A phrase that you are bound to hear many times at any ecology conference is “it depends”. We see context dependence – variation in the magnitude or sign of ecological relationships depending on the conditions under which they are observed (Fig. 1) – in just about every study and every system. Such variation, especially when unexplained, can lead to spurious or seemingly contradictory conclusions across studies, which can limit understanding and our ability to transfer findings across studies, space, and time. Because of the wide prevalence of observed context dependence and the critical need to tackle it, a group of us recently knocked heads (and read lots of fabulous papers!) about how it can be addressed. Our reading, thinking, talking, drawing and writing culminated in this open access paper in TREE.
In the paper, we identify two types of context dependence resulting from four sources (Fig. 2). Mechanistic context dependence occurs when a relationship, say between variables X and Y, fundamentally differs under different ecological and spatiotemporal conditions. Such relationships arise from (i) interaction effects of another variable, Z, which modifies the effect of X on Y, reflecting ecological processes. Apparent context dependence occurs when the relationship between variables X and Y does not differ but appears to due to: (ii) the presence of confounding factors that are either unaccounted for or are measured and accounted for in some studies but not others; (iii) problems of statistical inference where studies differ in sampling accuracy and precision, statistical power, or interpretation of statistical measures; and (iv) methodological differences among studies whereby studies observe and measure variables or relationships in different ways.
We illustrate our typology using examples from biological invasions, a field where context dependence is prominent and widely discussed, but we propose that the typology is applicable across all areas of ecology (and it may well extend to all natural and biological sciences…). We conclude the paper by outlining steps for addressing the different types and sources of context dependence, and provide a decision tree that outlines key actions likely to be helpful. We believe that by recognising the different ways in which context dependence can arise, we can better account for context dependence and reduce the prevalence of unexplained variation in ecology.
Full paper: Catford, J.A., Wilson, J.R.U., Pyšek, P., Hulme, P.E. & Duncan, R.P. (in press) Addressing context dependence in ecology. Trends in Ecology & Evolution. link (open access)