NATIONAL PARK SERVICE Mountain Goats in Olympic National Park: Biology and Management of an Introduced Species

Introduction of Mountain Goats

Eleven or 12 mountain goats were released at Lake Crescent and nearby Baldy Ridge along the northern edge of the Olympic range at various times from 1925 through 1929 (Fig. 17). Moorhead and Stevens (1982) compiled historical information on the releases from newspaper accounts, correspondence, and notes in the files of Olympic National Park and Olympic National Forest. Official records are scant.

Fig. 17. Mountain goat dispersal and colonization of the Olympic Mountains. Shows the distribution of goats from their release in the 1920's (starred) through the 1960's and the year that females with young-at-heel were first reported (parentheses). O symbol shows expanded distribution of the 1970's. Modified from Moorhead and Stevens (1982). (click on image for a PDF version)

Four goats from British Columbia's Selkirk Mountains were released at the Lake Crescent site in January 1925 (Webster 1925; Port Angeles Evening News, 2 Jan. 1925). The records of subsequent releases are less clear (Webster 1932). In either 1927 or 1929, seven or eight goats from Alaska (two from the Chugach Range, six from Juneau) were released at Lake Crescent (five or six animals) and, perhaps, Baldy Ridge (two). Apparently, one of the Alaska animals died enroute. Early accounts refer to the goats as pairs, but no records of sex or age exist.

Dispersal and Colonization

The dispersal and subsequent colonization (i.e., establishment of breeding subpopulations) were also traced by Moorhead and Stevens (1982). During the 1930's, groups of up to 12 goats were occasionally reported on the slopes of Mount Storm King above Lake Crescent (Scheffer 1949; Port Angeles Evening News, 4 Nov. 1937). Goats dispersed from the release area and were seen on Mount Appleton in 1929 (13 km south), Mount Angeles in 1933 (25 km east), Constance Pass in 1931 (56 km west), and Mount Constance in 1935 (58 km east; Fig. 17). The U.S. Forest Service estimated the goat population at 25 animals in 1937—1 year before establishment of the park (Scheffer 1949).

Accounts suggest that by the early 1940's mountain goats occurred in small numbers in the northern mountains and along the eastern park boundary. By 1947, goats were observed annually on Mount Anderson, 49 km south of Lake Crescent. Before 1960, occasional sightings were reported on the Olympic National Forest south of the park and annually thereafter (C. Anderson, M. Beckstead, Olympic National Forest, Olympia, Washington, personal communication, 1979). These limited records suggest a dispersal east and south from the release sites of 3-6 km/year. Goats reached the eastern Olympic Mountains in 10 years and the southernmost mountains in 30 years.

During the 1970's, goats were reported in virtually all sectors of the mountains (Fig. 17). Few goats were reported in the central Bailey Range or around Mount Olympus until the middle-to-late 1960's. This central massif receives an estimated 380-500 cm of precipitation (mostly as snow) and may represent marginal habitat for the mountain goat. The most distant dispersal was a single animal reported in 1973 near Grisdale, 79 km south of Lake Crescent. By the late 1970's, goats were thought to occupy about 1,800 km² of the Olympic Mountains (Moorhead and Stevens 1982).

Distinguishing dispersal from true colonization is difficult with the available information. Studies have shown that young males (ages 1-3 years) are most likely to disperse (Stevens 1983). If that were the circumstance for pioneer dispersal throughout the range, it may explain the lags between dispersal and subsequent colonization (Fig. 17). For example, goats were first reported on the Mount Angeles-Klahhane Ridge area in 1933. Apparently, they either did not persist, remained at low densities, or did not establish a breeding population until the 1950's, because park records and newspaper accounts were still reporting the "first goat" on Mount Angeles as late as 1952 (Newman 1952; C. B. Browne, Olympic National Park, unpublished report). Glacier Meadows, north of Mount Olympus, was colonized as recently as 1981 (V. Stevens, personal observation). The sparse records available suggest that patterns of dispersal and colonization were complex.

