patterns of apparent extirpation among isolated ... · persistence of animals in heterogeneous...

37

Journal of Mammalogy, 84(1):37–54, 2003

PATTERNS OF APPARENT EXTIRPATION AMONG ISOLATEDPOPULATIONS OF PIKAS (OCHOTONA PRINCEPS) IN THE

GREAT BASIN

ERIK A. BEEVER,* PETER F. BRUSSARD, AND JOEL BERGER

Program in Ecology, Evolution, and Conservation Biology/314, University of Nevada,Reno, NV 89557, USA (EAB, PFB, JB)

Department of Biology, University of Nevada, Reno, NV 89557, USA (PFB)Wildlife Conservation Society, Moose, WY 83012, USA (JB)

Present address of EAB: Forest and Rangeland Ecosystem Science Center, Biological ResourcesDivision, United States Geological Survey, 3200 SW Jefferson Way, Corvallis, OR 97331, USA

We conducted exploratory analyses to examine the relative roles played by natural andanthropogenic influences on persistence of a montane mammal. We revisited historicallocations of pikas (Ochotona princeps) within the hydrographic Great Basin during sum-mers of 1994–1999. Seven of 25 populations (28%) reported earlier in the 20th centuryappeared to have experienced recent extirpations. We assessed causative agents of faunalchange using several alternative, but not mutually exclusive, hypotheses. Higher probabilityof persistence was correlated with greater area of talus habitat at local and mountain-rangescales, higher elevation, more easterly longitude, more southern latitude, lack of livestockgrazing, greater distance to primary roads, and wilderness management. However, only areaof habitat in the mountain range, maximum elevation of talus habitat, and distance toprimary roads appeared in the most parsimonious model of persistence when we usedAkaike’s information criterion model-selection technique. These results suggest that relax-ation of montane faunas may occur more rapidly than previously expected; that biogeo-graphic models of species occurrence can be refined by including more proximate factors(e.g., grazing status, proximity to roads); and that habitat-based approaches to modellingvertebrate trends should be accompanied by field data because population loss can occurwith no apparent change in habitat.

Key words: American pika, biogeography, climatic effects, extinction, grazing, hydrographic GreatBasin, land management, montane alpine habitat, Ochotona princeps

Persistence of animals in heterogeneouslandscapes may vary as a result of habitatcomplexity or quality (Johnson 1975; Pul-liam 1988), initial population size (New-mark 1995; Pimm et al. 1988), spatial ar-rangement of patches and dispersal frequen-cy (Hanski 1991, 1998), frequency of andresponse to catastrophes (Mangel and Tier1994), and life-history characteristics ofspecies (Newmark 1995; Tracy and George1992). Few studies have assessed the rela-

* Correspondent: [email protected]

tive contributions of natural and current an-thropogenic factors to risk of extinction,and those that have done this have oftenexamined broad taxonomic groups (Cebal-los and Brown 1995; Cole et al. 1994; Kerrand Currie 1995; Newmark 1995) or lackedrecent empirical data on population trends(e.g., Ceballos and Brown 1995). A taxo-nomically coarse-scaled approach providesa summary of factors most frequently act-ing on species within a clade; however, itdoes not allow an in-depth examination ofextinction dynamics (e.g., interactions of

38 Vol. 84, No. 1JOURNAL OF MAMMALOGY

several threats). We attempted to investigatepersistence of a discontinuously distributedmammal species in relation to biogeograph-ic, climatic, and current anthropogenic fac-tors.

Isolated mountaintops in western NorthAmerica have been considered montanehabitat islands, and patterns of species oc-currence have been investigated in birds,butterflies, conifers, and mammals of theGreat Basin, as well as in montane mam-mals of New Mexico (e.g., Brown 1971,1978; Cutler 1991; Johnson 1975). Al-though montane mammal faunas of theGreat Basin have been used as models toexamine biogeographic patterns for nearly3 decades, there is no clear consensus onfactors that most strongly influence persis-tence in these communities. Island area, de-gree of isolation from local colonizingsources and mainland habitat, and habitatdiversity have been assessed, but it is oftendifficult to incorporate synecological fac-tors (e.g., competition) or life-history char-acteristics and habitat affinities of individ-ual species in models (McDonald andBrown 1992).

Brown (1971, 1978) first analyzed GreatBasin mountaintop faunas and concludedthat previously continuous populations havebeen undergoing slow but continuous extir-pations without concomitant colonizationsbecause intervening inhospitable lowlandhabitats were believed to limit dispersal be-tween ranges. Brown’s (1971) initial anal-ysis suggested an increase in montanemammal species richness as island (moun-taintop) area increased, but subsequent dis-coveries of additional species occurrencesand reanalyses have suggested a decreas-ingly important role of area (Brown 1978;Cutler 1991; Grayson and Livingston 1993;Lawlor 1998).

Most recently, Lawlor (1998) found thatdistributions of most species of mammalswith respect to mountaintop isolation andarea do not differ from random distributionsand suggested that pikas (Ochotona prin-ceps) exhibited one of the least nonrandom

distributions (Lawlor 1998). Lawlor (1998:1128) concluded that most species of thecurrent Great Basin montane fauna are ‘‘ex-tinction-resistant woodland species capableof considerable movement among mountainranges,’’ a result consistent with paleonto-logical evidence of recent (1,000–4,000years ago) cross-valley dispersal by thewoodrat (Neotoma cinerea; Grayson andMadsen 2000). Thus, results of Graysonand Madsen (2000) and Lawlor (1998) havechallenged, at least for some species,Brown’s (1978) assumption of no recentcolonizations by montane mammals of theBasin. However, Great Basin populations ofO. princeps have frequently been treated asbeing fully isolated at present (Grayson andLivingston 1993; Hafner 1994; Smith1974a).

North American pikas (O. princeps andO. collaris) likely evolved from one ormore Asian species of pikas that immigrat-ed across the Bering Strait (Dawson 1967;Kurten and Anderson 1980). Fossil evi-dence suggests that ochotonid lineages havepersisted in North America for at least500,000 years (Mead 1987). Ochotonareached its maximum distribution duringthe Wisconsin glaciation (Grayson 1987;Mead 1987). Subsequent warming duringthe mid-Holocene forced pikas to retreat tohigher latitudes and elevations (Grayson1987; Hafner 1993, 1994). This retreat setthe stage for the current relict, disjunct dis-tribution of pikas in the IntermountainWest.

Pikas provide a model system for inves-tigating recent faunal change in montanemammal species for several reasons. Pikasare diurnal and relatively easy to detect(Smith 1974a). When used collectively,sightings, species-specific calls, and pres-ence of fresh haypiles or feces provide ahigh probability of detecting the species ina single visit (Hall 1946). Thus, the prob-lem of possibly undetected populations ofmontane mammals mentioned by Graysonand Livingston (1993), Lawlor (1998), andMcDonald and Brown (1992) is reduced

February 2003 BEEVER ET AL.—EXTIRPATION DYNAMICS OF PIKA POPULATIONS 39

with pikas. In addition, life history, habitatspecificity, and behavior of pikas lendthemselves to testing several alternative hy-potheses regarding extinction dynamics(Hafner and Sullivan 1995; Smith and Wes-ton 1990). Because of their obligate asso-ciation with discontinuously distributed ta-lus habitat, vulnerability to warmer temper-atures, and tendency towards philopatry, pi-kas may act as early sentinels of changes inother montane mammal species. As an ex-ample, it has been predicted that of 13 spe-cies of montane mammals, only 3 would belost from more mountain ranges than pikaswould be if there were a rise of 38C in glob-al temperature (McDonald and Brown1992). Compared with other montane mam-mal species, many aspects of pika biologyhave received extensive study (Smith andWeston 1990). With knowledge of pikas’paleontological history, behavior, ecology,and dispersal, reproductive, and thermal bi-ologies, alternative hypotheses can thus beassessed with greater confidence and so-phistication.

Using the model of faunal flux occurringover tens to hundreds of millennia as abackdrop, we sought to determine if GreatBasin populations reported in the 20th cen-tury (Appendix I) have suffered extirpa-tions over decadal timescales. If extirpa-tions had apparently occurred, we investi-gated biogeography (isolation and area),thermal biology, and current anthropogeniceffects simultaneously as possible causes offaunal change across all populations.

