Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The ecology of movements made by Columbian black-tailed deer McNay, R. Scott 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1995-060217.pdf [ 3.79MB ]
Metadata
JSON: 831-1.0088371.json
JSON-LD: 831-1.0088371-ld.json
RDF/XML (Pretty): 831-1.0088371-rdf.xml
RDF/JSON: 831-1.0088371-rdf.json
Turtle: 831-1.0088371-turtle.txt
N-Triples: 831-1.0088371-rdf-ntriples.txt
Original Record: 831-1.0088371-source.json
Full Text
831-1.0088371-fulltext.txt
Citation
831-1.0088371.ris

Full Text

THE ECOLOGY OF MOVEMENTS MADE BY COLUMBIAN BLACK-TAILED DEERbyR. Scott McNayB.Sc.F., University of New Brunswick, 1981M.Sc., University of British Columbia, 1985A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Forestry)We accept this thesis as conforming to therequired standardTHE UNIVERSITY OF BRITISH COLUMBIAMarch 1995© Robert Scott McNay, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.Department of Fow-JThe University of British ColumbiaVancouver, CanadaDate 95.ôi. I(Signature)DE.6 (2/88)11ABSTRACTI used movements of 74 radio-collared black-tailed deer to investigatewhether an hierarchically-structured decision process constrains habitatchoices. Constraints on habitat can lead to rejection of the ideal-freedistribution hypothesis for black-tailed deer where their habitats are proneto large and rapid disturbances. I recorded 11,150 deer locations at 2 timescales (2-hourly and weekly) at 4 study areas on Vancouver Island, BritishColumbia from 1982-1991. I assessed the temporal and spatial independenceof these observations before examining the distance, frequency, timing, anddirection of movements.Deer that migrated every year (n = 12) occupied natal ranges at highelevations and used alternate ranges for >6 mo/yr. Migration routedirections varied and were used in the absence of snow. Alternate rangeswere established at mid-slope elevations on southern aspects and alwaysenclosed some old forest, the only forest type preferred more in winter thansummer. Areas of intense use on alternate ranges included old forests.Deer that migrated less consistently (n = 16) occupied natal ranges at mid-to low-slope positions and used alternate ranges for <4 mo/yr. All but 2 ofthese deer migrated in response to snow accumulation or ablation and alongroutes consistent with valley direction. They established alternate rangesat similar topographic positions as other migratory deer but did notnecessarily have access to old forests. Non-migratory deer (n = 44)occupied natal ranges at mid- to low-elevations and used whatever forestswere available. Areas of intense use tended not to include old forest. Iwas unable to trap migratory deer in young forests and the survival rate Iestimated for resident deer indicated a declining population.111Decisions concerning home- and seasonal-range establishmentconstrained more spatially specific decisions about habitat use. Further,these decisions were themselves likely to be limited by rigid tactics suchas philopatry and site fidelity.These conclusions were tested by logging old forest winter habitat in2 separate study areas. Fidelity, rather than habitat choice, dominated theinitial responses. Deer accepted remaining habitats rather than finding oldforest winter habitat elsewhere. Lack of freedom in choosing habitats hasimplications for habitat management, for deer response to habitat change,and for factors that affect population dynamics.TABLE OF CONTENTSivAbstractTable of Contents.11ivList of Tables viiList of FiguresAcknowledgementsForewordxxiiixivChapter 1 GENERAL OVERVIEWBackgroundThesis StructureLiterature Cited1145INDEPENDENCE OF OBSERVATIONS INMethodsDeer Location SamplesAnalytical TechniquesResultsIndependence of Location810101213Observations 14SPATIAL AND TEMPORAL SCALES OF RESOURCE USE:MOVEMENTSStudy AreasMethodsDeer Location SamplesFactors Used to Identify Spatial andTemporal ScalesAnalytical ProceduresResultsDeer Location SamplesDispersal: Life-time Scale . .Migration: Seasonal ScaleLocal Movement: Daily Scale . .Serial Movement: Hourly Scale .DiscussionMovements as Scalar Classes of Activity1432353838394249545656Chapter 2Chapter 3MOVEMENTSEffect of Migrations on Independence ofLocationsIndependence of Distance Between ConsecutiveLocationsDiscussionManagement ImplicationsLiterature Cited1818212326303232Movements as Hierarchically Structured DecisionsDecision Information Transfer andHierarchical FunctionImplications of Constraint on Movement DecisionsLiterature Cited• • 57• . 58• . 66• . 67VChapter 4 SPATIAL AND TEMPORAL SCALES OF RESOURCE USE:HABITAT PREFERENCE 75Study Areas 78Methods 80Deer Location Samples and Habitat Use 80Definitions and Habitat Features 81Analytical Procedures 83Results 86Sample Characteristics 86Home Range Preferences: Life-time Scale 89Habitat Preferences: Seasonal Scale 95Activity Nuclei Preferences: Daily Scale 110Discussion 113Hierarchy and Constraint of Habitat Choices . . . 113Interpretation of Habitat Preference 116Implications for Research and Management 120Literature Cited 121Chapter 5 RESPONSE TO LOGGING OF WINTER HABITAT 127Study Areas 128Methods 129Experimental Design 129Animal Capture and Monitoring 130Statistical Analysis 131Results 132Discussion 136Management Implications 137Literature Cited 139Chapter 6 MORTALITY CAUSES AND SURVIVAL ESTIMATES 143Study Areas 144Methods 146Mortality Causes 146Mortality and Survival Rate Estimation 147Results 148Estimates of Cause-specific Mortality 148Factors Affecting Mortality and Survival 151Discussion 156Causes of Mortality 156Factors Affecting Survival 157Management Implications 158Literature Cited 160Chapter 7 MANAGEMENT IMPLICATIONS OF CONSTRAINTS ON MOVEMENTS TACTICS 164Study Areas 165Methods 166Deer Location Samples 166Definitions and Analytical Procedures 167Results and Discussion 170Movements 170Habitat Use 176Response to Removal of Winter Habitat 179Survival Estimates 181Management Implications 182viLiterature Cited 185Chapter 8 GENERAL CONCLUSIONS 189Independence of Observations in Movements 189Spatial and Temporal Scales in Movements 190Spatial and Temporal Scales in Habitat Use 191Response to Winter Habitat Logging 192Mortality Causes and Survival Estimates 193Literature Cited 194viiLIST OF TABLESTable 2.1 Percentage of total, or seasonal, black-tailed deerhome ranges where the hypothesis of independence oflocation observations was not rejected. Values inparentheses are the number of total, or seasonal,home ranges tested from data collected during 1982-1991 on Vancouver Island, British Columbia . . . . 15Table 3.1 Physical features of 4 study areas located onVancouver Island, British Columbia 31Table 3.2 Straight-line distance between the estimated natalsite and the site occupied in their last natalseason for radio-collared, black-tailed deer caughtas subadults (<2-yr-old) at 4 study sites onVancouver Island, British Columbia, 1982-1991 43Table 3.3 Movement characteristics for 3 behaviour groups ofradio-collared, black-tailed deer on VancouverIsland, British Columbia, 1982-1991. Moves arestraight-line distances between successive locations(using 2-hour data for serial moves and weekly datafor local moves) or between arithmetic mean rangecentres (for migrations), range deviation is themean straight-line distance between arithmetic meancentres of successively used ranges of the sametype, nuclei dispersion is the mean squared distancebetween nuclei chosen from utilization distributionsfor individual deer ranges, frequency of nucleichanges is the mean daily frequency of movingbetween nuclei, and distance of nuclei changes isserial distance moved to change nuclei 46Table 4.1 Total sample sizes for deer and habitat samplesrecorded at 4 study areas located on VancouverIsland, British Columbia, 1982-1991. 79Table 4.2 Multivariate analysis of variance results for testsof study area, deer behaviour (migration tactics andrange types for selection of home ranges andmigration tactics for selection of seasonalhabitats), and the interaction of those maineffects, on habitata preference (a1)b estimates forradio-collared, black-tailed deer at 4 study areason Vancouver Island, British Columbia, 1982-1991 91Table 4.3 Habitat preference estimates for 28 migratory and 44resident, radio-collared, black-tailed deer at 4study areas on Vancouver Island, British Columbia,1982-1991 93Table 4.4 Habitat preference estimates for 26 migratory and 35resident, radio-collared, black-tailed deer at 3viiistudy areas on Vancouver Island, British Columbia,1982-1991 96Table 4.5 Habitat preference estimates for 26 migratory and 35resident, radio-collared, black-tailed deer at 3study areas on Vancouver Island, British Columbia,1982-1991 98Table 4.6 Habitat preference estimates for 28 migratory and 44resident, radio-collared, black-tailed deer at 4study areas on Vancouver Island, British Columbia,1982-1991 101Table 4.7 Habitat preference estimates for 26 migratory and 44resident, radio-collared, black-tailed deer at 3study areas on Vancouver Island, British Columbia,1982- 1991 104Table 4.8 Habitat preference estimates for 26 migratory and 44resident, radio-collared, black-tailed deer at 3study areas on Vancouver Island, British Columbia,1982- 1991 107Table 4.9 Maximum likelihood analysis of variance tables forfrequency of habitat types forming the primarycomponent of activity nuclei established byindividual deer. Nuclei were determined from deerlocations estimated weekly at 4 study areas onVancouver Island, British Columbia, 1982-1991, andwere pooled into 2 groups based on deer behaviour(migratory or resident) 111Table 5.1 Effect of forest logging on range sizes, use of oldforests, and range fidelity of radio-collared,black-tailed deer where logging occurred inside(treatment) or outside (control) their predisturbance home ranges. Deer were from 2 studyareas on Vancouver Island, British Columbia, 1988-1991 135Table 6.1 Monthly cause-specific mortality (%) for radio-collared, adult female, black-tailed deer at 4 studyareas on Vancouver Island, British Columbia, 1982-1991 150Table 6.2 Mortality (%) by month of year for 3 leading causesof mortality on radio-collared, adult female, blacktailed deer on Vancouver Island, British Columbia,1982-1991 152Table 6.3 Akaike’s information criterion (AIC) and likelihoodratio tests (LR x ) between competing models ofmonthly fate (cause-specific mortality or survival)of radio-collared, adult female, black-tailed deerixon Vancouver Island, British Columbia, 1982-1991.Hierarchical models of 4 categorical variables (A -study area, B- seasonal movement of deer, E -monthly mode of elevations used by deer, and M -month of the year) are listed from the most general(all factors- A’B.E’M) to the most reduced model(no factors - N) 153Table 6.4 Monthly survival (%) for 2 known, and 1 unknown,seasonal movement types of radio-collared, adultfemale, black-tailed deer at 2 broad elevations onVancouver Island, British Columbia, 1982-1991.Bracketed values are SE estimates 155Table 7.1 The sex, age classes, and number of relocations madefor a sample of radio-collared, black-tailed deer at4 study sites on Vancouver Island, British Columbia,1982-1991. Superscripted values are the number ofdeer that lived long enough (>10 mo) to classifyinto seasonal movement types 171xLIST OF FIGURESFigure 2.1 A chronological plot of Schoener’s (1981) Ratiocalculations (top) and distance between consecutivelocations (bottom) from location observationscollected weekly between 1982 and 1991 on a radio-collared, black-tailed deer (#NRC13402) at NanaimoRiver on Vancouver Island, British Columbia 16Figure 2.2 A chronological plot of Schoener’s (1981) Ratiocalculations (top) and distance between consecutivelocations (bottom) from location observationscollected weekly between 1982 and 1991 on a radio-collared, black-tailed deer (#N1M12901) at NimpkishRiver on Vancouver Island, British Columbia 17Figure 2.3 Scatter plot of distance (m) and time (days) betweenconsecutive observations of individual, radio-collared, black-tailed deer location estimatesobserved from 1982 to 1991 on Vancouver Island,British Columbia. Inset shows detail of therelationship at lower axis positions. Locationestimates were derived by maximum likelihoodestimation and distance was the straight-linedistance 19Figure 3.1 Cumulative frequency distribution (%) for thedistance between successively sampled locations (n =8,000) of radio-collared, black-tailed deer onVancouver Island, British Columbia, 1982-1991 40Figure 3.2 Number and timing of outlier moves (moves >1.2 kmthat were not dispersals or migrations) made byradio-collared black-tailed deer on VancouverIsland, British Columbia, 1982-1991 41Figure 3.3 The average time-per-trip spent away from the natalrange for migratory, radio-collared, black-taileddeer on Vancouver Island, British Columbia, 1982-1991. Individuals are ranked in their order ofduration 45Figure 3.4 Number of migratory moves made during each month ofthe year by 2 behaviour groups of radio-collared,black-tailed deer departing natal ranges (top) andreturning to natal ranges (bottom) on VancouverIsland, British Columbia, 1982-1991 48Figure 3.5 Directions of movement from natal ranges toalternate ranges for obligate migratory (solidarrows), and facultative migratory (dashed arrows),radio-collared, black-tailed deer at 4 study siteson Vancouver Island, British Columbia, 1982-1991 50xiFigure 3.6 A representation of activity nuclei as determinedfrom the harmonic mean utilization distribution for1 radio-collared black-tailed deer (#15001) on itsalternate range, Nanaimo River, British Columbia,November 1983 - March 1984 52Figure 3.7 Frequency of radio-collared, black-tailed deer,seasonal ranges having 1 or more activity nuclei 53Figure 3.8 Mean distance between successive, 2-hour,relocations of radio-collared black-tailed deer bythe hour of day as deer were departing nuclei (top),or within nuclei (bottom), at 4 study sites onVancouver Island, British Columbia, 1982-1991 55Figure 4.1 Forest seral age class abundance (total within studyarea), availability (total within home ranges), anduse (locations from radio-collared black-taileddeer) at 4 study areas on Vancouver Island, BritishColumbia, 1982-1991 87Figure 4.2 Elevation class abundance (total within study area),availability (total within home ranges), and use(locations from radio-collared black-tailed deer) at3 study areas on Vancouver Island, British Columbia,1982-1991 88Figure 4.3 Aspect class abundance (total within study area),availability (total within home ranges), and use(locations from radio-collared black-tailed deer) at3 study areas on Vancouver Island, British Columbia,1982-1991 90Figure 4.4 Observed frequencies and predicted proportionalfrequencies (maximum likelihood estimates) of foresthabitats (clear, young, or old) forming the primarycomponent of activity nuclei for individualmigratory or resident black-tailed deer at 4 studyareas on Vancouver Island, British Columbia, 1982-1991 112Figure 5.1 Habitat abundance (% of total) at 2 study areas onVancouver Island, British Columbia, 1988-1991 . . . . 133Figure 6.1 Monthly mortality by cause recorded for radio-collared, adult, female black-tailed deer onVancouver Island, British Columbia, 1982-1991.Values above histograms indicate the total samplesize in number of deer-months 149Figure 7.1 Percent of total locations for radio-collared,black-tailed deer of 3 movement types (obligatemigratory, facultative migratory, or resident) foundat different elevation bands (m asi) on VancouverxiiIsland, British Columbia, 1982-1991 . 174Figure 7.2 Habitats chosen by radio-collared, black-tailed deerat 4 study sites on Vancouver Island, BritishColumbia, 1982-1991. Habitats are: open, 0- to 5-yr-old forests; non-merchantable (NMF) rock, water,subalpine, and alpine; young, 6- to 45-yr-oldforests; or old, >250-yr-old forests 177Figure 7.3 Habitats used by radio-collared, black-tailed deerof 3 movement types (obligate or facultativemigratory, or resident) at natal ranges (top) and atalternate, winter ranges (bottom) on VancouverIsland, British Columbia, 1982-1991. Habitats werenon-merchantable (NMF) rock, water, sub- alpine, oralpine or forests of ages: open, 0- to 5-yr-old;young, 6- to 45-yr-old; or old, >250-yr-old 178Figure 7.4 The response of radio-collared black-tailed deer toremoval of their old forest, winter habitat at 2study areas on Vancouver Island, British Columbia,1988-1991 180xiiiACKNOWLEDGMENTSMy involvement with the Vancouver Island Deer Project began in 1981while I was a student at UBC. This association evolved into full-timeemployment with the BC Forest Service and, specifically, to this work whicheventually took me on a return path to UBC. This last role as student atUBC and CSU and as project leader for the Forest Service was most rewarding.I owe many people thanks for helping me to obtain this role and then forfilling it with a unique and rich learning experience. Fred Bunnell, RickEllis, and Brian Nyberg (my supervisors) demonstrated unprecedented patienceas I reached for opportunities they helped put before me. Contrasting hisalways vociferous approach to office life, Fred had a quiet andunpretentious way of guiding my direction with this study. I think hiscomplementary tacts were particularly well-chosen for my needs as a student.Also, Fred’s help in structuring and reviewing individual chapters isunmatched; my thanks for that assistance. Although my communication withLee Gass occurred only infrequently, he always left me with much to thinkabout. Lee’s questions helped me to sharpen my focus on many issues whichultimately clarified much of my writing. A. F. Nemec and 2 anonymousreviewers made helpful suggestions for improving chapter 2. R. M. Bartmann,K. R. Brunt, D. W. Janz, K. Mitchell, A. F. Nemec, J. B. Nyberg, R. E. Page,and 1 anonymous reviewer made helpful suggestions for improving chapter 6.Financial and logistical support came from the Integrated Wildlife -Intensive Forestry Research program (1981-1991), a cooperative endeavour ofBC Environment, BC Forest Service and the University of BC. Support inlatter years also came from the South Moresby Forest Replacement Account(1989-1993), from the BC Forest Service, and from 3 companies of the BCForest Industry: Canadian Forest Products, Englewood Logging Division(thanks to Al McLeod); Fletcher Challenge Canada, now TimberWest (thanks toBob Willington); and MacMillian Bloedel Ltd. (thanks to Ron MacLaughlin).Over the years there have been many people who, through insightfuldiscussions, have contributed to my thesis. Leading the way were mycolleagues Don Doyle, Tom Hobbs, Doug Janz, Jeff Morgan, and Rick Page. Twobooks had a notable impact on my thinking; one I stole from Knut Atkinson(The Ecology of Animal Movements) and another that formed the basis for adiscussion group headed by John Wiens at CSU (Perspectives in EcologicalTheory). Thanks to Knut and John for making me aware of these publications.Also, the work reported here has been part of a larger project involvingmany employees. As equal members of a research team, we collectivelygathered data so, in that respect, this product represents their efforts aswell as mine. This is especially so in the case of the quorum who, with me,formed the deer crew in latter years: Myke Chutter, Don Doyle, Jeff Morgan,and Joan Voller. The eleventh hour saw the typical yet notoriouslyunpredictable spot fires and I owe thanks to Marvin Eng, Line Giguère, LesPeterson, and Joan Voller for helping me attend to those problems.While I am genuinely grateful to all those who assisted me, I wantmost to thank my wife and friend Line Giguère, her brother Yves, and hersister Nicole. Whenever times were tough and I seemingly had no where toturn, the Giguères were always there; not just waiting to help, but eager aswell. Enthusiastic at any hour of the day or night and in any weather,their determination and pride of workmanship has left me with a deep respectfor their talent in working with wildlife. Their efforts are entangled inevery line herein and have contributed greatly to my completion of thisproduct.xivFOREWORDI chose to write each chapter of this thesis for publication inscientific journals and the available citations are listed below. As leadauthor, I led publication development from the inception of objectives andhypotheses through construction of analytical techniques, and interpretationof results, to writing of the final report. Co-authors, in their roles,contributed assistance in developing ideas, performing instructed analyses,and editing of early drafts of manuscripts. This description of effort isnot intended to belittle the significant contributions made by co-authorsbut only to clarify our respective roles for the purpose of evaluating myefforts.Chapter 2 - McNay, R. S., J. A. Morgan, and F. L. Bunnell. 1994.Characterizing independence of observations in movements of Columbianblack-tailed deer. J. Wildi. Manage. 58:422-429.Chapter 3 - McNay, R. S., and F. L. Bunnell. 1995. Spatial and temporalscales of resource use: Evidence of a functional hierarchy inColumbian black-tailed deer movements. Ecology 76: (in preparation).Chapter 4 - McNay, R. S., and F. L. Bunnell. 1995. Spatial and temporalscales of resource use: Indication of hierarchical effects on habitatuse by Columbian black-tailed deer. Ecology 76: (in preparation).Chapter 5 - McNay, R. S. 1995. Columbian black-tailed deer response tologging of their winter habitat: fidelity, range sizes, and habitatuse. J. Wildi. Manage. 59: (Submitted).Chapter 6 - McNay, R. S., and J. M. Voller. 1994. Mortality causes andsurvival estimates for adult female, Columbian black-tailed deer. J.Wildi. Manage. 59:138-146.Chapter 7 - McNay, R. S., and F. L. Bunnell. 1995. Behaviourial limits tomovement: the effect on habitat choices for Columbian black-taileddeer. Trans. Congr. Tnt. Union Game Biol. 21(2):295-303.1CHAPTER 1 - GENERAL OVERVIEWBACKGROUNDThis thesis is an extension of a major research initiative undertakenby the government of British Columbia (BC) to resolve what was once regardedas the wildlife issue of highest provincial priority. Black-tailed deer(Odocoileus hemionus columbianus) is a highly sought-after big-game speciesbut some deer populations in coastal BC have been reported to declinerecently from a combination of habitat loss (Harestad et al. 1982) andpredation (Jones and Mason 1983). Consequently, large areas of old forestswere delineated as winter range for deer and deferred from further logging.Similar events took place in southeast Alaska (Schoen et al. 1981).However, the deferrals in BC met with strong opposition from the forestindustry (Bunnell 1985). Implications of the deferrals were, among others,constraints on wood supply and increased road inventories and costs (Addison1978). Government needed to resolve this apparent conflict over use of oldforests and research was initiated to determine: if old forests areessential for deer, when young forests begin to be suitable habitat, and howdeer integrate habitat dispersion (Addison 1978).Attention toward old forests as winter habitat for black-tailed deerresulted from a series of mechanistic studies on various habitat componentssuch as: the amounts (Harestad 1979, Vales 1986) and quality (Rochelle1980) of rooted forage, the abundance of arboreal lichens (Stevenson 1978),the ability of forest canopies to intercept snow (Kirchhoff and Schoen 1987,McNay et al. 1988), the rates of shrub burial by snow (Jones 1975, Harestad1979), the depths to which deer sink in snow, and the energetic costs ofmoving in snow (Bunnell et al. 1990a,b). The knowledge resulting from thesestudies led to predictions about habitat values which I accepted in the2early stages of my research for the BC government; old forests were best fordeer and use of young forests would be for only brief periods and limited toareas of low snowfall (Bunnell 1985, Hanley et al. 1989).Specifically, I used these habitat values in association withoptimization theory (Schoener 1987) to study habitat use by radio-collared,black-tailed deer. Optimization theory leads to the prediction, forexample, that individual radio-collared deer should choose habitats thatmaximize their chance for survival and/or reproduction. Furthermore,Fretwell (1972) extended optimization theory to propose that animals wouldfill habitats in direct relation to their value provided they had perfectknowledge about habitats and were free to choose the best. This theoreticalconstruct is referred to as the “ideal-free” distribution of animals(Fretwell and Lucas 1970). On the basis of the ideal-free distribution, Iexpected to observe most deer in the best habitats and to observeindividual, radio-collared deer using habitats in direct association withtheir apparent value. After an initial period of research, however, Iconsidered some deer regularly made choices considered suboptimal underideal-free distribution theory. Suboptimal choices would have been expectedif the area was heavily populated by deer and optimal habitats were filled(Pulliam 1989) but, as stated above, deer populations were considered to bedeclining. I then proposed black-tailed deer may not be distributed in anideal-free manner and this became the major hypothesis or thesis of study.One reason I considered the ideal-free hypothesis not to hold wasbecause, in the event of rapid environmental change (e.g., disturbances fromwildfire or logging), individual deer are unlikely to be aware of all postdisturbance habitat choices (i.e., learning lags behind the pace of theseenvironmental changes). Choice of habitat, in such cases, may be3constrained by factors other than those typically related to survival and/orreproduction (Schoener 1987). Whether or not deer use habitats in theideal -free manner is not trivial. For example, Fagen (1988) assumed theideal-free distribution in demonstrating the effect of logging on deerpopulations in Alaska. Hobbs and Hanley (1990) provided strong rationale inopposition to Fagen’s assumption and, consequently, in his conclusions abouthabitat values based on habitat use/availability statistics. If the ideal-free distribution is constrained, animals may choose habitats for reasonsother than their value in which case use/availability observations may serveonly to undermine knowledge gained from more mechanistic studies. Empiricalobservations and tests concerning the ideal-free distribution of largeungulates are lacking and I considered it necessary to resolve this before Icould make sound recommendations for the future management of habitat forblack-tailed deer.I used estimated locations and movements to indicate specific habitatchoices by individual deer. I considered these choices in context of ahierarchical resource acquisition process (Johnson 1980, Senft et al. 1987,Levin 1992) which I proposed would expose a major constraint on the ideal-free distribution of black-tailed deer. I evaluated potential effects ofthis hierarchical process on habitat choices by investigating the nature ofdecision making by individual deer and how this decision making interactedwith specific physiographic characteristics (elevation, aspects, andclimate) of 4 study areas on Vancouver Island. I also designed andevaluated the effects of 2 large logging manipulations to test mypropositions about the effects of hierarchical decision making on the idealfree distribution (i.e., that habitat choices can be constrained by factorsother than habitat value). Finally, I used estimates of survival rates and4mortality causes to make inferences regarding the implications of suchconstraints on deer populations.THESIS STRUCTUREThis thesis consists of a series of papers, 6 after this generalintroduction. In the first paper (chapter 2), I evaluate the statisticalproperty of independence in the basic data which were deer locations inspatial and temporal dimensions. Recently, Swihart and Slade (1985) warnedof potential bias and improper power-of-test calculations due to thecollection of redundant or dependent animal location data. A contrastingview, however, is that few observations may reveal decidedly lessinformation than many observations (Reynolds and Laundre 1990). Myevaluation of independence was critical because deer locations werefundamental in testing all hypotheses about movement decisions, habitatchoices, and response to habitat changes.The second paper (chapter 3) is a report of my investigation todetermine the hierarchy of movements made by deer without regard to thehabitats being used or the environmental parameters that may have caused themoves. I considered that simple patterns in movements (distance, frequency,timing, and direction of travel) would reveal a hierarchy of decisions(i.e., subclasses of decisions nested within superciasses) which, in itself,implies that superclass decisions may constrain subclass decisions (Dawkins1976, Senft et al. 1987, Levin 1992). The issue of constraint is broughtinto deeper ecological significance in the third paper (chapter 4) where Iuse the hierarchy established in chapter 3 as a framework for organizing theinvestigation of habitat preferences.In the fourth paper (chapter 5), I assess the response of individual5deer to logging of their winter habitat. This investigation specificallyaddresses the assumption of the ideal-free distribution by imposing aprominent change in habitat on individual deer. I considered that aconsistency in habitat use (before and after disturbance) would support theideal-free hypothesis. Alternatively, fidelity to winter range sites, inthe event of this extreme disturbance, would support the notion thathierarchical decision making constrains the ideal-free distribution of deer.In the fifth paper (chapter 6), I investigate survival rates andmortality causes especially as they relate to conclusions about movement andhabitat use decisions considered in previous chapters and the sixth paper(chapter 7) contains a general discussion of results and a summary ofconclusions specifically reported in the context of applied management.Chapter 8 is a more general summary of conclusions.LITERATURE CITEDAddison, R. B. 1978. Research needs for integrated management of timber,deer, and elk in the Vancouver Forest district -- Report on Phase 1.Special Studies Div., British Columbia Minist. of For., Victoria.25pp.Bunnell, F. L. 1985. Forestry and black-tailed deer: conflicts, crises,or cooperation. For. Chron. 61:180-184._____F. W. Hovey, R. S. McNay, and K. L. Parker. 1990a. Forest cover,snow conditions, and black-tailed deer sinking depths. Can. J. Zool.68: 2403-2408.K. L. Parker, R. S. McNay, and F. W. 1-lovey. 1990b. Sinking depthsof black-tailed deer in snow, and their indices. Can. J. Zoo].68:917-922.Dawkins, R. 1976. Hierarchical organisation: a candidate principle forethology. Pages 7-53 in P. P. G. Bateson and R. A. Hinde, eds.Growing points in ethology. Cambridge Univ. Press, Cambridge,England.Fagen, R. 1988. Population effects of habitat change: a quantitativeassessment. J. Wildi. Manage. 52:41-46.6Fretwell, S. D. 1972. Populations in a seasonal environment. PrincetonUniv. Press, Princeton, N.J. 217pp._____and H. L. Lucas, Jr. 1970. On territorial behaviour and otherfactors influencing habitat distribution in birds. I. Theoreticaldevelopment. Acta Biotheoretica 19:16-36.Hanley, T. A., C. T. Robbins, and D. E. Spalinger. 1989. Forest habitatsand the nutritional ecology of Sitka black-tailed deer: a researchsynthesis with implications for forest management. USDA For. Serv.Gen. Tech. Rep. PNW-230. Portland, Oreg. 52pp.Harestad, A. S. 1979. Seasonal movements of black-tailed deer on northernVancouver Island. Ph.D. Thesis, Univ. of British Columbia, Vancouver.98pp.J. A. Rochelle, and F. L. Bunnell. 1982. Old-growth forests andblack-tailed deer on Vancouver Island. Trans. North Am. Wildl. Nat.Resour. Conf. 47:343-352.Hobbs, N. T., and T. A. Hanley. 1990. Habitat evaluation: do use/availability data reflect carrying capacity? J. Wildi. Manage.54:515-522.Johnson, D. H. 1980. The comparison of usage and availability measurementsfor evaluating resource preference. Ecology 61:65-71.Jones, G. W. 1975. Aspects of the winter ecology of black-tailed deer(Odocoileus hemionus columbianus Richardson) on northern VancouverIsland. M.S. Thesis, Univ. of British Columbia, Vancouver. 78pp.and B. Mason. 1983. Relationships among wolves, hunting, andpopulation trends of black-tailed deer in the Nimpkish alley onVancouver Island. British Columbia Minist. of Environ. Wildl. Rep. R7. Victoria. 26pp.Kirchhoff, M. D., and J. W. Schoen. 1987. Forest cover and snow:implications for deer habitat in southeast Alaska. J. Wildl. Manage.51:28-33.Levin, R. V. 1992. The problem of pattern and scale in ecology. Ecology73: 1943-1967.McNay, R. S., L. D. Peterson, and J. B. Nyberg. 1988. The influence offorest stand characteristics on snow interception in the coastalforests of British Columbia. Can. J. For. Res. 18:566-573.Pulliam, H. R. 1989. Individual behaviour and the procurement ofessential resources. Pages 25-38 in J. Roughgarden, R. May, and S.Levin, eds. Perspectives in ecological theory. Princeton Univ.Press, Princeton, N.J.Reynolds, T. D., and J. W. Laundre. 1990. Time intervals for estimatingpronghorn and coyote home ranges and daily movements. J. Wildi.7Manage. 54:316-322.Rochelle, J. A. 1980. The role of mature conifer forests in the winternutrition of black-tailed deer. Ph.D. Thesis, Univ. of BritishColumbia, Vancouver. 295pp.Schoen, J. W., 0. C. Walimo, and M. D. Kirchhoff. 1981. Wildlife forestrelationships: is a reevaluation of old growth necessary? Trans.North Am. Wildl. Nat. Resourc. Conf. 46:531-545.Schoener, T. W. 1987. A brief history of optimal foraging ecology. Pages5-68 in A. C. Kainil, J. R. Krebs, and H. R. Pulliam, eds. Foragingbehaviour. Plenum Press, N.Y.Senft, R. L., M. B. Coughenour, D. W. Bailey, L. R. Rittenhouse, 0. E. Sala,and D. M. Swift. 1987. Large herbivore foraging and ecologicalhierarchies. Bioscience 37:789-799.Stevenson, S. K. 1978. Distribution and abundance of arboreal lichens andtheir use as forage by black-tailed deer. M.S. Thesis, Univ. ofBritish Columbia, Vancouver. 148pp.Swihart, R. K., and N. A. Slade. 1985. Testing for independence ofobservations in animal movements. Ecology 66:1176-1184.Vales, D. J. 1986. Functional relationships between salal understory andforest overstory. M.Sc. Thesis, Univ. of British Columbia, Vancouver.164pp.8CHAPTER 2 - INDEPENDENCE OF OBSERVATIONS IN MOVEMENTS’The problem of temporal dependence of animal locations has beentreated by Slade and Swihart (1983), Swihart and Slade (1985a,b; 1986), andSwihart et al. 1988. Temporal independence of observations is important tohome-range size estimations because most parametric estimators requireanimal locations to be independent random samples (Ackerman et al. 1990).Home ranges will be consistently underestimated (biased) if based ondependent location observations (Dunn and Gipson 1977, Schoener 1981, Sladeand Swihart 1983). Swihart and Slade (1985b) documented a strong inverserelationship between estimates of home-range size and the degree ofdependence between location observations. Further, because dependent datacontain redundant information, less information is available in dependentdatasets compared with independent datasets of an equal size (Swihart andSlade 1985b). Consequently, dependent data are likely to produce biasedestimates even for nonstatistical measures.The central issue, however, is independence of observations ininferential statistics. Data are independent when the current observation(e.g., position at the current point in time t) is not a function of thelast observation (e.g., position at some time interval k previous to thecurrent time t). Alternatively, the variance between consecutiveobservations is proportional to the overall variance (von Neumann 1941).Consequently, if observations are independent, each observation contributessimilarly to the overall estimate of population parameters.While tests for independence of observations are known for data with 1dimension (see Box and Jenkins 1976), they are relatively unknown for data1 Published as: McNay, R. S., J. A. Morgan, and F. L. Bunnell. 1994.Characterizing independence of observations in movements of Columbian blacktailed deer. J. Wildi. Manage. 