(Contents)

{short description of image}

PREFACE

I have long had a strong desire to pursue a Ph. D. on roe deer habitat selection. I am therefore greatly indebted to my supervisors, professors Rolf A. Ims and Eivind Østbye, for all their help, back-up, inspiration and good friendship that made it possible! Without their generous and cooperative attitude and help, this thesis would never have been realized. The great hospitality of Vesla and Eivind Østbye in Lier has been appreciated during the course of the study. I will also thank Rolf A. Ims for sending me on a very inspiring trip to Lyon, France. I thank Nigel G. Yoccoz for letting me stay at his home, for his friendly attempts to "kill" a Scandinavian stomach at the restaurants in Lyon and my legs on a steep hike in the Alps! I further thank Jean-Michel Gaillard for sharing his knowledge of roe deer with me and for showing me his field site at Trois Fontaines.

Helpful expert advice about how to capture roe deer was given by Göran Cederlund, Petter Kjellander and Arild Reitan. To capture roe deer is time consuming and hard work. If it had not been for the extremely unselfish and generous voluntary help of my "hillbilly" friend Tom Einar Øverby in Øverskogen, this work would not have been as successful. Night or day, he has been there with his entusiasm and good humour! I wish also to thank Knut Arild Fuglerud, Leif Erik Horn and others in the Sylling community for help during various parts of the project and the landowners in Lier for their very cooperative attitude. I am also grateful to graduate students Per Kristian Larsen, Lise-Berith Lian, Kristin Lodgaard and Ann-Cathrin Nergård who helped me during part of the field work, and Bjørn Helge Bjørnsen who gave me unlimited access to his unpublished data!

Vidar Holthe and Morten Muus Falck gave me advice about wildlife and forest management and helped in raising funding for field work. I will thank Ivar Mysterud for lending me radio-telemetry and other field equipment, without it, this project would never have started. I also wish to thank all the nice folks at the Division of Zoology for making this period so enjoyable, especially Arild O. Gautestad, Ivar Mysterud, Iver Mysterud and Jerry Thomas Warren for lunch discussions, Ottar N. Bjørnstad, Rolf A. Ims and Nigel G. Yoccoz for bearing my (to them) silly statistical questions. I further thank Rolf A. Ims, Jerry Thomas Warren, Nigel G. Yoccoz and Eivind Østbye for valuable comments to the introductory chapter.

Financial support was given by the Norwegian Research Council through a three year scholarship (1996-1998), to which I am very appreciative. I am also appreciative to Viltfondet centrally (Directorate for Nature Management), the County Officers in Buskerud, Hedmark, Oppland, Telemark and Vestfold Counties, Utviklingsfondet for skogbruket, NORSKOG, and Norges Skogeierforbund who supported the field work.

I will thank my parents, Åse and Ivar Mysterud, for making me "love" the deep forests and showing me my first roe deer, and my brother, Iver, his wife, Marianne, and their four-legged "son", the dog Rask Mysterud, for friendship and many nice hikes over the years. Last but not least, I sincerely thank my wife, Rønnaug S. Mysterud, for her patience during long periods of field work, for her help during capturing animals and for always being supportive in every way!



Blindern, February 1998

Atle Mysterud



CONTENTS

PREFACE

CONTENTS

ABSTRACT

LIST OF INDIVIDUAL PAPERS

INTRODUCTION

The characteristics of the species (Ch. of sp.)

STUDY AREAS

The Lier valley, Buskerud

Flatdal, Telemark

FRAMEWORK, SAMPLING AND METHODS

Spatial scaling

Temporal scaling

Sampling and statistical analysis

Trapping

Estimating resource levels

Identifying trade-offs

Resource use

Habitat use/availability

RESULTS AND DISCUSSION

Resource use strategies

Bed-site selection during winter (paper I)

Bed-site selection during summer (paper II)

Forage site selection and food preference during winter (paper III)

Effect of snow on forage site selection (paper IV)

Habitat use/availability, trade-off and temporal scale

Functional responses in habitat use (paper V)

Resource availability and habitat rankings (paper VI)

Space use and social organization

Home range size and the mating system (paper VII)

Seasonal migration pattern and the clan system (paper VIII)

CONCLUSIONS

REFERENCES



ABSTRACT

The ecological processes and mechanisms that determine an animal's use of resources are scale dependent. For example, patch/habitat selection within the home range may depend on the shifting phases in the daily activity cycle of an animal, while the more infrequent home range shifts at the landscape scale may result from seasonal migrations. Patterns of habitat use may also reflect various trade-offs, e.g., between foraging and predation risk. However, these insights have only to a limited extent been incorporated into the analysis of habitat selection by cervids, such as the European roe deer ( Capreolus capreolus ), an important game species.

Habitat selection by roe deer was studied relative to the spatial distribution of food and cover using tracking, radio-telemetry or a combination thereof transcending different spatial and temporal scales mainly in Lier, Buskerud county, in Norway. In the first section, I tested several hypotheses regarding use of cover as a thermal strategy, an antipredator strategy and/or as a strategy to find areas with shallow snow. In the second section, a new method for analysing habitat selection in trade-off situations is presented, and I tested if habitat rankings based on roe deer habitat selection reflect resource availability. In the third section, patterns of space use are tested for sex-dependence.

During winter, roe deer foraged mainly in open habitats, but only bed-sites far from the forage area had denser canopy cover than that measured at random locations. Roe deer more often sought cover when bedding during cold rather than warm periods and when duration of the bedding period was long rather than short. During summer, roe deer left feeding areas more often before bedding during warm weather. However, roe deer did not select a more covered bed-site during periods of heat stress, rather they selected a more humid bedding substrate, probably to increase heat loss through conduction. Selection of forage site during winter was to a lesser extent influenced by temperature than was selection of bed-sites. Roe deer selected more open habitats at night, and avoided feeding close to human settlement during day. As snow depth increased, they foraged closer to human settlement, suggesting a trade-off between energetic demands and antipredator behaviour. Roe deer increased use of mature forest habitats as winter progressed, and use of mature forest was greater at higher compared to lower altitudes, corresponding to different snow depths.

The observations that roe deer balance foraging in open habitat with resting in cover habitat suggest that the relative use (selection) of these habitats may change with their relative availabilities. However, all current methods for evaluating habitat selection from use/availability data assume that use is directly proportional to availability. A new method was therefore developed that regresses use against availability on a logit scale. Different hypotheses may be framed in terms of the regression parameters. A slope of one (b=1) indicates that use is directly proportional to availability, whereas b=0 indicates that use is consistent with availability, e.g. if the animal forages in open habitat and rests in cover habitat. The prediction that food and cover alone may be poor predictors of roe deer habitat selection was tested with data on habitat use of 27 radiocollared deer and compared with measures of food and cover availability. As predicted, food availability did not predict roe deer habitat selection during summer. Though there was a correlation between roe deer habitat selection and canopy cover availability, this result may be dependent on the relative availability of food and cover. Use of forage-rich habitat was higher when deer were active as compared to when inactive, and tended to be higher at night than during the day.

Finally, data on seasonal migration pattern and home range of male and female roe deer are presented. Males were found to defend territories which were almost twice as large as female home range size during the mating season (summer), whereas the common pattern is equal range size among the sexes. It is hypothesized that this result may be due to the combination of low density (low costs) and high resource abundance in Lier (giving small female home ranges; high benefits). Roe deer in Lier had on average summer ranges at a higher elevation than their winter ranges. However, summer ranges at high elevations were large, suggesting these areas were of low quality. Snow depth is regarded as a major factor influencing migration patterns of temperate cervids, but it can be assumed that snow depth induces an equal migration pattern of males and females. However, there was a higher proportion of stationary males than females, and more females than males were long distance migrators (>10 km). This supports an earlier hypothesis that roe deer migration patterns are also influenced by social factors, even though grouping tendencies are less pronounced at low population density. The influence these sex-dependent patterns of space use may have on habitat selection is discussed.

LIST OF INDIVIDUAL PAPERS

This dr. scient. thesis is based on the 8 papers listed below. They are later referred to by their Roman numerals.
(*Note: References are updated relative to thesis)

I. Mysterud, A., and Østbye, E. 1995. Bed-site selection by European roe deer ( Capreolus capreolus ) in southern Norway during winter. Canadian Journal of Zoology 73: 924-932.

II. Mysterud, A. 1996. Bed-site selection by adult roe deer Capreolus capreolus in southern Norway during summer. Wildlife Biology 2: 101-106.

III. Mysterud, A., Lian, L.-B., and Hjermann, D.Ø. 1999. Scale-dependent trade-offs in foraging by European roe deer (Capreolus capreolus) during winter.Canadian Journal of Zoology 77: 1486-1493. *

IV. Mysterud, A., Bjørnsen, B.H., and Østbye, E. 1997. Effects of snow depth on food and habitat selection by roe deer Capreolus capreolus along an altitudinal gradient in south-central Norway. Wildlife Biology 3: 27-33.

V. Mysterud, A., and Ims, R.A. 1998. Functional responses in habitat use: availability influences relative use in trade-off situations. Ecology 79: 1435-1441.*

VI. Mysterud, A., Larsen, P.K., Ims, R.A., and Østbye, E. 1999. Habitat selection by roe deer and sheep: does habitat ranking reflect resource availability? Canadian Journal of Zoology 77: 776-783. *

VII. Mysterud, A. 1998. Large male territories in a low density population of roe deer Capreolus capreolus with small female home ranges. Wildlife Biology 4: 231-235. *

VIII. Mysterud, A. 1999. Seasonal migration pattern and home range of roe deer ( Capreolus capreolus ) in an altitudinal gradient in southern Norway. Journal of Zoology 247: 479-486.*

INTRODUCTION

One of the essential components of Darwin's theory of natural selection (Darwin 1859) is that only a fraction of each generation survives until maturity as a result of competition for scarce resources (Milinski and Parker 1991). A central issue for behaviour and population dynamics is therefore habitat selection, the study of where and how animals distribute themselves relative to resources (Fretwell and Lucas 1970, Rosenzweig 1981). A thorough understanding of habitat selection therefore also forms part of the basic platform necessary for a skillful management of game species, such as the European roe deer ( Capreolus capreolus ) (see below), and this has motivated this study.

