We Need (and Can Have) Expanded Horizons for Recovery
Recovery of grizzly bears means erring on the side of caution if we want to ensure long term viability and protection against future catastrophes and other environmental changes. (See Core Values). This is consistent with the requirements of the Endangered Species Act.
Models that estimate the viability of populations and risks of extinction can produce quite different results, but all of them rely on information in the Ecological Background section (Ecological Background). Which is why it is worth understanding some of this occasionally esoteric background.
Despite on-going debate and varied outputs from models, there is perhaps surprising consensus among scientists that a population of 2500-5000 bears is needed to ensure long term persistence. In practice, this would necessitate connecting all of the lower-48 grizzly bear populations, and recovering grizzly bears in the Selway Bitteroot and North Cascades Ecosystems (see maps) — along with bears in adjacent ecosystems in Canada. For smaller populations such as those in the Selkirk and Cabinet Yaak ecosystems, at least 250 grizzly bears is needed to achieve even the minimum standards for recovery.
So…a bit more about what undergirds these population targets. Population viability is, in theory, determined by a number of factors, most commonly: demographic variability, environmental variability, genetic variability (e.g., inbreeding, drift, and mutation), and catastrophes. The results of minimum viable population (MVP) and population viability analysis (PVA) simulations depend in large measure on the risk benchmarks, especially the probability of persistence that is specified, the time frame for this reckoning, and the kinds of factors that are considered. At one extreme, Shaffer and Sampson (1983) estimated that only 50-90 adult bears were needed to achieve 95% probability of persistence of a population over a 100 year period, but only considering demographic variation, which is the least consequential of all factors that could be considered. Weilgus (2002) similarly estimated that roughly 250 adult bears were needed to achieve a 95% probability of not falling below 100 (a “pseudo-extinction” threshold) over 20 years, but only considering demographic and environmental variation.
At the other extreme, when Reed et al. (2003) and Traill et al. (2007) considered all four sources of variation, they estimated that between 4200 and 5800 adult animals were needed to achieve a 99% probability of persistence over 40 generations which, for bears, translates into >500 years, a duration in accord with the recommendations of Boyce et al. (2001). Reed et al. (2003) based their estimate on PVAs for 102 species which they parameterized themselves, whereas Traill et al. (2007) based their estimate on a meta-analysis of PVAs done by other scientists for 212 different species.
Looking strictly at risks of inbreeding depression and loss of evolutionary potential, geneticists have long adhered to a rule of thumb of 50/500 effective population sizes for conservation, where the effective population (Ne) corresponds to the number of animals that contribute to the genetic material of succeeding generations. Ne is commonly estimated as being 1/10 of total population size (N), although this ratio for grizzly bears has been estimated to range from roughly 1/10 to 3/10. An Ne of 50 is thought to be enough to guard against inbreeding depression whereas an Ne of 500 is thought to be enough to balance genetic losses due to genetic drift with increases due to mutation. Concern about inbreeding depression is legitimized by observed declines in fecundity of inbred captive brown bears. Given the ratio of Ne to N, effective population sizes of 50 and 500 translate into total population sizes of 155-500 and 1,550-5,000 for grizzly bears. These rules of thumb still hold despite being developed over 30 years ago.
The known fates of isolated brown and grizzly bear populations are also instructive. Isolated bear ranges with the potential to support >200-500 bears all experienced population increases when aggressive hunting and other control measures were stopped, with evidence of resilience greatest of all for ranges with the apparent potential of supporting >2,000 bears. By contrast, bear populations in Eurasia and North America that have been constrained by growth of human populations and related infrastructures, and which have been reduced to <100 adult bears, have all failed to recover or have even continued to decline toward extirpation. The two grizzly bear ranges in North America with the apparent potential of supporting >300-600 bears – the Selway-Bitterroot and North Cascades– have not exhibited robust population growth simply because a sufficient inoculum of bears has been lacking.
Put together, this evidence from modeling, theory, and case studies is a basis for deriving the two sets of population goals to direct recovery of grizzly bear populations. At a minimum, recovery efforts should strive for populations of 250-500 adult bears to provide for reasonable prospects of genetic conservation and population survival in the face of normal demographic and environmental variation for periods of 20-100 years. However, recovery efforts to promote long-term (i.e., indefinite) persistence should strive for populations of >2,000-5,000 bears. Populations of this size would guard against environmental catastrophes and provide for evolutionary potential. In the case of populations smaller than 100 adult bears, recovery should, if possible, be premised on connection with other larger populations and, if not, involve concerted efforts to improve habitat conditions in areas large enough to potentially support at least 250 bears.
These goals can be achieved, but not with the current approach. Ultimately the grizzly bear’s future is about the values we hold and about choices we make today.
 Paetkau et al. 1998
 Miller and Waits 2003
 Lande 1995
 Laikre et al. 1996
 Spielman et al. 2004; even Jamieson and Allision 2012, despite their critique
 Mattson and Reid 1991, Servheen et al. 1999, Swenson et al. 2000, Jien and Harris 2006
 Servheen et al. 1999, Swenson et al. 2000, Haq et al. 2012
 Boyce and Waller 2003, Merrill 2005, Mowat et al. 2013
 Proctor et al. 2012