Habitat selection of adult moose Alces alces at two spatial scales in central Finland

The habitat selection criteria of moose Alces alces at several scales are the basic sets of information needed in moose management planning. We studied moose habitat use in central Finland during 1993–1996 using data from radio collared moose, satellite image based forest and land cover data, and applied the principles of compositional analysis. The habitat compositions of 54 home ranges (10 males during summer, six males during winter, 23 females during summer and 15 females during winter) were first compared with the overall landscape. The habitat compositions around moose locations within their home ranges were then compared with the habitat composition of the home range. Seasons and sexes were compared at both scales. In summer, there was only a slight difference between moose home ranges and the overall landscape. Based on tree species composition, home ranges are located in slightly more fertile areas than the overall landscape. Within their home ranges, moose favoured non-pine dominated habitats and mature forests, and avoided human settlements. In winter, the moose home ranges included significantly more pine-dominated plantations and other young successional stages than the overall landscape. The role of pine-dominated peatland forests/ shrub land was especially pronounced in winter. Winter home ranges included less agricultural land and human settlements than the overall landscape, probably due to the more distant location of important winter habitats from man-made landscapes. Within the home ranges, both sexes used non-pine dominated habitats more, and mature forests and human settlements less than expected. At the home range scale, there were no statistical differences between the sexes with respect to habitat use in either season. Within their home ranges, males and females used slightly different habitats during both seasons, suggesting spatially segregated habitat use by the individual sexes. The difference is more clear in winter when males tend to use more pine-dominated, young successional habitats than females. Compared to the situation in the summer, winter ranges are located in slightly more pine-dominated habitats with fewer settlements and agricultural fields. The shift in habitat use between the two seasons is more pronounced with respect to habitat use within the home range. Our results indicate that moose habitat selection criteria vary among different hierarchical levels of selection. We stress the importance of multi-scale assessment of the habitat and other resource selection of animals.

The habitat selection criteria of moose Alces alces at several scales are the basic sets of information needed in moose management planning. We studied moose habitat use in central Finland during 1993-1996 using data from radio collared moose, satellite image based forest and land cover data, and applied the prin ciples of compositional analysis. The habitat compositions of 54 home ranges (10 males during summer, six males during winter, 23 females during summer and 15 females during winter) were first compared with the overall land scape. The habitat compositions around moose locations within their home ranges were then compared with the habitat composition of the home range. Seasons and sexes were compared at both scales. In summer, there was only a slight difference between moose home ranges and the overall landscape. Based on tree species composition, home ranges are located in slightly more fertile areas than the overall landscape. Within their home ranges, moose favoured non-pine dominated habitats and mature forests, and avoided human settlements. In winter, the moose home ranges included significantly more pine-dominat ed plantations and other young successional stages than the overall landscape. The role of pine-dominated peatland forests/ shrub land was especially pro nounced in winter. Winter home ranges included less agricultural land and human settlements than the overall landscape, probably due to the more distant loca tion of important winter habitats from man-made landscapes. Within the home ranges, both sexes used non-pine dominated habitats more, and mature forests and human settlements less than expected. At the home range scale, there were no statistical differences between the sexes with respect to habitat use in either season. Within their home ranges, males and females used slightly different habitats during both seasons, suggesting spatially segregated habitat use by the individual sexes. The difference is more clear in winter when males tend to use more pine-dominated, young successional habitats than females. Compared to the situation in the summer, winter ranges are located in slightly more pine-d o m i n a t e d habitats with fewer settlements and agricultural fields. The shift in habitat use between the two seasons is more pronounced with respect to habi tat use within the home range. Our results indicate that moose habitat selec tion criteria vary among different hierarchical levels of selection. We stress the importance of multi-scale assessment of the habitat and other resource selec tion of animals.
M oose A lces alces has been a part of the Fennoscandian fauna since the glaciation period of the last Ice Age which ended about 10,000 years ago (Pulliainen 1987), and today the geographical distribution o f the m oose covers all the Fennoscandian countries (Cederlund & Markgren 1987, N ygrén 1987, Ø stgård 1987. In the 1970s and 1980s, the Fennoscandian m oose population increased from tens o f thousands o f animals to several hundred thousand anim als and, in some areas, the m oose pop ulation has probably been the densest in the w orld (Ce derlund & M arkgren 1987, Cederlund & Bergström 1996. Because o f the high population density, m oose has an am biguous position in Fennoscandian nature: on the one hand it is a valuable game animal (Mattson 1990), and on the other hand it is considered a severe pest in especially forest plantations (Lavsund 1987(Lavsund , H örnberg 1995. The rapid increase in population levels has been related to changes in hunting practises and forest m an agem ent m ethods after the 1940s-1950s (A hlén 1975, L avsund 1987, Ø stgård 1987) during w hich period clear-cutting becam e the established forest regenera tion m ethod, and pine becam e favoured over the oth er tree species. Furtherm ore, the econom ically optim al age class distribution from a forestry point o f view includes a high proportion of young forest plantations. It has been hypothesised that large areas o f young for est plantation in practise provide an unlim ited, con tinuous supply o f food, that allow s an effective repro duction o f m oose (Cederlund & M arkgren 1987).
One basic com ponent in moose management is infor mation about m oose habitat selection at several scales. T he general features o f the habitat requirem ents o f Fennoscandian m oose have been described by sever al authors (see e.g. Bergström & H jeljord (1987) and references therein). In boreal regions, m oose is adapt ed to the m osaic o f habitats created by natural distur bances, and moose especially favour young succession al stages (G eist 1974(G eist , Cederlund & B ergström 1996. E specially in w inter and spring, w hen m ost o f the m oose dam age in forest plantations occurs, the role of pine in the m oose diet is pronounced (Lavsund 1987). O lder forests are especially im portant in w inter due to their thinner snow cover, but also during snow free periods due to the presence o f im portant food plants in the d w arf shrub layer (H jeljord, H övik & Pedersen 1990).
A n an im al's habitat selection can be seen as a hier archical process in w hich an individual first selects its hom e range within the species' distributional range, and then selects betw een the various habitat types within the hom e range (Johnson 1980). Recognising this hier archical nature o f selection is essential as it affects our conclusions about the availability and usage o f differ ent resource com ponents (Johnson 1980). Studies on the habitat and other resource selections of the moose at different hierarchical levels have indicated that selec tion occurs at least at the region/landscape level (For bes & Theberge 1993), at the habitat level (e.g. C eder lund & O karm a 1988, H jeljord et al. 1990, Bø & H jel jord 1991, Heikkilä & H ärkönen 1993, Ball, Nordengren & W allin 2001, w ithin habitats in relation to food resource availability (Vivås & Saether 1987) and compo sition (D anell, E denius & Lundberg 1991) and, final ly, at the food item level (N iem elä & D anell 1988).
M ulti-scale assessm ents o f the habitat selection of moose from the home range to selection within the home range are still rare in the literature. Also, the habitat selection o f moose has been studied with radio telemetry on only a few occasions in Fennoscandia (Cederlund & O karm a 1988, H jeljord et al. 1990, Bø & H jeljord 1991, H eikkilä, N ygrén, Härkönen & M ykkänen 1996, B all et al. 2001. A nalytically (A ebischer, Robertson & K enw ard 1993), m ost o f the telem etry studies have assessed the habitat selection o f m oose w ithin hom e ranges or the landscape level in general.
In addition to hierarchical levels o f selection, the effect o f other factors such as season and sex should also be taken into account in resource selection studies (A eb i s h e r et al. 1993). Seasonal migrations o f m oose have been studied in several parts of the distributional range o f the species, and the distances betw een sum m er and w inter habitats have been reported to range from a few kilometers to several hundred kilometers (LeResche 1974, Pulliainen 1974, Kuznetsov 1987. The main fac tors underlying seasonal m igrations have been related to seasonal heterogeneity and availability of forage (Bergström & Hjeljord 1987), snow conditions (Sandeg r e n , B e r g s t r ö m & Sw eanor 1985) or a com bination of these factors (LeResche 1974). However, not all the stud ies have found differences in habitat selection between seasons (C ederlund & O karm a 1988), and in some areas, the m oose has been reported to be only partial ly m igratory (Ball et al. 2001). Studies on differences in m oose habitat use betw een sexes in Fennoscandia are lacking, but corresponding studies perform ed in northern A m erica report differences betw een m ales and females in this respect (Leptich & Gilbert 1989, M il ler & Litvaitis 1992, Thompson, Gilbert, Matula & M or ris 1995. T he aim o f our study was to analyse the habitat use o f m oose at tw o scales using telem etry data, satellite im age based land use and cover data, and applying the principles o f com positional analysis (A ebischer et al. 1993). W e first analysed w hether the habitat com po sition o f hom e ranges is different from the overall landscape habitat composition. Then, by using individual locations, w e analysed w hether m oose use habitats within their hom e ranges disproportionally to the habi tat com position o f their hom e range. T he analyses w ere m ade separately for individual seasons, i.e. w in ter and summer, and males and females were compared w ithin the seasons. The seasons were also com pared at both scales.

