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

Subalpine and Alpine Plant Communities

Plant communities were sampled with plots similar to those described by Franklin et al. (1971a) and Henderson et al. (1989) for forest vegetation. We used relatively small, variable-sized plots (20-100 m²) because extensive areas of homogeneous vegetation were uncommon at high elevations in the Olympic Mountains. The generally rectangular or square plots were located in vegetation that typified plant communities in the vicinity. Evidence of herbivore activity was not used as a criterion for locating individual plots, and plots were never placed near sites where salt had attracted mountain goats (Oreamnos americanus; see below).

Variables recorded included plant cover by species, substrate characteristics, slope, aspect, and elevation. Cover was estimated in classes of trace and percent ages of <1, 1-5, 6-25, 26-50, 51-75, 76-95, and 96-100. Small cover classes at each end of the scale prevented overestimates of species with low cover and underestimates of species with extremely high cover. The same classes were used to estimate cover of mosses, lichens, litter, organic matter, bare soil (<2-mm particles), total rock, bedrock, and five size classes of loose rocks (diameters of >2 mm—4 cm, 5-10 cm, 11-20 cm, 21-70 cm, and >70 cm).

Plant communities were classified using Two-Way Indicator Species Analysis (TWINSPAN). This method groups plots with similar cover and species composition (Hill 1979; Gauch 1982). Our data set included plant community survey plots (n = 705), permanent plots from Klahhane Ridge (n = 6) and Mount Constance (n = 6), and plot data from Belsky (1979; n = 11) and Pfitsch et al. (1983; n = 43). We sampled plots in 22 areas that represented regions of high and low precipitation as well as high (>3 goats/km²), medium (1-3 goats/km²), and low goat density (<1 goat/km²). Plots were concentrated in the eastern half of the park because most goat summer range occurred there (Houston et al. 1986).

Salt was used during 1972-84 on Klahhane Ridge to bait mountain goats into a single site where a dropnet was used to capture them (Johnson and Moorhead 1982). Goats were strongly attracted to salt for several weeks annually (approximately mid-June to early July), evidently for the sodium content (Hebert and Cowan 1971; Stevens 1979). During 1981-84, salt was maintained in a large rubberized tub to minimize the amount that escaped into the soil. Salt and tub were removed at the end of the trapping period. Some salt did escape into the soil (e.g., by spills or overflow during rains), and goats clearly devastated vegetation and soil within a 50-m radius of the site. We do not believe, however, that such intensified effects extended beyond the site. In particular, the salt did not increase goat densities in the Klahhane Ridge subpopulation either seasonally or year-round. Studies of 130 tagged goats on Klahhane Ridge showed that most were residents on the 3,900-ha area; they did not come from afar to consume salt. Klahhane Ridge provided excellent goat habitat, and goats certainly did not face rigors of that particular environment year-round merely to lick salt for a few weeks in summer. We believe that the location of the salt site had no effect on the overall goat densities on Klahhane Ridge or on studies of relations between goats and vegetation (i.e., the nearest vegetation plot was about 0.75 km distant from the salt site).

Baiting goats with salt also was attempted at eight other sites in the park (including Mount Dana) during the summers of 1982 and 1983. Baiting proved unsuccessful, so the salt was removed.

Interactions Among Herbivores, Plants, and Soils

Evidence of animal use was recorded in the plots used to classify plant communities in mountain goat summer range (Appendix A2). Animal sign in each plot was quantified by species to the extent possible. Scat piles, tracks, wallow or bedding sites, and marmot dens in each plot were counted. In addition, the number of wallow or bedding sites and marmot dens within 50 m of each plot and the number of trails crossing each plot were tallied. Animals in or near plots were noted.

A combination of factors was considered in determining which species of ungulate produced the sign. Roosevelt elk sign was identified by track and pellet size since both are considerably larger than those of mountain goats or black-tailed deer. Tracks were identified as having been produced by goats or deer based on shape. Deer tracks taper and are narrower toward the front while goat tracks have little taper, tending to be square or rectangular. Wallow or bedding sites and pellet groups were assigned to mountain goats or deer by the presence of other sign including identifiable tracks and the presence of hair (mainly goats). Mountain goat hair was a particularly useful indicator because goats annually shed winter coats in June and July, which left hair on rocks and plants in areas frequented by goats—hair did not persist through the subsequent winter (Schreiner, personal observation). Pellet groups, tracks, and wallows were classified as unidentified ungulate when the species responsible was uncertain. Deer and unidentified ungulate sign were combined during analysis because most unidentified sign was probably deer (i.e., we used a conservative approach with respect to identification of goat sign).

