Predation rate by wolves on the Porcupine caribou herd

Large migratory catibou {Rangifer tarandus) herds in the Arctic tend to be cyclic, and population trends are mainly driven by changes in forage or weather events, not by predation. We estimated daily kill rate by wolves on adult caribou in winter, then constructed a time and space dependent model to estimate annual wolf (Canis lupus) predation rate (P annual) on adult Porcupine caribou. Our model adjusts predation seasonally depending on caribou distribution: Pannual = SIGMAdaily* W *Ap(2)*Dp. In our model we assumed that wolves killed adult caribou at a constant rate (Kdaily, 0.08 caribou wolf1 day1) based on our studies and elsewhere; that wolf density (W) doubled to 6 wolves 1000 km2-1 on all seasonal ranges; and that the average area occupied by the Porcupine caribou herd (PCH) in eight seasonal life cycle periods (Dp ) was two times gteater than the area described by the outer boundaries of telemetry data (Ap /1000 km2). Results from our model projected that wolves kill about 7600 adult caribou each year, regardless of herd size. The model estimated that wolves removed 5.8 to 7.4% of adult caribou as the herd declined in the 1990s. Our predation rate model supports the hypothesis of Bergerud that spacing away by caribou is an effective anti-predatory strategy that greatly reduces wolf predation on adult caribou in the spring and summer.


Introduction
Migratory barren-ground caribou {Rangifer tarandus) herds show wide population fluctuations that have been explained by changes in forage, climate, predation and harvest (as reviewed in Klein, 1991).
Various researchers have pointed out the difficulty of separating interactions of forage-climate-predation when trying to determine the cause of change in caribou abundance (Gauthier & Theberge, 1986;Thomas, 1995;Adams etal., 1995;Bergerud, 1996;National Research Council, 1997).The effects of wolf {Canis lupus) predation on migratory barrenground caribou were poorly understood in the past, mainly because arctic wolves were migratory and difficult to follow (Kuyt, 1972;Stephenson & James, 1982).Recent studies in arctic Alaska (Dale et al, 1994;Ballard et al, 1997) and Canada (P. Clarkson, Government of the Northwest Territories, unpubl.;R. Hayes, unpubl.)provide new data about arctic wolf movements, range use and their killing rates on caribou.These data were required their to develop quantitative models for estimating predation rates on migratory caribou herds.
In this paper, we present data on winter kill rate by wolves on adult caribou when Porcupine numbers were high.We construct a simple predation rate model that includes constants for wolf density and kill rate that are applied to changing seasonal range use and densities of caribou.We discuss why predation by wolves is not the main force limiting the size of the Porcupine herd in the 1990s.

Study area
We conducted our predation rate research in 1989 in a 14 450 km 2 study area in the Northern Richardson Mountains.Predation studies that wintet were part of a larger study of wolf ecology conducted between 1987 and 1993 in the northern Yukon (R. Hayes, unpubl.).
Our study area straddled the northern boundary of the Yukon and Northwest Territories (NWT).
The main study area included the Northern  (878, Statistics Canada 1996).
We studied winter kill rate across 3 ecoregions (Oswald & Senyk, 1977): the Northern Mountains, the Coastal Plain, and Berry Creek.We have paraphrased descriptions of physiography and vegetation from Oswald and Senyk (1977) Moose density is low and most moose winter in the limited riparian forests along the Bell River (Smits, 1991).Few moose wintered in the north slope drainages, where we conducted most of predation studies.In the same area, Barichello et al. (1987)  Other large predators in the study area include brown bear (Ursus arctos) (Nagy, 1990), black bear (Ursus americanus) in the taiga, lynx (Lynx canadensis) and wolverine (Gulo gulo).Arctic fox (Alopex lagopus innuitus) are resrricted to coastal areas (Youngman, 1975).Ravens (Corvus corax) are the main scavengers that compete with wolves at kills.

