Mountain Goat Management in Olympic National Park
D. B. Houston, R. A. Hoffman, R. W. Olson, and B. B. Moorhead
The National Park Service goat management program launched in 1981 was based on several premises:
1. Mountain goats seemed to be changing the character of the park's subalpine vegetation.
2. The presence of goats and their effects on the vegetation were contrary to the congressionally mandated purpose of the park.
3. Goat removal was consistent with established policy on management of exotic species.
4. The time frame for defining exotic species by National Park Service policy had been tested in court with the Grand Canyon National Park burro issue (see Introduction).
Goat management during the 1980's occurred in two phases: an experimental management program (EMP) initiated under the 1981 environmental assessment (National Park Service 1981), with aspects continuing through 1987, and an operational management program (OMP) initiated in 1988 under provisions of the 1987 environmental assessment (National Park Service 1987) and originally scheduled to run through 1992. Both environmental assessments were prepared in compliance with National Environmental Policy Act (NEPA, 42 USC 4321 et seq.) requirements for public disclosure and full consideration of feasible alternatives.
Carlquist (1990) traces the socio-political development of the management program, including the two environmental assessments, public meetings, levels of media coverage, and so forth. He also describes compromises perceived necessary by managers to reconcile conflicts between National Park Service management objectives and concerns of various interest groups (particularly sport hunters, animal welfare groups, and a native plant society). Some of these conflicts are described below, but we confine this summary primarily to biological and management considerations.
Objectives of the experimental management program (EMP) were to
1. determine the relative feasibility and cost of various goat removal techniques,
2. conduct studies of goat density and distribution,
3. reduce goat densities in the Klahhane Ridge subpopulation in order to document the species' ability to withstand exploitation,
4. monitor effects of these reduced densities on vegetation and soils, and
5. evaluate sterilization as a means of population control.
Major findings, by objective, were
1. National Park Service biologists, resource managers, and associated personnel captured approximately 693 mountain goats from 1977 to 198919 as part of the park's research and management operations, thereby gaining considerable experience with capture and removal techniques. About 233 goats were captured by foot snare (208 by V. Stevens and M. Hutchins), 182 were captured in drop nets, 171 were taken by chemical immobilization (including 135 helicopter captures), and 81 were captured by aerial net-gunning (including 69 removed under contract in 1985). Eighteen goats were shot as biological specimens.
Each goat removal technique (Table 36) tested had particular strengths and weaknesses. For example, a drop net used effectively to capture goats on Klahhane Ridge required animals that were habituated to accepting bait. Attempts to bait goats at eight other sites in the park during 1982 and 1983 were generally unsuccessful. Darting from helicopters (using the immobilizing drugs M99 etorphine and carfentanil) and net-gunning from aircraft were deemed the capture methods of choice for the subsequent OMP. Shooting from helicopters was among the least expensive and the safest for the field teams. The costs presented in Table 36 represent start-up expenditures only; cost per animal captured rose steeply as goat densities were reduced and survivors became more dispersed and evasive. For example, the cost of drop-net capture on KR was around $400/goat during the initial 1981 removals. By 1984, when the population had been reduced by about 60%, costs were nearly $700/goat.
2-4. Studies of the distribution and density of goats, the response of the KR goats to exploitation, and the effects of goats on vegetation have been detailed previously.
5. Field sterilizations were attempted on 19 goats (13 females, 6 males; Hoffman 1987; Hoffman and Wright 1990). Males were rendered permanently sterile by chemical vasectomy. Implants of a pregnancy-inhibiting drug (melengestrol acetate [MGA]) apparently rendered females sterile for 3 to 5 years with about 90% effectiveness. Techniques currently available for sterilization, however, were deemed impractical for any large-scale application because of the need to selectively capture and treat each animal and to repeat treatments of females. An ideal sterilization treatment would be permanent, usable on all sex and age classes, and deliverable by dart from helicopter. By the end of the EMP, we had concluded that such a technique did not exist (but see Sterilization below).
