The Auk 115(3):694-705, 1998

RESPONSE OF GREAT HORNED TO EXPERIMENTAL "HOT SPOTS" OF SNOWSHOE DENSITY

CHRISTOPHROHNER • AND CHARLESJ. KREBS Centrefor BiodiversityResearch, Department of Zoology,University of BritishColumbia, Vancouver, V6T 1Z4,

ABSTRACT.--Predatorsthat aggregatein "hot spots"of high preydensity have been hy- pothesizedto synchronizepopulation cycles of smallmammals. During a peakand decline in a snowshoehare (Lepusamericanus) cycle, we createdartificial hot spotsof increasedhare abundanceby addingfood and excludingmammalian predators on three1-km 2 blocks and then recordedthe responseof radio-markedGreat Horned Owls (Bubovirginianus) to these food additions.Territorial owls showeda decreasein homerange size and patchinessof spatialuse as hare densitiespeaked and declined,although this was better explained by smallerterritory sizes due to a growingowl populationrather than a directbehavioral re- sponseto changingfood density.Experimental owls on food-enrichedterritories did not showa differencein conventionalmeasurements of home-rangesize and patchiness of spa- tial use comparedwith controls,but the distancesof locationsto treatmentblocks re- vealedconcentrations of spatial use on experimental hot spots.At a largerscale, neither ter- ritorialowls nor nonterritorial floaters showed a tendencyto leavepoorer patches and move towardhot spots,and the territorialsystem of GreatHorned Owls was largelyresistant to extremevariations in prey density.The effectof socialinterference between predators has beenassumed for severalmodels of predator-preyinteractions, but empiricalevidence has rarely beendemonstrated. Our resultssuggest that territorialbehavior, in additionto lim- iting the growthof a predatorpopulation, also prevents large aggregationsof predatorsat an intermediatespatial scale. Received 3 July 1997, accepted 29 December1997.

To WHATDEGREE do predatorsexploit local prey abundances,and thereforesynchronize concentrationsof prey, and what are the limi- the dynamicsof cyclicprey populationsover tationsof plasticityin theirbehavioral response large regions (e.g. Pitelka et al. 1955, Angel- to "hot spots"of prey abundance?These ques- stam et al. 1984, Lindstr6m et al. 1987, Yden- tions have been of intense interest in studies of berg 1987, Ims and Steen 1990, Korpim•iki predator-prey relationshipsfor several de- 1994, Steen 1995, Norrdahl and Korpim•iki cades. For example, differencesin predator- 1996). Also, possibleinterference between so- prey dynamicshave been related to variationin ciallyintolerant predators may require changes predatorforaging behavior. Most implications in the assumptionsfor classicpredator-prey for population dynamicswere derived from isoclines,for exampleallowing the predator analysesof functional responsesof predators isoclineto vary accordingto a ratio-dependent to prey density(Holling 1959,Fujii et al. 1986). model (e.g. Arditi and Ginzburg 1989,Hanski It is particularly the distinctionbetween the 1991). convextype-II responseand the sigmoidtype- Much work has focusedon how predators III responsethat changes the dynamicsof pred- optimizetheir foragingdecisions (see Stephens ator-prey interactions(Rosenzweig and Mac- and Krebs1986). Beside choosing optimal prey Arthur 1963,Holling 1965,Murdoch and Oaten items (Pulliam 1974), in a heteroge- 1975,Caughley and Sinclair1994). neousenvironment are confrontedwith spatial On the other hand, variations in social be- decisionsabout where to go and how long to haviorof predatorscan affectthe dynamicsof stayin specificpatches (Charnov 1976). Several predator-preyrelationships. Nomadic preda- foragingmodels have provided a mechanistic tors that are socially tolerant of one another approachto explain different functionalre- may track hot spotsof peak densitiesof prey sponsesin this context (Abrams 1982, 1987, rapidly and aggregatein suchareas, depress Mitchell and Brown 1990,Brown and Morgan 1995). Other models have analyzed the tem- E-mail:[email protected] poral and spatialdistribution of resourcesthat

694 July1998] PredatorResponse toPrey Hot Spots 695 will favor the evolution of nomadic or site-te- and movecloser to hot spotsas hare densities nacious life styles in predators (Andersson declinein the study area. 1980). Althoughaggregative responses of predators STUDY AREA AND METHODS may have a profoundeffect on predator-prey interactionsand communitystructure (Hassell Studysystem.--Great Horned Owls are large,long- and May 1974,Holt and Lawton 1994),very lit- lived predatorsthat feedmainly on lagomorphsand are territorial year-round (Donazar et al. 1989, Roh- tle empiricalinformation exists about the spa- ner 1995,1996). Great Horned Owls reachhigh den- tial andtemporal scales of suchresponses (e.g. sitiesin the Nearcticboreal forest, where they exhibit Keith and Rusch1988, Korpimfiki 1994), and a numerical responseto the snowshoehare cycle limitationsof predatorresponses due to social (Ruschet al. 1972,McInvaille and Keith 1974,Adam- interferencehave rarely been demonstrated cik et al. 1978, Houston 1987, Houston and Francis empirically.At present,there is virtually no in- 1995, Rohner 1996). formationon large carnivoresfrom food-sup- Snowshoehares show a regular 10-yearpopula- plementationexperiments (Boutin 1990). tion cycle acrossnorthern (Keith In this paper, we examinewhether Great 1963).They canhave more than threelitters per year and at peak numbersrepresent a higher biomass HornedOwls (Bubovirginianus) exhibit a spa- than that of any other herbivore in boreal forests tial responsein foragingeffort to changesin (O'Donoghueand Krebs1992, Boutin et al. 1995).In numbersof cyclicsnowshoe (Lepus amer- our study area,hare densitiesrecovered from a low icanus)in the boreal forest of the southwestern in the mid-1980s, increased more than 50-fold in , Canada. Our approachwas twofold. abundanceto peak levelsin 1989and 1990,and then First,we askedhow sizeand patchiness in use declinedrapidly from thewinter of 1990-91onwards of spaceby radio-markedowls changedover (seeKrebs et al. 1995). the peak and declineof a snowshoehare cycle, We worked at Kluane Lake (60ø57'N, 138ø12'W)in the southwesternYukon, Canada. The study area and what correlativefactors best explained comprised350 km2 of the ShakwakTrench, a broad thesechanges. Second, we increasedprey den- glacialvalley boundedby alpine areasto the north- sities by providing food for snowshoehares west and southeast.The valley bottom averages andby excludingmammalian predators (Krebs about 900 m abovesea level and is coveredmostly et al. 1995) and then recordedthe behavioralre- with spruceforest (Picea glauca), shrub thickets (Salix sponsesof owlsto theseartificially created hot spp.), aspen forest (Populustremuloides), grassy spotson 1-km2 blocks (ca. 20 to 30% of an av- meadowswith low shrubs (Betulaglandulosa), old erageowl territory;Rohner 1997). Lagomorphs burns,eskers, marshes, small lakes, and ponds. Definitionof terms.--Territorialowls were seden- are preferredprey items for Buboowls (Dona- tary, paired, and occupiedlong-term territories year- zar et al. 1989),and snowshoehares comprised round (Rohner 1996, 1997). Nonterritorial owls, or 83 to 92% of the diet of Great Horned Owls dur- floaters,were owls that were radio-markedas fledg- ing thepeak of thehare population cycle in our lings and monitored for up to 2.5 years, during study area (Rohner1995). According to the hy- which time they did not acquirea territory and did pothesisthat predatorsconcentrate their for- not breed (see Rohner 1997). Experimental blocks, agingeffort at hot spotsof prey densityat any food supplementation,and food additionwere ma- spatialscale, we addressthe followingpredic- nipulationsof prey densitiesas described in thenext section.Control blocks were otherplaces in the study tions:(1) owls on territorieswith hot spotswill area where controlprey densitieswere sampledfor decreasehome-range size (i.e. the area used comparison(see below). Experimental owls were ter- within a territoryduring a giventime will de- ritory ownerswith foodsupplementation (provided crease);(2) patchinessin use of spaceby terri- for hares)within their territories.Control owls were torial owls on enriched territories will increase; all other territorial owls with active radio transmit- (3) telemetrylocations of owlson enrichedter- ters,independent of whethertheir territories includ- ritories will be closer to food-addition blocks ed a samplingsite of controldensities of prey. Ter- ritory size was defined as the total area used by a than expectedfrom random locations;(4) ex- pair throughoutthe year.Territory sizes can be mea- perimentalowls with hot spots will have a sured by mapping hooting males and delineating higherportion of snowshoehares in their diets territorial boundariesin spring when territorial ac- than do control owls; and (5) territorial and tivity is most pronounced.Here, we used an approx- nonterritorialowls will leave poorer patches imationof territory sizethat wascalculated from the 696 ROHNERAND KREBS [Auk, Vol. 115