Study area
The study was carried out in the Province of Oulu in cen tral Finland (65°N, 25°30'E; Fig. 1). This area lies in the m iddle boreal region o f Finland (Ahti, H äm et-Ahti & Jalas 1968), and a high proportion o f the area is form er sea bottom that has gradually been exposed during the period follow ing the retreat o f the glaciers after the last Ice A ge, about 10,000 years ago. Locating the radio collared moose During 1993During -1996 moose (37 males and 36 females) w ere captured and radio collared (Heikkinen 2000). The m oose w ere radio collared in two different areas ca 45 km apart (see Fig. 1). A fter release, the anim als w ere located by triangulation using a Y agi-type hand held antenna. T he location o f the radio collared m oose was determined once a week all year around, except dur ing periods of intensive movement in spring and autumn w hen 2-3 locations per w eek were made. If the m oose w as not detected from the ground, an aircraft was used to track the anim als from the air. In such cases, how ever, the actual location was always determ ined on the ground. By the end o f 1996, the total num ber o f loca tions w as 4,544. D uring our study, w e adhered to the guidelines for the use o f animals in research and the legal requirem ents o f F inland and the Finnish G am e and Fisheries Research Institute.

Determination of home ranges
The beginning and the end o f the m ovem ent periods of m oose and their hom e range periods, respectively, vary between years, and to some degree also among individual m oose (Heikkinen 2000). Therefore, instead o f using fixed tim e periods, we distinguished between the home range periods and the intensive m ovement periods indi vidually for each m oose by visually exam ining the dis- tance, dispersion and timing between locations. W hen the winter hom e range period ends and the spring m ove m ents start, there is an abrupt increase in the distance o f locations to a cluster o f locations within the w inter hom e range. W hen the sum m er hom e range period ends and the autumn movements start, the locations start to disperse over a much larger area and at a greater dis tance from the cluster of locations that constitute the sum m er home ranges. Similarly, the start of the summer and winter home range periods were determined by examin ing w hen the locations started to cluster (Heikkinen 2000).
H om e range boundaries were determ ined using the harmonic mean method (Dixon & Chapman 1980) util ising a grid system with 40 X 40 m squares and Ranges V software (Kenward & Hodder 1996). W e used a m in imum of 20 locations to determine the home range of each sex as the num ber o f observations per season, espe cially in winter, was low (for male summer: mean = 30.9, SD = 11.9; for m ale winter: mean = 28.4, SD = 6.7; for fem ale winter: m ean = 24.3; SD = 5.2 and for female summer: m ean = 32.5, SD = 11.9). Also, an increase in the total hom e range size of m oose with an increasing num ber o f relocations could be the result o f gradual changes in range use over the course of time (Doerr 1983).
Because the 100% isopleths o f the hom e ranges are often influenced by infrequent and outlying locations, especially with a low num ber o f locations per hom e range, it is preferable to use low er isopleths to obtain a more accurate range representation (Harris, Cresswell, Forde, Trewhella, W oollard & W ray 1990). Therefore, we plotted the utilisation distribution for each m oose according to their radio locations, and looked for pos sible points o f inflection in order to determ ine centres o f activity (i.e. core areas). In most cases, the slope dis continuity was found in about 80% of the utilisation area (on average 30.1 % (1,200 ha) of the m axim um home range area), and therefore it was used to represent the hom e ranges o f the m oose (hereon referred to as hom e range). Coordinates o f the hom e range boundaries were imported to GIS using D X F-interchange files produced with Ranges V (Kenw ard & H odder 1996).

Digital maps and landscape variables
Land use and forest data were provided by the Finnish N ational Forest Inventory (NFI). In Finland, the NFI utilises L andsat TM 5 satellite im ages concurrently with field plots, as w ell as digital maps o f roads, agri cultural land and other non-forest land, to separate non forest land from forest land. The m ulti-source method (Tom ppo 1991(Tom ppo , 1996 uses the k-nn method for pro ducing estimates of e.g. timber volume for each tree spe cies for every pixel corresponding to 25 X 25 m land area. The original satellite im age was recorded in 1991. Digital m aps o f tim ber volum e estim ates for pine, spruce and deciduous trees, as w ell as digital maps o f fields and settlem ents, roads, w aters and peatlands, w ere im ported to GIS as separate layers. For the anal ysis, we com bined tim ber volume layers and other land use layers to a single land use and cover layer using total timber volume as a proxy o f forest age (Tomppo, Katila, M äkelä & Peräsaari 1998). In the resulting data, each pixel can belong to one o f 12 classes (Table 1).

Compositional analysis of home ranges
A ebischer et al. (1993) stated that there are four prob lems associated with the analysis o f m any habitat use data sets: 1) Proper determ ination o f sam ple units, i.e. individual animals should (usually) be used rather than individual radio locations; 2) the proportions of habi tats sum up to 1 over all habitat types (unit-sum con straint), w hich makes the habitat proportions non-inde pendent; 3) analysis should enable testing between-g r o u p differences by reference to within-group between  (Johnson 1980). To address the above-m entioned problem s 1 and 3, we calculated hom e ranges for each m oose separated further by sex (males and females) and season (summer and w inter). In the analysis, each hom e range for each animal and season is thus treated as one sample. M oose tend to show fidelity to their hom e range areas (C e derlund, S andegren & L arsson 1987, C ederlund & O karm a 1988) w hich was a possible source o f pseudo replication (H urlbert 1984) in our data, as some o f the individuals were follow ed for two or more seasons and the others for only one season. Therefore, although each m oose had to m ake a decision every year con cerning its habitat choice, w e took the conservative approach and checked w hether the hom e ranges of each m oose overlapped in the same season in consec utive years. If this was the case, only one o f the over lapping home range boundaries for each moose was ran dom ly selected for the analysis. As a result, our sam ple size was reduced to 33 sum m er (10 males and 23 fem ales) and 21 w inter hom e ranges (six males and 15 fem ales) w hich were used in the analyses.