Metapopulation

Observations of 237 tagged goats over 5 years led Stevens (1983) to view the goat population of the early 1980's as composed of reasonably discrete subpopulations. Dispersal occurred among subpopulations, and the comparative well-being of individual goats also differed among subpopulations. An assessment of the recent trend in numbers for the metapopulation (i.e., the overall population, composed of all the subpopulations; Andrewartha and Birch 1954) is available from rangewide aerial censuses conducted during 1983 and 1990.

Counting goats throughout an entire mountain range the size of the Olympics was formidable. The census design is described in detail elsewhere (Houston et al. 1986, 1991a). The census zone included all of the 50,063 ha of land free of glacial ice above 1,520 m in the Olympic Mountains. Most goats are seasonally migratory and visible mainly on subalpine ranges above 1,520 m during midsummer. About 87% of the area is within the park. A block count sampling method was used, and four census strata were recognized (Table 4). All known or suspected areas of high goat density and the Klahhane Ridge stratum were counted during both surveys. Known or suspected medium and low density areas were divided into blocks of about 500 ha each. Sample blocks were chosen randomly and censused. Aerial census is known to underestimate the actual number of mountain goats present, and a sightability bias of 0.66, developed from the goat removal program on Klahhane Ridge, was used to produce more accurate estimates (i.e., nearer the true total).

Table 4. Summary of mountain goat (Oreamnos americanus) censuses, Olympic Range, July 1983 and 1990.

Based on actual counts of 448 goats in 1983 and 172 in 1990, we estimated the goat population in the census zone at 1,175 ± 171 (SE) and 389 ± 106 for the respective years (Table 4; Fig. 18). Despite the large standard errors, the 1990 metapopulation estimate differed significantly from 1983 (t-test α0.05(2), t = 3.90; Norton-Griffiths 1978:80; Gasaway et al. 1986:62), and we concluded that goat numbers declined considerably over the 7 years (the observed exponential rate of increase was = -0.16; Caughley 1977). Based on the 1983 census, the metapopulation may have increased from the late 1920's introduction to 1983 at a rate of = 0.08.

Fig. 18. Mountain goats counted per census unit. Olympic Mountains, July 1983 and July 1990. Total count (diagonal hatch), Klahhane Ridge, medium density (shaded), and low density (open) units are indicated. Extensive areas of glacial ice are not shown. (click on image for a PDF version)

The contribution to the decline in goat numbers by the National Park Service removal program has been difficult to evaluate. Known goat removals totaled 326 from fall 1983 through fall 1989 (245 [75.2%] removed by National Park Service, 80 [24.5%] legally harvested outside the park, 1 [0.3%] illegally killed in the park). Annual removals were highly variable ( = 46.6 ± 34.7 SD goats/year, range 8-91).

Subpopulations

The quality of information on numerical trends differs markedly among subpopulations. We reconstructed reasonably complete pictures of growth trajectories for two subpopulations and provided general trends for others.

Klahhane Ridge

Goats inhabiting Klahhane Ridge (KR) range over about 3,900 ha year-round in the northeastern part of the park (Fig. 18). The goats migrate seasonally; some occupy winter ranges as low as 450 m in elevation. Summer ranges extend from about 1,400 to 1,968 m at the summit of Mount Angeles.

Goats seem to have colonized KR in the 1950's. By the early 1980's, the goats on KR were considered to be strongly limited by resources because they exhibited comparatively low reproductive rates, high dispersal, and slow body growth (Stevens 1983). The KR subpopulation was chosen for an experimental management program in 1981, largely because of the existence of background data, the high density of goats, and the relative ease of access. Goats were removed annually during 1981-84 as an experiment to measure response of the subpopulation to reduction (described below).

