Seasonal habitat selection of the red deer ( Cervus elaphus alxaicus ) in the Helan Mountains , China

We studied the seasonal habitat selection of the red deer, Cervus elaphus alxaicus Bobrinskii & Flerov, 1935, in the Helan Mountains, China, from December 2007 to December 2008. Habitat selection varied widely by season. Seasonal movements between high and low elevations were attributed to changes in forage availability, alpine topography, the arid climate of the Helan Mountains, and potential competition with blue sheep, Pseudois nayaur (Hodgson, 1833). The use of vegetation types varied seasonally according to food availability and ambient temperature. Red deer used montane coniferous forest and alpine shrub and meadow zones distributed above 2,000 m and 3,000 m in summer, alpine shrub and meadows above 3,000 m in autumn, being restricted to lower elevation habitats in spring and winter. The winter habitat of C. elaphus alxaicus was dominated by Ulmus glaucescens Franch. and Juglans regia Linnaeus, deciduous trees, and differed from the habitats selected by other subspecies of red deer. Cervus elaphus alxaicus preferred habitats with abundant vegetation coverage to open habitats in winter, but the reverse pattern was observed in summer and autumn. Red deer preferred gentle slopes (<10°) but the use of slope gradient categories varied seasonally. Red deer avoidance of human disturbance in the Helan Mountains varied significantly by season. Information on red deer habitat selection can help understand the factors affecting seasonal movements and also support decision making in the management and conservation of red deer and their habitats.

Habitat selection by animals is considered as an optimization process that involves factors such as food supply, conspecific population density, body size, competitors, predators, and landforms (MORRISON et al. 1998).Information on which resources are preferred or avoided by organisms improves our understanding of how they meet their requirements for survival and reproduction (MANLY et al. 2002).The distribution and availability of trophic resources are important factors that affect habitat selection.In most temperate habitats, food is scarce during the winter months and abundant in spring and early summer (MOEN 1976, SCHMITZ 1991).This forces animals in temperate regions to adapt to seasonal changes in food supply.Ungulates that inhabit temperate and boreal regions often exhibit cyclical seasonal movements between summer and winter ranges in response to environment factors (e.g., snow conditions, food availability), social constraints and predation risk (FRYXELL & SINCLAIR 1988, PÉPIN et al. 2008).Regular, round-trip movements between seasonal home ranges (WHITE & GARROTT 1990) have evolved to enable animals to avoid undesirable conditions at a particular

Seasonal habitat selection of the red deer (Cervus elaphus alxaicus)
in the Helan Mountains, China Seasonal habitat selection of the red deer in the Helan Mountains ZOOLOGIA 30 (1): 24-34, February, 2013 time of the year (VAUGHAN et al. 2000).Generally, seasonal movement patterns of ungulates include short-distance movements, dispersal, and migration (GROVENBURG et al. 2009).Seasonal migration of cervids involves dispersal to areas of lower elevation, particularly in winter, when the environment is less hospitable at higher elevations (ALBON & LANGVATN 1992, IGOTA et al. 2004, PÉPIN et al. 2008).Mixed strategies of migration have, however, been found among cervids that inhabit temperate and boreal mountainous regions (IGOTA et al. 2004, BRINKMAN et al. 2005, GROVENBURG et al. 2009).
Despite numerous studies of other subspecies of red deer, little is known about the seasonal habitat use or movement patterns of C. e. alxaicus during different seasons.The objective of this study was to: 1) compare habitats used by C. e. alxaicus during different seasons to document differences in habitat selection; and 2) examine environmental variables that affect the seasonal movement of C. e. alxaicus.

