IIIIIIIMIIIHH III II WINTER HABITAT STRUCTURE OF THE SNOWSHOE HARE “I I W N—‘l my . 0001 Thesis {or the Degree of M. S. MICHIGAN STATE UNIVERSITY Michael James Conroy 1976 I I1;Ilirmwtwwml”MW L I: “#18 M? nwctsiry 26.. [1.34 I NOV‘l 2, I?» 0 it ABSTRACT WINTER HABITAT STRUCTURE OF THE SNOWSHOE HARE By Michael James Conroy Snowshoe hare habitat structure was studied on diverse, partially clearcut areas in northern Michigan during January through March, 1976. Utilization, habitat, and weather variables were intensively measured on a 61 ha study area in order to develop a descriptive and predictive model; predictions from this model were tested during March, 1976 by surveying two 23 km2 extensive study areas. Results from the inten- sive study indicate that hare activity centered around lowland coniferous and alder (Alnus) types, but dispersed into adjacent upland coniferous-hardwood and clearcut hardwood communities where habitat interspersion was high. Distance from lowland coniferous-hardwood types and habitat interspersion were the two most important factors determining hare utilization. Utilization was heavy along several clearcut—conifer edges. Red maple (Acer rubrum) and speckled alder (AZnus rugosa) were the most frequently browsed species. Browse selection shifted to aspen (Pbpulus spp.), pine (Pinus Spp.), and blackberry (Rubus spp.) as these became available. The extensive surveys supported the conclusions about hare utilization made from the intensive study. Hare utilization decreased drastically farther than 200 m from lowland coniferous canopy cover, in both cut and uncut areas. Clearcuttings near lowland coniferous cover were utilized heavily, primarily along the edges. Clearcut communities very distant from lowland conifers were essentially Michael James Conroy non-utilized by hares. Cuttings managed for hares should be small or shaped so that canopy cover is within 100 m of all parts of the cutting. Slash left from cutting operations may act as supplemental cover if strategically concentrated along likely feeding and travel lanes. WINTER HABITAT STRUCTURE OF THE SNOWSHOE HARE By Michael James Conroy A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Fisheries and Wildlife 1976 ACKNOWLEDGMENTS I wish to thank my major advisor, Dr. Leslie Gysel, and the other members of my graduate committee, Glenn Dudderar, Dr. Rollin Baker, and Dr. Richard Hill. Carl Bennett and George Burgoyne of the Michigan Department of Natural Resources and Drs. John Gill and Ivan Mao of the Dairy Science Department at Michigan State University were of great assistance during the statistical design and analysis phases. I would also like to thank the staff of the Houghton Lake Wildlife Research Station for the use of the station's facilities, and for advice, assistance and encouragement during the course of my study. I am grateful to Mr. Harold Mayer of Maple Valley for allowing me to set up my weather equipment on his property. ii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES . . . . . . . . . . INTRODUCTION . . . . . . . . . . METHODS . Intensive Study Area . . . . . . . Estimation of Vegetative Parameters Measurement of Hare Utilization Browse Selection . Other Variables Data Analyses Utilization Map . . . . . Extensive Surveys RESULTS . . . . . . . . . . . . DISCUSSION CONCLUSIONS LITERATURE CITED APPENDIX iii iv 11 12 13 18 19 23 32 36 37 39 Number LIST OF TABLES Summary of vegetative estimates by community. Correlation matrix of untransformed utilization, weather, and habitat variables. Regression coefficients and proportion of variance explained by multiple regression models. Summary of the results of the discriminant analysis, using four utilization groups and three discriminant functions of weather and habitat variables. Percentage of total number of twigs browsed and stems for each community, grouped by species. Analysis of extensive track survey data. Percentage of total numbers of twigs browsed and stems barked in all communities, grouped by species. iv 17 25 28 29 39 Number LIST OF FIGURES Description of vegetative communities and illustra- tion of transect method. Density of total hare utilization index by community types. Utilization group means in standardized (Mean = 50, standard deviation = 10) discriminant space. Mean difference in transformed track index by treat- ments (cut vs uncut) and distance from cedar-fir for extensive track surveys. Distribution of hare utilization by community types. 