Great Basin Naturalist Memoirs

Volume 8 The Black-footed Ferret Article 13

5-1-1986 Determing minimum population size for recovery of the black-footed ferret Craig R. Groves Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209

Tim W. Clark Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209

Follow this and additional works at: https://scholarsarchive.byu.edu/gbnm

Recommended Citation Groves, Craig R. and Clark, Tim W. (1986) "Determing minimum population size for recovery of the black-footed ferret," Great Basin Naturalist Memoirs: Vol. 8 , Article 13. Available at: https://scholarsarchive.byu.edu/gbnm/vol8/iss1/13

This Article is brought to you for free and open access by the Western North American Naturalist Publications at BYU ScholarsArchive. It has been accepted for inclusion in Great Basin Naturalist Memoirs by an authorized editor of BYU ScholarsArchive. For more information, please contact [email protected], [email protected]. DETERMINING MINIMUM POPULATION SIZE FOR RECOVERY OF THE BLACK-FOOTED FERRET

Craifj; R. Groves' ~ and Tim W. Clark'

Abstract. —A minimum viable population (MVP) size is estimated for the critically endangered black-footed ferret by examining five basic methods: experiments, biogeographic jiatterns, theoretical models, simulation models, and genetic considerations. Each method is evaluated for its appiical)ilit\ to the ferret and endangered species in general with two criteria in mind: (1) potential research impacts to target species and (2) the value of scientific accuracy and precision in relation to short-term conservation needs. For the black-footed ferret, the genetic method proved to be the most useful, resulting in an MVP estimate of about 200 ferrets for maintenance of short-term fitness. For most endangered species, a combination of the simulation and genetic methods will probably yield the best estimate of MVP size. MVP estimates have direct implications for future research, management, and recovery efforts of endangered species.

Following the 1981 discovery of the critically (Lacava and Hughes 1984). In this paper, we endangered black-footed ferret (Miistela ni- estimate minimum ferret population numbers gripes) in northwestern Wyoming, the only pop- and, as a corollary, minimum area require- ulation of the species known, two primary goals ments. Our purpose is to establish guidelines for of a ferret study were established: (1) conserva- future ferret research and management and to tion of the population and (2) recovery of the provide impetus, direction, and encouragement species (Clark 1984a). Estimation of population to other endangered species programs to use the parameters such as distribution, birth rate, concept of minimum population size. death rate, and immigration/emigration was a major objective. In addition to providing valu- Concept OF Minimum Viable able life history data for the ferret, estimating Population Size these parameters would enable us to address a key question: What is the minimum ferret popu- The notion of a required minimum population lation size necessary for the species to persist size for species conservation was first embodied (i.e., avoid extinction)? of the apparent Because by Shaffer (1981) in his concept of minimum small size of the population in 1981, Wyoming viable population (MVP) size. He defined the the answer was paramount for successful conser- MVP for a species as the smallest isolated popu- vation and, ultimately, species recovery. lation having a 99% chance of remaining extant The minimum population size concept for spe- for 1000 years despite foreseeable effects of four cies conservation was recently introduced by types of stochastic events: (1) demographic Shaffer (1981), with potentially significant impli- stochasticity, or the chance events in the sur- cations for, endangered species programs. Al- vival and reproductive fitness of a finite number though many endangered species studies have of individuals; (2) environmental stochasticity, been conducted, few have tried to estimate min- or perturbations due to habitat parameters, imum population sizes (Shaffer 1978, 1981, La- competitors, predators, and disease; (3) natural cava and Hughes 1984, Salwasser et al. 1984). catastrophes, such as randomly occurring floods,

Although US DA Forest Service regulations re- fires, droughts, etc.; and (4) genetic perturba- quire that minimum viable populations of en- tions residting from changes in gene frequency. dangered species be maintained on Forest Ser- Shaffer (1981) stressed the tentative nature ofhis vice lands, these population numbers are not definition and emphasized the importance of estimated but are instead routinely established defining MVP with explicit but flc^xible criteria as those levels specified in recovery plans such as time frame and survival probability. For

Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209. Present address: The Nature Conservancy, 4696 Overland Road, Suite 51H. Boise, Idaho 8.370.5.

