INFLUENCE OF HABITAT, SEX, AGE, AND DROUGHT ON THE DIET OF TOWNSEND'S GROUND SQUIRRELS
BEATRICE V AN HORNE, ROBERT L. SCHOOLEY. AND PETER B. SHARPE
Department of Biology, Colorado State University, Fort Collins, CO 80523
We assayed diets of Townsend's ground squirrels (Spennophilus townsendii) in a semi-arid Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 shrub steppe in Idaho by analyzing 1,432 fecal samples during 1991-1994. Dietary com position differed between adults and juveniles but not between males and females. Sand berg's bluegrass (Poa secunda), a perennial bunchgrass, was the most common item in diets. Diets varied among habitats that differed in plant cover. Diets also broadened pro gressively during the active season of this hibernating species (February-June) to include more shrub material, annuals, and other foods. A drought at the end of the active season in 1992 caused lower body mass and survival of individuals because there were fewer annual plants during the mid- and late-season periods and fewer seeds in the late-season than in non-drought years. Sandberg's bluegrass was consumed in higher proportions during the late-season of the drought year, but most of it was senescent with reduced levels of essential fatty acids and other nutrients. Among native shrubs, winterfat (Ceratoides lanata) was preferred over big sagebrush (Artemisia tridentata). Adults in habitats that lacked shrubs had a lower probability of being recaptured the year following the drought. We conclude that preferred native shrub species provided a relatively constant food source under drought conditions, compared with perennial grasses and annual plants. Perennial grasses were the most commonly eaten food, and habitats dominated by them provided a good food resource except under drought conditions. Low gennination rates of annual plants under drought conditions and their short period of succulence make them a poor food source, and habitats dominated by them are unlikely to sustain viable populations of ground squirrels.
Key words: Spennophilus townsend;;, diet, fecal analysis, food preference, hibernation. survival, drought. shrub steppe
Ecologists considering if food limits dis manipUlations relative to background levels tribution and abundance of herbivores often of the resource may greatly influence the find an apparent plentitude of food avail results. Some studies have been designed to able. Food limitation of herbivores, how ascertain the relative importance of nutri ever, may be a complex process that varies ents, calories, and water in the choice of spatially among habitats within a region foods by ground squirrels. Ritchie (1988) and temporally among and within years. found Columbian ground squirrels (Sper Nutrients in food may change in ways that mophilus columbianus) to be largely energy render food insufficient as a source of nu maximizers. Females that did not forage op trition. Insights into food preferences can be timally relative to an ideal proportion of gained from laboratory "cafeteria" studies, monocots and dicots produced fewer suc but preferences may be confounded by cessful offspring (Ritchie, 1990). Eshelman availabilities of alternate foods under field and Jenkins (1989) indicated that Belding's conditions. Experimental manipulation of ground squirrels (S. belding;) selected the food base, usually through addition of plants on the basis of both water and pro food, can produce insights about the role of tein content, consistent with Pulliam's food limitation, although the timing of such (1975) model of "partial preferences."
Journal of MamlTUl!ogv. 79(2):521-537. 1998 521 522 JOURNAL OF MAMMALOGY Vol. 79, No.2
In a 4-year study we followed population als should show food preferences that may dynamics of Townsend's ground squirrels manifest themselves differently among sex (Spermophilus townsendii-Hoffmann et es, age classes, and habitats. A major ob a1. (1993) included this population in S. jective of our research was to analyze how mollis on chromosomal grounds, but we diet and dietary preference vary with sex, maintained original nomenclature for con age, and habitat. sistency with Rickarfs (1987) review and An extreme drought occurred near the with other publications from this study) in end of the second breeding season during Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 the Snake River Birds of Prey National our study, providing us with a natural ex Conservation Area (Birds of Prey Area, periment to investigate responses of vege hereafter). We expected that population tation and diets of ground squirrels to change would be linked closely to the drought. An additional a posteriori objec ground squirrels' food supply because they tive of our study was therefore to describe have a short active season of ca. 4 months, any habitat-specific differences in diet be followed by ca. 8 months in which they re tween drought and non-drought conditions. main underground and are largely torpid. Such variation in diet gave us additional in During the active season, both adults and sight into why populations declined during recently emerged juveniles generally dou drought and post-drought periods and if the ble their body mass prior to immergence pattern of decline varied among habitats. into torpor (Van Horne et aI., 1997), The diet is mostly herbaceous vegetation. a food METHODS source that is low in caloric content relative Study area.-The Birds of Prey Area (43°N, to animal matter or seeds. Adults emerge 1 16°W; 900-950 m above mean sea level) is his from the inactive seaSOn in late January or torically a shrub-steppe habitat, but it is increas early February, although female emergence ingly dominated by exotic annual grasses and may be delayed for 1-2 weeks when weath forbs, rendering it fire-prone when there is high er conditions are unsuitable. Young are net primary production followed by low precip born in early March and begin to emerge itation. Because those annuals inhibit re-estab from their natal burrows in early April. lishment of native shrub species, the area ap There is a short period during which both pears to be changing from shrub steppe to grass adults and juveniles are active above land (D. L. Yensen, in liU.). Sandberg's blue ground. Adult males immerge as early as grass (Poa secunda), a native perennial bunchgrass, is common in grasslands resulting late April. but adult females generally do from bums and in big sagebrush (Artemisia tri not immerge until after the first week of dentata) habitats, but less common in shrub hab May. Juveniles immerge during early- to itats dominated by winterfat (Ceratoides lanata). mid-June. Precipitation averages ca. 20 cm annually. Several approaches are needed to under Our study was structured in part to compare di stand factors influencing diet and the way ets of ground squirrels in native shrub habitats in which food affects demography. We have with those in grassland habitats. Native shrub investigated the role of fats in food. tem habitats included those dominated by winter poral and habitat-specific differences in fat-a low-stature shrub, big sagebrush-a taller body composition, and effects of experi shrub, and a mosaic of patches of winterfat and mental food supplementation (Van Home et big sagebrush. We studied ground squirrel populations on 20 al.. 1997). To gain further insight into the study sites of .:::;:9 ha each from 1991 to 1994. extent that food limits ground squirrels, we Ten of our study sites were in grassland habitats report the quantity and nutrient content of where fire had occurred 7-10 years previously. food from stomachs of ground squirrels. If Two of these had been reseeded artificially with food does indeed playa critical role in pop several species, including winterfat. Although ulation dynamics of this species, individu- coverage of winterfat was sparse, diets were af- May 1998 VAN HORNE ET AL.-DIETS OF TOWNSEND'S GROUND SQUIRRELS 523 fected sufficiently by reseeding that we chose to pinnata and D. sophia were not differentiated in analyze diets on these reseeded sites separately sampling, and cheatgrass-Bromus tectorum). from those in other grasslands. Six of our study Diet sampling.