Materials and methods
We used radiotelemetry techniques (Mech, 1974) to study predation behaviour of wolves.After we first We studied kill rates by monitoring the daily activities of seven radio-collared packs from 23 March to 16 April 1989 from a Maule LR7 aircraft.
We defined pack size as the mean number of wolves seen in the period (Messier, 1994;Dale et al, 1994;1995;Hayes et al, 2000).We defined kill rate as the number of caribou killed per wolf per day.The total biomass (kg) of caribou killed was used to measure consumption rates of wolves.Based on data from Skoog (1968) we estimated the live weights of adult caribou: male 107 kg, female 79 kg and unknown caribou 86 kg.We assumed the consumable biomass was 75% of caribou live weight (Ballard et al, 1987;1997).

Kill rate by wolves
We followed the daily activities of seven wolf packs for 17.1 ± 3.1 (standard error of the mean) days (Table 1).Traveling pack size was 4.4 ± 1.4, ranging from 2 to 12 wolves per pack.We found 23 wolf-killed caribou and we examined 13 carcasses in situ.All were adults (8M, 5F).The mean age of killed caribou was 6.1 ± 0.7 years-old.The lowest kill rate was for wolves in the Rat River II pack (Table 1) which scavenged from many hunter kills in the area.After excluding this pack, we estimated the wolf kill rate was 0.08 ± 0.03 caribou per day per wolf; or 7.5 ± 2.7 kg of caribou killed per wolf per day.Wolves consumed 5.6 ± 2.0 kg caribou each day in winrer.
We did not find a difference in the number of kills seen for morning-only sightings of Blow River wolves compared to the combined morning and evening sightings (n = 9 kills, 0.36 caribou per pack per day).We conclude that twice daily locations did not improve our ability to detect kills made by study packs.

Predation rate by wolves
Based on a daily kill rate of 0.08 adulr caribou (Kjaiiy), our model projected that wolves killed 7600 adult caribou from the Porcupine herd each year.
About 84% of the adults were killed during fall and    2.

Kill rate by wolves
The daily kill rate of our study wolves was similar to caribou-killing wolves in Alaska (0.08 caribou per wolf per day, Dale et al., 1994) and Northwest Territories (0.05 caribou, P. Clarkson, unpubl.),although our pack kill rates were more variable.We studied wolf kill rate in mainly small packs of 2-3 wolves (Table 1).Hayes et al. (2000) found wolves in small packs had much wider variation in kill rate of moose compared to larger packs, which could also explain our caribou predation data.
The mean daily consumption rate was 4.9 kg of caribou per wolf, above the range of 1.7 to 4.0 kg 54 required for survival (Mech, 1977;Thurber & Peterson, 1993) and above the 3.2 kg required for reproduction (Mech, 1977).Similar consumption rates were recorded for arctic wolves in northwestern Alaska (5.3 kg of moose and caribou, Ballard et al, 1997) and NWT (4.4 kg, P. Clarkson, unpubl.).