150 1986 Groves, Clark: Minimum Population 151

example, the survival probability could be set at logically simple, this approach is not plausible 95% or auy other level, aud the time frame could for many species, particularly endangered ones. be shortened or lengthened as appropriate for First, for many endangered species, there exists planning needs. no possibility of studying several isolated popu-

Shaffer (1981) proposed five basic methods for lations. In the case of the ferret, there is only one determining MVP size: experiments, biogeo- known population, and consequently there is no graphic patterns, theoretical models, simulation way to measure variability in persistence of dif- models, and genetic considerations. In the next ferent-sized, isolated populations. Second, time section, we elaborate on the application and util- is not an unlimited resource in most endangered ity of each method for estimating MVP size for species studies. The years necessary to monitor ferrets in particular and endangered species in persistence of populations for MVP estimates general. The usefulness of each method is prag- are simply not available for most endangered matically evaluated relative to: (1) acceptable species. Lastly, many wildlife species cannot be levels of potential research impacts on target censused sufficiently without some type of mark/ species and (2) the real value of scientific accu- recapture program. These techniques can have racy and precision in relation to short-term con- negative research impacts (e.g., trap mortality) servation needs. In endangered species recov- that some endangered species projects may not ery work, these two concerns must be addressed. be willing to risk, particularly in the early stages. Because endangered species programs in gen- The MVP concept is a rapidly evolving one. In eral possess high levels uncertainty, late 1984, the Forest Service and other agencies of there may be hesitancy to increase uncertainty sponsored a workshop to provide a state-of-the- to even greater levels (Clark 1984b). Such has been the art review of the MVP issue (M. L. Shaffer, case in the personal communication). A major purpose of Wyoming ferret project, where one research team chose initially not to the workshop was to consider revisions of Forest employ mark/recapture and radio telemetry procedures Service procedures for estimating MVP size. until other methods proved the population large The primary focus of the workshop was a discus- enough to withstand some stress (Clark sion of the relative importance of demographic, 1981). The experimental method could prove useful environmental, and genetic variability in deter- when extensive population data are available mining minimum population numbers as well as prior to a species becoming endangered. For the development of new simulation and analyti- example, mountain goat (Oreamnos ameri- cal models for estimating MVP size. Proceedings canus) and bighorn sheep {Ovis canadensis) of this workshop, which will be published in populations have been reintroduced to several 1986, may provide additional methods not dis- sites in the western United States in recent years cussed in this paper and/or may suggest revi- (Rideout 1978, Wishart 1978). If wildlife re- sions of methods outlined and discussed below searchers monitor the dynamics of these intro- for estimating MVP sizes of endangered species. duced populations, the data should eventually allow them to estimate MVP sizes for these spe- Determining MVP Size cies. Such information would be useful not only for future translocations of these species' popula- Experiments tions but also for management if, for example, an isolated population showed a decline in num- This method estabhshes isolated populations bers. of species and monitors their population dynam- ics through time. If censusing can be accom- Biogeographic Patterns plished without stressful manipulation of indi- viduals (as can occur in a capture/recapture By studying the distribution patterns of spe- program), this approach requires only low levels cies occupying patchy or insular habitats, an esti- of research activities. The only necessary data mate of MVP size and minimum area require- are population numbers and the area inhabited ments can be obtained. However, populations over a specific time period (i.e., time frame in examined should be in equilibrium and their MVP definition). approximate period of isolation known (Shaffer Determining MVP via experimental methods 1981). Under these conditions, researchers can brings up three concerns. Although methodo- estimate the smallest area inhabited by a species 152 Great Basin Naturalist Memoirs No. 8

Fig. 1. Distribution of 37 white-tailed prairie dog colonies on the Wyoming black-footed ferret study area, 1981-1984. and the percentage of patches of a particular size and minimum area requirements for the species. that a species occupies. The smallest population However, due to the time required to obtain that has persisted over an extensive time period these data, the biogeographical method will not can then provide a first estimate of MVP size. likely be useful for short-term management de- Because the ferret occupies a patchy habitat cisions concerning MVP size. Nevertheless, (i.e., prairie dog colonies with relatively discrete useful information on minimum area require- boundaries. Fig. 1), this approach can be ap- ments can be gleaned from the biogeographical phed, at least in part, to the Wyoming popula- approach. tion. Old time trappers have reported catching In the short run, researchers will need to de- ferrets in the 1920s and 1930s on the Wyoming cide what size of an area is necessary for ferret study area, thus indicating that ferrets may have reintroduction or translocation for recovery occupied the site for at least 50 years. Extinction planning (Forrest et al. 1985; Houston et al. and recolonization, although unlikely, could 1986). A recommendation for an area larger than have occurred during this period. Population the Wyoming site is warranted for two reasons. censuses over the last four years (Clark 1986) First, two habitat patches (e.g., Wyoming and suggest that the Wyoming ferret population is South Dakota) of equal size may not be the same not in equilibrium but is increasing. As a result, in habitat quality. For example, Wyoming fer- more sophisticated census techniques (e.g., rets occur on white-tailed prairie dog {Cynomys mark/recapture) are being used, leading to more leucurus) colonies, whereas a South Dakota precise and accurate population data. In the fu- population of ferrets occupied black-tailed ture, researchers should thus be able to deter- prairie dog (C. ludovicianus) colonies (Hillman mine with some degree of confidence whether and Clark 1980). White-tails usually form small, the Wyoming population is reaching an equi- sparsely populated colonies; black-tails form librium density. large, densely populated colonies (Hoogland

