The politics in environmental science: The Endangered Species Act and the Preble's mouse controversy

Abstract

Environmental science presupposes value judgements. Calls for sound, objective science are examined in the context of the US Endangered Species Act (ESA) and the controversy that surrounds the listing of Preble's mouse as a ‘threatened’ subspecies. Following an overview of this debate, its status as a controversy that can be resolved by objective knowledge alone is problematised by examining the normative underpinnings of species classifications, divergent risk orientations as they apply to concerns over classificatory errors, and the stating of testable hypotheses. Gleaning insights from the discussion, policy-relevant suggestions are provided.

Keywords:

Introduction

The ‘best available science’ mandate of the Endangered Species Act; the Data Quality Act of 2001 (which authorises the Office of Management and Budget to maximise its use of objective data); and countless references by politicians to ‘sound science’ – the assumption in each case is that good science rests upon objective, value-free data. By removing values from the decision-making equation, foundational truths can be revealed that will, in turn, lead the way for good policy. If only this were true. Unfortunately, the banner of objective science is often waved by politicians to mask some very subjective beliefs and assumptions. In stating this, I am not suggesting that ‘bad’ environmental science is that which is infiltrated by value statements; that otherwise objective science is being reduced to junk science by intentional acts of data manipulation. While science can (and occasionally is) manipulated, the argument of this paper is even more radical: namely, that environmental science presupposes value statements. Total objectivity, at least in the context of the environmental sciences, is a chimera (Jasanoff and Wynne 1998; Wynne 2002; Latour 2004). 1

The intention of this paper is thus to not speak of how science can become politicised. While the exposure of fraudulent knowledge claims (be they scientific or otherwise) is undeniably important, we must likewise be aware of what such arguments tacitly imply: that science and politics are inherently different entities. These arguments thus leave untouched beliefs relating to the objective authority of science. Indeed, if anything, they only help to further ingrain such views, for they imply that value-laden science is not really ‘science’ but ‘politicised science’ – an example of what others have referred to as ‘boundary work’ (Gieryn 1983; Jasanoff 1987). This paper instead seeks to illustrate the political choices that are confronted routinely within the environmental sciences. To speak of politicised science is thus a bit of a misnomer. For, in the end, environmental science requires the making of political, which is to say normative, decisions.

What, then, are we to make of those repeated calls by politicians for sound, objective science? In some cases at least, such calls no doubt reflect a naive understanding of science. Yet, in other cases, one must wonder if something significantly less sincere is happening, where the rhetoric of objective science is being used as a smokescreen to mask underlying value-laden statements. In doing this, this discourse of objectivity becomes a Trojan horse of sorts, which provides a convenient ruse for making normative judgements appear benign and rational (which has led one noted science policy scholar to proclaim that ‘science can make environmental controversies worse’ [Sarewitz 2004, p. 385]). After all, arguments for more and/or better objective knowledge are very difficult to counter. Who would dare question the place of such knowledge in the decision making process?

This paper takes a closer look at these calls, particularly as they emerge in the context of the US Endangered Species Act (ESA) and the controversy that surrounds the Preble meadow jumping mouse's listing as a ‘threatened’ subspecies. In doing this, specific attention is given to how values are an intimate part of this so-called objective scientific debate. In this case, we have a debate over whether or not Preble's mouse (as it is often called) constitutes a distinct subspecies. If it does, then its threatened status under the ESA holds, as do all the protections that accompany such a listing. But if it does not – that is, if it can be categorised with one of two unlisted subspecies of jumping mice (either the prairie jumping mouse or the Bear Lodge meadow jumping mouse) – then no such special protections would be granted.

To set the stage, a brief overview of this debate is provided. This is to highlight just how frequently the rhetoric of objective science is used by scientists and politicians alike. After briefly describing the case, I then move to problematise its status as a controversy that can be resolved by objective scientific data alone. In doing this, I reveal the numerous value judgements embedded within the seemingly objective claims that abound in this debate. For example, when working toward species delineations, regardless of whether it is within the context of the ESA, operational criteria must first be chosen. Yet, as later argued, such an act requires the making of value judgements, given that such criteria are not objectively given. Beyond this, delineating between species also requires a priori assumptions about risk orientations toward classificatory errors: that is, should a risk adverse stance be taken, whereby type I errors (false positives) are reduced, or should a more precautionary stance be embraced, whereby type II errors (false negatives) are minimised? One cannot answer this question without first relying upon value judgements. Even the stating of testable hypotheses is not a purely objective act, remembering also that which hypothesis is ultimately selected can be of significant political consequence.

