viernes, 15 de julio de 2011

Sobre Agendas de Investigación, Hipótesis Nulas y Decisiones Sobre Conservación y Manejo de Biodiversidad

Me contaron que en el contexto de una discusión sobre la conveniencia de permitir o no la pesca con palangres (i.e. "long-line fishing" en inglés) en aguas colombianas, la futura Ministra de Ambiente indicó que antes de tomar decisiones prohibitivas, habría que hacer estudios para determinar el efecto que esas prácticas tendrían sobre las poblaciones de los peces de la región. Las experiencias de otras zonas ya demuestran suficientemente que este tipo de pesca es muy difícilmente sostenible, por lo que el comentario atribuido a la futura ministra me llamó mucho la atención y me recordó un texto que escribimos con Iván Jiménez en 2003 y que nunca publicamos. Como más vale tarde que nunca, aunque ya alguien dijo algo muy parecido en la literatura hace un par años, hemos decidido publicarlo acá. Las ideas, por supuesto, también son relevantes para otras discusiones a las que se enfrenta el país, como las relacionadas con la minería. El texto está en inglés, pero esperamos que sea de interés y estaremos atentos a recibir comentarios:

NGOs and conservation science: a reply to da Fonseca

Iván Jiménez and Carlos Daniel Cadena

In the editorial of a recent issue of Conservation Biology, da Fonseca (2003) discussed the increasing production of conservation science by non-governmental institutions (NGOs), and pointed out associated problems and possible solutions. In particular, he suggested that the conservation community should strive to develop ways to handle two principal problems related to research conducted by NGOs: organizational control of research agendas and the "tweaking" of research results by NGO scientists. Both problems stem from the dependence of NGOs on their funders, and from the beliefs and values determined by the history of these organizations. In his editorial, da Fonseca stated that a third problem, the link between monetary gains and specific research outputs, is prominent in biomedical research carried out by private companies, but has little relevance for conservation research carried out by NGOs. Unlike da Fonseca, we believe that finances contingent on particular research outputs can influence conservation research by NGOs. Furthermore, contingent monetary profits can not only prompt "tweaking" of results, but can also influence the selection of null hypotheses and thereby affect research results and conclusions. Here we elaborate on these problems in the hope that they will be recognized and that appropriate safeguards are put in place.

Let us first consider whether or not having monetary gains linked to specific research outputs can be a significant issue in conservation science produced by NGOs. da Fonseca believes that "individual NGO scientists hold no real financial stakes in studies done under the auspices of their employers". We do not think this assertion is necessarily general. As pointed out by da Fonseca, NGOs are prominent generators of conservation research and are being increasingly recognized as authorities on environmental issues, especially in developing countries. Such recognition seems to drive a variety of corporations with dubious environmental reputation to seek financial agreements with NGOs in order to obtain advice and/or improve their environmental image. These agreements can take multiple forms, but here we focus on contracts through which NGOs are hired to perform environmental studies. NGOs participating in this type of agreement may face a dilemma: they could adhere tightly to their missions, which often reflect a history of conservationist world-views, and potentially put down the contracting corporation, thereby risking present and future contractual agreements. Alternatively, NGOs may forgo their mission and align with hiring corporations, an option that may be particularly appealing given the increasingly competitive nature of the biodiversity conservation market that da Fonseca describes. After all, NGOs' adaptability to the omnipresent quest for financial resources in a fierce competitive milieu might overcome the constraints imposed by their original beliefs and values.

An example of this type of relationship between NGOs and corporations is the contract through which Conservation International-Colombia (CI-Colombia) was hired as consultant by the Empresa de Acueducto y Alcantarillado de Bogotá (EAAB) for a project aimed at restoring the wetlands of the city of Bogotá, Colombia (CI-Colombia 2000). These wetlands were a major center of avian diversification in the Northern Andean highlands (Fjeldså 1985), and have harbored several endemic species and subspecies; some of these are already extinct and many are currently threatened by habitat destruction and degradation (Asociación Bogotana de Ornitología 2000, BirdLife International 2000). Grass-root groups and the local scientific community have serious concerns about the project designed by EAAB because, despite its environmentally-friendly rhetoric, it is essentially directed at urbanizing the wetlands for recreational purposes, with a substantial portion of its vast budget allocated to the construction of concrete structures such as bicycle paths and plazas with concomitant vegetation removal, and meager resources allocated to protection of habitats occupied by threatened species (Stiles 2003). Thus, in proposing how to manage the Bogotá wetlands, CI-Colombia may have chosen an urbanizing approach, supporting its contractor's substantial investment on hard scenic infrastructure for recreational purposes, or a more conservationist approach emphasizing the value of the wetlands' biological community. While CI-Colombia seems to favore the latter (CI-Colombia, page 34), this case clearly exemplifies the potential for NGOs and their scientists to have monetary gains linked, directly or indirectly through future agreements, to specific research outputs.