It is now widely accepted that ecological mechanisms are scale-dependent (Wiens 1989, Turner 1989, Kotliar and Wiens 1990, Bissonette 1997a), and different mechanisms determining an animal's use of resources seem to come into play at different spatial and temporal scales (Johnson 1980, Morris 1987, 1992, Senft et al. 1987, Orians and Wittenberger 1991). For example, at a fine spatial scale, habitat selection is suggested to be determined by the differential use of foraging locations within a home range, whereas at a coarser scale, habitat selection is determined by dispersal and the ability to relocate the home range (Morris 1987, 1992). Some authors have suggested up to 4-6 different spatial scales of large ruminant foraging (Senft et al. 1987, Bailey et al. 1996). Movement decisions at any level may be simultaneously affected by spatial structures at levels below and above in the hierarchy (Senft et al. 1987, Wiens 1989, Kotliar and Wiens 1990, Gautestad and Mysterud 1993, Ims 1995). With respect to temporal scale, there may be different habitat requirements between rest and activity periods, between night and day and, at a longer time scale, between seasons. The space and time dimensions in habitat selection are naturally linked (Fig. 1). Patch/habitat selection within the home range may depend on the shifting phases in the daily activity cycle of an animal, while the more infrequent home range shifts at the landscape scale may result from seasonal migrations. The mechanisms underlying roe deer habitat selection at a coarse scale, i.e. (density-dependent) dispersal and migration, have recently been thoroughly studied (Wahlström 1995, Wahlström and Kjellander 1995, Wahlström and Liberg 1995a, b), and the main focus of this thesis will therefore be at a within-home range scale.

scales

Figure 1. The approximate relationship between temporal and spatial scales of roe deer habitat selection considered in this study (adapted after Wiens 1989, Bissonette 1997b). Typically, long time scales are linked to broad spatial scales. However, a short time scale may give high apparent predictability regarding coarse spatial scales (Wiens 1989). The time dimensions are relatively easily defined, whereas the patch-habitat-landscape entities are difficult to define and often indicates a gradient from fine to coarse scale in this study.


Optimal patch use within a home range is under the simplest conditions predicted by the marginal value theorem (Charnov 1976). According to this, a forager should continue to exploit a patch until its harvest rate in the patch drops to its average over all patches (including foraging time in patches and travel time between patches). Most methods used to evaluate habitat selection from animal space-use observations infer habitat preference as the disproportional use of some habitats over others (Neu et al. 1974, Johnson 1980, Byers et al. 1984, Aebischer et al. 1993, Clark et al. 1993, Manly et al. 1993, Arthur et al. 1996, Otis 1997; see also Alldredge and Ratti 1986, 1992, Thomas and Taylor 1990, Cherry 1996), which is valid if the animal uses most of its time in the food-richest habitats as predicted by the marginal value theorem. However, the model used to derive the marginal value theorem makes a number of simplifying assumptions which limit its applicability (Kotler 1997). These include that foragers have no alternative activities; they only forage or look for new resources in the environment (Kotler 1997), and that there are no trade-offs between selection of different resources.

A problem commonly faced by animals is that many habitats do not have good mixes of patches that allow for all essential activities required for successful reproduction (Orians and Wittenberger 1991). Both food and cover have been described as important resources for habitat selection by roe deer (Dzieciolowski 1976, Papageorgiou 1978, Helle 1980, Henry 1981, Büttner 1983, Cederlund 1983, Stüwe and Hendrichs 1984, Cibien and Sempere 1989, Aulak and Babinska-Werka 1990a,b, Welch et al. 1990, Selås et al. 1991, Guillet et al. 1995, Latham et al. 1996, 1997, Tufto et al. 1996, Telleria and Virgos 1997) as well as by white-tailed deer ( Odocoileus virginianus ) (e.g. Gill 1966, Huot 1974, Armstrong et al. 1983a, b, Lang and Gates 1985). However, even though in particular Huot (1974), adequately classified cervids habitat situations as whether the distribution of food and cover were combined or distinct, and described that selection may vary with activity (Gill 1966, Huot 1974), these early studies were usually without much quantification or statistical testing (but see Armstrong et al. 1983a, b). On the other hand, the weaknesses of many recent studies is a consequence of statistical methods which do not take into account temporal scale or resource distribution (e.g. Tufto et al. 1996).

The main aim of this thesis is therefore to quantify roe deer habitat selection relative to the spatial distribution of food and cover transcending different spatiotemporal scales and mechanisms, and to suggest ways to solve some of the methodological challenges this approach presents. In the first section, I describe roe deer habitat selection relative to the spatial distribution of food and cover when selecting bed-sites (winter: paper I; summer: paper II) and forage sites (winter: paper III, IV). I further test some predictions regarding use of cover as a thermal strategy (paper I, II, III), an antipredator strategy (paper III) and as a strategy to find areas with shallow snow (paper III, IV). I also present a method to correct the bias that habitat selection may have on estimates of food preference (paper III). In the second section, the effect of temporal scale on habitat selection in trade-off situations is discussed in more detail. A new approach of evaluating habitat selection from animal space use observations pertinent under trade-off situations is presented (paper V). Based on these arguments, the prediction that food or cover availability alone may not be a good predictor of roe deer habitat selection is tested with data from 27 radio-collared roe deer and habitat rankings established with compositional analysis (Aebischer et al. 1993) together with information about food and cover availability in the habitats (paper VI). I end the thesis with a section on space use and social organization, since ecological factors may affect habitat selection of sex and age groups differently. I present data on space use during the breeding season (paper VII) and on seasonal migration pattern and the clan system (paper VIII), and discuss the influence sex-dependent patterns of space use may have on habitat selection.


The characteristics of the species (Ch. of sp.)

The European roe deer is a small (20-30 kg) cervid distributed all over Europe (except Ireland) and is now regarded as a separate species from the Siberian roe deer ( Capreolus pygargus ) (Danilkin and Hewison 1996). Roe deer is at its northernmost range in Scandinavia (Cederlund and Liberg 1995). The species colonized southern Scandinavia 7-9000 years BP (Hufthammer 1992, Hufthammer and Aaris-Sørensen 1998), but the present high density is historically more recent (Cederlund and Liberg 1995, Wahlström and Liberg 1995b).

During summer, roe deer feed mainly on herbs (Cederlund et al. 1980, Selås et al. 1991, Tixier and Duncan 1996), a forage which is rich in soluble sugars, has low levels of cellulose and hemicellulose, but contains high levels of tannins and lignin (Tixier et al. 1997). During winter, the roe deer diet consists mainly of varying amounts of bilberry ( Vaccinium myrtillus ) stems and twigs/buds from deciduous trees (Cederlund et al. 1980, Helle 1980, Mysterud and Østbye 1995), but they face severe constraints from their limited ability to digest these cellulose-rich winter forages (Holand 1992, 1994), probably because of their small body size (Demment and Van Soest 1985, Illius and Gordon 1987).

Ruminants typically have a simple foraging-resting-foraging type of time budget (Bunnell and Gillingham 1985, Mysterud 1998), and during winter, roe deer use about 55-60% of their time on rumination/rest distributed on about 8 daily bouts, whereas only 45-50% on 12 bouts during summer (Cederlund 1981, 1989, Jeppesen 1989). Changes in activity is mainly related to changes in digestibility of plants (Cederlund 1989), which is much higher during summer than winter (Cederlund and Nyström 1981, Holand 1993).

The mating system of roe deer is unique among the Cervidae since the males defend a (mating) territory from about beginning of April to the end of August, whereas the females come into estrus during late July/early August (Cederlund and Liberg 1995, Johansson 1996). The roe deer is only slightly sexually dimorphic (5-10%), suggesting low levels of polygyny (Loison et al. 1998), although female rut excursions are common in high density populations (Andersen et al. 1995, Johansson 1996). Females live solitary during summer, but with some spatial overlap with neighbouring females (Bjar et al. 1991, Andersen et al. 1995). The roe deer is polytocous, giving birth to between one and three (four) fawns in early June (Cederlund and Liberg 1995). Females usually breed for the first time as two-year-olds (Gaillard et al. 1992). The roe deer is classified as a typical "hider" with respect to the neonatal antipredator strategy the first few weeks after birth (sensu Lent 1974). Fawns spend 80% of the time inactive at well hidden places (Linnell 1994), probably as a strategy to reduce the high levels of predation from red fox ( Vulpes vulpes ) (Linnell et al. 1995, Aanes and Andersen 1996). As one-year-olds, a high proportion of both sexes leave their natal range and may disperse a long distance before settling (Wahlström and Liberg 1995a, b). Male yearlings are forced to leave the natal area by older territorial males (Strandgaard 1972, Wahlström 1994), whereas females leave voluntarily probably to gain access to more resources (Wahlström and Kjellander 1995). Since dispersal is often male-biased (Strandgaard 1972, but see Wahlström and Liberg 1995a), females produce male-biased litters when environmental conditions are limiting, possibly to lower local resource competition (Hewison and Gaillard 1996).

During winter, roe deer often congregate in small family groups (4-5), called clans, especially at high population density (Vincent et al. 1995). The mother and her calves form the core of this unit, often with female offspring from the previous year and an unrelated male (Linnell 1994). Roe deer in Scandinavia are prone to large die-offs in snowy winters (critical snow depth is considered to be 50 cm; Cederlund and Liberg 1995), and red fox may also kill adults under such conditions (Cederlund and Lindström 1983). In addition to autumn hunting (adult males: Aug. 16-Dec. 23; females and calves: Sep. 25-Dec. 23), lynx ( Felis lynx ) may kill all age and sex classes of roe deer in some areas of Norway (Linnell and Andersen 1995).



STUDY AREAS

The Lier valley, Buskerud

The main study area (paper I, II, III, VI, VII, VIII) is about 250 km 2 and situated in the Lier valley in southern Norway (between 59°52'-59°58'N and 10°14'-10°20E). Most of Lier is forested and situated within the boreonemoral region (Abrahamsen et al. 1977). Vegetation is varied and dominated by Norway spruce ( Picea abies ) mixed with Scots pine ( Pinus sylvestris ) on the drier and poorer locations. The forest has been commercially managed, and there are several clearcuts of varying sizes within the study area. Along the bottom of the valley on richer soil, deciduous forest predominates, fragmented by small cultivated fields which roe deer may feed on during winter (Kjøstvedt et al. 1998). In the deciduous forest, species such as hoary alder ( Alnus incana ) and bird cherry ( Prunus padus ) are dominant, mixed with elm ( Ulmus glabra ) and linden ( Tilia cordata ) on the richest sites.

The terrain is very hilly, rising from Lake Holsfjorden at 63 m a.s.l. to over 500 m 1.5-2.5 km from the lake (Fig. 1 in paper VIII). Some parts of the area are largely flat on a large scale. The bottom of the valley are undulating and hilly with many ravines on a fine scale (10-100 m between top and bottom) due to erosion in clay sediments.

The density of roe deer in the Lier valley during the winter 1992/93 was estimated to be 3-5 deer/km 2 (Mysterud 1993). There are red fox in the area, and occassional visits by lynx is also recorded, though the main cause of mortality is probably human hunting.


