Introduction

The global debate about the safety and environmental impact of transgenic crops is now at least twenty-five years old. In that time scientists working at the interface between science and policy have had to confront significant challenges to several long-held cherished ideals. These challenges include the notion that scientific evidence or fact is rarely absolute but invariably contingent and influenced by context, and that, in this Wiki world, the conviction with which a particular opinion is held is somehow to be regarded as being as important as the strength of the scientific evidence on which it is based. Additionally those scientists who refuse to proselytise or engage in a debate about the ‘unknown unknowns’ have occasionally had opprobrium heaped on their heads by their social science colleagues (Gray 2004).

Rather more weighty criticism has been levelled specifically at the science of ecology and its application to real-world problems. Two short quotations from R H Peters’ thought-provoking book capture this well: “academic ecology poses unanswerable questions”(p 13), and “the central constructs in ecology yield predictions with difficulty and these are often so qualitative, imprecise and specific that they are of little interest and less utility” (p 273) (Peters 1991). However such views are rarely reflected in the applied ecology literature which contains a good deal of optimism about the science’s ability to deliver solutions to practical problems. In a recent editorial statement entitled “Ensuring applied ecology has impact” the aim of The Journal of Applied Ecology was described as “to drive forward the field of applied ecology by providing a high quality evidence base for scientists, managers and policymakers” (Milner-Gulland et al. 2012)—a mission statement which embraces the idea of science as a ‘servant’ of policy. Evidence for such a relationship between ecological science and policy is widespread in the literature. For example a survey by Ormerod et al. (2002) indicated that more than half of the papers in two sample journals which included management or policy recommendations were taken up and utilised in some way by this target audience (57 % of papers in Journal of Applied Ecology 1999–2001, 51 % in Conservation Biology 1996–1998). Types of uptake included the design and management of nature reserves and forests as well as the determination of agri-environment policy. More than a third of the applications were to problems in agroecosystems.

Many practising ecologists can give examples where their research has provided evidence for action by managers and policymakers. Three examples from the author’s work are (1) simple regression models that relate the elevational range of intertidal plant communities to tidal amplitude and other factors being used to predict the distribution of mudflats and salt marshes following the construction of estuarine barrages or reservoirs (Gray 1992, Gray et al. 1995), (2) analyses of the genetic structure of populations of a range of rare grasses with contrasting breeding systems helping to formulate conservation and restoration policy for these species (Gray 1996a, b, 2002), and (3) the categorisation of intertidal habitats according to their sensitivity to water-borne oil providing the basis for plans to deploy barriers within a harbour in the event of oil spill (Gray 1985).

Despite many such examples, and their continuing and deep interest in questions of policy relevance (Sutherland et al. 2006), there is a perception among ecologists of a gap between science and policy. This paper, based on a keynote lecture delivered at IBSGMO 12, examines the relationship between science and policy as it pertains to the debate about the environmental impact and safety of transgenic crops, focusing on some problems of language and communication and on some issues surrounding the definition of ‘harm’.

The science: policy gap

In the author’s experience problems hindering engagement and communication between scientists and policy-makers derive mainly from the fact that they tend to work at different levels of generality (policy often deals with broad issues of value, science prefers specific questions and experiments—see also Sutherland et al. 2006), and complexity (policy-makers are looking for simple answers, ecologists tend to offer multi-factorial explanations). They also tend to work to different timescales (policy-makers want answers tomorrow, ecologists always seem to want time for more research). These differences are by no means unique to the debate about the cultivation of transgenic crops. Recent comments by Georgina Mace, President of the British Ecological Society, in the Society’s Bulletin (43, June 2012) illustrate the common problem of communication. Referring to a recent major international conference on global sustainability she observes “…the science was mostly in symposia, the policy discussions mostly in plenary. But making the two work together was hard to do, especially as most of the influential policy makers came in, read their 10 min view of the world, and left almost immediately (stopping briefly for a quick photo opportunity) …. So, while many scientists felt that the meeting was weak on science, many policy makers perceived it to be another event where scientists talk to each, and ask for more money for their research. What was missing was much real interaction”. Sadly such a scenario is all too familiar to scientists involved in the transgenic crops debate, where the problems are undoubtedly exacerbated by the subject’s political contentiousness.

