Predicting the Future of the Future: ESA, Climate Change, and the Evolving Contours of “Best Available Science”
Recent cases involving imperiled species present interesting case studies of the tensions inherent in the Endangered Species Act [ESA] in dealing with science, uncertainty, and the politics of climate change. The first of these cases addresses the ESA listing of the wolverine; the second, of the Beringia bearded seal. Both cases concern species dwelling on the front line of the impacts of climate change: the Arctic and near-Arctic. Both cases raise the question of whether model-based climate predictions constitute the “best scientific and commercial data available” upon which to base government listing decisions. Together, the cases highlight both strengths and weaknesses of the ESA to protect species in the realm of a warming climate.
The Wolverine Case: Defenders of Wildlife v. Jewel
The North American wolverine, the largest terrestrial member of the weasel family, relies on deep snow to den and reproduce. Climate models predict long-term decline in snow cover in the wolverines’ habitat. The wolverine, not unlike the polar bear, is expected to be affected directly over the coming decades by warming temperatures and declines in snow levels. Seeing this, the U.S. Fish and Wildlife Service (FWS) proposed to list the North American wolverine as a threatened species. This proposal, however, inspired a torrent of comments to FWS questioning or opposing the decision. The states of Colorado, Utah, Alaska, Idaho, New Mexico, Nevada, Oregon, Washington, and Montana all filed comments, either questioning the FWS analysis (Oregon and Washington), or outright opposing the listing (Utah, Alaska, Idaho, Nevada, and Montana). Following its receipt of these comments and a great deal of internal debate, FWS did an about-face and withdrew the proposed rule that it had issued 18 months earlier. The withdrawal led to the expected legal challenge filed by Defenders of Wildlife against the Secretary of the Interior, and the Director of FWS.
Ironically, the wolverine’s population in the United States is not obviously decreasing. So the case initially presented the question of whether it is appropriate to list a species as threatened based on long-term climate models when its present population is not necessarily in decline, and may even be increasing. Much of the case, however, turned on whether the FWS used the “best available science” as the ESA requires in listing decisions. The science relied upon by FWS for the proposed listing rule was complex – even cutting edge. It involved the down-scaling of an ensemble-average of long-term, coarse-scale, global climate models (“GPMs”), an averaging of selected emissions models, an hydrologic model which translated the climate models into snow-water equivalent and snow depth, and satellite imagery to correlate the GCM outputs to the ground. This enabled the mapping of known wolverine dens to current and projected snow cover. The modeling work came largely from two papers published in 2010 and 2011. The science was impressive, yet admittedly imperfect, incomplete, or uncertain in various respects, particularly in the spread of uncertainty in emission models over decades, since the modeling went out to 2045 and 2085, and in the downscaling of global models to attempt to make them locally predictive at a finer scale of resolution.
The climatemodels were those recommended by the Intergovernmental Panel on Climate Change (“IPCC”) and are used by scientists and governments the world over. There are approximately 20 of these models and the study relied on by FWS averaged ten of them. But despite the high competence accorded IPCC climate models, modeling Earth’s future climate with any degree of precision, particularly beyond 2050, is not surprisingly an uncertain enterprise. Predictive climate models depend on the unknown future level of greenhouse gases (carbon dioxide, methane and nitrous oxide) emitted into the atmosphere. “By mid-21st century, the magnitude of the projected climate change is substantially affected by the choice of emissions scenario.” And there is exceedingly wide variability in the projections. Emissions models, of which there are about 40, are really socio-economic-political assumptions incorporating various combinations of fossil fuel use, economic growth, transitions to renewable power sources, and global efforts to reduce emissions. The model used in the wolverine case adopted a scenario “reflecting a rapidly growing economy but with significant movement toward renewable power sources.”  The reviewing District Court characterized this choice as a “mid-range to conservative scenario.” Yet one may question if application of this scenario is appropriate now in light of the energy policies of the new administration. In any event, these not-strictly-scientific choices are inevitably components of the “best available science” used by the FWS and NMFS, and will take on increasing importance – and likely invite still more controversy – for any listing decisions relying on longer-range, post 2050, projections.
Partly on account of such factors, and likely because of resistance to federal listing decisions in general, Western states’ responses, summarized collectively below, strongly disputed whether the climate modeling constituted the best available science: 
- Unvalidated, non-rigorous, speculative, global CGM models are too coarse and imprecise to be used as predictors “at the landscape level”– the GCM’s are not “best available science”; the prediction is, rather, a “hypothesis”;
- The climate may simply be in an historically-recurring period of drought, through which the species has survived in the past;
- The use of untested or unverified models to predict where the wolverine may be, models “that speculate on a species’ possible future fate,” should be set aside and the service should use current demographics which are more reliable;
- Ascertaining the fate of the species over the next 30 to 75 years is not the “foreseeable future”; it is speculation;
- Because the population impacts [that] these models predict are highly uncertain, it is not necessary to immediately list the species;
- Loss of the sub-species in the contiguous United States would not represent a significant gap in relation to its entire range, which includes areas within the contiguous United States, Canada and Alaska. The population in the lower 48 states represents a small fraction of the entire range, such that it is insignificant when compared to its entire range.
