Blackwell Science, LtdOxford, UKEREEcological Research1440-17032004 Ecological Society of Japan2004192169178Original ArticleSeasonal migration of sika deerH. Igota et al.
Ecological Research (2004) 19: 169–178
Seasonal migration patterns of female sika deer in eastern Hokkaido, Japan Hiromasa IGOTA,1* Mayumi SAKURAGI,1 Hiroyuki UNO,2 Koichi KAJI,3 Masami KANEKO,3 Rika AKAMATSU4 and Koji MAEKAWA1 1
Laboratory of Boreal Forest Conservation, Field Science Center for Northern Biosphere, Hokkaido University, N9 W9 Kita-ku, Sapporo 060-0809, Japan, 2Eastern Hokkaido Wildlife Research Station, Hokkaido Institute of Environmental Sciences, 2-2-54 Urami, Kushiro 085-0835, Japan, 3Hokkaido Institute of Environmental Sciences, N19 W12 Kita-ku, Sapporo 060-0819, Japan and 4EnVision, N9 W4 Kita-ku, Sapporo 060–0809, Japan
Fifty-seven female sika deer (Cervus nippon yesoensis), captured at the wintering area in the Shiranuka Hills in eastern Hokkaido, Japan, were radio-tracked during 1997–2001 to examine the factors affecting seasonal migration at the individual-landscape level. Ten of the 57 deers migrated between low-altitude summer home ranges and intermediate-altitude winter home ranges (the upward migrants). Twenty-nine migrated between high-altitude summer home ranges and intermediate-altitude winter home ranges (the downward migrants). Twelve used the intermediate-altitude home ranges all year round (the nonmigrants). The remaining six were unknown. The summer home ranges of deer were widely scattered over an area of 5734 km2. Migration distances ranged between 7.2 and 101.7 km. Deer showed high site fidelities to their seasonal home ranges. The upward migrants wintered in areas of less snow, higher quality of bamboo grass, and more coniferous cover than their summer home ranges. The downward migrants wintered in areas of less snow, higher quality of bamboo grass, higher winter temperature, and more southern slopes, but less coniferous cover than their summer home ranges. The non-migrants used yearround ranges with little snow, high quality of bamboo grass, and sufficient coniferous cover. We suggest that snow cover and bamboo grass are the factors affecting seasonal migration of the population and that coniferous cover is another factor for the upward migration. Key words: Cervus nippon; geographic information system; radio-telemetry; seasonal migration; sika deer.
Introduction Cervids that inhabit temperate and boreal regions often exhibit seasonal migration between the summer and the winter ranges (e.g. elk [Cervus elaphus canadensis]: Adams 1982; white-tailed deer [Odocoileus virginianus]: Nelson & Mech 1987; mule deer [O. hemionus hemionus]: Garrott et al. 1987; Nicholson et al. 1997; blacktailed deer [O. hemionus columbianus]: Loft et al. 1984; caribou [Rangifer tarandus]: Cumming & Beange 1987; moose [Alces alces]: Histøl & Hjeljord 1993). Seasonal migration is a regular, round-trip movement of individuals between seasonal home ranges (White & Garrott 1990). It may have evolved to avoid undesirable conditions at a particular time of year (Vaughan et al. 2000). Seasonal migration might have allowed cervids to disperse toward areas of high altitude or latitude, where the environment seasonally changes drastically and becomes worse, especially in winter. In boreal mountainous regions, the activities of deer are physically
restricted by snow in winter (Verme 1973; McCullough 1985). Access to forage beneath the snow becomes difficult (Parker et al. 1984; Takatsuki 1992) and energy expenditure increases at low temperatures (Silver et al. 1971). These disadvantages may affect reproductive success and even cause higher mortality in years of heavy snow (Kaji et al. 1988; Takatsuki et al. 1994; Uno et al. 1998). Deer, therefore, may aggregate in limited refuges (i.e. wintering area) with less snow and warmer temperatures than in more widespread summer ranges where females give birth and nurse fawns (Adams 1982). Many studies have reported that deer descend to areas of lower elevation in winter and return to summer ranges of higher elevation in spring (i.e. altitudinal
*Author to whom correspondence should be addressed. Email: [email protected]
Received 9 April 2003, Accepted 4 September 2003.