To conclude, insights are gleaned from the aforementioned discussion to make policy-relevant suggestions. In particular, this paper calls upon politicians to be more open when it comes to expressing their underlying values. Rather than utilising terms such as ‘objectivity’ and ‘sound science’ to give an air of legitimacy to what are in reality very value-laden positions, politicians need to be more accountable to their constituents. In other words, they need to explain why they value and believe what they do, instead of relying upon the rhetoric of objective science to skirt such a discussion entirely.

The Preble's mouse debate and calls for objective science

Preble's mouse was first classified by Edward A. Preble in 1899. Its habitat is confined to the fast-growing Colorado Front Range and south-eastern Wyoming. In 1998, Preble's mouse was listed as ‘threatened’ under the ESA. By 2003, over 31,000 acres had been designated as critical habitat (Fish and Wildlife Service 2003); a regulatory move estimated by some to have cost developers, local governments and landowners as much as $100 million in lost revenues (Lehr 2004). At this point, the debate surrounding this mammal heated up. If the listing of this mouse had been economically inconsequential, it is quite likely that the story would end here. Yet because so much more was (and is) at stake than simply the survival of this isolated population of jumping mice, its subspecies status garnered (and continues to garner) a significant amount of national attention.

In December 2003, Rob Ramey, a biologist who at the time was employed by the Denver Museum of Nature and Science, published a report in which he was lead author (see Ramey et al. 2005). This article disputed the claim that the mouse is a distinct subspecies. Upon being published, the US Fish and Wildlife Service (FWS) initiated the process of delisting Preble's mouse as a threatened subspecies. During this time, in an attempt to resolve the issue, the FWS hired the Portland, Oregon-based Sustainable Ecosystem Institute to conduct its own independent analysis of Preble's mouse. In July 2006, the findings of this analysis were made public. They had concluded that Preble's mouse is a unique subspecies and that the earlier study by Ramey and colleagues was flawed (specifically, they believe Ramey et al. mixed up DNA samples from Preble's and Bear Lodge mice during analysis). And, at least as of March 2007, it appears that Preble's mouse will retain its ‘threatened’ status.

What is perhaps most interesting about this debate is how it has become near universally accepted among involved scientists, politicians and other stakeholders that it can be resolved with objective criteria alone – criteria that, in the end, will show us whether or not Preble’s mouse is distinct subspecies. The then US Congressman Bob Beauprez (R-CO), for instance, released a press statement proclaiming how the move by the FWS to begin the delisting process for Preble's mouse was an ‘example of letting good science dictate sound policy’. 2 He also commended both Roy Ramey, lead author of the aforementioned report that led to the initiation of the delisting process, and the FWS for their commitment to ‘sound science’. In fact, in this short press release of 337 words, Congressman Beauprez made reference to ‘sound’ or ‘good’ science on four separate occasions.

Similarly, Congresswomen Barbara Cubin (R-WY) praised the FWS for initiating the process of delisting Preble's mouse, proclaiming: ‘This entire episode highlights the problems of junk science within the Endangered Species Act’ (Casper Star-Tribune Editorial Board 2006, p. D3). Similar rhetorical posturing can also be found among those supporting the mouse's threatened status. For example, Jeremy Nichols, conservation director for Biodiversity Conservation Alliance, released a press statement in January 2005 criticising the move by the FWS to delist Preble's mouse, proclaiming: ‘This proposal is a devastating blow to open space across the Front Range of Colorado and Wyoming, to good science, and to the public interest.’ 3 Of course, these individuals are only evoking a language that has already been institutionalised within the ESA. The ESA repeatedly makes reference to the need for making regulatory decisions that rest ‘solely on the basis of the best scientific and commercial data available’ (16 USC §1533 [b] [1] [A]). This has been referred to by others as the best available science mandate of the ESA (see e.g. Doremus 2004).

This then raises the question: just what does ‘sound’, ‘good’ or the ‘best available’ science look like? While an answer to this question is far from obvious, we can be certain of one thing: that such a definition cannot be limited to purely objective criteria alone. As I now detail, to speak of objective science as it applies to ESA listing determinations is to engage in a form of rhetorical posturing, in that it gives an air of legitimacy to the side employing this strategy by cloaking its views about how the world should look in a discourse that appears rational and value free.