To address the above-mentioned problem 2, we trans form ed all habitat proportions to log-ratios using one of the habitat classes as denom inator (A ebischer et al. 1993). Thus, given that there are D habitats available and an individual's proportional habitat use is described by x 1, x2, ... xD, where xi is the proportion o f used habi tat o f type i, the log-ratios are calculated as y i = ln(xi /x j) where i = 1,2,... D, i???j, and Xj is the proportion o f avail able habitat o f type j. In this way, all the log-ratios, yi, which have the same denom inator are rendered linear ly independent (Aebischer et al. 1993). W e used the class Other (see Table 1) as denominator in the log-ratio trans formations.
The logarithm ic transform ations require that all the habitat types are recorded for all observations, i.e. that there are no zero proportions for any habitat class. In the case o f zero proportions, Aebischer et al. (1993) pro posed a m ethod in w hich the zero proportions are sub stituted by a value of an order of magnitude smaller than the sm allest recorded non-zero percentage in the actu al data. This method assumes that all the habitats are avail able and that zero represents use w hich is so low that it can not be detected (Aebischer et al. 1993). Following this method, w e replaced zero proportions with 0.001 w hich is a value that is an order o f m agnitude less than the sm allest recorded non-zero percentage in our data.
To address the above-m entioned problem 4, we first com pared the habitat com positions o f hom e ranges with the overall landscape (Johnson's second-order selection). In order to measure the overall landscape habi tat com position in the study area, we used the original hom e range boundaries (N = 54) and placed them ran dom ly in the area. H ow ever, we used two types o f cri terion based on our telem etry data in the random plac ing procedure. The home range boundaries were allowed to be located only within the area used by our study ani m als, i.e. the area w as restricted by the outerm ost boundaries o f all the actual hom e ranges. As our study area extends to the Botnian coast (see Fig. 1) and as there are several large settlements in the area, we also applied som e restrictions w hen random ly placing the hom e range boundaries within the study area. Moose have been reported to avoid m an-m ade landscapes (R olley & K eith 1980(R olley & K eith , Repo & L öyttyniem i 1985, and the ran dom ly placed hom e range boundaries were therefore restricted to include the class O ther (see Table 1) at a maximum o f 19.13%, human settlements at 11.75% and agricultural land at 36.32% . These values correspond to the m axim um values occurring in the actual home ranges.
W hen analysing w ithin-hom e range use (Johnson's third-order selection), w e com pared habitat com posi tions around the locations with the habitat compositions o f the hom e ranges. As the accuracy o f radio locations varies with, for instance, the distance to the located ani mal, topography and vegetation cover, we created a 200 m-wide buffer zone around each location and calculated the habitat class proportions from this area. In the anal ysis, we used only locations that fell within the bound aries o f the hom e ranges (i.e. for m ale sum m er hom e ranges 217 locations, for m ale w inter hom e ranges 73 locations, for female summer home ranges 628 locations and for fem ale w inter hom e ranges 186 locations).
As we had few er hom e ranges than dependent vari ables for some subgroups, w e did not use param etrised m ultivariate analysis o f variance tests (Tabachnick & Fidell 1996), but used random isation in all the com parisons (see also Pendleton, Titus, Lowell, D egayner & Flatten (1998) about statistical testing of compositional data). Following the presentation of Manly (1997), ran domisation is based on the idea that the mean values of observed data for com pared groups and the difference D 1 between groups are first calculated. Values from both groups are then random ly allocated to either o f the groups, and the difference between the means o f groups is recalculated. Repeating the second step numerous tim es gives an estim ate o f the distribution D (random isation distribution) that occurs by random ly allocating the values actually observed to either o f the groups. Finally, a decision about the significance o f the test is m ade by calculating the proportion o f all the observed values in the random isation distribution D that are greater than or equal to D 1 w hich corresponds to the P-v a l u e . W e used 5,000 randomisations for home range vs over all landscape com parisons, as w ell as for com parisons between locations, and 10,000 randomisations for home range vs location com parisons. All the com parisons w ere started by first checking whether m oose showed differences in their habitat use among seasons, i.e. sum m er and winter. W e then tested w hether m ales and females showed any differences in their habitat use, and if this was the case for either o f the subgroups, further analysis was m ade separately for each group. Otherwise, the seasons or sexes were grouped. Before deciding whether to use univariate tests o f habitats or not, we made a multivariate analysis based on randomisation to check w hether the m ean differences among groups for a com bination o f all habitats were likely to have occurred by chance. D epending on the data, different test statistics may yield different results (Manly 1997), and we there fore used W ilk 's lambda, sum o f log(F) and sum of squares (E-statistics) as parallel m easures for m ulti variate com parisons.
B ecause o f the num erous com parisons betw een dif ferent subgroups (sex and season) within and among scales and the great num ber o f habitat classes, w e limit ed our detailed consideration o f the differences mainly to those w hich were statistically significant (P < 0.05) and at the same tim e show ed a reasonable enough dif ference in absolute proportions for them to be ecolo gically meaningful. W e also calculated a correlation m atrix for the original class-proportion values in order to get an im pression o f w hich habitat classes w ere in ter-correlated. W hen com bined, this inform ation allow s a better focus on the differences in habitat util isation w hich are also likely to be a general phenom e non in m oose habitat selection.