72 • ,513(3) I•1 • 70I•ne• 5sl7(1) 495(1) I•1 ._•66 544(4) oo64 / / / • $39(2) >',o 62 4d4(1) / 521(1) "• / 503(1) 509(3) •'

6056 ...... 569(3) 6•6(1)•way 35 40 45 50 55 60 65 70 75

X-coordinates (kin) FIC. 1. Studyarea at KluaneLake, southwestern Yukon,with blocksof manipulatedsnowshoe hare 89 90 91 92 densities(open symbols), blocks of controlmeasure- YEAR ments of hare densities, and location of territorial owls(identification code, with numberof yearsmon- FIG.2. Manipulationsof snowshoehare densities itoredin parentheses). on 1-km2 blocks at Kluane Lake, Yukon. Shown are haredensities in springon three experimental blocks (opensymbols) and five controls(filled symbols), with averagedensities connected by lines.Symbols known populationdensity within 100 km2 (which as in Figure 1. wassimilar to mappedvalues; see Rohner 1997). We definedhome range as the fractionof a defendedter- ritorythat was actually used within a specifiedtime period.Home-range size was calculatedfrom telem- with neighborsat dawn and dusk,and obviousdis- etry locationsusing standard methods (see below). putesbetween hooting males or pairswere usedto Manipulationof prey densities.--We chose several 1- map territory boundaries.When necessary,play- km 2 blocks of undisturbed boreal forest for treat- backswere used to elicitresponses of territoryown- mentsand strivedto interspersetreatments and con- ers and their neighbors.Most maleswere individu- trolsin the studyarea (Fig. 1). Threeexperimental ally knownbecause of radio taggingand becauseof areaswere providedwith supplementalfood (com- their distinct vocalizations. Vocal differences were mercialrabbit chow, 16% protein, added weekly) ad laterverified with sonagramsfrom recordingsat the libitumyear-round, and oneof thosewas protected nest(C. Rohnerunpubl. data). Observations of ter- by an electricfence 2.2 m in heightto excludemam- ritorial activitywere made almost daily from early malian predators(fence checked daily for perfor- Februaryuntil late April (->300h eachyear). mance).All treatmentswere "press"experiments Nestswere found according to Rohnerand Doyle (sensuBender et al. 1984)that lastedthroughout the (1992), and during 1988 to 1991,radio transmitters studyperiod reported in this paper.Hare densities wereattached to 55 owletsbefore fledging. Of those, on blockswere estimatedby mark-recapturelive thesuccessful dispersers were later monitored inten- trappingduring four to five daysin late Marchand sively (3 born in 1988,11 in 1989,and 16 in 1990),and early April. Trappingwas conductedon permanent, nine remained as nonterritorial floaters in the area 36-hagrids on all blocks.We presentan averageof during this study (seeRohner 1996, 1997). Radio thedensity estimates from the program Capture and transmittersweighed 50 g, includinga shoulderhar- from the Jolly-Sebermethod (Pollocket al. 1990). nessof Teflonribbon for attachmentas a backpack Hare numberson the threeexperimental blocks in- (<5% of body mass; Kenward 1985); transmitters creaseddramatically and on averagewere 2.8 to 10.3 functionedfor 2 to 2.5 years. timeshigher than thoseon controlblocks (Fig. 2). In addition to juvenile , 21 territorial adult Five controls consisted of two untreated blocks and owlswere capturedwith mist netsand cagetraps three blocks treated with fertilizer and other treat- and equippedwith radio transmitters.The effective mentsthat did not resultin stronglyelevated hare samplesize of territorialowls used in eachyear var- numbers.Details are givenin Krebset al. (1995)and ied and was smallerthan the total samplesize be- Boutinet al. (1995). causeof logisticaldifficulties. We were ableto work Fieldtechniques for owl data.--GreatHorned Owls with oneto threeexperimental owls and two to six were censusedin late winter and early springon a controlsin eachof the four yearsof study(see Table 100-km2 plotwithin the mainstudy area. Individual 1, Fig. 3). To allow for better comparisons,we used pairswere identified when hooting simultaneously only femalesfor telemetrystudies, except in 1992, July1998] PredatorResponse toPrey Hot Spots 697

TABLE1. Summary of sample sizes and precision for telemetrylocations of territorialGreat Horned Owls at Kluane Lake, Yukon (including experi- (a) mental and control owls; seeMethods).