-We collected fecal samples sites were in big-sagebrush habitat, two were in from animals handled during regular trapping winterfat habitat, and two others were in sage near the time at which we sampled vegetation; brush-winterfat mosaic habitats. We livetrapped an early-season sample of adults collected be ground squirrels on each site throughout the ac tween 20 February-1O March, a mid-season tive season and uniquely marked each captured sample of both adults and juveniles collected be Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 animal with a passive integrated transponder tween 10-24 April, and a late-season sample of (Schooley et al.. 1993). juveniles collected between 18-31 May. Fecal Climatic pattems.-We obtained data from a samples were dried at 65°C for 48 h. Personnel weather station of the National Oceanic and At at the Compositional Analysis Laboratory at mospheric Administration located at Boise, Ida Fort Collins, Colorado, quantified diets of squir ho (Boise Airport Weather Service Forecast Of rels using a microhistological technique (Sparks fice; 43"34'N. 116"13'W; 865 m above mean sea and Malechek, 1968) in which dried samples level) to assess seasonal and annual climate pat were ground over a I-rum mesh screen, 20 fields terns during 1991-1994. We examined monthly were read from a prepared microscope slide, and cumulative precipitation and monthly averages plant species from the fecal samples were iden for maximum daily temperatures for January tified by matching epidennal patterns with those June and total snowfall for January-February, of plant tissues from reference material. Our and compared them to long-tenn means (1940- ability to distinguish sources of seed material 1990). Temporal patterns of climate were similar was variable, so we combined seed material into for Boise and the Birds of Prey Area. but actual a single diet category for statistical comparisons. precipitation at the Birds of Prey Area is ca. Evaluation of diet methodology.-We used 84% of that at Boise (Van Home et al., 1997). two approaches to assess rcHability of our fecal Vegetation.-We sampled vegetation and analysis for estimating diet. First. we evaluated plant phenology on each site to provide a con repeatability of the lab in estimating percent rel text for assessing dietary differences. We esti ative density of dietary items in our samples. In mated plant cover during two periods in 1991 1993. we selected three fecal samples from each (14 March-9 April, 17 April-2 May) and during of three habitats (grassland. sagebrush, and win three periods in 1992-1994 (19 February-l0 terfat) during early and mid-season and three March. 12-20 April, 22-30 May). We measured samples from two habitats (grassland, sage plant cover at 15-22 random locations in the brush) during late season. Thus. we used 24 trapping area on each site with point-intercept samples in our evaluation. We divided each of methods using a 3- by 0.5-m point frame as the the samples into three subsamples by splitting sampling unit (Floyd and Anderson. 1982). In individual fecal pellets into three equal pieces addition to recording the species encountered at and then placing each piece into one of the three each intercept, we noted if a grass blade or a subsamples. Choice of samples and placement forb leaf was green and succulent or brown and of material into subsamples was done in a ran desiccated. We used this status variable (succu dom manner. All subsamples were placed in lent or desiccated) to examine temporal changes sterile bags and labeled following normal pro in desiccation patterns for Sandberg's bluegrass. cedures, except that two of the three subsamples which was presumed to be the main perennial were given false identification numbers. The lab grass in the diet of the ground squirrels. For that was not notified of our evaluation, and thus it analysis. we used sites as replicates and exclud was a blind test. We calculated a similarity index (Anthony and Smith, 1977; Wydeven and Dahl ed those sites with ::s: 1.5% total cover of Sand berg's bluegrass. We summarized mean cover gren, 1982) for each set of three subsamples as: age of common plant species (> 1% horizontal coverage) for each habitat type using averages across years and sample periods. We also ex amined temporal changes in the coverage of the where Yi was the lowest percentage value of a three most common annuals (Russian thistle shared diet item and n was the total number of Salsola iberica, tansymustard-Descurainia items. We tested for differences in mean simi- 524 JOURNAL OF MAMMALOGY Vol. 79, No.2 larity indices among seasons with an analysis of Johnson, 1980) have been criticized for loss of variance (ANOYA) after arcsine-square root quantitative infonnation (Alldredge and Ratti, transfonnation of the data (SAS Institute Inc., 1986). Ratios. however. may be particularly im \989). precise when resource availability is low. Also, We were interested in the relationship be when animals prefer to include a given food as tween fecal samples and stomach contents. Wes a certain proportion in their diet regardless of toby et al. (1976) detailed some of the problems availability, the preference value will nonethe with using particle counts from stomachs to es less vary with the quantity available. Where this Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 timate herbivore diets. In general. taxa are over food is common. it may not be "preferred," estimated when abundant and underestimated even though it represents a large fraction of the when rare. Biases associated with differences in diet. In contrast, a food that animals consume digestibility are likely to be magnified in fecal only in small quantities, even when it is abun analyses, although Shalaway and Slobodchikoff dant, may appear preferred where it is rare. We (1988) and Wydeven and Dahlgren (1982) found calculated preference patterns to aid in interpre a high positive association between particle fre tation of dietary variation and thus did not test quencies in stomachs and in feces of Gunnison's for differences in preference statistically. We (Cynomys gunnisoni) and black-tailed prairie used the six most common foods to measure dogs (C. ludovicianus), respectively. To compare preference. Because we had only one measure similarity in diets from paired stomach and fecal of vegetation, and hence availability, for each samples from individual ground squirrels, we site, we used relative proportions of these six collected squirrels either through shooting or forages on each site as a measure of availability snap trapping in grassland and sagebrush habi tliat we compared with the mean diet on that tats during early and late seasons in 1993. Sam site. Each site rather than each animal was an ples from stomachs and feces were dried and observation. We thus report the mean (± 1 SE) analyzed following our nonnal procedures. We of preference across sites within age classes. compared correspondence in diets estimated years, and sample periods (early-, mid-, and late from paired stomach and fecal samples using the season). We restricted our measure of availabil previously defined similarity index and ANOVA ity to percent cover of green (not woody or se on arcsine-square root transfonned data (proc nescent) plants of a given species. This may GLM; SAS Institute Inc., 1989). have led us to underestimate availability, and Diet analysis.-To investigate differences in thus overestimate preference, on the few occa estimated diet among habitats, sexes, and ages, sions when animals consumed apparently senes we used multivariate ANOYA of arcsine-square cent grasses. root transfonned values (SAS Institute Inc., To investigate effects of the 1992 drought on 1989) of percent relative frequency for the most diet, we restricted analyses to sagebrush and common fragment types. We report results only grassland habitats, in which we had more sites for analyses with d.! ;;:: 20; in some years, in and thus larger numbers of samples. We ana dividuals of a given sex or age class were too lyzed dietary composition separately for habitat rare to provide sufficient numbers of fecal sam types, ages, and sampling periods so that year ples for robust analysis. To interpret significant was the only classification variable. results, we conducted separate ANOVA of diet Forage quality and quantity.