Figure 1 is an attempt to depict the science-policy gap as it specifically relates to transgenic crops. The evidence produced by scientific experiment and survey is on the left and the objectives of policy to which that evidence relates are on the right. Pathway (1) represents a simple relationship in which evidence is used to inform policy and policy looks to the evidence for guidance and direction. A ‘decision’ is made on the basis of both and is the driver for generating further policy and gathering evidence. This relationship reflects the ‘science as a servant of policy’ viewpoint mentioned above and is undoubtedly naïve. The single dimension of the diagram disguises the way in which science and policy are intimately bound together by a value-laden, messy and iterative process (hence the chicken and the egg metaphor) quite unlike that characterised by the ‘encounter problem –> do some science –> make policy’ model. Pathway (2) shows the more specific case of transgenic crops where science forms the basis of an Environmental Risk Assessment (ERA) which informs policy and in turn policy objectives define the key parameters of that ERA—for example what constitutes environmental ‘harm’ (see later).

Fig. 1
figure 1

The Science—Policy gap (see text for explanation of diagram)

Full size image

Pathway (2) recognises that, in an ideal world, a decision about the cultivation and management of a given transgenic crop is made by reference to the results of an ERA and the policy objectives of the country or region. It is sometimes difficult for scientists to accept that (3) is a legitimate, if politically generated, situation in which the decision is made without any reference to scientific evidence (e.g. in countries that wish to advertise that they are ‘GM free’).

Bridging the gap: facts and values

The messiness and complexity of the relationship between science and policy can be illustrated by a discussion of the topic ‘gene flow’. Defined technically as ‘the incorporation of genes into the gene pool of one population from one or more [different] populations’ (Futuyama 1998), gene flow has several recognisable stages leading to hybridisation (for plants this includes coincident flowering, pollen transfer, successful fertilisation, seed production, and germination and growth of the F1 hybrid) and introgression (further crossing and survival and spread of individuals possessing the genes of the F1 hybrid). A more or less ubiquitous process in natural populations, gene flow (or its absence) is a major driving force of evolution and, together with other factors such as the breeding system and population size, is the architect of population genetic structure and diversity.

The advent of transgenic crops gave new impetus to the study of gene flow—but in a quite restricted context. This is the concern that gene flow from transgenic crops to wild relatives might be an agent of ‘harm’; specifically that a crop-wild relative hybrid may become an invasive plant in natural habitats or a weed of agriculture (or in some cases may be undesirable in centres of diversity or origin of the particular crop species). Additionally crop-to-crop (transgenic to non-transgenic) gene flow may be viewed as a threat to the genetic integrity of traditional land races or the quality or purity of the non-transgenic crop, especially in a different agricultural system (e.g. ‘organic’ agriculture). These types of gene flow have been occurring since agriculture began and are not uncommon (Ellstrand 2003). What is different is they are now viewed as possible sources of environmental harm and therefore the risks from gene flow have to be assessed as part of the ERA. This imbues the process of gene flow with an additional significance that demands that the outcomes of scientific survey and experiment are evaluated from a perspective of what they tell us about ‘risk’ or ‘safety’. An ostensibly neutral and widespread biological phenomenon has been invested with value.

For example the fact that gene flow occurs between oilseed rape/swede (Brassica napus) and turnip (B. rapa ) has been known for a long time. Hybridisation between B. napus and B. rapa in agricultural environments was studied more than 50 years ago. Palmer (1962) described experiments which indicated low rates of hybridisation in swede fields when turnips were present, even in excess, but high rates in turnip fields (>50 % in 1:1 ratios), a contrast which reflects the self-incompatible breeding system of B. rapa. Very similar results were reported when this problem was revisited as a result of the transgenic crops debate, but the implications were now somewhat different (Jorgensen and Andersen 1994; Jorgensen et al. 1996). For example the high probability of producing herbicide tolerant hybrid B. rapa weeds in B. napus crops had to be evaluated as a potential harm.

The hybrid between B. napus as a crop and the wild B. rapa, wild turnip or Bargeman’s cabbage, has been found in the British flora with sufficient frequency to have been given a name—Brassica X harmsiana (Stace 1975). Again extensive studies of the distances over which hybridisation may occur and the frequency and survival of hybrids in the wild (in frequently-disturbed, ruderal and streamside habitats close to agriculture) were undertaken in order to quantify the risk of transgene ‘escape’ (see reviews in Lu 2008; Gray 2012). Raybould (2010) and others have argued that much of this research, whilst adding hugely to our knowledge of gene flow, has contributed little to improving the risk assessment. Furthermore, evaluation of the same data has led to different policy decisions. The production of herbicide-tolerant Brassica weeds in rotations involving herbicide-tolerant oilseed was judged to be unacceptable in some jurisdictions (e.g. many EU countries) but both acceptable and manageable in others (e.g. Canada, where the oilseed crop is currently more than 95 % herbicide-tolerant).