95%- error _. 4.5 o ß area 95%-error No. No. of (medi- area of locations an, (quartiles, Year owls per owla km2) km2) 1989 3 18.7 + 1.2 0.19 0.05 to 0.37 1990 5 20.0 + 0.0 0.10 0.02 to 0.26 1991 5 20.0 + 0.0 0.09 0.03 to 0.20 O•0.5 1992 6 21.0 _+ 0.0 0.05 0.02 to 0.15

aœ_+ SE. 90 91 92 whenfemale no. 503 emigrated from the studyarea and its mate (no. 564) was monitoredinstead. (b) Telemetrywork on useof spaceby territorialowls 0.8 wasconcentrated during periods of threeweeks each year, with locationsobtained on consecutivenights 0.6- for eachbird if possible(not achievedin the first year;thus, the total monitoring periods for eachbird ' in 1989were 27, 28, and41 days).These periods of 0.4- intensivemonitoring were conductedfrom 24 Julyto 8 September1989, 7 to 26 September1990, 5 to 26 September1991, and 12 Juneto 3 July1992. For more specificmeasurements of experimentalowls no. 544 and503 duringthe declinephase of the harecycle, all telemetrylocations taken on nightsbetween Sep- 89 90 91 92 tember1991 and September1992 were used.Telem- etry work to assessthe distanceof ninenonterritorial YEAR floatersfrom experimentalgrids involved weekly lo- cationsfrom January1990 to September1992 (see FIG.3. Characteristicsin use of spaceby Great Rohner1997). Horned Owls monitoredby radio telemetryduring Telemetry locationswere obtained by triangula- three-weekperiods in eachyear. Filled symbols are tionusing hand-held equipment. Topographic maps controls,open symbolsare experimentalowls on wereused in thefield to plot thelocations and assess food-enrichedterritories (sample sizes in Table1). thenumber of bearingsneeded for reliable estimates. Regressionlines include only controls:(A) Home- Thetriangulations were analyzed with theprogram rangesize, expressed as a minimumconvex polygon LocateII (Nares1990) for calculatingexact locations of the innermost70% of locations).(B) Patchinessin anddistances. Median 95%-error ellipses (Lenth es- spaceuse (expressed as the portion of themultinu- timator; Saltz and White 1990)are given in Table1. clear 70% clusters relative to the mononuclear con- The accuracyof telemetrylocations was assessed by vex polygongiven in [A]; seeKenward 1987). triangulationon five transmitters that were placed in treesat heightsof 4.5 to 5.5m. Thedeviation of these locations(error of 0.052+ SEof 0.018km 2) from the Use of spacewas measuredby utilization distribu- sitecoordinates obtained by a globalpositioning sys- tionsbased on clusteringmethods. From a centerof tem was 0.101 + 0.027 kin. closestlocations, an increasingpercentage of near- Summerdiets of territorial owls were sampled est-neighborlocations was added,resulting in a cu- during May to July (nine territoriesfrom 1989 to mulative increase of core area used. Mononuclear 1991, six territories in 1992). Pelletswere collected clusteringwas centered around the harmonicmean from breedingbirds at nestsand at roostsites locat- location,whereas multinuclearclustering allowed ed by telemetry.The resultsof pelletanalysis were for separateclusters wherever distances between lo- expressedas the percentageof a prey speciesin the cationswere closest.Home-range size was then de- totalprey biomass (see Rohner 1995). rived for different levelsof core percentages(Ken- Analysisof telemetrydata.--Calculations were per- ward 1987).Patchiness was calculatedas "part-ar- formed using RangesIV software (Kenward 1990). eas,"which are the areasused at a specificcore per- 698 ROHNERAND KREBS [Auk, Vol. 115 centageand expressedas a portion of the total area (see Kenward 1987). These proceduresallowed a moresensitive approach to recognizingbiases due to outliersand differentpatterns of spatialuse. For the 6.5. monitoringperiod in September1991, three of eight territorial owls were excludedfrom analysisbecause 6. of extremelong-distance movements during several days(these extraterritorial movements are described 5.5. in Rohner1996). Statisticalprocedures.--All arithmetic means are re- 5. ported _+one SE, and all probabilitiesare two-tailed unlessotherwise specified. Correlation coefficients 4.5. were calculatedas Spearmanrank correlations.For statistical testing, nonparametric tests were used 4 , whereverpossible (ANOVAs were calculatedwith 69 90 91 92 log-transformeddata, or with arcsine-transformed data for percentages).The testingof bootstraphy- YEAR pothesesfollowed Hall and Wilson(1991), and two- FIG. 4. Average territory size of Great Horned sided probabilitieswere derived from 500 simula- Owls during a peak and declineof snowshoehares. tions (seeRohner 1997). Mean territory sizes (filled circles)were calculated from a census of owl territories on 100 km 2 of the RESULTS main study area (vertical bars are minimum and maximum estimates,numbers of territoriesranging Useof space:Home-range size and patchiness.- from 14 to 15 in 1989 to 21 to 24 in 1992; Rohner Home-rangesizes of territorial owls decreased 1996). from 1989 to 1992 (rs = -0.64, n = 19, P < 0.001;Fig. 3A). Becausehome-range estimates Interestingly,the declinein territory size ex- are sensitive to outliers (Kenward 1987), a plained considerablymore of the variationin range of different core percentageswas ana- use of spaceby theseowls (r 2 = 0.85,F = 11.0, lyzed. The sametrend was consistentfor core df = 1 and2, P = 0.08).Because the samplesize percentagesof 60 to 95%,and we chosea level of only four years was small, and collinearity of 70%(i.e. discardingthe outermost30% of lo- in hare densityand territorysize of owlswas cations)for presentationin Figure3A to ensure high for all but one year, this correlativeap- a conservative and robust result. This trend proachcannot be takenas strongsupport for was also significantwhen experimentalowls the lack of a relationshipbetween prey density were excludedfrom analysis(rs = -0.63, n = and use of spaceby owls. 13, P < 0.05). Resultsfrom experimentalowls did not con- A similar trend was found for patchinessin firm a direct effect of prey densityon home- use of space (Fig. 3B). Not only did Great rangeuse. Mean home-rangesize of owls on HornedOwls use smaller areas within a given territorieswith hot spotsdid not differ from time, they alsoused this smallerspace more that of owls on controlterritories (ANCOVA, F uniformly (rs = -0.60, n = 19, P < 0.01).This < 0.1, df = 1 and 16, P = 0.99;Fig. 3A). Also, trend was alsosignificant for controlowls only patchinessin use of spacedid not differ be- (rs = -0.70, n = 13, P < 0.01). tween experimentaland controlowls (ANCO~ Because hare densities declined from 1990 to VA, F = 0.52, df = 1 and 16, P = 0.48; Fig. 3B). 1992(Fig. 2), oneobvious hypothesis is thatthe Althoughhare densitieswere 2.8 to 10.3times observedchanges in useof spaceby owlswere higher at treatmentsites than at controlsites, a directfunction of varyingprey density.Hare with no apparenteffect on owl home-range density,however, explained only a smallpor- size,it shouldbe notedthat the statisticalpow- tion of the variancein home-rangesize (r2 = er for demonstratinga differencewas low. 0.16, F = 0.39, df = 1 and 2, P = 0.59). But at Useof hot spots in experimentalterritories.--The the same time, the number of owls increased, contrastbetween prey density on the study leading to a more denselypacked array of ter- area in generaland on artificiallycreated hot ritoriesin the studyarea, which is equivalent spotsincreased drastically as the declinephase to a decreasein territorysize over time (Fig.4). of the snowshoehare cycle progressed (Fig. 2). July 1998] PredatorResponse to PreyHot Spots 699

