations "crashed" in late autumn. Drought conditions prevailed from late summer 1983 through late summer 1985. Both fawn survival and alternate prey populations were low during that period. Survival increased substantially for the 1986 and 1987 fawn cohorts, coinciding with improved forage conditions initiated by heavy rains and an autumn "green-up" of vegetation during 1985. This was followed by an irruption and subsequently high microtine population from summer 1986 through autumn 1987. Fawn Mortality in Relation to Winter Severity Mortality of fawns during winter varied greatly among years (Table 5.5). Multiple regressions did not indicate a linear relationship between winter severity one year and survival of the subsequent fawn cohort. Winter severity and fawn mortality during the same winter were positively related (r = 0.44, P < 0.05, N = 27), but the relationship had many exceptions . When 5 years that deviated most were omitted from the regression, the relationship between winter severity and fawn mortality strengthened (r = 0.655, P < 0.01, n = 22). Although winter fawn mortality might generally increase with winter severity (Fig. 5.14), prediction of fawn mortality based on winter severity alone could result in erroneous conclusions. Fawn mortality during the 2 most severe winters on record, 1977-78 and 1978-79, was about half that of 5 other winters, including 1975-76 and 1983-84, which were about 20% below the mean for winter severity. Winter severity was only slightly above the mean during 1984-85, but fawn mortality was the highest recorded. For most years when winter severity was near or below the mean (11.2), fawn mortality appeared to vary randomly between 4 and 30%. The winter severity index varied from below average (9.2) to very high (26.1) during winters when catastrophic mortality occurred. Prior forage conditions helped explain some anomalies in the in the relationship between fawn mortality during winter and winter severity. Fawn mortality was high during winter 1964-65 despite only slightly more than average numbers of deer. However, forage quantity and quality were below average (Fig. 3.9) during summer 1964 and, although forage conditions were above average during summer 1963, they also were below 146 100 90 <5 80 I 70- ^ 60- 1 50 O 2 40 c LL * 20 10-1 1984-85 1983-84 % 1975-76 1971-72 1964-65 1981-82 1968-69 • • 1978-79 0.0 10.0 20.0 Winter Severity Index 30.0 Figure 5.14 A plot of mule deer fawn mortality during winter against a winter severity index. average during both 1961 and 1962. Thus, high fawn mortality during winter 1964-65 occurred when forage conditions had been poor for 3 of the 4 prior summers and deer were probably in relatively poor condition entering winter. The winter of 1968-69 was very severe; the severity index was higher during only 3 other winters . Number of deer entering winter was at the long-term average and fawn mortality was moderately high, though not catastrophic. Forage conditions had been above average for 3 of the 4 prior summers (Fig. 3.9), and accumulated fat reserves probably enabled the deer population to get through winter without catastrophic mortality. 147 Winter severity during 1971-72 was similar to 1968-69; however, the number of deer in the population entering winter was above average. Catastrophic mortality of fawns and substantial mortality of adults occurred. Forage conditions were below average during each of the 3 previous summers (Fig. 3.9) and fawn production and survival had been high during the late 1960s suggesting both general body condition and fat reserves of deer were probably low at the onset of winter. The winter severity index was the highest on record during 1977-78 and 1978-79, yet fawn mortality was moderate (Fig. 5.14). Although forage conditions were below average during early summer 1977, a late summer "green-up" enabled deer to feed on high-quality forage for about 3 months prior to the onset of winter. Deer numbers were low and deer were in good condition entering winter. Forage conditions were excellent during summer 1978, deer numbers were low, and deer continued to build body condition and fat reserves and were in good condition at the onset of winter 1978-79. Although moderate fawn mortality occurred during winters 1977-78 and 1978-79, it was much less than was expected considering mortality that occurred during winters of equal or lesser severity (Fig. 5.14). Snow depth, which seldom exceeds 30 cm here, was greater than 75 cm more than 2 months during 1977-78 and exceeded 60 cm for more than 2 months during 1978-79. Persistent winds caused much drifting of snow and many shallow drainages were filled. Thus, snow depth precluded use by deer of most of the southern portion of the area. Even in areas where deer occurred, much of their normal food was inaccessible. Significant use of ponderosa pine needles by deer was detected only during those winters. Deer could only have survived by utilizing well-developed fat reserves. Winter 1983-84 was slightly below average in severity and winter 1984-85 was only slightly above average in severity, but fawn mortality was catastrophic during both winters; many adult females also died during 1983-84. Numbers of deer were above average at winter's onset during each of the 2 years. Also, forage conditions were below average for 3 of 4 summers preceding winter 1983-84, and for 4 of the 5 summers prior to winter 1984-85. Fawn production and survival was high starting in 1978, and continued relatively high through early 1983, which probably required mobilization of body reserves during 1982 and 1983. Because of dry conditions during late summer and autumn 1983, deer probably entered winter in poor condition, with very low fat reserves. The effect of winter severity on deer mortality appeared to be influenced by prior forage conditions. Numbers of deer 148 or density did not necessarily influence forage conditions, deer condition, or mortality. Fawn Production and Survival in Relation to Density Although one school of thought suggests that predators can control ungulate populations (Gasaway et al . 1983), the most widely accepted hypothesis is that a weather-forage complex regulates ungulate populations. This operates in a density-dependent manner through feedback mechanisms affecting fawn production and survival (Lack 1954, Caughley 1979, and Peek 1980). The latter holds that as density increases, increased demand for forage by ungulates ultimately has a detrimental effect on forage plants; declining forage quantity and quality subsequently lead to declining ungulate density. Implicit in this relationship are constant feedback and fluctuations in plant and animal components through which rates of increase in both are determined by ungulate density, assuming that everything else is equal. Because this system operates through a feedback loop, the density-dependent effect is often delayed. The existence of this density-dependent system is most apparent through changes in juvenile production and mortality because juvenile survival is most sensitive to reduced food supply (Caughley 1977). Theories involving density-dependence have usually assumed a more or less determinate carrying capacity (K) . More recently, Caughley (1977) recognized that density-dependence is most likely to operate in its purest form in areas where environmental fluctuation is minimal. Species or populations living in widely fluctuating environments such as the Missouri River Breaks would be least likely to exhibit density-dependent phenomena, but to date few species or populations have been placed in this category (Caughley 1977) . The data we collected over 28 years enabled us to examine the question of density-dependent regulation for this central Montana population and determine the extent to which variation in fawn production and survival was density-dependent. Initial production of fawns ( f awn-at-heel ratios) during 1976-1987 was not significantly (P > 0.05) density-dependent (Fig. 5.15), nor did the assumption of a 1-year lag effect result in statistical significance. Little change in fawn-at-heel ratio occurred until adult female numbers were more than 750, when a sharp decline took place. However, the decline which took place in 1984 and 1985 (Fig. 5. 15) may not have been related to density and the same decline could occur at any population level if an equally severe drought occured. Conception and pregnancy are probably less costly in terms of energy and resources than lactation, so initial production was 149 2.0 " a> 1.8 - c 3 ~3 1.6 - o co 1.4 - fc: Q) II + 1.2 - CM V) 1.0 - C 5 CO u. 0.8 - 0.6 - T- — r— — i 1 1 1 1 ' 1 i 100 200 300 400 500 600 700 800 900 Number of Females Figure 5.15 Fawn-at-heel ratio during June plotted against number of adult females in the population during June. expected to be less sensitive than survival to density. Thus, it is likely that initial production remains relatively stable except during very extreme conditions, such as the drought from late summer 1983 through 1985. Per capita fawn recruitment rate to December for breeding-age females was not statistically significantly related to the number of breeding-age females (r = -0.171, P > 0.05, Fig. 5.16). Although highest per capita recruitment rates occurred at low to moderate densities, low recruitment rates occurred at all densities. Recruitment deviated most from density-dependent expectations during 1973-1977 (furthest lower left points, Fig. 5.16). When data from those years were excluded, a more general decline in recruitment with density was apparent. However, fawn recruitment was substantially below expectations during some years of moderate density and above expectations during some years of high density. Overall, recruitment during almost half of all years deviated considerably from levels expected for the most extremely direct density-dependent linear relationship (Fig. 5.16) . 150 2.0 -I 1.8 - 1 ' r T ' 1 100 200 300 400 500 600 700 800 Number of 2+ Females - June (Nt) Figure 5.16 Fawns: mature female during December plotted against number of mature females in the population at birth pulse. Solid line is the actual regression line, the dashed line represents theoretical complete linear density- dependent fawn survival . Per capita recruitment of fawns to 1 year of age for breeding-age females was also not significantly (r = -0.052, P > 0.05) related to the number of adult deer present the previous spring (Fig. 5.17). A slightly more negative, but still non-significant (r = -0.246, P > 0.05), relationship occurred when per capita recruitment to 1 year of age was regressed against numbers of breeding-age females, rather than total adults. Although there may possibly be an underlying decline in recruitment with increasing density, it has little management application. From the standpoint of hunting management, the manager clearly can not depend upon high compensatory recruitment rates at low densities (Figs. 5.16, 5.17). Fawn mortality rates were also plotted against deer density for those who prefer data expressed in that manner ( Figs. 5.18 and 5.19). No significant linear relationship (P > 0.05) existed between fawn mortality rate to 1 year of age 151 1.4 -, 1.2 _ CO 1.0 E Q) U- 0.8 + CN 0.6 co c 0.4 2 (0 0.2 0.0 _L % + . • -• »- • • -//" 300 400 500 600 700 800 900 1000 1100 1200 Number of Adults - Spring (Nt) Figure 5.17 Number of fawns per adult female in spring plotted against the number of adults in the population at birth pulse when the cohort was born. Solid line is the actual regression line, the dashed line represents theoretical complete linear density-dependent fawn survival . and deer density at birth pulse (r = -0.06, n = 27, Fig. 5.18) or density 1 year prior to birth pulse (r = -0.33, n = 26, Fig. 5.19). The 5 data points with highest mortality and lowest density were for 1973-1977, when factors other than forage, density, or deer condition apparently influenced fawn mortality. When those years were excluded, a more general increase of mortality with density was apparent. However, even that relationship depended mostly on data points for the very highest and lowest densities; mortality was extremely variable through the middle range of densities. Other information suggested that data points at both low and high densities were influenced by coincident environmental conditions rather than by deer density. 152 . O ^ 0) CD ct CD c o CO *-* c c/> c 5 CD J_ CD >,_ 3 o c c 0.00 -1 .00 - -2.00 -3.00 -4.00 200 t— ■— Number of Adults - June 400 600 800 1000 1200 i Actual Theoretical - complete linear density dependence • N* Figure 5.18 Annual instantaneous rate of fawn mortality plotted against the number of adults in the population at birth pulse. Solid line is the actual regression line, the dashed line represents theoretical complete linear density- dependent mortality. "5 0.00 a> ■*— CO DC v> . -1.00 3 >• o£ a> co |° -2.00 2c — CD -u_ -3.00 CD 3 C C < -4.00 Figure 5.19. 200 t— •— Number of Adults - Previous June 400 600 800 1000 1 1 1 i 1200 Theoretical - complete linear density-dependence Annual instantaneous rate of fawn mortality plotted against the number of adults in the population 1 year prior to the birth of the measured cohort. Solid line is the actual regression line, the dashed line represents theoretical complete linear density-dependent mortality. 153 Numbers of recruits, rather than recruitment or mortality rate, must be plotted in order to determine the theoretical level of maximum sustained yield (MSY) . If complete, linear density-dependent recruitment rates applied, the maximum number of recruits would occur at the peak of a parabolic yield curve. The yield curve shown in Figure 5.20 was plotted with the following assumptions: 1) the highest observed recruitment rate occurred at the lowest observed density, 2) the lowest observed recruitment rate occurred at the highest observed density, and 3) the relationship was directly linear. 600 y- 500 CO *- la 4o° ll c "SO. 300 O T3 -Q 4-1 4-1 en a) m M C j^ o 4J -H •H 4-> at to 4-1 u o eO 0) cd E-i 3 a. o p. U CD Cfi Q CD r-l ■H en O 2: i-> ^H CD 3 C T3 3 < •-> rH O P ° U • (D 0, T) 0 n 0-, se ■n CD M m u tu cO ill 2: a O o H O T3 CD u a> id 0) x; Q • CD ti*a 0 O P-. £ •a CD M M u CD tfl CD 5: Q OO^OOCMOiHCOiHlO inoi^oomooeo-a-eNoo 0000000000 cooocicoincotnioin l/">r^00eOiH00,JD Ou">c«">c\loc"icMr^cMr~ OOOOOiHOiHtHO 0000000000 rH(N(v-maNCTioor^r~.o HHCM»OrHrHrovoin >j3in©r-»-3-m SNNNOOOOtOOOCOCO C7\C7\0%(7\CT*C7<(C7cO\CTtCy\ o •H 4-1 CO 1-1 I c •H I CD QO c a) x: u II Pd H o 157 Table 6.2. Annual and seasonal turnover of adult females as a proportion of 1 June adult female population, 1960-1986, Missouri River Breaks, Montana. Recruitment3 Mortality3 1 June-31 May 1 June-1 Dec. 1 Dec. -31 May Annual Hunting6 Winter 1960-61 0.346 0.187 0.168 0.019 1961-62 0.153 0.331*c 0.298* 0.033 1962-63 0.216 0.235* 0.196 0.039 1963-64 0.290 0.170 0.140 0.030 1964-65 0.063 0.214* 0.125* 0.089* 1965-66 0.126 0.158* 0.116 0.042 1966-67 0.196 0.207 0.163 0.044 1967-68 0.352 0.176 0.121 0.055 1968-69 0.262 0.206 0.131 0.075 1969-70 0.354 0.150 0.106 0.044 1970-71 0.250 0.154 0.088 0.066 1971-72 0.047 0.430* 0.182* 0.248* 1972-73 0.120 0.272* 0.174* 0.098 1973-74 0.154 0.244* 0.180* 0.064 1974-75 0.099 0.225* 0.169* 0.056 1975-76 0.048 0.113* 0.000 0.113* 1976-77 0.172 0.121 0.035 0.086 1977-78 0.148 0.049 0.016 0.033 1978-79 0.373 0.105 0.030 0.075 1979-80 0.353d 0.059 0.012 0.047 1980-81 0.273d 0.036 0.018 0.01 1981-82 0.221 0.191 0.118 0.073 1982-83 0.300 0.171 0.157 0.014 1983-84 0.076 0.114* 0.019 0.095* 1984-85 0.026 0.138* 0.072* 0.066* 1985-86 0.104 0.022 0.015 0.007 Mean 0.197 0.172 0.110 0.062 Range 0.026-0.373 0.022-0.043 0.000-0.298 0.014-0.248 3 Recruitment and mortality expressed as a proportion of starting, 1 June population. Recruitment equals number of yearling females 31 May /number adult females previous 1 June. All mortality expressed as number of adult females lost during the period/number of adult females 1 June, at start of year. b The period 1 June-1 Dec, called Hunting Mortality, also includes minor numbers (based on marked deer) of deer dying from causes other than hunting. c * - Indicates that mortality during the period exceeded annual recruitment . d Recruitment was actually 0.482 for 1979-80 and 0.336 for 1980- 81, but because of emigration of yearling females, recruitment to the population was at the figures listed in the table. 158 Table 6.3. Annual and seasonal turnover of adult males as a proportion of 1 June adult male population, 1960-1986, Missouri River Breaks, Montana. Recruitment3 Mortality3 1 June-31 May 1 June-1 Dec. 1 Dec. -31 May Annual Hunting*5 Winter 1960-61 0.740 0.520 0.500 0.020 1961-62 0.311 0.344*c 0.311 0.033 1962-63 0.373 0.322 0.288 0.034 1963-64 0.468 0.339 0.323 0.016 1964-65 0.100 0.414* 0.343* 0.071 1965-66 0.271 0.396* 0.354* 0.042 1966-67 0.429 0.429 0.405 0.024 1967-68 0.762 0.381 0.357 0.024 1968-69 0.500 0.448 0.397 0.051 1969-70 0.656 0.377 0.344 0.033 1970-71 0.436 0.423 0.372 0.051 1971-72 0.089 0.570* 0.456* 0.114* 1972-73 0.293 0.268 0.244 0.024 1973-74 0.310 0.310 0.286 0.024 1974-75 0.191 0.429* 0.357* 0.072 1975-76 0.125 0.500* 0.469* 0.031 1976-77 0.500 0.200 0.150 0.050 1977-78 0.385 0.500* 0.500* 0.000 1978-79 1.087 0.174 0.174 0.043 1979-80 0.954 0.488 0.488 0.024 1980-81 0.661 0.532 0.532 0.016 1981-82 0.435 0.580* 0.580* 0.029 1982-83 0.719 0.561 0.561 0.018 1983-84 0.185 0.246* 0.246* 0.077 1984-85 0.071 0.321* 0.321* 0.054 1985-86 0.359 0.385* 0.385* 0.025 Mean 0.439 0.413 0.375 0.038 Range 0.071-1.087 0.174-0.580 0.150-0.580 0.000-0.114 3 Recruitment and mortality expressed as a proportion of starting, 1 June population. Recruitment equals number of yearling males 31 May /number adult males previous 1 June. All mortality expressed as number of adult males lost during the period/number of adult males 1 June, at start of year. b The period 1 June-1 Dec, called Hunting Mortality, also includes minor numbers (based on marked deer) of deer dying from causes other than hunting. c * - Indicates that mortality during the period exceeded annual recruitment. 159 -autumn and 6.2% (50%), selective harvests of older and larger males, many of those most likely to be in poorest condition after the rut were not present during winter. The males least likely to be shot, yearlings and mature animals that were not dominant breeders, were those most likely to enter winter in better condition, with the best chance of survival. Causes of Mortality Females Two major causes of adult female mortality were identified: hunting and coyote predation. Of 77 marked adult females known to have died during 1976-1986, 26 (34%) were shot, and 6 others (8%) were presumed shot but not reported, based on time of disappearance and other evidence. Thus, a total of 42% of adult female mortality was categorized as hunting-related. Five (19%) of the 26 certain hunting-related deaths, were either crippling losses or illegal kills. Only 1 was known to have been illegally shot, though it is possible that some classified as crippling loss were intentionally left in the field. Nine deaths (12%) resulted from coyote predation during winter and early spring. Additionally, 27 deaths (35%) for which cause of death was not ascertained, occurred during winter-spring. Predation was ruled out in 2 of those, but was suspected to be the proximal cause of death for the remainder. All mortality of unmarked deer during winter-spring, for which cause was determined, was the result of coyote predation. Additionally, almost all mortality of marked white-tailed deer on the river bottoms adjacent to our study area and mule deer on the uplands of the NCRCA study area during winter 1975-76 was attributed to coyote predation (Knowles 1976). 161 Nine marked females (12%) died during summer and autumn, 2 as a result of coyote predation and 7 of unknown causes. Males Hunting was known or suspected as the cause of 91% of all deaths of marked adult males during 1976-1986. Of 46 known to have died, 32 (70%) were shot or crippling losses and 10 (22%) were suspected hunting mortalities based on time of disappearance. Additionally, almost all loss of males determined from population estimates (Table 6.3) occurred during the hunting season. Of 32 marked males verified shot, 2 (6%) were crippling losses. Only 4 (9%) males died during winter-spring. Cause of death was not assigned because the remains were not found. Though coyote predation was suspected as the proximal cause, wounding during the hunting season may have often been the ultimate cause. That was the case for 1 marked male assigned to the hunting loss category. He was wounded during hunting season and killed by coyotes during February. Age-specific Mortality Females Age-specific mortality rates were determined for 118 1-year-old and older females over 404 deer-years (Fig. 6.1). Only females aged to a specific year class at capture and those marked at the beginning of a biological year were included in the sample. Age-specific mortality was categorized as either hunting or natural. The few females that were suspected but not verified as being shot were included as hunting mortalities. Natural mortality included all other causes; most occurred during winter and spring. Total annual mortality was highest for females aged 12 and older, followed by females between the ages of 6 and 7, 9 and 10, and 1 and 2 years of age (Fig. 6.1). Relatively high mortality rates for the youngest and oldest adults were not unexpected, but high mortality rates for females in their 7th year of life was surprising. Most of the latter occurred during winter and was not hunting related (Fig. 6.1). As we noted elsewhere (Hamlin and Mackie 1987), the increased mortality for 6 year-old females was an aberration from the typical U-shaped mortality curve of mammals (Caughley 1966). The coincidental peaks in survival of fawns to 6 months and female mortality at 6 years suggested a relationship between reproduction and mortality. We hypothesized (Hamlin and Mackie 1987) that the peak in non-hunting mortality rate at 6 years was the average outcome of cumulative reproductive stress resulting from a strategy of maximum reproductive effort each year in a variable, unpredictable environment. 162 (0 r o 50- 45- 40- 35- 30- 25- 20- 15- 10- 5- o-^ • Total Mortality o Hunting Mortality o-- Natural Mortality 40 47 3 4 5 6 7 8 9 10-11 12 + Female Age at Start of Year 53 55 55 43 27 22 20 27 15 Sample Size Figure 6 . 1 Age-specific mortality rates for female mule deer in the Missouri River Breaks, Montana, 1976-1986. Except in the very best years, lactating and rearing fawns to 6 months of age probably draws on body reserves in addition to nutrition provided by current forage. Forage quality is often insufficient to fully replace body reserves following weaning, and on average, body reserves have reached a critical level following weaning in the fifth reproductive year. The more fawns a female has recruited during her first 5 reproductive years (cumulative reproductive stress), the poorer her condition at age 6 and the greater her vulnerability to death. Many females that do not die at 6 years either do not have fawns the next year or lose them shortly after parturition. This gives them the opportunity to recover body reserves, such that survival and fawn recruitment among 8+ year olds improves over that at 6 years. Reproductive histories were available for 7 of 9 females that died between age 6 and 7 from non-hunting causes. All had recruited fawns to weaning age at above average rates during the years leading up to their death. Four of the 7 had weaned fawns during the year of their death and for at least 2 years prior to that. The reproductive status for 2 was unknown during the year of their death, but they had both weaned fawns for at least the 2 years prior to their death. 163 One had weaned fawns during the year of her death and for at least 1 of 2 prior years. Although mortality of adult females was unusually high for 6 year old females in our sample, reproductive history rather than age, per se, may be the operative factor in reproductive "burnout". Marked females of all ages that died from non-hunting causes during the study had been much more (X2=26.1, 1 d.f., P>0.001) productive prior to their death than marked females that did not die, or died from hunter harvest. Twenty-eight 2+ year old females that died of non- hunting causes recruited fawns to weaning age during 76 (89%) of 85 observed opportunities prior to their death. Marked females that did not die, or were harvested, recruited fawns to weaning age during only 57 (55%) of 103 observed opportunities . On an individual basis, the relationship of high productivity and relatively early mortality is not perfect. At least some females were very productive for 5-6 years and have not yet died. On the average however, high productivity appeared to place the female at greater risk of death. Circumstances during the period the female is highly productive probably influence the outcome. Females productive during favorable summer and winter conditions were probably less vulnerable than those that recruited their last fawn(s) substantially on body reserves during a dry summer followed by a severe winter. Clutton-Brock et al. (1982) also noted that for red deer females, mortality rates for productive females of all ages were higher than for non-productive females, especially after 10 years of age. Hunting mortality rates were highest for females between 1 and 2 years of age and those 9+ years of age. No age-specific differences were apparent for females aged 3-8 years. Females with fawns generally occupied the most secure habitat. Yearlings, until they rejoined their mother's social group in late autumn or winter, were most likely to occupy areas on the fringes of their natal range or entirely new and unfamiliar home ranges. Thus, they were more likely to occupy relatively more open habitats where they were vulnerable to hunting. They also may be less wary than older deer. The relatively high mortality of 9+-year-old females as a result of hunting was not expected. Limited evidence, based on a few marked animals, suggested that older females without fawns may have lost secure fawn-rearing habitat to younger, productive females. This placed them, like yearlings, in more open habitats where they were vulnerable to hunters . A combination of additional factors also may help explain the relatively high hunting mortality for older females. After 7 years of age, females may undergo an alternating cycle of fawn 164 production and subsequent poor condition with no fawn recruitment the following year. During years when they do not wean fawns, females would be more likely to occur in larger groups (Chapter 7) and be more vulnerable to hunters. Also, older females in general are likely to be in poorer condition, which may cause them to be less alert and to lag in groups fleeing from hunters. In a non-hunted population, relatively high natural mortality rates would be expected for females 12-years-old and older. The relatively high hunting mortality for those females in this population removed many before they were exposed to natural mortality during winter. Thus, some hunting mortality of older females may not be additive to other mortality, thereby reducing observed natural mortality rates for older females to lower than expected levels. Based on population estimates, emigration of yearling females was apparently not balanced by immigration. Although this loss was relatively minor in total and most occurred for the 1979, 1980, and 1986 cohorts, it increased loss of yearling females to the population. When all losses to the population, including emigration, are considered, mortality of yearling females was 25% rather than the 20% (Fig. 6.1) observed for individuals . Males Because only small samples of 3+ year old males were available to determine age-specific mortality rates and the samples did not show differences, males 3-years-old and older were combined. Mortality rate for 33 males between 1 and 2 years of age was 36%. Between 2 and 3 years of age, 50% of 28 males died, and annual mortality was 56% for 36 males 3-years- old and older. As discussed previously, most deaths of adult males were hunting related. Only 9% of deaths of marked males were suspected to be the result of natural causes. The higher harvest rate of males aged 2 years and older than for yearlings apparently resulted from hunter selection for larger-antlered deer. Except for extreme drought years, most 2 year old males had small, 4X4-point antlers that were obviously larger than yearling antlers. Apparently, that antler size was enough to satisfy many hunters, because mortality rate of older and larger males did not increase substantially after 2 years of age. The fact that the mortality rates of 3 year old and older males did not increase beyond that of 2 year old males may also be related to older males being more wary than younger ones, at least prior to the rut . 165 Antler size of younger males also influenced relative harvest rates, depending on the structure of the general hunting season. During hunting seasons when only antlered males were legal game, a significantly greater proportion of yearling males with 2 antler points or more per side were harvested than occurred in the post-season population (X2=9.39, df=2, P=0.01; Table 6.4). Although contributions of each antler class to the total chi-square value were similar, selection against "spike"-antlered yearling males provided the greatest contribution. During hunting seasons when any deer could be harvested on the general license (A-tag) , the distribution of antler points per side for yearling males in the harvested sample was not statistically different from that of the post-season population (X2=1.99, df=2, P= 0.39). Table 6.4. Antler points per side of yearling males for harvested samples and post-season populations during antlered male only and any deer hunting seasons. Number of Antler Points Ma] .e Only Si easonsa Any Deer Seasonsb 1X1 1X2 2X2+ 1X1 1X2 2X2+ Harvest Sample 6 4 66 52 22 166 (0.08) c (0.05) (0.87) (0.22) (0.09) (0.69) Post- •Season Classification 41 29 155 136 66 363 (0.18) (0.13) (0.69) (0 • 24) (0.12) (0.64) a 1976-80 and 1985 and 1986 b 1960-64, 1971, and 1981-84 c Number (proportion) Although there may be some hunter selection for larger- antlered yearling males during all years, the lesser degree of selection apparent during any-deer seasons may, at least partially, result from small antlered "spikes" being shot as antlerless deer (Wood 1987). Conversely, during antlered male-only seasons, many "spikes", especially those some distance away, may not be perceived as males, thereby lowering their harvest rate. Dispersal may have affected mortality rates of yearling males. Only 6 (29%) of 21 yearling males that remained within the population were shot, while 6 (50%) of 12 yearling males that left the area were shot. Some of the yearling males that left the population unit and were shot had dispersed into obviously less secure habitat. Others, however, dispersed to 166 habitat that was not structurally different or obviously less secure than that from which they had dispersed. For those males that lived to 2 years of age, there was no apparent difference in known age at death between those remaining in the population (X = 3.6, n=12) and those that left the area (X = 3.5, n=5) . Based on population estimates, emigration and immigration for yearling males apparently balanced during most years, so dispersal did not contribute additional mortality to the population. Effect of Home Range Location on Mortality and Productivity of Females Because of indications that deer inhabiting parts of the study area with low topographic relief experienced greater mortality during winter than deer on other parts of the area, we examined the relationships between home range location and mortality and productivity of marked females. Home range and movements data (see Chapter 8) indicated 2 basic movement patterns for females: resident and migratory. Migratory deer most often occupied areas of low topographic relief during summer and moved at least 3 km to areas containing steeper terrain during autumn or winter. Some resident deer inhabited areas with low topographic relief yearlong, but most had home ranges that included at least some steep terrain comprised of north- and south-facing slopes. For analysis, low relief terrain was defined as those areas occupied during summer by marked migratory deer and all areas of equal or lesser relief. Mortality rates of marked females were examined during 3 winters of relatively high mortality (1981-82, 1983-84, and 1984-85). Data from prior winters were not included because few deer were resident to areas of low relief during population lows, and few of those were marked. Mortality rates were compared among 3 movement and home range categories: resident yearlong in areas with low topographic relief; occupying areas of low relief during summer but migrating to areas of steep terrain during autumn or winter; and resident yearlong in areas that contained some steep north- and south-facing slopes. Of 69 marked adult females that entered winter 1981-82, 9 (13%) died or disappeared and never were reobserved. In 1983-84, 12 (18%) of 67 marked adult females died, and in winter 1984-85, 6 (13%) of 47 marked females died. For the 3 winters combined, 27 deaths occurred (15%) over 183 marked adult female-winters. Mortality rates were significantly different among the 3 categories of home range use (X2=16.1, P<0.01, 2 df). 167 Eighteen (29%) of 62 females resident in areas of low topographic relief died. Only 1 (3%) of 37 migratory females and 8 (9%) of 85 females resident in areas that contained steep terrain with north- and south-facing slopes died. Females resident yearlong in areas of low topographic relief not only died at the highest rate overall, but also in each winter. When they were compared to all other females that used steeper terrain, whether resident or migratory, the difference in mortality rates remained significant (X-13.8, P<0.001, 1 df ) . There are several probable explanations for the higher winter mortality rates of females resident in areas of low topographic relief than for other females. During late summer and autumn, succulent forage was scarcer in areas of low relief than in steeper terrain both because of more intensive cattle grazing and because the shallow drainages were less protected from the desiccating effects of sun and wind. These areas also had less topographic and microsite diversity and therefore less diversity of plant species and plant phenology than areas of greater topographic relief. Thus, deer resident to areas of low topographic relief had shorter periods of access to high quality forage than other deer and females probably experienced more lactation stress, especially during dry summers. During winter, these deer had to spend more energy moving through deeper snow because of the lack of bare south-facing slopes and large, thickly-timbered, snow- intercepting north-facing slopes. Also, they did not have access to large, warmth-radiating south-facing slopes and the diversity or quantity of winter forage available in steeper, more diverse areas. Overall, deer resident to areas of low relief had a shorter period of positive energy balance and a longer, more severe period of negative energy balance than deer occupying areas with more diverse topography. They were more susceptible to all sources of natural mortality including coyote predation and malnutrition. We expected that in areas of low relief, fawn production and survival should be relatively low, and total deer numbers lower than on areas with greater topographic relief. This generally appeared to be the case, but because of the variability in movements of migratory deer between areas of low and high relief, interpretation of the data was difficult. At least some of the higher surviving migratory females had not moved to areas of steeper terrain at the time of some early winter surveys and some had not moved back to areas of low relief at the time of some spring surveys, thus fawn survival and total deer numbers in areas of low relief was difficult to determine. Also, if females in low relief areas died at a greater rate than other females during winters of high mortality, fawn survival for original numbers of females 168 in areas of low relief would be lower than apparent in sex and age classification data. Despite these considerations, which minimize differences in classifications, fawn survival was generally poorest in areas of low relief during both early winter and spring (Fig. 6.2). However, during years of higher fawn survival, fawn survival was as high or higher in areas of low relief as in areas of greater topographic relief. Numbers of adult females observed in areas of low relief during early winter helicopter surveys declined earlier and recovered later than in areas of greater relief. Numbers in areas of low relief declined from 1970 through 1978 and started to recover by 1979 (Fig. 6.3). The number of females in areas of greater relief did not decline until 1972, reached a low during 1976, and began to recover in 1977, 2 years earlier than in areas of low relief. The number in areas of low relief reached a peak in 1981 and declined following mortality during the winters of 1981-82, 1983-84, and 1984-85 (Figure 6.3). On the other hand, females in areas of greater topographic relief continued to increase through 1984 and then declined. These data substantiate those provided by marked deer in that deer numbers in areas of low relief were most affected during winters of relatively high mortality. These data are not precise comparisons because some year-to-year fluctuations in relative numbers between areas could have resulted from changing proportions of migratory females in the 2 areas at the time of the surveys and changes in relative observability between years. Relative numbers of total deer in the 2 types of habitat during spring were even more difficult to interpret than numbers during early winter. This occurred because there was even more variation between years in the location of migratory females at the time of spring surveys. During spring 1978, 1979 and 1982 (after severe winters) most migratory females remained in areas of high topographic relief through the time of spring surveys. During those years, total numbers of deer using areas of low relief during spring were underestimated, and numbers using areas of high relief were overestimated. The data (Fig. 6.4) do indicate, however, that there were very few deer resident yearlong to areas of low relief during 1978 and 1979. After 1982, spring surveys were flown later than previously and most migratory females had returned to areas of low relief at the time of the surveys. Although not ideally controlled, the data seemed sufficient to indicate that numbers of deer in areas of low relief decreased more than those in areas of greater relief. This differential decline occurred both during the population low of the mid-1970s and the population decline from winter 1983-84 through spring 1986. 169 130 -] 120- 110- 100- 90- 80- 70- 60- 50- 40- 30- 20- 10- 0- High Relief Areas Low Relief Areas — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — 1970 1972 1974 1976 1978 1980 1982 1984 1986 Year 100 90 80 " 70 - 60 " 50 - 40 - 30 - 20 - 10 0 B. -• — High Relief Areas -•— Low Relief Areas T T 1 1 1 1 1 1 1 1 1 1977 1979 1981 1983 1985 1987 Year Figure 6.2 Fawns: 100 females during early winter (A) and fawns: 100 adults during spring (B) compared with between high relief terrain and low relief terrain. 170 500 i to * 400 E o "■ o 300 - D C < > H. >» 200 0 — 0) HI 1 100 H 3 High Relief Areas Low Relief Areas — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i 1970 1972 1974 1976 1978 1980 1982 1984 1986 Year Figure 6.3. Number of adult females observed during early winter helicopter surveys compared between high relief terrain and low relief terrain. en c 600 -i Q. 500 H (f) ■ d> Q 3 300 i »— ■g 200- 3 100- v (0 o 1977 High Relief Areas Low Relief Areas *^~- 1 r 1979 — i 1 1 — 1981 1983 Year T 1 1985 1987 Figure 6.4 Number of mule deer observed during spring aerial surveys compared between high relief and low relief terrain. 171 Our data indicated that natural mortality and fawn production of adult females not only varied by age among individuals, but also varied across the area depending upon the specific strategy of habitat use by deer or location and characteristics of habitat within each home range. Areas with low topographic relief did not provide good survival habitat for mule deer across the observed range of environmental variability. Deer could survive and reproduce well on these areas during favorable periods but were more vulnerable to mortality during dry years or cold winters with deep snow. We also suspected that hunting mortality might be higher for females occupying more open, low relief terrain, but apparently that was not the case. Mortality rate of females owing to hunting was not significantly different among the 3 home range categories (X2=1.27, P=0.54, df = 2). The fact that we did not find that females in the more open, low relief areas more vulnerable to hunting may be related to the coincident lower accessibility of those areas. Most of the broad areas of low relief occurred in the southwestern and southcentral portions of the area, relatively distant from major roads. Access roads crossed large areas of habitat unsuitable for deer before reaching hunting areas . Hunters often bypassed open, low relief areas, preferring to travel on major trails until more typical deer habitat was reached. Thus, a substantial portion of low relief terrain received only minor hunting pressure. Our data did not allow us to determine if more deer were shot in low relief portions of generally high relief areas. Comparative Effects of Hunting on Females and Males Because of the normally lower proportion of males than females in the adult population, total numbers of males are lower than those of females and hunting mortality can have a different effect on male and female populations. Hunting of males occurred annually, so natural mortality rate for males in its absence remains unknown. If, in the absence of hunting, the natural mortality rate of males is higher than for females as observed by others (Martinka 1978, Gavin et al . 1984, Kie and White 1985), it may not be possible to maintain adult sex ratios much higher than 40-60 males: 100 females. Based on our data, it should be possible to maintain post- season ratios of 20-35 adult males: 100 adult females during most years without restricting the number of people hunting males. If hunting pressure increases, however, this expectation may be unrealistic. Following years of unusually poor fawn recruitment, unrestricted hunting of males resulted in post-season ratios of less than 20 males: 100 females. Past regulations have allowed hunters to select heavily for adult males, which were harvested at a rate 3-4 times 172 greater than adult females. Because male populations are lower than female populations and recruitment adds proportionally greater numbers to the male population, the male population can sustain a higher mortality rate (but not numbers) than females and remain stable. The degree to which males can be harvested more heavily than females depends upon the fawn recruitment rate and post-season male: female ratio the manager would like to maintain. As we have seen, neither fawn recruitment rates nor mortality rates of adults were stable, and averages were seldom useful for management decisions. A harvest strategy of quotas based on average recruitment and average natural mortality or constant hunting pressure would contribute to increased fluctuations of both males and females. Underharvests occur during periods of high recruitment and over-harvests during low recruitment (Beddington and May 1977). A "tracking strategy" (Caughley 1977) that accounts for annual variation could more successfully manage populations. The general season structure sets bag limits and sex restrictions on the A-tag (first tag) which an unlimited number of hunters can purchase over the counter. That season and its restrictions are set each year before winter survival is determined or a final estimate is made of the previous autumn's harvest. Thus, data used to establish the general season are 1 year behind and the efficiency of a "tracking strategy" is reduced. The number of B-tags (second tags specific for antlerless deer) can more closely "track" population trend because final quotas for the drawing of those tags are established after winter survival and the previous year's harvest have been determined. A model that demonstrates maintenance of different post- season sex ratios of adults under varying fawn recruitment and adult mortality rates is illustrated in Figure 6.5. Conclusions from the curves reflect no significant compensatory reproduction or mortality in the population. If the post-season adult sex ratio was 20 males: 100 females and if only 10 fawns were recruited per 100 females, any mortality over 20% for adult males during the following year reduces the male population and the male: female ratio to below 20:100 (Fig. 6.5). Because hunting mortality of adult males was almost always more than 20% under past regulations, a population decline for males over the next year can also be expected. If 50 fawns: 100 females were recruited, the male population could sustain 55% annual mortality (near the highest observed) and maintain 20 males: 100 females post- season. Annual mortality below 55% would result in the post- season sex ratio rising above an existing 20:100 and an increase in the number of males. 173 % Mortality c I 10 i 20 30 i i 40 50 60 70 80 90 i ii i ii lUU C 90" /' / °9 1 °0 In? 1 <0 A A* /£ /$ Q. CO 80- A A A A 1 o 70" /-•'' $: /i A /i / j? M /i A / i (0 60- 9 ■' / & /§ / $ / § E ^ ■' / £ CD LL O 50- 40- $ ■■' /i / i / f / § o / / ■^ s S 30- / s (/> .' c 20" • 5 j (0 LL 10- o- ■' ' S \*^ I I I 1 1 1 1 1 1 ( ) 10 20 30 40 50 60 70 80 90 % Mortality Figure 6.5 A simplified model illustrating adult female mortality rates that will maintain a stable female population and mortality rates for males that will maintain selected male: 100 female ratios given various recruitment rates. The model also indicates that within the normal range of adult sex ratios, stable populations of females cannot be maintained with a mortality rate similar to that of males. At 40 fawns recruited per 100 females, annual mortality rates for adult males of 25%, 33%, 50%, and 67% would maintain post- season sex ratios of 60 males:100 females, 40:100, 20:100, and 10:100, respectively. At the same recruitment rate, the female population would decline if annual mortality exceeded 17%. Fawn recruitment of at least 12-18 fawns: 100 females is necessary to replace the average annual natural mortality rate of 6-8% for adult females. A harvest rate of 33% for adult females would have reduced the female population during all years 1960-1986, except 1979-80, when numbers would have roughly stabilized rather than continued to increase. Also, because female populations are larger than male populations, an eguivalent harvest rate for the 2 sexes will result in a higher numerical harvest of females. 174 The importance of the information in Figure 6 . 5 is not in its precision, but that it can provide early insight to the likely effect of hunting regulations in a given year. If the goal is for the female population to increase, stay the same, or decrease and the fawn recruitment rate is known, the general level of antlerless tags, if any, necessary to achieve the objective can be estimated. It is also possible to determine when the male population is likely to decline, given general hunting of males. To keep the male population from declining or raise the male: female ratio, harvest of males may have to be restricted during some years of low recruitment. Overall, allowable harvest rates for females, in the absence of compensatory reproduction and mortality, are generally lower than previously believed. Also, in most areas, we are either harvesting males at higher rates or total male mortality from all causes is higher than previously believed. The mule deer herd at Sage Creek in central Montana, for example, is characterized by recruitment generally greater than 50 fawns: 100 females and post-season adult male: female ratios of 10:100 or less. This indicates that total annual loss of males to the population is usually 70-80%. Although it is probable that much of that mortality results from hunting, it also is possible that net emigration of yearling males from that open habitat contributes to high annual loss of males. 175 Chapter 7 SPATIAL AND SOCIAL DISTRIBUTION Spatial Distribution General Patterns Generally, mule deer occurred throughout the study area. Thus, our analysis primarily relates to patterns of dispersion or the spatial distribution of individuals within the population (Pennak 1964). Spatial and temporal patterns and factors influencing dispersion were determined by plotting the locations of deer observed during 32 aerial surveys from autumn 1976 through spring 1984 within 953 28.8-ha blocks generated for the area. Prior to 1976, deer observed during aerial surveys were assigned to 1 of 10 multi-sized subdivisions of the area. Those locations and locations of deer observed along vehicle routes during 1960-1964 within 3.9 km2 blocks (Mackie 1970) provided insight to general, long- term changes in distribution. The general pattern of dispersion or intensity of use of the area, based on relative numbers of mule deer observed in 28.8-ha blocks during all aerial surveys from 1976 through spring 1984 (Fig. 7.1) indicated that spatial distribution was neither uniform nor random. Rather, it followed a clumped or aggregated pattern. Both general and local aggregation, characterized by areas of relatively high deer density bordered by areas of declining density, was apparent. Across the area as a whole, moderate to dense concentrations generally extended across broken terrain along slopes and coulees between major ridgetops and major coulee or river bottoms. Highest contiguous densities occurred in the northwestern portion of the area between Sand Creek and the Missouri River. Blocks in which deer were never or only rarely recorded were primarily in open prairie habitat along the southern and southwestern fringes, larger expanses of sagebrush-grassland along major ridges, large open coulee bottoms within a few kilometers of the Missouri River, and bottomlands along the river. These areas did not comprise yearlong habitat for mule deer and received only occasional seasonal or transitory use. Within this general pattern, aggregated areas of high density were surrounded roughly concentrically by areas of lower, declining use. A quantitative representation of concentric distribution around high density blocks (Table 7.1) indicated that adjacent blocks were not randomly distributed among the 4 density categories (X2 = 164.6, 3 df, P < 0.0001). Other blocks of high density were significantly over 176 u $ ■o ^^ o CO ■D T3 + T" O T3 >, ■ (0 CO *"■ [> **^ (0 **-"' rn >. c >. ♦- Q *- n c o (A C O Q E 3 0) O x: T3 =: Q X s o _l o z HD E ^- t6t sh ■— : £1 ■a 0) c ■H e C a) +j a) -a m 03 to a) o ■H >1 +J ■H W C 1 H c •H G •H 1-1 3 c ■a o •H ta +J >i 3 > M O ■P OJ rn •H ^H T3 (0 ■H >^ M ■P OJ •H ra (fl C CM 0) m T) b Q) 0 > M •H »W +J (0 03 •-\ +J CD 03 0i 73 CD 1-1 ■H 177 Table 7.1. Relative density distribution of mule deer in the Missouri River Breaks, Montana, 1976-1984. Proportions of adiacent blocks that were : High Medium Low No Block Type Density Density Density Deer Use High Density 0.282 0.519 0.158 0.041 Medium Density 0.211 0.472 0.241 0.076 Low Density 0.125 0.472 0.238 0.165 No Deer Use 0.036 0.179 0.206 0.579 Overall Distribution 0.177 0.423 0.231 0.169 of blocks represented, while blocks in which deer were never observed during the aerial surveys were significantly under represented. Blocks of medium density occurred in greater than random proportion adjacent to high density blocks, while low density blocks occurred in lower than random proportion. Blocks surrounding medium, low, and zero density blocks also were not randomly distributed (X2 137.7, 3 df, P < 0.0001; r2 -, 2_ = 19.6, 3 df, P < 0.0001; X^ 553.7, 3 df, P < 0.0001; respectively) . In each case, blocks of the same density category were positively associated, while those 2 or 3 density categories higher or lower were negatively associated. These data together with Figure 7 . 1 indicate a general dispersion pattern characterized by a high degree of aggregation in and around "core" areas or habitats, and gradual rather than abrupt transitions in density distribution. Spatial Distribution by Sex and Age Spatial distribution by sex and age classes was quantified only during autumn when all deer were about equally visible and dispersion was not altered by years of varying winter severity. We tested by Chi-square and correlation analysis the null hypothesis that there were no differences in spatial distribution between sex and age classes. This hypothesis was rejected in all tests except for the comparison between females and fawns (Table 7.2), which, as expected, were not distributed differently. Spatial distribution of females and fawns, however, was different from that of both yearling and mature males, and yearling and mature males were also distributed differently. Correlation analysis indicated the degree and direction of association between sex and age classes (Table 7.3). 