NATIONAL PARK SERVICE Demography of Grizzly Bears in Relation to Hunting and Mining Development in Northwestern Alaska

Capturing Bears

We used helicopters to capture bears for radio-collaring or marking and immobilized the animals with a widely used procedure (Spraker et al. 1981; Ballard et al. 1982; Reynolds and Hechtel 1985; Miller et al. 1987). During 1986, we used phencyclidine hydrochloride (Sernylan, Bio-Ceutic Laboratory, St. Joseph, Missouri) or etorphine hydrochloride (M-99, Lemmon Co., Sellersville, Pennsylvania). After 1986, we used a mixture of tiletamine hydrochloride and zolazepam hydrochloride (Telazol, A. H. Robbins Co., Richmond, Virginia). The drugs were delivered from a dart projectile fired from a Cap-Chur gun (Palmer Chemical Equipment Co., Douglasville, Georgia) or by hand injection. A bear was considered immobile if it was lying on its sternum and was workable for processing. Induction was the time from initial injection to immobilization. The sex, weight, and measurements were recorded for each captured bear, and each was marked with from 1 to 3 lip tattoos and duflex or roto ear tags; they were also collared with radios manufactured by Telonics, Inc. (Mesa, Arizona) if judged to be more than 5 years old. Several subadult (probably 3-5-year-old) bears were radio-collared during the census with collars designed to fall off after several weeks. Two sets of standard hardware attachments, one on each end, were used instead of one. The ends were connected by inserting surgical tubing snugly under each attachment. Premolars were extracted from each immobilized bear judged to be over 1 year old.

All bears, except cubs of the year (COY), had more than one premolar extracted for age determination. Teeth obtained from 1986 to 1988 were cut, sectioned, stained, and the age determined using methods described by Goodwin and Ballard (1985). Beginning in 1989, all teeth were sectioned, stained with a Giemsa stain, and age was determined by Matson's Laboratory (Milltown, Montana). Blood samples were collected from each adult bear for determining percent hemoglobin and packed-cell volume. Sera were separated and frozen to be saved for future analyses of physical condition and for surveys for microbial pathogens. Each bear was administered an injection of antibiotic to reduce the risk of infection associated with capturing and handling. Following processing, each bear was left lying on its sternum. We used fixed-wing aircraft to observe the animal several hours after immobilization to determine whether it had moved from the capture site.

We attempted to capture all members of family groups including COY. We lightly sedated COY by hand injection with a syringe. Immobilized COY were easier to process, and abandonment rates were lower when both sows and COY were immobilized.

Telemetry Relocations

During 1987, we selected radio-collared bears for relocation weekly. We subjectively selected bears from areas that would be affected or altered by the Red Dog Mine. Other bears were relocated two or three times during summer to monitor status and survival of young and twice in late fall to locate den sites. At each relocation, the date, time, number, sex and age of associates, activity, and type of habitat were recorded on standard forms. Habitat classifications were based on overstory vegetation that could be identified from aircraft. Generally, vegetation was classified within 400 m of instrumented individuals using classifications of vegetation described by Viereck and Dyrness (1980). The vegetation type was assigned to reflect the dominant overstory vegetation. Prey and carrion observed while attempting to relocate bears were recorded. Prey carcasses observed at the location of radio-collared bears were considered to have been killed by that bear if there was fresh blood, an intact carcass, and absence of other bears or predators.

We determined slope, aspect, and elevational use by radio-collared grizzly bears by recording locations on 1:63,360-scale topographic maps after each flight. We determined elevation by extrapolating between contour lines to the nearest 15-m interval. We classified slopes into categories using contour line intervals: (1) flat, less than 10°; (2) gentle, 11-30°; and (3) moderate, more than 30°. Aspect was classified as 1 of 8 compass directions from a line perpendicular to the contour lines through the bear relocation point.

Data Analysis

We determined survival rates by Kaplan—Meier procedures (Pollock et al. 1989) using radio-collared adults and uncollared COY and yearlings accompanying radio-collared adult females. The procedure allows radio-collared animals to be entered at different time intervals (Pollock et al. 1989). Animals may be censored from the data set when a radio collar fails or the fate of the animal is not known. Assumptions of the method include: a particular sex and age class has been randomly sampled, survival times are independent for different animals, tagging and radio-collaring do not influence survival, censoring is not related to an animal's fate, and newly tagged animals have the same survival function as previously tagged animals. The procedure is simple and flexible and allows staggered entry of tagged animals (Pollock et al. 1989). A major disadvantage of the method is that precision is low when the sample size is less than 20.