The ESA: Looking beyond objective science to resolve conflicts

The best-available-science mandate of the ESA rests upon a linear view of science and policy, for it presupposes that ‘good’ policy comes from ‘good’ (which almost universally means objective) science (Pielke 2004). Yet, as detailed below, this view poorly captures the reality of the situation. In what follows, I unpack the so-called objective claims surrounding the Preble's mouse controversy so as to illustrate the numerous normative assumptions they rest upon.

The species problem

Ten years ago, a review of the modern biological literature found at least 24 different species concepts (Mayden 1997), and since then additional works on the subject appear at a steady pace (e.g. Schulter 2001; Isaac and Purvist 2004). The ‘species problem’, as it has come to be known among philosophers of biology and those in the biological sciences, is a product of an inherent ambiguity that surrounds defining and operationalising ‘species’. Divergent species concepts have emerged over the years as biologists continue to record properties that define diverging population lineages. The problem, however, is that they then assume that these characteristics are necessary properties of species. For example, to mention a few of the more widely recognised contemporary species concepts, the biological species concept centres on the property of reproductive isolation (e.g. Mayr 1942), whereas the ecological species concept emphasises speciation through ecological selection (Schluter 2001) and phylogenetic species concepts focus on the property of common descent (Ridley 1989). The problem with such concepts, however, is that they rest upon the assumption that speciation occurs in a fixed, universal manner (Sites and Marshall 2004). Yet while properties of each of the abovementioned species concepts have been recorded by biologists, to quote the renowned zoologist Kevin Queiroz (2005, p. 6600), they ‘are neither necessary nor sufficient for the definition of the species category’.

A case in point: the biological species concept (arguably the most widely used of all species concepts). As Mayr – who, along with Dobzhansky (1937), first put forward this concept – explains, ‘species are groups of interbreeding natural populations that are reproductively isolated from other such groups’ (Mayr 1970, p. 12; while Mayr's definition evolved over the years it remained centred on interbreeding). While having a certain theoretical elegance to it – after all, organisms delimited by interbreeding represent a seemingly clear mechanism for speciation – the biological species concept has faced stiff criticism over the years that it has failed to adequately address. One of the most significant critiques is that it is entirely useless when directed at organisms that never (or only occasionally) engage in sexual reproduction, such as those that reproduce by vegetative means or self-fertilisation.

Rather than resolve the species problem, the rise of genetic tests (e.g. mitochondrial DNA analyses) have only served to further highlight the limits of objective scientific knowledge when it comes to making such determinations. For example, Hendry and colleagues (2000) used genetic data to evaluate the validity of various bird species. Their findings highlight the inherent ambiguity of species classifications. As they explain, ‘‘For example, 25 of the 109 sister species of birds we considered showed less than 2% sequence divergence, whereas 11 of the pairs showed greater than 10% sequence divergence. The dichotomous nature of species delineation ignores quantitative differences between species, and effectively considers all equally distinct’ (Hendry et al. 2000, p. 73).

This brings us to how values inevitably play a role in species delineations. For example, given that most biologists allow for some gene flow between species when applying the biological species concept (Hey 2001), when should two populations be defined as distinct species – when they are 99%, 80%, 51% or 10% reproductively isolated? An answer to this question cannot be arrived at through objective measures alone. In short, all species concepts have to assume threshold levels of difference, none of which are objectively given (Hendry et al. 2000).

It is against this value-laden backdrop that we must place listing determinations under the ESA. Yet, given their location within the messy reality of regulatory politics, classifications under the ESA are even more problematic than the above mentioned discussion would suggest. That is because, while called the Endangered Species Act, its concerns are actually more broadly defined.

In 1978, US Congress amended the ESA's definition of species to include ‘any subspecies of fish or wildlife or plants, and any distinct population segment of any species of vertebrate fish or wildlife which interbreeds when mature’ (16 USC § 1532 [6] and [16]). Thus, in addition to the earlier mentioned species problem, we also face within the ESA the ambiguities associated with defining and operationalising ‘subspecies’ and a ‘distinct population segment’. Given that the ESA does not proscribe clear operational criteria from which to make species delineations, scientists are left to decide what a species (or subspecies or distinct population section) is. Yet what if the involved scientists come to implicitly rely upon different species concepts? Such appears to be, at least in part, a source of some of the conflict that has arisen in the case of Preble's mouse.