- There are too many unknowns as to how wolverines will respond to changes, indeed they may adapt.
The court’s reaction to these concerns from the states was blunt: “Of the numerous Western states which urged the Service not to list the wolverine and attacked the agency’s reliance on… [the study model], not one provided any scientific evidence directly rebuffing the study’s conclusions.” In other words, the comments were not in the Court’s view scientifically substantive. It appeared that the court instead saw the comments as per se objections against the use of climate projections in listing decisions.
As to whether the modeling was the best available science, the court held that FWS improperly “sought certainty beyond what is required by the ESA and case law interpreting it …” There were admittedly gaps in causal relationships between deep, persistent, spring snow and wolverine dens, and a lack of understanding why wolverines appear to require such conditions for denning. There were objections as to “limitations inherent in downscaled climate models,” and concern by some that climate models tend to predict temperature more confidently than precipitation. And two the authors of one of the studies had misgivings about its application and did not support the peer reviewers’ favorable conclusions. The judge, however, held that the service “must use the ‘best scientific data available,’ not the best scientific data possible.” In other words, the science needn’t be perfect – it can even be incorrect in respects, but if it’s the best available (and it may be the only available) then that’s the end of the story insofar as the ESA is concerned. There may be critical data gaps bearing on the question. There may be significant levels of uncertainty in projections in the near or far term that could well make a scientist or statistician in another context hold back from publication until the data gaps were filled and the analysis done more comprehensively. Yet the ESA is a species management tool, and the science, however imperfect, if it is the best available, will meet the ESA standard. In short, the court held that FWS should not have “ignored the best available science by demanding better science.”
Chief Judge Christensen determined that the FWS’s change of position, withdrawing its original proposed rule to list the wolverine, and discrediting the same science it had so meticulously used to support its original decision, was arbitrary and capricious. “No greater level of certainty is needed to see the writing on the wall for the snow-dependent species standing squarely in the path of global climate change.”
Put another way, the court’s message was that it is beyond dispute that the globe is warming, that northerly, snowy ranges, particularly at the lower latitudes of such ranges, will eventually be snow-free, and that this completely snow-dependent creature will be imperiled. “It doesn’t take a weatherman to know which way the wind blows…” and the best available science, even if incomplete or flawed, is certainly good enough, in the court’s view, to tell us what everybody already knew. The court seemed to imply that demands for extra-fine precision in the modeling were possibly disingenuous, concealing perhaps more fundamental, policy-oriented objections to the use of the ESA to conserve species in this long-term, projective way.
The Bearded Seal Case: Alaska Oil and Gas Association v. Pritzker
The wolverine case entailed two complete listing turnarounds. FWS’s listing was followed by its withdrawal of the listing, which was followed by judicial vacatur of the withdrawal, and remand. The bearded seal case involved a similar suite of listing turnarounds: the National Marine Fisheries Service (NMFS) adopted a final rule listing the Beringia bearded seal DPS. The Alaska District Court vacated the listing and the Ninth Circuit then reversed the District Court’s vacatur, and remanded.
NMFS proposed to list the bearded seal because of long-range climate change projections showing loss of essential ice habitat in the circum-Arctic oceans where the bearded seal lives. The distinct population segment (DPS) specifically inhabiting the Bearing and Chukchi Seas is called the Beringia DPS. Bearded seals are widespread, have low population densities, and spend much time underwater. So it is difficult to obtain reliable estimates of their population. The seals use non-contiguous ice floes in relatively shallow waters close to underwater food sources to whelp and nurse their pups, as platforms for resting, attracting mates, and avoiding predators. In this case, NMFS used six IPCC-recommended climate models to forecast the future of this sea ice habitat. NMFS also employed a Biological Review Team consisting of eight biologists, chemists and a climate scientist to review the status of the best available science data, and four independent scientists on top of that to conduct peer reviews of the Review Team’s report. The modelers used the same socio-economic-political scenario for emissions as was used by FWS for the wolverine (emissions from a rapidly growing economy but with significant movement toward renewable power sources). In addition they used an alternative, more aggressive scenario (representing heavy use of fossil fuels) to “represent a significant range of variability in future emissions.” NMFS’s climate model and listing rule established a 100-year projection of steady declines in sea ice and longer ice-free periods in May and June, the most critical months for mating, birthing and rearing the seals. “By the end of the 21st century, projections for the Bering Sea indicate that there will commonly be years with little or no ice in May, and that sea ice in June is expected to be nonexistent in most years.”
The predictions, as in the case of the wolverine, were considered reasonably supported by data through about 2050. This is because the model to mid-century will estimate the unfolding impacts of greenhouse gas emissions already known to be present in the atmosphere. NMFS, however, emphasized the greater uncertainty inherent in any longer-range prediction since it must rely upon estimating unknown future emissions.