21 27 31 Al INTRODUCTION The niche of a species may be defined in theory by its coordinates in n—dimensional space, each coordinate corresponding to a measurable environmental parameter (Hutchinson, 1957). Problems of applying this concept to field data include redundancy, non-additivity, and non- linearity among the parameters (Green, 1971). Analytical approaches to overcome these difficulties have included discriminant analysis and principal component analysis (Green, 1971; James, 1971; Whitmore, 1975)- In a very general sense, the niche of the snowshoe hare (Lepus americanus) is known. Grange (1932) described several types of habitats in Wisconsin in which hares were commonly found; hare utiliza- tion of these habitats varied seasonally, but included aspen (Pbpulus spp.) stands of moderate age adjacent to conifer swamps, alder (Alnus spp.) swamps, old burns, young Jackpine (Pinus banksiana) stands, and hardwood stands near conifer cover. In Minnesota, Aldous (1937) noted that during inclement weather hares remained close to forms or resting spots consisting of hollow logs, willow (ShZix spp.) clumps, or fallen trees. Based on pellet surveys in Montana, Adams (1959) found that hares occurred in the greatest concentrations where woody vegetation was thick, but there apparently is an optimum density of vegetation beyond which increasing density will diminish hare use of the habitat. More recent workers have attempted to isolate specific important factors in habitat structure. Bider (1961) showed that vegetation structure plays an important part in determining the size of home ranges in Quebec hares, and that climatic and physical factors may dampen or activate movements within those ranges. Brocke (1975) found that continuity of coniferous canopy seems to be essential for snowshoe hares in the Adirondack region of New York. Hares spend the day in "base cover" consisting of conifers averaging 3.5 m tall; "travel cover" (conifers 8.3 m tall) provides travel lanes when adjacent to base cover, but has no value in the absence of the latter. Hardwood browse was the most important winter food source. Keith (l97h) felt that specific vegetative parameters, particularly stems less than 3 mm in diameter, are a critical part of a hare-grouse- predator system of cyclic abundance. The objectives of this study were to, on diverse and partially clearcut areas: (1) built a descriptive and predictive model about hare utilization based on intensively measured utilization and habitat variables, (2) use this model to generate predictions about hare utilization on similar areas, (3) make a preliminary test of these predictions, using data from extensively surveyed areas, and (h) tentatively evaluate the effects of certain types of clearcut situations, based on the data collected above. METHODS Intensive Study Area Vegetative data were collected during the summer of 1975 and hare utilization data during January-March, 1976, on a 61 ha study area located in southeastern Roscommon County, Michigan. This part of Michigan is characterized by a relatively mild climate, with normal annual average temperatures of around 5°C and total annual precipita- tion of about 71 cm. The study area is part of a 23 km2 research unit (designated the Lanes Lake Unit by the Michigan Department of Natural Resources) that was partially clearcut in 1973 for a deer range management study. The geology of the area is primarily of glacial morainic origin. Three major soil types have been mapped (Veatch, 1929). Well-drained, low fertility Grayling sand on level to gently rolling terrain supports upland types in the southwestern third of the study area. A strip of Rubicon sand running from the northwest corner to the southcentral end of the area supports upland types in which small pockets of lowland types are interspersed. The east half of the area is predominately Newton loamy sand on level to gently sloping terrain, with many flat, low areas prevalent; large areas are very swampy, and some standing water occurs seasonally. The study area was surveyed during June, 1975, noting species compositions, canopy heights, and basal areas for the overstories and species compositions, heights, and percentages of cover for the understories (Figure 1). Overstory canopy heights were estimated using an altimeter, and basal areas were estimated using an angle gauge; several readings were taken in each community to get a rough range for each parameter. Understory heights and coverages were estimated by eye. Five