178 Table 7.2. Chi-square probabilities that distribution among 28.8 ha blocks for adult females, fawns, mature males, yearling males, and total males during autumn, 1976-1983 was not significantly different . Fawns Mature Males Yearling Males Total Males X2=86A.A P=0.000 df-521 X2=827.7 P=0.000 df=A71 Adult Females Fawns X2=250.2 P=1.000 df=AA5 X2=788.9 P=0.000 df=A89 X2=71A.8 P=0.000 df=A25 X2=720.2 P=0.000 df=A93 X2=687.2 P=0.000 df=A36 Mature Males X2=352.8 P=0.003 df=283 Table 7.3. Spearman correlation coefficients indicating degree and direction of association in distributions among 28.8 ha blocks for adult females, fawns, mature males, yearling males, and total males during autumn, 1976-1983. Adult Females Fawns Mature Males Mature Yearling Total Fawns Males Males Males 0.589 -0.098 -0.0A0 -0.05A P=0.0001 P=0.03 P=0.39 P=0.22 N=AA6 N=A90 N=A9A N=522 -0.2A5 -0.212 -0.20A P=0.0001 P=0.0001 P=0.0001 df=A26 df=A37 -0.225 P=0.0001 df=28A df=A72 179 Females and fawns, again as expected, were significantly positively associated; though the correlation coefficient did not approach 1.00 because not all females produced or reared fawns each year. The distribution of females was significantly negatively correlated with the distribution of mature males and was also negatively correlated with yearling males, but not significantly. The negative correlation between distributions of fawns and mature males was significant as was that between fawns and yearling males. Distributions of yearling and mature males also were significantly negatively correlated with each other. The stronger negative correlation between fawns and mature males than between females and mature males indicated that mature, productive females with fawns generally were distributed differently than mature males, whereas non- productive females often occurred in the same 28.8-ha blocks as males. The data also indicated that yearling males were more widely distributed and occurred in the same blocks as females more often than mature males . Temporal Changes in Spatial Distribution Comparison of Figure 7.1 with a more generalized map of density distribution of mule deer on the study area during 1960-1964 (Fig. 7.2, Mackie 1970) indicated that areas of highest and lowest densities were reversed for the 2 periods of intensive study. This difference may have been influenced by 2 major environmental changes. The first involved changes in timing of livestock grazing. The area of comparatively high mule deer density during 1960-1964 was centered around a private ranch on which cattle were grazed only during winter. In about 1970, the area was leased to other operators and thereafter heavily grazed during spring and occasionally grazed during other seasons. Winter grazing with supplemental feeding probably had little effect on the composition and abundance of spring and summer forage for deer and may have made spring growth more available. The absence of livestock in this diverse drainage-head habitat of moderate to low topographic relief may also have rendered succulent, high- quality forage more available through summer. Conversely, heavy grazing during spring over several years may have reduced the quantity and quality of summer forage and decreased quality of the area as fawn-rearing habitat. Grazing practices along the northern and northwestern portion of the area, which held relatively low deer densities in the early 1960s and high densities in recent years, followed the opposite trend. Headquarters ranches and winter livestock grazing and feeding operations along the Missouri River were phased out during the 1970s. There, habitats adjacent to the river that had been moderately to heavily 180 o r- cn 0) ■H u S E O u m en o en Cn C •H M a -a i +J ■H CO c CD 73 (1) > •H +J (0 ■-H • 1 1 4 * • •t£ * '• f* • * •« ■ • T * t in * i ■ ^ 1 • ■ • ■ ♦ 1 * » t w t 4 * 1 i ■ 1 4 * % • ■ % t * . ♦ £ ■A^ • * \ • 1 \ j i ■ * >— w» -f r H ^ ^ \ v% I %.\ I •. \ 1 \* LU mi CO < O V) w E in U Q) XI E 3 C c o •H ■p (0 rH O o c •H T3 w O i-H id g 0) M 0) TJ 0) .-I e T3 0) •H ft 3 U u o u o 00 CTv w 3 O 3 Oi •H I c r- O c^ N^ r» d) U 3 Cn •H 184 Table 7.4. Relative distribution of female mule deer densities at low (1976-1978) and high (1981-1983) density populations , Missouri River Breaks, Montana. Type of area Size Female Density Female Density at low densities of area 1976-1978 1981-1983 (Km2) (No. /Km2) (No. /Km2) Female core blocks 52.1 4.17 3.86 Scattered female blocks 15.0 4.48 3.86 Blocks within 0.8 km of 79.2 0.15 3.40 female core blocks Blocks used by deer at 81.7 0.0 2.51 high densities Blocks never used by deer 4.6 0.0 0.0 Entire Study Area 274.3 1.08 2.66 We were able to collect information that increases the knowledge about social organization of Rocky Mountain mule deer and its relationship to population dynamics. Additionally, by use of marked deer, we were able to determine the social relationships among maternally related deer. Group Size and Composition Group size statistics were computed for 4 categories of deer: total deer, adults only, adult females only, and adult males only. Group composition data were tabulated for 3 sex- age categories: yearling males, mature males, and adult females. Within each of those categories, 5 types of social groups were delineated. Using yearling males as an example, the social groups were: 1) solitary yearling male, 2) yearling male(s) with mature male(s), 3) multiple yearling males, 4) yearling male(s) with adult female(s), and 5) yearling male(s) in a mixed group containing at least 1 each of yearling male, mature male, and adult female. Mean group size for total deer, adults, and females increased from the fawning period through March (Table 7.5, Fig. 7.4). Data from mature female groups (Fig. 7.4) 185 Table 7.5. Mule deer group sizes during aerial surveys in the Missouri River Breaks, September 1976 through April 1984. Group Adult Adult Time All Deer Adults only Females only Males only July 1.59+0.04aAb 1.29+0.03A 1.08+0.02A 1.53+0.07A d-7)c (1-6) (1-4) (1-6) 2659d 1626 1163 548 Sept. -Oct. 2.4+0.07B 1.61+0.05B 1.46+0. 05B 1.55+0.08A (1-11) (1-8) (1-6) (1-7) 1400 1382 1009 480 Late Nov.- 3.04+0.25C 1.85+0.15C 1.53+0. 15B 1.10+0.07B Early Dec. (1-14) (1-8) (1-8) (1-3) (Rut) 190 186 161 88 Late Dec . - 3. 30+0. 11C 2.05+0.07C 1.92+0. ,07C 1.42+0.07A Early Jan. (1-16) (1-10) (1-10) (1-6) 1645 1593 1356 465 Mar. -Apr. 4.52+0.18D (1-20) 1061 1052 3.01+0.13D (1-13) a X + 952 CI b Means in columns followed by different letter are different, P<0.05. c Range of group sizes d Number of groups indicated that group size then declined through April and May to annual lows during June and July. The number of females per group was lowest during July (and probably late June) when an average 85% of all adult females did not occur in the same group with another female. Number of males per group peaked during summer and early autumn, but observations of marked males suggested that group size for mature males may have reached another high during mid-winter through spring. For a small sample of 59 observations of marked mature males during mid- January through mid-May, where the entire group sex-age composition was known, there were the fewest solitary mature males and greatest percentage of mature males with other mature males of any recorded period (Table 7.6). The smallest group size for males was during the rut when 82% of males did not occur in a group with other males. 186 Q. O O Q 6n 5- 4- 3- o -Q E 3 Z 1 2- Total Adults Females — i 1 1 1 1 1 1 1 1 1 1 1 Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Figure 7.4 Group size by month for 37 radio-collared 2- year-old and older females. During summer and early autumn, yearling males were almost equally distributed among all types of social groups (Table 7.6). However, during July, only 10 yearling males (2.4%) were seen with females that had fawns. Information from marked deer indicated that most of the females observed in association with yearling males during summer were either yearling females or non-productive mature females. The percentage of solitary yearling males declined from July through the remainder of the biological year as yearling males increasingly associated with females. The greatest degree of association between yearling males and mature males occurred during summer and declined thereafter. The lowest association with other males occurred during the rut. A small sample of 6 3 observations of marked yearling males indicated that most (74.6%) were associated with females during mid- January to mid-May (Table 7.6, Fig. 7.5). During summer and early autumn, mature males were typically solitary or grouped with other males (Table 7.6). Very few (<13%) were associated with females during summer or early autumn. During the rut, no mature males were observed with any other male, except when both were competing for the same female. Eighty percent of mature males observed during the rut were in groups containing females, the other 20% were solitary. During early winter, as rutting behavior 187 in O •H 1-1 id > c •H ro en CD I 3 . in X) (0 5^ fdpQ CD CD a> > i-H-H >-) « a 01 4-1 T3 CO -H ■J s I • c o cd Q >- 0) 4-1 l-l cO co ■J w I • CJ > 0) O Q Z OJ 4-1 1-1 CO CO .J w I 4-1 cu u cu o C/l r- ON 00 -J CM cm r^ <* 00 00 -3- VD r-~ r~ m r» t-n cm CO rH CN rH • b- Xi o cd 0) o S3 u U 4J Tj CO 3 0) rH CD 00 >%4J C l-l CO 4-1 -O -rl •H CO Si o < s r-1 4J l-l •H ^! 4: cd .— 1 4-1 4-1 4-1 4-1 0) O ■H •rl -rl -rl >4 CO » > > :* CO CM CT\ CO -J m o m m co as rH rH m rH U") N r. en r- id r~ VO O* CM CO VO CM rH rH CM rH CM m o o co cm CTi O O 00 CM rH rH VO rH -* CO CM O -*■ rH CI O >f CO 0 m rH rH CO CM CO -J- en co oi e r-i 0> CO tK en 33 OJ en us rH 0) 0) cd In rH en 53 D id 0) 4-1 a — i 00 cd 0) cd en C 33 fc 33 0) •H rH l-l 4-) 73 cd l-l OJ rH 0) S3 r-> «£ 3 X l-l 0) 4-1 T3 "rl ai (0 >* O < S3 i-i 4-> 3 •H 43 43 43 43 4J rH 4-> 4-1 4-> 4-1 CO o •rl -H -H -H » CO Is 8* D* t* O* rH O CM 00 rH . • . . o CM lO CO in CM VO CM m rH CD 00 33 cd 01 U-, C 33 25 rH l-l *-^ •H 0) cd o l-l C/l rH IH XI e ai g u d cu 0) S-%43 Cd 4-1 H fe V4 4-1 cfl ai cd ■H CO o bu >- z: S 4-> 4-1 rH -H 43 T- 3 r-l 4-1 C 4-1 4-1 4-1 ■0 o -H cfl •rl -r| •H < CO > o^ > > c* 188 en 100 a> 4-> w 80 o Q. Q. O 60 > 40 Q. O O 20- | Yearling Males [3 Mature Males Females Jul Sep - Oct Late Nov Rut Month Late Dec- early Jan Late Jan - May Figure 7.5. Percent yearling and mature males in groups containing females and percent females in groups containing males during various time periods. diminished, mature males were well distributed among all 5 types of groups. From mid-January through mid-May, 55.9% occurred in groups containing only mature males . Over 80% of all adult females observed were solitary or associated only with new fawns during July. This percentage varied among years, depending upon the percentages of yearling and non-producing mature females in the population. Most females that were not solitary during July did not have fawns. Only about 8% of mature females were observed with males during July (Table 7.6, Fig. 7.5). Yearling females, however, were grouped with other adult deer in 57% of 170 observations of 25 marked yearling females during June-August. By early autumn, females were about equally as commonly observed as solitary adults accompanied only by fawns or grouped with other females and fawns. Females were most commonly grouped with other females and fawns during and after the rut. At the peak of rut, 35% of all females observed were in groups containing a male (Fig. 7.5). The percentage of females observed with males declined thereafter. Of all males associated with mature females, the percentage that was mature males reached a peak during November and December (Fig. 7.6). Most males observed with females at other times of the year were yearlings. 189 CO E ■D CO o o en co < CO o (0 100 90 80 70 60 50 40 30 20 10 0 Sep N 29 y>/>>>j/?>ttftt nnvxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Y*******/*******//****/****/**,,,,, / / / , / ,>.s\'»xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx \ x x x x ********************************* ******* s\\xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx x x x x x x ********************************* ******* x x x x x \ s x % x x \ x x s x xxxxxxxxxxxxxxxx \ \ x x x x x \n - W/V/VAVVAV ^ ^ ^ •> ■> i i ^ ^ i ■> v„ V, V, V/VV, V, V , \,SS\,\,\, W/'/'/'/'f w ■• .. . A x x x x x x x x x x x X X ////////////A Yearling Males K'wavawa >*>>/*>>// >/>>*>*/>>/?/ xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx *^*^* ******************************* X X X X X ****** ********************************** ****** X X X X X X X Oct Nov Dec Jan Feb Month 35 50 39 20 28 Figure 7.6 Age of males associated with radio-collared 2- year-old and older females during September through February. Factors Influencing Group Size and Composition The increase in group size from June-July through March despite a decline in total deer density through the period indicated that seasonal change in group size resulted from changing sociability and habitat use rather than changes in deer density during the year. To determine whether seasonal group size changed from year-to-year as density of deer changed, correlations between annual deer numbers on the study area (density) and group size during autumn and early winter each year 1976-1985 were computed. The null hypothesis that group size did not increase as deer density increased was not rejected in any case. Correlation coefficients between deer numbers during autumn and group size were: total group size, r=-0.242, n=10; adult group size, r=-0.067, n=10; and female group size, r = -0.102, n=10 (all P's > 0.05). Coefficients between deer numbers during early winter and group size were: total group size, r=-0.263, n=9; adult group size, r=-0.121, n=9; and 190 female group size, r=-0.180, n=9 (all P's > 0.05). Deer numbers were 3.15 times higher at the high than the low, so the available range of densities should have been sufficient to show differences if they occurred. Although statistically non-significant, the negative correlation coefficients may indicate a tendency toward smaller group size as density increased. General observations suggested that mature females were socially oriented towards their fawn(s) first, and exhibited minimal relations with other adult deer. However, as the biological year progressed from fawning, females increasingly associated with other females, especially during years of poor fawn survival. This suggested that female group size was related to fawn survival. A test of that hypothesis with correlation analysis indicated a strong negative relationship (Figs. 7.7 and 7.8) between the number of females per group and fawn: 100 female ratio in the population during autumn (r=_0.907, P<0.01, n=14). A similar relationship was indicated with respect to numbers of fawns: 100 females in the population during early winter (r=-0.920, P<0.01, n=15). Data for autumn 1985 were not included in the analysis; that data point did not fit the linear regression line, probably because most fawn mortality had occurred by mid-summer such that females had a longer than usual time to reassociate prior to the autumn survey. These correlations indicated that the higher fawn production and survival through early winter, the less likely that females were associated with other females. Group size increased as fawn survival declined both within and between years . The same data displayed in a different manner (Fig. 7.9) led to additional conclusions. During early autumn, the largest decrease in fawn: female ratio occurred between single females and groups of 2 females. The degree of decline between these as compared to the decline between other sized groups indicated that the second female in the group was the most likely of all additional females to not have a fawn. Data from marked deer verified this and indicated that the second female in the group (first one added in autumn) was most likely an unproductive yearling daughter. For the early winter period, the change in fawn: female ratio was about the same from 2 to 3 females /group as it was from 1 to 2 females/group. Again, data from marked deer indicated that by early winter, the third female added to a group was most likely to be either a second yearling daughter that had not rejoined the family group by early autumn or an unproductive 2-year-old daughter. Little difference in fawn: female ratios occurred among female group sizes 3, 4, 5, and 6 during early winter, indicating that the fourth, fifth, and sixth females added to a group were about equally likely 191 a. 3 O O i_ o Q. CO o ro E o o (0 o E 3 2.0 -1 Y =2.0255 -. 00673 X 1.9 - • R = 0.823 1.8 - 1.7 - • 1.6 - • • • 1.5 - •^^, 1.4 - • • "*«» 1.3 - '^ * 1.2 - 1.1 - 1.0 - 1 1 r - i i -i - 1 1 — — I 1 1 1 1 10 20 30 40 50 60 70 80 90 100 110 120 130 Fawns : 100 Females - Autumn Figure 7.7. Relationship between number of females per group and the fawn: 100 female ratio during autumn Q. 3 O o Q. (0 0) E o Li. -Q E 3 3.0-1 2.8- 2.6 2.4-1 2.2 2.0 1.8- 1.6- 1.4- 1.2- 1.0 Y =3.1418 -.01 368 x R = 0.848 0 10 20 30 40 50 60 70 80 90 100 110 120 130 Fawns : 100 Females - Early Winter Figure 7.8. Relationship between the number of females per group and the fawn: 100 female ratio during early winter. 192 1.00 - 0.90 " 0.80 " o co 0.70 - E a 0.60 " en c 5 CD 0.50 " 0.40 - 0.30 - 0.20 " 0.10 - 0 Autumn Early Winter Autumn Early Winter -r— 1 1 1 1 r 1 2 3 4 5+ 12 3 4 5 6 Number of Females per Group 7+ Figure 7.9. Per capita fawn production in relation to female group size, 1976-1984. to have or not have a fawn. Some of those females were productive older daughters. Also, beyond the fourth generation (3 adult generations), daughters were most likely to establish their own matrilineal group. A total of 74% of all adult females observed during autumn occurred in groups of 1 or 2 adult females. Forty- three percent of all females observed did not occur in groups with other females and 31% occurred in a group of only 2 females. During early winter, approximately equal numbers of adult females were in groups containing 1, 2, or 3 adult females; those 3 group sizes accounted for 67% of all females observed. The percentages of all deer, of all sex and age classes, that were observed in each group size are shown in Figure 7.10. From autumn through spring, more deer occurred in groups of 3 than any other group size. Averaged for all 3 seasons, 20.0% of all deer observed were in groups of 3 and 15.7% were in groups of 4. Group sizes of 3 and 4 may be most observed because the most common social group, a female or 2 females and their fawns, are usually a group of 3 or 4 deer. 193 40 -i CD N in a 3 O 30 i_ o >» -Q i_ O <1> 20 Q ra »*- o ■*-> 10 c CD O 1— CD 0. Autumn Early Winter Spring -i — i — i — i — i — i — i — i — i — i — i — P — i — T"~1 — i — i — i — i 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Group Size Figure 7.10 Percent of all deer occurring in various sized groups during autumn, early winter, and spring. Associations of Genetically Related Individuals Mother-Son and Sister-Brother Relationships Very few mother-son or sister-brother combinations remained associated with each other after 1 year of age. At 12-14 months of age, 40 (70%) of 57 marked yearling males monitored, dispersed from their natal home range and 29 (51%) of those left the study area (Table 7.7). This meant that after 14 months of age (before their first breeding season), 70% of all males were no longer associated with their mother or sisters. Opportunity for the remaining males to associate or breed with relatives was also limited. Among 17 yearling males that had not left their natal home range by late August, at least 4 had no opportunity to breed with maternally related females. Three of these were shot before the rut, and 1 dispersed 5 km away prior to the rut and did not return until afterward. Five others used areas that included their natal range during their first breeding season. Of these, 1 used an area that included its 194 Table 7.7. Dispersal rates of yearling mule deer, Missouri River Breaks, Montana, 1976-1985. Males Females Z Dispersal Z Dispersal 1 from Population Birth from N; Year N Area 1976-78 9 100 1979 11 64 1980 11 04 1981 7 71 1982 11 55 1983 4 100 1985 4 50 Total 57 70 Z Dispersal Z Dispersal from Natal from Area Population 78 10 10 46 11 27 46 15 27 57 10 10 27 11 0 75 50 1 0 51 58 16 10 27 27 0 0 0 14 natal range through the second breeding season and 2 used the natal range through 4 breeding seasons. Radio failure or harvest by hunters precluded gathering further information on these 5 . Potential for breeding with maternally-related females was not determined for the remaining 8. Six of these were last observed prior to the rut and 1 during the rut when we ceased receiving signals from their radio transmitters. Ultimate fates of these 7 could have included: dispersal beyond search range, shot during the hunting season and not reported, or remained on the area undetected. Another was shot during November, but we did not know if it was prior to or during the rut . Five males remaining within their natal ranges had marked mothers and/or sisters. Male 3379 remained within his mother's (3578) home range (Fig. 7.11) throughout 22 months that his radio collar functioned. He was never observed with his mother after 1 year of age (16 observations) except at 13 months of age on 9 July 1980, when she was chasing him. No male with a fawn-type radio-collar or ear tags was ever observed with female 3578 during the remaining 25 months she was alive. Male 2581 and his twin sister 2381 were the offspring of marked female 2481. Brother and sister were together 4 of 6 times they were observed as yearlings. On 14 October they were observed together with their mother and her current fawn. What relationships, if any, that would have occurred during the rut were not determined because the sister was shot by hunters on 31 October and the mother on 9 November. 195 t N 2 km Mother (3578) Son (3379) • Relocation sites of both deer Figure 7.11. Home range boundaries and relocation sites for female 3578 and her son 3379 over a 22 month period. 196 Observations were made on all members of a set of triplets, 2 females (3480 and 3580) and a male (3680), through 2 years of age. After 1 year, the male was with either of the sisters during only 2 of 22 observations; both were after the rut in late January and mid-April. Male 3680 was not reobserved after that time. After male 3581 and his sister 3681 reached 1-year-old, they were observed together during only 2 of 9 observations . This male was last observed alone during the rut on 26 November. Male 2480 dispersed approximately 7 km from his natal home range on 3 June, but returned on 3 July. He was then seen with his sister 2580 during all 5 observations until shot on 3 November prior to the rut. Relationships of Brothers The available information indicated that brothers did not commonly associate with each other beyond 1 year of age. Brothers 2880 and 2980 were observed together during only 1 of 10 observations after dispersal began. On 29 May 1981, they were 1.5 km apart, but within their natal home range. By 3 June, they were about 3 km apart when both were located 7-10 km northwest of their natal range. On 4 June, both were back within the natal range, but about 0 . 5 km apart, and on 5 June, they were together, within their natal range. By 10 June, they had dispersed 30 and 40 km to the east where they remained apart until November, when 1 was shot and the radio-collar of the other ceased to function. Brothers 0981 and 1081 remained together for about 2 weeks after dispersing approximately 9.5 km northeast from their natal home range. Because both wore only neckbands, they could not be reobserved at will. Male 1081 was observed alone on 16 July; 0981 was not observed at or after that time and presumably left the study area. Male 1081 was observed periodically for about 1 year within the original dispersal area where his collar was found 4 years later. Brothers 4079 and 4179, sons of female 1577, wore non-functioning radio-collars at the time of dispersal. Although their ultimate fates are not known, they dispersed on different days and presumably did not remain together. Relationships of Mothers and Daughters Most mother-daughter combinations remained socially associated except during parturition and the fawn-rearing period, though degree of association declined with age when both recruited fawns. Eighty-four percent of marked yearling 197 females remained in the area of their natal home range (Table 7.7). Most, however, did not have marked mothers, so future relationships were not determined. The following are 6 case histories illustrating relationships between marked mothers and daughter (s). (1) Female 1277 and her twin female fawns, 1077 and 1177, were captured and marked in February 1977. The adult female was equipped with a radio collar and both fawns marked with neckbands. From breakup of the family unit about 1 June until 14 October, 1277 was observed 6 times, never with either daughter. She was observed with a new fawn during June and July, but was not accompanied by a fawn when observed after July. Her yearling daughters were observed together on 28 May, 17 June, and 20 June. Thereafter, neither was seen until 14 October, when one (1177) was with the mother. On 27 October, the other (1077) was with the mother, but sibling 117 7 was not observed. On 3 November, neither yearling accompanied 1277, but they were with her by 11 November. From then through 9 May 1978, the 3 were together in 12 of 13 observations. Daughter 1177 may have been hidden by dense timber on the 1 occasion she was not observed in the group. From 2 June 1978 until 9 October 1978, 1277 was observed 11 times, never with either 2 year-old daughter. The 2-year-old daughters were not reobserved until 9 October, when all 3 deer were together. None of the 3 females were observed with fawns during summer, but it appeared that 1177 was followed by a fawn on 9 October. From that time through 21 April 1979, the 3 were together a minimum of 10 of 13 times they were observed. However, dense timber may have prevented observation of all deer in the group on 2 of the 3 occasions they were not all observed together. The radio-collar on 1277 failed after May 1979, and all of these deer were observed infrequently after that time. From 28 May 1979 through 29 November 1979 (after which 1177 was never reobserved) , the sisters were not together any of the 5 times that 1177 was observed. Both 3-year-old females were accompanied by twin fawns throughout the period. Neither matriarch 1277, nor 1077 was reobserved until 31 December 1979; they were first observed together on 16 February 1980. They were observed together only 2 of the 8 times that either was observed during 31 December 1979 through 23 April 1980. Both females recruited twin fawns that biological year. During the next biological year, when 1277 was 11-years- old, and 1077 was 4-years-old, they were observed a total of 11 times, and only once (16 March 1981) were together. Both recruited twin fawns. 198 Thereafter, they were observed only once, on 21 September 1981, when they were together in a group of 6 adult females and 3 fawns. The matriarch, 1277, was shot by hunters on 26 October 1981. Daughter 1077 continued to used the same home range (Fig. 7.12) through 14 March 1983, after which she was never reobserved. This matriarch and daughters continued to use the same home range (though not at the same place and time, during fawn rearing periods) as long as all were alive. The mother, and daughter (1077) that remained alive the longest, used the same home range until the mothers death at age 12-1/2; the daughter was 5-l/2-years-old. After regrouping in autumn, mother and daughter were in the same group an average of 88% of the time until the daughter reached age 3 years . During the next 2 years, when both mother and daughter recruited twins, they were observed in the same group only 3 of 19 times. During the fourth year, they regrouped early, and the indications were that at least 1 did not have fawns. Although the boundaries of their home ranges as adults were nearly the same (Fig. 7.12), they only rarely occurred in the same group when both recruited fawns. There were 2 areas used by the matriarch during the fawn-rearing period in which the daughter was never observed during June through August after she was 1- year-old (Fig. 7 . 12) . (2) Female 0376 and her daughter 1680 were observed from the time 1680 was 6 months old until 44 months old. Female 0376 was a minimum of 6-l/2-years-old at the time 1680 was radio-collared. Observations on this relationship ceased on 10 February 1983 when 0376 lost her collar, but the presence of 0376 through at least 14 March 1983 could be verified because she retained a matted band of hair around her neck. After the initial breaking of the maternal bond, when 1680 was one year old, mother and daughter were not observed together until 2 September. During the remainder of that biological year, the mother, her new fawn, and daughter 1680 were together in 20 (95%) of 21 observations. During the next biological year, both 0376 and 1680 recruited fawns and were not observed in the same group until 3 November. For the rest of that biological year they were in the same group 12 of 13 times. Both 0376 and 1680 recruited fawns during the following biological year and were hot observed together until 14 December. From then until 0376 lost her collar, mother and daughter were together in 5 of 7 observations . 199 &> 1 km Mother (1277) Daughter (1077) Relocation sites for both, combined Areas used by matriarch 1277 during June - August but never used by daughter 1077 during that time after she was one-year-of age. Figure 7.12 Home range boundaries and relocation sites for female 1277 and her daughter 1077, including areas used exclusively by matriarch 1277 during June-August . 200 This mother-daughter pair remained associated with each other for a minimum of 3-1/2 years after the daughter's birth. Once they regrouped after the fawn rearing period each year, they were in the same group a total of 37 (90%) of 41 times. (3) We radio-collared the newborn fawns of neckbanded female 2378 in 2 successive years. Female 2378 was an estimated 5 years old when newborn 3079 was captured. After 2378 isolated herself to give birth to 2180, her previous daughter 3079 was not observed with them until 2 September. From that date through April, however, all 3 deer were in the same group during 15 of 16 observations. Daughter 3079 left the matrilineal group by 4 May at 23 months of age, while 2378 and her 11 month old daughter 2180 remained together until 3 June. On 5 June, the yearling was alone, but was with her 2-year-old sister on 10, 12, and 24 June. From that time, until 2180 lost her collar on 15 August, none of the 3 females were observed together. The radio collar on 3079 also ceased to function about that time. Female 2378 was next observed on 21 September 1981, with 2 other females and 2 fawns. One of the adult females was wearing a fawn-type radio-collar and was presumed to be 3079. These 2 deer were observed together all 5 times that 2378 was reobserved during the rest of that biological year. On 30 December 1982, the same 2 deer were observed together and the adult with the fawn-type radio-collar was recaptured, verified as 3079, and marked with an individually recognizable neckband. Reobservations were few from that time through spring 1987, but 2378 and 3079 were in the same group during 16 of 17 times they were seen from autumn through spring. They were not together during any of 3 summer reobservations, but were observed together by September in both 1983 and 1984. This mother-daughter combination continued to associate with each other through 8 years of age for the daughter. (4) Data on associations of female 2479 and her daughter 3679 were collected through 5 years of age for 3679. Mother and yearling daughter were no longer together by 8 June. The mother was always observed alone through 18 June and was not observed with new fawns. On 25 June and 3 July she was observed with a yearling male, indicating that she probably did not have fawns. Mother and daughter were first reobserved together on 8 August, but occurred in the same group during only 6 (30%) of 20 observations through the remainder of that biological year. Both mother and daughter reared fawns during the following year, however reobservations were few because neither retained a functioning radio-collar. They were first reobserved together, with their fawns, on 28 August and were 201 together during 6 of 9 observations through the rest of that biological year. They were observed together only 1 of 6 times during autumn through spring over the next 2 biological years . This mother-daughter combination, although continuing to use the same general home range area (Fig. 7.13), had the least degree of association of all marked mother-daughter pairs. They were in the same group during only 13 (37%) of 35 observations during autumn through spring periods for the daughter's ages of 1-4 years. (5) Female 6581 was marked with a neckband only, but her daughter 6381 wore a radio-collar. The daughter was first observed apart from her mother at 11 months of age on 13 May 1982. They remained separated through 24 June, after which radio contact with 6381 was lost. Because attempts to relocate 6381 were unsuccessful through July and August, she apparently had dispersed off the study area during that time. She may have returned to the area anytime after August, but was not re-observed until 26 December 1982. She was first seen with her mother on 2 February 1983. Thereafter, mother and daughter were together during all of 7 observations until 8 June when they again separated. They were next observed together on 4 August 1983 and each had a single fawn. From that time, until 23 February 1984, mother and daughter were together 15 of 19 times they were observed. The mother was never reobserved after 23 February 1984. (6) Female siblings 1582 and 1682 were the daughters of female 1681. From family breakup on 18 May 1983 through 22 August 1983, when they were first reobserved with their mother, the sisters were together only 4 of 11 times either was observed. Once reunited with their mother, they were together in 15 of 19 observations through 19 March 1984 when 1582 was killed by a coyote. During that period, 1582 was with her mother on 13 of 19 observations and 1682 was with her mother on 14 of 19 observations. When not with their mother, these siblings were together 3 times, alone once each, and with female 1481 (of unknown relationship, but usually with matriarch 1681) the remainder of the time. From the time 1582 was killed through 28 May 1984, 1682 was with her mother 4 of the 5 times she was observed. Female 1681 and her daughter 1682 were next observed together on 20 October 1984. They were in the same group during 3 of 4 subsequent reobservations . Association of Sisters Three cases involving the association of sisters were described with mother-daughter relationships. An additional 3 sets of twin females were broken up when one of the sisters 202 • • /, ' • • • /, ' • • • // • • • • ^» •••••• yV * • • • <• a • • • ••••• •• •••••• •• • • • • \ • • • • • • \ • • • • • • »_ • • • • • '/ Mother (2479) Daughter (3679) o Relocation sites for both, combined t N i 0.5 1 km Figure 7.13 Home range boundaries and relocation sites for female 2479 and her daughter 3679. 203 dispersed from the study area while the other did not. One case of marked sisters also occurred as part of a set of triplets . The relationship of triplet male 3680 with his sisters, 3480 and 3580 was discussed previously. The sisters were most often apart during the first month following family breakup but thereafter were usually in the same group through their second year. Family breakup occurred on about 23 May 1981. On that date, female 3580 and male 3680 were together, but female 3480 was separate from her siblings. All were separated on 29 May, but on 3 June female 3580 and male 3680 were back together. All 3 were separated on 5 , 8, 12, and 18 June, but the sisters were together on 10 June. Thus, during the period 23 May to 18 June, the sisters were together only 1 of 8 times that they were observed. From 24 June 1981 until 1 May 1982, near the start of their first fawning season, they were together 13 of 17 times that they were observed. During their third year (1 June 1982 through 31 May 1983), the sisters both had their first fawns; in each case, a singleton that appeared to survive through 1 year. They were observed 13 times that year, but never in the same group, although on 29 July 1982 they were observed about 100 m apart, each with her young fawn. During their fourth year (starting at 3 years of age) the sisters were observed 3 times during June and July but were never together. From 4 August through 14 December 1983, they were together 5 of 11 times they were observed. Both either had not had or, more likely, had lost their fawns by 4 August 1983. Female 3480 was collected on 14 December 1983, ending further possibilities of association. Female 3580 has remained on or near her natal area through 1987. Associations of Marked Deer of Unknown Family Relationship Two mature males, observed from January 1979 through October 1980, were closely associated with each other except during the rut. During January-October, the two were in the same group during 24 (59%) of 41 observations over the 2 years. The association was strongest during late February through May, when they were together 14 of 16 times either was observed. Although these 2 mature males were frequent associates, based on the dispersal rates of yearling males, it is unlikely that they were maternally related. There were several instances of marked females associating frequently with each other when the family relationship of the individuals was unknown. In the previously described relationship involving female 1277 and her daughters 1077 and 1177, another marked adult female 204 (1377) of unknown family background, was frequently in the same group. The 4 were captured as a group during February 1977. Female 1377 rejoined the group in mid-December 1977 and 1978, after 1277 and her daughters had regrouped. From spring 1977 through spring 1979, after the 2 older females had joined the group, they were together during 31 (84%) of 37 observations. On the occasions when the daughters of 1277 were not with her, they were often with 1377. Although the relationship of the 2 older females was unknown, based on their ages it is likely that 1377 was a younger sister or previous daughter of 1277. A similar situation existed in the previously described relationship of female 1681 and her daughters 1582 and 1682. Another marked mature female (1481) was often associated with this marked mother-daughter combination. From the time the 2 mature females reassociated with each other by early winter each year until the next parturition period, they were together during 38 (75%) of 51 observations. Again, when the twin yearling daughters were not with their mother, they were often with 1481. Based on the age of 1481 at capture, she could have been a previous daughter of 1681 (older sister of 1582 and 1682) . In 1980, twin male fawns (2880 and 2980) of female 1577 were captured and radio-collared. One of a set of twins (female 3080) of an unmarked female was captured and radio collared at the same time, near the same area. During summer and autumn, these 2 sets of twins and their mothers had overlapping home ranges (Riley and Dood 1984), but did not group together. From 13 February 1981 until 29 May 1981, all marked deer, and presumably the mother of 3080, were together 4 of 7 times they were observed. The yearling males both dispersed from the study area at 1 year of age. The yearling female was reobserved with female 1577, her new fawns, and other unmarked deer on 14 October and 3 November 1981, but lost her radio collar within a week of the last relocation. On 29 December 1982, 3080 was recaptured and refitted with a radio collar. For the next 2 winter-spring periods, 1577 and 3080 were together 11 of 17 times they were reobserved. Although 3080 was not the daughter of 1577, it is possible that her unmarked mother was maternally related to 1577. Summary and Discussion The spatial distribution deer on the area was clumped or aggregated and not uniform or random. "Core blocks" appeared to support relatively stable numbers of deer, and most fluctuation in deer numbers occurred in areas adjacent to and distant from those "core blocks". It was apparent that not all areas and habitats provided for equal production and survival of deer and that the stable "core blocks" observed 205 were areas that had been preempted by established matriarchs. Home ranges that included the stable "core blocks" were passed on through the matrilineal family. As long as the matriarch remains alive, however, female fawns must establish their own "parturition territory" (Ozoga et al . 1984) in areas adjacent to that of their mother. Thus, "core blocks" become larger "core areas" and include the "parturition territories" of several generations. The pattern of dispersion for the population indicated that in rapidly increasing populations, marginal habitat is filled as female fawns recruited in "core areas" must establish their own "parturition territories" at some distance away from that of the original matriarchs. All of this leads to a density distribution (Fig. 7.1) that appears as aggregated areas of high use surrounded by roughly concentric areas of lower, declining use. Mature males and yearlings of both sexes were relegated to areas not preempted by productive females . It was apparent that not only did different sex and age classes most often occur in different social groups, but they also tended to occupy different land areas during most of the year. Thus, for many analytical purposes, the different sex and age classes should be viewed as separate populations rather than as a homogeneous "mule deer population". Clutton- Brock et al . (1982) also indicated that male and female red deer, at least, should be viewed as separate populations. For some analyses, it may be valid to treat mature females and their fawns, non-productive females, yearling males, and mature males as 4 separate populations. Collectively, the information from marked deer indicated that matrilineal groups are the predominant social group for mule deer in the Missouri River Breaks. The predominance of matrilineal social groups and their structure was very similar to that reported for white-tailed deer by Hawkins and Klimstra (1970) and Ozoga et al . (1982a). The association of mother and daughter was known to last a minimum of 8 years . The relatively low level of dispersal by yearling females contributed to the formation of matrilineal groups. The degree of association between mother and daughter generally declined with age, but that was probably primarily related to the increasing likelihood that the daughter had her own fawns and became more likely to form her own matrilineal group. When both mother and daughter recruited fawns, reassociation often did not occur until mid-to-late-winter. Data on group characteristics from the entire population also indicated that females were less likely to group with other females when fawn survival was high. However, a 12-year-old female and her 7- year-old daughter were known to reassociate by early autumn when neither had living fawns. Matrilineal groups were known to contain a minimum of 3 generations; a matriarch, her daughter (s), and fawns of both the matriarch and her previous 206 daughters. We strongly suspected that some groups contained at least 4 generations; including at least 3 adult generations and the fawns of females 2-years or older. Sisters, both twins and sisters of different ages, were closely associated, often occurring in the same matrilineal group, especially if at least one did not have fawns. We were not able to determine the associations of sisters in cases where the matriarch was dead, but combined information from marked deer indicated that sisters were somewhat less closely associated than mother-daughter combinations. The only other social group we observed with a potential for relatively long-term stability occurred among mature males. Our data on these groups were limited, but males were commonly observed in groups from late January through early October, and some individuals were relatively closely associated with each other during that period. Agonistic behavior of mature males toward each other during the breeding season disrupted these groups, but they re-formed by late January or early February. High hunting mortality limited the opportunity for long-term association of individual males, thus lessening the likelihood of long-term stability. There was a tendency, at least during summer and autumn, for females with fawns and mature males to be socially and spatially separated. That indicated social and physiological requirements may differ between sexes (Verme 1988) and that competition for forage and cover between males and their offspring probably was reduced. Because 70% of yearling males emigrated from their natal home ranges, the opportunity for inbreeding was reduced considerably. Additionally, as a result of several causes, primarily pre-rut hunting mortality, a minimum of 4 of 17 non-dispersing males were known not to have bred with their mothers or sisters. Based on suspected additional hunting loss, that number probably was higher. Including the dispersing males, a maximum of 23% of yearling males could have possibly bred with their mothers or sisters. Although a small percentage of yearling males remained on or near their natal range, matrilineal groups did not contain males. In one case, a yearling male was in the same group as his mother in autumn prior to the rut, but no males were ever observed with their mother during or following their first rutting season. The low rate of dispersal by yearling females made father-daughter incest theoretically more likely than son- mother or brother-sister incest. The high rate of hunting mortality for mature males reduced that theoretical probability considerably, however. The father faces the last part of a hunting season after the daughter is conceived and 207 then 2 winters and a full and partial hunting season before she reaches breeding age. The statistical probability of the father living to the first breeding season for the daughter is 28%, based on average mortality rates observed during this study. Survival probabilities for the father decline considerably from that for subsequent breeding seasons. Probability of inbreeding is relatively small for this population, but the fact that at least some potential for inbreeding remains may be beneficial (Shields 1983). The occurrence of matrilineal groups probably had several survival advantages that outweighed the disadvantage of increased resource competition within an area. As population numbers increase, additional deer put increasing pressure on the resources of an area. When mature females allow their female offspring to rejoin them after early fawn-rearing, thereby forming matrilineal groups, they ensure that the resources within their home range, although somewhat diminished for them, are beneficial to and passed on to their genetic heirs rather than to unrelated deer. Kin-selection principles (Maynard-Smith 1964 and Hamilton 1964) can explain much of the rationale for the formation of matrilineal groups. Matrilineal groups are also especially beneficial to the offspring of deer that only move to "winter range" during the most severe winters. Continued association with the matriarch ensures that they will eventually learn the location of "winter ranges". The fact that yearling males are primarily associated with female groups also increases the chances that they will learn the locations of "wintering areas" even though they are usually genetically unrelated to the female groups. This may also be beneficial to the females because they are increasing the survival odds of a potential mate of new genetic stock. Our data enabled us to address the issue of factors determining group size as discussed for white-tailed deer by Hirth (1977). The general consensus has been that group size varies with habitat type and that relationship evolved as a means of minimizing losses to predation and possibly optimizing feeding efficiency. Our data indicated that other factors also are involved. Further, the general perception that mule deer are more gregarious than white-tailed deer (DeVos et al . 1967) may not result from inherent social differences, but rather is related to differences in habitat use and fawn survival . The habitat on our study area was a mixture of moderately dense cover interspersed with relatively open forest cover and very open shrub-grasslands. Deer had a variety of cover options, but dense cover was limited. Our data indicated that group size for adults was dependent on fawn survival. Females 208 tended to remain solitary as long as their fawns survived, although regrouping with yearling daughters was common after the fawn-rearing period. When both the matriarch and her daughters of fawn-bearing age recruited fawns, they often remained apart; if they regrouped, it was usually during late winter or spring. Fawns, the deer most vulnerable to predation, occurred in the smallest social groups; larger groups contained mostly adult deer. Because solitary females and their fawn(s) primarily used relatively denser cover (Riley and Dood 1984), the predator avoidance hypothesis that large ungulates are most likely to avoid detection by predators in dense cover when alone or in small groups seemed to hold. During the first 45 days of life, however, females and fawns made considerable use of open habitat (Riley and Dood 1984) and did not group with other deer. The only time that deer made extensive use of open habitats was during spring when feeding on ubiquitous new growth that appeared first in open habitats. For the limited time period that larger groups fed together on open areas, forage competition was not a problem. The most efficient way to use the high density of available food and detect predators was probably to feed in large groups. These groups were only temporary feeding aggregations and after feeding, the deer separated into smaller family groups when they moved back to denser cover to bed. We concur with Hirth (1977) that both predator avoidance and optimum feeding efficiency considerations may interact with habitat type to help determine group size. However, our data indicate that, in areas where a choice of habitat type is available, females with fawns prefer solitude and may preempt denser cover. The most observable groups are the larger groups that often use more open habitat, however, these larger groups did not necessarily prefer or always use open habitat. On our area, larger group size was more related to poor fawn survival than to the type of habitat the deer were using. Thus, in late summer and early autumn 1985, deer were in larger groups than usual even though they were not using more open habitat than normal. It is possible that the larger group sizes Hirth (1977) observed on the Welder Refuge as compared to the George Reserve were at least partially related to the poorer fawn survival on the Welder Refuge as compared to the George Reserve. We agree with Hirth (1977) that density of deer did not affect group size. On our area, the largest groups were observed during years of poor fawn survival that included years of both high and low deer density. However, we do not agree that mule deer are inherently more gregarious than white-tailed deer (DeVos et al . 1967). Very small family groups appeared to be the preferred social group, especially 209 as long as fawns survived. Both Wood (1987) and Hirth (1977) reported larger average and maximum group sizes for white- tailed deer in prairie habitat and open river bottom habitat, respectively, than we observed for mule deer. The perception that mule deer are more gregarious than white-tailed deer may reflect the fact that many reported studies of mule deer were in relatively open habitat and/or were conducted during periods of poor fawn survival. 210 CHAPTER 8 HOME RANGE AND MOVEMENTS Home range characteristics were determined from 6 395 relocations of 135 radio-collared deer monitored during 1976- 1984. Eighty-three of those deer were first radio-collared as newborn fawns; 30 of those also contributed home range information as adult deer. The remaining 52 radio-collared deer provided data on home range characteristics of adults. A total of 76 individual adult females and 6 adult males provided data for calculation of adult home ranges. Home range size and deer mobility were determined by 2 methods, the polygon home range (Mohr 1947) and average activity radius (Robinette 1966). The polygon home range (PHR) usually overstated home range size by including non-habitat within home range boundaries, but had the advantage of being a commonly used technique which facilitates comparisons. We used PHR to estimate home range size rather than various statistical methods because deer on this area did not use the environment in a bivariately normal manner and because statistical methods often "provide estimates that distort and obscure real biological phenomena" (Smith 1983). The average activity radius (AAR) was not a good measure of home range size or shape (Jennrich and Turner 1969) but gave a general measure of mobility by computing average distance of relocations from a geographic activity center. Home range size was calculated for 5 time periods. Life home range was calculated for 41 adult females and included only those deer monitored for 2 4 months or more. Mean number of relocations was 84.2 (range, 29-192) and mean number of months of continuous data was 46 (range, 24-85). Correlation analysis and plotted data indicated that once at least 29 relocations were made over a 24 month period, neither PHR nor AAR significantly increased (P>0.05) as number of relocations or number of months of observation increased. Once 24 months of radio-relocations were obtained, all observations of those deer were included in calculations of life home ranges, including those made after the radio-transmitter ceased functioning . Annual home ranges included relocations between 1 June and 31 May (biological year) . Sixty-seven annual home ranges were calculated for 36 females, and 6 were calculated for 5 adult males. Mean numbers of relocations for annual home ranges were 28.7 (range, 19-50) and 25.0 (range, 16-32) for adult females and adult males, respectively. Summer home ranges for adult deer included observations from 1 May through 30 November. Ninety-nine summer home ranges were calculated for 48 females and 7 for 5 adult males. The mean numbers of 211 relocations were 17.5 (range, 10-38) for adult females and 13.7 (range, 11-19) for adult males. Winter home ranges included relocations from 1 December through 30 April. One hundred-eleven winter home ranges were calculated for 70 females and 4 for 3 adult males. Mean number of relocations during winter was 10.6 (range, 8-20) for adult females and 10.5 (range, 8-13) for adult males. Summer home ranges for fawns included all relocations from 13 June through 22 September. Only those fawns that survived through at least late August were included in home range analysis. Mean number of relocations for 83 fawns was 18.9 (range, 8-31). Two types of movement patterns were determined for adult females. Yearlong residents could be found in any portion of their home range at any time of the year, though certain portions received more use during some seasons than others. Deer having distinct seasonal ranges that fit Baker's (1978) characterization of migratory movements moved regularly on a seasonal basis between areas separated by a minimum of 3 km. Migratory deer moved a maximum of 8.8 km between seasonal ranges. They were almost always deer that used the southern edge of the study area (near the prairie) during summer and moved during autumn and winter to areas near the river with much steeper topography and with both north- and south-facing slopes. Migratory adult females had significantly larger PHR and AAR (Mann-Whitney tests, all P<0.01) than yearlong resident adult females for life, annual, summer and winter periods (Table 8.1). A striking characteristic of home range measurements for all classes of deer and all periods, was the extreme individual variability in home range size and movements. The range of values in Table 8 . 