MATERIAL AND METHODS
This study was conducted over four seasons from December 2007 to December 2008 in the Helan Mountain region, which is located between the eastern Yinchuan plain in Ningxia Hui Autonomous Region and the western Alxa Plateau in Inner Mongolia Autonomous Region (105°44'-106°42'E, 38°21'-39°22'N) (Fig. 1).The Helan Mountain region, located in northwestern China, is on the transitional zone between steppe and desert regions of central Asia (TAKHTAJAN 1986).It generally lies at 2,000-3,000 masl, with a maximum elevation of 3,556 masl.It covers an area of 2,740 km 2 [including Ningxia Helan Mountain National Nature Reserve (2,063 km 2 ) and Inner Mongolia Helan Mountain National Nature Reserve (677 km 2 , Fig. 1)], with a north-south length of about 250 km and an east-west width of about 20-40 km (Z.S. Liu unpublished data).
The region has a typical continental climate, characterized by cool and dry conditions, with annual mean temperature of -0.9°C and mean annual rainfall of 420 mm.The local climate is influenced by the topography of the Helan Mountain, the low temperature center of the northern Ningxia.The maximum monthly mean temperature is 11.9°C, in July, and the minimum is -14.2°C, in January, 8.8 ~9.8°C lower than in Yinchuan, the capital city of Ningxia Hui Autonomous Region.Precipitation varies seasonally, with 62% falling as rain in summer.There is little precipitation in winter, about 10.1% of the annual total.Snow cover is limited in the Helan Mountains (GENG & YANG 1990).The vegetation distribution is strongly influenced by moisture conditions.The elevation differential between the Helan Mountains and the plains to the south and east is about 2100 m, which creates an elavational climatic gradient that results in the formation of four elevational vegetation zones.The mountain steppe zone (MS) occurs at 1,400-1,600 masl and covers an area of 1,241 km  ZOOLOGIA 30 (1): 24-34, February, 2013 From December 2007 to December 2008, we carried out four surveys, one per season, to determine the distribution of C. e. alxaicus throughout the Helan Mountain area.Each seasonal survey sampled 32 line transects established along the valleys and each survey took about one month to sample all topographic types.Differences in topographic relief prevented walking along the transects at the same velocity and length, therefore, transects ranged in length from 4.5 to 8.5 km, for a total of 350 km traversing the whole study area and covering all four elevational vegetation zones.The distance between any two transects was at least 2 km, to ensure the independence of each transect.
Because of the rarity of C. e. alxaicus in the area and their high sensitivity to anthropogenic disturbance (JEPPESEN 1987), we documented habitat use mainly by recording fresh signs, such as evidence of bedding or the presence of feces.We also observed deer in order to record their activities.It was easy to differentiate the signs left by C. e. alxaicus from those of other ungulates (e.g., P. nayaur, M. chrysogaster) in the Helan Mountains, because C. e. alxaicus is larger than P. nayaur and M. chrysogaster, and the size and shape of their feces are different (CHANG & XIAO 1988).To ensure the accu-racy of data, we recorded fresh feces only (3-5 days old as estimated by the color and water content).When deer were observed during line transect surveys, we first used telescopes to observe their feeding or bedding behavior, without disturbing the animals.After deer had departed the area, we carried out detailed sampling.
We recorded terrestrial coordinates according to a Global Positioning System (GPS) after a feeding or bedding habitat was identified.We then established a plot with five sample quadrants (Fig. 2) to collect data of 18 topographic and biological variables (Table I) using the methods described by LIU et al. (2002).The distance between any two plots was at least 500 m to ensure the independence of each plot (Fig. 2).Generally, research on resource selection requires a comparison of the habitats used by C. e. alxaicus (observed plots) with those that are available (expected plots, assuming no differential selection by deer).To provide comparison plots for the analysis of habitat selection, 617 randomly located plots were surveyed.The random plots were established along the survey transects, in areas with no obvious evidence of C. e. alxaicus use.The distance between random plots and occupied plots or any two random plots was at least 500 m to reduce the possibility of overlap between used and unused plots.Comparison plots were surveyed in each vegetation zone according to the proportion of used plots in each zone.Data were recorded for comparison plots using the same methods used in the occupied plots.
Data were analyzed to quantify habitat selection by C. e. alxaicus by season.To assess seasonal differences between the 18 factors recorded at used plots and comparison plots, we used a chi-square goodness-of-fit test within classified categories for each variable (MARCUM & LOFTSGAARDEN 1980).P-values less than 0.05 were considered statistically significant.Bonferroni confidence intervals were calculated by the following formula to identify variables that indicate preference or avoidance, following the method developed by NEU et al. (1974)  C. e. alxaicus showed no obvious selection (marked " = ") for category i when <0 and > 0.
Data for non-numeric ecological factors (VT, TO, DT, SL and SA; see Table I) were examined with chi-square tests.Data for the remaining numeric ecological variables were initially analyzed with one-sample Kolmogorov-Smirnov tests to determine if they were normally distributed.The normally distributed data were analyzed with independent-samples t-tests, while the non-normally distributed data were analyzed with Kruskal-Wallis H tests.