We used persistence data to test 7 hy-potheses (2 biogeographic, 2 climatic, and3 human influence). The minimum-area(1st) hypothesis (Lomolino 1986) assumesthat populations track a biogeographicmodel driven by extinction events and pre-dicts that only populations of Great Basinpikas existing in areas having more talusthan some minimum threshold should per-sist. The maximum-isolation (2nd) hypoth-esis (Lomolino 1986) assumes that popu-lations may be modelled by an incidencefunction driven by colonization events and

predicts that populations closest to othercolonizing sources (either the Sierra Neva-da or Rocky Mountain ‘‘mainland’’ or near-est-neighbor populations) will be most like-ly to persist, other factors being equal. A3rd hypothesis predicts that populations athotter, drier sites should have lower prob-abilities of persistence if thermal biology ofpikas dictates persistence of Great Basinpopulations. Similarly, a 4th hypothesissuggests that sites in ranges with the high-est-elevation talus should have highest pikapersistence if pikas are being forced tomove up-slope.

If humans have influenced pika popula-tions recently through recreational shootingor related disturbances, then probability ofpersistence should increase as distance fromthe nearest road increases. This human-dis-turbance (5th) hypothesis assumes thatroads (especially primary roads) providethe easiest access to taluses with pikas, po-tentially facilitating disturbance at thesesites. If feral or native herbivory adverselyaffects pikas, then populations in grazed ar-eas will be more likely to become extirpat-ed than are populations in ungrazed areas—our 6th, ungulate grazing, hypothesis. Fi-nally, we compared persistence of pika pop-ulations among current management juris-dictions. Assuming that more restrictivemanagement permits fewer activities pos-sibly detrimental to pika populations, wetested our 7th hypothesis that persistencewas higher in wilderness areas than in non-wilderness Forest Service and Bureau ofLand Management lands.

MATERIALS AND METHODS

Study area and fieldwork.—We adopted forour field sampling the hydrographic definition ofthe Great Basin (Grayson 1993) because it hasthe most clearly distinguishable boundary. Weexcluded data on sites in the Carson Range andWasatch Front (e.g., from Durrant 1952) becauseof their connection to the Sierra Nevada andRocky mountains. Although this definition cre-ates a domain different from that of previousbiogeographic studies of Great Basin mammals,it was suitable for our purposes because we were


TABLE 1.—Mountain ranges in the Great Basin of North America from which pikas had not beenpreviously reported and in which pikas were not detected in additional surveys. Letters in parenthesesrefer to identifications of ranges in Fig. 1. Range area denotes area above 2,286 m elevation.

Mountain range

Elevation (m)

Minimum MaximumLatitude

(8N)Range

area (km2)

Clan Alpine Mountains (A)Dogskin Mountain (B)Diamond Mountains (C)Humboldt Range (D)Jarbidge Mountains (E)Seven Troughs Range (F)Snake Range (G)Sonoma Range (H)Spring Mountains (I)

1,1661,5221,7661,2651,8231,2781,6101,330

853

3,0382,2893,2352,9973,2892,3733,9812,8643,633

39.4–39.939.8–39.939.5–40.140.2–40.741.7–41.940.3–40.738.7–39.640.5–41.035.6–36.5

1,14979

760697126335

2,109643

2,168

interested in comprehensively investigating per-sistence of isolated populations in a relativelyarid region. We obtained historical records ofGreat Basin pika populations from a variety ofsources (Appendix I). Although dispersal withinand among adjacent talus fields undoubtedly oc-curs to some degree within all extant popula-tions, dispersal capability of pikas is believed tobe limited to distances less than most distancesbetween our study populations (Grayson andLivingston 1993; Hafner and Sullivan 1995;Lawlor 1998; Peacock 1997; Smith 1974a).Therefore, exchange of individuals between sites(even between adjacent ranges) is not likely tohave occurred during the 10- to 100-year periodwe analyzed, and we thus considered each site(which we sometimes refer to as a ‘‘popula-tion’’) to be an independent replicate (Appen-dices I and II).

We visited sites during each summer (May–August) from 1994 to 1999. During 1996 wevisited all but 1 of Hall’s (1946) pika sites tominimize effects of interannual variation. In aneffort to augment sample size, we also searched9 ranges from which pikas had not been reportedpreviously but which had occupiable talus hab-itat (rock diameter 0.2–1.0 m) at elevations com-parable to those of Hall’s sites (Table 1).

We searched at each site for 8 h within andaround talus habitat and recorded locations ofpika individuals and talus patches without pikasby use of a handheld Global Positioning Systeminstrument. Area of talus searched was based ondensity and arrangement of talus in the mountainrange (Beever 1999). Within each talus area, wewalked parallel contour transects approximately15 m apart until the entire talus area was can-

vassed. Exhaustive searches of talus within 3.25km of historic locations were conducted exceptat 5 areas with vast amounts of appropriate talushabitat (0.2–1.0 m rock diameter—Tyser 1980)nearby, at which we sampled similar nearbypatches of habitat.

We recorded the presence of calling individ-uals, active haypiles, and individuals sighted toprovide evidence for the presence of pikas at asite. Presence of fresh fecal pellets helped leadus to other types of evidence. Adoption of stan-dardized methods such as that described abovefacilitates coarse-scale comparison of many sitesacross a broad geographical area. We also talliedthe number of defecations within 1 m of the pathwe walked as we moved within and between ta-lus areas (Appendix II). We classified each fecalpile as being produced either by cattle, nativeungulates (pronghorn antelope, elk, mule deer,or bighorn or domestic sheep), or horses (feraland domestic).

Extirpation or absence of a mobile animalfrom an area is difficult to demonstrate unequiv-ocally (Diamond 1987) and should be assertedwith great caution. We thus urge the reader toconsider our failures to detect populations as ap-parent extirpations. Because pikas may becomemore crepuscular and difficult to detect whenweather is hotter and drier and when a site con-stitutes more marginal habitat, we used severaltechniques to increase our confidence that fail-ures to detect pikas reflected extirpation events.To increase chance of detection, we visited pop-ulations believed to be extirpated on 2 separateoccasions and consulted other researchers whohad recently censused Great Basin pikas. To en-sure that pikas had not evaded detection by mov-


TABLE 2.—Results of univariate regressions on persistence (logistic regression) of Great Basinpikas. To avoid multicollinearity, only variables indicated were entered into AIC analyses, as de-scribed in Table 3. Results for variables followed by parentheses compare values of the level of thevariable in parentheses against values from all other sites. Chi-square values and their associated P-values are Wald chi-square tests on the coefficients, and test the hypothesis that the log odds ratiofor the independent variable (‘‘Factor’’) is 0.

FactorModel

coefficient x2

Prob-ability

Log likeli-hood ratio

test x2 Probability

Habitat in rangea (large) [Habitat]Maximum elevation, local talusa [Elevation]Nearest-neighbor (not in same range) isolationHabitat in range (medium)

28.4100.0060.261

16.150

0.004.753.300.00

0.970.030.070.98

19.1016.2413.7111.92

,0.0001,0.0001

0.00020.0006

Pika-equivalent elevationMinimum elevationLongitudeDistance to nearest primary roada [RoadDist]LatitudeGrazing statusa [GrazStatus]Management jurisdiction (wilderness)

0.0090.001

21.2070.712

20.936213.850

2.996

7.026.885.404.605.250.004.85

0.0080.0090.020.030.020.970.03

10.8410.1210.07

8.167.816.616.26

0.0010.00150.00150.0040.0050.010.012

Maximum daily temperature, Augusta {PRISM data}[MaxTemp]

Habitat in immediate area (large)Habitat in immediate area (medium)Habitat in range (small)Distance to nearest roadManagement jurisdiction (United States Forest Services)

20.0052.3002.485

14.2000.4761.792

3.343.933.220.001.772.27

0.070.0480.070.980.180.13

4.634.463.942.972.842.49

0.0310.0350.0470.0850.090.11

Mainland distanceAnnual precipitation {PRISM data}Total ungulate scatDistance to nearest populationCattle defecations

20.0080.0000.0020.0120.001

1.741.100.260.210.60

0.190.290.610.650.81

2.061.220.290.230.06

0.150.270.590.630.80

a Variable entered into models compared with AICc values; abbreviations in brackets appear in Table 3.

ing up-slope since the time of the original his-toric record, we searched in potential habitat allthe way to the summit and used the summit asa vantage point to locate (with binoculars) othertalus areas to search. Additionally, because pikasvary their daily activity patterns depending onclimate and weather (Smith 1974a; Verts andCarraway 1998), we sampled within 1 h of dawnor dusk (usually both) for populations in hotterareas, including all locations of apparent extir-pations. Pikas rely on vocalizations for conspe-cific attraction (Stamps 1988) and other purpos-es, especially when the talus is disturbed (ref-erences in Smith and Weston 1990; Verts andCarraway 1998). Thus, careful searches for sev-eral hours on taluses at locations of apparent ex-tirpations should have produced at least someevidence (i.e., sighting, call, fresh haypile, orfresh feces) of an extant population. Nonethe-less, despite our precautions, some unknown but

likely small proportion of the populations mayhave been incorrectly categorized as extant orextirpated.