Predation rate model
We verified our model assumptions by looking at caribou and wolf studies elsewhere.Our study, Dale et al. (1994) and P. Clarkson (unpubl.) reported kill rates of 0.05-0.08caribou wolf 1 day 1 .Thus, we believe that substantial changes to the value for variable K^.i, are not justified.Out study, Parker (1973), Kuyt (1972), Thomas (1995) and Clarkson & Liepins (unpubl.)all found a two-fold increase in wolf density on winter range.We had substantial telemetry data to evaluate seasonal PCH distribution for over twenty years.Thus, we could not justify increasing the areas of available caribou more than two-fold.Our model does not incorporate changing vulnerability to predarion, which Mech et al. (1998) found was an important function of wolf predation rate on the Denali caribou herd.
We next examined how our predation rate fit current knowledge of Porcupine caribou ecology.Fancy et al. (1994) found mean adult morrality rate for >3year-old caribou was 15% for females and 17% for Rangifer, Special Issue No. 12, 2000 males.Using our 1992 wolf predation rate estimate of 5.8%, our model projects that wolves were responsible for about 1/3 of the adult mortality in the early 1990s.
According to Fancy et al. (1994) and Walsh et al. (1995) the growth of the PCH is most sensitive to the survival rates of females three years and older, followed by production and survival rares of calves.Using different predation rate models, Dale et al. (1994) and Ballard et al. (1997) also determined that predation by wolves was not the main factor limiting caribou in northwestern Alaska.Ballard et al. (1997) estimated that wolves annually removed about 6-7% of the Western Arctic caribou herd.
Predation by wolves is an important factor limiting smaller caribou herds in Canada and Alaska (Gasaway et al, 1983;Bergerud & Elliot, 1986;Edmonds, 1988;Seip, 1992;Hayes & Gunson, 1995;Mech et al, 1998).Current knowledge suggests wolf predation acts in a depensatory fashion (i.e., it increases as herd size declined) where caribou are secondary prey to wolves that rely primarily on moose.Wolf predation does not appear to be the main cause of population change for large migratory caribou herds in the arctic (Messier, 1995;Crete & Huot, 1993;Thomas, 1995).Large migratory caribou herds tend to be cyclic, and previous population trends have been linked to changes in forage or weather events (Crete & Huot, 1993;Fancy et al, 1994;Messier, 1995).
The low effect of predation by wolves is supported by the hypothesis of Bergerud (1974), who has argued that the migratory behavior of caribou evolved as a predator-avoidance strategy.Bergerud (1992) believes that migratory caribou calve on small remote areas to 'space away' from predators.
By doing so, they can flood a large number of young in a small area where the per capita risk to being killed by any predator is lowest.
Our model does not estimate predation rate on calves, however, it does supports that 'spacing away' is also an effective anti-predatory strategy of adult caribou (Bergerud, 1974;1992;Thomas, 1995).In late spring and summer, Porcupine caribou concentrate on the coastal plain of Alaska and Yukon, where they occupy the smallest seasonal range, thereby reducing their exposure to predators (Table 2).Adult wolves are limited in their ability to travel there due to their requirement to feed pups at dens (Thomas, 1995;R. Hayes, unpubl. data).Fryxell et al. (1988)

Data quality
Although our estimate of mean daily kill rate was similar to other studies, it was bounded by a wide standard error.This could be because the sample size of packs was small, or the kill rate was undetestimated for some packs due to terrain or weather constraints.
We acknowledge some shortcomings with our predation rate model.Although the model fits current indices of the PCH, components of the model need further validation.First, we assumed that Kda,iy in the summer period was the same as for winter.
Wolves are reported to surplus kill neonatal and adult caribou (Miller et al, 1983;1988;C. Gardner, Alaska Dep. Fish and Game, pers. comm.).The effecr of wolf predation rate on changing calf recruitment rates of the Porcupine herd remains unknown, and we did not include this important population process in our model.
Second, the estimates of the area that caribou occupy seasonally are based on radiotelemetry loca-56 tions.There is a declining gradient outward from these areas where low density caribou will still be available to wolves.We estimated caribou-available areas to be twice the areas described by caribou telemetry, but the area might be even larger.
However, we needed to increase the caribou available area in our model by five-fold before wolves took 10% or more of the adults.Third, arcric wolves show strong preference for caribou, and wolves probably continue to search for caribou even when caribou appear to be absent (P.Clarkson, pers. comm.).If PCH wolves behave this way, then our estimates of seasonal predation rates could also be low. Nevertheless . Co., Douglasville, Ga.) equipment.Most wolves received a dose of Zoletil (A. H. Robins) at 8 mg/kg, based on an estimated average wolf weight of 40 kg.We attached conventional VHF radio-collars on wolves (Telonics, Mesa, Arizona).

