In 1831 the manager of a Hudsons Bay Company postin northern Ontario w - PDF document

1 In 1831 the manager of a Hudson’s Bay Co
In 1831 the manager of a Hudson’s Bay Company postin northern Ontario wrote to the head office in London.The local Ojibway Indians were starving,he reported,becauseofa scarcity of“rabbits,”and they were unable to trap for fursbecause they spent all their time fishing for food (Winter-halder 1980).These shortages ofso-called rabbits,which ap-parently occurred approximately every 10 years,are regularlymentioned in Canadian historical documents from the 18thand 19th centuries.Those rabbits were in fact snowshoe hares(Lepus americanus),and their 10-year cycle is one ofthe mostintriguing features ofthe ecology ofthe boreal forest.Ten-year cycles were first analyzed quantitatively whenwildlife biologists began to plot the fur trading records ofHudson’s Bay Company during the early 1900s.The Hud-son’s Bay Company,established in 1671,kept meticulousrecords ofthe numbers offurs traded from different postsspread across Canada.The most famous time series drawntogether from those records was that ofCanada lynx (Eltonand Nicholson 1942;Figure 1).The lynx is a specialist preda-tor ofsnowshoe hares,and the rise and fall in lynx numbersmirrors,with a slight time lag,the rise and fall ofsnowshoehare populations across the boreal region.The spectacular cycles ofsnowshoe hares and their preda-tors have captured the attention ofbiologists as well as his-torians.These cycles are highlighted in virtually all ecologytexts and are often cited as one ofthe few examples ofLotka-Volterra predator–prey equations,a simple modelwhich shows never-ending oscillations in the numbers ofpredators and their prey.Cycles seem to violate the implicitassumption ofmany ecologists that there is a balance in na-ture,and anyone living in the boreal forest would be hardpressed to recognize a balance among the boom and bust innature’s economy.The challenge to biologists has been to un-derstand the mechanisms behind these cycles,which has notbeen easy.One cycle lasts 10 years,and few PhD students orresearchers wish to take 10 years to obtain n= 1.Fortunately,over the last 40 years ecologists working in Alberta,theYukon Territory,and Alaska have put together an array ofstudies that have resolved most,but not all,ofthe enigmasbehind these cycles (Keith 1990,Boutin et al.1995).To understand any fluctuating population,one must firstknow in detail the mechanisms ofchanges in births,deaths,and movements that are the proximate causes ofthe changesin numbers.Before we describe these details,we shouldnote that these 10-year hare cycles tend to occur in synchronyacross broad regions.Indeed,hares across most ofCanadaand Alaska reached a peak in 1997–1999 during the most re-cent cycle.We explain the reasons behind this synchrony be-low,but let us note here that movements ofhares cannotexplain these population changes via immigration or emi-gration.Movements on a local level might be important,butat the regional level all populations rise and fall in unison.Population changes must be driven by changes in birthsand deaths.January 2001 / Vol.51 No.1• BioScience25 Articles Charles J. Krebs (e-mail: krebs@zoology.ubc.ca),an ecologist in the Department of Zoology at the University of British Columbia,Vancouver,B.C.V6T 1Z4,has been studying population cycles for 41 years; A. R. E. Sinclair,also with the Department of Zoology at the University of BritishColumbia,is a population and community ecologist who works extensively in East Africa on ungulate and bird populations and in the Yukon onhare population cycles. Rudy Boonstra,a population biologist in the Division of Life Sciences,University of Toronto,Scarborough,Ontario M1C1A4,is interested in the role of stress in population dynamics of mammals. Stan Boutin is an ecologist in the Department of Biological Sciences,University of Alberta,Edmonton,Alberta T6G 2E9; he has been studying northern mammal populations for 21 years. © 2001 American Instituteof Biological Sciences.What Drives the 10-yearCycle of Snowshoe Hares?CHARLES J. KREBS,RUDY BOONSTRA,STAN BOUTIN AND A.R.E. SINCLAIR THETEN-YEARCYC

2 LEOFSNOWSHOEHARES—ONEOFTHEMOSTSTRIKINGFE
LEOFSNOWSHOEHARES—ONEOFTHEMOSTSTRIKINGFEATURESOFTHEBOREALFOREST—ISAPRODUCTOFTHEINTERACTIONBETWEENPREDATIONANDFOODSUPPLIES,ASLARGE-SCALEEXPERIMENTSINTHEYUKONHAVEDEMONSTRATED Articles 26BioScience • January 2001 / Vol.51 No.1 Articles Reproductive and survival changesSnowshoe hares can have three or four litters over a summer,with five leverets on average in each litter.All hares begin tobreed in spring when they are 1 year old,so age at sexual ma-turity is a constant.Ifreproductive rates are to vary,only lit-ter size or the number oflitters can change.Lloyd Keith andhis students at the University ofWisconsin,working in cen-tral Alberta,supplied the first detailed description ofthe wayin which hares change their reproductive rate over a 10-yearcycle (Cary and Keith 1979).Reproductive output reachesits peak very early in the phase ofpopulation increase,whenfemales are producing 16–18 young per summer.It then be-gins to fall rapidly while numbers are still rising;it reachesa nadir in the year ofthe density peak or 1–2 years thereafter,during the decline phase ofthe cycle.The response ofre-productive output to hare density seems to lag 1–2 years,sothat,as shown in Figure 2,for example,the lowest repro-ductive output occurred in 1973,2 years after the densitypeak of1971.We found a similar but not identical patternofresponse in hares in the southwestern Yukon.The re-productive rate at the cyclic peak