Analytical techniques.—We tested competinghypotheses with univariate and information-the-oretic analyses, employing dummy variables forall categorical variables. To provide a descrip-tive summary here, we compared selected siteparameters of extant and apparently extirpatedpopulations using 1-way analysis of variance(ANOVA). We performed univariate logistic re-gressions on each predictor variable against per-sistence and assessed significance of factors (Ta-ble 2) with likelihood ratio tests (Sokal andRohlf 1995). Variables exhibiting possiblymeaningful relationships (P , 0.15—Hosmerand Lemeshow 1989) to persistence in univari-ate tests were entered into information-theoreticanalyses that involved Akaike’s information cri-terion (AIC—Burnham and Anderson 1998). We


used a less stringent critical value of alpha toensure that potentially important interactionswere not overlooked.

We divided variables with P , 0.15 into bio-geographic, climatic, and anthropogenic cate-gories and used a correlation matrix to identifyinstances of strongly multicollinear (r . 0.7)variables (both within each category and for thefinal model). We entered only the best univariatepredictor of persistence from each pair of cor-related variables in AIC analyses. The AIC mod-el-selection technique determines the most par-simonious model to describe biological phenom-ena and can be especially useful to comparemodels of varying complexity (i.e., numbers ofvariables). Values of AICc, a derivative of AICthat accounts for small sample sizes (our n 5 25sites), were used to compare strength of com-peting models.

Biogeographic effects.—Biogeographic per-spectives on montane mammals of the Great Ba-sin have followed the traditional island bioge-ography model, viewing mountaintops as islandsisolated from mainland populations in the SierraNevada or Rocky Mountains by a sea of inhos-pitable low-elevation valleys. To test for biogeo-graphic effects, we first used data on isolation-from-mainland distances of areas above 2,300 m(7,500 ft) from Lawlor (1998). For populationsnot included in Lawlor’s (1998) work, we de-fined distance from mainland as the measureddistance between the study site and the 2,286-mcontour of either the Sierra Nevada or RockyMountains (based on affinities indicated by Haf-ner and Sullivan’s (1995) genetic analyses) on a1:3,168,000-scale United States Geological Sur-vey base map. We additionally used geographicinformation systems to measure isolation of pi-kas from the absolute closest population and thenearest population not in the same mountainrange (‘‘nearest-neighbor’’ distance).

Because talus habitable by pikas occupies dif-ferent proportions of different ranges, we alsoanalyzed persistence using categorizations of ta-lus area at local (within 0.8-km radius) andmountain-range scales (Beever 1999). From pre-vious literature, these spatial scales are knownto correspond to dispersal distances commonlyreported within a season and over longer time-scales (Hafner 1994; Hafner and Sullivan 1995;Peacock 1997; Smith 1974a; Smith and Weston1990).

Climate change and pika thermal biology.—

To examine possible population losses related toclimate change and pika thermal biology, we ex-amined relationship of persistence to several val-ues: minimum elevation at which pikas were ob-served or reported; elevation midpoint of the re-ported population; and ‘‘pika-equivalent eleva-tion’’ (Hafner 1993), a variable represented bythe equation

E (m) 5 14,087 2 56.6 (8N) 2 82.9 (8W).

In the equation, E is the minimum elevation (inm) of O. princeps populations throughout NorthAmerica, and 8N and 8W are the latitudinal andlongitudinal positions (in decimal degrees) ofthe site. We also recorded maximum elevationof talus within a 3-km radius of each site andwithin the mountain range (or within 5 km forsites not in ranges) to test whether ability to mi-grate up-slope affected persistence. To more di-rectly assess climatic influence, for each site weobtained parameter-elevation regressions on in-dependent slopes model (PRISM)-modeled val-ues of annual precipitation and means of dailymaximum temperature for June, July, and Au-gust, averaged across 1961–1990 at a 4-km res-olution (PRISM—Daly et al. 1994). Here we re-port only the best univariate predictor of persis-tence from among these PRISM variables.

Anthropogenic and ungulate effects.—Wemapped study sites onto 1:100,000-scale mapsof management jurisdiction. Jurisdictional statusof pika sites (Appendix II) reflects the federalresource agency currently responsible for man-aging the area where populations occurred. Wil-derness sites had resource conservation as a pri-mary management objective and included 5 wil-derness areas administered by the United StatesForest Service, 1 wildlife refuge (sites 21 and22), and 1 wilderness study area administeredby the Bureau of Land Management (site 20).To examine more direct anthropogenic effectson persistence of populations, we measured dis-tance of each location with a historical popula-tion to the nearest road and to the nearest pri-mary road (Appendix II). We defined roads asany travel-way other than a trail that appearedon 1:100,000-scale topographic maps and pri-mary roads as roads that can be traveled in sum-mer without a 4-wheel-drive vehicle. Primaryroads appear distinct from more primitive roadtypes on topographic maps, and we verified thedistance between each site and nearest primary


road during our travel to and in the vicinity ofsites.

To gauge effects of introduced and nativegrazers, we characterized each site as grazed orungrazed (Appendix II) based on managementmandates and on conversations with local re-source personnel knowledgeable of the long-term grazing histories at our sampling areas.Grazed sites reflected instances in which signif-icant (i.e., more than occasional or trespass)grazing by cattle, native ungulates, or horses oc-curred at or near taluses for more than half theperiod since the historic pika record. We as-sumed that fecal tallies roughly indexed the rel-ative intensity of recent use by each group neartalus areas, and we used these tallies to confirmcharacterizations of grazing status. Because re-sponse to disturbance in alpine and subalpine ar-eas depends critically on severity of disturbance(Chambers 1997), we regressed persistence oncounts of defecations of cattle, horses, other un-gulates, and all ungulates combined.

RESULTS

Patterns of persistence and extirpa-tion.—We could not detect pikas at 6 of 25resampled populations (Fig. 1). At 1 addi-tional site that was comprehensivelysearched (Cougar Peak; Appendix II), wecould detect only 1 individual. Thus, weclassified that population as functionally ex-tirpated. This assumption of functional ex-tirpation was corroborated with a resurveyof the Fort Bidwell site, at which we de-tected 1 individual in 1996 but none in1999.

The 7 apparent extirpations occurred pri-marily in livestock-grazed areas (7 of 7 in-stances); at sites not in mountain ranges orin ranges with small amounts of talus hab-itat (6 of 7); in areas having small amountsof talus habitat within 0.8 km (4 of 7); andon lands administered by the Bureau ofLand Management (4 of 7 instances; Ap-pendix II). Six of the 7 apparent extirpa-tions, including all 3 recorded locations ofO. p. schisticeps from Nevada, occurred inthe northwestern corner of the Great Basin.Apparently extirpated populations were lo-cated at lower absolute (ANOVA, F 5

14.5, d.f. 5 1, 23, P 5 0.0009) and pika-equivalent elevations (ANOVA, F 5 15.7,d.f. 5 1, 23, P 5 0.0006) and nearer toprimary roads (ANOVA, F 5 8.1, d.f. 5 1,23, P 5 0.009) than were extant popula-tions. The 9 additional ranges with high-elevation talus we surveyed (Table 1) didnot contain pikas, although searches werenot comprehensive in the Jarbidge, Spring,or Snake ranges. Pikas have not been foundin an additional 27 other Great Basin rangesinvestigated by other researchers (C. Ray,in litt.).