If the Wyoming ferret population stabilizes in 1981). Consequently, a habitat patch that is large numbers, then data from this population will enough in one part of the range of the species provide the first field assessment of MVP size might be insufficient in another due to differ- 1986 Groves, Clark: Minimum Population 153

ences in habitat quality. Second, nothing is land is supported by three lines of evidence: (1) known about the frequency with which ferret several prairie dog colonies at the Wyoming site populations go extinct on habitat patches of vari- contained at one time or another only one adult ous sizes. Therefore, it would be inadvisable to ferret, suggesting that interbreeding of ferrets rely on the size of the Wyoming study area as an among prairie dog colonies is occurring, (2) exemplary model of area requirements for black- movement data reveal that ferrets range an aver- footed ferrets. age of 2.5 km between colonies, with the maxi- Although the biogeographical approach can- mum intercolony movement being 5.7 km (For-

not be used for endangered species with con- rest et al. 1985), and (3) aerial and ground tiguous distributions, it is applicable for species surveys indicate that potential ferret habitat de- with insular or patchy distributions. Its greatest clines significantly beyond the boundaries of this use should be for endangered species that oc- single colony complex, strongly suggesting that cupy patches of different size and similar quality these ferrets represent a distinct, isolated popu- and for which data on population numbers and lation. isolation periods are available. For example, the Data required in the Ti^ model are carrying Shoshone sculpin { greenei) is restricted capacity (K), per capita birth rates (\ ), and per to different-sized spring systems along a 45-km capita death rates (|x ). Direct observations of stretch of the Snake River in south central Idaho ferret litters in Wyoming indicated an average (Wallace et al. 1984). Relative and absolute den- htter size of 3.4, or 1.7 female young per adult sities of the sculpin in many of these spring sys- female (Forrest et al., in manuscript). We as- tems were determined during a status survey in sumed that the sex ratio approximated unity,

1980-1981 (J. Griffith, personal communica- natality did not vary with age, and all females tion). By monitoring the persistence of these bred. No data on juvenile mortality were avail- spring populations of this short-lived species able for ferrets; data fi-om other mustelids indi- over the next several years, researchers could cated a medium (60%) to high (80%) rate ofjuve- estimate MVP size and minimum area require- nile mortality (King 1980, M. nivalis ; King 1983, ments for this species. M. erminea; K. C. Walton personal communica- tion, M. putorius). We used a juvenile mortality Theoretical Models rate of 0.7, resulting in a X of 0.5. As with juve- nile mortality rates, no data on adult mortality Although there are several theoretical models rates were available for ferrets. Data from other that predict extinction probabilities and times mustelids indicated adult mortality ranged from for a population, most are too complex for the 15%-25% for Maries pennanti (Kelly 1977), M. simple data bases of endangered species. The americana (Strickland et al. 1982), and Mustela theory of island biogeography (MacArthur and erminea (Stroganov 1937), although King (1980) Wilson 1967), however, has received consider- reported mortality rates of 80%-90% for M. ni- able attention in the conservation field. Though

valis . We varied adult mortality |Ji) for ferrets originally developed in relation to true oceanic ( from 0.2 to 0.4. islands, this theory has been widely applied to probability of ferret populations of size n continental habitat "islands" or patches (e.g.. The the probability of extinc- Brown 1971). The theory's greatest potential ap- reaching a size where tion is nearly zero was calculated by plication has been in the design of nature pre- - serves for maintenance of species diversity (for P ~ 1 (fJL/X)"

review, see Margules et al. 1982). Much less where |x and X are the per capita death and birth attention has been devoted to those aspects of rates, respectively (MacArthur and Wilson the theory that predict the distribution of a sin- 1967, Richter-Dyn and Goel 1972). Two results gle, insular species (Smith 1974, Fritz 1979). emerged from this analysis (Table 1). First, as

Because the Wyoming ferret population is lo- |x/X decreases, the number of female ferrets

(i.e. prairie dog needed to colonize an area successfully de- cated on a large habitat "island" , as the colony complex in Fig. 1), we employed the Ty creases for a given probability. Second, model of island biogeography to predict proba- number of breeding females increases, a higher bilities of successful colonization and times to )x/X can still be tolerated with a high probability extinction for the ferret. Regarding the Wyo- of successful colonization. These results have a ming prairie dog colony complex as a single is- practical application in the reintroduction of 154 Great Basin Naturalist Memoirs No.