in the Yukon was about two-thirds ofthe maximum,compared with one-halfofthemaximum in the Alberta peak.But reproduction in both ar-eas was minimal in the decline phase (about one-third ofmaximum reproduction) and highest in the early increasephase ofthe 10-year cycle (Stefan 1998,Hodges 2000).Changes in mortality rates are the other driver ofchangesin hare numbers over the cycle.At Kluane Lake in the Yukonwe measured survival rates ofhares with radio collars dur-ing two population cycles.Additional data on survival ofhares over two cycles in central Alberta was obtained throughmark-and-recapture methods and radiotelemetry (Keith1990).The pattern ofchange for adult hare survival isshown in Figure 3.Adult survival rates begin to drop slowly Figure 1.Canada lynx fur returns from the Northern Department ofthe Hudson’s Bay Company from 1821 to 1910.TheNorthern Department occupied most ofwestern Canada.The cycle for these data averages 9.6 years.Data are from Eltonand Nicholson (1942).Photo: Mark O’Donoghue. Figure 2.Changes in the annual reproductive output offemale snowshoe hares in the Rochester area ofcentral Alberta,1962–1976.Reproductive output was measured in autopsy samples.Data from Cary and Keith (1979).Photo: Alice Kenney. as the population increases to a peak but then drops dra-matically for 1 to 2 years,part ofthe cause ofthe collapsein population numbers (Krebs et al.1986,Hodges 2000a).Once low numbers are reached,adult survival rates im-prove slowly but do not reach the maximum until 4 to 5 yearsafter the peak.Juvenile survival can be broken down into preweaning sur-vival for the first 30 days oflife,and postweaning survivalfrom 30 days to 1 year ofage the following April.Post-weaning survival follows the pat-tern already illustrated in Figure 3for adult hares.Preweaning survivalis more difficult to measure,and wehave data only from the Yukon forthis stage ofthe life cycle.We cagedpregnant hares caught in the wildwhen they were near term and thenradio-tagged the leverets immedi-ately after birth when the cage wasremoved.Figure 4 shows that sur-vival is very poor in this early stage.Relatively high survival occurs onlyin the first 2 to 3 years ofthe in-crease phase,and survival is alreadylow before peak hare density isreached.Preweaning survival re-mains low in the peak and at least 2years into the decline phase,and thelowest survival occurs near the endofthe decline when hare density isalready very low (Stefan 1998).The demographic pattern ofthehare cycle is remarkably clear andconsistent.Lloyd Keith and colleagues found the samechanges in central Alberta cycles that we found in

3 south-western Yukon cycles (Keith and Wi
south-western Yukon cycles (Keith and Windberg 1978).The keyfinding is that both reproduction and survival rates beginto decay in the increase phase ofthe cycle,2 years before peakdensities are reached.Maximal reproduction and highest sur-vival rates occur early in the increase phase ofthe cycle.Bythe late increase phase,reproduction has already slowedand survival rates ofboth adults and juveniles are falling.January 2001 / Vol.51 No.1• BioScience27 Articles Figure 3.Changes in adult hare survival rates over the 10-year cycle at Kluane Lake,Yukon,from 1977 to 1996.Hare density(histogram) in spring ofyear t is plotted along with survival rates averaged from spring ofyear t tot+1for radio-collaredhares in control areas.Too few hares were captured in 1985–1987 to estimate survival accurately. Figure 4.Preweaning survival ofsnowshoe hares over a population cycle at KluaneLake,Yukon Territority.Leverets were radio-tagged at birth and followed for 30days.It was impossible to obtain any juveniles in the summer of1993.Hare densityin spring is shown by the histogram. Both reproduction and survival rates continue to fall for 2to 3 years after the peak ofthe cycle,and over the low phasethey start to recover to high values.Causes ofthe cycleWhat causes these changes in reproduction and survival?There are three main factors that seem most likely to causehare cycles:food,predation,and social interactions.In ad-dition to these single-factor explanations,two multifactorexplanations have been suggested,one involving food andpredation,and the other—the most complex hypothesis—involving all three factors.Many other factors might affect snowshoe hare cycles,butthese seem more likely to be modifying influences than pri-mary causes.Disease and parasitism are two ecological fac-tors that might affect hare populations but do not seem tobe an essential cause ofcycles.Parasite loads,for example,might cause hares to be in poor condition and therefore moresusceptible to predators.Lloyd Keith and his students sur-veyed hare parasite loads for many years in Alberta andconcluded that none ofthe many parasites ofthe harescaused much direct mortality (Keith et al.1985).Experi-mental work with antihelminthics in field populations ofhares either had no measurable impact on reproductionand survival (Sovell and Holmes 1996) or produced mini-mal effects (Murray et al.1997).The conclusion is that dis-ease and parasites may affect some hare populationssporadically (see reports in Chitty 1948,1950,for example),but they cannot be an essential cause ofcycles.The food hypothesis is attractive because it can explainboth why reproduction changes over the cycle and why survival might change as well.There are two variants ofthe food hypothesis.First,hares may run out offood andstarve,or,second,the quality ofthe food may decline.Be-cause hares eat a variety ofgreen plants in the summer,noone has considered food shortage in summer to be an im-portant factor.Winter