Siberian polecats have been historically subjected to the potentially strong influences of plague even if Y. pestis itself has a fairly short evolutionary history [9]. A varying environment favors generalist predators [10] by preventing consistent advantage of preying on any single prey species, (assuming incomplete synchrony in fluctuations of prey species). Plague causes oscillations in populations of susliks and gerbils of central Asia (Fig. 1), and polecat populations appear to be affected [11]. The few polecats remaining after an epizootic may be forced to shift to less preferred prey species, similar to seasonal shifts in prey use by polecats in some areas [13] and to shifts in prey use by other generalist mustelids [14].

In contrast, the historical lack of plague in North America [15] seems to have allowed greater stability in populations of prairie dogs (Fig. 1), favoring specialization on them by ferrets. In part, the suitability of prairie dogs as prey for ferrets may stem from the high degree of coloniality and sociality in prairie dogs [16]. The prairie dogs' anti-predator warning system of alarm calls operates most efficiently when the prairie dogs are living at high densities. Although this defense works admirably against many predators, ferrets hunt mainly at night and usually kill prairie dogs below ground. Coloniality of prairie dogs benefits ferrets by providing high densities of prey, and the burrows dug by prairie dogs protect the ferrets against intraguild predators. Perhaps most importantly, the combination of locally abundant prey and shelter allows ferrets to minimize the need to move on the surface where they are exposed to predators [6].

Relative stability of resources may affect evolution of life histories as well as behaviours. These closely related Mustela provide an opportunity to contrast the relative predictions of r-K selection theory [17–19]. The prediction of greater fecundity in the r-selected form subjected to an unstable environment seems met by the larger average litter size of the Siberian polecat ($\overline{x}=8$; [13]), compared to the black-footed ferret ($\overline{x}=3.4$, [20]). In a stable environment, optimal litter size would be expected to be more constant, and greatest efficiency over the long term would be achieved by conceiving the number of young that could be consistently reared. The smaller litters of the black-footed ferret (compared to the polecat) thus suggest the K-selected maximization of energetic efficiency. As emphasized by Ricklefs [21], “The importance of r-and-K selection theory, relative to other sources of variation in life histories, depends on demonstrating a direct link between differences in population fluctuations and life history traits in pairs of otherwise similar organisms”. The direct link has not been proven, but circumstantial evidence is compelling for these ferrets and polecats.

Thus, an exotic disease and the dependence of black-footed ferrets on prey that were deemed to be pests resulted in declines to the point where the species became a charter member of the endangered species list after passage of the U.S. Endangered Species Act of 1973. At that time, a remnant population of ferrets existed, and was being studied, in southwest South Dakota [20,22]. However, the free-ranging population continued to decline, and had disappeared by the late 1970s. The first captive breeding of black-footed ferrets was attempted using nine ferrets taken from that South Dakota population during 1971–1973 [23], an effort which resulted in production of a few litters, but no surviving offspring. The last member of that captive population died in 1978.

Hope for black-footed ferrets was revived in 1981 [24], with the discovery of a new population living on white-tailed prairie dog colonies near Meeteetse, Wyoming. Intensive studies on this poorly understood carnivore began immediately, using radio-telemetry and other techniques [25–28], following recommendations from researchers in South Dakota. Canine distemper and plague simultaneously decimated the population in 1985 [29,30] transforming a somewhat tardy attempt to begin a captive breeding program into an emergency rescue effort [31]. During winter of 1985–86, ferret numbers dwindled to 10 known individuals (6 of which already were in captivity); the final free-ranging adults and their offspring were captured by March, 1987, bringing the total to 18 captives. The prairie dogs at Meeteetse continued to decline over a 10-year period due to the effects of plague [32], and the area is no longer suitable habitat for black-footed ferrets.