RESULTS
A total of 602 plots used by C. e. alxaicus (observed plots) were recorded and compared among the four vegetation zones (Fig. 3).Across the whole study period, 209 used plots in the coniferous forest mountain zone were measured, and this type was the most common vegetation type, followed by open mountain forest and steppe zone of 169 plots, mountain steppe zone of 108 plots and alpine shrub and meadow zone of 106 plots (Fig. 3).
In summer, 146 plots were used by deer (Fig. 3) and 167 comparison plots (Appendix 1) were surveyed.Deer preferred habitats with gentle, undulating slopes (<20°) on the south side of the MCF zone above 2,000 m and the ABM zone above 3,000 m, respectively, during summer.Habitats used in summer were on lower slopes near dense stands of trees (> 4 tree/100 m 2 ) with mixed tree species of 4-6 m height, near dense shrub stands (> 10 tree/100 m 2 ) taller than 1.3 m, with high herb coverage (> 80%), good hiding condition (hiding coverage lower than 50%) (Fig. 4) and far from bare rock (> 50 m).Distance to human disturbance did not affect habitat use by deer in summer (Appendix 1).

DISCUSSION
Cervus elaphus alxaicus in the Helan Mountains displayed a pattern of seasonal elevational migration similar to that of other red deer subspecies in mountainous areas (ALBON & LANGVATN 1992, JARNEMO 2008, PÉPIN et al. 2008).However, we observed differences between the habitat selection of C. e. alxaicus and that of other subspecies.HUTTO (1985) described the mechanisms determining habitat selection as: geographic restrictions, genetic evolution, influence of experience, and settlement decisions following exploration.
Vegetation type determines the composition and distribution of deer forage, and is determined by soil type, climate, sunlight, topography, landform and many microhabitat factors.The heterogeneous distribution of biotic and abiotic fac-tors in environments leads to spatial heterogeneity in vegetation types.Physiological and ecological requirements of deer are met to varying degrees by different vegetation types.Thus the geographic and seasonal variation in vegetation types affect red deer habitat selection.In the Helan Mountains, C. e. alxaicus range annually from the mountain steppe zone below 1,600 m to alpine shrub and meadow zone above 3,000 m (LIU 2009).Deer preferred habitats in the montane coniferous forest zone (> 2,000 masl) and alpine shrub and meadow zone (> 3,000 masl), in summer, and those in the alpine shrub and meadow zone (> 3,000 m) in autumn.Deer showed no preference for any vegetation type in spring or winter (Appendix 1), using habitats in proportion to their availability during those seasons.IGOTA et al. (2004) reported that deer rarely change their summer home ranges for breeding and nursing of offspring.We predicted that some C. e. alxaicus might migrate down from the alpine shrub and meadow zone during winter and spring, while some might stay at high altitude.