Characteristics of summer home ranges and habitat use within home ranges
In the m ultivariate com parison o f the sum m er and w in ter hom e range com positions, W ilk's A show ed a sig n ifican t difference betw een groups (0.74% o f ran domisations < original W ilk's ???), but the two other sta tistics did not (86.56% o f randomisations < original sum of log(F) and 26.40% for E statistics). Therefore, we ana lysed the hom e range level data of habitat compositions separately for sum m er and winter.
T he m ale and fem ale sum m er hom e range m ulti variate comparison did not show a significant difference betw een the sexes (45.12% o f random isations < orig inal W ilk 's ???, 71.90% for sum o f log(F) and 48.60% for E statistics), and therefore we pooled the data for fur ther analysis. In the multivariate com parison o f pooled m oose sum m er hom e range habitat com position to overall landscape, only W ilk 's ??? show ed a significant difference betw een groups (0.02% o f randomisations < original W ilk 's ???), but the tw o other statistics did not (23.56% of randomisations < original sum o f log(F) and 24.10% < original E -statistics). U nivariate com par isons o f habitats betw een hom e ranges and the overall landscape show that there was only one statistically sig nificant difference, i.e. non-pine dom inated thinning forests were more abundant in the home ranges ( Fig. 2A). Based on the tree species composition, and as there was a sim ilar trend for non-pine dom inated plantations, the hom e ranges o f moose in summ er contain slightly more fertile areas than occur on average in the overall land scape.  W ithin hom e ranges, fem ales used areas w ith m ore non-pine dominated plantations and all types of thinning forest (Fig. 3A). Pine-dom inated plantations on peatl a n d s / s h r u b land were significantly less frequent around the fem ale locations than expected on the basis o f the hom e range compositions. Contrary to the results o f the hom e range and overall landscape com parisons, female m oose use less pine-dom inated plantations on peatl a n d s / s h r u b land than expected on the basis o f the home range com positions.
In sum m er, m ales use areas w ithin the hom e ranges with significantly more mature forests and non-pine dom inated young forests (Fig. 3B) than expected on the basis of the home ranges habitat compositions. Males also tend to avoid hum an settlem ents and agricultural areas, but the difference is significant only for human settlements. A s for fem ales, within hom e ranges som e o f the habi tat classes were used close to the level or even less than expected, w hile in the hom e range vs overall landscape comparisons these habitats were used more than expect ed. Mature forest dominated areas are found significantly more frequently around male locations, and there are also trends towards pine-dominated thinning forest to be more abundant around m ale locations than expected on the basis o f the hom e range com positions.