Fig. 1 .
Fig. 1.Seasonal predation rate by wolves on PCH based on model.Seasonal periods correspond with numbers shown on Table2.
Hayes et al. (2000) adjusted kill rates to account for raven scavenging, estimating that ravens can remove up to half of consumable moose biomassRangifer, Special Issue No. 12, 2000 from small wolf packs (2-3 wolves).Five of our study packs were small and we commonly saw ravens ar caribou kills.However, we agree withBallard et al. (1997) who estimated that wolves lost less of their caribou kills to ravens because wolves can consume caribou carcasses more rapidly than they can consume moose -leaving less caribou biomass for scavengers.By back-tracking wolf trails,Dale et al. (1994) increased their estimate of kill rate because wolves killed then left the caribou carcasses before the next radio location.Hayes et al. (2000) underestimated kill rate by wolves on woodland caribou by locating packs once daily, and recommended back-tracking whenever possible.Clarkson and Liepens (unpubl.data) believed that arctic wolves remained close to their kills in order to protect them from other migratory packs, therefore, back-tracking was not useful in tundra areas.Without backtracking we recorded a similar kill rate as Dale et al. (1994) did with backtracking.We had the advantage of studying small migratory packs that traveled in open tundra areas, which probably remained near kills for defense purposes (P.Clarkson, unpubl.data).Increasing our observation rate to each morning and evening did not increase our ability to detect caribou kills made by a pack of 12 wolves.Despite the windblown conditions, we reasonably estimated kill rate of our study packs on Porcupine caribou winter range.

Fancy
et al. (1994)  speculated that the decline of the PCH after 1989 was related to a combination of low parturition rate of >3-year-old females in 1991, and lowered calf survival in March 1992.Using stochastic modeling,Walsh et al. (1995) showed that a survival rate decline of about 3% among adult females or 4% among calves could be enough to cause the Porcupine herd to decline.Our model projects that wolves would have to nearly double their predation rate to account for an addirional 3% decline in adult female survival.

Table 1 .
Killing rates by wolves on caribou in our study,March and April 1989.
Most packs traveled in the north slope drainages where snow conditions were heavily windblown in 1989-Wolves and their prey carcasses were difficult to see because of the contrasting mosaic of open ground and snow fields.Snow was usually too windpacked to backtrack wolves to determine their activities between location points.We located radio-collared wolves, then systematically searched for any kills in a 2-3 km 2 area, until we either found kills or we were confident wolves had not made a kill nearby.We estimated annual predation rate as the proportion of adult Porcupine caribou killed by wolves.To determine the rate of wolf predation on the Porcupine herd we needed a model that was based on reasonable ecological assumptions about wolves and caribou.From wolf surveys in the northern estimate predation rate by simply applying a fixed kill daily rate to the entire wolf population.To account for changing distributions of caribou and wolves, both in space and time, we constructed the model for estimating annual predation (J?annual)'.Pannual = ZKdai.y*W*Ap(2)*Dp.ineightseasonallifecycle periods (Dp , see Table2)was twice as large as the average area described by the outer boundaries of satellite telemetry data (Ap /1000 km 2 ; Int.Porcupine Caribou Board 1993).

Table 2 .
Variables and values used in modeling annual wolf predation rare on Porcupine caribou herd.Values for D;,and Ap -were provided by Inr.Porcupine Caribou Board (1993).
winter (Table2, Fig.1) when caribou use the largest areas, allowing more wolves to concentrate on fall and winter range.The remaining 16% of adults were taken in spring and fall when the herd's range is substantially compressed, and their availability to wolves is lowest (Table2, Fig.1).
, our results are consistent with other arctic wolf studies that found a uniquely migratory behaviour among wolves associated with barren-ground caribou, naturally low wolf densities, a preference for caribou prey, and moderate daily kill rates by wolves.The model we present is based on detailed knowledge of a dynamic seasonal range use pattern by Porcupine caribou that was available only after decades of radiotelemetry studies.Future predation research should be conducted to investigate whether the assumptions of our model hold in this period of declined herd size.