Table 1. Probability of a female black-footed ferret Table 2. Times to extinction (T|,) for black-footed fer- population of size n successfully colonizing and reaching ret populations with a carrying capacity (K) of 40 and 50, carrying capacity. For a birth rate (X) of 0.5 and a death and birth (\)/ death ((x) rates varied as shown. See text for rate (|x) varied from 0.2 to 0.4, the resulting jjl/X ranges details. from 0.4 to 0.8.

Population size 1986 Groves, Clark: Minimum Pofiilation l55

Simulation Models ity, Shaffer (personal conmiunication) simulated some preliminary estimates of Computer simulations may be the most useful MVP size for the ferret. Work is now in progress to refine method to estimate MVP size and minimum area these simulations, which will take into account requirements. Except for theoretical models both demographic and environmental stochasticity. like the T,. of island biogeography, computer As the Wyoming ferret study proceeds, models provide the only method in which a more age- and sex-specific natality/mortality data probability value for survival can be attached to may become available to incorporate into the simula- the MVP estimate. In addition, these models are tion. For short-term management needs, how- not confined to the mathematical assumptions of ever, a MVP estimate by simulation using data analytical models (e.g., density dependent na- from a closely related species should suffice for tality or mortality rates, lack ofage structure) and the black-footed ferret. provide a flexible mechanism for assessing the sensitivity of MVP estimates to changes in cer- Although not directly useful in estimating tain population parameters. MVP size, simulation models based on bioen- Shaffer (1978) employed the simulation ap- ergetics are useful in estimating minimum area proach to estimate MVP size and minimum area requirements. Two such models exist for the requirements for the grizzly bear (Ursiis arctos) ferret. One model estimates gestation, lactation, in Yellowstone National Park. This simulation and growth energy requirements for one female evaluated the effects of both demographic and ferret and her young (Stromberg et al. 1983); the environmental stochasticity and also indicated other estimates energy requirements based on which population parameters were most likely to experimental feeding studies of steppe polecats affect changes in survival probability. Watts and and observed activity patterns of the Wyoming Conley (1981) used both stochastic and deter- ferrets (Powell etal. 1985). Both models indicate ministic models to predict survival and extinc- the number of prairie dogs required to support a tion probabilities for a remnant population of given population of ferrets. Combined with esti- bighorn sheep in the southwestern United mates of MVP size, data from these models are States. Their simulations evaluated the effects of helpful in estimating minimum area require- demographic but not environmental stochasti- ments for the ferret, assuming that prey (prairie dog) densities can that is city. Although an estimate of MVP size was not be measured and there a close correspondence between prey density an explicit result of their eflfort, it could have been obtained from their simulations. and availability. The major disadvantage of the simulation ap- Computer simulations, though realistic in proach is the extensive population data it re- some aspects, cannot incorporate all the behav- quires. Minimum data requirements are the ioral and ecological adaptations of every species mean and variance of age-specific and sex- (Watts and Conley 1981). They should not be specific natality/mortality rates, age structure, used to predict actual population parameters but sex ratio, and the relationship of these variables instead should indicate a range of possibilities for to density (Shaffer 1981). For most species, ob- those parameters (Fowler 1981). In discussing taining these data requires an extensive and in- the role of computer simulations in wildlife sci- tensive mark/recapture and/or radio telemetry ence, Romesburg (1981) viewed these models as effort over several years. In his grizzly bear sim- comprehensive tools for integrating knowledge, ulation, Shaffer (1978) used the 12-year data common sense, hunches, and opinions. It is in base of Craighead et al. (1974). Such extensive this role that we feel computer simulations can data for any wildlife species are the exception, be an aid to estimating MVP size and minimum not the rule. However, for some species in which area requirements for endangered species. Shaf- data are lacking, it should be feasible to substi- fer's grizzly bear simulation is presently being tute data from a closely related (congeneric) spe- used by wildlife managers in Montana and Wyo- cies into the simulation. ming in this capacity. In addition, an adaptation

The black-footed ferret is a case in point. Al- of his simulation is currently being made avail- though some natality and sex ratio data are avail- able to the US DA Forest Service for use in the able (Forrest et al., in manuscript), mortality management of vertebrate species in their forest data are not. With ferret data on natality and planning process (Shaffer, personal communica- steppe polecat (M. eversmanni) data on mortal- tion). 156 Great Basin Naturalist Memoirs No.