The debate about transgenic crops is not unique in ascribing a value judgement to the outcome of gene flow. In different contexts gene flow is seen by wildlife conservationists to be either a ‘good’ thing (e.g. in preventing inbreeding depression, rescuing genetically depauperate or rare species populations) or a ‘bad’ thing (e.g. threatening rare or declining species, causing outbreeding depression in locally adapted populations). Thus on the one hand the objective of policy may be to ensure or enforce hybridisation or, on the other hand, to prevent it. Examples of the former include the rescue of the endangered dusky seaside sparrow (Ammodramus maritimus nigrescens) in Florida, USA, by captive breeding with other subspecies (Avise 1996) and of the rare lakeside daisy (Hymenoxys acaulis var. glabra) in Illinois where the remaining plants shared self-incompatibility alleles (DeMauro 1994). The many examples of gene flow prevention (see Gray 2002) include active programmes of control of the species that presents a threat—one such is the costly effort to eradicate the new world ruddy duck (Oxyura jamaicensis) in Europe to prevent hybridisation with the European white-headed duck (O. leucocephala) (Hughes 1996). A 5-year programme in the UK costing £3.4 m aims at ensuring the survival of the white-headed duck as a distinct species.

Bridging the gap: lost in translation

The Editorial in Journal of Applied Ecology mentioned above (Milner-Gulland et al. 2012) includes a table of articles from that journal which have featured in a news and information service for policy-makers in Europe (Science for Environmental Policy. http://ec.europa.eu/environment/integration/research/newsalert/index_en.htm). This service is designed to keep policy makers in the EU up-to-date with the latest research findings and includes headline summaries of scientific papers. For example an article by Bulleri and Chapman (2010) entitled ‘The introduction of coastal infrastructure as a driver of change in marine environments’ is given the headline news item ‘Coastal structures change marine ecosystems’, whilst one by Chytry et al. (2008) entitled ‘Habitat invasions by alien plants: a quantitative comparison among Mediterranean, subcontinental and oceanic regions of Europe’ has the headline ‘Which habitats are most at risk from invasive species?’. Also listed in the table is a paper by Lavigne et al. (2008) the original title of which is ‘How do genetically modified (GM) crops contribute to background levels of GM pollen in an agricultural landscape?’ The news item reporting the results of this paper for policy-makers is headlined ‘The bigger picture; GM contamination across the landscape’. It is difficult not to take the view from this headline that there is something harmful about GM pollen (The Oxford Thesaurus gives equivalents of the verb ‘contaminate’ as ‘defile, sully, pollute, corrupt, rot, stain, soil, taint, infect, poison, foul, spoil, befoul, debase, adulterate, vitiate, dirty, make impure, make radioactive’—all words indicating a negative impact, all implying harm). A public debate about pollen movement and gene flow thus becomes heavily laden with values about their desirability and impact in advance of any knowledge of their actual effects.

Indeed it has become difficult to avoid the implication of harm in discussions of gene flow (which may help to explain how non-scientists can arrive at the belief that gene flow is harmful per se). Although Lavigne et al. carefully and correctly define the parameter they are concerned with as ‘adventitious transgene presence’ (one of the rather clumsy alternative and more neutral phrases scientists prefer) they find it easier in the text to abbreviate it to ‘impurity rate’. The words ‘contamination’ and ‘impurity’ appear to have become an accepted part of the GM lexicon and even appear in the titles of scientific publications (e.g. Haslberger (2001) GMO contamination of seeds. Nature Biotechnology 19, 613; Sanvido et al. (2011) Status of feral oilseed rape in Europe: its minor role as a GM impurity and its potential as a reservoir of transgenic persistence. Environmental Science and Pollution Research 18, 111). Possibly to avoid accusations of pedantry, scientists appear to have accepted the wide use of such value-laden words, as well as terms such as ‘GM food’ which have either little meaning scientifically or have extremely broad prescriptions.