A grazing index (GI) for each species was obtained by estimating the proportion of plants with evidence of substantial grazing in each plot in classes of trace and percentages of 1-5, 6-25, 26-50, 51-75, and 76-100. An individual was considered to be substantially grazed only if ≥25% of the foliage or stems bad been cropped. In practice this meant that 25% of the culms in graminoids, 25% of the leaves in forbs, and 25% of the branches in shrubs had to be grazed before the proportion was estimated. Trace was entered when all plants (of a given species) in a plot had <25% of their foliage or stems cropped or if only flower stalks were cropped. It was not our intent to estimate the amount of standing crop removed by herbivores. Instead, we sought to create an index of grazing that could be compared among plots, areas, and plant communities. Armstrong and MacDonald (1992) noted that proportional estimates such as this are useful when comparing grazing levels among areas but that estimates of utilization (i.e., standing crop or biomass removed by herbivores) require intensive field studies.

Two indicators were used to characterize grazing on a plot basis. Average grazing provided an estimate of how intensely a plot had been grazed:

Percent of species grazed (PSG) was calculated by comparing the number of species with evidence of substantial grazing to the total number of species in a plot (i.e., species richness). High values of PSG indicated that herbivores grazed most of the species present in a plot and low values indicated that herbivores were more selective, grazing few of the available species. The index was calculated using only those species where more than 5% of the individuals of that species had been cropped at least 25% (i.e., GI >5%). Therefore, incidentally grazed species were not included. The formula was

An index of herbivore selection of particular plant species was calculated by comparing plant frequency to grazing frequency as follows:

Pfitsch and Bliss (1985) describe a similar index, also called species relative grazing frequency (SRGF), based on 1- x 1-m quadrats. They classified species as preferred (SRGF ≥50) and avoided (SRGF ≤25). However, because their index did not consider availability, it did actually estimate relative preference or avoidance in the conventional sense (Crawley 1983). To allow comparisons with their work, we used similar criteria but classified species as selected (SRGF ≥50) or nonselected (SRGF ≤25).

Each study area was classified according to the estimated density of mountain goats (see Fig. 32). Categories were high, medium, and low density based on estimates from 1986 except for two areas where we had additional information. The 1986 estimates reflected the 1983 mountain goat census (Houston et al. 1986), observations during live capture operations during 1984-85, and field observations by park staff (National Park Service 1987). Mount Angeles (1982) and Deer Park (1988) plots were assigned goat density levels based on more detailed information. We believe that there was good concurrence between the 1986 goat density estimates and goat density at the time of measurement because most plots (n = ~600) were established during 1983-86. Natural mortality and goat removals from 1988 to 1989 may have reduced correspondence between estimated and actual density for plots established during 1988-90. However, most of the 1988-90 plots were in areas already classified as low goat density. Goat density estimates used in the analyses were prepared independently of the process used to select study areas. Our initial selection of study sites included areas of high and low goat density based on the work of Stevens (1979, 1983) and perceptions of the field crew.

Responses of Vegetation to Reductions in Goat Density

Permanently marked plots were established in areas of homogeneous vegetation representative of plant communities in three areas. Plots were 100 m² unless a smaller 25-40-m² plot had to be used because the area of homogeneous vegetation was small (i.e., Avalanche Canyon where plant cover was sparse). Vegetation characteristics in each plot were estimated on 6-12 1- x 1-m quadrats selected using a stratified—random design.

Cover was estimated using plot frames gridded with string so that each square represented 1% of the quadrat.¹ The gridded frame helped minimize variability between estimates obtained by different investigators and provided for accurate estimates to the nearest 1%. Training sessions were conducted each year to ensure that crew members estimated plant cover similarly. Cover classes were trace = <0.1%, <1 = 0.1-0.5%, 1 = 0.5-1.0%, 2 = 1.1-2.0%, and so on to 100%. Grazing (i.e., GI) was estimated using grazing classes of trace and percentages of 1-5, 6-25, 26-50, 51-75, and 76-100.

