Dean E. Pearson,Yvette K. Ortega, Kevin S. McKelvey, and Leonard F. Ruggiero, USDA Forest Service, Rocky Mountan Research Staton Forestry Sciences Laboratory. PO Box 8089. Missoua. Montana 5'3807
Small Mammal Communities and Habitat Selection in Northern Rocky Mountain Bunchgrass: Implications for Exotic Plant Invasions
Abstract Agriculture and development have dramaticall) reduced the range of natire bunchgrass hahivats in heNarlhrrn Rocky Muun- lains. and the iniarion of exotic plants threatens to greatly alter the remaining pristine prairie. Small mammals play many impor- tant rules in ecmvstem functiuns. but little is known about small rnilrnrniil cummunitv comousition and structure in native bunch- composition and relative abundance were cunsistent among sites. with deer mice (Peromyscus nianiculatus) dominating, followed by muntilnr wlrs (Microrui montanus). which were uncommon. and montane shrews (Sorex mmicoiur),which were rare. Deer mice and monvanr voles eahihited complcmentar) hahilat separation. Deer mice tended lo selrcl open micro>iles and avoid sites will? high pcrccntagcs of vqctatirc covcr Male and female dccr mice dcmonstratcd strong habitat separation at two sites. but the hahilat wriablcs partitioned bcrivccn sexes differed hy site. Montane voles avoided open sites and selcctcd for concave microsites where lhe vplaliw corn was rclativcly densc. This information providcs an imponant hasclinc Cor undemanding pre~aettlr- mrnt small miimmal communities in lhr rapidly dwindling. natiie hunchgmia habitats of the Northern Rocky Mountilins.
Introduction where exotics can establish (Mielke 1977), dis- perse seeds of exotic plants (McMurray et al. 1997, Palouse prairie and sagebrush-steppe habitats of Pearson and Ortega in press), consume native and the Northwest have been dramatically altered by non-native plants and their seeds (Pyke 1986). grazing, agriculture, and development (Tisdale 1961, Mueggler and Stewart 1980, West 1996, and depredate biological control agents released see Flather et 81. 1999 for review), and most eco- to control exotic plants (Pearson 1999b, Pearson logically intact remnants persist as islands criti- et al. 2000). Reciprocally, exotic plant invasion cally threatened by the invasion of exotic plants of native grasslands can alter biomass and spe- (Daubenmire 1942, Tyser and Key 1988, Lacey cies composition of small mammal communities 1989, DeLoach 1991, Tyser 1992, Tyser and (Larrison and Johnson 1973), thereby modifying Worley 1992, Sheley et al. 1998). Valley bunch- ecological roles that small mammals play in grass- grass habitats, occurring at or below lower tim- land systems. Although little work has been done berline in the Northern Rocky Mountains. super- toexamineeffects of exotic plants on smallmam- ficially resemble the Palouse prairie grasslands ma1 communities (Lamson and Johnson 1973, of the Columbia River Basin, but are unique plant Ellis et al. 1997, Pearson et al. 2000), current re- communities differing in species composition and search suggests that exotic plant invasions can in the historical dominance of rough fescue alter small mammal ecology in complex ways. (Festucascnbrella) (Lynche 1955, Stickney 1961, Interpreting the degree to which the effects of Mueggler and Stewart 1980). Additionally, val- exotic plant invasions impact grassland ecosys- ley bnnchgrass habitats were historically much tems is contingent upon understanding conditions less extensive than Palouse prairie, and currently prior to invasion. However, a dearth of commu- only 10 to 20% of these habitats remain in west- nity-level studies of small mammals within val- ern Montana (Flather et al. 1999). ley bunchgrass habitats renders the composition Small mammals play important roles in eco- and structure of these communities uncertain. For system functions (see Pearson 1999a) and may instance, although some studies suggest that Mi- figure prominently in the ecology of exotic plant crotus are prominent species within many grass- invasions within xeric grasslands of the West. For land habitats in western Montana (Koplin and instance, small mammals create disturbed sites Hoffmann 1968. Stoecker 1972. Hodgson 1972,
Northwest Science. Vol. 75, No. 2, 2001 107 Douglas 19761, the relative impottance ofMi(.rotus effectively sampling the small mammal commu- in native bunchgrass habitats remains unclear. nity (Pearson and Ruggiero in review). Transects In this paper. we present data on species com- were spaced2100 m apart to ensure independence. pohition. relative abundance, and habitat use of The effective trapping area was therefore equal small mammals in three pristine valley bunch- at all sites. Traps were baited with peanut butter grass habitats of west-central Montana to estab- and whole oats and checked from 0700 to 1100 lish baseline information on small mammal com- hrs for four days. Small mammals were tagged munities within these ecosystems. with #I 005-1 monel ear tags (National Band and Tag Company, Newport, Kentucky 41072-0430), Methods and species, age, weight, sex, and reproductive condition were determined before release at the Study Sites trap station. Live Micratns montanus and Micro- tus penn~sjlvaizicusare not easily differentiated We sampled small niammal communities at three (Hoffmann and Pattee 1968). Therefore, since all locations in valley bunchgrass habitats of west- lnorlalities (20% of the total capture) were iden- central Montana: Bandy Ranch. Sieben Ranch, tified as montane voles (Micratus ~nontanus),and and Wildhorse 1sland.All sites fall in10 the rough fescueildaho fescue (Frstuca idnhoen.ris) habitat the habitats were gencrally too dry for M. penn.sjlvnnicus (Koplin and Hoffmann 1968, types (Mueggler and Stewart 1980), but domi- Hodgeson 1972), we assigned all