4
amount of food resources declined.Food availability is clearly critical for the nutrition of ungulates, especially for fawns and females (PETTORELLI et al. 2005).Adaptation to varying feeding conditions throughout seasons or years was also confirmed by GROOT & HAZEBROEK (1995).
Habitat selection by deer is determined by the presence of both food and hiding cover (BORKOWSKI & UKALSKA 2008).The role of hiding cover in habitat use may be especially important in winter, when cervids reduce their food intake and survive, to a large extent, using their fat reserves (PUTMAN 1988).PEEK et al. (1982) studied the role of cover in habitat selection.Two cover types were recognized: thermal cover, in which a forest overstory protects against weather and sun, and hiding (security) cover used to escape and avoid predators and humans.C. e. alxaicus in the Helan Mountains preferred habitats near cover to open habitats in winter, but the reverse was found in summer and autumn (Fig. 4).Habitats dominated by U. glaucescens and Juglans regia Linnaeus were frequently used in winter (Appendix 1).C. e. alxaicus are likely to be under low predation risk except from poaching in the Helan Mountains.Therefore, selection of habitat by deer mainly reflects their food requirements and need for protection against severe weather.The habitat selection process presumably results from demands to maximize their energy efficiency while minimizing their movements when searching for food, water and cover (QIAO et al. 2006).It differed from other subspecies of red deer, such as Cervus elaphus xanthopygus Milne-Edwards, 1867 in northeast China and Cervus elaphus nelsoni (Bailey, 1935) in the Rocky Mountains, which preferred coniferous forest and sapling trees and shrubs, feeding on shrub and epicormics in winter (UNSWORTH et al. 1998).C. e. alxaicus preferred montane coniferous forest and alpine shrub and meadow zones in summer and autumn, but preferred mountain steppe and mountain open forest and steppe zones in winter and spring, similar to Cervus elaphus scoticus Lönnberg, 1906(WELCH et al. 1990).
The special alpine topography and arid climate in the Helan Mountain region were the two important factors explaining the habitat selection of C. e. alxaicus.Given that precipitation is low in the Helan Mountain region, especially in winter, and the snow coverage is also very low (GENG & YANG 1990), the snow depth and coverage are not important factors initiating the habitat selection and movement of C. e. alxaicus.The local population of blue sheep is widespread and numbers more than 10,000 (LIU 2009, LIU et al. 2007b).Blue sheep inhabit the mountain open forest and steppe zone, and probably compete with deer for forage, especially in winter and early spring, when food is scarce (LIU et al. 2007a).We recorded migration of C. e. alxaicus during winter and early spring to the mountain open forest and steppe zone from higher altitude areas (Appendix 1).A reasonable assumption is that blue sheep density, to a certain degree, has a negative influence on C. e. alxaicus density, which might also explain deer movements to higher el-evations in summer.Further study on the niche overlap between blue sheep and C. e. alxaicus is underway.AGER et al. (2003) reported sympatric populations of Rocky Mountain elk (C.e. nelsoni) and mule deer, Odocoileus hemionus (Rafinesque, 1817), in northeastern Oregon, where elk exhibited strong daily and seasonal patterns of movements and habitat use under competition from mule deer.
Topography and slope gradient were important factors affecting habitat selection by C. e. alxaicus.Generally, deer preferred gentle (<10°), undulating slopes (Appendix 1), but this varied seasonally according to food availability and temperature (Fig. 5).Deer selection of lower slopes was also caused by the predominantly steep topography of the Helan range, which includes only a small proportion of the total area as gentle slopes (DI 1986).C. e. alxaicus preferred sunny slopes with direct solar radiation, especially during the cold winter and spring (Fig. 5).