Flatdal, Telemark

The other study area (paper IV), Flatdal, is about 6 km 2 and situated in Seljord municipality in Telemark county in south-central Norway (between 59°32'-59°34'N and 8°32'-8°36'E) and extends from 170 to 1000 m a.s.l. Slope varies between 15° and 30° (Bjørnsen 1985). The terrain is hilly also on a fine scale, with many large boulders and small cliffs. The area is forested and situated within the northern boreal zone, though it is close to the boreonemoral border-zone (Abrahamsen et al. 1977). Plant species representative for both vegetation zones are found within the study area. Norway spruce is the dominant tree species, but with scattered stands of Scots pine on poorer and drier sites. At the highest altitudes, subalpine birch ( Betula spp.) is dominant. The forest is commercially managed. There are no farms or houses within the study area. Roe deer density was about 2-4 deer/km 2 during the time of study. The area is rather undisturbed, but there is hunting during autumn (paper IV).



FRAMEWORK, SAMPLING AND METHODS

Animal movement patterns may be studied using several different bodies of theory (Ims 1995). Although not often explicitly stated, they usually address different spatial and temporal scales. Habitat selection theory was chosen a priori for this study, as is usually the case for this kind of field studies. Assuming that the theoretical framework is sound, the challenge becomes to find a sampling design and statistical methods that reflect this framework. Unfortunately, all statistical methods of habitat selection (Alldredge and Ratti 1992), as well as foraging models (Laca and Demment 1996), make explicit or implicit assumptions that limit their applicability (see also paper III, V). The chosen study design obviously also reflects limitations set by equipment, time and financial costs. Along with a few definitions and a brief presentation of methods, I therefore find it necessary to briefly comment upon some obvious pitfalls in this regard, whereas details of the sampling design are given in the individual papers.


Spatial scaling

The term "habitat quality" is often used rather loosely and with variable meaning. Scientists working with (large) cervids often think of habitat quality only in terms of food quantity and quality (e.g. Hanley 1984, Langvatn and Hanley 1993, Hjeljord 1997), whereas others suggest that habitat quality integrates habitat features such as resource levels, habitat structure, predation risk, social interactions and demographic parameters (Wiens et al. 1993). I, however, regard food (quantity and quality) and cover as the main habitat resources for roe deer in this study, based on earlier studies of roe deer habitat selection (e.g. Henry 1981, Stüwe and Hendrichs 1984). The entities patch, habitat and landscape were used as the basic, hierarchically ordered, spatial scales. However, the term habitat selection is seldom used in a scale-specific manner; e.g. I often include patch selection in habitat selection. When analysing selection in a hierarchy of spatial scales, it is common that what is availability at one scale, becomes use at the next (e.g. Schaefer and Messier 1995a). This is the case when I use compositional analysis (Aebischer et al. 1993), whereas elsewhere, the scales patch-habitat-landscape are seen as a continuum from local to coarse scale, and not as strictly defined entities.

The term selection is really intended for situations in which use differs from availability, whereas I do not always use this strict definition (see Johnson 1980, Thomas and Taylor 1990 for a discussion on terminology on preference/selection). Habitat selection is here evaluated in two distinct ways; either as what I term (1) resource use or (2) based on habitat use/availability data (details below). I also regard studies of space use such as scaling of home range size (paper VII) and migration patterns (paper VIII) as an indirect way of evaluating habitat selection at a landscape scale.

Resource use is evaluated based on measurements of resource levels at bed-sites or feeding sites (paper I, II, III), and neither patches nor habitats are demarcated. Habitat selection evaluated from more traditional data on habitat use relative to availability presumes demarcation of habitat types. In my case, habitat boundaries were determined according to the standard habitat classification scheme used in forestry in Norway (paper IV, VI). There are at least two problems with such an approach. The patch/habitat boundaries which cervids experience in most field situations can be both diffuse and distinct, and patch boundaries may have considerable effect on animals ability to assess resource levels (Stamps et al. 1987, Arditi and Dacorogna 1988, Kremsater and Bunnell 1992, Schmidt and Brown 1996). Further, animals may not view the world as we do. Wiens (1989) has urged for the development of operational ways of defining scale. This is fraught with difficulty due to the link between variability and scale (R.A. Ims, pers. comm.), but attempts to identify the "correct" number of habitats based on the animals habitat use have been made (Knight and Morris 1996). Once habitats are classified, there are further problems with identifying what is available to the animal, which is a considerable challenge. The specification of habitat availability is equivalent to specification of a spatial scale at which to study the selection process (Otis 1997). The study area scale is set arbitrarily, whereas the home range scale is defined from the animals use of space. Even though home range borders are not always easily defined (Spencer et al. 1990, Gautestad and Mysterud 1995), this is not likely to be a great problem except possibly at low habitat availabilities (paper V).


Temporal scaling

I have distinguished between rest and activity, night and day and between summer and winter in this study. Problems of definition are somewhat less difficult for temporal scales than for spatial scales. However, although time used for activities other than foraging and resting/ruminating is negligible during winter (less than 1% for moose: Renecker et al. 1978), males at least may spend a considerable amount of time defending and patrolling territories during summer (Johansson 1996). A further problem is that bout length may affect bed-site selection (paper I), and that the state of the animal may vary from hunger to near satiation within a foraging bout (Edwards et al. 1994, Gillingham et al. 1997). Possibly, future studies can combine radio-telemetry with physiological measures of the state of the animal (see MacArthur et al. 1979). Within the frame of this study, splitting the analysis on active/foraging and rest/ruminating is considered to be the best option.


Sampling and statistical analysis

Collection of data on roe deer space use was by either tracking (paper I, IV), by aid of radio-telemetry (paper VI) or by a combination thereof (paper II, III). Only radio-telemetry provides the opportunity to collect unbiased data in space and time in forested areas, and a total of 41 roe deer were therefore caught and fitted with radio-collars (see below). Fixes from radio-collared animals were obtained by standard triangulation from at least two directions, preferably within 300 m of the deer. Although a short distance between observer and deer will reduce the measurement error-rates, they do not eliminate them (Springer 1979, see also Saltz 1994). This reduces the power of the analysis, but Guillet et al. (1995) also showed that radio-tracking based methods were unable to determine preference of habitat-types which were smaller than the error polygon. I combine therefore radio-telemetry with tracking in two studies (paper II, III). Combining the two techniques also allows measurement of resource levels directly at feeding or bed-sites (see below).


Trapping

Animals used for analysis based on radio-telemetry (paper II, III, VI, VII, VIII) were caught during winter (February-March) in boxtraps of the "Öster-Malma"-type and with "drop-nets". Apples, carrots, grain-pellets etc. were used as bait. Animals were trapped at traditional artifical feeding sites (rare in the study area), but I mostly established new feeding sites. This was important in order to make unbiased samples of deer with respect to habitat use. However, all trap sites were at low elevations, and deer in some local areas seemed to be more difficult to catch. All deer were fitted with motion-sensitive radiocollars (Televilt TXE-3), i.e. they transmitted with a different pulse rate depending on whether or not the deer was active. Unfortunately, this function was not reliable on all collars.


Estimating resource levels

I estimated food quantity as coverage of main groups of forage plants (paper II, VI), whereas I used a food index that took into account both food quantity (availability) and quality (the selection index) in paper III. Plant quality may vary with site quality and shade conditions (reviewed in Jefferies et al. 1994), and the delayed phenological plant growth under cover may be an important factor for cervid habitat selection (Hanley 1982, Hjeljord et al. 1990). However, excluding plant quality will not likely bias the results, since I focus little on use per se, but rather on short term changes in use, e.g. between night and day, which could not be predicted from variations in plant quality.

I measured cover characteristics both as canopy cover closure and hiding cover (sighting distance along ground) with a spherical densiometer (Lemmon 1956, 1957, Nuttle 1997) and a cover board (Nudds 1977, Griffith and Youtie 1988), respectively.


Identifying trade-offs

Trade-offs can be identified in several different ways. If availability of one resource declines (or distance to resource increases) and availability of another resource simultanously increases e.g. when deer chooses a bed-site, then a trade-off between selection of these resources is suggested. In some cases, I correlate the two resources directly based on measurements of availability at bed-sites (paper II). A (perfect) trade-off situation gives a negative correlation (r=-1), whereas r=1 indicates no trade-off (Fig. 2). Given the same range of food and cover values, a correlation coefficient between this, e.g. r=0, indicates that there are patches with combinations of high/high, high/low, low/high and low/low availability of the two resources, i.e. local trade-offs may arise. Note that this measures trade-offs mainly at a patch (local) scale, though one usually discuss trade-offs related to habitat scale (see Introduction, paper V, VI). I also regard it as a sufficient indication of a trade-off if selection of only one resource varies with temporal scale (paper VI), since it is then apparently traded with some other (unknown) resource.


tradeoff

Figure 2. Selection of two resources may either be maximized at the same time (no trade-off, r>0) or traded-off (r<0). However, given the same variance in both resources over all r (e.g. both food and cover availability vary from 0-100%), the level of r indicates the extent of the trade-off. For example, if there is an equal number of patches with high/high, high/low, low/high and low/low availability of food and cover, respectively, the correlation will be r=0, which indicates that local trade-offs may arise. This assumes that the animal scales home range size to one resource only, so that it is unable to attain a sufficient amount of high food/high cover patches only (if they do, then the trade-off is at a coarser spatial scale).

Resource use

Resource use was evaluated in two distinct ways. Resource availability at bed-sites or feeding sites was compared with availability at a point 50 m away in a random direction (with Wilcoxon pair-tests). This distance was set arbitrarily, but comparing with a random location 10 m away showed a similar pattern (Mysterud 1993). It is assumed that this measures patch selection in the sense of a local choice. When evaluating how resource use correlates with environmental characteristics such as temperature, snow depth and period, resource availabilities at the feeding or bed-sites were correlated with the different predictor variables (mainly general linear models). This does not account for differences among individuals in availability or changing resource availability with time, but this is regarded as a small problem given a random sample of observations. This approach in itself does not say anything about the scale at which the selection process occurs. Ideally, differences between home ranges are removed before analysis by adjusting estimates with each individual means (paper III), so that any landscape level selection can be accounted for. Further, any contradictory result between the two analyses may indicate the scale at which selection is made. In paper III, this approach was brought a step further by using contrasted values (feeding site-random site) as response variable in a similar analysis.