1 indicated that not only do values overlap considerably between categories and seasons, but a wide range of values existed within each category in each season. Wide seasonal and annual variation of home range size was true even for individuals. Because of this extreme variation, only large differences in values were statistically significant. The annual PHR for migratory female 1577 varied from 4.55 to 16.64 km2 over a 5 year period, and the annual PHR for resident female 1480 ranged from 2.14 to 5.17 km2 over a 3 year period. The summer PHR of migratory female 1577 ranged from 3.69 to 9.77 km2 over 5 years, and the summer PHR of resident female 1580 varied from 1.68 to 4.48 km2 over a 3 year period. The variation in annual and seasonal home range size for individuals was not the result of differences in the number of relocations during the period. Similar numbers of relocations were made annually for most deer and PHR size did not increase with increased number of relocations for those deer at the extremes. For example, the smallest annual PHR for female 1577 (4.55 km2) was during the year we made the most relocations (50) of her. During the year we made the 212 fewest relocations (22), her PHR was twice as large (9.35 km2) Factors Affecting Home Range Size and Use Migratory deer had larger home ranges than resident deer, but we also examined the influence of deer density, sex and age of deer, fawning status of females, forage production, and habitat on home range size. Additionally, non-quantifiable influences such as tradition and chance were considered. Both PHR and AAR were significantly larger (Mann-Whitney tests, all P's <0.01) for adult males than for all adult females, annually and seasonally (Table 8.1). The home range size for a stag (castrated or non-descended testicles) was intermediate to those of adult females and adult males, but closer in size to those of adult males (Table 8.1). To reduce variability, we used only yearlong resident females to test the effect of reproductive status on home range and movements. Females with fawns were not as mobile as those without fawns (Table 8.2). Females with fawns that survived throughout summer had a significantly smaller mean PHR (2.34 km2, n=23) than those without fawns (4.50 km2, n=27; M-W Rank Sum = 369.0, P<0.01). Mean AAR also was significantly smaller (M-W Rank Sum = 418.5, P<0.01) for females with fawns (0.72 km) than for females without fawns (0.92 km). Because this relationship occurred across all years, densities, forage conditions, habitats, and age classes, the presence of f awns-at-side overrode all other factors in determining the home range size of adult females. Our data did not support the assumption that females would have smaller home ranges as density increased_ (Table 8.2). Four density classes were chosen; 1976-78 (X = 310 adult females, range 290-335)_, 1979-80 (X - 488 adult females, range 425-550), 1981-82 (X = 690 adult females, range 680-700), and 1983-84 (X = 775 adult females, range 760-790). Summer home range size of resident adult females was not significantly different among density classes (Kruskal-Wallis; PHR (X2=3.43, P=0.33, df=3) and AAR (X2=0.08, P=0.99, df=3)]. For most categories, sample size was too small to partition the effect of reproductive status and density. A sufficient sample was available to compare only resident females that did not have fawns in 1979-80 (Table 8.2, n=7, mean density = 488 females) to those that did not have fawns in 1983-84 (n=10, mean density = 775 females) . Neither PHR (X = 4.32 km2 and 4.29 km2) nor AAR (X = 0.94 km and 0.89 km) were significantly different for 1979-80 and 1983-84, respectively (Mann-Whitney, Rank Sum =72.0 P=0.41; Rank Sum=64 . 0 , P=0.96). 213 0) cr □ < c < S3 LE QJ £ O X t_ o> c tr 3 X a 2: cc QJ < a < c OJ Ct a> E O X u CU E E a: D X IS> Q- 2 -* ro Kl O r- -o oj ** «- ^* •O in a - OJ • lA ■ OJ • ro • fO • OJ O i O • O r O i + 100 + |r- + |0O £}£, ro ro T- -* •o to 00 OJ 00 • Irt o ■ 00 ■ • o - o • O • t— O w s- . «— «— OJ CO •o • OJ • OJ s* m -* • -4" ■ in 00 OJ O i T— 1 O i + 1 ' + io + |C0 + l>o «- o *— OJ • OJ - OJ • • o • o • o oj in r\j OJ — ' in •*-* K1 ^ «~ **^ o + 100 + 100 i1*- IO -4" t- ro ro ro O Od 00 ■ to ■ o • o> ■ • o • O • o • *— o — 00 r\j ro ro 00 ■ IT* *— ro oj r-- -o • a LP. f\J O I t— i O I + 1 I + IO + IO + IO «- -d- -* • o • >» • o ■ T— • o • o f*~ LA rO -^ in w V* *-* «- ■-* 00 00 o r- <\J LA •J- o ^ ^^ ^ ro O a r- r*- >o ro r>- CM IS. O «- O • (\j - IO • • f^ • 00 ■ ro O l O i O I O i + l>* + |nO + |nO + IO g^ f\J *d- o ■» «- 00 LO ** - T— ro • O • o o • o • «— o — *- *-* «- *-^ ro w *~ ro to O r- og O <— o • to ■ ro Kl LPt 1 *— i OJ 1 x— 1 + |oo + |m + |r-- + |m «— OJ ■ r- • O • -o r*. *~ m *— - 00 v O ""^ ro -' O r- ■o -4- OJ >o ^ ^, ^ OJ r- ro O* «■* c> O - ro ■ t— OJ OJ OJ O 1 • 1 O 1 + l^o O -J- + s o o + |Ch in 00 • in - o • O o O v* *- *-* «~ s-* XJ 00 r- T- ** ■-» r- • m • %* to 1 oj ro «— i + |00 + 1 i + |ro ^t m m Kt • • o • ** CO f- OJ ** — < L. < 4) E O — » 0C E O X c o CD ir» rn <>• ( + 1 (0 214 Table 8.2. Relationships of summer home range size and mobility of non- migratory adult females to various potentially influencing factors . Na PHRD AARC With fawns Without fawns 1976-78 (ave. density 1.1/km2) 1979-80 (ave. density 1.7/km2) 1981-82 (ave. density 2.5/km2) 1983-84 (ave. density 2.9/km2) Without fawns 1979-80, 1.7/km2 Without fawns 1983-84, 2.9/km2 Poor forage prod. Med. forage prod. Good forage prod. (1977,80,84) (1976,81,83) (1978,79,82) With fawns-poor forage prod. With fawns-good forage prod. 1-year-old females 2-year-old females 3-year-old females 4+year-old females 23 2.34 0.72 27 4.50 0.92 6 3.20 0.81 20 3.41 0.84 18 3.06 0.83 15 3.91 0.84 7 4.32 0.94 10 4.29 0.89 23 3.48 0.86 24 3.72 0.83 12 2.59 0.77 8 2.30 0.71 8 2.46 0.78 20 4.77 0.93 10 4.56 1.00 8 3.28 1.08 40 3.04 0.90 a number of adult females b km2 c km We tested the assumption that home range size and average movements should decrease as forage became more abundant (Table 8.2). Forage production as measured by clipped plots fell into 3 general levels; poor (1977, 1980, and 1984), medium (1976, 1981, and 1983), and good (1978, 1979, and 1982). Although home ranges generally were smaller during good forage years than during poor forage years, the difference was not statistically significant for PHR (K-W, X2=2.94, P=0.23, df=2) or AAR (K-W, X*=0.97, P=0.62, df=2). Females that had fawns during years of good forage production did not have smaller home ranges than those that had fawns during a poor forage production year (Table 8.2; PHR, Rank Sum=67.0, P=0.96; AAR, Rank Sum=76.5, P=0.40). Data in Table 8.2 indicated decreasing PHR size with increasing age, but no consistent relationship between AAR and 215 age. We believe that age and reproductive status were too autocorrelated (the chance that a female will have a fawn increases from age 1 to 5 , see Chapter 5) to enable us to determine any effects due entirely to age. Factors Affecting Summer Home Range Size of Fawns Home range size of fawns on this area from 1976-1980, as determined by the modified minimum area method (Harvey and Barbour 1965), was reported by Riley and Dood (1984). They found that fawn summer home range size decreased with increased population size during 1976-1980. We found that mean home range size for fawns increased in ascending order as follows: single females, twin females, single males, mixed sex twins, twin males (Table 8.3). Those differences were not statistically significant, however (Kruskal-Wallis; PHR, X2=1.39, P=0.85, df=4; AAR, X2=1.43, P=0.84, df=4). Combined, twins had larger home ranges than singles, but the significance level was marginal (Mann-Whitney; PHR, Rank Sum=1361.0, P=0.06; AAR, Rank Sum-1352.0, P=0.08). Mean home range size was larger for all males than for all females, but the difference was not statistically significant (Mann-Whitney; PHR, Rank Sum=874.5, P=0.43; AAR, Rank Sum=875.0, P=0.43). Although the general tendency for differences in movements by sex and litter size appeared to be as expected, the extreme variation in values (Table 8.3) precluded statistical significance. Home range size of fawns was compared among the same 4 density classes delineated for adult females, and a significant difference in home range size among density classes was observed (Kruskal-Wallis; PHR, X2=13.77, P<0.01; AAR, X2=9.27, P=0.03). Summer home range size generally declined as deer density increased. There may have been effects related to variation in sex and litter size among years, but sample sizes were too small to conduct tests which would eliminate those effects. The comparison of summer home range size for fawns eliminated the effect of reproductive status on home range size that we observed for adult females. Although total summer home range size of adult females did not decline with increasing density, it appeared that the size of the fawn- rearing areas or "parturition territory" did. There was no evidence that good forage production reduced the movements or summer home range size of fawns. Home range sizes were actually largest during 1978 and 1979 when forage production was best. Sample size was not sufficient to partition the effect of density and forage abundance. 216 Table 8.3. Summer home range size and mobility of fawns 1976-1984, Missouri River Breaks study area, Montana. Year Mean + SE PHRU AARC Range PHR AAR 1976 1977 1978 1979 1980 1981 1982 1983 1984 Total 4 2.3+0, ,7 10 2.3+0, ,4 9 4.8+0, ,8 11 2.5+0, ,5 8 1.9+0 .3 14 1.9+0 .5 11 1.3+0, .3 7 1.8+0 .3 9 1 . 8+0 .4 83 2.3+0, .2 0.73+0.07 0.71+0.06 0.85+0.07 0.66+0.05 0.73+0.13 0.61±0.08 0.60+0.08 0.53+0.03 0.77+0.14 0.68+0.03 (0.53-3.76) (1.12-5.19) (1.03-9.29) (0.91-5.95) (0.55-3.07) (0.24-6.12) (0.24-3.41) (0.84-2.98) (0.44-3.97) (0.58-0.93) (0.50-1.10) (0.46-1.18) (0.43-0.92) (0.30-1.10) (0.25-1.24) (0.22-1.04) (0.44-0.63) (0.32-1.67) (0.24-9.29) (0.22-1.67) Mean PHR AAR Range PHR AAR Single female 24 1.88 0 .61 (0, ,24-6, .12) (0, ,22-1, .07) Twin females 6 1.90 0 ,64 (0, ,74-3 .16) (0, .41-0, .82) Single male 22 2.06 0 .68 (0, .24-4 .33) (0, .25-1, .67) Mixed sex Twins 8 2.23 0, ,69 (0, ,44-5 .94) (0 .32-1, .24) Twin males 9 2.61 0, .73 (0, .42-9, ,29) (0, .35-1 .44) All females 30 1.88 0 .62 (0, .24-6 .12) (0, .22-1, .07) All males 31 2.22 0, .69 (0, ,24-9 .29) (0 .25-1 .67) All singles 46 1.97 0, .64 (0, .24-6. .12) (0 .22-1 .67) All twinsd 36 2.66 0, .74 (0, ,42-9, ,29) (0 .32-1, .44) a number of fawns b km2 c km d Includes 13 sets of twins for which only 1 member was captured and sexed. 217 Monthly Patterns of Movements Monthly AAR were calculated for all individuals with 2 or more relocations per month (range 2-13). AAR were partitioned by sex and by age in months and then pooled within those categories to form a population mean AAR for that sex and age. Because of the wide variation in values, monthly AAR of females were statistically different only for the most extreme differences in values (e.g. June at 13 months of age vs. June, July, and August for 24+ month old females). Relatively large AAR were recorded from 12 to 14 months of age (Fig. 8.1), when the strong social bond established between mother and young was broken as she prepared for and gave birth to a new fawn. This period was typified by wandering, and occasionally dispersal, of the yearling female. The smallest monthly AAR for 24+ month-old females coincided with parturition and fawn- rearing during June, July, and August. One other trend observed for established, reproducing females (movements of fawns also show this) was a seasonal peak in AAR during March, coinciding with movement to summer range by migratory deer. Movements to winter range did not appear as a sharp one month peak because they may occur throughout the period September-February and individuals may move between seasonal ranges several times during the period. The wide variation of individual values resulted in few significant differences between monthly AAR of males, but 2 significant peaks in AAR were apparent. The first occurred during May, June, and July as the male fawns reached 1 year of age (Fig. 8.2). A high proportion of yearling males dispersed from their natal home range at this time, accounting for an even greater degree of mobility of yearling males compared to yearling females. The second peak in movements was during the rut in November when the AAR for males was about 3 times that for females. Average activity radii may be larger for males than females during most months, but sample sizes were too small to verify that. Emigration Emigration rates of marked yearling males (Table 7.7) were not related to population density. Total numbers of yearling males in the fall population did not change much from that expected, based on recruitment, despite average emigration of 51% of all yearling males from the study area by 14 months of age. Thus, for yearling males, emigration was apparently balanced by immigration. 218 1.4 -i 1.2 1.0 ? ;* 0.8 < 0.6- < 0.4 0.2 0.0 4 5 I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 I I I 1 1 I I I I I I I I I I I I I I JJASONDJFMAMJJASONDJFMAMJJASONDJFMAMJJASONDJFMAM MONTH Birth to 36 Months- ». ^ All Females ^ 37 months + Figure 8.1 Monthly average activity radius for females, birth to 36 months, and then by month for all females 37 months and older. Bars represent the standard error of the mean and numbers represent the sample size per month. E DC < < 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 o.o- 40 41 36 30 o 12 » r I I 1 1 t 1" " !■ I 1 I JJASONDJFMAMJJASONDJFMAM Birth 14 mo. All Males 15 months + Figure 8.2 Monthly average activity radius for males, birth to 14 months, and then by month for all males 15 months and older. Bars represent the standard error of the mean and numbers represent the sample size per month. 219 Most dispersal of marked yearling females fit the description of pre-saturation dispersal (Lidicker 1978). Seven of 9 dispersals by marked yearling females occurred during the first 2 years of population increase. Only 2 radio- collared yearling females occurred among the 1986 cohort; both emigrated from the study area. Population estimates indicated that up to 49% of the 1986 female cohort emigrated during summer 1987 without an offsetting immigration. Dispersal during other years was low or non-existent. Contrary to what was observed for yearling males, emigration of yearling females did not appear to be balanced by immigration during any year, although some immigration may have occurred during autumn 1983. The average date of dispersal was earlier for yearling males than for yearling females (Fig. 8.3). Some yearling males dispersed before their mother was actively aggressive towards them, while few yearling females dispersed until their mothers aggressively chased them. Yearling females stayed in the same area as their mothers for some time, and active aggression on the part of the mother was necessary to break the close social bond that characterized doe-fawn relationships during the first year. Yearling males appeared to be inherently inclined toward dispersal but yearling females did not. c "55 a> Q. 0) 0 Q 60 "I 50" 40- 30- m- 20- 10" Males Females ill Mid Late Early Mid Late Early May May June June June July Late July Figure 8.3 Date of dispersal for yearling mule deer, 220 Dispersal distance averaged 25.7 km and ranged from 7 to 83 km for 21 yearling males (Fig. 8.4). For 7 yearling females, mean dispersal distance was 36.9 km, and the range was 5 to 69 km (Fig. 8.5). The direction of dispersal for yearling females was random (Rayleigh's Z=0.29, P>0.50), but that for yearling males was non-random (Rayleigh's Z=8.48, P<0.001). Dispersal by yearling males was bimodal; most males dispersed in westerly and easterly directions, staying within "river breaks" habitat. Individual Strategies To further understand factors affecting movements and home range size and shape, plots of home range size, shape, and intensity of use for individuals were instructive. Closely overlapping home range polygons for 5 consecutive years (age 5-9) of resident female 2679 are displayed in Fig. 8.6A. There was relatively little annual variation in home range size, shape, or pattern of use, but annual PHR size ranged from 2.7-4.8 km2. Intensity of use of the home range area was not uniform (Fig. 8.6B). A 1.7 km2 area in the southern portion of the home range (24% of 7.1 km2 life home range) contained 50.3% of all relocations of this female. The northern portion of her home range contained more steep, south-facing slopes than the southern portion and although the northern area was used yearlong to some extent, it was consistently used only during periods of winter when snow was deep. Statistical methods of home range analysis often reduce or overlook the importance of these less intensively used areas that are vital to the survival of deer. These "auxiliary winter ranges" were usually poor in forage quantity and quality and were less intensively used yearlong, but provided some forage, stable footing, and warm microsites during periods of cold temperature and deep snow. Life PHR for migratory female 1577 (age 3-4 and 6-10 years) was 19.9 km2 (Fig. 8.7). Annual PHR varied from 4 . 6 to 16.6 km2; the smallest and largest annual PHR occurred during years with the mildest and most severe winters, respectively. Female 1577 (Fig. 8.7) spent most of the spring-summer period in the most southern portion of her home range, but during autumn, a majority of her time was spent in an area about 3 km north of her "summer range". Most winter relocations were in or adjacent to the northern portion of the autumn area of use (Fig. 8.7). During periods of very deep snow, she moved to auxiliary wintering areas of very steep terrain north of her normal winter range. 221 E o CO o CM 03 CD r~\ i M o 4-1 (0 c Q) n c C ta 2 0 c c O M •-H 4-> id u CD X M +J •H H Q s 00 0 U 3 •H 222 to H 1 l-l 0 o CM ITS Q) l-i 13 03 +J (0 c CD ■p e o "0 (0 8.8, g to U c c •H 13 0 ♦J Q-, UJ >H XI 'U c Q) T3 C c 03 s 0 C c 0 X •p •p 03 u 0) X M +J •H •H Q ? in CO 0) H 3 tfl 223 A. B. N 1 2 3 4 5 6 Number of Relocations v— / Wi Winter locations 2 km Figure 8 . 6 A. Annual home range perimeters of resident female 2679 for 5 consecutive years (age 5-9). B. Annual home range perimeters and intensity of use plot for female 2679. 224 N ^— v J Auxiliary Winter Range "%/VV Normal Winter Range Autumn Range Normal Spring-Summer Range Life PHR 2 km Smallest Annual PHR • 1 • 2 Number • 3 of £ 4 Relocations • 5 Figure 8.7 Life home range perimeter, intensity of use plot, and seasonal ranges of migratory female 1577. 225 Seasonal use was not exclusive to an area, even for migratory deer. Rather, a preponderance of seasonal use was within an area. For example, 79% of spring-summer observations of female 1577 were in the southern portion of her home range and 81% of autumn-winter observations were in the northern portion of her home range. Half of the 21% of spring-summer relocations that were on "autumn-winter range" occurred during the dry summer of 1980 when she apparently made use of the more abundant succulent forage still available on steep, north-facing slopes not grazed by cattle on the northern area. The intensity of use pattern for this migratory deer (Fig. 8.7) was even less uniform than that for a resident deer (Fig. 8.6B). A total of 29.1% of all relocations were within a 1.4 km2 area of autumn-winter range and 39.6% of all relocations were within a 1.2 km2 area of spring-summer range. Combined, 68.7% of all relocations of female 1577 during 5 years were within an area that comprised 13.3% (2.6 km2) of the life PHR. Although not intensively used over the 5 year period, other areas were important to the survival of this deer. Auxiliary winter range areas in the northern portion of her home range were used during periods of deep snow. Those areas included the most northern portion of her PHR, used only during the extremely severe winter of 1977-78. Areas not intensively used by this deer were probably important in that they represented the process of finding alternate winter areas and areas where forage remained succulent longest. Almost all deer occasionally displayed unusual or unexplainable temporary movements outside of normal home ranges. Some movements may have been related to undocumented disturbances, but also may have been random exploratory movements that resulted in familiarity with alternative areas of use. We also may have classified some movements of deer as unusual because we did not know their complete history. For example, a female captured and marked as an adult may have dispersed to the area of capture as a yearling, but made occasional trips back to her natal area. A variation of the yearlong resident pattern also was evident for some deer (Fig. 8.8). During 3 years, female 0476 displayed a pattern typical of residents and her annual PHR varied from 1.7-4.8 km. She occurred within the same area during 2 other years, but moved to auxiliary winter areas during 2 very severe winters (points B, winter 1977-78 and points C, winter 1978-79; Fig. 8.8). During most years and snow conditions, her normal winter area was at the northeast corner of her home range. During 1977-78, she did not move to auxiliary winter range until early February, after 2 months of deep snow. Subseguent snowstorms resulted in her movement 226 Number of relocations 2 km Figure 8.8. Annual home range perimeters (5 years), intensity of use plot, normal auxiliary winter locations (B, 79) for female 0476. winter range, and 1977-78; C, 1978- 227 across a major drainage (Sand Creek) to a large complex of south- and north-facing slopes. She wandered throughout that complex during the remaining month of winter. During winter 1978-79, also very severe, she did not leave her normal winter area until late January, but when she did move, she immediately went to the last area she had used the previous winter and remained localized there for the remainder of winter. During most years, female 0476 occupied a yearlong home range, but when snow depth exceeded 60 cm during 1977-78 and 1978-79, her normal winter range was not adequate. Snowdrifts from unusual north and east winds often covered the smaller south-facing slopes of her normal winter range. She finally moved to an area that contained south-facing slopes with twice the elevational gradient of those on her normal wintering area. South-facing slopes on the auxiliary winter area bordered on a large open drainage bottom, with the closest major ridge to block the sun during early and late portions of the day more than 1.5 km away. On her normal winter area, smaller south-facing slopes were bordered by ridges on the south less than 200 m away, effectively blocking sun from much of the south-facing slope except during mid-day. The pattern of movements exhibited by this female during the severe winters of 1977-78 and 1978-79 indicated that she did not have an auxiliary winter range or that its location was a distant memory (she was a yearling during the previous severe winter of 1971-72). Her movements during winter 1977-78 indicated that she moved only when forced by extreme circumstances and then she did not go directly to an auxiliary winter range, but progressed through a series of more distant sites, apparently as the result of wandering movements. During winter 1978-79, however, once conditions became severe, she immediately moved to the best auxiliary winter range site she had encountered during her wanderings the previous winter. Other resident deer had similar patterns of movements during those severe winters. The speed and directness with which they moved to auxiliary winter range sites varied and may have depended upon their previous experience or that of the matriarch in their social group. Female 0476, as the others profiled, spent most of her time in a much smaller area than indicated by PHR size (Fig. 8.8). A total of 64% of the relocations of female 0476 were within an area that was 4.2% of her life PHR and 13% of her life PHR that excluded relocations during severe winters. This deer did not need an area of 21.9 km2 (total life PHR) to survive, but during some winters she did need to use areas 5 km from her normal home range. 228 The pattern of home range formation for resident female 3580 from birth until age 4 is shown in Fig. 8.9 (annual PHR 1-4). Annual PHR size varied from 3.1 to 5.4 km2. The AAR was smallest during the year she was a fawn (0.5 km), largest when she was a yearling (1.0 km), and intermediate during the years she was 2 to 4-years-old. Although exploratory movements took place and some important areas of use expanded slightly northeast, 70% of all relocations after 12 months of age were within the boundaries of her first year home range. Fifty-six percent of all relocations during the 4 years were within an intensively used area of 0.5 km2 that was 4% of her life PHR and 14.5% of her smallest annual PHR. Except during periods of deep snow when she moved to an area with larger south- facing slopes (Fig. 8.9), this female used the same areas during winter as during the rest of the year. For female mule deer on the study area, the majority of which do not disperse from ancestral home ranges as yearlings, the area in which they spend their first year becomes a major portion of their adult home range (Fig. 8.9). During fawn-rearing periods, however, mother and daughter seldom use the same location at the same time (Fig. 7.12). Female fawns of migratory females became migratory as well, and used the same general summer and winter areas as their mothers; though the timing of movements between areas was often independent. The general home range size, shape, and pattern of use appeared to be the result of learning and tradition, but daily movements were most often independent, especially from May through early autumn. Although the literature indicates that broad differences in habitat can alter home range characteristics, it is difficult to closely relate home range characteristics to habitat within one study area. We hypothesized that deer living in areas with large amounts of open area between islands of cover might have larger home ranges than those deer living in areas that included a lot of interconnected cover. If there were differences in home range size within our study area owing to varying habitat components, our gross analysis did not detect it. Nineteen summer home ranges were calculated for 9 resident adult females whose home range included or bordered on large amounts of sagebrush-grassland or pine-grassland cover types. Mean PHR for those deer was 3.11 km2 and mean AAR was 0.86 km. Mean PHR for all resident adult females was 3.44 km2 and mean AAR was 0.83 km. As discussed earlier, habitat probably largely determined whether deer were migratory or resident, and thus affected home range size. Overlapping home ranges of 3 resident and 4 229 i-& Relocations 2 km fi} Year of the Home Range Auxiliary Winter Range Figure 8.9 Annual home range perimeters of female 3580 from birth to age four. Intensity of use plot and auxiliary winter range location are included. 230 migratory females helped explain how habitat affected home range (Fig. 8.10). Polygon home range size for 3 resident females was much smaller than that of 4 migratory deer. There was no acceptable winter range adjacent to the spring-summer ranges of the 4 migratory deer, and they moved 4-6 km to other areas during autumn and winter. Three of the 4 migratory deer used the same areas during winter and the other used a different winter area (Fig. 8.10). Winter areas were less than 1 km from the center of the PHR of resident females. We believe that "parturition territoriality" (Ozoga et al . 1982a) and the limited availability of quality wintering sites led to the establishment of the migratory movement pattern. Because not all deer that used the area included in Figure 8.10 were marked, total home range overlap could not be mapped. It is likely, however, that territoriality during parturition and the early post-partum period limited the number of females that could establish residence near areas suitable as winter range. As the population increased, parturition territories were established at increasing distances from high quality winter range. Females establishing fawn-rearing territories at some distance from the winter range probably were fawns of females resident to winter range areas. After fawn rearing territories became less rigid during late summer and autumn, the younger females may have regrouped with their mothers and continued to use the same winter areas. As time passed and the older females died, the younger females continued to use separate summer and winter areas because of tradition and necessity rather than because of social ties. During mild winters, it was possible for deer to remain in the spring-summer range of females A, C, D, and E (Fig. 8.10). During winters with deep snow, however, it is unlikely that many deer, especially fawns, remaining in those spring-summer areas would survive. Over the long term, females that have spring-summer home ranges in areas of low relief near the prairie are unlikely to recruit many fawns (genes) or survive themselves if they do not maintain a migratory tradition. The resident pattern is probably the natural pattern because it is less costly in terms of energy expenditure and exposure to hazards. However, when an area is near maximum fill, at high population levels, a migratory movement pattern must develop. Some deer establish a resident pattern in low relief habitat, especially during long, mild weather cycles, but they are those most likely to die during severe winters. 231 2 km a N Resident females B,F,& G Migratory females A.C.D, & E ^A^N Normal Winter ^Ciy Range Spring -Summer Range A.C.D.E Auxiliary & Normal Winter Range E O Auxiliary Winter Range Figure 8.10 Life home range perimeters, summer, winter, and auxiliary winter range locations for three resident and four migratory females. 232 Some deer moved to new areas of use during autumn and remained near there during winter (Fig. 8.7). We believe that movements of autumn migrants may have been at least partially influenced by cattle grazing. The areas of low relief and more open cover types that those deer moved from were generally intensively used by cattle. The areas they moved to, while providing steep south-facing slopes and thermal cover on timbered north-facing slopes during winter, also provided advantages during autumn. The steeper terrain was less grazed by cattle and forage remained succulent longer on the more extensive north-facing slopes. As a special case, the autumn and winter ranges of females A, C, D, and E (Fig. 8.10), were within a "blind pocket" in the fencing system and received exceptionally light use by cattle. Coincidentally, the area they used during spring-summer may have been improved as spring range because of heavy use by cattle during summer and autumn. The lack of litter in those areas made early spring growth of grasses and forbs more available to deer. Discussion We identified 2 general types of movement patterns for adult females; resident and migratory. Within that framework, however, home range size and movement patterns were extremely variable, not only among deer and among seasons and years, but for the same deer among seasons and years. Much of this variation reflected seasonal and annual adaptation to fluctuations in environmental conditions. The form of these adaptations or adjustments depended upon the topographic, climatic, vegetational and land use characteristics of individual home ranges. Males generally had larger home ranges than females. Home range during summer was smaller for females with fawns than for those without fawns. Home range size of adult females was not affected by deer density or forage abundance, but summer home range size for fawns and thus parturition territory of adult females was smaller at higher densities. A home range develops as the result of tradition and cumulative movements to satisfy daily requirements. Any successful home range strategy must result in procurement of enough quality forage, warm microsites during winter, escape terrain from predators, hiding and thermal cover, and other amenities to enable the deer to survive and successfully recruit young. Generally, the smallest area to supply all those amenities should be the most efficient home range size. Initially, the matrilineal social structure of mule deer in this area provides the basis for home range formation by females. The mother aggressively breaks the social bond with her yearling daughter just prior to the birth of her new fawn(s). During most years, few yearling females disperse 233 from their natal range and although their mother does not tolerate their close presence during summer as long as her new fawn(s) survive, the yearling daughter maintains an overlapping home range with her mother. The yearling daughter does not use the same spot at the same time as her mother during this period but based on the learning and tradition established during her first year of life, continues to generally use the same home range as her mother. The yearling summer may also be characterized by exploratory movements beyond the boundaries of the mother's home range. Although those movements may eventually lead to small shifts in intensity of use patterns, colonization of new sites is often unsuccessful because the yearling encounters the parturition territories of other females and is chased from those. During autumn, or earlier if the mother's new fawns die, the yearling usually rejoins the mother and a matrilineal group is formed. Apparently, the resources within any home range are sufficient to support more than 1 female and the formation of matrilineal groups results in sharing of resources by relatives rather than unrelated deer. Any possible diminution of resources for the matriarch and her current fawns must be more than balanced by genetic advantages of kin-selection apparent in sharing with older daughters and granddaughters rather than allowing unrelated females to preempt portions of her home range. The daughter's chances for reproductive success are enhanced by her use of known resources on a successfully established home range. Dispersal, especially during periods of mid-high population density, could only result in establishment of a home range in marginal areas not yet used by successful females. Resources are marginal on these areas and their location must be learned by trial and error, exposing the disperser to greater peril from the weather, predators, and accidents. Those considerations may explain why most dispersal we observed by yearling females was at pre-saturation densities, when relatively better habitat was still unfilled. Once a home range is established, daily and seasonal movements within that area appear to be governed by actions that enhance the comfort and survival chances of the deer. A question to be answered by an examination of movements is: What types of areas best provide for the comfort and survival of deer? Although detailed analysis related to that question was not done, the available data did result in some obvious conclusions. Movements of deer during late summer and autumn indicated that the changing availability of high quality, succulent forage motivated those movements. The deer that shifted their areas of use during late summer and autumn most often moved from areas of low topographic diversity to areas that contained more topographic diversity and a wider variety of microsites. Broad north-facing slopes and the attendant 234 smaller drainage bottoms provided succulent forage longer than did other areas. Forbs and shrubs under the Douglas-fir canopy on most north-facing slopes not only were subject to less evapotranspiration stress during the heat of summer, but were protected from the earliest killing frosts of autumn by a temperature-moderating effect of forest canopy. Those deer living adjacent to the Missouri River bottom made more use of the river riparian type during autumn for similar reasons. During winter, forage was less of a driving force in determining movements and habitat use. In the northern United States and on this area, most of the forage available to deer during winter is of only maintenance guality or poorer (Bucsis 1974, Short et al . 1974, Mautz et al . 1976, Wallmo et al. 1977, Mautz 1978). Plant species making up the bulk of the winter diet on this area (big sagebrush and Rocky Mountain juniper) are abundant and ubiguitous . During mild or "normal" winters, deer can find adeguate guantities of most winter forage almost everywhere. However, rabbitbrush, a preferred species, was primarily available at the edge of ridges in the ecotones between sagebrush and pine and Douglas fir types and when snow depth was greater than about 4 5 cm, most rabbitbrush was unavailable. Although snow depth was reduced on the south- and north-facing slopes deer moved to during severe winters, forage availability did not appear to be the factor governing those movements. The south-facing slopes on the severe winter areas were usually either the pine- juniper-shale or greasewood and shale-longleaf sage vegetation types. Those types were the poorest of all types in forage guantity and guality (Tables 3.2 and 3.3). Use of the auxiliary winter areas provided deer with ease of movement and escape from predators, warm sites on sunny days on the south-facing slopes and relatively warmer sites on the timbered north-facing slopes at night and during cold, windy days. During spring, deer made the greatest use of the open sagebrush-grassland type and least use of the dense Douglas fir types. Such use was related to the greater availability of new green forbs and grasses in the open types. It reguired movement of only 100-200 m to open areas rather than major movements to entirely new areas of use. The home range of males should also contain adeguate resources for survival, but the reguirements of males differ from those of females (Verme 1988) . Our data indicated that males occupied lower guality habitat than productive females did, at least during summer and early autumn (Chapter 9). Data on social structure, distribution, movements, and habitat use indicated that males, especially during the fawn-rearing period, spend most of their time on sites not occupied by 235 productive females. Females with fawns aggressively chased males, as well as other adults, during the fawn-rearing period and most mature males were distributed in aggregations away from productive females during summer and early autumn. Additionally, data from marked deer indicated that most mature males and females were not associated during winter and spring. The relatively larger home range size of males compared to females is probably explained by 2 factors; breeding strategy and movement during the rest of the year among widely separated patches of habitat not occupied by many productive females. The greatest portion of the explanation for larger home range size of males may be related to breeding strategy. Because males do not face lactation stress during summer, their nutritional requirements during summer are probably less than that for productive females. They also do not require habitat with adequate hiding cover from predators that females protecting young fawns require. Rather than concentrate resources on raising a single litter of fawns per year, the reproductive strategy of the mature male is to have access to and breed as many females as possible each breeding season. A large home range results in access to more females (Fig. 8.11) . Many other unmarked males and females used the area presented in Figure 8.11, so that plot does not represent the home range of one male and all of his exclusively available females. The data do illustrate the relative degree of access of the male to different females compared to the potential for female interaction with each other. The home range of the male encompassed at least 3 subpopulations of females that had little or no interaction with each other. Familiarity with a large area provided the male with maximum potential breeding opportunities. Relocations made during the breeding season (Fig. 8.11) indicated that he moved throughout his entire home range during that period. Home range size, shape, and pattern of use for both females and males evolved as that which when used, resulted in the greatest number of recruited offspring. Females probably use the smallest home range that supplies all their needs. The matrilineal social structure, parturition territoriality, and changes in deer density all interact to determine whether a particular female can establish a home range that is optimal or marginal in its amenities. Males attempt to have access to many females during the breeding season, while using areas where they do not compete with their offspring for forage or space during the rest of the year. 236 2 km *»««»»»»»»»*« home range boundary - mature male home range boundaries - adult females # some locations of the male during the rut Figure 8.11 Relative distribution of home range perimeters for a mature male and eleven adult females. 237 Although often considered a species/population parameter, our data indicate home range size and shape as well as movement patterns are influenced to a large degree by individuality. They reflect an individuals attempt to effectively utilize the fixed habitat base in an area and survive under prevailing environmental conditions and fluctuations . 238 CHAPTER 9 HABITAT USE AND INTERSPECIFIC RELATIONS Food Habits Food habits of mule deer on and adjacent to the study area were emphasized in earlier work by Mackie (1970), Knowles (1975), and Komberec (1976). During 1976-1986, we obtained additional information by collecting rumen samples from hunter-killed deer during late October through November each year, as well as from coyote-killed deer, trapping casualties, road-killed deer, or deer deliberately collected for study. The following discussion generally summarizes our knowledge of food habits from both the earlier published reports and the additional information. On an annual basis, shrubs comprised 62%, forbs 33%, and grasses 5% of the mule deer diet during 1960-1964. Shrubs comprised 50% or more of the diet during August through March, and averaged 36% of the diet during May through July, their period of lowest use (Figure 9.1). Peak use of shrubs occurred during December and January (90%+). Forbs comprised one-third or more of the diet during April through September. Forb use was highest during May, June, and July, when it comprised 60-70% of the diet of mule deer. Lowest use of forbs occurred in December and January, coinciding with the greatest use of shrubs. Use of grasses was relatively minor, but peaks occurred during October and November of some years and late March through April in all years. Seasonal Patterns Summer Forbs comprised 45-75% of the summer diet of mule deer, averaging slightly over 50% (Mackie 1970, Knowles 1975). Yellow sweetclover was the major forb species used, comprising 36-59% of the diet. A variety of other forbs were also eaten, usually in minor amounts. The shrub, fragrant sumac, was the second-most important forage species during summer, comprising 10-36% of the diet. Both Mackie (1970) and Knowles (1975) noted that use of fragrant sumac increased as the abundance and succulence of yellow sweetclover decreased. During summer, use of fragrant sumac was primarily confined to leaves and buds. Other shrubs, including snowberry, rose, and chokecherry received increasing use as summer progressed. Grasses received minor, incidental use during summer. When sweetclover was scarce or absent, use of other forbs increased to a small degree, but most of the lack of yellow sweetclover was made up by increased use of fragrant sumac, snowberry, rose, and chokecherry. 239 b 100 90 80 70 60 50 40 30 20 10 0 Shrubs Forbs Grass Aug Sep Nov Dec Month r Feb Apr May Figure 9 . 1 Autumn Yearlong trend in mule deer use of browse, forbs, and grasses in the Missouri River Breaks, Montana, 1960-1964. General comments about forage use during autumn from Mackie (1970) and Knowles (1975) were supplemented by data from late autumn for 1976-1986 (Table 9.1). The species of forage used was highly variable during autumn, but use of shrubs generally exceeded use of forbs. Exceptions occurred when first-year growth of yellow sweetclover was available and remained succulent into autumn (1978, Table 9.1). Sweetclover was important as long as it remained green, even into winter. Use of grasses during autumn was also variable among years and considerable use of Sandberg bluegrass was recorded during autumn 1977, 1985, and 1986 (30-48% of diet). This increased use of grasses resulted from autumn "green-ups" that occurred after substantial late summer-early autumn rains . Autumn use of green grasses also occurred during 1961 and 1962 (Mackie 1970). Relatively heavy use of grasses (14%) also occurred during very dry autumns (1980 and 1984). The majority of that use was of dry grass, apparently eaten incidentally in attempts to eat the few remaining green leaves . The major forage species used during autumn of most years was rabbitbrush, primarily rubber rabbitbrush, but also green rabbitbrush. Use of rabbitbrush was seldom great prior to about the first part of October, when flowering was finished (Knowles 1975). Rabbitbrush not the most important forage species only during autumns when sweetclover remained green 240 $ •0 oo o s o r- ir» t> O o ro \ir> LT> \K> in s, s. -O \ o o T— o O o o o CO oo o woo ? no o ro \ f~l L. Kl no n- no *^ 8 O O O O o o O T- «— T— -J s\s S.V. 00 r%i ■a CO Kl •d* ro <— no *— s o «- r- oo o -o* oo oo \»- \i^ (\J >o \ O \C0 W \no •O ro no O no OJ f\l O o o o o «- K» CO O O O CO \ WO (M ~» \ r- i- >» W \0 >» 4-> «- r- «- k> »- 8 o O O O 00 O CO 00 \ v» -s, \ro c 8 8 $ 00 0> -o r^ -O o t_ r\J *_ N. ,_ K1 CO \o ^ \ \no w •" M h- ■or \o o -O \s ro W *» O ^ ro 00 o o O O o o r^ o o o O o o 00 O N \r- r*. l/> \~» w\,» \ o \>s w 8 \o K Li \fM *~ m ^ ** *~ 0> o o 8 88 S o o s 35 s r— O O •o s •-- \ \\ r- \ V, L. s T— r^ N- Og \ \00 (M *-- no «" r- 00 K1 <— in M no o o o r- o t/i t/"> u~< i/\ in W\ W irv *- -j- c- O O O- O Kl r (\Jr -Jt- ^f> \ WX3 INi O V. B LN \\ \(\l CO *J ~* m *- l\J K> -c CD 3 O o 00 o «- t- »- irv o k» o ki «- Kl o r~ *o O O \K> \>0 \C0 S. «- \~» \«- \no CO CO no K» t- ^» no Or n o o o o ro c\j no S.SS o o O r^ ro ro 8 o o o OOO fr f> \ \-» ir> "S.V. >0 ro ro «- «- ro 00 r» o K» Kl K> O I O i o o Kt tn o «- no CO *- \ \C0 V. 3 V — I 3 ^ lis N- tr> \ ro ro f\j rj it* u-» 8 8 «- oo *o f\I ■ \ \Kt CO \ O O \ \r^ O <\J f\J t- ro O Q O O i/MTi OO «- f\J «M «- «- lA *J W fNJ t- ? o- c I/) OJ 01 01 ■ E *J 3 ■ L. a *~> Cl C a c a i/i o ui 3 T3 a ■ yn Ul 0 u a C 3 u C LO L. E ra ra a ta l^ E . ■•■ c> r« Q. — ' (/) (- C D tf» U « c S° I 0J v t_ E •«- 3 • I. M 01 »- O \ 0 « « 1. ■ L. «-> 3 01 -> 03 i O e > r i- c *< »« 241 and/or a "green-up" of grasses occurred. Snowberry ranked second in both use and occurrence among species used during autumn. Its use was especially heavy during good fruit- bearing years when the berries were probably a source of high energy. Rose, while not usually a major food item, was used consistently and was also most heavily used when rose hips, possibly high in energy (Welch and Andrus 1977), were abundant. Use of fragrant sumac was usually negligible after leaf drop in September. Autumn use of Rocky Mountain juniper and big sagebrush was relatively minor, but was more important when sweetclover or green grasses were not available. Winter More than 80% of the diet during winter consisted of shrubs during most years. Use of forbs was generally low except during periods of deep snow, when desiccated forbs, protruding above the snow, were eaten, and during warm, snow-free periods, when some forbs "greened up" on south-facing slopes. Grasses were generally used only during late March as new growth started. Forage use during winter usually depended upon 2 major factors: snow depth and forage use during autumn. Big sagebrush and Rocky Mountain juniper were major forage species during winter in either case. Rabbitbrush was preferred and heavily used whenever it was available, but deep snow reduced both availability and use. Essentially, all current annual growth of rabbitbrush was utilized each year. When green forage was available during autumn and use of rabbitbrush was reduced, much of the use of rabbitbrush was delayed until early winter, when it received heavy use. Use of rabbitbrush then declined through winter as use reduced availability. Rabbitbrush may be depleted earlier in winters with high deer populations than during winters with low populations . Approximately 28% of the diet was rabbitbrush during February 1977 (470 deer), 20% during February 1980 and 1981 (1020-1030 deer), and only a trace during February 1984 (1545 deer). However, the latter periods also coincided with dry conditions, leading to heavy autumn use of rabbitbrush. There was more than 30 cm of snow on the ground for 97 and 112 days and more than 45 cm of snow on the ground for 85 and 96 days during winters 1977-78 and 1978-79, respectively. Winter 1981-82 was less severe than either 1977-78 or 1978-79, nevertheless, snow depth was 23 cm or more for 58 days. Rabbitbrush was essentially unavailable those winters and received little use regardless of population level (Table 9.2). Fragrant sumac was also largely unavailable and was used much less than in more open winters. Rocky mountain juniper and big sagebrush were heavily used and very heavy use 242 oo a\ r-t A OO 3 o u XI 4-1 r>» r> » .-i \o r- o\ H . £ u • u cd n) C r: cd »J ^ C H o m S cd cu • CO x: M 00 ti 3 0) o M M pg x: 4-> u QJ M > QJ •H X3 oi a QJ ■H CJ M QJ 3 Q O co Qj CO * s » CO M •H dJ CO QJ ^ T3 .— 1 m Hi C »H to 3 s c QJ >^ a XJ 3 m ■a oi ^ co xj 3 T3 co QJ OJ C ■H ■H u E ai U p, QJ en 4-> 0) TD -a O O co CM OJ r-H X) cfl E- 14-1 O o c to co u 01 QJ H 4-1 4-> 0 3 O CO c M CO OJ 4-1 P. c 01 •H OJ ^ ■a TO N 00 H ao CTv CD I O OO a. o oo i o-> oo 0> oo I i CO Oi CM CM 0> c/> QJ •H u QJ P. CO CO 00 o -^ m o \ cm a\ o o 00 O CM \ lO \ U0 "-- w .-I CM 4-1 o CM ■D <3\ 00 4J CO •H id •-{ •H CO 'H C o o 4-1 -H ■H U OO -H C 4-1 O 4-1 .-H O cd en •H 3 C0 4-1 -H O a -h CO 4-1 .