170 H. Igota et al.
migration, Loft et al. 1984; McCullough 1985; Schoen & Kirchhoff 1985; Garrott et al. 1987). Sika deer in eastern Hokkaido also exhibit such downward migration (Uno & Kaji 2000; Sakuragi et al. 2003a). However, some individuals in the population migrate between low-altitude summer ranges and intermediatealtitude winter ranges (Sakuragi et al. 2003a). This reverse altitudinal migration (upward migration) has never been reported in other cervid populations. Moving toward areas of higher altitude in winter seems unreasonable, because the snow deepens and temperature decreases as the altitude increases. In order to examine the reasons why the reverse altitudinal migration exists together with other migratory type in the population, we need to compare the environmental characteristics of the seasonal home range between summer and winter for each migratory type based on the individual-landscape level. Although several studies have compared the characteristics of the seasonal ranges at the population level (Sakuragi et al. 2003a), few studies have examined the individual decisionmaking to migrate or not (Ball et al. 2001). Ball et al. (2001) compared the snow condition and the composition of the vegetation of the seasonal home range of individual migratory moose between summer and winter, but detected no differences. The annual variations of weather conditions, such as snow depth, may conceal the potential differences of the characteristics of the seasonal home ranges between summer and winter in a given year (Ball et al. 2001). Sika deer in Hokkaido became endangered due to over-harvesting and record snowfalls in the 1870s and their distribution was restricted to several small pockets (Nagata et al. 1998; Kaji et al. 2000). The population, thereafter, gradually recovered due to the legislated protection and their distribution has expanded rapidly over the entire area of eastern Hokkaido between 1950s and 1980s (Kaji et al. 2000). The present patterns of sea-
Ecological Research (2004) 19: 169–178
sonal migration of the population are considered to have been developed during the process of its expanding distribution. Therefore, in order to assess the present patterns, including the reverse altitudinal migration, we need to compare the environmental characteristics of the seasonal home ranges with the trend in the decades. The main aims of the present study were: (i) to describe the patterns of seasonal migration of sika deer in eastern Hokkaido and (ii) to examine the factors affecting seasonal migration of the population with a focus on the reverse altitudinal migration at the individual-landscape level.
Methods Study area The study area was approximately 7500 km2 in eastern Hokkaido, Japan. The capture site (43∞13¢N, 143∞53¢E, Fig. 1) was located along the upper stream (200-m elevation) of the Shoro River in the Shiranuka Hills, where there is a representative wintering area of sika deer in eastern Hokkaido (Hokkaido Institute of Environmental Sciences 1995). The mean annual precipitation was 1399 mm and the mean annual temperature at the capture site was 5.2∞C with monthly means of -8.2∞C in February and 18.6∞C in August (National Land Agency of Japan 1992). During the study years, the mean temperature in Naka-teshibetsu in February near the wintering area varied between -7.9 and -12.2∞C, while the value of normal years was -7.6∞C. Snow cover persists from December to late March or late April there. Mean snow depth in February in the wintering area in the study years varied between 34 and 80 cm. These were 58–138% of the value of an average year (58 cm). The major vegetation of the whole study area is mixed forest with evergreen conifers and
Fig. 1. Distribution of individual summer home ranges (n = 51) of radio-collared sika deer in eastern Hokkaido, Japan, 1997–2001. () The summer home ranges of the downward migrants (n = 29), () the summer home ranges of the upward migrants (n = 10), () the summer home ranges of non-migrants (n = 12). Arrow indicates the capture site in the wintering area in the Shiranuka Hills.