I was unable to find a single instance where ‘subspecies’ was clearly defined among the various peer-reviewed reports debating whether or not Preble's mouse is a distinct subspecies. Could it be that the scientists are in part talking past each other due to divergent understandings of what constitutes a distinct subspecies? While this certainly does not capture the debate in its entirety, one can find hints within this literature that divergent species/subspecies concepts are at least adding to the controversy.

For example, while Ramey and colleagues (2006) admit that Preble's mouse is a reproductively isolated population, they nevertheless conclude, based upon physical (cranial morphology) and genetic (mitochondrial DNA) features, that the mouse is not a valid subspecies. Conversely, Vignieri and colleagues (2006, p. 236), in an article written in response to the initial Ramey report, ‘regard its [Preble's mouse] geographic and genetic isolation … as [one] operative hypotheses that must be explicitly disproven’. Thus, whereas Vignieri and colleagues indicate working, at least in part, from a species concept that centres on reproductive isolation (in other words, the biological species concept), Ramey and colleagues do not view the property of reproductive isolation as necessary in their definition of species. In short, while scientific methods are no doubt being debated in these pieces, a closer read also reveals a far less objective source of contention, which rests upon divergent beliefs about how ‘species’ should be defined and operationalised.

Nor is the Preble's mouse controversy unusual in this regard. The species problem, broadly defined, has been at the heart of a host of ESA listing debates. Perhaps the most famous of these debates within the US conservation community centres on whether or not hatchery and wild salmon are to be included in the same distinct population segment (the genetic independence of grey wolves in Yellowstone National Park is another well known case). Briefly, under the ESA, the National Marine Fisheries Service (NMFS) is charged with protecting many endangered species of fish and marine mammals. Rather than species, however, the NMFS protects what is known as ‘evolutionarily significant units’ (ESUs), which ‘is defined as a genetically distinct segment of a species, with an evolutionary history and future largely separate from other ESUs’ (Myers et al. 2004). Yet ESUs, like species, are not objectively given. This helps explain the endless debate that continues to swirl around how best to conceptualise and operationalise ESUs for purposes of assessing the extinction risk to salmon in the wild. Critics of hatcheries, for example, note that while genetic similarity can be used to classify hatchery salmon as part of the ESU from which they were derived – thus making them the genetic equivalent to wild salmon – such a focus ignores other biological differences between the two that can greatly impact the survivability of hatchery fish in the wild (Waples 1999). In the end, underlying this debate are different conceptions about what constitutes a ‘salmon’ in the first place; a question that cannot be answered wholly by science (Scarce 2000).

Confidence limits

As in all scientific debates that involve a degree of uncertainty, thresholds have to be decided upon that will dictated the level at which hypotheses are either accepted or rejected. Yet these thresholds are not objectively given, which is to say they are matters of politics. To introduce how divergent risk orientations (or thresholds, confidence limits, etc.) can shape how one assesses hypotheses in the face of uncertainty, let us first look briefly at the field of epidemiology.

As members of a social network – with its own norms, identity structures and social conventions – epidemiologists' understanding of ‘evidence’ is different, at least to a degree, from those embedded within different social networks (e.g. community activists). Among professional epidemiologists, social conventions strongly discourage the making of false positives – that is, concluding that there is a causal link between contaminant X and a cancer cluster when there is not. This explains why a higher confidence limit (of typically 90% or higher) is usually the norm within this professional field. Scientists have a reputation and identity to uphold as a result of their membership to a larger professional community. This ‘expert’ reputation and identity could be severely undermined in the event of proclaiming a causal link when, in fact, none exists. Such type I errors (also called false positives) could thus not only be embarrassing but potentially damaging to one's future membership within the community (and the resources associated with it, e.g. access to grants). Moreover, if such errors occurred with great frequency across the social network, it could risk undermining the legitimacy of the entire profession.

On the other hand, community activists, as a result of their different network connections and interests, have equally compelling reasons to avoid at significant costs false negatives (also called type II errors). For them, erring on the side of caution, if it means saving their lives and the lives of loved ones, is well worth making a type I error every now and again (Kleinman 2005). This in part explains why proponents of what has been called ‘popular epidemiology’ prefer a much lower confidence limit when examining causal relationships between pollutants and cancer clusters (Brown and Mikkelsen 1997).