Microtu,~cap- nant grasses varied at microsites among rough tures to M. montaniis. fescue. Idaho fescue. and hluebunch wheatgrass (Agropyron spiratum) as a function of moisture Small mammal habitat was assessed by visu- and aspect. All study areas were 25 ha. The Bandy ally estimating percent cover for several vegeta- Ranch site is located about 8 km northwest of tion categories within a 5-m radius of each trap Ovando. MT at 1370 m elevation. Uplands at the stalion. We estimated (I) total bunchgrasses - pri- Bandy Ranch are dominated by bunchgrass and marily bluebunch wheatgrass, Idaho fescue, and big sage (Artrmisia rridentatu) habitats and in- rough fescue, but also June grass (Koleriu crisfara), terspersed by glacial potholes and their associ- needle and thread grass (Stil~acomara), and other ated wetland vegetation. Trapping at this site was less abundant native bunch grasses: (2) total non- restricted to upland hunchgrass habitats, which bunchgrasses - mostly Poa species: (3) shrubs - included some sage and some concave microsites lnostly big sage and rabbit bmsh (Chrssothnmnu.~ with mesic vegetation. Sieben Ranch is about 8 spp.); (4) native forbs - primarily arrowleaf kmeast of Lincoln. MT at 1400 m elevation. Sieben balsamroot (Balsumorhizu sayittata) and lupine Ranch is dominated by bunchgrasses, but also (Lupinus spp.); and (5) exotic plants - spotted contains patches of big sage. Wildhorse Island is knapweed, leafy spurge (Euphorbia esula), Eur- the largest island in FlatheadLake, Lake County, asian toadtlax species (Linaria dalrnatica and L. MT, covering nearly 10 km' and as such appears vulguri.~),and cheatgrass (Bromus rectorum). to function essentially as a mainland for small Additionally, physiographic and abiotic features m;immal communities. Previous trapping stud- measured included topography of the microsite ies on islands in Flathead Lake suggest that many (i.e.. concave = 1 or not = 0; as an indicator for grassland small mammals have become established low, moist sites). percent cover of bare ground, on the various islands (Plopper 1968) and are likely and percent cover of rocks = 10 cm dia. Cover all present on the substantially larger Wildhorse estimates were made by the same two observers Island. The trapping area on Wildhorse Island at all sites. ranged from 900 to 1030 m in elcvation. Analytca Methods Field Methods We present frequency distributions of the num- We lrapped small manlmals in August 1998 us- ber of individuals captured hy species for each ing 25 Shem~anlive traps placed a1 10-m inter- site. We used logistic regression to determine which vals along eight transects per site. Transects were habitat variables best differentiated trap stations used because they generate morecaptures andmore that captured small mammals from those that did species than grids of equal size, thereby more not. Traps having multiple captures were included
108 Pearmn, Onega. McKel\ey, and Ruggiero only once in the analysis for each species. By whether categorizing capture sites by sex (i.e., employing thc gcncralircd cstimating equations only males captured, only females captured, males capability in SAS GENMOD, we developed un- and fcmales captured) resulted in better habitat suuctured correlation models that relax the stan- use models than those disregarding sex. This analy- dard assumption of independence among obser- sis effectively examines the degree to which habitat vations within transects (SAS Institute 1990). use by males and females directly overlaps (100%. Counting trap stations only once and controlling overlap results in no habitat separation), testing for spatial correlations in the data reduce the for differences in habitat variables among the three pseudoreplication that can occur in data sets due potentially distinct groups of capture sites. Be- to repeated captures of a subset of individuals cause CATMOD does nolgeneratc likclihood ratio within apopulation since individuals are most often tests for coefficients. we present Wald tests for recaptured in the same or adjacent traps. Equa- this analysis. Correlation was not built into the tions produced by logistic regression serve as re- multinomial models for habitat selection by sex source selection functions an; the coefficients senre because CATMOD does not offer this option for as selection coefficients (p) whose values indi- multinomial regression models. cate the relative importance of the predictor vari- ables in the equation (Manly et al. 1991). Posi- tive coefficients indicate selection for the resource Results and negative coefficients indicate avoidance when Bunchgrass dominated the herbaceous cover at coefficients differ significantly from zero (Manly all study sites ranging from approximately 40% et al. 1991).We chose likelihood ratio tests over cover on Wildhorse Island to 65% on Bandy Ranch Wald tests to determine the significance of selec- (Figure 1). Sieben Ranch had the lowest cover of tion coefficients because of their greater reliabil- native forbs, nonbunchgrasses, and shrubs and the ity (Hauck and Donner 1977). highest cover estimates for bare ground. Wildhorse We used a model selection procedure (Manly Island had the highest densities of nonhunchgrasses et al. 1993. Asthur et al. 1996, Ruggiero el al. and native forhs. Exotic plants were generally rare 1998) to determine whether habitat selection for to absent on all sites. Spotted knapweed. toad- each species differed among sites. thereby indi- flax, and leafy spurge cover estimates averaged cating whether site data should be pooled or ana-
Small Mamnals in Native Bunchgrass 109 Sieben Ranch
Bandy Ranch
Forb Ground Nonbunch Rock Shrub Totbunch Figure 1. hlcan percent cover rslimatcs for vegetatiun, rocks. and hare ground within a 5-m radius of each trap buion.