-I XJ M Ij l-i < s o b o CM CM O 00 00 iH m o\ to rl N H CM oo CO CO <3\ Oi >* CT> r~ O M .H M M. IT) CTi \ I-- CM \ i-H CO CM CM VO U0 O m m m m CM o r^ r^ r~ r~ m ~-~^ H ^•^ ^^ -~-~ M CM CT **s. •j- in VO P~ ~~v H r». l~^ rH CO rH rH o o o 0 O P^ O lO H H lO H CA CO 01 1C lO H MW CTl CM CM CM . O CO r-i o vD o r^ CD *■ — LO ■^ rH rH o ^^ « 0) m 3 a" QJ M CO *~s O cd oi m a U 3 rH n o > u 4J K( IS J3 CJ "O 243 was also made of ponderosa pine needles. Despite this, mortality, especially of adults, was relatively light during those winters and fawn mortality was less than during some milder winters. Overall, it appeared that rabbitbrush was the preferred winter forage and that use of big sagebrush and Rocky Mountain juniper increased only as a result of declining availability. Similar conclusions were made by Dusek (1975) for mule deer north of our study area. These conclusions may be consistent with Nudds (1980) hypothesis that deer are forage generalists when winter conditions are severe. Spring Use of forbs increased throughout spring and reached an annual peak during May. Shrub use usually declined by late March when "green-up" of new growth was initiated, though in some years "green-up" did not begin until early April. Use of grasses increased when new growth was initiated and reached an annual peak of around 20% of the diet in April, declining thereafter as forbs became more abundant. The major grass species used during spring was Sandberg bluegrass . Important forbs used during spring were bastard toad flax (Comandra umbel latum) , wild onion (Allium textile) , oyster plant (Tragopogon dubius) , wild parsley (Musineon divaricatum) , lomatium (Lomatium foeniculaceum) , and American vetch (Vicia americana ) . Discussion Green, succulent forage was preferred whenever it occurred. When sufficient late summer and/or autumn rainfall and relatively warm autumn temperatures resulted in maintenance of succulence of forbs or promoted regrowth of forbs and grasses, use of normal autumn browse species was delayed. These autumn browse species, especially rubber rabbitbrush, were then available for longer periods into winter. Heavy dependence by mule deer on Rocky Mountain juniper and big sagebrush was delayed further into winter. An autumn green-up prolonged the period of good to adequate nutrition not only through autumn, but into winter. Conversely, drought led to earlier maturation and desiccation of all herbage and in turn to heavy utilization of preferred browse, decreasing its availability during winter. Generally, a shorter winter season occurred on our study area than in most mountainous areas of the Rocky Mountains. Regrowth often began by late March although it was sometimes delayed until mid-April . During occasional winters of deep, continuous snow cover, much of the area was not inhabitable by deer. Many deer using 244 the southern portion of the study area migrated northward to areas of greater relief with steep south-facing slopes. Those areas were characteristically poor in forage resources, regardless of deer numbers. The most preferred winter forage, rubber rabbitbrush, was covered by deep snow and was usually absent on the steeper areas where deer are forced to winter. Paradoxically, effectively long winters (periods of negative energy balance), were often the result of hot dry springs, summers and autumns rather than deep snow and cold temperature during the calendar months of winter. Dry conditions during spring through autumn resulted in earlier than normal and heavy use of rubber rabbitbrush and snowberry which led to unseasonably early dependence on big sagebrush and Rocky Mountain juniper. The nutritional quality of the mule deer diet was influenced at least as much by the density-independent factor of climate as by the number of deer competing for forage. Weather certainly acted with more regularity than intraspecif ic competition on the nutritional quality of the deer ' s diet . General Patterns of Habitat Use Earlier studies (Mackie 1970, Knowles 1975, Komberec 1976) defined general patterns of habitat use by mule deer. Therefore, data obtained during 1976-1984 were primarily used to evaluate and refine previous conclusions about habitat selection and habitat factors influencing population ecology of deer in riverbreaks habitat. Aerial surveys ensuring complete coverage of the study area during July, September or October, December or January, and March or April provided the most complete data for analysis. Locations from individual surveys, recorded as the mid-point of 3.2 ha cells, were compiled seasonally and interpreted to represent habitat use and selection during summer, autumn, winter, and early spring. A "block analysis" (Porter and Church 1987, Wood 1987) was employed. This involved (1) compiling numbers of deer observed during aerial surveys within 28.8-ha blocks to determine relative seasonal density or intensity of use of each of the 953 blocks on the area and (2) relating those data to various habitat attributes measured within each block. The latter included the kind and amount of each vegetation cover type, the number of different cover types, the number of different patches of cover, a topographic relief index, and distance to free water during wet and dry years. Habitat parameters were measured from a vegetation cover type map developed from aerial photographs and ground-truthing 245 as overlays to orthophoto and topographic quadrangle maps of the study area. The kinds and amounts of vegetation cover types in each 28.8-ha block were measured by superimposing a dot grid of 25 regularly spaced points on each block and recording the type under each dot. Each cover type was assigned a value from 0 to 25 in each block, weighing its relative importance. The number of different cover patches was counted as the number of discrete units of all types in the block. The topographic index was determined by counting the number of contour lines intersected by 1 line drawn horizontally and 1 drawn vertically through the center of each block. Each block was then assigned to 1 of 4 TOPOINDEX categories: 1 = 0-69 m relief /km (X =_34 m/km) ; 2 = 70-149 m/km (X = 114 m/km); 3 = 150-230 m/km (X = 195 m/km); and 4 = 231-321 m/km (X = 275 m/km). Distance to water was measured from the center of the block to the nearest water source known to exist seasonally during wet and dry years within one of 5 distance classes: 1 = 0-0.402 km; 2 = 0.403 - 0.806 km; 3 = 0.806 - 1.61 km; 4 = 1.62 - 2.41 km; and 5 = > 2.41 km. Use and Selection of Vegetation Cover Types Mule deer used blocks containing forested cover types more than expected during all seasons (X2 = 129.9 to 550.4, 1 df, P < 0.005). Forest types comprised 51.5% of the study area and 63.4%, 61.4%, 66.9%, and 60.9% of blocks used by deer during summer, autumn, winter, and spring, respectively. Deer use of forested types was greater during winter than all other seasons (X2 = 8.41 to 36.7, 1 df, P < 0.005). Deer also used forested types proportionately more during summer than during spring (X2 = 4.49, 1 df, P = 0.04). Data concerning relative use of individual vegetation cover types by season (Table 9.3) are presented as relative risk-odds ratios (Everitt 1977), which were used as a preference index. Values greater than 1.00 indicated that more deer than expected (P < 0.05) used blocks containing a particular type. The higher the value, the more the apparent selection. Values less than 1.00 indicated that fewer deer than expected (P < 0.05) used blocks containing that type. Where no value is listed, deer use was not significantly different from that expected based on relative availability. Mule deer use of specific vegetation cover types differed seasonally (Table 9.3; X2 = 378.76, 60 df, P < 0.0001). Blocks used by deer during summer contained more burned Douglas fir- juniper type than expected. They also used blocks containing moderate and open density classes of the Douglas fir- juniper and pine- juniper-shale types, and the shale-longleaf sage and grassy bottom types more than expected. Conversely, blocks where the sagebrush-grassland, 246 Table 9.3. Relative risk ratios for seasonal mule deer use of vegetation cover types. Values > 1.00 indicate selection for that type, values < 1.00 indicate avoidance, blanks in Table indicate use not significantly different than expected. Vegetation Cover Type Year- Summer Autumn Winter Spring long Douglas f ir-Juniper-moderate Douglas f ir-Juniper-open Douglas f ir-Juniper-scattered Douglas f ir-Juniper-burned Pine -Juniper- Shale -mode rate Pine -Juniper -Shale -open Pine -Juniper- Shale -scattered Pine- Fir- Juniper- mode rate Pine -Fir- Juniper- open Pine -Fir- Juniper- scattered Pine -Fir- Juniper-burned Pine -Juniper -Grass -mode rate Pine -Juniper- Grass -open Pine -Juniper-Grass -scattered Pine -Juniper-Grass -burned River Riparian Sagebrush-Grassland Shale-Longleaf Sage Greasewood Silver Sagebrush Grassy Bottoms 1.48 1.43 1.84 1.40 1.68 0.06 0.59 1.36 0.41 1.31 1.58 2.62 2.00 1.67 1.81 0.76 1.41 0.42 0.55 1.52 0.64 0.15 46 88 62 98 ,42 ,46 1.78 1.18 1.20 0.14 0.48 1.33 0.70 0.25 1.34 0.59 1.29 1.38 1.55 1.28 1.50 0.03 0.73 0.52 0.40 37 52 46 76 33 56 1.27 1.68 0.32 0.15 0.58 1.24 0.59 0.28 247 greasewood, and river riparian types predominated were used less than expected during summer. The Douglas fir-juniper and pine- juniper-shale types were often closely associated on adjacent north- and south-facing slopes such that the block method of analysis gave weight to both types even if one was preferred more than the other. Riley and Dood (1984) reported that the Douglas fir- juniper type was used proportionately more by mule deer fawns than the pine- juniper-shale type during summer. Overall, deer use was more widely distributed among types (i.e., there was less negative selection or avoidance of types) during summer than any other season. Deer use of blocks containing the Douglas fir-juniper types remained high relative to other types during autumn, but overall use of blocks containing the pine- juniper-shale types declined from summer. The 2 Douglas fir types with the least overhead cover received the greatest use. Also, the scattered density pine- juniper-grass type received proportionately greater usage than the more dense pine- juniper-grass types. The river riparian type received more use during autumn than other seasons, but use remained less than expected based on availability. During winter, deer selected for a variety of types, although Douglas fir-juniper types received their highest combined use of any season. Blocks containing the Douglas fir-juniper types received their lowest seasonal use during spring, and fewer deer than expected were observed in association with the scattered density Douglas fir-juniper cover type. The scattered density pine- juniper-shale and scattered density pine- juniper-grass cover types were the most selected types during spring. The sagebrush-grassland type, although not selected, received its greatest use during spring as influenced by the availability of forbs in that type (Table 3.3) . On a yearlong basis, deer were more closely associated with and apparently showed more selection for Douglas fir types, collectively, than for any other vegetation types on the area. Among cover types, the burned Douglas fir- juniper type received the greatest use relative to availability. The pine- juniper-shale vegetation type ranked second, but this may have been due at least partially to its close topographic association with the Douglas fir types and the method of analysis . Any assessment of habitat selection based on observed use relative to availability is subject to possible shortcomings that must be addressed to place results in perspective and avoid false conclusions. The block analysis, in common with 248 several other methods of quantifying habitat selection, assumed that all vegetation cover types were equally available to all deer on the study area. That assumption was not met. For example, the Douglas fir types were distributed primarily on the western 60% of the area, where all other types also occurred. Deer on the eastern portion did not have the option of using fir, but had to choose among pine-juniper and other apparently less selected or preferred types. Because of this, the fir types probably were even more preferred and selected for than indicated by our analysis. Distributional biases also were apparent in the relatively low selection indicated for other types, especially the river riparian and sagebrush-grassland types. The former was usually available only to deer ranging within about 3 km of the northern boundary of the area and could be entered only from one direction. Although our analysis indicated the river riparian type was used less than expected based on availability, general observations and movements of marked deer suggested it received relatively heavy use at least during autumn, by deer whose home ranges included some river bottomland. Our analysis also indicated that the sagebrush-grassland type was not preferred during any season, whereas Mackie (1970) showed heavy and apparently preferential use of the type, especially during winter and spring. This type covered more of the study area than any other vegetation type. It occurred throughout, interspersed among forested and other types and as large vegetationally homogeneous blocks on major ridgetops and along the southern and southwestern fringes . Lacking diversity, such large blocks were not attractive to and received little use by deer. Further analysis showed that more deer than expected were observed in blocks containing 1% to 49% aerial coverage of sagebrush-grassland, and fewer than expected (P < 0.0001) occurred only in blocks containing more than 75% coverage (Fig. 9.2). This indicated that, although large blocks of the type were avoided, small patches interspersed with forested types were important to deer. This was especially true during spring green-up when general observations and data from earlier studies (Mackie 1970) indicated that a majority of the deer observed feeding were in sagebrush-grasslands . In reporting earlier studies on habitat use by mule deer on the study area, Mackie (1970) noted that observability bias probably resulted in underestimates of deer use and the relative importance of the timbered pine-juniper and Douglas fir-juniper vegetation types as compared with open types such as sagebrush-grassland. We compared relative use of major vegetation types as determined by observations along vehicle routes during 1960-1964, by direct recording of vegetation 249 26 increments of coverage by sagebrush-grassland 5 increments of coverage by sagebrush-grassland 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 % of Block Covered by Sagebrush-grassland Habitat 25% 26-50 % 51-75% 76-100% Figure 9.2 Use of sagebrush grassland type by mule deer in relation to amount of that type per 28.5 ha block. types in which mule deer were observed during aerial surveys from July 1978 through April 1980, and block analysis of all aerial observations during 1976-1984 (Table 9.4). All 3 methods showed similar use of the pine-juniper types, but use of Douglas fir- juniper types was higher and use of sagebrush-grassland was lower based on direct aerial observations and aerial observations submitted to block analysis than determined by observations along vehicle routes. All 3 methods also indicated that use of sagebrush-grassland was highest and use of Douglas fir-juniper types was lowest during spring. However, both direct aerial observations and the block analysis indicated the highest seasonal use of Douglas fir-juniper types occurred in winter when use as determined by vehicle routes was lowest. Observability differences between timbered and open habitats for vehicle routes and aerial observations must have been important, but some of this difference might have been influenced by winter severity. Approximately 54% of all mule deer observed from the air during the severe 1978-79 winter were in Douglas fir- juniper types compared to only 26% during the relatively mild 1979-80 winter. All winters during 1960-1964 were mild and 250 Table 9. A. Seasonal percentage use of vegetation type by mule deer, compared for three methods of observation and analysis. Vegetation Type Method Season Summer Fall Winter Spring 29 42 56 70 7 5 9 41 26 25 23 31 48 33 32 21 58 52 47 46 40 38 41 42 11 10 2 1 19 29 39 7 20 18 22 16 12 15 10 8 16 14 5 6 14 19 14 11 Sagebrush-Grassland Pine-Juniper Douglas fir-Juniper All Others Vehicle Routes Aerial Observations Block Analysis Vehicle Routes Aerial Observations Block Analysis Vehicle Routes Aerial Observations Block Analysis Vehicle Routes Aerial Observations Block Analysis snow-free. Overall, the comparison indicated that our use of aerial survey data greatly reduced bias due to differential observability of deer in different vegetation types. It also confirmed Mackie's (1970: p. 33) suggestion that overall use of forested types was much more intensive than observations indicated and that relative use of sagebrush-grassland and other open types was minor, especially during summer and autumn . Influence of Vegetational Diversity on Habitat Use Both the number of different cover types and the number of cover polygons per block were considered indicative of habitat diversity. /Amount of "edge" within blocks increased with number of cover types, while both edge and interspersion increased with the number of cover polygons; i.e., a block containing 2 polygons of each of 2 vegetation cover types (4 cover units) was considered more diverse than a block containing 1 polygon each of 3 different types (3 cover units ) . Deer were not distributed randomly in proportion to the number of cover types available per block during any season (summer: X2 = 582.9, 5 df, P < 0.00001; autumn: X~ df, P < 0.00001; winter: X2 = 1042.0, 5 df, P spring: X2 = 621.1, 5 df, P < 0.00001). Fewer deer than expected were observed in blocks containing 1 or 2 cover types (Fig. 9.3), while blocks containing 3 cover types were used as ! = 642.1, 5 < 0.00001; 251 0.35 T3 Q) > 0.30 »— CD CO 0.25 O a> 0.20 CD a o 0.15 c o 0.10 ■*- im o Q. O 0.05 0.00 Expected Summer Autumn □ Winter Spring H Entire Year 6-8 Number of Different Cover Types per Block Figure 9 . 3 Annual and seasonal distribution of mule deer in relation to number of different cover types per 28.8 ha block. expected, and blocks with 4 or more different cover types received greater than expected use. Once 4 cover types were included, the relative degree of selection increased only slightly to the maximum of 8 types found in any block. Deer use of the least diverse blocks ( 1 or 2 cover types) was highest in spring, coinciding with increased use of sagebrush-grassland vegetation. Similarly, deer were not distributed randomly in proportion to the number of cover polygons available per block during any season (summer: X2 = 659.0, 7 df, P < 0.00001; autumn: X2 = 583.6, 7 df, P < 0.00001; winter: X2 = 1269.0, 7 df, P < 0.00001; spring: X2 = 783.3, 7 df, P < 0.00001). Fewer deer than expected occurred in blocks containing 1 to 3 cover polygons (Fig. 9.4), blocks containing 4 polygons were used as expected, and those containing 5-12 polygons were used by more than expected numbers of deer. Influence of Topographic Diversity on Habitat Use Deer did not use the 4 TOPOINDEX categories in equal proportion to availability during any season (summer: X2 = 654.3, 2 df, P < 0.005; autumn: X 2 _ 709.5, 2 df, P <0.005; winter: X' = 1316.7, 2 df, P < 0.005; spring: 252 465.0, 2 df, ■a > 0) JQ O i_ 0) >_ CD CO -Q O 0.5-, 0.4- I Expected ^ Summer H Autumn □ Winter Ei3 Spring 0 Entire Year 0-6 7-13 14-20 Topoindex Value 21-28 Figure 9 . 5 Annual and seasonal distribution of mule deer in relation to the degree of topographic relief. Increasing topoindex = increasing topographic relief (see text). We re-examined the possible influence of free water on distribution and habitat use of mule deer on the area through the block analysis of data for 1976-1984. There were 63 known water sources on the area during that period as compared to 56 during 1960-1964. Of the 63, 44% were located in open sagebrush-grassland or broad coulee bottom habitat, 38% were on the edge of sagebrush and timbered vegetation types, and only 18% were within timbered types. Because of this, the visibility bias described earlier against observations of deer in timbered types from vehicle routes would also be expected to indicate greater than actual proportions of deer use close to water sources . Analysis of deer dispersion during the wet summers of 1978 and 1979 showed that 43.9% occurred within 0.805 km of a water source and 89.9% were within 1.61 km. During autumn, those percentages were 41.4% and 88.4%, respectively. During the dry summers of 1980 and 1983, only 20.0% of the deer observed were within 0.805 km of a known water source, while 66.1% were within 1.61 km. During the subsequent autumn, those percentages increased to 30.6% and 73.9%, respectively. 254 Although the percentage of deer observed within 0.805 km of water in the current analysis was lower than reported for 1960-1963, this was expected based on changes in visibility bias between the 2 periods. The percentages of deer observed within 1.61 km of water was nearly the same throughout. The high percentage of deer observed within 0.805 and 1.61 km of water during most seasons and years was not different from a random distribution. Very little of the study area was more than 1.61 km from water. During wet years, when all sources held water, 44.9 % of the 953 28.8-ha blocks were within 0.805 km of water and 89.0% were within 1.61 km. In addition, ephemeral water sources occurred throughout the area. During dry years, 31.9% and 74.4% of the blocks were within 0.805 km and 1.61 km, respectively, of a known water source and ephemeral water rarely occurred. The observed distribution of deer relative to distance from water was not different from that expected based on a random distribution during wet summers (X2 = 0.76, 1 df, P = 0.40), wet autumns (X2 = 2.30, 1 df, P = 0.14), or dry autumns (X2 = 0.94, 1 df, P = 0.36). During dry summers, however, the observed distribution of distances at which deer were observed from water was significantly non-random (X2 = 107.6, 1 df, P < 0.001). Examination of cell contribution to the total chi-square value indicated that significantly fewer deer than expected were observed in blocks within 0.805 km from a water source and significantly more than expected occurred in blocks more than 1.61 km distant from water. Overall, proportions of deer observed at various distances from water were significantly different between wet and dry years both in summer (X2 = 141.2, 3 df, P < 0.005) and autumn (X2 = 64.3, 3 df, P < 0.005). Deer were distributed further from water during dry years than wet years. This indicated that deer did not move toward remaining water sources as reservoirs dried up as also was the case with deer distribution between the very dry year of 1961 as compared with 1960 (Mackie 1970). This does not mean that free water was not important to mule deer. Rather, deer apparently can easily move 2.41 km or more to obtain free water if needed and water was sufficiently abundant and distributed to meet all seasonal and annual needs even during driest periods. During such periods, several marked deer were seen drinking from reservoirs up to 3.2 km from their normal home range. When disturbed they immediately returned to that home range. Thus, though deer may occasionally move long distance to water during dry periods, such movements are specific to that purpose and do not signify a general shift in home range closer to water sources. 255 Even if deer preferred to range close to sources of free water, several other factors may have led to their distribution at random or greater distances than expected from water. As stated earlier, 44% of all water sources on the study area were located within broad open sagebrush-grassland and coulee bottom areas . The lack of vegetational and topographic diversity may have precluded deer from spending more than minimal amounts of time in the vicinity of water in those habitats. However, deer spent considerable time in the open sagebrush-grassland type during spring when new, succulent green growth occurred in abundance. They also made increased use of the type following green-up in autumn and in foraging on rubber rabbitbrush during autumn and winter. This suggested that a lack of preferred forage also may have limited use of larger blocks of sagebrush-grassland, especially during hot, dry summers when they were distributed at greater than random distances from water. At that time, they probably preferred to remain near steep, timbered, north- facing slopes that retained at least some succulent forage. There were few water sources and thus reduced use by cattle near those sites, which left the succulent forage remaining in those areas primarily available to deer. Livestock grazing patterns probably were also an important factor influencing deer distribution in relation to water. The broad open areas of sagebrush-grassland and coulee bottoms in which nearly half of all water sources were located comprised primary range for cattle (Mackie 1970). During most periods, but especially during hot dry summers, cattle were closely associated with water such that all areas, including timbered sites, within 0.805 km and up to 1.61 km or more from a water source were heavily used for feeding and resting. Although deer may have avoided such areas in part for social reasons, little quality deer forage remained available by mid summer. In the absence of cattle, deer may have made greater use of those areas and have been distributed closer to water, at least during hot, dry summers and autumns. Patterns of Habitat Selection by Sex and Age Class Vegetation Cover Types Seasonal use of vegetation cover types was not statistically different among adult females, fawns, yearling males, and mature males (summer: X2 = 51.7, 60 df, P = 0.76; autumn: X2 = 47.7, 60 df, P = 0.88; winter: X2 = 39.7, 60 df, P = 0.98). Possible differences could have been obscured by that test because non-productive females were often associated with males, and all females and fawns were often associated with yearling males during autumn and winter. To determine whether differences occurred between productive females and mature males, we further compared use of vegetation types by 256 fawns and mature males. Fawns and mature males used vegetation types differently during summer (X2 = 16.8, 4 df, P < 0.005). Some difference may also have occurred during autumn (X2 = 8.9, 4 df, P = 0.07), but there was no difference in winter (X2 = 3.6, 4 df, P > 0.50). Both fawns (and by inference productive females) and mature males used blocks containing Douglas fir-juniper types in greater proportion than availability, but selection for those types was strongest by fawns (Table 9.5). Table 9.5. Observed proportional use of blocks containing various vegetation types by fawns and mature males compared with expected proportional use. Douglas Pine- Pine- fir- Juniper- Juniper- Sagebrush All Juniper Shale Grass Grassland Others Summer Fawns 0.207 0.183 0.218 0.258 0.134 Mature Male 0.159 0.255 0.150 0.286 0.152 Expected 0.132 0.147 0.186 0.377 0.158 Autumn Fawns 0.210 0.172 0.218 0.240 0.160 Mature Male 0.196 0.223 0.176 0.280 0.125 Expected 0.132 0.1/4 7 0.186 0.377 0.158 Winter Fawn 0.217 0.202 0.224 0.223 0.134 Mature Male 0.187 0.214 0.201 0.238 0.161 Expected 0.132 0.147 0.186 0.377 0.158 Similarly, both selected blocks containing pine- juniper-shale types, but here the strongest selection was by mature males. The pine- juniper-grass type was selected by fawns/productive females, but avoided by mature males. The differences were greatest during summer, when fawns were youngest and females with fawns generally isolated themselves from other deer, and declined through autumn and winter. Possible differences in use of vegetation cover types between females and males were also tested by comparing vegetation composition of blocks most strongly selected by adult females with that of blocks most strongly selected by males through summer and autumn. Female-selected blocks were those in which the number of females observed was at least 3 times that expected with a regular distribution and no males were ever observed or, if males were observed, the ratio of females to males was at least 5 times greater than expected on 257 the basis of proportional distribution. Male-selected blocks were based on the same criteria, except that males predominated. Fifty two blocks were designated as female selected and 61 were male selected (Fig. 9.6). The vegetational composition of female- and male-selected blocks differed significantly (X2 = 378.1, 19 df, P < 0.001). Further examination indicated that the greatest difference was in occurrence of the pinus- juniperus-shale type, which was used more than expected by males and less than expected by females. The open and moderate density Douglas fir- juniper, scattered density pine- juniper-grass , moderate density pine- juniper, and river riparian cover types, in decreasing order of importance, were all used more than expected by females and less than expected by males. Non-timbered types, including shale-longleaf sage, sagebrush-grassland, and greasewood were more abundant in male-selected blocks, although individual contributions to the total chi-square value were not high. These differences in selection of vegetation cover types between productive females/fawns and adult males were expected on the basis of cover and forage availability. Douglas fir- juniper and scattered density pine- juniper-grass types, selected by productive females/fawns, provided the greatest seasonal quantity and diversity of forage for deer (Tables 3.2 and 3.3). The pine- juniper-shale type produced the lowest quality and diversity of forage of any timbered type on the area and had greatest representation in blocks used predominantly by males. The Douglas fir- juniper and moderately dense pine- juniper cover types, most strongly selected by females, also provided the greatest amount of hiding cover among all major types. Vegetational Diversity and Topography Fawns and productive females used blocks containing more cover types than those used by mature males during summer (X2 = 10.64, 5 df, P = 0.06), but not during autumn (X2 = 7.62, 5 df, P = 0.19). Mature males were more likely to occur in blocks containing only 1 or 2 cover types than were fawns and productive females. Fawns and productive females also selected more diverse blocks than mature males based on number of cover polygons per block during summer (X2 = 24.07, 7 df, P < 0.005), but not during autumn (X2 = 9.05, 7 df, P = 0.25). There were significant differences in distribution of fawns, females, and males among topographic relief classes during summer (X2 = 21.2, 6 df, P = 0.002) and autumn (X2 = 13.4, 6 df, P = 0.04), but not during winter (X2 = 1.7, 6 df, P = 0.94). Fawns and productive females avoided areas of both extremely low and extremely steep terrain. Proportionately 258 355 ^7 7 j 1 ♦j_ -f' / - F^s ■ / ~ JL* "* '"^S Tl_ ^-^ ? Wmm \ '•' / '• i - & * hi rri' » 1 / 1 Hi * W 1 • Jt4 s— " ^w 3 ^ j E3 _j ^^ HB^% J 1 &\ •'mm L t% ^ S M I _S„SS E ; V V \ m 1 • ^i""^ « ^# T 3\ ?T >?5"""tf i 3"^"^ m * \ :_i • ^ i v 2 VJ.3 VJ- - . ^ ^-^ Z. % ]h ■ \ 1 *•• ^^ ■ < > V^ \ ■ < . ft - * 4- \- -i^.' X ( 1 < < I 22 / \u \ »/ JL^£ / fL ''s* • ■ / n* f «Sk • ■ £ -r t r -S z5E~k 1 ■_ \ M IB \ rw_ s * * i -I 5 £ 1^5 t A 3E ^ ■ t • E v,-^ £W ■ M ' bsHs^a ^\i PC -/^ \ sgv^ i^r " / J * ,; JP r X V |fl V : >E«^E „ - j t \ a • it v WT — 1 \** \N V * -i WLA\ \% 1 ^tft#l fflW a 03 TO TO ID LL LU mi col < O tfi f* E 4-> X ^ £1 T3 G (0 w i -a p 0 Q) .—I In ft 0) u 1-1 o> ■U cd (D U 00 n) u •H 4-1 • ■H u (3 01 au 01 ■H T3 CO 0) Jfl .-1 *J 3 •rH 0 3 >N en u Ol o rH 4-1 .a m cd rH •H 01) i-i •H ad 0 > 14-1 *j 0 cd 4J en -H a) ■O 0) cd u H. (0 43 Ol ai (D > 01 U C id o C oo ■H CO cd ai u < i-i 01 e 0 3 CO in cd 0) u < u ai cd .a in cd cd E-i > o o OOOOrHOOHoO O OOOOOOOOOiO O CI OOOOOOOOOOiH • O oooooooooooo V o OOOOOOOOOOOOi II v v v v v v v v ii ii n - a, - . . . » . . . . . . . CN 1-HcOOCNOOir^OkDr^lOiH H CM vor^ooin-3-r~.io-> I 1-1 •H 4-1 oo 3 O Q C ai ex o 1-1 ai a. ■H C 3 >-> i 1H ■H 14-1 1j 01 ft ■rH a 3 *-> l 0) .H cd CO i 0) r-H m JC CO 1 IH 0) ft •-H a Ol 00 1/1 -H 1H 3 cd cd <4H 01 "-J CO ji rH ft l en OO 01 •H 01 4-1 3 c 3 cd C a cd lH cd O rH 3 •H 01 J3 ■H Q oo i-j a, r-H Ol t-l D I OO X) OO cd T3 O ai T3 c 0 cd ft 01 Q C 01 O o CO ■H iH ■H u J 3 Cd 0) ■a Oh Ol l 01 u 4-1 ai 4-1 0) Ul 01 IJ 4-1 c C 4-> r-H cd > 0) en a C 01 ca cd Ol r-H > 01 U 3 a. u -C 1H ■-H >H ft en -Q o en C/l o CO Cd >> ^- Vj 0) > o o c 0 ■H 4-1 cd 4-1 0) OO 01 > en en cd lH tfl 1 en en 1-1 cd 0) !-. a o -a •H i G c U cd 3 01 r-l -"J ft en 1 •rH en en 01 C cd E c 3 l-i O ■rH r") OO 4-1 a, 1 4-J Ol -C 0 Ol c en PQ 4-1 ■H 3 cd Oh U ^ In -Q en 0) C 01 en -a 0> OO cd o ft cd U E o CO o O o O o o O o o O o o O • • • o o o V V V Oh Cm 04 * » ■> >tf ^D m O en m rH II ii II CM CM CM CO OJ a ^ ^ ■u u o u in rH O) OJ 0) .a 3 > > rH o o lH cd u u ai ft > 4-1 ,* 4H X O O o en w o C Q U r-H u o 2 01 ,Q 01 OO M XI X ^ O 0 Vh 0 rH PU 3 OJ 3 O o z a. s ft E-« 262 polygons per block and steeper terrain than summer range areas . Migratory deer generally moved from areas of low relief at the heads of drainages along the southern edge of the area to areas of steep terrain near the Missouri River. Thus, most of the difference in habitat characteristics between seasonal ranges was apparent without statistical tests. The interpretive problem related more to separation of cause and effect and elimination of possible spurious relationships. Steeper terrain containing adjacent north- and south-facing slopes appeared to be the most important factor distinguishing winter from summer ranges of migratory deer. Differences in vegetation cover types and diversity followed differences in terrain and thus did not appear to be the primary determining factors in deer preference for those areas. For example, the 2 eastern wintering areas lacked the Douglas fir type, 5 of the 7 lacked the silver sagebrush type, and 3 lacked the river riparian type. On the other hand, steep terrain generally held lesser snow depths and provided thermal environments more favorable to deer survival than level or rolling terrain. North-facing slopes usually were dominated by timbered types with dense overhead cover that further reduced snow depth and provided thermal cover. South-facing slopes generally were more open, providing for rapid snow melt as well as warm areas for resting and foraging on sunny days . Mule Deer Habitat Use in Relation to Elk and Cattle A prior publication (Mackie 1970) focused specifically on habitat use relationships between mule deer, elk, and cattle on the study area during 1960-1964. Although we do not propose to review those relationships in detail, data obtained from subsequent studies, especially during 1976-1984, enabled us to re-examine previous conclusions based on additional information and different analytical techniques. Data on numbers and general distribution of elk were recorded during all aerial surveys in winter 1964-1975. During 1976-1984, numbers and locations of all elk observed during seasonal aerial surveys were recorded identically to those for mule deer and were compiled and analyzed using block analysis. Time and other considerations precluded us from collecting quantitative data on cattle distribution or habitat use after 1963. Thus, conclusions relating to mule deer-cattle relationships can only be reviewed qualitatively on the basis of general observations and additional studies conducted on the supplementary NCRCA during 1972-1975 (Knowles 1975, Komberec 1976, Campbell and Knowles 1978) and during our studies in 1976-1986. 263 The range use and food habits data obtained during 1960-1964 indicated that opportunities existed for competition between mule deer and elk and between mule deer and cattle on the study area. The probable degree of competition was light, except for some seasons. The greatest overlap in range use and food habits of deer and elk occurred during the period from April to September, while that between mule deer and cattle occurred in April and May. Opportunities for competition in spring centered around intensive use of the sagebrush-grassland type by all 3 species. The potential for competition between mule deer and elk during summer was related to significant use of timbered vegetation types and similar forage plants by both species. In addition, competition could become more severe; 1) during periods of drought, 2) if stock water sources were constructed on terminal portions of large ridges or on smaller ridges within timbered breaks and extended intensive grazing to those areas, or 3) if numbers of elk using the area increased greatly. Considering relative distribution and varying intensity of use of the area by mule deer, elk, and cattle on a yearlong basis, there was little overlap between primary cattle range areas and primary mule deer habitats (Fig. 9.7). Primary areas for use by cattle were the broad open plains, major open ridgetops, and larger coulee bottoms. Mule deer primarily used steeper, diverse or timbered terrain between major ridgetops and major coulee bottoms. Comparison of overall spatial distribution of mule deer and elk also indicated relatively little overlap, especially when seasonal patterns were considered. Mule Deer and Elk A comparison of the yearlong density distribution (intensity of use) pattern for elk during 1976-1984 (Fig. 9.8) with that for mule deer (Fig. 7.1) indicated little overlap in areas of greatest intensity of use for both species. A simultaneous plot of the 2 highest categories of intensity of use for both species (Fig. 9.9) indicated only 11.2% of the high density blocks for both species were the same. Of high density elk blocks, 26.9% were also high density mule deer blocks and 19.1% of high density mule deer blocks were also high density elk blocks. The general pattern of dispersion of elk on the study area, however, was similar to that for mule deer. That is, moderate to dense concentrations generally extended across broken terrain along slopes and coulees between major ridgetops and major coulee or river bottoms. Thus, the primary habitat for elk and mule deer centered in and around essentially the same land areas, although areas used intensively by elk extended on to primary cattle range more so 264 w 03 0) U 03 0) P c •H P 03 U •H ■0 c •H 0) ^H P 4J 03 U -o C 03 M (1) 0) -o 3 0) fc= a 4-4 en 0 X! U 03 d) CJJ P p >i id -u Cu T3 c Q) n W ■H 3 P O >i a .-t ■rH 0) M > P ■H in W •H c T) 0) p H C 03 >H M 0) P c en 0) 0 u e 161 SO 0) M Cn •H 265 1 a ■ » • # ■ j^k ■ ■ ^t • i lis! 1 neS * s -, $ ! * I i :■:"" * J "': :':: * i » i 4 i || i I I : i KM ..::: ™1I • i :: I l 1 I i 1 * 4 :: ■l t 1 ' 1 H mi * * • * • . l ■ 1 81 1 ih a t < *■ 1 « | ;|:j # * \ r^ ■ # ' #" i 1 f 4 * * * 1 '• ,* * * • * 1 1 " *-; i ^M S ' A o" in * 1 ■ ^1 • ^J * >* ■ # in ■ (g * 03 P i: * TD 4 • i X • ■V i i \ • * f 1 * • i !:::: f • ■ 1 t * i ■Kr~ » * > P ffl 1 ♦ * ■ 1 • ■ • • 1 1 ;; i \% * • » \% • <\ t \\ 1 \* CO C CM E ■p rtJ X) >1 XI XJ d) C ■H e u ai +j CD •a ^ H CD n o 4-1 CD CO 4-1 o .p *** CD CTi iH I CTl C CJ1 •H 'H X) T3 ■H ■P CO >1 co (D ■H > X) ^ >i «> -P •H -H CO «3 C -H Q) S-J T3 CD CD > CM •H ro ■P <0 £ ■-* o CD M Pi 4-1 CO X en CD H 3 tj> •H 266 c/> -Q E 3 1600-1 1400 1200- 1000- 800- 600- 400 200 H 1960 400 n B LU O »- 200 1 year) females in the population during June, and about 50 male fawns were added for 292 every 50 adult males assuming an approximate 1 male: 2 female adult sex ratio. Without additional mortality, the highest observed reproductive rate would result in a doubling of a population of 100 adult females and 50 adult males (an instantaneous growth rate of 0.69); at the lowest observed reproductive rate, population growth would be only slightly reduced to an instantaneous rate of 0.51. Because natural mortality was low, averaging only 7.2% in years when females were not hunted, the observed "reproductive restraint" (including both intrinsic and extrinsic factors) was not, by itself, sufficient to halt growth of the female population. As is the case for most populations, substantial mortality of fawns and greater than average mortality of adults must occur to achieve a stable population. Other data on reproduction and age-specific mortality of adult females also indicated that little, if any, effective "reproductive restraint" was inherent in the population. Essentially all adult females initially produced fawns at the maximum rate their body condition allowed. A high rate of initial production apparently is less costly than rearing fawns (Short et al . 1969), and the initial effort can be terminated prior to or during early lactation if necessary. It also provided maximum opportunity for successful reproduction during years when quality of habitat resources for rearing fawns were good. In highly variable and unpredictable environments such as our study area, the trait of initially producing maximum numbers of fawns during all years apparently maximizes the opportunity for females to pass on their genes. Recruitment of fawns in alternate years (Mundinger 1981) or producing fewer fawns during all years, are apparently not successful strategies in this environment because these females are no more successful than others in rearing fawns during years of poor conditions. Such strategies of reproductive restraint might be selected for in harsh, but stable and predictable environments where conditions for fawn rearing are seldom exceptionally good. The environment of our study population is not only variable and unpredictable, but includes an effective natural predator. Living conditions for deer can swing rapidly from exceptionally good to exceptionally poor, and both favorable and unfavorable conditions often prevail over several consecutive years. In this situation, the best reproductive strategy appears to be to initially produce young at maximum rates each year to take advantage of favorable conditions for recruitment as they occur. Given the relatively short reproductive life of mule deer on the area, it seems most adaptive for a female to have the maximum potential to respond to good conditions, should they occur during her lifetime. 293 The longer the relative reproductive life span, the more likely "reproductive restraint" becomes adaptive. Population theorists, beginning with Malthus and Darwin, traditionally have looked at the impressive reproductive potential of most populations and concentrated their efforts on the role and importance of mortality in population regulation. This view forms the basis for much current population theory. In harsh, variable, and unpredictable environments, however, it may be equally valid to assert that high reproductive potential is necessary to cope with inherent regular or periodically high mortality (Fisher 1930, White 1978). Although changes in reproductive rates, by themselves, were not sufficient to significantly regulate population growth of deer on our study area, reduced reproductive rates usually coincided with high fawn mortality after birth; they often coincided with increased adult mortality as well. If reduced reproductive rates and high post-partum mortality rates often resulted from the same ultimate factor(s), variations in the operation of such factor or factors may explain much of the variation in deer numbers. Variation in Mortality Rate The relative intensity of operation or relaxation of factors affecting mortality rates of fawns and adults appeared to exercise the greatest impact on population growth rates and trends. It should be noted, however, that from the standpoint of population dynamics and regulation, the role and impact of fawn mortality differed from that of adult mortality. Factors affecting fawn mortality determined annual recruitment to the adult segments of the population, whereas those affecting adult mortality determined losses. Also, the relative importance of various mortality factors differed between fawns and adults. This, plus the fact that our data indicated recruitment to the reproductive population was not effectively achieved until individuals were at least 1 year old, required that mortality rates and causes for fawns and adults be addressed separately. Fawn Mortality Annual mortality rate of fawns varied considerably among years (17-94%) and population trend generally followed the trend in fawn mortality by 1 year. Most mortality occurred during summer ( ave . 65%), but seasonal rates varied widely among years . Predation by coyotes was the single most important mortality factor for fawns, accounting for 88% of all deaths of radio collared fawns and 85% of deaths in summer and autumn. Overall, coyote predation on fawns was generally 294 lower during autumn than other times of the year. Only 9.1% of all deaths of marked fawns during 1976-1986 occurred during autumn. During the early-mid 1970s, 1983, and 1984, coyote predation on fawns was relatively high during autumn. It may also have been high in some earlier years such as 1961. With the possible exception of 1961, hunter harvests had only negligible impact on fawn mortality. Over-winter mortality of fawns was important, but quite variable. Of the fawns that died, 26.3% died during winter; however, over-winter mortality rates for individual years ranged up to 82%. Coyote predation was the proximate cause of almost all (95%) of the fawn mortality during winters 1976-1986. Winter severity influenced fawn mortality rates, but not in a simple manner. During most years, fawn mortality increased with winter severity in a density-independent manner (Fig. 6.14). Some years were exceptions, however, and very high mortality occurred in both relatively mild and severe winters, independent of density. Fawn mortality was relatively low, even during severe winters, when microtine populations remained high over winter (1968-69, 1978-79, 1982-83, and 1986-87). In either case, the highest winter fawn mortality rates occurred in years preceded by 1-3 dry summers that resulted in deer entering winter in poor condition. Conversely, moderate fawn mortality during very severe winters was preceded by one or more summers of excellent forage conditions. This occurred when deer densities were both high (1968-69) and low (1978-79). Thus, general body condition and level of fat reserves as affected by summer and autumn forage conditions, winter severity, coyote and alternate prey population levels, all interacted to influence winter mortality rates. Fawn mortality rates were not significantly correlated with coyote density, but other factors, including forage conditions, winter severity, and population levels of alternate prey were related to fawn mortality rates. Such other factors, at the least, influenced predation rates. This indicated that coyotes and predation, although important, probably were not always the ultimate or overriding factor in fawn recruitment or population dynamics and regulation. Rather, the role and importance of predation was tied to other biological and environmental factors which simultaneously influenced recruitment and numbers of deer on the area. Fawn survival to early winter was closely correlated with forage production during most years. Because deer density did not affect that relationship, forage quality (defined as the succulence of vegetation) and the timing and length of the period when green, succulent forage was available appeared to be the controlling factor. Calculations of the quantity of forage available, information on antler size and quality among 295 yearling males, and the lack of density relationships involving antler size also supported that interpretation. All of this indicated that fawn survival/mortality was functionally determined by factors influencing forage quality and the physical condition of deer during fawning and fawn-rearing. How does the relationship between forage quality and fawn survival to winter reconcile with the fact that 85% of the summer-fall mortality resulted from predation? Were, as some other studies suggest, deer in poorer condition and more susceptible to predation when forage quality was low? Although there may be basis for such supposition, our data indicated that the relationship between deer condition, coyote predation, and fawn survival was not simply one of cause and effect, especially for young fawns. Regular observations of fawns during summer, including those killed by coyotes, indicated that few, if any were in obviously poor condition. The only exception occurred in 1984, when several fawns, including 2 abandoned by their dams, appeared to be in relatively poor condition at the time of capture. Most other radio-collared fawns killed by coyotes appeared in sufficiently good condition to have survived in the absence of predation; all were active and playful in interactions with each other and their mother when last observed. Moreover, measurements and general observations of deer condition during the years 1973-1977, when fawn mortality was consistently high, indicated that all deer were in at least average, and probably excellent physical condition. They certainly were in better condition during those years than during 1984 and 1985 when some reduction in reproductive effort occurred. Although no experiments involving coyote control were conducted on our study area in recent years, studies in other western states during the 1970s (Beasom 1974, Kie et al. 1979, Smith and LeCount 1979, Stout 1982) indicated that predator control increased fawn survival or that fawn survival was higher in predator-free exclosures than in adjacent areas. Such results implied that coyotes killed fawns that were healthy enough to survive in the absence of predation. We noted earlier (Chapter 3) that coyote control efforts on our study area during the 1950s may have influenced the high fawn survival that occurred even during years of poor forage conditions. Similarly, circumstantial evidence from the Sage Creek area, about 96 km southwest of our study area, indicated that fawn survival might have been higher in the absence of coyote predation during the mid-to-late 1970s. There, poor fawn survival and a population decline similar to that on our study area also occurred during and following the 296 severe 1971-72 winter (C.R. Watts, unpublished data). In contrast with our study area, however, fawn survival subsequently increased sharply after 1972 to a high of 109 fawns: 100 adults in winter 1974-75. We did not observe increased fawn survival until 1978-79 despite good to excellent forage conditions through the period on both areas. The major apparent difference between the areas was that coyote control involving intensive private aerial hunting during winter occurred at Sage Creek; no equivalent control occurred on our study area. Other data also suggested that coyote predation on deer may have been abnormally high during 1973-1975, especially in relation to forage conditions on the study area. Normally, when data were obtained for all species, annual production and survival of young followed similar trends in mule deer, white-tailed deer, antelope, and elk on the area. For example, survival of young was the highest ever recorded for all 4 species during 1979 and was above average for all during 1986. All species experienced below-average survival of young in 1976, 1977, 1984, and 1985. During 1973-1975, survival of mule deer fawns was much below average each year; survival of whitetail fawns was below average during both years for which data were obtained; and survival of antelope fawns was below average during 1973 and 1974 and average during 1975. In contrast, elk experienced average calf survival in 1973 and above average survival in 1974 and 1975. Forage production and presumably fawn hiding cover was above average during all 3 years, especially in 1974 and 1975, and density of all species was low. Thus, the only difference between elk and mule deer, white-tailed deer, and antelope must have been related to the relative vulnerability of their young to predation by coyotes. The larger size of elk calves and the tendency of elk to form large nursery groups probably rendered them less susceptible to coyote predation. There was some evidence that non-density-related forage deficiencies can contribute to the intensity of coyote predation on fawns. Forage-vegetation conditions appeared to influence behavioral patterns of deer and the vulnerability of fawns to predation. They also influenced coyote behavior with respect to predation on deer, especially through their effects on alternate prey populations . Coyotes hunt deer primarily by sight, and it usually is necessary for them to maintain visual contact with the deer to successfully complete a kill. Because of this and the tendency of fawns to remain bedded and well hidden when not nursing during the first few weeks of life (Riley and Dood 1984), most attempts at predation occur after coyotes have observed a fawn active around the doe during nursing bouts 297 (Hamlin and Schweitzer 1979). The amount of time fawns spend active with the doe increases as they get older. Thus, much mortality of fawns on the area occurred after they passed their most sedentary and cryptic stage ( Dood 1978, Riley 1982, Hamlin et al . 1984) . General observations suggested that annual forage conditions and the timing of forage desiccation influenced the degree of coyote predation on fawns by altering activity patterns of both females and fawns. For example, it appeared that during dry years, females spent more time foraging away from their fawn and may have produced less milk (Rowland 1944, Thomson and Thomson 1953) because of difficulty in locating succulent forage. At the same time, hungry fawns spent more time foraging on their own, exposing themselves to observation and attack by coyotes . Fawns normally spend some time foraging on their own during all summers, and most are at least somewhat active independent of the dam by about 1 August (Riley 1982). Observations during the dry summers of 1983, 1984, and 1985 indicated that fawns were more active, independent of their mothers, during the day than in previous years. Similarly, data obtained by Dood (1978) and Riley (1982) indicated that fawns began foraging about 1-2 weeks earlier during the relatively dry summers of 1977 and 1980 than during 1976, 1978, and 1979. Fawns were born about 6 days later, on average during 1977 and 1980 than during 1979, further indicating abnormally advanced, independent foraging by younger fawns during those dry years . These observations would seem to justify the hypothesis that dry conditions resulting in early desiccation and reduced quality of forage result in behavioral changes in fawns that increase their susceptibility to predation. Such a hypothesis can explain at least some of the interaction between fawn mortality, forage quality, and coyote predation rates; however, more extensive data on fawn behavior during wet and dry summers and relationships to predation rate are needed. Because coyotes are facultative rather than obligatory predators on deer, alternate prey population levels also played a major role in the relationship between forage production and fawn mortality. Microtine population irruptions coincided with years of high forage production and abundant ground cover. The availability of abundant, easily captured prey and its increased use as food by coyotes (Hamlin et al. 1984) greatly reduced the intensity of coyote predation on mule deer fawns during those years . The fact that fawn survival during 1980 and 1981 was higher than expected based on both forage conditions and microtine population levels, probably reflected increased availability of deer mice and lagomorphs during those years. Both of these groups were more 298 abundant during 1980 and 1981 than during the previous, 1976-1977 low in microtine populations. Quantitative data on small mammal populations on the study area were not collected prior to 1976, but data existed from other studies in central Montana (Cada 1968, Tschache 1970, Reichelt unpubl . ) . Also, qualitative information from a variety of sources such as narratives of game range personnel (Appendix D) , annual reports of various individuals and agencies, area newspapers, and field notes provided basis for estimating probable trends in abundance back to the 1930s. Murie's (1935) notations, for example, indicate that both jackrabbits and microtines had probably reached simultaneous peaks around 1932 or 1933. Thereafter the narratives indicated peaks in jackrabbit populations at 10-12 year intervals (early 1930s, early 1940s, 1949-1950, late 1950s, and 1970-1971). Lows generally were noted near the middle years of each decade; ca. 1945, 1955-1957, 1965-1966, and 1973-1977. Our data indicated a subsequent peak in 1983 and a low during 1985-1987. Microtine population highs can be documented during 1963-64, 1968-69, 1978-79, 1982-83, and 1986-87. Moderately high numbers apparently occurred during 1974-75, but the irruption was much less distinct and the peak lower than in 1978-79. Information on deer mouse populations was less complete, but highs apparently occurred in central Montana in 1968 and 1973. Populations on the study area increased from lows during 1977 to a peak in 1983, crashed abruptly in 1984, and subsequently increased through 1986. Although microtine populations apparently began to increase in summer 1974 and were moderately high during early 1975, fawn survival was poor in contrast to other periods of microtine abundance. However, because the microtines never became exceptionally abundant and deer mice and lagomorph populations, especially jackrabbits, were at cyclic lows during 1973-1977, the combined small mammal prey base was relatively low and did not deter predation on deer. Some coyote control using compound 1080 was practiced on the area until 1972, although the extent and intensity of efforts was considerably reduced from the late 1940s and 1950s. Narrative reports and field notes indicated that poison bait stations became less effective after the early-mid 1950s when coyotes either learned to avoid the stations and/or poisoning selected for survival of individuals that killed, rather than scavenged for food. Thus, despite at least some effort at control and additional mortality to hunting and trapping, coyotes were common, if not abundant on the area through the 1960s and into the early 1970s. A peak in coyote numbers may have occurred in 1973 as a result of simultaneous peaks in microtine and jackrabbit populations during 1969-1971 and high overwinter mortality of deer in 1971-72. Such a peak 299 in the coyote population, apparently in response to an abundance of food, was also documented during 1984 (Fig. 5.12), following peaks in microtine and jackrabbit populations in 1982-1983 and high overwinter mortality of deer during 1983-84. Although coyote numbers declined during 1974-1977, they apparently remained sufficiently abundant during 1973- 1975 to put extra predation pressure on deer despite the minor peak in microtines during 1974-75. Our interpretation of the apparent unusually heavy impact of coyote predation on the low deer populations of 1972-1977 may be similar to the relationship described by Gasaway et al . (1983) for wolf and caribou (Rangifer tardandus) populations in Alaska. That is, the effects of predation may be inversely density dependent and, when combined with harvests (predation by humans), may maintain deer populations at low levels unless or until environmental conditions change significantly. Similar observations were reported for wildebeest and their predators in Kruger Park, South Africa (Smuts 1978, Walker and Noy-Meir 1979) . Collectively, trends in fawn survival relative to vegetation conditions, coyote and alternate prey populations, and winter severity for the period 1960-1975 indicate similar interactions of all factors on fawn mortality as previously described for the period of intensive studies (1976-1987, Chapter 5). Thus, interpretation of the role and importance of any single environmental factor on fawn mortality and recruitment on our study area and similar environments cannot be determined independently of all others. Although relative body condition of deer may influence coyote predation rates on fawns, not all deer killed by coyotes were in noticeably poor condition, and even many of those killed when in relatively poor condition might have survived in the absence of predation. Several factors other than body condition and alternate prey population levels were involved in coyote predation on deer during winter. For example, both deer behavior, based on age and experience, and local environmental and site conditions played a role. Observations of coyotes attempting to prey on deer as well as observations of deer behavior during drive-trapping indicated that, under pressure, fawns are the most likely of all deer to either break away from the group or to stop and stand in confusion. This behavior was also noted by Griffith (1988). Thus, although fawns are generally in somewhat poorer condition than most other deer during any year (Runge and Wobeser 1975, Dusek 1987), it is often their behavior when attacked that makes them vulnerable to predation. Deer in relatively poor condition may tire sooner and stop or break away from a group, but fawns exhibited this behavior even 300 during years when they were in relatively good condition. Inexperienced behavior might also explain the relatively high predation rate we observed for yearlings as compared with older deer. Many deer killed by coyotes were found on ice along the Missouri River or Fort Peck Reservoir, but many were also found in uplands. The incidence of coyote kills on the river and reservoir apparently peaked in the mid 1970s, because relatively fewer remains of deer were found on those sites after 1976. Most upland kill-sites were located at the base of steep slopes and in sharply cut drainage bottoms. Evidence indicated that the kills were usually accomplished when the deer fell down on the steep, icy slope or bottom. Snowfall followed by periods of melting and severe cold that caused ice to form on those sites often preceded abnormally severe predation on deer (Knowles 1976 and personal observations). These conditions could cause the fall and perhaps injury of any running deer. Often, when investigating these kill sites, we had to use hand-holds on shrubs to move across the slope. Because the deer apparently were in relatively good condition, the high losses during winter 1975-76 (Knowles 1976) appeared to be related to icy conditions that prevailed as well as the low alternate prey populations that year. Similar icy conditions prevailed during the relatively mild winters of 1983-84 and 1984-85. Despite poor body condition and relatively high mortality, most deer survived both of those winters in worse physical condition than many deer killed by coyotes in other years. Similarly, Shaller (1967) reported that predation by tigers... "is not confined to the young, old, sick, and surplus animals, but that prime ones are also readily taken" . He also pointed out that the method of hunting is less important in determining the class of animals killed than the characteristics of the prey population. Mortality of Adult Females Hunting was the major known cause of death among adult females . Coyote predation ranked second among known causes and was suspected to be at least the proximal factor in most mortality for which cause was undetermined. Most hunting mortality of adult females apparently was additive to other mortality. During 13 years (1960-1972) with a hunter bag limit of 2 deer of either sex, total annual mortality averaged 22.2% for adult females. For 6 years with a 1 deer, either sex, bag limit (plus a very small number of second tags during 4 years), the average was 17%. During 8 years in which only males were legal (except for 0.08 and 0.16 antlerless tags/mi2 during 1985 and 1986), average annual mortality of adult females was 7.2%. During these same 301 periods, winter mortality of adult females averaged 6.8%, 6.1%, and 5.4%, respectively. Averaged over long periods, hunting mortality increased average annual mortality, but did not result in a "compensatory" reduction in winter mortality rates. A plot of natural mortality rate against hunting mortality rate for individual years (Anderson and Burnham 1976) did not indicate decreasing natural mortality with increasing hunting mortality for adult females (Fig. 11. 