The Problem of Evidence in Ecology

The good news is that the general public are becoming more concerned about the problems of wildlife management and conservation in general. The bad news arising from this interest is the lack of understanding exhibited by many of the comments in the media about ecological problems. This leads to a suggestion that we need an ecological “fact checking” team that looks at what is said about broad scale environmental issues and points out how much evidence there is for what is stated in the media. My interest in this issue is driven by so many news stories that are stated as fact with very little scientific understanding. Too many well-meaning reports fly around the media that border on complete error or complete nonsense. One consequence of this problem is a failure of evidence-based decision making for ecological problems (Christie et al. 2022).

This is not of course a problem confined to ecological science as you can see by reading nonsense claims about medical issues like Covid. It will not go away and with the climate crisis the number of ‘experts’ has multiplied. The problem comes down to the issue of evidence and how we evaluate evidence. A partial solution to this is better education about what is evidence in ecology as well as all of science. We need to teach workshops or courses on concrete examples of what is suggested to be evidence in ecological papers. The first step might be to analyse one or a few papers with the following procedure:

  1. What is the major conclusion of the paper?
  2. What data are presented to reach this conclusion?
  3. What background assumptions are being made to move from data to conclusions?

These questions lead us back to basic questions illustrated well by statistical inference. What is the ‘population’ to which the major conclusions apply? There is very little discussion of this in most ecological papers and the consequence can be overgeneralizations. Suppose for example we are examining the hypothesis that the geographic range of a species set is moving toward the poles because of a warming climate. We must for practical purposes restrict our study to a small set of species, so this is a major assumption that the species selected are a random sample of the biota under discussion. Another limitation is that it may be difficult to isolate climate change without considering for example human disturbances to the landscape from forestry and agriculture. A consequence of these complications is that our major conclusion for all this research rests on minimal data. So, a conclusion might be that we need to design further extensive studies. But perhaps of the 6 species under study, 4 are moving as the climate hypothesis predicts, but one is not moving at all, and one is moving in the opposite direction to what is predicted. Do we now turn our attention to these anomalous species that do not follow our major hypothesis? Or should we be happy that most of our candidate species follow the rule specified in our major conclusion?

       By doing manipulative experiments ecologists attempt to insert more rigor into their conclusions, but many of the generic questions mentioned above apply equally to these experimental designs. If we do a set of experiments in Iowa and in Germany, should we get the same results? We are back to the question of generality in all our studies. We hope for global rules, but experiments are all limited in time and space.

Can we escape all these bottlenecks with models that capture the generality and behave according to our assumptions? But models suffer from the same problems that make empirical studies difficult – what are the hidden assumptions? Taper et al. (2021) discuss the problem of errors arising from model misspecification in evaluating empirical data. Perhaps every ecological publication should end with an additional short section listing the assumptions made in reaching the major conclusions of the research.

These points come to the fore when we attempt to predict future environmental changes. A simple example is the hypothesis that, by humans increasing CO2 in the atmosphere, plants will increase photosynthesis and thus negate part or all the effects of climate change on our current ecosystems. This has caused much discussion ranging from planting more trees to alleviate climate change to relying on engineering solutions to climate change.

The bottom line that we should all recognize is that our predictions in ecology and our understanding of ecosystem changes are more limited than we admit. We know that we cannot rely on the old adage of the equilibrium hypothesis that “Mother Nature will take care of the earth” so all will be well. Wisdom always relies on critical evaluations which are too often lost in the media of our current world.

An important alternative approach is illustrated by the Conservation Evidence Journal and the approaches recommended by Sutherland et al. (2022) to specify local actions that can improve the conservation status of particular species or groups of species, for example by reintroducing birds to islands or areas from which they have been extirpated. The dichotomy here is a divide between the particular and the general, from short-term local questions to long-term general questions (Saunders et al. 2020). The hope is that progress on local questions will gradually inform the dominant global theories of ecology to bring them together and support the “devil in the details’ approach that can define ecological progress in our time (Sutherland et al. 2021).

Christie, A.P., et al. (2022) Principles for the production of evidence-based guidance for conservation actions. Conservation Science and Practice, 4, e579.doi: 10.1111/csp2.12663 .

Saunders, M.E., Janes, J.K. & O’Hanlon, J.C. (2020) Moving on from the Insect Apocalypse Narrative: Engaging with Evidence-Based Insect Conservation. BioScience, 70, 80-89.doi: 10.1093/biosci/biz143.

Sutherland, W.J., Downey, H., Frick, W.F., Tinsley-Marshall, P. & McPherson, T. (2021) Planning practical evidence-based decision making in conservation within time constraints: the Strategic Evidence Assessment Framework. Journal for Nature Conservation, 60, 125975.doi: 10.1016/j.jnc.2021.125975.

Sutherland, W.J. et al. (2022) Creating testable questions in practical conservation: a process and 100 questions. Conservation Evidence Journal, 19, 1-7.doi: 10.52201/CEJ19XIFF2753.

Taper, M., Lele, S., Ponciano, J., Dennis, B. & Jerde, C. (2021) Assessing the global and local uncertainty of scientific evidence in the presence of model misspecification Frontiers in Ecology and Evolution, 9, 679155.doi: 10.3389/fevo.2021.679155.

Leave a Reply

Your email address will not be published. Required fields are marked *