Assessment of factors and contexts in previous ERH syntheses
Categorising meta-analyses by the contexts and factors they examine reveals reasons for previous inconsistencies (Table S1). We examined 16 ERH meta-analyses and meta-syntheses to ascertain whether existing syntheses capture the component factors of the ERH and the extent to which they consider context. Our results revealed that individual syntheses provide only partial tests of the ERH, and many cannot be directly compared (Fig. 5; Table S1). For example, Mitchell & Power (2003) tested how enemy diversity changes after invasion, using the metric of species richness, whereas González-Browne et al. (2016) tested whether enemy impact varies between exotic and native plants, using the metric of reproductive potential (Fig. 5). Both meta-analyses provide valuable information but analyse quite different things. Several meta-analyses combine different metrics into a single effect size (and, in two cases, different factors into a single effect size: Lamarqueet al. 2011; Felker-Quinn et al . 2013; Fig. 5), likely increasing uncertainty and variance around that effect size due to underlying methodological differences, which can result in apparent context dependence (Catford et al. 2022).
Ecological context seems to have been under-explored in the 16 meta-analyses (Fig. 5). Not all contexts are relevant for all factors, and some meta-analyses implicitly account for certain contexts in their design by, for example, accounting for resource availability by only including common garden comparisons. However, explicit tests were included in just 17 of 88 cases (19%) where context could affect meta-analysis results (Fig. 5). In 15 of those 17 cases (88%), they were found to be a significant moderator of effect sizes (Fig. 5). Ecological context therefore has huge potential to explain variation both within and between meta-analyses yet is rarely accounted for.