major groups of vegetative communities occur on the study area (Figure 1). Communities IVa to IVe, formerly composed of oak (Quercus spp.), aspen, and pines (Pinus spp.) 12 m to 15 m tall were clearcut between January and April, 1973; they thus had three complete growing seasons as of the commencement of this study. Moderate to heavy reproduction of aspen, red maple (Acer rubrum), oak, cherry (Prunus spp.), and juneberry (Amelanchier sp.) dominates these communities. Large slashpiles left from the cutting operations are scattered throughout. Although Figure 1 indicates that the overall percentages of low cover are similar for all five of the clearcut communities, the distributions of low woody cover and slash are quite heterogeneous. Communities IVa and IVd, particularly the former, have many rather open patches, and cover tends to be relatively sparse near the edges. Communities IVb, IVc, and IVe, while having some open areas toward the centers of the communities, tend to be more densely covered toward the edges. These latter communities also tend to have a higher density of slash, especially around the edges, than do communities IVa and IVd. Forestry records indicate that community IVf was commercially cut for oak in 1961; although there are no further records, it appears to have been burned about five years prior to the time of the study. Scattered 10 cm to 20 cm diameter fir (Abies balsamea), aspen, red no mnwwon opdou magaoxm .wH.H pd menu was :.H oumawonooo .Amopwawonooo uoaoafio mo moo adv sons go one apnea pm uhdpm amazon hawsn 8.8 #4 5m .eoosmoe £32 .nua . «you? .> G ham Humans Oblom rum hpuopooSn .hhpono .oamoa .xdo .ommm< manma .xoo con .ocwm ooh oonoppoom .m mblom :IH huuopoqsw .hhhono .oamoa .Mmo .comm¢ .olo seagufio Bandeau .3 O omum TH Eu 2.23 §§8v 8258 £32 NA ,1an :8 €me .sao esoanmse cookoumm camakoq .HHH@ new and Medan .oamma main MIH .psooo .hhuono .oamoa .nmm .hooa< Hana PHINH ooh .pmooo opens .hfim Somawm oUOUOOOQ new nmw Mooan .oamms mlo mna .nmooo .hupono .oamms .nmo .pocad mHINH omuma ooh .aoooo mews: .hfim_aomawm .d cookohwmnhomwooo 693564 .HH . sodas .A.mmm omnom mIH maumaobv pdcaoumn .mamofi .Mwo .coam¢ ml: wauma x60 ooh .oawm ooh .ocwmxodb .o magma 041mm mIH huponoqzn .oamma .hhhono .Mmo .comm< mum mHINH own .oommw .qunxoon .oawm oom .o manna ooh omnow mIH huhopondn .mamwa .hhaoSo .moo .dommd win mHINH .oommo .oofimxown .onHm ooh .Mdo com .9 omuma mna hypopoosn .oamoa .hpnono .Mmo .momm< ml» wHINH onwmxoon .ooflm ooh .xdo com .d homwnooévoosondm 98.3: .H @ uo>oo Aav mowoomm Away “av a .pm . sous .pm Hommm bogmnmdcb repmhms nzmomq .oonpoa poomsonp mo oofipdnpmsaaw one moapaqsaaoo obwpopomo> no acapmwuomon .H madman PRIVATE PROPERTY IIIIIIIIIIIIIIII -_F_ I...1I1,, . " , 1,, 1711.1 11m“, I.II«11IIIIH IIIIIIIIII E—W Coordinates (SO—m units) m mamnm.o III Rmzaz. w «mmaamm spzaommaa &mmummmoz o> mwmfia.m Ramam.ma Ramuz. w emaaummoa Romammaw Rpmuomwaw m>H mmmaw.m ammum.ma ammaa.a ammaammm Ammamsm: ammummmmm 6>H momnw.m Romne.ma anmflo.s aammwmoa awmflmaws ammnmmawm 6>H mmnflm.m Ramaw.ma upmam. m Rmmmammm Rmsaommw amaammmam m.p>H mmmam.a eamns.ma Rmm Ho. m ammgam emmnmsww umHHmmHom w>H mnmaa.m Ramam.ma &Hmam.oa xmhammwa Rahaoomm mmmawmzmm HHH uwmam.a «Hmuo.oa u:ma:.ma “smumama awmaawmm «mswmmmsm o.e.oHH mmmao.m wmmaw.m mmaam.ma xawamazw Roaaamrwa Rmmawwamm QHH wwwnm.a awoaa.m amam.ma “waummmm Rmmmnmw usaammmm aHH mwmflm.a Rmmu:.ma Rmaaw.m Ramaamwa mmmamsmm mmmwmmHOm 6H unmum.a Roano.aa ammnm.oa Romammom ammnomea mamuomsam 6H mooaao.a wmmam.aa emmam.oa Rawwmzmm stwmamm gmgammamm pH wmomam.o ammum.aa fibaam.ma fiomammma scrammma eazawmmma 6H oxmoow execs“ Asv Aos\mswpmv a w.:nw.a a w.alm.o hpwndaaoo -ao>oo sofiposupmpo pswfion hpwmooo Awg\msopmv Smmam Hoaopoq hmocmo haopmhoso Swansea haouaaoocb .Amcooa map mo owwpcooaom mo commoaaxov mam>popcw oocoofimsoo gem meow: "hpflooasoo an mopdafipmo o>flpopowo> mo assassm .H canoe 11 recorded by the coordinate of the starting point (Figure 1). Along each 50-m sampling unit, the following were counted: (1) recent (one- to-two-day-old) trails (designated TL in the analyses), (2) areas with many trails crisscrossed (designated TM), (3) areas covered with indistinguishable trails (designated TH), and (A) recently used (one- to—two-day-old) runways, divided into three subjective categories by intensity of use (designated RL, RM, and RH, for low, medium, and high intensity, respectively). In addition, the first trail or runway encountered in each 50-m sampling unit was followed