1A). These findings appeared contrary to current concepts concerning compensatory mortality in deer, which hold: 1) that harvest removes animals that would otherwise die from another cause; and 2) that harvest removal of any animal from a population increases the chance of survival (reduces mortality) of remaining animals and/or those of a subsequent cohort in a density-dependent manner. Because of this, it seems essential to further examine both concepts as they apply to mortality patterns and rates of adult females. It is possible that the low average natural mortality rate of adult female mule deer on the area (X = 7.2% with no hunting) allowed little opportunity for compensation to occur. Most work supporting the concept that harvests remove animals that would otherwise die has involved species populations like birds that commonly experience natural mortality rates of 50% or more per year. If a relatively low percentage of all adult females is destined to die over winter (X = 6.2% in this study) , and those most likely to die are not selected for or against by hunters, only 6.2% of the hunter harvest can be "compensatory" (substitute for natural loss). The remainder, or more than 9 of every 10 adult females shot on the average, is additive and will contribute toward a decline in the population. Thus, annual survival of adult females declined in an almost straight line with hunting mortality rate (Fig. 11. IB) . Overwinter mortality rates may vary considerably (0.7-24.8% during 1960-1986), however, such that the exact degree to which harvest losses in any given year may substitute (be "compensatory") or be additive will also vary. It also may depend on a number of other variables, including the age structure of the adult female segment of the population. The apparent higher harvest rate for older (8.5 +) females than younger females may indicate a slightly greater average degree of substitution than 6.2% for that segment or for the female population in years when animals in that age group are relatively common. Somewhat greater than average compensation could also occur when emigration of yearling females during some years is taken into account. Vacant home ranges of some harvested adult females could be occupied by yearling females which might otherwise have left the population. 302 A. 1.00 -I 0.90 - 0.80 - 0) 0.70 - CO CE 0.60 - >» +-* "CD 0.50 - ** i_ O 0.40 - 2 To 0.30 - v. D • 03 0.20 - z 0.10 -J • • n no •• • • •• • . • •• • .•: • • u.uu .10 .20 i .30 i .40 B. 1.00- 0.90 -i 0.80- * • •• . 0) 0.70- • "co • DC 0.60- CTJ • > "> 0.50- l_ 13 CO 0.40- 75 3 0.30- c c < 0.20- 0.10- u.uu i i .10 .20 Hunting Morta i .30 ity Rate i .40 Figure 11.1. A. Annual natural mortality rate of adult female mule deer plotted against their annual hunting mortality rate. B. Annual survival rate of adult female mule deer plotted against their annual hunting mortality rate. 303 The lack of a relationship between deer density (across the range of densities we observed) and mortality rates may be another reason that there was little evidence of compensatory mortality among adult females. Over-winter mortality of adult females was not significantly related to either total number of deer (r=0.03, P>0.50, df=27) or total number of adult females (r=0.08, P>0.50, df=25) entering winter. It also was not related to the number of adult females in the population the previous summer (r=0 . 15 , P=0.47, df=27). Increased survival of deer remaining after harvest is most often based on the assumption that winter forage is in short supply and limiting deer numbers. It also assumes that harvesting deer prior to winter increases the quantity of forage available per capita for remaining deer, thereby increasing their condition and survival. If winter forage supplies are not limiting, as in this study, neither assumption holds, and much of the rationale for compensatory mortality is gone. Although it may also be argued that reducing deer numbers through hunting could increase per capita forage during other times of the year, we also found no evidence that the quantity of forage during other seasons was limiting, nor that mortality responded to deer density at any time of the year, or in a delayed manner. To clear up possible objections or confusion over terms, we note here that we have already (Chapter 5, earlier discussion this chapter) indicated that we had no evidence for density-dependent compensatory reproduction. Thus, hunting mortality of adult females resulted in little substitution (compensatory mortality) , no measurable decrease in mortality as the result of increased per capita forage, and no compensatory reproduction. Although hunting mortality was a major cause of mortality for adult females, hunting mortality rates were low to moderate, and generally below long-term average recruitment rates. Only during 1961, 1964, 1971-74, and 1984 was hunting mortality of adult females, by itself, greater than subsequent recruitment. During those years, however, recruitment was much below average. A stable population can be maintained with some adult mortality in all years because at least some fawn recruitment always occurs. In some years, however, the natural mortality rate among adult females was higher than recruitment. Thus, the availability of a "harvestable surplus" of adult females varies among years. It could also vary with management goals; i.e, to increase, stabilize, or decrease numbers of deer in the population. The ease with which harvest can decrease population size during years or periods of low recruitment and/or high adult mortality was demonstrated by population trends during 1961-1962 and 1972-1974. 304 Exceptionally heavy harvest of adult females and fawns in connection with low overall fawn recruitment in 1961-62 resulted in a sharp, though relatively modest population decrease to winter, spring, and summer 1962. Despite a subsequent reduction in the harvest, continued low recruitment resulted in stabilization of population size to the following winter-spring and provided the base for population increase and a larger "surplus" in 1963 when good recruitment occurred. The combined effects of low fawn recruitment, severe winter mortality, high overwinter mortality of adult females, and relatively high harvest mortality of adult females resulted in a very sharp and substantial decrease in both total population size and the adult female population from autumn 1971 to spring and summer 1972. Continued liberal harvests of adult females along with low fawn recruitment influenced further reductions in total population and adult females through 1974. Overwinter mortality of adult females was relatively low (5-10%) in each of those 3 years. However low recruitment and a female age structure comprised primarily of individuals with a very low natural mortality rate (susceptible females were eliminated during the severe 1971-72 winter), essentially precluded any "harvestable surplus" throughout the period. Under those circumstances, all mortality was "additive" and reduced the female population. Such a low population level was reached that the absolute number of fawns recruited was insufficient to stabilize or increase population size. It was not until 1978, when fawn recruitment increased as a result of decreased coyote predation, that the population could begin to recover. The potential for a similar impact of hunting existed during 1964-1965, when severe winter conditions also occurred. Although fawn mortality was heavy during the 1964-65 winter, adult female mortality was relatively low. Harvest mortality of adult females was also relatively low in autumn 1964. Thus, while total deer numbers on the area declined considerably to spring 1965, the decrease in adult female numbers was not extreme. Also, fawn recruitment exceeded adult female harvest rates and was only slightly below total female mortality in 1965 and 1966 so total female numbers remained relatively high and stable until fawn recruitment increased sharply in 1967. Generally, in this environment, conditions that resulted in below-average fawn recruitment rates also resulted in above-average natural mortality rates among adult females, and vice versa. When recruitment rate is declining and female mortality is increasing, and a management goal is to stabilize or increase numbers of deer, the only option is to curtail hunting mortality of adult females as rapidly as possible. Previously, it was assumed that recruitment was declining and 305 adult mortality was increasing because of deficiencies in the quantity of deer forage on the area. Thus, further reduction in deer numbers by continued heavy hunting (as occurred during 1972-74) should have improved forage conditions at least on a per capita basis and resulted in compensatory increases in recruitment and decreased natural mortality. Our data indicate that neither occurred to a significant degree; rather hunting mortality was mostly additive to other mortality, and served to further reduce the population. Although predation rates were generally low for adult females, predation was the second most common cause of mortality for adult females. Most coyote predation on adult females occurred during winter and spring; though at least some occurred during summer and autumn. Over-winter mortality for adult females, from all sources, was usually less than 10%, exceeding that level only during 1971-72 (24.8%) and 1975-76 (11.3%). Even if all winter mortality was the result of coyote predation, its impact normally would be relatively low. However, over-winter mortality and predation rates of more than 5% on adult females, although by itself not of great impact, indicated that mortality and predation rates for fawns were usually high and that recruitment would be low. Adult females in both relatively good and relatively poor condition were killed by coyotes. Those in poor condition may have died in the absence of coyote predation and their loss may have substituted for other mortality. Most of the adult females killed by coyotes during 1975-80 were in relatively good condition. They probably would not have died in the absence of coyote predation and their deaths probably helped contribute to the low, stable populations of the mid-late 1970s. They were certainly in better condition than many females that survived winters 1983-84 and 1984-85. Both hunting mortality and coyote predation reduced the numbers of adult females, but, during most years, hunting had a greater numerical impact than coyote predation. During years that harvests of antlerless deer were legal, hunting usually removed at least twice and occasionally up to 5 times as many adult females from the population as other sources of mortality. The effect of varying levels of adult female mortality on population numbers depended upon fawn mortality and recruitment rates. Even relatively low mortality rates for adult females contributed to declining or stable populations when recruitment rates were as low as or lower than female mortality rates. During 1972-1977, because of very low recruitment of fawns (15-55 yearling females/year), even relatively low coyote predation rates on adult females contributed significantly to declining or stable populations. This was 306 especially true during 1972-74, when losses of adult females to predation were added to hunting mortality. Even at the low densities of 1972-77, compensatory increases in recruitment did not occur. Mortality of Adult Males Hunting was the major cause of mortality for adult males; few died from other causes . Harvest rates for adult males averaged 37.5% from 1960-1986, ranging as high as 58%. During the years that marked males were present, composite samples indicated that mortality from hunting averaged 36.4% for yearling males and 53% for males 2 years old and older. Sex ratios of adults averaged 42 males: 100 females pre-hunting season (range, 20-64:100) and 31 males: 100 females post-hunting season (range 13-50:100). Male:100 female ratios for some unhunted deer populations (Martinka 1978, Gavin et al. 1984, Kie and White 1985) were no higher than for this hunted population during many years. Gavin et al. (1984) indicated that the annual mortality rate for adult male Columbian white-tailed deer was 40%, even in the absence of hunting mortality. They suggested that increased energy utilization associated with the breeding season resulted in poor condition of males and subsequent winter mortality. Similar conclusions were made by Flook (1970) to explain the higher natural mortality rate of adult male elk compared to adult female elk in unhunted populations. The relatively low rates of winter mortality for adult males in the population during 1960-1986 (3.8%) and for marked males (8.7%) during 1977-1986 (marked sample was older than the male population as a whole, 1977-86), suggested that hunting mortality may be less additive to winter mortality for adult males than for females or fawns. Harvest rates were higher for males than for antlerless deer, so a greater portion of those destined to die over-winter were harvested. The older, dominant males, most likely to be in poorest condition after the breeding season, were also those most heavily harvested by hunters . During many years, especially those preceding severe winters, lower harvest rates for males might not result in appreciably lower total annual mortality rates. We do not know what the natural mortality rates of adult males are in this population in the absence of hunting, but it is unlikely the annual increase in survival would exactly match the decline in harvest rates. If, in the absence of hunting, annual natural mortality rates of adult males in our population were as high as the 40% observed by Gavin et al . (1984), then the majority of hunting mortality of adult males at current levels (especially that of older males) may not be additive to winter mortality. However, because not all 307 hunting mortality of males is likely to be compensatory, at least some increase in survival of males should occur following a reduction in harvest rate. Emigration and Immigration Relatively high rates of emigration (>50%) occurred among yearling males during all years. Because our data did not indicate annual declines in male populations owing to emigration, yearling males from adjacent areas must have immigrated to the study area in nearly equal numbers to emigrants. Dispersal of yearling males can, thus, be viewed as a mechanism for redistribution of males for genetic interchange and allocating resources by sex, but not as a factor that increases or decreases numbers of yearling males within our study area. Emigration of yearling females was much more variable and apparently not balanced by immigration. The majority occurred at low population densities, among yearlings in the first 2 numerically large cohorts recruited following the population low in the mid 1970s (those born in 1979 and 1980). Neither emigration nor immigration were detected among marked deer or in population estimates and classifications during years of mid to high population densities (among cohorts born between 1981 and 1985). A probable temporary movement of adults and fawns onto the study area from surrounding prairie during the dry autumn of 1983 was not considered equivalent to permanent immigration or dispersal of yearlings. A few females and their fawns were also observed to temporarily move off the area for a few weeks during some years . Net emigration apparently also occurred among yearling females born in 1986. One of the 2 radio-collared yearlings on the area in spring 1987 was known to have dispersed to the north off the area; the fate of the other was unknown, but radio contact was lost during the period when dispersal occurs. Although this sample was too small to indicate the degree of dispersal during spring 1987, it did indicate that emigration occurred. Subsequent population estimates not only documented that emigration of yearling females had occurred, but also that it was extensive and involved up to 49% of all females surviving from the 1986 cohort. Overall, about 215 females "disappeared" between spring 1987 and December 1987. About 100 of these could be accounted for as hunting mortalities during autumn 1987, most of the remaining 115 had to have emigrated. Total female density, including all yearlings from the 1986 cohort, was the highest ever recorded (3.3/km ) in spring. Net emigration reduced that to about 2.8 adult females/km2 by autumn, equivalent to the previous high in 1983. 308 The available evidence indicated that net emigration of yearling females occurred at the lowest and highest female densities observed during the study, but not at intermediate densities. Considering the small samples of marked yearlings, it was possible that dispersal occurred but was not detected during years of moderate to relatively high female density. However, because female population estimates generally were closely consistent with numbers of yearling females recruited, any emigration that occurred during those years had to have been balanced by immigration. The observed pattern of emigration among yearling females was similar to the pre-saturation and saturation dispersal proposed by Lidicker (1978). It was not consistent with the "social fence" hypothesis proposed by Hestbeck (1982) in that it did not explain the net emigration at high population density. This inconsistency may be explained by the habitat and social behavior of deer on our study area. The southern, southeastern, and southwestern portions of the area were bounded by blocks of very marginal or non-deer habitat, across which yearlings could disperse and not encounter aggressive "neighboring" deer until they were a significant distance from their natal range. Also, variation in size of social groups apparently depended more on fawn survival than on deer density, and dispersal did not necessarily consist of movement to just the nearest available, open habitat. At both low and high densities, dispersers moved long distances within a day or two, by-passing interaction with close neighbors. Saturation dispersal (Lidicker 1985), occurs at densities at or above "carrying capacity"; pre-saturation dispersal occurs before "carrying capacity" has been reached. Reasons advanced for pre-saturation dispersal include genetic selection and availability of unfilled habitat. Although it is possible that "habitat fill" was the motivating factor behind dispersal of yearling females at high densities, previous social experience could also explain their dispersal at both low and high densities. Occupation of vacant habitat should not occur at 100% efficiency up to the point of habitat fill. Thus, during all years, at least some animals should not find habitat suitable to establish home ranges until they have passed the study area boundaries. If habitat fill was the only factor motivating emigration at high densities, it would follow that at least some, and probably an increasing amount of emigration would occur as density increased (Lidicker 1978). Similar to Gaines and McClenaghan (1980), we did not find that emigration increased proportionally with density. However, it is possible that more intrapopulation dispersal or small shifts in home range occurred with increasing density. 309 Past social experience may explain the variation in rates of emigration among yearling females. Petrusewicz (1963) found that experimentally induced disruptions to social structure could result in changes in density of stable laboratory populations of mice. Population growth could be induced by both removal and addition of several new mice to the population. This suggested that different social structures may tolerate different levels of density. Terman (1962, 1963) for deer mice, Calhoun (1963) for Norway rats, and Hunter and Davies (1963) for Scottish blackface sheep have shown that early social experience is important in determining subseguent spacing behavior. Both periods of high emigration during our study were preceded by at least 2 years of poor fawn recruitment, when few yearling females were added to matrilineal social groups. Under those circumstances, mature females enjoy relative solitude and little social interaction at the time of parturition. A sudden increase in the number of yearlings recruited (spring 1980, 1981, and 1987) could elicit very aggressive behavior by mature females around their parturition territory. On the other hand, after several years of increasing recruitment and the addition of yearling females to social groups during autumn and winter (1981-83), mature females may be more used to the presence of other deer such that yearlings are tolerated nearby as long as they don't approach the new fawn. Thus, past social experience, as influenced by recruitment of yearling females, could explain both pre- saturation and saturation dispersal. During this study, both periods of dispersal occurred during good forage and weather conditions that followed several years of poor fawn recruitment. These conditions served to enhance the opportunity for success by dispersers (colonizers). Regardless of the reason for its occurrence, emigration of yearling females reduced the growth rate of the female population during years that it was documented. During 1979-80, net emigration of yearling females reduced the instantaneous growth rate of the female population from 0.353 to 0.258; during 1980-81 the reduction was from 0.262 to 0.212; and during 1986-87 the reduction was from 0.220 to 0.085. Even though many of those yearlings were incorporated into populations occupying adjacent marginal habitats, their loss reflected "mortality" to the mule deer population on our study area. Overall, emigration did not appear to occur in a manner or to the extent that it would singly control mule deer numbers. Rather, it was one of many factors that at times, reduced populations in core areas below those which would have 310 occurred based on recruitment rates . Perhaps the most important role of dispersal and emigration was to maintain mule deer in vacant available habitat on and off the study area and maintain genetic diversity. "Dispersal is increasingly viewed as having significant demographic causes and consequences" in microtines (Lidicker 1985). Observations of the "fence effect" for at least 6 species of Microtus is evidence that dispersal prevents or slows population growth as long as a dispersal sink is available (Lidicker 1985). Thus, it is possible that some conclusions regarding the effects of increasing density on reproduction and mortality based on studies of deer in enclosures (Kie et al. 1979, McCullough 1979, Ozoga et al . 1982b) may not be entirely applicable to a natural environment where emigration is not impeded. In most studies of "penned" deer, density reaches much higher levels than observed in adjacent habitat where dispersal could act to keep densities below levels that result in density-related nutritional and social problems affecting reproduction and mortality. Similar conclusions might be made about studies (Verme 1965, 1967, 1969) that "artificially" raise density by decreasing forage quantity and quality of penned deer to levels "seldom encountered by free-ranging whitetails" (Verme 1965). Klein and Strangaard (1972) observed that food did not appear to directly regulate roe deer (Capreolus capreolus) numbers in Denmark through malnutrition or starvation except in situations where areas were fenced. Instead, they indicated that dispersal by young animals kept roe deer populations below densities where forage supplies become a factor in population regulation. However, they also noted that social factors controlling roe deer populations would not apply to North American deer because the latter are not territorial and deer habitats are relatively continuous with no low density areas to absorb the dispersing surplus. We now know that those assumptions were premature. Findings of Ozoga et al. (1982a) and this study indicate that some North American deer can be functionally territorial. Similarly, many North American deer populations, especially in the west, are surrounded or interspersed by marginal or low-density habitat capable of absorbing dispersers . Until recently, the potential impacts of functional territoriality and dispersal have been ignored or underestimated. We suggest these impacts must be further examined and considered in research and management of free-ranging populations. Interaction of Reproduction and Mortality Our data (Chapters 5 and 6, Hamlin and Mackie 1987) indicated that a condition cycle in females related to reproduction may contribute to periodic low fawn recruitment 311 and high adult female mortality. Lactation stress associated with rearing fawns to weaning age may be severe (Short et al. 1969, Clutton-Brock et al . 1982), even during relatively good forage years. Females that have recruited fawns to early winter do not accumulate as much body fat as females that did not have fawns or lost them early. Successfully rearing fawns for 2 successive years or more probably requires drawing on body reserves in addition to the nutrition supplied by current forage . The cumulative effect of several years of successful fawn recruitment is relatively poor body condition for the female. The "average" decline in fawn production and recruitment rates that we observed after 6 years of age and increase in female mortality between ages 6 and 7 probably reflected this cumulative reproductive stress. Obviously, not every female on the area died between ages 6 and 7 or did not rear a fawn at age 7. Based on the evidence obtained, however, those females which had successfully reared the most fawns prior to 6 years of age were in poorest condition and thus most vulnerable to either death or poor recruitment of fawns thereafter. The exact age of the female may not be the most critical factor, but rather her reproductive history. On "average", 6-year-old females are most vulnerable, but a 5-year-old female that has recruited fawns for 4 successive years is probably more vulnerable than a 6-year-old female that has only recruited 2 fawns in 5 years. Prevailing environmental conditions, especially during summer and autumn may be important in these relationships. If a dry summer and/or autumn occurs after 2 to 3 years of above average fawn recruitment for the population, females that rear a fawn to weaning, or close to weaning, during the dry year are often in very poor condition. They may be in too poor of condition to breed or to breed at maximal rates such that initial fawn production declines, as was the case during 1983-84 and 1984-85. Additionally, many females may be in such poor condition that they die over-winter. Thus, because of declining female condition resulting from cumulative reproductive stress, a series of years of high fawn recruitment carries within it the seeds of its own destruction. Similarly, the older and dominant breeding males, which are most active during the autumn breeding season, are in poorest condition entering winter and are the males most likely to die over-winter (Gavin et al . 1984). Their deaths occur because of excessive energy expenditure immediately prior to winter, not because of intraspecif ic competition for 312 forage that is inadequate in quality to increase the condition of any deer. Physical condition of females can also decline because of reproductive drain unrelated to increased intraspecif ic competition for food. This scenario can occur even over a long series of years of relatively good forage quality, but is especially likely to occur when a series of good years is followed (as is often the case) by a very bad year. All females are more vulnerable than usual to mortality and low productivity when a very dry year occurs, but those which had been most productive in previous years are most vulnerable. Thus, poor female condition following extended periods of high productivity can explain both the unusually high female mortality that occurred during winters 1964-65, 1971-72, 1983-84, and 1984-85 and the low fawn recruitment observed during the following summers. Fawn recruitment was generally high to weaning during the prior years of 1963-64, 1967-70, and 1977-83, respectively. Often, recruitment remained relatively high during the first year of poor forage following good years. That occurred because, when current forage conditions were not extremely bad, females can rear 1 more fawn crop by utilizing body reserves. This, however, only leads to further deterioration in body condition, even poorer fawn recruitment, and increased adult female mortality the following year. Although superficially, these condition cycles often appeared to coincide with changes in density, density and intraspecif ic competition for food were not the operative factors. Physical condition of females that successfully recruited fawns declined even during what were considered good forage quality years with low deer density on our study area. A decline in recruitment, often drastic, is triggered by a cumulative decline in female condition and/or an abrupt change in forage quality unrelated to an increase in deer density. Condition cycles in adult females are more related to density- independent changes in quality of forage and cumulative fawn recruitment rates than to deer density and forage quantity. Increases in deer density result because fawn survival increases and adult mortality decreases relative to previous levels. Random variation of environmental conditions can result in patterns of fawn recruitment that will mimic what has been viewed as density-dependent recruitment. By random chance, a good fawn recruitment year or series of years is most likely to be followed by at least 1 year somewhat worse than the good year(s). Fawn-rearing conditions can vary considerably across a spectrum, but generally fall within certain limits, depending on environmental variation in the area. Extremely wide environmental fluctuation does not 313 increase variability in fawn recruitment beyond certain finite mathematical and biological limits. Fawn recruitment cannot be less than zero no matter how bad the year and practically cannot be higher than about 1.8 fawns per 2-year-old and older female, no matter how good the conditions. Because of those considerations, we can place fawn-rearing conditions and recruitment levels into limited categories. If forage quality and fawn-rearing years are rated on a scale of 1-10, with 10 being the best and 1 the worst, we can illustrate scenarios indicating that random variation can mimic properties of density-dependence. For example, at the most extreme end, a year that was at the best level for fawn-rearing and recruitment (10) can only be followed by a year that is equal to it (10) or worse (1-9). Random chance dictates that the year following the best year will usually (90% probability) be worse. If year A was an 8 for fawn-rearing conditions, or years A and B were 9 and 8, by random chance the next year is more likely to be worse (1-7, 70%) than equal or better (8-10, 30%). Random environmental variation can lead, over time, to any possible successive combination of years, but regression to the mean and random chance dictate that, on average, relatively poorer fawn-rearing years will follow relatively good fawn-rearing years and vice versa. Thus, it will appear than an increase in deer density (resulting from good years) is followed by a decrease in recruitment. Also, the increase in non-productive yearlings automatically lowers population recruitment rates to some extent. Alternatively, a year or 2 of good fawn recruitment often follows several years of declining deer density (resulting from poor recruitment) . When observed over the short-term, these scenarios appear to represent density-dependent fawn recruitment. The same scenarios, however, can result from randomly varying environmental conditions and regression to the mean. Any data purporting to prove density-dependent relationships must show that it occurs to a greater or more precise degree than would appear to occur because of random variation. Longer-term studies can place this phenomena in perspective and indicate, at least on this area, that similar levels of increases and of decreases in recruitment occurred over a wide range of densities. The importance of long-term studies (20 years +) in interpretation of cause and effect can be illustrated by our data. A study on our area starting in 1973 and continuing for 7 years (through 1979) would have indicated that fawn survival increased as density increased (Fig. 11.2). We do not mean to imply here that fawn survival does increase with density, only that it is not necessarily related to density. Most wildlife/population ecologists consider 7 years to be a long-term study. Certainly most published studies, including most indicating support for 314 density-dependent regulation are shorter than 7 years. Only recently have studies of adequate length been available that properly question existing concepts. In the Breaks environment, the physical condition of females declined with age and following several years of above average recruitment rates, regardless of forage conditions. We found no evidence that intraspecif ic competition for forage played a major role in that condition cycle. Reduced initial reproductive effort and poor survival of the subsequent fawn crop often go hand-in-hand because of the female condition cycle and because, on average, poor forage years tend to follow good forage years. Obviously, those females attempting to recruit fawns during periods of poor forage conditions will be in worse physical condition than during times of good forage conditions, but body condition for productive females declines even during good forage years if they rear fawns. On the other hand, females that do not rear fawns and do not undergo lactation stress can maintain and often even improve body condition during relatively poor forage years. The length of winter through its effect on plant dormancy and the severity of winter through its effect on energy expenditure also affected deer condition and the interaction of reproduction and mortality. 120 -i 110- i- CD 100- E CD 90- O CO 80- Q ■ 70- CO CO 60- E Q> 50- U_ o 40- o *" 30- CO c 20- 5 CO 10- r = 0.93 0 400 500 600 700 800 900 1000 Total Number of Deer - December 1100 Figure 11.2 Fawns observed per 100 females during December plotted against the total number of mule deer in the population during December, 1973-1979. 315 Summary To summarize, then, many different factors influenced and/or interacted to influence fawn production and recruitment and adult mortality. Summer forage condition, especially forage guality as determined by the availability of succulent green vegetation during spring and summer, was extremely important. Forage guantity was generally adeguate during all seasons and only incidentally related to population performance. Both forage guantity and guality appeared to be controlled primarily by climatic variation and succession; deer densities and grazing or browsing by deer had no detectable effect. Climatic factors also determined the timing and length of periods of both positive and negative energy balance. Weather conditions, especially precipitation received during the 9 months prior to initiation of plant growth and temperature during the growing season, determined total forage production and length of time that forage plants remained succulent and provided guality forage. During some years, heavy late summer or early autumn rains result in an autumn "green-up, " prolonging the period of positive energy balance. Weather factors influencing variability of onset and end of winter and severity of winter all affected the length and severity of the period of negative energy balance. Forage conditions from spring through autumn, together with current and past history of lactation stress, established the condition of the female entering winter. The consequence of the condition in which deer enter winter during any particular year depends .on the length and severity of the winter period. Length and severity act together, but length acts primarily in determining how long deer must use maintenance guality or poorer forage, whereas severity determines the degree of energy deficit relative to available forage and fat reserves. Possible summer-winter energy balance combinations are as variable as the weather. In one sense, net recruitment of deer to the population is the result of a net annual positive energy balance, and net deaths in the population are the result of a net annual negative energy balance. Predation rates on mule deer by coyotes appeared to be closely related to cycles in microtine rodent and lagomorph populations, which served to buffer predation on deer during periods of relative abundance. However, high prey populations may also result in high coyote populations that persist for a year or 2 after the "crash" of the small mammals. Within those underlying cycles, the exact level of lows and highs in alternate prey populations is probably set by forage/cover conditions influenced by climatic variation. In addition, the level of coyote predation during winter can be influenced by variable snow and ice conditions that determine accessibility 316 of alternate prey and create advantageous hunting and killing conditions for coyotes preying on deer. Net emigration of yearling females during some years, although enhancing adjacent populations in marginal habitat, resulted in "mortality" or reduced recruitment to the female population on the study area. Induced primarily by behavioral factors, emigration may act periodically as a natural population regulating mechanism by reducing population growth rate following several years of low fawn survival at either low or high female densities. Overall, however, it did not appear to be of singular importance in limiting numbers and the population dynamics of mule deer on the area. Rather, dispersal in the early years of increase following population declines or lows may benefit recovery by rapidly refilling vacant habitats capable of sustaining and producing deer during periods of favorable environmental conditions, both on and off of the study area. This, at high densities, may also serve to maintain deer in more marginal habitats during years of good conditions. Considering all other factors affecting the population, hunter harvests that were not adjusted to periodically low recruitment and increased natural adult mortality rates directly reduced the population and influenced trends and dynamics over periods of one to several years. Although hunting generally did not significantly influence fawn recruitment rates, its role in reducing the number of adult females on the area, in conjunction with other mortality factors, did influence absolute numbers of fawns/yearlings recruited and, thus, also population trends during some periods. The fact that hunting mortality, especially among adult females, was largely additive to other mortality indicated that hunting can be an important regulatory factor or mechanism in the population at times. However, its overall effect in determining trends through the years was minimal relative to environmental variation. Although any attempt to model the role or importance of various single factors on the population is unrealistic, considering all interactions involved, it may be somewhat indicative of the relative contribution of some factors to trends we observed. For example, for 7 years from December 1980 through December 1987, the population increased from 1,175 to 1,405 mule deer, though not in a straight-line manner (Fig. 4.2). Despite this growth, many "potential" deer were lost to the population. If no mortality other than a few deaths from accidents and old age had occurred and reproduction had been at the maximum observed rate each year, simple arithmetic modeling shows that the population could have increased to 38,840 deer 317 at the end of 7 years. Coincidentally , this is also an average annual growth rate of r=0.50, equivalent to the maximum expected for the population (see Chapter 4). By additional "modeling," we can compare the relative contribution of each of several major mortality factors in limiting population growth from the potential to that actually realized. If no emigration of yearling females had taken place during 1980-1987, but all other mortality factors had operated at rates observed annually, the mule deer population would have been 1,725 in December 1987 (1.23 times greater). If reproduction had occurred each year at the maximum observed rate, but all mortality factors had remained as observed, the population would have reached 1,780 deer in 1987 (1.27 times greater) . Similarly, assuming that all other mortality factors operated at observed rates, if there had been no hunting mortality of antlerless deer, the population would have reached 2,155 deer in December 1987 (1.53 times greater). With no hunting of any deer, it could have reached 2,940 (2.09 times greater) . With no coyote predation on any deer, it should have reached 9,540. Although coyote predation was the proximal cause of loss of most of the potential, and in its absence the population could have been 6.8 times higher in 1987 than actually observed, the interaction of all factors was even more important. Without all factors interacting, the population would have been 27.6 times higher than actually occurred. Obviously, all other mortality rates would not have remained the same if any factor had been eliminated. The elimination of even the factor with the least influence, emigration, would have resulted in higher populations in December than we ever observed on the area. Thus, results equivalent to "penned" deer studies may have occurred, and at higher population levels, we may have observed density-related effects on reproduction and mortality. The important implication is that the combination of all factors (multi- factorial approach, Lidicker 1988) apparently has kept the population below levels where density-dependent mortality is easily observable. Probably, no 1 factor can be said to regulate the population unless we call the total variable environment 1 factor. 318 CHAPTER 12 POPULATION REGULATION Thus far, we have discussed what the population did, how it did it, and to a limited extent, why it did what it did. We now concentrate more fully on the latter and the questions of why the population behaved in the observed fashion, why abundance varies from place to place, and why no population increases without limit (fluctuates). In the context of population ecology these are the essence of population regulation or limitation. The population we studied displayed an initial growth pattern (Fig. 4.7) that was similar to a "typical" logistic growth curve, implying density-dependent processes. Indeed, the representative of density-dependent regulation. The key point, however, is that almost all the theoretical work focuses on the initial "irruptive", logistic growth phase. Subsequent fluctuations, if considered at all, are passed off lightly as "minor" fluctuations about "carrying capacity" (K). Seemingly, it is assumed that similar processes operate within these fluctuations as within the initial growth phase. Our data collection and mathematical tests occurred entirely within the fluctuating phase of an established population. What may be termed minor fluctuation by theorists or mathematical ecologists is of major concern to population managers. During 1972-77, for example, neither deer managers nor hunters in eastern Montana were pleased with ongoing "minor" fluctuations in mule deer numbers. Compensatory reproduction and mortality (density-dependent processes) were nowhere evident. Thus, the question was not what processes operate in founding populations, but what determines fluctuations of established populations. Fluctuations in established populations might not result from density- dependent regulation. If that is so, the implications to management are entirely different than those inherent in density-dependent regulation theory. Factors regulating populations can generally be placed into 2 categories; intrinsic and extrinsic, though there are variations and combinations (see Chapter 1). Proponents of population regulation by intrinsic factors or processes hold that there is self -regulation by the population that prevents it from reaching densities where it harms or destroys its resource base. Various intrinsic mechanisms, operating in a variety of ways have been proposed—behavior , physiological change, genetic selection. All key on population density as 319 the triggering cue and thus, all are variations of density- dependent population regulation. We could find no evidence to support the theory of self- regulation by populations. Mule deer on our study area employed a range of intrinsic behavioral and physiological mechanisms to deal with the variability of extrinsic factors. Variable emigration, food habits and habitat use, and movement strategies were all examples of such mechanisms. However, these intrinsic mechanisms by themselves, did not function to regulate population growth. Rather, they dampened the effects of variable extrinsic factors. Thus, population growth was not as fast nor decline as severe as might have been the case in their absence. It was also unnecessary to invoke intrinsic mechanisms to explain major changes in the population. Behavior that contributed toward slowing population growth (parturition territoriality and emigration) occurred at all densities and did not halt population growth. A condition cycle (another possible intrinsic factor??) that explained some variations in reproduction and mortality was influenced to a large extent by variable weather (an extrinsic factor). Our conclusion is that intrinsic mechanisms interacted with extrinsic factors to moderate population growth and decline, but did not directly regulate population growth. The role and importance of extrinsic factors in regulating or limiting population growth has received the greatest attention from population biologists . Various extrinsic factors (e.g., weather, food, space, predators) have been suggested to regulate populations in either a density- dependent or density-independent manner, or both. Regulation by resource shortages (especially food) through density- dependent feedback (Nicholson 1933) is widely accepted. However, limitation of population growth through density- independent changes in weather (Andrewartha and Birch 1954) also has its adherents. We found that weather affected the deer population in various ways, both directly and through influence on other extrinsic factors such as food, space, and predation. Our data clearly showed that fawn recruitment was directly influenced by forage conditions (see Chapter 5). Through a delayed condition cycle, adult mortality was also influenced by forage conditions. We found no evidence, however, that forage production, availability, or conditions were substantially influenced by deer density. Thus, we also found no evidence for population regulation through a density- dependent feedback loop. Rather, the primary influence on forage conditions was density-independent changes in weather. This finding is strongly emphasized because it is a major 320 departure from the explicit or implicit assumptions in density-dependent theory that the number of animals grazing or browsing is a primary factor influencing forage production and quality. The influence of weather on vegetation production also influenced cover for deer and forage and cover for alternate prey (microtines and rabbits), thereby influencing coyote predation rates and deer mortality. Predation rates were also influenced by variable snow and ice conditions. Weather also influenced the energy balance equation of deer both through forage production, quality, and availability and through its influence on energy demand by deer. Through interaction with the fixed habitat base, weather (wind, snow depth, forage quantity and quality) affected the kinds (reproductive, non-reproductive, long-term, temporary) and amounts of effective deer habitat (space). Deep, drifted snow resulted in some areas becoming non-habitat (nutritionally, excessive energy use, vulnerability to predation) during some winters. Similarly, during drought, some areas could not provide even maintenance quality forage, or their use required behavioral changes in deer that increased their vulnerability to predation. Because of the pattern of deer dispersion at habitat fill, more deer occupied marginal areas at high than at low densities. Thus, mortality could be high at high densities (density-related) under certain weather conditions. However, this relationship was not deterministic or consistent. When weather conditions remained favorable, mortality was low even at high densities. Overall, the variable weather, interacting with the fixed habitat base was the most important factor influencing the deer population. This occurred because weather also influenced all other factors including food, energy demand, predation, available habitat (space), and even hunter harvest. Within limits of the fixed habitat base, population growth was limited by the time that weather conditions were favorable and population decline was limited by the length of time that weather conditions were unfavorable. The role and importance to population dynamics of proximate factors such as predation and hunter harvests has been discussed earlier (Chapters 5, 6, and 11). Coyote predation was the major proximal cause of deer mortality. Although, it appeared that predation often functioned as an agent of variable weather, the absence of predators would probably result in at least temporarily higher deer densities. At these higher densities, it is possible that density- dependent effects on reproduction and mortality might be more 321 observable. Because hunter harvest may be extraneous to the development of "natural" regulation of populations, it will not be discussed here. However, hunter harvest is the major factor we can control or manage and will be further discussed in Chapter 13-Management Implications. The finding that variable weather is the major factor influencing forage production and quality and population dynamics of herbivores is not new (Talbot and Talbot 196 3, Teer et al. 1965, du Plessis 1972, Hirst 1975, Sinclair 1977, Medin and Anderson 1979, Croze et al . 