-To estimate items and used least squares-means for pairwise forage quality in different habitats and at differ comparisons among years or habitats. We also ent times of the year, we used stomach contents measured similarities between male and female from ground squirrels collected for the stomach diets using the similarity index. feces similarities. Contents were analyzed for We used Savage's (1931) forage ratio and its acid detergent fiber. neutral detergent fiber, lig associated variance (Manly et al.. 1993:38) to nin, nitrogen, gross energy, and lipid content by examine preference among common foods and the Nutritional Analysis Laboratory at Colorado how preference changed with season and age State University. We used the resulting data to class, a ratio> I indicated a food was being con detennine the proportion of the sample com sumed in higher proportions than its availability. posed of nearly or completely indigestible cell Measures of preference based on differences in wall components (hemicellulose. cellulose, and rank of available and consumed foods (e.g .• lignin) and highly digestible components (pro- May 1998 VAN HORNE ET AL.-DIETS OF TOWNSEND'S GROUND SQUIRRELS 525 tein, lipids, digestible carbohydrates-Robbins, A c 1983; Van Soest, 1963, 1967). a 250 1993 ~ Assimilation depends on both forage quality l!I and the quantity consumed. Dry mass of gut .1i 200 contents is an indicator of the latter when move ."w_ .... X " E 150 .•.. ment through the digestive system is constant. ~E 1991 ~- 1994 Unusually large gut capacity may be an indica ~ .!l! 100 1992 tor of low forage quality (Gross et al., 1985) and ~ E Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 perhaps vulnerability to food limitation. We cal ~ 50 <) culated wet masses of gut contents for four 0 squirrels by subtracting wet mass of the empty J F M A M J digestive tract (stomach, cecum, intestines) from Month mass of the full digestive tract. Mean mass of empty digestive tracts was 7.3% ± 1.0% of body mass. We compared the measured wet mass of gut contents with that predicted by the interspe B cific equation for herbivores: wet mass of con tents (kg) = 0.0936 WI.03 where W was body E mass in kg (Demment and Van Soest, 1985; ~ 1c E E Gross et aI., 1985). For 29 squirrels, we had only .~ .g summer weights of digestive tracts and gut con ~ - tents. For those individuals, we assumed that 'Oct:c empty digestive tracts weighed 7.3% of total a ~ ~ro 2 body mass, and we compared. estimated wet ·5 w mass of gut contents to that predicted by the iE interspecific equation. " J!l
JFMAMJ JFMAMJ J RESULTS 1991 1992 1993 1994 Climatic patterns.-Extreme differences FIG. I.-Climatic patterns for January-June in climate among years should influence 1991-1994 from data recorded at the weather vegetation in this dry region. In 1991, pre station located at Boise, Idaho: A) monthly cu cipitation was somewhat below average, mulative precipitation relative to the long-tenn but maximum temperatures were near nor mean (1940-1990), and B) deviation from nor mal, except in February (Fig. 1). In 1992, mal (1940-1990 mean) for monthly means of a severe drought occurred during the active daily maximum temperatures. season of ground squirrels; there was little precipitation from February through May, and maximum temperatures were above av in the other 3 years. In 1994, precipitation erage for January-June (Fig. I). Precipita was below nonnal and maximum tempera tion that occurred in June 1992 fell after tures were above nonnal, but conditions perennial grasses were donnant and nearly were not as severe as in 1992 (Fig. 1). all of the ground squirrels had immerged. Therefore, the main climatic patterns during In 1993, cumulative precipitation was our study were a severe drought in 1992, above average for March-June, and maxi prolonged winter weather in early-season mum temperatures were generally below 1993, and unusually high precipitation dur average (Fig. 1). Cold temperatures in early ing most of 1993. 1993 were associated with above average Plant cover.-On the two winterfat sites, snowfall in January (371 mm, 50-year winterfat was the most common vegetation mean = 168 mm) and February (264 rrun, 1ype (X ~ 16%) followed by mosses (Tor 50-year mean = 88 rrun). Monthly snowfall tula ruralis and others; 10%), Sandberg's during January and February was ~96 mm bluegrass (5%). and budsage (Artemisia 526 JOURNAL OF MAMMALOGY Vol. 79, No.2 spinescens; 1 %). On the six sagebrush sites, >'00 big sagebrush was the most common 0.75 "" groundcover = 23%), followed by moss eX 0.50 es (15%), Sandberg's bluegrass (9%), bur buttercup (Ranunculus testiculatus; 5%), 0.25 cheatgrass (5%), and squirreltaiI (Sitanion 000 hystrix; 3%). Big sagebrush was also the u most common groundcover on the sage .!! Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 brush-winterfat mosaic sites (X = 19%), .,.1l m followed by Sandberg's bluegrass (12%), u w winterfat (10%), mosses (9%), and tansy ~ u mustard (Descurainia; 8%). Common spe m .,.~ cies on grasslands, including reseeded sites, w were Sandberg's bluegrass (X == 17%), 1'l cheatgrass (6%), Russian thistle (4%), and ~ ~ bur-buttercup (3%). Reseeded winterfat 8 m covered <0.6% of the area on the reseeded w ~ sites. &. Vegetation dynamics.-Any pronounced 15 c temporal changes in timing and coverage of a vegetation would be expected to influence a 0. diet of the largely herbivorous Townsend's '"e "- ground squirrels. There were substantial " among-year differences in the temporal pat .. terns of desiccation of Sandberg's bluegrass (Fig. 2). The drought of 1992 was associ ated with early drying of Sandberg's blue grass; a mean of 43% of Sandberg's blue grass was classified as desiccated in April of 1992 compared with ::=;12% in the other Time of sampling 3 years. Late winter of 1993 was associated FIG. 2.-Seasonal and annual patterns in the with a delayed green-up of Sandberg's blue proportion of Poa secunda that was classified as grass compared with other years (Fig. 2). desiccated at the Birds of Prey Area, 1991- The above-normal precipitation during the 1994. Each bar represents a mean (+ 1 SE) for spring of 1993 also was related to a delay multiple study sites (n = 16 in 1991; n = 18 in in the desiccation of Sandberg's bluegrass; 1992-1994); most SEs were ~O.03 and are not a mean of 71 % of Sandberg's bluegrass was displayed; width of each bar along the x-axis classified as desiccated in late May 1993 indicates the time period when sampling was compared with 99% in 1992 and 1994 (Fig. conducted. 2). Several species of annual plants had a but by late season in 1993, it averaged growth flush during the wet spring of 1993. 14.8% on grassland sites and 8.0% on re The coverage of succulent tansymustard seeded sites. The only other time with mod was typically ::=;1.5%, but by late season erate coverage of Russian thistle was in late 1993, it averaged 10.2% on winterfat sites, season of 1994 (grassland = 5.6%, reseed 3.4% on sagebrush sites, 21.0% on mosaic ed = 7.2%). Russian thistle did not exhibit sites, 10.7% on grassland sites, and 16.8% a growth flush on any of the shrub-domi on reseeded sites. Coverage of succulent nated habitats. Coverage of succulent cheat Russian thistle also was generally < 1.5%, grass also peaked in late season of 1993 on May 1998 V AN HORNE ET AL-DIETS OF TOWNSEND'S GROUND SQUIRRELS 527
TABLE I.-Similarity indices (SI) for comparisons of diet items of Spermophilus townsendii during early-, mid-, and late-season in 1993. Comparisons are for triplicate subsamples of individual fecal samples and for paired stomach contents and feces.
