Groups versus individuals in the determination of caribou distribution

Studies of caribou (Rangifer tarandus) habitat selection based on group analyses have led to erroneous conclusions. Convenient designations such as «male-» or «female-dominated» group encompass a wide array of possible sizes and compositions which change continuously and erratically. Whenever individuals of at particular sex/age class can occur in more than one group type, and/or whenever groups within a type vary in size, an analysis based on groups alone is fallacious. Data must be based on individual caribou for most, if not all, determinations of distribution.


Introduction
Caribou (Rangifer tarandus) are social animals and usually occur in groups.Over short time periods, at least, caribou groups are cohesive units.Behavior of individuals within a group tends to be uniform and synchronous, and groups often respond to various stimuli as a unit.Also, many groups are relatively homogeneous, e.g., predominantly adult males or predominantly cow/calf pairs.Thus, it appears reasonable to analyze caribou behavior or distribution patterns by studying caribou groups.
There is another, more practical reason for recording data on groups rather than individual caribou.It is easier to assign an entire group some general category of composition, activity, or location than it is to record the same data for each individual in the group.This is particularly true when groups are large or far away, when the observer is flying rapidly past the group in an aircraft, and/or when the terrain does not permit full view of all individuals in a group.
However, certain characteristics of caribou groups render them inappropriate for describing the distribution of a population.In this paper we demonstrate how failure to consider variations in group size and composition has led to erroneous conclusions on differential habitat use by male and female caribou.Lent (1965) and Bergerud (1974) concluded that caribou groups are loose social units characterized by temporary and tenuous social bonds.Recent radio-telemetry studies in Alaska and the Yukon Territory indicate that caribou captured together frequently change groups and seldom reassociate with each other (Valkenburg et al, 1983;D. Russell and R. Farnell, pers. comm.) Miller (1974) and Miller et al (1975), on the other hand, concluded that small groups, or «winter bands», are the basic units of caribou social organization, and that affinity to these bands persists for many years, perhaps for life; caribou groups observed during the rest of the year are either aggregations or fragments of these winter bands.Thus, the concept of winter bands is not entirely inconsistent with the general conclusion that caribou groups are dynamic.

Caribou groups
Rangifer, SpeciaMssue No. 1,1986 Virtually all studies of caribou social organization confirm that group size and composition change seasonally and that, within any one season, groups vary considerably in both size and composition.Large groups generally behave differently from small groups, and male-dominated groups behave differently from femaledominated groups.Moreover, convenient categories of group type may sometimes be inappropriate.For example, the distribution of males cannot be determined by examining only the distribution of male-dominated groups because males occure in female-dominated groups as well.

Differential habitat use and reactions to disturbance by male and female caribou
Recently, Carruthers et al (1984) and Curatolo (1985) concluded that female and calf caribou in the Central Arctic region of Alaska avoid riparian habitats.These authors further suggested that one would therefore expect to find few cows and calves along the Trans-Alaska Pipeline System (TAPS) where it is closely associated with the Sagavanirktok River.Such conclusions conflict with our earlier findings that the proportion of cows and calves among caribou along the Dalton Highway (TAPS haul road) was lower than that among caribou observed along all major rivers and the coastline within the range of the Central Arctic Herd (CAH) (Cameron et al., 1979;Cameron and Whitten, 1980;Whitten and Cameron, 1983).If the conclusions of Carruthers et al. (1984) and Curatolo (1985) are valid, we should have seen relatively few cows and calves anywhere along our survey route, except possibly along the coast.This was decidedly not the case.How then could our conclusions be so different?

Analyses based on groups
Both Carruthers et al. (1984) and Curatolo (1985) assumed that the distribution of male and female caribou could be determined by sampling the distribution of male-and female-dominated groups.Both used similar categories of group type.Curatolo (1985) defined a cow group as comprising more than 70% cow/calf pairs and a bull group as more than 70% bulls.Carruthers et al. (1984) defined a female group as one in which more than 67% of the caribou were classified, and more than 67% of the classified adults were female; in a male group, more than 67% were classified, and of those, more than 67% were males.Under either definition, some bulls could occur in cow groups, and vice versa.In fact, by the Carruthers et al. (1984) definition, a male or female group could theoretically contain mostly members of the opposite sex (e.g., 0.67 x 0.67 = minimum of 44% males in a male group).Differences in group size were not considered in either study, and all groups were weighted equally in determining habitat use.Both studies demonstrated that a significantly higher proportion of male groups than of female groups was found in riparian habitat.
Riparian habitats covered 9% of the regional aerial strip transects used by Carruthers et al. (1984) to determine caribou distribution.More than 9% of the male groups were in riparian habitat during all seasons, and less than 9% of the female groups used riparian areas during most seasons (Table 1).Furthermore, female groups were, on average, farther than expected from riparian habitats (based on mean distance available) while male groups were usually closer.Thus, Carruthers et al (1984) concluded that females avoided areas within and near riparian habitats.Curatolo (1985) reached the same conclusion based on the observation that 37% of bull groups were in riparian habitat during summer, versus only 19% of the cow/calf groups.
These results can be extrapolated to the population as a whole only if individual male and female caribou were distributed similarly to male and female groups.The problems outlined above suggest that such an assumption is, at best, tenuous.Curatolo (1985) reported only summarized results, but Carruthers et al. (1984) presented sufficient data to test this assumption.

