In the previous post in this series, I outlined the major scientific papers that are at the heart of a discussion about both wolverine ecology, and the proposed listing rule. The USFWS is obliged to use “the best available science” in creating decisions for listing under the Endangered Species Act. In this post, I’ll talk about the conflicting interpretations of the papers explored in the previous post, and contentions that each interpretation represents “the best available science.” Those conflicting interpretations create different stories about what is going on with wolverines in the Rockies. Some of the stories clearly justify listing. Others suggest that either everything is fine, or that trapping is fine, or that there’s too much uncertainty out there and we ought to give up on listing because we just don’t know what’s going on. I’ll discuss these stories in a third post.
During the review process, two of the seven reviewers opined that the listing decision was not valid because it was not based on the best available science. Montana Fish, Wildlife, and Parks also submitted comments expressing disapproval of the decision to list and offering arguments that the science used in the decision was erroneous. These different reviews and comments were largely in agreement with each other, but with some significant variations in the scientific arguments that they emphasized, so I’m presenting them individually. The reviews were detailed, and I have summarized them here, so some precision is lost. I apologize in advance for any errors, omissions, or accidental misinterpretations.
Audrey Magoun’s concerns, expressed in edits to the language of the listing decision as well as a separate 33-page analysis, revolve around the fact that there are recorded instances of wolverines denning outside the snow model, denning in shallow snow, and weaning kits prior to the May 15th date used in the Copeland et al 2010 snow model paper. She uses this to argue that Copeland et al, which defines wolverine habitat in the Rockies, is invalid, and that the McKelvey et al 2011 paper, which is based on Copeland et al and is the foundation for the finding that wolverine habitat will diminish, is also therefore invalid. This, in Magoun’s assessment, invalidates the finding that wolverines in the US Rockies are endangered by climate change.
Although she is a co-author on the Copeland et al 2010 paper, she says that at the time of writing, she expressed concerns about the choice of May 15th as the weaning date for wolverine kits – and hence the snowmelt date for the paper. In her analysis of the listing decision, she cites an example of a den found outside the snow model in Ontario, gives a detailed description of the snow conditions in that situation, and then adds examples of den abandonment prior to May 15th to build a case that persistent spring snow (>1 meter depth and lasting until May 15th, as she defines it) is not necessary for wolverines to reproduce successfully. She suggests that the very high fit between dens and the snow model is an artifact of several biases in the data, including a search bias (ie, wolverine researchers look for dens in snowy regions, so this is where they find them) and a research bias (ie, and that most of the published literature is on wolverines in mountain and tundra habitat, leaving out populations in boreal forest regions where the species may behave differently). The critique includes a number of details on how MODIS satellites map snow cover (to argue that wolverines may be capable of utilizing small or remnant snow patches, which may be underrepresented in MODIS data, for denning) and on thermal thresholds for the insulating properties of snow (to argue that thermal cover, or the idea that wolverines den in snow partially to provide insulation and warmth for their kits, would not be necessary at the end of the denning period, when outside air temperatures are warmer, and also to argue that deep spring snow would not be necessary for insulation, since the maximum insulating properties of snow are achieved at 30-50 cm depth.)
Magoun also takes issue with the contention that wolverine dispersal will be inhibited or restricted due to climate change. Her arguments again rest on the idea that the snow model in Copeland et al was inappropriately parameterized and that late spring snow is not the governing factor in defining wolverine habitat. If female wolverines are not tied to snowbound home ranges, this argument goes, then dispersing juveniles will not have to travel farther to establish new home ranges. She points out that wolverines are not restricted to snowcovered areas in order to disperse, and questions whether earlier snowmelt would inhibit connectivity among wolverine population nodes.
With snow dependence and connectivity eliminated as concerns, the argument concludes, there is no justification for listing wolverines as endangered.
Bob Inman’s stance echoes Magoun’s in the assertion that USFWS did not use the best available science to reach the listing decision. He begins with an analysis of a 2007 paper, Aubry et al’s Distribution and Broadscale Habitat Relations of the Wolverine in the Contiguous United States, which looks at historical distribution of wolverines and concludes that the Great Lakes and northeast never (within historical times) harbored reproducing populations of wolverines. The conclusion in the paper supports the idea that wolverines are constrained by snow, since these regions are outside the current snow model. Inman takes issue with this, suggesting that the paper engaged in faulty and circular logic by using historical records to define wolverine habitat, and then using that definition to determine which historical records represent reproducing populations and which represent dispersers. He adds that human interference by the time historical records were being kept (19th century) could as easily account for wolverine absence from the northeast and the Great Lakes as climate constraints. Inman advocates for a food-storage-based hypothesis for constraints on wolverine distribution, as opposed to an obligate snow-denning hypothesis, and by throwing the historical distribution into question, he builds a case for alternatives to the snow model.
Inman then goes on to reiterate a number of Magoun’s arguments about Copeland et al 2010 and McKelvey et al 2011. Inman explicates something that Magoun discussed in broader terms, namely that wolverines may be able to den in small snow patches and that the McKelvey et al paper, in looking at the entire area mapped by Copeland et al as wolverine habitat, failed to assess threats to the areas of the snow map that would be most relevant to wolverine denning – ie, high, northfacing slopes, where all Rocky Mountain wolverine dens have been found to date. In short, he contends that all the snow in the Rockies could melt out during the wolverine denning period with no effects on wolverines, so long as snow remained on these northfacing slopes.