SI (%) for triplicates SI (%) for of fecal samples stomach-feces comparisons Season X SE Range n X SE Range n
Early 86 4.8 64-100 9 92 5.3 55-100 7 Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 Mid 86 3.9 59-98 9 Late 58 9.6 22-85 9 67 6.4 41-97 10 both grassland eX = 10.8%) and sagebrush ilar pattern and reported a mean similarity sites (8.1 %). of 63% between stomach and fecal samples, Evaluation of diet methodology.-The which was lower than ours. However, their mean similarity index for triplicate subsam samples contained more species than our pIes of individual fecal samples was 72 ::t samples (averages of 9.1 and 9.7 species in 4.1 % (n = 24). Repeatability of dietary de stomach and fecal samples, respectively), terminations was similar in early- and mid so we would expect their similarities to be season, but it declined in late season (P = lower if similarity declines with increased 0.008; F ~ 6.10, d.f ~ 2,21; Table I). Total numbers of species. number of dietary items in fecal samples Dietary variation.-We analyzed 457 fe was similar in early season (X = 2.4, range cal samples in 1991, 403 in 1992, 264 in = 1-5) and in mid-season (3.0, range = 2- 1993, and 308 in 1994 for a total of 1,432 5), but it increased in late season (7.8, range samples. More than 80% of the mean diet = 6-9). There was a negative association across years and seasons in each habitat between total number of diet items in fecal consisted of four plant species: Sandberg's triplicates and their similarity (rs = -0.76, bluegrass, winterfat, big sagebrush, and P ~ 0.0001, n ~ 24). Russian thistle (Appendix I). The mean similarity index of dietary To assess overall differences in diets items in stomach contents and feces of among habitats, sample periods, and ground squirrels was 82 ± 5.0% (n = 17). drought and non-drought years, we used However, similarity in diets between stom broad classifications of dietary items to de achs and feces was greater in early season velop a summary graph (Fig. 3). Statistical than in late season (F ~ 11.92, d.f ~ 2,14, tests for differences were conducted on dis P = O.cX>4; Table 1). Average number of tinct dietary items, not on those broad clas dietary items in stomachs increased from sifications. There was little indication of di early season eX = 1.2, range = 1-2) to late etary differences between drought and non season (4.0, range = 2-6). Number of items drought years in early-season samples (Fig. in fecal samples also increased from early 3); effects of the severe drought of 1992 season (X = 2.3, range = 1-4) to late sea appeared in mid-season and were most pro son (5.0, range = 3-7). Overall, there was nounced by late-season. By mid-season, a negative correlation between total number consumption of annuals was relatively low of dietary items in a paired sample and the in grassland habitats during the drought, similarity index (r, = -0.90, P = 0.0001, and by late-season, the decrease in con n ~ 17). sumption of both annuals and seeds was Fecal samples of Townsend's ground pronounced, except in winterfat habitats squirrels tended to have more species of where those items were always rare in the forage items than did stomach samples. diet. Wydeven and Dahlgren (1982) found a sim- We expected that diets might vary be- 528 JOURNAL OF M AMMALOGY Vol. 79, No.2
Non-drought Drought Early s.eason all habitats III each of these groups based on all 12 dietary categories. We detected no '" differences in diets between either adult " males and females (P ;;:: 0.205 for 15 com parisons, d.f. =::: 20) or juvenile males and females (P =::: 0.179 for 13 comparisons).
Mid season Mean similarity of male and female diets was 74.4% for adults and 60.8% for juve Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 niles. Inspection of the mean diets revealed no systematic differences, so we concluded th at male and female diets were similar and I~ mu Ii ' Jill[ combined them for all remaining analyses. ~ 0 "" Late season To compare diets of adults and juveniles. we performed a MANOVA using the six most common mid-season dietary catego ries (tansymnstard replaced cheatgrass as . ~ Il[ ~! ~nr~ the sixth most common item) for individu wnena! MosaIC ResoodDd Wuterf81 MoS<\IC Reseedec als blocked by habitat in each of the years Sagebrush Grassland S~ Grasslard s~parately. Diet differed significantly in I_smm DHe Vegetation cover Diet Winterfat "0"",." •• Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 Sagebrush Sagebrush-winterfat mosaic •• 91 92 93 94 Grassland Reseeded grassland 91 92 93 94 91 92 93 94 • POSE § CELA 0 ARTR 0 SAIB rn DESC 8 BR TE 0 Other FIG. 4.-Mean percentage of vegelation cover across sites within habilat types for the six moSI common plants (POSE = Sandberg's bluegrass, CELA = winterfat, ARTR = big sagebrush, SAIB = Russian thistle, DESC = tansymustard. and BRTE = cheatgrass) found in diets of Townsend's ground squirrels in the Birds of Prey Area. Idaho, and relative frequency of Ihese items and all other dietary ilems combined (Other); bars represent vegetation and dietary samples for early- and mid season in 1991, and early-. mid-, and late-season in all other years; bars arc omiued where low densities precluded diet estimation; age classes were pooled for ca1culating diet means; mid- and laic-season samples in 1992 when drought effeclS were most severe are indicated with aSJerisks. 530 JOURNAL OF MAMMALOGY Vol. 79, No.2 TABLE 2.~Dietary preference (means and standard errors) across sites for the six most common food items of adult arul juvenile Townsend's ground squirrels at different times in the 1991-1994 active seasons in the Birds of Prey Area. Foods with means >1 (asterisk) comprised a higher mean proportion of the diet than their mean proportionate coverage among the six food types on sites and may have been preferred. "High" and "low" indicate the item was apparently preferred, but the mean availabLe resource covered <0.1% aflhe ground so that we have little confidence in the actual preference value. "High" items comprised >10% and "low" items <10% of the mean diet for a given age class, time, and year. Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 Sundberg'S Big Russian Tumble- bluegrass Winterfat sagebrush thistle Cheatgrass mustard (perennial) (perennial) (perennial) (annual) (annual) (annual) Ago Season Year Ii SE Ii SE X SE X SE X SE X SE Adult Early 1991 1.3* 0.D4 1.4* 1.24 0.2 0.02 1.0 0.71 0.0 low* 1992 1.6* 0.03 0.6 0.13 0.0 0.0 0.0 low* 1993 1.8* 0.08 0.1 0.01 0.2 0.01 0.1 0.01 0.0 low* 1994 4.3* 1.12 0.1 0.00 0.0 0.0 low* 0.0 Mid 1992 1.3* 0.03 1.5 0.14 0.6 om 0.2 0.01 0.0 0.8 0.12 1993 0.8 0.02 4.8* 10.87 2.3* 0.20 1.5* 0.08 0.0 1.6* 1.14 1994 0.1 2.58 3.7* 0.18 3.1* 1.01 1.1 * 0.01 low* 0.3 0.03 Juvcnile Mid 1991 0.9 0.01 1.6* 0.16 0.9· 0.01 low 0.4 0.08 low* 1992 1.5* om 1.3* 0.07 0.2 0.00 0.2 0.00 0.0 0.5 0.06 1993 1.0 0.03 3.2* 10.04 1.4* 0.04 1.9 0.15 0.0 0.9 0.04 1994 7.4* 7.53 2.8* 0.20 0.4 0.03 0.2 0.01 2.4* 3.39 0.1 0.00 Latc 1992 high* 1.4* 0.05 0.2 0.00 0.1 0.00 0.0 low* 1993 3.8* 1.19 3.2* \.43 1.1* 0.00 1.0 0.10 0.7 0.38 0.4 0.02 1994 high* 0.5 0.