Habitat use by singing voles and tundra voles in the southern Yukon

Oecologia (Berlin) (19t35) 6ó:410 416 Oecologiaí.q¡ Springer-Verlag 1985

Habitat use by singing voles and tundra volesin the southern YukonCarlos Galindo and Charles J. KrebsDepartment olZoology and lnstitute of'Animal Resource Ecology, University olBritish Columbia, Vancouver 8.C., Canada, V6T-lW5

Summary. We investigated how far cornpetitive interactionsinfluence the use of habitats and relative abundance of twospecies of Mit'rotus in the southwestern Yukon. We workedin the ecotone between alpine tundra and subalpine shrubtundra where populations of singing voles (Microtus miur-ns) and tundra voles (M. oeconomus) overlap little.

We removed tundra voles lrom shrub tundra on onelive-trapping area to look at the effect on the contiguouspopulation of singing voles in alpine tundra. The removalof tundra voles did not affect the distribution or relativeabundance ofsinging voles. The spatial distribution ofthesespecies and their movements within habitats suggest thatthey have a strong habitat preference.

Populations of small mammals in the area are extremelydynamic and the relative importance of competitive interac-tions may change as density varies. At present we haveno evidence that competition affects habitat use in M. ntiur-us.

In areas where prairie voles (Microtus ot'hrogaster) andmontane voles (Microtus montanus) are absent, meadowvoles (M. pennsylttanit'ars) occupy a wider range of habitats(Findley 1954). This early observation, consistent with theemerging view of the importance of interspecific competi-tion in natural communities, led Findley (1954) to suggestthat competition among closely related microtines (Arvico-lidae) was at least partly responsible lor such distributionalpatterns. Since then. a number of studies, some on pairsof microtine species, have been undertaken to look intothe role that interspecific interactions play influencing thedistribution and abundance of these species (Grant 1972a,1978). The morphologic and dietary similarities among mi-crotines make them potential candidates lor a strong com-petitive interaction.

A number of microtine species use habitats similarly.Many species occupy contiguous areas along well-dehnedecotones between structurally different habitats (Koplinand Hoffman 1968; Guthrie 1971; Stoecker 1972; Iversonand Turner 1972; Morris and Grant 1972; Turner et al.1975; Hawes 1976; Randall 1978). However, while moststudies have suggested that competition for space mediatedthrough aggressive behavior is responsible for the distribu-tional pattern, the evidence has been unconvincing. Fur-

thermore, other studies have found no negative effectsamong pairs of species (Hawes 1976; Krebs 1977) or havesuggested that in some areas habitat preferences may besulflcient to explain the patterns documented (Douglass1976).

We studied two morphologically similar species of mi-crotines: tundra voles and singing voles. Tundra voles (Mi-crotus oeconomus) occur in moist tundra areas, sedge andcottongrass marshes (Banfield 1974;' Youngman 1975). Inthe area of study they have been lound in alpine and subal-pine tundra and marshes as well as in spruce forest (Krebsand Wingate 1976). Singing voles (M . miurus), on the otherhand, are mainly confined to alpine tundra (Banfield 1974;Youngman 1975) but have also been recorded at subalpinetundra, marshes and spruce forest (Krebs and Wingate1976).

During 1981, we studied the habitat use by these speciesand tested experimentally the hypothesis that competitiveinteractions influence their habitat distribution and relativeabundanCe. If interspecific interactions actively influencethe spatial distribution of these species, the hypothesis pre-dicts that removal of one species will result in a habitatchange and density increase by the other species.

Study urca

The field work was carried out during the snowlree seasonfrom April to September 1981, in the Kluane Ranges, 7kilometers south of the south corner of Kluane Lake,Yukon (61 N, 138 W). We worked in two vegetation zones:alpine tundra and subalpine shrub tundra (Hoefs et al.1975), which we considered to be different habitats.

Alpine tundra occupies the mountain tops above ap-proximately 1500 m. Dwarfed vascular plants, moss andlichens dominate this community. Shrubs are very scarceor completely absent. Some of the common plant speciesare heaiher (Cassiopc). dwarl willow (Satix spp.¡. ur.ii. lr-pine ( Lup inus), blueberry (V ac c in ium), bearberry (A r c t o s t a-phylos), dwarf birch (Betula) and a variety of sedges andgrasses.