Again the transgenic crops debate is by no means unique in the field of applied ecology in being subject to the power and subtleties of language. Webb and Raffaelli (2008) analyse the language used by the various exponents in a debate about the proposal to cull hedgehogs (Erinaceus europaeus) on three Scottish islands. Introduced to South Uist in 1974 to control garden pests, hedgehogs have become significant predators of the eggs of ground-nesting shorebirds and the publically funded conservation organisation Scottish Natural Heritage proposed a cull of hedgehogs to halt the decline in shorebird numbers. This was vehemently opposed by animal welfare and hedgehog conservation groups (who generally advocated trapping and translocating hedgehogs) and the conflict was widely debated in the media. Webb and Raffaelli’s analysis of the texts generated by this debate illustrates how different stakeholders tend to use different language such that in extremis a pro-wader view might speak of ‘the cull of Erinaceus’, a pro-hedgehog view of ‘the killing of hedgehogs’ and the media of ‘the slaughter of Mrs Tiggywinkle’. Words such as ‘cost’, ‘expense’, and ‘taxpayer’s money’ or ‘biodiversity’, ‘welfare’ and ‘control’ were deployed in ways which reflected the underlying values of these three classes of stakeholder.

It is important to acknowledge that, whilst the media versions of the debate tend to be colourful, there is a sense in which all three are correct. More scientific evidence, unless it significantly altered our understanding of the impact of hedgehogs on shorebird populations (not disputed here), is unlikely to help. In fact, as in other conflicts, including aspects of the GM debate, further evidence can be used by both sides to ‘make environmental controversies worse’ (Sarewitz 2004).

One of the most misused of the value-laden words in the GM debate, and one which science has a limited role in defining, is ‘harm’. In particular the use of ‘harm’ in the context of ERA for transgenic crops seems to create a number of problems.

Where’s the harm?: linking science and policy objectives

Scientists involved in ERA for transgenic crops sometimes find their engagement with policy- and decision-makers clouded by misunderstandings about what one should expect from the other. Not the least of these is that science can define ‘harm’. An obvious generic objective of environmental policy is not to cause harm, but deciding exactly what constitutes environmental harm is not a trivial exercise. Regulatory and advisory bodies around the world have attempted to provide working definitions of harm, often in the context of their local environments (e.g. ACRE 2002; EC 2004, CBD 2006). A common feature of such definitions is that harm involves a ‘significant’ impact on a population of an organism (usually a decline in numbers of a ‘valued’ species but occasionally an increase as in the case of pest outbreaks), and the advent of a problem formulation approach to ERA has facilitated the recognition and definition of such populations in terms of specific protection goals (Raybould 2006; Wolt et al. 2010; Gray 2012). But, as Sanvido et al. (2011) point out, particularly in the context of monitoring post-market environmental change, the decision about how much the numbers have to change to constitute an ‘adverse’ effect, and indeed which are the key species populations to measure in the first place, depends as much on ethical, political, social and economic considerations as on any parameters that science can offer. Defining harm is truly an exercise for policy-makers to engage with scientists but one in which they must consider a range of other factors.

The difficulty of deciding which environmental changes are harmful, and should therefore be addressed during the formulation of policy objectives, can be illustrated by a consideration of the UK’s Farm Scale Evaluations (FSEs) of genetically modified herbicide tolerant (GMHT) crops.

The FSEs were carried out on a representative sample of UK farms between 2000 and 2005 and involved a comparison of around sixty pairs of fields for each of four crops—spring- and winter oil seed rape, sugar beet and forage maize. The paired fields comprised a GMHT and a conventional cultivar of each crop species (glyphosate-tolerant beet and gluphosinate ammonium-tolerant rape and maize). This huge, and costly, experiment was arguably the largest single experiment on farmland biodiversity ever undertaken and is certainly the largest single exercise in applied ecology designed to underpin ERA for transgenic crops. The trials were undertaken against a background of concern about the decline in the UK in farmland biodiversity, particularly in populations of several bird species dependent on farmland, during the last 20 years. Because birds were involved the FSEs had a very high public profile in the UK, whilst the science aimed simply to test the formal null hypothesis that there would be no differences between GMHT and conventional crops in terms of the abundance of weeds, weed seeds and invertebrates (lower parts of the farmland food chain for birds).

