Community effects of praying mantids: a meta-analysis of the influences of species identity and experimental design

1. Generalist arthropod predators are ubiquitous in terrestrial ecosystems but experimental studies have yielded little agreement as to their effects on prey assemblages. Drawing on results from a suite of experimental field studies, a meta-analysis was conducted of the impact of praying mantids (Mantodea: Mantidae) on arthropod assemblages in order to identify predictable and unpredictable effects of these extremely generalised predators. 2. Results across different experiments were synthesised using the log response ratio framework, with a focus on quantifying net mantid impacts on arthropod density across taxonomic orders and trophic levels of arthropods, paying special attention to the contribution of mantid species identity and experimental design variables, such as the use of cages, length of experiment, and manipulated mantid density. 3. Calculated on a per mantid-day basis, the net impacts of Tenodera sinensis on arthropod density were generally weaker but more predictable than the effects of Mantis religiosa. Mantids in general had weak negative effects on density for most taxa but exhibited strong negative and positive effects on some taxa. Tenodera sinensis tended to have negative effects on Homoptera, Diptera, and Hemiptera and herbivores as a group, however M. religiosa exhibited greater variation in response of different taxa that appeared to be affected more strongly by experimental design. The effects of Stagmomantis carolina tended to be negative or non-significant. 4. Experimental cages had little influence on either the sign or magnitude of net community impacts for T. sinensis <x>, however cage experiments reversed the sign of the mean effect for two of six taxonomic orders when the experimental predator was M. religiosa. Cages also increased the variability of effect size greatly for M. religiosa but not for T. sinensis. 5. It was concluded that it is possible to use log response ratios to determine general, predictable trends in a well-studied system. Similar meta-analyses of generalist predator effects in other systems should produce predictions of how these predators influence food webs, an important step towards defining more clearly the influences of generalist predators on community structure and dynamics.