The pooled selection model for deer mice estimated because values were zero for one cat- provided a significant improvement over the no- egory At the Bandy Ranch deer mice avoided arras selection model (x' = 100.03. df = 7, P < 0.001), with higher cover of bunchgrass (x2= 5.57. df = 1, indicating deer mice exhibited habitat selection P = 0.018) and nonbunchgass (x2= 13.99. df = I, (Table 1). Furthermore, aggregation of the site- P < 0.001) and selected xeas with more bare ground specific selection models resulted in a model that (x2= 8.35. df = 1, P = 0.004). On Wildhorse Island was significantly improved over the pooled se- deer mice responded to all variables measured ex- lection model (x' = 92.89, df = 15, P < 0.001). cept topography. Mice on Wildhorse avoided most indicating that deer mouse habitat selection dif- vegetative cover such as native forbs (p= 9.30. df =I, P = 0.002), nonbunchgrass (x2= 5.62, df = I, P
~-~~~ = = = , ~~~, 0.01 8), and shrubs (x2 4.75, df - I, P 0.029) Ranch deer mice avoided areas with higher forb in favor of areas with higher percent cover of bare (x' = 8.23. df = 1. P = 0.004) and bunchgrass ground (x' = 26.21. df = I, P < 0.001) and rock (x2 cover (x2= 13.54, df = 1. P < 0.001) and exhib- = 14.90. df = I. P c 0.001). However, deer mice on ited marginally significant selection for rocky areas Wildhorse selected for bunchgrass cover (f = 6.92, (x' = 3.57, df = 1, P = 0.059). A selection coeffi- df = I, P = 0.009), whereas they avoided it at other cient for shrubs at Sieben Ranch could not be sites.
110 Pearson. Ortega. McKelvey, and Ruggiero I Deer Mouse lmMontane Vole
Bandy Ranch Siehen Ranch Wildhorse Island Figure 2. Species composition and relative abundance of small mammals capturcd in three location\ within valley hunchgas* habitats of a,rst-cmtral Mvluntana.
TABLE 1. Selection cuefficients for deer mice in pristine valley hunchgrass hahitats of WCSI~C~~,~hluntii~ based on logirric regression analysis. Parenthese, indicate the number of stations capturing mice. A iclccdon coefficient was not calculaud Tor ahruba at Sirhrn Rmch due to singularity of the variable.
Sclccrion cocfficicnts Modcl Forh Ground Nunbunch Ruck Shrub Bunch TODO Deviance df P
Null 0.00 0.00 0.00 0.00 0.00 Poolcd (199)" 0.011* 0.028** 0.003 0.05hx* -0.015" Siehen (48)' -0.134'"0.003 -0.010 0.145 Bandy (XI* 0.016 0.041** -0.013** 0.019 0.015 Wildhorsc(126)** 0.019** 0.036** 0.013* 0.117** -0.146* Sire deviance totals Null vi, pooled (sslscrion) Pooled (ielectionl Vs site elru rut ion)
* indicates rigniticant diffcrcncc a1 a = 0.05 ** mdicates rigniiicant difference at n = 00l
Sex-based habitat selection models for deer cant for Bandy Ranch (x' = 13.12, dl = 7. P = mice weresignificantforSiebenRanch(~'= 17.33. 0.069). Females at Sieben Ranch used sites with df = 6, P = 0.008) and Wildhorse Island (x' = more bare ground (x'= 4.85, df = I, P = 0.027) 24.09. df = 7, P < 0.001) and marginally signifi- and more total bunchgrass cover (x' = 7.75, df =
Small Mammals in Native Bunchgrass 11 1 1, P = 0.005) than males. At Wildhorse Island, MacCracken et al. 1985, Groves and Steenhof females wed sites that were less concave (x'= 1988, Koehler and Andcraon 199 1. Elliot et al. 6.17. df = I, P = 0.013) and had less shrub cover 1997). Although some species that could occur than male5 (x2= 88.3, df = I, P = 0.004). and in these habitats as uncommon or rare commu- females selected microsites with lower shrub cover nity member5 such as meadow voles (M. (X:=4.15, df = 1, P = 0.042) than stations where p~rms~1vnnicu.s).jumping mice (Za~iuspf-inceps). both males and females were captured. At Bandy vagrant shrews (S. vugrarrs), and cinereus shrews Ranch, habitat partitioning between sexes was not (S. cinereus) may have gone uncaptured, most significant for any individual comparisons (P > species described for adjacent grasslands such as 0.140). Chi-square goodness of fit tests assum- the prairie vole (M. oclvoguster), sagebrush vole ing equal sex ratios indicated that scx ratios for (Lemi.rcus currutus), grasshopper mouse (Ono- Wildhorse Island were male biased (1.40 males com?.s la~ogu.ster),plains pocket mouse (Per-og~ to females; x'= 6.61, df = 1, P = 0.010), hut the narhus prrrvus), olive-backed pocket mouse (P sex ratio for Bandy Ranch was not biased (1.59 faiciarus), western harvest mouse (Reithrodon- males to females; x2= 2.27, df = 1, P = 0.132). tomys megaloris), and kangaroo rat (Dipodon~yc nor was the sex ratio for Sieben Ranch (0.88 males ordii), do not occur within western Montana or to females; x'= 0.38. df = I, P = 0.540). are extremely rare (only a few records) (Hoffmann The pooled seleclion model for montane voles and Pattie 1968). One possible explanation for provided a significant improvement over the no- the lower species richness of valley bunchgrass selection model (x2= 18.00. df = 7. P = 0.012). habitats is that the presence of valley glaciers and However. the site-specific model provided little historic Lake Missoula (Ah and Hyndman 1986) improvement over the pooled model (x' = 22.94. extirpated more spccialized grassland species from df = 16, P = 0.1 15), indicating either that habitat the region. Forested mountain ranges surround- selection by montane voles did not differ among ing these grasslands may further serve to prevent sites. or that we lacked sufficient power to detect recolonization. such differences due to low captures of this spe- Conclusions regarding species composition and cies. Overall. montane voles selected concave relative abundance based on a single year's data micrositcs (x' = 6.30, df = I, P = 0.01 2) and avoided require some consideration of annual variability areas with high percentages of bare ground (x' = given the multiannual fluctuations exhibited by 7.18, df = I, P = 0.007). but did not exhibit selec- many small