food plants are the small terminaltwigs ofwillow,birch,and small trees,as well as othershrubs;most studies have concentrated on the possibility thatwinter foods are limiting to hares (Keith et al.1984).Like allherbivores,snowshoe hares have preferred winter foodsand may browse a large fraction ofthese preferred plants atthe peak ofthe cycle.There is little evidence from our Yukon studies that over-all food quantity is limiting at any time.We measured foodabundance over the cycle by quantifying edible forage andwe assessed consumption rates ofmarked twigs.Con-sumption increased markedly during the peak (Smith et al.1988,Hik 1995,Krebs et al.2001),with 80% to 90% oftheavailable bog birch being consumed.However,only 20% to40% ofthe grey willow—the dominant shrub in the area—was consumed,thus indicating that food was not limitingat any time in the cycle.Ifthe absolute abundance offoodwere limiting,we should have found hares that had starved,but only about 3% ofthe hare mortalities could be directlyattributed to starvation.Alternatively,food quality could change over the

4 cycle.John Bryant at the University ofAl
cycle.John Bryant at the University ofAlaska suggested one at-tractive qualitative food hypothesis based on secondarychemicals:Shrubs and small trees can fight back againstbrowsers by increasing their content ofsecondary chemicalssuch as tannins and resins,which deter digestion in herbi-vores (Bryant 1981).Indeed,experimental browsing ofshrubs in Alaska has shown that the plants can respond todamage by increasing their secondary chemical defenses(Bryant et al.1985).The key question is whether these plantchanges can influence the hare cycle.To answer this question,we conducted five food-additionexperiments during two hare cycles in the southwesternYukon.In four we provided high-quality rabbit chow with-out limit to hares;in the fifth experiment we added high-quality natural food to a declining hare population (Krebset al.1985,1995,Sinclair et al.1988).The response ofharesto rabbit chow is classic:Hares move into the food-additionareas and their density increases approximately two- tothreefold in comparison with control areas.But once the den-sity increases on the food-addition areas,the hare cyclecontinues unchanged.Hares decline in number at the sametime and at the same rate on the food areas as on unma-nipulated controls.Artificial food-addition experiments have been criticizedbecause the added food is high quality and not natural.Wetried to address this criticism by supplying natural food toone declining hare population.Tony Sinclair and JamieSmith had shown that snowshoe hares largely avoided smallwhite spruce trees because the needles contain camphor(Sinclair and Smith 1984,Rodgers and Sinclair 1997).Con-sequently,small spruce seedlings were the least preferred foodin cafeteria trials with hares.But foliage from large whitespruce trees with branches beyond the reach ofhares con-tain no camphor,and these branches become highly pre-ferred food when supplied in a cafeteria trial.Thisobservation was dramatically verified when a large whitespruce tree was blown over by a windstorm:Hares devouredthe fallen branches.We therefore decided to feed a popula-tion ofhares through a decline by cutting down whitespruce trees and thus providing natural,highly preferred foodto a collapsing hare population.Stan Boutin and Scott Gilbert did this experiment overthree winters,with the results shown in Figure 5.The extranatural food produced no detectable effect on the rate ofpopulation collapse.The failure ofthis extra food to affectthe hare population decline was shown clearly on five areasin two cycles in the southwest Yukon.Such results imply thatfood shortage by itselfis not the explanation for the hare cy-cle.Whatever secondary chemical changes occur in winterfood plants,they are at most a contributing factor,not theprimary cause ofthe hare collapse.Another way ofapproaching the possible role offood inhare cycles is to improve the quality ofthe vegetation byadding nutrients in fertilizer.We ran this experiment from1987 to 1996 on two areas in the southwestern Yukon,each28BioScience • January 2001 / Vol.51 No.1 Articles 1 km2.On each area we added NPK fertilizer each spring toincrease the availability ofsoil nutrients.Boreal forest soilsare typically impoverished in nutrients,particularly nitro-gen,and nutrients added through fertilizer are immedi-ately taken up by plants.We measured plant responses tonutrient additions and found large increases in individualplant growth in grasses,shrubs,and trees (Turkington et al.1998).None ofthis plant improvement resulted in moresnowshoe hares on the fertilized areas,relative to the con-trols;we therefore concluded that the dynamics ofthe harecycle cannot be changed by nutrient additions to the ecosys-tem.The message seemed to be repeated:The hare cycle isnot driven primarily by plant–herbivore interactions.The predation hypothesis is the next most likely expla-nation for the hare cycle.In our studies with radio-collaredhares,the immediate cause ofdeath of95% ofthe hares waspredation by a variety ofpredators,the main ones in theYukon being lynx,coyotes

5 ,goshawks,and great horned owls(Rohner a