Goat numbers were determined using ground and aerial census (Houston and Stevens 1988). The 1977-81 estimates of goat numbers were calculated from ratios of tagged (n = 130) to untagged goats during ground censuses. The index manipulation index (IMI) method (Caughley 1977) was used to estimate the population during 1981-84, while goats were being removed. Subpopulation size was estimated by counts taken from helicopter (the index) following the June birth pulse before and immediately after a known number of goats was removed by trapping (the manipulation). On average, only 66% of the goats present were observed during the aerial censuses. Subsequent population estimates were made by adjusting the number actually counted by this value. Sufficient information was available for three early counts (1959, 1960, 1964) to also estimate sightability biases.

Thirty censuses of the KR population have been conducted from 1959 through 1990 (Table 5). About 180-200 goats were present in fall populations during 1977-80. Stevens (1983) considered that the KR subpopulation was essentially at ecological carrying capacity (K_l; Caughley 1976a) and that the goats did not undergo the eruptive overshoot and crash typical of colonizing ungulates (Caughley 1970a). Population growth from establishment through 1980 may have occurred as shown in Fig. 19; r was about 0.09 for 1959-80. Similar calculations using the three adjusted counts for the 1959-64 period of initial population growth returned an r of 0.24 ± 0.14 (SE). This represents a crude estimate of r_m, the intrinsic rate of increase (Caughley and Birch 1971), which is a measure of population growth when, presumably, no resource is in short supply.

Table 5. Mountain goat (Oreamnos americanus) numbers and removals, Klahhane Ridge, 1959-1990.

Fig. 19. Population estimates and removals for the Klahhane Ridge mountain goats.

Different census techniques produced comparable results. Ground counts and calculations using tagged goats produced an estimate of 183 goats during fall 1981, reasonably close to the IMI estimate of 171 for the mid-July population (Table 5; Fig. 19). The IMI estimate of the 1981 birth pulse population also seems reasonable when compared with fall populations in 1977-80. Fall estimates reflect declines due to dispersal and kid deaths (Stevens 1983).

Park records showed that 247 goats were removed from the KR subpopulation in 1959-89 (Table 5; Fig. 19). About 93% were live-trapped and translocated, 6% were collected as biological specimens, and 1% were illegal kills. Forty goats were removed between 1972 and 1980 ( = 4.4 ± 2.9 SD/year). If birth-pulse populations were around 220, then removals represented annual human predation of about 2%.

During 1981-84, we attempted to exploit the KR goats at a level equal to the estimated mean production of young in the prereduction populations; 48 ± 25 SD kids occurred in late June populations in 1977-80, and about 46 ± 20 in fall populations. About 29 ± 5 yearlings occurred in June populations from 1977 to 1980. Beginning with the June 1981 population of 229 goats, an average of 52 (range 51-52) goats was removed annually for 3 years. Removals exceeded the average June estimate of young by four, and the fall estimate by six. Removals exceeded the mean number of yearlings in June by about 23 animals. Attempts to remove another 52 goats in 1984 failed when only 19 were captured. At this level of removal, numbers declined at = -0.33 to an estimated 80 goats in the preremoval population of June 1984. Four additional goats were collected for necropsy in late winter 1985. The birth-pulse populations of 1981 declined at = -0.40 ± 0.06 SE to an estimated 41 goats by 1986.

Management direction changed from experimental to operational in 1988, and 20 additional goats were removed by 1989. Overall, the population of 1981 declined at = -0.49 to an estimated three goats by 1990. National Park Service removals were the main force driving the decline of the KR goats, but environmental factors played a role, as goat numbers seemed to continue to decline even when removals were suspended during 1985-87.

Conceivably, a large reduction in numbers could occur without enhancing per capita resources if goats have strong affinities for local sites (e.g., Kuck 1977). However, our observations indicated that even at low densities, the KR goats continued to distribute themselves throughout the area in summer and winter, albeit in smaller groups. This suggested that per capita resource levels were much greater at the low goat densities, a circumstance that had important implications for the population's dynamics (see Chapter 5).

