In a paper published online today in Methods in Ecology and Evolution, Malcolm Nicoll and Ken Norris look at a controversial issue, that of detecting effects of predators on bird populations. This is controversial because some predators, especially raptors, were formerly rather scarce, but have become more abundant in recent years – in the case of raptors because organochloride chemicals are not used any more. At the same time mammal predators have also increased in numbers. This has led to suggestions that increases in predators may be a contributory factor to declines in some groups of birds, such as farmland birds, and there has been a great deal of discussion and debate over the issue.

Unfortunately this is a hypothesis that is not very easily addressed, unsurprising given the spatial and temporal scales that may be involved. Experimental approachs would be the ‘gold standard’, but these are difficult. Gradients of predator abundance have to be created by means of barriers or removal, which is expensive, logistically challenging and potentially expensive.

More usually opportunistic, observational evidence has to be relied upon. For example, this might take the form of statistically comparing populations in birds in areas with high numbers of predators with those with low numbers of predators. Whilst more practicable, such studies can suffer from the possibility of confounding: if a third ‘hidden’ variable also varies between sites, then this could generate problems for the interpretation.

The previous literature has presented mixed results: some studies have demonstrated effects of predators on prey populations, others have not. Whether this variation has an ecological basis is not clear.

In this new study, Nicoll & Norris re-evaluate previous observational studies by means of meta-analysis. They look at the effects of the quality of data and the number of predator species studied on the outcome of analyses. Importantly they find that the probability of detecting an effect depends on both the quality of data and the number of predators studied.

There are several implications of this work. Studies with poor quality data and that include small numbers of predators cannot reliably tell whether there are effects of predators or not, they are simply inconclusive. Nicoll & Norris go as far as to say that one should be skeptical about any short-term observational study that reports no effects of predators. They also suggest that the combined effects of predators are likely to be more important than that of any single predator, and that future studies should account for this.

Finally this study highlights the importance of methods in ecology: one cannot interpret evidence unless the method used is shown to be reliable and fit for purpose. Nicoll & Norris’ study is a good example of how re-evaluation of methods can help improve ecological understanding.