Habitat use/availability

Most methods designed to evaluate habitat selection compare some measure of use relative to some measure of availability (see Introduction), except for experimental situations when availability can be kept equal (Elston et al. 1996). A crucial step toward a sound analysis are to have availability well defined or demarcated. When data is from unknown individuals and hence pooled (paper IV), availability is calculated for the whole study area (design 1, sensu Thomas and Taylor 1990), and I used the Bonferroni Z-statistics (Neu et al. 1974, Byers et al. 1984). In such a case, it is not possible to separate selection of a home range (landscape scale) from selection within home range (habitat scale), although attempts to partly account for this were made (paper IV). When data are from radiocollared deer (paper VI), one can calculate availability both for the study area and for the individual home ranges (design 3), and hence selection of home range and habitats within home range can be separated. I then used the compositional analysis of log-ratios (Aebischer et al. 1993). This method is currently regarded as the best option for evaluating habitat selection at two spatial scales since it also incorporates the unit-sum constraint, has a proper level of replication and allows testing for effects of biological relevant factors such as sex and age (Aebischer et al. 1993). Compositional analysis takes into account much of the criticism earlier methods treating use/availability data for habitat evaluation have received (i.e. that such data only apply for ideal-free foragers after equilibria is set, Van Horne 1983, Hobbs and Hanley 1990), at least at the home range scale. Note, however, that these habitat types alone do not say anything about resource levels. I therefore either combine this approach with separate measurements of food and cover in transects (paper VI), or make qualitative assessments of resource levels (paper IV).



RESULTS AND DISCUSSION

Resource use strategies

Both food and cover have been termed important for habitat selection by smaller temperate cervids such as roe deer, white-tailed deer and mule deer ( Odocoileus hemionus ) (e.g. Armstrong et al. 1983a, b, Lang and Gates 1985, Schmitz 1991, Armleder et al. 1994). Obviously animals must eat to acquire energy, but the effect of cover on habitat use is less understood (reviewed in Peek et al. 1982, Mysterud and Østbye 1998). Several different hypotheses (not mutually exclusive) have been proposed about the mechanisms that can explain variation in the use of cover. It may be (1) a strategy to find areas with a favorable thermal regime (e.g. Moen 1968, 1973), and hence, use of cover is predicted to increase during severe cold (paper I) or heat (paper II). It may also be (2) a strategy to lower the risk of predation (e.g. Lima and Dill 1990). It is more difficult to assess whether, or to what degree, the use of cover is influenced by risk of predation, since predation pressure is not easily quantified. However, use of cover can be predicted to be higher during day than night when chances of visual detection are higher, if the thermal regime can simultanously be statistically controlled for (paper III). Use of cover may also be (3) a strategy to find areas with shallow snow (e.g. Armleder et al. 1994). The use of cover areas can be predicted to increase with increasing snow depth for two reasons (paper IV): (a) Snow depth increases proportionally between habitats (e.g. a snow depth of 20 cm in dense and 40 cm in open habitat increases to respectively 40 and 80 cm) and (b) energetic costs of movement increases at a much higher rate when snow depth exceed breast height (Parker et al. 1984). Note that all of these predictions assume some food-cover trade-off, which is discussed in more detail in the next section.

Bed-site selection during winter (paper I)

During the snow-poor winter 1992/93, snow tracking of roe deer in Lier indicated that they foraged mostly in open habitats (average 1% canopy cover) such as small agricultural fields, deciduous forests and clear-cuts (paper I). Distance from feeding sites to bed-sites varied from 0 to over 200 m, with fewer bed-sites as distance from feeding sites increased. Only bed-sites far from feeding sites were more covered than random sites. Seeking cover for bedding was hence the same as moving further away from the feeding site, and there was apparently a trade-off between selection of food when foraging and of cover when resting.

As predicted, there was more canopy cover above roe deer bed-sites during cold than during mild periods, whereas temperature had no effect on cover characteristics at feeding sites (paper I). Though the more detailed study also showed an effect of temperature on selection of feeding sites (paper III), relatively little energy is used for thermoregulation compared to that which can be gained by choosing high quality forage (Moen 1968). The effects of temperature and wind-chill were not separated, but roe deer had more hiding cover in direction of incoming wind, suggesting that windchill was an important part of the thermal environment. Studies on white-tailed deer and red deer ( Cervus elaphus elaphus ) also conclude that the combination of low temperature (correlated with radiation) and high windspeed may be detrimental for the energy budget during winter (Moen 1973, Staines 1976). It was further shown that roe deer sought more cover when bedding for long rather than short periods of time (the estimates of duration of bedding period was based on observations of degree of snow melt in the bed-site, and hence this result should be interpreted with some caution). That roe deer use cover more often when resting for long periods and in colder weather, suggests that use of cover is linked to the energetic costs of reaching cover habitats. This is further evidenced by Stüwe and Hendrichs (1984) who reported that roe deer in an open field habitat sougth cover for bedding only when distance from feeding site to cover was short.


Bed-site selection during summer (paper II)

Bed-site selection of 3 male and 2 female roe deer was studied during summer 1994 (paper II). There was a higher coverage of herbs at bed-sites than at random sites, suggesting that bed-site selection is influenced by selection of feeding site. There were no correlation between cover and herb availability. Hence, there seemed to be no general trade-off between selection of food and cover during summer (see also Tufto et al. 1996), although such may exist at a local level for some patches/habitats.

During warm weather, the coverage of herbs at bed-sites was lower than during cooler periods, indicating movement away from feeding sites (paper II). However, this movement was not correlated with an increase in canopy cover above bed-sites (but see Belovsky 1981, Schwab and Pitt 1991, Demarchi and Bunnell 1995), but rather roe deer selected areas that provided a more humid substrate. The conductive heat loss, i.e. the direct transfer of energy between the deer and substrate, may be the most important component of the energy budget for a bedded ungulate (Moen 1973). This has been measured to approach 30% of sheep's ( Ovis aries ) minimum heat production on cold, poorly insulated ground (Gatenby 1977). This strategy has also earlier been suggested as part of the thermal strategy for several other ungulates (sheep: Gatenby 1977; white-tailed deer: Jacobsen 1980; elk Cervus e. canadensis : Merrill 1991; mule deer: Sargeant et al. 1994), but without any quantitative measures of humidity.

Forage site selection and food preference during winter (paper III)

Feeding site selection and food preference of 5 male and 5 female radio-collared roe deer was studied by snow-tracking during winter 1996 (paper III). A forager's diet results from the two-step decision process of selecting where to seek food (habitat selection) and then, once there, selecting what to eat or pursue among available foods (e.g. Nudds 1980, Senft et al. 1987, Brown and Morgan 1995). Despite this theoretical knowledge, empirical studies of hierarchical foraging by cervids are few (Hanley 1997; but see Edwards et al. 1994, Ward and Saltz 1994, Schaefer and Messier 1995a, b for bovids), and current methods for evaluating food selection based on use/availability data from field observations have not incorporated this element (e.g. Neu et al. 1974, Aebischer et al. 1993). We therefore present a new way of using a method originally presented to control for autocorrelation in habitat selection (Arthur et al. 1996). The use of this method allowed us to compare the chosen diet (use) with availability within each foraging site in succession, rather than to some home range or study area average (Neu et al. 1974, Aebischer et al. 1993). There was no evidence of variation in food preference between males and females or between day and night, though statistical power was low.

As predicted, roe deer used more open areas at night than during the day. They also avoided feeding close to human settlement during day. As snow depth increased, they foraged closer to human settlement, suggesting a trade-off between energetic demands and antipredator behaviour. Also, observations that the deer seemed very shy and selected bed-sites that were well hidden compared to random plots both during winter (paper I) and summer (paper II), suggest that the antipredator aspect of cover use may be very important for how roe deer select habitats. However, the strategy will likely vary with the predators hunting strategy. For example, it has been observed that roe deer seek open fields in areas with lynx (V. Holthe and S.R. Gjems, pers. comm.), a typical ambush predator (Murray et al. 1995).

Roe deer's selection of forage sites was random with regard to cover when measured against a random location only 50 m away, whereas selection of cover varied with temperature and period. In contrast, there was more forage at the chosen patches than at the random locations, but there was no effect of either temperature or snow depth. This suggests that the selection of cover is at a coarser level (habitat scale) as compared to food (patch scale). This was further supported by the use of contrasted values (feeding site-random site).

Effect of snow on forage site selection (paper IV)

Snow decreases availability of forage to deer, and increases the cost of locomotion profoundly, especially when snow depth exceed breast height (Parker et al. 1984). Shallow snow may hence also ease escape from predators (Messier and Barrette 1985), and is therefore regarded as a major factor determining habitat selection of North-American cervids (Huot 1974, Kucera 1976, Armstrong et al. 1983b, Beier and McCullough 1990, Pauley et al. 1993, Armleder et al. 1994). Although snow depth influenced distance to human settlement, no effect of increasing snow depth on selection of canopy cover was found in Lier (paper III). However, this may be due to the shallow snow that did not reach the critical limit (see prediction 3b). Hence, I reanalysed a data set from Flatdal in Telemark in which snow depth (Fig. 4 in paper IV) in some habitats exceeded the critical limit (50 cm, Cederlund and Liberg 1995).

Consistent with the predictions, use of open habitats tended to decrease as winter progressed and was lower at higher than at lower altitudes in accordance with differences in snow depth. Snow depth also seemed to affect habitat selection at a coarser scale, since use of higher altitudes decreased as winter progressed (paper IV).

Habitat use/availability, trade-off and temporal scale

In several papers (paper I, V, VI), I assume that the trade-off are between foraging (in open habitat) and resting (in cover habitat), or between selection of night and day areas (paper III, VI). Even though this approach demonstrates that strength of trade-offs varies with temporal scale (paper I, VI, see also Cowlishaw 1997a, b), it does not quantify the trade-off per se. Based on estimates of forage quantity and quality, one may estimate the trade-off by contrasting the predicted and observed use of each habitat. However, the spatiotemporal variation in resource levels and possible confounding factors, such as differing availability, limit the value of this approach. A different approach is to measure giving-up-densities (GUDs) on artificially planted depletable food trays (Brown 1988, see various applications in Brown 1992, 1996, Brown et al. 1994, Kotler et al. 1994, Morgan and Brown 1996, Schmidt and Brown 1996, Kotler 1997, Morgan et al. 1997, Morris 1997), or to titrate food and safety by controlling levels of energy content in forage presented in different habitats (e.g. Kotler and Blaustein 1995). This allows for much more controlled measures of trade-offs, and I believe further progress on habitat use by cervids can come from adapting these methods that until now have been used mainly for rodents (but see Kotler et al. 1994 for Nubian ibex Capra ibex ). However, radio-telemetry and use/availability data will obviously have their central place also in the future, especially when studying sex or age dependent patterns of habitat selection (see next section). In this section, a new method that can be used to test for trade-offs for such data is presented (paper V), and I explore to what extent use/availability data can be predicted to reflect resource levels (paper VI).