Below the alpine tundra and above the treeline, from1100 to 1500 m, the subalpine shrub tundra forms a broadbelt characterized by tall shrubs (up to 3 m), mostly willow(Salix glauca and S. reticulata) and dwarf birch (Betulaglandulosa).In some areas the cover is very dense, .formingimpenetrable thickets. Vascular plants such as heather (Cas-siope) and grasses (Festuca) are common in the understory.Olfprint requests /o. Carlos Galindo

431

Methods

Habitat use patterns

We used snap-trap lines to determine where the species live Num.b,e,

and how abundant they are. These lines consisted of 20 rn¿ri¡'¿uars

stations spaced at 10 m intervals with three 'museum spe- causht

cial' traps per station (Krebs 1964; Krebs and Wingate1976). The i.up, *.r. baited with peanut butter and ríere f;;1,?S,"left in position for two days. Lines were trapped throughoutthe summer.

Liue-tr apping te chnique

We set two long, rectangular live-trapping grids: a controland an experimental. Each occupied nearly 7 ha with 102'Longworth' live-traps arranged in a 6 x 17 grtd. The dis-tance between traps was 10 m. The grids were located atthe ecotone between alpine tundra and subalpine shrub tun-dra, with approximately half the traps in each habitat. Thegrids were set perpendicular to each other. The closest trapsbetween grids were 40 m (129 ft) apart. We set the trapsfor two nights every second week and left them locked openbetween trap sessions to allow voles to move in and outfreely. The traps were provided with whole oats (as bait)and cotton batting (as bedding). All animals captured forthe first time were ear-tagged with numbered fingerling fishtags. Every trap session we recorded: species, tag number,sex, breeding condition, weight and trap location. Thebreeding condition of individuals was assessed in the follow-ing way: for males the position of their testes was recordedas either scrotal (breeding) or abdominal (not breeding).For females we recorded if their vagina was perforate ornot, nipple size (small, medium, large), pubic symphysiscondition (open, slightly open, closed) and evident pregnan-cies.

Exper i men t a I man i pu la ! ion

From the experimental grid we removed all tundra volesevery trap session, starting in the second one (June 29)and continuing throughout the summer. The aim of theremoval experiment was to reduce the population of tundravoles and keep the shrubby habitat vacant for the adjacentpopulation of singing voles.

On the live-trapping grids we estimated the proportionof shrub cover (Salix glauca) within a circle of 6 m diametercentred around each trap station. Two perpendicular diam-eters were used in each circle to estimate cover by the line-intercept method (Mueller-Dumbois and Ellenberg 1974).We classified stations under five categories of shrub cover:100o/o , 7 50Á , 5004, 25o/o and 0oA .

We used simple measures of niche breadth and nicheoverlap (Levins 1968; Colwell and Futuyma 1971; South-wood 1978; formulas 13.30 and 13.31 p445 and p446) locompare the spatial distribution between experimental andcontrol areas.

We used both Cole's (C7) and Hurlbert's (C8) coefh-cients of interspecific association to indicate the strengthof the associations. Both range from - 1 when there is astrong negative association to f 1 when the association isstrongly positive (Cole 1949; Hurlbert 1969; Ratliff 1982).

ALPINE ECOTONETUNDRA

Fig. I. Density indices for singinghabitats for May to August 1981

SUBALPINESHRUB

TUNDRA

and tundra voles in dilferent

Table 1. Resource matrix (number of individuais per trap-nighttimes 1000), habitat niche breadth and habitat overlap for bothspecies ofvoles using snap-traps. (Based on 15 singing voles and27 Íundra voles during 1200 trap-nights)

Species Resource states

Alpinetundra

Ecotone Subalpineshrub tundra

Singing voleTundra vole

27.082.08

29.16

11.1111.1122.22

0.044.4444.44

38.1 9

57.6395.82

Habitat niche breadth:

Singing vole 1.70Tundra vole 1.58

Actual habitafniche overlap (postcompetitive) : c: 0.23

Results

Patterns of habitat use

The ecotone between alpine tundra and subalpine shrubtundra is not sharply defined. It is characterizedby a mosaicof shrub and open meadow patches where singing volesand tundra voles show a remarkably small overlap in theirhabitat use. Captures from the snap-trap lines in these twohabitats show that singing voles occur mainly in the openmeadows of the alpine tundra, but are also caught at theecotone with the adjacent shrubby habitat (Fig.1). Froma total of 15 animals caught in 1200 trap-nights, 13 (87%)were in the open meadows while the remaining 2 (13%)were trapped in the ecotone between both habitats. Nosinging voles were caught in the subalpine shrub tundra.