for 20 m in the hare's direction of travel (if indeterminable, decided by coin flip), counting woody twigs browsed and stems barked by species, noting forms with signs of recent use, and counting any additional trails or runways encountered. These additional trails or runways were included in the count taken on the main part of the transect. Each 50-m sampling unit thus had two major types of information: (1) levels of use, determined by trail and runway counts, and (2) categories of utilization, determined by following trails. The latter type of information was intended to reduce the problem of interpreting transect data that occurs when animals pass through an area on their way to another, possible more preferred, area utilizing the measured area only for travel. Sampling was done on 28 days during January through March, 1976. Browse Selection Browsings were classified by species of plants browsed and major community groups in which browsing occurred (Table 5). Only those l2 browsed species for which 5% or more selection occurred were included in Table 5; a complete listing of species browsed is provided in Table A-1. Mean browsing index per community was computed by dividing the total browse index (twigs browsed plus stems barked) by the number of observation points taken in that community over the study period. Numbers of stems available per community can distort browsing canparisons, by artificially inflating intensity in areas with few stems and deflating intensity in areas with many stems. Therefore, mean browsing index was adjusted for numbers of stems available by multiplying by the number of stems available, and dividied all figures by 106. Other Variables Several variables were determined from each coordinate on the community map (Figure 1). These were (1) habitat interspersion (numbers of communities within 100 m), (2) distance from lowland conifer communities, (3) distance from alder swamps, (A) distance from upland hardwood-coniferous communities, and (5) distance from clearcut communities. Distances were measured from the coordinate to the closest edge of any community of the appropriate designation. Although not of primary interest in this study, it is known that weather fluctuations can greatly affect the activity patterns of hares (Bider, 1961). Furthermore, snow conditions may affect the use of runways (O'Farrell, 1965). In order to minimize unexplained variation in any model describing utilization over time, windchill, cloud cover, barometric pressure, precipitation, and snow conditions 13 were measured and included in the analyses. A windchill meter (Verme, 1968) was set up approximately 1.2 km from the study area to record daily windchill index. Cloud cover, precipitation, and barometric pressure were taken from the daily records of the U. S. Weather Bureau at Houghton Lake. Snow depth was measured with a meter stick, and snow compaction was measured with a 9 kg cm2 compaction gauge (Verme, 1968). Snow conditions were sampled at regular intervals along the transects to provide at least 6 readings per day for each community group; variation was generally slight within these groups. Daily averages were obtained from these samples for each community group, and these averages were used in the analyses. Since the moon was visible on relatively few nights during the study, moon phase was not considered to be an important factor. Data Analyses The two analytical approaches used were multiple regression and discriminant analysis. The multiple regression (Draper and Smith, 1966) attempted to describe the types of utilization as functions of the independent variables: habitat structure and weather conditions. Three different models were developed, using the measured independent variables (Table 2) and three dependent variables: track index, browsing index (twigs browsed plus stems barked), and number of forms used, at each coordinate. The track index was computed by weighting the observations in each trail and runway category by arbitrary coefficients: specifically, Track indexsTL + ZDM + 3TH + 2RL + 3RM + hRH. _ ~.~_.u..:> a... 4: . 7...; —.——-Q -It_‘ .qu-E --ub 1h no>oo nmem 0m poosooHo aoum monopmHQ >HQ psmHog hmocdo md :oHpopHmHoohm emu ooHQIxoo 60pm moswpmHQ H0 thmooo hAOpmao>0 00 madmmoam oHapoEOHmm m0 As m.zlm.Hv no>oo USOHQ A00 EHMIAoooo Bong monopmHQ HHQ thmooo haopmaooob man xoUGH HHHnoocHs moz mason mm :onpommhoch pmpHpom Hm AB w.HIm.0v coHpoomEoo 30cm ozm xoocH omzoam mm coHpooppmpo Honoqu m0 thmcov snowmuooob H00 npmov 3o:m 02m xoosH moose me moHpmHao> pprpom moHpoHno> nonpooz moHpoHns> coHpouHHHpD 90m m0.1 >H0 :0. mm.1 H0 No.1 mm. :m.1 >0 00.1 m2. >m.1 HP. HHQ mo. mm.1 Ho. :m.1 sm.1 Hm :0. mm.1 Hm. mm.1 mm.1 m0. om H0.1 om.1 00. mo. 00. 