58:422-429.9with more than 1 dimension such as animal location data (Schoener 1981) thatusually are expressed as x and y spatial coordinates. A second importantdistinction of location data is that ordering of the sampled dataset isthrough a third dimension, time. Location data can, but do not necessarilyhave to, represent a rate of travel. Additionally, location data can bepresented in 1 dimension, an example being the distance between consecutivelocations (Fitch 1958). Even though distance between consecutive locationsrepresents only 1 dimension it still is intimately connected to time and isa rate variable. Reynolds and Laundre (1990) found, however, that increasesin the time interval between observations leads to poorer information aboutthe true distance travelled during the interval.Swihart and Slade (1985a) examined Schoener’s Ratio statistic (1981)as a potentially useful measurement of independence when observationsinvolve 2 spatial dimensions. Also, Schoener (1981) suggested the ratio mayhelp determine the number of samples necessary for parametric estimation ofhome-range size. Subsequent to testing Schoener’s Ratio, Swihart and Slade(1985a) suggested further uses of the statistic to (1) determine the timeinterval necessary to obtain independent sample observations, (2) identifyshifts or patterns in the use of space, and (3) make comparisons of the rateat which different animals use space.Since 1985, however, there has been little use of Schoener’s Ratio inthe manner intended by Swihart and Slade. Hoizenbein and Marchinton (1992)used Schoener’s Ratio to assess independence of observations of white-taileddeer (0. virginianus) locations but presented no documentation of results.They assumed 4 hr, or greater, between observations to be sufficient for adeer to move to any point in its home range. Other researchers demonstratedloss of biological information when using only those animal locations that10were judged to be independent by Schoener’s Ratio (Reynolds and Laundre1990). Finally, Kremsater and Bunnell (1992), while recognizing theimportance of independence of observations, developed alternative techniques(to Schoener’s Ratio) to address specific questions about deer use oflandscape mosaics. Kremsater and Bunnell (1992) also argued against testingfor independence when location data are to be used for assessing conditionalprobabilities of decision making.The lack of use of Schoener’s Ratio despite compelling arguments ofSwihart and Slade (1985a) prompted us to evaluate independence in ourobservations of black-tailed deer locations. Our primary interest was theapplication of Schoener’s Ratio when observations included movements ofmigratory deer. We also wanted to compare assessments of independencebetween the 2 related measures of animal movements: (1) animal locations inspace and (2) distance moved between consecutive locations. Our specificobjectives were to: (1) evaluate independence in observations of black-tailed deer locations using Schoener’s Ratio; (2) assess the influence ofmigrations on Schoener’s Ratio; and (3) evaluate independence inobservations of distance between consecutive locations.METHODSDeer Location SamplesWe obtained location estimates for a sample of radio-collared deermonitored for another study (McNay and Doyle 1990) at 4 sites on VancouverIsland, British Columbia. We used triangulation (White and Garrott 1990) tolocate deer with no less than 3 bearings recorded at separate and permanentstations marked at 100-rn intervals along roads. Bearings for an individualdeer location were usually collected in <10 mm at sites that were line-of-11sight with, and close to (<400 m), the transmitter being monitored.We located deer from January 1982 to June 1984 on an ad hoc schedulethat generally resulted in each deer being located once per week. AfterJune 1984 until project completion at June 1991, sampling was standardizedso that, during a calendar month, each deer was located at least once perweek and once within each quarter of a calendar day. At specific times(usually once per month) we established sessions of comparatively moreintensive monitoring; sampling was increased to once every 2 hr forpredetermined periods (usually from 3 to 5 days).We estimated final deer locations by 2 different techniques. Duringinitial years of study, we plotted triangulation data and determined thelocation as the centroid of the polygon that resulted from overlappingbearings (Hupp and Ratti 1983). In 1984 and subsequent years, bearinginformation was retained and analyzed with the maximum likelihood estimatorpresented by Lenth (1981). We modified a SAS program (SAS Inst. Inc. 1985)by White and Garrott (1990:64) to accept bearings from any 3- to 5-samplingstations of known Universal Transverse Mercator grid coordinates. Using theresultant x2’ goodness-of-fit test for all bearings contributing to eachlocation and the location’s 95% error ellipse size, we made a finaljudgement on the quality of individual locations a posteriori. If theprobability of observing poorer goodness-of-fit than that calculated was<0.10, the location estimate was considered poor. We made exceptions whenthe bearing set was collected for a location close to the observer’slocation (antenna-to-animal distance <100 m). In cases of poor bearing fit,proximity was identified by error ellipses <1 ha. We plotted all locations(Borland Tnt. Inc. 1992) as a check against field data to identify anyrecording or coding errors.12Analytical TechniquesWe iteratively sampled data to construct 5 individual datasets(Swihart and Slade 1985a). The first 2 datasets were constructed using datacollected during intensive monitoring sessions while the last 3 datasetsincluded data collected on a weekly basis. First, we used all datacollected during intensive monitoring sessions. Second, we omitted anyintensive monitoring data for which the time interval between samples was <4hr. Third, we omitted data if the time interval between samples was <1 day.Fourth, we disallowed time intervals <17 days and, fifth time intervals <38days. In the latter 3 datasets we wanted to obtain average samplingintervals of approximately 1, 3, and 6 wk respectively. Henceforth, we willrefer to these datasets as 2-hour, 4-hour, weekly, 3-week, and 6-week.We calculated Schoener’s Ratio (Schoener 1981) for each deer-session(2- and 4-hour datasets) and for each deer (weekly, 3-, and 6-weekdatasets). We calculated the critical value of Schoener’s Ratio usingmethods of Swihart and Slade (1985a,b) to test the null hypothesis that deerlocations were independent. We chose a = 0.25 (Swihart and Slade 1986)unless stated otherwise.We used 2 methods to examine the influence of migrations using theweekly, 3-, and 6-week datasets. First, to identify the effect that eachlocation had on the overall statistic, we recalculated Schoener’s Ratio eachtime a new location was added to the dataset. Second, we omitted migratorymovements and calculated Schoener’s Ratio for each spatially exclusive,seasonal home range. We determined migratory movements by visual inspectionof chronological location plots for each deer. Inspection of each movementallowed us to identify those composing regular trips with predictable returnmoves (Sinclair 1984).13To examine independence of distances between consecutive locations, wemeasured the straight-line distance from the last location to the currentlocation. We first looked for indications that data reflected a rate oftravel by plotting distance between consecutive locations against timebetween consecutive locations and by testing for linear trends usingcorrelation analysis (SAS Inst. Inc. 1985). Secondly, we used the meansquare successive difference test, alternatively known as the V statistic(von Neumann et al. 1941), to evaluate independence in observations.RESULTSWe monitored 44 resident and 28 migratory deer for 253 deer-years and12,572 locations. We sampled 42 of those deer during 24 intensivemonitoring sessions for a total of 133 deer-sessions. Not all deer weresampled during each intensive monitoring session. Intensive monitoringaccounted for 4,039 of the locations. Complete bearing information wasavailable for 9,234 of the locations and with those data we calculatedgoodness-of-fit and error ellipse sizes for each location estimate. Twopercent of the locations were generated from bearings with poor goodness-of-fit (x2 P 0.10) and large error ellipses (>1.0 ha). Generally, locationshad a 95% error ellipse of <1 ha ( = 0.98 ha, SD 6.5; n = 12,103).The average time interval between samples was 2.0 hr (SD = 1.6; n =3,905) in the 2-hour dataset, and 5.4 hr (SD = 2.2; n = 1,613) in the 4-hourdataset. Time intervals between less intensive samples were 8.1 days (SD =10.3; n = 8,464) in the weekly dataset, 24.6 days (SD = 14.2; n = 2,865) inthe 3-week dataset, and 46.4 days (SD = 16.9; n = 1,506) in the 6-weekdataset.14Independence of Location ObservationsThe hypothesis of independence was rejected (P < 0.25) for most deerlocation datasets regardless of the time interval between locations (Table2.1), especially for deer that migrated. With a 1-week time intervalbetween locations, none of the datasets for migratory deer were independent,but 10 and 18% of the datasets were independent at the 4-hour and 6-weekintervals, respectively. The highest percentage of independent datasets(41% or 18 of 44 deer) came from resident deer with 6 wk between locationsamples.Effect of Migrations on Independence of LocationsMigratory movements affected Schoener’s Ratio. The first occurrenceof a migration in each dataset caused Schoener’s Ratio to drop indicatinglack of independence. Subsequent migrations, however, were largelyundetectable (Fig. 2.1). Migrations were not the only movements that led torejection of independence although they were the most conspicuous. Datasetsfor resident deer (or for migratory deer within a seasonal range) alsobecame dependent if the deer moved to a unique place at the periphery of itsrange. Alternatively, in cases where no migrations or outlier locationswere recorded, results of Schoener’s Ratio tests often oscillated betweenindependence and dependence in an unpredictable pattern (Fig. 2.2). Only 5of 345 tests revealed datasets that were judged independent throughout thedata collection period.When we removed migratory movements and recalculated Schoener’s Ratiofrom spatially distinct seasonal ranges, the independence statistic improved(Table 2.1). In more than half of the seasonal range datasets (26 seasonalranges of 48) we did not reject (P > 0.25) independence of locationTable2.1.Percentageoftotal,orseasonal,black-taileddeerrangeswherethehypothesisofindependenceaoflocationobservationswasnotrejected.Valuesinparenthesesarethenumberoftotal,orseasonal,homerangestestedfromdatacollectedduring1982to1991onVancouverIsland,BritishColumbia.RangeDeer%independentrangesintimeintervaltypebehaviour2-hour4-hourWeekly3-week6-weekSchoener’sRatioTestTotalResident7(71)4(70)9(44)27(44)41(44)TotalMigratory6(62)10(61)0(28)7(28)18(28)SeasonalMigratory21(56)54(48)49(41)vonNeumann’sVTestTotalResident45(71)61(70)7(44)43(44)54(44)TotalMigratory42(62)82(61)25(28)57(28)68(28)SeasonalMigratory34(56)50(48)56(41)aTwotestswereused:(1)Schoener’sRatio(Schoener1981)testsindependenceofobservationsin2-dimensionalspace,and(2)vonNeumann’sV(vonNeumannetal.1941)testsindependenceofobservationsofdistancesbetweenconsecutivelocations.160CuU)Ia)0.0(1)Calculated valueCritical value3.532.521.510.5A)I I I I I I I4,00012,000.210,0008,000g 6,000• 4,000C)2,00:- denotes migration,‘1 ,II,e B)II I II1)IIII’II0 01 II:1 00 II III I III I ,I‘I ,I‘I II0 II I)‘I II IIII II I•III ( IIII0 III0•IIi\ • I‘,‘ II I — .L.1 h —.. — r i —MAMJJASONDJFMAMJJAMonthFigure 2.1 A chronological plot of Schoener’s (1981) Ratio calculations(top) and distance between consecutive locations (bottom) from locationobservations collected weekly between 1982 and 1991 on a radio-collared,black-tailed deer (#NRC13402) at Nanainio River on Vancouver Island, BritishColumbia.17Calculated valueCritical valueA)IIII—I2.82.62.42.221.81.61.41.21,6001,4001,2001,000800600400200I—S0Cu00Cl)—I—i.1 BI— II IIII—IIa it••I iiI •i I I— II t II1 •i SIjI•t 51 I%4 iiIllSp St•I I— III•• gI •l . I I I I4gI titi • I IIlgI lbI III • • I ii ,I gIt t— •J S ISi I,%II % • •I I I,I S i %jI... — -• S I’I I—I I I I I I I I I I IAMJJASONDJFMAMJJASMonthFigure 2.2 A chronological plot of Schoener’s (1981) Ratio calculations(top) and distance between consecutive locations (bottom) from locationobservations collected weekly between 1982 and 1991 on a radio-collared,black-tailed deer (#N1M12901) at Nimpkish River on Vancouver Island, BritishColumbia.18observations when the time interval between observations was 3 wk (Table2.1).Independence of Distance Between Consecutive LocationsDistance between consecutive locations was poorly associated with timebetween those locations (r = 0.26; n = 12,437; P = 0.0001). At most timeintervals, except the shortest (intensive monitoring), deer travelled a widerange of distances from 0 to 14 km (Fig. 2.3).Von Neumann’s V indicated more consistent independence amongobservations of distances than was achieved among spatial locations (Table2.1). We did not reject (P > 0.25) independence in 82% of the intensivemonitoring sessions recorded for migratory deer using 4 hr between locations(50 of 61 deer-sessions). The weekly dataset showed the poorest percentageof independent datasets (7 to 34%). There was no improvement in going froma 3-week time interval to a 6-week time interval nor did eliminatingmigrations improve independence of data collected on migratory deer (Table2.1).DISCUSSIONMost of our datasets on black-tailed deer were composed ofstatistically dependent observations (Table 2.1). If our objectives were tomeasure the amount of space used by deer, or the average distance travelledby deer, we would conclude that our estimates would likely be biased low(Schoener 1981, Swihart and Slade 1985b). The bias would be the result ofthe sample containing redundant observations.Migrations led to dependence in datasets for migratory black-taileddeer because they indicated relatively infrequent moves to different sites.1915,00014,000 0• 12,00013,000 10,000012,000 8,000 00000 6,000 8 8- 0008.2 4,000 0.10,0002,000 89,000__________8,000O°2 4°6-’g 7,000CI •.3,000 ::: : ..2,000 ——_.•_-‘1 000 •* • . .— . .0 !_‘II..i..I.r:.I.II I II• I I I I I i 1i i1 i Ii I0 28 56 84 112 140 168 196 224 252Time between consecutive locations (days)Figure 2.3 Scatter plot of distance (m) and time (days) between consecutiveobservations of individual, radio-collared, black-tailed deer locationestimates observed from 1982 to 1991 on Vancouver Island, British Columbia.Inset shows detail of the relationship at lower axis positions. Locationestimates were derived by maximum likelihood estimation (Lenth 1981) anddistance was the straight-line distance.20Such movements expanded the overall variance in one, or both, spatialcoordinates and hence, daily use of sites within a seasonal range becamecomparatively redundant once a migration was made. In this respectSchoener’s Ratio performed well as a measure of a significant, first-timechange in the use of space. Because the statistic is calculated fromaverage deviations, however, subsequent changes in use of space wentundetected (Fig. 2.1).Although our observations of distance between consecutive locationshad marginally less dependence, we found similar results to those in ourinvestigation of how deer use 2-dimensional space. Again, in at least halfthe cases, black-tailed deer infrequently made larger than normal moves,most of which were migrations. Those large moves effectively caused themore common moves (generally <250 m) to become comparatively redundant.The conclusion of dependence in both of the above cases, however, isbased on a little mentioned, yet important, assumption of the 2 analyticaltechniques. Both tests require normal data distributions because they arecalculated from average deviation of the samples (von Neumann 1941, Schoener1981). Application of the tests in circumstances of skewed datadistributions could lead to an apparent lack of independence even whenindependence is achieved. Swihart and Slade (1985a,b) were careful toensure their constructed datasets came from a normal distribution. Whenthey dealt with data collected from real observations, Swihart et al. (1988)dropped, from their calculations, any dataset in which the animal shiftedits activity centre. Swihart and Slade (1985b) alluded to this effect ofmigrations by indicating that temporal rhythmicity in movements may reducelikelihood of independence. We concur, noting that migrations, or othertemporal rhythmicity, would lead to non-normality and hence an apparent lack21of independence.Normality of location data is influenced by temporal use of space (ordistance travelled). If velocity was always constant, then distancemeasurements at specific time intervals would tend to be normallydistributed. Movements, however, are the essence of the behaviour of mobileanimals and animals may choose to move at a variety of rates from running tono movement at all. It is unlikely that movements could ever be expected tobe normally distributed. Lack of normality in our data was mostconspicuously caused by migrations that could occur within the time intervalbetween most location sampling (Fig. 2.3). In fact, with the exception oflocation sampling during intensive monitoring, deer had time to travelanywhere in their home ranges (i.e., our distance measurements were notlikely to be indicative of any specific rate of travel). We concluded thatbecause variation in behaviour contributed to the lack of a normaldistribution in location observations, it impaired our ability to find anappropriate sampling interval for black-tailed deer and led to an apparentdependency in the data. Had all our data been more indicative of a rate oftravel (e.g., the 2- and 4-hour datasets; Fig. 2.3, inset) the arbitraryculling of data (see Methods) should have produced better tests ofindependence and likely less dependence between samples.MANAGEMENT IMPLICATIONSWe concluded that testing our location data for independence employedtechniques that were not robust to skewed data distributions that can becaused by migrations between seasonal ranges. To avoid apparent dependenceof observations, we would have had to disregard about 90% of our data to endup with an average of 8 locations/deer-year. Doing so would have resulted22in samples sizes below that required for many analyses and would haveeliminated information about the dynamic manner in which black-tailed deeruse space.In an operational sense, the primary concern about independence shouldfocus on whether an animal has had time to move to any location within itshome range before the next observation is taken (Lair 1987). That was theinterpretation adopted by Holzenbein and Marchinton (1992) when they chose asampling interval of 4 hr for their observations of white-tailed deer.White and Qarrott (1990:148) expanded on that principle by suggesting thereal issue was to properly sample the time interval over which an estimateis to apply. A systematic sample over specific time periods eliminates theeffects of bias due to redundant data (White and Garrott 1990:148) but stillinflates n causing variance to be underestimated. Biased variance, however,would be of little concern in home range estimates because it is nevercalculated for a single home range.Furthermore, choosing appropriate time intervals to sample animallocations appears to be more a problem of study objectives than ofstatistical independence, provided samples are obtained systematically.Lair (1987) observed that minimum time intervals to statistical independencecan be long enough to preclude investigations of home-range characteristics.While samples taken close together in time may be redundant statistically,they reveal comparatively better behaviourial information on use of spacethan samples taken farther apart in time (Lair 1987, Reynolds and Laundre1990). In this study, black-tailed deer were observed to travel even thelargest distances in less time than our sampling intervals in all but theintensive monitoring (Fig. 2.3). For that reason we concluded that ourweekly observations were likely to have biological independence (Lair 1987)23even though they may be declared statistically dependent by the tests weused. We regarded the datasets to have apparent dependence rather thanactual dependence because the data violated the assumption of normalityrequired for the independence tests.We recommend that investigators strive to achieve biologicalindependence in systematic observations of animal movements rather thanpassing the criteria of statistical tests that assume normality. Datadistributions of animal movements will rarely follow a normal distributionbecause movements reflect behaviourial decisions. Any test of independenceusing the non-normal data would likely lead to conclusions of apparentredundancy forcing elimination of important biological information. Oursuggestions imply that investigators assume spatial and temporal dependence,rather than independence, in animal location data. Making that admissioncould help advance traditional data analysis (Tukey 1977) beyond exploratorytechniques to those that acknowledge and model statistical dependence(Houston et al. 1988, Rossi et al. 1992).LITERATURE CITEDAckerman, B. B., F. A. Leban, M. D. Samuel, and E. 0. Garton. 1990. User’smanual for program Home Range. For. Wildl. and Range Exp. Stn. Tech.Rep. 15. Contrib. 259. Univ. of Idaho, Moscow. 8Opp.Borland Tnt. Inc. 1992. Quattro Pro for windows: user’s guide. BorlandInternational, Inc. Scotts Valley, Calif. 448pp.Box, G. E. P., and G. M. Jenkins. 1976. Time series analysis: forecastingand control. Holden-Day, San Francisco, Calif. 575pp.Dunn, J. E., and P. S. Gipson. 1977. Analysis of radio telemetry data instudies of home range. Biometrics 33:85-101.Fitch, H. S. 1958. Home ranges, territories, and seasonal movements ofvertebrates of the Natural History Reservation. Univ. Kans. Mus. Nat.Hist. 11:63-326.Holzenbein, S., and R. L. Marchinton. 1992. Spatial integration of24maturing-male white-tailed deer into the adult population. J. Mammal.73:326-334.Houston, A. I., C. W. Clark, J. M. McNamara, and M. Mangel. 1988. Dynamicmodels in behaviourial and evolutionary ecology. Nature 332:29-34.Hupp, J. W., and J. T. Ratti. 1983. A test of radio telemetrytriangulation accuracy in heterogeneous environments. Tnt. Conf.Wildi. Biotelem. 4:31-46.Kremsater, L. L., and F. L. Bunnell. 1992. Testing responses to forestedges: the example of black-tailed deer. Can. J. Zool. 70:2426-2435.Lair, H. 1987. Estimating the location of the focal centre in red squirrelhome ranges. Ecology 68:1092-1101.Lenth, R. V. 1981. On finding the source of a signal. Technometrics23:149-154.McNay, R. S., and D. D. Doyle. 1990. The Integrated Wildlife-IntensiveForestry Research (IWIFR) program deer project. Northwest Environ. J.6:389-390.Reynolds, T. D., and J. W. Laundre. 1990. Time intervals for estimatingpronghorn and coyote home ranges and daily movements. J. Wildl.Manage. 54:316-322.Rossi, R. E., D. J. Mulla, A. G. Journel, and E. H. Franz. 1992.Geostatistical tools for modelling and interpreting ecological spatialdependence. Ecol. Monogr. 62:277-314.SAS Inst. Inc. 1985. SAS user’s guide: basics, version 5 edition. SASInstitute Inc., Cary, N.C. 584pp.Schoener, T. W. 1981. An empirically based estimate of home range. Theor.Pop. Biol. 20:281-325.Sinclair, A. R. E. 1984. The function of distance movements invertebrates. Pages 240-258 in I. R. Swingland and P. J. Greenwood,eds. The ecology of animal movement. Claredon Press, Oxford, England.Slade, N. A., and R. K. Swihart. 1983. Home range indices for the hispidcotton rat (Sigmodon hispidus) in northeastern Kansas. J. Mammal.64:580-590.Swihart, R. K., and N. A. Slade. 1985a. Testing for independence ofobservations in animal movements. Ecology 66:1176-1184._____and_____. 1985b. Influence of sampling interval on estimates ofhome-range size. J. Wildl. Manage. 49:1019-1025.and . 1986. The importance of statistical power when testingfor independence in animal movements. Ecology 67:255-258.25_____ _____and B. J. Bergstrom. 1988. Relating body size to the rateof home range use in mammals. Ecology 69:393-399.Tukey, J. 1977. Exploratory data analysis. Addison-Wesley, Reading,Mass. 688pp.von Neumann, J. 1941. Distribution of the ratio of the mean squaresuccessive difference to the variance. Ann. Math. Stat. 12:367-395.R. H. Kent, H. R. Bellinson, and B. I. Hart. 1941. The mean squaresuccessive difference. Ann. Math. Stat. 12:153-162.White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-trackingdata. Academic Press, San Diego, Calif. 383pp.26CHAPTER 3 - SPATIAL AND TEMPORAL SCALES OF RESOURCE USE: MOVEMENTSWe consider aspects of the general question: are patterns of resourceuse products of an hierarchically-structured decision process? We usedmovements as a device for questioning because no point in space provides allnecessary resources (Greenwood and Swingland 1984) and resources themselvesare dynamic, often through temporally specific events (Van Home et al.1988). Movements are not only the mechanism for resource use but reflectboth specific decisions and broad tactics.Generally, studies of optimization in ecology have sought tounderstand evolution and processes of natural selection (Maynard-Smith 1978,Mayr 1983, Stevens and Krebs 1986, Pulliam 1989). More specifically,organisms use resources, allocate energy gained to specific life-functions,and accumulate a surplus of energy to use in procreation. Organisms thataccomplish that best (optimally) are regarded the most fit by evidence oftheir survival and reproductive output (Schoener 1971). Studies on optimaluse of resources are, however, not without criticism (Gould and Lewontin1979, Zach and Smith 1981, Bunnell and Gillinghani 1985, Ollason 1987) andmany lack precise quantitative support (Pyke 1984). Also, some applicationsof the theory have been problematic (Westoby 1974, Owen-Smith and Novellie1981) especially when considering optimal choice of forage patches ratherthan specific diet choices (Schoener 1987).We concur with Schoener (1987) that extending optimal dietformulations to patch selection, or to non-laboratory conditions (Zach andSmith 1981), is not straight forward. The concept of optimality, however,remains an appealing basis for evaluating resource selection. Recently,interest in optimality theory has been restored by qualifying classicalforaging models (e.g., MacArthur and Pianka 1966, Charnov 1976) within the27framework of constraints external to diet, thereby extending theory acrossseveral scales of ecological resolution (McNamara and Houston 1980, Real andCaraco 1986, Mangel and Clark 1986, McNamara and Houston 1986, Houston eta!. 1988, Huston et a!. 1988, Rosenzweig 1991). Foraging decisions can beset within the context of different ecological goals, the relativeimportance of which may vary among different ecological scales (Dawkins1976, Pierce and Ollason 1987, Senft et a!. 1987, Orians and Wittenberger1991). Placing observations of foraging decisions within the context ofother, perhaps more general, scale-determined decisions helps refineconclusions about optimality of resource use (e.g., Orians and Wittenberger1991).Scale-determined decisions may invoke a nested, functional hierarchywith its attendant criteria of containment and constraint of scalaractivities (Dawkins 1976, Allen et al. 1984). Activities are scalar ifmeasurements of the phenomenon can be isolated into distinct classes alongsome dimension, with isolation often achieved by activity rate changes(Wiens 1976, Senft et a!. 1987, O’Neill et al. 1989, Pulliam 1989).Movement for many terrestrial vertebrates is scalar because it can beisolated into several velocity-based classes: resting (stationary),walking, or running. Furthermore, these scalar classes of movement arehierarchical if we consider the classes as immobility or mobility (withwalking and running as subclasses of mobility). Describing movements inthis way, we invoke a nested hierarchy because subclasses (walking andrunning) are contained within superciasses (mobility) (Dawkins 1976).Constraint in hierarchies is more generally referred to as informationtransfer among scales (Gass 1985, Levin 1992) and is identified differentlydepending on 2 opposing perspectives; 1 from the bottom of the hierarchy28looking up, the other from the top looking down. Upward transfer ofinformation occurs when subclass activities define initiating conditions forsuperclass activities and downward transfer of information occurs whensuperclasses constrain the range of activities within subclasses (Dawkins1976, Senft et al. 1987, O’Neill et al. 1988, Powell 1989, Levin 1992).A shift in perspective from simple rate of movement to factorsinstigating movement (e.g., resource use) can potentially provide an exampleof a functional hierarchy. Senft et a!. (1987) provided compellingarguments for considering resource use by large ungulates to havehierarchical function in time and space. Such consideration implies adifferent meaning ecologically than if resource use was hierarchial only byclassification or was simply a scalar activity with no hierarchy. In theexample of Senft et a!. (1987: Table 1), diets provide the basis upon whichfeeding areas are chosen, which in turn provide the basis for home rangeestablishment. Alternatively, home ranges constrain the availability offeeding areas, which in turn constrain diets. Implicit in this example isthat foraging (resource use) goals have varying degrees of importance amongscales and that information is transferred to help form scale-dependentdecisions about resource use. Key aspects of each goal are: for whatduration and to which specific locations does any particular decision committhe individual (Orians and Wittenberger 1991) and how quickly are animalsable to adapt their decisions to new resource conditions (Gass 1985)?Levin (1992) proposed that, to understand natural systems efficiently,the examination of scales should be focused on “how information istransferred from one [class of a scalar activity] to another”. Gass (1985)considered such communication links to be key in understanding the abilityof organisms to function efficiently. Dawkins (1976) viewed transfer of29information among classes to be the basic characteristic of hierarchicalfunction. Allen et al. (1984) proposed that “the inter[class] relations inhierarchical organization of complex systems, in part, explains why suchsystems usually total more than the sum of their parts”.We examined movements made by black-tailed deer assuming theirmovements indicated decisions about resource use. Resources sought aretypically considered to include: 1) nutritious forage (Gates 1968, Rochelle1980), cover for hiding from predators (Kufeld et al. 1988), cover fromextremes in thermal environments (Parker 1988), cover from deep snow duringwinter (Bunnell and Jones 1984, Bunnell et al. 1990a,b), and access to matesand other family members (Hirth 1977, Bunnell and Harestad 1983, Hamlin andMackie 1989). We questioned whether resource use by deer followed thehierarchical structure conceptualized by Senft et al. (1987) for largeungulates and, if so, what implications would hierarchical function have onoptimization of resource use?Early studies on black-tailed deer were by direct observation andconcerned local movement patterns in the context of foraging habits and dietchoices (e.g., Cowan 1945). Studies since the advent of radio-telemetrywere usually based on indirect observation and were primarily concerned withlong movements and relative use of habitat patches (Sanderson 1966).McCullough (1985) classified movements of large terrestrial mammals asdispersal, nomadism, migration, and local movements but discussed only thefirst 3 classes independently of each other. Our interest was in howclasses of movement were related or linked (Allen et al. 1984, Levin 1992)and the role of these linkages in the optimal use of resources.Specifically, we wanted to: 1) clarify the empirical justification forscalar classes of movements described by others, 2) assess whether or not30the classes are hierarchical, and 3) identify if, and perhaps how,information is transferred among classes thereby assessing the existence ofa functioning hierarchical decision process. Clarifying the decisionprocess for movements is relevant to understanding resource choices ingeneral (e.g., Orians and Wittenberger 1991), management of landscapes(O’Neill et al. 1988), and preservation of biodiversity (Franklin 1993).STUDY AREASWe studied movements made by black-tailed deer at 4 locations onVancouver Island, British Columbia (Table 3.1). The Chemainus, Nanaimo, andNimpkish rivers are characterized by open, relatively flat-bottomed valleys(U-shaped) while Caycuse River ranges less in elevation but has steeperslopes and least area at lower elevations (V-shaped). All study areas werelogged extensively by clearcutting resulting in habitats ranging fromrecently deforested (0- to 5-yr-old) to old (>250-yr-old) forests.Arrangement of habitats was typical of coastal logging with initial harvestscoming from the bottom and downstream end of valleys and with subsequentharvests coming from the mid-slopes, from the headwaters, and most recentlyfrom higher elevations. Our study areas were in a late stage of harvestwith most of the valley bottom in young (6- to 45-yr-old) forests and themid-slopes deforested or in remnant patches of old forests.Climate in the Coastal Western Hemlock zone (Meidinger and Pojar 1991)on Vancouver Island is temperate and wet. No month has a mean temperaturebelow 0 C and the mean temperature of the warmest month is 17 C. In eachyear there are usually 291 frost free days, an average of 820 mm of snow,and a mean precipitation of 2140 mm (Meidinger and Pojar 1991).Table3.1.Physicalfeaturesof4studyareaslocatedonVancouverIsland,BritishColumbia.StudyNorthLatitudeAreaForestCoverElevationDrainageAreaWestLongitude(km2)Ratioa(masi)DirectionbCaycuseRiver48°48’-124°30’11144:36:20:00200-12492700ChemainusRiver48°56’-124°05’3306:82:10:02300-1541125°NanaimoRiver49°02’-124°12’14529:50:16:04300-154150°NimpkishRiver50°08’-126°30’4143:32:12:12200-1821315°aForestcoverratioisthepercentageofstudyareainold(>250-yr-old),young(6-to45-yr-old),recentlydeforested(0-to5-yr-old),orsubalpinehabitats.bNorthis0°and360°.32METHODSDeer Location SamplesWe presented details about catching, collaring, and monitoring deer inMcNay et al. (1994). We aged deer as fawns (0- to 1-yr-old), yearlings (1-to 2-yr-old), or adults (>2-yr-old) at time of collaring by inspection oftooth wear and replacement (Robinette et al. 1957) and by body size andfacial appearance. About 25% of the ages were assessed and confirmed laterbased on cementum annuli analysis (Thomas and Bandy 1973). Briefly, wemonitored radio-collared deer at each study area by standard triangulationtechniques (White and Garrott 1990). Initially, we sampled deer locationsweekly on an ad hoc schedule but, in 1984, we standardized sampling so that,during a calendar month, each deer was monitored at least once-per-week andonce within each quarter of a calendar day. We also monitored deer moreintensively, once every 2 hr (2-hour), for 3- to 5-day periods duringspecific weather events at Nanaimo and Nimpkish rivers or once monthly atCaycuse River. In rare situations when deer moved rapidly, triangulationwas impossible and data were not recorded. When deer moved slowly (<150 m)during sampling, we accepted potential errors up to 50 m on locations(Schmutz and White 1990). Although we used 2 techniques to estimate deerlocations (see McNay et al. 1994), most (74%) locations were maximumlikelihood estimates (Lenth 1981).Factors Used to Identify Spatial and Temporal ScalesWe assessed straight-line distances and directions between successiverelocations for spatial and temporal patterns after confirming that sampleinterval (weekly data) had little effect on estimated distance travelled(McNay et al. 1994). We examined 4 attributes of movements in space and33time: magnitude (distance), frequency, timing, and direction. Patternswere identified when movements could be described as periodic andrepetitive.Local movements were designated as those in which distance betweenlocations was not greater than the 95th percentile of all distances,excluding 2-hour samples. The 2-hour samples were designated as serialmovements because they were considered to be serially dependent (McNay etal. 1994). We grouped movements above the 95th percentile of all weeklydistances into 3 categories. Dispersals were moves away from, and with noreturn to, the original range (Howard 1960, Bunnell and Harestad 1983).Migrations (adapted from Sinclair 1984) were moves that: (1) were followedby 2 or more local moves before a return to the original location (i.e., useof a spatially distinct range for 2 wk), or (2) were immediately followedby a return to the original range but were repeated at other times (i.e.,predictable returns between spatially distinct ranges). Outliers were movesabove the 95th percentile that could not be defined as dispersal ormigration.Seasons were termed natal (May and June), summer (July and August),hunting and rut (September through November), winter (December throughFebruary), and spring (March and April). Annual periods began during onenatal season (June 15) and ended the next natal season (June 14).We assumed fawns followed their mothers for the first year of life(Harestad and Bunnell 1981, Nelson and Mech 1981) and dispersal could occurin any subsequent natal season (Bunnell and Harestad 1983, Hamlin and Mackie1989, Nelson and Mech 1992). Our definition of the natal season was basedon empirical observations of others (Cowan 1956, Golley 1957, Thomas 1970,Salwasser and Holl 1979, Livezey 1991). We assumed that locations during34that season indicated the natal area for non-dispersing deer (Masters andSage 1985, McCullough 1985, Hamlin and Mackie 1989) and would be wheresubsequent offspring were produced. Non-dispersing, non-migratory deerremained on their natal ranges. We termed that area the natal range whileany different ranges used by migratory deer, usually during other seasons,were alternate ranges.We segregated areas occupied by individual deer into separate rangesif they were spatially exclusive and if deer made migrations between thethem. We used the presence or absence of migrations to classify deer asmigratory or nonmigratory (resident). We classified migratory deer asobligate migratory deer if migrations occurred consistently among years andas facultative migratory deer if migrations occurred more irregularly.We termed isolated patches of concentrated use as activity nuclei(after Don and Rennolls 1983). Each nucleus (from 1-4 in each range) was aspatial clustering of deer locations (i.e., many locations in closeproximity). The centres of nuclei, determined for statistical convenience,were the highest point(s) chosen from a graphical representation of thedistribution of deer locations in space; the z dimension represented thedegree of spatial clustering. Dixon and Chapman (1980) called thisgraphical representation the utilization distribution which formally is thefirst inverse areal moment of the harmonic mean (Samuel and Garton 1987):H.