Functional responses in habitat use (paper V)

During winter, roe deer balance foraging in open habitat against bedding in cover habitat (paper I). This suggests that the relative use (i.e. selection) of food and cover areas may change with their relative availabilities (paper V). However, all current methods (see Introduction) for evaluation of habitat selection from use/availability data implicitly assume that use is directly proportional to availability. We therefore present a new approach to evaluate the effects of availability on relative use. We propose to use logistic regression to regress proportional use against proportional availability for a sample of individuals. Different statistical hypotheses may be framed in terms of the regression parameters. A slope of 1 (b=1) will indicate that use is directly proportional to availability. If, however, an animal spends a consistent amount of time in each habitat with changing availability (e.g. bedding in cover habitat and foraging in open habitat), the slope is expected to be equal to 0 (b=0). There may also be several other types of what we call a functional response in habitat use, i.e. a change in selection with availability of one of two main habitat types (paper V).

For this method to be applicable, availability must vary between individuals in the population. This is probably closely related to how animals scale their home range. When food is the limiting factor, the animal may adjust home range size to include a certain minimum amount of food. For example, if satiation is reached for the feeding habitat at 50 ha and for the cover habitat at 5 ha, and if the availability of feeding habitat is <=10%, then the animal will establish the home range so as to include at least 50 ha of feeding habitat. Within this home range, the amount of cover habitat will vary, but often will exceed 5 ha if habitats are sufficiently mixed. Assuming all home ranges include the same amount of feeding habitat, then with increasing home range size, the ratio of food to cover in the home range decreases (Fig. 2 in paper V). Thus availabilities of the different habitats will vary even though the amount of feeding habitat in all home ranges is constant. If the time budget remains stable with availability, then the strength of selection or the correlation between habitat and activity must change. Note that this also suggests that the way compositional analysis (Aebischer et al. 1993) evaluates habitat selection at broad scales is moot, since preference for a habitat will be totally dependent on the arbitrary definition of the study area availability.

A limitation with the method presented is that it requires a composition of only two habitat types. This is, however, a rather common situation if one considers functional habitat types (sensu Armstrong et al. 1983b). For example, it may not be important which of four habitats that is used for cover. Any unjustified lumping of habitat categories will be identified as a lack-of-fit of the data to the model (i.e. an overdispersed error term). This was the case for the roe deer data set presented in paper VI, and the logistic regression approach could therefore not be used. For true multiple habitat situations, an option is to incorporate temporal scales into the MANOVA-models of compositional analysis. Even though this is a clear improvement over pooled data and may provide valuble insight, the MANOVA approach assumes that use is directly proportional to availability for each of the temporal scales (i.e. b=1 for each factor, or equivalently, b=0 for the pooled data if two temporal scales). Compositional analysis with several temporal scales also requires a large number of radio-locations per individual, especially if there are many habitat types, and a different approach was therefore chosen in the following (paper VI).


Resource availability and habitat rankings (paper VI)

Based on the arguments above and in the Introduction, it is expected that food or cover alone may be poor predictors of roe deer habitat selection. We tested these predictions with radio-telemetry data on habitat selection by 27 roe deer, and contrasted this with similar data of 10 free-ranging domestic sheep on sympatric forest range in a mix of 12 habitat types with different amounts of food and cover. The rationale for including two species was to contrast one species (roe deer), where a trade-off was expected to play a role in determining habitat rankings, with another (sheep) for which no trade-off was expected. Habitat selection by sheep in Norway has earlier been studied in several areas (Warren and Mysterud 1991, Warren et al. 1993). Sheep are originally mountain animals with grouping and use of precipitous terrain as their main antipredator tactic, and their habitat selection seem to be little influenced by cover availability (Warren and Mysterud 1991, Warren et al. 1993).

As predicted, habitat rankings of sheep, as estimated with compositional analysis, were highly correlated with food (grass) availability at both the home range and study area scales (paper VI). In contrast, there was no correlation between roe deer habitat rankings and food (herb) availability. Even though there was significant correlation between habitat rankings based on roe deer selection and cover availability, this result may be dependent on the relative availability of food and cover areas (paper V). Roe deer used forest habitats with more forage when active than when inactive, and tended to use more forage rich habitats at night. This indicates that the selection of food is traded with some other resource. However, the predictions (Table 1 in paper VI) that roe deer use more open habitat when active or at night were not clearly supported (using hiding cover gave similar results, A. Mysterud, unpubl. data). Although the trend was in the predicted direction, this may reflect that also other resources than food and cover, such as bedding substrate (paper II), influence habitat selection by roe deer.

Space use and social organization

I have so far presented data on how ecological factors such as food and cover, and mechanisms such as temperature and snow depth, affect habitat selection by roe deer. I have to this point assumed that these measures have direct and equal benefits to all individuals. This is probably not the case, partly due to the social setting in which these animals live. In the following, I will present data on space use of male and female roe deer during summer, which hence is a study of the mating system (see Ch. of sp.), and also on how the clan system may affect migration patterns of male and female roe deer differently. The relevance of these sex-dependent patterns of space use for habitat selection are discussed.

Home range size and the mating system (paper VII)

In mammals, only female reproductive success is limited by access to resources, whereas male reproductive success is limited by access to females (Trivers 1972, Davies 1991). The spatial distribution of females therefore generally reflects the spatial and temporal resource distribution, whereas distribution of males is influenced more by the distribution of receptive females (Trivers 1972, Ims 1988), at least during the breeding season.

Male roe deer are territorial during the summer, whereas females live solitarily with some spatial overlap to neighbouring females (see Ch. of sp.). Range sizes of males and females are typically of equal size at densities varying from medium to high (>10 deer/km 2, see Ch. of sp. and references in paper VII). In Lier (3-5 deer/km 2, Mysterud 1993), however, males defended territories that were almost twice as large as female home range size during the same period (summer) (paper VII). Earlier, it has been shown that male territory size, but not female range size, decreases with increasing density (ratio male/female range size was 1.5 and 1.1, respectively, at 5-7 and 25 deer/km 2, Vincent et al. 1995). The magnitude of the difference was nevertheless surprising, especially since the size of male territories was similar to the size of female home ranges in another low density area in Sweden (3-4 deer/km 2, Cederlund 1983). In the Swedish study area, however, female home range size was large, and hence resource levels probably low when compared to Lier. I therefore hypothesize that the large territories of males in Lier, seen relative to female home range size, may be due to the combination of low density (costs) and high resource levels/small female home ranges (benefits) (the female dispersion hypothesis, paper VII), although a skew in the sex ratio also has the potential to explain this pattern (the male density hypothesis, paper VII).

This result thus provides some support to the suggestion that male space use, and therefore possibly male coarse scale habitat selection during summer, is more influenced by the distribution of females than the food and cover distribution. All studies that are not experimental must face the fact that there may be confounding factors underlying the observed patterns (Ims and Yoccoz 1995). It is thus possible that males only track female habitat use during summer also at finer scales, and that the rather similar pattern of selection (paper VI), hence, arises for very different reasons.

It has also been questioned whether use of cover conveys any direct benefit to adult females due to low levels of adult mortality due to predation during summer (Tufto et al. 1996). The set of behaviours associated with the hider strategy of fawns (see Ch. of sp.) also include changes in the mothers behaviour (Linnell 1994). It has been suggested that predators may use the mother as a cue to find the hidden fawn, and that this is why females use cover during summer (pronghorn Antilocapra americana : Byers and Byers 1984; roe deer: Tufto et al. 1996). Although data are limited, observations of yearling females (without fawn) do suggest that their use of cover is rather similar to adult females (with fawns). During winter, roe deer with following young tended to use more well hidden feeding sites than those without (paper III, see also references therein for other ungulates). However, the males also responded to period in the use of cover, even though predation rates on adults are low in shallow snow (Cederlund and Lindström 1983). Low predation rates may indicate that the behavioural antipredator tactic is effective, and not necessarily that predation is unimportant for behaviour. I therefore conclude that the mother-infant relationship may not be necessary to explain why adult roe deer use cover during summer, although it may affect the strength of the food-cover trade-off.


Seasonal migration pattern and the clan system (paper VIII)

Typically, migrating cervids in the temperate region choose a high elevation summer range and a low elevation winter range (see paper VIII and references therein), and roe deer in Lier were no exception (paper VIII). The downhill migration during autumn is regarded as a strategy to find areas with less snow (e.g. Nelson 1995), whereas the uphill migration during spring is less understood. It has been suggested that high altitude (coastal) summer ranges have a higher food quality for red deer (Albon and Langvatn 1992), whereas the opposite may be the case for moose (in the inland, Hjeljord 1997). The increasing home range size of roe deer with increasing altitude in Lier, and the low proportion of the population migrating to high elevations, suggest that these areas are of low quality.

If snow depth is the only mechanism affecting migration, then migration pattern of males and females should be equal, since there is minimal sexual size dimorphy in roe deer (Loison et al. 1998). However, I found a lower frequency of stationary animals and a higher frequency of long-distance migrators among female than male roe deer in Lier (paper VIII); several females migrated much further than expected from local snow depth gradients (see also Holand et al. 1998). It has been reported earlier that female white-tailed deer migrate further than males in some populations (Nelson and Mech 1981, Nixon et al. 1991), and Wahlström and Liberg (1995a) observed that only female roe deer migrated in flat terrain in Sweden. A possibility is of course that the females choose habitats for their following (shortlegged) calves, though, the propensity to migrate seem to be most common in single females (Wahlström and Liberg 1995a).

A much more plausible hypothesis is linked to the dispersal pattern and the winter clan system (see Ch. of sp.) of these cervids (Nelson and Mech 1981, Wahlström and Liberg 1995a). It has been observed that yearling female deer that disperse from their natal range in spring, return in autumn and become part of their mothers winter clan (Nelson and Mech 1981, Wahlström and Liberg 1995a, paper VIII). Dispersing males often join a winter clan at their new location, probably due to their higher social rank. It was therefore interesting that this pattern also occured in Lier, since grouping tendencies was less evident at low (5-7 deer/km 2) than at high (25 deer/km 2) density in a forested habitat in France (Vincent et al. 1995). Further, some individuals continue to migrate after having their own calves (Wahlström and Liberg 1995a, paper VIII). This suggests that other factors, such as familarity with the range (Clarke et al. 1993, Stamps 1995), may also play a role (Wahlström 1995).