Genetic Considerations f=l (i-^)' Increasing attention is being given to estimat- 2N, ing minimum population numbers from a ge- where / equals the inbreeding coefficient, N^ netic standpoint (Soule and Wilcox 1980, equals the genetically effective population size,

Frankel and Soule 1981, Schonewald-Cox et al. and t equals generation time. breeders 1983, Lacava and Hughes 1984, Lehmkuhl have noted a significant reduction in fecundity

1984, Salwasser et al. 1984). The critical ques- when/approaches 0.5-0.6 (Soule 1980). Setting tion is what population thresholds are necessary Np at 50 and/at 0.6 in the above equation results to maintain short-term and long-term (evolu- in t equal to about 90 generations. If the genera- tionary) fitness. Based on the extensive experi- tion time for the ferret is 1 year (time from birth ence of animal breeders, Franklin (1980) and of a female kit to birth of her first litter), it would Soule (1980) determined that the maximum al- take about 90 years for an effective population of lowable rate of inbreeding necessary to avoid 50 ferrets to reach an inbreeding coefficient that short-term effects of inbreeding depression is could cause extinction solely on genetic 1% per generation. This rule of thumb translates grounds. If the N^ for the Wyoming ferret popu- to a genetically efiective size of 50 for mainte- lation is now substantially less than 50 , nance of short-term fitness. this short-term genetic threshold could be The concept of a genetically effective popula- reached in considerably fewer than 90 years. tion (NJ is important. Such a population is de- The amount of time the population has already fined as one in which all individuals mate ran- been isolated may be several decades. domly, the sex ratio approaches unity, variance Franklin (1980) suggested that a genetically in family size is zero, and generations do not effective population of 500 animals is necessary overlap. Obviously, few if any vertebrate popu- for long-term genetic variation required by the lations meet these criteria. Thus, a genetically evolutionary process. As with short-term fitness, eSective size of 50 translates into a larger actual a Np of 500 translates into a much larger number (census) number of breeding adults for most spe- of breeding adults when dealing with real, as cies. opposed to ideal, populations. As an alternative Several mathematical expressions have been to this rule of thumb approach, Frankel and derived for determining the effects of variance in Soule (1981) suggested monitoring genetic varia- progeny, unequal sex ratios, overlapping gener- tion in target species by electrophoretic tech- ations, and population fluctuations on the genet- niques that estimate percentage polymorphisms ically effective population size (e.g. , Kimura and and percentage heterozygosity in a population. Crow 1963, Emigh and Pollak 1979). Lehmkuhl Data from natural populations suggest that rela- (1984) developed a procedure, based on the tively heterozygous individuals have greater vi- above formulae, for adjusting the genetically ef- ability and fecundity than individuals with lower fective population of 50 to arrive at the census percentages of heterozygosity and polymor- number of breeding animals. With Lehmkuhl's phism (Soule 1980). procedure, we estimated the number of breed- For many species, particularly endangered ing ferrets (N) necessary to maintain a geneti- ones, electrophoretic techniques can be imprac- cally effective population (N^) of 50 (see the Ap- tical for several reasons. First, they require po- pendix for calculations). Results of this exercise tentially injurious tissue sampling. Second, indicated that a MVP of 214 breeding ferrets is small sample sizes of most studies can verify the necessary for maintenance of short-term genetic presence but not the absence of rare alleles and fitness. This MVP estimate is five times the preclude the use of statistics to test departures known number of adults in the Wyoming popu- from Hardy-Weinburg frequencies. Third, lation (Forrest et al., in manuscript). Minimum many higher vertebrates lack polymorphic ge- area requirements for this MVP estimate would netic markers that these techniques can detect. be about 10,700 ha of white-tailed prairie dog Nevertheless, if these techniques become more habitat. efficient and automated in the future, they could A question arises as to what constitutes short- be a valuable management tool for working with term versus long-term genetic fitness. This threatened and endangered species. If test re- question can be satisfactorily answered with the sults indicated a reduction in genetic variability, following equation then restorative steps, such as increasing Np, 1986 Groves, Clark: Minimum Population 157 could be taken (Frankel and Soule 1981). Base- servation biologist should be to use both empiri- variation in line data on genetic the Wyoming cal as well as theoretical approaches such as is- ferret population are now being gathered (Kil- land biogeography and genetic considerations. patricketal., 1986). Each of the methods for estimating MVP size Although both empirical data and theory have relies on several assumptions. Which method(s) contributed to developing the genetic approach then are most appropriate? Looking at this same to estimating MVP size, this method is not with- question from another angle, one might ask to out its shortcomings. Both the "50 and 500 rules" which type of perturbation is the target species which are being advanced are based on simple, or population most susceptible. For most endan- analytical models of population genetics that do gered species, it is unlikely that the latter ques- not account for either age structure or environ- tion will be answerable; it is likely that the target stochasticity. Previous mental work indicates species will be subject to some combination of that the exclusion of structure age can dramati- demographic, environmental, and genetic per- cally decrease MVP size (Shaffer 1978) and in so turbations. Therefore, those methods that take doing optimistically deflate the MVP estimates. into account these perturbations should proba- bly carry the most weight. Obviously, the avail- Discussion and Conclusion able data will also dictate to some extent which methods can be used. At the outset of this paper, we indicated that For the majority of endangered species the five methods for estimating size would MVP studies, empirical data necessary for estimat- evaluated in relation to research impacts be on ing MVP size with the experimental and bio- target species value of pre- and the accurate and geographical methods will not be available. If cise data when time is a critical factor. In the population data on natality and mortality are black-footed ferret study, we adopted the prag- obtainable, then a combination of the simula- matic philosophy of Frankel and Soule (1981) tion and genetic approaches may yield the that conservationists must often employ rough best estimate of MVP size. The simulation approximations of critical parameters instead of method can account for both demographic waiting for precise data that may never be ob- and environmental perturbations, and the ge- tained. Assuming that this philosophy is sound netic method can account for demographic for most endangered species studies, in practice stochasticity as it relates to genetic drift. On it leads us to conclude that the experimental and the other hand, theoretical methods like the simulation methods for estimating MVP size T^ model of island biogeography only account have serious limitations. The former requires a for demographic perturbations. A combina- lengthy study period, and the latter an extensive tion of the simulation and genetic approaches set of population data. Neither time nor exten- will likely produce a range of MVP estimates. sive population data are goods in great supply for Results of the recent Forest Service workshop most endangered species programs. As previ- indicated that environmental and demo- ously pointed out, though, it should be possible graphic variability may be more important at times to use less than "perfect" data in the than genetic variability in setting the lower simulation approach to produce a meaningful limit to population viability (M. L. Shaffer, MVP estimate. personal communication). One possible im- Although Shaffer (1981) underscored the im- plication of these results is that the final MVP portance of defining MVP size with explicit but estimate should be biased toward that esti- flexible criteria (i.e., time frame and survival mate produced by the simulation method. probabihty), only two of the five methods exam- However, management constraints will also in establishing size ined (Tj^ and simulation) yield results that in- be a primary factor MVP estimates. clude both of these criteria. Are we then to dis- somewhere within the range of miss the other approaches as meaningless When no population data are available and exercises? We think not. When applicable, both empirical approaches have also been ruled out, the experimental and biogeographical methods then genetic considerations must take priority in provide empirical data for estimating MVP size, estimating MVP size. This is the case for the although no survival probability can be attached ferret. Until more data are obtained for simula- to their estimates. A prudent path for any con- tions, we must rely on our estimate of about 200 158 Great Basin Naturalist Memoirs No.