Biogeographic and climatic hypothe-ses.—Extinction patterns in Great Basinpika populations were consistent with pre-dictions of the minimum-area (1st) hypoth-esis but not with predictions of either amaximum-isolation or compensatory pat-tern (Fig. 2a; see Lawlor 1998:1119 forgraphical predictions for each hypothesis).In contrast to populations not in ranges orin ranges with little habitat, populations inranges having moderate or large amounts oftalus usually remained extant (17 of 18 in-stances). Amount of talus habitat in therange was the strongest univariate predictorof population persistence (Table 2). Thisfactor also appeared in the most parsimo-nious AIC-selected model (Table 3). Dis-tance to mainland (Rocky Mountain or Si-erra Nevada) or absolute nearest colonizingsources did not significantly predict persis-tence (P . 0.15; Table 2). Nearest-neighbordistance strongly predicted persistence butin a manner contrary to expectations: extir-pated sites had closer nearest-neighborsthan did extant populations (X 5 19.9 ver-sus 42.5 km). We assumed this result to bespurious, and did not enter the variable intoinformation-theoretic analyses.

Extirpated populations were located atsignificantly lower minimum-encounter andpika-equivalent elevations. Five of the 7sites with apparent extirpations had amongthe most negative residuals of elevationwhen regressed against latitude (Fig. 2b).Extirpated sites received 19.6% less annualprecipitation (ANOVA, P 5 0.30) and aver-


FIG. 1.—Locations of archaeological and contemporary records of pikas (Ochotona princeps) andpopulations presumed to be extirpated in the Great Basin–Mojave Desert region (as defined in Brus-sard et al. 1998), United States. Although the region boundary differs slightly from the hydrographicGreat Basin in the far southwestern and southeastern corners, no sampling locations occurred in ornear areas of discrepancy. Lettered mountain ranges possessed talus habitat at high elevations, butwe found no pikas there during exploratory searches (also see Table 1). Lighter shading denotesmountain ranges defined by McLane (1978). Numbers refer to site number in Appendix I. Extantlocations in Utah were reproduced from Durrant (1952). These sites were not sampled in this study(see text) but are presented for greater completeness. Sites on the eastern slope of the Sierra NevadaMountains were too numerous to include with clarity.


FIG. 2.—a) Relationship between degree ofisolation from mainland (i.e., Sierra Nevada orRocky Mountains) and categorical estimation oftalus habitat within the mountain range for ex-tant (open circles) and apparently extirpatedpopulations (closed circles) of pikas (Ochotonaprinceps). Numbers refer to site numbers in Ap-pendix I. b) Relationship between minimum el-evation at which pikas were encountered or (ifextirpated) reported and latitude for extant (opencircles) and apparently extirpated populations(closed circles) of O. princeps in the Great Ba-sin. Regression is described by the equationminimum elevation (m) 5 9943.3 2 188 (lati-tude, 8); r2 5 0.61. Dashed lines denote 95%confidence interval for the linear regression, andnumbers near selected locations refer to sitenumbers in Appendix I.

aged daily maximum temperatures 7.7–10.2% higher than those of extant sites dur-ing June, July, and August (P 5 0.04 to0.08), consistent with predictions of the 3rdhypothesis. Maximum elevation of talusnear sites (4th hypothesis) predicted pikapersistence better than any other climate-

related variable did (Table 2) and, amongmodels we compared, appeared in all of the9 models with the greatest strength of evi-dence in support of them (Table 3).

Human-influence hypotheses.—Greaterdistances from the nearest road (of anytype) correlated with persistence (P 5 0.09;Table 2). Increasing distance to the nearestprimary road exhibited even stronger cor-relation with persistence, significantly (P 50.004; Table 2) increasing probability ofpersistence and appearing as 1 of 3 factorsin the best model in information-theoreticanalyses (Table 3), consistent with predic-tions of the 5th hypothesis. Although extir-pated populations were significantly closerto primary roads (X 5 1.75 versus 4.63 km;P 5 0.009) than were extant populations,we detected abundant evidence of direct hu-man influence at only 3 of 7 sites no longersupporting pikas. At 1 site, about half of thetalus area was excavated and used as a‘‘borrow pit’’ for road maintenance. At an-other site, the talus area apparently wasused extensively as a dump site. Carvingsin aspen tree trunks suggested extensive hu-man use of Smith Creek since at least the1930s, and we found numerous gun shellson taluses there.

Results of the effects of presence and in-tensity of grazing on pika populations (6thhypothesis) were mixed. Grazing statusshowed a significant (P 5 0.01) negativecorrelation with persistence (Table 2), as all7 extirpations occurred in livestock-grazedareas (n 5 14 sites), compared with 1 ex-tirpation out of 11 ungrazed sites. Further-more, the highest-ranking model that in-cluded grazing status in AIC analyses,which was 3rd-highest overall, had delta ,2 (Table 3), suggesting that the model has‘‘substantial evidence’’ to support it (Burn-ham and Anderson 2001:114). However,the highest-ranking model had 2.4 times asmuch support as did this model (Table 3).Absence of the grazing-status variable inthe 2 highest-ranking models may have oc-curred in part because ungrazed sites werelocated at higher minimum elevations (P ,


TABLE 3.—Comparison of ability of various models to predict probability of population extirpation,ranked in order of increasing AICc values (Akaike’s information criterion, adjusted for small samplesize). Abbreviations of model variables refer to variables with a superscript letter a in Table 2.‘‘Model’’ contains the variables included in each model, maximum log-likelihood (log L) refers tothe value of the maximized log-likelihood over unknown parameters (given the data and model), kis number of estimated parameters in the model, AICc is value of AIC corrected for small samplesizes, Delta is difference of AICc for a particular model from the best model, and Weight is a factorthat reflects how good a model is relative to the rest.

ModelLog-likelihood(22)·(log L) k AICc Delta Weight

Elevation Habitat RoadDistElevation HabitatElevation RoadDist GrazStatusElevation RoadDistElevation Habitat RoadDist GrazStatusElevation

4.8528.4146.625

10.3004.701

13.408

434352

14.85215.55716.62517.44317.85917.953

0.0000.7051.7732.5913.0073.101

0.26270.18470.10830.07190.05840.0557

Elevation GrazStatusElevation Habitat MaxTempElevation Habitat GrazStatusHabitat RoadDistElevation MaxTempElevation RoadDist MaxTempElevation GrazStatus MaxTempHabitat MaxTempHabitat

10.8928.2048.410

11.48212.353

9.87010.30617.29720.728

344334432

18.03518.20418.41018.62419.49619.87020.30624.44025.273

3.1833.3523.5583.7734.6445.0185.4549.588

10.421

0.05350.04920.04440.03980.02580.02140.01720.00220.0014

0.05) and tended to have larger amounts ofnearby habitat (Appendix I). Neither totalnumber of ungulate defecations nor cattledefecations alone at sites predicted popu-lation persistence (Table 2). Neither thenumber of horse defecations nor the num-ber of other ungulate defecations showedany relation to persistence of pikas.

Lands administered by the Bureau ofLand Management experienced more pop-ulation extirpations (4 of 6) than did landsadministered by the United States ForestService (2 of 8 populations) or wildernessareas (1 of 11 populations; Appendix II),consistent with predictions of the 7th hy-pothesis.

DISCUSSION

This is 1 of few documented instances inwhich a medium- to small-sized mammal inNorth America has apparently experiencedextirpations at a bioregional scale over thespan of only a few decades (55–86 yearssince last record—Appendix I). In contrast

to previous analyses of montane mammals,this analysis provides a step toward merg-ing explanations for persistence that spantemporally from human generation times toecological scales to evolutionary time-scales. As such, we were able to investigatethe interaction of natural stochasticity pre-sent in a rare, discontinuously distributedorganism with external climatic and human-mediated influences.

Because it was not obvious to us a prioriwhich combination of biogeographic, cli-matic, and anthropogenic variables wouldbest explain metapopulation dynamicsacross the Great Basin, we used univariateregressions as a filter to select variables toenter into AIC analyses. Our naturally lim-ited sample size was insufficient to addresseven a truncated set of combinations of allpotentially meaningful variables. However,Burnham and Anderson (2001) recommendthat biological, rather than statistical, con-siderations drive selection of the combina-tions of variables to include in models to


be compared with information-theoreticmethods. Consequently, we caution that ourapproaches should be considered explorato-ry, and they do not provide cause-and-ef-fect or confirmatory evidence about factorsthat may explain the apparent extirpations.