= 1J (3.1)x=1 ixcalculated for each grid point j on an arbitrary plotting grid. H1 is the35harmonic mean value at grid point j, p is the number of animal locations,and is the distance between location x and grid point j. A potentialnucleus was accepted if it had >4 locations associated with it and wastopographically isolated from other potential nuclei. Additional locationswere assigned to a nucleus if <250 m from its centre. Activity nuclei usedduring night hours (18:00 hr to 09:00 hr) were referred to as night-timenuclei; those used at other times were day-time nuclei.Analytical ProceduresWe assessed dispersal distances for deer, collared as fawns oryearlings, that survived at least their next natal season. Dispersaldistance was calculated as the straight-line distance from the natal area tothe position held during the last July (July being the first month after thenatal period) of the deer’s life or during July 1991 (project end).Dispersers and deer that could not be classified as resident ormigratory were omitted from analyses of migratory, local, and serialmovements. These omissions included: (1) 2 deer that dispersed from theirnatal areas as yearlings; (2) 2 adults that made a single, outlier move justbefore death; and (3) 34 adults and 16 juveniles for which we had limitedinformation due to brief data collection intervals ( = 102 d, SE = 12 d).We calculated migration distances and directions based on straightlines between arithmetic mean centres of successively observed ranges. Wemeasured fidelity to ranges (alternatively, deviation in range use) as thestraight-line distance between arithmetic mean centres of successive use ofthe same range or between centres of successive annual ranges for residentdeer. Departure and arrival dates from one range to another were assumed tobe half-way between the date first observed on the new range (end of36migrations) and the date last observed on the former range (start ofmigrations).We assumed that within-range dispersion of locations could berepresented by local movements measured as straight lines between successivelocation samples. We estimated range areas (total area and the compositearea of segregated ranges) using the SAS program of White and Garrott (1990:343) for the minimum convex polygon (Mohr 1947) excluding 2-hour samples.We weighted harmonic mean calculations by time spent at each location(Samuel and Garton 1987), where time spent was the duration between last andcurrent locations. Because the harmonic mean has been criticized (Worton1989) as overly sensitive to choice of grid scale, we used 1 grid scale forall deer. We investigated temporal patterns in use of nuclei with bothlocal and serial moves. The former moves were used to assess repetitive useof nuclei because these moves were temporally unrelated while the latter,temporally related, moves were used to assess duration of use of nuclei.Additional quantification of ranges was provided by the number anddispersion of nuclei within them. Dispersion (r2) was estimated as the sumof the mean squared distances between each individual nucleus and thearithmetic mean of all nuclei (after Swihart and Slade 1985):r2=l1(X1_)2÷’ (Y1- )2; (3.2)where nuclei are numbered I = 1, 2, 3,..., n and X and Y are the spatialcoordinates representing nuclei centres.Samples of 2-hour data for migratory deer on natal ranges were sparseso we limited analyses of serial movement to winter months only. We used a37trigonometric model to describe cycles of serial movements and tested foramplitude and phase differences in mean distance moved among groups of deer.We also tested for temporal effects through winter using calendar month as afactor. Cycle lengths were visually estimated using PROC TIMEPLOT (SASInst. Inc. 1985).We assessed data distributions using techniques in PROC UNIVARIATE(SAS Inst. Inc. 1985). We used Bartlett’s test to assess homogeneity ofvariance after checking normality of data distributions (Zar 1984) andassessed distribution of directional or temporal data with Rayleigh’s z-test(Batschelet 1981). We assessed independence of observations in deermovements (Swihart and Slade 1985) in a related study (McNay et al. 1994)and noted that movement data were unlikely to be independent. Therefore, wemade the assumption that the distance between consecutive weekly samples wasrepresentative of local moves because they were collected systematically andbecause deer could have travelled anywhere in their home range during thetime interval between most samples (McNay et al. 1994). The 2-hour samples,by comparison were intentionally collected as time-dependent samples(Reynolds and Laundre 1990, Kremsater and Bunnell 1992).We tested for differences in population parameters betweenbehaviourial groups of migratory deer by either a two-sample t-test, whenpopulation distributions were normal, or by the normal approximation to theMann-Whitney U-test when population distributions were not normal (Zar1984). The Kruskall-Wallis test (Zar 1984) was used when unequal varianceswere encountered and when population comparisons included resident deer aswell as both behaviour groups of migratory deer. Because resident deer didnot use alternate ranges, we first compared all migratory classes (resident,facultative, and obligate) on natal ranges. Then we compared migratory deer38(facultative or obligate) on both ranges (natal or alternate). If data werenormal and homoscedastic (or could be transformed so) an F-test was used(SAS Inst. Inc. 1985), otherwise we used the Kruskall-Wallis test. We usedWatson’s U2-test (Zar 1984) to compare directional or temporal data.RESULTSDeer Location SamplesWe caught and collared 30 juvenile deer but 13 (10 fawns and 3yearlings) died before their next natal period leaving a total of 17 deerfor analyses of dispersal. For all migration and local movement analyses,we treated 72 non-dispersing, collared deer (69 9 and 3 cc) monitored for atotal of 253 deer-years and 8,533 weekly locations. We described patternsin serial movements using only 2-hour data collected during winter on 23non-dispersing deer in 11 separate monitoring sessions for a total of 73deer-sessions and 2,526 locations.Most collared deer were at Nanaimo River (n = 40) including 32 adultfemales, 7 yearling females, and 1 adult male that was released to the wildafter being raised in captivity for another study (Bunnell et al. 1990a,b).We captured the remaining 32 deer, mostly adult females, at Caycuse River (n= 13), Nimpkish River (n = 11), and Chemainus River (n = 8). Two of thecollared deer at Caycuse River were male fawns; we collared 1 female fawneach at Caycuse, Chemainus, and Nimpkish rivers.Initially (1982-1988), we sampled deer only at Nanaimo River. Wecollected most data during winter months and from day-break through tomidnight. Most (98%) triangulation produced locations estimated with a 95%error ellipse (White and Garrott 1990: 72) of less than 1 ha and anacceptable goodness-of-fit for the bearings used (McNay et al. 1994).39Radio-collars functioned (i.e., produced an identifiable signal on alive deer) for an average 3.4 yr (SE = 0.2; n = 72). At Nanaimo River 1deer provided 9.0 yr of data and 4 others each provided 7.0 yr of data.Although the average number of weekly locations per deer was 173 (SE = 14; n= 72), 18 deer were represented by less than 70 samples and 13 deer had morethan 150 samples. Average time intervals between weekly samples was 8.0 d(SE = 01; n = 8,461) with a mode of 7.0 d. The average time intervalbetween 2-hour (serial) samples was 2.0 hr (SE = 0.03; n = 2,453).Analysis of straight-line distance between successive weekly locationsshowed that relatively few (n = 461 or 5%) of the total sample exceeded 1.2km (Fig. 3.1). Because that distance represented only 9% of the maximumdistance observed, it also represented a sharp deflection in the cumulativefrequency distribution (Fig. 3.1). We used that distance to distinguishlocal movements (<1.2 km) from other movements. Twenty-eight of the 72 deermade a total of 256 moves that satisfied our definition of migration andhence were called migratory deer. We classified the remaining non-localmoves as outliers (n = 185) or as fragments of migrations (n = 20).Outliers represented 2.1% of locations and reflected moves to traverseoccupied ranges or moves off the range to locations used only once ( = 1.7km, SD = 0.6; n = 185). Those moves occurred most frequently in June andNovember (Fig. 3.2) and were made at least once by all deer. We could notdetect any effect on distance of outlier moves due solely to migratory class(F = 0.80; df = 1,37; P = 0.3743) or to class of range (F = 0.32; df = 1,37;P = 0.5733) being used at the time the movements occurred.Dispersal: Life-time ScaleOf 17 deer that could have dispersed (Table 3.2) only 2 (12%) did so.40120100U)080CuC.)0Cu0‘I60a)C.)I0.a)CUEC-)2000 2 4 6 8 10 14Distance between successive locations (km)Figure 3.1 Cumulative frequency distribution (%) for the distance betweensuccessively sampled locations (n = 8,000) of radio-collared, black-taileddeer on Vancouver Island, British Columbia, 1982-1991.1241Co00E00I,0I..0.0EzFigure 3.2 Number and timing of outlier moves (moves >1.2 km that were notdispersals or migrations) made by radio-collared black-tailed deer onVancouver Island, British Columbia, 1982-1991.35302520151050J F MAM J JASON DMonth42None of the 7 fawns dispersed as yearlings; 2 died as yearlings and none ofthe remaining 5 dispersed as 2-year-olds. Two of the 10 deer trapped asyearlings dispersed in their second natal season and 3 others died in theirsecond year without dispersing. The remaining 5 yearlings lived into theiradult years but never dispersed. The average distance between the estimatednatal site, for non-dispersing deer, and the site occupied in their lastnatal season (Table 3.2) was 0.7 km (SE= 0.1; n = 15).The 2 deer (#13592 and #15504 in Table 3.2) that dispersed werecollared during the winter of 1988-89 and dispersed from the collaring siteduring the following natal period. They were last recorded close to theircapture site on June 23 and were found together on July 04, >7 km to theeast where they resided until September. During the months of September andOctober, both deer travelled a further 10 km east stopping for brief staysat each of 2 major rivers along the way. On October 3 they reached theirmaximum distance from the original site of collaring (about 25 km), stayed 2wk, then began to return west. One disperser was hit and killed by avehicle on October 13, 1989. The surviving disperser finally settled at thesame site she used during the previous summer, still >7 km from the site ofcollaring. She stayed there until the end of the study, and no furthersigns of dispersal or migration were recorded through the following 1.5 yrof monitoring.Migration: Seasonal ScaleA total of 28 radio-collared deer migrated at least once: 2 of 8 atChemainus River, 2 of 11 at Nimpkish River, 7 of 13 at Caycuse River, and 17of 40 at Nanaimo River. All 3 males migrated and, of the 25 migratoryfemales, 1 was a fawn who migrated along a similar route as her radio-43Table 3.2. Straight-line distance between the estimated natal site and thesite occupied in their last natal season for radio-collared, black-taileddeer caught as subadults (<2-yr-old) at 4 study sites on Vancouver Island,British Columbia, 1982-1991.Deer Migrator Sex Age at Age at Distancenumber capture death moved(yr) (yr) (km)15301 Yes 0.75 2.17 1.317912 Yes 9 0.67 2.17 0.117991 No 9 1.50 2.42 0.518412 Unknown d 1.83 2.42 0.519311 Unknown 0.75 1.17 0.619312 Yes 0.75 2.25 1.915101 No 9 0.75 3.17 0.213592 Unknown 9 1.75 2.33 8.115411 No 9 0.67 2.17 0.215504 No 9 1.58 3.17 7.99101 Yes 9 1.58 4.42 1.115001 Yes 9 1.58 6.75 1.015401 No 9 1.75 7.83 1.016102 Yes 9 1.83 4.75 0.116601 Yes 9 1.92 6.33 0.717701 No 9 1.92 6.67 0.517803 Unknown 9 0.75 1.58 0.444collared mother. Associated with the migrations were 12 movements composing6 visits (<4 wk stays subsequently followed by longer stays), possibly toassess conditions at other ranges, and occasional moves between closelyassociated ranges along a single migration route (n = 42) leaving a total of202 full migrations (n = 94 moves away from natal ranges and n = 108 returnmoves to natal ranges).The use and pattern of, or lack of, migrations revealed a differenceamong deer which led us to classify 3 behaviour types. Resident deerclearly chose different tactics than migratory deer (i.e., their tacticswere devoid of large movements). Among migratory deer, one group migratedannually and spent long periods away from their natal range (obligatemigratory) while a second group migrated less regularly and were away fromtheir natal ranges for comparatively short and variable durations(facultative migratory). Time spent away from the natal range by the 2groups (Fig. 3.3), which were first discriminated by Duncan’s and LSD (df =59; p < 0.05), differed (F = 7.58; df = 27,59; P = 0.0001). Facultativemigratory deer left their natal range for <80 d per trip while obligatemigratory deer stayed away >148 d per trip. On an annual basis, the averagetime spent on alternate ranges was also different (t = 61.14; df = 65; P <0.0001); >6 mo for obligate migratory deer and <4 mo for facultativemigratory deer (Table 3.3). All 7 migratory deer at Caycuse River, 1 of the2 migratory deer at each of the Chemainus and Nimpkish river sites, and 7 ofthe 17 migratory deer at Nanaimo River were facultative migratory deer.Magnitude.- We summarized distances between ranges for migratory deer inTable 3.3. On average, migration distances were different between thebehaviour types (U = 8095; n =95,119; P < 0.05). More than half theobligate group travelled >5 km while only 2 of 16 facultative deer migrated45>‘C’)Cua)0)CCuL.CuCUCE0.4-CUCuCa)a.Cl)a)E0)CUI-C)‘IIndividual radio-collared deerFigure 3.3 The average time-per-trip spent away from the natal range formigratory, radio-collared, black-tailed deer on Vancouver Island, BritishColumbia, 1982-1991. Individuals are ranked in their order of duration.mean+SEmeanI mean-SE‘•i. 1•11 -.300250200150100500IIITable3.3.Movementcharacteristicsfor3behaviourgroupsofradio-collared,black-taileddeeronVancouverIsland,BritishColumbia,1982-1991.Movesarestraight-Linedistancesbetweensuccessivelocations(using2-hourdataforserialmovesandweeklydataforLocalmoves)orbetweenarithmeticmeanrangecentres(formigrations),rangedeviationisthemeanstraight-linedistancebetweenarithmeticmeancentresofsuccessivelyusedrangesofthesametype,nucleidispersionisthemeansquareddistance(afterSwihartandSlade1985)betweennucleichosenfromutilizationdistributions(DixonandChapman1980)forindividualdeerranges,frequencyofnucleichangesisthemeandailyfrequencyofmovingbetweennuclei,anddistanceofnucleichangesisserialdistancemovedtochangenuclei.DeerBehaviourObligateFacultativeResidentPoolednSDnSDnSDnSDMigrationduration(d/yr)2093451663358Migrationdistance(km)5.6953.73.01192.04.22143.2Rangedeviation:natal(km)0.3300.30.4330.30.21060.10.21690.2a’ternate(km)0.5350.90.6321.10.6671.0Localmoves:natalrange(km)0.45440.30.314390.20.449240.20.369070.2alternaterange(km)0.47790.20.33140.30.410930.2Rangearea:total(km2)11.0127.27.4166.21.9441.84.6725.6natal(km2)1.0120.81.7161.81.9441.81.7721.7alternate(km2)2.5124.92.0162.72.2283.7Nucleidispersion:natal(km)0.4100.10.4110.20.4350.10.4520.2alternate(km)0.460.20.650.50.4150.3Nucleiuse:frequency(f/d)4.682.42.1271.32.6461.62.6811.7distance(km)0.3990.20.31620.20.33400.20.36010.2Serialmoves:total(km)0.11200.10.16290.20.110880.10.118370.1outsidenuclei(km)0.1290.20.12810.20.22810.20.25910.2insidenuctei(km)0.1910.10.13480.10.18070.10.112460.10)47that far.Frequency.- Deer used individual ranges repetitively and displayed strongfidelity to specific range locations. The deviation in centres of rangeswas <0.5 km (Table 3.3) regardless of migratory behaviour (F = 2.5; df =1,126; P = 0.1166) or range type (F = 0.81; df = 1,126; P = 0.3695).Resident deer also showed strong affinity for their range locations, thecentre of which varied <0.2 km on an annual basis ( = 0.2, SE = 0.1; n =106).Timing. - Departures from, and arrivals at, natal ranges were concentratedaround November and May, respectively (Fig. 3.4). Obligate migratory deerexhibited the most concentrated timing of moves overall (z = 35.43; n = 45;P < 0.05) when they returned to their natal ranges (May 26). By comparison,facultative migratory deer returned to their natal ranges earlier (February21) but the timing was not different than random (z = 10.03; n = 63; P >0.05). On average, obligate migratory deer left their natal ranges onOctober 20 (z = 24.33; n = 38; P < 0.05) while facultative migratory deerdeparted on December 10 (z = 20.19; n = 56; P < 0.05). The migration dateswere significantly different between the 2 groups for both returns to (U2 =2.19; n = 63,45; P < 0.05), and departures from (U2 = 1.89; n = 56,38; P <0.05), natal ranges.We rarely were able to follow individuals as they migrated. Ourestimate of the elapsed time for each trip, an average of 5.0 d (SD = 6; n =204), estimates maximum duration because it is similar to the average timeinterval between location samples for all weekly data.Direction.- We found a significant pattern to the direction of migrationfor facultative migratory deer (z = 5.02; n = 119; P < 0.05) but no patternfor obligate deer (z = 0.29; n = 95; p > 0.05). Most facultative deer48035>25020C,Figure 3.4 Number of migratory moves made during each month of the year by2 behaviour groups of radio-collared, black-tailed deer departing natalranges (top) and returning to natal ranges (bottom) on Vancouver Island,British Columbia, 1982-1991.30Co00E>1‘-200Cu.2’l 5E‘I0100E5zJ F MAM J JASON DB)FacultativeObligate01:1ZI ..IIIIIIIIIIIII ...:iJ F MAM J JASON DMonth49migrated along their main river valleys (Fig. 3.5). Seven of 12 obligatemigratory deer made notable deviations to valley directions by travellingacross 1 or 2 valleys to an alternate range; only 2 of 16 facultativemigratory deer made similar deviations.Local Movement: Daily ScaleMagnitude. - Most (95%) weekly locations were <1.2 km apart ( = 0.4 km, SD= 0.2; n = 8000) and we never recorded any combination of local moves thatresulted in a switch from one range to another. Distances betweensuccessive locations on natal ranges did not differ (F = 0.65; df = 1,6904;P = 0.4198) between obligate and resident deer (Table 3.3) but, forfacultative deer, that distance was significantly less (F = 10.16; df =1,6904; P = 0.0014). Similar results occurred on alternate ranges whereconsecutive locations for obligate deer were even more dispersed than thoseof facultative deer (F = 27.44; df = 3,3072; P 0.0001). Observations ofmovements (2-hour data) between activity nuclei (see Frequency below)resembled dispersion of weekly observations (Table 3.3). Again, facultativemigratory deer made smaller movements on average (F = 14.06; df = 1,598; P =0.0002).Another index of local movement is home range size (Fitch 1958).Total, or life-time, home range size for the 1 surviving disperser was 0.4sq km at the site of collaring (January-June, 1989), 21.3 sq km duringdispersal (July-November, 1989), and 0.3 sq km for the period followingdispersal (December 1989 - June 1991). By comparison, the average totalhome ranges for non-dispersing deer ranged from 1.9 to 11.0 sq km.Migratory deer had total home ranges that were 3 times (Ffacultative = 36.68;df = 1,69; P < 0.0001) to 5 times (Fobllgate = 60.46; df = 1,69; P < 0.0001)Figure3.5Directionsofmovementfromnatalrangestoalternaterangesforobligatemigratory(solidarrows),andfacultativemigratory(dashedarrows),radio-collared,black-taileddeerat4studysitesonVancouverIsland,BritishColumbia,1982-1991.1.001.02.0—KIIornetrescuseRic.en51larger than those used by resident deer (Table 3.3) yet smaller than thatused by the disperser.Generally, size of individual seasonal ranges averaged 1.0 to 2.5 sqkm (Table 3.3). Although we could not detect differences in seasonal rangesizes between the 2 groups of migratory deer (F = 0.70; df = 3,51; P =0.5543), estimates of natal ranges were smaller for obligate migratory deerthan for resident deer (F = 4.94; df = 1,69; p = 0.0296).Still another index of local movement is distance between activitynuclei (Fig. 3.6). Dispersion of nuclei (Table 3.3) was not different amongbehaviour types (F = 0.58; df = 2,49; p = 0.5563) or between range types (F= 0.80; df = 1,37; p = 0.3743), averaging 0.4 km (SE = 0.2; n = 67).Freciuencv. - We monitored deer occupying separate ranges (i.e., separated inspace, such as natal or alternate ranges, and separated in time, such as adeer using a natal range after having used an alternate range) on 336different events. One hundred of these events were represented by <10locations and in these cases of low sample size, and in some other cases ofextremely dispersed locations (n = 16), we were unable to detect specificactivity nuclei. The remaining sample (n = 220) represented 87 of thepossible 100 individual deer x range combinations and 71 of the 72 non-dispersing deer. Most deer x range combinations had at least 2 nuclei (Fig.3.7) and 67 showed repeated use of nuclei with successive periods of use.Although 2 nuclei (each being 39.2 ha by our definition) represents only 16%of the average range size (Table 3.3), they typically enclosed 61% of thelocations. Only 5 deer x range combinations had unique nuclei and 20 had noreplication of range use upon which to evaluate repetitive use of nuclei.When repetitively used, nuclei were displaced by 0.1 km (SE = 0.1; n = 153)on average.—-CDc-TIcc-CD——CUtilizationdistributionoCD—c-s.c-s.CD)-“D —--c-s.—a.CD(DN(i’5JCDc-s.z—c-s.0)CDc-s.,0_0—CDc-s.0CD-I,rD)C-)D)c-s.c-s. ———-“r1o—‘<C,)—3CDCD,-.-Q_,—-5,—CDoC-.)0o-a..CLO)co—%_-1CD-oo—ICD—JCDD--.oco1WCDC)coow-—I‘0)-1DCDCD0)Qc-s.-CD—I30)——C_)CDc-CD•0)s.—-—ICD3CDCDCD-C,C)CD53Cl)C)a)C)C)•aII0I..C).0Ez1008060402004Number of nuclei in rangeFigure 3.7 Frequency of radio-collared, black-tailed deer, seasonal rangeshaving 1 or more activity nuclei.1 2 3 5 654Timing. - Observations of movements in 2-hour intervals during winterrevealed that deer changed activity nuclei 2-3 times daily (Table 3.3).Obligate migratory deer changed nuclei twice as often as other deer (F =10.14; df = 1,78, P =.0021). We could not detect seasonal periodicity indistance between locations once the effect due to behaviour type was removed(F = 0.15; df = 4,7985, P =.9640). Similarly, we could not detectperiodicity in the day-time or night-time use of nuclei (only 12 of 48comparisons were significant [x2 < 0.05]; the remaining 39 deer x rangecomparisons had insufficient samples to complete the x2 analysis).Observations of movements (2-hour data) between activity nuclei occurredmost frequently at crepuscular hours (Fig. 3.8 top). Sample size wasinsufficient to evaluate periodicity in seasonal use of nuclei.Direction. - Directions for local movements based on weekly data did notdiffer from random (P > 0.05) for any deer on either range.Serial Movement: Hourly ScaleDuring winter, serial movements were dependent and cyclic in time andcould be represented in the following trigonometric form:D = 148.6—37.2sin(1 * HR) + 76.9OBL + 56.4DEC; (3.3)where D is mean distance between consecutive locations, HR is hour of day,OBL is a dummy variable for obligate migratory deer, and DEC is a dummyvariable for the month December (r2 = 0.12; = 11.36; F = 18.26; P <0.0001; n = 411). Largest movements typically occurred at 08:00 and 20:00hr with least movement at 12:00 and 24:00 hr.All deer moved more in December than either January or February (P <55Hour of dayFigure 3.8 Mean distance between successive, 2-hour, relocations of radio-collared black-tailed deer by the hour of day as deer were departing nuclei(top), or within nuclei (bottom), at 4 study sites on Vancouver Island,British Columbia, 1982-1991.A)5004003002001000500400300200C.)z0)t0a)•0a)C.)Cu4-’CO•0Cua)a)C.).4-’C.)Cu4-’Cl)•0CuG)A02 468 10121416182022L B)F1000 02468 10121416182022560.001) and obligate migratory deer generally moved further than eitherfacultative migratory or resident deer (P < 0.001). Distance of serialmovements, excluding those made as deer departed specific nuclei, weresimilar for all deer (Duncan’s; P > 0.05, Table 3.3) but distance travelledoutside nuclei was twice that travelled inside nuclei (F = 66.25; df =1,1822; P < 0.0001). Movements inside nuclei during crepuscular hoursgenerally were the longest (Fig. 3.8 bottom).DISCUSSIONMovements as Scalar Classes of ActivityWe found that movements could be isolated into distinct classes alongseveral dimensions and hence were scalar (Levin 1992). Movements wereclearly scalar in their frequency of occurrence; dispersal once in a lifetime, migration usually twice each year, local movements several times aday, and finally, serial movements more-or-less continuously in definedperiods of activity. Movements also differed, but less distinctly, alongother dimensions. Migrations and dispersals tended to be longer and moredirectional than local or serial movements. Timing differed as well withdispersal occurring during the natal season, migrations occurring in early-and late-winter, local movements occurring mostly at crepuscular hours, andserial movements throughout a calendar day.Our quantification of these scalar classes of movements supports thecommon use of terms described elsewhere. Other workers also have reporteddeer dispersal as a single event (once in a life-time) covering longdistances and associated with the natal season (Harestad and Bunnell 1981,Schoen and Kirchhoff 1985, Hamlin and Mackie 1989, Nelson and Mech 1992).Migrations were similar in distance to those reported for black-tailed deer57in other studies (Harestad 1979, Schoen and Kirchhoff 1985, Livezey 1991)but were remarkably less than those reported for mule deer (0. h. hemionus)migrations (17-110 km in Garrott et al. 1987, 29 km in Kufeld et al. 1989,27-51 km in Loft et al. 1989, 4-66 km in Brown 1992). The difference couldbe due to the topographically insular nature of study areas throughout therange of black-tailed deer. Coastal watersheds in Washington, BritishColumbia, and southeast Alaska tend to be steeper sloped and narrower thaninterior watersheds inhabited by mule deer, thereby creating potentialconstraints on migration while proximity of seasonal habitats is increasedthrough topographic relief. We found no other reports specifying patternsof serial movements by deer but our observations of local movements and homerange sizes were similar to those reported elsewhere (Harestad 1979, Schoenand Kirchhoff 1985, Livezey 1991).Movements as Hierarchically Structured DecisionsMovements revealed that resource use was not only a scalar activitybut appeared to be based on decisions that were nested hierarchically.Local movements occurred as regular periodic activities and were alwayscontained temporally and spatially within decisions made at higher classesof movement. This physical aspect of movement was described as a nestedhierarchy by Gautestad and Mysterud (1993). They used an hierarchicalmovement process as the basis for area utilization assessments to offer “amore solid platform for parameter estimates and statistical tests than thetraditional [home range] protocols”. Their premise was that, bydistinguishing between physical aspects (i.e., nested subclasses withinsuperclasses) and biological aspects of animal movements, local and temporalvariation in movements could be more clearly correlated with environmental58parameters (Gautestad and Mysterud 1993). For black-tailed deer, specificdecisions about seasonal ranges were contained within the more generaldecision about home-range settlement (the latter being a decision between 2choices, dispersal or philopatry). Similarly, when we could identifyspecific nuclei, local moves between them were always on the same seasonalrange and thus contained within the most recent decision to migrate.Further, serial movements showed a cyclic pattern of smaller movementswithin activity nuclei and thus were contained within decisions of localmovements between nuclei.Decision Information Transfer and Hierarchical FunctionBecause movements by deer are scalar, and because the decisions theyrepresent show containment, we concluded the resource use problem is anested hierarchy. Furthermore, we consider this particular hierarchy to bemore than a simple classification framework because it embodies a functional(information transfer) aspect. We contend the pattern (magnitude,direction, frequency, and timing) of movement decisions reveal thatsuperclass decisions: (1) were based on a synthesis of subclass decisionsand (2) once made, constrained the breadth of choices for those subclassdecisions. These are standard, physical implications of typical nestedhierarchical function (Dawkins 1976).Beyond the physical implications of the hierarchy, we also notedspecific tactics for decisions about resource use at several scales. Thesetactics generally implied that black-tailed deer use a conservative approachto dealing with resource changes; that conservatism tends to reinforceconstraints within the hierarchy of decisions about resource acquisition.We present these tactics at each scale of the hierarchy below.59Dispersal: Life-time Scale.- The decision to disperse or to be philopatricis an irrevocable commitment, made once, to a long-term tactic and,therefore, to a range of more specific tactics for subsequent resource use(Dawkins 1976, O’Neill et al. 1989, Levin 1992). This decision at the lifetime scale is one of choosing home ranges and, among deer in our study,appeared to constrain, for example, tactics for seasonal range selection.The lowest proportion of migratory deer (highest proportion of residentdeer) came from Nimpkish River where our study was located close to valleybottom. Kufeld et al. (1989) found that most deer in the Rocky Mountainfoothills of north-central Colorado were resident; deer at higher elevationsin Colorado were reported to migrate seasonally (Garrott et al. 1987).Obligate migratory deer in our study came mostly from Nanaimo River, arelatively mountainous area (Table 3.1). Harestad (1979) reported mostlymigratory deer at Davie River, located close to Nimpkish River but moremountainous. These observations illustrate that the decision to migrate isnot only contained within the decision about where to live (dispersal) butmay also be constrained by it.In comparison to that top-down perspective, synthesis of resource usedecisions accumulated on both alternate and natal ranges could guide thedecision of philopatry or dispersal. Choice of home ranges, regardless ofthe impetus (e.g., social interactions, population density, habitat quality,genetics), ultimately must be based on some form of information regardingcomposite resource conditions. Because so much information needs to beassessed, Orians and Heerwagen (1990) considered this process to be unwieldyand proposed that animals use environmental cues to act as indicators ofoverall suitability. The long distance travelled by the 2 dispersers weobserved (Table 3.2; see also 1-tarestad and Bunnell 1981) and the circuitous60route taken suggests considerable exploration and testing to determine afinal site to settle. These activities would be necessary only if knowledgeabout finer scales of resource use at new sites was incomplete yetimportant.The effort, incomplete knowledge, and risk involved in dispersalapparently contribute in a proximal way to home range selection among deerand taken together represent important mechanisms for developingevolutionary responses (Orians and Heerwagen 1990). While the structure ofthe hierarchy implies a transfer of information between levels concerningthe decisions of dispersal and migration, empirical observation revealsspecific tactics that also guide the final choices. Loft et al. (1989)suspected a high level of philopatry in a population of mule deer; aconclusion commonly reported from other studies of social interactions amongdeer (Hirth 1977, Nelson and Mech 1981, Hamlin and Mackie 1989). Weobserved only 2 dispersals and although we can not be certain, we believemost of our sample deer remained in matriarchical groups that overlapped, orat least were adjacent to, their true natal areas. It seems that stayingclose to family is a strong component of the tactics for settling homeranges.Philopatry implies an adaptive advantage in that young deer rely onmaternal expectations of future resources (i.e., the maternal home range iswell suited to the resource conditions). Because the decision is madeinfrequently, the advantage may not be realized if resource conditions asthey are evaluated at the home range scale are, or become, unstable (Allenand Starr 1982, Gass 1985, Senft et al. 1987). Forest growth and renewalis, for the most part, relatively slow in coastal forests (Franklin andSpies 1984, Bunnell 1995) and although wildfire may effect large areas,61frequency of fires in coastal British Columbia tends to be high only inshort periods isolated by 100 yr intervals (Schmidt 1970).Migration: Seasonal Scale. - The cyclic nature and consistent timing ofmigrations made by obligate migratory deer in our study resembled thepatterns described elsewhere (Harestad 1979, Garrott et a!. 1987, Loft eta!. 1989, Brown 1992). The facultative behaviour we observed is commonlyreported in studies of some birds (e.g., Terrill and Ohmart 1984) but hasnot been acknowledged as a distinct behaviour in populations of deer eventhough it has been described previously (Fairman 1966, Harestad 1979,Garrott et al. 1987, Hamlin and Mackie 1989, Brown 1992). Obligatemigratory deer migrated further, followed less predictable directions, andused alternate ranges longer and more consistently than facultativemigratory deer (Fig. 3.5 and Table 3.3). The choice of migration tactics(obligate, facultative, or no migration) commit individual deer to 1 tactic,at least for a season, resulting in a limited range of tactics for morespecific decisions about habitat use. This decision at the seasonal scaleinvolves selection of seasonal habitats. Similar to the way that migrationtactics apparently depend on the choice of home ranges, we found that localmovements (especially area of use and frequency of local moves; Table 3.3)apparently depend on specific migration behaviour. Because we observedconsistent differences in local moves among classes of migratory behaviour,we concluded the choice of migration tactics constrains the breadth oftactics for local movements and subsequent resource use decisions.Previous attempts to understand migration patterns have led toexplanations based primarily on changes in weather (McCullough 1964,Harestad 1979) and/or changes in condition and abundance of forage (Klein1965, Garrott et al. 1987). Such changes in resource condition could be62recognized by animals only through synthesis of an accumulation of finer-scale resource use decisions. For example, facultative migratory deer movedto alternate ranges along valleys (only as far away as valley bottoms) andstayed for relatively short periods. We considered this pattern to indicatea tactic guided by a synthesis of environmental conditions. It seemsdoubtful, for example, that migrations could be as inconsistent temporallyif the tactic was imposed by higher scale (home range) decisions.We found the highest proportion of facultative migratory deer camefrom Caycuse River, a coastal watershed with a high degree of maritimeinfluence, little high elevation alpine habitat, steep slopes, and littlelow elevation flat topography. This movement pattern for facultativemigratory deer, combined with that of obligate migratory deer (summarizedabove) and the lack of migration by resident deer, suggests that localclimate and geography at natal ranges strongly influences the need to moveto alternate ranges in winter. In more mountainous areas than Caycuse forexample, deer would be more likely to migrate each year, as obligatemigratory deer do by definition. The timing of migrations made by obligatemigratory deer indicate avoidance of severe winter conditions that arelikely to occur on high-elevation natal ranges during winter months. Atmid-elevations, or where winter weather is moderated by coastal climate,deer may choose not to migrate except in times of severe weather -- thepattern of facultative migration. Finally, at lowest elevations (e.g., ourNimpkish River site or valley bottoms at other sites) deer remain residentwith no migration.Our study was not designed to address the adaptive significance ofthese tactics for selecting seasonal ranges. For example, we are unable toconclude if these are only partially migratory populations resulting from a63mixed strategy where groups remain reproductively isolated or from aconditional strategy where individuals choose a gradation of tacticsdepending on their individual circumstances (Swingland 1984).Deer displayed a tactic of fidelity to alternate ranges (Table 3.3,Garrott et al. 1987) indicating an element of tradition in the choice ofactual sites. Loft et a!. (1989) took such fidelity a step further bysuggesting that the selection of winter areas was based largely upon familymembers following each other. Other studies of spatial and behaviourialorganization provide similar suggestions (Hirth 1977, Hamlin and Mackie1989). If choice of alternate ranges in winter was guided by a matriarchthen offspring should conform to the migration tendencies of their mothers.Sweanor and Sandegren (1988) reported similarity of behaviour amongrelatives in a population of partially migratory moose (Alces alces). Wefound no data to test the hypothesis for black-tailed deer; parenthood wasunconfirmed, but female fawns appeared to follow their mothers duringmigration in this study. This tactic must act in conjunction with the moregeneral tactic of maintaining suitable seasonal ranges where suitability isbased upon resource conditions for a variety of needs but primarily seasonalforage quality and availability.Choices of seasonal ranges appeared to be strongly habitual, an actionargued to be more common in static or predictable environments (Gass 1985).Again, as in philopatry and for the same reasons, we view fidelity toalternate ranges to be an adaptive advantage in old, coastal forests(especially when the choice is usually that of the matriarch).Local Movement: Daily Scale. - Local movements represent decisions to movebetween activity nuclei within which more refined decisions about resourceuse (serial moves) will be made. At the daily scale, local moves indicate64habitat selection and specific decisions about resource use should be madewithin that choice of habitats. Because habitats range widely in theirability to meet the needs of deer we assume that habitat choices reflect afocus on 1 or several, but not all, needs (e.g., Miller 1970). Local moves,therefore, constrain the types of decisions about actual resources that maybe acquired within the specific site chosen. At this daily scale, specificgoals in resource use becomes more differentiated than at higher scales ofresource use.Morgan (1994) and others (Klein 1965, Miller 1970) discussed intraseasonal adjustments in the locations of activity centres in relation tophenological changes in forage implying that deer move to new locations,expanding their seasonal ranges, as forage changes. Morgan (1994) alsonoted use of special activity sites located on the periphery or outsideseasonal ranges and attributed movements to those locations for the purposesof parturition. His conclusion was based largely on the timing of themovements. Others have similarly considered such movements to unusuallocations as temporary escape from predators (Kufeld et al. 1988, Hoizenbeinand Schwede 1989). Our data (Fig. 3.2) also illustrate movements toisolated sites, presumably to seek specific resources.We also noted resident and facultative migratory deer used natal areastwice the size of the areas used by obligates. Because we assume few, ifany, barriers to movement during summer months, we concluded that the natalranges of obligate migratory deer have the highest density of all resourcessought by deer (Miller 1970). Contrary to use of space on natal ranges,obligate migratory deer used 25% more space on alternate ranges than eitherfacultative migratory or resident deer. On average they also had morenuclei and changed between nuclei twice as often. In winter, we assumed65barriers to movement would exist when interception of snow by forestcanopies (Kirchhoff and Schoen 1987, McNay et al. 1988) was insufficient toarrest development of deep snowpacks. Because snow buries forage we assumedthat mobility to locate remaining forage and windthrown arboreal foragewould be an asset in forage rich communities (Harestad et al. 1982, Bunnell1985) and a detriment where little or no forage was available (Parker et a!.1984). By evidence of their greater mobility, we concluded obligatemigratory deer likely had access to comparatively better habitats than otherdeer during periods with snow and local moves were primarily instigated byforaging decisions.Black-tailed deer have adopted local movement tactics that includediurnal timing and a tenacity for specific activity sites. Similar tophilopatry and fidelity, these choices again represent habit although with amarked variation (Figs. 8 and 9). Some of this variation has beenattributed to daily weather patterns (Miller 1970), to reproductive status(Hoizenbein and Schwede 1989), and to the occurrence of predators (Kufeld etal. 1988). Presumably, because decisions on local movements and habitatselection are made more frequently than those concerning home ranges orseasonal ranges, the ability to adapt to new resource conditions is greaterthan is apparent at those higher levels of resource use (Staddon 1983, Gass1985, Senft et al. 1987).Summary. - Field studies that discuss movements in context of the completeprocess of resource use are rare. Orians and Wittenberger (1991) discussed2 different spatial and temporal scales of habitat selection and concludedthat goals about resource use were different at different scales. Wedescribe 4 scales (temporal and spatial) of movement that form a functionalhierarchy of resource use decisions. Considered from the top of the66hierarchy, we observe constraints that occurred both as a function of thehierarchy and as a function of specific tactics. Viewed from the bottom ofthe hierarchy, we interpret a synthesis of resource use decisions that couldform proximate instigation for higher scale decisions. Decisions at thehourly scale were frequent, flexible, and likely intended to achieve goalsthat were perhaps overlapping but not simultaneous. For example, foragingcan be at the expense of thermal suitability, can increase exposure topredators, or can sacrifice conspecific connections (Stephens and Krebs1986). Goals at the daily scale are likely more general, somewhat flexible,and must integrate the future value of resources over an entire season.Goals at the seasonal scale appear focused toward even more generalconditions of resources, specifically those that change seasonally (e.g.,snow conditions); facultative migratory behaviour strongly implies a focuson avoidance of extended temporal and spatially unfavourable environmentalconditions. Finally, at the life-time scale, choice of home range mustincorporate a synthesis of expectations for life-long resource conditions.Implications of Constraint on Movement DecisionsPhilopatry, a constraint on species’ range expansion, on colonization,and on gene flow, also constrains the choice of migration tactics. Fidelityto migration tactics constrains the range of habitat choices. Tenacity forspecific sites constrains the range of resources and resource procurementtactics. Together, this collection of tactics unifies different scales ofhabitat selection. But these tactics are, at the top of the hierarchy,based on strong family bonds which limit mobility of individual black-taileddeer and thus amplify constraints on learning about altered resourceconditions beyond those resulting from the hierarchical nature of resource67use. Family groups, or demes, could be thought of as metapopulations (e.g.,Harrison et al. 1988). Hence, response to large-scale habitat alterationbecomes a function of metapopulation dynamics rather than the usualassumption of individual learning and subsequent redistribution based onhabitat preferences (Pulliam and Danielson 1991). Supporting evidence forthis effect can be found by considering the conclusions reached by others onsuch issues as dispersal and migration (Hamlin and Mackie 1989), genetics(Manlove et al. 1976), and general social behaviour (Hirth 1977).Matrilineal groups may imply competitive benefits because continuedassociation with a matriarch and her resource choices ensures offspring willadopt and practise tactics that have worked in the past. We expect thesetactic are beneficial when habitats change slowly. By far the mostwidespread, rapid, and consistent alteration of deer home ranges is causedby logging and since the late 1960’s, this activity has been commonthroughout Vancouver Island. When habitats are altered extensively andrapidly, potential lags in response could lead to inappropriate andsuboptinial resource choices (Orians and Wittenberger 1991).LITERATURE CITEDAllen, T. F. H., R. V. O’Neill, and T. W. Hoekstra. 