Biological and application-oriented factors influencing plant disease suppression by biological control: a meta-analytical review.

ABSTRACT Studies to evaluate the effectiveness of biological control in suppressing plant disease often report inconsistent results, highlighting the need to identify general factors that influence the success or failure of biological control in plant pathology. We conducted a quantitative synthesis of previously published research by applying meta-analysis to determine the overall effectiveness of biocontrol in relation to biological and application-oriented factors. For each of 149 entries (antagonist-disease combinations) from 53 reports published in Biological & Cultural Tests between 2000 and 2005, an effect size was calculated as the difference in disease intensity expressed in standard deviation units between the biocontrol treatment and its corresponding untreated control. Effect sizes ranged from -1.15 (i.e., disease strongly enhanced by application of the biocontrol agent) to 4.83 (strong disease suppression by the antagonist) with an overall weighted mean of 0.62, indicating moderate effectiveness on average. There were no significant (P >0.05) differences in effect sizes between entries from studies carried out in the greenhouse versus the field, between those involving soilborne versus aerial diseases, or among those carried out in conditions of low, medium, or high disease pressure (expressed relative to the disease intensity in the untreated control). However, effect sizes were greater on annual than on perennial crops, regardless of whether the analysis was carried out for all entries (P = 0.0268) or for those involving only soilborne diseases (P = 0.0343). Effect sizes were not significantly different for entries utilizing fungal versus bacterial biocontrol agents or for those targeting fungal versus bacterial pathogens. However, entries that used r-selected biological control agents (i.e., those having short generation times and producing large numbers of short-lived offspring) were more effective than those that applied antagonists that were not r-selected (P = 0.0312). Interestingly, effect sizes for entries that used Bacillus spp. as biological control agents were lower than for those that applied other antagonists (P = 0.0046 for all entries and P = 0.0114 for soilborne diseases). When only aerial diseases were considered, mean effect size was greater for entries that received one or two sprays than for those that received more than eight sprays of the biocontrol agent (P = 0.0002). This counterintuitive result may indicate that investigators often attempt unsuccessfully to compensate for anticipated poor performance in antagonist-disease combinations by making more applications.