The details of the FSEs, including the methods, design, statistical analysis and results have been fully published (see Firbank et al. 2005 and references therein). Summarised below are four general results, sufficient for a discussion of how ‘harm’ has been used in their interpretation.

  • In all four crops the null hypothesis was falsified by clear and significant differences between treatments (GMHT and conventional fields).

  • The differences were solely attributable to the management of the crops, i.e. the herbicide regime, rather than the GM crop itself.

  • When compared to their conventional counterparts, fields of GMHT beet and spring and summer oilseed rape had lower biodiversity as measured by the abundance of weeds and most invertebrates. Forage maize by contrast showed an increase in biodiversity in GMHT fields associated with the later application of the broad spectrum herbicide.

  • The differences between different crops were as great, or greater, than those between GMHT and conventional treatments within the same crop species. Both treatments of spring-sown oilseed rape were more biodiverse than all other crops.

These results have unleashed a wide range of reactions and comment. Official bodies, especially in the EU, have deemed them to contain sufficient evidence of harm to invoke a ban on GMHT cultivars of the crops in which biodiversity was reduced (notably spring-sown oilseed rape and sugar beet—the results in winter-sown oilseed rape were less clear cut). In a strict technical sense where the policy objective, or protection goal, is to halt or reverse the decline in farmland biodiversity such a decision appears sound. ‘Abundance of farmland birds’ for example is a clear assessment endpoint that could be negatively affected if these GMHT crops replaced the conventional counterparts, at least on any large scale (see below). However, as the UK’s advisory body, ACRE (Advisory Committee on Releases to the Environment), has pointed out, the results of the FSEs raise some key issues for future applications to cultivate transgenic crops (ACRE 2007). Not the least of these is the fourth finding listed above, that the differences in biodiversity were greater between crops than between GMHT and conventional versions of the same crop. Such were these differences that the policy objective of conserving farmland biodiversity would be better achieved by, for example, replacing all current forage maize crops with GMHT spring-sown oilseed rape! Although an absurd suggestion, this example emphasises the more important point that wider changes in UK agriculture, including the replacement of pasture by arable forage crops such as maize, have had a huge impact on biodiversity. Trends such as the change from spring to winter crops and from hay to silage have significantly reduced the abundance of weeds and weed seeds available for farmland birds and are acknowledged to be key factors in their decline (Donald et al. 2001; Edwards and Hilbeck (2001).

Other regulatory issues raised by the FSEs include those of scale (how to incorporate differences in scale, farmer take-up, and severity of effect into the prediction of future scenarios), valid comparators (how to make comparisons between the effects of transgenic versus non-transgenic crops in a shifting agricultural background) and mitigation (how to deal with transgenic cropping systems which preserve biodiversity by conserving field margins or band-spraying row crops (e.g. Dewar et al. 2003; Pidgeon et al. 2007). These uncertainties led ACRE, who also point out that other systems of regulating agriculture permit applicants to balance possible harms with any positive benefits of cultivation, to propose that ERAs for transgenic (and all novel) crops, should be widened to include a comparative sustainability assessment that considers both costs and benefits of the agricultural systems of which they are a part (ACRE 2007 –see also Pretty 2008).

Not unexpectedly the wide range of contrasting reaction and comment on the FSE results has reflected the value-systems of the many stakeholders in this highly politicised debate. From politicians there was a plea for more time—“longer term trials, at least 4 years long, are needed, especially of maize” (Report of the House of Commons Environmental Audit Committee). This uncharacteristic request for more science may have been prompted as much by the wish to further delay a decision on the cultivation of maize as by the desire to remove uncertainties (the 4-year moratorium on GM cultivation which accompanied the trials having protected politicians from making a decision which would be unpopular with a hostile public). Breeders of sugar beet pointed out that GMHT crops could be deployed to farm in a wildlife-friendly way; “the FSEs demonstrate an opportunity to manage sugar beet for the benefit of farmland birds” (see Dewar et al. 2003). In contrast an employee of the Royal Society for the Protection of Birds was moved to say on BBC radio that “[the FSEs prove] that GM crops harm birds”. These and other perspectives will no doubt prove fertile ground for social science. From an applied ecologist’s viewpoint, however, the FSE results speak directly to the policy-maker in a situation where halting the decline in farmland biodiversity is enshrined in the national legislation (as it is in the UK in the form of statutory Biodiversity and Habitat Action Plans).