Conversely, tundra voles occur mainly in the denselycovered shrub tundra where 24 (89%) were caught. Of theremainder, 2 (1%) were in the ecotone while one individual(4oA) was caught in the alpine tundra. The resource matrixfor the snap-trap data indicates that these species have simi-lar habitat niche breadths and that their habitat niche over-lap is small (Table 1).

A more detailed view of the microhabitat distributionof both species at the ecotone is provided by the analysis

432

SINGING VOLES

N umbe rol

Captures

30

20

10

'16

't4

Number '12

of10

250Cover

TUNDRAVOLES

RemovedTUNDRA VOLES 7O

60

50Nu mber

ol 40Captures

3o

20

10

i 4-6

,,'1.2

>5o 25 oShrub Cover

Fig. 2. Microhabitat use by singing voles and tundra voles on thecontrol grid. Singing voles use open areas whiie tundra voles use

areas with higher shrub cover. The expected value is obtained byassigning the captures proportionally to the number of traps ineach shrub category

Tzble 2. Resource matrix (number ol captures per trap-night times1000), niche breadth and niche overlap lor both species of voleson control grid for all summer

Species Resource states: shrub cover categories

MJJASFig. 3. Number of tund¡a voles removed. The average number olindividuals removed declined within trap sessions. Closed circlesindicate the lirst night and open circles indicate the second night.Arrow represents the beginning of the manipulation

the rest were not used. The species are negatively associated(X2:5.17, P< 0.05) and the strength of the association is

moderate (C7:-0.41 , C8:-0.47). There were no trapsaturation problems. Even when numbers were highest inboth species, only 32oA of the traps were used.

Remoual experiment

A total of 35 tundra voles was removed throughout thesummer from the experimental grid. After the removal ofwhat we considered the resident population, an averageof 5.8 tundra voles per session was caught (Fig. 3). Betweenthe first and second night of each trap session the averagenumber of animals caught and removed declined from 4.6to 1.2. This reduction plus the fact that they were untaggedindividuals, indicates that the grid was almost free of tundravoles by the end of every session. Animals caught subse-quently were dispersers from elsewhere. The removal re-sulted in a feduction of habitat use by tundra voles of atleasf 1006 (estimated from the number of captures com-pared with that on the control).

Spatial distribution

If tundra voles actively influence the spatial distributionof singing voles, then the removal of the former will resultin the use of subalpine shrub tundra (niche shift; ecologicalrelease) by singing voles. This ecological release would be

reflected in any of the following ways. First, an increasein the number of traps shared by both species, due to theuse of traps by singing voles of those sites from which tun-dra voles had been removed. Second, an increase in thefrequency of captures of singing voles in those traps for-merly used by tundra voles. Third, a shift in the kind oftrap station used from open sites to trap stations underhigher shrub cover. As a consequence of this shift bothmicrohabitat niche breadth and overlap should be largerin the absence of the suspected competitor (virtual or pre-competitive niche breadth).

Were more traps shared by both species on the experi-mental grid? On the experimental grid 24 singing voles and35 tundra voles were caught a total of 59 and 35 timesrespectively over 2142 trap nights. From a total of 102

traps, 20 were used by Microtus miurus alone, 20 by Micro-tus oeconomu,r alone and only 3 by both species. The restof the traps were not used. Fisher's exact test did not reject

1005025

Singing vole 53.22Tundra vole 8.40

61.62

51.58 12.82 8.65 0.0 126.2'l47.61 64.10 73.59 95.23 288.93

99.19 7 6.92 82.24 95.23 415.20

Habitat niche breadth

Singing vole 2.78Tundra vole 3.97

A c t ua I hat¡itar niche overlap (po stcompetitive) c : 0. 3 6

of lrequency of captures at live-traps under different coverconditions on the control grid (Fig. 2).