1984. Interlevelrelations in ecological research and management: Some workingprinciples from hierarchy theory. USDA For. Serv. Gen. Tech. Rep. RM110. Ft. Collins, Cob. llpp.Batschelet, E. 1981. Circular statistics in biology. Academic Press,N.Y. 371pp.Brown, C. G. 1992. Movement and migration patterns of mule deer insoutheastern Idaho. J. Wildl. Manage. 56:246-253.Bunnell, F. L. 1985. Forestry and black-tailed deer: conflicts, crises, orcooperation. For. Chron. 61:180-184._____• 1994. Forest dwelling vertebrate fauna and natural fire regimes:patterns and implications for conservation. Cons. Blob. 8:In press.68_____and M. P. Gillingham. 1985. Foraging behaviour: dynamics of diningout. Pages 53-79 in R. J. Hudson and R. G. White, eds. Bioenergeticsof wild herbivores. CRC Press, Roca Baton, Fla.and A. S. Harestad. 1983. Dispersal and dispersion of black-taileddeer: models and observations. J. Mamm. 64:201-209.and G. W. Jones. 1984. Black-tailed deer and old-growth forests --a synthesis. Pages 411-420 in W. R. Meehan, T. R. Merrell, Jr., andT. A. Hanley, eds. Fish and wildlife relationships in old-growthforests. Proc. Symp. April 1982 in Juneau, Alaska. Am. Inst. Fish.Res. Biol., Morehead City, N.C.F. W. Hovey, R. S. McNay, and K. L. Parker. 1990a. Forest cover,snow conditions, and black-tailed deer sinking depths. Can. J. ofZool. 68:2403-2408.K. L. Parker, R. S. McNay, and F. W. Hovey. 1990b. Sinking depthsof black-tailed deer in snow, and their indices. Can. J. Zool.68:917-922.Charnov, E. L. 1976. Optimal foraging: the marginal value theorem. Theor.Pop. Biol. 9:126-136.Cowan, I. McT. 1945. The ecological relationships of the food of theColunibian black-tailed deer, Odocoileus hemionus colunibianus(Richardson), in the coast forest region of southern Vancouver Island,British Columbia. Ecol. Monogr. 15:111-139.1956. Life and times of the coast black-tailed deer. Pages 523-617 in W. P. Taylor, ed. The deer of North America. Stackpole Press,Harrisburg, Penn.Dawkins, R. 1976. Hierarchical organisation: a candidate principle forethology. Pages 7-53 in P. P. G. Bateson and R. A. Hinde, eds.Growing points in ethology. Cambridge Univ. Press, Cambridge,England.Dixon, K. R., and J. A. Chapman. 1980. Harmonic mean measures of animalactivity areas. Ecology 61:1040-1044.Don, B. A. C., and K. Rennolls. 1983. A home range model incorporatingbiological attraction points. J. Anim. Ecol. 52:69-81.Fairman, L. L. 1966. Movement, populations, and behaviour of deer on awestern Montana winter range. M.S. Thesis, Univ. of Montana,Missoula. ll7pp.Franklin, J. F. 1993. Preserving biodiversity: species, ecosystems, orlandscapes. Ecol. Appl. 3:202-205.and T. A. Spies. 1984. Characteristics of old-growth Douglas-firforests. Pages 328-334 in Society of American Foresters, eds.Proceedings of the Society of American Foresters national convention,69October 16-20, 1983. Soc. of Am. For., Washington.Garrott, R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A. W.Alidredge. 1987. Movements of female mule deer in NorthwestColorado. J. Wildl. Manage. 51:634-643.Gass, C. L. 1985. Behaviourial foundations of adaptation. Pages 63-107 inP. P. G. Bateson and P. H. Klopfer, eds. Perspectives in ethology.Vol. 6. Plenum Publishing Corp., N.Y.Gates, B. R. 1968. Deer food production in certain seral stages of thecoast forest. M.S. Thesis, Univ. of British Columbia, Vancouver.lOSpp.Gautestad, A. 0., and I. Mysterud. 1993. Physical and biologicalmechanisms in animal movement processes. J. Appl. Ecol. 30:523-535.Golley, F. B. 1957. Gestation period, breeding and fawning behaviour ofColumbian black-tailed deer. J. Mamm. 38:116-121.Gould, S. J., and R. C. Lewontin. 1979. The spandrels of San Marco andthe Panglossian paradigm: a critique of the adaptionist programme.Proc. Roy. Soc. London Biol. 205:581-598.Greenwood, P. J., and I. R. Swingland. 1984. Animal movement: approaches,adaptations, and constraints. Pages 1-6 in I. R. Swingland and P. J.Greenwood, eds. The ecology of animal movement. Oxford Univ. Press,N.Y.Hamlin, K. L., and R. J. Mackie. 1989. Mule deer in the Missouri RiverBreaks, Montana: A study of population dynamics in a fluctuatingenvironment. Montana Dept. of Fish and Wildl., Missoula. 4Olpp.Harestad, A. 5. 1979. Seasonal movements of black-tailed deer on northernVancouver Island. Ph.D. Thesis, Univ. of British Columbia, Vancouver.98pp._____and F. L. Bunnell. 1981. Dispersal of a yearling male black-taileddeer. Northw. Sci. 57:45-48.J. A. Rochelle, and F. L. Bunnell. 1982. Old-growth forests andblack-tailed deer on Vancouver Island. Trans. North Am. Wildl. Nat.Resourc. Conf. 47:343-352.Harrison, S.,, D. D. Murphy, and P. R. Ehrlich. 1988. Distribution of thebay checkerspot butterfly, Euphydryas editha bayensis: evidence for ametapopulation model. Am. Nat. 132:360-382.Hirth, D. H. 1977. Social behaviour of white-tailed deer in relation tohabitat. Wildl. Monogr. 53. 55pp.Holzenbein, S., and G. Schwede. 1989. Activity and movements of femalewhite-tailed deer during the rut. J. Wildl. Manage. 53:219-223.70Houston, A. I., C. W. Clark, J. M. McNamara, and M. Mangel. 1988. Dynamicmodels in behaviourial and evolutionary ecology. Nature 332:29-34.Howard, W. E. 1960. Innate and environmental dispersal of individualvertebrates. Am. Midl. Nat. 63:152-161.Huston, M., D. DeAngelis, and W. Post. 1988. New computer models unifyecological theory. BioScience 38:682-691.Kirchhoff, M. D., and J. W. Schoen. 1987. Forest cover and snow:implications for deer habitat in southeast Alaska. J. Wildl. Manage.51:28-33.Klein, D. R. 1965. Ecology of deer range in Alaska. Ecol. Monogr.35:259-284.Kremsater, L. L., and F. L. Bunnell. 1992. Testing responses to forestedges: the example of black-tailed deer. Can. J. Zool. 70:2426-2435.Kufeld, R. C., D. C. Bowden, and D. L. Schrupp. 1988. Influence ofhunting on movements of female mule deer. J. Range Manage. 41:70-72.__________and_____. 1989. Distribution and movements of female muledeer in the rocky mountain foothills. J. Wildi. Manage. 53:871-877.Lenth, R. V. 1981. On finding the source of a signal. Technometrics23:149-154.Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology73:1943-1967.Livezey, D. B. 1991. Home range, habitat use, disturbance, and mortalityof Columbian black-tailed deer in Mendocino National Forest. Calif.Fish and Game 77:201-209.Loft, E. R., R. C. Bertram, and 0. L. Bowman. 1989. Migration patterns ofmule deer in the central Sierra Nevada. Calif. Fish and Game 75:11-19.MacArthur, R. H., and E. R. Pianka. 1966. On the optimal use of a patchyenvironment. Am. Nat. 100:603-609.Mangel, M., and C. W. Clark. 1986. Towards a unified foraging theory.Ecology 67:1127-1138.Manlove, M. N., M. H. Smith, H. 0. Hillestad, S. E. Fuller, P. E. Johns,and D. 0. Straney. 1976. Genetic subdivision in a herd of whitetailed deer as demonstrated by spatial shifts in gene frequencies.Ann. Conf. Southeastern Assoc. Game and Fish Commissioners 30:487-492.Masters, R. D., and R. W. Sage, Jr. 1985. White-tailed deer fawn/daminteractions and fawn home range establishment. N.Y. Fish and Game J.32:93-94.71Maynard-Smith, J. 1978. Optimization theory in evolution. Ann. Rev. Ecol.Syst. 9:31-56.Mayr, E. 1983. How to carry out the adaptionist program. Am. Nat.121:324-334.Mccullough, D. R. 1964. Relationship of weather to migratory movements ofblack-tailed deer. Ecology 45:249-256._____1985. Long range movements of large terrestrial mammals. Pages444-465 in M. A. Rankin, ed. Migration: mechanisms and adaptivesignificance. contrib. Marine Sci., Suppl. vol. 27.McNamara, J. M., and A. I. Houston. 1980. The application of statisticaldecision theory to animal behaviour. J. Theor. Biol. 85:673-690.and_____. 1986. The common currency for behaviourial decisions.Am. Nat. 127:358-378.McNay, R. S., J. A. Morgan, and F. L. Bunnell. 1994. characterizingindependence of observations in movements of Columbian black-taileddeer. J. Wild]. Manage. 58:422-429.L. D. Peterson, and J. B. Nyberg. 1988. The influence of foreststand characteristics on snow interception in the coastal forests ofBritish columbia. Can. J. For. Res. 18:566-573.Meidinger, D., and J. Pojar. 1991. Ecosystems of British columbia.British columbia Minist. of For. Special Rep. Ser. 6. Victoria.330pp.Miller, F. L. 1970. Distribution of black-tailed deer (Odocoileushemionus columbianus) in relationship to environment. J. Mamm.51:248-260.Morgan, J. A. 1994. Factors influencing summer habitat use of black-taileddeer on south-central Vancouver Island. M.S. Thesis, Univ. of Britishcolumbia, Vancouver. 64pp.Mohr, c. 0. 1947. Tables of equivalent populations of North American smallmammals. Am. Mid]. Nat. 37:223-249.Nelson, M. E., and L. D. Mech. 1981. Deer social organization and wolfpredation in northeastern Minnesota. Wild]. Monogr. 77.and . 1992. Dispersal in female white-tailed deer. J. Mamni.74:891-894.Ollason, J. G. 1987. Foraging theory and design. Pages 549-562 in A. c.Kamil, J. R. Krebs, and H. R. Pulliam, eds. Foraging behaviour.Plenum Press, N.Y.O’Neill, R. V., B. T. Milne, M. G. Turner, and R. H. Gardner. 1988.Resource utilization scales and landscape pattern. Landsc. Ecol.722:63-69._____A. R. Johnson, and A. W. King. 1989. A hierarchical framework forthe analysis of scale. Landsc. Ecol. 3:193-205.Orians, G. H., and J. H. Heerwagen. 1990. Evolved responses to landscapes.Pages 555-578 in J. Barkiow, L. Cosmides, and J. Tooby, eds. Theadapted mind: evolutionary psychology and the generation of culture.Oxford Univ. Press, N.Y.and J. F. Wittenberger. 1991. Spatial and temporal scales inhabitat selection. Am. Nat. 137:S29-S49.Owen-Smith, N., and P. Novellie. 1981. What should a clever ungulate eat?Am. Nat. 119:151-178.Parker, K. L. 1988. Effects of heat, cold, and rain on coastal black-tailed deer. Can. J. Zool. 65:20-23.C. T. Robbins, and T. A. Hanley. 1984. Energy expenditures forlocomotion by mule deer and elk. J. Wildi. Manage. 48:474-488.Pierce, G. J., and J. G. Ollason. 1987. Eight reasons why optimalforaging theory is a complete waste of time. Oilos 49:111-125.Powell, T. M. 1989. Physical and biological scales of variability inlakes, estuaries, and the coastal ocean. Pages 157-176 in J.Roughgarden, R. M. May, and S. A. Levin, eds. Perspectives inecological theory. Princeton Univ. Press, Princeton, N.J.Pulliam, H. R. 1989. Individual behaviour and the procurement ofessential resources. Pages 25-38 in J. Roughgarden, R. M. May, and S.A. Levin, eds. Perspectives in ecological theory. Princeton Univ.Press, Princeton, N.J.and B. J. Danielson. 1991. Sources, sinks, and habitat selection:a landscape perspective on population dynamics. Am. Nat. 137:S50-S66.Pyke, G. H. 1984. Animal movements: an optimal foraging approach. Pages7-31 in I. R. Swingland, and P. J. Greenwood, eds. The ecology ofanimal movement. Oxford Univ. Press, N.Y.Real, L., and T. Caraco. 1986. Risk and foraging in stochasticenvironments. Ann. Rev. Ecol. Syst. 17:371-390.Reynolds, T. D., and J. W. Laundre. 1990. Time intervals for estimatingpronghorn and coyote home ranges and daily movements. J. Wildi.Manage. 54:316-322.Robinette, W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Noteson tooth development and wear for Rocky Mountain mule deer. J. Wildi.Manage. 21:134-153.Rochelle, J. A. 1980. The role of mature conifer forests in the winter73nutrition of black-tailed deer. Ph.D. Thesis, Univ. of BritishColumbia, Vancouver. 295pp.Rosenzweig, M. L. 1992. Habitat selections and population interactions:the search for mechanism. Am. Nat. 137:S5-S28.Saiwasser, H., and S. A. Holl. 1979. Estimating fetus age and breedingand fawning period in the North Kings River deer herd. Calif. Fishand Game 65:159-165.Samuel, M. D., and E. 0. Garton. 1987. Incorporating activity time inharmonic home range analysis. J. Wildl. Manage. 51:254-257.Sanderson, G. C. 1966. The study of mammal movements - A review. J.Wildl. Manage. 30:215-235.SAS Inst. Inc. 1985. SAS user’s guide: basics, version 5 edition. SASInstitute Inc., Cary, N.C. 584pp.Schmidt, R. L. 1970. A history of pre-settlement fires on Vancouver Islandas determined from Douglas-fir ages. Pages 107-108 in Tree-ringanalysis with special reference to northwest America. Proceedings ofa conference February 19-20, 1970. Univ. of British Columbia, Fac.For. Bull. 7. Vancouver.Schmutz, J. A., and G. C. White. 1990. Error in telemetry studies:Effects of animal movement on triangulation. J. Wildl. Manage.54: 506-510.Schoen, J. W., and M. D. Kirchhoff. 1985. Seasonal distribution and home-range patterns of Sitka black-tailed deer on Admiralty Island,southeast Alaska. J. Wildl. Manage. 49:96-103.Schoener, T. W. 1971. Theory of feeding strategies. Ann. Rev. Ecol. Syst.2:369-404._____1987. A brief history of optimal foraging ecology. Pages 5-68 inA. C. Kamil, J. R. Krebs, and H. R. Pulliam, eds. Foraging behaviour.Plenum Press, N.Y.Senft, R. L., M. B. Coughenour, D. W. Bailey, L. R. Rittenhouse, 0. E. Sala,and D. M. Swift. 1987. Large herbivore foraging and ecologicalhierarchies. BioScience 37:789-799.Sinclair, A. R. E. 1984. The function of distance movements invertebrates. Pages 240-258 in I. R. Swingland and P. J. Greenwood,eds. The ecology of animal movement. Oxford Univ. Press, N.Y.Staddon, J. E. R. 1983. Adaptive behaviour and learning. Cambridge Univ.Press, Cambridge, England.Stephens, D. W., and J. R. Krebs. 1986. Foraging theory. Princeton Univ.Press, Princeton, N.J. 247pp.74Sweanor, P. Y., and F. Sandegren. 1988. Migratory behaviour of relatedmoose. Holarc. Ecol. 11:190-193.Swihart, R. K., and N. A. Slade. 1985. Testing for independence ofobservations in animal movements. Ecology 66:1176-1184.Swingland, I. R. 1984. Intraspecific differences in movement. Pages 102-115 in I. R. Swingland, and P. J. Greenwood, eds. The ecology ofanimal movement. Oxford Univ. Press, N.Y.Terrill, S. B., and R. D. Ohmart. 1984. Facultative extension of fallmigration by yellow-rumped warbiers (Dendroica coronata). Auk101:427-438.Thomas, D. C. 1970. The ovary, reproduction, and productivity of femaleColumbian black-tailed deer. Ph.D. Thesis, Univ. of British Columbia,Vancouver. 2llpp._____and P. J. Bandy. 1973. Age determination of wild black-tailed deerfrom dental annulations. J. Wildl. Manage. 37:232-235.Van Home, B., T. A. Hanley, R. G. Cates, J. D. McKendrick, and J. D.Homer. 1988. Influence of seral stage and season on leaf chemistryof southeastern Alaska deer forage. Can. J. For. Res. 18:90-99.Westoby, M. 1974. The analysis of diet selection by large generalistherbivores. Am. Nat. 108:290-304.White, C. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, San Diego, Calif. 383pp.Wiens, J. A. 1976. Population responses to patchy environments. Ann.Rev. Ecol. Syst. 7:81-120.Worton, B. J. 1989. Kernal methods for estimating the utilizationdistribution in home range studies. Ecology 70:164-168.Zach, R., and J. N. M. Smith. 1981. Optimal foraging in wild birds? Pages95-113 in A. C. Kamil and T. D. Sargent, eds. Foraging behaviour.Garland STPM Press, N.Y.Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, EnglewoodCliffs, N.J. 718pp.75CHAPTER 4 - SPATIAL AND TEMPORAL SCALES OF RESOURCE USE: HABITAT USEAfter seminal studies by Cowan (1945) and Einarsen (1946), forestsexceeding 250 yr-of-age were considered relatively poor habitat for black-tailed deer. Abundant forage and elevated deer use found in comparativelyyoung forests appeared to support this consideration (Brown 1961, Gates1968, Smith 1968). However, research conducted through more severe winters(Willms 1971, Smith 1973, Jones 1975, Bloom 1978, Harestad 1979, Rose 1982,Schoen et al. 1985) demonstrated the importance of older forests. Oldforests provide shelter from snow (Kirchhoff and Schoen 1987, McNay et al.1988) which reduces costs of locomotion (Parker et al. 1984, Bunnell et al.1990a,b) and rates of food burial (Jones 1975, Harestad 1979). Some oldforests also provide abundant arboreal forage seldom found elsewhere(Stevenson 1978).We may assume from the notions above that, if winter habitats werechosen optimally, black-tailed deer would collectively spend most time inold forests (Schamberger and O’Neil 1986), especially in years of heavysnowfall. Optimality, in this context, is largely a theoretical constructof individual behaviour (reviewed by Schoener 1987) but can be integratedover populations of deer within another theory, the ideal-free distributionof animals (Fretwell and Lucas 1969, Fretwell 1972). Ideal-free refers tothe main assumption in this model of habitat choice, that animals are freeto choose habitats conferring greatest individual fitness.Little has been done, however, to test the notions of optimal habitatchoices or ideal-free distributions in populations of black-tailed deer.Recent studies of habitat use (Jones 1975, Harestad 1985, Schoen andKirchhoff 1990), for example, offer conclusions limited to the quality ofold forests because data came from areas with only recent, little, or no76forest harvesting. Although Yeo and Peek (1992) studied a wider range offorest ages, and thus habitat choices, they were limited to low elevationsand to non-migratory deer. These limitations are not trivial because, atminimum, animals must be sampled randomly, and all habitats must beavailable, before habitat use data can be used for inference about habitatquality or the optimality of habitat choices.Although it is tempting to consider habitat use as an index of habitatquality, we note that Hobbs and Hanley (1990) challenged Fagen’s (1988)direct association between habitat preference and habitat quality. Withsimulation modelling, they demonstrated that such an association assumes (1)the ideal-free distribution and (2) a long-term stable relationship betweendeer populations and the quality/quantity of their habitats. Fagen (1988)assumed only the former criterion while Hanley and Hobbs (1990) claimed thelatter is rarely satisfied in nature. We consider these criteria to beclosely related. If, for example, deer respond slowly to rapid changes inhabitat, not only would instability exist in the population/habitatassociation but the ideal-free distribution would be unlikely to hold.Habitat selection by individual deer would indicate past, rather thanpresent, habitat conditions (Van Home 1983, Hobbs and Hanley 1990) makinginference about optimal habitat choices unreliable. Long lags in responseto change would make these inferences unjustifiable (Hobbs and Hanley 1990).We considered decisions that guide movements made by black-tailed deerto be constrained in 2 major ways, making their response to environmentalchange difficult (chapter 3). First, because the decision process appearsto be hierarchical, superclass decisions could limit the breadth of choicesfor subclass decisions therefore producing a scalar constraining effectthroughout the decision-making framework (Dawkins 1976, Allen et al. 1984,77Senft et al. 1987, Levin 1992). Second, at each scale in the hierarchy, wefound evidence of specific tactics (i.e., philopatry and site fidelity) thatreinforced this scalar constraint on movement decisions (Gass 1985, Fahrigand Paloheimo 1988). These constraining effects led to our supposition thatblack-tailed deer are unlikely to be distributed among habitats in an ideal-free manner and hence individual habitat preferences should range widely asindicators of optimality. This variance of habitat preference should beexplained in part by study area and/or by behaviourial effects which wouldotherwise not be revealed from data pooled over these factors (Thomas andTaylor 1990). Here we have 2 broad objectives: (1) to examine habitatpreferences in context of the movement hierarchy to determine if there aregeneral scalar constraints on habitat preferences and (2) to evaluate ifknowledge of those scalar constraints confers greater insight or betterinterpretation of (i) an ideal-free distribution of black-tailed deer, (ii)apparent individual preferences for habitats, and (iii) habitat managementinitiatives for coastal British Columbia.These objectives are particularly important because, in BritishColumbia and southeast Alaska, old forests are ardently sought by the forestindustry and by those that manage black-tailed deer (Schoen et al. 1981,Bunnell 1985). Although management options have been proposed to resolvethis conflict in British Columbia (Bunnell 1985, Nyberg et al. 1986), thespecific balance between the management options remains unclear anduntested. Our objectives will help clarify the relative quality of youngand old forests as deer habitat as a first step toward specifying theapplication of habitat management options.78STUDY AREASWe studied habitat selection by black-tailed deer at 4 locations onVancouver Island, British Columbia from February 1982 through June 1991(Table 4.1). Specific physiographic parameters of study areas are providedin chapter 3 (Table 3.1). Nanaimo and Chemainus rivers are in neighbouringvalleys situated 43 km northeast of Caycuse River and 202 km southeast ofNimpkish River. The Chemainus, Nanaimo, and Nimpkish rivers arecharacterized by open, relatively flat-bottomed valleys (U-shaped) whileCaycuse River ranges less in elevation but has steeper slopes and the leastflat area at lower elevations (V-shaped).Extensive logging occurred throughout all study areas resulting in arange of habitats from recently clear-cut sites to old forests. The spatialarrangement of habitats was characteristic of historic, coastal loggingpatterns. Initial harvests came from the bottom and downstream end ofvalleys with subsequent harvests coming from the mid-slopes, and last fromthe headwaters and higher elevations. Our study areas were all in a latestage of harvest leaving most of the valley bottom in young, 6- to 45-yr-oldforests and the mid-slopes deforested (0- to 5-yr-old clear-cuts) or inremnant patches of old forests.The dominant “zonal ecosystems” (Meidinger and Pojar 1991) onVancouver Island are the Coastal Western Hemlock (CWH) zone at lowerelevations and the Mountain Hemlock (MH) zone at higher elevations. Westernhemlock (Tsuga heterophylla) is the most common tree species in the CWH,especially in old forests or in young forests at high elevations or coolaspects. Western red cedar (Thuja plicata) and Douglas fir (Pseudotsugamenziesii) are widespread and young forests are dominated by Douglas fir.Western hemlock is also the dominant tree species in the MH zone but79Table 4.1. Total sample sizes for deer and habitat samples recorded at 4study areas located on Vancouver Island, British Columbia, 1982-1991.Study AreasSamples Caycuse Chemainus Nanaimo Nimpkishdeer 13 8 40 11deer locations 1,582 784 5,261 860deer home ranges (ha) 1,877 1,117 11,906 1,552study areas (km2) 111 33 145 41years studied 89-91 89-91 82-91 89-9180amabilis fir (Abies ainabilis) and yellow-cedar (Chamaecyparis nootkatensis)are common. Climate in both zones is temperate and wet. In the CWH zone,there is no month with a mean temperature <0 C and the mean temperature ofthe warmest month is 17 C (Meidinger and Pojar 1991). On average there are291 frost free days and 82 cm of snow each year. The driest month averages65 mm of precipitation and the mean annual precipitation is 2,140 mm.METHODSDeer Location Samples and Habitat UseMethods for capturing deer, attaching radio-collars, and monitoringdeer locations are described in detail in chapter 3. Weekly samples forcollared-deer locations began in February 1982 at Nanaimo River and inFebruary, or March, 1989 at other study areas and continued until death ofthe deer or June 1991. In 1984, sampling was standardized so that, during acalendar month, each deer was located at least once-per-week and once withineach quarter of a calendar day.Triangulation data (White and Garrott 1990) were collected frompermanent sampling stations at 100-rn intervals along forest roads. In thefield, if at least 3 bearings intersected at 1 general site then bearinginformation was recorded along with signal frequency, time of day, and date.Because each study area had an extensive network of roads, we were able tocollect data quickly and in close proximity to collared deer. As a result,error polygons (Lenth 1981) were usually <1.0 ha (McNay et al. 1994). Homeranges were estimated as the 95% minimum convex polygon (White and Garrott1990:343).Habitat use and availability was estimated by querying forest coverand topographic maps, stored on Geographic Information Systems (Terrasoft;81Digital Resources, Nanaimo, B.C. and PAMAP; PAMAP Technologies Corp.,Victoria, B.C.), at specific Universal Transverse Mercator grid co-ordinatesrepresenting polygons (bounded by study area corners or by home rangevertices) and points (individual deer locations).Snow depth was recorded daily at airports near each study area andwere supplied to us by the Atmospheric Environment Service (EnvironmentCanada, Vancouver, B.C.).Definitions and Habitat FeaturesWe defined natal ranges as areas occupied during the natal period,which for deer in coastal British Columbia is usually late-May through June(Cowan 1956, Thomas 1970). We determined that annual migrations occurredjust prior to the natal period (chapter 3) so we assumed that the natalrange included, or was adjacent to, the birth-site for non-dispersing deer(Masters and Sage 1985, McCullough 1985, Hamlin and Mackie 1989) and wouldbe where subsequent offspring were produced each year. Spatially separateranges occupied at other times were identified by migrations made betweenthose areas and the natal range, and were termed alternate ranges.Migrations differed from dispersal or nomadic movements in that, wheneverdeer migrated, they would make return migrations to the original location(Sinclair 1984). Dispersal involved no such predictable return to theoriginal location (chapter 3). Deer behaviour types were determined on thebasis of seasonal movements where migratory deer made migrations andresident deer did not. We defined activity nuclei (chapter 3) after Don andRennolls (1983) as isolated patches of relatively concentrated andrepetitive use within individual home ranges. Activity nuclei had 200-mradii, the centres of which were the highest point(s) chosen from a82graphical representation of the harmonic utilization distribution (Dixon andChapman 1980, chapter 3).Habitat types were characterised by 3 different vectors: (1) forestswere old if >250-yr-old, young if 6- to 45-yr-old, or open if either 0- to5-yr-old, non-commercial forest (subalpine or alpine), or non-forest (rockor water); (2) elevations asl were <400 m, 401-600 m, 601-800 m, or >800 m;and (3) aspects were north if 316° to 45°, east if 46° to 135°, south if136° to 270°, west if 271° to 315°, or flat. Chosen in this way, habitatpolygons typically exceeded 60 ha (e.g., statistics for the forest vector atNanaimo River were: = 65.5; n = 199, SE = 0.08) or >60 times the size ofmost error polygons surrounding estimated locations for deer. Summer wasdesignated May through October and winter designated November through April.Annual periods were from May 1 through April 30 of the following calendaryear.Terms used to describe habitat selection vary and have been usedindiscriminately (Thomas and Taylor 1990). We followed Johnson’s (1980)recommended interpretations. Specifically, abundance was the quantity ofhabitats in the environment (study area), availability was the habitats’accessibility to deer (that within individual home ranges), use was thequantity actually visited within a time period (seasonal locations forindividual deer), selection was the process of choosing to visit habitats,and preference reflected the likelihood of choosing a particular habitat ifall habitats were offered equally. Our study was based on multiple spatialscales of habitat use: (1) preference for home ranges within broad studyareas, (2) preference for seasonal habitats within home ranges, and (3) useof “activity nuclei” (after Don and Rennolls 1983) within seasonal habitatsby migratory and by resident deer both within and among study areas.83Preference for home ranges was assessed on the basis of the quantity ofstudy area habitats while preference for seasonal habitats was related toaccessibility of habitats to individual deer within their home ranges.Although study area boundaries were subjectively delineated, weconsidered our study areas large enough to negate the effects of habitatpattern on estimates of abundance (Porter and Church 1987, Verbyla and Chang1994). To corroborate this, we estimated relative diversity (Shannon 1948)for the forest vector at Nanaimo River beginning with many 1-ha study areasthen increasing study area sizes gradually to 11,000 ha, close to actualstudy area size (14,500 ha). Habitat diversity for such a study area was78% of the potential diversity obtainable if all forest types had beenequally abundant. The index began to oscillate and plateau, however, at ahypothetical study area size of 1,600 ha. We concluded that our study areaswere large enough to capture most of the forest type diversity at NanaimoRiver and assumed this to be the case for other habitat vectors there and atother study areas.Analytical ProceduresWe assessed data distributions using PROC UNIVARIATE (SAS Inst. Inc.1985) and assessed independence of observations in deer movements (Swihartand Slade 1985) in a related study (McNay et al. 1994). Based on thoseanalyses we assumed that locations taken once-per-week were independent,systematic samples of use of space. Chi-square tests (Sokal and Rohlf 1981)were used to assess homogeneity of habitat types among study areas.We attempted to counter unequal availability of habitat types (Johnson1980, Thomas and Taylor 1990) using Manley’s measure of preference, a,(Manley 1974, Chesson 1978, 1983), which we interpreted as the proportion of84habitat use that would be of type i if all habitats were equally abundant oravailable. The calculation is:Iin.1 1=1 (4.1)where r1 is the number of times habitat type I is selected, n1 is therelative abundance or availability of habitat I, and m is the number ofhabitats considered. Furthermore, we kept m low so at least some of eachtype would be available to all deer (forests m = 3, elevations m = 4, andaspects m = 5), and to limit the potential for experimentwise error (Thomasand Taylor 1990). Since a values must sum to 1.0 (Chesson 1983), random useof seral age habitats occurred if = a2 = a3 = 0.33. Similarly, random useof aspect and elevation habitats occurred if a = 0.20 and 0.25,respectively. We evaluated variation in habitat selection tactics amongindividual animals (Thomas and Taylor 1990) by testing homogeneity ofpreferences within groups of deer (Manley 1974):S()2(4.5)x23’ , j1,.. .VAR(c1)where s is the number of deer within the group and other symbols are as85above.We tested for potential effects of study area and a deer behaviourtype x range type variable (migratory natal ranges, migratory alternateranges, or resident ranges), and the interaction of those effects, onpreference for home ranges within study areas using multivariate analysis ofvariance (MANOVA, SAS Inst. Inc. 1985). Separate MANOVAs were used for eachhabitat vector because our sample sizes were insufficient to consider theinteractions of habitat vectors. MANOVAs for elevation and aspect vectorsdid not include data for Nimpkish River because they were unavailable. Weused the same approach to test for similar effects (study area and deerbehaviour as either migratory or resident) on preferences for seasonalhabitats within individual home ranges. Because habitat vectors werelinearly dependent (Manley’s a values sum to 1.0) we obtained overall Festimates (Wilks’ A) by omitting 1 a value producing a new vector length ofin -1. Parameters for the omitted a were determined in a subsequentunivariate analysis of variance (ANOVA, SAS Inst. Inc. 1985). When effectswere significant (P < 0.05), differences between adjacent means (e.g., amonghabitats within a single vector) were compared using Duncan’s multiple ranget-tests with P = 0.10. All means are least-squares estimates (Searle et al.1980). Paired-sample tests (Zar 1984) were used to test for significantdifferences in habitat preferences between range types for migratory deerand between seasons for all deer.We tested for the effect of year (Schooley 1994), and it’s interactionwith deer behaviour type, on seasonal habitat preferences at Nanaimo Riverwhere we had the longest data set (9 yr; Table 4.1). Within a singlevector, individual dependent variables with significant annual variationwere correlated with total annual snowfall (Nanaimo airport), proportion of86annual deer samples that were migratory deer, total number of deer sampledannually, and year of study using Spearman’s Rank correlation (SAS Inst.Inc. 1985).Finally, the potential effects of study area and deer behaviour typeon use of activity nuclei was assessed using logistic regression (CATMOD,SAS Inst. Inc. 1985). We obtained maximum likelihood estimates for theprobability of specific habitats being the primary (i.e., the highestpercentage) component of individual nuclei.RESULTSSample CharacteristicsWe collared most deer (56%) and collected most samples of deerlocations (62%) at Nanaimo River where our study lasted longest (Table 4.1).Our sample was dominated by resident deer (n = 44 or 61%), especially atChemainus and Nimpkish rivers where only 2 deer at each site were migratory(chapter 3).Abundance of habitats varied significantly among study areas forforests (x2 = 1016.75; df = 6; P < 0.001), for elevations (x2 = 2994.44; df =6; P < 0.001), and for aspects (x2 = 1863.26; df = 8; P < 0.001). Althoughyoung forests dominated each study area (Fig. 4.1), notable deviationsoccurred primarily at Chemainus River where more young, and less old, forestexisted than would be expected under the assumption of homogeneity. Otherdeviations from homogeneity were the lack of open forest and the abundanceof old forest at Nimpkish River and the abundance of open forest at CaycuseRiver (Fig. 4.1). Caycuse River had less area >800 m and more area <400 mthan other study areas (Fig. 4.2). Also, little area existed <400 m atChemainus River and, at Nanaimo River, more area existed >800 m than would87a)C.)0Seral age classesFigure 4.1 Forest seral age class abundance (total within study area),availability (total within home ranges), and use (locations from radio-collared, black-tailed deer) at 4 study areas on Vancouver Island, BritishColumbia, 1982-1991.Caycuse River 80$todyo,,eHome reogosLoretion, 60Chemainus River5040300’00Open Young20100 INanaimo River Nimpkish River80706050403020iciliOpen YoungOld88Chemainus RiverElevation classes (m asi)Figure 4.2 Elevation class abundance (total within study area),availability (total within home ranges), and use (locations from radio-collared, black-tailed deer) at 3 study areas on Vancouver Island, BritishColumbia, 1982-1991.Caycuse River$hdy.rea505040302010040302010Cu000C.)000>800<400401-60050403020100<400 601 -800401 -600 >800Nanaimo River[1<400 601 -800401 -600 >80089be expected under the assumption of homogeneity. The most notabledeviations from homogeneity of aspects was in Caycuse River where there wasthe least flat habitat and the most habitat with a western aspect (Fig.4.3).Home Range Preferences: Life-time ScaleHomogeneity of Preferences. - Generally, deer preferred a wide range ofhabitats. We rejected homogeneity of preferences in all tests concerningforest types (n = 3 habitats x 3 range types x 4 study areas = 36 tests).In only 3 of 36 elevation tests (4 habitats x 3 range types x 3 studyareas), and in only 8 of 45 aspect tests (5 habitats x 3 range types x 3study areas), were deer considered to have a similarity of preferences (P >0.05).Analysis of Main Effects. - Deer preferences for forests varied stronglywith study area and marginally with migration tactics and the type of rangebeing used (Table 4.2). Migration tactics and the type of range stronglyeffected deer preferences for elevations while study area, and theinteraction of study area and range type, had little effect (Table 4.2).Finally, deer preferences for aspects did not vary significantly with any ofthe main effects that we assessed (Table 4.2).Deer generally preferred young forests over old and preferred openforests least (pooled home range types: Table 4.3). This lack of preferencefor open forests was most evident at Nimpkish River (a = 0.02; Table 4.3)where abundance of that type was low (Fig. 4.1). Migratory deer deviatednotably from these general trends in their preference for old forests onalternate ranges (pooled study areas: a = 0.47; Table 4.3), a consistenttrend across all study areas. Resident deer, by comparison, generally did90Cu0.4-0a)C.)I0.Caycuse River$h.dyomaHome rangesLonoee4.3 Aspect class abundance (total within study area), availabilitywithin home ranges), and use (locations from radio-collared, black-deer) at 3 study areas on Vancouver Island, British Columbia, 1982-706050403020100706050403020100Chemainus RiverFlat East West Flat East WestNorth South North South70 Nanaimo River6050403020Flat EUL WestNorth SouthAspect classesFigure(totaltailed1991.Table4.2.Multivariateanalysisofvarianceresultsfortestsofstudyarea,deerbehaviour(migrationtacticsandrangetypesforselectionofhomerangesandmigrationtacticsforselectionofseasonalhabitats),andtheinteractionofthosemaineffects,onhabitatspreference(aj)bestimatesforradio-collared,black-taileddeerat4studyareasonVancouverIsland,BritishColumbia,1982-1991.HabitatvectorsSelectionlevelSeratageElevationAspectandvariablesFdf(n,d)PFdf(n,d)EFdf(n,d)PHomeRangeStudyarea3.436,1720.0031.146,1480.3401.428,1460.192Rangetype2.314,1720.0602.806,1480.0131.348,1460.227Interaction1.0312,1720.4261.3312,1960.2020.7816,2240.286SeasonalHabitats:SurmierStudyarea3.296,3340.0041.816,3020.0973.028,3000.003Behaviour1.762,1670.1754.353,1510.0060.804,1500.527Interaction1.386,3340.2213.726,3020.0011.948,3000.055SeasonalHabitats:WinterStudyarea2.456,3900.02410.856,340<0.0013.398,3340.001Behaviour6.492,1950.0028.883,170<0.0011.474,1670.214Interaction1.616,3900.1434.436,340<0.0013.148,3340.002ccI.Table4.2.Continued.HabitatvectorsSelectionlevelSeralageElevationAspectandvariables°Fdf(n,d)PFdf(n,d)PFdf(n,d)PSeasonalHabitats:PairedsampledifferenceStudyarea1.626,2860.1410.606,2540.7271.028,2500.418Behaviour3.642,1430.0292.173,1270.0950.594,1250.672Interaction1.576,2860.1572.076,2540.0571.498,2500.160Habitatvectorsformedthedependentvariablesfor3separatemultivariateanalysisofvarianceswherehabitatsvectorswere:(1)forestseralage(open,young,orold);(2)elevation(<400m,401to600in,601to800m,or>800in);and(3)aspect(316°to45°,46°to135°,136°to270°,271’to315°,orflat).bPreferenceestimatesforeachhabitatwerecalculatedforindividualdeeraccordingtoChesson(1983)firstcoirparinghomerangehabitatswithstudyareahabitats(homerangeselectionlevel)andsecond,comparinghabitatsatdeerlocationswithhomerangehabitats(seasonalhabitatselectionlevel).Independentvariablesassessedinthemultivariateanalysisofvariancewerestudyarea(Caycuse,Chemainus,Nanaimo,andNinkishRivers)andadeerbehaviourxrangetypevariable(migratorydeernatalranges,migratorydeeralternateranges,orresidentdeernatalranges)atthehomerangelevelorstudyareaanddeerbehaviour(migratoryorresidentdeer)attheseasonalhabitatlevel.r’3Table4.3.Habitatspreference(a,)”estimatesfor28migratoryand44resident,radio-cot lared,black-taileddeerat4studyareasonVancouverIsland,BritishColuitia,1982-1991.