Uso de metawin como herramienta de análisi de metadatos proveniente de más de 60 artículos relacionadas con plagas de cultivos en sistemas agroecológicos Predictive theory on how plant diversity promotes herbivore suppressionthrough movement patterns, host associations, and predation promises a potential alternativeto pesticide-intensive monoculture crop production. We used meta-analysis on 552experiments in 45 articles published over the last 10 years to test if plant diversificationschemes reduce herbivores and/or increase the natural enemies of herbivores as predicted byassociational resistance hypotheses, the enemies hypothesis, and attraction and repellencymodel applications in agriculture. We found extensive support for these models withintercropping schemes, inclusion of flowering plants, and use of plants that repel herbivores orattract them away from the crop. Overall, herbivore suppression, enemy enhancement, andcrop damage suppression effects were significantly stronger on diversified crops than on cropswith none or fewer associated plant species. However, a relatively small, but significantlynegative, mean effect size for crop yield indicated that pest-suppressive diversification schemesinterfered with production, in part because of reducing densities of the main crop by replacingit with intercrops or non-crop plants. This first use of meta-analysis to evaluate the effects ofdiversification schemes, a potentially more powerful tool than tallies of significant positive andnegative outcomes (vote-counting), revealed stronger overall effects on all parametersmeasured compared to previous reviews. Our analysis of the same articles used in a recentreview facilitates comparisons of vote-counting and meta-analysis, and shows thatpronounced results of the meta-analysis are not well explained by a reduction in articlesthat met its stricter criteria. Rather, compared to outcome counts, effect sizes were rarelyneutral (equal to zero), and a mean effect size value for mixed outcomes could be calculated.Problematic statistical properties of vote-counting were avoided with meta-analysis, thusproviding a more precise test of the hypotheses. The unambiguous and encouraging resultsfrom this meta-analysis of previous research should motivate ecologists to conduct moremechanistic experiments to improve the odds of designing effective crop diversificationschemes for improved pest regulation and enhanced crop yield