0H.1 0:. m0 H0. mm. NH.1 H0.1 H0.1 NH. mm.1 Fm.1 ad H0.1 pm. Hm.1 m0. m0. NH. mm.1 20.1 mm. 00 No.1 mm.1 m0. :0. 00. mH.1 Hm. mm. 00.1 no.1 man m0. H0.1 00. mm.1 0H.1 :0.1 ma. mm. 00.1 00.1 so. H00 0H. H0.1 m0. m0.1 :0.1 00. 0H. No.1 H0. N0. H0.1 H0.1 m0. :0.1 m0. :0. m0.1 00.1 H0.1 H0.1 :0. m0. m0. :m. mm MH.1 mH. 00. No. no.1 so. m0.1 :H.1 0H. NH. HH.1 EH.1 m0. :H.1 mo. 00.1 m0. Hm. pm. me mam >HQ H0 >0 HHQ Hm 0m mo 5. 00 man H00 EB A00 mo: ozm sz E mm .moHpmHnw> popHpmn was .aonpmos .coHpmNHHHp: ooahommcdnpqs mo xthoa ooHpoHouhoo .m «See 15 The correlation matrix of all variables showed high correlations among 5 of the independent variables: density of understory (in both height classes), density of overstory, canopy height, and lateral obstruction (Table 2). In order to reduce problems in analysing and interpreting such intercorrelated variables (Green, 1971), only density of overstory was used in the regression analyses instead of all five intercorrelated variables. This variable was selected over the others because: (1) It succinctly expresses the structural changes occurring along the light gradient from densely canopied to open areas; the other variables are partly redundant. (2) It is most correlated with the track index. (3) In a future study, it would be one of the easiest variables to measure. Prior to the regression analyses, the dependent variables were transformed according to the formula TRANS(Y) = /TR:_675 in an attempt to meet assumptions of normality (Sokal and Rohla, 1969); however, the transformed variables still failed to meet assumptions of normality (Kolgomorov—Smirnov test significant, p < .05). In this situation, estimation of parameters (means, regression coefficients) is still valid, but hypothesis testing is not (Searle, 1971); this was acceptable for my study since the primary interest was in descrip— tion (estimation) and not in hypothesis testing. Since the trans- formation did make the distributions of the dependent variables more closely resemble the normal distribution, the transformed variables were used in the regression analyses. Browse index and forms were measured cumulatively because browsings and forms counted on one day could be recounted on successive days. Since fluctuations over time were not relevant to such cumulatively 16 measured variables, these variables were not regressed on the weather independent variables, but only on habitat variables. Treating the browse index and form measurements as cumulative variables could conceivably affect the analyses, if particular browsings and forms were multiply counted, due to artificial magnification of among- community differences. However, I feel that cumulating these variables were justified for two reasons: (1) Browsings and forms were located at points along hare trails followed away from transect lines, and the probability of recounting them.was low (I estimate less than 10%). (2) Any minor effects on browse index and forms measurements due to recounting should have enhanced the analyses, since these variables were less likely to be sampled than trails and provided fewer data for comparisons between locations. The best regression equation for each dependent variable was selected by means of a stepwise procedure (Draper and Smith, 1966). Entry criteria of F = 3.00 for track index and browse index and F = 1.00 for forms were selected by trial and error to yield the best equations (Draper and Smith, 1966; Nie et al., 1975). The regression equations were poor predictors of utilization, accounting for only 9%, 3%, and 0.3% of the total variation in track index, browse index, and forms, respectively (Table 3). These models were probably inadequate partly because of reasons pointed out by Green (1971): particularly violations of the assumptions of additivity and linearity among the parameters. In discriminant analysis, groups (any logical units of animal distribution, activity, behavior, etc. defined by an ecologist) are separated in k-dimensional space (where k is equal to the number of 17 .H 00 ooHpoH>oo osmooopm .0 mo new: * moo H mm m.o + mEHOIh\ m.0 + NoooH onmsohm\ 000. u mm m.o + amen“ aoaaex chHmmonwom an oocHmexm GOHHwHMm> mo copromoam szo. poon 809m ooqumHQ . nooHo Nwmo aonm monopmHo . pneumoHo