---

¹We selected cover as a measure of plant abundance over time because reliable estimates can be made rapidly by experienced observers. Thus we optimized time and accuracy. Most other measures of plant abundance (e.g., frequency, density, standing crop, basal area) suffer from one or more shortcomings. For example, basal area and density are suitable parameters for plants with individual stems (or clumps) but are difficult to apply to rhizomatous plants or forbs with poorly defined "basal area." An advantage of cover is that estimates can be applied to all growth forms (Mueller-Dombois and Ellenberg 1974). We employed quadrat frames to estimate cover because we judged the pin drop method as too time consuming (i.e., we needed to sample a greater area than this method would have allowed). See Daubenmire (1968) and Mueller-Dombois and Ellenberg (1974) for additional discussion.

---

Klahhane Ridge

Klahhane Ridge is located in the north end of Olympic National Park, about 12 km south of the Strait of Juan de Fuca. Pfitsch and Bliss (1985) identified nine plant community types ranging from windswept ridges dominated by cushion plants to late snow areas dominated by Carex nigricans. Two communities, Phlox—Festuca and Phacelia scree (unstable herb meadow), account for most of the unforested vegetation in the area. Both communities become snow-free by late May or early June and compose 94% of available meadow vegetation in early summer and 59% in late summer.

Five permanent plots were established on Klahhane Ridge and one at Deer Park (as a control) in 1983 and were studied until 1990. Plots were established initially in areas of high, moderate, and low goat use by previous observations of goat behavior. Moderate- and low-use plots were combined for analyses. Cover values represented the maximum of mid-July, early-August, and late-August samplings by species.²

---

²We used maximum cover values measured each year for among-year comparisons. Comparisons based on similar dates (e.g.. 24 August) were inappropriate because measurements were influenced strongly by temporal variation in growing season (i.e., extent of drought, time of snowmelt, etc.; Schreiner 1987). We believe maximum cover values approximated peak annual standing crops.

---

Herbivores observed utilizing Klahhane Ridge included mountain goats, marmots, and black-tailed deer. Goats were observed mainly from 1983 to 1986, and black-tailed deer and marmots were seen from 1983 to 1990. Deer seemed to increase utilization of the Klahhane Ridge area as goat density was reduced.

Mount Dana

Mount Dana is located in a high precipitation region, 17 km east of Mount Olympus. Plant communities represented a subset of those found throughout goat summer range. The TWINSPAN classification of plots (n = 189) revealed eight plant community types. Dry site communities found elsewhere in the park (e.g., Astragalus—scree) were not encountered at Mount Dana. Additionally, some communities found in other areas (Carex—Valeriana, Delphinium—scree, Senecio—scree) were not well represented. Typical communities included Phyllodoce empetriformis, Carex spectabilis—Lupinus latifolius, and Phlox diffusa—Carex phaeocephala.

One permanent plot in each of five community types was established in 1984. Cover was estimated in late July or early August 1984-86, 1990, and 1992. Results were predisposed to greater phenological variability because sites could be sampled only once per season. Plots at Mount Dana were divided into high and low initial goat use classes by using the number of species grazed (NSG) more than 5% (i.e., GI >5%) in each 10- x 10-m plot as an indirect measure of the intensity of goat use. The number of species grazed in high use plots at Klahhane Ridge was initially greater than that in low use plots and declined as goat density was reduced. The value of NSG in the high use plots ultimately was equal to or less than that in low use plots. Consequently, we used NSG as an estimate of goat use levels at Mount Dana (and Avalanche Canyon).

Herbivores utilizing the Mount Dana area included elk, mountain goats, deer, marmots, and bear (actually an omnivore). Relatively limited observations during the study period (i.e., ~1 week/year in late July or early August, 1983, 1984, 1990, 1992) indicated that mountain goats utilized snow patches, meadows, and rocky outcrops along the east and north flanks of Mount Dana and open areas just south and west of the summit. On hot summer days, mountain goats utilized east-facing rock outcrops and debris chutes as low as 800 m. Extensive Carex spectabilis meadows below and southwest of the summit were used mainly by elk, bear, and marmots. Deer, their beds, and their tracks were most commonly observed in subalpine fir forests and among subalpine fir clumps above the Carex spectabilis meadows and below the summit of Mount Dana. Thus there was overlap in the sites used by goats and marmots but little overlap between sites used by goats and those used by elk, deer, and bear.