In a 2009 paper, Michael Schwartz and the team at RMRS used genetic analysis to place the US Rocky Mountain wolverine effective population size – that is, the number of wolverines out of the population at large that are actually reproducing - at about 35, with a range between 28 and 50, out of a total population of about 300. This low number has caused concern because it is a very low number, lower than the proposed minimum for long-term persistence of a population on a landscape. Inman suggests that the analysis in the 2009 paper intentionally excluded samples that would have given a higher effective population size. He also points out that we don’t actually have samples from the entirety of occupied or proposed-occupied range in the Rockies, which may also bias the effective population estimates downward. This leads him to conclude that the concern about the low number is misplaced, and not a reason for listing.
Inman then addresses two human impacts on wolverines. The first is infrastructure development, which was ranked of relatively little concern in the listing decision. The second is trapping, which was found to be a secondary threat. Inman thinks that infrastructure development could have greater negative consequences for wolverines than climate change, and suggests that the proposed listing rule dismisses this potential threat without adequate reasoning, citing several studies that show that large chunks of wolverine dispersal habitat are in private ownership and that development could become a problem.
Trapping, on the other hand, is sustainable, in Inman’s view. The crux of this argument rests on determining whether wolverine mortality is compensatory – that is, wolverines that are killed in traps are excess, non-contributing members of the population who would die without reproducing anyway – or additive – that is, wolverine deaths by trapping remove more animals than would normally die, and therefore reduce the population below carrying capacity. Inman argues that trapping in Montana is compensatory, not additive. He looks at two papers that are frequently cited to suggest that trapping mortality is additive (Krebs et al 2004, Squires et al 2007), and refutes them on the grounds of statistics and some behind-the-scenes social factors that, he says, provoked trappers to a sort of revenge trapping frenzy in the study area after being insulted by a researcher’s anti-trapping attitude. He also suggests that survivorship estimates in researched wolverine populations are biased low because much of the research is conducted on “front country wolverines,” wolverines that are more accessible, therefore closer to humans, therefore likely to have higher mortality rates in traps (because trappers, like researchers, aim to make things easier on themselves by trapping in relatively accessible locations.)
Inman also offers detailed comments on the Colorado reintroduction plans and the 10(j) rule. Since those are not relevant to the immediate debate about the relationship between wolverines and snowpack, I’m going to gloss over this for now. He concludes that the rule as written doesn’t reflect the best available science and that conservation actions would be difficult because the decision identifies climate change (which cannot be regulated under the ESA) as the primary threat, while relegating infrastructure, development, and human activities (which can be regulated under the ESA) to a secondary concern, or not a concern at all.
Montana Fish Wildlife and Parks
MFWP criticized the listing rule on the grounds that it represents a repudiation of Montana’s management of wolverines, which MFWP contends has been responsible and conservative. Asserting that the population is healthy and still expanding, they advocate for an “adaptive management” approach that monitors wolverines until the population reaches a threshold that would trigger listing. They cite many of the same arguments and points addressed by Magoun and Inman, but frequently carry these arguments a few steps further. Harking back to the politicized and spurious argument in the 2008 decision against listing, they suggest that wolverines in the Rockies are not a DPS and therefore cannot be listed, since listing under the ESA relies on a determination that the listable population is distinct. They emphasize that the proposed relationship between wolverines and spring snowpack, as put forth in Copeland et al 2010, is simply a hypothesis, not a proven fact, and pluck at a statistical point that Magoun and Inman also reference: the fact that 69% of the dens in the Copeland et al paper were located in areas that retained deep spring snowpack in 6-7 years out of 7, which MFWP interprets to indicate that wolverines den outside of the snow model 30% of the time. They pose Inman et al 2012 as an alternative to Copeland et al 2010, saying that the discrepancies in the snow model can be explained by considering a food-based hypothesis for distribution. They also go so far as to suggest that earlier melt out will benefit wolverines because of increased productivity in wolverine habitat.
MFWP refutes the idea that observed wolverine expansion over the past few decades is the result of a reduction in mortality. It’s not entirely clear to me whether they think that recolonization following extirpation never happened, or whether they are taking issue with the idea that recent reductions in the trapping quotas led to accelerated expansion, but they do state that the listing decision offers no proof that expansion following reduced mortality is a true explanation for recolonization They state that wolverines in Montana are “at or near capacity” to indicate that there is an excess population that can tolerate trapping. They emphasize, repeatedly, that wolverine populations have grown in “…size and distribution, concurrent with declines in spring snow cover and trends toward earlier runoff,” which they equate with conditions that will exist in the future. They also quarrel, however, with the predictive powers of the climate models that were used to determine that wolverine habitat is threatened.
MFWP concludes with the fact that the ESA cannot regulate climate change and says that therefore wolverines shouldn’t be listed, since status under the ESA wouldn’t do anything for them anyway.
What are we supposed to do with all of this information? This is the question that faces decision-makers when they try to interpret science to create policy. Obviously, it’s not an easy task. In the next post, I’ll address the different stories about wolverine ecology that arise from these different scientific analyses, and then discuss the different policy prescriptions that would be appropriate if a given story were true.
In the meantime, a bit of media: here’s a story about the reopening of the comment period which also gives some details on the study in the North Cascades.
And perhaps of note, for the past week I’ve been keeping an eye on wolverine literature published in 2013. This morning I noted for the first time a paper by Inman and several co-authors, with an October 2013 publication date, proposing yet another habitat model and making suggestions for wolverine management at a meta-population scale. I haven’t yet had time to read this paper in detail, but will post more here in coming days, as my schedule allows. For anyone with academic access, the article is available here.