Singing voles were captured more frequently at live-trapstations surrounded by little (25%) or no shrub cover(Gadj.:32.28, df.2, P<0.01). Few animals were caughtaf trap stations under more lhan 50oA shrub cover. Tundravoles also differed in their habitat use significantly fromrandom (Gadj.:25.55, df . 2, P<0.01) but in the oppositedirection. They mostly used traps with high shrub coverand avoided traps in open habitat. Tundra voles use a high-er number of microhabitats than singing voles on the con-trol grid. The actual microhabitat niche overlap (postcom-petitive) is larger than the habitat overlap but the differenceis relatively small (Table 2).

There were more traps used exclusively by either of thetwo species than traps shared by both. Throughout the sum-mer 23 individual singing voles and 33 individual tundravoles were caught on the control grid a total of 69 and110 times, respectively, over 2142 trap-nights. From a totalof 102 trap stations, 20 were used only by singing voles,39 were used only by tundra voles, 6 by both species and

aIt.... ...IIIo

.._........_o-o..\ \a/o

-.4.. -...,..

433

Observed

40Number

of 30Captures

20

)so 2s oShrub Cover

Fig. 4. Microhabitat use by singing voles on experimental grid.The use of shrub categories by singing voles is similar in bothgrids regardless of the absence of tundra vo1es. Expected valueswere obtained by assigning captures proportionally to the distribu-tion in the control grid

Table 3. Resource matrix (number of captures per trap-night times1000), niche breadth and niche overlap for both species of voleson experimental grid for all summe¡

Species Resource states: shrub cover categories

sI_l

\N

,/

/.

suba¡pineshrubl

,/ ."

:do'

CONTBO!

:

Fig. 5. Movements between grids. Closed figures represent singingvoies and empty figures represent tundra vo1es. The subindicesindicate the sequence ofcaptures. Both species move between gridsto their corresponding habitats

grid with a similar habitat to that from which they came.Such areas involved a greater distance to travel than thatrequired to move to a different habitat on the same grid.This is more apparent in the movements of singing voles,since the areas of alpine tundra between the grids werefarther apart. In alpine tundra the closest distance betweentraps in different grids was 150 m (485 ft).

Three singing voles travelled between grids. Two adultfemales moved from the control grid to the experimentalgrid, a distance of 230m (744 ft) and 220m (712 ft). Anadult male poved from the experimental to the controlgrid and back, travelling2l0 m (680 ft) and 260 m (841 ft).The trip back was done overnight (Fig. 5). The movementswere recorded during June and early July, and by mid-Julythese individuals had disappeared from both grids.

Similarly, three tundra voles moved from the controlgrid to the experimental grid, where they were removed.Two adult females and an adult male travelled 130 m(421 ff),160 m (518 ft) and 170 m (550 ft) respectively. Thetrip by the male was done overnight. These movementswere recorded during June (Fig. 5).

Mouements within grids

On the control grid tundra voles occupied two differentareas, the west half (Rows 9 to 17) and a small part ofthe east corner (Rows I Io 4). Trap stations in these areashad high shrub cover. They were absent from the openmeadows area between these two which was occupied bysinging voles. In spite of the relatively small distance be-tween the areas (50 m), none of the tundra voles caughtin one area were caught in the other area. This distancebetween shrub habitat patches is relatively small when com-pared with: a) the distance between the most widely sepa-rated capture points of individuals within each area (10to 80 m) and b) the distances travelled by individuals fromgrid to grid within the same subalpine shrub tundra habitat(greater than 130 m).

Therefore, both species can move to the adjacent habi-

rvvv4

10

EXPEFIMENfAL

100755025

Singing vole 27.47 35.14 25.79Tundra vole 9.15 7 .93 19 .84

36.12 43.07 45.63

0.061.9061.90

88.4098.82

187.22

0.00.00.0

Habitat niche breadthSinging vole 2.94Tundra vole 2.23

Vir tual habitat niche overlap (precompetitive) c : 0.37

the hypothesis of independence (P:0.34). The coefhcientsof association give a moderate negative association (C7:-0.42, C8: -0.42). More traps were used by both specieson the control grid. But in fact the number of traps usedby Microtu,s miurus alone and shared wjth Microtus oecono-mus is not signihcantly different between experimental andcontrol grids (Gadj. :1.42, df.1, P> 0.10).