ammo scum oocwpmHQ . oonnmmmpopoH mafia savanna . oGHQIHwo ommH 80am moompmHm pooHonmooo oHpoHaw> *ooNHonmooopm pooocomoooH 00:0. gumbo zoom mMHH.1 ho>oo osOHo osma.1 soapspnmuomam . osHmlnmooo NPOH sonm moswpmHm . psoaooHo mmmm scum ooadpmHQ . oonuommnopoH amen papfipam pooHonmooo oHpmHho> avouHopoocdpm pooocomoocH .1 uHQIMoomo ammo scum mommpmHn .1 menauaso nzmo scam oocopmHQ 00m0.1 thono haoumno>o ommo. uo>oo nmon . pneumoHo mmmo 80am mocoumHn . connmmmhopoH mmmo vopHpsm pcoHonmmoo oHpmHaw> acoNHonmocopm powwowmoooH m.0 + mappm\ m.0 + soooH monsommx moHnoHad> pnoooomom m.o + sauna auaaex .Apxop oomv onsooooum omHsmopm 0 hp oopomHom mo mnoHvdndo poop you one mpooHonmooo .mHoooa sonmonwma onHpHoa hp oocHonxo ooeoHaw> mo cothomonm one mpnoHonmooo oOHmmonwom .m mewe 18 groups defined, minus one) by functions of the environmental variables that the ecologist has chosen to measure. This approach to analysing utilization removes two of the problems inherent in the regression approach used above by (1) eliminating the need to assume linear additive relationships between environmental factors and utilization, and (2) simplifying interpretations by reducing the model from m dimensions to k discriminant functions, where m is the number of environ- mental parameters considered (in this study, 17 habitat and weather variables) (Green, 1971). Four utilization groups to be analysed by discriminant methods were defined as follows: Group 1 (Non—utilized) consisted of all observations (each sampling unit on each day sampled) in which there were no signs of hare activity. Group 2 (Travel) consisted of observa- tions in which trails or runways were recorded, but no browsings or forms. Group 3 (Feeding) consisted of observations in which both trails and forms in use were counted; this category also included non-feeding observations, but in most feeding also occurred. The analysis maximized the distances among groups along each of three discriminant function axes (Cooley and Lohnes, 1971). Computations for all analyses were performed on the Michigan State University CDC 6500 computer, using a packaged statistical program (Nie et al., 1975). Utilization Map A descriptive map of utilization was constructed by means of a total utilization index. The index was intended to express total utilization, but to emphasize feeding and form use over travel. This 19 was accomplished by weighting mean track index 1, browsing index 2, and forms 20. Forms were weighted heavily because I felt that 10 instances of browsing were roughly equivalent to 1 instance of form use, in terms of total intensity of utilization. The weighted track, browse, and form indices were summed to give a total utilization index, expressed as utilization per hectare (Figure 2). Extensive Surveys In order to broaden the scope of this study, and as a preliminary test of some predictions from the above discriminant model, hare trail surveys were made on two 23 km2 study areas in March 1976. Both areas had been partially clearcut during 1972 and 1973 and thus had under- gone 3 to h growing seasons by the time of this study. The areas are predominately mixed upland conifers and hardwoods (oak, jackpine, red pine) 12 m to 15 m tall, with scattered lowland coniferous-hardwood types (cedar, fir, maple, ash). Area I (designated as the M-18 Unit by the Michigan Department of Natural Resources) is located approxi- mately 20 km southwest of the intensive study area, and was 25% clear- cut. Area II (designated as the Lanes Lake Unit) encompasses the intensive study area and was 50% clearcut. The areas were sampled by cruising roads and logging trails and counting hare trails and runways to derive a track index as computed by the formula given earlier. Route segments of varying lengths adjacent to mapped cover types were classified by factors determined to be important from the discriminant model (RESULTS section). These were: (1) treatment (uncut v8 cut), (2) distance from lowland conifer-hardwood canopy cover (level 1: 0 m to 200 m, level 2: 200 m to hOO m, level 3: farther 2O Apxop oomv .momhp mpaqssaoo an xoooH ooHpmNHHHp: one: Hopop mo thmooa .m 693a 21 'd {-1 o x <1) 0 0 (H 3 rd 'H 'd c: E: 3.. H d O d E S O p: 0 c: \ I I Id 9, 0" o '5 $4 0 m n-l-p (I) O o o ‘1 ax O 'H 3 . 63.0) v E 3 .4 U A, m. 