Canopy , snow , and lichens on woodland caribou range in southeastern Manitoba

I examined the relationships among snow cover (api), l ichen abundance, and canopy composition on the range o f the Aikens Lake population o f woodland caribou (Rangifer tarandus caribou) i n southeastern Mani toba. Percent cover o f forage lichens (Cladina spp.) was positively correlated w i t h max imum total thickness and w i t h max imum vertical hardness o f api. M i x e d communities o f trembling aspen (Populus tremuloides), spruce (Picea spp.), and balsam fir (Abies balsamea) showed the most favourable nival conditions for caribou but had l o w lichen abundance; those dominated by jack pine (Pinus banksiana) were the converse. The results suggest an energetic compromise for woodland caribou when foraging for terrestrial lichens. D u r i n g winter, caribou exhibited significant selection for jack pine communities whereas mixed communities were avoided.


Introduction
Api (snow on the ground) and lichens are perhaps the most salient habitat features that influence the behaviour of caribou (Rangifer tarandus) in winter.
Caribou are well-adapted to both.Their broad hooves, low foot loading, and single-file mode of travel are characteristics that qualify them as chionophiles (Russell & Martell, 1984).During winter, lichens provide a unique, highly-digestible energy source and are often exploited as a principal forage (Luick, 1977;Russell & MarteU, 1984).
For Rangifer, api and hchens are opposing entries on the energetic balance sheet: travelling on and cratering through snow cover are often energetic investments for obtaining terrestrial lichens.Recent studies have begun to explore these energetic costs (Boertje, 1985;Fancy & White, 1985, 1987) and nutritional benefits (Person et ai, 1980;Thomas et ah, 1984) but, as yet, we know little of the association between these 2 important range features.
Such knowledge seems fundamental for managing habitat and modelling caribou foraging behaviour.
Here, I examine the relationships between api characteristics, lichen abundance, and canopy composition on the taiga range of the Aikens Lake population of woodland caribou (R. t. caribou) in Manitoba.

Materials and methods
Field investigations were conducted in the vicinity of Taiga Biological Station, Wallace Lake, in southeastern Manitoba (51°01'N, 95°40'W).The study area, part of the Northern Coniferous Section of the Boreal Forest Region (Rowe, 1972), was dominated by Precambrian outcrops with numerous intervening bogs and small lakes.Sampling was based on a map recognising 5 vegetation communities (Schaefer & Pruitt, 1991): mature spruce bog, jack pine forest, mixed coniferous forest, jack pine outcrop with intervening black spruce bog, and mixed deciduous forest.
The density of trees was determined at each quadrat by tallying individuals (> 2.5 m high) within a plot of radius 2.0 m.For analysis, each site was treated as an individual sampling unit and was represented by the average tree density and lichen coverage from 18 plots.
To display relationships between canopy composition and other habitat features, principal components analysis (PCA) was performed using SPSS/PC+ (Norusis, 1986) on the density of tree species.The effectiveness of PCA lies in its ability to reduce the number of dimensions to display such multidimensional data while retaining much of the original information (Pielou, 1984).Using PCA, therefore, density data of each of 5 tree species Betula papyrifera, and Abies balsamed) were reduced to one axis, the largest principal component (i.e., PCA axis 1).

Api Conditions
I restricted the analysis to the time when api thickness and hardness were maximal (early March) in 1985 and 1986.I selected a 6 x 50 m site, representative of each community type (Schaefer & Pruitt, 1991), where api profiles were placed haphazardly.
In each profile, vertical hardness (g-cnr 2 ) of the sur- face layer, using spring penetrometers, and total thickness (cm) were recorded (Pruitt, 1959).During 1985, 5 profiles were taken per site; during 1986, 4 profiles were taken.For each community type, the 2-year averages of maximum total thickness and maximum vertical hardness were extrapolated to all sites of the same habitat type.The lack of independence of api observations means that such sampling is pseudoreplicated (Hurlbert, 1984).Hence, although I used a Bonferonni-corrected a of 0.005 as an indication of significant linear correlations between variables (Norusis, 1986), these statistical inferences should be treated cautiously.

Results
Api variables, canopy composition, and the abundance of Cladina lichens were interrelated in the study area (Table 1).Maximum total thickness and maximum vertical hardness of the api were negatively correlated with the density of balsam fir (Abies balsamed) and positively correlated with that of jack pine (Pinus banksiana).Coverage of forage lichens (Cladina spp.) was similarly related; more lichen was associated with increased density of jack pine, less lichen with with increased density of fir.
Correspondingly, greater lichen abundance was positively correlated with both harder and thicker api (Table 1).We can compare the extremes in upland communities: on jack pine outcrops, where coverage of Cladina spp.averaged 17%, api was 80 g-cnr 2 (more than 100%) harder and almost 10 cm (32%) thicker than in mixed coniferous forests where lichen coverage was only 1.5%.
The first principal component from the ordination of overstorey species -tending to separate sites dominated by jack pine from mixed stands of balsam fir, spruce (Picea spp.), and trembling aspen (Populus tremuloides) -accounted for 34% of the total variation.When plotted against Cladina spp.abundance, the above interrelationships can be visualised (Fig. 1).The ordination axis, with mixed communities (types 1, 3, and 5) tending to the negative (left) end of the axis and jack pine habitats (types 2 and 4) to the positive (right) end, was positively correlated with lichen coverage (r = 0.69).The superimposed api parameters, hardness and thickness, show relationships to both overstorey and Cladina abundance (Fig. 1).Vertical api hardness was better correlated to the canopy ordination axis (r = 0.67), however, than was total thickness (r = 0.38).