mammal species (Krebs 1996). Ex- tion with regard lo othcr variables measured. amination of the literature strongly suggests that deer mouse domination of xeric grasslands is the Discussion norm within much of the Northwest (Granter al. 1982. MacCracken et al. 1985, Pyke 1986. Groves Community Cornpostion and Steenhof 1988. Elliot et al. 1997). However, The small mam~nalcommunity we identified for deer mice can periodically decline to hecome less valley bunchgrass habitats in west-central Mon- abundant than montane voles, especially if vole tana was remarkably consistent with regard to populations increaseconcurrently with thcdecline. species composition and relative abundance among Montane voles becameessentially codominant with the three distinct locations [rapped. Dcer mice deer mice in the second year of Pyke's (1986) dominated each native bunchgrass site. Montane study in Washington. and montane voles clearly voles were consistently captured. hut relatively outnumbered other small mammals at Grant et uncommon. Montane shrews were rare at two al's (1982) ungrazed site in the Bridgcr Moun- locations, and none were captured on Wildhorse tains of Montana. Although shrews fluctuate and Island. Valley bunchgrass habitats of west-cen- can be common in moist habitats (Spencer and lral Montana appear to contain fewer species than Pettus 1966). shrews are not reported as common small mammal communities described for neigh- within dry habitats of the Northwest (Negus and boring grasslands and shrub-stcppe of eastern Findley 1959, Rickard 1960, Clark 1973. Pearson Washington. eastern Oregon, southern Idaho. and 1999a). Our results considered in the context of the westcrn Great Plains of eastern Montana small mammal studies within other grasslands of (Lanison and Johnson 1973, Rogers and Hedlund the Northwest suggest that deer mice generally 1980. Gano and Rickard 1982, Grant et al. 1982. dominate valley bunchgrass habitats. with voles
112 Pcarson, Ortega, McKelvep. and Ruggiero being uncommon and shrews rare. However. et al. 2000), may be morc effective at preying on montane voles may periodically dominate these particular insects in open habitats. or their pre- communities during population highs. Montane fel~edpreymay attain higher densities within such voles may also play a more prominent role in hlgher habitats. Deer mice also tended to select rockier elevation, more mesic bunchgrass hahitats (e.g., areas, though this relationship was only margin- Grant et al. 1982). ally significant at Sieben Ranch where rocks were relatively rare. Rocks may provide escape cover Habtat Selection that the predominantly herbaceous vegetation docs Deer mice are habitat generalists at the macro- not. However, most rocks wcrc relatively scat- habitat scale. as indicated by their viability within tered, partially buried, and small (approximately a wide range of habitats (Handley 1999). How- I to 3 dm). Possibly, rocks provide cover for se- ever, their ability toefficiently exploit a wide range cure burrow sites as deer mouse burrow cntranccs of resources can rcnder them quite speciali~edin were sometimes observedin association with rocks. their selection of microsites within habitats that We observed habitat partitioning between male differ in resource availability (e.g., Pearson 1994). and female deer mice at Sieben Ranch and In native valley bunchgrass habitats of west-cen- Wildhorse Island. Variables separating male and tral Montana, we found that deer mouse selec- female deer mice differed by site, with females tion of microhabitats varied by site. This differ- at Sieben Ranch using microsites having more cntial sclcction for rcsourccs huch as bunchgrass bare ground and more bunchgrass cowthan males. (selected for at Wildhorse Island, but avoided at and females at Wildhorse Island using microsites Bandy and Siehen Ranches) probably retlects 1) with less shrub cover than malcs. Females at the fact that the distribution of these resources Wildhorse Island also used concave sites less of- varies by site and 2) the tict that deer mice are ten than males. Since deer mice, as a species, responding to a complex of resources, many of urhich were not measured. For example, as bunch- avoided bunchgrass at Sieben Ranch and shrubs grass distributions change so may the abundance at Wildhorse Island, thc xxual separation on these and therefore relative value of associated resources variable axes would appear to result from a fur- huch as sccdh and insects that dccr micc arc for- ther partitioning of these resources hetween the aging on. However, at all sites, deer mice tended sexes. At Wildhorse Island, the species-level re- to avoid heavy vegetative cover in favor of more sponse to topography may have hccn masked by open and in some cases, rockier hahitats. Elliott habitat partitioning between the sexes, as males et al. (1997) sin~ilarlyreponed that dccr mice exhibited no avoidance of concave sites. whereas selected microhabitats with more hare ground and topography values at female trap stations and sta- less grass cover in southeastern Wyoming grass- tions where both malcs and females were cap- lands. That deer mice increase in response to graz- tured indicated relatively strong avoidance. At ing (Lamison and Johnson 1973, Grant et al. 1982, Sichen Ranch, a similar phenomenon was ob- Rosenstock 1996) may also be evidence of se- served: males exhibited no selection for bare ground lection for more open vegetation, as fra7ing de- while females differed significantly from the males creases the total vegetative biomass on a sire (Grant in their seleclion for bare ground. In general. fe- ct al. 1982). males appeared to be more selective of micro- Dccrnmusc dcction forharc ground in hunch- habitats than were males. At Wildhorse Island thc grass habitats, which are inherently low in veg- skcwcd