,goshawks,and great horned owls(Rohner and Krebs 1996,O’Donoghue et al.1997).Fewhares died with signs ofmalnutrition,and those that didstarve were found more often in the increase and peakphases rather than in the decline (Hodges 2000a).In con-trast to adult hares,leverets are killed by a variety ofsmallraptors,such as boreal owls,red-tailed hawks,kestrels,andhawk owls,and by small mammals,particularly red squir-rels and ground squirrels.Ofthe leverets with radio tags,81%died because ofpredation (O’Donoghue 1994).These ob-servations ofnatural history support the contention that pre-dation by a variety ofbirds and mammals plays an importantrole in the hare cycle.For mammalian predators in winter,we used snow track-ing to monitor density changes and kill rates oflynx and coy-otes.All hare predators showed strong numerical changes thatlagged behind the hare cycle 1–2 years (Boutin et al.1995).In addition,both lynx and coyotes killed more hares per dayin the peak and decline phases than during the increase.These kill rates were well above previous estimates and wellin excess ofenergy demands.Surplus killing seems to be acharacteristic feature ofthese predators.To test the predation hypothesis,we excluded mammalianpredators from two areas by constructing electric fences inthe Yukon,each 1 km2.We also attempted to exclude avianpredators from smaller areas inside the electric fence bymeans offish netting and monofilament fishing line strungin trees,but these proved impractical and ineffective.Thefence was permeable to hares and other small mammals,which could move in and out at will.In one fenced area wealso added food,so we had a combined treatment manip-ulating predation pressure and food supplies (Krebs et al.1995).Because ofthe size ofthe exclosures and the main-January 2001 / Vol.51 No.1• BioScience29 Articles Figure 5.Changes in snowshoe hare numbers on control (1050,red) and food-supplemented (blue) areas during thepopulation decline of1981–1983 at Kluane,Yukon.The natural feeding experiment was begun in October 1981 (bluetriangle).Summer months are shaded yellow.Data are from Krebs et al.(1985). tenance they required,we were unable to replicate thesefence treatments.The main impact ofthe electric-fence predator exclo-sures was to increase the survival rate ofradio-collaredhares.Figure 6 shows that the collapse in survival that nor-mally occurs in the peak and decline phases ofthe cycle wasnearly eliminated by the exclusion ofmammalian predators.There is a slight decrease in survival during the declinephase with the combined predator exclosure and food treat-ment,but the impact ofadditional food on survival duringthe decline is small.The mortality rate is almost entirely dri-ven by predation.Two further conclusions follow from these experiments.Inasmuch as avian predators had access to the two preda-tor exclosures,avian predation by itselfis not sufficient toexplain the changes in survival rates.There is a slight re-duction in survival inside the predator exclosures during thedecline (Figure 6),and this is a measure ofthe impact ofbirdpredation by itselfon hare survival.Since many species ofpredators are involved in causing the collapse in prewean-ing,juvenile,and adult hare survival,we cannot pinpoint therole ofany individual predator species.In particular,thesnowshoe hare cycle is not strictly a lynx–hare cycle,asmany textbooks claim,and ifthe lynx is removed from thepredator community—as it is on Anticosti Island in the St.Lawrence River in eastern Canada—the hare cycle contin-ues unchanged because ofpredator compensation (Stensethet al.1998).Ifwe can explain the changes in mortality rates by predation,how can we explain the changes in reproductionthat accompany the cycle? We have attacked this question experimentally by providing hares on two open areas withsupplemental food year-round.We measured annual re-productive output for female hares by multiplying preg-nancy rates and average litter sizes for each ofthe summerlitters and summing these for each breeding season.Figure

6 7 summarizes our data.Reproductive outpu
7 summarizes our data.Reproductive output at KluaneLake was severely reduced in the decline phase,as Lloyd Keith(Cary and Keith 1979) found in Alberta (Figure 2).Addingfood during the peak ofthe cycle in 1989 and 1990 had noimpact on reproductive output (O’Donoghue 1994).How-ever,during the decline phase in 1991 and 1992,the preda-tor exclosure plus food treatment caused a dramatic increasein reproductive output over that found in control hares(Hik 1995,Stefan 1998).Because we were unable to obtainthese reproductive data on food addition sites in 1991 and1992,we do not know ifit is the additional food or the ex-clusion ofmammalian predators,or a combination ofthesefactors,that produced the results shown in Figure 7.Indirect effectsHow might predator exclusion affect reproductive output?There has been much interest in the indirect effects ofpreda-tors on prey (Lima 1998).One way in which predatorsmight cause a decline in reproductive output is by stressinghares through high rates ofencounter and repeated unsuc-cessful attacks during the peak and decline phases (Boon-stra et al.1998a).Chronic stress has many direct detrimentaleffects on mammals,including reduction in reproductiverate,mobilization ofenergy reserves,and increased sus-ceptibility to diseases and parasites.Stress effects may alsobe indirect and long term,affecting adult brain function(Lupien and McEwen 1997) and offspring viability and be-havior (Matthews 2000).Figure 8 illustrates these ideas fora hypothetical hare cycle in which predation pressure causeschronic stress.Unfortunately,we have not yet been able tomeasure stress levels through an entire hare cycle,but the datawe have from declining and low populations is consistentwith this hypothetical scheme.What is the role ofhigh-quality food in the hare cycle?Population density increased on areas provided with sup-plemental food.Hares on the areas provided with supple-mental food did not lose weight over winter and