Avalanche Canyon

Avalanche Canyon is located near the eastern border of Olympic National Park just west of Mount Constance. The region is incredibly rugged and is known for its scree slopes, loose rock, and numerous prominent peaks (Olympic Mountain Rescue 1988)—vegetation is sparse. Plants occurred mainly on scree, talus, rock outcrops, and a few relatively stable benches. Species associated on talus slopes with delayed snowmelt (e.g., Elmera racemosa) were more common in Avalanche Canyon than at Klahhane or Dana. Individual plots were dominated by Luetkea pectinata, Phyllodoce empetriformis, Carex spectabilis, Phlox diffusa, Phacelia hastata, or Delphinium glareosum. Meadows characteristic of goat summer range elsewhere in the park were essentially absent.

Six permanent plots were established during 1988. Four plots were located in relatively unstable scree slopes dominated by Phacelia hastata or Delphinium glareosum and two on relatively stable sites dominated by Phlox diffusa. Cover was measured once in 1988 and in late July and late August 1989-92. The maximum cover value for each species in each year was used for analysis.

Attempts to classify plots as high or low initial goat use were unsuccessful because NSG values fluctuated considerably. Therefore, we used a crude approximation of goat use derived from the grazing index. Plots were divided into two groups: decreasing use where grazing occurred only during 1988-89 (n = 2) and stable or increasing use where grazing remained constant, increased, or occurred only during 1990-92 (n = 4).

Herbivore use in Avalanche Canyon was mainly by mountain goats. Elk and bear or their sign were not observed; marmots occurred occasionally, and deer were seen infrequently. Observations of goats, hair, and grazing in permanent plots and the surrounding area indicated they were the major herbivore present and that overall goat use remained more or less constant during the study.

Photographic Comparisons: A Qualitative Appraisal of the Influence of Climate and Disturbances on Vegetation, 1915-1990

Sources searched for photographs included the Clallam County Historical Society, park files, and records of two O'Neil expeditions (1885 and 1890), early mountaineers, pioneers, rangers, and others known to travel the mountains (e.g., E. B. Webster, the Cameron family, Doc Ludden). Many photographs were located during a historic resources study (Evans 1983).

We analyzed general features of photographic pairs (see Plates 1-11 in Chapter 11), then described and tallied differences (if any) in each community. The following categories were used:

1. Total plant cover;

2. Percent cover of disturbance-oriented species (Achillea millefolium, Cirsium edule, Phacelia hastata);

3. Cover of plants indicating community stability (Festuca idahoensis);

4. Cover of shrubs (mainly Juniperus communis);

5. Number and height of trees;

6. Cover of bare ground and rock;

7. Number of trails and wallow or bedding sites;

8. Obvious changes in cover of fine soils and pebbles <2 cm in diameter; fines were either exposed or replaced by gravel or cobble-size rocks; and

9. Soil erosion or deposition as indicated by increased or decreased exposure of roots and large rocks.

Most scenes contained more than one plant community, so comparisons of several communities were often possible from a single pair. The foreground, for example, might show a ridgetop community while the background showed a north-facing slope of Abies lasiocarpa. Additionally, if two different photographic pairs contained views of the same site from a different angle, only one community view was included in the analysis.

Temperature and precipitation data were obtained from the Port Angeles weather station, 12 km north of the study area (National Oceanic and Atmospheric Administration 1910-91). Snow depth data are for April from the Hurricane Ridge snow course, 3 km southwest of Mount Angeles (Soil Conservation Service 1960, Olympic National Park records). April records were used because they were closest to the growing season and therefore gave a crude indication of soil moisture available to plants the following summer. Climatic trends were plotted using a technique called Locally Weighted Regression Scatter Plot Smoothing (LOWESS; Chambers et al. 1983).

Rare Plants—Astragalus

Habitat sampling was conducted in the Blue Mountain and Mount Angeles areas where most Astragalus occurred. All subpopulations, with the exception of one containing approximately 200 plants north of Blue Mountain, were sampled—nearly one-third of all Astragalus plants were included. Plots were located in areas of physical site homogeneity that included at least one Astragalus individual and were distributed over the complete range of environmental conditions observed in each subpopulation. Plant cover, grazing, animal use, and environmental parameters were recorded using the same methods described previously for plant communities and herbivore—plant community interactions. Additionally, the number of Astragalus seedlings and mature plants in each plot were counted.

In 1985, 10 of the 25-m² plots were marked permanently to measure Astragalus population trends. Plot sides were marked temporarily with short steel nails at 1-m intervals, and cotton string was used to grid the entire 5- x 5-m plot into 25 1- x 1-m squares. Astragalus individuals in each square meter were mapped and counted and grazing or other physical injuries to each plant noted. Seedlings also were counted and mapped in 1987, 1988, 1990, and 1991.