Did singing voles more frequently use traps from whichtundra voles had been removed? Singing voles were caught54 times in traps used exclusively by them and 5 timesin shared traps. Their frequency olcaptures at shared trapsis lower in the experimental grid than in the control(Gadj.:6.17, df. 1, P<0.02), and not higher as was pre-dicted.

Did singing voles increase their use ol the microhabitatfrom which tundra voles had been removed? Accordingto the shrub categories used, the proportion of singing volecaptures after the removal is not different from that onthe control grid (Gadj.:1.51, df. 2, P>0.30) (Fig. a). Acomparison of the actual (control) and virtual (experimen-tal) microhabitat niche breadth of singing voles shows al-most no difference between them. Accordingly, actual andvirtual microhabitat overlap are strikingly similar (Table 3).

In summary, the removal of tundra voles had no effecton the spatial distribution of singing voles.

Mouements between grids

A few individuals ofboth species were caught on both grids.Members of each species moved to the area in the new

434

tat, since the movements normally recorded are far largerthan the distance required to get tá the contiguous habitát. t':::1:

VOLES

Population dynamics

Two or more species compete if an increase in any oneresults in a corresponding decrease in the others (MacAr-Ihur 1972). Even when the predictions from this definitiondo not discriminate other interactions from competition(predation, parasitism), it is useful to analyze the effectof one species on another. If one of these species has anegative effect (exploitation, interference) on the second,

removal of tundra voles should result in an increase innumbers of singing voles. The increase would be producedeither by higher reproductive rate, lower mortality, higherimmigration, lower emigration, or a combination of these

parameters. However, since singing voles did not utilizethe area vacated by tundra voles, there should be no differ-ences in their demographic parameters between grids. Infact, the following analysis shows that the populations wereremarkably similar.

Numbers

Is the population of singing voles higher in the absence

of a contiguous population of tundra voles? On the controlgrid the population of overwintered singing voles beganat a very low density during June. The onset of breedingwas not documented, as some females were already preg-nant when trapping began. The first juveniles were trappedduring the second week ofJuly. From then on, the popula-tion increased continuously until the first week of Sep-

tember, reaching 37 individuals per hectare when the trap-ping ended. Throughout the summer a total of 23 individ-uals was caught; 8 adults (6 females and 2 males) and 15juveniles (10 females and 5 males).

The singing vole population was also very low on theremoval grid during June. With the recruitment ofjuvenilesthrough July, they increased at a faster rate than on thecontrol. Nevertheless, they remained at a lower level thanon the control during August and September. A total of21 individuals was caught with similar sex and age ratiosto those of the control population: 8 adults (5 females and3 males) and 13 juveniles (8 females and 5 males).

A comparison of the minimum number of animalsknown to be alive on both grids (Fig. 6) shows that thepopulation of singing voles on the control grid increasedto a higher level than that on the removal grid. This is

contrary to what would be expected if tundra voles werehaving a negative influence on singing voles.

The total number of singing voles caught throughoutthe summer is very similar on both grids (Gadj.:0.08, P>0.70) as are the age and sex ratios (Gadj.:0.28, df . 2, P>0.50). The adult classes were pooled to obtain a joint class

with an expected frequency greater than 5. There was abias in the sex ratio towards females on both grids. Onthe control grid 7 5oA of the adults and 66oA of the juveniles

were females. On the experimental grid 620A of the adultsand 61oA of the juveniles were females.