The initial failure to produce any young ferrets during the first season of the new captive breeding program (1986), considered in light of the failed program a decade earlier, was disconcerting. During subsequent years, however, breeding success improved as knowledge increased [33]. Although kit production rates recently have reached their highest point (P. Marinari, pers. comm.), the number of young reared per female in the captive population of about 250 adults at 6 zoos and a federal facility (1.9 kits) remains significantly lower (P=0.004) than the historic production by free-ranging females at Meeteetse (3.2 kits per female). Production in the captive population is hampered by females that either (1) do not breed, (2) become pseudopregnant, or (3) do not successfully rear their litters; the average size of captive litters is similar to the average litter size of their wild ancestors. Finding a solution could greatly increase the efficiency of captive propagation.

Another important question raised early in the program was whether or not the kits produced in captivity would be suitable for release (behaviourally and physically). Initially, we addressed this question using Siberian polecats as surrogates to investigate the development of behaviours [34]. Rearing conditions and learning affected predator avoidance behaviours [34]. Also, effects of rearing polecats in cages were cumulative, with decreasing cautiousness through four generations [6]. The apparent transmission of boldness (or lack thereof) seemed to be mediated by social learning between dam and offspring, interacting with environmental conditions. A single generation of rearing in quasi-natural outdoor pens partially reversed the process, an effect that was measurable through two subsequent generations (D. Biggins, unpublished data). These Lamarckian phenomena are not a recent discovery [35], but socially-mediated attributes are receiving renewed attention in regard to adaptive significance and evolution [36,37]. Experimental releases of Siberian polecats confirmed that rearing environments had significant effects on behaviours and that behavioural differences measured in laboratory trials were relevant to survival in the wild [6]. Studies on captive black-footed ferrets confirmed effects of environment on behavioral development [38–40] Further experimentation with quasi-natural rearing environments in the form of outdoor pens demonstrated significant effects on behaviours (Fig. 2) [41] and survival (Fig. 3) [42] of released black-footed ferrets. Predation has accounted for 93% of the losses in released polecats and ferrets monitored via radio-telemetry (n=130), underscoring the critical importance of development of predator avoidance behaviours.

The issue of quality versus quantity of animals released was debated vigorously in the early years of the black-footed ferret program. It is relatively cost-efficient to mass-produce animals in relatively simple environments, but efficiency must take into account post-release survival rates. An example taken from the first several seasons of reintroductions at the Charles M. Russell Wildlife Refuge in Montana illustrates the net effect of releasing animals of varying backgrounds on a single site. Because of differences in short-term and long-term survival [42], and reproductive success during the first two breeding seasons post-release, the proportion of the population with pen-experience in their lineage became amplified and the cage-reared representation nearly disappeared (Fig. 4).

Although the generational changes in behaviour of captive ferrets seemed to be mostly due to cultural evolution coupled with an environment providing unnatural stimuli, genetic change also is possible. Even if we are avoiding unintended directional selection for phenotypes that flourish in captivity (which is doubtful), we have removed the captive population from stabilizing selection, allowing variation to expand. Rapid evolution in captive populations of vertebrates is common, including changes wrought by artificial selection during the process of “domestication”. These processes are overlaid upon a foundation of low genetic variability in black-footed ferrets [43] compared to Siberian polecats and many other carnivores, perhaps due to repeated genetic bottlenecks in the isolated remnant wild populations [44]. The final bottleneck was defined by about 8 founder equivalents represented in the 18 individuals taken from the Meeteetse population. Furthermore, it is unlikely that the Meeteetse population, even in its historic condition, would have well-represented the genetic variability of a species whose range spanned 20 degrees latitude and 2500 m elevation.

The quality of the habitat at potential ferret reintroduction sites attracted considerable attention before releases began, and a system for ranking sites, based on ferret energetics and prey density, was developed [45]. We later discovered that prairie dog density seemed to effect movements and dispersal of released black-footed ferrets [41], but many questions remain unanswered regarding the effects of habitat and diseases on the success of ferret reintroduction. Canine distemper and plague are particularly perplexing, but perhaps it does not happen by chance that the most vigorous populations of black-footed ferrets have been those released in South Dakota where plague is absent and prairie dog densities are highest (Fig. 5). Vaccines are being developed and tested for both diseases, but some biologists had not envisioned a level of management of reintroduced ferret populations that includes delivery of vaccines in perpetuity.