Biogeographic hypotheses.—Because oftheir obligate association with discontinu-ously distributed talus habitat, pikas exhibitgreater isolation than does perhaps any oth-er montane mammal in the Great Basin andare isolated at both within-range and be-tween-range scales. Our finding of no re-lationship between isolation from mainlandor nearest neighbors and persistence, pre-dicted by the 2nd hypothesis, suggests thatcontemporary populations of Great Basinpikas are not a colonization-driven system,a result consistent with previous work (e.g.,Lawlor 1998) and other lines of evidence.Pikas are highly philopatric, and only 25%of all juveniles may attempt dispersal(Smith 1987). Additionally, the likelihoodof successful dispersal across nontalus hab-itat may be low for pikas (Smith 1974a,1974b, 1980). Maximum dispersal distanc-es are typically 3 km (Smith 1974b), withthe vast majority consisting of much shorterdistances. Hafner (1994) estimated thatvery few colonizations across distances of.20 km have occurred over the last 6,000years in the southern Rocky Mountains.Given these observations, natural recoloni-zation of extirpated localities in the GreatBasin (assuming they are still suitable forpikas) under current climatic conditionsseems improbable.

Longer dispersal events may be possible,however, in more mesic, higher-elevationhabitats such as the Sierra Nevada Moun-tains of California (one 2-km dispersal ob-served per year—Peacock 1997) or theRocky Mountains (Hafner 1994). Usingcomparisons of genetic similarity, Hafnerand Sullivan (1995) found that pika meta-populations are separated by 10–100 kmand postulated that maximum dispersal dis-tance for an individual pika would probablybe 10–20 km. The degree to which these

distances are traversed currently by pikas inthe Great Basin could be assessed most pre-cisely by use of combined mark–resight andgenetic techniques (Peacock 1997).

Amount of talus habitat at the mountain-range scale was the strongest univariate pre-dictor of persistence. Additionally, amountof talus habitat occurred in the best modelfor persistence using AIC model selection,lending further support to the 1st hypothe-sis. The importance of habitat area has alsobeen demonstrated at smaller spatial scalesby Smith (1980), who found that all casesof population extinctions occurred on smallor medium-sized patches at Bodie in the Si-erra Nevada of California. The vast differ-ence in spatial scale complicates compari-sons between Smith’s (1980) investigationsand ours, however. For example, differentmechanisms or combinations of factors mayoperate at different scales to produce thesimilar pattern of higher rates of disappear-ance in smaller patches. Whereas habitatarea may influence species richness of birdsor mammals by sustaining greater habitatdiversity (Brown 1978; Johnson 1975), ob-ligate association of pikas with talus pre-cludes the importance of coarse-scale hab-itat diversity for pika persistence. Morelikely, persisting in ranges with minimalhabitat or at locations not in ranges is dif-ficult because potential rescue effects(Brown and Kodric-Brown 1977) are un-likely. Furthermore, in those cases popula-tions are limited to small sizes, which in-creases their vulnerability to stochasticevents (Caughley 1994; Newmark 1995).

Because talus area remains constant overecological timescales, space seems an un-likely direct determinant of extinction overthe short timescales we investigated, exceptto the extent that it forces pikas into asmall-population dynamic (Caughley1994). This latter possibility, of synergisticinfluence with other threats, could well bemagnified if increasing temperatures reducethe effective area of habitable talus in anarea. Occurrence of extirpations without de-tectable change in (talus) habitat abun-


dance, quality, or spatial pattern highlightsthe importance of including a broad rangeof variables in and field-truthing habitat-based models of trend for vertebrate popu-lations.

Because previous biogeographic investi-gations have included only a subset of theavailable Great Basin records for any par-ticular species, they provide an incomplete(and perhaps biased) picture of how speciesrespond to factors driving persistence or ex-tinction within a region. These studies ad-mittedly had goals very different fromthose of this study, though, and have in factcontributed to a better understanding oftrends in montane faunas.

Climatic hypotheses.—Although therehas been increasing awareness of long-termtrends in climatic variables (e.g., changes inglobal temperature), effects on vertebratesare just beginning to be understood. Ourfinding that maximum elevation of talushabitat occurred in all of the 9 highest-rank-ing (yet in none of the 8 lowest-ranking)models for persistence in information-the-oretic analyses (lending support to the 4thhypothesis) suggests that thermal effectshave influenced recent persistence trajecto-ries of Great Basin populations of pikas.Thus, warmer temperatures seem likely tobe contributing to apparent losses that haveoccurred at a pace significantly more rapidthan that suggested by paleontological re-cords. Maximum elevation of talus at localand mountain-range scales relates to cli-matic influence because it denotes how farup-slope pikas can migrate in relativelycontiguous taluses under increased temper-atures. Importance of thermal biology issupported more forcefully by the fact thatextirpations occurred in 3 low-elevation ar-eas in close proximity to high-elevationpopulations that remained extant (i.e., atsites within the Desatoya, Hart, and Rubymountains). In the Ruby Mountains, wesearched Thomas Canyon around elevationsat which pikas were collected in summer1956 and detected abundant pikas at 2,743and 2,895 m elevation but none at 2,375 m.

Climatically induced thermal stress onGreat Basin pikas could influence their dis-tributions in several ways. Rapidly increas-ing temperatures could change the compo-sition or relative abundance of plants in andaround talus areas to a mix with which pi-kas have not coevolved. Quantitative veg-etation data collected over decades wouldbe necessary to address this possibility rig-orously. Alternatively, higher summer tem-peratures could permit less midday foragingtime (Smith 1974a), perhaps preventing pi-kas from gaining sufficient body mass andcollecting sufficient hay to overwinter suc-cessfully. Compared with other montanemammals, energetic demands are exacer-bated for pikas, in part because they are ac-tive year round. Thus, pikas make up to 13haying trips per hour to create average-sized haypiles (Dearing 1997) and need tofill their stomach 9 times daily (Smith andWeston 1990). Alternatively, rises in sum-mer temperature may modify the thermalclimate of talus to the point that the rela-tively low upper lethal temperature (Smith1974a) or capacity of pikas to thermoreg-ulate behaviorally is exceeded. This seemsespecially likely in structurally simple (i.e.,small-diameter rocks of homogeneous size)or shallow talus, where thermal refugia areless well buffered from ambient environ-mental conditions. The relationship be-tween extant populations and climatic fac-tors across the range of pikas in NorthAmerica was suggested strongly by Hafner(1993), who found that species of Ochotonaare restricted to cool, moist microhabitatswithin regions having short summers andfreeze-free periods, long winters, and meanannual precipitation .30 cm. Increasingtemperatures could cause the range of po-tential pika habitat in a mountain range tomove up-slope (Peters and Darling 1985),altering the spatial distribution of the meta-population by rendering unsuitable previ-ously habitable talus patches or by frag-menting higher-elevation talus.

If extirpations of pika populations in thepast few decades were driven by climatic


changes, then rapid climatic alteration sim-ply accelerated (greatly) the rate of popu-lation loss observed through paleontologi-cal evidence in the Holocene and Pleisto-cene periods (Grayson 1987, 1993). Previ-ously extirpated populations, considered‘‘extralimital’’ to current distributions, oc-curred at elevations between 730 and 2,070m and were extant 7,200–34,000 years ago(Grayson 1987, 1993). Latitudinal gradientsin pika distribution and persistence, pre-sumably tied to gradients in climate, haveexisted at least since the onset of the Ho-locene (Grayson 1993; Grayson and Liv-ingston 1993). Similarly, because northernpopulations presumably have been separat-ed from the mainland for shorter time pe-riods, northern and southern populationsmay not exhibit comparable extinction rates(Grayson and Livingston 1993). If this werethe case, the direct prediction would be thatsouthern populations, separated for a longertime period, will exhibit greater populationlosses. However, we found the opposite:probability of persistence decreased signif-icantly (P 5 0.005) with increasing latitude(Table 2). An alternative interpretation ofthis result is that marginal southern popu-lations were extirpated long ago, but north-ern populations are still on the steeper partof the exponential curve of extinction rateover time. This interpretation supports a rel-ict climatic effect on population persistenceand suggests that northern populations hada greater extinction debt (Tilman et al.1994).

Human-influence hypotheses.—Althoughbiogeographic analyses generally have con-sidered Great Basin mountaintops to be oa-ses of low anthropogenic influence becauseof their isolation, at least 12% of tundraecosystems in the western United Stateshad experienced some level of historic-erahuman-induced disturbance by 1978(Brown et al. 1978). The short growing sea-son, variable precipitation, relatively lowprimary productivity, temperature fluctua-tions, high wind speeds, and shallow, weak-ly developed soils of alpine and subalpine

systems can compound effects of distur-bance and make these ecosystems amongthe most difficult to restore (Butler 1995;Chambers 1997).