Reproduction

There are no evident differences in singing vole reproduc-tion between grids. Females were already pregnant by thefirst trapping session on both grids. On both experimental

.-. CONTROLo...o REMOVAL

t-.\aao\ o/

s , \/.-

16

l4

12

10MinimumNumberAlive

TUNDRAVOLES

M¡n¡mumNumberAlive

16

14

12

10

4

2

MJJASFig. 6. Changes in minimum number alive lor both singing and

tundra voles on the experimental and control grids. Singing volesreached a higher density on the control grid

and control grids the hrst juveniles were caught in the scc-

ond week of July. Therefore the onset of breeding was syn-chronous. There seemed to be two litters during the breed-ing season of 1981, the first born in mid-June and the sec-

ond in mid-July.On both grids half of the tagged adult females were

caught only once or twice during June and early July anddisappeared for the rest of the summer. We considered themtransient individuals. Because the rest were caught in 3 to5 trapping sessions we considered them residents. Followingthe breeding condition of the individual resident females,

we estimated a minimum of 3 litters was produced on eachgrid. The number of juveniles recruited during July andAugust per adult female for the control and experimentalgrids was 2.5 arrd 2.6 respectively, or 5 and 6.5 when onlythe resident females are taken into account.

Discussion

Recent reviews have attempted to assess the role of competi-tion by looking at the experimental evidence across diversetaxa (Schoener 1983; Connell 1983). In the following dis-cussion we will concentrate on studies carried out with mi-crotines (Arvicolidae) in an attempt to look for patternsin a more homogeneous subset.

A number of factors may influence the distribution ofthese two microtine species. While the well-def,rned patternof non-overlapping populations could result from interfer-ence competition (Guthrie 1971.), the structural differenceof the vegetation from open meadows to a dense cover

.-. CONTROLo...o REMOVAL i

lI-i'o.,.

.o'o'

of shrubs affects many other biotic and abiotic componentsof the community. The results from this study do not sup-port the hypothesis that the habitat distribution of singingvoles is influenced by competitive interactions. The use oftrap stations indicated that both species showed a stronginverse distributional pattern in relation to habitat structure(shrub cover), but the removal of one species did not affectthe distribution of the other. The resource matrix analysissupports this result. If competitive interactions were in-fluencing the spatial distribution of singing voles, then thehabitat niche breadth of this species should be larger inthe absence of tundra voles. Similarly, virtual niche overlapshould be larger than actual niche overlap. None of thesepredictions were supported. Both the habitat niche breadthand overlap of singing voles are very similar regardless ofthe removal of tundra voles. Furthermore, individuals ofboth species moved large distances within and betweengrids. This clearly shows that they could have moved tothe adjacent habitat. Nevertheless, they were never caughtthere.

Several studies have dealt with habitat separation andinterspecific interactions among microtines. Koplin andHoffmann (1968) experimentally tested the hypothesis thathabitat segregation of meadow voles (M. pennsybanicus)and montane voles (M. montanus) is maintained by inter-specihc interactions. Montane voles made exploratorymovements into the grassland after the removal of meadow

.voles. However, the experiment was criticized since the ex-perimental plot was fenced and not grazed by bison, andhad a 4 to 5 times higher density than the control (Grant1972; Connell 197 5; Birch 1979). The differences in habitatuse by montane voles in experimental and control plotsmay have been a consequence of the different populationdensities and not ofthe absence ofmeadow voles. Stoecker(1912) did the inverse experiment of removing montanevoles and found that meadow voles used the drier habilafin the absence of the former species. Neither studies wereable to determine whether individuals that moved into thevacated areas remained there or were merely transients.Both studies lacked replicates and adequate controls. Thestudies suggested behavioral interactions as a mechanismfor the habitat segregation.

Competitive interactions have also been invoked to ex-plain habitat segregation among other pairs of related spe-cies. A number of distributional studies have suggested thatactive interference results in habitat segregation by red-backed voles (Clethrionomys) and meadow voles (Miuotus)(Morris 1969; Grant 1972). The former live in woodlandsand the latter in adjacent grasslands (Morris 1969). Experi-ments inside enclosures showed that Clethrionomys oc-curred more often in the grassland in the absence of Micro-lers (Grant 1969) and fhal Microtus occurred more oftenin the forest in the absence of Clethrionomys (Morris andGrant 1912). Both colonization of grassland habitat by red-backed voles and of spruce forest by meadow voles havebeen documented (Iverson and Turner 7972;Turner efal.1975). In both cases the colonization took place duringwinter, once the breeding season was over, and animalsdisappeared from the colonized habitat at the onset ofbreeding. The authors suggested that the decrease in in-traspecific aggression during the non-breeding season per-mitted these species to coexist. However their conclusionis based on a correlation. No experiment was done toconhrm a causal relationship.

