We examined food items cached by arctic ground squirrels (Spermophilus parryii) from boreal-steppe meadows of southwest Yukon Territory, Canada. Caches recovered from two sites are dominated by fruits and seeds of either northern comandra (Geocaulon lividum) or prickly rose (Rosa acicularis). These two taxa are relatively rare in the local flora at the study sites (Site 1: ≥32 available taxa, and Site 2: ≥39 available taxa), suggesting they are selectively cached as preferred items. Cache selectivity may be related to perishability, fruit size/seed abundance, and predation risk. These caches are of significantly different composition than caches from present tundra sites and Pleistocene fossil arctic ground squirrel nests and caches (middens) recovered from central Yukon. These findings suggest that although arctic ground squirrels evolved in open tundra, they can subsist on a variety of cache items and may have the ability to adapt to and select a profitable cache within a variety of boreal and tundra habitats.
Northern environments are characterized by winters with limited forage access and low food abundance. To cope with predictable forage scarcity, northern herbivores have adapted various strategies, such as migration, hibernation, fat accumulation, and food storage (Marchand, 1996). Hibernating mammals enter a state of highly reduced body temperatures and metabolic activity, and rely on fat reserves when energy requirements cannot be met through foraging (Galster and Morrison, 1976; Barnes, 1996; Buck and Barnes, 1999a, 1999b). Food caching is a common strategy for many non-hibernating mammals living in the north, including red squirrels (Tamiasciurus hudsonicus; Hurly and Robertson, 1987), collared pikas (Ochotona collaris; Morrison et al., 2004), and voles (Microtus spp. and Clethrionomys spp.; Vander Wall, 1990). Fat and food caches enable animals to store reserves when food availability is high for use during periods when food availability is low (Vander Wall, 1990).
Arctic ground squirrels (Spermophilus parryii) inhabit tundra and boreal habitats of northern North America and have adopted a mixed hibernation and food cache strategy that differs between sexes (McLean and Towns, 1981; Buck and Barnes, 1999b). Female arctic ground squirrels are not known to cache food (McLean and Towns, 1981; Buck and Barnes, 1999b), an observation consistent with other species of ground dwelling sciurids (S. saturatus; Kenagy et al., 1989; S. richardsonii; Michener, 1993; S. columbianus; Shaw, 1926). Males enter hibernation in late August–early September and emerge in late March to early April (McLean and Towns, 1981). Males sequester themselves underground for a one- to two-week pre-emergence euthermic interval (Buck and Barnes, 1999b) where they feed on seeds cached during the previous summer and autumn (McLean and Towns, 1981; Gillis et al., 2005b). Cache consumption by males replaces the 30–48% body mass lost during hibernation (Galster and Morrison, 1976; Gillis, 2003; Buck and Barnes, 1999a) and facilitates post-torpor sexual maturation (Barnes, 1984, 1996; Barnes et al., 1987). The sexual differences in cache behavior are related to intense male-vs.-male competition for females during the 3–4 week spring mating season in which competitive advantage correlates positively with body size and foraging opportunities are limited (Carl, 1971; Green, 1977; Buck and Barnes, 1999b; Gillis, 2003). Thus, the size and quality of a male's cache plays an important role in its reproductive fitness (Barnes, 1996).
Few studies have documented arctic ground squirrel forage preferences for consumption or for caching (Table 1). S. parryii are considered generalist foragers that consume a wide variety of graminoids and forbs with some species taken preferentially, such as legumes (Batzli and Sobaski, 1980; McLean, 1985; Frid and Turkington, 2001). McLean and Towns (1981) observed increased seed foraging by males in late summer and autumn to accumulate food caches. Mayer (1953) documented a food cache dominated by bulbils (fruits) of alpine bistort (Polygonum viviparum) from a site near Point Barrow, Alaska. Krog (1954) documented a nest recovered near Kotzebue, Alaska, that contained discreet caches of green willow leaves (Salix sp.), spikes of wheatgrass (Agropyron latiglume), and capsules containing ripe seeds of rush (Juncus balticus). The most detailed study on cache selection is from an alpine tundra site in southwest Yukon Territory in which cached food was inferred from cheek pouches contents (Gillis et al., 2005b). At least 25 cached plant taxa were documented, dominated by Polygonum viviparum rhizomes and fruits, from within a local community of at least 100 available vascular plant taxa (Gillis et al., 2005b).
Vascular plant species recorded in caches of arctic ground squirrels. Plant parts: A = achene, B = rhizome, C = seed capsule, F = floret or flower, L = leaf, P = perigynia, S = seeds, SC = silicle, SP = spikelet, and SQ = silique.