On a fine scale, Gilbert and Bateman (1983) found that the presence of other individuals was the most important factor for bed-site selection of white-tailed deer in an experiment where cover characteristics were manipulated. Although conspecifics can be used as a cue to ensure the choice of a suitable habitat (Smith and Peacock 1990, Reed and Dobson 1993, Lima and Zollner 1996), it is likely that deer receive direct benefits from group living through a good trail systems in periods of deep snow (Messier and Barrette 1985) and increased vigilance (Linnell 1994). However, it is yet to be tested whether roe deer in groups use habitats in a different way than single individuals.



CONCLUSIONS

This study provides empirical support to the hypothesis that spatiotemporal scaling is crucial to the study of habitat selection by roe deer. For example, the temporal scale (active/inactive, night/day) clearly affected the trade-off between selection of food and cover. Reliable predictions regarding the temporal and spatial scales of habitat selection can only be inferred when the two scale dimensions are seen together. For example, cover affected roe deer selection of bed-site at both a patch and habitat scale, but cover only affected selection of feeding site at a habitat scale. The strength of the food and cover trade-offs, and hence the demographic consequences, remains to be explored. More data on sex (and reproductive status) and age dependent food-cover trade-offs are also needed to further advance our understanding of the adaptive basis for roe deer habitat selection.



REFERENCES

Aanes, R., and Andersen, R. 1996. The effects of sex, time of birth, and habitat on the vulnerability of roe deer fawns to red fox predation. Can. J. Zool . 74: 1857-1865.

Abrahamsen, J., Jacobsen, N.K., Kalliola, R., Dahl, E., Wilborg, L., and Påhlsson, L. 1977. Naturgeografisk regioninndeling av Norden. Nordiska Utredningar Ser. B 34: 1-135. (In Norwegian).

Aebischer, N.J., Robertson, P.A., and Kenward, R.E. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology 74: 1313-1325.

Albon, S.D., and Langvatn, R. 1992. Plant phenology and the benefits of migration in a temperate ungulate. Oikos 65: 502-513.

Alldredge, J.R., and Ratti, J.T. 1986. Comparison of some statistical techniques for analysis of resource selection. J. Wildl. Manage . 50: 157-165.

Alldredge, J.R., and Ratti, J.T. 1992. Further comparison of some statistical techniques for analysis of resource selection. J. Wildl. Manage . 56: 1-9.

Andersen, R., Linnell, J.D.C., and Aanes, R. 1995. Rådyret i kulturlandskapet . Fagrapport 10, NINA, Trondheim. (In Norwegian with English summary).

Arditi, R., and Dacorogna, B. 1988. Optimal foraging on arbitrary food distributions and the definition of habitat patches. Am. Nat. 131: 837-846.

Armleder, H.M., Waterhouse, M.J., Keisker, D.G., and Dawson, R.J. 1994. Winter habitat use by mule deer in the central interior of British Columbia. Can. J. Zool. 72: 1721-1725.

Armstrong, E., Euler, D., and Racey, G. 1983a. Winter bed-site selection by white-tailed deer in central Ontario. J. Wildl. Manage . 47: 880-884.

Armstrong, E., Euler, D., and Racey, G. 1983b. White-tailed deer habitat and cottage development in central Ontario. J. Wildl. Manage . 47: 605-612.

Arthur, S.M., Manly, B.F.J., McDonald, L.L., and Garner, G.W. 1996. Assessing habitat selection when availability changes. Ecology 77: 215-227.

Aulak, W., and Babinska-Werka, J. 1990a. Preference of different habitats and age classes of forest by roe deer. Acta Theriol . 35: 289-298.

Aulak, W., and Babinska-Werka, J. 1990b. Use of agricultural habitats by roe deer inhabiting a small forest area. Acta Theriol . 35: 121-127.

Bailey, D.W., Gross, J.E., Laca, E.A., Rittenhouse, L.R., Coughenour, M.B., Swift, D.M., and Sims, P.L. 1996. Mechanisms that result in large herbivore grazing distribution patterns. J. Range Manage . 49: 386-400.

Beier, P., and McCullough, D.R. 1990. Factors influencing white-tailed deer activity patterns and habitat use. Wildl. Monogr . 109: 1-51.

Belovsky, G.E. 1981. Optimal activity times and habitat choice of moose. Oecologia 48: 22-30.

Bissonette, J.A. 1997a. Wildlife and landscape ecology. Effects of pattern and scale . Springer-Verlag New York Inc., New York.

Bissonette, J.A. 1997b. Scale-sensitive ecological properties: historical context, current meaning. In: Bissonette, J.A. (ed.). Wildlife and landscape ecology. Effects of pattern and scale . Springer-Verlag New York Inc., New York. pp. 3-31.

Bjar, G., Selås, V., Lund, L.O., and Hjeljord, O. 1991. Movements and home range dynamics of roe deer, Capreolus capreolus L. in southeastern Norway. Fauna norv. Ser. A 12: 12-18.

Bjørnsen, B.H. 1985. Rådyret (Capreolus capreolus (L.)) sitt vinterhabitat i Flatdal, indre Telemark . M.Sc. thesis. Univ. of Oslo, Oslo. (In Norwegian).

Brown, J.S. 1988. Patch use as an indicator of habitat preference, predation risk, and competition. Behav. Ecol. Sociobiol . 22: 37-47.

Brown, J.S. 1992. Patch use under predation risk: I. Models and predictions. Ann. Zool. Fennici 29: 301-309.

Brown, J.S. 1996. Coevolution and community organization in three habitats. Oikos 75: 193-206.

Brown, J.S., Kotler, B.P., and Mitchell, W.A. 1994. Foraging theory, patch use, and the structure of a Negev desert granivore community. Ecology 75: 2286-2300.

Brown, J.S., and Morgan, R.A. 1995. Effects of foraging behavior and spatial scale on diet selectivity: a test with fox squirrels. Oikos 74: 122-136.

Bunnell, F.L., and Gillingham, M.P. 1985. Foraging behavior: dynamics of dining out. In: Hudson, R.J., and White, R.G. (eds.). Bioenergetics of wild herbivores. CRC Press, Inc., Boca Raton, Florida. pp. 53-79.

Büttner, K. 1983. Winterliche Fährtenzählungen beim Rehwild in Abhängigkeit von Witterung, Waldstruktur und Jagddruck. Z. Jagdwiss . 29: 82-95. (In German with English summary).

Byers, C.R., Steinhorst, R.K., and Krausman, P.R. 1984. Clarification of a technique for analysis of utilization-availability data. J. Wildl. Manage. 48: 1050-1053.

Byers, J.A., and Byers, K.Z. 1983. Do pronghorn mothers reveal the locations of their hidden fawns? Behav. Ecol. Sociobiol. 13: 147-156.

Cederlund, G. 1981. Daily and seasonal activity pattern of roe deer in a boreal habitat. Swed. Wildl. Res . 11: 315-353.

Cederlund, G. 1983. Home range dynamics and habitat selection by roe deer in a boreal area in central Sweden. Acta Theriol . 28, 30: 443-460.

Cederlund, G. 1989. Activity patterns in moose and roe deer in a north boreal forest. Holarct. Ecol. 12: 39-45.

Cederlund, G., and Liberg, O. 1995. Rådjuret. Viltet, ekologin och jakten . Svenska Jägareförbundet, Solna. (In Swedish).

Cederlund, G., and Lindström, E. 1983. Effects of severe winters and fox predation on roe deer mortality. Acta Theriol. 28, 7: 129-145.

Cederlund, G., Ljungqvist, H., Markgren, G., and Stålfelt, F. 1980. Foods of moose and roe deer at Grimsö in central Sweden - results of rumen content analysis. Swed. Wildl. Res. 11: 167-247.

Cederlund, G., and Nyström, A. 1981. Seasonal differences between moose and roe deer in ability to digest browse. Holarct. Ecol. 4: 59-65.

Charnov, E.L. 1976. Optimal foraging: the marginal value theorem. Theor. Pop. Biol. 9: 129-136.

Cherry, S. 1996. A comparison of confidence interval methods for habitat use-availability studies. J. Wildl. Manage . 60: 653-658.

Cibien, C., and Sempere, A.J. 1989. Food availability as a factor in habitat use by roe deer. Acta Theriol . 34, 7: 111-123.

Clark, J.D., Dunn, J.E., and Smith, K.G. 1993. A multivariate model of female black bear habitat use for a geographic information system. J. Wildl. Manage. 57: 519-526.

Clarke, M.F., da Silva, K.B., Lair, H., Pocklington, R., Kramer, D.L., and McLaughlin, R.L. 1993. Site familarity affects escape behaviour of the eastern chipmunk, Tamias striatus . Oikos 66: 533-537.

Cowlishaw, G. 1997a. Refuge use and predation risk in a desert baboon population. Anim. Behav . 54: 241-253.

Cowlishaw, G. 1997b. Trade-offs between foraging and predation risk determine habitat use in a desert baboon population. Anim. Behav . 53: 667-686.

Danilkin, A.A. in association with Hewison, A.J.M. 1996. Behavioural ecology of Siberian and European roe deer . Chapman and Hall, London.

Darwin, C. 1859. On the origin of species . Murray, London.

Davies, N.B. 1991. Mating systems. In: Krebs, J.R., and Davies, N.B. (eds.). Behavioural ecology. An evolutionary approach . Blackwell Sci. Publ., London. pp. 263-294.

Demarchi, M.W., and Bunnell, F.L. 1995. Forest cover selection and activity of cow moose in summer. Acta Theriol. 40: 23-36.

Demment, M.W., and Van Soest, P.J. 1985. A nutritional explanation for body-size patterns of ruminant and nonruminant herbivores. Am. Nat. 125: 641-672.

Dzieciolowski, R. 1976. Roe deer census by pellet-group counts. Acta Theriol . 21, 26: 351-358.

Edwards, G.R., Newman, J.A., Parsons, A.J., and Krebs, J.R. 1994. Effects of the scale and spatial distribution of the food resource and animal state on diet selection: an example with sheep. J. Anim. Ecol . 63: 816-826.

Elston, D.A., Illius, A.W., and Gordon, I.J. 1996. Assessment of preference among a range of options using log ratio analysis. Ecology 77: 2538-2548.

Fretwell, S.D., and Lucas, H.L. 1970. On territorial behaviour and other factors influencing habitat distribution in birds. I. Theoretical development. Acta Biotheor . 19: 16-36.

Gaillard, J-M., Sempere, A.J., Boutin, J-M., Van Laere, G., and Boisaubert, B. 1992. Effects of age and body weight on the proportion of females breeding in a population of roe deer ( Capreolus capreolus ). Can. J. Zool. 70: 1541-1545.