435

Hawes (1976) showed thal Microtus oregoni was morerestricted in its habitat use after the introductton of M.townsendíi. However, in the same study the removal of M.oregoni did not influence habitat utilization by M. longicau-dus. Similarly, Krebs (1977) found no evidence of negativedemographic effects between Miuotus pennsyluanicus arrdM. ochrogaster living in the same grassland areas.

Alternatively, there is support for the hypothesis thathabitat segregation in microtines is related to habitat prefer-ences. Differences in use ofhabitats related to physiologicaltolerances have been documented for some species. Mon-tane voles are often found in drier places (even shrub-steppecommunities) while meadow voles occupy more mesic habi-tats (Murie 19'7 1 ; Banfield 197 4). Water balance in meadowvoles is not well adapted for dry situations (Getz 1963).Douglass (1976) demonstrated that these species show dif-ferent habitat preferences and concluded that such prefer-ences are probably the most important factor responsiblefor their segregation. He suggested that the relative impor-tance of social interactions and habitat preferences maychange along a gradient of habitats. At one extreme, whenhabitats are sharply dehned, the distribution will be theresult of habitat preferences, whereas if the habitat is uni-form behavioral interactions may have a stronger influence.

In addition, habitats may become more similar not onlybecause of the resemblance of their structural characteristicsbut because of the effect of population density. As the den-sity of a species in a habitat increases, its suitability willdecrease, reducing its difference from less suitable habitats(Fretwell and Lucas 1970; Merkt 1981). If populations ofsinging voles and tundra voles fluctuate, their interspecificinteractions may be contingent on density changes. Cyclesin populations of tundra voles have been documented inFinland and Alaska (Taitt and Krebs 1985). In the latterlocality peak densities reached approximately 70 to 80 volesper hectare. While there have been no studies on the demog-raphy of singing yoles, Krebs and Wingate (1976) reportedchanges in their abundance from year to year. Furthermore,the total disappearance of both species from the study siteduring the winter of 1981 indicates that they do fluctuatein the area. At present, however, neither species has beenfound in the area in higher numbers than those reportedhere (Krebs and Wingate 1976), and therefore densities mayrarely be high enough for competitive interactions to occur.

Acknowledgements. We thank the Natural Sciences and Engineer-ing Research Council of Canada, Consejo Nacional de Cienciay Tecnologia de Mexico, the Arctic Institute of North Americaand the Department of Indian Allairs and Northern Developmentfor financial support. We thank Margriet De Poorter, Stan Boutin,Scott Gilbert and Jean Carey for their field assistance and an anon-ymous reviewer lor suggestions.

References

Banfieid AWF (1974) The mammals of Canada. Univ of TorontoPress, Toronto

Birch LC (1979) -Ihe effect of species of animals which share com-mon resources on one another's distribution and abundance.Fortschr Zo ol 25 : 197 -221

Cole LC (1949) The measurement of interspecific association. Ecoi-ogy 30:411-424

Colwell RK, Futuyma DJ (1971) On the measurement of nicheb¡eadth and overlap. Ecology 52:567-576

Connell JH (1975) Some mechanisms producing structure in natu-

436

ra1 communities. In: Ecology and evolution ol communities,Cody and Diamond (eds). Belknap Press, Harvard Univ Press

CambridgeConnell JH (1983) On the prevalence and relative importance of

interspecific competition: evidence from held experiments. AmNat 122:661 696

Douglass RJ (1976) Spatial interactions and microhabitat selec-

tions of two local1y sympatric voles, Microtus montanus aÍdMicrotus pennsyluanicus. Ecology 57 :346-352

Findley JS (1954) Competition as a possible lactor in the distribu-tior of Microtzs. Ecology 35:418-420

Fretwell SD, Lucas HL (1970) On territorial behavior and otherfactors influencing habitat distribution in birds. I. Theoreticaldevelopment. Acta Biotheriol 19:16 36