We describe results from a preliminary study of Spermophilus parryii food caches from boreal-steppe meadows near Kluane Lake, southwestern Yukon Territory. We analyze cache contents from two sites and compare these with local vegetation to examine cache forage selectivity. Since limited work has been conducted on arctic ground squirrel cache foraging, the intent of this paper is to present these data, compare them with others, and hypothesize tentative explanations for cache forage selection. These are the first data on arctic ground squirrel cache contents from the boreal forest and add to the limited knowledge of foraging behavior for this herbivore. An objective of this study is to supplement ongoing investigation of Pleistocene arctic ground squirrel fossils nests and caches (middens) recovered from permafrost sediments in west-central Yukon (Zazula et al., 2005). These fossil middens have yielded diverse cache assemblages that include seeds, fruits, and leaves from at least 60 tundra and steppe taxa dating to the last glaciation, ca. 25,000 radiocarbon years ago (Zazula et al., 2005, in review). However, their efficacy for paleoecological reconstruction relies on some understanding of modern S. parryii foraging ecology.
Study Area and Methods
Arctic ground squirrel nests and caches were excavated by G. Zazula and R. Mathewes from two sites on the east side of Kluane Lake, southwest Yukon Territory on 10, 16, and 17 August 2004. The study area consists of vegetation dominated by white spruce forests (Picea glauca), with an understory of willow shrubs (Salix sp.), shrub birch (Betula glandulosa) and various forbs, and trembling aspen stands (Populus tremuloides) (Krebs et al., 2001). Grassland or steppe meadows form pockets of azonal vegetation within the boreal forest of southwest Yukon, and are restricted to dry hills or ridges and south-facing slopes (Laxton et al., 1996). The study was conducted in the latter part of the growing season when many of the graminoid and forb plant taxa were in fruit. The boreal-steppe in southwest Yukon was chosen for study because some aspects of arctic ground squirrel ecology have been studied previously in the region (McLean and Towns, 1981; McLean, 1985; Karels et al., 2000; Hubbs and Boonstra, 1997; Karels and Boonstra, 1999; Hik et al., 2001; Boonstra et al., 2001; Krebs et al., 2001). Furthermore, these steppe meadows are considered by many researchers to contain refugial vegetation that may be analogous to communities that were widespread in central Yukon and Alaska during Pleistocene glaciations (Laxton et al., 1996).
Field and Laboratory Methods
The study sites are meadows, 15 and 19 km north of the Alaska Highway on the Cultus Bay Road, respectively (61°08.361′N, 138°25.690′W, 813 m a.s.l.; and 61°09.401′N, 138°25.008′W, 799 m a.s.l.). The study sites were chosen because we noted physical signs of recent burrow use. Burrow systems were excavated using garden shovels by following burrows from their entrances. At each site, we defined vegetation types based on visually determined dominant plant cover and structure and all species found in that type were assigned a cover value using the Braun-Blanquet (1927) approach (Appendix A). Nests and caches were collected in the field, placed in sample bags, air dried, and examined in the laboratory with a dissecting microscope. Plant nomenclature follows Cody (2000).
Site 1 is a crescent-shaped meadow bordered by the Cultus Bay road to the east and spruce forest to the north, west, and south (Fig. 1). We observed freshly excavated soil surrounding at least 2 of 12 burrow entrances at the ground squirrel colony. Five vegetation types were recognized with 32 plant taxa recorded (Appendix A).
A hibernaculum was recovered approximately 75 cm below surface within silt, immediately above the underlying gravel (Fig. 2a). The hibernaculum consisted of a main spherical chamber (25 cm wide, 25 cm deep, 25 cm high) that housed a nest, and a small antechamber (10 cm wide, 10 cm deep, 10 cm high) with a food cache (Site 1, Cache 1). A side burrow next to the nest contained a latrine with wet feces. The grassy nest was built upon an old nest that contained moldy, wet material and decaying remains of eight juvenile arctic ground squirrel carcasses. A small food cache was also found associated with the moldy nest (Site 1, Cache 2). Several beetles were observed in the hibernacula. Further excavations within the colony failed to yield additional nests or caches.
The nest weighs 209 g and is composed primarily of graminoid foliage, with foxtail (Hordeum jubatum), wild rye (Elymus trachycaulus ssp. subsecunduns), reed bent grass (Calamagrostis purpurescens var. purpurescens), sedge (Carex filifolia), and prairie crocus (Pulsatilla ludoviciana), lichen, moss, and horse hair. Cache 1 contains seeds, fruits, and other remains from 12 vascular plant species and weighs 13.86 g (Appendix B). Geocaulon lividum berries dominate the cache, representing 92.7% of the total mass. Cache 2 weighs 5.52 g and is of similar composition, with 11 plant species dominated by Geocaulon lividum berries (60.1%). In total, Geocaulon lividum berries account for 83.4% of the total dry mass of cached food at Site 1 (Fig. 3).