Another way of thinking about this distinction could be in terms of ‘normal science’ and ‘precautionary science’. 4 When speaking of normal science, I am referring to the cultural orthodoxy of science, which, for various reasons, is averse to type I errors. Normal science represents that mythical view of science we are taught in grade school, where science is the fountainhead of truth and scientists the paragon of objectivity. Type I errors risk undermining such myths and are thus often vigorously avoided by scientists through their recognising as ‘evidence’ only that which meets thresholds of confidence of (typically) 90% or higher. There are no essential or universal criteria saying science must assume high thresholds of confidence – that it must, in other words, be averse to type I errors. Rather, it is simply assumed that this is how science should operate (for reasons mentioned earlier). This is not to suggest that all scientists prescribe to this epistemic viewpoint. Depending, for example, upon the situation or the ‘epistemic culture’ (Knorr-Cetina 1999) they locate themselves in, scientists find reasons to believe that a less strict threshold should be applied. In such instances, we can say that precautionary science is being practised. Again, neither normal nor precautionary science is inherently ‘better’ than the other. Rather, which stance one should take depends upon what it is that one values. To position this within the broader science and technology studies (STS) literature, questions over levels of confidence reflect a broader debate around when to accept evidence as either fact or fiction – what is often referred to as the ‘problem of demarcation’ (see e.g. Collins and Evans 2002; Evans 2005; Lahsen 2005).

Returning to the aforementioned Preble's mouse controversy, we can find examples where conflicting scientific conclusions have partial grounding within divergent views about which threshold levels should be applied in cases of hypothesis testing. According to Ramey and colleagues (2005), molecular techniques reveal Preble's mouse to be 99.5% genetically similar to other subspecies of mice – not enough, they argue, to constitute a distinct subspecies. Yet this begs the question: how much of a genetic difference is ‘enough’ to constitute a distinct subspecies and what objective criteria determine this cut-off point? Given the best-available-science mandate of the ESA and that the FWS began the process of delisting this animal as a direct result of this group's research, such criteria must exist – right?

Rather than resting upon objective criteria, these delineations are instead a product of value statements. In other words, they rest upon the belief that a strict epistemic threshold that limits type I errors should be applied in this case. In an interview to the Rocky Mountain News (a Denver-based newspaper), Rob Ramey hinted at this when he said: ‘If you're willing to keep this [Preble's mouse] listed as a subspecies, then how far are you willing to go?’ (Erickson 2006, p. A2)? In this statement, we find Ramey's reason for why 99.5% genetic similarity is not enough to justify calling Preble's mouse a distinct subspecies: because such a standard would unduly allow type I errors to be committed, where populations are defined as distinct species/subspecies when in fact they are not.

Conversely, in their response to the 2005 study conducted by Ramey et al., Vignieri and colleagues (2006) appear to take a more precautionary posture when it comes to making species classifications. In their own words, ‘[W]e must seek to maximize evolutionary potential through the protection of populations on separate evolutionary trajectories. … [T]he most important aspect of preserving biodiversity is protecting evolutionary potential’ (Vignieri et al. 2006, p. 237). In other words, they argue that populations should be delineated between not just in instances where clear genetic differences exist, but even in those cases where there is the potential for such differences. Clearly, this better-safe-than-sorry position is far more accepting of type I errors than the stricter level of confidence called for by Ramey and colleagues.

Is it any surprise, then, that these groups of scientist disagree when it comes to determining whether or not Preble's mouse constitutes a distinct subspecies? In the end, the debate does not just rest purely on the evidence of the matter. Also significant are the divergent value judgements that actors bring with them into this case, which ultimately shapes not only how evidence is viewed but also what constitutes ‘evidence’ in the first place.

While the ESA calls for making decisions based on the best available science, this should not be viewed as a clear policy directive in favour of normal science (where type I errors are reviled). Numerous examples can likewise be found both within the ESA itself as well as among the numerous reports published by the FWS and the NMFS, to support the taking of a precautionary science position. Take, for example, a non-binding handbook published in 1998 by the FWS and the NMFS. This handbook explains that, in cases where the FWS and the NMFS cannot fulfil their duty of providing a biological opinion because of insufficient data (and where the agencies do not agree to provide more time to collective for information), they can proceed to ‘develop the biological opinion with the available information giving the benefit of the doubt to the species’ (US Fish and Wildlife Service and National Marine Fishers Service 1998, p. 6).