Gatenby, R.M. 1977. Conduction of heat from sheep to ground. Agric. Meterol . 18: 387-400.

Gautestad, A.O., and Mysterud, I. 1993. Physical and biological mechanisms in animal movement processes. J. Appl. Ecol . 30: 523-535.

Gautestad, A.O., and Mysterud, I. 1995. The home range ghost. Oikos 74: 195-204.

Gilbert, F.F., and Bateman, M.C. 1983. Some effects of winter shelter conditions on white-tailed deer, Odocoileus virginianus , fawns. Can. Field-Nat . 97: 391-400.

Gill, J. 1966. What's a deer yard in Maine? The Northern Logger 14 (8): 14-15, 28.

Gillingham, M.P., Parker, K.L., and Hanley, T.A. 1997. Forage intake by black-tailed deer in a natural environment: bout dynamics. Can. J. Zool. 75: 1118-1128.

Griffith, B., and Youtie, B.A. 1988. Two devices for estimating foliage density and deer hiding cover. Wildl. Soc. Bull. 16: 206-210.

Guillet, C., Bergström, R., Cederlund, G., Bergström, J., and Ballon, P. 1995. Comparison of telemetry and pellet-group counts for determining habitat selectivity by roe deer ( Capreolus capreolus ) in winter. Gibier Faune Sauvage, Game Wildl. 12: 253-269.

Hanley, T.A. 1982. Cervid activity patterns in relation to foraging constraints: western Washington. Northwest Sci. 56: 208-217.

Hanley, T.A. 1984. Habitat patches and their selection by wapiti and black-tailed deer in a coastal montane coniferous forest. J. Appl. Ecol. 21: 423-436.

Hanley, T.A. 1997. A nutritional view of understanding and complexity in the problem for diet selection by deer (Cervidae). Oikos 79: 209-218.

Helle, P. 1980. Food composition and feeding habits of the roe deer in winter in central Finland. Acta Theriol . 25, 22: 395-402.

Henry, B.A.M. 1981. Distribution patterns of roe deer ( Capreolus capreolus ) related to the availability of food and cover. J. Zool .: 271-275.

Hewison, A.J.M., and Gaillard, J-M. 1996. Birth-sex ratios and local resource competition in roe deer, Capreolus capreolus . Behav. Ecol. 7: 461-464.

Hjeljord, O. 1997. Moose ( Alces alces ) habitat quality - an empirical approach. Ph.D. thesis. Agric. Univ. of Norway, Ås.

Hjeljord, O., Hövik, N., and Pedersen, H.B. 1990. Choice of feeding sites by moose during summer, the influence of forest structure and plant phenology. Holarct. Ecol . 13: 281-292.

Hobbs, N.T., and Hanley, T.A. 1990. Habitat evaluation: do use/availability data reflect carrying capacity? J. Wildl. Manage. 54: 515-522.

Holand, Ø. 1992. Winter digestive strategy of a concentrate selector in Norway: the European roe deer. Can. J. Zool . 70: 1331-1335.

Holand, Ø. 1993. In vitro dry matter disappearance using roe deer inocula from summer and winter. Rangifer 13: 143-147.

Holand, Ø. 1994. Seasonal dynamics of digestion in relation to diet quality and intake in European roe deer ( Capreolus capreolus ). Oecologia 98: 274-279.

Holand, Ø., Mysterud, A., Wannag, A., and Linnell, J.D.C. 1998. Roe deer in northern environments: physiology and behaviour. In: Andersen, R., Duncan, P., and Linnell, J.D.C. (eds.). European roe deer: the biology of success . Scandinavian Univ. Press, Oslo. In press.

Hufthammer, A.K. 1992. Prehistorical and early historical distribution of roe deer Capreolus capreolus (L.) in Norway. Fauna norv. Ser. A 13: 35-37.

Hufthammer, A.K., and Aaris-Sørensen, K. 1998. Late- and postglacial European roe deer. In: Andersen, R., Duncan, P., and Linnell, J.D.C. (eds.). European roe deer: the biology of success . Scandinavian Univ. Press, Oslo. In press.

Huot, J. 1974. Winter habitat of white-tailed deer at Thirty-one Mile lake, Quebec. Can. Field-Nat . 88: 293-301.

Illius, A.W., and Gordon, I.J. 1987. The allometry of food intake in grazing ruminants. J. Anim. Ecol. 56: 989-999.

Ims, R.A. 1988. Spatial clumping of sexually receptive females induces space sharing among male voles. Nature 335: 541-543.

Ims, R.A. 1995. Movement patterns in relation to spatial structure. In: Hansson, L., Fahrig, L., and Merriam, G. (eds.). Mosaic landscapes and ecological processes. Chapman and Hall, London. pp. 86-119.

Ims, R.A., and Yoccoz, N.G. 1995. Ecological methodology: study design and statistical analysis . Technical report. Univ. of Oslo, Oslo.

Jacobsen, N.K. 1980. Differences of thermal properties of white-tailed deer pelage between seasons and body regions. J. Therm. Biol . 5: 151-158.

Jefferies, R.L., Klein, D.R., and Shaver G.R. 1994. Vertebrate herbivores and northern plant communities: reciprocal influences and responses. Oikos 71: 193-206.

Jeppesen, J.L. 1989. Activity patterns of free-ranging roe deer ( Capreolus capreolus ) at Kalø. Dan. Rev. Game Biol. 13 (8): 1-32.

Johansson, A. 1996. Territorial dynamics and marking behaviour of male roe deer . Ph.D. thesis. Univ. of Stockholm, Stockholm.

Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61: 65-71.

Kjøstvedt, J.H., Mysterud, A., and Østbye, E. 1998. Roe deer Capreolus capreolus use of agricultural crops during winter in the Lier valley in Norway. Wildl. Biol. 4: in press.

Knight, T.W., and Morris, D.W. 1996. How many habitats do landscapes contain? Ecology 77: 1756-1764.

Kotler, B.P. 1997. Patch use by gerbils in a risky environment: manipulating food and safety to test four models. Oikos 78: 274-282.

Kotler, B.P., and Blaustein, L. 1995. Titrating food and safety in a heterogenous environment: when are the risky and safe patches of equal value? Oikos 74: 251-258.

Kotler, B.P., Gross, J.E., and Mitchell, W.A. 1994. Applying patch use to assess aspects of foraging behaviour in Nubian ibex. J. Wildl. Manage . 58: 299-307.

Kotliar, N.B., and Wiens, J.A. 1990. Multiple scales of patchiness and patch structure: a hierarchical framework for the study of heterogeneity. Oikos 59: 253-260.

Kremsater, L.L., and Bunnell, F.L. 1992. Testing responses to forest edges: the example of black-tailed deer. Can. J. Zool . 70: 2426-2435.

Kucera, E. 1976. Effects of winter conditions on the white-tailed deer of Delta Marsh, Manitoba. Can. J. Zool. 54: 1307-1313.

Laca, E.A., and Demment, M.W. 1996. Foraging strategies of grazing animals. In: Hodgson, J., and Illius, A.W. (eds.). The ecology and management of grazing systems. CAB International, Wallingford, UK. pp. 137-158.

Lang, B.K., and Gates, J.E. 1985. Selection of sites for winter night beds by white-tailed deer in a hemlock-northern hardwood forest. Am. Midl. Nat . 113: 245-254.

Langvatn, R., and Hanley, T.A. 1993. Feeding-patch choice by red deer in relation to foraging efficiency; an experiment. Oecologia 95: 164-170.

Latham, J., Staines, B.W., and Gorman, M.L. 1996. The relative densities of red ( Cervus elaphus ) and roe ( Capreolus capreolus ) deer and their relationship in Scottish plantation forests. J. Zool . 240: 285-299.

Latham, J., Staines, B.W., and Gorman, M.L. 1997. Correlations of red ( Cervus elaphus ) and roe ( Capreolus capreolus ) deer densities in Scottish forests with environmental variables. J. Zool . 242: 681-704.

Lemmon, P.E. 1956. A spherical densiometer for estimating forest overstory density. For. Sci. 2: 314-320.

Lemmon, P.E. 1957. A new instrument for measuring forest overstory density. J. For. 55: 667-669.

Lent, P.C. 1974. Mother-infant relationships in ungulates. IUCN Publ. new series 1 (24): 14-55.

Lima, S.L., and Dill, L.M. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Can. J. Zool. 68: 619-640.

Lima, S.L., and Zollner, P.A. 1995. Towards a behavioral ecology of ecological landscapes. Trend. Ecol. Evol . 11: 131-135.

Linnell, J.D.C. 1994. Reproductive tactics and parental care in Norwegian roe deer . Ph.D. thesis. Nat. Univ. of Ireland, Cork.

Linnell, J.D.C., Aanes, R., and Andersen, R. 1995. Who killed bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildl. Biol . 1: 209-224.

Linnell, J.D.C., and Andersen, R. 1995. Høy rådyr-dødelighet i gaupeland. Jakt og Fiske 124 (10): 50-52. (In Norwegian).

Loison, A., Gaillard, J-M., Pelabon, C., and Yoccoz, N.G. 1998. What factors shape sexual size dimorphism in ungulates? Evol. Ecol .: in press.

MacArthur, R.A., Johnston, R.H., and Geist, V. 1979. Factors affecting heart rate in free-ranging bighorn sheep: a physiological approach to the study of wildlife harassment. Can. J. Zool. 57: 2010-2021.

Manly, B.F.J., McDonald, L.L., and Thomas, D.L. 1993. Resource selection by animals: statistical design and analysis for field studies . Chapman and Hall, London.

Merrill, E.H. 1991.Thermal constraints on use of cover types and activity time of elk. Appl. Anim. Behav. Sci. 29: 251-267.

Messier, F., and Barrette, C. 1985. The efficiency of yarding behaviour by white-tailed deer as an antipredator strategy. Can. J. Zool. 63: 785-789.

Milinski, M., and Parker, G.A. 1991. Competition for resources. In: Krebs, J.R., and Davies, N.B. (eds.). Behavioural ecology. An evolutionary approach. Blackwell Sci. Publ., London. pp. 137-168.

Moen, A.N. 1968. Energy exchange of white-tailed deer, western Minnesota. Ecology 49: 676-682.

Moen, A.N. 1973. Wildlife ecology . W.H. Freeman and Company, San Francisco.

Morgan, R.A., and Brown, J.S. 1996. Using giving-up densities to detect search images. Am. Nat . 148: 1059-1074.

Morgan, R.A., Brown, J.S., and Thorson, J.M. 1997. The effect of spatial scale on the functional response of fox squirrels. Ecology 78: 1087-1097.

Morris, D.W. 1987. Ecological scale and habitat use. Ecology 68: 362-369.