Distance of pika populations from pri-mary roads significantly predicted popula-tion persistence in univariate analyses andwas a component of the best model usingAIC model-selection techniques, consistentwith predictions of the 5th hypothesis. Be-cause extirpated populations also occurredat lower elevations and at areas with lesshabitat, however, we cannot refute the pos-sibility that apparent influence of proximityto roads was confounded by higher temper-atures and sparser habitat at lower eleva-tions.

All apparent population extirpations oc-curred in areas open to livestock grazing,consistent with predictions of the 6th hy-pothesis. However, grazed areas occurred atlower elevations and had less talus habitatthan did ungrazed areas. These results will,we hope, stimulate further research on therelationship between domestic and feral un-gulate grazing and pika populations. Be-cause pikas exhibit behavior consistent withcentral-place foraging, intensity of pika for-aging generally decreases, and costs asso-ciated with foraging generally increase withincreasing distance from talus (Huntly et al.1986). Furthermore, risk of predation in-creases as pikas must forage greater dis-tances, and pikas can safely use broader ar-eas when artificial talus cover is provided(Huntly et al. 1986). Therefore, cattle orhorses may negatively influence pikas if theungulates graze in areas within 20–50 m oftalus. If significant interaction is occurring,it need not be through exploitative compe-tition for food. Rather, indirect influencecould occur via trampling of soils or veg-etation, which occurs in both cattle (Weltzet al. 1989) and horse grazing (Beever et al.2003).

In contrast, grazing should be implicatedwith great caution when interpreting de-clines of pikas and other talus inhabitants.First, the solid nature of talus rock may pre-


vent direct interaction between large her-bivores and pikas; we observed an activehaypile directly under a well-traveled horsetrail in the Desatoya Mountains and severalhaypiles near other trails. Similarly, cattleappeared to avoid all opportunities to crosstalus, especially on steep slopes. Second,grazing status of very broad areas may cor-respond only loosely with grazing pressurein the immediate vicinity (i.e., within 50 m)of talus areas because steep terrain or rockformations may largely prevent livestock orferal horses from accessing talus margins.Third, whereas cattle and horses primarilyconsume graminoids (Hanley and Hanley1982), pikas are generalist herbivores.Fourth, it may be difficult to model diet andbehavior of generalized herbivores such asthe pika because of possible nutrient-bal-ancing constraints unrelated to interspecificinteractions and inter- and intraspecific var-iability in plant chemistry (Rapport 1980).

Although we failed to reject the hypoth-esis that persistence and extant populationsizes should be highest at wilderness sites(7th hypothesis), this finding is confoundedby greater available habitat at wilderness-area sites. This confounding is not uniqueto our study; Norton (1999) similarly foundthat protected reserve areas are concentrat-ed in the high-elevation and steep, infertileparts of landscapes. In North America, mostexisting national parks were established toprotect scenic and geological (rather thanbiological) wonders, which are often locat-ed at higher elevations and have more talus-producing formations (Wagner et al. 1995).

Management influence on Great Basinpikas is difficult to assess for other reasonsas well. Although Hart Mountain (n 5 2sites) was established as a Refuge in 1936,7 of our 11 wilderness sites did not receiveformal wilderness designation until 1989.Thus, because no surveys were performedacross sites before 1989, it remains uncer-tain how wilderness management has af-fected Great Basin pikas during the 20thcentury. Furthermore, in the Great Basin,the Bureau of Land Management was as-

signed jurisdiction of lower-elevation lands,which were often in poorer condition thanwere Forest Service lands at the inceptionof the Bureau of Land Management in1934.

Pikas in the Great Basin appear to haveundergone significant losses (.25% of his-toric sites) during the last half century. Theinclusion of some anthropogenic as well asnatural variables in models selected usingAIC methods in this exploratory analysissuggests that current anthropogenic influ-ences (i.e., grazing status, proximity toroads) may have combined with factors act-ing over longer timescales (e.g., climate,habitat area) to produce fairly rapid appar-ent extirpations of pikas in the Great Basin.Thus, anthropogenic factors should be usedto refine biogeographic and metapopulationmodels of species incidence because theycan affect the fundamental biogeographicprocesses of immigration, extinction, andevolution.

However, there are caveats to this inter-pretation. Long-term studies are needed toverify these patterns of apparent extirpationand to more firmly establish their causes.Furthermore, to fully evaluate the relativeutility of management actions for restora-tion or conservation of vulnerable popula-tions in the Great Basin, we recommendmanipulative experiments to partition nat-ural variability more clearly from anthro-pogenic influence. Although some combi-nation of factors investigated here may af-fect pika populations in other geographicregions, genetic divergence among subspe-cies resulting from long-term isolation de-mands that the domain of inference for ourfindings be restricted to the Great Basin.

ACKNOWLEDGMENTS

B. Pritchard, T. Nevius, B. McProud, J. Speck,C. Olson, J. Pence, M. Bulthius, and R. Gormleyof the United States Forest Service and Bureauof Land Management provided assistance in de-termining grazing status (especially historically)at sites. We thank D. Grayson, D. Hafner, S. Jen-kins, B. Longland, S. Mensing, B. J. Verts, and


an anonymous reviewer for commenting on ear-lier drafts. We also thank A. Smith, N. Huntly,L. Otterman, and R. Hess for sharing useful in-sights at various stages of the research, J. Land-messer and R. Beever for providing field assis-tance, M. Huso for assistance with AIC analy-ses, N. Slade, C. Ludwig, and L. Carraway forassistance with specimen dates, and B. Mc-Menamy and B. Elston for creating Fig. 1. Spe-cial thanks are due to M. Peacock and C. Ray,who provided the greatest assistance, advice,and helpful feedback. Thanks are also due to theUnited States Forest Service (Humboldt, Toiya-be, and Fremont national forests) and Bureau ofLand Management in Nevada and Oregon, Haw-thorne Army Ammunition Depot, and HartMountain National Wildlife Refuge for permit-ting pika surveys on lands under their jurisdic-tion. This work was supported by grants to E.A. Beever from the Nevada Biodiversity Re-search and Conservation Initiative, Nevada Ag-ricultural Experiment Station, American Muse-um of Natural History, and the KosciusczkoFoundation.

LITERATURE CITED

BEEVER, E. A. 1999. Species- and community-level re-sponses to disturbance imposed by feral horse graz-ing and other management practices. Ph.D. disser-tation, University of Nevada, Reno.

BEEVER, E. A., R. J. TAUSCH, AND P. F. BRUSSARD.2003. Characterizing grazing disturbance in semiaridecosystems across broad spatial scales using ecosys-tem components. Ecological Applications 13:119–136.

BROWN, J. H. 1971. Mammals on mountaintops: non-equilibrium insular biogeography. American Natu-ralist 105:467–478.

BROWN, J. H. 1978. The theory of insular biogeogra-phy and the distribution of boreal birds and mam-mals. Great Basin Naturalist Memoirs 2:209–227.

BROWN, J. H., AND A. KODRIC-BROWN. 1977. Turnoverrates in insular biogeography: effect of immigrationon extinction. Ecology 58:445–459.

BROWN, R. W., R. S. JOHNSTON, AND K. VAN CLEVE.1978. Rehabilitation problems of Arctic and alpineregions. Pp. 23–44 in Reclamation of drastically dis-turbed lands (F. W. Schaller and P. Sutton, eds.).American Society for Agronomy, Madison, Wiscon-sin.

BRUSSARD, P. F., D. A. CHARLET, AND D. S. DOBKIN.1998. Great Basin-Mojave Desert region. Pp. 505–542 in Status and trends of the nation’s biologicalresources (M. J. Mac, P. A. Opler, C. E. PuckettHaecker, and P. D. Doran, eds.). United States Geo-logical Survey, United States Department of the In-terior, Reston, Virginia.

BURNHAM, K. P., AND D. R. ANDERSON. 1998. Model

selection and inference: a practical information-the-oretic approach. Springer-Verlag, New York.

BURNHAM, K. P., AND D. R. ANDERSON. 2001. Kull-back-Leibler information as a basis for strong infer-ence in ecological studies. Wildlife Research 28:111–119.

BUTLER, D. R. 1995. Trampling, wallowing, and ge-ophagy by mammals. Pp. 82–107 in Zoogeomor-phology: animals as geomorphic agents (D. R. But-ler, ed.). Cambridge University Press, New York.

CAUGHLEY, G. 1994. Directions in conservation biol-ogy. Journal of Animal Ecology 63:215–244.

CEBALLOS, G., AND J. H. BROWN. 1995. Global patternsof mammalian diversity, endemism, and endanger-ment. Conservation Biology 9:559–568.

