Are modellers trying to steal your data?
Field ecologists not bothering to read your equations?
If so, you’re not alone, because the authors in Heuschele et al 2017 share your concern. They reckon that ecological research is being limited by a lack of communication and collaboration between modellers and experimentalists.
My lab QAECO is made up of a diverse bunch of researchers, who use many different approaches to skin a cat. So this week in our fortnightly reading group, I bought this paper in to see what everyone thought. Turns out, quite a bit.
The authors of this paper conducted an online survey of ecologists, a bibliometric analysis of highly cited papers, and examined the background of highly cited ecologists. In doing so, they identified two key aspects that seem to be preventing collaboration between modellers and experimentalists: journal articles being written in “cryptic” ways that make it difficult for their counterparts to decipher, as well as a lack of data being exchanged. They showed that the recipe for being a highly cited paper/author, was to model, or use a combination of experimental and modelling approaches (but not just experimental).
Despite a couple people not being convinced by their review methods (and they didn’t use any modelling… tsk tsk), these concerns did seem to ring true in our discussion. The majority of experimentalists they surveyed were keen to share their data, however, had reservations about modellers using their data inappropriately (i.e. not recognising the limitations of their methods) and wanted appropriate acknowledgement. These results prompted a debate amongst us about what constitutes authorship, should collecting the data automatically mean you are a co-author? Unsurprisingly, our modelling folks tended not to think so, and generally those who have conducted experiments did. However, we all agreed that it was context specific and that those who collected the data should at least get an opportunity to contribute to the manuscript. Regardless, it was clear that like the authors, we all agreed that collaboration and sharing data is essential to make sense of the complex systems we study. Although, this shouldn’t just be a one-way street of experimentalists handing over their data without apprehension, modellers should also seek to collaborate in the experimental design process.
But hang on a second, what the blooming heck is an experimentalist and who is a modeller? We were a little confused with their definitions (for example, they excluded statistical modelling from the definition of being a modeller, yet classified papers using statistical models of non-manipulative experiments as modelling papers in their bibliometric analysis). We also felt that this paper tended to reinforce the stereotypes of being either a modeller or an experimentalist. These labels can stick very strongly in science and can dictate what you work on in the future. In some sections, the paper seemed discouraging of using both approaches – something that many of us in QAECO strive to do.
A particular concern we found from this research was that experimentalists do not seem to be drawing inspiration from modelling papers. Surveyed experimentalists stated that these papers were difficult to understand and that they were sceptical about the model being a realistic representation of the system. A key recommendation the authors put forward was to increase mathematical teaching in ecology, which is of course, well justified and often called for. However, a few of us also thought that there is more that modellers could be doing to bridge this gap that didn’t get mentioned in the paper (and rarely does). For example, modellers could head into the field to see how this data is being collected, talk to land managers, communicate their findings directly to experimentalists that would be interested (especially if your highlighting an information gap), encourage them to test your model and collect their data in a way that would increase the modelling opportunities.
This article promotes an important discussion about how we can improve our field. We feel very lucky at QAECO to have such a wide spectrum of experimental-modelling approaches being used under one roof. It is clear that increasing collaboration and communication in ecology needs to be better encouraged, luckily there are feasible steps we can all take to bridge this gap (which are probably much easier than catching thousands of bandicoots or deriving new algorithms).
This blog post was originally posted on the QAECO website here.