In addition to monetary profit associated to research outputs, but in close connection with it, there is another important potential problem related to research produced by NGOs: inappropriate selection of null hypotheses. Null hypothesis selection has major consequences for research results and conclusions because it determines what should be assumed in the absence of further study or insufficient statistical power to detect a particular effect size. This is a distinct issue from that of "tweaking of results", because exactly the same data can produce very different conclusions depending on null hypothesis selection (MacKenzie and Kendall 2002, Smallwood 2002). It is fundamental for conservation scientists to realize that the selection of a particular null hypothesis places the burden of proof on showing that such hypothesis is incorrect and, therefore, has major consequences for the conclusions reached by any study. The null hypothesis of a study should epitomize the assumptions made by researchers in the absence of new data and, thus, should be based on all previous knowledge of the study system as well as on theory. In addition, the world-views of scientists have a legitimate role in determining null hypothesis selection (MacKenzie and Kendall 2002, Smallwood 2002). Indeed, scientists in the field of natural resource management often apply the precautionary principle for conservation biology (Shrader-Frechette and McCoy 1993) by choosing null hypotheses that minimize the error of failing to detect a negative effect on the environment (e.g., United States Environmental Protection Agency 1989). Yet, world-views are dependent on prospects of economic profit, as exemplified by a recent legal challenge to the USA endangered species act (Greenhouse 1997). Therefore, given that world-views are important determinants of null hypotheses selection, and that world-views of NGOs' scientists, like those of anyone, can be modified by prospects of economic profit, the appropriateness of null hypotheses should be thoroughly scrutinized.

To illustrate the paramount importance of null hypothesis selection, let us consider once more the EAAB project of restoring the wetlands of Bogotá. This project involves removal of extensive tracts of aquatic vegetation occupied by endemic and endangered bird species (Stiles 2003). What should be the appropriate null hypothesis regarding the impact of the EAAB project on these avian species? We argue that based on the endemicity and endangered status of these bird species, their patterns of habitat use, the identification of habitat destruction as the main threat for their survival (Asociación Bogotana de Ornitología 2000, BirdLife International 2000), population ecology theory (Soulé 1987), and the precautionary principle for conservation biology, it should be: "the EAAB project for the Bogotá wetlands will reduce the population size of endemic and endangered bird species through reduction of suitable habitat and, thus, will increase their probability of extinction". Specific predictions can be derived from this null hypothesis to produce statistical null hypotheses (e.g., Fig. 1, top panel), which are distinct from, but depend on, theoretical hypotheses (Krebs 2000). Such predictions can be examined statistically through equivalence testing, a procedure that is well established for environmental studies (McBride 1999, Manly 2001) and for the kind of biomedical research to which da Fonseca draws parallels (West-lake 1973, Metzler 1974).

In sharp contrast to the null hypothesis we proposed above, defenders of the EAAB project published the following statement in a major Colombian newspaper: "The concern of environmentalists is that building bicycle paths and walkways destroys bullrushes, native forests and shrubs where many species dwell, especially birds. The truth is that there is no study demostrating damage or loss of species." (Cabrera-Puentes and Ortega 2003). Such statement amounts to the following null hypotheses: "the EAAB project of altering the wetlands of Bogotá will not reduce the population size of endemic and endangered bird species through reduction of suitable habitat". Anyone using this null hypothesis will assume that the EAAB project does not affect the endangered bird species of the wetlands despite extensive removal of their habitat, unless sufficient evidence is compiled to demonstrate it. Thus, regardless of general knowledge of the conservation status and habits of these birds, lack of detailed data on population dynamics would result in neglecting an obvious threat to these species. The sharp contrast between this approach and the null hypothesis that we proposed above is shown in Figure 1. The null hypothesis advocated by defenders of the EAAB project is analogous to that used in criminal cases, where defendants are presumed to be innocent until proven guilty and, thus, it is appealing to many, especially those that see great benefits in the creation of large scenic water bodies deprived of vegetation, appropriate for recreational activities.