scx ratio could have facilitated this dif- etative cover, may result from predator influences ferential habitat selection between sexes by al- on deer mice or deer mice respondins to their lowing the proportionally fewer females to bemore own prey-base. For example, use of open habitat selective of microhabitat space. However, one could indicate that predation risk from weasels. might expect that the more abundant male> would which favor heavy ground cover, cxcccds that from overwhelm such an effect by encompassing all raptors, which prefer to hunt where ground cover female microsites as well as additional hahitats. is minimal (e.g.. Korpiniaki et al. 1996).Alterna- Moreover, sex ratios did not differ from random tivcly, dccr micc. which arc primarily inxctivo- elscwhcrc, and so could not be invoked to explain rous in westem passlands (Sieget al. 1986. Pearson differential habitat selection at Sieben Ranch.
Small Mammals in Native Bunchgrass 113 Determining which sex triumphs when a re- and Randall and Johnson (1979) showed that source is partitioned within a spccics is contin- montane voles favored bunchgrass habitat over gent upon correctly assessing the resource gradi- hhrubby habitats in Washington, but overflowed ent. Bowers and Smith (1979) produced a into shrubby habitats when populations increased. defensible argument for female deer mice out- Collectively, these studies indicate that montane competing males by showing that females domi- voles avoid open habitats in favor of microsites nated microsites having higher measured soil with higher vegetatix covcr and higher soil mois- moisture in habitats where water was clearly lim- ture. Montane voles may also avoid shrubs, with ited. Seagle (1985) and Belk et al. (1988) con- the possible exccption of big sage. but since we tend that structural features such as logs and woody attempted to exclude sage from our sampling and vegetation are the preferred resources in some other shrubs were rarc on the study sites, we could habitat5 and have shown that females are more not effectively assess these relationships. closely associated with these resources than are Deer mice and montane volcs appear to oc- males. They therefore argue that females effec- cupy complemmtary niches in valley bunchgrass tively dominate the preferred resources in these habitats of west-central Montana. Whereas deer systems. This argument is reasonable, but untested. Woody strucLurc provides one important vector mice generally sclected for dryer, more open habi- along a complex of resource gradients. but un- tats with relatively little vegetative cover, mon- less it is the limiting resource. and the addcd hcn- tane voles avoided open areas in favor of moister efits of cover do not compromise other resource concave micrositcs with higher herbaceous cover. nceds (i.e.. the collective resource gradient is lin- Such complementary habitat selection could be ear and correlated with structure), it is unclcar construed as habitat partitioning resulting from whether the tradeoff, in other resource vectors interspecific competition. as has been argued for associated with increasing cover will lead to in- deer mice and other Microtus species in grass- creased fitness. Similarly, without a defensible lands of British Columbia (Redfield el al. 1977) measure of resource quality. we do not attempt and Colorado (Abramsky et al. 1979). However. to assess which sex dominated the preferred re- resource competition for food seems unlikely as sources within the bunchgrass community that we deer mice are primarily insectivorous and gnmivo- studied. However, based on our results. we con- rous in this rcgion (Sieg et al. 1986, Pearson et clude that I) resources were sufficiently limited al. 20001, and montane voles are herbivorous to warrant intraspecific partitioning within the (Zimmernian 1965. Lindroth and Batzli 1984). bunchgrass community (c.g., Bowers and Smith InLcrfcrence competition is similarly reduced by 1970). 2) resources were sufficiently variable to dihparatc activity patterns that help separate noc- allow for their partitioning among conspccifics turnal dccr mice (Falls 1968) from the largely (see Seagle 1983, and 3) intraspecific resource diurnal montane voles (Drabek 1994). Further- selection pressures differed among sites. more, although voles avoid open habilats u,here Fcw studies have examined microhabitat se- deer mice abound, deer micc arc smaller than lection of montane voles in the Northern Rocky Miuotus (by 30% on avcragc in this study) and Mountains. In our balky bunchgrass habitats. thus unlikely to exclude voles through interfer- montane voles strongly avoided trap stations in ence competition. Although decr micc avoid the open areas wilh more barc ground and selected heavier herbaceous cover, thcy arc ncither excluded concave microsites where total herbaceous cover from it nor from the concave rnicrosites where was high. The higher herbaceous cover at con- voles abound. Complementary habitat selection cave microsiles appcarcd lo rcsult from higher between deer mice and montane voles in vallcy soil moisture. Hodgson (1972) found that mon- bunchgrass habitats may arise from noncompcti- tane voles were negativelycorrelated with shrubs. tive coexistence as has been shown for dccr mice with the cxception of big sage, and positively and southern red-backed voles (Clefhriorzontys correlated with soil moisture and higher graminoid gnpperi) in Virginia (Wolff and Dueser 1986). cover in southwestern Montana. Similarly, Belk Determining the causal mechanism for habitat el al. (1988) ohscrvcd that montane volcs selected separation betwccn dccr mice and montane voles habitats with higher herbaceous cover and lower within vallcy hunchgrass habitats would require shruh cover within a mixture of habitats in Utah: reciprocal removal studies.
