Scale-specific spatial density dependence in parasitoids: a multi-factor meta-analysis

Summary Within a landscape, the risk of an insect being attacked by a parasitoid varies with the local density of the host species. This relationship should be strongest when observed at medium extents and resolutions with respect to parasitoids’ foraging range, and turn negative at fine resolutions. The relationship is also hypothesized to depend on certain traits of the host and parasitoid taxa – for example being more positive for more specialized hosts or parasitoids and more negative for mobile hosts or gregarious parasitoids. Building on earlier literature reviews, it is now possible to investigate these hypotheses using meta-analysis. We performed a multi-factor meta-analysis on 151 analyses of parasitism rates with respect to host densities at specified scales, from 61 empirical studies published from 1988 to 2012. We explored how the correlation between host density and parasitism rate may be related to the explanatory variables already mentioned, plus parasitoid body length and various other characteristics of both hosts and parasitoids. Correlations (Pearson's r) between host density and parasitism rate ranged from –0·88 to 0·98 (mean 0·16, standard deviation 0·39). The correlation was more often negative where the host was exotic or in the orders Lepidoptera or Diptera, where the parasitoid was larger or exotic, or where the study was conducted at a finer grain size. Hymenoptera and Homoptera were the most likely host orders to reveal positive associations, with Coleoptera and Diptera intermediate. The fact that increased observational grain size had similar effects to decreased parasitoid body length could be taken as evidence that parasitoids’ foraging ranges increase with their body length. However, the hypothesis about scale-specific foraging was not supported by studies that compared multiple scales. We conclude that parasitism most commonly produces positive (compensatory) spatial density dependence, but ecological context is all-important. These findings should help improve the design and interpretation of field experiments on parasitism as well as their application to the modelling of population dynamics and the practice of biological control.