Objectives for the OMP were to eliminate mountain goats from the 3,250-km2 core area of the park (Fig. 44) and to control (i.e., reduce) densities along the eastern boundary of the park (about 300 km2) adjacent to state-managed lands where goats were hunted. Live capture was to be used exclusively for 3 years; shooting would then be used when capture became too expensive, inefficient, or hazardous, The effects of goats on subalpine vegetation were to be monitored throughout the program. An inventory of the abundance and distribution of endemic and rare plants was also to be conducted (see Chapter 12). An advisory committee composed of representatives from agencies and major interest groups was established to monitor the program.
During 1981-89, under provisions of the two environmental assessments, the National Park Service removed 407 goats from the park, including 360 (88%) captured and translocated from the park, 28 (7%) capture-related deaths, and 19 (5%) collected as scientific specimens (Table 37). Two hundred sixty of these animals were removed during 1981-87, under the EMP; 147 were removed during 1988-89, under the OMP. In addition, three known illegal kills occurred in the park, and 111 goats were harvested legally by recreational hunters (archers) outside the park during 1981-89 (Washington Department of Wildlife 1982-90).
Table 37. Mountain goat (Oreamnos americanus) removals by location in Olympic National Park, 1981-1989.
Removals during the abbreviated OMP warrant further comment. Based on availability of personnel and funds, park managers set removal goals of 80 goats for 1988 and 120 for 1989. Deaths caused by capture were to beheld at 5%. Eighty goats were removed during 1988, with 8.7% capture-related deaths. Only 67 goats were removed during 1989 with 19% deaths (Table 37; Fig. 45). Capture operations were plagued by poor flying weather both years; efforts were canceled 19 of 30 days (67%) between 20 May and 19 July 1988, and 14 of 27 days (52%) between 21 June and 26 July 1989. (Operations usually were not scheduled during weekends and holidays to reduce conflicts with park visitors.)
Number of goats captured per day of operations averaged 5.0 ± 2.37 SD during 1988 and 6.7 ± 2.06 during 1989. Goat captures per hour of flight time were 1.3 ± 0.38 and 1.3 ± 0.28 during the respective years. These measures of catch per effort did not differ between years (Mann-Whitney U-tests, two-tailed, α = 0.05).20 In 1989, capture attempts were sometimes launched in more difficult terrain than during 1988, and mortality increased (mostly from falls sustained during drug induction or netting; R. W. Olson, personal observation). The capture team viewed the increased mortality in 1989 as real, although the proportion of animals dying during capture did not statistically differ between years (x2 test, α = 0.05). In any event, the goal of 5% capture-related goat mortality was, in retrospect, unreasonable given the circumstances of capture.
As described in Chapter 4, the 1990 estimate of goat numbers in the metapopulation (389 ± 106) was significantly lower than the 1983 estimate (1,175 ± 171). The smaller 1990 population resulted from removals by the National Park Service, perhaps combined with relatively more severe winters (i.e., from exceptionally mild to periodic "average" snowpacks).
The capture operation scheduled for 1990 was canceled by the superintendent of Olympic National Park when two independent assessments indicated that continued efforts posed unacceptable risks to the capture team (Machlis et al. 1990; Peterson 1990). Moreover, a full environmental impact statement, rather than a less comprehensive environmental assessment, was judged necessary to continue with the removal program primarily because of the expected controversy surrounding the option of shooting (M. Finnerty, Superintendent, Olympic National Park, personal communication, 1990).
The National Park Service is preparing an environmental impact statement on goat management in the Olympic Mountains jointly with the U.S. Forest Service (Olympic National Forest) and the Washington Department of Wildlife. Negotiations are difficult because agency management objectives and mandates differ greatly. In addition, the process is being tracked by at least three disparate interest groups, including recreational hunters who wish to retain goats outside the park for sport hunting, a native plant society that wishes to see all goats removed, and an animal welfare group that wishes to see no goats killed. Predicting the outcome of this potentially volatile process demands a prescience that we lack, but a discussion of the management options available to the park in the near term is appropriate.