Pressure from predators and human disturbance are often considered to be important factors that influence ungulate behavior and habitat selection (BONENFANT et al. 2004, BORKOWSKI 2004).However, historically, the main predators of C. e. alxaicus, the snow leopard, Uncia uncia (Schreber,1775); the gray wolf, Canis lupus Linnaeus, 1758; and lynx, Lynx lynx (Linnaeus, 1758), became extinct during the 1980s in the Helan mountain region (WANG & SCHALLER 1996).Numerous studies have demonstrated that deer behavior and habitat use are mostly influenced by humans (BORKOWSKI & UKALSKA 2008), including hunting, other human activities (LICOPPE 2006), travel routes ZOOLOGIA 30 (1): 24-34, February, 2013 and road traffic (YOST & WRIGHT 2001). JIANG et al. (2007) reported that C. e. xanthopygus in northeastern China showed some behavioral plasticity in response to human influences, and the effect of habitat loss or fragmentation caused by human activities are expected to be great (JAEGER et al. 2005).However, in the Helan Mountain region, we found that the deer neither preferred nor avoided human disturbance (p > 0.05) (Appendix 1), and they profited from the programs of Forest-Grass Conservation Project and Shelterbelt Forestry Project and Returning Husbandry to Forestry (Grass) launched in 1996 (ZHAO et al. 2000) and 1999(ZHAO et al. 2004), respectively.Wild animal hunting and livestock grazing have been forbidden in most of the Helan Mountain region and habitats are generally wellpreserved.Therefore, C. e. alxaicus has not suffered much human disturbance, apart from regular ranger, activities whose impact is probably negligible.However, in our data, there was a significant amount of seasonal variation in deer avoidance of human disturbance in the Helan Mountains (Table II).The migration of C. e. alxaicus individuals to lower altitudes for foraging in winter and spring has brought deer closer to areas that are impacted by humans.Deer preferred habitats far from human disturbance at higher elevations during summer and autumn.During our survey, human-related C. e. alxaicus fatalities were low compared to those of other studies (LICOPPE 2006, PÉPIN et al. 2008); therefore, limited persecution and food availability may have facilitated and encouraged C. e. alxaicus individuals to use human-impacted areas in winter and spring.
Many studies on ungulate seasonal movement and habitat selection are based on data provided by GPS-collars or radio-collars, which are very precise.However, because of the difficulties involved in capturing C. e. alxaicus and the rugged topographic conditions in the Helan Mountains, we conducted our surveys by comparing use versus availability of habitats to quantify the seasonal habitat selection of C. e. alxaicus.This study was based on several years of observation and one year of data collection, but the biodiversity and natural conditions in the Helan Mountains have remained stable for years due to its continental climate.Understanding habitat selection by the red deer and its strategies of seasonal movement can be used for decision making on appropriate management and conservation measures (GUISAN & THUILLER 2005) for C. e. alxaicus and its habitat.The symbols (=), (+) and (-) located beside the observed values signify that those values were found to be (according to Bonferroni Intervals) in equal, greater, or lesser proportion than the respective expected values.(+) indicates that red deer preferred category i; (-) indicates that red deer avoided category i; (-) indicates that red deer showed no obvious selection according to category i. Difference between used and comparison plots is statistically significant at p< 0.05 based on a chi-square goodness-of-fit test.