animals by the genetics method to serve as the Step 2 : Adjustment for unequal se.x ratio. Nm = (Ne + [male:female ratio X Nj)/4 estimate for black-footed ferrets. In the MVP where Np, = number of males. final analysis, this number may be reduced as Nf = N^ X femaleimale ratio a result of compromising between estimates N = N„ + Nf N/50 = increase in N^ fi-om the simulation and genetics methods. N = increase x 70 (from Step 1) The preceding discussion on genetic ap- For the black-footed ferret: (sex ratio data from 1984 to maintenance of proaches is relevant only census — Forrest et al., in manuscript) short-term fitness. For fitness on a long-term N^ = (50 + 18/25(50))/4 = 21.5 X 25/18 = 29.9 scale, we must try to establish several ferret Nf=21.5 N = 29.9 + 21.5 = 51.4 populations, each of which maintains a short- 51.4/50 = 1.02 term MVP size. Although such goals may now 1.02x70 = 71.4 appear beyond the vision of what resource Step 3 : Adjustment for overlapping generation. managers can do to conserve the ferrets in the Double the census number from Step 2. near term, we must never lose sight of the fact For the black-footed ferret: 2 X 71.4 = 142.8 that large, viable populations of ferrets and Step 4 : Adjustment for population fluctuations. between empirical high and low population cen- other endangered species are necessary for Ratio suses indicates factor with which to increase Step 3 result the recovery and conservation of these species for MVP estimate. fi-om an evolutionary standpoint. For black-footed ferret: 43/28 =1.5 1.5 X 142.8 = 214 = MVP (1983 census = 28 adults, 1984 = 43, Forrest et