\3 0.9 0.09 0.7 0.02 1.4* 0.44 0.3 0.04 creased in 1993 and 1994 in grassland and nuals (Fig. 4), so preference for perennials reseeded-grassland habitats. reflected higher consumption rather than Some of the dietary variance may have lower availability. Winterfat and big sage indicated preference or avoidance. In sage brush were preferred by adults at mid-sea brush-winterfat mosaic habitats, winterfat son in 1993 and 1994 and juveniles in 1993 was relatively more common than big sage when Sandberg's bluegrass was not pre brush in diets, although coverage of those ferred (Table 2). Preference patterns for pe plants was similar (Fig. 4). The perennial rennial shrubs may have been associated plants-Sandberg's bluegrass, winterfat, with relatively wet years and timing of new and big sagebrush, tended to be more com shrub growth. During the drought of mid mon in diets than annual plants-Russian and late-season in 1992, winterfat was pre thistle, cheatgrass, and tansymustard. ferred food. Fonnal calculations of preference were Effects of drought on diet.-During the consistent with those patterns. Among drought, Townsend's ground squirrels did foods common in the diet, mean prefer not follow nonnal patterns of mass gain. ences for perennials generally were higher Compared to 1991, pre-immergence masses than those for annuals (Table 2), although at the end of the above-ground season in it was likely that annuals had a shorter pe 1992 were 43% less for adult males, 44% riod in which they were commonly eaten less for adult females, 52% less for juvenile (and we may not always have sampled pre males, and 35% less for juvenile females. cisely at this time). Among the plants used Fewer animals were recaptured between in preference calculations, availabilities of 1992 and 1993 than between 1991 and perennials were greater than those of an- 1992. We measured persistence of individ- May 1998 VAN HORNE ET AL-DIETS OF TOWNSEND'S GROUND SQUIRRELS 531 Ga POSE 0 ARTR 0 SAIB c:;:) DESC o Other D CELA D SEED fimI BRTE _ ARTH Sagebrush 28 24 25 33 27 26 11 7 17 25 7 20 ~';~ll Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 !!l 0 lUll m liDW Grassland ~ 27 105 53 43 41 22 -III 100 .E; 75 !!l Q) '" Early Mid lUllEarly Mid Mid Late Early Mid Mid Late Adults Ju .... eniles Adulls Ju .... eniles Adults Ju .... eniles 1991 1992 1993 1994 FIG. 5.-Mean relative frequencies of food types in diets of adult and juvenile Townsend's ground squirrels in the Birds of Prey Area, Idaho, during 1991-1994 in sagebrush and grassland habitats (POSE = Sandberg's bluegrass, CELA = winterfat, ARTR = big sagebrush, SEED = seed material, SAIB = Russian thistle, BRTE = cheatgrass, DESC = tansymustard, ARTH = arthropods, and Other = other fragment types); sample sizes are indicated above bars. uals through the inactive period as the per habitat, age, and sample period, the MAN centage of marked individuals present dur DVA with year as an effect was significant ing the immergence period that were recap (P < O.()(x)l). All differences discussed be tured in the following season; that number low had P < 0.05 for least-squares means reflected both survival and emigration from comparisons of separate ANDVA for each study plots (Van Horne et aI., 1997). Per dietary item. Early in the season, consump sistence declined from 55% to 34% for tion of Sandberg's bluegrass by adults in adult males, from 69% to 12% for adult fe sagebrush habitats was relatively low in males, from 29% to 1 % for juvenile males, 1994, but consumption of Sandberg's blue and from 51 % to 1 % for juvenile females grass in grassland habitats was relatively (Van Home et aI., 1997). high in 1992 and consumption of Russian Because effects of drought on change in thistle was relatively low in 1992 and 1994 body mass and plant phenology began to (Fig. 5). In mid-season, consumption of appear in mid-season of 1992 and were Sandberg's bluegrass and seeds by adults in most evident by late-season that year, we sagebrush habitats was relatively high in were interested in dietary differences be 1994. Similarly, adults in grassland habitats tween 1992 and other years during those ate relatively more Sandberg's bluegrass periods. We did not expect to see drought and less Russian thistle in both 1991 and effects on early-season samples. Because 1992, and more seeds in 1992. we used only sagebrush and grassland hab Differences in diet for juveniles in mid itats that were not reseeded to test for season of 1992 were similar to patterns of among-year differences in diet, winterfat adults. Juveniles in sagebrush habitats in was rare. We used Sandberg's bluegrass, mid-season also ate more seeds in 1992 big sagebrush, seed, and Russian thistle as than in the other years, and they ate more dependent variables in MANDVA. For each Sandberg's bluegrass in both 1992 and 532 JOURNAL OF MAMMALOGY Vol. 79, No.2 1993 than in 1991 and 1994. They ate less TABLE 3.-Composition of forage removed Russian thistle in both 1991 and 1992 than from stomachs of Townsend's ground squirrels in in 1993 and 1994 and more big sagebrush the Birds of Prey Area, Idaho, 1993. Sample sizes in 1991 than in the other years. Juveniles are 14 for early season (March) and 9 for late in grassland habitats ate more Sandberg's season (May). Means that differ significantly be tween early and late season have an asterisk. bluegrass and less Russian thistle in 1991 and 1992 than in 1993 and 1994. High seed Early season Late season Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 consumption by adults in both habitat types Component X SE X SE and juveniles in sagebrush at mid-season of Cellulose (%) 11.3 0.23 12.9 1.40 1992 may have been related to earlier phe Lignin (%) 3.2* 0.32 6.2* 0.76 nology, especially of grasses, associated Hemicellulose (%) 22.9 0.49 22.5 3.54 with the warm early spring and subsequent Silica (%) 2.6 0.45 2.0 0.46 onset of drought. High consumption of Protein (%) 34.9* 0.82 19.7* 1.90 Sandberg's bluegrass, given its high cov Lipid (%) 7.4 0.71 7.S 1.07 Gross energy erage value in these habitats, was probably (kcal/g) 4.45 0.Q4 4.32 0.14 associated with lower availability of alter nate foods such as new growth of big sage brush, and Russian thistle, a moisture-sen sitive annual whose coverage was reduced Gut contents.