Morris, D.W. 1992. Scales and costs of habitat selection in heterogenous landscapes. Evol. Ecol . 6: 412-432.

Morris, D.W. 1997. Optimally foraging deer mice in prairie mosaics: a test of habitat theory and absence of landscape effects. Oikos 80: 31-42.

Murray, D.L., Boutin, S., O'Donoghue, M., and Nams, V.O. 1995. Hunting behaviour of a sympatric felid and canid in relation to vegetative cover. Anim. Behav. 50: 1203-1210.

Mysterud, A. 1993. Rådyrets (Capreolus capreolus) liggeplasser og leieatferd om vinteren . M.Sc. thesis. Univ. of Oslo, Oslo. (In Norwegian).

Mysterud, A. 1998. The relative roles of body size and feeding type for activity time of temperate ruminants. Oecologia 113: 442-446.

Mysterud, A., and Østbye, E. 1995. Roe deer Capreolus capreolus feeding on yew Taxus baccata in relation to bilberry Vaccinium myrtillus density and snow depth. Wildl. Biol. 1: 249-253.

Mysterud, A., and Østbye, E. 1998. Cover as a habitat element for temperate ungulates: effects on habitat selection and demography. Ms. (Submitted to Wildl. Soc. Bull.).

Nelson, M.E. 1995. Winter range arrival and departure of white-tailed deer in northeastern Minnesota. Can. J. Zool. 73: 1069-1076.

Nelson, M.E., and Mech, L.D. 1981. Deer social organization and wolf predation in northeastern Minnesota. Wildl. Monogr . 77: 1-53.

Neu, C.W., Byers, C.R., and Peek, J.M. 1974. A technique for analysis of utilization-availability data. J. Wildl. Manage . 38: 541-545.

Nixon, C.M., Hansen, L.P., Brewer, P.A., and Chelsvig, J.E. 1991. Ecology of white-tailed deer in an intensively farmed region of Illinois. Wildl. Monogr. 118: 1-77.

Nudds, T.D. 1977. Quantifying the vegetative structure of wildlife cover. Wildl. Soc. Bull. 5: 113-117.

Nudds, T.D. 1980. Forage "preference": theoretical considerations of diet selection by deer. J. Wildl. Manage . 44: 735-740.

Nuttle, T. 1997. Densiometer bias? Are we measuring the forest or the trees? Wildl. Soc. Bull. 25: 610-611.

Orians, G.H., and Wittenberger, J.F. 1991. Spatial and temporal scales in habitat selection. Am. Nat . 137: 29-49.

Otis, D.L. 1997. Analysis of habitat selection studies with multiple patches within cover types. J. Wildl. Manage. 61: 1016-1022.

Papageorgiou, N.K. 1978. Use of forest openings by roe deer as shown by pellet group counts. J. Wildl. Manage. 42: 650-654.

Parker, K.L., Robbins, C.T., and Hanley, T.A. 1984. Energy expenditure for locomotion by mule deer and elk. J. Wildl. Manage. 48: 474-488.

Pauley, G.R., Peek, J.M., and Zager, P. 1993. Predicting white-tailed deer habitat use in northern Idaho. J. Wildl. Manage. 57: 904-913.

Peek, J.M., Scott, M.D., Nelson, L.J., Pierce, D.J., and Irwin, L.L. 1982. Role of cover in habitat management for big game in northwestern United States. Trans. N. Am. Wildl. Nat. Resour. Conf. 47: 363-373.

Reed, J.M., and Dobson, A.P. 1993. Behavioural constraints and conservation biology: conspecific attraction and recruitment. Trend. Ecol. Evol. 8: 253-256.

Renecker, L.A., Christopherson, M.K., and Arelis, C. 1978. Effects of posture, feeding, low temperature and wind on energy expenditures of moose calves. In: Cumming, H.G., Bubenik, A.B., Crete, M., Franzmann, A.W., Jordan, P.A., Lankester, M.W., Patton, A.E., and Telfer E.S. (eds.). Proc. of the North American moose conference workshop . Thunder Bay, Ontario. pp. 126-140.

Rosenzweig, M. 1981. A theory of habitat selection. Ecology 62: 327-335.

Saltz, D. 1994. Reporting error measure in radio location by triangulation: a review. J. Wildl. Manage. 58: 181-184.

Sargeant, G.A., Eberhardt, L.E., and Peek, J.M. 1994. Thermoregulation by mule deer ( Odocoileus hemionus ) in arid rangelands of southcentral Washington. J. Mammal. 75: 536-544.

Schaefer, J.A., and Messier, F. 1995a. Habitat selection as a hierarchy: the spatial scales of winter foraging by muskoxen. Ecography 18: 333-344.

Schaefer, J.A., and Messier, F. 1995b. Winter foraging by muskoxen: a hierarchical approach to patch residence time and cratering behaviour. Oecologia 104: 39-44.

Schmidt, K.A., and Brown, J.S. 1996. Patch assessment in fox squirrels: the role of resource density, patch size, and patch boundaries. Am. Nat. 147: 360-380

Schmitz, O.J. 1991. Thermal constraints and optimization of winter feeding and habitat choice in white-tailed deer. Holarct. Ecol . 14: 104-111.

Schwab, F.E., and Pitt, M.D. 1991. Moose selection of canopy cover types related to operative temperature, forage, and snow depth. Can. J. Zool. 61: 3071-3077.

Selås, V., Bjar, G., Betten, O., Tjeldflaat, L.O., and Hjeljord, O. 1991. Feeding ecology of roe deer, Capreolus capreolus L., during summer in southeastern Norway. Fauna norv. Ser. A 12: 5-11.

Senft, R.L., Cougenhour, M.B., Bailey, D.W., Rittenhouse, R.L., Sala, O.E., and Swift, D.M. 1987. Large herbivore foraging and ecological hierarchies. Bioscience 37: 789-799.

Smith, A.T., and Peacock, M.M. 1990. Conspecific attraction and the determination of metapopulation colonization rates. Cons. Biol . 4: 320-323.

Spencer, S.R., Cameron, G.N., and Swihart R.K. 1990. Operationally defining home range: temporal dependence exhibited by hispid cotton rats. Ecology 71: 1817-1822.

Springer, J.T. 1979. Some sources of bias and sampling error in radio triangulation. J. Wildl. Manage. 43: 926-935.

Staines, B.W. 1976. The use of natural shelter by red deer ( Cervus elaphus ) in relation to weather in north-east Scotland. J. Zool . 180: 1-8.

Stamps, J. 1995. Motor learning and the value of familiar space. Am. Nat. 146: 41-58.

Stamps, J.A., Buechner, M., and Krishnan, V.V. 1987. The effects of edge permeability and habitat geometry on emigration from patches of habitat. Am. Nat . 129: 533-552.

Strandgaard, H. 1972. The roe deer ( Capreolus capreolus ) population at Kalø and the factors regulating its size. Dan. Rev. Game Biol . 7 (1): 1-205.

Stüwe, M., and Hendrichs, H. 1984. Organization of roe deer ( Capreolus capreolus ) in an open field habitat. Z. Säugetierkd . 49: 359-367.

Telleria, J.L., and Virgos, E. 1997. Distribution of an increasing roe deer population in a fragmented Mediterranean landscape. Ecography 20: 247-252.

Thomas, D.L., and Taylor, E.J. 1990. Study designs and tests for comparing resource use and availability. J. Wildl. Manage . 54: 322-330.

Tixier, H., and Duncan, P. 1996. Are European roe deer browsers? A review of variations in the composition of their diets. Revue d'Ecologie (La Terre et la Vie) 51: 3-17.

Tixier, H., Duncan, P., Scehovic, J., Yani, A., Gleizes, M., and Lila, M. 1997. Food selection by European roe deer ( Capreolus capreolus ): effects of plant chemistry, and consequences for the nutritional value of their diets. J. Zool. 242: 229-245.

Trivers, R.L. 1972. Parental investment and sexual selection. In: Campbell, B. (ed.). Sexual selection and the descent of man . Aldine Press, Chicago. pp. 136-179.

Tufto, J., Andersen, R., and Linnell, J.D.C. 1996. Habitat use and ecological correlates of home range size in a small cervid: the roe deer. J. Anim. Ecol . 65: 715-724.

Turner, M.G. 1989. Landscape ecology: the effect of pattern on process. Annu. Rev. Ecol. Syst . 20: 171-197.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. J. Wildl. Manage. 47: 893-901.

Vincent, J.P., Bideau, E., Hewison, A.J.M., and Angibault J.M. 1995. The influence of increasing density on body weight, kid production, home range and winter grouping in roe deer ( Capreolus capreolus ). J. Zool. 236: 371-382.

Wahlström, L.K. 1994. The significance of male-male aggression for yearling dispersal in roe deer ( Capreolus capreolus ). Behav. Ecol. Sociobiol . 35: 409-412.

Wahlström, L.K. 1995. Natal dispersal in roe deer - an evolutionary perspective . Ph.D. thesis. Univ. of Stockholm, Stockholm.

Wahlström, L.K., and Kjellander, P. 1995. Ideal free distribution and natal dispersal in female roe deer. Oecologia 103: 302-308.

Wahlström, L.K., and Liberg, O. 1995a. Patterns of dispersal and seasonal migration in roe deer ( Capreolus capreolus ). J. Zool. 235: 455-467.

Wahlström, L.K., and Liberg, O. 1995b. Contrasting dispersal patterns in two Scandinavian roe deer Capreolus capreolus populations. Wildl. Biol . 1: 159-164.

Ward, D., and Saltz, D. 1994. Foraging at different spatial scales: Dorcas gazelles foraging for lilies in the Negev desert. Ecology 75: 48-58.

Warren, J.T., and Mysterud, I. 1991. Summer habitat use and activity patterns of domestic sheep on coniferous range in southern Norway. J. Range Manage. 44: 2-6.

Warren, J.T., Mysterud, I., Stensli, O.M., Berg, M., and Movik, E. 1993. Use of forest vegetation by sheep as recorded by radio telemetry. Acta Agric. Scand.; Sec. A. Anim. Sci. 43: 190-192.

Welch, D., Staines, B.W., Catt, D.C., and Scott, D. 1990. Habitat usage by red ( Cervus elaphus ) and roe ( Capreolus capreolus ) deer in a Scottish sitka spruce plantation. J. Zool . 221: 453-476.

Wiens, J.A. 1989. Spatial scaling in ecology. Funct. Ecol . 3: 385-397.

Wiens, J.A., Stenseth, N.C., Van Horne, B., and Ims, R.A. 1993. Ecological mechanisms and landscape ecology. Oikos 66: 369-380.