-HabitatvectorsStudyareaandOpenYoungOldhomerangesna,SEa,SEa,SECaycuseRiverM-A70.120.070290.110.590.1OAM-N70.320.O7AB0.580.11A0.100.1OBR-N60.320.080.410.110.270.10Pooled200.250.040.430.060.320.06ChemainusRiver1.1-A10.410.190.130.280.460.26M-N20.130.140.520.200.350.18R-N60.310.080.520.110.170.10Pooled90.280.080.390.120.330.11NanaimoRiverM-A170.270.050.420.070.310.06M-N170.350.050.450.070.200.06BR-N230.390.040.410.060.200.05BPooled570.340.030.420.040.240.03TabLe4.3.Continued.HabitatvectorsStudyareaandOpenYoungOldhomerangesna,SESEa1SENimpkishRiverM-A20.00O.14B0.490.200.510.18M-N20.020.140.97O.20A0.010.18R-N90.03O.06B0.580.090.390.09Pooled130.02O.07x0.680.100.300.09PooledStudyAreasM-A270.200.060.330.090.470.08wM-N280.200.050.630.080.170.07R-N440.260.030.480.050.260.04Habitatswere:openifrock,water,subalpine,alpine,orforestedbut<6yrold;youngif6-to45-yr-old;andoldif>250-yr-old.Meanpreferenceestimatesforindividualdeerwithinstudyareasandbehaviourgroups(n)werecalculatedaccordingtoChesson(1983),comparinghomerangehabitatswithstudyareahabitats,andsumto1withinahabitatvector(row).With3habitats,a=0.33indicatesrandomusewhilegreatera’sindicatepreferenceandlessera’sindicatelackofpreference.Differentletterswithinarow(uppercase),orwithinacolumnforpooledestimates(lowercase),indicatedifferencesinmeans(Duncan’s-adjustedt-tests;E<0.10).Homeranges,NfornatalareasandAforalternateareas,wereforresidentdeer,R,andmigratorydeer,M.cO95not prefer old forests except at Nimpkish River where the percent abundanceof that particular forest type was greatest (Fig. 4.1). Deer generally didnot prefer elevations <400 m or >800 m (pooled study areas: Table 4.4). Onnatal ranges, however, migratory deer contradicted that trend by stronglypreferring elevations >600 m. Although only significant in particular cases(Table 4.5), preference for southern aspects tended to dominated all homeranges as did a general lack of preference for western aspects.Paired Difference in Preferences for Home Ranges by Migratory Deer. -Migratory deer preferred habitats differently when occupying alternateranges than when occupying natal ranges. While their preferences for openforest did not change (a = -0.09, T = -1.59; n = 27; P = 0.123), preferencefor old forests increased (a = 0.25, T = 3.29; n = 27; p = 0.003) andpreference for young forests decreased (a = -0.16, T = -2.63; n = 27; P =0.014) on alternate ranges (Table 4.3). Also while on alternate ranges,migratory deer preferred elevations <400 m more (a = 0.18, T = 2.43; n = 24;P = 0.023) and elevations >800 m less (a = -0.27, T = -3.72; n = 24; P =0.001) than on natal ranges (Table 4.4). Their preference for southernaspects (Table 4.5) on alternate ranges also increased (a = 0.24, T = 2.22;n = 24; P = 0.036) while their preference for northern aspects decreased (a= -0.16, T = -2.12; n = 24; P = 0.045).Habitat Preferences: Seasonal ScaleHomogeneity of Preferences.- We had the opportunity to assess homogeneityof preferences for forests in 208 tests (habitats x years x movement types xstudy areas) but only 78 of these had >5 deer/group. Homogeneity wasrejected in 57 of these 78 tests (P < 0.05). The cases in which deer didhave similar preferences were resident deer preferences for old forests (12Table4.4.Habitatpreference(a1)”estimatesfor26migratoryand35resident,radio-collared,black-taileddeerat3studyareasonVancouverIsland,BritishColumbia,1982-1991.-HabitatvectorsStudyareasand<400m401-600m601-800m>800mhomerangescnSEa1SEa1SEa,SECaycuseRiverM-A70350.110.350.100.260.080.040.08M-N70.000.110.060.100.440.08A0.500.08AR-N60.550.12A0.330.1OAB0.110.O9BC0.010.09CPooled200.300.070.250.060.270.050.180.05ChemainusRiverM-A20.060.210.150.180.430.150.360.15M-N20.020.210.430.180.260.150.290.15R-N60.190.120.350.100.220.090.240.09Pooled100.090.110.310.090.300.080.300.08NanaimoRiverM-A170.230.O7AB0.350.06A0.320.05A0.100.05BN-N170.110.070.280.060.300.050.310.05R-N230.160.06C0.450.05A0.310.05B0.080.04CPooled550.170.040.360.030.310.030.160.03ccTable4.4.Continued.HabitatvectorsStudyareasand<400m401-600m601-800m>800mhomerangesna,SEa,SEa,SEa,SEPooledStudyAreasM-A260.210.080.280.070.340.060.170.06M-N260.040.08x0.250.070.340.060.370.06wR-N350.300.060.380.050.210.040.110.04Habitatswereelevationsinmabovesealevel(asl)asindicated.Meanpreferenceestimatesforindividualdeerwithinstudyareasandbehaviourgroups(n)werecalculatedaccordingtoChesson(1983),comparinghomerangehabitatswithstudyareahabitats,andsunto1withinahabitatvector(row).With4habitats,a=0.25indicatesrandomusewhilegreatera’sindicatepreferenceandlessera’sindicatelackofpreference.Differentletterswithinarow(uppercase),orwithinacoluirforpooledvalues(lowercase)indicatedifferencesinmeans(Duncan’s-adjustedt-tests;P<0.10).Homeranges,NfornatalareasandAforalternateareas,wereforresidentdeer,R,andmigratorydeer,M.Table4.5.Habitatapreference(a)t’estimatesfor26migratoryand35resident,radio-collared,black-taileddeerat3studyareasonVancouverIsLand,BritishCoLumbia,1982-1991.-HabitatvectorsStudyareasandNorthEastSouthWestFlathomeranges’nSEa1SEa1SEa,SEa1SECaycuseRiverM-A70.030.080.110.070.480.12A0.170.050.2109H-N70.210080.300.070.210.120.210.050.0709R-N60.270.080.120.070.500.13A0.040.050.0710Pooled200.170.050.180.040.400.070.140.030.1106ChemainusRiverM-A20240.150.010.120.460.230.040.090.2518M-N20.360.150.050.120.410.230.010.090.1718R-N60.000.080.230.070.560.13A0.010.050.2010Pooled100.200.150.090.060.480.120.020.050.2109NanaimoRiverH-A170.130.O5BC0.130.O4BC0.580.08A0.010.03C0.1506BM-N170.270.O5AB0.180.O4BC0.360.08A0.090.03C0.1006CR-N230.100.O4BC0.150.04B0.550.07A0.030.03C0.1705BPooled550.170.150.150.020.500.040.040.020.1403co cxTable4.5.Continued.HabitatvectorsStudyareasandNorthEastSouthWestFlathomerangesna,SEa,SEa,SEa,SEa,SEPooledStudyAreasM-A260.130.060.080.050.510.090.080.040.2007M-N260.280.060.180.050.330.090.100.040.1107R-N350.130.040.170.040.540.070.020.030.1405Habitatswereaspects:north(3160.450),east(4601350),south(1360.2700),west(271°-315°),orflat.Meanpreferenceestimatesforindividualdeerwithinstudyareasandbehaviourgroups(n)werecalculatedaccordingtoChesson(1983),comparinghomerangehabitatswithstudyareahabitats,andsumto1withinahabitatvector(row).With5habitats,a=0.20indicatesrandomusewhilegreatera’sindicatepreferenceandlessera’sindicatelackofpreference.Differentletterswithinarow(uppercase),orwithinacoluiriforpooledestimates(lowercase),indicatedifferencesinmeans(Duncan’s-adjustedt-tests;P<0.10).Homeranges,NfornatalareasandAforalternateareas,wereforresidentdeer,R,andmigratorydeer,M.c)100of 20 tests) and migratory deer preferences for open forests (4 of 6 tests).We had enough deer to assess homogeneity of preferences for elevationsin 92 of 224 cases (habitats x years x movement types x study areas). Only13 of these 92 tests failed to reject the hypothesis of homogeneity (P >0.05); these were mostly resident deer preferences for elevations >800 m (8of 16 tests).Similarly, we had enough deer to assess homogeneity of preferences foraspects in 115 of 280 cases (habitats x years x movement types x studyareas). In 40 of these 115 tests, we failed to reject the hypothesis ofhomogeneity (P > 0.05) and only 3 of those 40 were associated with aspectsother than north, west, or flat. Resident deer preference for north, west,or flat aspects composed most (29 of 40) cases of homogeneous preference.Summer. - Deer did not vary their summer-time preferences for habitat acrossyear-of-study for forests (F = 1.07; df = 16, 206; P = 0.390), forelevations (F = 0.82; df = 21, 293; P = 0.696), or for aspects (F = 0.67; df= 28, 366; P = 0.900). Their preferences for forests and aspects did varysignificantly with study area while their preferences for elevations variedsignificantly with migration tactics and with the interaction of thosetactics and study area (Table 4.2).Generally, deer preferred young forests and avoided old forests insummer (Table 4.6). The effect of study area was largely due to a lack ofpreference for open forest by deer at Nirnpkish River, where little openforest was available (Fig. 4.1). As a result, deer at Nimpkish showed acomparatively strong preference for young forest (Table 4.6). Deergenerally used all elevations in a more balanced fashion within their summerranges than was indicated by their preferences for elevations at the homeranges level (compare pooled study areas: Tables 4.4 and 4.7). MigrationTable4.6.Habitatspreference(a,)bestimatesfor28migratoryand44resident,radio-collared,black-taileddeerat4studyareasonVancouverIsland,BritishColitia,1982-1991.-HabitatvectorsStudyareaandOpenYoungOldseasonalrangesa1SEa,SEa,SECaycuseRiverM-S140.470.070.490.080.040.05BR-S100.460.090.440.090.100.06BPooled-S240.470.060.470.060.060.0414-U130.300.060.360.080.340.08R-W80.500.080.410.100.090.1OBPooled-U210.400.050.390.060.210.06ChemainusRiverM-S40.140.140.510.150.350.10R-S110.420.080.510.090.070.06BPooled-S150.280.080.510.090.210.06M-W60.370.090.230.120.400.11R-W120.240.070.670.08A0.090.08Pooled-U180.310.060.450.070.240.07I—icDTable4.6.Continued.HabitatvectorsStudyareaandOpenYoungOldseasonalranges’na,SEa,SEa,SENanaimoRiverM-S530.350.04B0.500.04A0.150.03CR-S680.390.030.460.040.150.03BPooled-S1210.370.030.480.030.150.02M-W570.220.03C0.460.04A0.32O.04BR-U830.320.0280.48O.03A0.20O.03CPooled-U1400.270.020.470.020.260.02NimpkishRiverN-S30.000.160.95O.17A0.050.11R-S130.190.080.69O.08A0.120.05Pooled-S160.090.09x0.820.09w0.090.06M-W50.070.1080.570.13A0.360.12A8R-W200.170.050.650.06A0.180.06Pooled-U250.120.Oóx0.610.070.270.07ITable4.6.Continued.HabitatvectorsStudyareaandOpenYoungOldseasonalranges’na1SEa1SEa,SEPooledStudyAreasM-S740.240.060.610.060.150.04R-S1020.370.040.520.040.110.03M-W810.240.040.410.050.350.05wR-W1230.310.030.550.04w0.140.04Habitatswere:openifrock,water,subalpine,alpine,orforestedbut<6-yr-old;youngif6-to45-yr-old;andoldif>250-yr-old.bMeanpreferenceestimatesforindividualdeerwithinstudyareas,behaviourgroups,andseasons(n)werecalculatedaccordingtoChesson(1983),comparinghabitatsatestimateddeerlocationswithhomerangehabitats,andstinto1withinahabitatvector(row).With3habitats,a=0.33indicatesrandomusewhilegreatera’sindicatepreferenceandlessera’sindicatelackofpreference.Differentletterswithinarow(uppercase),orwithinacolumnforpooledseasonalvalues(lowercase),indicatedifferencesinmeans(Duncan’s-adjustedt-tests;E<0.10).SeasonalrangeswereSumer(S),MaythroughOctober,andWinter(W),NovemberthroughApril,forresidentdeer,R,andmigratorydeer,ii.Table4.7.Habitat’preference(a)”estimatesfor26migratoryand35resident,radio-collared,black-taileddeerat3studyareasonVancouverIsland,BritishColumbia,1982-1991.-HabitatvectorsStudyareasand<400m401-600m601-800m>800mseasonalrangesnSESEa1SEaSECaycuseRiverM-S140.000.080.130.080.540.07A0.330.08BR-S100.530.09A0.320.1OAB0.150.O9BC0.000.09CPooled-S240.270.060.220.060.340.060.170.06M-W130.270.080.320.080.310.070.100.04R-W80.490.1OA0.340.1OAB0.170.O9BC0.000.05CPooted-W210.380.07w0.330.060.240.060.050.03ChemainusRiverM-S40.000.140.320.150.300.140.380.15R-S110.140.090.240.090.200.080.420.09Pooled-S150.070.080.280.090.250.080.400.09M-W60.000.12C0.180.110.290.100.530.06AR-W120.070.080.530.08A0.250.070.150.04Pooled-W180.030.070.360.070.270.060.340.04wI—’cDTable4.7.Continued.HabitatvectorsStudyareasand<400m401-600m601-800m>800mseasonalrangesna,SEa,SEa1SEa,SENanaimoRiverM-S520.130.04B0.280.040.300.040.290.04R-S680.150.030.340.04A0.340.03A0.170.04Pooled-S1200.140.030.310.030.320.030.230.03M-W570.170.04C0.490.04A0.260.03B0.080.020R-W820.150.03C0.440.03A0.360.03B0.050.02DPooled-W1390.160.020.470.020.310.020.070.01PooledStudyAreasM-S700.050.060.240.060.380.050.330.06R-S890.270.04w0.300.050.230.040.200.04M-W760.140.050.330.050.290.040.240.02wR-W1020.230.050.440.040.260.040.070.02HabitatswereelevationsinmabovesealeveL(asl)asindicated.MeanpreferenceestimatesforindividuaLdeerwithinstudyareas,behaviourgroups,andseasons(n)werecalculatedaccordingtoChesson(1983),comparinghabitatsatestimateddeerlocationswithhomerangehabitats,andsumto1withinahabitatvector(row).With4habitats,a=0.25indicatesrandomusewhilegreatera’sindicatepreferenceandlessera’sindicatelackofpreference.Differentletterswithinarow(uppercase),orwithinacolumnforpooledseasonalestimates(lowercase),indicatedifferencesinmeans(Duncan’s-adjustedt-tests;<0.10).SeasonalrangeswereSumer(5),MaythroughOctober,andWinter(W),NovemberthroughApriL,forresidentdeer,R,andmigratorydeer,M.I-,C),106tactics affected preferences for elevation during summer because migratorydeer generally exhibited a strong lack of preference for elevations <400 mwhile preferring elevations >600 m (pooled study areas: Table 4.7). Bycomparison, residents were less responsive to elevation except at Nanaimowhere they preferred elevations 401-600 m and 601-800 m (Table 4.7).Relative to deer preferences for aspects in entire home ranges (pooled studyareas: Table 4.5), preferences for aspects within summer ranges werecomparatively more balanced between northern, eastern, and southern aspectswith deer again showing lack of preference for western aspects and flatareas (pooled study areas: Table 4.8). Study area affected summer-timepreferences for aspects largely because deer at Chemainus River had a strongpreference for eastern aspects which were otherwise used randomly (Table4.8).Winter. - Deer varied their winter habitat preferences with year-of-studyfor all habitat features: forests (F = 2.58; df = 18, 238; P = 0.001),elevations (F = 1.86; df = 27, 342; P = 0.007), and aspects (F = 1.62; df =36, 429; P = 0.014). Study area also affected deer preferences for allhabitat features (Table 4.2) and preferences for forests and elevations alsovaried significantly with migration tactics (Table 4.2). The interactionbetween study area and migration tactics was significant in preferences forelevations and aspects but not for forests (Table 4.2).The annual variation was specifically significant in preferences foropen forests (F = 4.36; df = 9,139; P < 0.001) and old (F = 2.68; df =9,139; P = 0.007), for 601-800 m (F = 2.07; df = 9,138; P = 0.037) and >800m (F = 2.99; df = 9,138; P = 0.003) elevations, and for north aspects (F =3.06; df = 9,136; P = 0.003). We could not explain these variations on thebasis of total annual snowfall (all P > 0.354) or on the proportion of deerTable4.8.Habitatspreference(a,)’estimatesfor26migratoryand35resident,radio-collared,black-taileddeerat3studyareasonVancouverIsland,BritishColuiia,1982-1991.-HabitatvectorsStudyareasandNorthEastSouthWestFlatseasonalrangescna1SEa,SEa,SEa1SEa1SECaycuseRiverM-S140.180.070.260.070.210.090.280.060.0706R-S100.410.08A0.080.080.440.1OA0.070.070.0008Pooled-S240.290.050.170.050.330.070.180.050.0305M-W130.180.O5BC0.110.O7CD0.410.1OA0.290.O5AB0.00060R-W80.200.06B0.100.O9BC0.700.12A0.000.06C0.0007CPooled-U210.190.04w0.100.050.560.080.150.04w0.0004ChemainusRiverN-S40.230.130.450.120.200.170.000.110.1312R-S110.070.08C0.570.07A0.220.1OB0.000.O7BC0.1407BPooled-S150.150.080.510.07w0.210.100.000.060.1307N-U60.050.070.180.100.660.14A0.000.070.1108R-W120.000.050.300.07A0.420.1OA0.000.050.2806APooled-U180.02O.Ol,y0.240.060.540.090.000.040.2005wTable4.8.Continued.HabitatvectorsStudyareasandNorthEastSouthWestFlatseasonalrangesna,SEa,SEa1SEa1SEa,SENanaimoRiverM-S520.270.040.230.030.290.050.080.03B0.13038R-S680.180.03B0.190.03B0.430.04A0.100.030.0903Pooled-S1200.230.020.210.020.360.030.090.020.1102M-W550.120.020.190.030.550.05A0.030.02C0.1203R-W820.090.020.160.0380.620.04A0.070.020.0602Pooled-W1370.100.Olx0.180.020.580.030.050.010.0902PooledStudyAreasN-S700.220.050.310.050.240.060.120.040.1105R-S890.220.040.280.040.370.050.050.030.0804M-W740.120.030.160.040.540.060.100.030.0803R-W1020.100.030.180.040.580.050.020.030.1203Habitatswereaspects:north(316’-45’),east(46°-135’),south(136-270°),west(271’-315),orflat.Meanpreferenceestimatesforindividualdeerwithinstudyareas,behaviourgroups,andseasons(n)werecalculatedaccordingtoChesson(1983),comparinghabitatsatestimateddeerlocationswithhomerangehabitats,andsumto1withinahabitatvector(row).With5habitats,a=0.20indicatesrandomusewhilegreaterasindicatepreferenceandlessera’sindicatelackofpreference.DifferentLetterswithinarow(uppercase),orwithinacolumnforpooledseasonalestimates(lowercase),indicatedifferencesinmeans(Duncan’s-adjustedt-tests;E<0.10).SeasonalrangeswereSumer(S),MaythroughOctober,andWinter(W),NovemberthroughApril,forresidentdeer,R,andmigratorydeer,M.109sampled annually that were migratory (all P > 0.125). Average preferencefor old forest at Nanaimo River varied with year-of-study (Spearman’s Rank R= 0.96; n = 10; P < 0.001) as did preference for open forest (Spearman’sRank R = -0.90; n = 10; P < 0.001) and 601-800 m elevations (Spearman’s RankR = -0.90; n = 10; p < 0.001). The latter 2 were themselves stronglycorrelated (P’s < 0.002) with preferences for old forest indicating probablecovari ance.Generally, deer preferred young forests during winter (pooled studyareas: Table 4.6), although migratory deer also preferred old forests.Study area was important in preference for forest types again primarilybecause of deer lacked preference for open forests at Nimpkish River (Table4.6). In winter, migratory deer maintained their preference for elevations>800 m while resident deer avoided those areas (pooled study areas: Table4.7). Deer at Chemainus, however, generally preferred higher elevationsmore than deer at other study areas (Table 4.7) thereby contributing to asignificant study area effect. Also contributing to the study area effect,deer at Caycuse River preferred elevations <400 m more than deer at otherstudy areas. An interaction effect (study area x migration tactics) likelyoccurred because migratory deer at Nanaimo River preferred the 401-600 melevation more than resident deer. Deer at Chemainus River preferred flathabitats and lacked preference for northern aspects more than elsewhere(Table 4.8). Deer at Caycuse preferred western and northern aspects morethan elsewhere. An interaction between study area and migration tacticsoccurred for preference of aspects during winter largely because migratorydeer at Caycuse maintained seasonal preference for western aspects (Table4.8).Paired Differences in Preferences for Seasonal Habitats. - Paired sample110tests of difference between seasonal habitat preferences indicated that deersignificantly increased their preferences for open forest (a = 0.07, T =3.17; n = 152; P = 0.002), and decreased preference for old forest (a = -0.10, T = -4.68; n 152; P < 0.001), in summer. These differences variedsignificantly with migration tactics (Table 4.2) primarily because theaverage decrease in preference for old forest was 0.05 (SE = 0.02; n = 90)for resident deer and 0.17 (SE = 0.04; n = 62) for migratory deer.Differences in preferences for other habitats occurred more similarlyfor all deer because we found no effects due to study area or to migrationtactics (Table 4.2). Deer decreased preferences for 401-600 m habitats (a =-0.15, T = -5.98; n = 135; P < 0.001) and increased preferences for >800 mhabitats (a = 0.17, T = 7.26; n = 152; P < 0.001), during summer. Duringsummer, deer increased preferences for eastern (a = 0.06, T = 2.36; n = 134;P < 0.020), northern (a = 0.12, T = 5.45; n = 134; P < 0.001), and western(a = 0.03, T = 2.08; n = 134; P < 0.039) aspects and decreased preferencesfor southern aspects (cx = -0.22, T = -7.10; n = 134; P < 0.001).Activity Nuclei Preferences: Daily ScaleGenerally, the frequency of activity nuclei that were characterized byspecific habitat features varied (P = 0.337) with study area and migrationtactics for forests but not for elevations or aspects (Table 4.9). Exceptat Chemainus and Nimpkish Rivers, where sample sizes were relatively low,activity nuclei used by migratory deer were consistently more likely to bein old forests and less likely to be in open forests than they were forresident deer (Fig. 4.4). Other characteristics of activity nuclei weremore similar between the behaviour types and among study areas except for111Table 4.9. Maximum likelihood analysis of variance tables for frequency of habitats features forming theprimary component of activity nucleib established by individual deer. Nuclei were determined from deerlocations estimated weekly at 4 study areas on Vancouver Island, British Columbia, 1982-1991, and werepooled into 2 groups based on deer behaviour (migratory or resident).Habitat vectorand variance source df X2Forest HabitatIntercept 2 44.18 <0.001Study area 6 17.91 0.007Behaviour 2 10.24 0.006Likelihood Ratio 6 6.82 0.337Elevation HabitatIntercept 3 12.17 0.007Study area 9 32.30 <0.001Behaviour 3 24.06 <0.001Likelihood Ratio 9 37.20 <0.001Aspect HabitatIntercept 3 69.99 <0.001Study area 6 6.02 0.420Behaviour 3 11.57 0.009Likelihood Ratio 6 14.98 0.020Habitats for the forest vector were: open if rock, water, subalpine, alpine, or forested but<6- yr—old; young if 6- to 45-yr-old; and old if >250-yr-old. Habitats for the elevation vector were in masi: <400, 401—600, 601-800, or ‘800. Habitats for the aspect vector were: north (316—45), east (46—135’), south (136’—270’), or west (271’—315’) and flat.Activity nuclei had 200 m radii, the centres of which were established by peaks in the harmonicutilization distribution for home ranges (Dixon and Chapman 1980, chapter 3).1121 Caycuse River 1 Chemainus River12 obn,d 15MiSrto,y•.SEprdiddUO 12ovod____Rido4 SE1106 0660.4 ...1 0.4i o:IIjo:j1__Open Young Old Open Young Old1 Nanaimo River 1 Nimpkish River0 130t 0.8 0.8o 510. 48o06 060418 0402 02 — IOpen Young Old Open Young OldSeral age classesFigure 4.4 Observed frequencies and predicted proportional frequencies(maximum likelihood estimates) of forest habitats (clear, young, or old)forming the primary component of activity nuclei (chapter 3) for individualmigratory, or resident, black-tailed deer at 4 study areas on VancouverIsland, British Columbia, 1982-1991.113the use of open forests by deer at Caycuse River (Fig. 4.4).DISCUSSIONWe drew 4 conclusions from patterns in variance of habitatpreferences. First, home range establishment appears to resolve generalhabitat needs relative to overall habitat abundance. Second, habitatselection at subclass levels of use (e.g., within home ranges) isconstrained by superclass-level decisions (e.g., home range establishment).Third, because of this constraint, its reinforcement through specificbehaviourial tactics (chapter 3), and the rapid pace of recent habitatalterations, deer cannot be considered to have an ideal-free distribution(sensu Fretwell 1972); mean preference for habitat is unlikely to equate tohabitat value (Hobbs and Hanley 1990). Fourth, consistent changes inseasonal habitat preferences indicated old forests, at mid-elevations, onsouthern aspects as relatively optimal winter habitat for individual deer.Hierarchy and Constraint of Habitat ChoicesExcept for the lack of open forest at Nimpkish River, other uniquefeatures of study areas had little effect on what habitats deer includedwithin their home ranges (Table 4.2). Alternatively, home rangecompositions varied more consistently with migration tactics and range typeand, as a result, deer preferences indicated at least 2 different tacticsfor establishing home ranges. Migratory deer occupied natal ranges athigher elevations than did resident deer (Table 4.4). Conversely, alternateranges of migratory deer and natal ranges of resident deer were similar inelevation and aspect but differed in forest characteristics; migratory deerhad a strong preference for old forests (Table 4.3). We note from other114studies that preference for aspects was usually less variable within, thanamong, different regions (Kucera and McCarthy 1988, Garrott et al. 1987,Schoen and Kirchhoff 1990) revealing a tendency toward warm, dry aspects asregulated by regional climates. The general trend for elevational migrationis more common (Harestad 1979, Garrott et al. 1987, Schoen and Kirchhoff1990) and resident deer usually live at relatively low elevations (Kufeld etal. 1989, Schoen and Kirchhoff 1990, Brown 1992, Yeo and Peek 1992). Adivergence in preference for forest types among groups of deer is notcommonly reported, although in southeast Alaska, Schoen and Kirchhoff (1990)demonstrated a disproportionately greater use of older forests than itsabundance by a sample of mostly migratory deer and Yeo and Peek (1992), alsoin southeast Alaska, documented a preference for clear-cuts and youngforests by resident deer. The latter study only considered preference forseasonal habitats within home ranges.At the level of seasonal habitat selection, these effects due to thetactics of migration or residency were weak to nonsignificant and effectsdue to study area were common (Table 4.2). We interpreted this to mean thatalthough tactics for selecting home ranges within study areas were evident,more specific use of habitats was constrained by availability (i.e., theresources chosen when establishing home ranges). Deer did not prefer oldforests within home ranges in either season (Table 4.6). Deer preferredelevations similarly within home ranges as within study areas although thissimilarity became obscured by interactions between study area and migrationtactics, particularly at mid-elevations (Table 4.7). The basic tactic ofpreference for southern aspects persisted at both the study area and homerange levels of analysis. Schoen and Kirchhoff (1990) did not reportcomparisons of use within study areas and Harestad (1985) provided only115pooled estimates of seasonal use so we could not compare our results tothose studies. Our results depart from those of leo and Peak (1992) only byour observations of migratory deer increasing their preference for oldforests in winter. Although we are not certain, it appeared that Yeo andPeek (1992) had few if any migratory deer in their sample.Where 21% of the activity nuclei used by migratory deer were composedof old forests, the same forest type occurred in only 8% of the activitynuclei used by resident deer (Fig. 4.4). Considering the combination ofstrong site fidelity exhibited by most deer (Schoen and Kirchhoff 1985,Garrott et a!. 1987, Brown 1992), which we interpreted as philopatry(chapter 3), the regulation of seasonal movements by conditions at natalranges, and the usual forest harvesting pattern in mountainous terrain(lower elevations first), it is not surprising that residents lackpreference for activity nuclei in old forests; most resident deer in ourstudy had little to none of that forest type available (Figs. 4.1 and 4.4)and so had little chance to display such preference.We concur with Garrott et al. (1987) who proposed that seasonalmovements of mule deer are driven by seasonal changes in energy needs andthe quality and quantity of available forage. This proposal is consistentwith hierarchical habitat selection in that superclass decisions are basedon an integration of subclass activities (Dawkins 1974, Senft et a!. 1987,O’Neill et a!. 1988, Levin 1992). The integration of day-by-day resourceuse by deer born at high elevations, for example, ultimately leads to theneed for selecting winter habitat elsewhere since topographic and climaticconditions limit forage quantity and quality in winter. Deer born at lowelevations do not need to migrate. Similarly, the integration of seasonalconditions determines the need, or lack thereof, to disperse and to116establish entirely new home ranges. Although the seasonal nutritionhypothesis is a straight-forward basis for migration decisions, it is lessstraight-forward what forms the basis for dispersal (Robinette 1966, Hawkinsand Klimstra 1970, Kammermeyer and Marchinton 1976, Bunnell and Harestad1983, Masters and Sage 1985). For ecological hierarchies, the bottom-upsynthetic perspective involves functional mechanisms basic to habitatselection (e.g., forage acquisition). We consider the top-down,constraining perspective to be important, however, because in combinationwith tactics such as philopatry and site fidelity (chapter 3), it limitshabitat availability for individual deer which then constrains theestablishment of activity nuclei to a specific subset of habitats. Theclearest example of this constraint in our study came from the case ofresident deer, most having relatively little or no access to old forests intheir home ranges (Table 4.3), consequently lacking preference for oldforests (Table 4.6), and choosing to establish activity nuclei mostly inother forest types (Fig. 4.4).Interpretation of Habitat PreferenceIdeal-Free Distributions and Black-tailed Deer. - We believe there issufficient evidence to indicate that habitat selection tactics, especiallythose for winter habitat, are not in equilibrium with dynamics of habitatchanges as those dynamics have taken place on Vancouver Island. First,logging has changed the basic spatial pattern of deer habitat by reducingthe total amount of old forests at low elevations and by isolating theremaining old forests into widely separated patches. Simultaneously, broadexpanses of open and/or young forests are created. Second, this changehappens much faster than the more natural rate of change in an undeveloped117forest (Franklin and Spies 1984). Third, constraints on habitat choicesforce adaptation to changed habitats to be based more on generationaldynamics than solely by individual learning. O’Neill et a!. (1988)suggested that resources clumped in space, limited in abundance, orproviding a specialized need (all of which apply to old forests as winterhabitat for deer) demand large scales of resource use. Resident deeroperate at relatively small scales by definition. So, for example, not onlydoes habitat change occur more rapidly than changes in behaviourial tactics,resident for example are usually unaware of the condition of habitats thatchange (e.g., year by year knowledge about the location of remnant oldforest patches).Philopatry and site fidelity for alternate ranges (Garrott et a!.1987, Brown 1992, chapter 3) further suggest that even migratory deer maynot be free to choose habitats because social relations may be stronger thanhabitat needs in terms of home range establishment. Philopatry, forexample, means that establishment of new home ranges has as much or more todo with the mother’s historical choice than with an assessment, made byyoung deer, of current habitat quality. Consequently, the freedom ofmovement assumed by Fagen (1988) for Sitka black-tailed deer (0. h.sitkensis) is unlikely to hold for Columbian black-tailed deer on VancouverIsland.Finally, although we have reason to believe there are strong seasonalpatterns in the manner of habitat selection by deer, we also have reason tobelieve that a longer-term pattern exists as well. During severe winters(as opposed to severe winter weather), the requirement for old forests hasbeen known to be extreme not only on Vancouver Island (Smith 1973) but insoutheast Alaska as well (Schoen and Kirchhoff 1990). By comparison to the118winters of 1968-69 and 1971-72, winters during this study were relativelymild hence conferring more freedom for deer to use young and open forests.Individual Preferences for Habitats. - Individuals used habitats with adiversity of tactics which ultimately led us to conclude a general lack ofhomogeneity in habitat preferences. This could have resulted as aconsequence of the lack of freedom in establishing home ranges (discussedabove). For example, if home ranges were settled on the basis of thematriarch’s historical choice and habitats were subjected to persistent,rapid changes from logging, then availability of habitats, and tactics forusing habitats, would range widely for individual deer.Cases where homogeneity was not rejected were mostly those wherepreferences were low for particular habitats (e.g., migratory deerpreferences for open forests, resident deer preferences for old forests,resident deer preferences for high elevations, and resident deer preferencesfor north, west, and flat aspects). We interpreted this to imply that ourconclusions about lack of preference are likely more robust and general thanconclusions associated with preferred habitats.Relative Habitat Qualities of Young and Old Forests. - In moving from summerhabitats to winter habitats, deer generally preferred open forests less andold forests more, they preferred elevations >800 m elevations less andelevations at 400 to 600 m more, and they preferred north and east aspectsless and southern aspects more. Because we were unable to assess thepossible interaction among the main habitat vectors, one could argue thepreference for old forests, for example, is only a correlate with a moreproximate need for deer to choose habitat at lower elevations during winter.Logging patterns in coastal forests, however, tend to produce a marked lackof old forests at low elevations so we concluded that the preference for old119forests was likely a strong determinant of habitat selection.Lack of behaviour (migration tactics) or study area effects ondifferences in seasonal preference for the topographic vectors led us toconclude these preferences (Tables 4.7 and 4.8) as general seasonal tactics.The attributes of old forests includes a long list of mechanistic benefitsto deer: more abundant forage (Harestad 1979, 1985), higher quality forage(Van Home et a!. 1988, Happe et a!. 1990), an alternative and/or additional(to rooted) forage supply (Stevenson 1978, Rochelle 1980, Dawson et a!.1990), and reduced snow depths (Kirchhoff and Schoen 1987, Bunnell et a!.1990a) hence reduced cost of locomotion (Parker et al. 1984). But thesebenefits primarily accrue in winter months. Our data (Table 4.6), and thoseof others (Harestad 1985, Schoen and Kirchhoff 1990, Yeo and Peek 1992),show a strong avoidance of old forests in summer. In an evolutionary sense,these data indicate migration is instigated as much to avoid old forests insummer as to preferentially select them in winter. If migrations weredriven solely by topographic factors then we should have seen some migratorydeer preferring a matrix of young and clear forests in winter; we foundnone. Further, in periods of mild weather (sometimes lasting a number ofyears), some otherwise migratory deer failed to migrate (chapter 3) and as aresult, preferential use of old forests in winter was marginal (Table 4.6)even though it was highly preferred as a component of alternate ranges.While, the benefits of old forests to individual deer confirm it’shigh quality in winter, we concluded it was relatively avoided in all butsnowy weather. Hence, old forests represent a special habitat in coastalclimates that seemingly cannot be replaced by young forests but is neededonly when winter weather is severe; otherwise, a matrix of clear and youngforests appears preferential and may negate the need to migrate if they120occur at mid- to low-elevations. On a mechanistic basis (primarily forageabundance and quality), however, we can forecast the quality of youngforests in coastal British Columbia to decline with age considerably asthese forests age beyond 30- to 40-yr-old (Cowan 1945, Gates 1968, Alaback1982).IMPLICATIONS FOR RESEARCH AND MANAGEMENTWe concluded a lack of association between habitat preference andhabitat value and yet we also assessed the value of old and young forestsbased on the same preference estimates. The opportunity to confront these 2objectives with the same data resulted from maintaining data in a segregated(rather than pooled) condition, from exploiting the significant variancecontributed by effects such as behaviour and study area (Thomas and Taylor1990), and by structuring analysis on a hierarchical framework correspondingto levels of habitat acquisition decisions (Johnson 1980, Senft et al.1987). We believe these are important concepts in analysis of mostuse/availability data.The relative value of old and young forests should not be consideredstrictly discrete (Hobbs and Hanley 1990). Although the importance of oldforests cannot be denied, young forests in coastal British Columbia likelyprovide better winter habitat in most places during most years. Twoimportant qualifications are necessary, however: (1) young forests willinevitably grow older and temporarily decrease in quality and (2) migratorydeer appear to retain association with old forests and so their role in theoverall population must be considered before relaxing any constraints onharvest of those habitat types.If our conclusions are correct then at least 3 predictions result: (1)121as less old forest exists at mid- to low-elevations, seasonal habitatselection tactics will become dominated by residency and fewer deer willestablish natal ranges at high elevations; (2) populations in general shouldbe less resilient to perturbations caused by severe winter weather; and (3)as open forests age into young forests and as currently young forests ageolder, deer populations in general should decrease.LITERATURE CITEDAlaback, P. B. 1982. Dynamics of understory biomass in Sitka spruce-western hemlock forests of southeast Alaska. Ecology 63:1932-1948.Allen, T. F. H., R. V. O’Neill, and T. W. Hoekstra. 