-For the four ground by dry conditions. squirrels with actual weights of gut con Late-season samples in 1992-1994 con tents, wet masses of gut contents were 144 tinued the same trends, except that seed ± 9.4% higher than those predicted using consumption was lower than in 1991. Con the interspecific equation based upon body sumption of seeds was lower on both hab mass (e.g., actual mass 37 g, predicted mass itats in 1992 and 1993 than in 1994. Less 15 g). For the remaining 29 squirrels for Russian thistle was consumed in 1992 than which we had estimated weights of gut con in 1993 and 1994 on grassland sites, but tents, gut contents weighed 84% more than consumption was lower on sagebrush sites would be predicted by body mass (e.g., ac in both 1992 and 1993 in comparison to tual mass 27 g, predicted mass 15 g). Ac 1994. Consumption of Sandberg's bluegrass tual increases in gut capacity over predicted by juveniles late in the season was higher values were probably higher than those re in 1992 than in the other years in both hab ported because it was unlikely that all di itats. Again, this likely was associated with gestive tracts were filled to capacity. shortage of alternative foods. Forage quality.-Although ground DISCUSSION squirrels in grassland and sagebrush sites General dietary patterns.-The impor had different diets, they selected diets of tance of Poa secunda in diets of Town similar quality in 1993 (Table 3). Nutrition send's ground squirrels in the Birds of Prey al quality of the stomach samples did not Area, especially early in the season, is con differ (P > 0.20) between ground squirrels sistent with indications from other research collected in grassland and sagebrush sites ers that Townsend's ground squirrels are in 1993, although the percentage of lignin highly dependent on grasses. Rogers and increased (P = 0.(012) and protein de Gano (1980) found Poa, Descurainia, and creased (P < 0.0001) between March and Lupinus to be the most common foods of May (Table 3). Caloric values of stomach Townsend's ground squirrels in southcentral contents (gross energy) also were similar to Washington. Koehler and Anderson (1991) those reported by Rickart (1982) for Town reported that Townsend's ground squirrels send's ground squirrels in Utah eX = 3.8- ate a large proportion of crested wheatgrass 4.6 kcal/g). (Agropyron desertorum) at a site in south- May 199X VAN HORNE ET AL-OIETS OF TOWNSEND'S GROUND SQUIRRELS 533 eastern Idaho (this grass was not similarly unsaturated fatty acids cannot be synthe available in the Birds of Prey Area). sized endogenously, except from specific Yensen and Quinney (1992) reported that fatty-acid precursors (1. C. Corn, in litt.). grasses comprised 37-87% of the diet of Animal adipose tissue content of these es Townsend's ground squirrels in the Birds of sential fatty acids is correlated closely with Prey Area, with winterfat and sagebrush dietary content, although somewhat higher also being important components. As we (J. C. Corn, in litt.). Winterfat and sage found for juveniles, animals in their study brush are similar in linolenic acid content Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 consumed more seeds prior to torpor. Our but higher relative to Sandberg's bluegrass. data also are consistent with their sugges Those shrubs are likely to be a better source tion that seed consumption decreased under of essential fatty acids than are grasses. drought conditions. We did not, however, Lack of difference between male and fe find consumption of cheatgrass as high as male diets is somewhat surprising, given that reported by Yensen and Quinney the size dimorphism between sexes and the (1992). Because cheatgrass is an annual, its need for females to bear most of the ener coverage varies greatly with rainfall pat getic and nutrient costs of reproduction late terns. Coverage was higher during their in their short active season. However, males study, averaging 11% in 1987 and 33% in have high energy demands associated with 1988 on sites dominated by native perennial early arousal and searching for mates early bunch grasses. In contrast, mean coverage in the active season and may need to store of cheatgrass during our study varied from extra fat late in the previous active season 3% to 14% on native grass sites during to meet those demands. Variation in avail 1991-1994. ability among years, seasons, and habitats Leaves of shrubs are consumed by Town also may constrain the optimal diet for send's ground squirrels when available, al males and females in a similar fashion. though it appears that consumption of big In contrast to the male-female pattern, sagebrush is limited largely to new growth. there were differences in diet between ages Cook (1977) reported that ether extract (es shortly after juvenile emergence-the only sential oils), protein, and gross energy of time at which we sampled both adults and big sagebrush and winterfat in the Great juveniles. Such differences could result Basin of western Utah were higher in April from a lack of foraging experience causing and May when those browse species were juveniles not to be selective or to respond actively growing than in January or July more directly to availability than did adults, when they were maturing or quiescent. New or they may be reluctant to forage widely growth of shrubs may provide some nutri from the burrow entrance because of extra ent value not available from Sandberg'S predation risk caused by lack of experience bluegrass. According to Cook (1977), those and physical immaturity. It is also possible browse species were higher in protein, lig that their diet is no less optimal than that nin, and phosphorus than were grass species of adults, but they are either relatively in (Indian ricegrass, Oryzopsis hymenoides, tolerant of the secondary chemistry in sage and squirreltail grass), which were higher brush leaves, or they avoid protein-com in cellulose. plexing properties of secondary chemistry Shrubs also may provide more fat than of sagebrush during their rapid growth grasses. Linoleic acid can decrease energy phase. costs of hibernation in marmots (Florant et Effects of the drought on diet.