in generalwere in better body condition than control hares.Foodquality clearly limits body condition and population den-sity in hares.In addition to these direct effects,changes infood quality could have subtle indirect effects on hares,contributing to chronic stress and making them more or less30BioScience • January 2001 / Vol.51 No.1 Articles Figure 6.Changes in survival rates ofsnowshoe hare numbers on controland fenced areas during thepopulation cycle of1988–1996 atKluane,Yukon.The arrows show themean survival rate for eachtreatment.The survival rate per 30days is averaged over each year,with90% confidence limits,for radio-collared hares. able to avoid predators,parasites,and disease (Hik 1995,Boonstra et al.1998a,Hodges et al.1999).Poorer body con-dition in the peak ofthe cycle is one consequence when haresovergraze their preferred winterfoods and suffer from chronicstress,causing them to mobilizeenergy.This loss ofcondition maylead to reduced ability to avoidpredators and thus to an increase inhares’stress levels.The impact ofextra food would therefore not di-rectly affect reproduction and mor-tality rates,but it would actindirectly to make hares less sus-ceptible to stress.Indirect effectshave limited power to affect popu-lation dynamics,however,and allour feeding experiments show thatextra food does not make hares lesssusceptible to predation and cannotprevent the hare population de-cline.Another possible source ofstressfor hares is social interactions.Snowshoe hares are not particu-larly good candidates for social reg-ulation ofpopulation size,andmost hypotheses about hare cycleshave ignored the social dimension(Krebs 1986).Hares are not territorial,and aggression seemsto be restricted to males competing in mating chases and tofemales resisting attentive males.There are no records ofin-January 2001 / Vol.51 No.1• BioScience31 Articles Figure 7.Reproductive output offemale snowshoe hares over a population cycle at Kluane Lake,Yukon.The blue line showshare density in control populations.Supplemental food (green bars) does not affect reproductive

7 output at the peak ofthecycle,but in the
output at the peak ofthecycle,but in the decline phase of1991 and 1992 high reproductive output was sustained on the experimental area fencedfrom mammalian predators and provided with supplemental food (red bars).We were unable to measure reproductiveoutput in every year for all treatments,and thus we do not know for the critical decline years of1991 and 1992 whetherreduced reproductive output is caused by food or predation or both. Figure 8.The chronic stress hypothesis as an explanation ofthe changes in reproductionthat accompany the snowshoe hare cycle.Chronic stress is postulated to arise from signsofpredator abundance (odor,visual sightings,tracks,unsuccessful chases),and is thusrelated to predator densities.According to this hypothesis,any physiological measure ofchronic stress,such as cortisol levels,should be maximal late in the decline phase ofthehare cycle.The immune system would also be compromised because ofchronic stress.We do not yet know whether this hypothesis is the correct explanation for the reducedreproduction that occurs during the hare cycle. fanticide in hares,and without territoriality and infanti-cide there would seem to be little room for social interactionsto play a role in population dynamics.Nevertheless,there remain some puzzles.One is the lackofan explanation for the low phase,which follows the de-cline and lasts 2–4 years (Boonstra et al.1998b).A secondpuzzle is evidence implicating the role ofmaternal effects inreproductive performance.Tony Sinclair maintained a lab-oratory colony ofhares for 14 years in Vancouver to test forimpacts ofsecondary chemicals on food choice.Females weremaintained in individual cages with high-quality food andwere treated for potential parasites.Hares in captivity reg-ularly live 5–7 years,much longer than they live in nature.Therefore the colony at times consisted ofa mixture offe-males trapped live in the Yukon from peak populations andfrom low populations.Figure 9 compares the reproductiveoutput in captivity ofthese two groups ofhares (A.R.E.Sinclair,personal communication,1999).Hares taken fromlow populations maintained a lifetime reproductive outputmore than double that ofhares taken from peak populations,suggesting that female hares are already programmed fortheir reproductive success early in life,and no matter howfine an environment is provided,they cannot change.These results are completely serendipitous:This was nota planned comparison,females could not be matched for age,and this reproductive measurement was not the reason thecolony was maintained.There was no correlation in thesecolony data between the reproductive output ofmothers anddaughters,suggesting that this effect was not a genetic.Aplanned comparison ofcolony hares should be carried outto see whether the effect can be verified in another popula-tion.At present,these results must be considered unproven,but tantalizing.We are now close to understanding the snowshoe hare cy-cle.The 10-year cycle is a result ofthe interaction betweenpredation and food supplies,but ofthese two factors,pre-dation is clearly the dominant process.The impact offoodis felt largely in winter and it is mostly indirect.Hares do notusually die directly ofstarvation or malnutrition—the im-mediate cause ofdeath is virtually always predation.But foodquality and quantity affect body condition and in this waymay predispose hares to predation,increased parasite loads,and higher levels ofchronic stress.These indirect effects ofpredation and food are the probable cause ofreduced re-productive output.Hares in peak and declining popula-tions must trade offsafety and food,and these behavioraltradeoffs define the dynamics ofthe decline (Hik 1995,Hodges 2000b).The result is a time lag in both the indirecteffects and the direct effects ofpredation,which causes thecyclicity.The low phase ofthe cycle is the combined resultofcontinuing predation mortality and slowly recoveringreproductive potential.Synchrony in the hare cycleNot only do snowshoe hare numbers fluctuate in 10-year cy-cles,but these