We used the McPAAL software package (M. Stuwe and C. E. Blohowiak, Conservation Research Center, National Zoological Park, Smithsonian Institution, Front Royal, Virginia, personal communication) to estimate home range sizes by the following methods: convex polygon (Mohr 1947), concave polygon, 95% ellipse (Jennrich and Turner 1969; Koeppl et al. 1975), and harmonic mean (Dixon and Chapman 1980). For the harmonic mean transformation we calculated home range sizes based on 80, 90, and 95% of the relocations. Home range sizes were only calculated for bears that had more than 10 relocations. We used the convex polygon method for comparisons unless stated otherwise.

Satellite Telemetry

In early June 1988, transmitters compatible with the Argos Data Collection and Location System (hereinafter referred to as satellite transmitters) were deployed on 6 adult females that had been monitored for 1-2 years using conventional very high frequency (VHF) telemetry. The transmitter package, including a separate VHF transmitter attached to a neck collar, weighed 1.8 kg and was manufactured by Telonics, Inc. (Mesa, Arizona). The satellite transmitters, known as platform transmitter terminals (PTT's), were programmed to transmit 6 h/day between 25 May and 10 October and were expected to operate during a second 5-month season after bears emerged from their dens. The history and current use of PTT's for locating and monitoring behavior of wildlife in Alaska has been described by Fancy et al. (1988, 1989, 1990) and Harris et al. (1990).

Each PTT can be programmed to transmit at varying intervals for up to four different transmission schedules. A 6-h transmission is thought to be an optimum length to allow the satellite sufficient opportunity to consistently fix at least one accurate relocation while maximizing battery life (W. P. Burger, Telonics, Inc., Mesa, Arizona, personal communication). Users are provided, on a monthly basis, microcomputer diskettes that contain all of the relocations and data from other sensors on the collars. Users can usually obtain relocations by modem within 6 h after a satellite overpass.

In addition to the information needed to determine the bear's location, the transmitted signal included the ambient temperature and data from a thermistor and a mercury tip-switch within the transmitter canister that provided information concerning the animal's activity. The 60-s activity index, which ranged from 0 to 60 s, indicated the number of seconds during the minute before transmission that the mercury tip-switch was activated. The 24-h activity index accumulated the 60-s counts for a 24-h period. The orientation of the tip-switch that best discriminates between activities differs for each species (Harris et al. 1990; Ballard et al. 1991). The anterior end of the tip-switch used in our PTT's was aligned —8° relative to the circuit board. No data are available from grizzly bears to relate activity counts to specific activities or to determine if —8° is the optimum angle for bears.

Density and Population Estimates

Except where stated, the method for censusing bears is mark—resight using radiotelemetry to correct for population closure (number and identification of individual radio-collared bears that were either in or out of individual count areas; Miller et al. 1987). A fixed-wing aircraft is used to thoroughly search (without aid of telemetry) individual count areas until a bear or bears are spotted. Once spotted, radiotelemetry is used to determine whether the animal is marked (i.e., radio-collared). Only sightings of bears with functioning radio collars are considered as resightings of marked individuals; however, for some sets of population and density estimates that are identified later, we considered young accompanied by their mothers to have the same status as their mothers. If a bear did not wear a functioning radio collar, we considered it unmarked. If unmarked, the location of the bear was transmitted to a nearby helicopter crew that immobilized it. Once immobilized and radio-collared, the bear was potentially available as a marked individual in subsequent searches.

Effort was made to capture all unmarked adult bears but not the subadults accompanying their mothers. All unmarked adults were captured, with the exception of one adult female accompanied by one 2.5-year-old (estimate based on size) that escaped. Because the census occurred during the breeding season, adults were sometimes observed together. These sightings were treated as independent observations.

Equations for calculating population size, density, and associated confidence intervals were provided by Miller et al. (1987:25-26).

Calculation of population estimates followed Seber (1982), where

However, instead of using the daily values of n₁, n₂, and n₂, as would be done if the population were closed, we obtained values used for these parameters by cumulating the daily values recorded during the capture period. This resulted in a different population estimator, N_d*. We defined N_d*, conceptually, as the total bear-days our search area was occupied during the search period. The average number of bears that inhabited the search area during a search period of n days was then (N_d*/n). Substituting N_d* for N* in equation 1 required redefining the parameters of equation 1 as

n₁ = radio-marked bear-days in the study area during a study period of n days as determined by telemetry (1 radio-marked bear verified in the study area during 1 day = 1 marked bear-day present);

n₂ = bear-days observed by spotters in planes during a study period of n days (1 bear, marked or unmarked, seen in any 1 day = 1 bear-day observed); and

m₂ = radio-marked bear-days observed by the spotters in planes during a study period of n days.

