Monthly Archives: November 2018

Blaming Climate Change for Ecological Changes

The buzz word for all ecological applications for funding and for many submitted papers is climate change. Since the rate of climate change is not something ecologists can control, there are only two reasons to cite climate change as a reason to fund current ecological research. First, since change is continuous in communities and ecosystems, it would be desirable to determine how many of the observed changes might be caused by climate change. Second, it might be desirable to measure the rate of change in ecosystems, correlate these changes to some climate variable, and then use these data as a political and social tool to stimulate politicians to do something about greenhouse gas emissions. The second approach is that taken by climatologists who blame hurricanes and tornadoes on global warming. There is no experimental way to trace any particular hurricane to particular amounts of global warming, so it is easy for critics to say these are just examples of weather variation of which we have measured much over the last 150 years and paleo-ecologists have traced over tens of thousands of years using proxies from tree rings and sediment cores. If we are to use the statistical approach we need a large enough sample to argue that extreme events are becoming more frequent, and that might take 50 years by which time the argument would be made too late to request proper action.

The second approach to prediction in ecology is fraught with problems, as outlined in Berteaux et al. (2006) and Dietze (2017). The first approach has many statistical problems as well in selecting a biologically coherent model that can be tested by in a standard scientific manner. Since there are a very large number of climate variables, the possibility of spurious correlations is excessive, and the only way to avoid these kinds of results is to be predictive and to have a biological causal chain that is testable. Myers (1998) reviewed all the fishery data for predictive models of juvenile recruitment that used environmental variables as predictors and data was subsequently collected and tested with the published model. The vast majority of these aquatic models failed when retested but a few were very successful. The general problem is that model failures or successes might not be published so even this approach can be biased if only a literature survey is undertaken. The take home message from Myers (1998) was that almost none of the recruitment-environment correlations were being used in actual fishery management.

How much would this conclusion about the failure of environmental models in fishery management apply to other areas in ecology? Mouquet et al. (2014) pointed out that predictions could be classified as ‘explanatory’ or ‘anticipatory’ and that “While explanatory predictions are necessarily testable, anticipatory predictions need not be…….In summary, anticipatory predictions differ from explanatory predictions in that they do not aim at testing models and theory. They rely on the assumption that underlying hypotheses are valid while explanatory predictions are based on hypotheses to be tested. Anticipatory predictions are also not necessarily supposed to be true.” (page 1296). If we accept these distinctions, we have (I think) a major problem in that many of the predictive models put forward in the ecological literature are anticipatory, so they would be of little use to a natural resource manager who requires an explanatory model.

If we ignore this problem with anticipatory predictions, we can concentrate on explanatory predictions that are useful to managers. One major set of explanatory predictions in ecology are those associated with range changes in relation to climate change. Cahill et al. (2014) examined the conventional hypothesis that warm-edge range limits are set by biotic interactions rather than abiotic interactions. Contrary to expectations, they found in 125 studies that abiotic factors were more frequently supported as setting warm-edge range limits. Clearly a major paradigm about warm-edge range limits is of limited utility.

Explanatory predictions are not always explicit. Mauck et al. (2018) for example developed a climate model to predict reproductive success in Leach’s storm petrel on an island off New Brunswick in eastern Canada. From 56 years of hatching success they concluded that annual global mean temperature during the spring breeding season was the single most important predictor of breeding success. They considered only a few measures of temperature as predictor variables and found that a quadratic form of annual global mean temperature was the best variable to describe the changes in breeding success. The paper speculates about how global or regional mean temperature could possibly be an ecological predictor of breeding success, and no mechanisms are specified. The actual data on breeding success are not provided in the paper, even as a temporal plot. Since global temperatures were rising steadily from 1955 to 2010, any temporal trend in any population parameter that is rising would correlate with temperature records. The critical quadratic relationship in their analysis suggests that a tipping point was reached in 1988 when hatching success began to decline. Whether or not this is a biologically correct explanatory model can be determined by additional data gathered in future years. But it would be more useful to find out what the exact ecological mechanisms are.

If the ecological world is going to hell in a handbasket, and temperatures however measured are going up, we can certainly construct a plethora of models to describe the collapse of many species and the rise of others. But this is hardly progress and would appear to be anticipatory predictions of little use to advancing ecological science, as Guthery et al. (2005) pointed out long ago. Someone ought to review and evaluate the utility of AIC methods as they are currently being used in ecological and conservation science for predictions.

Berteaux, D., Humphries, M.M., Krebs, C.J., Lima, M., McAdam, A.G., Pettorelli, N., Reale, D., Saitoh, T., Tkadlec, E., Weladji, R.B., and Stenseth, N.C. (2006). Constraints to projecting the effects of climate change on mammals. Climate Research 32, 151-158. doi: 10.3354/cr032151.

Cahill, A.E., Aiello-Lammens, M.E., Fisher-Reid, M.C., Hua, X., and Karanewsky, C.J. (2014). Causes of warm-edge range limits: systematic review, proximate factors and implications for climate change. Journal of Biogeography 41, 429-442. doi: 10.1111/jbi.12231.

Dietze, M.C. (2017). Prediction in ecology: a first-principles framework. Ecological Applications 27, 2048-2060. doi: 10.1002/eap.1589.

Guthery, F.S., Brennan, L.A., Peterson, M.J., and Lusk, J.J. (2005). Information theory in wildlife science: Critique and viewpoint. Journal of Wildlife Management 69, 457-465. doi: 10.1890/04-0645.

Mauck, R.A., Dearborn, D.C., and Huntington, C.E. (2018). Annual global mean temperature explains reproductive success in a marine vertebrate from 1955 to 2010. Global Change Biology 24, 1599-1613. doi: 10.1111/gcb.13982.

Mouquet, N., Lagadeuc, Y., Devictor, V., Doyen, L., and Duputie, A. (2015). Predictive ecology in a changing world. Journal of Applied Ecology 52, 1293-1310. doi: 10.1111/1365-2664.12482.

Myers, R.A. (1998). When do environment-recruitment correlations work? Reviews in Fish Biology and Fisheries 8, 285-305. doi: 10.1023/A:1008828730759.