8 cycles tend to occur in synchrony acros
cycles tend to occur in synchrony acrossmuch ofthe boreal forests ofCanada and Alaska.Synchronyis not absolute,however;different regions within Canada candrift out ofphase.Smith (1983) used questionnaire data onsnowshoe hare abundance from 1931 to 1948 to measuresynchrony across Canada (Figure 10).He concluded thatthere was a large area ofcyclic synchrony in northern Man-itoba and Saskatchewan that was 2 years ahead of“average,”and areas in southern Ontario and Quebec that were up to2 years later than average for this time period.Populationsin the Yukon were also up to 2 years later than average.Insome sense,the peak ofthe hare cycle for this time periodwas like a traveling wave in a pond,with the center in themiddle ofthe boreal forest.The result ofthis kind oftrav-eling wave is that synchrony falls offwith distance,as shown32BioScience • January 2001 / Vol.51 No.1 Articles Figure 9.The annual reproductiveoutput ofsnowshoe hare femalesmaintained in a colony atVancouver.Individual females weretaken from field populations andclassified by the state ofthe fieldpopulation at the time ofinitialcapture.For 4 years there wereenough females to comparesimultaneously the population offemales taken from the peak phaseofthe cycle with those taken fromthe low phase.Low-phase femalesmaintain a lifetime reproductiveoutput much greater than peak-phase females; these lab resultsmimic the observed changes in fieldpopulations.Numbers above barsare numbers offemales. 2 2342834 in Figure 11 for lynx fur return data from the Canadianprovinces for the period 1920–1987 (Ranta et al.1997).Because Canada lynx fur harvests are highly correlatedwith snowshoe hare numbers,the fur trading records ofthe Hudson’s Bay Company have been used extensively toanalyze synchrony in the 10-year cycle (Stenseth et al.1999,Haydon and Greenwood 2000).Regions across Canadashowed a higher degree ofsynchrony in the 19th century thanduring the 20th century,for reasons that are not clear.Within Canada,Stenseth et al.(1999) showed,there werethree broad regions within which the lynx cycle was simi-larly structured (Figure 12).These three regions coincidedwith the climatic zones defined independently from theNorth Atlantic Oscillation (NAO),a climatic oscillationsimilar to the El Niño in the Pacific Ocean.The reasons forthese correlations are not clear,though Stenseth et al.(1999)suggest that the changing snow levels associated with theNAO might affect the hunting efficiency oflynx.What ecological factors might cause synchrony acrosslarge geographic regions? Two general models for synchronycan be suggested,one driven by weather and one by dispersalmovements.The most famous model to synchronize the harecycle across North America is the weather model related tothe sunspot cycle (Sinclair et al.1993).Sunspots and harenumbers are highly correlated for three time periods dur-ing the past 250 years:1751–1787,1838–1870,and1948–1986,which were all periods ofhigh-amplitudesunspot fluctuations (Sinclair and Gosline 1997).Becausesunspots affect broad weather patterns,it is possible thatsunspots might,through weather,entrain snowshoe hare cy-cles across the continent when solar activity is unusually high.The main problem with this explanation is that there is atpresent no mechanism for sunspot-affected weather patternsto translate into demographic impacts on hares.All ourstudies in the Yukon suggest the predominant role ofpreda-tors in driving the hare cycle.As mentioned above,Stensethet al.(1999) suggested an indirect impact ofsnow depth onhunting success ofmammalian predators as one way inwhich synchrony might be coordinated,but there are no datato test this suggestion.There is also a problem in reconcil-ing the global nature ofsunspot activity with the local vari-ations in synchrony shown in Figure 10.Dispersal movements would appear to be the more likelyexplanation for synchrony.Hares disperse on a local scale ofa few kilometers (Gillis 1997,Hodges 2000b),so hare move-ments seem incapable ofaffecting synchrony except on a lo-cal scale

9 .Predator movements are more likely the
.Predator movements are more likely the key elementcausing synchrony within the boreal region.Long-rangeJanuary 2001 / Vol.51 No.1• BioScience33 Articles Figure 11.Synchrony among pairs ofprovinces in Canada for lynx furreturn data from 1920 to 1987.Thegeographic midpoint ofeach provincewas used to calculate the distancebetween provinces.The synchronyindex is the simple correlationcoefficient between the pairs ofprovinces.Synchrony falls offwithdistance and then rises again becauseBritish Columbia and the Yukon aswell as eastern Canada tend to peak2–3 years later than the average ofcentral Canada.The dashed line is asimple polynomial regression.Datafrom Statistics Canada,as used byRanta et al.