Confidence intervals for N_d* were similarly calculated by substituting the previously defined values of n₁, n₂, and m₂ into the appropriate equations provided by Seber (1982). These were approximations to the distribution based on the binomial or normal distributions. Seber (1982) recommended criteria for choosing which distribution to use based on the values of n₂ and p*, where p* was estimated as (m₂/n₂).

When the normal approximation was appropriate according to the criteria, the variance of N_d* was calculated according to the formula given by Seber (1982):

Confidence intervals for circumstances when the binomial approximation to the hypergeometric distribution was appropriate, according to criteria given by Seber (1982), were calculated using Clopper—Pearson graphs (example in Overton and Davis 1969:413). Using p* as the entering variable on the x-axis of the Clopper—Pearson graph, corresponding values for upper (p_u) and lower (p_l) limits that were associated with the isoclines for n₂were read from the y-axis of the Clopper—Pearson graph. Then the upper and lower limits of the confidence interval were, respectively,

(N_d*)_u = n_l/p_u *

and

(N_d*)_l = n_l/p_l *

These limits, as well as the estimate for N_d*, can be converted from bear-days to bears by dividing by n and the number of days in the search period.

During this study, we did not use Clopper—Pearson graphs as described by Miller et al. (1987). Instead, we used a DBASE microcomputer program that calculates the binomial confidence intervals for the 80, 90, 95, and 99% levels (D. Reed and J. Venable, Alaska Department of Fish and Game, Fairbanks, personal communication). These values were then entered on a Lotus worksheet (S. D. Miller, Alaska Department Fish and Game, Anchorage, personal communication), and the confidence intervals for bear-days, numbers of bears, and density were calculated automatically.

The 2,207-km² census area was initially divided into 12 sample units (i.e., count areas, ranging in size from 161 to 202 km²; Fig. 4). Natural landmarks such as streams and ridgetops were used as boundaries between count areas. We eliminated count areas 11 and 12 after the first survey because we didn't know whether the entire census area could be adequately covered each day with available personnel and aircraft.

Twenty individuals from three agencies, two private companies, and Noatak participated in the census that was conducted from 29 May through 4 June 1987. Six fixed-wing aircraft and one helicopter (Bell Jet Ranger 206B) were used during the census. Fixed-wing aircraft used for surveying included three Piper PA-18's, one Piper PA-12, and one Arctic Tern. A Cessna 185, used primarily for radio-tracking to determine degree of population closure, was also used for surveying. In both instances, we assessed population closure after we had searched the assigned count areas. During other days, radio-tracking and surveys took place simultaneously. Depending on location of survey aircraft and availability of the helicopter, personnel in the tracking aircraft monitored unmarked bears that were spotted from the survey aircraft and subsequently captured and radio-collared. This relieved staff on the survey aircraft from the tedious task of watching bears until the helicopter became available; they could continue surveying with minimum delay. The tracking aircraft was careful not to transmit the identity or whereabouts of any radio-collared bears.

Survey aircraft, pilot—observer teams, and assigned count areas were rotated daily. Pilot—observer teams did not discuss the location of sighted bears during or after the census so that bias in the search efforts in succeeding days was minimized. Personnel in the tracking aircraft were not rotated. One biologist was assigned permanently to the helicopter to ensure consistency in immobilization and handling procedures.

Statistical Tests

We determined differences among means, medians, ranks, and survival rates by t-test, one-way ANOVA, Kruskal—Wallis, or Mann—Whitney tests (Sokal and Rohlf 1969). Residual and normal plots were examined with MINITAB to determine if assumptions of equality of variances and normality had been met. Sex ratio data were tested by chi-square analysis. Proportion data were analyzed using chi-square contingency tables and by comparing ratios of means using t-tests (Satterthwaite 1946; Cochran 1977).

We were interested in determining if hunters could take certain measurements on bears that might allow managers to determine sex and age without examining the skull or hide. Consequently, we examined relations among morphometric measurements and ages of bears with multiple regression equations in a LOTUS worksheet.