Although the fresh nest was built on what may be a former female natal nest, we think the seeds and fruits (dominated by fresh Geocaulon lividum berries; Fig. 2a) represent a recent male cache because females are not known to cache seeds and fruits (Buck and Barnes, 1999a; Gillis et al., 2005b), and the study was conducted when females have typically already entered hibernation (McLean and Towns, 1981; McLean, 1985).
Site 2 is on a ca. 150-m-long northwest-southeast–trending ridge that slopes down about 20° to the south. Burrow entrances were observed across the ridge, but dense colonies were at the northwest end and near the middle (Fig. 4). Six vegetation types were recognized with 39 plant taxa recorded (Appendix A).
Several of the burrow entrances at Site 2 had small piles or scatters of partially eaten Rosa acicularis hips (fruits sensu lato), rose achenes (seeds sensu lato), and/or bastard-toadflax (Comandra umbellatus) berries (Fig. 2b). Excavation was attempted at several of these entrances but many tunnels ended abruptly, suggesting these were temporary duck holes (Carl, 1971). We partially excavated a colony near the middle of the study site, yielding one nest with cache and two other individual caches. The three caches were recovered in the burrow system within a ca. 3-m radius of each other (Fig. 4). Cache 1 was 25 cm below the surface within an oval chamber (25 cm wide, 25 cm deep, and 15 cm high) located alongside a main tunnel (Fig. 2c). Cache 1 consisted of two discreet piles; one with unopened rose fruits and the other dominated by rose seeds and milk-vetch (Astragalus williamsii) legumes, both resting on a small nest (Fig. 2d). Both Cache 2 and Cache 3 were found in small round chambers (50 cm and 40 cm below surface, respectively) along side main burrows.
The nest associated with Cache 1 weighs 29.5 g and is dominated by graminoid foliage, with some Populus tremuloides leaves, lichen, moss, deciduous woody twigs, nylon string and plastic wrap. Cache 1 is the most diverse with 17 vascular plant species and weighs 37.05 g (Appendix B). Rosa acicularis fruits (63.4%) and seeds (31%) dominate the mass of Cache 1. Although less diverse in composition, Cache 2 (17.96 g; 9 species, 8 families) and Cache 3 (30.23 g; 11 species, 8 families) were similar in that they are also dominated by rose hips (89.6% and 96.2%, respectively). Rosa acicularis fruits and seeds account for 95.5% of the combined mass of the three caches at Site 2 (Fig. 5).
Selection of Food Cache Items
Although our study is limited to caches recovered from two sites, these data suggest Geocaulon lividum and Rosa acicularis are preferred cache items for male arctic ground squirrels in the boreal-steppe meadows of southwest Yukon. Because these taxa were relatively rare in the local flora and there were abundant potential forage plants available (Site 1: ≥32 available taxa, and Site 2: ≥39 available taxa), the overwhelming dominance of these two cache items suggests high cache forage selectivity.
Because our data is limited, and little is known about arctic ground squirrel foraging behavior, our discussion of potential forage selection factors can only be considered tentative. Unfortunately, we did not obtain or are not aware of any nutritional data for Geocaulon lividum or Rosa acicularis fruits. Thus, we refer to some aspects of foraging theory to hypothesize several factors that may be important for cache selection at our study sites. In general, forage selectivity is influenced by intrinsic characteristics of food items, such as caloric value, specific nutrients, perishability, or size, and extrinsic factors of the forage, such as the spatial distribution of vegetation, predation risk, or social environment (Stephens and Krebs, 1986; Vander Wall, 1990).
For many food cachers, perishability is a key factor in a forager's assessment of food cacheability and determines whether an animal will immediately consume a food item or cache it for later (Eshelman and Jenkins, 1989; Vander Wall, 1990; Gendron and Reichmann, 1995; Hadj-Chikh et al., 1996; Kotler et al., 1999; Gerber et al., 2004). Consumption of perishable fleshy fruits such as Rosa acicularis and Geocaulon lividum may enable arctic ground squirrels to meet current water or nutrient requirements while the enclosed seed can be cached over-winter. No other available plants (with the exception of Arctostaphylos uva-ursi) at our sites produce fleshy fruits with cacheable seeds. Thus, perishability may offer a plausible explanation for the dominance of Geocaulon lividum and Rosa acicularis fruits and seeds in our caches. Cache preparation to increase storability (Vander Wall, 1990) is suggested by separate piles of Rosa acicularis fleshy fruits and seeds in Cache 1 at Site 2 (Figs. 2c, 2d) and scattered partial fruits and seeds near some tunnel entrances (Fig. 2b).