(1997). Figure 10.Synchrony in snowshoe hare cycles acrossCanada,1931–1948,as measured by questionnaires(Chitty 1948,1950).The average peak phase acrossCanada was scaled as 0.0,and the contour lines indicatepeaks occurring earlier than average (red,negativecontours) or later than average (green,positive contours).During this period,hare peaks were reached earliest inthe central boreal region ofnorthern Saskatchewan andManitoba (Smith 1983). dispersal movements ofradio-collared lynx have been de-scribed in several studies (Mowat et al.2000).Documenteddispersal movements up to 1100 km have been recorded,with15 movements greater than 500 km.These dispersal move-ments were detected only because the dispersing lynx werecaught by fur trappers,and thus constitute a biased (trun-cated) set oftrue dispersal movements.Houston (1978) re-ported similar data for movements ofgreat horned owls,with36 banded owls moving 265 to1415 km before being shot,trapped,or found dead.These data for great horned owls andlynx indicate the potential for predator dispersal to syn-chronize hare cycles over large areas ofthe boreal forest re-gion (Ydenberg 1987,Ims and Steen 1990).Ifpredatorsmove from food-poor to food-rich regions on a scale of1000km,they could readily bring local regions into synchrony inthe general pattern shown in Figure 10.The predator-dispersal hypothesis for synchrony ofhare cycles is attrac-tive because it is parsimonious,since predation plays suchan important role in driving the hare cycle through its direct effects on hare mortality and its indirect effects on hareLynx,which have been declared a threatened species in thecontinental United States,and the dynamics ofits interac-tion with snowshoe hares in states with southern borealforests,such as Montana,are in critical need ofstudy (Rug-giero et al.2000).Ifhare populations are synchronized re-gionally by predator movements,the fragmentation ofhareand lynx habitat in the southern parts ofCanada and in thenorthern United States could break the ebb and flow ofthese linkages.The present shortage ofgenetic informa-tion on the geographic structure ofboth snowshoe harepopulations and lynx populations needs to be remedied todefine these spatial issues more clearly for conservation.Lotka and Volterra werepartly correct when theyguessed that the snowshoehare cycle was a predator–prey oscillation,but theymissed the critical point thatthe cycle can be understoodonly by analyzing threetrophic levels rather thantwo.The hare cycle is pro-duced by an interactionbetween predation and foodsupplies,and its biologicalimpacts ripple across manyspecies ofpredators and preyin the boreal forest.The boreal forest is one ofthe great ecosystems oftheearth,and the 10-year snow-shoe hare cycle is one ofthe most striking features ofthisecosystem.After 70 years ofquestionnaire research,timeseries analyses,and field experiments,we have a goodunderstanding ofthe dynamics behind the hare cycle andthe importance ofpredation and food supplies in regulat-ing that cycle.The snowshoe hare is a critical species in theboreal forest:Ifit should disappear,many species ofpredators would go with it,and the structure ofthe plantcommunity would be altered substantially.The boreal forest ecosystem and the 10-year cycle insnowshoe hares are clearly resilient to a variety ofnaturaldisturbances,f

10 rom forest fires to short-term climatic
rom forest fires to short-term climatic fluc-tuations,but it is unclear whether they can withstand thechanges that humans impose without collapsing.Theimpact ofdirectional climate change on the boreal forestcommunity will probably be less in the short term than theimpact offorestry and other human activities such as oiland gas exploration.We should explore the limits ofresilience ofthe snowshoe hare cycle as forest harvestingextends north into the boreal forest,lest we compromisethis fascinating system.ReferencesBoonstra R,Hik D,Singleton GR,Tinnikov A.1998a.The impact ofpredator-induced stress on the snowshoe hare cycle. Ecological Mono- graphs 68:371–394. Boonstra R,Krebs CJ,Stenseth NC.1998b.Population cycles in mammals:The problem ofexplaining the low phase. Ecology 79:1479–1488. Boutin S,et al.1995.Population changes ofthe vertebrate community dur-ing a snowshoe hare cycle in Canada’s boreal forest. Oikos 74:69–80. Bryant JP.1981.Phytochemical deterrence ofsnowshoe hare browsing by ad-ventitious shoots offour Alaskan trees.Science 213:889–890.Bryant JP,Wieland GD,Clausen T,Kuropat P.1985.Interactions ofsnowshoehare and feltleafwillow in Alaska. Ecology 66:1564–1573. 34BioScience • January 2001 / Vol.51 No.1 Articles Figure 12.Climatic regions(shaded) withinCanada,definedby the NorthAtlanticOscillation.Thelarge circles definethe three regionswithin which thelynx cycle is mostsimilar.Theseregions fit withinthe independentlydefined climaticzones.Stenseth et al.(1999). Cary JR,Keith LB.1979.Reproductive change in the 10-year cycle ofsnow-shoe hares.Canadian Journal ofZoology 57:375–390.Chitty H.1948.The snowshoe rabbit enquiry,1943–46.Journal ofAnimalEcology 17:39–44.———.1950.The snowshoe rabbit enquiry,1946–48.Journal ofAnimal Ecol-ogy 19:15–20.Elton C,Nicholson M.1942.The ten-year cycle in numbers ofthe lynx inCanada.Journal ofAnimal Ecology 11:215–244.Gillis EA.1997.Natal dispersal and post-weaning survival ofjuvenile snow-shoe hares during a cyclic population increase.Master’s thesis.Univer-sity ofBritish Columbia,Vancouver,BC.Haydon DT,Greenwood PE.2000.Spatial coupling in cyclic population dy-namics:Models and data. Theoretical Population Biology 58:239–254. Hik DS.1995.Does risk ofpredation influence population dynamics? Evi-dence from the cyclic decline ofsnowshoe hares.Wildlife Research 22:115–129.Hodges KE.2000a.The ecology ofsnowshoe hares in northern boreal forests.Pages 117–161 in Ruggiero LF,Aubry KB,Buskirk SW,Koehler GM,KrebsCJ,McKelvey KS,Squires JR,eds.Ecology and Conservation ofLynx inthe United States.Denver (CO):University Press ofColorado.