Getz LL (1963) A comparison of the water balance of the prairieand meadow voles. Ecology 44:202-207

Grant PR (1969) Experimental studies of competitive interactionin a two species system.I. Microtus and Clethriononys speciesin enclosures. Can J Zool 41:1059-1082

Grant PR (1972) Interspecific competition among rodents. AnnRev Ecol Syst 3:79-106

Grant PR (1978) Competition between species of small mammals.In: Populations of small Mammals under Natural Conditions,Snyder DP (ed) Pymatuning Lab Ecol Symposium Vol 5 Pitts-burg, Pa:38-51

Guthrie RD (1971) Factors regulating the evolution of mic¡otinetooth complexity. Sonderdruck ats Z. f. Sáugetierkunde Bd. 36H. 1, pp 37-54

Hawes DB (1976) Experimental studies of competiton among fourspecies of voles. PhD thesis. Dept Zoology, UBC Vancouver

Hoefs M, McT.Cowan I, Krajina VJ (1975) Phytosociological anal-ysis and synthesis of Sheep Mountain, southwest Yukon Terri-tory, Canada. Syesis 8: 125-228

Hu¡lbert SH (1 969) A coefficient of interspecific association. Ecol-ogy 50:1 9

Iverson SL, Turner BN (1972) Winter coexistence of Clethríonomysgapperi and Microtus pennsyluanícus in a grassland habitat. AmMidl Nat 88:440_'445

Koplin JR, Hoffman RS (1968) Habitat overlap and competitiveexclusion in voles (Microtus). Am Midl Nat 80:494 507

Krebs CJ (1964) The lemming cycle at Baker Lake, NorthwestTerritories, during 1959-62. Arctic Institute of North AmericaTech Paper 15

Krebs CJ (1977) Competition between Microtus pennsyluanicuí andMicrotus ochrogaster. Am Midl Nat97:4249

Krebs CJ, Wingate I (1976) Small mammal communities of theKluane Region, Yukon Territory. Can Field Nat 90:379-389

Levins R (1968) Evolution in changing environments. PrincetonUniv Press, Princeton, NJ

MacArthur RH (1972) Geographical ecology. Harper and Row,NY

Merkt JR (1981) An experimental study ol habitat selection bythe deer mouse, Peromyscus maniculat¿¿s, on Mandarte Island,BC. Can J Zool 59:589 597

Morris RD (1969) Competitive exclusion between Miuotus andClethrionomys in the aspen parkland of Saskatchewan. J

Mamm 50:291 301Morris RD, Grant PR (1972) Expeimental studies of competitive

interaction in a two species system .IY . Microlus and Clethrion-ornys species in a single enclosure. J Anim Ecol 41 :275-290

Mueller-Dombois D, Ellenberg H (1974) Aims and methods invegetation ecology. John Wiley & Sons. NY

Murie JO (1971) Behavioral relationships between two sympatricvoles (Microtus): relevance to habitat segregation. J Mamm52:181 186

Randall JA (1978) Behavioral mechanisms of habitat segregationbetween sympatric species of Mícrotus: habitat prelerence andinterspecihc dominance. Behav Ecol Sociobiol 3:187 202

Ratliff RD (1982) A corection of Cole's C7 and Hurlbert's C8coefficients of interspecihc association. Ecology 63 : 1605 1606

Schoener TW (1983) Field experiments on interspecific competi-tion. Am Nat 122:240-285

Southwood TRE (1978) Ecological methods. 2nd ed Chapman andHa1l, London

Stoecker RE (1972) Competitive relations between sympatric popu-lations of voles (Microtus montanus and M. pennsyluanicus).J Anim Ecol 47:311 329

Taitt MJ, Krebs CJ (1985) Population dynamics and cycles. In:Biology of New World Microt¿¡s. Tamarin RH (ed) (In press)

Turner BN, Perrin MR, Iverson SL (1975) Winter coexistence ofvoles in spruce lorest: relevance of seasonal changes in aggres-sion. Can J Zool53:1004 1011

Youngman PM (1975) Mammals of the Yukon territory. Nat Mu-seums ofQanada, Publs Zoology 10

Received November 5, 1984