Discussion
Snow cover is influenced by the type and density of the trees above it, a relationship which has been widely documented (e.g., Bunnell et al., 1985).In Total thickness (cm)

Canopy (PCA axis 1)
Fig. 1.Maximum total thickness and vertical hardness of api superimposed on a scattergram of lichen (Cladina spp.) abundance and the first principal component of canopy species.Jack pine communities tend to the right, mixed spruce/fir/aspen communities to the left.
general, greater coniferous cover reduces api thickness by providing greater snow interception and qali (Pruitt, 1957) formation.It reduces the surface hardness through wind interference and shading of the forest floor (Bunnell et ai, 1985).
In this study, the relation between maximum thickness of snow cover and jack pine abundance was counter to this general trend and, initially, appears counter-intuitive.The result may stem from the low sample size, but I offer 2 further, tentative explanations.First, the form of the jack pine -api relationship may not be strictly linear across the full range of jack pine densities.The extent of densities studied here (approx.0.5 -1.0 x 10 3 stems-ha' 1 ) was well below that in younger (37-year-old) stands in the study area (approx.2.5 -3.5 x 10 3 stems-ha -1 [Schaefer, unpubl. data]).Such dense, intermediate-aged stands, although not used by woodland caribou, appeared on cursory inspection to have thinner snow cover than old-growth communities.Coverage of Cladina spp., however, was similar (Schaefer & Pruitt, 1991).
Second, and perhaps more fundamental to our understanding, is that the relationships outlined here are probably scale-dependent.Results from many ecological studies (reviewed by Wiens, 1989) suggest that such correlations may depend on scale, i.e., the size of the sampling unit and the extent of sampling chosen by the investigator.Schaefer & Messier (1995), for example, found that correlations between arctic vegetation and snow cover (upsik) appeared stronger when larger sampling units were used in analysis.In this light, the positive association of jack pine and api may be counter to expectations only at the finer, microsite level.The extent of sampling for vegetation in this study was at the 'stand' level; i.e., I used a potential sampling area of 50 x 100 m at each site.At this scale, the increasing amount of jack pine at the expense of other species may be indicative of a relatively more open canopy.
LaPerriere & Lent (1977) have noted a hierarchical nature of habitat selection by caribou: animals first choose regional areas (such as a watershed), feeding areas (such as an outcrop), and finally a crater site (such as a lichen patch).This suggests that the present study of range relations has implications for the feeding area level of habitat selection, but may not validly be extrapolated to higher or lower scales.In general, there is a need for studies of how scale affects the interpretation of ecological phenomena.See Wiens (1989) for an excellent discussion on spatial scaling in ecology.
Nonetheless, the present results are in agreement with previous reports that tree species differ in their efficacy at altering snow cover (Bunnell et ah, 1985).
In the study area, balsam fir appears to be the most effective species at providing favourable api for Rangijer.This accords with the findings of Lull & Rushmore (1961, in Bunnell et at, 1985: 323) who found that balsam fir intercepted a relatively high proportion in snow.
Yet, interestingly, those habitats dominated by fir also contain a paucity of Cladina lichens as well as other winter forages for woodland caribou (Schaefer, 1988;Schaefer & Pruitt, 1991).Jack pine communities in old-growth condition had abundant lichen (Cladina spp.) but more severe api conditions.This is consistent with Stevenson & Hatler's (1985) generalisation that, for caribou, a trade-off exists between snow cover and forage abundance.
Rangifer, Special Issue No. 9,1996 How do woodland caribou deal with such a energetic compromise?Previous investigations (Pruitt, 1959;Bergerud, 1974) Pruitt, 1991).These results imply that added investment of cratering and travelling on harder, thicker api (Fancy & White, 1985;1987) (Harris, 1992), but removal of the canopy by logging can result in thicker and denser api (KirchofF & Schoen, 1987;Schwab et al, 1987) and, inferentially, greater energy costs for foraging caribou.In the study area, Schaefer & Pruitt (1991) found that api hardness and thickness were substantially greater in burned (i.e., 5-year-old) than in old-growth stands, even though much Cladina lichen remained intact within the burn.
The results here suggest an energetic compromise for caribou while foraging for Cladina lichens during winter.Further documentation of the relationship between forages and snow may improve our understanding of caribou foraging behaviour and management of their habitat.
density and lichen coverage of 5 vegetation communities, 90 -160 years of age, were sampled 12 June to 31 July 1985.Discrete areas of habitat within an accessible area were selected randomly.Within each discrete area, 2 sampling sites were picked randomly with a 1:15840 aerial photo and a grid overlay of points 50 m apart.At each site a 50m transect was established.Its exact origin was determined by the toss of a stone, its orientation (north or south) by the toss of a coin.At 3 random points along the transect, a 50-m branch transect was established with a random east or west directi-on.The abundance of lichens of the genus Cladina

Table 1
. Correlations between density of tree species, api characteristics, and lichen abundance in the Wallace-Aikens lakes study area in southeastern Manitoba, 1984-86.All correlations were significant (Bonferonni-corrected P < 0.005)." Lichen coverage is expressed as In (% cover of Cladina spp.+ 1).
have indicated that caribou select feeding sites primarily for abundance of forage, secondarily for suitability of nival conditi- The results may carry implications for the management of woodland caribou habitat.In particular, because of the influence of trees on snow cover, alteration of the canopy by fire or logging may indirectly affect the winter foraging behaviour of caribou.Logging operations, for instance, may have only negligible effects on the biomass of Cladina lichens