1981, and Rogers 1987) However, the tendency has been to assume almost perfect operation of density-dependent reproduction and mortality despite non-density related changes in forage conditions. Under this assumption, reduction in density always increases nutritional level and survival. In contrast, we maintain that under unfavorable conditions the condition and fates of individual animals vary as individuals, not as l/N. Thus, during drought, a reduction of density by one-half as a result of harvest does not mean that the remaining animals will all survive because they now have twice as much to eat. In some drought situations, any surviving animals have only sub- maintenance quality of forage to subsist on and quantity is not important. A careful reading of numerous papers (especially in combination) indicated to us that weather related changes in forage abundance and quality and animal distribution often influenced large mammal population dynamics regardless of density. Low density populations did not perform well during unfavorable conditions, but high density populations could perform well during favorable conditions. This seemed to be true across a wide variety of species and locations: wildebeest in Masailand (Talbot and Talbot 1963) and Kenya (Croze et al. 1981), white-tailed deer in Texas (Teer et al . 1965), blesbok in South Africa (du Plessis 1972), African buffalo in the Serengeti (Sinclair 1977), black bear in Minnesota (Rogers 1987), mountain sheep in California (Wehausen et al . , 1987), and mule deer in Montana. Talbot and Talbot (1963) reported that drought was the primary factor controlling reproduction of wildebeest in western Masailand. They stated (p. 71): "Low nutrition plane in wildebeests appears to be caused by lack of green forage. Wildebeests in western Masailand, being free to migrate widely following rains, showed no evidence of low nutrition. Wildebeests on the Athi-Kapiti plains, however, being restricted throughout the year to their former wet-season range, show evidence of low 322 nutrition affecting reproduction even under relatively mild conditions." Similarly, Teer et al. (1965:56) stated: "Changes in carrying capacity occur rapidly and frequently, and are often independent of density of grazing herbivores." Du Plessis (1972) noted that "Blesbok lose condition during the dry season." Rogers (1987) reported that reproduction in black bears was controlled mainly in a density-independent manner by fruit and mast supplies that fluctuate in abundance from year to year. Although the influence of various factors on population dynamics sometimes varied with density, density was not a consistently regulating factor nor did it function deterministically . Often, density was only a coincident factor, thus regression analyses and correlations can result in false conclusions about cause and effect. Population growth and decline were both limited by the length of time that extrinsic factors (especially weather) remained either favorable or unfavorable. This, together with the behavioral and biological capabilities of the animals, determined the "distribution and abundance" of the deer. It is clear that application of regulation or limitation theory must take variable exogenous factors into account. These factors, for many populations, tend to hold them below densities where density-dependent processes might operate. To date, with rare exceptions, too little emphasis has been placed on variable environmental conditions and heterogenous habitat. The natural tendency of man is to lump, type, or categorize, thus variability has been considered the exception. We maintain that variability and heterogeneity are more pervasive and applicable than stable equilibria. The importance to population dynamics of heterogeneity of the environment in both space and time interacting with characteristics of the animal species will perhaps be more clearly portrayed by use of illustrations. Heterogeneity of habitat quality and quantity in space is illustrated by Figure 12.1. The relative mix and juxtaposition of types of habitat quality and quantity will vary from area to area, by species, and by ecological amplitude of the species. Despite and because of this variability, the major point of Figure 12.1 remains; habitat is heterogeneous in space, not all space is equal in habitat quality and quantity. The performance and fates of individual animals occupying individual home ranges will vary as individuals, not as l/N. Habitat quality and quantity are also heterogeneous in time (Fig. 12.2). Habitat quality and quantity in the 3 different time periods illustrated (Fig. 12.2 A, B, and C) do 323 o ••; unoccupied yearlong, long-term deer habitat occupied yearlong, long-term deer habitat deer habitat during favorable environmental conditions and/or with use of seasonal migratory behavior (unoccupied) as above, except occupied core area Figure 12.1 A schematic representation of the variability in space of habitat quality and quantity. not necessarily follow consecutively. Any of the scenarios illustrated could follow any other.- Variable weather can result in rapid changes in habitat quality and quantity from year to year and season to season. Variability in weather not only results in rapid changes in forage quality and quantity (Chapter 3) and thereby fawn survival (Chapter 5), but also by interacting with the fixed habitat base, it influences whether space is reproductive, survival, or non-habitat (Chapters 3, 5-10). Thus, habitat quality and quantity are variable in time, especially as influenced by the inherent variability of the weather, but also as influenced by succession, fire, and many of man's activities. This heterogeneity of habitat in space and time interacts with the inherent characteristics of dispersion and dispersal for any animal species to further influence population dynamics. A hypothetical species distribution at a population low is illustrated in Figure 12. 3A. This distribution and population low could have resulted from a variety of causes (e.g., drought, severe winters, or overharvests ) . Given favorable environmental conditions and/or low harvest rates, the distribution of the species can progressively expand toward "habitat fill" as illustrated by Figure 12. 3B and C. 324 B. • - reproductive habitat O - non-reproductive survival habitat .;;;. - non-reproductive, adult mortality vulnerable habitat .;•;. - reproductive and/or survival habitat with specialized movement strategies Legend as in A. Legend as in A. Figure 12.2 A schematic representation of the variability in time of habitat quality and quantity. Figures A, B, and C represent independently different time periods, they are not necessarily consecutive periods. 325 o B. occupied yearlong, long-term habitat unoccupied yearlong, long-term habitat habitat during favorable environmental conditions and/or with use of seasonal migratory behavior (occupied) as above (unoccupied) Legend as in A. C. Legend as in A. Figure 12.3 A schematic over time conditions . habitat by represented population near habitat fill (C) representation of habitat fill given favorable environmental The progressive filling of dispersion and dispersal is in consecutive order from a low (A) through mid-fill (B) to 326 The species social structure and behavior (Geist 1971) along with its reproductive potential and morphological, physiological, and genetic characteristics determine its rate of dispersion and dispersal. Thus, given favorable environmental conditions, distribution of microtine rodents might progress from A to C and beyond (Fig. 12.3) in a year or less. For deer (Odocoileus spp.) and moose (Alces spp.), 4 to 5 years might be necessary for expansion of their distribution from that represented by Figure 12. 3A to that in Figure 12. 3C. An equivalent expansion in distribution might take decades for species such as mountain sheep (Ovis spp.) (Geist 1971). Clearly then, the species rate of dispersion and dispersal is another time factor important in population dynamics, particularly in variable environments. Species with rapid rates of dispersion have the potential to reach "habitat fill" (Fig. 12. 3C and beyond) during periods when environmental conditions remain favorable. These species are more likely to at least occasionally reach "habitat fill" and perhaps display "density-dependent" dynamics. Species with relatively slower rates of dispersion may seldom display density-dependent dynamics because environmental conditions suddenly may become unfavorable before sufficient time has passed for them to reach "habitat fill". The relative stability of different environments is also an important consideration in determining population dynamics. Thus, relatively stable environments are more likely to provide adequate time for "habitat fill" to be achieved while more variable environments provide less time under favorable conditions for "habitat fill" to occur. This combination of factors leads us to the hypothesis that species with rapid rates of dispersion that live in stable environments are more likely (though not guaranteed) to reach densities where a significant degree of "density-dependent population dynamics might be observed than species with slower rates of dispersal that live in variable environments. A counter to the foregoing argument might be that species with slower rates of dispersion and dispersal are more likely to increase within place, placing increased pressure on local resources. Thus, these slower dispersing species might be most likely to exhibit density-dependent dynamics. This argument certainly needs more consideration and investigation. Tentatively, however, our observation is that slower dispersing species are usually those which also have lower reproductive output. Thus, environmental conditions usually turn unfavorable before local densities have the opportunity to reach extreme levels. Also, species populations that live in variable environments tend to have rapid, not slow rates of dispersal. Dispersal rate may also vary within a species, depending on the environment in which they live. 327 An alternative way of viewing population dynamics in variable environments is that "carrying capacity" (in whatever manner it is defined) fluctuates widely and unpredictably. Thus, some may assert that although measurements using absolute density may not show predictable density-dependent relationships, population dynamics may respond to relative density. Our data did not show a response to relative density for the Missouri River Breaks mule deer population. Furthermore, if fluctuation in "carrying capacity" is primarily caused by density-independent factors (as we observed) , then there is no feedback loop, hence no regulation. Why was this population limited by "random" or chaotic extrinsic factors rather than density-dependent regulation? Actually, our results are not unusual for free-ranging populations. There has been no substantial evidence presented to "prove" density-dependent regulation for free-ranging "natural" populations (see Chapter 1). The only substantial evidence for density-dependent regulation has come from laboratory, penned, introduced, or island populations and possibly from populations for which predators had been eliminated. Cohen et al . (1980) recognized this when they stated (p. xv) : "Laboratory studies ... tend to present a picture of populations well regulated and largely self- regulated. .. Because food, predation, and disease [also weather and emigration] are not allowed to act as limiting factors, exogenous controls appear unimportant, and since random environmental events are prevented from impinging, eguilibrium states are readily observed." They further stated (p.x): "In fact, an informed consensus of the participants in the symposium, pre-selected in large part for their interest in internal regulation, was that there is enormous variation both among species and among habitats in the relative importance of self- regulation; that .. .mechanisms of self-regulation are at some level probably fairly common among animal species; but that for most species they may only rarely come into play in periods when exogenous forces do not first intervene." Similarly, Botkin et al . (1981:373) state: "One of the problems with much existing population theory for large mammals is its failure to view the animals within an ecosystem context. Standard 328 approaches ... logistic, Lotka-Volterra, Leslie Matrix are deterministic and abstract, ... populations are treated independently of most of the variability of the environment and biographical forces . " Thus, although various studies provided evidence for density-dependent regulation under precisely specified and often unrealistic conditions, management application of the principles derived has often proven inappropriate or even disastrous for free-ranging populations. Wildlife managers have actually paid too much attention to population theorists. Similarly, impressions left by early writers, when left unchallenged for decades, became unconsciously assumed "law". For example, one of the major themes of the writings of Malthus and Darwin is that of the tremendous reproductive potential of species. We have been unconsciously left with the impression that populations are always straining against "carrying capacity". This often unrecognized assumption has been inherent in much of population theory despite the fact that "carrying capacity" was not defined or measured. When "carrying capacity" was defined or measured, it often turned out that populations were not at or near "carrying capacity". Just awareness of the possibility that populations are not always pressing against "carrying capacity" or "trying" to exceed it can lead us to new ways of thinking about population regulation and especially management. We believe we have shown that a multitude of factors can and do affect population dynamics. The relative impact of these factors can vary among species and among species populations inhabiting different environments. For the mule deer population inhabiting the widely fluctuating environment we studied, important limiting factors included: the length of time green forage was available, the energy demand by deer, predation, emigration rate, and hunter harvest. All these factors were influenced or controlled to some degree by weather, the first 3 especially so. All limiting factors interacted such that the opportunity to observe density- dependent regulation did not occur. It is possible that conditions leading to density- dependent regulation could occur, given changes in the environment. For example, limitation of emigration by fencing or lack of dispersal sinks, annihilation of predators, and/or cessation of hunting singly or in combination could result in density-related resource shortages leading to poor condition. Thus, population dynamics could be altered toward density- dependent regulation. We believe, however, that many, if not most, free-ranging "natural" large mammal populations are usually limited by variable extrinsic factors below densities 329 where potential density-dependent regulation might be observed . The conclusions we have reached about the lack of a neat and tidy "balance of nature" or "regulating mechanisms" may be unsettling to some. There is, however, no reason for panic; "chaos" is not as intimidating as it first seems. It is, perhaps, a desire of most of us that there be order, measurability , and predictability in our world. This desire may have even affected Albert Einstein's lack of acceptance of the uncertainty inherent in the guantum theory that eventually evolved. This view reportedly led him to state emphatically to Niels Bohr that "God does not play dice [with the universe]" (Calder 1979). More recently, Joseph Ford, a physicist at the Georgia Institute of Technology, has answered Einstein (guoted by Gleick 1988 p. 314): "God plays dice with the universe. But they're loaded dice. And the main objective of physics [or ecology] now is to find out by what rules were they loaded and how can we use them for our own ends . " Explanations for the behavior of weather, a major factor influencing population dynamics, have come more and more to rely on "chaos" theory (Gleick 1988). What at first appears to be chaos, is upon further examination, channelized randomness or ordered disorder (Gleick 1988). This "channel" of variation (or loading of the dice) is measurable and environmental variation and population level and dynamics will remain within the bounds of that "channel". The lack of evidence for deterministic and consistent self-regulation or other forms of density-dependent regulation of many populations need not be controversial or disquieting. Before, during, and continuing after the times we have argued about mechanisms and theories of regulation, populations have generally varied within natural channels. We have not found the "balance of nature" we sought because we looked in the wrong places. A "balance" does occur--it is in the environment. There, over time, unfavorable conditions are balanced by favorable conditions. Rarely, natural catastrophes may perturb a population outside its "natural" channel. In that case, another natural phenomenon may take place — local or large scale extinction. More frequently of late, it seems, man's activities may perturb environmental conditions beyond the natural channel of animal adaptability. Extinctions that may result in these cases are preventable and thus, are cause for concern. Reliable information on natural environmental variability and animal adaptability and fixed habitat requirements will be necessary to address the impacts of human activities. 330 CHAPTER 13 MANAGEMENT IMPLICATIONS Management Philosophy and Theory As indicated in Chapter 1, most population and habitat management for ungulates has been based on the belief that density-dependent regulatory processes are in fact operating on/within the population and that they apply at all population levels (are logistic). It is most commonly assumed that intraspecif ic density-dependent competition for resources (especially food) defines "carrying capacity" (K). In most models, K is displayed as relatively stable and influenced primarily by the animals through grazing and browsing pressure. Generally, because of their reproductive potential, populations are portrayed as always straining to exceed "carrying capacity" (inherently irruptive). Some of the more common concepts or "principles" derived from density-dependent theory and applied in deer management include: compensatory reproduction, compensatory mortality, maximum sustained yield, intraspecif ic competition for food, winter "bottleneck", "doomed" surplus (insignificance of predation) , finite "carrying capacity", the "law of diminishing returns", and various approaches to habitat management. Many of these concepts, "principles", or "laws" have come to have a life of their own, without recognition of the assumptions behind them. For mule deer in the Missouri River Breaks (MRBMD), density-dependent processes were not consistent, precise, or predeterministic (if they occurred at all). Variable effects of limiting factors led to both low recruitment/high mortality and further population declines at low densities and high recruitment/low mortality and further increases at high densities. "Carrying capacity" was not easily or simply definable whatever criteria were used. Forage production and quality as well as usable space fluctuated independently of density. Population growth rate declined to zero or below at low as well as high densities. It also was explosive at both relatively high and low densities. Expected compensation in reproduction and mortality was not observed. Gradually, it became obvious that management concepts based on density- dependent population regulation seldom applied. Our discussion of management implications and recommendations centers on mule deer in fluctuating environments that include predators and dispersal sinks. We suspect that many of our findings and concepts will apply in some degree to other areas, and perhaps to other species in similar environments. However, we caution the reader against 331 a common problem in converting science to management; the overapplication of findings from one area or species to others . Management Concepts The implications of our findings to management are dissimilar to those usually encountered. Therefore, we believe it is necessary to first discuss some common management concepts in relation to our findings before making recommendations . Management of ungulates has usually centered around the concept of a forage based "carrying capacity" (K) . Because much of the observable natural mortality occurred during winter, it has further been assumed that there was a winter "bottleneck" and that the quantity and quality of winter forage was a key factor determining "carrying capacity" . Thus, "carrying capacity" and population dynamics in deer have most commonly been interpreted to directly relate to the quantity and utilization of winter forage, especially "key" browse plants . Additionally, deer populations have been assumed to be inherently irruptive, which led to "over-utilization" of forage and a reduction in "carrying capacity". Management strategies emphasized the importance of hunter harvests to bring deer numbers into "balance" or "equilibrium" with forage supplies (especially winter forage) and prevent overuse of browse plants. Similarly, it has been believed that populations held somewhat below "forage carrying capacity" were the most productive and would yield maximum sustained harvests . A reduction of the number of deer on "over-used" range would reduce browsing pressure and result in improved forage condition and production, thereby improving fawn production and survival for the remaining deer. Thus, hunter harvests were deemed to be compensated for by increased reproduction (compensatory reproduction) and decreased mortality (compensatory mortality) among survivors. Because deer were considered to be inherently irruptive, both hunter harvests and kills by predators were considered to substitute for (not add to) mortality of the "doomed surplus" from starvation or other causes. For our study area and mule deer population, forage production and condition were primarily determined by factors other than deer browsing pressure (Chapters 3 and 10). Forage production and quality were determined mainly by variation of temperature and precipitation. Deer browsing pressure did not result in long-term deterioration of forage plants and in many cases actually stimulated increased production of browse plants (Mackie 1973b, Peek et al . 1978). Measures of "forage 332 carrying capacity" varied many-fold among years independently of deer density; forage quantity was not limiting during any year or season. Variation in forage quality (nutritional content) among years and the length of time nutritionally adequate forage was available in any year was of major importance to mule deer population dynamics. Because quality of winter forage was typically only of maintenance or lower quality, we determined that forage conditions during spring through autumn were of vital importance. The overall physical condition of deer and amount of fat that they carried into winter comprised a major component of "winter range". Length and severity of winter were important as they influenced the rate of use of fat reserves and the length of time deer were forced to subsist on maintenance or sub-maintenance quality forage . Although the length of time that high quality forage was available had major influence on population dynamics, there was no evidence that intraspecif ic competition for food played a major role in determining nutritional plane of deer. Thus, management concepts tied to intraspecif ic density-dependent competition for food, finite carrying capacity, and "winter bottlenecks" had little applicability to this population. As evidenced by the combination of a variety of literature cited throughout this report (Chapters 1, 3, 5, 10-12), similar conclusions may also be drawn for some other areas and populations . Because intraspecif ic competition for food was not a major influence on population dynamics, the concepts of compensatory reproduction and mortality also are questionable, at least for this population. We were, in fact, unable to measure significant compensation. When nutritional deficiencies are primarily independent of the number of deer on the area (density), reductions in the number of deer do not improve nutritional conditions. It follows that, if nutritional conditions are not improved, reproduction and survival do not improve and there is little or no "compensatory" (increased) reproduction or "compensatory" (decreased) mortality among surviving deer. An additional and separate aspect of the issue of compensation is the question of whether hunting mortality substitutes for or is additive to other forms of mortality. This is the "compensation" typically referred to across all wildlife species. For example, game birds are rarely, if ever, considered to destroy their food base, yet most of their mortality by hunting is considered "compensatory mortality" in the sense that it generally substitutes for other mortality. To keep this form of compensation separated (as it should be) from that addressed in the previous paragraph, we will use the term substitution. Thus, we frame the question: does hunting 333 mortality substitute for or add to natural mortality? Also, does mortality from coyote predation substitute for emigration or mortality from starvation and "old age"? Obviously, at least some hunting mortality substitutes for natural mortality that would occur in any case; the question is how much? Our data indicated that this varies among sex and age classes and probably among locations and species. Because forage conditions for survivors were not improved as the result of death of individuals, substitution probably only functioned on an exact basis. That is, if the specific animal harvested was one that was "doomed" to die overwinter, then that hunting mortality substituted for natural mortality and was not additive to other mortality. On the other hand, the harvest of a healthy, fat 3-year-old female could not and did not prevent a malnourished fawn (or 16-year-old female) from dying. Similarly, the harvest of deer in core habitat did not guarantee survival of deer in marginal habitat. Thus, the degree of substitution depends upon the varying annual percentage of the population vulnerable to "natural" mortality. The impact of additive hunting mortality to the population depended upon subsequent recruitment rates that determined replacement. Thus, even though a varying degree of hunting mortality was additive to other mortality, population declines did not occur when recruitment was adequate to replace that mortality. Because recruitment rates varied substantially among years, the effect of any given rate of hunting mortality also varied among years. Pre-season male: female ratios on our study area were not much different in many years than those reported for unhunted populations. Thus, much hunting mortality of adult males appeared to substitute for high natural mortality which would probably occur in the absence of hunting. Our data suggest that the degree to which pre-season male: female ratios are below about 40:100 may indicate the extent to which hunting mortality of males is additive to natural mortality. Hunting mortality of adult females did not substitute for natural mortality to the same extent as for males because of lower natural mortality among females than males. Also, the removal of adult females by hunting potentially affected future population dynamics more than removal of males because it more directly affected future numbers of fawns produced. Because emigration by yearling females probably increased at high densities, there is likely an increased degree of substitution for hunting mortality of adult females by yearlings remaining in or returning to the population. 334 Because of their higher natural mortality rate, a greater percentage of the hunting mortality of fawns substituted for natural mortality than occurred with adult females. However, the low hunting mortality rate of fawns resulted in little realized substitution on a population basis. As a result of the above, the degree of substitution of hunting mortality for natural mortality (compensation) in the Breaks population probably varied quite widely. Substitution may be greater at high densities or when natural mortality rates are high. There may be almost no substitution when deer numbers are low or deer are in excellent condition. The degree of substitution also varies among sex and age classes and among individuals within those classes. Again, when analyzing all population processes, it is important to recognize that deer are individuals (not l/N) . The impact of predators on the population can be viewed similar to that of hunter harvest, although there are some differences. Like hunting, the extent to which a predator kill substitutes for other mortality will often also depend upon the exact animal killed. In the Missouri River Breaks, coyotes killed much higher proportions of fawns than did hunters. Similarly, coyotes killed higher proportions of older animals that were wounded, sick, or otherwise predisposed to mortality. Thus, a higher proportion of the deer killed by coyotes than by hunters substituted for losses that would have resulted from malnutrition, old age, or emigration. Not all mortality of deer by coyote predation substituted for other mortality, however. Results of this and other studies indicated that coyotes can and at times do kill significant numbers of healthy deer that would have survived in the absence of predation. The concept that predators kill only the "doomed surplus" and do not cause population declines is based on the erroneous assumption that reproduction always produces more animals than the habitat will support. Clearly, mule deer populations in the Missouri River Breaks were not always near habitat fill, and predators did kill animals that were not "doomed surplus". Similar to the effects of hunter harvests, predation by coyotes appeared to have its greatest impact when deer densities were low. This probably occurred because there is little opportunity for substitution at low densities (few vulnerable, old, etc.) nor can good nutritional conditions and good fawn recruitment be guaranteed. Why do our conclusions differ from traditional views about the occurrence and application of compensatory reproduction and mortality? The "principles" of compensation have generally been based, and especially applied, on the assumption that nutritional deficiencies resulting in poor 335 reproduction and survival were always dependent upon density of the study population. Our findings plus those becoming increasingly common in the literature indicate that nutritional deficiencies are not always, or even often, related to density in free-ranging populations. Many people (especially lay persons) have recognized that substitution probably doesn't occur on a 1:1 basis. Early management programs, however, in practice appeared to assume 100% substitution and guaranteed compensatory reproduction and mortality at all densities and times. There was no database to challenge those assumptions until hunting mortality rates began to rise above natural mortality rates. Differences in interpretations about the importance of compensation and substitution are also rooted in the inherent characteristics of the species studied. In hindsight, it is easy to understand some conclusions resulting from early studies of insects, fish, and birds, all of which had high annual or generational rates of natural mortality. Only recently have long-term data become available for larger, longer-lived species with relatively low rates of natural mortality. For a game bird population with an annual mortality rate of 50-80% in the absence of hunting and for which young of the year are not selected against by hunters, a high proportion of hunting mortality might be expected to substitute for other mortality. Should the same degree of substitution be expected for adult female mule deer in the Missouri River Breaks with an average annual natural mortality rate of 7-8%? Obviously not. Thus, the conclusions of early studies were safe until relatively recent times, when hunting mortality began to exceed natural mortality rates by a significant degree in more and more places . The higher degree of substitution (compensation) likely for game bird populations does not mean that additive effects of hunting mortality can be ignored for these species. First, game birds as well as deer must be considered as individuals. Chances of natural mortality vary at least somewhat by sex and age class as well as place of residence. Thus, even for a species with an annual natural mortality rate of 70%, substitution isn't 100% efficient all the way up to 70% hunter mortality. At least some fraction is additive. Second, similar to what occurred with big game populations (and similarly still unrecognized) , harvest rates may be creeping up to and exceeding the natural mortality rate in some locations. If an "unexplained' crash occurs in some local areas in the future, it may not all be explainable by loss of habitat or bad weather. 336 The concept of maximum sustained yield assumes that populations maintained below maximum density are more productive and yield greater sustained harvests than high density populations. We observed 2 major problems with application of this concept in management. First, net recruitment may not be related to population density; for the MRBMD population it certainly couldn't be predicted from density. Thus, yield was not predictable (and sustainable) at any population level. Second, even if density-relationships applied, forage-based "carrying capacity" fluctuated so much that attempts to achieve stable population levels and sustainable yields were impractical. Yields were not sustainable at any population level, but tended to be higher at higher population densities (Chapter 5). The "law" of diminishing returns also needs further discussion. At times, we have heard this concept confused with compensatory mortality. The statement usually goes something like this: "There is compensatory mortality because if deer become too scarce in one area, hunters compensate by moving to another area, thereby lowering mortality in the former area." The concept of diminishing returns and the extent to which it may be operative need to be updated. Long ago, Aldo Leopold (1933) recognized that the level of operation of "diminishing returns" and hunter effectiveness are variable when he stated: "gasoline has not lengthened the tether of the bobwhite". Habitat security is much more than vegetation composition, interspersion, and topography. Changes in access, hunter affluence, mobility, leisure time, equipment, and desire can alter habitat security as surely as changes in vegetation composition. What was "secure" habitat yesterday may be insecure today even without environmental change. Reliance on the "law of diminishing returns" to affect changes in hunter distribution and hunting pressure can result in hunter dissatisfaction (Wood et al . 1989). Use of the "law of diminishing returns" as a management tool will result in unhappy customers and would not be employed in competitive businesses. Finally, we address concepts of habitat management. Because of the overriding influence of the concept of a forage-based carrying capacity, most habitat management for ungulates has focused on practices that "improve" forage conditions, particularly on winter range. Indeed, the philosophy of reducing deer populations to at least somewhat below "carrying capacity" has been considered habitat management because it is intended to improve forage production and the condition of forage plants. The U.S. Forest Service often justifies timber cutting on the basis that it improves forage production for ungulates. Many State Game Departments buy land (especially winter range) for the purpose of preserving and enhancing forage for ungulates. Although 337 forage is an important consideration in habitat management, the ways in which it is important include more than simply quantity, and other aspects of habitat may be of equal importance in management. For example, more recently, with the advent of heavier hunting pressure, the importance of cover has also received increased recognition. General Approach to Management The general approach to deer management we recommend remains very similar to that outlined in earlier Montana Deer Studies reports (Mackie et al . 1980, 1985). Here, we discuss the conceptual basis for a general approach to deer management. Findings of this study together with those of other Montana deer studies indicate that each deer population is a unique product of its total environment. The range of characteristics and responses for each important population can be established, indicating the range of potential management options . Our model of population-habitat relationships and dynamics (Fig. 1.4) indicates that the level and variability of resource outputs interacting with animal strategy determines population dynamics. Thus, the first step of management is to gather data on characteristics and variability of both the deer population and the environment. Though both types of data are desirable, one can often be inferred from the other. Population characteristics that should be measured include: population size, sex and age composition, harvest size and composition, seasonal distribution, and mortality patterns. Habitat variables to be similarly quantified include: seasonal and annual temperature and precipitation, snowfall, winter severity, and plant phenology, especially timing and length of the "green" period. In most cases, these data will be available for 10 years or more; if not, collection should begin immediately. Most of the necessary environmental data will usually be available for correlating environmental events and characteristics with population responses. Such correlation will result in some degree of predictability for population responses. Management strategies can be based on expected population behavior and thus, if desirable, can be specific to individual populations. Population Evaluation and Management Harvest Management Harvest strategies and management will vary among individual managers, with the desires of local hunters, landowners, and outfitters, and with local environmental characteristics. Thus, management depends as much upon psychological, sociological, and economic considerations as upon ecological considerations. For instance, some managers and hunters prefer consistency of hunting regulations across 338 areas and from year-to-year, despite the resulting increased fluctuation of populations. Others may accept variable regulations if they result in moderated population fluctuation which make expectations for hunting success more consistent. Although psychological, sociological, and economic aspects of management cannot be ignored, our main purpose is to assure that the biological and ecological background to management is sound. Thus, management decisions will have at least considered realistic biological potentials, and probable population responses of management actions will not be unanticipated. In variable environments like our study area, a harvesting strategy based on average recruitment and mortality rates or the role of compensatory reproduction and mortality will seldom track population trend. During extended periods of below-average recruitment and above-average natural mortality, the population will decline under an "average" harvesting strategy. On the other hand, during extended periods of above-average recruitment and below-average natural mortality, the population will increase. If sufficient dispersal to vacant habitat has occurred and the goal is to stabilize populations, above-average harvest rates must be implemented during increasing phases. In variable environments, the use of an "average" harvesting strategy increases the fluctuations of a population, rather than dampening those fluctuations. A "tracking" harvest strategy (Caughley 1977), with specific numbers of tags valid for antlerless deer would more closely respond to population trend. This concept and other discussion of harvest strategy were discussed in more detail in Chapter 6 - Adult Mortality. If managers prefer consistence of regulations from year to year or across environmental types, they must be prepared to accept the implications of populations occasionally declining below the normal low. Social, economic, and political factors will often dictate "crisis" management measures when deer populations fluctuate to "unacceptable" levels. Harvesting deer in variable environments is not a simple scientific or mathematical exercise of calculating stable averages, yield curves, and tag numbers using computer models. Extrinsic factors, including hunting mortality, do not act in an average manner and there are no average deer. Natural mortality rates are different for different sex and age classes and among areas and years. Fawns have higher mortality rates than adult females. A female that has recruited fawns for 2 or more years in a row is more likely to die over-winter than one that lost her fawn in June and recovered body condition during summer and autumn. An older, 339 dominant breeding male is likely to be in poorer condition entering winter than a yearling male. All males usually enter winter in poorer condition than females. Adult males are harvested at a higher rate than yearling males and all antlerless deer, adult females are harvested at higher rates than fawns, and different age classes of adult females are harvested disproportionately. Because adult females have a lower natural winter mortality rate than fawns, the potential level of compensatory mortality between age classes is further reduced. Although fawns have a higher natural mortality rate than adult females, most of their annual mortality takes place between birth and the hunting season. Thus, hunting mortality can only potentially substitute for the fraction of fawn mortality that occurs over winter. Hunting tags valid only for fawns could potentially increase the operative level of compensatory mortality (substitution) in populations. The actual level would, of course, vary among populations and years, but would be somewhat higher if more fawns were harvested. Deer were not harvested equally across our study area even though there was reasonably good access to all parts of the area. Deer in more marginal, open habitat, close to roads, and on ridge tops were harvested more heavily than those whose home ranges were in rugged, diverse core areas. This also, at least partially, explains why hunters on this area don't perceive many deer until the marginal habitat begins to be filled. That typically occurred when the area- wide density exceeded about 3.9 deer/km2 Thus, maintenance of deer in marginal habitat was important to maintaining hunter satisfaction, but it reduced the degree of "substitution" that could occur. All of these variations among sex and age classes, among individual deer, and in hunter response to them, when combined with the unpredictability and annual variation in weather patterns and predation levels, made yield models impractical. Instead, improvements in harvest efficiency can be made only by directing kill to certain sex and age classes and to certain areas or habitats . Changes in harvest strategies must be made quickly and as near the opening of the season as possible to be most efficient. If such changes are impractical or if we desire consistency and continuity of regulations, then we must resign ourselves to either a greater degree of fluctuation in deer populations or more conservative hunting seasons than might otherwise be possible. Because generalized density-dependent compensatory mortality and reproduction did not operate effectively in the MRBMD population, we must not allow ourselves to be deceived by the implied precision of existing models that incorporate 340 such processes. Harvesting strategies that are currently practical do not even closely mimic the culling and "harvesting" practices of domestic livestock operations which make use of "substitution". Normally, domestic cattle ranchers attempt to cull and sell older, unproductive, and unthrifty cows before they die. Similarly, "cow-calf" operators sell the major portion of the crop of young-of-the-year shortly after weaning, retaining the best calves to replace adults that were culled. Although not 100% efficient, generally those animals most likely to die or decline in productivity are sold for a return rather than retaining them in the population. If on 1 October each year, the rancher randomly, rather than selectively, removed 20-30% of his cows and calves for sale, some additional mortality would almost certainly occur during the next year among vulnerable animals not selectively culled. In this case, the rancher employed a harvesting strategy similar to most "enlightened" wildlife harvesting strategies that harvest a portion of the females and young. The point of this example is that even for areas or species in which some density-dependent compensatory mortality and reproduction are considered to be a valid component of models, we should not assume anywhere close to 100% efficiency of operation. At least some harvesting loss will always be additive to other losses. Recruitment must match harvests plus additional natural loss to maintain stable populations . Where habitats contain interspersed public and private lands, the distribution of harvest in both space and time has important implications to population dynamics. Hunters seem to prefer to not ask permission to hunt on private land if public land is available. Part of the mystique of hunting is the sense of freedom and "providing for yourself". In Montana at least, asking permission violates both desired products. At some point, however, overcrowding on public land reduces the sense of freedom and the deer population enough to increase the attractiveness of hunting on private land. This situation aids those interested in "privatization" of wildlife and argues against overharvests of deer on public lands. It may also indicate that high sustained harvests and maximum opportunity for everyone is not a unanimous public goal for management on public lands. Because of the above and other considerations, uncontrolled distribution of hunting pressure often results in overharvests of "roadside" deer and deer on public lands. Thus, core populations in secure habitat and populations on private lands with limited access can remain relatively unchanged while hunter dissatisfaction generally increases. 341 Self -distribution of hunters seldom helps solve game damage problems until extreme levels are reached. Early prevention of game damage is certainly not achieved by free-choice distribution of hunting effort. Our studies have shown that varied and specialized movement and home range strategies are necessary if habitat fill is to occur in complex, diverse, and variable environments. On our study area, autumn and winter migratory movements were necessary if deer were to successfully occupy shallow relief areas from late spring through autumn. Similarly, mule deer populations in mountainous areas also have migratory segments . Some of these may be long-distance migrants, occupying summer range far from the nearest winter range. Significant numbers of these migratory deer are often necessary to achieve habitat fill and maximize population level on many areas . Long-distance migratory segments often move to winter range earlier than other deer; thus, it is possible for various regulations and/or non-directed distribution of hunter kill to impact some segments longer than and more than others. This learned migratory behavior, passed from mother to daughter, may take some time to re-establish if excessive harvests remove significant numbers of these long-distance migrants. This problem may be especially important in mountain deer populations where low recruitment rates prevail. It is possible that very heavy harvests of antlerless deer from some western Montana populations in the late 1950s and early 1960s almost entirely removed early migratory segments with traditions of long-distance migration. Thus, some summer range at long distances from winter range remains unfilled, and total population size has never recovered. For riverbreaks and prairie populations that move to "winter range" only during very severe winters, excessive antlerless harvest during the intervening years may result in little "population memory" of wintering areas. Although the remaining deer may eventually find these areas by wandering during subseguent severe winters, additional mortality may result because few deer remain that remember the appropriate immediate response. In many cases, population managers may determine that changes in regulations affecting distribution of kill in time and/or space are impractical or undesirable. However, they should at least be aware that broadbrush approaches to management can affect subtle changes which may have long- lasting impacts on the population. Our findings also had some implications with respect to the current concern of some segments of the public for 342 managing for larger and/or trophy males. Our data, like findings of many other studies, indicated that males and females used habitat differently, and to some extent used different areas. This partitioning of habitat and location appeared to have its basis in parturition territoriality of females. It suggested a hypothesis that potential male: female ratios were at least partially determined by the relative proportions of reproductive and non-reproductive habitat within an area. The difference between potential ratios and actual ratios were primarily determined by hunter harvests and recruitment rates. Some population-habitat units, composed primarily of reproductive habitat, may not have the potential to support high numbers of males. In such areas, "trophy" management would likely be unsuccessful. Habitat security will also affect the potential of an area for "trophy" management. Areas with low habitat security will not produce older males with even moderate hunting pressure. The extensive emigration and immigration by yearling males on our study area indicated that attempts to "carry over" males within a single, small hunting unit may be unsuccessful beyond certain limits. The level of carry-over will vary with population and habitat, but beyond that the "saved" males may disperse to adjacent hunting units. In addition to age of the animal, antler size was clearly influenced by forage conditions. Drought and low quality forage resulted in smaller antlers for all age classes. Thus, results of "trophy" management in variable environments will be variable. During some years, antlers will be relatively smaller even when older males are present. Male: female ratios could be increased in populations which historically had higher ratios than currently observed. For this to occur, however, substantially fewer males must be shot than now occurs. Whether this is accomplished by limited permits or other means, relatively few hunters can harvest males if significant numbers of older males are to be maintained. Even when "trophy" management is implemented, there is no guarantee of improvement beyond a certain point. Dispersal and varying nutritional conditions will also play important roles in reducing the degree of success of "trophy" management. A clear portrayal of what is lost versus what is gained should be made for all potential management actions in each population/habitat unit. Finally, our data provide for comment about aspects of harvest strategy relevant to game damage problems . Many short-term solutions have been applied toward game depredation problems including various types of fencing, repellents, scareguns, and damage hunts. Stack yards for hay and certain types of fencing have been used for longer-term solutions, but 343 the most common solution is to attempt reduction of populations through harvest, preferably during the regular hunting seasons. Many types of damage, such as to seed alfalfa or pasture, are more dispersed and do not lend themselves to solutions such as fencing or repellents. It has become increasingly recognized that many damage problems are chronic and are not necessarily related to population level. In many cases, drought draws deer or other wildlife into the few remaining areas of green forage or to haystacks. This occurs regardless of population level and may involve only a few animals that cause significant damage or the perception of damage. Generalized reductions in deer populations across broad areas do not necessarily stop damage. For some situations where large numbers of deer are not causing the problem, selective local killing by Department personnel of the few deer causing the problem may be the most effective solution. We should also recognize that in areas of variable environment, many agricultural areas contain the best natural habitat and the best, most dependable sources of high quality forage. Depredation problems in those areas will be chronic. For more than 50 years, we have been encouraging modes of operation and improved land management practices by agriculturists that benefit wildlife. In many cases, improvement has been substantial and deer and other wildlife have benefitted even when increasing wildlife population was not a goal or even desirable to the landowner-manager. Thus, we must recognize that many depredation problems will be chronic and will not be solved by generalized population reductions. Instead, efforts must concentrate on long-term solutions such as permanent stackyards, properly designed fences, and where hunting is allowed, possibly some level of monetary reimbursement or purposeful habitat alterations to reduce deer occurrence. Predicting Fawn Survival Advance knowledge of recruitment rates could help deer managers more closely match hunting regulations to population trend. For the Missouri River Breaks study area, we developed a regression equation that accurately predicted fawn survival from data collected by the end of May, just prior to birth. Temperature during May and precipitation during July-May (Chapters 3 and 5) predicted fawn survival during 21 of 27 years for which data were available. Although level of fawn survival wasn't accurately predicted during 6 years, trend was accurately predicted. Our success in predicting trend and level of fawn survival indicates that similar potential exists for predicting fawn survival in other areas. Wood et al . (1989) indicated that similar factors influenced fawn survival 344 for both mule deer and white-tailed deer in a prairie environment . Predictions of fawn survival on our study area could not be calculated soon enough to help set general hunting season structure in Montana. However, predictions were available in time to help establish B-tag levels. The relationship between temperature, precipitation, and fawn survival, although not a precise mathematical tool available at the ideal time, does represent a significant improvement in management capability for early response to fluctuating population/environmental characteristics . Existing weather and deer classification data may be sufficient to attempt to establish similar predictive regression models for other areas. Relationships similar to those described by this study and studies of Wood et al . (1989) might be expected for other non-mountainous areas east of the Continental Divide subject to periodic drought. The exact timing of important periods of temperature and precipitation may vary slightly among areas, but should be generally similar. For areas west of the Divide and other mountainous areas, summer-autumn forage conditions are more dependable and drought is less of a problem. For these areas, length and severity of winter may be more important factors determining fawn survival. Although summer-autumn forage may be of consistent quality in these areas, the length of time deer are able to use that forage may vary with the timing of onset and end of winter. Severity of winter would affect rate of fat depletion. Predator Management It was clear both from direct and circumstantial evidence gathered during this study that predation by coyotes influenced mule deer population dynamics. Other recent work on predation also has indicated that the impact of predation is more than just "removal of those animals that were going to die anyway" (Bergerud 1971, Keith 1974, Beasom 1974, Bergerud 1978, Stout 1982, and Gasaway et al . 