CHAMBERS, J. C. 1997. Restoring alpine ecosystems inthe western United States: environmental con-straints, disturbance characteristics, and restorationsuccess. Pp. 161–187 in Restoration ecology andsustainable development (K. M. Urbanska, N. R.Webb, and P. J. Edwards, eds.). Cambridge Univer-sity Press, New York.

COLE, F. R., D. M. REEDER, AND D. E. WILSON. 1994.A synopsis of distribution patterns and the conser-vation of mammal species. Journal of Mammalogy75:266–276.

CUTLER, A. 1991. Nested faunas and extinction in frag-mented habitats. Conservation Biology 5:496–505.

DALY, C., R. P. NEILSON, AND D. L. PHILLIPS. 1994. Astatistical-topographic model for mapping climato-logical precipitation over mountainous terrain. Jour-nal of Applied Meteorology 33:140–158.

DAWSON, M. R. 1967. Lagomorph history and thestratigraphic record. Pp. 287–316 in Essays in pa-leontology and stratigraphy (C. Teichert and E. L.Yochelson, eds.). Department of Geology, Univer-sity of Kansas, Special Publication 2:1–626.

DEARING, M. D. 1997. The function of haypiles of pi-kas (Ochotona princeps). Journal of Mammalogy78:1156–1163.

DIAMOND, J. M. 1987. Extant unless proven extinct?Or, extinct unless proven extant? Conservation Bi-ology 1:77–79.

DURRANT, S. D. 1952. Mammals of Utah: taxonomyand distribution. Museum of Natural History, Uni-versity of Kansas, Lawrence.

GRAYSON, D. K. 1987. The biogeographic history ofsmall mammals in the Great Basin: observations onthe last 20,000 years. Journal of Mammalogy 68:359–375.

GRAYSON, D. K. 1993. The desert’s past: a natural pre-history of the Great Basin. Smithsonian InstitutionPress, Washington, D.C.

GRAYSON, D. K., AND S. D. LIVINGSTON. 1993. Missingmammals on Great Basin mountains: Holocene ex-tinctions and inadequate knowledge. ConservationBiology 7:527–532.

GRAYSON, D. K., AND D. B. MADSEN. 2000. Biogeo-graphic implications of recent low-elevation recol-onization by Neotoma cinerea in the Great Basin.Journal of Mammalogy 81:1100–1105.

HAFNER, D. J. 1993. North American pika (Ochotonaprinceps) as a late Quaternary biogeographic indi-cator species. Quaternary Research 39:373–380.

HAFNER, D. J. 1994. Pikas and permafrost: post-Wis-


consin historical zoogeography of Ochotona in thesouthern Rocky Mountains, U.S.A. Arctic and Al-pine Research 26:375–382.

HAFNER, D. J., AND R. M. SULLIVAN. 1995. Historicaland ecological biogeography of Nearctic pikas (La-gomorpha: Ochotonidae). Journal of Mammalogy76:302–321.

HALL, E. R. 1946. Mammals of Nevada. University ofCalifornia Press, Berkeley.

HANLEY, T. A., AND K. A. HANLEY. 1982. Food re-source partitioning by sympatric ungulates on GreatBasin rangeland. Journal of Range Management 35:152–158.

HANSKI, I. 1991. Single-species metapopulation dy-namics: concepts, models and observations. Biolog-ical Journal of the Linnaean Society 42:17–38.

HANSKI, I. 1998. Metapopulation dynamics. Nature396:41–49.

HOSMER, D. W., JR., AND S. LEMESHOW. 1989. Appliedlogistic regression. Wiley-Interscience, New York.

HUNTLY, N. J., A. T. SMITH, AND B. L. IVINS. 1986.Foraging behavior of the pika, Ochotona princeps,with comparisons of grazing versus haying. Journalof Mammalogy 67:139–148.

JOHNSON, N. K. 1975. Controls of number of bird spe-cies on montane islands in the Great Basin. Evolu-tion 29:545–567.

KERR, J. T., AND D. J. CURRIE. 1995. Effects of humanactivity on global extinction risk. Conservation Bi-ology 9:1528–1538.

KURTEN, B., AND E. ANDERSON. 1980. Pleistocenemammals of North America. Columbia UniversityPress, New York.

LAWLOR, T. E. 1998. Biogeography of Great Basinmammals: paradigm lost? Journal of Mammalogy79:1111–1130.

LOMOLINO, M. V. 1986. Mammalian community struc-ture on islands: the importance of immigration, ex-tinction, and interactive effects. Biological Journalof the Linnaean Society 28:1–21.

MANGEL, M., AND C. TIER. 1994. Four facts every con-servation biologist should know about persistence.Ecology 75:607–614.

MCDONALD, K. A., AND J. H. BROWN. 1992. Usingmontane mammals to model extinctions due to glob-al change. Conservation Biology 6:409–415.

MCLANE, A. R. 1978. Silent cordilleras: the mountainranges of Nevada. Camp Nevada, Monograph No.4. Reno, Nevada.

MEAD, J. I. 1987. Quaternary records of pika, Ocho-tona, in North America. Boreas 16:165–171.

NEWMARK, W. D. 1995. Extinction of mammal popu-lations in western North American national parks.Conservation Biology 9:512–526.

NORTON, D. A. 1999. Forest reserves. Pp. 525–555 inMaintaining biodiversity in forest ecosystems (M. L.Hunter, ed.). Cambridge University Press, Cam-bridge, United Kingdom.

PEACOCK, M. M. 1997. Determining natal dispersalpatterns in a population of North American pikas

(Ochotona princeps) using direct mark–resight andindirect genetic methods. Behavioral Ecology 8:340–350.

PETERS, R. L., AND J. D. S. DARLING. 1985. The green-house effect and nature reserves. BioScience 35:707–717.

PIMM, S. L., H. L. JONES, AND J. DIAMOND. 1988. Onthe risk of extinction. American Naturalist 132:757–785.

PULLIAM, H. R. 1988. Sources, sinks, and populationregulation. American Naturalist 132:652–661.

RAPPORT, D. J. 1980. Optimal foraging for comple-mentary resources. American Naturalist 116:324–346.

SMITH, A. T. 1974a. The distribution and dispersal ofpikas: influences of behavior and climate. Ecology55:1368–1376.

SMITH, A. T. 1974b. The distribution and dispersal ofpikas: consequences of insular population structure.Ecology 55:1112–1119.

SMITH, A. T. 1980. Temporal changes in insular pop-ulations of the pika (Ochotona princeps). Ecology61:8–13.

SMITH, A. T. 1987. Population structure of pikas: dis-persal versus philopatry. Pp. 128–142 in Mammaliandispersal patterns: the effects of social structure onpopulation genetics (B. D. Chepko-Sade and Z. T.Halpin, eds.). University of Chicago Press, Chicago,Illinois.

SMITH, A. T., AND M. L. WESTON. 1990. Ochotonaprinceps. Mammalian Species 352:1–8.

SOKAL, R. R., AND F. J. ROHLF. 1995. Biometry: theprinciples and practice of statistics in biological re-search. 3rd ed. W. H. Freeman and Company, NewYork.

STAMPS, J. A. 1988. Conspecific attraction and aggre-gation in territorial species. American Naturalist131:329–347.

TILMAN, D., R. M. MAY, C. L. LEHMAN, AND M. A.NOWAK. 1994. Habitat destruction and the extinctiondebt. Nature 371:65–66.

TRACY, C. R., AND T. L. GEORGE. 1992. On the deter-minants of extinction. American Naturalist 139:102–122.

TYSER, R. W. 1980. Use of substrate for surveillancebehaviors in a community of talus slope mammals.American Midland Naturalist 104:32–38.

VERTS, B. J., AND L. N. CARRAWAY. 1998. Land mam-mals of Oregon. University of California Press,Berkeley.

WAGNER, F. H., ET AL. 1995. Wildlife policies in theU.S. national parks. Island Press, Covelo, California.

WELTZ, N., M. K. WOOD, AND E. E. PARKER. 1989.Flash grazing and trampling effects on infiltrationrates and sediment yield on a selected New Mexicorange site. Journal of Arid Environments 16:95–100.

Submitted 27 September 2001. Accepted 17 July 2002.

Associate Editor was Thomas O’Shea.