A recent, perceptive analysis of attempts to control or eradicate feral goats on oceanic islands and habitat islands in New Zealand is relevant to mountain goat management on the Olympic Peninsula (Parkes 1990a, 1990b). From Parkes' perspective, "One of four strategic [management] options may be adopted: no actiongiving a stable but undesirable outcome [adverse effects on indigenous biota]; eradicationgiving a stable and desirable outcome; annual sustained controlgiving a stable and desirable outcome but only if inherently fickle government effort is sustained; and occasional sustained controlgiving unstable outcomes with or without the desired results depending on the goals and hunting frequency" (Parkes 1990a:335).
Moreover, to achieve eradication, "Four conditions must be met: all animals must be at risk; there must be no colonisation; the harvesting rate must exceed the rate of increase of the population; and those attempting the task must believe it possible" (Parkes 1990b:297). Four conclusions reached following his analysis of the New Zealand situation seem particularly germane: "Policies of eradication are counterproductive unless they are possible; short intensive campaigns are best; managers should measure and record costs, efforts, and kills to improve planning; [and] cynics are useful if only to check claims of success" (Parkes 1990b:297).
Recognizing conflicting interagency objectives, managers on the Olympic Peninsula have only four broad options for management of mountain goats:
1. Do nothing and allow the entire goat population to achieve ecological carrying capacity (Kl). This would accept goats as part of the fauna and also accept the consequent changes in vegetation, which could include increased risk of extinction for populations of rare plants. Forego hunter harvest on subpopulations outside the park and manage goats for nonconsumptive (viewing) values.
2. Eliminate mountain goats from the entire Olympic Mountain Range. The metapopulation of goats on the peninsula is isolated, and all animals could be put at risk of removal.
3. Eliminate goat subpopulations from the park but maintain those on the Olympic National Forest and continue to extract a sustained yield by hunting. Because all animals are not at risk, the park must be prepared to control immigrants in perpetuity. Once removal of the park subpopulations was initially completed, control of immigrants would seemingly be a relatively small-scale program, if control were sustained diligently and funding were secure.
4. Annual sustained control and harvest in perpetuity; park subpopulations maintained at low density (e.g., perhaps 0.20 goat/km2 or lower on summer ranges) and subpopulations on the forest maintained at economic carrying capacity (Kc, see below) by maximum sustained yield harvests.
Different combinations or levels of these options are possible (e.g., park accepts goats at Kl with national forest subpopulations at Kc, or park eliminates subpopulations only from areas with highest densities of endemic-rare plants), but the basic management scenarios available are captured by the four alternatives. Acceptance and execution of any particular strategic option are different matters and depend only to a limited degree on the available removal tactics and expense.
A brief digression into the principles of harvesting large mammals is needed before the options can be evaluated. The following material, greatly simplified, is drawn primarily from Caughley (1977) and Macnab (1985). Consider an ungulate population colonizing a new area. Initially, the population grows at the maximum rate possiblethe intrinsic rate of increase (rm)because resources are unlimited. The equilibrium eventually reached between the herbivores and their food supply represents the maximum sustainable population levelecological carrying capacity (Kl). The rate of increase of the animal population at Kl is zero, by definition. To extract a sustained yield (SY) harvest, the density of the population must be reduced below Kl and a positive rate of increase generated. In the simplified logistic harvest model, the maximum sustained yield (MSY) would be generated when the population was reduced to half its equilibrium density (Kl/2) and harvested at half the intrinsic rate of increase (rm/2). Thus, a population with Kl of 1,000 and rm of 0.30 would be reduced initially to 500 animals (economic carrying capacity, Kc) and could then be harvested annually to produce MSY of 75 animals (500 x 0.15). At harvest rates less than rm, the population would stabilize at a density between Kl and Kc; if greater, the population would decline to extinction.
What does this mean for mountain goat management? The simplified logistic model would likely overestimate MSY harvests from goat populations, and maximum sustained yields might occur at densities nearer Kl (Fowler 1981a, 1981b; Hoffman 1987; Houston and Stevens 1988). Nevertheless, the model has considerable heuristic value, and we use it to explore the management options outlined above. We then consider a more refined model developed specifically for the Olympic mountain goats.