I mention this point not to make an argument for either normal or precautionary science. Each has its place within regulatory and conservation politics. Yet, as the Preble's mouse controversy illustrates, this is something that requires discussion and debate. As things stand now, at least in the context of the ESA, individuals are working from divergent presumptions as to what threshold of evidence should be used to evaluate hypotheses and make species delineations. These presumptions need to be made more explicit, recognising that such is a matter of policy and politics, not objective science.

This is not to suggest, however, that this choice rests only upon normative politics. An important epistemological point is often lost in debates surrounding whether or not society should adopt a normative commitment to precaution: namely, the complexity of many of today's environmental problems is such that causalities can rarely be established to a high level of confidence. As environmental scientists are increasingly called upon to understand and predict complex systems – from, for example, global climate change to earthquake occurrence, asteroid/comet impact hazards and groundwater movement (see e.g. Sarewitz et al. 2000) – we must revisit our understanding of what constitutes ‘acceptable’ levels of confidence.

The experimental sciences have long been interested in studying the (largely) immutable characteristics of material phenomena, devoid of natural context, which allows for their reduction into purely mathematical terms. This gives way to the ‘laws’ of the universe – although even these constants of the universe, such as ‘alpha’, are beginning to be understood as not entirely fixed (Choi 2002). And when working with such ‘closed’ systems, with entities that are independent of time and place, normal science thresholds are relatively unproblematic.

Environmental problems, however, involve ‘open’ systems, which inevitably introduces uncertainty in the modelling process. As Sarewitz and Pielke (1999, p. 125) point out, ‘while traditional physical science isolates phenomena from their context in nature in order to understand the invariant characteristics of the phenomena, the integrative earth sciences study the context itself’. Thus, whereas the traditional physical sciences reduce phenomena into precisely described mathematical terms, the earth sciences can at best only identify statistically significant precursors to the ‘event’ being studied. These statistically significant levels, however, may never reach a level of confidence of 90 per cent or higher.

Thus, unless one dare claim that the earth sciences are non-scientific, this highlights how normal science thresholds are not inherent to the enterprise of science. Given the object of study within the environmental sciences – namely, the very context that is removed within the experimental sciences – significant levels of confidence may not always be possible. All of which should cause suspicion towards those seemingly rational calls for high levels of statistical certainty; for such certainty may forever be elusive in light of the inherently variable, complex nature of the systems being modelled (see e.g. Oreskes 2004).

Stating hypotheses

While philosophers of science have long recognised that how hypotheses are stated can be of significant methodological and empirical consequence (Kuhn 1970), this point is often forgotten in analyses that seek to problematise the notion of objectivity within the environmental sciences. When discussing cause-and-effect relationships, such as between an environmental toxin and cancer, the null hypothesis always assumes that any such relationship could have arisen through chance. Yet, within the realm of environmental science, and conservation science in particular, the issues often cannot be reduced to such cause-and-effect relationships. The practice of stating hypotheses thus becomes more open to political contestation, given that the weight of scientific tradition does not weigh heavily on any one particular hypothesis frame.

In the case of the ESA, two perfectly reasonable hypotheses could be ‘the species is endangered’ or ‘the species is not endangered’ (see also Ruhl 2004). Yet each hypothesis frames the debate in radically different ways. If the former is selected (‘the species is endangered’), regulatory action can only take place in those instances where the hypothesis has been confirmed. Such a hypothesis thus stacks the deck in favour of a status quo, business-as-usual environment. It would also lead property rights groups (and ESA foes more generally) to argue in favour of taking a normal science stance when it comes to evaluating evidence, given how such a position requires a greater level of certainty before a hypothesis is said to be confirmed. Environmental and conservation groups, by contrast, would argue for precautionary science so as to lower the required threshold of evidence and thus make such cases of hypothesis confirmation easier to attain.

Conversely, if the latter hypothesis is selected (‘the species is not endangered’), regulatory action will take place except in those instances where the hypothesis has been confirmed. This hypothesis thus stacks the deck in favour of a highly regulated environment. In this case, the methodological posturing is reversed from that described above. Here, it is the environmental and conservation groups that would argue in favour of taking a normal science position, whereas property rights groups would prefer the employment of more precautionary thresholds.