Characteristics of winter home ranges and habi tat use within home ranges
The m ultivariate com parison o f m ale and fem ale w in ter hom e ranges did not show a significant difference betw een the sexes (42.70% o f random isations < orig inal W ilk 's ???, 42.62% for sum o f log(F) and 46.80% for E statistics), and the data for both sexes were pooled Table 2. Correlations between proportions of habitat classes in m oose home ranges (Spearm an's rho; N = 54). * indicates a correlation sig nificant at the 0.05 level, ** P < 0.01 and *** P < 0.001. for further analysis. M ultivariate com parison o f the pooled moose winter home ranges and overall landscape showed a significant difference or a trend for such a dif ference (0.02% o f random isations < original W ilk's ???, 7.20% for sum o f log(F) and 3.22% for E statistics). According to the univariate habitat comparisons, moose w inter hom e ranges included significantly more pine-d o m i n a t e d , young habitats than expected on the basis o f the overall landscape habitat com position (Fig. 2B). The role of pine-dominated plantations on peatland/shrub land is especially pronounced in m oose w inter ranges. Moose winter home ranges are also located in areas with significantly less human settlem ents and agricultural fields than occurred on average in the overall land scape. The fact that m oose w inter ranges are charac terised by pine-dom inated young successional habi tats is further supported by the significant negative correlations betw een these and non-pine dom inated habitats, m ature forests and settlem ents (Table 2).
Flabitat com positions around the w inter locations of fem ales and m ales w ere significantly different (3.12% of random isations < original W ilk's ???, 5.62% for sum o f log(F) and 3.41% for E statistics), and further com parisons were made separately for each sex. Multivariate com parisons o f habitat com positions around locations with the hom e ranges show ed a significant difference or a trend for both fem ales (0.01 % o f random isations < original W ilk 's ???, 4.03% for sum o f log(F) and 1.28% f o r E statistics) and males (0.01% o f random i sations < original W ilk's ???, 5.72% for sum of log(F) and 3.68% for E statistics).
There w ere significantly more non-pine dom inated plantations and thinning forest and less hum an settle ments and agricultural fields around the female locations than expected on the basis o f the habitat distributions o f the hom e ranges (Fig. 3C). Fem ales also tend to use less pine-dominated peatland habitats within their home ranges than expected, but the differences in absolute pro portions are small. For m ales, the only statistically sig nificant difference in forested habitats was found for non pine dom inated plantations w hich w ere used slightly m ore w ithin the hom e ranges than expected (Fig. 3D).

Comparison of habitat use between females and males
O ur data did not show differences in hom e range lev el habitat use between the sexes, but several statistically significant differences were found in within home range habitat use both for sum m er and w inter (Table 3). However, taking into account also home range level re sults and habitat use w ithin hom e ranges (see D is cussion), m ales and fem ales did not show drastic dif Table 3. Differences in within home range habitat use between male and fem ale m oose within seasons and between winter and summ er for each o f the sexes. P-values show the probability o f the random ised distribution o f group differences being the same as the observed dif ference between groups (5,000 randomisations). Randomisations were made for the log-transform ed ratios o f Class;/Classj (i???j) using the class Other as a denom inator. See Table 1 for a description of the habitat classes. ferences in within home range habitat use in neither sum m er nor winter. In summer, males are located somewhat more in pine-dominated plantations on mineral soils and females more in pine-dominated thinning forests on peat lands. In winter, males are found m ore in pine-dom i nated forests on mineral soils than females.

Comparison of moose habitat use in winter and summer
As suggested by the contradictory results obtained in the different m easures o f significance in the m ultivariate com parison o f the sum m er and w inter hom e range compositions, there were only a few differences between seasons. In the univariate com parisons, the only sig nificant differences were that there are less hum an set tlem ents and agricultural fields in the w inter hom e ranges than in sum m er hom e ranges (P < 0.001 for both sexes). The negative correlations between the pro portions o f habitat characteristics o f w inter ranges and settlem ent areas (see Table 2) further suggest that w in ter ranges are located further from settlem ents than sum m er ranges.
M ultivariate com parisons between the habitat distri butions around w inter and sum m er locations show ed a significant difference for both sexes (for females 0.01% < original W ilk 's ???, 8.59% for sum o f log(F) and 2.99% fo r E statistics; for m ales 0.01% < original W ilk's ???, 2.64% for sum o f log(F) and 0.03% for E sta tistics). C om pared to winter, in summer, female moose are located more often in areas with more mature forests and pine-dominated thinning forests on mineral soils (see T able 3). There are also less agricultural areas around female winter locations than around female summer loca tions. For males, there were more significant differences in habitat com positions around sum m er and w inter locations than for fem ales (see Table 3), and the dif ferences are also som ew hat larger in absolute propor tions. Considering also the hom e range vs overall land scape com parisons, all types o f pine-dom inated habi tats are found m ore frequently around male locations in w inter than in summ er. N on-pine dom inated, young forests are im portant during both seasons fo r both sexes.