They are also of value in situations where the environmental objectives are quite different. In Australia, for example, and arguably in most countries around the globe, the presence of weeds in arable fields is not regarded as ‘a good thing’ since it is likely to affect crop yields and quality [especially in Australia where most of the arable weeds are exotic non-native species of no biodiversity ‘value’ (CSIRO 2003)]. In the FSEs the infield weeds are not ‘weeds’ in the widely accepted sense (i.e. a plant in the wrong place) since they support farmland biodiversity and thus do not cause ‘harm’. Indeed the concern about farmland biodiversity is almost uniquely an issue in the UK and Europe where most of the landscape is farmed, there is very little ‘wilderness’, and valued habitats tend to be plagioclimax communities with many centuries of management by grazing or forestry. Additionally food security is high and the population includes many affluent, well-fed and risk-averse people who have leisure time (to watch birds for example). In other countries, especially those where food security is low or agricultural intensification has not been an issue, the role of GMHT crops in weed control will be viewed very differently. And almost certainly not as a ‘harm’.

The Farm Scale Evaluations provide a powerful example of the problem highlighted earlier; the tendency of policy to deal with broad issues whilst science poses specific questions. Such differences in generality mean that there is unlikely to be a one-to-one correspondence between any single piece of science (the trials) and the policy it is designed to inform (the protection of farmland biodiversity). Many uncertainties remain. The question arises as to whether attempting to remove some of the residual scientific uncertainties (e.g. the long term impact of different types of weed control on the provision of seed for granivorous adult birds versus invertebrate food for chicks) is likely to provide value for money when the major effects may depend on factors such as farmer take-up and crop management. Models provide insight but may also have limited generality. For example Watkinson et al. (2000) model of the impact of reduction in the seed output of a single farmland weed species, Fat Hen Chenopodium album, on a single bird species, Skylark Alauda arvensis, was used to make general assertions about the overall affect of GMHT crops on farmland biodiversity. While criticised for its simplicity (e.g. Firbank and Forcella 2000; Carpenter et al. 2002) this model indicated that the overall impact on skylarks would be significantly determined by the extent and pattern of farmer take-up, specifically whether farmers with weed-rich or weed-poor fields were more likely to adopt GMHT technology. This key factor introduces uncertainties in prediction which a better scientific understanding of the underlying processes in the ecosystem concerned is unlikely to resolve. More ecology may not be helpful.

In highlighting the general problem of linking specific scientific investigation with broader policy objectives, an exception should perhaps be made for the exemplary study of the impact of Bt corn pollen on the Monarch butterfly (Sears et al. 2001). In this case a well-resourced, well-designed and thorough study of the range of variables that determine the field exposure of Monarch larvae to lepidoptera-active Cry1 proteins, and therefore the impact on the species’ populations of cultivating Bt corn, produced an ERA which directly informs a policy to avoid harm to a valued species.

Concluding remarks

Notwithstanding the problems discussed in the Introduction, applied ecology has a proud record of problem solving, especially in the fields of nature conservation and environmental policy (Milner-Gulland et al. 2013). Arguably the record of ecological research undertaken in support of ERA for transgenic crops is no worse than that in other policy areas. It suffers from the same systemic problems which beset the science, such as the difficulty of generalising from specific cases or experiments and, conversely, of applying generalised models to local situations or populations. It has the same problems of defining environmental harm, of value-laden language and of the need to improve communication between different stakeholders.

If the role of applied ecology in the transgenic crops debate can be characterised as being different in any way from that in other areas of environmental policy it may be with respect to three aspects of that debate. First the GM debate is especially contentious, highly politicised and inclined to polarisation between entrenched pro- and anti- GM factions, ensuring that discussions about science are frequently underpinned by a fierce debate about values (Sarewitz 2004). Second, the science serving the debate has usually been undertaken as part of an ERA, in which the emphasis is on the ‘risk’ of harm; rarely is it about ‘benefit’ or weighing risk against benefit. This contrasts with most applied research in the field of nature conservation which is aimed at problem solving. Finally, the objectives of environmental policy in this area seem to be particularly broadly stated. Generic policy statements which invoke difficult concepts such as ‘sustainability’ and ‘biodiversity’, and aspire solely not to cause harm to the environment, provide a significant challenge to ecologists working in the field.