2.5- ..... OWL544 'r]• 25- OWL 544 ...... OWL5•4 .•,.'."'" 20- t • obs

15- •') 1-

n• 0.5- ....o....exp 10-

5-

o

PERCENT OF LOCATIONS 0 • O ..1 0 04 0.8 1,2 1.6 2 2.4 FIG. 5. Home range sizes of two experimental owls during June1992 compared with four controls (œ _+SE), as determinedfrom telemetrylocations. The areaof the maximumconvex polygon are shown 25- as a functionof an increasingcore percentage of in- OWL 503 nermost locations. 20- .... o .... exp

Therefore,we investigatedowls on two exper- 15- • obs imental territories that were accessible for work in more detail in 1992, when the responseof 10- ßo'"ø'",o...o, _ predatorswas expectedto be greatest.Food shortagesfor Great Horned Owls during this year were indicated by a complete lack of 5 ~ breeding attemptsand elevatedrates of emi- gration and mortality in resident owls (see Rohner 1996). A more detailed analysisof home-rangesize was performed across all core 0 0.4 0.8 1.2 1.6 percentages(Fig. 5). As we found in the mul- tiyear analysis,there were no consistentdiffer- DISTANCE FROM 'HOT SPOT' (km) encesbetween experimentaland control owls FIG. 6. Concentrationsin spaceuse of two exper- (repeated-measuresANOVA, F = 0.21, df = 1 imental owls relative to the center of 1-km 2 blocks and 4, P = 0.67). with increasedprey density.Locations were grouped A less conventionaland more specificap- into distanceclasses, and shownas frequency distri- proachwas taken to detectpotential concentra- bution of observedlocations (filled symbols)and ex- tionsin use of space.A null hypothesisof uni- pectedfrequencies from randomlocations in the spe- form spatial use was modeledby randomly cificterritories (open symbols). generating 5,000 points within the known boundaries of these territories. The distances mean of positivedeviations from the expected from thesepoints to the centerof the territory valueswas 5.35 (P < 0.01); for owl no. 503 this (geometricmean) were then calculated and mean was 9.20 (P = 0.02). groupedinto classesof 200 m, thus represent- Werethese concentrations in useof spacere- ing an expectedfrequency distribution of uni- lated to hot spotsin hare abundance?Teleme- form spatialuse within a territory.The actual try locationswere thencompared to the exper- telemetrylocations of experimentalowls were imentalblocks of manipulatedprey densities, thencompared with this expecteddistribution. and the distances to the center of these 1-km 2 The resultsclearly showedthat someregions blockswere calculated(Fig. 6). Both experi- were used more frequentlythan expectedfor mental owls showeda positiveresponse com- both experimentalowls. For owl no. 544, the pared with the expecteddistributions as cal- 700 ROHNERAND KREBS [Auk,Vol. 115 culatedfrom pointsrandomly generated with- TABLE2. Shiftsin use of spaceby nonterritorial in the boundaries of their territories. For owl GreatHorned Owls relativeto experimental"hot spots"of snowshoehares. Values are mediandis- no. 544, the distancewas 0.846 q- 0.050km (n tances(km) of weekly telemetrylocations to the = 152)from the centerof the hot spot(mean centerof theclosest experimental block during the expecteddistance 1.534 km; bootstrapP < peak (D•; January1990 to May 1991)and decline 0.001).For owl no. 503, this distancewas 0.741 (D2;June 1991 to September1992) phases of the -+ 0.087km (n = 41, mean expecteddistance snowshoehare cycle. 1.179km; bootstrapP < 0.001). Owl Data from pellet analysissupported the no- no. n•a n2a D• D 2 D 2 - D• pb tionthat concentrations in use of spacewere re- 406 71 20 10.858 11.070 0.211 0.153 lated to foraging activity. Summer diets of 407 61 5 5.221 7.091 1.876 0.092 Great Horned Owls consisted of 83 to 86% 415 72 14 4.454 4.967 0.513 0.674 snowshoehares from 1989 to 1991, when hare 417 25 22 4.564 3.545 --1.019 0.258 425 52 15 8.723 11.380 2.657 0.003 abundancewas high (n = 13 territories;see 433 60 13 5.178 6.115 0.937 0.076 Rohner1995). In 1992,the proportion of hares 505 25 31 4.890 12.700 7.810 0.001 in thediet dropped to 13 -+9% (range 0 to 42%, 515 38 19 1.800 3.667 1.868 0.021 n = 5 territories).We collectedpellets from one • Number of telemetrylocations during peak (n0 and decline(n2) experimentalowl during summer1992. In con- phasesof harecycle. trastto controlowls, the diet of experimental • P-valuefrom two-tailed Mann-Whitney U-tests based on D2 - D v owl no. 544 consistedof a high proportion (65%) of hares. Predatormovements from poorpatches to rich sole 1980, Schoener1983). In contrastto this patches.--Duringthe declinephase of the hare prediction, an analysisbased on conventional cycle,we predictedthat territorial owls would measuresof home-rangesize and patchiness in abandonpoor territories and intrude (at least useof spacesuggested that Great Horned Owls temporarily)into territorieswith hot spotsof decreasedtheir use of spacein responseto de- creasingprey density.However, a morespecific prey abundance.Despite intensive monitoring, examination led to a somewhat different result. we did not observeany owls that had vacated their territories to use food-addition blocks or During the first year of the hare decline,Great changedto a nomadicstrategy on the study HornedOwl numberswere increasing because area (seeRohner 1996). the responseof owlsto increasinghare densi- Althoughnonterritorial floaters were present tiesoccurred with a two-yeartime lag (Rohner in thestudy area, these owls did not aggregate 1995, 1996).Consequently, during the decline in harenumbers, the home-range size of settled in areaswith foodenrichment, and despite in- owls actually decreased(contrary to theory) tensive monitoring, we were unable to detect becauseof the more denselypacked array of an associationof floaterswith hot spotsof prey abundance. Table 2 summarizes the results on owl territories. Large territories may allow owlsto be moreselective for goodpatches over potentialshifts of eight nonterritorialowls to- a larger area,but lossesin territory sizeto new- ward hot spotsduring a time whenthe abun- comersmay thenforce owls to hunt moreeven- danceof haresin thesepatches increased by 3.8 ly in a smallerarea (see also Village 1982). to 10.3 times comparedwith controllevels in A preferencefor prey concentrationswithin the studyarea (see Fig. 2). Therewas no dis- a predator'sterritory is supportedby the re- tinctpattern, and the only significantshifts in sponseof Great HornedOwls to experimental useof spaceby threeowls were opposite to the hotspots on their territories. Although conven- predicted direction. tionalmethods of estimatinghome-range size and patchinessin use of spacefailed to detect DISCUSSION a differencebetween experimental and control owls, direct measurements of owl distances to Responsesofterritorial owls and measures ofspa- hot spotsshowed a clearresponse. The distri- tial use.--Home-rangesizes of many animals butionof owl locationswithin knownterritory arethought to vary inversely with food density boundaries and the distance of locations rela- (e.g.Myers et al. 1979,Davies 1980, Carpenter tive to food-additionblocks were significantly 1987b, Temeles 1987, Boutin 1990; but see Eber- differentfrom a randomprediction in two rep- July 1998] PredatorResponse to PreyHot Spots 701