114 Pearwn. Ortega. McKelley, and Ruggiero Conclusions formation on small mammal community compo- sition, rclativc abundance, and habitat use in the Exotic plants have transformed vast rcgions ol rapidly dwindling native hunchgrass habitats of western grasslands and will continue to impact the Northern Rocky Mountains. However, addi- the remainder of these systems with potentially tional research will he necesvary to understand complex ecological consequences. However, it is ecological roles of small mammals innative bunch- difficult to oredict the ~otentialinloacts that ex- -grass habitats, and comparative studies will be otic plants may have on small mammal commu- imperative to deciphering the complex ways in nities hecausc little is known about small mam- which small manunal community composition and mal ecology in intermountain grasslands. Our small mammal roles are changing in these sys- rcsults emphasize the habitat specificity of mon- tems as a result of exotic plant invasions. tane voles and the habitat flexibility of deer mice in thc lowelevation intermouutain grasslands and Acknowledgements leadus to hypothesize that generalist spccies such as deer mice may he favored over habitat and di- Kersy Foresman and Rudy King reviewed earlier etary specialists such as voles and shrews given drafts of this manuscript. Rudy King provided invasions of such exotics as spotted knapweed, statistical advice and greatly assisted in data analy- Icafy spurge, and Eurasian toadtlax species. Bc- sis. We also thank two anonymous reviewers for cause small mammals play important species- their helpful comments on earlier drafts. Todd specific roles in ecosystem functions. changcs- in Musci and Dan Cariveau collected the field data. 3mall mammal \pecics composition will likely We thank The University of Montana for use of affect prcdator communities and the ecological thc Bandy Ranch: the Montana Department of roles that small mammals play with regard to Fish. Wildlife. and Parks for use of Wildhorsc herhiwry. sccdandinsect predation, seed dispersal, Island; and the Baucus family for use of the Sieben etc. This paper presents important baseline in- Ranch.
Literature Cited Drahek. C. M. 1441. Surnnw and iulumn acti~ilyor Ihc montane vole (Micmnr~rmontiinu~) in lhc Ficld. North- Absiirns$. Z.. M. I. Dycr. and P. 11. Harrison. 1079. Compe- wrblSciencc 68:178181. lilion among mall mammals in crpsrirnrnlidly per Flliart.A. G.. S.H.Anderson.and W.A. Hubert. 1097. >lam- turhed area of the shorlgniss prairie. Ecology 60:530~ mill use olihorlgrass praric and aswciarcd riparian 530. hahitat in Wyommg. lntzrmountain Journid of Sciences Alt, 0.. and U. W. Hyndmim. 1986. Rosdbidc gcolopy of 3:117~123. hlunlunii. Mounia~n PTCSSPuhliihing Company. Elli5. L. h.1. C. S. Crnuford. and LI. C. Mullrb, J. 1997. Ro~ Miwoula. \4ontana dcnl communities in native and erotic ripanno veg- Arthur. S. M.. B. F. J. Manly. L. L. hlclhoald, and Ci. W. etation in the Middle Riu Griinde Valley ul crnlriil Garner. 1996,A~sosinghnhitat selection when uuail- Uru Mexico. SorNhwcircrn Naturalist 12:13-19, abilil) changcs. Fcnlogy 77:215-227. Falls. J. R. 1968. Actirlty. Pages 543-570, lii J.A. King (ed.) Rclk. \l. C.. H. U. Smith. and J. Liiumn. 1988. Urc and par^ Biology ofPrro,riyscii.~(Rudrnlia). Special Puhlicil- titioningofrnontmr hahiwrhy iniall mnmmals. Journal lions. Thc Arncrican Society of Mammalog~sti2:)- ol hlammalogy h0:6XX-695. 503. Flillhcl: C. 11.. S. J. Hrady. and hl. S. Knowlei. 1990. Wild- Boae~s.>l. A. and H. D. Smilh. 1979. Diffcicntial hahirat hfe resource trends in the United Slnlo: a lcchnical urilitalion hy sexes of the decmmaose. Ptmm~icu~ document supporting the 2000 USDAFose\t Servre nioniculori~s.Ecology 60:869~875. General Technical Rrpuit RMRS~GTR-33.Rock) Clark. T. W. I973 Distrihulion and reproduction nf shrews Mounlain Rcscarch Station. Fort Collinr. Colorado. in Grilnd Tcnm Salional Park. Wyommg. Nonhivcst Gano. K. A, and W. H. Kickard. 