1983). We found that the degree of predation on deer by coyotes was influenced by vegetation and forage production and quality which influenced deer behavior and also levels of alternate prey populations. Certain types of snow and ice conditions also influenced predation rates. In some years and some circumstances, coyotes killed many deer that would have survived in the absence of predation. Predisposition of deer by poor physical condition was not necessary for significant mortality by predation to occur. Similarly, 0 ' Gara and Harris (1988) found that a sample of deer killed by coyotes and mountain lions (Felis concolor) contained more prime-aged deer 345 and deer in better condition than a sample killed in collisions with vehicles. Thus, predation involved more than simply "weeding out the young, the sick, and the old". Coyotes are facultative predators that do not depend upon one major prey species; they successfully use a wide variety of food. Because of this, at low deer densities, coyote numbers will not necessarily decline and deer populations may be subject to significant predation, depending upon the availability of alternate prey. Predation, combined with even low-moderate levels of other mortality (e.g. hunting), can be sufficient to further reduce low density deer populations or at least keep them from increasing. A change in circumstances, such as an increase in alternate prey populations, is necessary to release the deer population from the influence of predation and result in an increase in density. From observations such as this, Haber (1977) proposed a multiple "equilibrium" model in which predators can hold a prey population at low densities, but have little impact at high prey densities. The implications of these findings to deer population management obviously will depend upon local circumstances, population goals, management philosophy, and political, sociological, and economic factors. At high densities of deer, when alternate prey populations are nearing or at highs, and/or when hunter harvests are insufficient to remove the annual recruitment, managers may be unconcerned with predation or even welcome it. However, if a large decline in the deer population has occurred owing to severe weather or hunting, the manager may want to consider the potential for management actions . As suggested by Gasaway et al. (1983), management options are limited when a large decline in ungulate populations has occurred, predators are present, and the management goal is to increase ungulate populations. We perceive 4 basic options. 1. No action - The manager can wait for a natural change of circumstances that releases predation pressure on deer. These changes may include a large increase in the availability of easily caught alternate prey, a natural decline in predator populations, and/or favorable changes in weather conditions . These changes could occur very slowly. For example, peaks in jackrabbit populations in this area may be 10+ years apart. This no action option may not be considered viable by many hunters, landowners, and/or managers . 2. Limited action - Here, the manager could also reduce or eliminate harvest of deer by man and wait for the same natural changes in circumstances. However, this 346 alternative by itself may only halt further population decline or allow a very slow increase in the deer population. 3. Active intervention - The manager could attempt to hasten the increase in deer numbers by reducing predator numbers through various control techniques. Several studies (Beasom 1974, Stout 1982) have shown that predator control can significantly increase fawn survival. This option may be vigorously opposed by some segments of the public and enthusiastically supported by others. 4. "Preventative" management - Where significant predator populations are present, the manager can try to ensure that hunter harvests do not reduce deer populations to such low levels that they become significantly affected by predation. For the Missouri River Breaks mule deer population, that low was about 50-60% of observed population highs. Of the 4 options, "preventative" management may be the ideal, but significant population decline may also result from severe weather or other non-hunting factors uncontrolled by the manager. In such cases, only the other 3 options are available. The choice among those depends upon many circumstances. Where alternate prey populations seem to be cyclic, it is advisable to know where those populations are within cycles. For example, it might not be necessary to initiate a predator control program when alternate prey populations are naturally increasing. On the other hand, if alternate prey populations have recently "crashed", predator control measures may be more effective or important in maintaining prey population stability. The option of predator control is particularly subject to political, sociological, and economic considerations. Like biology, any or all could support either side of the issue depending upon location and circumstances. Given the probable difficulty of implementing predator control, it is unlikely that it could often be accomplished on a timely basis. Nevertheless, predator control should not be ruled out of the management repertoire if more rapid population increases are desired. Also, the success of attempts to transplant, re- establish, or otherwise increase low density populations of some species may also depend upon local predator control. 347 Population Evaluation Population Estimates Our data (Chapter 2) indicated that reliable estimates of mule deer populations could be made in riverbreaks habitat. Similar findings were reported for other eastern Montana populations (Wood et al . 1989, Watts unpubl . , Knapp unpubl . , Jackson unpubl . ) . Use of complete-coverage aerial surveys and Lincoln Indexes with at least 20-25 well-distributed radio- collared deer was an appropriate method. Observability indexes and other results were generally consistent when the same pilot and observer were used. Once the spectrum of survey conditions had been covered, reasonable estimates of the observability of deer on any particular flight could probably be made by an experienced observer even without marked deer. Surveys with the Bell 47 helicopter were less sensitive to variations in weather conditions and ground cover than those with the Piper SuperCub. Helicopter surveys and multiple fixed-wing surveys may often be too expensive and time consuming for widespread use by management personnel. When that is the case, and where an estimate of recruited population size is the most important information desired, we recommend full-coverage, early spring surveys with a Piper SuperCub. Under proper conditions, early spring is the most efficient time for survey, and it also most accurately measures net annual recruitment. We counted similar proportions of the population during early spring surveys with the SuperCub as we did with the helicopter in either early winter or spring. For best results, correct timing of these spring surveys is very important. The objective is to time flights such that the majority of deer are feeding on open ridges and south facing slopes where new green growth first appears. We obtained best results in flying when new growth of grasses and spring forbs had only recently started. The highest proportion of the population was observed in the open, and deer spent more time during the day in the open when flights were made before new growth had started under the forest canopy and before large amounts of new growth were available in open areas. Under these conditions, effective surveys can be flown during all hours of daylight except immediate mid- day. Our data on deer distribution and dispersal patterns had important implications to the proper establishment of study areas and trend areas for measurement of population trend. When establishing these areas, marginal deer habitat must be included along with core habitat. On our study area, 348 population increases beyond a certain point were not measurable in core habitats. Much of the latter stages of population increase occurred in marginal habitat. In like manner, population declines generally occurred first and to the greatest extent in habitats marginal to core area, thus, trend areas that do not include marginal habitat may lack the capability to provide early warning of true population trend. Age Structure and Sex and Age Ratios Age structure was not useful in predicting population trends. It was most valuable in verifying past population history. Age structure has often been considered very important because it was easily collected at check stations, and little, if any, data were collected on population size or fawn recruitment. Often, the only information some states had on fawn recruitment was data for past years based on age structure of hunter-killed samples. Adequate sex and age classifications and population estimates provide better data for most management purposes than age structure from hunter- killed samples. There are also other reasons, at least for populations in variable environments, why age-structural data may not be as useful to management as once thought. It has been commonly assumed in wildlife management that a hunted population has a younger age structure than an unhunted population. Similarly, healthy, growing populations had a younger age structure than unhealthy, high density populations. This was not necessarily the case in the Missouri River Breaks. It apparently was also not the case in some other northern Montana habitats . Data from the mule deer population on the Brinkman Preserve, which had been unhunted for many years (Rosgaard, unpubl. data), indicated a younger age structure than our hunted population. The expectation that younger age structures occur in hunted populations may be partially valid for populations with stable recruitment rates. On average, hunted animals will not live as long as unhunted animals. Thus, the average age of adults should be less in hunted populations if recruitment is stable. However, where recruitment rates vary considerably from year to year, age structures may not differ measurably between hunted and unhunted populations . A major reason why the assumption of a younger age structure for hunted populations does not hold for variable environments is its basis in the "principle" of compensatory reproduction. Because fawn survival on our study area was not directly related to number of deer in the population, a reduction in numbers by hunting did not necessarily result in increase fawn recruitment. Therefore, it was not surprising that age structure of the female population on our study area 349 was not younger during years of female harvest than during years when females were not harvested. In comparing population ecology and dynamics among studies, care must be taken in expressing and interpreting various methods of reporting population composition. It is always most informative to be able to estimate and report total numbers in each category rather than report only ratios or percentages. This is especially true when determining the implications to management and to hunters. For example: assume that we implement a new season type and report that 10% of mule deer population was males last year and 14% was males this year. Did the new season result in an increase in survival of males? Not necessarily. The total number of males and females could have remained exactly the same, but fawn survival was much lower. Thus: Year t 20 males (10%) 100 females 75 fawns Year t+1 20 males (14%) 100 females 20 fawns When examining the possibilities, it can be determined that survival of males could actually have declined under the new season even though they comprised a greater percentage of the population post season. Similar problems exist when only ratios are reported. Assume we report that post-season fawn: doe ratios have remained at 70:100 for 3 years and the statewide hunter questionnaire has reported an exactly stable harvest of bucks, does, and fawns for each of the last 2 years. Have we achieved the manager's dream of stability and is everything wonderful? Possibly not! A potential result of observed stable levels of harvest and fawn recruitment in post-hunting season populations could be: Year t - post season 20 males 100 females 7 0 fawns 190 total deer Year t+2 - post season 5 males 65 females 45 fawns 115 total deer 350 Thus, stable numerical harvests and stable recruitment do not necessarily mean stable populations. The main point of these examples is to show the pitfalls in the path of anyone making decisions based on data reported only as ratios or percentages. Estimates of population numbers or trend are almost essential for confidence in the meaning of ratios and percentages . Interpreting Population Trend Trends are one of the most widely used and abused tools of all aspects of life. If correctly used, trend information can be valuable, especially if data are sufficient to provide long-term perspective. When quick answers or decisions are required, we all usually rely on the trend ( "Go with the trend" - "The trend is your friend"). However, trend data are also subject to misuse and overuse, and we often place too much confidence in its reliability. At minimum, only 2 data points are required to establish a trend. Although most people are justifiably reluctant to place confidence in a 2 data-point trend, many decisions are based on no more than that. A longer-term data set is necessary to place annual changes in perspective. However, the particular data set or its specific length can greatly influence that perspective and the conclusions drawn. Examples of both the value and the pitfalls inherent in trend data are illustrated by changes in numbers for the Missouri River Breaks mule deer population over time (Fig. 13.1). Five years of trend information, starting in 1960 (Fig. 13. 1A), indicated that both short- and long-term trend for the population were up. However, addition of only 1 more year of data (Fig. 13. IB) indicated just the opposite conclusion; both short- and long-term trend were down. Expansion of the study to a "long-term" 10-year data set (Fig. 13. 1C) indicated that, given a "longer-term" perspective, the "real" long-term trend was up. Again, however, once this uptrend was clearly established, the population immediately turned downward, and by 1975 after 16 years (Fig. 13. ID), the obvious trend was toward increased fluctuation, with higher highs and lower lows. After the lower low was established (1975), the population almost immediately began an uptrend (Fig. 13. IE). After 23 years, it appeared the trend for higher highs was coming to an end, but once again, one year of additional data (Fig. 13. IF) falsified trend projections by re-establishing the trend of higher highs. Finally, after 28 years (Fig. 13. 1G), the trend of lower lows and higher highs appears to have been halted. Can we trust this interpretation of the trend? Is the population now fluctuating around a higher "equilibrium" level than that of the 1960s? A higher 351 CO l_ CO CD >» CO CM UJ CO co o >» co CO 0> O \ CO CO to CD 5> CO CO d) >» in V> in i O H VO d) <7\ > H o ^ fO (0 u >i 0) T3 ■p 3 3 +-> ■H H M in it) >i CN •H IX 3 ■H 03 H 3 ^ 0 tT (/) •H s CN Q) Xi ^^ +-> U c 0 rn M CN 0) ^ 0) T3 O <^, -~ 0) .H r- 3 i-i E ^ 4-1 ,,_^ 0 CJ ■^. ,-. H 0) o .a .H E 3 >. 3 . — . CQ 3 ■ — ^ •H m> -a • 3 ^r- a) ^-co u h ^-.H Q) U 3 fa 352 equilibrium level with reduced fluctuation has been a management goal after the 1970s. Whether that goal has been achieved and will be sustained can only be determined by additional years of testing. Interpretation of trends and results and conclusions of studies may often depend upon such chance events as the particular year data collection starts and how long it continues. For example, for our study (Fig. 13. 1G) , the reader can choose a variety of 10 consecutive year periods that will indicate vastly different, even opposite, population trends. These could also provide vastly different interpretations of population ecology and regulation. The initial year of investigation makes all the difference in 10- year trend. Similarly, it is obvious that 10 years does not necessarily provide a long-term perspective. The degree of environmental variability within an area may be a major consideration in determining valid study length and reliability of trend projection. One year's change did not make a trend, but at about the time a long-term trend seemed to become obvious, the trend was often about to change. This seemed to occur because extremes tended to make the past trend obvious and "extremes" were unlikely to occur for 2 or more years in succession. Thus, population or any other measured data tend to reverse after extremes . We suggest that trends in 1 population on 1 area provide insight to dynamics only after long periods. To best or most quickly measure the results of any management practice, the trends of 1 "treated" and 1 "untreated" population within similar environmental types should be measured simultaneously. This paired population approach is necessary to determine how much of the trend was "natural" and how much resulted from management action. Habitat Evaluation and Management The importance of healthy, productive habitat to healthy, productive animal populations has been justly recognized for years. However, concurrence on the meaning of healthy habitat and habitat requirements of species is lacking. This results partly from the different perspectives and objectives of different individuals, groups, and agencies. It is also the result of the lack of solid information about and understanding of the interactions between habitat and animal populations despite years of various types of studies. The assumption that winter habitat ("range") was of overriding importance to big game populations in the northern United States may have resulted in inadequate data collection on other aspects of deer ecology. This led to lack of long-term 353 monitoring of animal-habitat relationships and confusion about those relationships. For example, despite millions of dollars spent on years of habitat manipulation including burning, cutting, chaining, spraying, plowing, planting, and etc., we are not aware of published data documenting benefits to big game populations in terms of survival or increased numbers . The lack of published information may indicate little or no monitoring of animal populations; perhaps it also indicates that beneficial results were not observed. Some studies have shown that various animal populations have made increased or decreased use of treated areas, at least for a time. If increased use occurs, the assumption has usually been that the animals have benefitted. Documentation has remained lacking, however. Even much of the usage data is at best neutral or even negative about the effects of habitat manipulation on deer (Clarly et al . 1974, Short et al. 1977, McCulloch 1974). Cost effectiveness of these habitat manipulation projects is also guestionable (Regelin 1975, Nellis 1977). In one of the few studies that measured population response to habitat manipulation, Klinger et al . (1989) found that, although black-tailed deer use of burned chaparral was temporarily higher, there was no significant change either in population density or fawn survival resulting from the burn. The lack of definitive results from habitat management for deer may be related to multiple causes. Factors may include: 1.) lack of accurate knowledge about species habitat requirements; 2.) assuming too much knowledge (e.g., exclusive concentration on winter habitat/forage); 3.) larger home ranges of large mammals than small game result in manipulations impacting only a few animals; 4.) much of our effort is spent in trying to maintain existing habitat; and 5.) it is difficult to balance the requirements of multiple species and at the same time make changes sufficient to result in measurable benefits for 1 species. Essentials of Mule Deer Habitat Traditionally, the essence of habitat has been summarized as the amount and distribution of food, water, and cover (Dasmann 1971). For deer, emphasis has been almost entirely on food and most often on winter forage (see Chapter 1). Our data clearly show that plant phenology and forage quality are important to deer population dynamics. However, density- independent, rather than density-dependent factors played the greatest role in deer-nutritional relationships. Also, winter forage relationships were not of overriding importance; forage relationships during other seasons were of at least equal, and possibly greater importance. In this respect, Short (1981: p. 127) has summarized seasonal forage requirements well: 354 "Summer forage should allow adequate milk production by does and permit all deer to achieve adequate growth and fat storage. Autumn forage should be abundant and of good quality to delay the depletion of fat stores. Winter range should provide forage that minimizes energy deficits and fat depletion. Spring range should offer feed that permits early recovery from stresses of winter." We view the first essential of mule deer habitat as space, a place to exist. Although this may seem basic or trite, it is important to establish that houses, highways, airports, parking lots, mines, other animals, etc. usurp space for living. Similarly, deep, drifted snow reduces space for living. Thus, not all space is usable throughout the year. What types of space best provide for the requirements of mule deer? We found that mule deer were positively associated with topographic and vegetational diversity. Across their species distribution, mule deer occur in a wide variety of habitats and vegetation complexes. This is true even within Montana. Generally, however, relatively steep, broken terrain on at least a portion of the area is a common feature of all mule deer habitat. Gentle terrain was uninhabitable by deer during severe winters on our study area. Rough terrain provided south- facing slopes that were relatively snow free and timbered north-facing slopes that provided thermal cover and uncrusted, undrifted snow cover of lesser depth than on open areas. Rough terrain also helped provide vegetational diversity and microclimatalogical diversity which provided forage diversity and extended the period of availability of succulent forage during late summer and autumn. Habitat diversity also interacted with the social system of mule deer to provide more parturition territories in more diverse areas. Rough terrain and thick timber limited hunter access, thereby increasing security for deer. It also provided mule deer better hiding and escape cover from coyotes by limiting the ability of coyotes to maintain visual contact with the deer. Overall, physiographic and vegetational diversity provided the best natural system to provide quality forage for the longest periods. It also provided more alternative use areas to compensate for the variety of environmental and man-induced events that occurred. Often ecological reports spend many pages discussing animal use of and importance of specific habitat-land types. We will not do that here. Rather, to understand and affect mule deer population dynamics, we believe it more important to understand why mule deer used some types more than others and some at different times than others. As noted previously, mule deer live in a variety of environments, thus the specific 355 type is of less importance than the function it provides. Effective management, even on a specific area, needs to be based on the why of use, not just the fact of use. At some point in time, for some purpose, mule deer used all vegetation types on our area. The importance of some types or sites to the long-term health of the mule deer population was evident in some cases only after long-term monitoring. Assenting to the "sacrifice" of some of those areas based on limited data could have detrimental effects on the population over the long-term. Mule deer on our study area appeared to strongly prefer Douglas fir habitat types yearlong. However, Douglas fir habitat was absent from the eastern portion of our study area and further east in Montana. Obviously, mule deer in those areas somehow survive and even thrive in the absence of Douglas fir. Similarly, mule deer on our study area seem to survive without the rocky outcrops prevalent on many Ponderosa Pine winter ranges in western Montana. Thus, deer prefer certain areas or types at certain times for what they provide rather than what they are. On our study area, Douglas fir types are relatively moist types within a drought-prone environment. Where available, they provide hiding cover and succulent forage during summer and autumn and thermal cover, lower snow depths, and softer snow during winter. All of these contribute toward the attainment of a positive energy balance on an annual basis. As noted earlier, mule deer live well on many areas without Douglas fir types. Something(s) else obviously substitutes for the Douglas fir types and fulfills its function in maintaining a positive energy balance for deer. A common substitution factor in much of central and eastern Montana is agricultural forage. Many crops, such as alfalfa, remain succulent longer than natural forages. Others, such as barley, corn, or sugar beets provide higher energy forage than naturally available. These forages are also more dependable than natural forages. Enough may be available for deer even when drought results in an economic crop failure for the farmer or rancher. Thus, these high quality, high energy agricultural crops can often substitute for a lesser degree of thermal cover, for example. In most "natural" situations, there is an interaction between forage and cover in the energy balance equation. That is, adequate thermal cover during winter can help ameliorate the lack of quality forage because it helps reduce energy use. Similarly, availability of high quality agricultural forage may ameliorate the absence of adequate thermal cover. The importance to deer populations of small, seldom used areas or types and areas that often received use for a very 356 limited time was evident from this study. This type of information would seldom be discovered during short-term investigations. For example, for about a month during April, mule deer would occasionally bed within the Douglas fir type, but feed almost exclusively within the Sagebrush-Grassland, Shale-Artemisia, or other open types. The latter are the first sites on which new, succulent green forage becomes available. Thus, substantial use of these types for only 1 month out of 12 may significantly increase the energy balance equation of the deer. Similarly, data collected during the severe winters of 1977-78 and 1978-79 (Chapter 8) indicated that small, seldom used areas provided "critical" winter range during those years. The importance of those areas certainly could not have been predicted based on the poor quantity and quality of forage on the sites. Our interpretation of the relative importance of habitat types and areas is vastly different from that usually inherent in the "key species-key area" concept. A variety of areas and types were critical to survival of deer over time. Except on a frequency of operation basis, one could not be separated out as more critical or "key" than another. Some were "critical" every year, but others, perhaps critical only once every 10 years, were just as necessary to long-term survival of existing deer populations. Allowing oneself to be badgered into designating certain areas as more critical than others by resource development- extraction agencies or groups results in sacrifice of areas which can be of equal importance. The sacrifice of areas that are of less frequent critical importance leads to insidious long-term declines or declines that may occur abruptly long after the causative event. The essentials of mule deer habitat, then, are a place to live and adequate diversity to provide options for maintaining a positive energy balance. In the natural environment, we cannot over-emphasize the preservation of diversity. Deer populations will fluctuate with the environment, but diversity will ensure that any one event is less likely to be catastrophic. Population characteristics and dynamics will generally follow diversity. The greater the diversity in an environment, the more stable the dynamics. Populations with fewer options to counter severe environmental events will be subject to wider fluctuations. Dynamics can also be stabilized by the "domestication" of intensive management. For more intensively managed situations, there are substitutes for natural diversity. We can, for example, raise deer in pens in parking lots by 357 feeding them appropriate forage and employing veterinarians . We can also enclose by fence and selectively cull animals, essentially simulating domestic livestock operations. These latter options, however, do not appeal to many people. Habitat Maintenance, Enhancement, and Development As can be surmised from our previous discussions, we believe that maintenance of diversity is important to mule deer. Deer are very effective at taking advantage of situations, areas, etc. that increase their chance of survival. In habitat management, it seems important that we leave them as many options as possible. Earlier in the century, increased diversity could often be provided by various land uses (e.g., timber cutting or agricultural development). As is typical, however, very little of anything is done in moderation. Increasingly, diversity has peaked as we passed from diversity that included some natural monocultures to an increasing proportion of manipulated monocultures. Thus, habitat management by necessity has increasing become "anti development", promoting maintenance of existing habitat rather than additional "development" or "enhancement" . Given the lack of a proven beneficial track record by previous "habitat development" projects, we are reluctant to specifically recommend many manipulative developments. Also, increased ecological knowledge gained by this study and others has cast doubt on the necessity, benefits, or cost- effectiveness of many manipulations. Certainly, preserving winter range space by precluding subdivision will provide meaningful long-term benefit, but attempts to increase forage quantity for deer on winter range may not produce measurable benefits . Generally, both the stable and variable properties of the environment are difficult to change, especially to benefit species such as mule deer. In some situations, however, we believe our knowledge is sufficient to recommend potential improvements for mule deer habitat. The applications of these suggestions will depend, among other things, upon the importance of and goals for other species in the area. In most cases, alteration of topography is uneconomic and impractical. However, in cases where proposed mining activities will move large amounts of overburden, reclamation of the site could be done in ways that might benefit mule deer. Most mining reclamation plans require that the site be restored to its original contour or possibly to even less topographic relief than originally occurred to reduce potential erosion. We proposed that if it is desirable to benefit mule deer, the overburden could be deposited, 358 oriented, and contoured in such a way that topographic diversity and specific exposures were more advantageous to mule deer after mining than before. Erosion potential could be addressed by various techniques with the understanding that some degree of erosion is natural and helps add to diversity of slope and exposure. Similarly, planned fires may alter the habitat structure in ways beneficial to mule deer. Burned Douglas fir-juniper habitat type was preferred by mule deer on this area during summer-winter. Only very limited burning of small patches of that type are recommended, however, because mule deer make important use of thickly timbered Douglas fir-juniper habitat during severe winters. Also, Rocky Mountain elk, a species of great interest and concern on the area, require these thickly timbered areas for security during hunting season. The planting and maintenance of alfalfa in small patches throughout public lands in arid eastern Montana would probably be beneficial to mule deer. This nutritious legume, with its extremely well developed root system, is competitive for water and generally remains green and succulent longer than most native forage species. The availability of alfalfa would increase the length of time that quality forage was available to mule deer in many areas and years. It also would provide more stable, dependable forage, potentially resulting in less population fluctuation. Additionally, planting alfalfa on public lands within deer habitat and/or at the margin of public deer habitat and adjacent private agricultural lands could act as a lure crop, reducing deer damage to private agricultural crops during dry years. Based on the data of Mackie (1970) and this study, we do not believe that further water development within the riverbreaks habitat will benefit mule deer. Adequate water sources seem to exist, and further development could only result in increasing overlap between distributions of cattle and mule deer. Lewis and Clark as well as other "oldtimers" present before 19 05 reported "vast" numbers of mule deer in the riverbreaks before any water development occurred. Generally, some degree of cattle grazing has likely been beneficial to mule deer habitat values over the long term. Locally, and for some shorter term periods, it has probably been detrimental. For example, some information presented in this report (Chapters 7, 8, and 9), indicated that heavy early spring grazing by cattle might be detrimental to mule deer. To benefit mule deer, we recommend that spring grazing by cattle be deferred to as late a date as possible. Intensity of grazing by domestic livestock has varied tremendously over the years, but stocking rates have generally 359 been much more reasonable in recent years than earlier. We believe that as long as overall grazing pressure by domestic livestock is light to moderate, the existing season-long grazing system provides a patchwork of diversity, including heavily grazed, moderately grazed, lightly grazed, and ungrazed areas within the Breaks type. This diversity results in a wide variety of plants species adapted to each situation. The wider the variety, the greater the spread of maturation dates and the more likely some species of value to deer will occur under all environmental conditions. Both totally ungrazed and heavily stocked areas would likely result in lower diversity. Similarly, highly managed, "efficient beef production grazing systems", often also promoted as good for vegetation and soil may also result in monocultural vegetation patterns. This, together with increased social disturbance associated with intensively managed systems, would likely be detrimental to mule deer. Finally, we caution that where riparian habitats comprise an important portion of an area, conclusions about appropriate grazing systems might be different than for "Breaks" habitat type. 360 APPENDIX A Appendix A, Arithmetically modeled population estimates for mule deer on the Missouri River Breaks study area, 1960-1987. Period Total Adults Fawns Females Males October 1960 1225 785 480 535 250 December 1960 955 570 385 445 125 May 1961 925 555 370 435 185 620 120 185 305 October 1961 1235 925 310 620 305 December 1961 850 645 205 435 210 May 1962 805 615 190 415 95 510 200 95 295 October 1962 1055 805 250 510 295 December 1962 850 620 230 410 210 May 1963 810 590 220 390 110 500 200 110 310 October 1963 1295 810 485 500 310 December 1963 1000 640 360 430 210 May 1964 910 620 290 415 145 560 205 145 350 October 1964 1260 910 350 560 350 December 1964 1020 720 300 490 230 May 1965 715 645 70 440 35 475 205 35 240 October 1965 895 715 180 475 240 December 1965 725 575 150 420 155 May 1966 670 545 125 400 60 460 145 65 210 October 1966 920 670 250 460 210 December 1966 730 510 220 385 125 May 1967 665 485 180 365 90 120 90 455 210 361 Appendix A. (Continued) Period Total Adults Fawns Females Males October 1967 1100 665 435 455 210 December 1967 935 535 400 400 135 May 1968 825 505 320 375 160 535 130 160 290 October 1968 1335 825 510 535 290 December 1968 1080 640 440 465 175 May 1969 870 585 285 425 140 565 160 145 305 October 1969 1480 870 610 565 305 December 1969 1250 705 545 505 200 May 1970 1070 670 400 480 200 680 190 200 390 October 197 0 1530 1070 460 680 390 December 197 0 1290 865 425 620 245 May 1971 1140 800 340 575 170 745 225 170 395 October 1971 1510 1140 370 745 395 December 1971 1130 825 305 610 215 May 1972 665 595 70 425 35 170 35 460 205 October 1972 830 665 165 460 205 December 1972 675 535 140 380 155 May 1973 600 485 115 335 55 150 60 390 210 October 1973 780 600 180 390 210 December 1973 640 470 170 320 150 May 1974 565 440 125 295 60 355 145 65 210 October 1974 685 565 120 355 210 December 1974 535 430 105 295 135 May 1975 470 395 75 275 35 120 40 310 160 362 Appendix A. (Continued) Period Total Adults Fawns Females Males October 1975 695 470 225 310 160 December 1975 535 395 145 310 85 May 197 6 390 355 35 275 15 290 80 20 100 October 197 6 565 390 175 290 100 December 197 6 490 365 125 280 85 May 1977 435 335 100 255 50 305 80 50 130 October 197 7 675 435 240 305 130 December 197 7 520 365 155 300 65 May 1978 450 355 95 290 45 335 65 50 115 October 197 8 800 450 350 335 115 December 1978 760 420 340 325 95 May 1979 640 390 250 300 125 425 90 125 215 October 197 9 1135 640 495 425 215 December 1979 1020 530 490 420 110 May 1980 910 500 410 400 150 550 105 205 310 October 1980 1355 860 495 550 310 December 1980 1175 685 490 540 145 May 1981 1080 670 410 530 150 680 135 205 345 October 19 81 1555 1025 530 680 345 December 1981 1215 745 470 600 145 May 1982 985 685 300 550 150 700 135 150 285 October 1982 1490 985 505 700 285 December 1982 1200 715 485 590 125 May 1983 1115 700 415 580 210 790 120 205 325 363 Appendix A. (Continued) Period Total Adults Fawns Females Males October 198 3 1715 1115 600 790 325 December 1983 1545 1020 525 775 245 May 1984 1040 920 120 700 60 760 220 60 280 October 1984 1365 1040 325 760 280 December 1984 1135 895 240 705 190 May 1985 870 830 40 655 20 675 175 20 195 October 1985 1075 870 205 675 195 December 1985 975 785 190 665 120 May 1986 915 775 140 660 70 730 115 70 185 October 1986 1480 915 565 730 185 December 1986 1355 805 550 715 90 May 1987 1230 760 470 675 235 910 85 235 320 October 1987 1720 1115 605 795 320 December 1987 1405 850 555 695 155 May 1988 1200 810 390 665 145 364 Appendix B Table Bl. Method of calculating estimated forb production on the study area using 1982 as an example. Total Usable Relative Vegetation Area Area Abundance kg/ha Total kg Year Type (ha) (ha) Couma Coum Coum 1982 AAb 10.328 6,899 1.00 47 324,253 PJG 5,106 4,883 0.39 18.3 89,359 PJS 4,024 3,955 0.00 0 0 DFJ 3,591 3,515 0.00 0 0 GW and SA 1,796 1,524 0.06 2.8 4,267 All others 2,555 1,988 0.00 0 417,879 Relative Abundance kg/ha Total kg 6,899 Meof Meof Meof AA 1.00 26.4 182,134 PJG 4,883 2.42 63.9 312,024 PJS 3,955 0.23 6.1 24,126 DFJ 3,515 2.08 54.9 192,974 GW and SA 1,524 1.62 42.8 65,227 All others 1,988 1.00 Relative 26.4 828,968 52.483 Total kg Abundance kg/ha Other Other Forbs Forbs Forbs AA 6,899 1.00 158 1,090,042 PJG 4,883 0.23 36.3 177,253 PJS 3,955 0.02 3.2 12,656 DFJ 3,515 1.09 172.2 605,283 GW and SA 1,524 0.17 26.9 40,995 All others 1,988 0.17 26.9 1,979,706 TOTAL KG ALL FORBS 3,226,553 53.477 For plant name abbreviations, see Appendix C. For vegetation type abbreviations, see Table 3.2 and 3.3 365 Appendix B (continued) Table B2. Method of calculating estimated shrub production on the study area using 1982 as an example. Total Usable Relative Vegetation Area Area Abundance kg/ha rotal kg Year Type (ha) (ha) Rhara Rhar Rhar 1982 AAb 10,328 6,899 0.00 PJG 5,106 4,883 1.00 25 122,075 PJS 4,024 3,955 0.11 2.8 11,074 DFJ 3,591 3,515 1.00 25 87,875 GW and SA 1,796 1,524 0.00 — -- All others 2,555 1,988 0.00 — 221,024 Usable Relative Total Relative Total Vegetation Area Abundance kg/ha kg Abundanc e kg/ha kg Type (ha) Rosa Rosa Rosa Symp. Symp. Symp. AA 6,899 0.04 2.1 14, ,488 0. 00 PJG 4,883 1.39 74.4 363, ,295 0. 36 5.9 28,810 PJS 3,955 0.15 8.0 31, ,640 0. 01 0.2 791 DFJ 3,515 1.00 53.5 188, ,052 1. 00 16.5 57,998 GW and SA 1,524 0.04 2.1 3, ,200 0. 00 -- -- All others 1,988 0.00 0. 00 — 600, ,675 87,599 Relative Vegetation Abundance kg/ha Tota 1 kg Total kg All Type Prvi Prvi Prvi Shrubs AA 0. 00 -- - - PJG 0. ,11 4.7 22, 950 PJS 0. .00 -- - - DFJ 1, ,00 42.5 149, 387 GW and SA 0. ,00 - - All others 0. ,00 _ _ 1 1,081 172, 33- ,635 For plant name abbreviations, see Appendix C. For vegetation type abbreviations, see Table 3.2 and 3.3, 366 Appendix C Common and scientific names of plants for which abbreviations are occasionally used in tables. Abbreviation Common Name Artr Big sagebrush Chna Rubber rabbitbrush Chvi Green rabbitbrush Coum Bastard toadflax Jusc Rocky Mountain juniper Meof Yellow sweetclover Pipo Ponderosa pine Prvi Chokecherry Psme Douglas fir Rhar Aromatic sumac Ribes spp. Currant and Gooseberry Ronu Nootka rose Rosa spp. Rose Syoc Western Snowberry Symp. spp. Snowberry Scientific Name Artemisia tridentata Chrysothamnus nauseosus Chrysothamnus viscidiflorus Comandra umbellata Juniperus scopulorum Melilotus officionalis Pinus ponderosa Prunus virginiana Pseudotsuga menziesii Rhus aromatica various species Rosa nutkana various species Symphorocarpos occidentalis various species 367 Appendix D. Excerpts from quarterly narratives of CMRNWR personnel and DFWP reports, 1940-1959. 1940 - "The western end of the Range [present study area] is the only high grade deer country. ...but, deer are scarce. " "Rabbits are on the upswing of the cycle." "No plan has as yet been even tentatively set up for predator control." 1941 - "Mule deer though scarce, may be increasing a little. " "Everything indicates a small upswing in mule deer production. " "We rarely see deer, though my contacts with several coyote trappers who work within or near our boundaries indicate that deer are definitely increasing. " "Range conditions for big game are ideal as to forage in the range as a whole." "In spite of tremendous numbers of coyotes being taken, I have seen twice as many coyotes as I saw last year. " "Both Lepus and Silvilagus [sic] are picking up. The latter look as if they may be approaching a cycle peak now. Mice, we repeat are numerous, and jackrabbits as well as cottontails are on the upgrade. In the country in general, however, jackrabbits have been hunted extensively for their skins, which have been worth up to $0.42 each." 1942 - "It is obvious that a good population of blacktailed [sic] deer exist on the Game Range. Contacts with stockmen and farmers living near the Game Range indicate that this population is and has been increasing through the last three or four years." "With weather conditions as they have been for the past four years, range conditions are at present better than at any time within the memory of the old-timers . " "Coyotes are frequently observed. " "Numerous complaints were received through this report period relative to the abundance and damage being done to sheep [domestic] by coyotes." 1943 - Text in a 1943 report, along with an enclosed map, indicated that 1 . 3 mule deer per square mile were estimated for the portion of the Game Range including the present study area. This was the highest estimated density for the Game Range. However, no indication was provided as to how these figures were derived. 368 "We have continued to receive complaints of an excessive population of coyotes on the Game Range from livestock operators around the area." 1944 _ "Range conditions were excellent and animals observed through this period were in excellent condition." "Conditions for deer through this period were ideal. " Holibaugh (1944), after a reconnaissance trip in the vicinity of our present study area, stated: "It also becomes more evident with each trip into the breaks country that the deer herd is increasing at a tremendous rate... Deer are abundant, sufficiently so to be a good hunting unit." "On October 20, Mr. DuBeau and myself worked out one of the tributaries of Sand Creek located in Sections 23, 24, 25, and 26, Township 21 North, Range 24 East [on the present study area]. Approximately four square miles were intensively covered. This included the main stream and all of the wooded coulees tributary to it. Thirty-one different mule deer does and three different mule deer bucks were observed in this area. It is estimated that between 60 and 70% of the deer in the area were observed." [Author's Note:] The 8.5 mule deer /mi2 observed in 1944 indicated a high deer population by that time. The actual density may have been higher. Although people making a deer drive through an area may recount the same deer more than once, it is also likely that they miss many deer (McCullough 1979). Aerial observations during autumn 1979-1986 on this same 4 mi2 area indicated 17-31 mule deer ( 4 . 3-7 . 8/mi2) observed per flight. Observability indexes (Lincoln Indexes) during the same period indicated that actual densities ranged from 11 to 17 mule deer /mi2. It is possible that mule deer density on this area was quite similar during 1944 to densities observed during 1979-1986. 1945 - "Mule deer - Trend up. Range excellent." "The coyote population has become unusually heavy over the past 5 years . " "The low price of coyote pelts, two to five dollars, along with a critical labor shortage, probably has been the cause of a constant uptrend in the population of coyotes over all of eastern Montana." 1946 - "Mule deer - Trend is a strong up over all of the Game Range. Carrying capacity of the Game Range in Petroleum, Fergus, and Phillips Counties will have to be closely watched if a dry cycle starts, as our population here is high." "It appears as tho [sic] part of the range in Phillips, Fergus, and Petroleum Counties should be 369 opened to hunting within the next two or three years . " "Jackrabbits - Trend is no longer down, but believed now to be steady, with an increase probable." "Coyotes - Trend still up, with a brighter future. More effective control is now possible since more men are available. " "It is believed that we loose [sic] many deer each year to coyotes but since their population is high enough to withstand extensive depredation, they still increase. " "We are pleased to announce for the first time that an intensive coyote control program is underway." (September-December, 1946). "The program planned was a duplicate of the one conducted in Nevada in 1945-46, and consisted of the distribution of small cubes of strychnine enclosed in a small ball of horse tallow which was coated on the outside with sugar. This is called a poison pill. Distribution was accomplished by plane for economy in funds and efficiency in operation. Plans were made to poison the entire Game Range area and 1,200,000 acres adjacent to it. " "Manufacture of the poison pills started about the first of November and by the end of the first week production was up to about 9,000 per day, for the three man crew." 1947 - "The first extensive control of these [coyotes] was done through the winter, through the use of poison distributed by airplanes. Results in the areas treated were above expectations. A conservative estimate of the poison kill stands at 1,500." "Mule Deer - Trend is rapidly up on the west end . . . The population on the western thirty miles of the Game Range [includes the present study area] is nearing the point of critical forage consumption, indicating the necessity of an open season soon. " [Author's Note:] The first substantial guantitative data on mule deer population size on the area were collected by aerial census during September, 1947 (Brown 1947). Brown and an observer in a fixed-wing aircraft flew 40 miles of strip census lines in northeast Fergus County, including our study area. They counted 4 8 mule deer and assumed that they had observed deer within a l/4-mile-wide strip for a total of 10 sguare miles of coverage. These data convert to an observed density of 4.8 mule deer/mi2 (1.85/km2). Our observability indexes for fixed-wing aircraft surveys during autumn have ranged from 0.265 to 0.488 (see Chapter 4). Because Brown (1947) noted that the " air was extremely rough," the percentage of the deer present that was observed at the time 370 of their flight was probably at the lower end of the range. The application of the observability factor to their data indicates a density of 18 mule deer/mi2 (4.8/0.265). This figure probably should be corrected again because the area they flew was that which, in recent years, had the highest density on the study area. Data collected from 1964-1986 indicated that the density figure from Brown (1947) should be multiplied by a correction factor of 0.70 to convert to a figure applicable to the entire study area. This final conversion indicated a density of 12.6 mule deer/mi2 (4.86/km2) for September, 1947. 1948 - "The statement that continual outflux of deer from the Game Range is seeding all of Eastern Montana is heard many times from many quarters . Leo Coleman who was there when 2 men killed one of the last Grizzly Bears on the Game Range in the Timber Creek breaks in 1905 states emphatically that there are many more deer along the river in the Game Range than at any time in their memory. This statement is always agreed upon by old-timers . " "Our deer population on many areas is nearly to the point where 50% of the annual growth on the most palatable species is being used by wildlife. Deer populations must be watched closely to prevent range damage. The time is not too far distant when the unmentionable proposition of opening the season on does will have to be discussed." "All animals observed and checked were in fine condition with fine glossy coats. Bucks handled were all free from external parasites and loaded with fat." "One remark common to all livestock operators and old timers is, "This years crop of antelope [or deer] fawns is the largest crop of twins I have ever seen." It is believed that last winters 1080 program is responsible for the quotation." [Author's Note] It is likely that the coyote poisoning program initiated during winter 1946-47 did result in increased fawn survival. There was little consistent information, with large sample sizes, available for fawn/female or fawn/adult ratios prior to 1949. However, narratives indicated that pronghorn antelope fawn/adult ratios increased from an "average" of about 0.3 fawns/adult prior to coyote poisoning during winter 1946-47 to 0.63 fawns/adult during 1948. The narratives implied that fawn/adult ratios for mule deer increased to a like or greater degree. The fawn/100 female ratios reported for mule deer during 1949-1956, indicated consistently higher fawn/ 100 female ratios than observed during any other time period. 371 [Author's Note:] Another strip census flight covering an estimated seven square miles on the study area was flown on 24 February, 1948. Brown (1948) observed 40 mule deer (5.7/mi2 observed) and estimated that "it is possible that not more than 50% of deer in this area were observed." The estimate of 50% observability for that time of year from a fixed-wing aircraft is reasonable, based on our more recent work. Again, the area flown was what is presently high density habitat, and we applied a correction factor that indicated a density of eight mule deer/mi2 (3.09/km2). The September, 1947, and February, 1948, estimates appear to be consistent with each other, especially when considering autumn and overwinter mortality. We believe that eight to 12 mule deer/mi2 (3.1- 4.6/km2) represents a reasonable minimum density estimate for early winter, 1947-48. [Author's Note:] The first hunting season for bucks in 12 years within the CMRNWR was held during autumn 1948. The season had been open for bucks since 1933 on areas off the CMRNWR. Checking stations for hunters were established for the first two days of the season on two roads leading to the study area. The 36% hunter success rate recorded for the opening two days compares favorably with opening day success rates in 1960-64 (20-47%) and 1979-80 (23-26%), when deer populations were at preseason levels of 10-13+/mi2. That further substantiated a population estimate of 8-12 deer/mi2 during February 1948. 