Hoffman (1987) estimated that Kl = 1,400 goats for all subpopulations in the park; this estimate was derived by considering the 1983 goat census, the extent of goat removals that had occurred to that date, and the areas of potential habitat as yet uncolonized by goats (combined with his considerable field experience in the Olympic Mountains). This seems a reasonable guess, because about 15% of the goat habitat is outside the park, by extrapolation Kl = 1,650 for the metapopulation. Two growing subpopulations showed initial rates of increase (rm) around 0.24, but rm may have been only 0.10 or less for the metapopulation (see Chapter 4). We cannot predict rm for the established goat population (and it would likely vary by subpopulation; see Chapter 4), but it would probably lie between these extremes. These estimates of Kl and rm are used to explore the available management options, and we will disregard, temporarily, the low accuracy and precision of the values. Populations growing at = 0.24 would double in size about every 3 years; at = 0.10, doubling time is about 7 years. We also have assumed that the goat population will indeed rebound from current low numbers, which were imposed, to a large degree, by cropping. This is by no means certain; for reasons that were not entirely clear, cropped populations of mountain goats elsewhere have remained at low levels or continued to decline (e.g., Kuck 1977).
Allow the entire population to achieve Kl. This is the easiest to carry out but would be inconsistent with the policies of the National Park Service and the other agencies.
This alternative specifies elimination of the entire goat population. At rm = 0.24, a theoretical MSY harvest of 99 goats could be extracted annually from the metapopulation if numbers were reduced to Kc of 825; with rm = 0.10, MSY would be only 41 goats. These removals would have to be exceeded in order to drive the population to extinction. Presumably, however, this option would not start with the metapopulation at Kl. Therefore, the effort theoretically required to eliminate initial populations of 1,000 or 500 goats at the two extreme rates of increase is explored when half the animals present are removed annually (Table 38).
Table 38. Approximate offtake, years of effort, and cost required
to eliminate populations of 1,000 and 500 mountain goats (Oreamnos
americanus) based on logistic growth models.a
The offtakes from these paper populations could be achieved by either increasing effective mortality (shooting, live capture), reducing fecundity (sterilizations), or some combination. In practice, it would be impossible to halve the population annually using sterilization (see below, and indeed, kids represent only about 20% of the population).
In the worst-case scenariopopulation of 1,000 goats, ?? ?rm = 0.24, and initial offtake costs of $100/animal doubling each time numbers are halvedelimination requires removal of 1,237 animals over 9 years at a cost of nearly $1.4 million; costs approach $6.8 million using an initial removal expense of $500/animal. At the other extreme500 goats, rm = 0.10, and costs constant at $100/animal removed547 goats are removed in 7 years at cost of $55,000; campaign costs are $274,000 at the figure of $500/animal. These extreme scenarios are unrealistic; removal costs will surely not remain constant as the population is reduced, but it is difficult to see them doubling annually in the latter stages of a program if the increased cost is due mainly to additional helicopter search timeeven recognizing that removal of the last few animals would be enormously expensive. Cost increases of 10-50% annually over the first half of a campaign to some plateau in the second half seem more likely. (Note that Table 38 may be modified easily to accommodate different starting or ending population sizes, offtake rates, or cost factors. An initial cost of $1,000/animal treated or removed is entirely possible, as shown in Table 36.)
None of the scenarios qualifies as the short intensive campaign recommended by Parkes (1990a, 1990b), which we define arbitrarily here as 5 years. On paper, short campaigns are easy to accomplish: for the population of 1,000 goats (rm = 0.24), if the targeted removals for years 4 and 5 were increased to 150 and 70 goats, respectively (versus 106 and 65), the population would be extinguished in 5 years. Shortening the prolonged effort by 4 years results in enormous monetary savings for the scenarios where removal costs escalate rapidly; for example, where costs double annually, the total expenditures of 5-year programs are only 30% of those for campaigns lasting 9 years.