licates.In supportof a concentrationof hunting measuresmay be designedas experiments (e.g. effortat hot spots,the proportionof snowshoe Walters1986, Caughley and Sinclair1994). haresin the diet of oneexperimental owl was Territorialityand spatialaggregation of preda- muchhigher than that of controlowls, accord- tors.--Wedid not observeany aggregationsof ing to analysis. GreatHorned Owls at experimentalhot spots. However, these results have to be considered Also, we did not observeany movementsby with somecaution. First, our sample sizes were owlsin thevicinity that suggested elevated vis- verysmall andalthough we succeededin rep- itation rates, which possiblycould have in- licatingour experiments,further evidence from creasedpredation rates due to predatorsbeing other studies would be desirable to confirm attractedto theserich patches.We cannotex- suchresponses to hot spotsof prey within a cludethe possibilitythat owls other than our predator'sterritory. Second, we havepresented marked individualsaggregated at hot spots. thenull hypothesis that use of spaceby exper- Assumingthat these other owls did notbehave imental owls is uniform if not influencedby differentlyfrom our monitoredbirds, however, foodhot spots. Many other factors, such as veg- our sample size of marked animals was suffi- etation structure, distribution of favored cient at least to detect such movements had perchedsites, or proximityto potentialintrud- theytaken place. ers, may influencethe spatialdistribution of Two explanationscan accountfor this result. raptors(e.g. Janes 1985). Although we found First,the spatialscale of our experimentalhot that owl locations were concentrated at food spotsmay not have been large enough to pres- hot spots,alternative interpretations must be ent a detectablepatch (or a patchof sufficient kept in mind. For future studies,the bestpro- rewards) to attract Great Horned Owls other cedure to test our preliminary resultswould than the specificterritory owners. Although consistof temporalcontrols, i.e. by monitoring thisexplanation may apply to neighboringter- experimentalowls before(and possiblyafter) ritory holders,nonterritorial floaters ranged manipulationsof fooddensity (e.g. Rohner and widelyin thestudy area and were temporarily Smith1996). As a third concern,we addressthe locatedat manydifferent sites, including at or relationshipbetween hare abundanceand hare nearhot spotsfor shortperiods of time. availability.Previous studies have shownthat A more likely explanationis that Great prey abundanceand prey availabilityto raptors Horned Owls did not follow an ideal free dis- do not alwaysvary directly(e.g. Southern and tribution (sensu Fretwell 1972, Milinski and Lowe1968, Bechard 1982, Preston 1990). Snow- Parker1991), and territoryholders prevented shoehares are more vulnerable to predationby aggregationsof other owls at hot spots.Al- Great Horned Owls in open habitats(Rohner thoughfloaters overlapped broadly with each and Krebs1996), and becausehares tend to use other and with establishedterritory holders, openhabitats less frequently at low densities they occurredmore often at the peripheryof (Hik 1995),they may be lessavailable to owls territoriesthan expectedby chance(Rohner at the low phaseof the populationcycle. How- 1997).Other evidence also suggests that terri- ever,in bothreplicates, the distributionof owls torial behavior affected subordinate birds. For was concentratedat hot spotsof hare abun- example,during 1988to 1993,inverse density- dance,suggesting that hare availability did not dependentgrowth rates in the territorialpop- deviatesubstantially from hare abundance. ulation, density-dependent accumulation of Thelogistical effort required to createexper- floatersin the studyarea, and replacements of imentalhot spots of preyat a scalemeaningful territorial vacanciesconfirmed the hypothesis to raptors is enormous.Because many conser- that social behavior limited the number of owl vationand managementprojects rely on telem- territories (Rohner 1995; see also Newton etry data, this raisesconcern about the detect- 1992).Stability in thenumber of breeding pairs ability of responsesof organismsto environ- despitelarge variationin prey densitywas also mentalchange. We strongly advocate the use of observedfor Tawny Owls (Strix aluco)in tem- specificquestions and experimentsinstead of perateforests (Jedrzejewski et al. 1996). the applicationof standardanalyses of home- Althoughsome theoretical models have as- range size. If manipulationsare not possible, sumedthat interference between predators will then managementactions and conservationreduce predationrates (e.g. Arditi and Ginz- 702 ROHNERAND KREBS [Auk, Vol. 115 burg 1989,Hanski 1991,Caughley and Sinclair irruptive movementsof Northern Goshawks 1994), such effectshave rarely been demon- (Accipitergentilis) and Great Horned Owls may stratedempirically. Our data suggestthat the causeelevated rates in southernCan- territorial structure of Great Horned Owls is ada and the northernUnited States(Keith and robust to extreme variations in prey density, Rusch1988). Although a considerablepropor- and that interferencebetween predators was tion of nonterritorial owls can live in an area commonboth at peak densitiesof a prey cycle with establishedterritories (Rohner 1996), and (seeRohner and Smith 1996)and at artificially predation rates may increase considerably createdhot spotsof food abundance.Accord- whenturnover rates of nomadicowls are high, ing to territory economicsand the threshold this has not been demonstratedempirically. model of territoriality (Brown 1964, Davies Limitationsto aggregativeresponses by so- 1980, Carpenter1987a), territories should be cial interferencemay not apply to otherpred- abandonedwhen food and intruder pressure ators.Korpim•iki (1994) found that avianpred- increasebeyond an upper thresholdwhere the ators in Finland tracked peaks in cycles costs of defense exceed the benefits of exclusive rapidly and reachedvery high breedingden- access.Evidence suggests that immature Gold- sities without delay,and experimentalreduc- en (Aquilachrysaetos) aggregate in al- tions of breeding pairs resulted in elevated areaswhere ungulate carcasses occur fre- prey densities(Norrdahl and Korpim•iki1996; quently,thus creatingsuch a high intruder but seeSteen 1995). In our studyarea, the num- pressurethat thesehot spotscannot be defend- ber of Great territories dropped ed by territorialeagles (Jenny 1992). In the case with a timelag of twoyears after the hare peak, of Great Horned Owls, this was not observed but NorthernGoshawks responded immediate- (seeRohner 1996). Although intrusions by non- ly with decliningnumbers (Doyle and Smith territorial owls were frequent,the robustter- 1994).Northern Goshawks can have large and ritorial systemof the owls imposedself regu- overlappinghome ranges and may showag- lation (sensuCaughley and Sinclair1994) on gregativeresponses to hot spotsof prey den- the growthof theirpopulation and alsorepre- sity (Kenwardet al. 1981).Comparisons of ag- senteda ceilingto spatialaggregation of pred- gregativeresponses of predatorswith different atorswhere prey was very abundant. socialsystems deserve more study. This direc- Evidencefor thepredator hot spot hypothesis?