-Differ aI., 1993), and the polyunsaturated fatty ac ences in nutrient levels of foods may help ids linoleic and alpha-linolenic may be es explain continued low reproduction and sential for successful hibernation in ground survival during and immediately following squirrels (J. C. Corn, in litt.). Those poly- the drought (Van Home et aI., 1997). New 534 JOURNAL OF MAMMALOGY Vol. 79, No.2 growth of grass is a better source of essen logical requirements for essential fatty ac tial fatty acids than mature grass. Linolenic ids, but diets in grasslands apparently did acid is higher in green plants than in senes not (1. C. Com, in litt.). Consumption of cent plants and is also relatively high in annuals was always relatively lower for seeds (1. C. Com, in litt.). Both new growth both adults and juveniles in these shrubby of grasses and numbers of seeds were at habitats, so that rarity of annuals during low levels late in the 1992 season. Plants in drought had relatively less effect on overall 1993 had relatively low content of linoleic diet. During drought, proportions of foods Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 acid compared with pre-drought conditions. in the diet were least affected on the win Effects of drought on diet varied with terfat sites, and individuals on the mosaic habitat type. Alternative foods were an im sites sought out shrub material (mostly win portant component of the diet in winterfat, terfat) under drought conditions. It thus ap sagebrush, and mosaic habitats when Sand pears that habitats dominated by native berg's bluegrass senesced early and seeds shrub species may support a more stable matured early in the drought year of 1992. food source under drought conditions than Juvenile ground squirrels consumed rela habitats dominated by Sandberg's bluegrass tively high amounts of Sandberg's blue and exotic annuals. grass late in the season despite its senescent Our study indicates that ground squirrels state, and did not gain mass as they did dur ~e dependent on perennial plants, especial ing non-drought years. Only a few juveniles ly early in the active season, and that in sagebrush habitats and none in grassland drought may drastically decrease availabil habitats were recaptured after the interven ity and consumption of annual plants late ing inactive season; clearly, the diet of se in the active season. As a consequence, we nescent Sandberg's bluegrass was inade believe that any habitats dominated by an quate. Adults in sagebrush habitats were re nual plants that lack substantial coverage of captured the year following the drought at perennial plants cannot support stable, in relatively higher rates than in grassland creasing, or source populations of ground habitats (30% for adult males and 28% for squirrels for any substantial time period. adult females in sagebrush; 17% for adult Regional extinctions of ground squirrels are males and 10% for adult females in grass likely to occur, especially during drought, if land excluding reseeded and food-supple such habitats become widespread. These mented sites; Van Home et al., 1997). Loss annual-dominated habitats are a product of of annuals as a major diet component, es human disturbance and mixing of biotas, pecially on grassland and reseeded sites and ground squirrels may not be able to during the mid-season period of the drought adapt to them. Even if large blocks of year, may explain the relatively low persis shrub-steppe habitat are not converted to tence of animals on these sites as compared annual-dominated grassland, we predict that to sagebrush sites, where animals consumed increased coverage by annuals and de more shrub and other foods during the mid creased coverage by perennials will in season of the drought year. crease sensitivity of Townsend's ground Animals in sagebrush habitats consumed squirrel populations to low moisture con relatively more shrub material in both ditions. drought and non-drought years than those in grassland habitats. This alternative food ACKNOWLEDGMENTS source may have provided an important This research was funded by the United States source of water and nutrition during Bureau of Land Management (BLM) through drought. Because animals in sagebrush ate Contract YA651-CTO-340032 to B. Van Horne. more shrub material, diets in sagebrush It was part of a cooperative project with the habitats in 1993 appeared to meet physio- BLM, Idaho Army National Guard, and United May 1998 VAN HORNE ET AL.-DIETS OF TOWNSEND'S GROUND SQUIRRELS 535 States National Biological Service (NBS), We PuLLIAM, H. R 1975. Diet optimization with nutrient thank >40 technicians who assisted with trap constraints. The American Naturalist, 109:765-768. RICKART, E. A. 1982. Annual cycles of activity and ping, field work, and data entry, and J, Com who body composition in Spermophilus towm'endii mol provided input throughout. S. Knick, K. Steen lis. Canadian Journal of Zoology, 60:3298-3306. hof, and M. Kochert of the NBS Raptor Re ---. 1987. Spennophilus townsendii. Mammalian search and Technical Assistance Center and J. Species, 268:1-6. RITCHIE, M. E. 1988. Individual variation in the ability Rotenberry provided many types of infonnation of Columbian ground squirrels to select an optimal and assistance. D. Quinney, Idaho Army Nation diet. Evolutionary Ecology, 2:232-252. Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 al Guard, assisted with the identification of plant ---. 1990. Optimal foraging and fitness in Colum species. J. Wiens reviewed earlier versions of bian ground squirrels. Oecologia, 82:56-67. this manuscript. ROBBINS, C. T. 1983. Wildlife feeding and nutrition. Academic Press, Inc., Orlando, Florida, 343 pp. ROGERS, L. E., AND K. A. GANO. 1980. Townsend LITERATURE CITED ground squirrel diets in the shrub-steppe of south ALLDREDGE, 1. R, AND J. T. RATII. 1986. Comparison central Washington. The Journal of Range Manage of some statistical techniques for analysis of re ment, 33:463-465. source selection. The Journal of Wildlife Manage SAS INSTITUTE INC. 1989. SAS/STAT® user's guide, ment, 50:157-165. version 6, Fourth ed. SAS Institute Inc., Cary, North ANTHONY, R G., AND N. S. SMITH. 1977. Ecological Carolina, 2:1-1686. relationships between mule deer and white-tailed SAY AGE, R. E. 1931. The relation bctween thc feeding deer in southeastern Arizona. Ecological Mono of the herring off the east coast of England and the graphs, 47:255-277. plankton of the surrounding waters. Fishery Inves COOK, C. W. 1977. Effects of season and intcnsity of tigation, Ministry of Agriculture, Food and Fisher use on desert vegetation. Utah Agricultural Experi ies, Series 2,12:1-88. ment Station Bulletin, 483:1-57. SCHOOLEY, R L., B. VAN HORNE, AND K. P. BURNHAM. DEMMENT, M. W., AND P. J. VAN SOEST. 1985. A nu 1993. Passive integrated transponders for marking tritional explanation for body size patterns of rumi free-ranging Townsend's ground squirrels. Journal of nant and nonruminant herbivores. The American Mammalogy, 74:480-484. Naturalist, 125:641-672. SHALAWAY, S., AND C. N. SLOBODCHIKOFF. 1988. Sea ESHELMAN, B. D., AND S. H. JENKINS. 1989. Food se sonal changes in the diet of Gunnison's prairie dog. lection by Belding's ground squirrels in relation to Journal of Mammalogy, 69:835-841. plant nutritional features. Journal of Mammalogy, SPARKS, D. R, AND J. C. MALECHEK. 