1984. Interlevelrelations in ecological research and management: some workingprinciples from hierarchy theory. USDA For. Serv. Gen. Tech. Rep. RM110. Ft. Collins, Cob. llpp.Bloom, A. M. 1978. Sitka black-tailed deer winter range in the KadashanBay area, southeast Alaska. J. Wildl. Manage. 42:108-112.Brown, E. R. 1961. The black-tailed deer of western Washington.Washington Dept. Game Biol. Bull. 13. l24pp.Brown, C. G. 1992. Movement and migration patterns of mule deer insoutheastern Idaho. J. Wildl. Manage. 56:246-253.Bunnell, F. L. 1985. Forestry and black-tailed deer: conflicts, crises,or cooperation. For. Chron. 61:180-184._____and A. S. Harestad. 1983. Dispersal and dispersion of black-taileddeer: Models and observations. J. Mamm. 64:201-209.F. W. Hovey, R. S. McNay, and K. L. Parker. 1990a. Forest cover,snow conditions, and black-tailed deer sinking depths. Can. J. Zool.68:2403-2408.K. L. Parker, R. S. McNay, and F. W. Hovey. 1990b. Sinking depthsof black-tailed deer in snow, and their indices. Can. J. Zool.68:917-922.Chesson, J. 1978. Measuring preference in selective predation. Ecology59: 211-215.1983. The estimation and analysis of preference and itsrelationship to foraging models. Ecology 64:1297-1304.Cowan, I. McT. 1945. The ecological relationships of the food of the122Colunibian black-tailed deer, Odocoileus hemionus columbi anus(Richardson), in the coast forest region of southern Vancouver Island,British Columbia. Ecol. Moriogr. 15:109-139._____1956. Life and times of the coast black-tailed deer. Pages 523-617 in W. P. Taylor, ed. The deer of North America. Stackpole,Harrisburg, Penn.Dawkins, R. 1976. Hierarchical organisation: a candidate principle forethology. Pages 7-53 in P. P. G. Bateson and R. A. Hinde, eds.Growing points in ethology. Cambridge Univ. Press, Cambridge,England.Dawson, R. J., H. M. Armleder, and M. J. Waterhouse. 1990. Preferences ofmule deer for Douglas-fir foliage from different sized trees. J.Wildi. Manage. 54:378-382.Dixon, K. R., and J. A. Chapman. 1980. Harmonic mean measures of animalactivity areas. Ecology 61:1040-1044.Don, B. A., and K. Rennolls. 1983. A home range model incorporatingbiological attraction points. J. Anim. Ecol. 52:69-81.Elnarsen, A. S. 1946. Management of black-tailed deer. J. Wildi. Manage.10:54-59.Fagen, R. 1988. Population effects of habitat change: a quantitativeassessment. J. Wildi. Manage. 52:41-46.Fahrig, L., and J.H. Paloheimo. 1988. Determinants of local populationsize in patchy habitats. Theor. Pop. Biol. 34:194-312.Franklin, J. F., and T. A. Spies. 1984. Characteristics of old-growthDouglas-fir forests. Pages 328-334 in Society of American Foresters,eds. Proceedings of the Society of American Foresters nationalconvention, October 16-20, 1983. Soc. of Am. For., Washington.Fretwell, S. 1972. Populations in seasonal environments. Princeton Univ.Press, Princeton, N.J. 217pp.and H. L. Lucas, Jr. 1970. On territorial behaviour and otherfactors influencing habitat distribution in birds. I. Theoreticaldevelopment. Acta Biotheoretica 19:16-36.Garrott, R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A. W.Alldredge. 1987. Movements of female mule deer in northwestColorado. J. Wildi. Manage. 51:634-643.Gass, C. L. 1985. Behaviourial foundations of adaptation. Pages 63-107 inP. P. G. Bateson and P. H. Klopfer, eds. Perspectives in ethology.Vol. 6. Plenum Publishing Corp., N.Y.Gates, B. R. 1968. Deer food production in certain seral stages of thecoast forest. M.S. Thesis, Univ. of British Columbia, Vancouver.123lO5pp.Hamlin, K. L., and R. J. Mackie. 1989. Mule deer in the Missouri RiverBreaks, Montana: A study of population dynamics in a fluctuatingenvironment. Montana Dept. Fish and Wildi., Missoula. 4Olpp.Happe, P. J., K. J. Jenkins, E. E. Starkey, S. H. Sharrow. 1990.Nutritional quality and tannin astringency of browse in clear-cuts andold-growth forests. J. Wildl. Manage. 54:557-566.Harestad, A. S. 1979. Seasonal movements of black-tailed deer on northernVancouver Island. Ph.D. Thesis, Univ. of British Columbia, Vancouver.98pp._____1985. Habitat use by black-tailed deer on northern VancouverIsland. J. Wildl. Manage. 49:946-950.Hawkins, R. E., and W. D. Klimstra. 1970. A preliminary study of thesocial organization of the white-tailed deer. J. Wildl. Manage.34:407-419.Hobbs, N. T., and T. A. Hanley. 1990. Habitat evaluation: douse/availability data reflect carrying capacity? J. Wildl. Manage.54:515-522.Johnson, D. H. 1980. The comparison of usage and availability measurementsfor evaluating resource preference. Ecology 61:65-71.Jones, G. W. 1975. Aspects of the winter ecology of black-tailed deer(Odocoileus hemionus columbianus Richardson) on northern VancouverIsland. M.S. Thesis, Univ. of British Columbia, Vancouver. 78pp.Kammermeyer, K. K., and R. L. Marchinton. 1976. Notes on dispersal ofmale white-tailed deer. J. Mamm. 57:776-778.Kirchhoff, M. D., and J. W. Schoen. 1987. Forest cover and snow:implications for deer habitat in southeast Alaska. J. Wildi. Manage.51:28-33.Kucera, T. E., and C. McCarthy. 1988. Habitat fragmentation and mule deermigration corridors: a need for evaluation. Trans. Western Sec.Wildl. Soc. 24:61-67.Kufeld, R. C., 0. C. Bowden, and D. L. Schrupp. 1989. Distribution andmovements of female mule deer in the Rocky Mountain foothills. J.Wildl. Manage. 53:871-877.Lenth, R. V. 1981. On finding the source of a signal. Technometrics23:149-154.Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology73:1943-1967.Manley, B. J. F. 1974. A model for certain types of selection experiments.124Biometrics 30:281-294.Masters, R. D., and R. W. Sage, Jr. 1985. White-tailed deer fawn/daminteractions and fawn home range establishment. N.Y. Fish and Game J.32:93-94.McCullough, D. R. 1985. Long range movements of large terrestrialmammals. Pages 444-465 in M. A. Rankin, ed. Migration: Mechanismsand adaptive significance. Contrib. Marine Sci., Suppl. vol. 27.McNay, R. S., J. A. Morgan, and F. L. Bunnell. 1994. Characterizingindependence of observations in movements of Columbian black-taileddeer. J. Wildl. Manage. 58:422-429._____L. D. Peterson, and J. B. Nyberg. 1988. The influence of foreststand characteristics on snow interception in the coastal forests ofBritish Columbia. Can. J. For. Res. 18:566-573.Meidinger, D., and J. Pojar. 1991. Ecosystems of British Columbia.British Columbia Minist. of For. Special Rep. Ser. 6. Victoria.330pp.Nyberg, J. B., F. L. Bunnell, D. W. Janz, and R. M. Ellis. 1986. Managingyoung forests as black-tailed deer winter ranges. British ColumbiaMinist. of For. Land Manage. Rep. 37. Victoria. 49pp.O’Neill, R. V., B. T. Milne, M. G. Turner, and R. H. Gardner. 1988.Resource utilization scales and landscape pattern. Lands. Ecol. 2:63-69.Parker, C. T. Robbins, and T. A. Hanley. 1984. Energy expenditures forlocomotion by mule deer and elk. J. Wildl. Manage. 48:474-488.Porter, W. F., and K. E. Church. 1987. Effects of environmental patternon habitat preference analysis. J. Wildi. Manage. 51:681-685.Robinette, W. L. 1966. Mule deer home range and dispersal in Utah. J.Wildl. Manage. 30:335-349.Rochelle, J. A. 1980. The role of mature conifer forests in the winternutrition of black-tailed deer on northern Vancouver Island. Ph.D.Thesis, Univ. of British Columbia, Vancouver. 295pp.Rose, C. L. 1982. Deer response to forest succession on Annette Island,southeast Alaska. M.S. Thesis, Univ. of Alaska, Fairbanks. 56pp.SAS Inst. Inc. 1985. SAS user’s guide: basics, version 5 edition. SASInstitute Incorporated, Cary, N.C. 584pp.Schamberger, N. L., and L. J. O’Neil. 1986. Concepts and constraints ofhabitat-model testing. Pages 5-1- in J. Verner, M. L. Morrison, andC. J. Ralph, eds. Wildlife 2000: modeling wildlife-habitatrelationships of terrestrial vertebrates. Univ. of Wisconsin Press,Madison.125Schoen, J. W., and M. D. Kirchhoff. 1990. Seasonal habitat use by Sitkablack-tailed deer on Admiralty Island, Alaska. J. Wildl. Manage.54:37 1-378._____M. D. Kirchhoff, and M. H. Thomas. 1985. Seasonal distribution andhabitat use by Sitka black-tailed deer in southeastern Alaska. AlaskaDept. of Fish and Game, Federal Aid in Wildi. Res. Final Rep., ProjectW-17-11, W-21-1, W-21-2, W-22-3, and W-22-4., Juneau. 44pp.0. C. Wallmo, and M. D. Kirchhoff. 1981. Wildlife-forestrelationships: is a reevaluation of old growth necessary? Trans.North Am. Wildl. Nat. Resourc. Conf. 46:531-545.Schoener, T. W. 1987. A brief history of optimal foraging ecology. Pages5-68 in A. C. Kamil, J. R. Krebs, and H. R. Pulliam, editors.Foraging behaviour. Plenum Press, N.Y.Schooley, R. L. 1994. Annual variation in habitat selection: patternsconcealed by pooled data. J. Wildl. Manage. 58:367-374.Searle, S. R., F. M. Speed, and G. A. Milliken. 1980. Population marginalmeans in the linear model: an alternative to least square means. Am.Stat. 34:216-221.Senft, R. L., M. B. Coughenour, D. W. Bailey, L. R. Rittenhouse, 0. E. Sala,and D. M. Swift. 1987. Large herbivore foraging and ecologicalhierarchies. Bioscience 37:789-799.Shannon, C. E. 1948. A mathematical theory of communication. Bell SystemTech. J. 27:379-423.Sinclair, A. R. E. 1984. The function of distance movements invertebrates. Pages 240-258 in I. R. Swingland and P. J. Greenwood,eds. The ecology of animal movement. Claredon Press, Oxford, England.Smith, I. D. 1968. The effects of hunting and seral succession uponVancouver Island black-tailed deer. M.S. Thesis, Univ. of BritishColumbia, Vancouver. l4Opp.1973. Our declining winter ranges. Western Fish and Game Magazine.January. 4pp.Sokal, R. R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman and Company,N.Y. 859pp.Stevenson, S. K. 1978. Distribution and abundance of arboreal lichens andtheir use as forage by black-tailed deer. M.S. Thesis, Univ. ofBritish Columbia, Vancouver. 148pp.Swihart, R. K., and N. A. Slade. 1985. Testing for independence ofobservations in animal movements. Ecology 66:1176-1184.Thomas, 0. C. 1970. The ovary, reproduction, and productivity of femaleColurnbian black-tailed deer. Ph.D. Thesis, Univ. of British Columbia,126Vancouver. 2llpp.Thomas, 0. L., and E. J. Taylor. 1990. Study designs and tests forcomparing resource use and availability. J. Wildl. Manage. 54:322-330.Van Home, B. 1983. Density as a misleading indicator of habitat quality.J. Wildl. Manage. 47:893-901.Verbyla, D. L., and K. Chang. 1994. Potential problems in using GIS forwildlife habitat research. Pages 271-277 in Polaris ConferencesIncorporated, eds. GIS’94 Volume 1: Decision making with GIS thefourth dimension, Proceedings of a conference February 21-24,Vancouver, British Columbia. Minist. of Supply and Serv. Canada,Ottawa, Ontario.White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-trackingdata. Academic Press, San Diego, Calif. 383pp.Willms, W. D. 1971. The influence of forest edge, elevation, aspect, siteindex, and roads on deer use of logged and mature forest, northernVancouver Island. M.S. Thesis, Univ. of British Columbia, Vancouver.184pp.Yeo, J. J., and J. M. Peek. 1992. Habitat selection by female Sitka blacktailed deer in logged forests of southern Alaska. J. Wildl. Manage.56:253-261.Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, EnglewoodCliffs, N.J. 718pp.127CHAPTER 5 - RESPONSE TO LOGGING OF WINTER HABITATRecent studies of black-tailed deer have indicated that old forests(>250-yr-old) provide important habitat during winter (Jones 1974, Bloom1978, Harestad 1979, Rose 1981, Schoen and Kirchhoff 1985). This importancewas interpreted from a preference for old forests by deer, particularlythose born in regions of high snowfall (chapter 4). Old forests provide anabundant source of arboreal forage seldom found in younger forests(Stevenson 1978) and provide shelter from snow (Kirchhoff and Schoen 1987,McNay et al. 1988) which permits reduced costs of locomotion (Parker et al.1984, Bunnell et al. 1990a,b) and lower rates of food burial (Jones 1975,Harestad 1979). For these reasons, old forests, especially those at lowelevations, are generally considered to be the highest valued winter habitatfor black-tailed deer (Walimo and Schoen 1980, Bunnell and Jones 1984,chapter 4). Furthermore, black-tailed deer are philopatric and developstrong fidelity to specific sites (Schoen and Kirchhoff 1985, Weckerly 1993,chapter 3). Therefore, loss of old forests (e.g., wildfire or logging) onwinter ranges could lead to conflict within affected deer between abehaviourial tendency for fidelity to their winter-range site and preferencefor old forests.The response of these deer to loss of winter habitat through loggingis poorly understood. Affected deer could: (1) remain in their existingwinter range where they are either in, or surrounded by, new habitatconditions; (2) remain in their existing winter range but shift habitat useaway from the affected site; or (3) move to a new winter range which may becomposed of at least some old forest. I consider these potential responsesto vary from one dominated by site fidelity (response 1) to anotherdominated by habitat preference (response 3). The intermediate response 2128could be considered site fidelity with a range size change. My objectivewas to assess which of these responses would be used by individual deerafter controlled logging of old forests within their winter range.Knowledge of actual responses to logging may help explain why deer use awide variety of habitats during winter (Yeo and Peek 1992, chapter 4) andmay also help illuminate the population-level consequences of individualresponses to habitat change. This knowledge is also basic to planningsilvicultural prescriptions for restoration of winter habitat in youngforests (e.g., Nyberg et al. 1986). The specific tests were: (1) that sitefidelity did not differ between comparisons in which 1 year involveddisturbance (i.e., logging) and comparisons without disturbance, or betweendeer subjected to disturbance and those not subjected to disturbance; (2)that annual variation in winter range sizes did not differ for deersubjected to disturbance compared to those not subjected to disturbance; and(3) that use of old forests in winter did not differ among years for thosedeer subjected to disturbance as a result of the disturbance.STUDY AREASI studied responses of black-tailed deer to logging in Chemainus(48°56’N, 124°05’W) and Nimpkish (50°08’N, 126°30’W) river valleys onVancouver Island, British Columbia from 1988 to 1991. Each study area was1,600 ha in size and, in each, I identified an individual stand of oldforest that was slated for logging 1-2 yr after the initiation of study.The stands met criteria for typical winter habitat (e.g., Bunnell 1985,Nyberg et a!. 1986, Bunnell 1990) and were isolated from similar stands.Hence, adjacent habitats were young, Douglas-fir forests (6- to 45-yr-old),recently logged areas (0- to 5-yr-old), or remnant areas of old forest129unsuitable for winter habitat because of topographic position (e.g.,unsuitably high in elevation and/or on northern aspect). No habitat existedbetween the ages of 46- and 250-yr-old.A proposed, or recent, logging area is locally referred to as a cut-block and I use that term. The Chemainus River cut-block was 60 ha with anaverage tree volume of about 1,000 in3/ha and was scheduled for logging duringthe summer of 1989. Elevation in the cut-block ranged from 800-1,000 m withan aspect of 2100. Tree species composition was dominated by Douglas-firand western hemlock with a minor component of western redcedar. Yellow-cedar occurred at higher elevations.The Nimpkish River cut-block was situated about 202 km northwest ofthe Chemainus River site. It was 47 ha with an average tree volume of about700 m3/ha and was scheduled for logging during the fall of 1990. Elevationin the cut-block ranged from 300-650 m with an aspect of 160°. Tree speciescomposition was similar to the Chemainus River site.METHODSExperimental DesignStudy design was based on 3 criteria beyond the fact that candidatestands for logging had to be typical of deer winter habitat. First, Iwanted to force a distinct and dramatic habitat change on individual deer.The treatment deer (i.e., those with logging within their home ranges) wouldhave to make a clear choice between fidelity to their winter range site orcontinuity of their habitat choices. I defined fidelity as overlap inconsecutive winter ranges and defined continuity of habitat choice asinsignificant (x2 > 0.05) differences in the use of old forests duringconsecutive winters (November through April). To make results for this130comparison mutually exclusive, I aimed to have the candidate stands loggedcompletely and to retain other stands of similar characteristics accessible(2-4 km away), but not directly adjacent to, each logged stand.Second, I expected a strong source of annual variation in observationsof fidelity and habitat use. Therefore, I monitored a second group of deertermed control deer. Control deer, for comparisons of fidelity and winterrange size, were monitored close to each group of treatment deer (<2 kmaway). Control deer, for habitat-use comparisons, were deer that used oldforests during the same years but in areas remote from any current loggingactivity (although logging had occurred in these areas in the past); wechose 2 deer from data collected at Nanaimo River which is a neighbouringvalley to Chemainus River (chapter 4).Third, and again because I expected strong annual variation, Iarranged the study so that the 2 forest stands were logged in successiveyears. As a result, I had 2 pre-disturbance years at Nimpkish River and 2post-disturbance years at Caycuse. Comparisons of fidelity and use of oldforests between these years were termed non-disturbance comparisons whileother comparisons were termed disturbance comparisons.Animal Capture and MonitoringDuring the winters of 1988-1989 and 1989-1990 at Nimpkish River, and1988-1989 at Chemainus River, traps (Clover 1956) were placed within thestands of old forest identified for logging and in nearby stands. Oncetrapped, female deer were restrained manually and collared with radiotransmitters (Telonics Inc., Mesa, Ariz., USA and Lotek Inc., Newmarket,Ontario, Canada).Collared deer were located once each week with triangulation data131obtained at permanent stations marked at 100-rn intervals along roadsthroughout each study area. Deer locations were estimated using the maximumlikelihood estimator described by Lenth (1981) and provided in a SAS program(SAS Inst. Inc. 1985) by White and Garrott (1990:64). I determined thaterror polygons (Lenth 1981) for estimated deer locations were usually <1.0ha (McNay et al. 1994).I expressed winter ranges as the 95% minium convex polygon (Mohr 1947)based on locations of deer collected during winter. I excluded migrationsand locations of migratory deer if they were still on summer ranges duringthis time (chapter 3). I assessed independence of location observations ina related study (McNay et al. 1994) and considered locations sampled once-per-week were biologically independent, systematic observations of use ofspace. Range sizes, therefore, were an adequate index for comparisons amongdeer through predetermined time intervals. Individual deer were theexperimental units and I assumed they were independent of each other.Habitat use was determined by overlaying deer locations as query points toforest cover maps in a Geographic Information System (Terrasoft; DigitalResources, Nanaimo, B.C. and PAMAP; PAMAP Technologies Corp., Victoria,B.C.). Habitats were: open if 0- to 5-yr-old; young if 6- to 45-yr-old;old-low if >250-yr-old and below, or at the same elevation as, the loggedstand; or old-high if >250-yr-old and above the elevation of the loggedstand.Statistical AnalysisFor statistical convenience I constructed an index of fidelity basedon a x2 test of independence between consecutive winter ranges (White andGarrott 1990:136). Positioned on the arithmetic centre of one range, I used132concentric circles with radii iteratively reduced by 100 m, starting at aradius of 1,500 m, to find the first test where deer locations in the nextrange were considered to be independently distributed (P < 0.05) from thosein the previous range. This radius was termed the maximum radius forindependence (MRI) and small MRI values indicated strong fidelity. Habitatuse was the percent of location observations in old-low forests where datawere transformed using an arcsine transformation for percentage data (Zar1984). I used a repeated measures, analysis of variance based on year (forrange size and habitat use) or based on disturbance (for fidelity) to testfor the potential effect of logging (control versus treatment deer) nestedwithin study areas (Chemainus or Nimpkish). All reported means are least-squares estimates (Searle et al. 1980).RESULTSLogging was most successfully carried out at Nimpkish River because noold-low forest was left adjacent to the cut-block. About 180 ha of old-highhabitat remained adjacent to the block on its upper boundary but this waspoor quality winter habitat. Old-low habitat was available in a 15 ha standabout 0.3 km below the bottom boundary of the cut-block and was separatedfrom it by a stand of young forest. At Chemainus River, the logging was notcompleted as planned. Half the cut-block was harvested in 1989 and theother half in 1990. A 30 ha block of old-low habitat remained adjacent tothe cut-block because a gully impeded further logging. There was another57-ha stand of old-low forest about 1.3 km away from the cut-block. In eachstudy area, half of the total old-low habitat was logged (Fig. 5.1).Twenty female deer were monitored in the 2 study areas for 2 or 3winters between 1988-1989 and 1990-1991 for a total of 52 deer-winters. Six133Chemainus RiverNimpkish RiverLogged habitat807060s.00’Cud)F°3020100II]Open Young Old-low Old-highHabitat typesWaterFigure 5.1 Habitat abundance (% of total) at 2 study areas on VancouverIsland, British Columbia, 1988-1991.134deer in each study area (3 treatment deer and 3 control deer) were monitoredfor all 3 years (Table 5.1). Each deer was monitored 16 times per winter onaverage (n = 52, SD = 8). Two other deer that had access to old forests atNanaimo River were monitored during the same study period.Fidelity to winter ranges was relatively stronger at Nimpkish Riverwhere the maximum radius for independence (MRI) between pre- and post-disturbance comparisons was 0.3 km compared to 0.5 km at Chemainus River(Table 5.1). Fidelity measurements for deer with 3 years of data (i.e.,both non-disturbance and disturbance comparisons) was significantly strongerin non-disturbance comparisons (F = 5.44; df = 1; P = 0.048) but this wasnot associated with differences between treatment and control deer (F =0.23; df=3; P = 0.871).Home range sizes did not vary consistently through time for deer thatwere followed for 3 years at Chemainus River (F = 2.11; df = 2; P = 0.184)or at Nimpkish River (F = 1.30; df = 2; p = 0.325). Also, there was noevidence of a disturbance effect (F = 0.66 and 0.63; df = 2; P = 0.463 and0.470 at Chemainus and Nimpkish rivers, respectively) even though home rangesizes did increase (F = 6.90; df = 1; P = 0.018) when all deer wereconsidered for a comparison of 1-yr-pre- to 1-yr-post-disturbance (Table5.1). In this latter comparison, there was no treatment effect (F = 0.98;df = 3; p = 0.425).Use of old-low habitat by treatment deer that were monitored for 3years did not differ (F = 3.15 and 3.96; df = 1; P = 0.174 and 0.141 atChemainus and Ninipkish rivers, respectively), on average, from the controldeer. In both study areas, however, there was a temporal effect (F = 13.21and 23.55; df = 2; P = 0.006 and 0.001 at Chemainus and Nimpkish rivers,respectively) caused by treatment deer using less old-low habitat afterTabLe5.1.Effectofforestloggingonrangesizes,habitatuse,andsitefidelityofradio-collared,black-taileddeerwhereloggingoccurredinside(treatment)oroutside(control)pre-disturbancehomeranges.Deerwerefrom2studyareasonVancouverIsland,BritishCott.mtia,1988-1991.StudyAreaRangesize(ha)bOldforest(%oftotaluse)FidelityindexdYearControlTreatmentControlTreatmentControlTreatmentNimpkishRivere4727471988-8920.2(7.4)31.7(7.4)67.4(5.2)38.5(4.2)1989-9051.2(15.1)37.8(15.1)38.9C8.9)42.3C7.3)0.23C0.16)0.17(0.16)1990-9154.2(13.8)34.4(13.8)42.3(10.3)9.2(8.4)0.40C0.29)0.33(0.29)1-yr-pre47.2(8.8)34.2(6.7)38.3C6.1)42.0C4.0)1-yr-post47.9(22.2)60.3(16.8)30.7C8.1)5.7C5.3)0.30(0.21)0.34C0.16)ChemainusRivern’5424541988-8912.0C8.3)24.7(8.3)67.4(5.2)51.7C8.4)1989-9055.8(41.3)77.4(41.3)38.9C8.9)37.7C7.3)0.83(0.29)0.53C0.29)1990-9124.1(22.2)57.3(22.2)42.3(10.3)11.7C6.2)0.66C0.16)0.20C0.16)1-yr-pre10.4C7.9)18.8C8.8)38.3C6.1)47.6(5.3)1-yr-post48.6(19.8)81.1(22.2)30.7C8.1)32.0(7.0)0.56C0.19)0.52C0.21)Loggingoccurredonapre-selectedstandofoLd-forestmeetingtypicalwinterhabitatcharacteristics(Nybergetal.1986)in1989atChemainusRiverandin1990atNimpkishRiver.Rangesizewasa95%minimumconvexpolygonusingLocationsfromNovember-April.Locationsofmigratorydeeronsumerrangeswereomitted.%oldforestvaluesarearcsinetransformationsofthepercentoftotallocationscollectedduringonewinterseason(NovemberthroughApril)thatwerefoundinoldforestsatelevationsequaltoorbelowtheloggedstand.Anindexoffidelitywasexpressedasahomerangeradiuswhich,whenusedtorepresentthepre-disturbancehomerange,wasjudgedindependentofthepost-disturbancehomerangebyax2testofindependence(P=0.05).Samplesizesprovidedarefor1-yr-pre-disturbanceversus1-yr-post-disturbancecomparisons.Samplesfor3-yrannualstatisticscamefrom3controland3treatmentdeerineachstudyarea.(‘3 0,136logging compared to control deer (Table 5.1). This temporal effect was onlyevident in the second year after disturbance for treatment deer at ChemainusRiver and in the post-disturbance year for treatment deer at Nimpkish River.DISCUSSIONMy conclusion from this study is that fidelity, rather than habitatchoice, dominates initial (i.e., 1-2 yr post-disturbance) responses by deerto large, striking changes in habitat condition (i.e., 45-60 ha clear-cutlogging of old forests). Relative to the other potential responses Imeasured, fidelity remained the most consistent between treatment andcontrol deer through the 3 years of this study (Table 5.1). Use of old-lowhabitats, although steady in non-disturbance years, was reduced considerablyafter disturbance even when this same type was available within 0.5 km ofthe cut-block at Nimpkish River. I noted, however, that when old-lowhabitat was left adjacent to the cut-block at Chemainus River, use of thattype continued. After further logging at Chemainus River, however, use ofold-low habitat decreased similar to the response at Nimpkish River. Winterhome range sizes were variable but did not vary in association with thedisturbance.Other studies have shown similar results demonstrating strong fidelityto home sites during extreme disturbances in habitat (Hood and Inglis 1974;Hershey and Leege 1982; Edge et al. 1985; Gasaway et al. 1989). Edge et al.(1985) found that logging caused elk (Cervus elaphus nelsoni) to shifthabitat use within the home range with no significant outward shift in homerange size or site. Fidelity is likely advantageous in stable environmentsbecause a mother’s site-specific knowledge about food, cover, and potentialhazards can be easily demonstrated to offspring (Edwards 1976, Hamlin and137Mackie 1989:233-238). Alternatively, fidelity may indicate a simple lag indecision-making while individuals obtain enough knowledge to evaluate tradeoffs provoked by rapid changes in habitat (Gass 1985). Lags themselves maybe advantageous because quick selections of new home ranges, if deer wereprone to such decisions, could decrease survival due to unknown hazards inunfamiliar areas (O’Bryan and McCullough 1985). At very least, such forced,quick decisions could decrease foraging efficiency (Provenza and Balph1987), although persistence of such detrimental effects may not last long(Gillingham and Bunnell 1989). However, fidelity to sites that have beendisturbed undoubtedly presents deer with new challenges in learning toadjust to the new habitat condition, at least in the short-term.MANAGEMENT IMPLICATIONSFidelity to specific sites has important management implications.First, Fagen’s (1988) assumption about freedom of movement between habitats(i.e., the ideal-free distribution of animals) was invalidated for oursample of deer. If old-low forests were the best winter habitat for black-tailed deer, my sample of deer failed to alter winter range sites toaccommodate the reduction in range quality caused by logging. Rather,treatment deer simply increased the proportional use of other habitatswithin their winter ranges. Even if this were simply a delay in decisionmaking, I am led to question the meaning of habitat use observations sampledin areas of extensive habitat disturbance. Are habitats chosen as a matterof preference, as a matter of habit, or as a default from fidelity to sites?I consider the data here to indicate that site-fidelity may play a largerole in determining habitat preferences in post-disturbance situations.Consequently, this may help explain the wide variance in reports of habitat138preferences by black-tailed deer (Yeo and Peek 1992, chapter 4).In cases of continuous and contiguous logging, some deer areinevitably faced with no old forests to use. In recent studies involvingsuch cases, none of the study deer trapped in (chapters 3 and 4), orimmobilized in or near (Yeo and Peek 1992), young forests were migratory.Provided that site fidelity constrains habitat choice as my results appearto indicate, this absence of migratory deer in young forests could onlyoccur if: (1) resident deer were superior competitors in the remaining youngforests, or (2) the offspring of migratory deer in these situations developnon-migratory tactics. Alternatively, if the site fidelity I observed wasonly a lag in decision making then migratory deer may eventually seek outold forests elsewhere (i.e.; move to new winter range sites). Although datahere are insufficient to address these hypotheses, I noted that deer atChemainus River shifted home ranges to use old-low habitat when that typewas adjacent to the cut-block. Until further data are available, it wouldseem that some deer (perhaps only migratory deer) may be able to makeappropriate habitat adjustments when winter ranges are only partiallylogged. If this were true, I would expect deer living in extensivelymodified areas to be mostly resident, as our observations suggest (Yeo andPeek 1992, chapter 4). In turn, this would mean migratory deer are forcedinto increasingly smaller areas of old forests. Where logging was carriedout preferentially in areas of prime winter range (Schoen and Kirchhoff1985), the density problem intensifies because an increasing density of deerwould occur in habitats of decreasing value. Based simply on carryingcapacity theory, we would expect migratory deer populations to decline(Hobbs and Hanley 1990).Another implication of site fidelity constraints on habitat choices is139that management of young forests specifically to improve their suitabilityas winter habitat for deer (e.g., Nyberg et a!. 1986) could simply benefitdeer in the immediate vicinity, at least initially. This is not to say thatsuch improvements are fruitless, however, because in highly philopatricspecies, benefits would accrue to offspring and subsequent generations. Theimportant point is that habitat improvements may be most beneficial longafter the initial management efforts. Such a lag in results from habitatimprovement attempts have been documented for moose in Alaska (Gasaway etal. 1989) and we consider our results to indicate the same would happen whenmanaging forests for black-tailed deer.LITERATURE CITEDBloom, A. M. 1978. Sitka black-tailed deer winter range in the KadashanBay area, southeast Alaska. J. Wildl. Manage. 42:108-112.Bunnell, F. L. 1985. Forestry and black-tailed deer: conflicts, crises, orcooperation. For. Chron. 61:180-184._____1990. Black-tailed deer ecology and forest management. Pages 31-63in J. B. Nyberg and D. W. Janz, eds. Deer and elk habitats in coastalforests of southern British Columbia: a handbook for forest andwildlife managers. British Columbia Minist. of For. Special Rep. Ser.5. Victoria.and G. W. Jones. 1984. Black-tailed deer and old-growth forests: asynthesis. Pages 411-420 in W. R. Meehan, T. R. Merrell, Jr., and T.A. Hanley, eds. Fish and wildlife relationships in old-growthforests. Proc. Symp. April 1982 in Juneau, Alaska. Am. Inst. Fish.Res. Biol., Morehead City, N.C.F. W. Hovey, R. S. McNay, and K. L. Parker. 1990a. Forest cover,snow conditions, and black-tailed deer sinking depths. Can. J. Zool.68:2403-2408.K. L. Parker, R. S. McNay, and F. W. Hovey. 1990b. Sinking depthsof black-tailed deer in snow, and their indices. Can. J. Zool.68:917-922.Clover, M. R. 1956. Single-gate deer trap. Calif. Fish and Game. 42:199-201.Edge, W. D., C. L. Marcum, and S. L. Olson. 1985. Effects of logging140activities on home-range fidelity of elk. J. Wildi. Manage. 49:741-744.Edwards, J. 1976. Learning to eat by following the mother in moosecalves. Am. Midl. Nat. 96:229-232.Fagen, R. 1988. Population effects of habitat change: a quantitativeassessment. J. Wildl. Manage. 52:41-46.Gasaway, W. C., S. D. Dubois, R. D. Boertje, D. J. Reed, and D. T. Simpson.1989. Response of radio-collared moose to a large burn in centralAlaska. Can. J. Zool. 67:325-329.Gass, C. L. 1985. Behaviourial foundations of adaptation. Pages 63-107 inP. P. G. Bateson and P. H. Klopfer, eds. Perspectives in ethology.Vol. 6. Plenum Publishing Corp., N.Y.Gillingham, M. P., and F. L. Bunnell. 1989. Effects of learning on foodselection and searching behaviour of deer. Can. J. Zool. 67:24-32.Hamlin, K. L., and R. J. Mackie. 1989. Mule deer in the Missouri RiverBreaks, Montana: a study of population dynamics in a fluctuatingenvironment. Montana Dept. of Fish and Wildl., Missoula. 4Olpp.Harestad, A. S. 1979. Seasonal movements of black-tailed deer on northernVancouver Island. Ph.D. Thesis, Univ. of British Columbia, Vancouver.98pp.Hershey, T. J., and T. A. Leege. 1982. Elk movements and habitat use on amanaged forest in north-central Idaho. Idaho Dep. Fish Game Wildl.Bull. 10. 24pp.Hobbs, N. T., and T. A. Hanley. 1990. Habitat evaluation: douse/availability data reflect carrying capacity? J. Wildl. Manage.54:515-522.Hood, R. E., and J. M. Inglis. 1974. Behaviourial responses of white-tailed deer to intensive ranching operations. J. Wildl. Manage.47:664-672.Jones, G. 1974. Influence of forest development on black-tailed deerwinter range on Vancouver Island. Pages 139-148 in H. C. Black, ed.Wildlife and forest management in the Pacific Northwest. Oregon St.Univ., Corvallis._____1975. Aspects of the winter ecology of black-tailed deer(Odocoileus hemionus columbianus Richardson) on Vancouver Island.M.Sc. Thesis, Univ. of British Columbia, Vancouver. 78pp.Kirchhoff, M. D., and J. W. Schoen. 1987. Forest cover and snow:implications for deer habitat in southeast Alaska. J. Wildl. Manage.51: 28-33.Lenth, R. V. 1981. On finding the source of a signal. Technometrics14123:149-154.McNay, R. S., J. A. Morgan, and F. L. Bunnell. 1994. Characterizingindependence of observations in movements of Columbian black-taileddeer. J. Wild]. Manage. 58:422-429._____L. Peterson, and J. B. Nyberg. 1988. The influence of forest standcharacteristics on snow interception in the coastal forests of BritishColumbia. Can. J. For. Res. 18:566-573.Mohr, C. 0. 1947. Tables of equivalent populations of North American smallmammals. Am. Midi. Nat. 37:223-249.Nyberg, J. B., F. L. Bunnell, D. W. Janz, and R. M. Ellis. 1986. Managingyoung forests as black-tailed deer winter ranges. British ColumbiaMinist. For. Land Manage. Rep. 37. Victoria. 49pp.O’Bryan, M. K., and D. R. McCullough. 1985. Survival of black-tailed deerfollowing relocation in California. J. Wild]. Manage. 49:115-119.Parker, K. L., C. T. Robbins, and T. A. Hanley. 1984. Energy expendituresfor locomotion by mule deer and elk. J. Wild]. Manage. 48:474-488.Provenza, F. D., and D. F. Baiph. 1987. Diet learning by domesticruminants: theory, evidence and practical implications. App]. Anim.Beh. Sd. 18:211-232.Rose, C. L. 1981. Deer response to forest succession on Annette Island,southeast Alaska. M.S. Thesis, Univ. of Alaska, Fairbanks. 56pp.SAS Inst. Inc. 1985. SAS user’s guide: basics, version 5 edition. SASInstitute Inc., Cary, N.C. 584pp.Schoen, J. W., and M. D. Kirchhoff. 1985. Seasonal distribution and home-range patterns of sitka black-tailed deer on Admiralty Island,southeast Alaska. J. Wildi. Manage. 49:96-103.Searle, S. R., F. M. Speed, and G. A. Milliken. 1980. Population marginalmeans in the linear mode]: an alternative to least square means. Am.Stat. 34:216-221.Stevenson, S. K. 1978. Distribution and abundance of arboreal lichens andtheir use as forage by black-tailed deer. M.S. Thesis, Univ. ofBritish Columbia, Vancouver. l48pp.Walimo, 0. C. and J. W. Schoen. 1980. Response of deer to secondary forestsuccession in southeast Alaska. For. Sd. 26:448-462.Weckerly, F. W. 1993. Intersexual resource partitioning in black-taileddeer: a test of the body size hypothesis. J. Wildi. Manage. 57:475-494.White, G., and R.A. Garrott. 1990. Analysis of Wildlife radio-tracking data.Academic Press Inc., San Diego, Calif. 383pp.142Yeo, J. J., and J. M. Peek. 1992. Habitat selection by female Sitka blacktailed deer in logged forests of southeastern Alaska. J. Wildl.Manage. 56:253-261.Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, EnglewoodCliffs, N.J. 718pp.143CHAPTER 6 - MORTALITY CAUSES AND SURVIVAL ESTIMATES2Interactions among wolves (Canis lupus), black-tailed deer, and deerhunters have been assumed to dominate the predator-ungulate system onVancouver Island (Janz and Hatter 1986). Declines of deer, resulting frompredation (50-70% from 1976-82) and hunter harvests have been contrary tomanagement objectives. Because predators were presumed the major cause ofthe declines (Jones and Mason 1983), the argument for retention of winterhabitat (Bunnell 1985) was difficult to make to those wishing to use theland for other purposes (Janz and Hatter 1986). Managers were forced toreduce deer mortality before addressing habitat concerns. Consequently,attention has been focused on population modelling (Hatter and Janz 1994) tofoster management initiatives that help restore deer populations. Also, baglimits for deer hunters on Vancouver Island have been reduced and, in someareas, limited control of wolves has been provisionally instigated (Janz andHatter 1986).Population management requires information on factors that causepopulation changes, principally survival and reproductive rates (Caughley1976). Although reproductive rates for black-tailed deer are documented(Taber 1953, Cowan 1956, Golley 1957, Thomas 1970, Thomas and Cowan 1975,Thomas 1983), survival rates are not. Survival is usually measuredindirectly in surveys of recruitment and gross population changes (Harestadand Jones 1981, Janz 1989; see Hatter 1988 for an exception), and mortalitystudies have focused on single, rather than all, mortality causes (Klein andOlson 1960, Smith 1968, Hebert et al. 1982, Jones and Mason 1983, vanBallenberghe and Hanley 1984).2 Published as: McNay, R. S., and J. M. Voller. 