Figure 1. Null and alternate hypotheses about the effect of removing aquatic vegetation on the endemic and endangered bird species of the Bogotá wetlands. Both hypotheses are represented by a distribution of means (Zar 1999, page 76) from random samples that estimate the difference in bird density between wetland areas where aquatic vegetation was removed and wetland areas were it was not. Top panel: The null hypothesis, depicted with a continuous line, states that removal of aquatic vegetation does cause a population decline higher than or equal to the critical threshold E at which extinction probability increases. The alternate hypothesis, represented with a discontinuous line, assumes that removal of aquatic vegetation causes no population reduction. Bottom panel: The null hypothesis, depicted with a continuous line, assumes that removal of aquatic vegetation causes no population reduction. The alternate hypothesis, represented with a discontinuous line, states that removal of aquatic vegetation causes a population decline higher than or equal to the critical threshold E at which extinction probability increases. Note that in the top panel the probability of failing to detect a negative effect on the environment is a type I error. In contrast, in the bottom panel, the probability of failing to detect a negative effect on the environment is determined by type II error.

We welcome the increasing production of conservation science by NGOs. Furthermore, we believe that conservation science produced by universities and other academic institutions is subject to much the same problems associated with NGOs. Yet, the conservation community should be well aware of what exactly these problems are in order to place effective safeguards in place. Peer review and collaboration between NGOs and academic institutions are likely to increase the quality and effectiveness of conservation research, as indicated by da Fonseca. We suggest that reviewers should watch carefully for appropriate justification of particular null hypotheses. In cases in which current knowledge does not clearly justify the selection of a particular null hypothesis, several should be employed (e.g., MacKenzie and Kendall 2002), and results interpreted in light of them all. However, as indicated by da Fonseca, many studies that have major influence on governments and funding agencies are not subject to peer review. How should the quality of such unpublished but important research be controlled? It would seem that review by contestant NGOs is an equivalent mechanism to that employed by peer reviewed scientific journals, and should channel the competitive relationships among conservation NGOs towards increasing the quality of the science they produce, rather than towards aligning NGOs and corporations with associated departures from NGOs' original values and beliefs.

Literature Cited

Asociación Bogotana de Ornitología. 2000. Aves de la Sabana de Bogotá, guía de campo. ABO and CAR, Bogotá, Colombia.

BirdLife International. 2000. Threatened birds of the world. Barcelona and Cambridge, UK. Lynx Edicions and BirdLife International.

Cabrera-Puentes, M. and M. Ortega. 2003. El renacer de los humedales. El Tiempo March 3, Sección Medio Ambiente.

CI-Colombia. 2000. Estrategia para la recuperación de los humedales Bogotanos. Santa Fé de Bogotá D.C. (

Da Fonseca, G. A. B. 2003. Conservation Science and NGOs. Conservation Biology 17: 345-347.

Fjeldsa, J. 1985. Origin, evolution, and status of the avifauna of Andean wetlands. Ornithological Monographs 36: 85-112.

Greenhouse, L. 1997. Court ruling expands lists of species act challengers: Economic interest allowed as basis for suits. New York Times March 20, 146: A11, B10.

Krebs, C. J. 2000. Hypothesis testing in ecology. Pp. 1-14 in: Research techniques in animal ecology: controversies and consequences. Boitani, L. and T. K. Fuller eds.. Columbia University Press, NY.

Manly, B. F. J. 2001. Statistics for environmental science and management. Chapman and Hall/CRC, Boca Raton, Florida, USA.

MacKenzie, D. I. and W. L. Kendall. 2002. How should detection probability be incorporated into estimates of relative abundance? Ecology 83: 2387-2394.

McBride, G. B. 1999. Equivalence tests can enhance environmental science and management. Australian and New Zealand Journal of Statistics 41: 19-29.

Metzler, C. M. 1974. Bioavailability - a problem of equivalence. Biometrics 30: 309-317.

Shrader-Frechette, K. S., and E. D. McCoy. 1993. Method in ecology: Strategies for conservation. London: Cambridge University Press.

Smallwood, K. S. 2002. Habitat models based on numerical comparisons. Pp. 83-96 in: Predicting species occurrences: issues of accuracy and scale. Scott, J. M., Heglund, P. J., Samson, F., Haufler, J., Morrison, M., Raphael, M., Wall, B. eds. Island Press, Covelo, CA.

Soulé, M. E. (ed.). 1987. Viable populations for conservation. Cambridge University Press, Cambridge, UK.

Stiles, F. G. 2003. Un humedal sin aire. Edición del 16 de febrero, Periódico Universidad Nacional de Colombia, Bogotá, Colombia.

United States Environmental protection Agency. 1989. Methods for evaluating attainment of cleanup standards. Volume 1: soils and soil media. EPA report 230/02-89-042, Office of Policy, Planning and Evaluation, Washington DC.

Westlake, W. J. 1973. Use of statistical methods in evaluation of in vivo performance of dosage forms. Journal of Pharmaceutical Sciences 62: 1579-1589.