al. , in manuscript) Acknowledgments

We thank the following groups for financial Literature Cited support in our Wyoming ferret study: World Brown. H. 1971. Mammals on niountaintops: nonequi- Wildlife Preservation J. Wildlife Fund—U.S., librium insular biogeography. Amer. Nat. 105:467- Trust International, New York Zoological So- 478. ciety's Wildlife Conservation International, Clark. T. W. 1981. A proposal for a research grant in support of the Meeteetse black-footed ferret conservation National Geographic Society, Charles A. studies; for the period 15 Feb. 1982-14 Feb. 1983 (12 Lindbergh Fund, National Wildlife Federa- months). 85 pp. tion, The Nature Conservancy, the Joseph 1984a. Strategies in endangered species conservation: Henry Fund of the National Academy of Sci- a research view of the ongoing black-footed ferret conservation studies. Pages 145-154 in Symp. on Is- ences, Defenders of Wildlife, Humane Soci- sues in Technology and Management of Impacted States, Xi, the Fre- ety of the United Sigma Western Wildlife, Steamboat Spring, Colorado, mont County (Wyoming) Audubon Club, and November 1982. Thorne Ecol. Inst. Tech. Publ. No. the Bethesda-Chevy Chase Chapter of the 14. 1984b. Biological, sociological, and organizational Izaak Walton League of America. S. Forrest, challenges to endangered species conservation: the C. Fowler, R. Miller, Shaffer, M. Soule, M. black-footed ferret case. Human Dimensions in and J. Terborgh provided constructive criti- Wildl. Newsl. 3(1):10-15. cisms of earlier drafts. Discussions with M. 1986. Some guidelines for management of the black- footed ferret. Great Basin Nat. Mem. 8:160-168. Shaffer improved the manuscript consider-

Clark. T. W , L Richardson. S C. Forrest, T M Campbell ably. III, D Casey, and K A Facerstone. 1985. Black- footed ferret prey base. Pages 7. 1-7. 14 in S. Ander- son and D. Inkley, eds.. Black-footed Ferret Work- Appendix shop Proc, Laramie, Wyoming, September 18-19, 1984. Wyoming Game and Publ., Cheyenne. Calculation of actual adult census number Cr.\ighead, 1974. Craighead, J. J., J Varney, and F C Jr. A (N) for black-footed ferrets through adjust- population analysis of the Yellowstone grizzly bears. ment of genetically effective population size Mont. For. and Conserv. Expt. Sta. (Missoula) Bull. No. 40. 120 (NJ (Lehmkuhl 1984). pp. Crowell, K L 1973. Experimental zoogeography: introduc-

Step 1 : Adjustment for variation in progeny number. tions of mice to small islands. Amer. Nat. N = (2N, V^ + 1)/K' 107:535-558. where Np ^ effective population size of 50, V^ ^ variance Emigh.T H .andE Pollak 1979. Fixation probabilities and in number of young, K = mean number of young per effective population numbers in diploid populations female. When lacking data on V;,, multiply 1.4 times N^. with overlapping generations. Theoret. Pop. Biol. For the black-footed ferret: 50 X 1.4 = 70. 15:86-107. 1986 Groves, Clark: Minimum Population 159

Forrest, S. C, T. W Clark. L. Richardson, and T M. Richter-Dyn N, and N. S. Goel. 1972. On the extinction Campbell III. 1985. Black-footed ferret habitat: of a colonizing species. Theoret. Popul. Biol. some management and reintroduction consider- 3:406-433. ations. Wyo. Bur. Land Mgt. Wild. Tech. Bull. RiDEouT.C B 1978. Mountain goat. Pages 149-159 in J. No. 2. Cheyenne. 49 pp. L. Schmidt and D. L. Gilbert, eds.. Big game of Fowler, C W 198L Comparative population dynamics North America: ecology and management. Stack- in large mammals. Pages 437-456 /nC. W. Fowler pole, Harrisburg. 494 pp. andT. D. Smith, eds.. Dynamics of large mammal RoMESBURG, H C 1981. Wildlife science: gaining reliable populations. J. Wiley & Sons, New York. 477 pp. knowledge. J. Wildl. Manage. 45:293-313. Frankel. O. H , AND M E SOULE 198L Conservation Salwasser, and evolution. Cambridge University Press, Cam- H , S P. Mealey, and K. Johnson 1984. bridge. 327 pp. Wildlife population viability: a question of risk. Trans. Franklin, I. R. 1980. Evolutionary change in small popu- North Amer. Wildl. Natur. Resour. Conf. lations. Pages 135-150 in M. E. Soule and B. A. 49:421-437. Wilcox, eds.. Conservation biology: an evolution- Schonewald-Cox, C M., S M Chambers. B Mac- ary-ecological perspective. Sinauer, Sunderland. Bryde, and L. Thomas, eds. 1983. Genetics and 395 pp. conservation. Benjamin/Cummings, Menlo Park. Fritz, R. S. 1979. Consequences of insular population 722 pp. structure: distribution and extinction of spruce Shaffer. M L 1978. Determining minimum viable pop- grouse populations. Oecologia 42:57-65. ulation sizes: a case study of the grizzly bear (Ursus Gilbert, F. S. 1980. The equilibrium theory of island arctos L.). Unpublished dissertation, Duke Uni- biogeography: fact or fiction? Biogeogr. J. versity, Durham. 190 pp. 7:209-235. 1981. Minimum population sizes for species con-