The size and/or quantity of cacheable seeds on a particular plant are important for cache foraging decisions because they directly affect feeding rates and handling times (Hurly and Robertson, 1987; Vander Wall, 1990; Garb et al., 2000; Gerber et al., 2004). Plants that produce abundant fruits and/or large cacheable food items minimize foraging cost and exposure to predators because sufficient forage can be obtained in one excursion without the need to travel between several plants or patches. Our site vegetation data suggest Geocaulon lividum and Rosa acicularis are the largest fruits with the largest or most abundant enclosed seeds available. Individual Rosa acicularis stems typically produce one large fruit which encloses many seeds (15–30), while Geocaulon lividum plants produce 2–4 individual berries that each contains a single large seed. Thus, fruit size and seed abundance may be factors affecting selectivity for cache items at our study sites.
Varying levels of risk and security from predators associated with different vegetation types should influence foraging decisions and cache contents (Ivins and Smith, 1983; Brown, 1988; Andrusiak and Harestad, 1989; Vasquez et al., 2002; Brown and Kotler, 2004). Because Geocaulon lividum and Rosa acicularis do not occur directly on the colonies at our sites, ground squirrels must accept greater foraging distances and exposure to predators to obtain these cache items (Figs. 1, 4; Appendix A). However, Geocaulon lividum and Rosa acicularis occur in vegetation types of higher stature (shrubby vegetation) and denser ground cover than steppe vegetation; hence these forage patches provide increased security cover and reduced predation risk. The possible reduction in predation risk while collecting Geocaulon lividum and Rosa acicularis may be an important factor for their dominance of caches contents.
Arctic Ground Squirrel Cache Foraging Adaptations
Cache contents at our two study sites and between those from previously published studies (e.g. Mayer, 1953; Krog, 1954; Gillis et al., 2005b), have markedly different contents. These results suggest that cache foraging strategies for arctic ground squirrels may be site or habitat specific based on local plant availability and distribution. The principal difference between our study and others is that we examined cache contents from low elevation boreal-steppe meadows, while others are from high latitude arctic tundra (Mayer, 1953; Krog, 1954) or high elevation alpine tundra (Gillis et al., 2005a, 2005b). Potential forage items differ substantially between tundra and boreal meadows. Because both Geocaulon lividum and Rosa acicularis are boreal plants that do not occur within tundra habitats, they are not potentially available forage items in those habitats (e.g. Mayer, 1953; Krog, 1954; Gillis et al., 2005b). Also, these plants probably did not occur within Pleistocene habitats in Beringia occupied by arctic ground squirrels (Zazula et al., 2005). Thus, modern caches excavated from the boreal-steppe meadows cannot be considered good analogues for preferred plant taxa cached by Pleistocene arctic ground squirrels in Beringia (Zazula et al., 2005; Zazula, unpublished data), although some selective factors (e.g., seed abundance) may apply during both periods. Furthermore, the similarity of dominant food items between fossil Pleistocene caches from Beringia (Zazula et al., 2005, in review) and modern tundra caches (Mayer, 1953; Gillis et al., 2005b) support other hypotheses (Hik et al., 2001; Gillis et al., 2005a) that suggest arctic ground squirrels evolved in and are adapted to open steppe-tundra habitats. However, because arctic ground squirrels can subsist on a variety of cache items in both tundra and boreal forest, they may have the ability to adapt to and select a profitable cache within a variety of northern habitats.
Funding to Zazula is provided by the Northern Scientific Training Program, the Geological Society of America, a Simon Fraser University Graduate Fellowship, and the Alberta Provincial Scholarship Program's James Lougheed Award of Distinction. Funding to Mathewes is provided by a Natural Sciences and Engineering Research Council of Canada operating grant. We thank Tracey Smith (University of Alberta) for assistance in the field. We thank Tim Karels and Brian Barnes for constructive reviews of our manuscript.
Cover (abundance) (after Braun-Blanquet, 1927) of vascular plant species recorded in each vegetation type at study sites near Kluane Lake. Numerical scale: first number designates cover value, second number designates measure of grouping. Cover value: * = sparsely or very sparsely present with cover very small; 1 = plentiful but of small cover value; 2 = very numerous or covering up to 5% of area; 3 = any number of individuals covering 25–50% of area; 4 = any number of individuals covering 50–75% of area; 5 = covering more than 75% of area. Grouping value: 1 = growing singly or isolated individuals; 2 = group or tufted; 3 = in small patches or cushions; 4 = in small cushions. ***Trees present but not quantified.
Vascular plant taxa recovered from caches near Kluane Lake. % represents percent of total mass of cache. Plant parts: A = achene, B = berry, C = seed capsule, CS = cone scale, F = floret or flower, H = hip (fleshy hypanthium or fruit), L = leaf, LG = legume, N = needle, S = seed, SC = silicle, and T = twig.