1995. Mortality causesand survival estimates for adult female, Columbian black-tailed deer. J.Wildl. Manage. 59:138-146.144Typically, winter weather has been advanced as a factor influencingsurvival, having direct and indirect effects. Snow covers forage andimpedes locomotion (Harestad et al. 1982), creating a direct effect onindividual energy balances and, hence, survival. Shifts in use of habitatresulting from accumulations of snow can modify survival rates indirectlydue to consequent shifts in predator efficiency and/or prey vulnerability(Mech 1977, Messier and Barrette 1985, Nelson and Mech 1991). Withoutspecific information on the range of mortality causes, on mortality andsurvival rates, and on environmental and behaviourial factors affectingsurvival, changes in population management will be ad hoc.We investigated survival rates and mortality causes for adult, femaleblack-tailed deer at 4 study areas on Vancouver Island. Mortality data camefrom deer that were radio-collared for another study (McNay and Doyle 1990)and were used to (1) document causes of mortality, (2) estimate averagemonthly cause-specific mortality (N), (3) estimate average monthly survival(.), and (4) assess the relative effect of 5 variables on 11 and , (studyarea, seasonal movement types of deer, average elevation used by deer(monthly), month of year, and mean monthly snowdepth).STUDY AREASThe study took place at 4 locations on Vancouver Island: Nanaimo(49°02’N, 124°12’W), Chemainus (48°56’N, 124°05’W), Nimpkish (50°08’N,126°30’W), and Caycuse (48°48’N, 124°30’W) river valleys. Nanaimo andChemainus rivers are neighbouring valleys 43 km northeast of Caycuse Riverand 202 km southeast of Nimpkish River. Study areas were 200-300 km2 withvalley bottoms located at 200 m above sea level (asi). Peaks ranged from1,249 m asl at Caycuse to 1,821 ni asl at Nimpkish. The Chemainus, Nanaimo,145and Nimpkish rivers were relatively open, flat-bottomed valleys (U-shaped),while Caycuse River had steeper slopes and least flat area at low elevations(a V-shaped valley).All study areas were extensively logged by clearcutting resulting inhabitats ranging from recently deforested (0- to 5-yr-old) to old (>250-yr-old) forests. Arrangement of habitats was typical of coastal logging withvalley bottoms in young (6- to 45-yr-old) forests, most mid-slopesdeforested, and higher elevations and headwaters in old forests. Generally,Chemainus River had the least old (about 6%) and most young (about 82%)forests. Caycuse and Nimpkish sites had similar amounts of old (40%) andyoung (30%) forests. Nanaimo River had about 30% old forest and about 50%young forest. Each study site was 10 to 20% deforested.Wolves, cougars (Felis concolor), and humans preyed on adult deer onVancouver Island (Jones and Mason 1983, Janz 1989). Some wolves wereremoved from all study areas as part of ongoing predator management,although the greatest effort occurred at Nanaimo (35 removed since 1982) andNimpkish (44 wolves removed from 1987 to 1991) rivers. Local trapperssporadically removed wolves at Caycuse (1 wolf removed in 1985 and 2 in1986). We removed 1 cougar from Nanaimo River in 1988. All areas were opento buck deer and cougar hunting during their respective seasons. Inaddition, there was a limited entry hunt for antlerless deer at NanainioRiver during weekends each November.Vancouver Island is temperate and wet; in the dominate ecologicalzone, no month has a mean temperature <0 C, and mean temperature of thewarmest month is 17 C. Over most of the island there is usually 291 frostfree days, an average 820 mm of snow, and a mean precipitation of 2,140 mmeach year (Meidinger and Pojar 1991).146METHODSWe fitted deer with radio transmitters containing mortality sensorsfrom 1982-1988 at Nanaimo River, in 1989-1990 at Nimpkish and Caycuserivers, and in 1989 at Chemainus River. In 1988-1990, we captured most deerduring winter in Clover traps (Clover 1956) and manually restrained themwhile fitting radio collars. During 1982-1987, and when trapping wasunsuccessful, we immobilized deer with powdered anectine delivered in dartsshot from a 32-gauge CAP-CHUR gun. We monitored radio-collared deer 1time/week for 4 yr after capture or until death, provided we could maintainradio contact. Deer dying 12 days after capture (n = 3) were excluded fromanalyses because we could not positively rule out capture myopathy (Harthoon1977).We established age classes of deer by tooth wear and replacement atthe time of collaring (Robinette et a!. 1957). This technique wassufficient to classify deer as fawns (<1-yr-old), yearlings (between 1- and2-yr-old), or adults (>2-yr-old). We assigned all subadults common birthdates of 15 June, to determine when they reached adult status.Mortality CausesUpon monitoring mortality signals, usually 1-4 days after death, wesought deer and determined likely mortality cause. Death was by 1 of 6causes: predation by wolves or cougars, human related (legal hunting andpoaching), malnutrition, accident, or unknown. We identified predation,legal hunting, and poaching by state of the corpse (Roy and Dorrance 1976).We classified fresh kills with neck or head injuries, clean incisions at thegut, and partially buried remains as predation by cougars. Kills withconsiderable mid- and hind-section injury and comparatively more scattered,147unburied remains were classified as predation by wolves. When there wasonly a small amount of carcass remaining and no definite evidence of othermortality causes, we considered wolf predation as the likely cause of death.We determined nutritional status from examination of bone marrow (Cheatum1949). Red, gelatinous bone marrow suggested malnutrition was thepredisposing cause of mortality.Mortality and Survival Rate EstimationWe used logistic regression (SAS Inst. Inc. 1985) to compute maximumlikelihood estimates for monthly II and ,, while simultaneously assessing theeffects of 5 variables. Four of the independent variables were categorical:(1) study area had 4 levels (Caycuse, Chemainus, Nanaimo, and Nimpkish), (2)seasonal movement types of deer had 3 levels (migratory, resident, orunknown), (3) average elevations used within each month had 3 levels (<600 masl, >600 m asl, and unknown), and (4) month had 12 levels corresponding toeach month of the year. The fifth variable was mean monthly snow depthmeasured daily at the closest airport to each study area (Atmos. Environ.Serv., Environ. Can., Vancouver, BC). We censored deer with failed radios(n = 3), or that lost collars prematurely (n = 1), from analyses in themonth during which they could no longer be monitored. We conducted testsamong individual survival estimates, or between estimates of 11, using theCONTRAST statement in PROC CATMOD (SAS Inst. Inc. 1985).Our philosophy for choosing the model that most closely mimicked theobserved data followed White and Garrott (1990:222). We posed differenthierarchical subsets of the main factors as models. In the most generalmodel, all factors contributed to II and &, levels of the dependent variable.We compared this model (the null hypothesis) with simpler nested models, or148subsets, of the main factors (alternate hypotheses) using likelihood-ratiotests (White and Bartmann 1983). We continued iterative testing of nestedmodels until the null hypothesis was rejected, thereby revealing whichalternate model was most economical and consistent with the observed rates.Choice of the best model was corroborated on the basis of the AkaikeInformation Criterion (AIC; Chatfield 1992:197), a statistic used to assessparsimony of model construction.RESULTSWe captured and monitored 95 adult, female deer through 1982-91.During the same period, we caught 12 female yearlings and 12 female fawns,of which 9 yearlings and 1 fawn aged into the adult cohort and were added toour sample. Although we caught deer in each month, most captures (71%)occurred from January through March reflecting our greater effort andincreased capture success during these months. Collectively, 105 deeryielded 2,182 deer-months of data.Estimates of Cause-specific MortalityWe recorded 54 mortalities and 4 collar failures, most of which (64%)occurred either from April through June or during November (Fig. 6.1).Average monthly mortality due to predators was 1.5%, with predation bycougars being the most frequent (Table 6.1). Predation by cougars wasgreater (P = 0.012) in Caycuse and Chemainus than in Nanaimo or Nimpkish(Table 6.1). Predation by wolves was similar to that by cougars, however,we found no difference in predation by wolves among study areas (P > 0.614).Human-related mortality was 0.4% monthly, or half the mortality caused byeither cougars or wolves.a)a)II0I.0EzFigure 6.1. Monthly mortality by cause recorded for radio-collared, adult,female black-tailed deer on Vancouver Island, British Columbia, 1982-1991.Values above histograms indicate the total sample size in number of deer-months.14914121086420J FMAM JMonthJ ASONDTable6.1.Monthlycause-specificmortality(%)aforradio-collared,adult,femaleblack-taileddeerin4studyareasonVancouverIsland,BritishColumbia,1982-1991.StudyareasCaycuseChemainusNanaimoNimpkishTotalb11SEnNSEnNSEnNSEnNSECougar52.10•9A52.00•5A60.402B20.805B180.80•2CWolf10.40.420.80.3100.70.220.80.5150.7O.2Human31.30.710.40.450.30.2090.40.1Unknown10.40.420.80.630.20.1060.30.1Malnutrition10.40.4020.10.1030.10.1Accident10.40.4010.10.110.40.430.10.1aMaximumlikelihoodestimatesofmonthlymortality(%)werecomputedusinglogisticregression(CATMOD;SASInst.Inc.1985).Estimateswiththesameornoletteramongstudyareas,oramongcausesfortotals,arenotdifferent(P>0.10).bNo.ofdeerkilledbythespecificmortalitycause.I-.0,151Generally, monthly change in N was most evident in the leading causesof mortality (Table 6.2). Peaks in N occurred in February and from Aprilthrough July due to wolves, from March through May due to cougars, and inNovember due to humans.Factors Affecting Mortality and SurvivalThe simplest, acceptable model that appeared to explain 11 and . usedthe seasonal movement of deer as the independent variable (Table 6.3). Thatmodel was slightly more parsimonious (AIC = 624.02) than the model usingmovement and elevation (AIC = 640.72).Average monthly . was estimated to be 97.5% (SE = 0.3), or 74%annually. Survival was lower (P = 0.024) for resident deer (. = 97.8%, SE =0.4%, or 77% annually) than for migratory deer (. = 99.2%, SE = 0.3%, or 90%annually). Deer of unknown seasonal movement type had poorest survivaloverall (. = 71.2%, SE = 5.6%) but were insufficient in number to reverse ornullify the effect of seasonal movement. Had all unknown deer beenmigratory, behaviourial differences still would have been important (P =0.055).Some migratory deer, however, survived at rates more comparable withresident deer once elevation was considered. Although . for migratory deerat low elevations was not different (P = 0.991) from . for resident deer athigh elevations (Table 6.4), the opposite comparison (migratory deer at highelevations versus resident deer at low elevations) was significant (P =0.015). Survival did not differ between elevations for resident (P = 0.142)or for migratory (P = 0.184) deer.In August, September, and December through March, deer . rarelydropped below 99%, or 89% annually (Table 6.4). Survival for resident deer152Table 6.2. Mortality (%)a by month of year for 3 leading causes of mortalityfor radio-collared, adult, female black-tailed deer on Vancouver Island,British Columbia, 1982-1991.Deer Mortality causeMonth months Cougar Wolf Humanb N SE n II SE n 11 SEJan 160 1 0.6 0.5 1 0.6 0.5 0Feb 175 1 0.6 0.4 2 1.1 0.7 0Mar 192 3 1.6 0.8 1 0.5 0.5 1 0.5 0.4Apr 221 5 2.3 1.0 2 0.9 0.5 2 0.9 0.5May 211 4 1.9 0.9 3 1.4 0.8 0Jun 208 1 0.5 0.5 2 1.0 0.7 1 0.5 0.5Jul 170 1 0.6 0.4 2 1.2 0.7 0Aug 171 0 0 0Sep 169 0 0 1 0.6 0.4Oct 170 0 1 0.6 0.6 0Nov 171 1 0.6 0.6 0 4 2.3 1.2Dec 164 1 0.6 0.6 1 0.6 0.6 0a Maximum likelihood estimates of monthly mortality (%) were computedusing logistic regression (CATMOD; SAS Inst. Inc. 1985).b No. of deer killed by the specific mortality cause.153Table 6.3. Akaike’s Information Criterion (AIC) and likelihood-ratio tests (LR x)° between competingmodels of monthly fate (cause-specific mortality or survival) of radio-collared, aduLt, female black-tailed deer on Vancouver Island, British Columbia, 1982-1991. Hierarchical models of 4 categoricalvariables (A = study area, B = seasonal movement types of deer, E = monthly elevations used by deer, and N= month of the yr) are listed from the most general (all factors = A•B•E•M) to the most reduced model (nofactors = N).General AIC Reduced LR x2 df EA•B•E•M 720.19 A•B•E 71.33 66 0.3051A•B•M 7.98 12 0.7867B•EM 16.73 18 0.5417A•E•M 16.48 12 0.1702A.BE 659.52 A•B 7.39 12 0.8308A•E 15.54 12 0.2132BE 17.20 18 0.5094A•B•M 704.17 AB 70.74 66 0.3225A•M 73.28 12 0.0000B•M 16.56 18 0.5535B•E•M 700.92 B•E 71.80 66 0.2917B•M 7.81 12 0.7998E•M 15.18 12 0.2317A•E•M 712.67 A•E 70.39 66 0.3330A•M 64.78 12 0.0000E•M 15.43 18 0.6323A•B 642.91 A 82.12 12 0.0000B 17.11 18 0.5156A•E 651.06 A 73.97 12 0.0000E 16.17 18 0.5807B•E 640.72 B 7.30 12 0.8372E 14.51 12 0.2693A•M 753.45 A 79.58 66 0.1217N 22.18 18 0.2241Table 6.3. Continued.General AIC Reduced LR df PB.M 684.73 B 71.29 66 0.3063M 78.90 12 0.0000E•M 692.10 E 71.13 66 0.3110M 71.53 18 0.0000A 701.03 N 24.39 18 0.1427B 624.02 N 89.40 12 0.0000E 631.23 N 82.19 12 0.0000M 739.63 N 81.79 66 0.0910AIC indexes model parsimony (Chatfield 1992) and is calculated as (-2 X ln[max. likelihood] + 2x [no. of independent parameters]).Likelihood—ratio tests were (—2 X ln[max. likelihood reduced model])- (—2 X ln[max. likelihoodgeneral model]).154Table6.4.MonthLysurvival(%)‘for2knownand1unknownseasonalmovementtypesofradio-collared,adult,femaleblack-taileddeerat2broadelevationsinmabovesealevel(asl)onVancouverIsland,BritishColuthia,1982-1991.Monthlymodeofelevationsusedbydeer<600masl>600maslUnknownResidentMigratoryResidentMigratoryUnknownbSEnSEnSEn£SEn£SEJan6298.41.23599.20.73999.30.62299.60.4283.011.3Feb6597.91.33799.00.84499.10.72599.50.4481.210.6Mar7498.21.03599.10.63798.70.83499.50.41277.59.1Apr8796.61.43696.02.83797.71.24399.20.61859.29.5May7296.41.65097.21.73398.50.94699.10.61066.910.5Jun7397.41.34897.91.72598.70.95599.60.3774.510.7Jul6096.61.84398.80.92298.41.14299.50.4374.112.4Aug5999.01.14599.90.02399.40.84299.90.0292.89.7Sep6099.30.74199.90.02199.11.04599.90.0289.011.9Oct7298.51.23097.82.41999.70.44799.90.2284.213.4Nov8393.52.42299.50.43893.83.72699.70.3251.316.9Dec7398.51.12599.20.73799.30.62799.60.4283.811.2Maximumlikelihoodestimatesofmonthlysurvival(%)wereNo.ofdeerkilledbythespecificmortalitycause.computedusinglogisticregression(CATMOD;SASInst.Inc.1985).(Tic-fl156at low elevations from April through July, by comparison, was rarely >97%(73% annually). By comparison, & for migratory deer at low elevations was<97% only during April (Table 6.4).Except in November, survival in most winter months remained high(Table 6.4) and, consequently, we could demonstrate only a weak relationshipbetween . and mean monthly snow depth (P = 0.098). Mean monthly snow depthduring the study was 11 cm (n = 106, SD = 18 cm) and ranged 0-66 cm.DISCUSSIONCauses of MortalityNo estimates of cause-specific mortality rates have been published forblack-tailed deer. Our estimate of predation by wolves (0.7% mon.) was halfthe 17% annual mortality reported by Nelson and Mech (1986) for predation onadult, female white-tailed deer by wolves but greater than that reported forpredation on adult, female mule deer by coyotes (Canis latrans) (Hamlin andMackie 1989). The rate in our study is likely, in part, a reflection ofongoing removal of wolves (Janz 1989). Still, wolves and cougars wereprimary causes of mortality for our radio-collared deer (Table 6.1) and,together, created a mortality rate similar to that reported by Nelson andMech (1986) for wolves alone. Klein and Olson (1960) reported starvation asthe most frequent cause of mortality for Sitka black-tailed deer but mostmortality reported in their study came from areas having more severe winterweather than we had in our study. In the absence of severe winter weatherand relatively heavy predation by wolves, cougars, and humans (78% of alldeaths) it is not surprising that malnutrition was rarely observed.Cougars established activity centres, especially from March throughMay, and killed deer in isolated stands of old forests, most of which were157reserved as winter habitat for deer (see Janz 1989). Predation by cougarshas been considered unimportant (Janz and Hatter 1986), but our dataindicate they can have strong local effects that are intense in late wintermonths. Predation by cougars may have increased during the mid- to late-1980’s concomitant with removal of some wolves if predation wascompensatory. However, Hamlin and Mackie (1989) considered compensatorymortality unlikely in the adult, female segment of a mule deer population(average annual mortality was only 6.2%). Alternatively, we may simplynotice predation by cougars more now because kills occur in winter rangesthat, through time and continued forest harvesting, have become moreisolated in space.Wolves, by comparison, generally appeared to be less site specific andless seasonal in their kills even though most kills occurred during winterand spring. A notable lack of adult mortality by any cause existed in latesummer. Both predators likely concentrate on fawns rather than adults asprey during summer months (Scott and Shackleton 1980, Hatter 1988).Factors Affecting SurvivalEstimates of survival, excluding deer of unknown movement type, rangedfrom 73% annually for resident deer when they were at low elevations to 95%annually for migratory deer when they were at high elevations. Survivalrates for adult, female deer from other studies are similar to ourfindings: 78% for white-tailed deer in Montana (Dusek et al. 1989), 79% forwhite-tailed deer in Minnesota (Nelson and Mech 1986), 57-97.8% for muledeer in Montana (Hamlin and Mackie 1989), and 76-100% for mule deer inColorado (White and Bartmann 1983, White et al. 1987, Bartmann et al. 1992).We found that survival related more closely to seasonal movement of158deer, or to elevations used by deer, than to study area, month, or averagemonthly snow depth. Migratory deer exhibited highest S, which is especiallyrelevant considering timing and primary causes of mortality. Predators, themost dominant mortality agent, concentrated on adult, female deer fromFebruary through July when differences between survival of migratory andresident deer were greatest. Harestad (1979) found that most migratory deerdeparted winter ranges during March, coincident with the onset of highpredation rates. After March, migratory deer were at higher elevations, insteeper terrain, and in habitat with fewer roads compared with the habitatsof resident deer (Harestad 1979, Schoen and Kirchhoff 1985, McNay unpubi.data). We concluded that most migratory deer, either coincidentally orpurposefully, reduced risk of mortality due to predators by leaving lowelevation winter ranges as soon as they could in spring.Our findings contrast with those of Nelson and Mech (1991) for white-tailed deer in Minnesota. Their data showed that migratory deer sufferedgreater mortality due to high risk during fall migrations. Migrationdistances were 16 km in Minnesota but were 8 km for black-tailed deer onVancouver Island (Harestad 1979) or southeast Alaska (Schoen and Kirchhoff1985). Perhaps more important, migration is likely a more predictable eventin continental regions as a result of regular winter weather. In coastalclimates with more ephemeral winter weather and insular valleys, migrationis shorter and less predictable (Harestad 1979, Schoen and Kirchhoff 1985).In the former scenario, wolves may be more able to exploit the vulnerabilityof deer during migration.MANAGEMENT IMPLICATIONSAnnual survival rates for resident deer at low elevations (73%) were159unlikely adequate to sustain their populations. We constructed a simpleLeslie matrix (Leslie 1945) to assess recruitment necessary to stabilizesuch a population. We used age-specific productivity rates reported byThomas (1983), adult survival rates from this study, and held yearlingsurvival constant at 60% while varying fawn recruitment within reportedlimits from Janz (1989). At the upper level of reported recruitment (about25%), all but the low elevation resident population increased. To stabilizethat population with 27% adult mortality, it was necessary to have about 30%recruitment from the fawn population; a level rarely observed on VancouverIsland (Janz 1989). Black-tailed deer are not as fecund as otherconspecifics (Thomas 1983) and are therefore more sensitive to adultmortality (van Ballenberghe and Hanley 1984). Furthermore, recruitment intothe adult population is low in the presence of predation by wolves (Jonesand Mason 1983, Hatter 1988, Janz 1989). Although we could not estimatesubadult survival in this study, we judged it low. Of the 24 female, radiocollared, subadults only 1 fawn and 9 yearlings lived to become adults. Weconcluded that risk of mortality to adults at low elevations likelyoutweighed potential benefits in habitat quality (Gates 1968) derived fromthe early seral forests that follow forest-harvesting operations. Roadconstruction undoubtedly provides easy access throughout lower valleyelevations for wolves, cougars, and humans and, hence, direct access todeer.Forest harvesting also creates an isolation of old-forest winterranges and may directly influence predation by concentrating prey andfocusing predators’ attention to specific sites, especially during latewinter when deer are most vulnerable (Nelson and Mech 1986). Isolation ofwinter habitat seems particularly important because no migratory deer were160caught in young forests during winter. Although migratory deer had highannual survival (95%), we remain suspicious about the vitality of thatportion of the population. Their absence in young forests could imply 1 ora combination, of several processes: (1) migratory deer concentrateincreasingly in diminishing areas of old forest with eventual reduction oftheir survival due to deteriorating range condition, (2) mortality ofmigratory deer in subadult age classes is high, or (3) subadults abandonmigratory tactics in favour of resident habitat selection patterns. Theimplication of losing migratory deer could be a reduction in populationresiliency because the remaining resident deer are subject to comparativelyhigher mortality.In contrast to the weakening of rationale for retention of old-forestwinter habitat for deer because of declining deer populations (Janz andHatter 1986), we consider our results to indicate that a retention of older,intact forests is basic to rebuilding deer populations. Forest harvesting,hence, road building and spatial isolation of winter habitats, may intensifypredation on 1 segment of deer populations (resident deer) and indirectlyimpede recruitment to the other segment (migratory deer). The result likelycontributes to declines in deer populations and an overall loss ofpopulation resiliency.LITERATURE CITEDBartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatorymortality in a Colorado mule deer population. Wildl. Monogr. 121.39pp.Bunnell, F. L. 1985. Forestry and black-tailed deer: conflicts, crises,or cooperation. For. Chron. 61:180-184.Caughley, G. 1976. Wildlife management and the dynamics of ungulatepopulations. Appi. Biol. 1:183-246.161Chatfield, C. 1992.London, England.The analysis of time series. Chapman and Hall,241pp.Cheatum, E. L. 1949. Bone marrow as an index of malnutrition in deer. N.Y.State Conserv. 3:19-22.Clover, M. 1956. Single-gate deer trap. Calif. Fish and Game 42:199-201.Cowan, I. McT. 1956. Life and times of the coast black-tailed deer. Pages523-617 in W. P. Taylor, ed. The deer of North America. StackpoleBooks, Harrisburg, Penn.Dusek, G. L., R.Popul ationRiver, USA.Gates, B. R. 1968.coast forest.J. Mackie, J. D. Herriges, Jr., and B. B. Compton. 1989.ecology of white-tailed deer along the lower YellowstoneWildl Monogr. 104. 68pp.Deer food production in certain seral stages of theM.S. Thesis, Univ. British Columbia, Vancouver. lO5pp.Golley, F. B. 1957. An appraisal of ovarian analyses in determiningreproductive performance of black-tailed deer. J. Wildi. Manage.21:62-65.Hamlin, K. L., and R. J. Mackie. 1989.Breaks, Montana: a study of populenvironment. Montana Dep. Fish,Wildl. Restor. Final Rep., Proj.Harestad, A. S. 1979.Vancouver Island.98pp.Mule deer in the Missouri Riveration dynamics in a fluctuatingWildl., and Parks. Fed. Aid inW-120-R-7-18. Helena. 4Olpp._____and G. W. Jones. 1981. Use of night counts for censusing black-tailed deer on Vancouver Island. Pages 83-96 in F. W. Miller and A.Gunn, eds. Proc. symposium on census and inventory methods forpopulations and habitats. For., Wildl., and Range Exp. Stn. Contrib.217. Univ. Idaho, Moscow.J. A. Rochelle, and F. L. Bunnell. 1982. Old-growth forests andblack-tailed deer on Vancouver Island. Trans. North Am. Wildl. Nat.Resour. Conf. 47:343-352.Harthoon, A. M. 1977. Problems relating to capture. Anim. Regul. Stud.1:23-46.Hatter, I. W. 1988. Effects of wolf predation on recruitment of black-tailed deer on northeastern Vancouver Island. British ColumbiaMinist. of Environ. Wildi. Rep. R-23. Victoria. 82pp.and D. W. Janz. 1994. Apparent demographic changes in black-taileddeer associated with wolf control on northern Vancouver Island. Can.J. Zool. 72:878-884.Seasonal movements of black-tailed deer on northernPh.D. Thesis, Univ. British Columbia, Vancouver.Hebert, D. M., J. Youds, R. Davies, H. Langin, D. Janz, and G. W. Smith.1621982. Preliminary investigations of the Vancouver Island wolf (C. 1.crassodon) prey relationships. Pages 54-70 in F. H. Harrington and P.C. Paquet, eds. Wolves of the world. Noyes Publ. Park Ridge, N.J.Janz, D. W. 1989. Wolf-deer interactions on Vancouver Island: a review.Pages 26-42 in D. Seip, S. Pettigrew, and R. Archibald, eds. Wolf-prey dynamics and management. British Columbia Minist. of Environ.Wildl. Rep. WR-40. Victoria._____and I. W. Hatter. 1986. A rationale for wolf control in themanagement of the Vancouver Island predator-ungulate system. BritishColumbia Minist. of Environ. Wildl. Bull. B-45. Victoria. 35pp.Jones, G. W., and B. Mason. 1983. Relationships among wolves, hunting,and population trends of black-tailed deer in the Nimpkish valley onVancouver Island. British Columbia Minist. of Environ. Wildl. Rep. R7. Victoria. 26pp.Klein, D. R., and S. T. Olson. 1960. Natural mortality patterns of deerin southeast Alaska. J. Wildl. Manage. 24:80-88.Leslie, P. H. 1945. On the use of matrices in certain populationmathematics. Biometrika 33:183-212.McNay, R. S., and D. D. Doyle. 1990. The Integrated Wildlife-IntensiveForestry Research (IWIFR) program deer project. Northwest Environ.6:365-366.Mech, L. D. 1977. Wolf pack buffer zones as prey reservoirs. Science198:310-321.Meidinger, D., and J. Pojar. 1991. Ecosystems of British Columbia.British Columbia Minist. of For. Special Rep. Ser. 6. Victoria.330pp.Messier, F., and C. Barrette. 1985. The efficiency of yarding behaviourby white-tailed deer as an antipredator strategy. Can. J. Zool.63:785-789.Nelson, M. E., and L. D. Mech. 1986. Mortality of white-tailed deer innortheastern Minnesota. J. Wildl. Manage. 50:691-698.and_____. 1991. Wolf predation risk associated with white-taileddeer movements. Can. J. Zool. 69:2696-2699.Robinette, W. L., 0. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Noteson tooth development and wear for Rocky Mountain mule deer. J. Wildl.Manage. 21:134-153.Roy, L. D., and M. J. Dorrance. 1976. Methods of investigating predationof domestic livestock. Alberta Agric., Edmonton. 54pp.SAS Institute Inc. 1985. SAS user’s guide: statistics. Version 5. SASInst. Inc., Cary, N.C. 956pp.163Schoen, J. W., and N. D. Kirchhoff. 1985. Seasonal distribution and home-range patterns of Sitka black-tailed deer on Admiralty Island,southeast Alaska. J. Wild]. Manage. 49:96-103.Scott, B. N., and D. N. Shackleton. 1980. Food habits of two VancouverIsland wolf packs: a preliminary study. Can. J. Zool. 58:1203-1207.Smith, I. D. 1968. The effects of hunting and seral succession uponVancouver Island black-tailed deer. M.S. Thesis, Univ. BritishColumbia, Vancouver. l4Opp.Taber, R. D. 1953. Studies of black-tailed deer reproduction on threechaparral cover types. Calif. Fish and Game 39:177-186.Thomas, D. C. 1970. The ovary, reproduction, and productivity of femaleColumbian black-tailed deer. Ph.D. Thesis, Univ. British Columbia,Vancouver. 2llpp._____1983. Age-specific fertility of female Columbian black-tailed deer.J. Wildl. Manage. 47:501-506.and I. M. Cowan. 1975. The pattern of reproduction in femaleColumbian black-tailed deer, Odocoileus hemionus columbianus. J.Reprod. Fert. 44:261-272.van Ballenberghe, V., and T. A. Hanley. 1984. Predation on deer inrelation to old-growth forest management in southeastern Alaska.Pages 291-296 in W. R. Meehan, T. R. Merrell, Jr., and T. A. Hanley,eds. Fish and wildlife relationships in old-growth forests. Am. Inst.Fish. Res. Biol., Morehead City, N.C.White, G. C., and R. M. Bartmann. 1983. Estimation of survival rates fromband recoveries of mule deer in Colorado. J. Wild]. Manage. 47:506-511.and R. A. Garrott. 1990. Analysis of Wildlife radio-tracking data.Academic Press, San Diego, Calif. 383pp._____R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987.Survival of mule deer in northwest Colorado. J. Wildl. Manage.51:852-859.164CHAPTER 7 - MANAGEMENT IMPLICATIONS OF CONSTRAINTS ON MOVEMENT TACTICS3Prior to 1970, research on black-tailed deer was restricted largely toareas of low snowfall. Findings suggested that deer populations respondedpositively to forage increases generated by timber harvesting and forestrywas assumed beneficial to deer (Bunnell 1985). Initial research in areas ofhigher snowfall documented contrary results; old forests were found to bebeneficial to black-tailed deer (Walimo and Schoen 1980, Bunnell and Jones1984, Bunnell 1985). Potential conflict between forest harvesting andmaintenance of black-tailed deer populations was exposed. That conflicthelped inspire 2 large research efforts: the Integrated Wildlife- IntensiveForestry Research program (IWIFR) and Managed Stands for Deer Winter Range(MSDWR). Both were designed to guide the type and distribution of forestrypractices that would maintain deer populations. Black-tailed deer, however,are relatively plastic and, among their apparent adaptations, show differentmovement patterns that initial analyses suggested were related to broadhabitat features (Bunnell 1990). For management to be effective thesepatterns had to be documented well enough that responses to habitatalterations could be predicted under different conditions. This paperintegrates results of several studies to document how behaviourial featuresof black-tailed deer limit their use of habitats and their responses torapid alterations of habitat. It documents general movement patterns ofblack-tailed deer on Vancouver Island, relates these movements to broadenvironmental features including local climate and topography and to largescale habitat alteration, and notes how they interact with major causes ofmortality.Published as: McNay, R. S., and F. L. Bunnell. 1994. Behaviouriallimits to movement: the effect on habitat choices for Columbian black-taileddeer. Trans. Congr. Tnt. Union Game Biol. 21(2):295-303.165STUDY AREASWe studied movements, habitat use, and survival of black-tailed deerat 4 locations on Vancouver Island, British Columbia, 1982-1991. We alsostudied individual responses to logging winter habitat at 2 of these studyareas, 1989-1991. The most northerly study site was located at the NimpkishRiver valley (50°08’N, 126°30’W). The Nimpkish River flows primarilynorthwest (315°) giving the general study site a southern aspect and themain valley ranges in elevation from about 200 m to 1,821 m asl. TheNanaimo River (49°02’N, 124°12’W) and the Chemainus River (48°56’N,124°05’W) are neighbouring valleys on southeastern Vancouver Island. Bothwatersheds have a considerable southern exposure; the Nanaimo River flows ina direction of 500 while the Chemainus River flows more to the south at125°. These watersheds range in elevation from about 300 m to 1,540 m asl.The last and most maritime study site was located at the Caycuse River(48°48’N, 124°30’W) on southwestern Vancouver Island. The Caycuse Riverflows due west but has many sub-drainages flowing south or north. Thisvalley has the least area of flat valley bottom and ranges least inelevation from about 300 m to 1,249 m asl.The dominant ecosystem on Vancouver Island is the Coastal WesternHemlock (CWH) zone while the less prevalent Mountain Hemlock (MH) zoneexists at high elevations (Meidinger and Pojar 1991). While theseecosystems were represented at each study site, the Caycuse River valley hadonly minor amounts of MH. Extensive harvest of trees had occurred at eachstudy site but the resulting mosaic of forest seral-age classes differedamong sites. Generally, the Chenialnus River site had the least amount ofarea remaining in old (>250-yr-old) forests (6%) and the most area in young(6- to 45-yr-old) forests (82%). The Caycuse and Nimpkish River sites were166similar in that each had about 40% of their area in old forest and 30% inyoung forest. The Nanairno River site was represented by about 30% oldforest and 50% young forest. Each study site had from 10 to 20% of theirarea in deforested (0- to 5-yr-old) clearcuts.Climate on Vancouver Island is characterized by temperate, wetweather. In the CWH zone, no month has a mean temperature <0 C and the meantemperature of the warmest month is 17 C. Usually there are 291 frost freedays, an average of 820 mm of snow, and a mean precipitation of 2,140 mmeach year (Meidinger and Pojar 1991).METHODSDeer Location SamplesDeer were captured using Clover traps (Clover 1956), or byimmobilization using succinylcholine, and collared with radio transmitters(McNay and Voller 1995). Collaring took place at Nanaimo River from 1982 to1988, at Chemainus River in 1989, and at both the Caycuse and the NimpkishRivers in 1989 and 1990.Radio-collared deer were located by triangulation (White and Garrott1990) using >2 bearings, each recorded at separate and permanent recordingstations marked at 100-rn intervals along roads. Bearings were usuallycollected in <10 mm at sites that were line-of-sight with, and close to(<400 m), the radio-transmitter being monitored. Samples were generallytaken once-per-week on each collared deer. In 1984 the sampling schedulewas standardized so that, during a calendar month, each deer was located atleast once-per-week and once within each quarter of a calendar day. Deerlocations were estimated using the maximum likelihood estimator (Lenth 1981)presented in a SAS program (SAS Inst. Inc. 1985) by White and Garrott.167(1990:64).Definitions and Analytical ProceduresMovements and Habitat Use.