BES_conference_talk:From pedigrees, through divorce, to microbes and CO2: how can fast growing data-landscape help ecological and evolutionary synthesis? .pptx

Meta-analysis is one of the main drivers of progress in ecology and evolutionary biology: it summarises the current knowledge on the topic and it allows for identification of methodological flaws, knowledge gaps, and areas of further research or intervention. Meta-analysis is traditionally conducted on the set of published (primary) studies. As such, it is sensitive to the publication bias and the quality of primary studies. Open Science is increasing the number and the quality of datasets available to incorporate into meta-analysis. These datasets can be used to verify (or supplement) the results of primary studies, or as data-points themselves. Furthermore, different types of datasets (sometimes from different disciplines) can be combined. However, this great opportunity has not yet been fully and equally seized by the ecological and evolutionary disciplines. We evaluate the current state and the potential benefits (and issues) of open research practices to meta-analysis by using several case studies from a spectrum of subdisciplines within ecology and evolutionary biology: animal evolutionary ecology, microbial ecology, soil carbon ecology, climate change, and aquatic ecology. Our case studies use two different approaches to constructing the set of effect sizes for the meta-analysis: one searches for published studies, and the other one for datasets. We found that search efficiency, and the completeness and the quality of the data-landscape differ between the subdisciplines. We also find that we would need to give up on some of our meta-analysis had we used the published studies only. Finally, we estimate the value of the datasets submitted with studies to the quality of the meta-analysis. We hope this work will encourage better utilisation of the dynamic, heterogeneous, and fast-growing data-landscape within ecology and evolutionary biology, and lead to better meta-analysis for improved decisions based on previous research, and for better integration of research outputs from different disciplines.