This exercise illustrates some important points. First, elimination of a large population with high rates of increase would be difficult and costly; the converse also is true. If eradication is the option selected, then initiating the campaign when populations are at low levels, for whatever reason, makes the most biological and financial sense. Second, short intense campaigns where removals greatly exceed increases are indeed best, if for no other reason than cost containment. We appreciate the great disparity in cost among the scenarios, but at this time (April 1994), we do not know when or if a removal program will be launched or what removal tactics will be considered acceptable. We can be sure that none of the models will mimic exactly the response of the metapopulation to control.
As indicated, the logistic model overestimates a mountain goat population's ability to respond to removals. Hoffman (1987) developed a stochastic (i.e., parameters were allowed to vary randomly within prescribed limits) simulation model of population growth for the metapopulation of the Olympic goats that incorporated age specific fecundity and mortality, age distributions, and dispersal rates among 21 subpopulations (Fig. 46). Parameters for the model components were obtained from field surveys (his own and those made earlier, including Chapter 5 and Stevens 1983) and the literature. Each decision in the model (dying, dispersing, etc.) was made using Monte Carlo methods (Hammersley and Handscomb 1965), where possible outcomes were assigned probabilities of being chosen based on the predicted probability of occurrence in the system being modeled. The model then randomly selected one outcome based on these predetermined probability distributions. Simulated outcomes were used to test a variety of management scenarios (e.g., random vs. systematic removals with regard to the spatial distribution of subpopulations, May vs. August removals, and male vs. female sterilizations).
The model provided insight into the reduction alternatives; little difference occurred in time to extinction between random or systematic removals or with the time of year of removals (Hoffman 1987). Removals of 100 animals/year from an initial population of 1,000 goats required an average of 15.2-16.1 years to drive the model population to extinction (depending on the removal scenario; Fig. 47). The model was designed primarily to evaluate the relative effectiveness of live removals and sterilization (as opposed to shooting), so only scenarios involving treatments of up to 100 goats/year (a most ambitious goal) were tested. Direct comparisons with the logistic model are limited, but at a removal rate of 100/year, it also predicted that populations with rm = 0.10 would persist for about 15 years. But at rm = 0.24, goats would persist essentially indefinitely, as nearly 950 would remain following 40 years with these constant removals.
Sterilization scenarios showed that at least 50 animals/year (either males or females) would have to be treated and at least 15 years would be required to drive the population to extinction (Fig. 48). Insufficient information on goat breeding behavior was available to choose the most appropriate scenarios of male and female sterilization. Although we cannot predict with certainty which model (logistic or stochastic) would most closely mimic the actual population's response to removals, both demonstrated that elimination would require formidable effort.
Elimination from the park while maintaining sustained yields from the remaining population on the national forest is a variation on option 2 that would leave perhaps 200 goats and seemingly commit the National Park Service to removal of immigrants in perpetuity. Assuming that park subpopulations could be removed to initiate this option, at the effort and cost described above, keeping the park goat-free over the long term might be more difficult (and expensive) than first anticipated. This pessimism results from the difficulty of tracking immigration. Even though the number of annual immigrants would be small (perhaps up to 20 individuals if, at a worst case, 10% of the remaining subpopulations immigrated; Stevens 1983), goats could disperse across a large area and would be difficult to detect. If half the park summer range (about 21,000 ha) was searched annually for immigrants at the intensity of the goat census (0.04 min/ha), then 14 h of helicopter search time would be required plus perhaps an additional 6 h for such items as ferry time and refueling trips. At $550/h, the annual cost would be $11,000. Additional costs (personnel, equipment, etc.) could increase this estimate to perhaps $20,000/year. Several years of low returns might convince financially hard-pressed administrators that funds would be better spent elsewherepossibly setting the stage for recolonization.
However, the goats remaining outside the park would not occur as a single cohesive subpopulation but would be fragmented into small numbers along the eastern edge (primarily) of the park. The ability of such small groups to persist long-term is unknown but may be doubtful (e.g., Berger 1990) even without hunting. (The potential Kl might be raised for subpopulations that reside outside the park by deliberately enhancing winter habitats with prescribed burning [e.g., Johnson 1983:55], thus easing the risk of extinction.)