In the case of the ESA, Congress dictates these hypotheses through policy directives. Typically, the hypotheses they state are inclined to promote the status quo. This is why pro-industry, free-market and private property groups overwhelmingly proclaim the need for sound and good science to inform regulatory and conservation policies: because these terms imply the use of normal science confidence levels when evaluating evidence. No doubt, if Congress were to choose to restate these hypotheses, thereby creating a policy environment sympathetic to regulation, these groups would suddenly have a change of heart and call for the use of more precautionary measures. Examples of a shifting methodological posture can be found within Congress.

While Congress has voted on numerous so-called ‘sound science’ bills in recent years to fix what many there view as being a broken ESA, such bills apply largely to agency decisions when it comes to listing species under Section 4. Rarely (if indeed ever), however, have these legislative prescriptions been applied to decisions to allow development, such as in cases of issuing incidental take permits under Section 10 (Ruhl 2004). In the light of such discrepancies in the application of normal science thresholds, one must question the sincerity of those who claim to have no agenda when touting sound science.

Turning once again to the Preble's mouse controversy, we find indications that the debate over whether or not this mouse constitutes a distinct subspecies is in part an effect of differing hypotheses statements between scientists. Arguing in favour of its distinct subspecies status, Vignieri and colleagues (2006, p. 237) state that the mouse's ‘geographic and genetic isolation, occurrences in an ecoregion distinct from that of conspecifics, and formally described distinctive phenotypes of pelage and skull shape [should be taken] as operative hypotheses that must be explicitly disproven for synonymy to be accepted’. In other words, their hypothesis is ‘Preble's mouse is a distinct subspecies,’ and it is up to science (and Ramey et al. in particular) to prove otherwise.

Ramey and colleagues, in contrast, approached the subject with a different hypothesis in mind: ‘Preble's mouse is not a distinct subspecies.’ While they set out to test a number of hypotheses, synonymy appears to underlie each: for example,

[W]e retested the original quantitative basis of Krutsch's (1954) conclusions to split Z. h. campestris into three subspecies (Z. h. preblei, Z. h. campestris and Z. h. intermedius) … [and also] tested genetic and ecological exchangeability (Crandall et al., 2000) of Z. h. preblei relative to other subspecies to determine if it should be considered a distinct population (Ramey et al. 2005, p. 330).

Couple this with a normal science evaluatory threshold – e.g. ‘If defensible data are lacking and a protected organism is not distinguishable with a high degree of certainty from neighbouring, non-threatened relatives …’ (Ramey et al. 2005, p. 329, my emphasis) – and we find a much different epistemological starting-point from that described earlier by Vignieri and her collaborators.

Conclusion

The purpose of this paper is not to argue that science is of no value to conservation and regulatory efforts. Environmental policy needs science. Yet, importantly, the science it needs is not the mythical view of science that we are taught about in grade school, which rests upon the holly trinity of eternal truths, objectivity and precision. That science does not exist. It is important for us to understand this point. For as long as we continue to buy into this myth, it will continue to be drawn upon by some for very disingenuous ends.

As mentioned earlier, a number of ‘sound science’ bills have been debated in Congress in recent years: e.g. the Sound Science for Endangered Species Act Planning Act of 2002; the Sound Science for Endangered Species Act Planning Act of 2003; and HR 3824 (titled, ‘To amend and reauthorize the Endangered Species Act of 1973 to provide greater results conserving and recovering listed species, and for other purposes’). In each, the goal is to raise the evidentiary bar by requiring more demanding burdens of proof when testing hypotheses, often by way of institutionalising further layers of peer review before data can be acted upon. The most recent of these bills is HR 3824, which passed the US House of Representatives in September 2005. The bill then moved on to the Senate, where it was referred to the Committee on Environment and Public Works (where it still sits at the time of writing). In addition to increasing evidentiary threshold levels and prolonging the peer-review process, HR 3824 would also limit the use of population models that are widely used by biologists when making recovery goals (Trombulak et al. 2006).

When dealing with issues that are not time-sensitive, in a regulatory and political environment with unlimited resources, perhaps such rigorous burdens of proof and time-consuming review processes would be justified. Yet such methodological rigour must be balanced against the fact the some species simply do not have the time to wait for such strict evidentiary requirements to be met. In addition, we must recognise that such increases in burdens of proof and time consuming peer-review processes must also be met with additional funding. To continue to underfund the ESA – the median expenditure per listed species in 2004 was a mere $5,500 (although, in actuality, 4% of listed species received 84% of all funds) (Trombulak et al. 2006) – while increasing the procedural and financial burdens of the FWS and NMFS suggests there is more to these bills than concerns over producing quality data.