Ecological Research (2004) 19: 169–178
deciduous broad-leaved trees typical of the transitional areas between temperate and subarctic zones (Tatewaki 1958). The dominant tree species are Acer mono, Tilia japonica, Quercus crispula, and Abies sachalinensis, with undergrowths of Sasa senanensis (Ss), S. nipponica (Sn), S. kurilensis (Sk), and S. borealis (Sb) (Toyooka et al. 1983). The areas higher than 700 m are covered by evergreen coniferous forests of Abies sachalinensis, Picea jezoensis, and P. glehnii. The areas lower than 200 m are covered by deciduous broad-leaved forests of Acer mono, T. japonica, Q. crispula, and Q. dentata. Plantation forests composed of Abies sachalinensis and P. glehnii are scattered in the mountainous area. The eastern part of the study area is flat lowland (< 200-m elevation), dominated by pastures. In this part, the riversides are covered by deciduous broad-leaved forests and plantation forests mainly composed of deciduous conifer, Larix leptolepis, are also scattered in some part. Capture and radio-tracking A total of 60 female deer were captured and equipped with radio-collars in April 1997, March 1998, March 1999, and March and April 2000 (Sakuragi et al. 2003a). The matriline of deer continue to use the same area over generations with minimal dispersal (Nelson & Mech 1999), suggesting that the movements of females reflect a traditional migration patterns in a population. Therefore, we focused on the patterns of female migration. Three of the 60 deer captured died just before tracking commenced. The remaining 57 deer were categorized as fawns (n = 11), yearlings (n = 6), and adults (n = 40)(≥ 2years old) according to teeth condition (Ohtaishi 1980). We collected 7333 location points between April 1997 and May 2001. Monitored duration for each individual averaged 656 ± 57 (SE) days (range = 20–1511). Radio-collared deer were relocated by ground triangulation (White & Garrott 1990) or visual observations. The inherent error of triangulation was estimated at 196 m (Sakuragi et al. 2003a). Locations were plotted on a 1 : 25 000 map in the field and inputted onto a GIS. Tracking was carried out within every 7 days for each deer. When deer were missed on the ground, aerial tracking was conducted by airplane following the methods of Mech (1983). Migration patterns We categorized radio-collared deer into three types (the upward migrants, the downward migrants, and the non-migrants) depending on the migration pattern. We considered a deer ‘migrant’ if the seasonal home ranges did not overlap, and ‘non-migrant (NM)’ if the seasonal home ranges overlapped. The upward migrant (UM)
Seasonal migration of sika deer 171
had the summer home range lower than or at similar altitude as its winter home range (i.e. the reverse altitudinal migration), while the downward migrant (DM) had the summer home range higher than its winter home range. We defined the seasonal (summer or winter) home ranges for individual migrants as those areas that demonstrated stabilized use during each season except migration periods. We based the winter and summer home ranges of the non-migrants on May– October and November–April locations, respectively. We determined migration directions as the bearings from the capture site to the summer home ranges and categorized them into ‘north’ (northwest–northeast), ‘east’ (northeast–southeast), ‘south’ (southeast– southwest), or ‘west’ (southwest–northwest). We calculated the geographic centers of activity (COA) as the average Universal Transverse Mercator of locations of individual within each seasonal home range (Hayne 1949). We calculated straight-line distances between individual seasonal COA as the migration distances. Mann–Whitney U-tests were used to compare migration distances between UM and DM. We also calculated the seasonal home range sizes by the minimum convex polygon method for each deer located more than 30 times in a season. Site fidelities to the seasonal home ranges were assessed by whether individual summer or winter home ranges overlapped or not in successive years and by calculating the distances between individual COA in successive years for summer and winter. c2 tests were used to compare the proportions of individuals with overlapping and non-overlapping seasonal home ranges. Kruskal–Wallis tests were used to compare the distances between individual COA in successive years among the three migratory types for each season. Mann–Whitney U-tests were used to compare the distances between individual COA in successive years between summer and winter. Factors affecting seasonal migration The characteristics of the seasonal home ranges (elevation, snow depth, bamboo grass forage value, coniferous cover ratio, winter temperature, and southern slope ratio) at the individual-landscape level were calculated for each ‘COA grid (a 1-km grid that included an individual seasonal COA). Elevation was determined as the mean elevation of 50-m grids within a COA grid (National Land Agency of Japan 1992). Snow depth was determined as the average during the 30-year period (1955–1984) of the mean snow depth in February in a COA grid (National Land Agency of Japan 1992). Bamboo grass species are important components in the winter diet of sika deer in eastern Hokkaido (Yokoyama et al. 2000). Bamboo grass forage value was