Maintain park populations at low density (0.20 goat/km2 or around 100 animals) and national forest populations at Kc (perhaps 150-200 goats) for a metapopulation of 250-300. This option suffers from the difficulties of option 3 (questionable viability of remaining low-density subpopulations and fickle agency commitment of scarce funds in perpetuity), plus the added problem of low precision in determining goat numbers (i.e., wide confidence intervals surrounding estimatesto be expected based on all earlier counts). Low precision means that tracking actual subpopulation numbers closely enough to maintain MSY harvests on the national forest and the narrowly defined limits of acceptable densities in the park would be nearly impossible. We suspect that, for reasons of cost and the likelihood of shifting budget priorities over time, this program could easily shift from the proposed annual sustained control to occasional sustained control (Parkes' option 4) by default.
Only three broad tactics are available to control or eliminate the goat population: live capture, field shooting, and sterilization (other lethal possibilities such as use of toxicants are considered to be unacceptable, as are habitat manipulation schemes such as fencing goats out of selected endemic plant populations). Effectiveness and costs of any management program hinge on which tactics or combination of tactics are selected. After 8 years of effort, park personnel have good understanding of the usefulness and limits of various techniques in the formidable Olympic Mountains.
Of the 10 capture techniques tested (Table 36), only net-gunning and darting from helicopters were feasible for the removal of relatively large numbers of goats over extensive areas. Although several subpopulations were essentially eliminated using live capture (including Klahhane Ridge where a drop net was used initially and where the two goats remaining in 1990 were males), the limits of the techniques, in terms of human safety, were judged to have been reached by 1989. Accessible animals and subpopulations had been removed, but at least 300 goats remained in the park. The remaining goats occurred in small groups or in several large subpopulations that occupied exceedingly precipitous terrain that was considered essentially unworkable for live capture (Fig. 49). Hence, the metapopulation cannot be eliminated using aerial live capture. Conversely, if the population is allowed to rebound, animals will reoccupy terrain from which they could be live-captured but still with appreciable risk to the capture teams.
Our experience with aerial shooting is limited to collecting eight biological specimens and to one simulated shooting sortie conducted to estimate costs in 1983 (Olson 1984, 1992). Aerial shooting is among the least expensive removal options tested (Table 36). Because capture techniques alone exceeded rates of increase and dramatically reduced several subpopulations, the combination of ground and aerial shooting, with the possibility of working essentially all terrain, seemingly could greatly reduce or eliminate many other subpopulations. Whether or not the metapopulation could be driven to extinction by shooting is unknown because removal of the last few animals would be difficult and expensive. Shooting from the ground has been used elsewhere to eliminate feral goats (Parkes 1990a, 1990b), and aerial shooting for commercial harvest apparently devastated populations of thar, a mountain ungulate introduced to New Zealand (Caughley 1983).
A panel of five scientists recently conducted an independent, on-site review to evaluate the potential for contraceptives or sterilants to control or eliminate mountain goats in the park (Warren et al. 1992). The panel reviewed current technology and the previous experimental sterilization efforts (e.g., Hoffman 1987) as well as the overall goat management program. They concluded
The panel estimated that a sterilization or contraception program could cost as much as $1-2 million10 times more than a shooting program (by their estimate)and would still not eliminate the population. They concluded that "at best, currently available contraceptives or sterilants could be used to (1) maintain a population at a stable level, (2) reduce the rate of population increase, or (3) slowly reduce population numbers over a long period of time (i.e., the life expectancy of a mountain goat; possibly as long as 10-15 years)."
Finally, "It is likely that research developments within the next 10-20 years may provide a 'one-shot' permanent sterilant that may be applicable for mountain goat management at ONP [Olympic National Park]. However, even if such a sterilant is developed in the future, it will likely never eliminate mountain goats from ONP."
The bottom line of this overview of management strategies and tactics shows that all control or elimination options will be difficult and expensive to achieve in practice. As difficult and costly as elimination of the metapopulation could be (option 2), sustained control (options 3 and 4) will be more so, especially if reduction in fecundity (sterilization) is the tactic applied. Any management program selected will surely test the stamina and commitment of agency managers.
Last Updated: 12-Dec-2007