Hillman, C N , AND T Clark 1980. Mustela ni- W servation. Bioscience 31:131-134. gripes. Mammal. Species No. 126. 3 pp. Shaffer, M. L., and F B. Samson 1985. Population size HOOGLAND, J L. 1981. The evolution of coloniality in and extinction: a note on determining critical pop- white-tailed and black-tailed prairie dogs (Sciuri- ulation sizes. Amer. Nat. 125:144-152. dae: Cynomys leticurus and C. Itidovicianus). Ecology 62:252-272. Smith, A T 1974. The distribution and dispersal of pikas: consequences of insular population structure. Houston, B R , T W Clark, and S C Minta 1986. Habitat suitability index model for the black- Ecology 55:1112-1119. footed ferret: a method to locate transplant sites. SouLE. M. E. 1980. Thresholds for survival: maintaining Great Basin Nat. Mem. 8:99-114. fitness and evolutionary potential. Pages 151-169 Kelly, G.M 1977. Fisherbiology in the White Mountain in M. E. Soule and B. A. Wilcox, eds.. Conserva- National Forest and adjacent areas. Unpublished tion biology: an evolutionary-ecological perspec- dissertation. University of Massachusetts, Am- tive. Sinauer, Sunderland. 395 pp. herst. 178 pp. SouLE, M. E., AND B. A. Wilcox. 1980. Preface in Conser- KiLPATRicK, W., S. C. Forrest, W. Clark. 1986. C andT vation biology: an evolutionary-ecological per- Estimating genetic variation in the black-footed spective. Sinauer, Sunderland.395 pp. ferret—a first attempt. Great Basin Nat. Mem. Strickland. M. A., C W. Douglas. M. Novak, and N. P. 8:145-149. HUNZINGER. 1982. Marten. Pages 599-628 in J. A. Kimura, M , AND J F Crow 1963. The measurement of Chapman and G. A. Feldhamer, eds.. Wild mam- effective population number. Evolution 17: Hopkins University 279-288. mals of North America. Johns Press, Baltimore, Maryland. 1147 King. C. M 1980. Population biology of the weasel pp. Mustela nivalis on British game estates. Holarctic Stroganov, S. U. 1937. A method of age determination Ecol. 3:160-168. and an analysis of age structures of ermine popula- 1983. Mustela errninea. Mammal. Species No. tions (M. errninea). Pages 93-108 in Biology of 195. 8 pp. mustelids: some Soviet research, vol. 2. Science Lacava, J.,andJ Hughes. 1984. Determining minimum Inform. Div., Dept. Sci. Ind. Res., New Zealand. viable population levels. Wildl. Soc. Bull. 148 pp. 12:370-376. 1983. Stromberg, M. R , L E Rayburn, andT. W Clark

Lehmkuhl, J. F. 1984. Determining size and dispersion Black-footed ferret prey requirements: an energy of minimum viable populations for land manage- 47:67-73. balance estimate. J. Wildl. Manage. conservation. Envir. ment planning and species P. Con- Wallace, R. L., J. S. Griffith, D. M. Daley. J. Manage. 8:167-176. nolly, AND G. B. Beckham. 1984. Distribution of

MacArthur, R , AND E O Wilson 1967. The theory of the Shoshone sculpin {Cottus greenei : ) in island biogeography. Princeton Univ. Press, the Hagerman Valley ofsouth central Idaho. Great Princeton. 186 pp. Basin Nat. 44:324-326. Margules.CJ A Higgs.andR. W Rape 1982. Modern Watts, T, AND WCoNLEY. 1981. Extinction probabilities biogeographic theory: Are there any lessons for in a remnant population of Ovis canadensis mexi- nature reserve design? Biol. Conserv. 24:

. 26:393-405. 115-128. cana Acta Theriol. Bighorn sheep. Pages 161-172 in L. Powell, R. A., T W Clark. L Richardson, and S. For- WiSHART, W 1978. J. rest. 1985. Black-footed ferret {Mustela nigripes) Schmidt and D. L. Gilbert, eds., Big game of food consumption, energy expenditure, and prey North America: ecology and management. Stack- requirements. Biol. Conserv. 33:1-15. pole, Harrisburg. 494 pp.