- tion also seems promising for mammalian The aggregativeresponse of Great Horned predatorsthat respondto prey cycles,because Owls to hot spotsof prey densitywas depen- manyof themare highlymobile and canbe so- denton spatialscale. At the smallestscale of an cially intolerant even between species (e.g. individual territory, owls appeared to concen- Keith et al. 1977, Ward and Krebs 1985, Litvaitis trate their foraging effort on experimental et al. 1986,Thompson and Colgan1987, Gese patches,as predicted.At an intermediatespa- et al. 1988,Frafjord et al. 1989,Breitenmoser et tial scale of 350 km 2,neither territorial owls nor al. 1993,O'Donoghue et al. 1997). nonterritorialfloaters aggregated near experi- mental hot spots. Although intrusionsoc- ACKNOWLEDGMENTS curred regularly,it is likely that resourcede- Thisresearch is part of a collaborativeproject, and fenseby territory ownersprevented large ag- we thank all members of the Kluane team for their gregationsof Great Horned Owls. At this in- tirelesseffort, particularly for settingup and main- termediate scale, densities are taining the manipulationsof hare densitiesand for synchronous(Fig. 2), andit is unlikelythat nat- assistingwith live-trappingof hares.Frank Doyle, C. ural hot spotsthat exceededour experimental Schmid, P. Heaven, J. Stroman, B. Zimmermann, S. densitiesoccurred elsewhere near the study Wagnitre,B. Delehanty,C. Kullberg,T. Wellicome,K. area.We cannotexclude aggregative responses Russenberger,R. Mueller, and C. and U. Breitenmo- at the very largescale of wholeregions within ser helpedwith the telemetrywork. We thank Tony Sinclair and Deb Wilson for inspiring discussions, theboreal forest, particularly because substan- and B. Kotler, D. Ward, C. Preston, and an anony- tial emigrationof owls occurredas hare den- mousreviewer for helpful commentson the manu- sitiesdeclined to very low levels(Rohner 1996). script. Funding was provided by the Natural Sci- Suchemigrations are also documentedfrom encesand EngineeringCouncil of Canada(grants to (Houston and Francis1995), and C.J.Krebs and J.N.M. Smith), a J.R. ThompsonWild- July1998] PredatorResponse toPrey Hot Spots 703 life Fellowship,and a GraduateFellowship from the Complexitiesand future directions.American Universityof BritishColumbia. The manuscriptwas Zoologist27:401-409. completedwith supportof the SwissAcademy of CAUGHLEY,G., AND A. R. E. SINCLAIR.1994. Wildlife Sciences.This is publicationnumber 120 of the Klu- ecology and management.Blackwell Scientific ane BorealForest Ecosystem Project. Publications, Oxford. CHARNOV,E. L. 1976.Optimal foraging and the mar- LITERATURE CITED ginal valuetheorem. Theoretical Population Bi- ology 9:129-136. A•RAt•S,P. A. 1982.Functional responses of optimal DAVIES,N. B. 1980. The economicsof territorial be- foragers.American Naturalist 120:382-391. haviour in birds. Ardea 68:63-74. A•RAt•S, P. A. 1987. The functional responsesof DONAZAR, J. A., E HIRALDO, M. DELIBES,AND R. R. adaptiveconsumers of two resources.Theoreti- ESTRELLA.1989. Comparative food habits of the cal PopulationBiology 33:262-288. EagleOwl Bubobubo and the GreatHorned Owl ADAMCIK, R. $., g. W. TODD, AND L. B. KEITH. 1978. Bubovirginianus in six Palearcticand Nearcticbi- Demographicand dietary responsesof Great omes. Ornis Scandinavica 20:298-306. HornedOwls duringa snowshoehare cycle. Ca- DOYLE,F. I., ANDJ. N.M. SMITH.1994. Population re- nadian Field-Naturalist 92:156-166. sponsesof NorthernGoshawks to the 10-year ANDERSSON, M. 1980. Nomadism and nest site te- cyclein numbersof snowshoehares. Studies in nacity as alternative reproductive tactics in Avian Biology 16:122-129. birds. Journalof Ecology49:175-184. EBERSOLE,J.P. 1980. Food density and territory size: ANGELSTAM, P., E. LINDSTR(SM,AND P. WIDI•N. 1984. An alternative model and a test on the reef Roleof predationin short-termpopulation fluc- Eupomacentrusleucostictus. American Naturalist tuations of some birds and in Fenno- 115:492-509. scandia.Oecologia 62:199-208. FRAFJORD,K., D. BECKER,AND A. ANGERBJ(SRN.1989. ARDITI,R., AND L. R. GINZBURG.1989. Coupling in Interactions between and red in predator-prey dynamics: Ratio-dependence. Scandinavia--Predation and aggression.Arctic Journalof TheoreticalBiology 139:311-326. 42:354-356. BEeHARD,M.J. 1982.Effect of vegetativecover on for- FRETWELL,$. D. 1972.Populations in a seasonalen- aging site selectionby Swainson'sHawk. Con- vironment. Princeton University Press, Prince- dor 84:153-159. ton, New Jersey. BENDER,E.g., t. J. CASE,AND M. E. GILPIN. 1984. Per- FujII, K., C. S. HOLLING, AND 1>. M. MACE. 1986. A turbation experimentsin community ecology: simplegeneralized model of attackby predators Theoryand practice.Ecology 65:1-13. and parasites.Ecological Research 1:141-156. BOUTIN,$. 1990.Food supplementation experiments GESE, E. M., O. J. RONGSTAD,AND W. R. MYTTON. with terrestrialvertebrates: Patterns, problems, 1988.Home rangeand habitatuse of coyotesin and the future. CanadianJournal of Zoology68: southeasternColorado. Journalof Wildlife Man- 203-220. agement52:640-646. BOUTIN, S., C. J. KREBS,R. BOONSTRA,M. R. t. DALE, HALL,P., AND S. R. WILSON.1991. Two guidelinesfor S. J. HAINON, K. MARTIN, A. R. E. SINCLAIR,J. bootstraphypothesis testing. Biometrics 47:757- N.M. SMITH, g. TURKINGTON,M. BLOWER,A. BY- 762. ROM, E I. DOYLE,C. DOYLE,D. HIK, L. HOFER,A. HANSKI,I. 1991.The functionalresponse of preda- HUBBS, T. KARELS, D. L. MURRAY, V. NAMS, M. tors: Worries about scale.Trends in Ecologyand O'DONOGHUE, C. ROHNER, AND $. $CHWEIGER. Evolution 6:141-142. 1995.Population changes of the vertebratecom- HASSELL,M.P., AND R. M. MAY. 1974. Aggregation munity during a snowshoehare cyclein Cana- of predatorsand insectparasites and its effects dds boreal forest. Oikos 74:69-80. on stability. Journalof Animal Ecology43:567- BREITENMOSER,U., B. G. , AND C. BREITEN- 594. MOSER-WORSTEN.1993. Predators of cyclicprey: HiK, D.$. 1995.Does risk of predationinfluence pop- Is the victim or profiteer of the ulationdynamics? Evidence from the cyclicde- snowshoehare cycle?Oikos 66:551-554. of snowshoe hares. Wildlife Research 22: BROWN,J. L. 1964.The evolutionof diversityin avian 115-129. territorial systems.Wilson Bulletin 76:160-169. HOLLING,C. $. 1959.Some characteristics of simple BROWN,J. S., AND R. A. MORGAN. 1995. Effects of for- typesof predationand parasitism. Canadian En- agingbehavior and spatialscale on diet selectiv- tomologist91:385-398. ity: A testwith .Oikos 74:122-136. HOLLING,C. $. 1965. The functional responseof CARPENTER,EL. 1987a. Food abundance and terri- predatorsto prey densityand its rolein mimicry toriality: To defend or not to defend?American and populationregulation. Memoirs of the En- Zoologist 27:387-399. tomologicalSociety of Canada45:1-60. CARPENTER,F. L. 1987b.The study of territoriality: HOLT,R. D., ANDJ. H. LAWTON.1994. The ecological 704 ROHNERAND KREBS [Auk, Vol. 115