Figure 1 .
Figure 1.Location and distribution of the study area and transects in the Helan Mountain region, China.Grey shading represents Helan Mountain Nature Reserve in Inner Mongolia Autonomous Region, while the unshaded area represents the reserve in Ningxia Hui Autonomous Region.

Figure 2 .
Figure 2. Diagramatic presentation of the survey transects, plots and sample quadrats used in this study.
(MS); Mountain open forest and steppe zone (MOFS); Mountain coniferous forest zone (MCF); Alpine bush and meadow zone (ABM) VT Topography Categorized by the slope and fault of a hillside, divided into 5 levels: Smooth undulating slope; Moderately broken slope; Distinctly broken slope; Scree/landslide; Cliff TO Dominant tree The tree covers 70% of the density in the 10×10 m plot.It usually was Ulmus glaucescens, Ziziphus jujube, Salix spp., Juniperus rigida, Pinus tabulaeformis, Picea crassifolia Mixture or Open land with no tree DT Tree density (trees/100 m 2 ) The total number of trees in the 10×10 m plot TD Tree height (m) The mean height of trees in the 10×10 m plot TH Distance to the nearest tree (m) Distance from the center of the 10×10 m plot to the nearest tree DtT Shrub density (trees/100 m 2 ) The number of shrubs in the 10×10 m plot SD Shrub height (m) The mean height of shrubs in the 10×10 m plot SH Distance to the nearest shrub (m) Distance from the center of the 10×10 m plot to the nearest shrub DtS Herb coverage (%) The mean herb coverage of the 5 sample quadrats in the 10×10 m plot HC Slope gradient (°) Slope gradient of the hillside where the spot located measured with military compass SG Slope location A visual assessment of the site location relative to the macroslope which is usually from valley bottom to ridge top, classed as: lower slope (includes valley bottom and flat), middle slope and upper slope (includes ridge top) SL Slope aspect Aspect was surveyed to eight compass points, translated as 0°, 45°, 90°, 135°, 180°, 225°, 270° and 315° from North, as 0° is equivalent to 360°.And the slope aspect was grouped into 3 main directions: sunny slope (135°~225°), partial shade slope (45°~135° and 225°~315°) or shady slope (315°~45°) SA Distance to water resource (m) The distance from the spot to the nearest water resource DtW Distance to human disturbance (m) The distance from the spot to the nearest place of human activity such as highway, road and shelter forest station, etc DtH Distance to bare rock (m) The distance from the spot to the nearest bare rock DtR Hiding cover (%) The coverage of the hiding conditions.Percent hiding cover was determined by visually estimating the percent of a deer or a substitute (a 1 m stick) obscured at 30 m in the four cardinal directions (KUNKEL & PLETSCHER 2001) HiC

Figures 3
Figures 3-4.(3) Abundance and proportion of plots used by Cervus elaphus alxaicus in different vegetation type zones among four seasons, in the Helan Mountain region, China.(MS) Mountain steppe zone, (MOFS) Mountain open forest and steppe zone, (MCF) Mountain coniferous forest zone, (ABM) Alpine bush and meadow zone.(4) Seasonal changes in the usage ratio of hiding coverage on C. e. alxaicus in the Helan Mountain region, China.

Table I .
Variables collected in feeding and bedding habitat plots used by Cervus elaphus alxaicus in the Helan Mountain region, China.
Further study with GPS collars on C. e. alxaicus has been conducted to test the prediction.CUI et al. (2007) and CHANG et al. (2010) reported that U. glaucescens, Populus davidiana Dode, P. monglica, Potentilla spp., Graminoids (Stipa spp., Poa spp.), Caragana spp.were important in the winter diet of C. e. alxaicus.In summer, C. e. alxaicus ate 18 plant species of 11 families, which were mostly Salix microstachya var.bordensis (Nakai) C.F. Fang, P. davidiana, U. glaucescens and Agropyron cristatum (L.) Gaertn.which were mainly distributed in the mountainous open forest and steppe zone, ranging from 1,600 m to 2,000 m altitude, and C. e. alxaicus of migratory group migrated down from alpine shrub and meadow zone during winter and spring (Table II) as the

Table II .
Characteristics of 13 ecological factors in the habitats used by Cervus elaphus alxaicus during the year in the Helan Mountain region, China.Significant P-values: * p р 0.001.The proportion of 18 ecological factors in used (Obs) plots by Cervus elaphus alxaicus and comparison (Exp) plots during different seasons in the Helan Mountain region, China.Selection was determined according to the method of BYERS et al. (1984).Factors: (AL) Altitude, (VT) Vegetation types, (TO) Topography, (DT) Dominant tree, (TD) Tree density, (TH) Tree height, (DtT) Distance to the nearest tree, (SD) Shrub density, (SH) Shrub height, (DtS) Distance to the nearest shrub, (HC) Herb coverage, (SG) Slope gradient, (SL) Slope location, (SA) Slope aspect, (DtW) Distance to water resource, (DtH) Distance to human disturbance, (DtR) Distance to bare rock, (HiC) Hiding cover.