Analyses based on individuals
Reanalysis of the data presented by Carruthers et al. (1984) indicates that the distribution of individual male and female caribou was quite different from the distribution of male-and female-dominated groups.On average, female groups were larger than male groups.Assuming that group sizes were similar among habitats, caribou in female groups would have outnumbered those in male groups in riparian areas during much of the year, even though a smaller proportion of female groups than of male groups was observed in riparian habitat (Table 1).Carruthers et al. (1984) reported numbers and composition of individual caribou for only 1 year of their 3-year study.In 1983, they compared caribou observed during regional aerial surveys with caribou observed along a transect flown directly over TAPS, but unfortunately reported data on caribou within 2 km of riparian habitat rather than in riparian habitat only.An analysis based on the distribution of individuals shows that caribou consistently preferred riparian areas, both along TAPS and in the surrounding region (Table 2).Use of areas in or near riparian habitats by all caribou and by calves was high relative to availability of riparian habitat.The higher proportions of individual caribou in or near riparian habitats (Table 2), than of caribou groups in riparian habitats (Table 1), could indicate that groups in riparian habitats were larger, or that many groups occurred close to (<2 km), if not within, the riparian zone.Either case negates the conclusion that one would expect to see few cows and calves along a major river.
Most striking, however, is that percentage calves among caribou along TAPS was consistently lower than in the surrounding region (Table 2).Within or near riparian habitat, calves were less abundant along TAPS except during calving.In nonriparian habitats, calves were less abundant along TAPS except during August dispersal and pre-rut.
In summary, there is no clear evidence that calves (and, by inference, their mothers) consistently used riparian habitats differently than did other caribou.All caribou preferred riparian habitats, and cow/calf pairs avoided all habitats along TAPS.This reanalysis of the findings of Carruthers et al. (1984) corroborates our own conclusions (based on data from individual caribou) that cow and calf caribou are underrepresented along TAPS, but not in similar habitats elsewhere.Local disturbance by traffic and construction activity is likely the major cause of this cow/calf avoidance (Cameron et al, 1979;Cameron and Whitten, 1980;Whitten and Cameron, 1983).Curatolo (1985) reported the distribution of individual CAH caribou in demonstrating that calf percentages in study sites along the Kuparuk River were only ca.40% of those in nonriparian study sites farther west near Oliktok Point.Petroleum development was considered to have influenced distribution in both areas equally.Therefore, Curatolo (1985) concluded that the relative scarcity of cows and calves along the Kuparuk River was due to their natural avoidance of riparian areas, a conclusion supported elsewhere in his paper by the analysis of caribou group distribution.Curatolo failed to note, however, that the area near the Kuparuk River had been disturbed for at least 10 years before his study began in 1982, whereas the Oliktok Road in the nonriparian area had been built just that year.Furthermore, calf percentages in the Kuparuk River area declined from levels similar to regional estimates in 1978 to ca. 40% of regional estimates by 1980 (Cameron et al. 1981).Thus, we believe disturbance to be the primary cause of cow/calf underrepresentation along the Kuparuk River, rather than any natural avoidance of riparian areas.
The cause hypothesized by both Carruthers et al. (1984) and Curatolo (1985) for cow/calf avoidance of riparian habitats in general (i.e., instinctive avoidance of predator-concealing habitat) is not operative in the range of the CAH.While predators may ambush caribou in tall riparian willow (Salix spp.) stands in the ranges of many herds (Curatolo, 1975;Bergerud, 1974;Boertje, 1981), both predators and tall willows are scarce on the summer range of the CAH.Riparian areas are more often characterized by terraces with cushion tundra vegetation consisting mainly of Dryas, legumes, forbs, and low (<0.5 m) willows.Tussock or wet sedge tundra often extends to the very edge of a watercourse, and islands in braided streams often have typical tundra vegetation.Lack of predators and a mosaic of habitat types, including both foraging areas and vegetation-free areas for insect relief, may in fact explain the preference for riparian areas noted here for all caribou, including cows and calves.

Conclusions
Data based on groups are not appropriate for describing population distribution or habitat preference.Our reanalysis of the data of Carruthers et al. (1984) demonstrates that individual bulls and/or cow/calf pairs were not distributed similarly to male-and female-dominated groups.Extrapolations based on treatment of groups as equivalent units yielded erroneous results.By inference, Curatolo's (1985) conclusions based on similar methodology are also suspect.
Few researchers would make the mistake of treating unequal areas as equivalent sample units.For example, no one would compare caribou density in two areas by determining the number of caribou per linear km using 1-km wide transects in one area and 2-km wide transects in the other.Yet this is precisely the sort of error made when caribou distribution is described by weighting all groups equally.
Caribou groups vary in size and composition, and members of the same sex/age class usually occur in more than one group type.Therefore, an analysis based on groups alone will not be representative of a population.Individual caribou must serve as the basis for most, if not all, studies of distribution.
are grateful to J.D. Wadland, Fairbanks; R.D. Boertje, W.L. Regelin, and W.T. Smith, Alaska Department of Fish and Game; M.C.S. Kingsley, A.M. Martell, and J .Smith, Canadian Wildlife Service, for their critical reviews of the manuscript.

Table 1 .
Distribution of caribou in male and female groups relative to riparian habitat, Central Arctic Slope, Alaska.3Occurrence in riparian habitat b b9% riparian habitat available.c (number of groups) X (mean group size).Does not take into account possible differences in group size between riparian and nonriparian habitats.d Group type.

Table 2 .
Distribution of individual caribou relative to habitat type along the Trans-Alaska Pipeline (TAPS) and in the surrounding region, Central Arctic Slope, Alaska 1 .