- To document general movement patterns andhabitat use, we examined consecutive locations of individual, radio-collareddeer to find the magnitude (distance), frequency, timing, direction, andhabitat choices resulting from movements. Natal ranges were defined asthose areas occupied during the natal period, which for deer in coastalBritish Columbia is anywhere from late-May throughout June (Cowan 1956,Thomas 1970). We assumed that the natal range was close to the reallocation of birth (Masters and Sage 1985, McCullough 1985, Hanilin and Mackie1989) and would be where offspring were produced. Spatially separate rangesoccupied at other times were identified by migrations made to go betweenthem and the natal range and were termed alternate ranges. Migrations anddispersal differed in that migrations had predictable return migrations tothe original position (Sinclair 1984) while dispersals did not (Howard 1960,Bunnell and Harestad 1983). Migratory deer were termed obligate migratorydeer if they exhibited consistent annual migrations and facultativemigratory deer if migrations occurred irregularly. Non-migratory deer weretermed residents.Forest habitats were classified as: old (>250-yr-old), young (6- to45-yr-old), open (unforested, 0- to 5-yr-old forests), or non-merchantable(rock, water, subalpine, and alpine). No forests existed between 46- to250-yr-old. Elevations above sea level (asl) were considered in 200-rn bands(201-400 m, 401-600 m, etc.) except in the study of deer survival whereelevations were considered in 2 strata; above and below 600 m.All distances and directions were measured as a straight line, or168bearing, from the last recorded location to the current location. Departureand arrival dates from one range to another, or from one location toanother, were taken to be the half-way point between the date last observedto the date of the current observation.We assessed independence of location data in a related study (McNay etal. 1994) and assessed the distributional properties of circular data (e.g.,dates and directions) using Raleigh’s z test (Batschelet 1981). Theproportion of migrations initiated with or without snow was tested formovement type (obligate or facultative) effects using Fisher’s exact test(Sokal and Rohif 1981). Fidelity to natal ranges was tested for movementtype effects using the straight-line distance between arithmetic centres ofconsecutively used natal ranges in an F-test (Sokal and Rohlf 1981).Comparisons of habitat use between movement and range types were made with x2tests (Sokal and Rohif 1981).Winter Range Removal. - We assessed responses by both control and treatmentdeer to clear-cut logging of winter habitat. For each deer, we comparedpre- and post-disturbance: (1) fidelity to winter ranges, (2) winter rangesizes, and (3) use of old forests. Candidate stands of old forest that metcharacteristics of typical winter habitat (e.g., Bunnell 1985, Nyberg et al.1986) were selected for harvest at the Chemainus and Nimpkish river sites(for detail see chapter 5). These stands were isolated from stands ofsimilar characteristics and were to be entirely clear-cut logged. Trapswere placed within the selected stands of old forest to trap treatment deerand at sites within nearby stands to trap control deer. Control deer forhabitat use comparisons were monitored at Nanaimo River. Analysis treatedonly data for the winter season (November through April) prior to, anddirectly following, logging (i.e., 1 yr pre- and post- disturbance). We169omitted migrations and data for migratory deer if they were on their natalranges during winter months. We defined fidelity as overlap in pre- andpost-disturbance winter ranges and defined continuity of habitat choice asinsignificant (x2 I P > 0.05) differences in the use of old forests duringwinter. Home ranges were 95% minimum area polygons using data from winter.For statistical convenience we constructed an index of fidelity based on atest of independence between consecutive winter ranges (White and Garrott1990:136). Positioned on the pre-disturbance range centre, we usedconcentric circles with radii iteratively reduced by 100 m, starting at aradius of 1,500 m, to find the first test where deer locations in the post-disturbance range were considered to be independently distributed (x2 I P <0.05). This radius was termed the maximum radius for independence (MRI) andsmall MRI values indicated strong fidelity. Habitat use was the percent oflocation observations in old forests situated below, or at the sameelevations as, the cut-block. Old forests above the cut-block did not meetcriteria typical of deer winter habitat (Nyberg et al. 1986) and wereconsidered a poor alternative to the logged habitat. Habitat use data weretransformed using an arcsine transformation for percentage data (Zar 1984).We used a repeated measures (pre- and post-disturbance), analysis ofvariance to test for the potential effect of logging (control versustreatment deer) nested within study areas (Chemainus or Nimpkish). Allreported means are least-squares estimates (Searle et al. 1980).Survival Estimation. - Deer were aged at the time of collaring and followeduntil death at which time the cause of mortality was determined (for detailssee McNay and Voller 1995). We used logistic regression (SAS Inst. Inc.1985) to compute maximum likelihood estimates for monthly mortality (H) andsurvival () of adult, female deer, while simultaneously assessing the170effects of 5 variables. Four of the independent variables were categorical:(1) study area had 4 levels (Caycuse, Chemainus, Nanaimo, and Nimpkish), (2)seasonal movement types of deer had 3 levels (migratory, resident, orunknown), (3) average elevations used (on a monthly basis) had 3 levels(<600 m asl, >600 in asl, and unknown), and (4) month had 12 levelscorresponding to each month of the year. The fifth variable was meanmonthly snow depth measured daily at the closest airport to each study area(Atmos. Environ. Serv., Environ. Can., Vancouver, BC). We conducted testsamong individual survival estimates, or between estimates of N, using theCONTRAST statement in PROC CATMOD (SAS Inst. Inc. 1985).RESULTS AND DISCUSSIONAnalysis treated 126 deer collared at 4 study areas (119 females and 7males) monitored for a total 76,693 deer-days. The majority of the deerwere collared as adults at Nanaimo River (Table 7.1). One deer that made asingle abnormal move just before death and 51 deer for which we had <10 moof data could not be classified by seasonal movement type and were omittedfrom analysis of movements and habitat use (Table 7.1). We collected atotal of 8,624 weekly locations on the remaining 74 deer (Table 7.1). Only20 adult, female deer were alive long enough to be considered in theresponse to removal of winter habitat and all adult females were treated inthe analysis of monthly survival rates (n = 105).MovementsThe only recorded dispersal movements were made by 2 female deercollared as juveniles at Nimpkish River. Both deer left the collaring sitetogether late in June and spent much of the summer at a site 7 km to theTable7.1.Thesex,ageclassesa,andnumberofrelocationsmadeforasampleofradio-collared,black-taileddeerat4studysitesonVancouverIsland,BritishColumbia.Superscriptedvaluesarethenumberofdeerthatlivedlongenough(>10mo)toclassifyintoseasonalmovementtypes.AdultYearlingFawnTotalNumberofStudySited9d9d9d9LocationsbCaycuse024(10)115(2)2(1)6(2)27(11)1582Chemainus0150104(1)020(8)784Nanainio1(1)44(32)0051(1)57(39)951Nimpkish012(10)02(2)01(1)015(13)5307aAgeclasses,determinedold),oradult(>2-yr-old).attimeofcollaring,were:fawn(<1-yr-old),yearling(between1-and2-yr-Locationsareforsuperscripteddeeronly.bI-.172east. During the fall they began moving again and travelled another 25 kmeast before 1 of the 2 was killed in a collision with a vehicle. Theremaining disperser continued travelling only to return and settle at thesame location it had spent the previous summer. No further indication ofdispersal or migration was recorded. Dispersal distances were longer thatmost migratory moves (e.g., McNay and Doyle 1987).Migratory moves (n = 202) occurred at all study sites with movementsto alternate ranges occurring at October 20 (z = 24.33; n = 38; p < 0.05)for obligate migratory deer and December 10 (z = 20.19; n = 56; P < 0.05)for facultative migratory deer. Return migrations to the natal rangesgenerally occurred before the birth of fawns for both types of migratorydeer; about May 26 (z = 35.43; n = 45; p < 0.05) for obligate migratory deerand about February 21 (z = 10.03; n = 63; P > 0.05) for facultativemigratory deer. Obligate deer travelled to alternate ranges every year,leaving before snowfalls began (only 2 of 38 moves were in snow) andreturning after snow at low elevations had melted (0 of 45 moves were insnow). Facultative migratory deer, by comparison, usually departed foralternate ranges only after snowpacks had already accumulated on natal areas(42 of 56 moves were in snow) and returned before snow ablation hadcompleted (16 of 63 moves were in snow). The proportion of time that moveswere made in snow were significantly different (Fisher’s P < 0.001; n = 94leaving, and P = 0.016; n = 108 returning to, the natal range) between the 2groups. If no snow fell, facultative migratory deer usually did not leavetheir natal range. Two exceptions occurred; 1 deer at both Nanaimo andChemainus left the natal range before winter (October) each year for short(<3 wk) intervals. Obligate deer spent >6 mo ( = 209 d; n = 34, SE = 8.77)away from the natal range each year while facultative deer spent <3 mo away173( = 66 d; n = 33, SE = 10.05). Obligate migratory deer showed noconsistent directional tendencies in their movements; facultative migratorydeer usually followed the direction of the valley (chapter 3).Facultative deer appeared reluctant to leave the natal area and did soonly if snowfall modified their habitat. These deer had established theirnatal areas at mid- to high-elevations (Fig. 7.1) where snowpacks wereephemeral. Obligate migratory deer established their natal ranges at higherelevations (Fig. 71) and appeared to move in anticipation of inclementwinter weather. That observation is consistent with the proportion ofbehaviour types observed at each study site. For example, the Caycuse areahad the most maritime climate, the lowest maximum elevation, and the leastarea of flat valley bottom; it had no obligate migratory deer, only 1resident deer, and 7 facultative migratory deer. Nanaimo River, bycomparison, was the most variable study site with both low elevationhabitats that were free of snow in some years through to mountainoussubalpine habitats that received at least some snow every year. Our sampleof deer there consisted of 23 resident deer, 7 facultative migratory deer,and 10 obligate migratory deer. Studies from other locations support thenotion that tactics for seasonal movements depend on local topography(Kufeld et al. 1989, Brown 1992) and can be initiated by specific weatherrelated phenomena (Mccullough 1964, Loft et al. 1989), snow depth (Richens1967, Gilbert et a!. 1970), or the condition of seasonal forage (Garrott etal. 1987). We concluded that local topography and climate at the natalrange provide a continuum in the need to use alternate ranges. Whereinclement winter weather can be expected in most years, deer migrateannually to alternate ranges. At lower elevations, where snowpacks areephemeral, deer are afforded the opportunity to leave the natal area only17470Facultative Obligate • Resident6000.;50400Cooo30200l:IH i.2-4 4-6 6-8 8-10 10-12Elevation bands (m asl * 100)Figure 7.1 Percent of total locations for radio-collared, black-tailed deerof 3 movement types (obligate migratory, facultative migratory, or resident)found at different elevation bands (m asl) on Vancouver Island, BritishColumbia, 1982-1991.175when necessary. Finally, at lowest elevations deer are resident becauseinclement winter weather occurs infrequently and no topographic opportunityexists to provide any escape from inclement winter weather when it doesoccur.Similar to other studies (Linsdale and Tomich 1953, Bunnell andHarestad 1983, Garrott et al. 1987, Hamlin and Mackie 1989), dispersal wasobserved infrequently. We therefore concluded that establishment of a natalrange to be dependent on, or at least associated with, the location of thematernal natal range. That has been observed to be the case in at least 2other studies (Hirth 1977, Hamlin and Mackie 1989) and we believe it to bethe case in our study. If that is the case, and provided that the need tomigrate is dependent on local topography and climate (above), then offspringshould adopt similar migration tactics as their mothers and ranges should beused with a high degree of fidelity.Fidelity to the natal area was virtually unvarying among periods ofuse ( = 0.3 km; n = 66, SE = 0.3) and we could not detect any effect due toseasonal movement type (P > 0.05). Other studies report similar results(Garrott et a!. 1987). Early mortality precluded unequivocal documentationof fawns’ migration tactics relative to their mothers but, based onobservations of surviving fawns, we believe fawns mostly adopted theirmother’s tactics for selecting seasonal ranges. Although we could find noconfirmation for deer, Sweanor and Sandegren (1988) reported an exactcorrelation between movements of offspring and movements of mothers in apartially migratory population of moose.We conclude that the need to select an alternate range depended on thelocal topographic and climatic conditions of the maternal natal range.Dispersal could act to reset natal range establishment but that would only176occur infrequently.Habitat UseDeer used young forests 65-75% of the time at all sites except atCaycuse River where deer used this type about 45% of the time; the balanceof locations at Caycuse River occurred in open habitats (Fig. 7.2). Use ofopen habitats elsewhere ranged from 7% to 15%. Use of old forests rangedfrom 7% at Caycuse to 14% at Nimpkish.Habitat use differed (x2 = 410.8; df = 3; p < 0.001) between natal andalternate ranges primarily in a switch of seral age classes (Fig. 7.3).Natal ranges were characterized by less use of old forests (9% versus 29%)and greater use of open habitats (21% versus 12%). This difference wasclearer to us when we considered the frequency of deer using these habitatson each range type. All 12 obligate migratory deer, and 10 of 16facultative migratory deer, used old forests while on alternate ranges (Fig.7.3). Resident deer, by comparison, did not have as much access to oldforests and only 25 of 44 resident deer used it. Of 6 facultative deer thatdid not use old forests, 2 migrated outside the winter season (October) inthe absence of snow, apparently for reasons other than to find winterhabitat. The remaining 4 facultative deer never had access to old forestsbecause none was available along their restricted path of movement, down inelevation and out the main valley. Furthermore, of the 39 deer that weretrapped, all those trapped in young forests were resident deer (n = 11) butless than half of those trapped in old forests were resident.We conclude that old forests were the preferred habitat when inclementwinter weather occurred. Summer habitats were composed primarily of youngforests and open habitat combinations. Resident deer, being lower inCu C) 1 Cu >1•0 Cl) Cl) C 0 4.d Cu C.) 0 Cu 0 0 C C) C) C) aFigure7.2HabitatschosenbyBritishColumbia,1982-1991.water,subalpine,andalpine;radio-collared,black-taileddeerat4studysitesonVancouverIsland,Habitatsare:open,0-to5-yr-oldforests;non-merchantable(NMF)rock,young,6-to45-yr-oldforests;orold,>250-yr-oldforests.—4100 80 60 40 20 0ClearYoungOldNMFHabitat types860-T:40-Cu 2O-4+4•+_____________________::.+C.2-SEIMeang•:•::•:::<.:::::•::•:::::•:•:::•::•::•:•:•:•::::::•A+SETAFacultativeISI•ObligateC40—.::;.;IResidenttiI t+0OpenYoungOldNMFHabitat typesFigure7.3Habitatsusedbyradio-collared,black-taileddeerof3movementtypes(obligateorfacultativemigratory,orresident)atnatalranges(top)andatalternate,winterranges(bottom)onVancouverIsland,BritishColumbia,1982-1991.Habitatswerenon-merchantable(NMF)rock,water,subalpine,oralpineorforestsofages:open,0-to5-yr-old;young,6-to45-yr-old;orold,>250-yr-old.cc179elevation than other deer and consequently in a zone of greater forestharvest, used primarily young habitats year-round and only where remnantstands of old forests occurred did they use that habitat type. Thoseresults are similar to those of Yeo and Peek (1992) in south-east Alaska.We conclude again that the mothers’ selection of natal range limitssubsequent breadth of habitat availability on alternate ranges duringinclement winter weather.Response to Removal of Winter HabitatThe most complete test of deer response to removal of their winterhabitat occurred at Nimpkish River because the stand of old forest (47 ha)was completely logged as scheduled in the fall of 1990. A stand of oldforest remained untouched by logging about 300 m below the bottom of thecut-block at Nimpkish River. At Chenialnus River, the stand of old forest(90 ha) was partially logged (30 ha each) during the spring of 1989 leaving60 ha of old forest adjacent to the cut-block. We found deer to retainstrong fidelity to their winter ranges at both study areas (Fig. 7.4) withno indication that logging altered this attraction to sites (F = 0.59; df =3; p = 0.639). Winter ranges were larger after logging than before logging(F = 6.90; df = 1; P = 0.018) but there was again no evidence of an effectdue to logging (F = 0.98; df = 3; p = 0.425) between treatment and controldeer (Fig. 7.4). Finally, use of old forests was lowest during post-loggingmonitoring (F = 29.05; df = I.; P = 0.002) but only the treatment at NimpkishRiver had an effect on use of old forests (F = 5.91; df = 2; P = 0.018).Our observations of deer response to logging are preliminary in thatthey come from only one post-logging year. Nevertheless, results show thatfidelity to site is a strong factor involved in habitat use followingUpperse•ChemainusRiverPre-disturbanceControl deer•Mean•NimpkishRiverILowerseANanaimoRiverPost-disturbanceTreatment deer100Otocu¶-.-120Ø0QA.CaoC.................:....-.....-.•.•....:......I I:::-:;•::::::::C.0I00560•ii...:....:.:.:....I:.:.:.:...:...:.:.::::I::.9i•Tc2OXG)c0 00....................................................................________________Figure7.4Theresponseofradio-collaredblack-taileddeertoremovaloftheiroldforest,winterhabitatat2studyareasonVancouverIsland,BritishColumbia,1988-1991.I-. co181logging. Even though old forest was still available closeby at NimpkishRiver, deer never left their winter ranges to access that type and as aresult would have adjusted their habitat use in proportion to the amount ofold forest lost through logging. Deer at Chemainus River, however, stillhad old forest left adjacent to the cut-block and appeared to adjust theirwinter range sizes (but only marginally) to use that habitat type. Similarresults were obtained by Gasaway et al. (1989) when they studied response bymoose to habitat improvements; only moose having previous experience in, oradjacent to, the manipulated area increased their use of that area aftermanipulation. Others found no tendency for deer (Hood and Inglis 1974) orelk (Hershey and Ledge 1980, Edge et a!. 1985) to move away from habitatdisturbances that were considered to be detrimental.Survival EstimatesOf the 12 female yearlings and 12 female fawns that we collared, 9yearlings and 1 fawn aged into the adult cohort and were added to the sampleof 95 deer collared as adult females (Table 7.1). We recorded mortalitiesfrom about half of that sample (54 deaths and 4 collar failures of 105collared deer) most of which occurred between February and June. Predationwas the most frequent cause of death with an average monthly mortality of1.5% (McNay and Voller 1995). Human-related mortality occurred at 0.4%monthly and most frequently during the hunting season in November.The simplest, acceptable model that explained survival used theseasonal movement of deer as the independent variable (McNay and Voller1995). A slightly better, but less efficient, model used both seasonalmovement type and elevation. Average monthly survival was 97.5% (SE = 0.3),or 74% annually. Survival was lower (P = 0.024) for resident deer (97.8%,182SE = 0.4%) than for migratory deer (99.2%, SE = 0.3%). Although survivalfor migratory deer at low elevations was not different (P = 0.991) from thatfor resident deer at high elevations (McNay and Voller 1995; Table 4), theopposite comparison (migratory deer at high elevations versus resident deerat low elevations) was significant (P = 0.015). In August, September, andDecember through March, deer survival rarely dropped below 99%, or 89%annually. Survival for resident deer at low elevations from April throughJuly, by comparison, was rarely >97% (73% annually). By comparison,survival for migratory deer at low elevations was <97% only during April(Table 6.4).We concluded that obligate migratory deer could have beencoincidentally or purposefully in transit to, or from, their natal rangesduring the periods of highest mortality which tended to be due to predatorsand concentrated at lowest elevations (McNay and Voller 1995). Thatconclusion contrasts the results found by Nelson and Mech (1986) for white-tailed deer in Minnesota where migratory deer had lower annual survivalcompared to residents. The overall annual survival rates comparedfavourably with those reported for white-tailed deer in Montana (Dusek etal. 1989) and mule deer in Colorado (White and Bartmann 1983).MANAGEMENT IMPLICATIONSWe draw 3 major management implications from these findings. Thefirst concerns the interaction of decisions about resource use by blacktailed deer and their loyalty to specific movement tactics (chapter 4).Because decisions about resource use create constraints that tend to cascadedown through an hierarchy of decisions (e.g., home range choices constrainseasonal range selections), and because deer show a consistency of movement183patterns (e.g., philopatry, fidelity to seasonal ranges), their habitatchoices and movements will not be easily changed or new tactics developed.This stands as a major constraint on habitat selection whenever habitats arealtered in any rapid and extensive manner (e.g., wildfire or logging). Theconstraint, in turn, presents a lack of freedom in choosing habitats;freedom-of-choice being the fundamental assumption in the ideal -freedistribution hypothesis of habitat selection (sensu Fretwell 1972). Thatis, deer will not behave as though omniscient with full knowledge of theirsurroundings, selecting and filling the best habitat first before occupyingless favourable habitat. Responses to changes in their habitat will lag intime by some unknown amount, possibly generations. We had some indicationof this in the experiment that involved removal of winter habitat. Thatdelay has serious implications to researchers and managers attempting tointerpret patterns of deer habitat use. It is unclear when and what tomeasure as a response to changes in habitat, such as those induced byforestry practices. Interpretation of use/availability measures isobscured, evaluation of habitat is obscured (Hobbs and Hanley 1990), andresearcher credibility is potentially undermined. For example, costlyefforts have been undertaken to create black-tailed deer winter range inmanaged stands (e.g., Bunnell 1985, Nyberg et al. 1986). If deer respond tothose efforts with no more alacrity than they responded to winter rangeremoval, efforts to create winter range may be successful without an obviousresponse by deer. Conversely, deer may remain in areas where somemanagement action has created unfavourable habitat.The second major implication concerns old forests. Black-tailed deeruse young forests and appear to do reasonably well there (e.g., Fig. 7.3).Our findings here, though, corroborate those of other researchers (see184reviews of Bunnell and Jones 1984, Bunnell 1985, 1990) that deer using oldforests do best. When snow is on the ground only resident deer are found inyoung forests. Facultative and obligate migratory deer more often attainbetter habitat in snowy winters and also show higher annual rates ofsurvival. It is difficult to predict the consequences as more old forest isharvested from lower elevations; much of it is already gone. Ourobservations suggest that as old forest winter habitat is removed there willbe fewer and fewer migratory deer until only 1 movement tactic, the leastsuccessful over long periods, is retained. We therefore expect resilienceof deer populations to decline as the amount of old forest winter habitatdeclines. Deer currently present in young forests during winter may reflectonly that portion of the initial population that is non-migratory.Tactics for maintaining wildlife in forested ecosystems involve bothstand prescriptions and watershed or landscape level planning (Bunnell andKremsater 1993). The seasonal movement types reported here have clearimplications to landscape level planning. For example, it is apparent thatfacultative migratory deer move down and out of valleys. Similarly,obligate migratory deer select winter ranges on south-facing, old forests atmid- to high elevations (e.g., Fig. 7.1). Reserves of old forests toprovide winter ranges, and management of young forests to imitate winterrange in managed stands (e.g., Nyberg et a!. 1986), should be few but largeon southern aspects at mid-elevations. That approach should accommodateboth facultative and obligate migratory deer (who will eventually find theareas) and reduce the effects of predators, such as wolves and cougars,which likely concentrate their efforts in small areas. Because residentdeer move little and occur mostly at low elevations, reserves andsilvicultural treatments at those elevations should be spread over many,185smaller areas. Combined, this distribution of management actions across thelandscape will maintain all 3 seasonal movement types and resilience withindeer populations.LITERATURE CITEDBatschelet, E. 1981.37lpp.Circular statistics in biology. Academic Press, N.Y.Brown, C. G. 1992. Movement and migration patterns of mule deer insoutheastern Idaho. J. Wild]. Manage. 56:246-253.Bunnell, F. L. 1985. Forestrycooperation. For. Chron.and black-tailed deer: conflicts, crises, or61:180-184_____1990. Black-tailed deer ecology and forest management. Pages 31-63in J. B. Nyberg and D. W. Janz, eds. Deer and elk habitats in coastalforests of southern British Columbia: a handbook for forest andwildlife managers. British Columbia Minist. For. Special Rep. Ser. 5.Victoria.and A. S. Harestad. 1983. Dispersal and dispersion of black-taileddeer: models and observations. J. Mamrn. 64:201-209.and G. W.synthesis.A. Hanley,old-growthJones. 1984. Black-tailed der and old-growth forests - aPages 411-420 in W. R. Meehan, T. R. Merrell, Jr., and T.eds. Proc. symposium on fish and wildlife relations inforests. Am. Inst. Fish. Res.and L. L. Kremsater. 1993. Tactics for maintaining biodiversity inforested ecosystems (this proceedings).Clover, M. 1956. Single-gate deer trap. Calif. Fish and Game 42:199-201.Edge, W. D., C. L. Marcum, and S. L. Olson. 1985. Effects of loggingactivities on home-range fidelity of elk. J. Wild]. Manage. 49:741-744.Fretwell, S. D. 1972. Populations in a seasonal environment. PrincetonUniv. Press, Princeton, N.J. 217pp.Cowan, I. McT. 1956.523-617 in W. P.Co., Harrisburg,Dusek, G. L., R.Popul ati onRiver, USA.Life and times of the coast black-tailed deer. PagesTaylor, ed. The deer of North America. StackpolePenn.J. Mackie, J. D. Herriges, Jr., and B. B. Compton. 1989.ecology of white-tailed deer along the lower YellowstoneWildl. Monogr. 104. 68pp.Garrott, R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A. W.186Alidredge. 1987. Movements of female mule deer in NorthwestColorado. J. Wildl. Manage. 51:634-643.Gasaway, W. C., S. D. Dubois, R. D. Boertje, D.J . Reed, and D. T. Simpson.1989. Response of radio-collared moose to a large burn in centralAlaska. Can. J. Zool. 67:325-329.Gilbert, P. F., 0. C. Walimo, and R. B. Gill. 1970. Effect of snow depthon mule deer in Middle Park, Colorado. J. Wildl. Manage. 34:15-23.Hanilin, K. L. and R. J. Mackie. 1989. Mule deer in the Missouri RiverBreaks, Montana: A study of population dynamics in a fluctuatingenvironment. Montana Dept. of Fish and Wildl., Missoula. 4Olpp.Hershey, T. J. and T. A. Leege. 1982. Elk movements and habitat use on amanaged forest in north-central Idaho. Idaho Dep. Fish Game Wildl.Bull. 10. 24pp.Hirth, D. H. 1977. Social behaviour of white-tailed deer in relation tohabitat. Wildi. Monogr. 53. 57pp.Hobbs, N. T., and T. A. Hanley. 1990. Habitat evaluation: douse/availability data reflect carrying capacity? J. Wildl. Manage.54:515-522.Hood, R. E. and J. M. Inglis. 1974. Behavioral responses of white-taileddeer to intensive ranching operations. J.Wildl. Manage. 47:664-672.Howard, W. E. 1960. Innate and environmental dispersal of individualvertebrates. Am. Midl. Nat. 63:152-161.Kufeld, R. C., D. C. Bowden, and D. L. Schrupp. 1989. Distribution andmovements of female mule deer in the rocky mountain foothills. J.Wildi. Manage. 53:871-877.Lenth, R. V. 1981. On finding the source of a signal. Technometrics23:149-154.Linsdale, J. M. and P. Q. Tomich. 1953. A herd of mule deer. Univ.California Press, Berkeley. 567pp.Loft, E. R., R. C. Bertram, and D. L. Bowman. 1989. Migration patterns ofmule deer in the central Sierra Nevada. Calif. Fish and Game 75:11-19.Masters, R. D. and R. W. Sage, Jr. 1985. White-tailed deer fawn/daminteractions and fawn home range establishment. N.Y. Fish and Game J.32:93-94.McCullough, D. R. 1964. Relationship of weather to migratory movements ofblack-tailed deer. Ecology 45:249-256._____• 1985. Long range movements of large terrestrial mammals. Pages444-465. in M. A. Rankin, ed. Migration: Mechanisms and adaptive187significance. Contrib. Marine Sci., Suppi. Vol. 27.McNay, R. S. and D. D. Doyle. 1987. Winter habitat selection by black-tailed deer on Vancouver Island: a job completion report. BritishColumbia Minist. Environ. Parks and Minist. For. IWIFR-34, Victoria.9Opp._____and J. M. Voller. 1995. Survival and cause-specific mortality ofColumbian black-tailed deer on Vancouver Island. J. Wild]. Manage.59:138-146.J. A. Morgan, and F. L. Bunnel]. 1994. Characterizing independenceof observations in movements of Columbian black-tailed deer. J.Wildl. Manage. 58:422-429.Meidinger, D. and J. Pojar. 1991. Ecosystems of British Columbia.British Columbia Minist. For. Special Rep. Ser. 6. Victoria. 330pp.Nelson, M. E. and L. D. Mech. 1986. Mortality of white-tailed deer innortheastern Minnesota. J. Wildl. Manage. 50:691-698.Nyberg, J. B., F. L. Bunnell, D. W. Janz, and R. M. Ellis. 1986. Managingyoung forests as black-tailed deer winter ranges. British ColumbiaMinist. For. Land Manage. Rep. 37. Victoria. 49pp.Richens, V. B. 1967. Characteristics of mule deer herds and their range innortheastern Utah. J. Wildi. Manage. 31:651-666.SAS Inst. Inc. 1985. SAS user’s guide: basics, version 5 edition. SASInstitute Incorporated, Cary, N.C. 584pp.Searle, S. R., F. M. Speed, and G. A. Milliken. 1980. Population marginalmeans in the linear model: an alternative to least square means. Am.Stat. 34:216-221.Sinclair, A. R. E. 1984. The function of distance movements invertebrates. Pages 240-258 in I. R. Swingland and P. J. Greenwood,eds. The ecology of animal movement. Claredon Press, Oxford,England.Sokal, R. R. and F. J. Rohlf. 1981. Biometry. W. H. Freeman and Company,N.Y. 859pp.Sweanor, P. Y. and F. Sandegren. 1988. Migratory behaviour of relatedmoose. Holarctic Ecol. 11:190-193.Thomas, D. C. 1970. The ovary, reproduction, and productivity of femaleColumbian black-tailed deer. Ph.D. Thesis, Univ. of British Columbia,Vancouver. 2llpp.Wallmo, 0. C. and J. W. Schoen. 1980. Response of deer to secondary forestsuccession in southeast Alaska. For. Sci. 26:448-462.White, G. C. and R. M. Bartmann. 1983. Estimation of survival rates from188band recoveries of mule deer in Colorado. J. Wild]. Manage. 47:506-511._____and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data.Academic Press Inc., San Diego, Calif. 383pp.Yeo, J. J. and J. M. Peek. 1992. Habitat selection by female Sitka blacktailed deer in logged forests of southeastern Alaska. J. Wildl.Manage. 56:253-261.Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, EnglewoodCliffs, N.J. 718pp.189CHAPTER 8 - GENERAL CONCLUSIONSAfter preliminary field work observing habitat choices made by deer atNanaimo River, I began to question the strictly economic properties of theoptimal foraging theory (Schoener 1971) that formed a basic assumption inour study design. Initially, I questioned the notion of optimalitygenerally because it appeared vague to me but, eventually, I couldn’t ignoreits implicit and appealing association with the theory of natural selection.Then I considered that animals likely have goals beyond those implied informulations of optimality established by Schoener (1971). Indeed, inrecent years the whole aspect of animal behaviour has entered the minds ofresearchers more strongly and has led to an emergence of new kinds ofresearch on optimality beyond an accounting of costs and benefits offoraging actions (e.g., Schoener 1987, Gass and Roberts 1992). Primarily,this new research is addressing the question of - when do animals optimize -and draws into consideration the fact that, in the past, the effects oftemporal and spatial scales on resource use problems has been neglected.This is specifically where I directed the objectives for the IWIFR DeerProject and what my conclusions are meant to address.INDEPENDENCE OF OBSERVATIONS IN MOVEMENTSBecause most hypotheses or investigations would be based on deerlocation data (chapter 1), it was necessary to assess the statisticalindependence of the data I collected. This need was particularly strongbecause of recent emphasis on the potential dangers in working withtemporally dependent data (Swihart and Slade 1985). I found most data weretemporally related and, further, a systematic elimination of data to createlonger time intervals between samples did not correct the problem. I190determined that the lack of independence was only an apparent problembecause the distributional properties of my data invalidated the publishedtest statistic. The particular distributional properties I observed aroselargely from migrations and other occasional, long movements.Once I considered this distributional phenomenon of the data, Irealized that movements result from behaviourial-related decisions made bydeer and therefore are unlikely to ever be judged independent on that basis;each new movement will always be related in some way to previous movements.Rather than eliminate data in an attempt to achieve independence, I proposedthat researchers should attempt to gather as much data as possible (moreinformation) but to ensure these data are collected systematically in time.SPATIAL AND TEMPORAL SCALES IN MOVEMENTSThe importance of spatial and temporal scales in resource use problemswas presented by Senft et a!. (1987). But I didn’t see the clear link tooptimality until considering some thoughts by Schoener (1987) and Levin(1992). Finally, and I admit this was only recently, these thoughts weresolidified by Gass and Roberts (1992) with their consideration ofinteractions between temporal scales and optimization. Although noteloquent perhaps, I approached this problem of scales by making the explicitrecognition that resource use is based on an hierarchical decision-makingframework (Chapters 3 and 4). High elevation natal ranges cannot providesuitable habitat conditions during winter months so, eventually, the manydecisions deer must make about how to use a resource that diminishes throughsummer months, are synthesized into higher level decisions about migration.Gass and Roberts (1992) referred to this hierarchical phenomenon as anupward cascading of fine-scaled actions. But the higher level decision to191migrate then constrains habitat use decisions within the general site chosenfor winter range. I found, for black-tailed deer, this constraint isemphasized by a general tactic for fidelity to specific sites and likely bythe fact that those sites are chosen by a matriarch. In fact, affinity forsites was a general rule expressed as philopatry at the level of home range,fidelity at the level of seasonal ranges, and a tenacity for specific sites(we referred to these as nuclei) within habitats at the level of habitatselection. Together, the hierarchical nature of decision-making and thebehaviourial tenacity I observed, lead to individuals with a generallystatic approach to habitat use. I expected this static nature of decisionmaking (habit) could form the basis for lags in response to rapid alterationof habitats (Chapter 5). Therefore, optimality of resource use, althoughcorrectly measured at fine-scale choices, must be considered and interpretedat, or least set in context of, all scales of resource use.SPATIAL AND TEMPORAL SCALES IN HABITAT USEI consider this static approach to resource use also partiallyexplains the variety of habitat preferences observed in chapter 4. Ifindividuals maintained a single tactic for resource use while environmentsfluctuated (e.g., winters variably moderated by maritime weather patternsand/or rapid and extensive logging) I would record and interpret variablehabitat preferences. Even so, I observed consistent trends (paired samplecomparisons between summer and winter) among these preferences to indicatethe superiority of old forests as winter habitat. Other trends wouldindicate the superiority of young forests and even open habitats wheneverwinter weather was not severe. I also noted that migratory deer preferredold forests more than resident deer; an observation strengthened by the fact192that I trapped no migratory deer in young forests (Chapter 7). Conclusionsabout optimality of habitat choices is out-of-reach, however, unless thesechoices are set within the context of both site- and time-specific events aswell as the larger issues of ultimate survival and productivity differencesthat may exist between the 2 seasonal movement groups.RESPONSE TO WINTER HABITAT LOGGINGAs in many large projects that take place in uncontrolledenvironments, problems of strict logistics seriously weakened my ability todraw firm conclusions on several initiatives; probably the most seriousbeing the question concerning lags in response to logging (Chapter 5). Thisflaw is fundamental to my central thesis about constraints to the ideal-freedistribution hypothesis. A short-lived study species (due mostly topredator efficiency), a limited technology (4-yr radio collars), and plannedmanipulations that were large and politically unstable led to small samplesizes and barely suitable manipulations. Still, there is qualified supportfor the notion that individual deer change their habitat use decisionslittle, even when subjected to large and abrupt changes in habitatstructure. I expected some deer to make dispersal-type decisions when theirwinter ranges were essentially destroyed. This did not occur. The notionthat deer respond to logging by making subtle shifts in habitat use withintheir original home range is a notion firmly held in the minds of those thathave experience observing black-tailed deer. Until we have new informationto reject this notion, I must conclude that deer commit themselves early inlife to specific tactics for seasonal range use and altering this commitmentis difficult at best. The implications this loyalty has is that subsequentresource use decisions become increasingly more a function of habitat change193than of habitat quality. Adaptation is limited to learning about newresource conditions within the established home range rather than aboutother (may include better) habitats outside the home range. Indirectlythen, this loyalty to specific tactics, implies a constraint to the ideal-free distribution of deer.MORTALITY CAUSES AND SURVIVAL ESTIMATESI could not avoid the logistical problems associated with observingproductivity of individual deer and so, although I did observe adult femalesurvival, this led to another major weakness in making conclusions about theimplications of constrained resource use tactics. However, I conclude fromthe survival data that migratory deer survived better than resident deer.This differential in survival was primarily caused by residents being athigh risk to predation from wolves and cougars in late-winter and spring,and to hunting from man during fall. The higher risk was a consequence ofresident deer making up the greater proportion of the population living, inlate-winter, at low elevations where access was perhaps enhanced by loggingroads.This conclusion about differential survival, and its interaction withthose from other chapters, led to 3 major management implications (Chapter7). First, because I expect that response by deer to any habitatmanipulations will lag in time, it is unclear what and when to measure as aresponse. Hence, I bring into question previous interpretations ofuse/availability measures and habitat evaluations. Second, I am concernedthat, as whole valleys are changed from old forests to young forests, fewerand fewer migratory deer will exist. With only resident deer remaining,resilience of deer populations will likely decline. Third, and on a more194positive note, the behaviour of deer populations would appear to berelatively predictable to the habitat manager based only on variablesassociated with topography, local climate, and the spatial locations andgeneral ages of forest stands. This knowledge could facilitate landscapelevel planning (Chapter 7) to aid the maintenance of both migratory andresident deer within the same population thereby retaining a degree ofpopulation resilience against both predators and severe winter weather.LITERATURE CITEDGass, C. L., and W. M. Roberts. 1992. The problem of temporal scale inoptimization: Three contrasting views of hummingbird visits toflowers. Am. Nat. 140:829-853.Levin, R. V. 1992. The problem of pattern and scale in ecology. Ecology73:1943-1967.Schoener, T. W. 1971. Theory of feeding strategies. Ann. Rev. Ecol. Syst.2:369-404._____1987. A brief history of optimal foraging ecology. Pages 5-68 inA. C. Kamil, J. R. Krebs, and H. R. Pulliam, eds. Foraging behaviour.Plenum Press, N.Y.Senft, R. L., M. B. Coughenour, D. W. Bailey, L. R. Rittenhouse, 0. E. Sala,and D. M. Swift. 1987. Large herbivore foraging and ecologicalhierarchies. BioScience 37:789-799.Swihart, R. K., and N. A. Slade. 1985. Testing for independence ofobservations in animal movements. Ecology 66:1176-1184.

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0088371/manifest

Comment

Related Items