———.2000b.Proximate factors affecting snowshoe hare movements dur-ing a cyclic population low phase. Ecoscience 6:487–496. Hodges KE,Krebs CJ,Sinclair ARE.1999.Snowshoe hare demography dur-ing a cyclic population low. Journal ofAnimal Ecology 68:581–594. Houston CS.1978.Recoveries ofSaskatchewan-banded great horned owls.Canadian Field-Naturalist 92:61–66.Ims RA,Steen H.1990.Geographical synchrony in microtine population cy-cles:A theoretical evaluation ofthe role ofnomadic avian predators. Oikos 57:381–387. Keith LB.1990.Dynamics ofsnowshoe hare populations.Current Mam-malogy 4:119–195.Keith LB,Windberg LA.1978.A demographic analysis ofthe snowshoehare cycle.Wildlife Monographs 58:1–70.Keith LB,Cary JR,Rongstad OJ,Brittingham MC.1984.Demography andecology ofa declining snowshoe hare population.Wildlife Monographs90:1–43.Keith LB,Cary JR,Yuill TM,Keith IM.1985.Prevalence ofhelminths in a cyclicsnowshoe hare population. Journal ofWildlife Diseases 21:233–253. Krebs CJ.1986.Are lagomorphs similar to other small mammals in their pop-ulation ecology? Mammal Review 16:187–194.Krebs CJ,Boutin S,Gilbert BS.1985.A natural feeding experiment on a de-clining snowshoe hare population. Oecologia 70:194–197. Krebs CJ,Gilbert BS,Boutin S,Sinclair ARE,Smith JNM.1986.Populationbiology ofsnowshoe hares.I.Demography offood-supplemented pop-ulations in the southern Yukon,1976–84. Journal ofAnimal Ecology 55: 963–982. Krebs CJ,Boutin S,Boons

11 tra R,Sinclair ARE,Smith JNM,Dale MRT,Ma
tra R,Sinclair ARE,Smith JNM,Dale MRT,Mar-tin K,Turkington R.1995.Impact offood and predation on the snow-shoe hare cycle. Science 269:1112–1115. Krebs CJ,Boutin S,Boonstra R,eds.2001.Ecosystem Dynamics ofthe Bo-real Forest:The Kluane Project.New York:Oxford University Press.Lima SL.1998.Nonlethal effects in the ecology ofpredator–prey interactions. Bioscience 48:25–34. Lupien SJ,McEwen BS.1997.The acute effects ofcorticosteroids on cogni-tion:Integration ofanimal and human-model studies. Brain Research Reviews 24:1–27. Matthews SG.2000.Antenatal glucocorticoids and programming ofthe de-veloping CNS. Pediatric Research 47:291–300 Mowat G,Poole KG,O’Donoghue M.2000.Ecology oflynx in northernCanada and Alaska.Pages 265–306 in Ruggiero LF,Aubry KB,BuskirkSW,Koehler GM,Krebs CJ,McKelvey KS,Squires JR,eds.Ecology andConservation ofLynx in the United States.Denver (CO):UniversityPress ofColorado.Murray DL,Cary JR,Keith LB.1997.Interactive effects ofsublethal nema-todes and nutritional status on snowshoe hare vulnerability to predation. Journal ofAnimal Ecology 66:250–264. O’Donoghue M.1994.Early survival ofjuvenile snowshoe hares. Ecology 75: 1582–1592. O’Donoghue M,Boutin S,Krebs CJ,Hofer EJ.1997.Numerical responses ofcoyotes and lynx to the snowshoe hare cycle. Oikos 80:150–162. Ranta E,Kaitala V,Lundberg P.1997.The spatial dimension in populationfluctuations. Science 278:1621–1623. Rodgers AR,Sinclair ARE.1997.Diet choice and nutrition ofcaptive snow-shoe hares (Lepus americanus):Interactions ofenergy,protein,and plantsecondary compounds. Ecoscience 4:163–169. Rohner C,Krebs CJ.1996.Owl predation on snowshoe hares:Consequencesofantipredator behaviour. Oecologia 108:303–310. Ruggiero LF,Aubry KB,Buskirk SW,Koehler GM,Krebs CJ,McKelvey KS,Squires JR,eds.2000.Ecology and Conservation ofLynx in the UnitedStates.Boulder (CO):University Press ofColorado.Sinclair ARE,Gosline JM.1997.Solar activity and mammal cycles in theNorthern Hemisphere. American Naturalist 149:776–784. Sinclair ARE,Smith JNM.1984.Do plant secondary compounds determinefeeding preferences ofsnowshoe hares? Oecologia 61:403–410. Sinclair ARE,Krebs CJ,Smith JNM,Boutin S.1988.Population biology ofsnowshoe hares.III.Nutrition,plant secondary compounds and food lim-itation. Journal ofAnimal Ecology 57:787–806. Sinclair ARE,Gosline JM,Holdsworth G,Krebs CJ,Boutin S,Smith JNM,Boonstra R,Dale M.1993.Can the solar cycle and climate synchronizethe snowshoe hare cycle in Canada? Evidence from tree rings and ice cores.American Naturalist 141:173–198.Smith CH.1983.Spatial trends in Canadian snowshoe hare,Lepus americanus,population cycles. Canadian Field-Naturalist 97:151–160. Smith JNM,Krebs CJ,Sinclair ARE,Boonstra R.1988.Population biologyofsnowshoe hares.II.Interactions with winter food plants .Journal ofAn- imal Ecology 57:269–286. Sovell JR,Holmes JC.1996.Efficacy ofivermectin against nematodes infectingfield populations ofsnowshoe hares (Lepus americanus) in Yukon,Canada. Journal ofWildlife Diseases 32:23–30. Stefan CI.1998.Reproduction and pre-weaning juvenile survival in a cyclicpopulation ofsnowshoe hares.Master’s thesis.University ofBritish Co-lumbia,Vancouver,BC.Stenseth NC,Falck W,Chan KS,Bjornstad ON,O’Donoghue M,Tong H,Boonstra R,Boutin S,Krebs CJ,Yoccoz NG.1998.From patterns toprocesses:Phase and density dependencies in the Canadian lynx cycle. Proceedings ofthe National Academy ofSciences 95:15430–15435. Stenseth NC,et al.1999.Common dynamic structure ofCanada lynx pop-ulations within three climatic regions. Science 285:1071–1073. Turkington R,John E,Krebs CJ,Dale MRT,Nams V,Boonstra R,Boutin S,Martin K,Sinclair ARE,Smith JNM.1998.The effects offertilization onthe vegetation ofthe boreal forest.Journal ofVegetation Science 9:333–346.Winterhalder BP.1980.Canadian fur bearer cycles and Cree-Ojibway hunt-ing and trapping practices.American Naturalist 116:870–879.Ydenberg RC.1987.Nomadic predators and geographical synchrony in mi-crotine population cycles. Oikos 50:270–272. January 2001 / Vol.51 No.1• BioScience35 Art