1968. Estimating 70:846-852. percentage dry weight of diets using a microscopic FLORANT, G. L., L. HESTER, S. AMEENUDDlN, AND D. technique. Journal of Range Management, 21:261- A. RINTOUL 1993. The effect ofa low essential fatty 265. acid diet on hibernation in marmots. American Jour VAN HORNE, B., G. S. OLSON, R. L. SCHOOLEY, J. G. nal of Physiology, Regulatory Integrative Compar CORN, AND K. P. BURNHAM. 1997. Effects of drought ative Physiology, 264:R747-R753. and prolonged winter on demography of Townsend's FLOYD, D. A., AND J. E. ANDERSON. 1982. A new point ground squirrels in different shrubsteppe habitats. interception frame for estimating cover of vegeta Ecological Monographs, 67:295-315. tion. Vegetatio, 50: 185-186. VAN SOEST, P. 1. 1963. Use of detergents in the analysis GROSS, J. E., Z. WANG, AND B. A. WUNDER. 1985. of fibrous feeds. II. A rapid method for the deter Effects of food quality and energy needs: changes mination of fiber and lignin. Journal of the Associ in gut morphology and capacity of Microtus och ation of Official Analytical Chemists, 46:829-835. roga.rter. Journal of Mammalogy, 66:661-667. ---. 1967. Development of a comprehensive sys HOFFMAN, R S., C. G. ANDERSON, R. W. THORINGTON, tem of feed analyses and its application to forages. JR., AND L. R. HEANEY. 1993. Family Sciuridae. Pp. Journal of the American Scientist, 26: 119-128. 419-465, in Mammal spccies of the world: a taxo WESTOBY, M., G. R ROST, AND 1. A. WEls. 1976. Prob nomic and geographic reference (D. E. Wilson and lems with estimating herbivore diets by microscop D. M. Reeder, eds.). Second ed. Smithsonian Insti ically identifying plant fragments from stomachs. tution Press, Washington, D.C., 1206 pp. Journal of Mammalogy, 57:167-172. JOHNSON, D. H. 1980. The comparison of usage and WYDEVEN, P. R, AND R B. DAHLGREN. 1982. A com availability measurements for evaluating resource parison of prairie dog stomach contents and feces preference. Ecology, 61:65-71. using a microhistological technique. The Journal of KOEHLER, D. K., AND S. H. ANDERSON. 1991. Habitat Wildlife Management, 46: 1104-1108. use and food selection of small mammals near a YENSEN, E., AND D. L QUINNEY. 1992. Can Town sagebrush/crested wheatgrass interface in SOutheast send's ground squirrels survive on a diet of exotic ern Idaho. The Great Basin Naturalist, 51 :249-255. annuals? The Great Basin Naturalist, 52:269-277. MANLY, B. E J., L. L. McDONALD, AND D. L. THOMAS. 1993. Resource selection by animals: statistical de Submitted 7 October 1996. Accepted 3 October 1997. sign and analysis for field studies. Chapman & Hall, London, United Kingdom, 177 pp. Associate Editor was James G. Hallett. 536 JOURNAL OF MAMMALOGY Vol. 79, No.2 ApPENDIX I Fragment types identified in fecal samples from Townsend's ground squirrels in the Birds of Prey Area, /dalw. The mean percentage of total fragments for individuals within habitat types across years (1991-1994) and across sampling periods (early, middLe, late) is given. Columns represent mean total diet for a habitat and thus sum to 100, P = number of sample periods and n = mean number of ground squirrels sampled per sampling period. Items are listed in decreasing order of mean percentage across habitat types. Items comprising <0,1% of mean total fragments are indicated as "trace"." Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 Habitat Sage- Winterfat brush Mosaic Grassland Seeded (P = 7; (P = 11; (P = 8; (P = 11; (P = 10; Dietary component Category n = 10) n = 30) n = 18) n = 52) n = 18) Sandberg's bluegrass (Poa secunda) Perennial 30.8 53.7 40.8 73.6 59.9 Winterfat (Ceratoides lanata) Shrub 61.2 0.4 42.1 0.8 9.2 Big sagebrush (Artemisia tridentata) Shrub 1.7 26.2 4.7 1.4 0.2 Russian thistle (Salsola iberica) Annual trace 0.9 0.0 8.0 13.0 Grass seed Annual, Perennial l.l 5.8 2.5 3.9 4.1 Pinnate tansymustard (Descurainia pinnata) Annual trace 0.6 1.7 2.8 2.1 Boraginaceae seed Perennial 0.0 0.0 2.7 0.8 2.2 Unidentified seed Annual, Perennial 0.0 2.0 1.4 0.7 1.3 Cheatgrass (Bromus tectorum) Annual 0.0 0.7 0.1 3.5 0.6 Lappula occidentalis Perennial trace trace 2.8 0.2 1.3 Unidentified arthropod Other l.l 0.4 0.5 0.6 0.4 Unidentified forb Annual, Perennial 0.0 2." 0.0 trace trace Crested wheatgrass (Agropyron deser- torum) Perennial trace 1.2 0.2 0.2 0.6 Thistle (Cirsium at"en~·e) Annual 0.0 0.9 0.0 0.7 0.5 Medicago sativa Perennial 0.0 0.0 0.0 0.3 1.5 Atriplex Shrub 1.6 0.1 0.0 0.1 0.0 Thurber's needlegrass (Slipa thurberi- ana) or Indian ricegrass (Oryzopsj~· hymenoides) Perennial 0.1 1.3 trace 0.1 trace Rabbitbrush (Chrysothamnus viscidi- florus or C. nauseosus) Shrub 0.0 1.0 0.0 0.1 0.5 Tumblemustard (Sisymbrium altissi- umum) Annual 0.1 trace trace 0.2 Coleoptera Other 0.7 trace trace 0.2 0.1 Kochia scoparia Shrub 0.0 0.0 0.1 0.6 trace Horsebrush (Tetradymia glabrata) or spiny horsebrush (T. spinosa) Shrub 0.0 0.6 0.0 trace trace Squirreltail (Sitanion hystrix) Perennial 0.0 0.1 0.0 trace 0.4 Prairie juncgrass (Koeleria pyramida- ta) Perennial 0.0 trace 0.0 trace 0.4 Coleoptera or Hymenoptera Other 0.4 0.0 0.0 0.0 0.0 Tapertip hawksbeard (Crepis acumi- nata) Perennial 0.2 0.2 trace 0.0 trace Moss (Tortula ruralis) and other moss- Perennial 0.2 trace trace trace Ime, Orthoptera" Other 0.2 trace 0.1 trace trace Phlox (Phlox aculeata) Perennial 0.0 0.2 0.0 0.13 trace Rower Annual, Perennial 0.0 0.3 0.0 trace 0.0 Greasewood (Sarcobatus vennicula- tus) Shrub 0.1 0.1 0.0 trace trace Endogone-fungus Other 0.0 0.1 0.0 0.0 0.2 May 1998 V AN HORNE ET AL.-DIETS OF TOWNSEND'S GROUND SQUIRRELS 537 ApPENDIX I.-Continued. Habitat Sage- Winterfat brush Mosaic Grassland Seeded (P = 7; (P = 11; (P = 8; (P = 11; (P = 10; Dietary component Category n = 10) n = 30) n = 18) n = 52) n ~ 18) trace 0.0 Shadscale (Atriplex eonfertifolia) Shrub 0.2 0.0 trace Downloaded from https://academic.oup.com/jmammal/article/79/2/521/852626 by guest on 29 September 2021 Blazing star (Mentzelia albieau/is) Perennial 0.0 0.0 0.2 0.0 0.0 Unidentified bark Shrub 0.0 0.1 0.0 traco trace Compositae Annual, PerenniaJ 0.0 trace trace 0.1 trace Brassicaceae seed Perennial 0.0 0.1 0.0 traco 0.0 Diptera Other 0.0 0.0 0.0 0.0 0.1 Lepidoptera Other 0.0 0.0 0.0 0.0 0.1 • Items that comprised <0.1 % of !he mean diet in all habitat types included, in decreasing order of commonness; the annuals storksbill (Erodium ('kutarium), Lepidiwn perfoliatum or L papilliferum, Panicum. Draba vema. six weeks' fescue (Vulpia octoj/tJra), and Medicago or Melilotus; annual or perennial unidentified plant material; the perennials Sphaeralcea munroano., lichens, prickly lettuce (Lactuca serriola), and unidentified Umbelliferae; Artemisia seed; and lepidoptera larvae.