'U C: L. d «4 .. O 31 Ha) lg: .5: .51 Z :1: .5.) .pm \ \ \ II. '0 'U :5 DV 0 O 0 'd a q 0 m d) ..:r J a: I H “3 H s. ,3 H H d a) Al Al V D 1.: ,4 O z! B I II IIIIIII II IIIIIIIIIIIII IIIIIIIIIII IIIIIIIIIIIII IIIIIII II III "III III “II ,II,1IIIII..111III ,III III III IIIIIIII . . . . ,III. III 1' ”I ' 111 /' 41.11 1 .III1 IIIII W IIII III III ,|"1IIII 111 I III PRIVATE PROPERTY . . .,.1111' 11,. . . 1.. 11.1.. . .1... ... IIIIIIIIIIIIIIIIII IIIIIIIIIIII IIIIIIIII'I ’ I111 ”. II'IIIIIIIIIII'II‘I III 11 I“. :02 111111111111 11 1 ,, ”III!" III“ II IImmI II ’(v.0.0, v.5. IIIIIIIIII11111IIIIIIIIIIIIIII IIII’I3' “"3 I ‘3‘ III I,/«"”-,'v 9.” 00‘" 099 ' 9"” 9090 ,,%:3 "%v~%"¥¥&fl§b.4/ ’0.0:. A. .20. 22 than hOO m), and (3) distance from upland hardwood—conifer canopy cover (same levels as above). The track index for each segment was converted to an index per kilometer of the traveled route. Data were also classified by areas (I vs II). Weather conditions were used as a blocking factor, grouping the four days on which sampling was done into two groups of two days each, one of mild weather, the other inclement. Before analysis, the data were transformed to /?—:—ET§ to correct for non-normality. The distribution of the transformed data was not significantly different from normal (p > .05) when tested by the Kolgomorov-Smirnov method (Nie et al., 1975). The null hypotheses of no effect on utilization due to areas, treatments, distance from lowland conifer-hardwood, and distance from upland hardwood-conifer were tested, using weather conditions as a blocking factor, by a five- factor analysis of variance (Nie et al., 1975). RESULTS From examination of Figure 2, it is evident that hare utilization was most concentrated in cedar-fir, oak-pine, and alder communities. It is also apparent that utilization tended to be away from the centers of these communities, and toward regions of high habitat interspersion. Clearcut communities were less utilized than canopied communities (except community III), but utilization was heavy around theedges of communities IVb, IVc, IVe, and particularly IVf. Community IVa was very sparsely utilized, and appeared to be acting as a barrier to movement between the oak-pine and cedar-fir communities. Precipitation and cloud cover appeared to have depressing effects on utilization (Table 3). Most browsing occurred away from the oak— pine communities. Form use was essentially unpredictable from the variables measured (Table 3). Examination of the residuals from the regression equations plotted over time (Draper and Smith, 1966) revealed no apparent time trends in utilization. Three discriminant functions were able to classify correctly 39.2% of the observations into utilization groups; the apparent error (Lachenbruch, 1975) of prediction was thus 61.8%. The main habitat factors determining function I were distance from cedar-fir, habitat interspersion, and distance from clearcut communities. The main habitat factor determining function II was distance from oak-pine. Overstory density and distance from oak-pine were both important in determining 23 2h function III. The three functions accounted for 58.1%, 28.3%, and 13.6% of the among-group variation, respectively (Table h). These functions are combined in a three—dimensional representation of habitat volume (Figure 3). According to this representation, hare utilization was centered in diverse areas near cedar-fir canopy cover and away from clearcut communities; within these areas, hares fed farther from oak-pine canopy cover, rested and fed closer to oak-pine, and traveled between. Table 5 indicates that red maple and speckled alder were the most frequently selected woody browse species for the entire study area. Browse selection shifted from pine and maple in the upland hardwood and conifer communities to maple and alder in the lowland communities and aspen, maple, and blackberry in the clearcut communi- ties. Browsing intensity was highest in the clearcut and alder communities, and lowest in the lowland hardwood communities. Most hare utilization in the extensive study areas occurred close to or in lowland coniferous communities. However, heavy utilization often occurred along lowland coniferous-clearcut edges, especially on the canopied sides. There was no significant (p > .05) difference in hare utilization ascribable to areas or distance from upland conifer-hardwood types, while there was a significant (p < .05) difference in utilization due to treatments (uncut v8 cut) and distance from lowland conifers (Table 6). There were significant (p < .05) interactions between treatment and distance to lowland conifer-hardwood, and between treatment and distance to upland hardwood-conifer. These tests may be biased because of significant (chi-square = lh.68, p < .05) heterogeneity of variance as determined by a Bartlett's (1937) test. 25 maama.u oommo. ammmm. coapwpflasomsm mmozm. mmmam.l wmwwm.l mswmlxmo scum mosmpmwm mmoma. mawoa.