1082. Small mammnls of a Scicncc 17:128131 hitterbrosh-chciagriisb cun~n~unily.Nollhwol Scicncc IDauhenmirs. R. F. I942 An ecolugicid rlody of thc vcpeta~ 56:1L7. tiun uT wulhcil,rcm Washington and adlacent Idiiho. Gram W. E.. E. C. Birnry. N. R. French. and D. M. Swill. Ecological Monographs 133~79 19x2. Structure and reproductiilty of grassland small DzLuach, C. J. 199 1. Pax wcccss and current prospect, 111 miln~miilcommunirics rclarcd to praling~induccd rhc biological control of weeds in the Unilcd Slam changes in vegetati\e cover Jwmd of Milrnmalugy and Canada Niituriil Arcilr Journal 11: 129~J2. 63:24X-200. Doueli~s.R. J. I976 Soadal mlcractinnc and microhahirat Groir,. C. R.. and K. Stccnhof. 1088. l Small Mammals in Native Bunchgrass 115 Hundlry.C. 0. Jr 1999. Deer mouse 1 P~~ruw~scusmiini~ulntus. Murgglcr. W. F.. and W-. L. Stewarl. 1980 Grassland and Pages 575~577.I,, The Smithivniiln Book of Norlh shrubland habilat types nf acitrm Montana. USDA American Mammals ID. E. Wilson and S. Ruff, cds.). Furat Service. ticnrral Technical Rrpun INT~66.In- Smithsonian Institute Presr. Wabhington D. C., USA. tcrmou~itiiinRangc and Enperimmt Station. Ogdcn, Hash. H. S. 1973. Muvrments and food habits of the thei.ochu Utah. elk. M.S. Thais, University ufldaho, Moscow. Idaho. Negus. N. C., and 1. S. tindlcy. 1959. Mammals of Jack~on W. W. Jr., and A. l3onner. 1977 Wald's rrbt as ap- Hole. Wyoming. Journal of Mammalogy 4037 1407. plied to hypothcici in logit analysis. Juuinid of thc Peairon. D. E. 1994. Habitat uie by thc southern red-hacked Atmerican Staliilical Association 72:851-853 vole ~Clerlt~n~~omysgappen): response of an old^ Hodgsun, 1. R. 1972. Local distcibution ofMimxuc ,normnu& growth associated spcciri to SUCCCSS~O~.MS. The- i~lldM.~~enn$~Iv(inicusin southw~strmMontana. Jour~ sis. Lnivcriity of Montana, Missoula. Montilna. nil1 of Mammalogy 53:487-499. . 1999a. Small mammals of the Bitrenoot U\laLional Hoffmilnn. R. S.. and D. L. fi~trie.1968. Aguide to Montana Forest: a litcraturr review and annotated hlhliugra- mimmals: idcnlitication, hahitat, distribution and ahun- phy. UDSA Forril Service, Gcnrrid Technical Repun dance. Umversity of Montana, Missoula. Monlanii. RMRS-GTR~Z?.Rocky Mountain Research Station, USA. Fun Collin,. Culwado. Koehlcr. D. K., and S. H. Anderson. 1991. Hdhitilt use and -. 1999h. lkci muuae predation on the Biologicid Con- rvud selection of smilll mamnlali ncdr a sagehn~~hl trol Agent. Urophomspp.. Introduced to Control Spot- crrhted aheatgias inte~facein ioulhraatern Idaho. red Knapweed. Nonhv,estern Nalurdist 80: 26~29. Grria Basin Naturalist 51:2-19-255. Koplin. J. R.. and R. S. Hofhmn. 1968. Hahirat uverhp and Pearson, D.E., K. S. McKcl\cy. L. F. Ruggicru. 2000 Non~ competitive exclusion in voles (,Wicmius). American target effects of an introduced biological cuntml agent Midlimd Naturalist 8ll.194-507. on dccr moue ecology. Orcologia 122:121-128. Kolpimaki. E.. V Koivunen. and H. Hakkilraiocn. 1996. mi^ Pearson. 11, t.,and L. F Ruggicro. In ievicw. Trilnsect icr~ crohahital use and behavior oivules tinder wcabel and \us grid mimgemenn fur sampling small mammal raptor predation rirk: prcdalur fisilitation'! Brhai iim comnruniliss. Jounlal of Wildllfe Management. Ecology 7:30. Parson, D. F., and Y. K. Ortcga. In press. An indirect dis~ Krebi. C. J. 1990. Popularion cycles mirited, Juurnal of perval pathway for spotrcd knilpaeed ~ecds\iil deer V1,lammalugy 77:X-24. mice and great~hornedowli. Camdim Fisld~Naturiilist. Lxey C. 1989. Knapweed managcmrnt a decadc of change. Plopper. C. E. 1968. Insular and cmainland pupulationr of in Fay P. K., Lacey R. J. (cds.1 Knapwccd Sympo- Pemngscu n,iiriiculniu at Flathcad Lake, Montanil. icun~Proceedings EB45. Plant and Soil Dzpanment Masws The~is,Uniwrsity of Montana. Misoula. and Extension Scrvice. Montana Slats Univcrrity. Montma. Boreman, Montana 1-7. Pykc, 13. A. 1986. Demographic responses oiBm,nir.t reitonini 1,orriwn. E. L.. and D. R. Johnson. 