Baldy Ridge

One of the first subpopulations of goats to become established in the Olympics was in the Baldy Ridge area, about 5 km from the site of introduction. A substantial portion of the Baldy Ridge subpopulation frequented the extensive cliffs along Hughes Creek during winter and spring, where they were observed easily from the Elwha River road. Most goats leave the cliffs by late spring, but we know little about summer or fall distribution. None of the 254 goats tagged or radio-instrumented elsewhere in the park during summers from 1977 to 1986 was observed wintering on the Hughes Creek cliffs (R. Hoffman, personal observation; V. Stevens, personal observation).

Sixty-two ground counts were available for November-April periods from 1953 to 1988, including 12 years with multiple counts (Table 6; Fig. 20). Although the variation among annual counts was sometimes great, numbers observed declined substantially over the 37 years. The rate of decline was calculated as = -0.08 based on the maximum annual counts for years with multiple counts. Five aerial counts confirmed the trend observed from the ground. This decline could represent either reduced observability as forest cover increased on Baldy Ridge, a real decline in numbers of goats, or both.

Table 6. Mountain goat (Oreamnos americanus) numbers and removals, Baldy Ridge-Hughes Creek, 1953-1990.

Fig. 20. Ground counts and National Park Service removals for Baldy Ridge-Hughes Creek mountain goats. Counts made during November-April periods; number of annual counts in parentheses.

The ridge burned during the late nineteenth century, and a photo taken in March 1907 compared to 1983 shows a substantial increase in forest cover (Fig. 21). Additionally, park records show that a small fire occurred on the crest of Baldy Ridge between 1916 and 1939 and that in 1950 another 120 ha burned on the north slopes below the ridge top. Lack (1962) reported that grasses throughout the burns had been heavily grazed by goats during 1962. It is likely that the decline in goats was real and that K_l declined as forest cover increased. The absence of kids in winter populations since 1982 also suggests a declining resource base.

Fig. 21. Mount Baldy from the Elwha River on 10 March 1907 (upper) and 16 March 1983, showing early burn and the increase in forest cover during the intervening 76 years on the goat winter range. (Asahel Curtis collection, Washington State Historical Society; retake by E. G. Schreiner)

Only one goat has been removed from the area by the National Park Service (Table 6). Goats have been hunted legally outside the park each year in this area since 1971, but removals averaged fewer than 1/year through 1983 and 1.5 + 1.2 SD/year from 1984 to 1989 (R. Johnson, Washington Department of Wildlife, Olympia, personal communication, 1983; Washington Department of Wildlife 1982-90). Harvest likely contributed to the decline, but the timing and magnitude of change suggest that the decline was driven largely by changes in resource availability. The scenario described for the Baldy Ridge subpopulation, where fire and plant succession apparently played major roles, does not seem to be important in subpopulation trends elsewhere in the Olympics. Coincidental enhancement of this winter range by fire near the site of goat introduction, however, may have facilitated mountain goat establishment in the Olympics.

Mount Dana-Mount Wilder

This subpopulation was well established on the steep, east-facing escarpment of Mount Dana when we conducted our first helicopter count in 1983. The count of 45 animals was much higher than anticipated and added considerably to the standard error of the rangewide goat census (Houston et al. 1986). We know little about the seasonal distribution of these goats; occasional sightings suggest that they remained on the extensive east- and south-facing cliffs of Dana and Wilder throughout winter (R. W. Olson, personal observation).

Eight aerial counts were obtained from 1983 through 1990 (Table 7; Fig. 22). The subpopulation declined at = -0.34 as 39 goats were removed during 1985-89, including 25 taken in 1985. Only five to six animals were estimated to remain in 1990.

Table 7. Mountain goat (Oreamnos americanus) numbers and removals, Mount Dana-Wilder, 1982-1990.