1949 - "All deer killed were in excellent condition and almost every hunter had some comment on how fat his buck was . " "Although we experienced one of the most severe extended winters for many years, we have no evidence to dispute our casual observations, and information from local residents on and near the Game Range, to the effect that our game animals came through the winter in good condition and with very little loss." "Practically no coyotes, or the effects of them, have been noted on the Refuge. Nearly all local livestock men and sportsmen speak highly of the progress made in coyote eradication. "Jackrabbits and cottontails - Both are increasing considerably, especially the cottontail which seem to be almost everywhere." [Author's Note:) A sample of 144 mule deer classified during early winter, 1949-50, indicated 96 fawns/100 females. That high level of fawn production and survival indicated little initial impact of the severe winter of 1948-49 on the 1949 fawn cohort. 372 1950 - "Reports of heavy deer losses on the Game Range and vicinity have been investigated and almost without exception we find these reports to be greatly exaggerated..." (winter 1949-50) "Upon receipt of reports from the Hedman Brothers, local ranchers in Petroleum County in mid-March that winter loss of deer in that area was extremely heavy and would probably amount to 30 or 40 percent of the fawns and yearlings, we again conducted an aerial survey in Petroleum County on March 30 through cooperation of Don Brown and the State Fish and Game Department. Therefore, a resurvey was made of the entire area and the count came out so close to that of the previous census in February and due to the fact that only one dead deer was seen from the plane, we came to the conclusion that the loss could not be nearly so heavy as reported." "It is quite possible that winter losses in certain localities of the Game range may have been heavier last winter than was indicated in previous reports." (summer-fall 1950) "The most noticeable fact is that many of the [deer] fawns were dropped later than usual this summer." (May-August 1950) . "At the close of the report period [Aug. 1950], a high percent of the fawns are still heavily spotted. We feel that the exceptionally severe past winter is closely tied in with this fact." "Coyotes - The time is here when this species could be dropped from mention in the Narrative Report. They are now reduced to such an extent on the Game Range that observing one is cause for special comment." [Author's Note:] An aerial mule deer census was again conducted during early February 1950. Viewing conditions were good because of 16 inches of snow on the ground. The survey area was larger than that of previous surveys and included areas that, at present, have mule deer densities about half that on the study area. One hundred thirty-two (132) mule deer were observed on 157 miles of strips that covered an estimated 39.25 mi2. Observed density was 3.4 mule deer/mi2. If 50% of the mule deer present were observed, the estimated density was 6.7/mi2. If mule deer density distribution was the same in 1950 as at present, then the density estimate for our current study area may have been higher, at about 8- 10/mi2. We believe that 7-10 mule deer/mi2 ( 2 . 7-3 . 9/km2) is a reasonable estimate for February 1950. The recensus flown on 30 March, 1950, apparently had nearly the same results, but the data for that flight were not found. These data indicated a probable decline in density, compared to 1947-48. Such a decline is reasonable because the February 1950 census occurred after the severe winter of 1948-49 and during the 373 severe winter of 1949-50. The narratives indicated the probability of winter loss, at least during 1949-50. 1951 - "Food and cover conditions for big game on the area are excellent. " [Author's Note:] The final census flight of the 1950s was flown on 13 February, 1951. Sixty-eight miles of strips were flown, covering an estimated 16 mi2. Eighty mule deer were counted and, assuming a 50% efficiency of observation, a density estimate of 10 mule deer/mi2 (3.86/km2) was indicated. [Author's Note:] Classifications for population composition were made during both early October and early December, 1951. For 519 mule deer classified during early October, 125 fawns/100 females and 86 adult males/100 females were observed. It is likely that classifications made during the first week of October somewhat overestimated the proportion of males in the population. For 470 mule deer classified during early December, 125 fawns/100 females and 38 adult males/100 females were observed. If there were around 10 mule deer/mi2 during spring 1951, then the observed level of fawn production and recruitment, by itself, would have resulted in pre-hunting season populations of 15-18 mule deer/mi2. Based on data from hunter checking stations and post-season questionnaires, we estimate that the number of mule deer harvested on the study area was 200-300. Mule deer population density during December 1951 may have been in the range of 12-15/mi2. 1952 - "There is every indication that our mule deer herd is one of the healthiest and most productive in the country. While there aren't any widespread signs of overuse as yet, we should be constantly on guard, "... "Bobcats are probably the most numerous predator on the Game Range. Their numbers are far above normal on the west end of the area in the most heavily populated big game locality. " [Author's Note:] Pre-hunting season classification of 1,191 mule deer during 1952 indicated that there were 94 fawns/100 females and 48 adult males/100 females. A post-season classification of 860 mule deer was comprised of 91 fawns/100 females and 29 adult males/100 females. Data from hunter checking stations and post-season questionnaires indicated that 250-350 mule deer may have been harvested from the study area. An estimated 791 mule deer were harvested from the northern end of the county, with much of the pressure directed toward the vicinity of the study area. 1953 - "Due to favorable weather conditions during the spring and early summer, grasses, weeds, herbaceous plants, trees, and shrubs all made an excellent growth." ... 374 "the leaves and seed heads on many species remained green and succulent for a much longer than normal period. " [Author's Note:] Reports of widespread deer mortality during late summer and autumn, 1953, especially along the Missouri River, were investigated (Chaffee 1954). Based on more recent information and experience, we believe that the reported mortality during summer and autumn, 1953, represented an outbreak of epizootic hemorrhagic disease (EHD). During more recent years, mortality from EHD has been mostly confined to white-tailed deer and few mule deer have died. At least some mule deer living near the Missouri River died during late summer and autumn, 1953, but it is unlikely that the outbreak of EHD had a major impact on mule deer populations. 1954 - "The mule deer population in the areas that have been open to hunting, is lower than at any period since 1949. " "All browse species, as well as grasses and weeds, made an excellent growth this season. " "Game range personnel have seen only 2 Coyotes this period. The bobcat population over the entire area is believed to be high." (May-Aug. 1954) "A total of 6 coyotes were observed on the Game Range this period [Sept. -Dec. 1954]. This is the largest number recorded in any period report since 1949. There have been two complaints of coyote depredation on domestic sheep bands running in or adjacent to the Game Range this period." [Author's Note:] During August, 112 fawns/100 females were observed, and a classification of 684 mule deer during October indicated 94 fawns/100 females. 1955 - "The latter part of the summer was extremely dry"... "Coyotes - This animal is definitely on an increase on the Game Range but are by no means present in alarming numbers . Our personnel have seen as high as four animals in one day. " [Author's Note:] During August 1955, 128 fawns/100 females were observed. 1956 - "When drought conditions occur over a large area as they have during the past two seasons, the deer tend to concentrate within a five to six mile strip along the river or reservoir." "This movement back to the river breaks was first noticed in locally dry areas late in the fall of 1955. " 375 "Mule deer range showed some signs of over-use in the vicinity of the game range boundary and along the heads of the many drainages at a considerable distance back from the river. As is normally the case, during periods of severe drought and poor market, livestock operators find it difficult to reduce their herds to the carrying capacity of the range. Heavy use by livestock results in utilization of browse species not normally consumed during the regular grazing season. " "When ranges outside [the Game Range] are overutilized by livestock during drought periods such as we have experienced in the past two years, both cattle and sheep utilize most, if not all, of the preferred browse species even during the summer months forcing game onto other ranges . " "At the extreme west end of the game range, [our study area] grasses and sedges made no growth whatsoever this season or for that matter we can also include the season of 1955." "A few more coyotes have been observed on the area this period than for a number of years. Hunters during the deer season have likewise reported seeing more coyotes than in the past. Complaints are more frequent from sheep operators in or near the game range . " - During August, 116 fawns/100 female mule deer were observed. 1957 - "A number of mule deer checked were not in as good condition as in previous years and a few animals were found to be thin and a considerable number harbored lice. " "During the report period a control campaign has been undertaken to clean up or at least reduce the population of predators in that section of the Game Range lying west of the mouth of the Musselshell [includes our study area]. The PARC trapper on the south side of the river either poisoned with 1080, killed with cyanide or trapped a total of 54 coyotes and 19 bobcats during the period [September-December 1957] . " "The rabbit population, especially the jackrabbit, is down to the lowest point in game range history. It is becoming a rare occurrence when personnel report seeing a jack" . 1958 - "In our opinion, deer generally, as well as other big game species present, suffered to a considerable extent from lack of water and were probably in poorer physical condition at the end of the report period [January-April] than during any winter since the Game Range was established." 376 "... dried-out range lands did not offer a very succulent diet." (May-August). "On the West Unit, a long arid summer resulted in many waterholes and reservoirs drying up"... (Sept. -Dec.) [Author's Note:] During February 1958, an attempt was made to remove all deer from a 8.13 km2 pasture that had been fenced to receive a transplant of Rocky Mountain bighorn sheep (Janson 1958). This pasture was approximately 9 . 7 km west of the study area and within similar habitat. Fourteen mule deer were shot, two were known to have escaped through the fence, and a minimum of eight deer were known to have survived and remained in the pasture. These data provided a minimum density estimate of 7.7 mule deer /mi2 (2.95/km2) for an area near our study area during February 1958. 1959 - (Jan. -Apr.) ..."Severe winter conditions and particularly deep snow caused hardship for most all wildlife. By the latter part of February, many mule deer were in poor condition. Body weight appeared down and they seemed to lack the vigor and alertness of normal animals." "The mule deer situation remains precarious over the West Unit. The area was found to be uniformly overutilized to the point we are in danger of losing a considerable portion of the palatable forage available that deer are dependent upon." Data obtained during the winter indicates our doe-fawn ratio to be about 65 fawns per 100 does which is an indication the productivity of the herd has already degenerated to a serious point." "An early spring with abundant new growth of forage." (May-Aug. ) "The limited use of the 1080 baits by coyotes last winter [less than 5% eaten] should indicate that the population is relatively low. However, almost any night these animals can be heard singing at any location along the river". "Both cottontails and white-tailed jackrabbits are present in larger numbers. Increased road kills and more sight records seem to support this theory." "Body condition [of mule deer] was excellent and during the early part of the hunting season all animals had a wide layer of fat coating the body cavity and internal organs." 377 LITERATURE CITED Allen, E. O. 1968. Range use, foods, condition, and productivity of white-tailed deer in Montana. J. Wildl. Manage. 32:130-141. Anderson, A. E., D. E. Medin, and D. C. Bowden. 197 4. Growth and morphometry of the carcass, selected bones, organs, and glands of mule deer. Wildl. Monogr. 39:1-122. Andrewartha, H. G., and L. C. Birch. 1954. The distribution and abundance of animals. Univ. of Chicago Press, Chicago . Baker, R. R. 1978. The evolutionary ecology of animal migration. Holmes and Meier Publ . , Inc. New York, NY. 1012 pp. Barrett, M. W., J. W. Nolen, and L. D. Roy. 1982. Evaluation of a hand held net-gun to capture large mammals. Wildl. Soc. Bull. 10:108-114. Bartholomew, G. A. 1986. The role of natural history in contemporary biology. BioScience. 36:324-329. Bartmann, R. M., L. H. Carpenter, R. A. Garrott, and D. C. Bowden. 1986. Accuracy of helicopter counts of mule deer in pinyon- juniper woodland. Wildl. Soc. Bull. 14:356-363. Basile, J. V. and S. S. Hutchings. 1966. Twig diameter- length-weight relationships for bitterbrush. J. Range Manage. 19:34-38. Beasom, S. D. 1974. Relationships between predator removal and white-tailed deer net productivity. J. Wildl. Manage. 38:854-859. Beasom, S. L. , W. Evans, and L. Temple. 1980. The drive net for capturing western big game. J. Wildl. Manage. 44:478-480. , F. G. Leon III, and D. R. Synatzske. 1986. Accuracy and precision of counting white-tailed deer with helicopters at different sampling intensities. Wildl. Soc. Bull. 14:364-368. Beddington, J. D., and R. M. May. 1977. Harvesting natural populations in a randomly fluctuating environment. Science. 197:463-465. 378 Bergerud, A.T. 1971. The population dynamics of Newfoundland caribou. Wildl. Monogr. 25:1-55. . 1978. Caribou. Pp. 83-101 in Big game of North America. J.L. Schmidt and D.L. Gilbert (eds.). Stackpole Books, Harrisburg, Pa. Birch, L. C. 1957. The role of weather in determining the distribution and abundance of animals. Cold Spring Harbor Symp. on Quant. Biol. 22:203-218. . 1971. The role of environmental heterogeneity and genetical heterogeneity in determining distribution and abundance. Pp. 109-126 in Dynamics of Populations: Proceedings of the Advanced Study Institute on Dynamics of Numbers in Populations. P.J. den Boer and G.R. Grodwell (eds.). Centre for Agricultural Publishing and Documentation, Wageningen, the Netherlands. Birney, E. C, W. E. Grant, and D. D. Baird. 1976. Importance of vegetative cover to cycles of Microtus populations. Ecology. 57:1043-1051. Blaisdell, J. P. 1958. Seasonal development and yield of native plants on the Snake River plains and their relation to certain climatic factors. U.S.D.A. Tech. Bull. No. 1190. Washington, D.C. 68 pp. Botkin, D. B., J. M. Mellilo, and L. S. -Y. Wu . 1981. How ecosystem processes are linked to large mammal population dynamics. Pp. 373-388, in Dymanics of Large Mammal Populations. C.W. Fowler and T.D. Smith (eds.) John Wiley and Sons. New York. 477 pp. Bowyer, R. T. 1984. Sexual segregation in southern mule deer. J. Mammal. 65:410-417. Brown. D. L. 1947. Aerial census of deer and range inspection on Fort Peck Game Range. Mont. Fish and Game Dept., P-R Quart. Rept . 7:1-8. . 1948. Aerial inspection of Fort Peck Game Range (Phillips and Fergus Sub-Units). Mont. Fish and Game Dept., P-R Quart. Rept. 8:314-320. Bucsis, R. A. 1974. Ecological characteristics of the Armstrong mule deer winter range, Bridger Mountains, Montana. Unpubl . M.S. Thesis. Mont. State Univ., Bozeman. 104 pp. Burk, C. J. 1973. The Kaibab deer incident: a long- persisting myth. BioScience. 23:113-114. 379 Burroughs, R. D. 1961. The natural history of the Lewis and Clark expedition. Michigan St. Univ. Press, Lansing. 340 pp. Byers, C. R., R. K. Steinhorst, and P. R. Krausman. 1984. Clarification of a technique for analysis of utilization- availability data. J. Wildl. Manage. 48:1050-1053. Cable, D. R. 1975. Influence of precipitation on perennial grass production in the semidesert southwest. Ecology. 56:981-986. Cada, J. D. 1968. Populations of small mammals in central Montana with special reference to tentative sagebrush control sites. Unpubl . M.S. Thesis, Mont. State Univ., Bozeman. 52 pp. . 1985. Evaluations of the telephone and mail survey methods of obtaining harvest data from licensed sportsmen in Montana. Pp. 117-128 in S.L. Beasom and S.F. Roberson, eds . Game Harvest Management. Caesar Kleberg Wildlife Research Inst. Kingsville, Tex. Calder, N. 1979. Einstein's Universe. The Viking Press, New York, N.Y. 154 pp. Calhoun, J. B. 1963. The ecology and sociology of the Norway rat. U.S. Dept. Health, Education, and Welfare. Public Health Service. U.S. Govt. Print. Off., Bethseda, MD, 288 pp. Campbell, R. B. and C. J. Knowles . 1978. Elk-hunter- livestock interactions in the Missouri River Breaks . Mont. Fish and Game Dept. Job Final Report. Pro j . W-130- R-9, Job 1-6.5. 153 pp. Caughley, G. 1966. Mortality patterns in mammals. Ecology 47:906-918. . 1970. Eruption of ungulate populations with emphasis on Himalayan Thar in New Zealand. Ecology. 51:53-70. . 1976a. Plant-herbivore systems. Pp. 94-113 in R.M. May (ed.), Theoretical ecology: principles and applications. Blackwell, Oxford. . 1976b. Wildlife management and the dynamics of ungulate populations. Pp. 183-246 in T.H. Coaker, ed. Applied biology, Vol. I. Academic Press, London, U.K. . 1977. Analysis of vertebrate populations. John Wiley and Sons, New York. 234 pp. 380 1979. What is this thing called carrying capacity? pp. 2-8 in North American Elk: Ecology, Behavior, and Management (M.S. Boyce and L.D. Hayden-Wing, eds . ) , Univ. Wyoming Press, Laramie. 1980. A population of mammals. A review of: The George Reserve deer herd. Population Ecology of a K- selected species. Dale R. McCullough. Univ. of Michigan Press, Ann Arbor, 1979. 272 pp. Science 207:1338-1339. 1981a. Comments on natural regulation of ungulates (what constitues a real wilderness?). Wildl. Soc . Bull. 9:232-234 1981b. What we do not know about the dynamics of large mammals. Pp. 361-372 in Dynamic of large mammal populations. Charles W. Fowler and Tim D. Smith (eds.). John Wiley and Sons, New York. 477 pp. _, and J. Goddard. 1972. Improving the estimates from inaccurate censuses. J. Wildl. Manage. 36:135-140. Chaffee, G. B. 1954. An investigation of deer mortality in the Missouri Breaks. Mont. Fish and Game Dept . , P-R Rept. W-59-R-1, Job No. III-C, 4 pp. Cheatum, E. L. 1949. The use of corpora lutea for determining ovulation incidence and variations in fertility of the white-tailed deer. Cornell Vet. 39:282- 291. Chitty, D. 1955. Adverse effects of population density upon the viability of later generations. Pp. 57-68, in The Numbers of Man and Animals. J.B. Cragg and N.W. Pirie (eds.). Inst. of Biol., Oliver and Boyd, Ltd. Edinburgh, 152 pp. 1960. Population processes in the vole and their relevance to general theory. Can. J. Zool . 38:99-113. 1967a. The natural selection of self-regulatory behavior in animal populations. Proc . Ecol . Soc. Aust. 2:51-78. 1967b. What regulates bird populations? Ecology, 48:698-701. Christian, J. J., and D. E. Davis. 1964. Endocrines, behavior, and population. Science. 146:1550-1560. 381 Clark, C. 1976. Mathematical bioeconomics : the optimal management of renewable resources. J. Wiley and Sons, New York. 352 pp. Clary, W. P., M. B. Baker, Jr., P. F. O'Connell, T. N. Johnsen, Jr., and R. E. Campbell. 1974. Effects of pinyon- juniper removal on natural resource products and uses in Arizona. USDA Forest Service Research Paper RM- 128. 28 pp. Clutton-Brock, T. H., F. E. Guinness, and S. D. Albon. 1982. Red deer - Behavior and ecology of two sexes. Univ. of Chicago Press, Chicago. 378 pp. . 1983. The costs of reproduction to red deer hinds. J. Animal Ecology 52:367-383. Coblentz, B. E. 1977. Comments on deer sociobiology . Wildl . Soc. Bull. 5:67. Cockburn, A., and Lidicker, W. Z. Jr. 1983. Microhabitat heterogeneity and population ecology of a herbivorous rodent, Microtus calif ornicus . Oecologia. 59:167-177. Cohen, M. N., R. S. Malpass, and H. G. Klein. 1980. Introduction. Pp. ix-xx in Biosocial Mechanisms of Population Regulation. M.N. Cohen, R.S. Malpass, and H.G. Klein (eds.). Yale University Press, New Haven. 406 pp. Cole, G. F. 1958. Range survey guide. Montana Dept . of Fish and Game booklet. 18 pp. Multilith. . 1959. Key browse survey method. Proc . West. Assoc. Game and Fish Commissioners. 39:181-186. . 1961. Big game population dynamics and season setting. Montana Fish and Game Dept. Information Bulletin No. 14. Helena, MT. 7 pp. Conner, M. C, R. A. Lancia, and K. H. Pollock. 1986. Precision of the change-in-ratio technique for deer population management. J. Wildl. Manage. 50:125-129. Cook, R. S., M. White, D. 0. Trainer, and W. L. Glazener. 1971. Mortality of white-tailed deer fawns in south Texas. J. Wildl. Manage. 35:47-56. 382 Croze, H., A. K. K. Hillman, and E. M. Lang. 1981. Elephants and their habitats: How do they tolerate each other. Pp. 297-316 in Dynamics of Large Mammal Populations. C.W. Fowler and T.D. Smith (eds.). John Wiley and Sons. New York. 477 pp. Dahl, B. E. 1963. Soil moisture as a predictive index to forage yield for the sandhills range type. J. Range Manage. 16:123-132. Darwin, C. 1872. The Origin of Species. (Reprint of 6th Ed.). Random House, New York. Dasmann, R. F., and R. D. Taber. 1956. Behavior of Columbian black-tailed deer with reference to population ecology. J. Mammal. 37:143-164. Dasmann, W. 1971. If deer are to survive. The Wildlife Management Institute and Stackpole Books, Harrisburg, Pennsylvania. 128 pp. Davis, D. E. 1950. Malthus - a review for game managers. J. Wildl. Manage. 14:180-183. ., and R. L. Winstead. 1980. Estimating the numbers of wildlife populations . Pp. 221-246 in Wildlife Management techniques manual, 4th ed. (S.D. Schemnitz, ed.). The Wildl. Soc, Washington, D.C. Devos, A., P. Brokx, and V. Geist. 1967. A review of social behavior of the North American cervids during the reproductive period. Am. Midi. Nat. 77:390-417. DeVoto, B. 1953. Journals of Lewis and Clark. Houghton- Mifflin Company, Boston. Dood, A. R. 1978. Summer movements, habitat use, and mortality of mule deer fawns in the Missouri River Breaks, Montana. Unpubl. M.S. Thesis, Montana State Univ., Bozeman. 55 pp. Dusek, G. L. 1975. Range relations of mule deer and cattle in prairie habitat. J. Wildl. Manage. 39:605-616. . 1987. Ecology of white-tailed deer in upland ponderosa pine habitat in southeastern Montana. Prairie Nat. 19:1-17. . 1988. Population ecology and habitat relationships of white-tailed deer in river bottom habitat in eastern Montana. Job Final Rept., W-120-R-12-18 , Program I, Study BG-2.0, Job 3. 116 pp. 383 , R. J. Mackie, J. D. Herriges, Jr., and B. B. Compton . 1989. Population ecology of white-tailed deer along the lower Yellowstone River. Wildl. Monogr. 104:1-68. du Plessis, S. S. 1972. Ecology of Blesbok with special reference to productivity. Wildl. Monogr. 30:1-70. Edwards, R. Y. , and C. D. Fowle. 1955. The concept of carrying capacity. Trans. N. Am Wildl. and Nat. Res. Conf. 20:589-598. Ehrlich, P. R., and L. C. Birch. 1967. The "balance of nature" and "population control". Amer. Naturalist. 101:97-107. Eichorn, L. C, and C. R. Watts. 1984. Plant succession on burns in the river breaks of central Montana. Proc . Mont. Acad. Sci. 43:21-34. Eustace, C. D. 1971. Mule deer food habits and browse use study. Montana Fish and Game Dept . Job Final Report, Proj. W-130-R-1 and 2, Job. No. 1-7.1. 25 pp. Everitt, B. S. 1977. The analysis of contingency tables. Monogr. on Appl . Prob. and Stat., Chapman and Hall, London. 125 pp. Fisher, R. A. 1930. The genetical theory of natural selection. Oxford: Oxford University Press. Flook, D. R. 1970. Causes and implications of an observed sex differential in the survival of wapiti. Canadian Wildl. Ser. Rept . , Ser. 11. 71 pp. Floyd, T. J., L. D. Mech, and M. E. Nelson. 1979. An improved method of censusing deer in deciduous-coniferous forests. J. Wildl. Manage. 43:258-261. Foote, L. E. 1971. Wildlife management principles. Pp. 52- 55 in A Manual of Wildlife Conservation (R.D. Teague, ed.). The Wildl. Soc . , Washington, D.C. Fowler, C. W. 1981. Density dependence as related to life history strategy. Ecology 62:602-610. French, C. E., L. C. McEwen, N. D. Magruder, R. H. Ingram, and R. W. Swift. 1956. Nutrient requirements for growth and antler development in the white-tailed deer. J. Wildl. Manage. 20:221-232. Gaines, M. S., and L. R. McClenaghan, Jr. 1980. Dispersal in small mammals. Ann. Rev. Ecol . Syst., 11:163-196. 384 Gasaway, W. C, R. O. Stephenson, J. L. Davis, P. E. K. Shepherd, and 0. E. Burris . 1983. Interrelationships of wolves, prey, and man in interior Alaska. Wildl. Monogr. 84:1-50. Gavin, T. A., L. H. Suring, P. A. Vohs , and E. C. Meslow. 1984. Population characteristics, spatial organization, and natural mortality in the Columbian white-tailed deer. Wildl. Monogr. 91. 41 pp. Geist, V. 1971. Mountain Sheep - A Study in Behavior and Evolution. The Univ. of Chicago Press, Chicago. 383 pp. . 1981. Behavior: adaptive strategies in mule deer. Pages 157-223 in O.C. Wallmo, ed. Mule and black-tailed deer of North America. Univ. of Nebraska Press, Lincoln. 605 pp. Gieseker, L. F., C. B. Manifold, A. T. Strahorn, and 0. F. Bartholomew. 1953. Soil survey, central Montana. U.S.D.A. Montana Agric . Expt. Sta., Series 1940, No. 9:1- 133. Gilbert, F. F. 1966. Aging white-tailed deer by annuli in the cementum of the first incisor. J. Wildl. Manage. 30:200-202. Gleick, J. 1988. Chaos - Making a New Science. Penguin Books, New York, N.Y. 352 pp. Greene, H. W. 1986. Natural history and evolutionary biology. Pp. 99-108, Chap. 7 in Predator-prey relationships: Perspectives and approaches from the study of lower vertebrates . (M.E. Feder andG.V. Lauder, eds . ) Univ. of Chicago Press, Chicago. 198 pp. Griffith. B. 1988. Group predator defense by mule deer in Oregon. J. Mammal. 69:627-629. Haber, G. C. 1977. Socio-ecological dynamics of wolves and prey in a subarctic ecosystem. Unpubl . Ph.D. Thesis, Univ. of British Columbia, Vacouver. 817 pp. Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure, population control, and competition. Amer. Naturalist. 94:421-425. Hamilton, W. D. 1964. The genetical theory of social behavior: I. and II. J. of Theoretical Biology 7:1-52. 385 Hamlin, K. L. 1977. Population dynamics and habitat relationships of mule deer in the Bridger Mountains, Montana, 1955-1976. Ann. Conf . N.W. Section, TWS . , Kalispell, MT Feb. 1977. , and L. L. Schweitzer. 1979. Cooperation by coyote pairs attacking mule deer fawns. J. Mammal. 60:849-850. , R. J. Mackie, J. G. Mundinger, and D. F. Pac. 1982. Effect of capture and marking on fawn production in deer. J. Wildl. Manage. 46:1086-1089. , S. J. Riley, D. Pyrah, A. R. Dood, and R. J. Mackie. 1984. Relationships among mule deer fawn mortality, coyotes, and alternate prey species during summer. J. Wildl. Manage. 48:489-499. , and R. J. Mackie. 1987. Age-specific reproduction and mortality in female mule deer: implications to population dynamics. Proc . XVIIIth Congr. Int. Union Game Biolog., Jagielonian Univ., Krakow, Poland. In Press. Harvey, J. J., and R. W. Barbour. 1965. Home ranges of Microtus ochrogaster as determined by a modified minimum area method. J. Mammal. 46:398-402. Hawkins, R. E., L. D. Mortoglio, and G. G. Montgomery. 1968. Cannon-netting deer. J. Wildl. Manage. 32:191-195. ., and W. D. Klimstra. 1970. A preliminary study of the social organization of white-tailed deer. J. Wildl. Manage. 34:407-419. Hayne, D. W. 1949. Calculation of size of home range. J. Mammal 30:1-17. Hestbeck, J .B. 1982. Population regulation of cyclic mammals: the social fence hypothesis. Oikos 39:157-163. Hirst, S. M. 1975. Ungulate-habitat relationships in a South African Woodland/savanna ecosystem. Wildl. Monogr. 44:1-60. Hirth, D. H. 1977. Social behavior of white-tailed deer in relation to habitat. Wildl. Monogr. No. 53. 55 pp. Hobbs, N. T., D. L. Baker, J. E. Ellis, D. M. Swift, and R. A. Green. 1982. Energy- and nitrogen-based estimates of elk winter-range carrying capacity. J. Wildl. Manage. 46:12-21. 386 Hollibaugh, G. 1944. Fergus Unit - Quarterly Report for the 2nd Quarter - 1944. Mont. Fish and Game Dept . files, typewritten rept . 9 pp. Hornaday, W. T. 1908. Diversions in picturesque game-lands, Grand bad-lands and mule deer. in Scribner's Magazine July 1908. Howard, L. 0., and W. F. Fiske. 1911. The importation into the United States of the parasites of the gypsy moth and the brown-tail moth. U.S. Bur. Entom. Bull. No. 91. 344 pp. Hudson, P., and L. G. Browman . 1959. Embryonic and fetal development of the mule deer. J. Wildl . Manage. 23:295- 304. Huf faker, C. B., and P. S. Messenger. 1964. The concept and significance of natural control. pp. 74-117 in Biological Control of Insect Pests and Weeds. P. DeBach (ed.). Chapman and Hall, London. 84 4 pp. Hunter, R. F., and G. E. Davies . 1963. The effect of method of rearing on the social behavior of Scottish blackface hoggets. Anim. Prod. 5:183-194. Janson, R. G. 1958. Deer surveys. Montana Fish and Game Dept. Job Comp. Rept., Pro j . W-74-R-3, Job No. A-l. 13 pp. Jennrich, R. I., and F. B. Turner. 1969. Measurement of non-circular home range. J. Theoret . Biol. 22:227-237. Jorgensen, H. E., and R. J. Mackie. 1976. Evaluation of range survey methods, concepts, and criteria. Job Progress Report, Fed. Aid Pro j . W-120-R-7, Montana Dept. of Fish and Game. 6 pp. Keith, L. B. 1974. Some features of population dynamics in mammals. Proc . Internatl. Cong. Game Biol. 11:17-58. Kie, J. G., M. White, and F. F. Knowlton. 1979. Effects of coyote predation on population dynamics of white-tailed deer. Proc. Welder Wildl. Found. Symp. 1:65-82. , and M. White. 1985. Population dynamics of white- tailed deer (Odocoileus virginianus) on the Welder Wildlife Refuge, Texas. The Southwestern Naturalist. 30:105-118. 387 King, M. M., and H. D. Smith. 1980. Differential habitat utilization by the sexes of mule deer. Great Basin Natur. 40:273-281. Klein, D. R. 1970. Food selection by North American deer and their response to overutilization of preferred plant species, pp. 25-46 in Animal populations in relation to their food resources (A. Watson ed.) Blackwell Scientific Publications, Oxford. 477 pp. . 1985. Population ecology: the interaction between deer and their food supply. Biol. Deer Prod. 22:13-22. , and H. Strandgaard. 1972. Factors affecting growth and body size of roe deer. J. Wildl. Manage. 36:64-79. Klinger, R. C, M. J. Kutilek, and H. S. Shellhammer. 1989. Population responses of black-tailed deer to prescribed burning. J. Wildl. Manage. 53:863-871. Knowles, C. J. 1975. Range relationships of mule deer, elk, and cattle in a rest-rotation grazing system during summer and fall. Unpubl . M.S. Thesis, Mont. State Univ., Bozeman. Ill pp. . 1976. Observations of coyote predation on mule deer and white-tailed deer in the Missouri River Breaks, 1975- 1976. Pp. 117-138 in Montana deer studies, Job Progress Rept., P-R Pro j . W-120-R-7. Mont. Fish and Game Dept . , Helena. 170 pp. Knowlton, F. F. 1976. Potential influence of coyotes on mule deer populations. Pp. 111-118. in Mule deer decline in the West - a symposium. Utah State Univ., Logan. Koch, E. 1941. Big game in Montana from early historical records. J. Wildl. Manage. 5:357-370. Komberec, T. J. 1976. Range relationships of mule deer, elk, and cattle in a rest-rotation grazing system during winter and spring. Unpubl. M.S. Thesis. Mont. State Univ., Bozeman. 7 9 pp. Krebs, C. J. 1972. Ecology: the experimental analysis of distribution and abundance. Harper and Row, New York. 693 pp. . 1978. A review of the Chitty Hypothesis of population regulation. Can. J. Zool . 56:2463-2480. Lack, D. 1954. The natural regulation of animal numbers. Clarendon Press, Oxford, London. 388 . 1966. Population studies of birds. Oxford Univ. Press, London. 341 pp. Leopold, A. 1933. Game management. New York: Scribner's. 4 81 pp . Leopold, A. S. 1966. Adaptability of animals to habitat change. Pp. 69-81 in Future Environments of North America. (F.F. Darling and J. P. Milton, Eds.). Natural History Press, Doubleday and Co., Inc. LeResche, R. E., and R. A. Rausch. 1974. Accuracy and precision of aerial moose censusing. J. Wildl. Manage. 38:175-182. Levins, R. 1966. The strategy of model building in population biology. Am. Sci. 54:421-431. Lidicker, W. Z., Jr. 1978. Regulation of numbers in small mammal populations -- historical reflections and a synthesis. Pp. 122-141 in Populations of small mammals under natural conditions (D.P. Snyder, ed.). Spec. Publ . Ser., Pymatuning Lab. Ecol . , Univ. Pittsburgh, 5:1-237. . 1985. Dispersal. Pp. 420-454 in Biology of New World Hicrotus (R.H. Tamarin, ed . ) . Spec. Publ. Amer. Soc . Mamm., 8:1-893. . 1988. Solving the enigma of microtine "cycles". J. Mammal. 69:225-235. Loveless, C. M. 1967. Ecological characteristics of a mule deer winter range. Tech. Bull. No. 20. Colorado Game, Fish and Parks Dept . , Denver. 124 pp. Lund, R. E. 1983. A user's guide to MSUSTAT — An interactive statistical analysis package. CP/M Version. Research and Development Institute, Inc., Montana State University, Bozeman. Lyon, L. J. 1968. Estimating twig production of serviceberry from crown volumes. J. Wildl. Manage. 32:115-119. Mackie, R. J. 1965. Range ecology and relations of mule deer, elk, and cattle in the Missouri River Breaks, Montana. Unpubl . Ph.D. Thesis, Montana State Univ., Bozeman. 229 pp. . 1970. Range ecology and relations of mule deer, elk and cattle in the Missouri River Breaks, Montana. Wildl. Monogr. 20:1-79. 389 1973a. What we've learned about our most popular game animal. Montana Outdoors. 4(5): 15-21. 1973b. Response of four browse species to protection. Statewide big game range research. Job. Prog. Rept . , Fed. Aid Pro j . W-120-R-4, Study No. 28.01, Job B6-2.01. Montana Dept . Fish and Game, Helena. 30 pp. Multilith. 1975. Evaluation of the Key browse survey method. Paper presented at the Fifth Mule Deer Workshop, Albuquergue, New Mexico, February 1975. 17 pp. Available from author. 1976. Mule deer population ecology, habitat relationships, and relations to livestock grazing management and elk in the Missouri River Breaks, Montana. Pp. 67-94 in Montana deer studies. Job Compl . Rept., P-R Pro j . W-120-R, Study BG-1.4, Jl . Montana Fish and Game Dept., Helena. 170 pp. 1978. Natural regulation of mule deer populations. Pp. 112-125 in The Natural Regulation of Wildlife Populations (F.L. Bunnell, D.S. Eastman, and J.M. Peck, eds . ) . Forest, Wildl. and Range Exp. Sta., University of Idaho, Moscow. Proc . No. 14: 112-125. _, K. L. Hamlin, and J. G. Mundinger. 1976. Habitat relationships of mule deer in the Bridger Mountains, Montana. Montana Fish and Game Dept., Job Prog. Rept., Proj. W-120-R-6, Job No. BG-2.01 (Supplement). 44 pp. _, D. F. Pac, and H. E. Jorgensen. 1979. Population ecology and habitat relationships of mule deer in the Bridger Mountains, Montana. Pp. 69-113 in Montana deer studies. Job Prog. Rept., Proj. W-120-R-10. _, K. L. Hamlin, H. E. Jorgensen, J. G. Mundinger, and D. F. Pac. 1980. Management implications and recommendations. Pp. 192-205, in Montana Deer Studies. Job Prog. Rept., Fed. Aid Proj. W-120-R-11. Montana Dept. Fish and Game, Helena. 205 pp. _, , and D. F. Pac. 1981. Census methods for mule deer. Pp. 97-106 in Symposium on census and inventory methods (F.L. Miller and A. Gunn, eds.). For., Wildl. and Range Exp. Sta., Univ. of Idaho, Moscow. 217:97-106 _, G. L. Dusek, K. L. Hamlin, H. E. Jorgensen, D. F. Pac, and A. K. Wood. 1985. Management Recommendations, pp. 1-8, in Montana Deer Studies. Job Prog. Rept., Fed. Aid Proj. W-120-R-16. Montana Dept. of Fish, Wildl. and Parks, Helena. 132 pp. 390 Maelzer, D. A. 1970. The regression of log Nn+1 on log Nn as a test of density dependence: an exercise with computer- constructed, density-independent populations. Ecology. 51:810-822. Martin, P. R. 1972. Ecology of skunkbush sumac (Rhus trilobata Nutt.) in Montana with special reference to use by mule deer. Unpubl . M.S. thesis. Mont. State Univ., Bozeman. 97 pp. Martinka, C. J. 1978. Ungulate populations in relation to wilderness in Glacier National Park, Montana. Trans. North Am. Wildl. Nat. Resour. Conf . 43:351-357. Mautz, W. W. 1978. Sledding on a brushy hillside: the fat cycle in deer. Wildl. Soc . Bull. 6:88-90. , H. Silver, J. B. Holter, H. H. Hayes, and W. E. Urban, Jr. 1976. Digestibility and related nutritional data for seven northern deer browse species. J. Wildl. Manage. 40:630-638. Maynard-Smith, J. 1964. Group selection and kin selection. Nature. 201:1145-1147. McCulloch, C. Y. 1974. Control of pinyon- juniper as deer management measure in Arizona. Final report P-R Project W-18-R, WP4 , J2 and 7. Phoenix: Arizona Game and Fish Department. 32 pp. McCullough, D. R. 1979. The George Reserve deer herd, population ecology of a K-selected species. Univ. of Michigan Press, Ann Arbor. 271 pp. . 1983. Rate of increase of white-tailed deer on the George Reserve: A response. J. Wildl. Manage. 47:1248- 1250. . 1984. Lessons from the George Reserve. Pp. 211-242 in White-tailed deer ecology and management. L.K. Halls (ed.). The Wildlife Management Institute, Stackpole Books, Harrisburg, Pa. McEwen, L. C., C. E. French, N. D. Magruder, R. W. Swift, and R. H. Ingram. 1957. Nutrient requirements of the white- tailed deer. Trans. North Am. Wildl. Nat. Resour. Conf. 22:119-132. McNab, J. 1985. Carrying capacity and related slippery shibboleths. Wildl. Soc. Bull. 13:403-410. 391 Mech, L. D., and G. D. Delgiudice. 1985. Limitations of the marrow-fat technique as an indicator of body condition. Wildl. Soc. Bull. 13:204-206. Medin, D. E., and A. E. Anderson. 1979. Modeling the dynamics of a Colorado mule deer population. Wildl. Monogr. 68:1-77. Miller, F. L. 1970. Distribution patterns of black-tailed deer (Odocoileus hemionus columbianus) in relation to environment. J. Mammal. 51:248-260. . 1974. Four types of territoriality observed in a herd of black-tailed deer. In. The behavior of ungulates and its relation to management (V. Geist and F.R. Walter, eds.). 2:644-660. IUCN New Series Publ . No. 24. Morges, Switzerland: IUCN. Milne, A. 1957. Theories of natural control of insect populations. Cold Spring Harbor Symp. on Quant. Biol. 22:253-271. Mohr, C. 0. 1947. Table of equivalent populations of North American small mammals. Am. Midi. Nat. 37:223-249. Morton, M. A. 1976. Nutritional values of major mule deer winter forage species in the Bridger Mountains, Montana. Unpubl . M.S. Thesis, Montana State University, Bozeman. 104 pp. Mundinger, J. G. 1981. White-tailed deer reproductive biology in the Swan Valley, Montana. J. Wildl. Manage. 45:132-139. Murdoch, W. W. 1966. "Community structure, population control, and competition" - a critique. Amer. Naturalist. 100:219-226. Murie, 0. J. 1935. Report on the Fort Peck Migratory Bird Refuge. Typewritten report, Charles M. Russell Nat. Wildl. Refuge files, Lewistown, Montana. 33 pp. Murphy, D. A., and J. A. Coates . 1966. Effects of dietary protein on deer. Trans. N.A. Wildl. and Nat. Res. Conf . 31:129-138. Murray, B. G., Jr. 1979. Population Dynamics - Alternative Models. Academic Press. New York. 212 pp. 392 Nellis, C. H. 1977. Evaluate response of forage species and evaluate response of deer to irrigation/fertilization. Project Completion Report, P-R Project W-160-R-5, Study- No. 5, J2 , 3, 4. Boise: Idaho Dept . Fish and Game. 5 pp. Nelson, M. E., and L. D. Mech. 1981. Deer social organization and wolf predation in northeastern Minnesota. Wildl. Manage. 77:1-53. Nicholson, A. J. 1933. The balance of animal populations. J. An. Ecol. 2:132-178. , and V. A. Bailey. 1935. The balance of animal populations. Part I. Proc . Zool Soc . London, pp. 551- 598. Nudds , T. D. 1980. Forage "preference": theoretical considerations of diet selection by deer. J. Wildl. Manage. 44:735-740. O'Gara, B. W., and R. B. Harris. 1988. Age and condition of deer killed by predators and automobiles. J. Wildl. Manage. 52:316-320. Ordway, L. L., and P. R. Krausman. 1986. Habitat use by desert mule deer. J. Wildl. Manage. 50:677-683. Ostfeld, R. S., W. Z. Lidicker, Jr., and E. J. Heske. 1985. The relationship between habitat heterogeneity, space use, and demography in a population of California voles. Oikos, 45:433-442. , and L. L. Klosterman. 1986. Demographic substructure in a California vole population inhabiting a patchy environment. J. Mammal. 67:693-704. Overton, W. S., and D. E. Davis. 1969. Estimating the numbers of animals in wildlife populations. Pp. 403-456 in Wildlife management techniques, 3rd ed. (R.H. Giles, Jr., ed. ) . The Wildl. Soc., Washington, D.C. Ozoga, J. J., L. J. Verme, and C. S. Bienz . 1982a. Parturition behavior and territoriality in white-tailed deer: impact on neonatal mortality. J. Wildl. Manage. 46:1-11. , and . 1982b. Physical and reproductive characteristics of a supplementally-fed white-tailed deer herd. J. Wild. Manage. 46:281-301. 393 , and . 1984. Effect of family-bond deprivation on reproductive performance in female white-tailed deer. J. Wildl. Manage. 48:1326-1334. Pac, D. F. 1976. Distribution, movements and habitat use during spring, summer, and fall by mule deer associated with the Armstrong winter range, Bridger Mountains, Montana. Unpubl . M.S. Thesis, Montana State University, Bozeman. 120 pp. . 1979. Statewide carcass collection. Montana Fish and Game Dept., Job Prog. Rept., Pro j . W-120-R-10. 17 pp. , R. J. Mackie, and H. E. Jorgensen. (in prep.). Mule deer population organization, behavior, and dynamics in a northern Rocky Mountain environment. Peek, J.M. 1980. Natural regulation of ungulates (what constitutes a real wilderness?). Wildl. Soc . Bull. 8:217-227. . 1986. A review of wildlife management. New Jersey: Prentice-Hall. 486 pp. , F. D. Johnson, and N. M. Pence. 1978. Successional trends in a ponderosa pine/bitterbrush community related to grazing by livestock, wildlife and to fire. J. Range. Manage. 31:49-53. Pellew, R. A. 1984. Food consumption and energy budgets of the giraffe. J. Appl . Ecol . 21:141-159. Pennak, R. W. 1964. Collegiate Dictionary of Zoology. Ronald Press Co., New York, 583 pp. Peterson, R. 0. 1977. Wolf ecology and prey relationships on Isle Royale. U.S. Natl. Park Serv. Sci. Monogr. Ser. 11. 210 pp. Petrusewicz, K. 1963. Population growth induced by disturbance in the ecological structure of the population. Ekologia Polska, A, XI, 87-125. Pieper, R. D. 1978. Measurement techniques for herbaceous and shrubby vegetation. Multilith. New Mexico State Univ., Las Cruces . 148 pp. Pimentel, D. 1961. Animal population regulation by the genetic feed-back mechanism. Amer. Naturalist. 95:65- 79. 394 Porter, W. F., and K. E. Church. 1987. Effects of environmental pattern on habitat analysis. J. Wildl. Manage. 51:681-685. Pyrah, D. 1984. Social distribution and population estimates of coyotes in north-central Montana. J. Wildl. Manage. 48:679-690. . 1987. American pronghorn antelope in the Yellow Water Triangle, Montana. Montana Dept. of Fish, Wildlife and Parks Tech. Bull. 121 pp. Ransom, A. B. 1965. Kidney and marrow fat as indicators of white-tailed deer condition. J. Wildl. Manage. 29:397- 398. Rasmussen, G. P. 1985. Antler measurements as an index to physical condition and range quality with respect to white-tailed deer. N.Y. Fish and Game Jour. 32:97-113. Ratcliffe, F. N. 1958. Factors involved int the regulation of mammal and bird populations. The Australian J. of Sci. 21:79-87. Ray, A. A. 1982. SAS User's Guide. SAS Inst., Inc., Cary, North Carolina. 921 pp. Regelin, W. L. 1975. Middle Park cooperative deer study: Junction Butte wildlife habitat improfement project. In Game Research Report, July 1975, part 2, pp. 265-294. Denver: Colorado Division of Wildlife. 314 pp. Rice, W. R., and J. D. Harder. 1977. Application of multiple aerial sampling to a mark-recapture census of white- tailed deer. J. Wildl. Manage. 41:197-206. Riley, S. J. 1982. Survival and behavior of radio-collared mule deer fawns during summers, 1978-1980, in the Missouri River Breaks, Montana. Unpubl . M.S. Thesis, Montana State Univ., Bozeman. 59 pp. , and A. R. Dood. 1984. Summer movements, home range, habitat use, and behavior of mule deer fawns. J. Wildl. Manage. 48:1302-1310. Riney, T. 1955. Evaluating condition of free-ranging red deer (Cervus elaphus) , with special reference to New Zealand. New Zealand J. Sci. and Technol . 36 (Sec. B):429-463. . 1982. Study and management of large mammals. J. Wiley and Sons. New York. 552 pp. 395 Robinette, W. L. 1966. Mule deer home range and dispersal in Utah. J. Wildl. Manage. 30:335-349. , D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. J. Wildl. Manage. 21:134-153. , C. H. Baer, R. E. Pillmore, and C. E. Knittle. 1973. Effects of nutritional change on captive mule deer. J. Wildl. Manage. 37:312-326. , N. V. Hancock, and D. A. Jones. 1977. The Oak Creek mule deer herd in Utah. Resource Publication 77-15. Salt Lake City: Utah Division of Wildlife. 148 pp. Robson, D. S., and H. A. Regier. 1964. Sample size in Petersen mark-recapture experiments. Trans. Amer. Fish Soc. 93:215-226. Rogers, L. L. 1987. Effects of food supply and kinship on social behavior, movements, and population growth of black bears in Northeastern Minnesota. Wildl. Monogr. 97:1-72. Rogler, G. A., and J. H. Haas. 1947. Range production as related to soil moisture and precipitation on the northern Great Plains. J. Am. Soc. Agron . 39:378-389. Rowland, S. J. 1944. The occurrence in winter of milk with a low content of solids - not fat. J. Dairy Research. 13:261-266. Runge, W., and G. Wobeser. 1975. A survey of deer winter mortality in Saskatchewan. Wildl. Rep. 4. Regina: Saskatchewan Department of Tourism and Renewable Resources. 22 pp. Salwasser, H., S. A. Hall, and G. A. Ashcraft. 1978. Fawn production and survival in the North Kings River deer herd. Calif. Fish and Game. 64:38-52. SAS Institute, Inc. 1985. SAS user's guide: statistics. Version 5 ed. SAS Inst., Inc., Cary, North Carolina. 956 pp. Shaller, G. 1967. The deer and the tiger. Univ. of Chicago Press, Chicago. Shields, W. M. 1983. Optimal inbreeding and the evolution of philopatry. Pp. 132-159 in The ecology of animal movement. I.R. Swingland and P.J. Greenwood (eds.). Clarendon Press, Oxford. 396 Shiflet, T. N., and H. E. Dietz. 1974. Relationship between precipitation and annual rangeland herbage production in southeastern Kansas. J. Range Manage. 27:272-276. Short, H. L. 1981. Nutrition and metabolism. Pp. 99-127 in O.C. Wallmo, ed. Mule and black-tailed deer of North America. Univ. of Nebraska Press, Lincoln. 605 pp. , J. D. Newsom, G. L. McCoy, and J. F. Fowler. 1969. Effects of nutrition and climate on southern deer. Trans. N. Am. Wildl . Conf . 34:137-145. , R. M. Blair, and C. A. Segelquist. 1974. Fiber composition and forage digestibility by small ruminants. J. Wildl. Manage. 38:197-209. , W. Evans, and E. L. Boeker. 1977. The use of natural and modified pinyon pine-juniper woodlands by deer and elk. J. Wildl. Manage. 41:543-549. Silver, H., J. B. Holter, and H. H. Hayes. 1969. Fasting metabolism of white-tailed deer. J. Wildl. Manage. 33:490-498. Sinclair, A. R. E. 1977. The African buffalo: A study of resource limitation of populations. Chicago: University of Chicago Press. . 1981. Environmental carrying capacity and the evidence for overabundance. Pp. 247-257 in Problems in Management of Locally Abundant Wild Mammals. P. A. Jewell and S. Holt (eds.). Academic Press, New York. Slobodkin, L. B., F. E. Smith, and N. G. Hairston. 1967. Regulation in terrestrial ecosystems, and the implied balance of nature. Amer . Naturalist. 101:109-124. Smith, H. S. 1935. The role of biotic factors in the determination of population densities. J. Econ. Entomol. 28:873-898. Smith, C. A. 1976. Deer sociobiology - some second thoughts. Wildl. Soc. Bull. 4:181-182. Smith, R. H., and A. LeCount . 1979. Some factors affecting survival of desert mule deer fawns. J. Wildl. Manage. 43:657-665. 397 Smith, T. D., and C. W. Fowler. 1981. An overview of the study of the population dynamics of large mammals. Pp. 1-18 in Dynamics of Large Mammal Populations. C.W. Fowler and T.D. Smith (eds.). John Wiley and Sons. New York. 477 pp. Smith, W. P. 1983. A bivariate normal test for elliptical home-range models: Biological implications and recommendations. J. Wildl. Manage. 47:613-619. Smoliak, S. 1986. Influence of climatic conditions on production of Stipa-Bouteloua prairie over a 50-year period. J. Range Manage. 39:100-103. Smuts, G. L. 1978. Interrelations between predators, prey, and their environment. BioScience. 28:316-320. Snedecor, G. W., and W. G. Cochran. 1967. Statistical methods. 6th ed. Ames: Iowa State University Press. Sneva, F. A. 1982. Relation of precipitation and temperature with yield of herbaceous plants in eastern Oregon. Int. J. Biometeor. 26:263-276. St. Amant , J. L. S. 1970. The detection of regulation in animal populations. Ecology. 51:823-828. Steigers, W. D., Jr., and J. T. Flinders. 1980. Mortality and movements of mule deer fawns in Washington. J. Wildl. Manage. 44:381-388. Stout, G. G. 1982. Effects of coyote reduction on white- tailed deer productivity on Fort Sill, Oklahoma. Wildl. Soc. Bull. 10:329-332. Talbot, L. M., and M. H. Talbot. 1963. The Wildebeest in Western Masailand, East Africa. Wildl. Monogr. 12:1-88. Tanner, J. T. 1966. Effects of population density on growth rates of animal populations. Ecology. 47:733-745. Teer, J. G., J. W. Thomas, and E. A. Walker. 1965. Ecology and management of white-tailed deer in the Llano Basin of Texas. Wildl. Monogr. 15:1-62. Terman, C. R. 1962. Spatial and homing consequences of the introduction of aliens into semi-natural populations of prairie deermice. Ecology. 43:216-223. . 1963. The influence of differential early social experience upon distribution within populations of prairie deermice. Anim. Behav. 11:246-262. 398 Thompson, W. R. 1929. On natural control. Parasitology. 21:269-281. Thompson, C. B., J. B. Holter, H. Hayes, H. Silver, and W. Urban, Jr. 1973. Nutrition of white-tailed deer. I. Energy requirements of fawns. J. Wildl. Manage. 37:301- 311. Thomson, W., and A. M. Thomson. 1953. Effect of diet on milk yield of the ewe and growth of her lamb. British J. Nutrition. 7:263-274. Trainer, C. E., J. C. Lemos, T. P. Kistner, W. C. Lightfoot, and D. E. Toweill. 1981. Mortality of mule deer fawns in southeastern Oregon, 1968-1979. Oreg. Dept . Fish, Wildl., Res. and Dev. Sect. Wildl. Res. Rep. 10. 113 pp. Trout, R. G. 1978. Small mammal abundance and distribution in the Missouri River Breaks, Montana. Unpubl . M.S. Thesis, Montana State Univ., Bozeman. 64 pp. Tschache, 0. P. 1970. Effects of ecological changes induced by various sagebrush control techniques on small mammal populations. Unpubl. M.S. Thesis, Montana State Univ., Bozeman. 51 pp. Uvarov, B. P. 1931. Insects and climate. Trans. Ent . Soc . Lond. 79:1-247. Van Ballenberghe, V. 1980. Utility of multiple equilibrium concepts applied to population dynamics of moose. Proc . N. Am. Moose Conf. Works. 16:571-586. Verme, L. J. 1962. Mortality of white-tailed deer fawns in relation to nutrition. Pp. 15-38. In Proc. Nat ' 1 . White-tailed Deer Disease Symp., I. Univ. of Georgia, Athens . . 1965. Reproduction studies on penned white-tailed deer. J. Wildl. Manage. 29:74-79. . 1967. Influence of experimental diets on white-tailed deer reproduction. Trans. N. Am. Wildl. Nat. Resour. Conf. 32:405-420. . 1969. Reproductive patterns of white-tailed deer related to nutritional plane. J. Wildl. Manage. 33:881- 887. . 1988. Niche selection by male white-tailed deer: an alternative hypothesis. Wildl. Soc. Bull. 16:448-451. 399 , and J. C. Holland. 1973. Reagent-dry assay of marrow fat in white-tailed deer. J. Wildl. Manage. 37:103-105. , and D. E. Ullrey. 1984. Physiology and nutrition. pp. 91-118 in (L. K. Halls, ed. ) White-tailed deer ecology and management. The Wildlife Management Institute and Stackpole Books, Harrisburg, Pennsylvania. Wallmo, 0. C, L. H. Carpenter, W. L. Regelin, R. B. Gill, and D. L. Baker. 1977. Evaluation of deer habitat on a nutritional basis. J. Range Manage. 30:122-127. , and W.L. Regelin. 1981. Rocky Mountain and intermountain habitats . Part 1 . Food habits and nutrition. Pp. 387-398 in O.C. Wallmo, ed., Mule and black-tailed deer of North America. Univ. of Nebraska Press, Lincoln. Walker, B. H., and I. Noy-Meir. 1979. Stability and resilience of savanna ecosystems. In Symposium on dynamic changes in savannah ecosystems. S.C.O.P.E., Pretoria. Watson, A., and R. Moss. 1970. Dominance, spacing behavior and aggression in relation to population limitation in vertebrates. Pp. 167-218 in Animal Populations in Relation to their Food Resources. A. Watson (ed.). Blackwell Scientific Pub., Oxford. Wehausen, J. D., V. C. Bleich, B. Blong, and T. L. Russi. 1987. Recruitment dynamics in a southern California mountain sheep population. J. Wildl. Manage. 51:86-98. Welch, B. L., and D. Andrus . 1977. Rose hips — a possible high-energy food for wintering mule deer? U.S.D.A., Forest Service, Res. Note INT-221. 6 pp. Intermountain For. and Range Exp. Stat., Ogden, Utah. White, T. C. R. 1978. The importance of a relative shortage of food in animal ecology. Oecologia. 33:71-86. White, L. M. 1979. Relationship between meteorological measurements and flowering of index species to flowering of 53 plant species. Agricultural Meteorology. 20:189- 204. Wielgolaski, F. E. 1974. Phenology in agriculture. Pp. 369- 381 in Phenology and seasonality modeling. H. Lieth (ed.). Springer-Verlog, New York. 444 pp. 400 Wolff, J. 0. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecol. Monogr . 50:111-130. Wood, A. J., I. M. Cowen, and H. C. Nordan. 1962. Periodicity of growth in ungulates as shown by deer of the genus Odocoileus . Canada J. Zool . 40:593-603. Wood, A. K. 1987. Ecology of a prairie mule deer population. Ph.D. Thesis, Montana State Univ., Bozeman. 205 pp. , R. J. Mackie, and K. L. Hamlin. 1989. Ecology of sympatric populations of mule deer and white-tailed deer in a prairie environment. Montana Dept. Fish, Wildlife and Parks Technical Bulletin. 97 pp. Workman, G. W. , and J. B. Low, eds . 1976. Mule deer decline in the west: A symposium. Utah State Univ., College of Nat. Resources, and Agr. Exp. Stat., Logan. 134 pp. Wynne-Edwards, V. C. 1965. Self-regulating systems in populations of animals. Science. 147:1543-1548. Youmans, H. B. 1979. Habitat use by mule deer of the Armstrong winter range. Unpubl . M.S. thesis, Montana State Univ., Bozeman. 66 pp. Zar, J. H. 1984. Biostatistical analysis, second edition. Prentice-Hall, Inc. 718 pp. 401