Summer home ranges and habitat use within the home range
Relatively few studies in Fennoscandia have addressed the habitat selection o f m oose in sum m er (Cederlund & O karm a 1988, H jeljord et al. 1990, Bø & Hjeljord 1991, Heikkilä et al. 1996. Furthermore, keeping in mind the hierarchical nature o f the resource selection o f ani mals (Johnson 1980), m ost o f the m oose radio telem e try studies have been based on location vs available habi tat com parisons within hom e ranges or the landscape in general, and the characteristics o f entire hom e ranges have seldom ly been assessed (Cederlund & Okarm a 1988). As the criteria between habitat selection on a dif ferent scale may vary, the direct com parison of differ ent scale studies should be m ade with caution and, for this reason, the discussion o f the hom e range level re sults is kept at a rather general level.
A ccording to our results, the habitat com positions of sum m er hom e ranges lie relatively close to that o f the overall landscape w hich suggests that m oose are able to utilize a variety o f habitats in summer instead of being strictly adapted to certain types o f habitat (Hjeljord et al. 1990). There were significantly more only non-pine dom inated thinning forests in the sum m er ranges, and a sim ilar trend was also found for non-pine dom inated plantations and pine-dom inated plantations on mineral soils. The im portance o f non-pine dom inated forests in sum m er was further supported by the fact that, within the hom e ranges, m oose used areas w ith significantly m ore non-pine dom inated forests. In our habitat classi fication, non-pine dom inated forests com prise all the com binations from mixed tree species forests to pure deciduous or spruce dom inated forests. In the study area, pine is the predominant species in about 75% o f the forests, and the remaining forests are either spruce-dom inated (~13% ) or birch-dom inated (~10%) or mixed forests (Statistical Yearbook of Forestry 1996). As a large proportion of deciduous trees or spruce in the forests usu ally indicates m ore fertile soils, our data suggest that, at the summer home range level, moose favour areas with m ore fertile habitats than occur on average in the over all landscape (Bergström & H jeljord 1987, Hjeljord et al. 1990. M ature forests have been considered im portant dur ing snowfree periods due to the presence o f food plants in the dw arf shrub layer (Markgren 1974), and especially in late summ er due to the delayed phenological changes in food plants (Hjeljord et al. 1990). On the other hand, C ederlund & O karm a (1988) found that, at the hom e range level, the utilisation o f m ature forests was close to the expected level. A ccording to our results, the pro portion o f m ature forests in the m oose sum m er hom e ranges did not differ from that in the overall landscape. W ithin hom e ranges, on the other hand, m oose are fre quently located in areas with significantly more mature forests than expected on the basis of the home range habi tat com position. This either implies that m ature forests do not play an im portant role in habitat selection at the hom e range level, or that the proportion o f m ature forests em bedded in forestry-modified landscapes dom inated by young successional stages is still above a threshold at which the summer range selection of moose is not affected.

Winter home ranges and habitat use within the home range
The qualitative and quantitative distribution of food in different habitats largely determ ine the activity pat terns and habitat use o f large ruminants. In winter, the m ajority o f the browse o f m oose is nutritionally o f low quality, mainly pine (Cederlund, Ljungqvist, Markgren & Stälfelt 1980), and moose spend less time feeding than in sum m er (Cederlund 1989). Therefore, moose should seek areas with relatively densely distributed feeding hab itats in w inter (Cederlund 1989). In our study, a large proportion of the habitats in the winter home ranges were pine-dom inated young successional stages. The role of pine-dom inated plantations in peatlands/shrub land was especially pronounced at the hom e range level. W ithin the hom e ranges, in contrast, m oose are more frequently located in non-pine dom inated habitats, and pine-dom inated habitats are used close to the expected level or even less. As a conclusion, our findings seem to support the idea that moose respond to the food quan tity at the hom e range level (Cederlund 1989, W allace, Turner, Rom m e, O 'Neill & W u 1995, but within the ranges they seek habitats with a mixture o f tree species in addition to pure pine-dom inated habitats also in w in ter (Hjeljord et al. 1990).
It has been suggested that, due to the increased moose population size, the increasing incidence o f young forests and over-brow sing, there has been a sw itch in habitat use, and therefore peatland habitats would not be as im portant as previously believed (Bergström & H jeljord 1987). In Finland, how ever, more peatlands have been found in areas with higher moose densities (Heikkilä & H ärkönen 1993), and the evidence from telem etry studies indicates that forested peatland hab itats are im portant for m oose all year around (Heikkilä et al. 1996). O ur results also indicate that forested peat land habitats are im portant determ inants o f m oose ranges, especially in winter. It has been hypothesised that drained peatlands w ould be especially suitable habitats for m oose due to the accelerated m obilisation o f nutrients affecting the secondary m etabolite pro duction and/or increase in the growth o f trees and oth er browse (Heikkilä & Härkönen 1993). The proportion of drained peatlands is high in our study area, and it is likely that drained peatlands com prise a part o f the forested peatlands also within the home ranges. However, no geographic inform ation about drained peatlands was available for our analysis, and the role of different types o f peatland habitats could not be further assessed.
According to Johnson (1980), the preference in selec tion can only be reflected if the resource com ponent is relatively scarce. Com ponents that are vita), but also abundant at the same tim e, m ight lead to the erroneous conclusion that these habitats are o f little value. Also, if som e habitat types are im portant at the hom e range level, the use o f these habitats within the hom e ranges may appear equal or lower when com pared to the avail ability w ithin hom e range, w hich could lead to the erroneous conclusion that these habitats are o f little value in within hom e range habitat selection (Thom as & T aylor 1990). This can partly explain the contradic tory results o f the im portance o f pine-dom inated peat land forests between home range level and within home range use. A ccording to the statistical tests, w ithin hom e ranges pine-dom inated peatland forests do not seem to be important or are even used less than expect ed. H ow ever, as the am ount o f forested peatland hab itats is high at the hom e range level, and the differences betw een hom e range and within hom e range use are small, it is more likely that peatland habitats are im por tant at both scales (for a more detailed discussion o f the interpretation o f results am ong scales see the section M oose habitat selection at the two levels o f scale).
In w inter, m oose have been reported to select areas w ith less hum an settlem en ts and ag ricu ltu ral land (Rolley & Keith 1980). According to moose damage in ventories, dam aged forest plantations have also been found further from settlem ents and roads (R epo & Löyttyniem i 1985, Heikkilä 1990, but see also Ball & D ahlgren 2002), and this has been interpreted to mean that m oose avoid hum an disturbance. H ow ever, as the summer home ranges did not show a similar pattern, the fact that w inter ranges include less settlem ents and agricultural fields probably reflects the more distant loca tion o f im portant w inter habitats from m an-m ade land scapes rather than disturbance. H um an settlements and agricultural fields are often located in m ore productive land originally consisting o f other than pine-dom inat ed habitats or peatlands. This is further supported by the significant negative correlations between both settlements and agricultural areas and the proportion o f im portant w inter habitats (see Table 2).