The foregoing discussion highlights a sample of biological challenges to ferret conservation. Several socio-political problems also have surfaced during this conservation program. Even among those individuals whose motivations are altruistically ferret-oriented, opinions differ on the best means to reach widely-accepted goals. Individual working groups are understandably concerned primarily with growth rates of their own populations and protection of those populations from risks. Although those concerns should also be important to individuals and agencies responsible for the overall recovery program, the perspectives are understandably somewhat different. For example, in 1983, the cost in terms of risk to the Meeteetse population of removing ferrets for translocation or captive breeding were deemed to be too high to warrant that action. Conservation and protection of the Meeteetse population took such high priority that alternative options necessary for recovery could not be considered [31]. We do not suggest that the nearly disastrous result of that philosophy early in the program portends the program's future in any way, but simply that perspectives of scale have tremendous influence over decisions. The Fish and Wildlife Service, which oversees the recovery program, must look at recovery over the long term and maximum geographic scale. We have made this argument for research activities that were perceived to have some costs at the local and short-term scales, but could have potentially large benefits for the program as a whole. Translocation of ferrets is similar from the standpoint of perspective and priority.

Translocation is an example of a recently-tested tool that is biologically attractive but poses political problems. Data from South Dakota suggest that about 40% more wild-born kits than captive-born kits remain 30 days after groups of both are released onto new habitat. This difference (or even a less dramatic difference) could result in considerable monetary savings given the cost of producing a ferret in captivity (ca. $4000 U.S.). We do not imply that money is the only measure of efficiency (or even the best measure). Other considerations include the probability of establishing a self-sustaining population in a time frame that will not test the patience of working groups, getting populations back under the influence of natural selection, and reducing time spent in a genetic bottleneck.

Game managers have long held to the notion that highest productivity of a population occurs well before its numbers reach the carrying capacity of the habitat. Although years of experience has taught us a great deal more about the complexities of ecological systems, that basic tenet remains generally valid. Managing ferrets at carrying capacity of the habitat (if we can figure out what that is) may provide a buffer against losses, but probably does not make most efficient use of animals produced as long as there are new populations to be established. Local managers will doubtless be protective of the populations they have reestablished, especially if translocations cross jurisdictional boundaries. Like the Meeteetse example, the risks will be impossible to articulate exactly, leaving much room for debate. Nevertheless, high security of individual populations must be balanced against achieving the highest growth rates possible in the program as a whole.

Even after all parties come to agree on the point in population growth at which removals may begin, the financial burden of those operations must be equitably distributed. Costs include monitoring the donor population to annually assess its status and capability to supply ferrets, capture of animals selected for translocation, and quarantine/transport of ferrets. Fairness dictates that the group managing the site receiving ferrets (the immediate beneficiary) and the recovery program (the secondary beneficiary) should find ways to collectively participate in these tasks to further the cause of the black-footed ferret (the ultimate beneficiary). The group or agency managing the donor site should not be expected to support the entire operation, even if that agency has a mandate to contribute to recovery of endangered species.

Other socio-political problems are long-standing (e.g., attitudes of the agricultural community regarding prairie dogs) or have surfaced during various phases of the recovery program (e.g., conflicts over use of “experimental nonessential” provisions of the Endangered Species Act), and are described elsewhere [31,46,47]. Despite these biological and social challenges, recovery of black-footed ferrets has progressed. The captive population reached the capacity of facilities (200–250 adults) in 1995, and has been supplying an expanding reintroduction program (Fig. 5). Rapidly-increasing populations of naturally-reproducing ferrets have been reestablished in South Dakota (Fig. 5). The future of ferrets is less secure in parts of the range impacted by plague, but research is addressing possible tools for plague management including vaccines and pulicides.