Fig. 22. Population estimates (from helicopter counts only) and National Park Service removals for the Mount Dana-Mount Wilder mountain goats.

Mount Appleton

Goats were reported from the Mount Appleton area shortly after their introduction to the Olympics, and 12 censuses were available from 1979 to 1991 (Table 8; Fig. 23). Relocations of seven radio-instrumented females showed that the subpopulation was seasonally migratory and that some goats wintered on wooded cliffs along Cat, Schoeffel, and Boulder creeks.

Table 8. Mountain goat (Oreamnos americanus) numbers and removals, Mount Appleton, 1979-1991.

Fig. 23. Population estimates and National Park Service removals for Mount Appleton mountain goats. The ranges of population size for 1980 and 1981 were based on ground counts of a smaller area (see text).

Comparisons of ground and aerial counts were difficult, because aerial counts included a larger area. Steven's estimate of 35-50 goats in the subpopulation by 1980 was based on ground counts of the Mount Appleton-Cat Creek area (about 500 ha). Aerial counts covered about 1,082 ha that included Mount Appleton, High Divide, and Bogachiel Peak. Counts were adjusted using a sightability of 0.71 determined for the specific area (Hoffman 1987). Based on the helicopter counts, the population declined precipitously (r = -0.33) until, by 1991, only three adult males were found in the area. The major decline seems to have occurred between 1983 and 1984, perhaps associated with the intervening winter. (We have no measure of winter severity from the area. Snow course records from Hurricane Ridge, 20 km distant, showed 198 cm and 145 cm of snow during April 1983 and 1984, considerably more than the 69 ± 25 cm mean of the previous 5 years.)

The role in the decline of human predation and an experimental sterilization program were examined. Eleven goats were removed from the area by the National Park Service from 1983 to 1988, including 9 in 1988. Hoffman (1987) attempted to sterilize seven females in the subpopulation (three in 1983, four in 1984). Although three of the females subsequently produced young, sterilization seemed to be partially effective in reducing reproduction by 1986 (Hoffman and Wright 1990). Human predation and sterilization no doubt contributed to the overall decline, but the steepest drop from 1983 to 1984 was apparently environmentally driven because it occurred before intensive human intervention.

Sawtooth Ridge

Stevens (1980) documented the initial increase in this subpopulation at = r_m = 0.23 from 1977 to 1981 (Table 9; Fig. 24). Summer range for these goats straddled the eastern park boundary, and animals were thought to winter outside the park (Stevens 1980). Four counts from the 1985-90 period suggested relatively constant numbers. The National Park Service removed seven goats in 1989. The goats were hunted outside the park, but numbers removed cannot be determined for the subpopulation because the harvest was reported for a much larger management unit. Harvests from the larger unit from 1981 to 1989 averaged 3.9 ± 1.3 goats/year (Washington Department of Wildlife 1982-1990).

Table 9. Mountain goat (Oreamnos americanus) numbers and removals, Sawtooth Ridge, 1977-1990.

Fig. 24. Population estimates and National Park Service removals for Sawtooth Ridge mountain goats.

Mount Constance

The rugged terrain around Mount Constance and Charlia Lakes in the northeastern Olympic Mountains had been colonized by mountain goats by the 1940's. The area was impossible to census thoroughly from the ground, and the subpopulations during 1979-81 were estimated from ground counts at 60-120 goats (Table 10; Fig. 25). Two helicopter counts over the 3,546-ha area (a considerably larger area than covered by ground censuses; V. Stevens 1991, personal communication) suggested that a major decline in goat numbers occurred between 1983 and 1990—this was associated with the removal of 13 animals by the National Park Service. Animals were hunted outside the park, and known hunting removals averaged 6.2 ± 2.0 goats/year from the larger goat management unit from 1981 to 1989 (Washington Department of Wildlife 1982-1990).

Table 10. Mountain goat (Oreamnos americanus) numbers and removals, Mount Constance area 1977-1990.