Habitat use between seasons
According to our data, the compositions of moose sum m er and w inter hom e ranges did not show a statistical difference for any type o f habitat other than agricultu ral fields and settlem ents, which w ere found less fre quently in w inter ranges than in sum m er ranges. In this respect, our results are in accordance with those of Cederlund & Okarma (1988) and Ball et al. (2001), who found no drastic differences between summ er and win ter habitats. A ccording to the overall landscape com parisons, how ever, there is a slight shift towards pinedom inated habitats in w inter (Bergström & H jeljord 1987). Furtherm ore, the difference in habitat com po sitions around sum m er and w inter locations seems to suggest that the change in m oose habitat use between seasons is more pronounced in within home range habi tat use than at the hom e range level.
The migration o f moose between winter and summer ranges has been related to the snow cover, and several studies have suggested that for w inter m oose migrate to areas with less snow (Coady 1974, Sandegren et al. 1985. In our study area, the average snow cover is rela tively thin (40-70 cm ) and perhaps not thick enough to substantially restrict m oose m ovem ent (Coady 1974). In the autum n, m oose mainly migrate from the coast towards the inland, and thus m ove at least in principle to areas with deeper snow cover than in areas closer to the coast. A lso, the relatively flat terrain and the short average migration distances between summ er and win ter hom e ranges do not seem to support the idea that migration is related to the snow depth gradient at least in the autumn. On the other hand, the moose winter home ranges w ere located in areas containing significantly m ore thinning forests than the overall landscape, and m oose also selected areas with more thinning forests within their hom e ranges. Thus, in addition to the avail able side-branch forage, at least part o f the importance of thinning forests could be related to snow cover depth or snow quality in within hom e range habitat selection (Ball et al. 2001). However, as there were no data avail able on the variation in the depth or the quality o f snow for the area, the possible m echanism s related to snow and migration from summer ranges to winter ranges can not be further assessed here.
In our study population, the migration distances were generally short, i.e. 15-25 km (Heikkinen 2000), which indicates that m igration does not reflect any substantial variation in any habitat characteristic. In spring, moose mainly m igrate tow ards the northw est, i.e. from the inland closer to the coast, although opposite directions have also been reported (Heikkinen 2000). The start of spring migration o f northern tem perate ungulates has been linked to snow m elt and the em ergence o f fresh green vegetation (LeResche 1974). It is thus possible, although it can not be assessed using our habitat data, that also in our study population the proxim ate reason for m igration from w inter to sum m er ranges is linked to the earlier em ergence o f vegetation close to the coast. On the other hand, there is evidence that the migratory traditions o f the moose can lag far behind envi ronmental changes (Andersen 1991), and considering the relatively short duration o f intensive changes in the for est landscapes due e.g. to forestry, it is possible that the migratory behaviour of moose still reflects earlier habi tat conditions.
It has been hypothesised that intensive forestry prac tices increase m oose philopatry by creating landscapes in which the different successional stages are evenly dis tributed over small distances and areas smaller than the home ranges o f moose (Cederlund & Sand 1992). Also, Ball et al. (2001) have reported m oose populations to be only partially m igratory, and they found no drastic differences in habitat com positions between migrants and non-migrants. In our study population, the summer and winter home ranges of the same individuals partially overlapped, especially in the case o f fem ales, and the migratory behaviour could thus resemble the behaviour described in Swedish studies. D ue to the low num ber o f observations in subsequent years, the effect o f dif ferent migratory behaviours on habitat use could not be com pared in our study. H ow ever, as both sum m er and winter home ranges differed significantly from the over all landscape, our data suggest that, at the scale of the moose home ranges (about 1,200 ha) at least, there is var iation in habitat com position that allow s a hom e range level selection. As the intensity and history of forestry and other land use practices varies among countries, and even betw een regions w ithin countries, the contradict ing results o f a num ber o f studies might merely reflect differences in the landscape structure and land use his tory betw een the areas studied.

Habitat use of male and female moose
Com parison o f the habitat use by the sexes (basically, the same applies for com paring seasons or other sub groups) is com plicated by the fact that statistical dif ferences betw een the sexes do not necessary indicate a preference for or an avoidance o f habitats by either of the sexes. Statistical differences betw een the sexes indicate different habitat distributions, but judgem ents about w hether sexes favour or avoid different habitat types require more inform ation about the habitat use of both sexes separately. To be ecologically meaningful, either hom e range level habitat use (hom e range vs overall landscape com parisons) or within hom e range hab itat use (hom e range vs location com parisons) should indicate the same type o f pattern for the same habitat types for either o f the sexes as found in between sex com parisons. This is illustrated by the sum m er use o f pine-dom inated peatland plantations/shrub land. Comparisons o f the home range to the overall landscape do not show preference or less use o f this type o f habi tat and neither does the within home range use o f males. Instead, fem ales tend to use significantly less pine-d o m i n a t e d peatland plantations/shrub land within their hom e ranges than expected. Still, a betw een sex com parison show s that pine-dom inated young peatland forests/shrub land are significantly more frequent around fem ale locations than around male locations. K eeping in m ind the results for the hom e range level and w ith in home range habitat use, we can conclude that this type o f peatland habitat does not seem to be im portant for either sex in summer.
Female and m ale moose are spatially segregated dur ing m ost o f the year, and their patterns of habitat use are also likely to be different due to their different nutritional requirem ents (Cederlund & Sand 1994). According to within hom e range habitat use, males and females used somewhat different habitat environments which suggest that the sexes are also spatially segregated in our study population. The difference is most clear in winter, when male m oose use pine-dom inated young forest habitats more than females. Because pine is the main browse spe cies for m oose in w inter (Cederlund et al. 1980), our result supports the hypothesis that, due to their larger body size, male moose should seek areas with good food availability even at the expense of the food quality (Ce derlund & Sand 1994). However, it should be kept in mind that the differences between the sexes found in our study are relatively small, and other possible factors, such as the effect o f offspring on habitat selection between the sexes, could not be assessed from our data.