Likewise, it need not only be the endangered or threatened species that is on a time-sensitive schedule. While this point is often forgotten about by conservationists, economic and land development issues can also be time-sensitive due to ever-changing market conditions. Thus the application of more rigorous standards to habitat conservation plans (also known as HCPs) for evaluating the impacts from ‘taking’, while unlikely to offend environmentalists, would enrage property rights groups. And this, in turn, could risk further undermining the legitimacy of the ESA in Congress (Ruhl 2004).

An argument could in fact be made that Congress's so-called ‘sound science’ bills are in reality ‘anti-science’ in their spirit. Philosophers of science – from the rationalist Karl Popper to the anti-rationalist Paul Feyerabend – have not agreed on much over the decades, save for perhaps this one point: science is an never-ending dialogue. Sound science bills seek to squash this dialogue, by portraying science as something that speaks only in certainties, proofs and ultimate truths. In avoiding uncertainties, then, Congress is making new, groundbreaking research all the more difficult to attain given its preference for ‘puzzle-solving’ Kuhnian normal science – to the exclusion of what Kuhn (1970) called revolutionary science. As long as science is cast as something that only deals in certainties, powerful interests will continue to exploit this misunderstanding of science for personal gains by engaging in what Michaels (2005) have called the ‘manufacturing of uncertainty’. From the tobacco industry denying the links between smoking and cancer (e.g. Michaels 2005) to big oil and the global warming debate in the US Congress (e.g. McCright and Dunlap 2003), the strategy of creating uncertainty has already paid significant dividends for commercial interests in its ability to ‘scientifically’ justify the status quo.

And manufacturing uncertainty is exactly what is taking place in many of these cases (versus, say, emphasising uncertainty). In their study of the scientific debate that surrounds the Steller sea lion off the coast of Alaska, for example, Mansfield and Haas (2006, p. 81) note how uncertainty cannot be defined as an objective state alone but is a product of broader social relations: ‘People produce uncertainty when they agree on what is not known … , but also when they disagree on what is known … or even when they agree on what is known but interpret that knowledge in different ways.’ With this point in mind, the authors then go on to demonstrate how powerful interests were able to shift the scale at which the Steller sea lion debate was framed to introduce further uncertainty into scientific models. Specifically, the fishing industry successfully framed the scale of the debate to the level of macro-regional drivers, such as climate change (versus the NMFS who wanted to focus on local interactions between individual fisheries and the Steller sea lion). And, not surprisingly, as ‘the system’ studied expanded in its scope, the scientific models employed rested upon an expanding degree of assumptions and unknowns – thus uncertainty was manufactured.

To conclude, it is important that we understand that the presence of values need not disqualify a knowledge claim from being ‘scientific’ (Jasanoff 2005). Science often – and some would say always (Latour 1987) – requires the making of value judgements. Recognising this will make it more difficult for politicians to skirt discussing what it really is they are arguing for when they evoke such terms as ‘sound’ and ‘good’ science. No longer having these rhetorical veils to hide behind, politicians will be forced to reveal their true political motivations. This will compelled them to discuss, for example, why they think normal or precautionary science thresholds should be employed, why a hypothesis should be stated a particular way (thus giving favour to certain policy outcomes and stakeholders), and/or why one particular species concept should be given greater operational value over another?

Some may argue that these questions are best left up to scientists to decide, given their scientific expertise. Yet, in light of the ‘is-ought’ problem first detailed by David Hume centuries ago (which recognises the logic problems that arise when statements of fact are taken for statements of value), the normative nature of these questions undermines that epistemic authority. A scientist is thus no better suited to deal with these questions of value than is a non scientist (Epstein 1996; Fischer 2000). And given the non-democratically elected nature of scientists, we could do far worse than to settle these questions in the political arena.

Notes

1. I am not claiming that there is anything inherent to environmental science that makes it qualitatively different from, say, high-energy particle physics. I simply do not know as much about other fields of science to extend my argument to them.

2. Available online at http://www.house.gov/beauprez/news/20050128.htm, accessed 20 September 2006.

3. Available online at http://www.voiceforthewild.org/wildspecies/news/n28jan05.html, accessed 20 September 2006.

4. Normal science is a term first popularised by Kuhn (1970). Like Kuhn, normal science in this paper is meant to represent a paradigm of sorts that encourages scientists, through norms and processes of socialisation, to be averse to false positives (type I errors).