Can publication bias affect ecological research? A case study on soil respiration under elevated CO 2

Literature surveys are a prerequisite to enhance scientific knowledge as they allow us to separate systemic from idiosyncratic mechanisms and processes, and thus provide insight at a higher level than can be gained from individual studies. In ecological research, statistical synthesis of literature surveys, using meta-analysis, has become a powerful tool to quantify global mean responses to a changing climate (Curtis & Wang, 1998; Medlyn et al., 1999; Rustad et al., 2001; Treseder, 2004, 2008; Knorr et al., 2005; de Graaff et al., 2006; Luo et al., 2006; Janssens et al., 2010). Such analyses have greatly improved our understanding of ecosystem functioning and the parameterization of models. Unfortunately, there is also a downside to quantitative review methods such as meta-analysis, as they can easily be affected by publication bias (Møller & Jennions, 2001). Publication bias can be defined as the selective publication of articles showing certain types of results over those showing other types of results. The most commonly suspected publication bias is the tendency for authors and journals to only publish studies with statistically significant results, which has been termed the ‘file-drawer problem’ (Rosenthal, 1979). Moreover, researchers are under increasing pressure to publish frequently, and it is much easier to publish results that can easily be explained or support widely accepted hypotheses (Jarvis et al., 2001), than having to fight a time-consuming battle with conservative and suspicious referees (as they perhaps should be)