The district court seized upon this long-term uncertainty in vacating the listing rule as arbitrary and capricious. “Under the facts in this case, forecasting more than 50 years into the future is simply too speculative and remote to support a determination that the bearded seal is in danger of becoming extinct.” In the court’s view there was no significant threat to the bearded seal before 2090. “[The] farther into the future the analysis extends, the greater the uncertainty.” Significantly, the court also held that there was insufficient data regarding the actual impact on the population from the gradual loss of sea ice, and that additional studies are needed.
The Ninth Circuit rejected this insistence upon the need for more certainty, and reversed the Alaska District Court’s decision. The court noted NMFS’s projected monthly sea ice concentrations between March and July for each decade from 2025 to 2095; the results (presumably in the near term) performed reliably well when applied to observational data in the Arctic. The court was possibly also influenced by a dramatic new study showing that temperatures in the Arctic were 30 years ahead of schedule and that the Arctic would be nearly ice free in summer by mid-century. The Ninth Circuit held that the IPCC climate models constituted the “best available science” and “reasonably supported the determination that a species reliant on sea ice would likely become endangered in the foreseeable future.” The projection up to 2050 was a period in which “specific data” supported the IPCC climate projections. What about the reliability of projections after that time, when the choices among emissions scenarios cause the results to diverge more dramatically?
According to the court, NMFS’s projections for the second half of the century were also reasonable, scientifically sound and supported by evidence:
The fact that climate projections for 2050 through 2100 may be volatile does not deprive those projections of value in the rulemaking process. The ESA does not require NMFS to make listing decisions only if the underlying research is ironclad and absolute. … NMFS provided a reasonable and scientifically supported methodology for addressing volatility in its long-term climate projections, and it represented fairly the shortcomings of those projections – that is all the ESA requires.
The Ninth Circuit also noted there is no ironclad definition of “foreseeable future,” and that different time frames are appropriate for different species and different threats. The court was emphatic that the ESA does not require an agency to obtain data which does not exist. It is merely to use the best available data, and explain its shortcomings.
The court also addressed the lower court’s conclusion that NMFS was unable to provide any evidence of the actual impacts on the bearded seal population, or of an extinction threshold, or of the probability of reaching that threshold. In other words, to the lower court there was a great deal of inference – but no evidence – that predicted losses of sea ice would cause the decline in seal populations. While this would be considered a fatal flaw in some areas, the Ninth Circuit held that under the ESA, all that was required of NMFS was a reasoned explanation of why it believed there existed a causal link between the loss of habitat and a species’ survival: “NMFS has demonstrated that it “considered the relevant factors and articulated a rational connection between the facts found and the choices made.” In other words, professional opinion, rationally explained, could substitute for evidence of a causal connection. This, too, is part of the evolving meaning of “best available science.”
Curious Similarities Among the Cases
There are similarities worth noting between these two cases:
- Both entailed listings of reclusive/remote northerly species where field data is difficult or impossible to obtain;
- Both involved species dependent for their critical life events upon continued snow (wolverine) or ice (bearded seal) habitats expected to be at risk from foreseeable temperature and precipitation trends;
- Both involved species whose populations are now limited but believed to be stable;
- Both cases used an ensemble of sophisticated global computer models to predict climate and habitat changes over extremely long periods of time;
- Both cases used peer reviewers to review the results of the science;
- Both encountered arguments against the use of climate change projections in ESA listings;
- Both encountered arguments that 50 and particularly 100 year projections are speculative and not within the “foreseeable future” within the meaning of good science or the ESA rules;
- Both court cases ultimately upheld the use of decades-forward climate change projections incorporating emissions scenarios as the best available science, if not the best possible science.
These cases together also illustrate the strengths and weaknesses of the ESA in the realm of a warming climate. Strengths: there is actually a statutory framework (the ESA listing process) whereby threats to species caused by environmental changes, on great scales of space (global effects) and time (decades) can actually be considered in an orderly administrative setting. This can be accomplished with scientific input and peer review, subject to judicial oversight. Weaknesses: the exhaustive, expensive, lengthy, back-and-fourth history of each of these cases, and evident lack of universal consensus for the use of climate science in listing decisions, particularly beyond mid-century.
The wolverine case tested the best available science concept, as well as how that concept can be manipulated by decision-makers when a particular outcome is sought. The bearded seal case, in particular, tested the limits of what is “foreseeable” and how tenuous the line is between predictability and chance.
Both cases also show the limits of the ESA’s necessarily constrained enforcement role in climate-affected spheres. When the FWS proposed to list the wolverine, it flatly stated in the rule that it would not regulate greenhouse gas emissions. Similarly, no one would expect NMFS to attempt to regulate emissions to slow the loss of sea ice and conserve the bearded seal. If this is the case, then what is all the fuss about? Why go to all the trouble both to list (and to resist listing) if nothing, really, can be done? Why list in such cases? FWS put it this way:
[FWS] expect[s] that it will indirectly enhance national and international cooperation and coordination of conservation efforts, enhance research programs, and encourage the development of mitigation measures that could help slow habitat loss and population declines.
It is likely that the tensions in the ESA listings between the science and best available science, between evidence and reasoned opinion, will themselves continue, far into the future.
This article is not a substitute for legal advice. Please consult with your legal counsel for specific advice and/or information. Read our complete legal disclaimer.