1973 Densitv chance and ,redlings nf .4grqnro,i ipunen to gra~ingby small mammals: occurrence and rcbcrity of grazing. Journal dEculogy 74:739-7% Randall. J. A.and R. E. Johnion. 1979. Population densities meadow rolc (Microrurpcnri~~lvnnicu~r)in hlucgras and habitat occupancy by Micmtus longicu~lr~sand and prairic hahitilts Journal or Mammalogy 65:60(1- M. mmrunus. Juumal of Mammalogy 60217-219, 606. IUedf~cld,J. A, C. J. Krcbs. and M. J. Tditt. 1977. Competi- Lynchc. D. 1955. Ecolngy of Ihr aspen giovcland in Glacier tion hstnccn Perontpm, rnnriiculoticr and Microruc Cuunty. Montana. Eculugical Mooogrnphs 25:32 1311. ioii',iienr/ii in grilsshnds of coastal Britirh Columbia. Manly. B. F. J, L. I.. McDonnld. and D. L. Thomai. 1943. Juumal of Animal Ecology. 46:007-616. Resource Sclcctiun by Animals. Chapman and Ilall. Kickard. W H. 1960 T3c distribution of small mammals in London, Lnitcd Killgdum. relation to thc climax vegetation moliiic in eawm MacCrilckcn. 1. Ci.. D. U: Uresk. and R. M. Haosen. IYX5. Washinglun and cnortlicm Idaho. Ecology 41 :99-106. Rugcrs. L. E.. and J. D. Hedlund. 1980. A cornparkon of small rnamrnal populations occupying thrcc dihtinct ihruh~ ateppc communities in riiitern Oicgon. Northwcit Scicncr 5d: 183- 186. Rmcnstock. S. S. I996 Shnhgraislmd small inanma1 and vsgstatio~~rrhpunses to rat fium graling. Journal of Range Managcmrnt 49: 19Y-203. \I L. I1 \I . " \I I..I I. ..I.. , ,! Ih, .< .I... : p, :,- Ruggiero. I.. F. D. E. Pearson. idS. E. Hcnry. 1998. char^ I,.I.. ,I. 1 ,ll~l,.,l,_l . I. .,I I i.LI.., I, I, aclrristici ofAmcrican marten dcn sites in Wyoming. \ : .. 1.. , I ,,I..' ,I>, !: :. .. .\ I :.I. Jo~malor Wildlifc Management 62663~673. SAS lnslitutc. 1990. SASIST.4Tuiers guide. Ver 1 16 Peanon. 0rtcg;t. McKehey, and Ruggirn~ Seaplr, S. W. 1985. Parrerms of small mammal microhahitat Tyiei. K. M' 1992. Vesrtillion associated with two alizn plant utihriltmn in ccdar glade and deciduuus lurest hahi- sprciss in afescue graslandin Glacicr National Park. tat. Journal of Milmmalog) 66:22-35. Montana. Great Basil, Niiluralisl 52:lXY 193. Shrlry. R. I.., S. J. Jocobi, and M. F. Carpinelli. 1998. Diw- Tyscr, R. W.. and K. m'. Key. 1988. Sputtrdknapwecd in natural bution. biology and rnanagemeni oidifirc knapiveed area fcicue grasslands: an ecolugiciil ilisr,m~enl. (Cnirnureo difiisu) and spotted knapaeed IC~ntuu~ Nurth*est Scicncs 62: 15 1-160. ira r,iociiiovii). Weed Technolug) 12353~362. Tyrer. R. Mr., and C. A. Work). 1992. Alicn flora in grasi- Sicg. C. H.. D. W. Uresk. and R. hl. Hanssn. 1986. Seasunill lands adjacent to roads and trail corridurs in Glacicr diets ol'dccr mice on hentonlts mine spoils and sage- Nutionill Park Montana (U.S.A.). Conservation Bi- hrmh griaslands in Soulhcastem Montana. Kurthwerl ology h:253-262. Scicncs 60:X-89. Wmt. N. E. I996 Strategies for mainlenancc and repair of Spencer. A. W.. and 11. Psttus. 1966. Hiihilal prcfcrsnces of biotic cocnmunity dixersity onmngelandi. Pagci 326~ five sympauic qxcies nf long-tailed hrrv,. Ecology 216 In Szaro. R. C.. iind Johnsron, D. W. (cds.). l7:676-683. Biodivcriityin managed landscitpes: thew) i~ndprilc~ Stichq. P. F. 1961. Kange of tough fescue (Fesruco vcahnlln tlce Oxford Unireriity Prcsi. New York, USA. Ton.) In Momma. Proceedings of theMontana Acild~ Woll'f. 1.0..and R. D. Dueser. I986 Nuncompelitivc coo- en,v of scicncc 20: 12-17, islence bctwccn Pemmxyiu specie? and C/eriiriononi,s Slocckcr. It. E. 1972. ComprLinrc relations hstaeen sympil~ gapprri. Ciinildian Ficld-Naturalist 100: I 86- 19 1 uic popularionr of roles IMicrorm niontunui and ,W. Limmerman. E. G. 1965. A cornparison of hahirar and food pe,i,is~ivnnicu~).Journal of Animal Eculugy 11:3I1 of two species of Mirrorus. Journal of Mammalop ~729 - 46:605-612. Tisdalc.E. W. I961 Ekological changes in the Palouse. North- a,e\t Scmtce 35:134~138. Recerved I1 Jrtnuan 2000 Acc?pled 25 October 2000 Small Mammals in NaLive Bunchgrass 117