consequencesof shared natural enemies.An- AND R. TURKINGTON.1995. Impact of food and nual Reviewof Ecologyand Systematics25:495- predationon the snowshoehare cycle.Science 520. 269:1112-1115. HOUSTON,C. S. 1987.Nearly synchronouscycles of LINDSTR•)M, E., P. ANGELSTAM, P. WIDI•N, AND H. the Great Horned Owl and snowshoe hare in ANDRI•N.1987. DO predatorssynchronize vole Saskatchewan.Pages 56-58 in Biologyand con- and fluctuations?--An experiment. Oi- servationof northern forest owls (R. W. Nero, R. kos 48:121-124. J. Clark, R. J. Knapton, and R. H. Hamre, Eds.). LITVAITIS,J. A., J. A. SHERBURNE,AND J. A. BISSONET- Forest Service General Technical TE.1986. habitat use and home range size Report RM-142. in relation to prey density.Journal of Wildlife HOUSTON, C. S., AND C. M. FRANCIS. 1995. Survival Management 50:110-117. of Great Horned Owls in relation to the snow- MCINVA1LLE, W. B., AND L. B. KEITH. 1974. Predator- shoehare cycle.Auk 112:44-59. preyrelations and breeding biology of the Great IMS,R. A., ANDH. STEEN.1990. Geographical syn- Horned Owl and Red-tailed in central Al- chronyin microtinepopulation cycles: A theo- berta. Canadian Field-Naturalist 88:1-20. retical evaluation of the role of nomadic avian M1L1NSKI,M., AND G. g. PARKER.1991. Competition predators.Oikos 57:381-387. for resources.Pages 137-168 in Behavioural JANES,S.W. 1985.Habitat selectionin raptorial birds. ecology:An evolutionaryapproach, 3rd ed. (J.R. Pages159-188 in Habitat selectionin birds (M.L. Krebs and N. B. Davies, Eds.). BlackwellScien- Cody, Ed.). AcademicPress, Orlando, . tific Publications, Oxford. JEDRZEJEWSKI,W., B. JEDRZEJEWSKA,g. SZYMURA, MITCHELL,W. A., ANDJ. S. BROWN.1990. Density-de- AND K. ZUB. 1996. Tawny Owl (Strix aluco)pre- pendentharvest rates by optimalforagers. Oikos dation in a pristine deciduousforest (Bialowieza 57:180-190. National Park, Poland). Journal of Animal Ecol- MUROOCH, M. W., AND g. OATEN. 1975. Predation ogy 65:105-120. and populationstability. Advances in Ecological JENNY,D. 1992. Bruterfolg und Bestandsregulation Research 9:2-131. einer alpinen Populationdes SteinadlersAquila MYERS, J.P., P. G. CONNORS,AND E PITELKA. 1979. chrysaetos.Der OrnithologischeBeobachter 89:1- Territorysize in wintering Sanderlings:The ef- 43. fects of prey abundanceand intruder density. KEITH,L. B. 1963.Wildlife's ten year cycle.University Auk 96:551-561. of Press, Madison. NAMS, V. O. 1990. Locate II. Pacer, Truro, Nova Sco- KEITH, L. B., AND D. H. RUSCH. 1988. Predation's role tia. in the cyclicfluctuations .Pages NEWTON,I. 1992. Experimentson the limitation of 699-732 in Acta XIX CongressusInternationalis numbersby territorialbehaviour. Biological Ornithologici(H. Ouellet,Ed.). Ottawa,, Reviewsof the CambridgePhilosophical Society 1986. National Museum of Natural Sciences,Ot- 67:129-173. tawa. NORRDAHL, K., AND E. KORPIM.•KI. 1996. DO nomadic KEITH, L. B., A. W. TODD, C. J. BRAND,R. S. ADAMCIK, avian predatorssynchronize population fluctu- ANDD. H. RUSCH.1977. An analysisof predation ations of small mammals?A field experiment. during a cyclicfluctuation of snowshoehares. Oecologia107:478-483. Proceedingsof the International Congressof O'DONOGHUE,M., S. BOUTIN,C. J. KREBS,AND E. J. Game Biologists13:151-175. HOFER.1997. Numerical responsesof KENWARD,g. E. 1985. Raptor radio-trackingand te- and lynx to the snowshoehare cycle.Oikos 80: lemetry.ICBP Technical Publication 5:409-420. 150-162. KENWARD,g. E. 1987.Wildlife radio-tagging:Equip- O'DONOGHUE,M., AND C. J. KREBS.1992. Effects of ment, field techniques,and data analysis.Aca- supplementalfood on snowshoehare reproduc- demic Press, London. tion and juvenilegrowth at a cyclicpopulation KENWARD,R. E. 1990. RangesIV. Institute of Terres- peak. Journalof Animal Ecology61:631-641. trial Ecology, Furzebrook, Wareham United PITELKA, F. A., P. Q. TOMICH, AND G. W. TREICHEL. Kingdom. 1955.Ecological relations of jaegersand owlsas KENWARD, R. E., V. MARCSTR•)M,AND M. KARLBOM. lemming predators near Barrow, Alaska. Eco- 1981. Goshawk winter ecology in Swedish logicalMonographs 25:85-117. pheasanthabitats. Journal of Wildlife Manage- POLLOCK,K. H., J. D. NICHOLS, C. BROWNIE,AND J. E. ment 45:397-408. HINES.1990. Statistical inference for capture-re- KORPIM;•KI,E. 1994. Rapid or delayed tracking of captureexperiments. Wildlife MonographsNo. multi-annual vole cyclesby avian predators? 107. Journalof Animal Ecology63:619-628. PRESTON,C. J. 1990.Distribution of raptor foraging KREBS,C. J., S. BOUTIN, R. BOONSTRA,A. R. E. SIN- in relationto preybiomass and habitat structure. CLAIR, J. N.M. SMITH, R. t. DALE, K. MARTIN, Condor 92:107-112. July1998] PredatorResponse toPrey Hot Spots 705