AP

PE

ND

IXI

Cha

ract

eris

tics

ofsi

tes

inth

ehy

drog

raph

icG

reat

Bas

in,

Uni

ted

Sta

tes,

atw

hich

popu

lati

ons

ofpi

kas

(Och

oton

apr

ince

ps)

wer

ere

port

eddu

ring

the

20th

cent

ury.

Sit

enu

mbe

rsid

enti

fypo

pula

tion

sin

Fig

s.1

and

2.C

rite

ria

for

cate

gori

esof

occu

piab

leta

lus

foll

owB

eeve

r(1

999)

,an

dpi

ka-

equi

vale

ntel

evat

ion

foll

ows

Haf

ner

(199

3).

‘‘M

ainl

and’

’re

fers

toei

ther

the

Sie

rra

Nev

ada

orR

ocky

Mou

ntai

ns,

asde

scri

bed

inte

xt.

Sit

enu

mbe

rL

ocat

ion

Sou

rcea

Spe

ci-

men

date

Mou

ntai

nra

nge

Am

ount

ofoc

cupi

able

talu

s

Inim

med

iate

area

(0.8

kmra

dius

)In

mou

ntai

nra

nge

Mea

nla

ti-

tude

(8N

)

Mea

nlo

ngi-

tude

(8W

)

Pik

a-eq

uiva

lent

elev

a-ti

on(m

)

Min

imum

elev

atio

nof

enco

un-

terb

(m)

Dis

tanc

eto

mai

n-la

nd(k

m)

1 2 3 4

Duf

fer

Pea

kc

Tob

yC

anyo

nS

mit

hsC

reek

c

Sum

mit

Lak

ec

a a a a

1935

1934

1934

1936

Pin

eF

ores

tD

esat

oya

Des

atoy

aN

otin

ara

nge

Sm

all

Lar

geL

arge

Sm

all

Sm

all

Med

ium

Med

ium

Not

ina

rang

e

41.6

539

.39

39.3

541

.61

118.

7411

7.75

117.

6811

9.07

1,88

5.7

2,09

5.6

2,10

3.8

1,86

1.0

2,56

02,

484

2,28

61,

798

203.

113

3.0

133.

023

9.0

5 6 7 8 9 10 11

For

tB

idw

ellc

Lon

gC

reek

Thr

eeL

akes

Mus

tang

Mou

ntai

nP

inch

otC

reek

Arc

Dom

eS

outh

Tw

inR

iver

a a a a a a a

1941

1928

1928

1933

1933

1898

1915

Not

ina

rang

eR

uby

Rub

yW

hite

Whi

teT

oiya

beT

oiya

be

Med

ium

Sm

all

Lar

geL

arge

Med

ium

Lar

geL

arge

Not

ina

rang

eL

arge

Lar

geL

arge

Lar

geL

arge

Lar

ge

41.9

040

.54

40.5

937

.90

37.9

238

.83

38.8

3

119.

9611

5.47

115.

3911

8.31

118.

2911

7.35

117.

31

1,77

1.4

2,22

0.2

2,22

4.1

2,13

3.5

2,13

4.5

2,16

0.7

2,16

4.4

1,68

02,

188

2,55

02,

865

2,57

63,

139

2,52

5

209.

423

7.0

237.

016

.016

.017

6.0

176.

012 13 14 15 16 17 18

Moh

awk

Can

yon

Gre

enm

onst

erC

anyo

nM

ount

Jeff

erso

nP

eter

son

Cre

ekS

teel

sC

reek

Big

Indi

anM

ount

ain

Tho

mas

Cre

ek

a a a a a b b

1931

1933

1933

1934

1929

1947

1956

Toi

yabe

Mon

itor

Toq

uim

aS

hosh

one

Eas

tH

umbo

ldt

Was

suk

Rub

y

Med

ium

Med

ium

Lar

geM

ediu

mL

arge

Lar

geL

arge

Lar

geS

mal

lL

arge

Med

ium

Med

ium

Med

ium

Lar

ge

38.9

838

.75

38.7

539

.19

40.9

038

.49

40.6

2

117.

3311

6.57

116.

7211

8.30

115.

1111

8.79

115.

40

2,15

4.2

2,23

0.4

2,21

7.4

2,06

2.3

2,22

9.3

2,06

0.8

2,22

0.9

2,70

12,

621

2,72

02,

423

2,58

32,

713

2,74

3

176.

018

2.0

182.

017

6.0

237.

057

.823

7.0

19 20 21 22 23 24 25

Cur

rant

Mou

ntai

nK

iger

Gor

geS

tock

ade,

War

ner,

Goa

tC

reek

s20

mi

NE

Ade

lc

Cou

gar

Pea

kd

Tho

mas

Cre

ekR

ange

rS

tati

onc

Cra

neM

ount

ain

c d d d d d d

1990

1916

1937

1913

1925

1925

1930

Whi

teP

ine

Ste

ens

Har

tN

otin

ara

nge

Not

ina

rang

eN

otin

ara

nge

War

ner

Med

ium

Med

ium

Lar

geS

mal

lL

arge

Sm

all

Lar

ge

Med

ium

Med

ium

Lar

geN

otin

ara

nge

Not

ina

rang

eN

otin

ara

nge

Med

ium

38.9

142

.77

42.4

242

.40

42.3

142

.40

42.0

8

115.

4011

8.58

119.

7611

9.64

120.

6312

0.60

120.

24

2,31

8.1

1,83

6.1

1,78

5.0

1,76

9.1

1,69

2.2

1,68

9.7

1,73

7.50

2,70

11,

951

1,79

81,

760

2,07

31,

768

2,31

6

240.

037

6.0

316.

031

5.8

305.

631

2.5

275.

0

aS

ourc

es:

a,H

all

(194

6);

b,U

.C.

Ber

kele

yM

useu

mof

Ver

tebr

ate

Zoo

logy

reco

rds;

c,L

awlo

r(1

998)

;d,

Ver

tsan

dC

arra

way

(199

8).

bT

helo

wes

tel

evat

ion

atw

hich

we

dete

cted

pika

sin

surv

eys

duri

ng19

94–1

999

or(i

fun

dete

cted

)pi

kas

wer

ere

port

ed.

cE

xtir

pate

d.d

Fun

ctio

nall

yex

tirp

ated

(onl

y1

indi

vidu

alde

tect

ed).


APPENDIX II

Data used to test human-influence hypotheses for pika populations of the Great Basin. In non-wilderness areas, ‘‘jurisdiction’’ refers to the land agency administering management at the area.Primary roads were roads passable in summer without a 4-wheel-drive vehicle.

Sitenum-ber Location

Defecationsa

CattleOther

ungulate HorseManagementjurisdictionb

Grazingstatusc

Distance to nearest

Road(km)

Primaryroad (km)

123456

Duffer Peakd

Toby CanyonSmiths Creekd

Summit Laked

Fort Bidwelld

Long Creek

10081

2692

6053

83119

010

2

0429120

341

10

BLMBLMBLMBLMBLMUSFS

GGGGGG

2.751.250.252.000.505.50

2.756.000.253.250.506.75

789

10111213

Three LakesMustang MountainPinchot CreekArc DomeSouth Twin RiverMohawk CanyonGreenmonster Canyon

70

44126

2197449

3181081

434

120

172352

279

00

WildernessUSFSBLMWildernessWildernessUSFSUSFS

UUGUUGG

1.500.502.257.758.252.752.75

1.501.252.507.758.253.255.50

14151617181920

Mount JeffersonPeterson CreekSteels CreekBig Indian MountainThomas CreekCurrent MountainKiger Gorge

18117

1227

48291

1137073

217

327

0000000

WildernessUSFSWildernessWildernessWildernessWildernessWilderness

UGUUUUG

5.250.751.250.503.501.750.25

5.502.753.508.503.505.257.25

2122232425

Stockade, Warner Crks.20 mi NE Adeld

Cougar Peake

Thomas Creek R.S.d

Crane Mountain

598

4627

29119

359

60401

WildernessWildernessUSFSUSFSUSFS

UUGGG

0.752.000.000.250.00

1.504.500.750.252.75

a Defecations reflect counts of fecal piles in standardized encounter surveys.b BLM 5 Bureau of Land Management; USFS 5 United States Forest Service; ‘‘Wilderness’’ defined in text.c G 5 grazed by domestic cattle or sheep for .50% of the period between historic record of pikas and our sampling; U 5

ungrazed.d Extirpated.e Functionally extirpated (only 1 individual detected)

patterns of apparent extirpation among isolated ... · persistence of animals in heterogeneous...

Documents