Fig. 25. Population estimates (from helicopter counts only) and National Park Service removals for Mount Constance mountain goats.

Miscellaneous Subpopulations

We have limited information on trends and removals for four additional subpopulations from 1983 to 1990 (Table 11). The Chimney Peak subpopulation occupies the steep canyon walls along the upper Quinault River entirely within the park. We were dismayed to count 52 animals during the 1983 census, and this high count, like the one on Mount Dana, added substantially to the standard error of the rangewide census. Numbers apparently declined by 1990. The Royal Basin subpopulation intermingled with goats from Mount Constance and also ranged outside the park (Stevens 1980). The Royal Basin subpopulation had declined sharply by 1990. The Buckhorn subpopulation occurs outside the northeastern corner of the park, and the Pershing-Washington subpopulation is present outside the southeastern corner. Both subpopulations were hunted and seemed to have declined.

Table 11. Number of mountain goats (Oreamnos americanus) counted in four areas of the Olympic Range during July 1983 and 1990 censuses. Numbers are actual counts and were not adjusted for sightability.

Interpretation

The review of dispersal, population trends, and the experimental manipulation of the KR animals provided insights into mountain goat ecology as well as information useful for park management. Mountain goats colonized the varied landscapes of the Olympic Mountains in about 60 years. Dispersal rates of around 3-6 km/year are comparable to those documented for mountain ungulates elsewhere (Caughley 1977). The pattern of subpopulation establishment was complex, patchy, and probably influenced locally by climatic events and by the presence of suitable winter range.

The possible influence of recent climatic events on colonization is notable. Initial colonization of the range coincided with the period of reduced precipitation (the 1920's through the 1940's), including the most prolonged spring-summer drought in 300 years (see Chapter 2). Occupancy of the Bailey Range began during the 1960's and could have been facilitated by the period of reduced winter snowpacks that have generally continued to the present. We do not know to what extent, if any, these climatic trends aided colonization, but subpopulations in the Bailey Range may not persist if high winter snowpacks return.

Body weight suggests that r_m for the mountain goat should approach 0.30 (Caughley and Krebs 1983). Several introduced goat populations have approached this rate (Houston and Stevens 1988), including the subpopulations on KR (0.24) and Sawtooth Ridge (0.23) in the Olympics. Rates of increase approaching r_m apparently do not persist for lengthy periods, as mountain goat populations may be especially sensitive to resource limitations (see Chapter 5).

Unfortunately, growth trajectories of the metapopulation and most subpopulations were not closely tracked, and we have incomplete records. The incomplete picture suggests quasi-independent subpopulations growing and declining asynchronously across the range, notwithstanding the recent rangewide decline. The pre-1986 variations in subpopulation sizes seemed driven by forces as varied as drought, winter severity, postfire plant succession, and levels of human predation. Striking differences in demographics of subpopulations and condition of individual goats underscore this diversity (Stevens 1983). Growth or decline of subpopulations at different rates across areas of small geographic scale may contribute to difficulties encountered elsewhere in managing mountain goat populations for sustained-yield harvests.

Trends in numbers of KR goats provided information useful for their management in the park and elsewhere. For example, a partial compensation model (PCM) and a harvestable surplus model (HSM) dominate ungulate harvest theory, although these competing models have not been adequately tested (Caughley 1976, 1985). Under the PCM, cropping to produce a sustained yield always reduces population density and maintains it below K_l. Under the HSM, hunting mortality and natural mortality are presumed to be compensatory, and the population is not reduced below K_l if it is cropped at a level equal to the production of young or yearlings (depending on the version of the model). Goat removals permitted a rough test of the HSM. If the HSM was appropriate, then the population should have declined at about r = -0.02 to -0.03—the extent that cropping exceeded the number of young in the initial population (June and fall). An r of -0.12 should have occurred for the yearling cropping version. Instead, numbers plummeted at = -0.40. The HSM is clearly inappropriate for the KR goats. Removals drove the population below even a theoretical maximum sustained yield for the PCM at economic carrying capacity (Kc = K_l/2; Caughley 1976a). This would have been a disaster if sustained harvest had been the management goal. However, these results provided part of the framework for assessing goat management alternatives in the park (see Chapter 14).