172 H. Igota et al.
Ecological Research (2004) 19: 169–178
determined by composition of the four bamboo species in a COA grid, following the method proposed by Kaji et al. (2000). Bamboo grass species were ranked in the order: Sn > Ss > Sb > Sk based on availability, grazing tolerance and nutritional value. The value (scores in parentheses) were none (0), Sk (1), Sk + Ss (2), Sb or Sn + Sb or Ss + Sb (3), Ss (4), Ss + Sn (5), Sn (6). Coniferous cover ratio (%) was determined by calculating the proportion of 30-m grids of coniferous forest or mixed forests in a COA grid (Environment Agency of Japan 1997). Winter temperature was determined as the average during the 30-year period (1955–1984) of the minimum temperature in February in a COA grid (National Land Agency of Japan 1992). Southern slope ratio (%) was determined by calculating the proportion of 50-m grids of slope aspect of 135–225∞ in a COA grid (National Land Agency of Japan 1992). Because the present patterns of seasonal migration of the population are considered to have been developed during the process of its expanding distribution between 1950s and 1980s after the bottleneck in 1870s (Kaji et al. 2000), we used the average of the 30-year period between 1955 and 1984 for the snow depth and the winter temperature. These environmental parameters were calculated as mean values during tracked years for each individual. Wilcoxon matched-pairs signed ranks tests were used to compare the paired values of the individual characteristics of the seasonal home ranges between summer and winter. Kruskal–Wallis tests were used to compare the characteristics of the seasonal home ranges among the three migratory types. If a Kruskal–Wallis test proved significant, multiple comparisons were carried out using the Dunn’s method.
Results Migration patterns Ten (18%) of 57 radio-collared deer were UM, 29 (51%) were DM, and 12 (21%) were the NM (Table 1). The remaining six (11%) indicated migratory behaviors but died or were missed due to radio-collar difficulties before their summer home ranges were determined.
UM and DM migrated from the winter home ranges to the summer home ranges between March and April and migrated from the summer home ranges to the winter home ranges between October and January. The summer home ranges of UM and DM were widely scattered over a total area of 5734 km2 (Fig. 1), while the winter home ranges were concentrated in an area of 821 km2. Migration direction of most UM (90%) was east, while most DM (86%) moved north (Table 1). Migration distance of all migrants averaged 35.1 ± 3.6 km (mean ± SE, n = 39, range = 7.2–101.7 km). Upward migrants migrated significantly greater distances than did the DM (U = 43.0, Z = -3.28, P = 0.001, Table 1). Mean seasonal home range sizes of the three migratory types varied from 66 to 125 ha in summer and 74–197 ha in winter (Table 1). We monitored 33 deer in two or more summers and 24 deer in two or more winters. As the proportion of individuals with overlapping and non-overlapping seasonal home ranges did not differ among the three migratory types in summer (d.f. = 2, c2 = 0.76, P = 0.68) or winter (d.f. = 2, c2 = 0.75, P = 0.69), they were pooled for each season. Thirty-two (97%) of 33 deer had overlapping summer home ranges in successive years. Seventeen (71%) of 24 deer had overlapping winter home ranges in successive years. The proportions of individuals with overlapping or non-overlapping seasonal home ranges differed between summer and winter (d.f. = 2, c2 = 7.87, P = 0.005). As the distance between individual COA in successive years did not differ among three migratory types in summer (d.f. = 2, H = 1.87, P = 0.39) or winter (d.f. = 2, H = 2.79, P = 0.25), they were pooled for each season. The distance between individual COA in two successive years in summer (mean ± SE = 814 ± 626 m, n = 58) were significantly shorter than in winter (mean ± SE = 2340 ± 768 m, n = 42; U = 840, Z = -2.64, P = 0.008). The winter home range characteristics and migration distances of the individuals with non-overlapping winter ranges (n = 7) did not vary among years (Friedman’s test, d.f. = 3, elevation: c2 = 3.20, P = 0.36; snow depth: c2 = 2.79, P = 0.42; winter temperature: c2 = 7.3, P = 0.06; coniferous cover ratio: c2 = 1.30, P = 0.73; southern slope ratio:
Table 1 Migration patterns of the upward migrant, the downward migrant and the non-migrant of sika deer in eastern Hokkaido, Japan. Migration distance (km) Migratory type
Mean ± SE
Upward migrants Downward migrants Non-migrants Unknown‡
10 29 12 6
East 9, West 1 North 25, East 4 – –
58.7 ± 8.9 27.0 ± 2.3 – –
18.4–101.7 7.2–53.4 – –
Summer home range size (ha)† Mean ± SE (n) Range
Winter home range size (ha)† Mean ± SE (n) Range
125 ± 25 (9) 123 ± 29 (19) 66 ± 4 (11) –
74 ± 19 (9) 107 ± 17 (14) 197 ± 36 (7) –
26–264 19–602 48–94 –
9–163 27–230 69–353 –
Seasonal home range sizes were calculated when deer located more than 30 times in a season. They showed migratory behavior but were not counted due to radiocollar failures or because they died before their summer ranges were found.
Seasonal migration of sika deer 173
Ecological Research (2004) 19: 169–178
c2 = 0.75, P = 0.86; migration distance: c2 = 1.70, P = 0.64).
with deeper snow than its summer home range. There was no significant difference in snow depth between the seasonal home ranges of NM (Table 2). The snow depths in the summer home ranges differed among the three migratory types (d.f. = 2, H = 32.84, P < 0.0001), and those of DM were significantly higher than those of NM or UM (P < 0.0167). The snow depths in the winter home ranges were similar among the three migratory types (d.f. = 2, H = 1.09, P = 0.58). According to the weather data of the Japan Meteorological Agency, the snow depths during November and May in Naka-Shibetsu in the eastern part were higher than those of Naka-Teshibetsu in the Shiranuka Hills (Wilcoxson signed ranks tests, Z = -2.20, P = 0.03), although the former (50 m) is in a lower elevation than the latter (80 m). The bamboo grass forage values in the summer home ranges of both UM and DM were significantly lower than in their winter home ranges, respectively (Table 2). Statistical comparison could not be done because the bamboo grass forage values of all seasonal home ranges of NM were the same (Table 2). The bamboo grass forage values in the summer home ranges differed among the three migratory types (d.f. = 2, H
Factors affecting seasonal migration The summer home ranges of UM were at significantly lower elevations than their winter home ranges, while DM had the summer home ranges at significantly higher elevations than their winter home ranges (Table 2). The summer and winter home ranges of NM did not vary in elevations (Table 2). The elevations in the summer home ranges differed among the three migratory types (d.f. = 2, H = 38.46, P < 0.0001), while the elevations of the winter home ranges were similar among the three migratory types (d.f. = 2, H = 3.53, P = 0.17). The snow depths in the summer home ranges of UM were significantly higher than in their winter home ranges (Table 2). Although only one UM migrated to a winter home range with deeper snow than its summer home range, the difference was only 1 cm. The snow depths in the summer home ranges of DM were significantly higher than in their winter home ranges (Table 2). No DM migrated to a winter home range
Table 2 Comparison of the characteristics (mean ± SE) of the seasonal home ranges of the upward migrant (UM, n = 10), the downward migrant (DM, n = 29), and the non-migrant (NM, n = 12) of sika deer in eastern Hokkaido, Japan Summer