Moose habitat selection at the two levels of scale
In general, our results support the hypothesis that ani mals select their habitats using different criteria at dif ferent hierarchical levels (Johnson 1980, Senft, Coughe nor, Bailey, Rittenhouse, Sala & Swift 1987. At the hom e range level (Johnson's second order selection), in winter especially, moose prefer areas with a high pro portion o f pine-dom inated young successional habi tats or thinning forests, but within their hom e ranges (third order selection), m oose select areas with a vari ety o f habitats. This is in agreement with Hjeljord et al. (1990), w ho concluded that a m osaic o f older and younger forests, rather than large areas o f young suc cessional stages, is important for moose. However, ac cording to our results, the younger successional stages, especially pine-dom inated habitats, seem to direct the habitat selection o f m oose at the hom e range level, i.e.
the am ount o f young successional stages has to be high enough to provide a suitable environm ent for moose. W hen the proportion o f younger successional stages is high enough at the home range level, the mixture o f oth er habitat types becom es im portant in w ithin hom e range habitat selection.
According to our results, the actual differences in habi tat distributions between home ranges and those around the locations w ithin the hom e ranges are relatively small in both summer and winter, but the significant dif ferences in univariate com parisons still indicate that m oose do select habitats within their hom e ranges. This is in contrast with H jeljord et al. (1990), who re ported that m oose m ovem ents and habitat selection within sum m er hom e ranges were random in relation to the food resource distribution in the area. Also, W al lace et al. (1995) suggested that ungulates feed randomly within forage patches (at a fine scale), but they respond to forage abundance at a larger landscape scale. There are at least two possible explanations for the discrep ancies obtained in the various studies. The first one is related to the definition o f available habitat (Johnson 1980(Johnson , A ebisher et al. 1993. The general method for defining available habitats has been to use m ore or less arbitrary study area boundaries. If, however, the habi tat selection o f the animal also happens at the home range level, i.e. the habitat com position is different from the habitat com position o f the study area, it can affect the significance o f the statistical tests and thus the detec tion o f habitat preferences at different scales. Second, when interpreting the differences in habitats favoured at the hom e range scale, which are used equal ly much or even less within the home ranges than expect ed by the overall hom e range habitat distribution, it should be kept in m ind that the preference may be con ditional on availability (M ysterud & Ims 1998). In such a situation, the use o f some habitat is not directly pro p o rtio n al to the availability o f that habitat, but changes as the am ount o f that particular type o f habi tat increases or decreases. As a consequence, the inter pretation o f preference for or less use o f habitats w ith in hom e ranges m ight be obscured if e.g. som e thresh old value for the am ount o f favoured habitat is reached or exceeded at the hom e range level. In our data, this m ight at least partly explain the differences found be tween the hom e range level and the within home range use o f habitats that are found in excess at the home range level, such as e.g. pine-dom inated peatland habitats. M ysterud & Ims (1998) proposed an approach to test for a change in relative use with changing availability o f habitats w hen two habitat categories are considered. So far, tests for m ultiple habitat situations, like in our data, have not yet been developed (M ysterud & Ims 1998), and the possible effect o f changing availability on the change in habitat use o f moose remains to be assessed in future studies.

Methodological aspects
Com positional analysis is based merely on the ratios of habitat proportions, and it is the m easure o f only one landscape structure component, i.e. the composition. As the same am ount o f a specific habitat can be spatially distributed over the same area in several patterns, m ea sures o f the spatial arrangem ent o f different habitat classes, i.e. the landscape configuration, could have potential in explaining the habitat selection o f animals (M ysterud & Ims 1998). Although the spatial distribu tion of moose locations within the home ranges partially addresses the question of the location distribution of dif ferent resources within home ranges, it does not answer these questions at the hom e range level. The location data used in our study were relatively few over time, and the inform ation about the quality o f the observation (m oose activity) was limited. Constructing the habitat requirem ents o f m oose from the landscape level to a single habitat patch requires more intensive (in time) and accurate (location) tracking of moose habitat use. Global positioning system (GPS) equipped radio collars (Lynch & Shum aker 1995), together with autom atic data col lection o f the activity of moose, are potential tools for providing such data. Linking more accurate data about m oose m ovem ents with landscape structure indices, such as patch density, patch size distribution and inter patch distances, could provide more insight into the pro cesses affecting habitat selection criteria at different scales (M ysterud & Ims 1998). However, studies that utilise spatially and tem porally more accurate telem e try data probably also require more accurate data o f hab itats in addition to data derived from e.g. satellite images (Saveraid, Debinski, K indscher & Jakubauskas 2001).
Finally, as our sample sizes for seasonal ranges (par ticularly for males: six w inter and 10 sum m er ranges) were not large, and as selection may only be detectable if im portant habitat com ponents are not excessively present (Johnson 1980), we recom m end and look for w ard to carrying out further studies using com posi tional analysis in other geographical areas.