It was difficult to assess the relative contributions of the National Park Service removal program and other forces to the decline in the metapopulation from 1983 to 1990. Exploratory regression using six park subpopulations (KR, Appleton, Royal Basin, Dana-Wilder, Constance, Chimney) showed a nonsignificant relation (Y = -0.66x +0.66, r² = 0.27, P > 0.10) between the proportions of the 1983 subpopulations removed and the proportions of the initial populations remaining by 1990 (1991 for Appleton). Thus, either the effects of the removals were complex and variable among subpopulations, or the census precision was too low to measure the effects of removals at the subpopulation level. Further, some subpopulations declined without removals and others declined more precipitously than expected if removals were the primary agent of decline. (These unexpected declines might have reflected, in part, the effects of demographic dependence, where immigration from donor subpopulations maintained goat densities in recipient subpopulations. Reducing donor subpopulation sizes produced fewer dispersing goats, which contributed to declines in recipient subpopulations)

Known goat removals from 1983 to 1990 were highly variable and averaged 46.6 ± 34.7 goats/year, only 4% of the estimated 1983 population. This seemingly low value may be misleading because additional mortality may have been associated with the National Park Service goat captures, and removals were not distributed across the metapopulation in relation to goat densities but were sometimes concentrated on particular subpopulations. In relation to additional mortalities, aerial darting and net gunning were used to capture goats from 1985 to 1989 (Houston et al. 1991b); only rarely were kids captured with these techniques during July (primarily) operations. About 66% of the 102 adult females captured were lactating, and we suspect that few of their kids survived (data on orphaned kid survival elsewhere are scant and contradictory; Foster and Rahs 1982). Moreover, some yearlings may continue to depend heavily on maternal females (Hutchins 1984). Consequently, the population effects of the National Park Service removals could have been greater than indicated.

In relation to subpopulations, National Park Service removals were probably the force driving the decline of the KR goats, but even there, goat numbers continued to decline when removals were discontinued from 1985 to 1987. Similarly, park removals probably contributed to substantial declines at Dana-Wilder and perhaps Royal Basin. Declines at Mount Appleton over the same period were likely due to winter mortality. Such patterns are not unusual for mountain goats. Smith (1984), for example, documented strong density-independent mortality among goat populations in southeastern Alaska during severe winters. Although National Park Service removals were a strong influence on the decline of certain subpopulations, the metapopulation decline between 1983 and 1990 was also influenced by environmental forces—most likely the return to winters with substantial snowpacks.

Caughley (1970a, 1976a) developed a model of ungulate population growth based on earlier work by Riney (1955). Under the Riney—Caughley model, "When ungulates are introduced into a previously unoccupied area they increase to a peak density and then crash steeply to a considerably lower level. Subsequent oscillations are heavily dampened and density finally steadies well below the initial peak. The vegetation follows a reciprocal trajectory..." (Caughley 1976a: 197).

Our information on mountain goat subpopulation growth was too fragmentary to provide a test of this model. The pattern of colonization seemed to be complex, and we lacked reliable dates for the establishment of key subpopulations. The decline in the Baldy Ridge subpopulation (among the first established) was consistent with the model, but this was confounded by postfire plant succession. Several of the most recently established subpopulations seemed to be the fastest growing, their growth rates possibly accelerated by yearling breeding (see Chapter 5) as expected from the model, but most were also hunted. Mountain goats affected plant cover, but neat reciprocal relations between goats and vegetation density were difficult to discern across the Olympic Mountains because of the dissimilarities in plant communities (see Chapters 2 and 9-12) and the infrequent goat censuses.