| mowmm.l sodam scum mosmpmwa mmama. bmmoa. Hmow:.| pdoswmao scum moswpmwm :ommm.l MMOOH. NQQMJ. Hamlhmdmo SCAM wocdpmwm mmmmm. mmama.l m»~m:.n nonummonvcfi pmpfipmm momoo.u Haooa. mmoma. sm>oo nmmam Hozzm.| Pwomm. szom.l hpfimsmc myopmum>o o:omo.: w»o:m.n wmmmo. mudmmmsm owspmaosdm wzwmm. mzmoa. mmoma. nm>oo csoao ammmm. mamwo. mmwsm.u Hafinoeawe Nmmmm. mamma.l mzawz. cofipommaoo 30cm Foams. chem». ammmm.n spams scam HHH noapocsa HH acapoqsm cospocsa .mmmmmmmm AH mo sofipwfi>mc usmvsdpm .0 mo sdwzv mpsmHOfimmooo soaposdm pawswawsomwn omufiosmcsmpm Hmmmm. mamma. moomm.u manommmuwnfipmmm mpmso. mma:m.n mamzm.| wnwuoOK wmmom.l mzoom. mOmHm.l Ho>th omboo.l mhmdo.l NHHOM. douwawpdlnoz HHH coupossa HH seapocsa H acapocsa AH mo cprmw>mu opquwpm .0 mo qwmzv madam pcwsfiefiuomwm cmufiwsmvcmpm cw mono: macho :oprNwprD m.mH m.mm H.wm HHH uoapocsm HH qupocsm H cowponpm nom cmunfiooo< soapmfiam> maouu wsoa< yo pcoonom .Apxwp mmmv moanwfinm> pmpwpwn can umnpwms mo mnofipondm padswswsomwc mounp and masosw GOprNwprs nsom wean: .mwmhadsd psdcfiafihomwc can no mpHSmmu on» mo hhdaadm .: wands 26 N.mm Hmpoe H.m; msfiommm Imsfipmom m.m: b.wm msfiommm Ho>mhe O.H: emufiaspsncoz guano aespwaflawps mpHSmmm cowpdofimfimmmao vowmfimmdao hapomunoo pnoonom "Ao.pnoov : manna 27 .mmapmwsw> msfipmGHsHsomHu MOnwE a“ omwonoqw mo soHpothc mumowosfi msohh< .oommm unacfiEwsomwc AOH u cofipdfl>oc andcsmpm .om u adozv vmnwcumcswpm 2H mamma macaw cofipmufiafipn .m onsmfim .5 ’ Qéwb ,Amx awav chub? ’9‘ (fimethspxk o 663. «raw 04 No .~. S #3 03.0 0: O b O\ 0 xx 9 z Ava on £65 or... one I. 3780 so m» and ’ 06, an» .. c Yam ’ O . d O . n A a a m a m om~_._.5..zo. u u 2 av r S mrm U A w . . O rm m. m. s a A“... TV m. .0 625%: -quMEr h 28 OH t mapmaww>d mampm a x hpflsdasoo ca mpsfiom sofipw>pmmpo % m wswmsopp ampoeo .spaquSOO an as a x AH mapwav an\msmpm an .H wanna afiecmda< as empmfiqw m:.o m.mH m.o m.m m.m ma.o ospflmcmpcs mcfimsosm HHH.H©~.N mam.:m 0mm.mmm.H omm.awa mmm.:mm amm.emm pmapmafla>a mewpm a Awmmavooa “moavooa Ahmmvooa Ammavooa Ammzvooa Aoamvooa Hmpoe . . . . . . Anode mm Azwzva mm Amavm 2a Amoavm PH Amavm ma Amoava Hm Amgvm :H can» Mandy pmnpo Ammavm.» a s * * m.mm A.mmm manwmv mafia .mmm m3 Abaav:.w * Awwv:.:H * s t A hhhmMMWWHm Aommva.:a o Ampavo.mm AJHVN.HH Aomva.m * A.Qmm m3~3m9mv nomad Ammmva.sa Aamvm.sm AowvH.OH Amsvm.mm Aooavs.om o amass emaaommm Angvm.bm A:va.ma Ahmavo.mm Ammv:.om Amwavo.mm flowvm.mm magma com mmmn< HH< sooa< psonwmao gnolnm< hamlumwmo mcfimlxwo mowoomm museum > >H HHH HH H hpwqdaaoo .cmmsoup msonadc u .mofiommm hp ummsoum .hpficssaoo comm pom noxamn mampm can vomSOAD mmwbp mo gonads HmpOp yo swapsoohom .m manna 29 .mo. v a .pcaofiaacmHm 8. Amsowpomnmpsfi honwwn 0mm.OH P: omm.mm: can uouum vmaoomv Hdddwmmm Hmm. mom.m ma oms.ema aszuosp scape . . . osfimnxdo scam *mom m :mw mm m mm: hm ooqdpmwn N mpcmapwmne . . . namnudvoo scum *mam m mmm mm m war Pm mocwpmwo x mucospwosa msowpomsmpaH hotness mam. smm.m m amm.m oqsmuaao song monapmHa *mww.> mmm.ow m :zw.ama haulhdvoo aosw oosdpmwn .806 5.3 a 5.8 $3 .383 3858; mac. boa. H NwH. mdoh< mpommwm saw: mmm.:om H wmm.:ow Amcowpwosoo unannouv mxooam a vanadm ado: mo mondsvm Mo Sam nodeth> mo oohdom .m.o + ax\xoucH xowhfix ma mapmwad> pcocsmmom .mpmc hm>h3m Momhp o>wmsmpxo mo mwmhadq< .m wanna 30 Therefore, selected contrasts among cell means were made, using a Scheffe (1953) test. These indicate that increasing distance from lowland conifers has a significant (p < .05) depressing effect on hare utilization in uncut communities. The response is parallel but non-significant (p > .05) for clearcut communities. The only significant (p < .05) difference between cut and uncut response occurred during inclement weather and farther than 1:00 m from lowland conifers (Figure h). Clearcut communities, regardless of their location, had less utilization as entities than did canopied areas; however, clearcuttings near lowland coniferous cover had significantly higher utilization than those away, and did not differ significantly from canopied areas. 31 .mhm>sdm xowup m>fimcopxo you hHMIhadoo scum moswpmwc and Apsoss an psov mpcmapdwnp hp sows“ xodsu wwahommcdnp aw moosonmmmfiu ado: .: madman A 3!. 25:53 uozmmuhma 7:-.. + wmzoawmm z. wozmmwmua z