PULLIAM,H. R. 1974.On the theoryof optimaldiets. SCHOENER,t. W. 1983. Simple models of optimal American Naturalist 108:59-75. feeding-territory size: A reconciliation.Ameri- ROHNER, C. 1995. Great Horned Owls and snowshoe can Naturalist 121:608-629. hares:What causesthe timelag in thenumerical SOUTHERN,H. N., ANDV. P.W. LOWE.1968. The pat- responseof predatorsto cyclicprey? Oikos 74: tern of distributionof prey and predation in 61-68. Tawny Owl territories.Journal of Animal Ecol- ROHNER,C. 1996.The numericalresponse of Great ogy 37:75-97. HornedOwls to the snowshoehare cycle:Con- STEEN,H. 1995.Untangling the causesof disappear- sequencesof non-territorial "floaters" on de- ance from a local population of root ,Mi- mography.Journal of Animal Ecology65: 359- crotusoeconomus: A test of the regionalsynchro- 370. ny hypothesis.Oikos 73:65-72. ROHNER, C. 1997. Non-territorial "floaters" in Great STEPHENS,D. W., ANDJ. R. KREBS.1986. Foraging the- Horned Owls: Spaceuse during a cyclicpeak of ory. Princeton University Press,Princeton, New snowshoe hares. Animal Behaviour 53:901-912. Jersey. ROHNER, C., AND F. I. DOYLE. 1992. Methods of lo- TEMELES,E. J. 1987.The relativeimportance of prey catingGreat Horned Owl nestsin theboreal for- availabilityand intruderpressure in feedingter- est.Journal of RaptorResearch 26:33-35. ritory sizeregulation by harriers,Circus cyaneus. ROHNER,C., ANDC. J.KREBS. 1996. Owl predationon Oecologia74:286-297. snowshoehares: Consequences of antipredator THOMPSON, I.D., AND P. W. COLGAN. 1987. Numeri- behaviourOecologia 108:303-310. cal responsesof martensto a food shortagein ROHNER, C., AND J. N.M. SMITH. 1996. Brood size northcentral Ontario. Journal of Wildlife Man- manipulations in Great Horned Owls Bubovir- agement 51:824-835. ginianus:Are predatorsfood limited at the peak VILLAGE,A. 1982.The home range and densityof of prey cycles?Ibis 138:236-242. Kestrels in relation to vole abundance. Journal of ROSENZWEIG, M. L., AND R. H. MACARTHUR. 1963. Animal Ecology51:413-428. Graphicalrepresentation and stability condi- WALTERS,C. J. 1986. Adaptive managementof re- tions of predator-prey interactions.American newable resources. Macmillan, . Naturalist 97:209-223. WARD, R. M. E, AND C. J. KREBS.1985. Behavioural RUSCH,D. H., E. C. MESLOW,L. B. KEITH, AND P. D. responsesof lynx to decliningsnowshoe hare DOERR.1972. Responseof Great Horned Owl abundance.Canadian Journal of Zoology 63: populationsto changingprey densities.Journal 2817-2824. of Wildlife Management36:282-296. YDENBERG,R. C. 1987.Nomadic predators and geo- SALTZ,D., AND G. C. WHITE. 1990. Comparisonof graphicalsynchrony in microtinepopulation cy- different measuresof telemetry error in simu- cles. Oikos 50:270-272. lated radio-telemetrylocations. Journal of Wild- life Management54:169-174. Associate Editor: K. L. Bildstein