Analysis and modeling of time-dose-mortality of Melanoplus sanguinipes, Locusta migratoria migratorioides, and Schistocerca gregaria (Orthoptera: Acrididae) from Beauveria, Metarhizium, and Paecilomyces isolates from Madagascar.

Abstract A complementary log–log (CLL) model was used to model time–dose–mortality relationships from bioassay tests of 26 fungal isolates mostly from Madagascar, Africa, against three acridid species, all referred to here as “grasshoppers.” The fungal pathogens included 15 isolates of Beauveria bassiana, 9 isolates of Metarhizium flavoviridae, and 2 isolates of Paecilomyces spp. Grasshopper species tested included Melanoplus sanguinipes, Locusta migratoria migratorioides, and Schistocerca gregaria. The scaled deviance, mean deviance, Pearson X 2 statistic, Hosmer–Lemeshow (H–L) C statistic, and three-dimensional time–dose–mortality graphs were used to assess extra-binomial variation, data points that were potential outliers, conformance of the scaled deviance statistic and Pearson's X 2 statistic to a χ 2 distribution, and the fit of the CLL model. The H–L C statistic also was found to be useful in showing the goodness of fit of the CLL model for the fungal isolates prior to modeling the extra-binomial variation. After the extra-binomial variation was modeled using Williams' method, the slope from maximum likelihood estimation, modified log(LD 50 ) estimates (which were corrected for background mortality using the CLL model), a dynamic ranking of the log(LD 50 ) values over time, and a three-dimensional plot of time, dose, and mortality of the three grasshopper species were used to evaluate the effectiveness of the fungal isolates. In general, the CLL model provided a rather poor fit of the fungal isolates which had a large number of replicate trials in the bioassay tests (i.e., a large sample size) due to extra-binomial variation. The CLL model provided an excellent fit of the time–dose–mortality relationships of such isolates after the extra-binomial variation was modeled and included in the CLL model. Metarhizium isolates MFV and SP5 were found to be the most virulent isolates tested against M. sanguinipes, followed by Metarhizium isolates: SP8, SP7, SP9, SP6, and SP1, and Beauveria isolate S33B. Metarhizium isolates SP3, SP5, SP6, and SP9, and Beauveria isolates SP11, SP12, SP13, and SP16 showed higher levels of virulence against L. migratoria migratorioides over more of the time periods tested than the other pathogen isolates examined. Metarhizium isolates SP9 and SP5 were the most effective isolates tested against S. gregaria. In general, the Metarhizium isolates were more virulent against the grasshoppers than the Beauveria isolates, which were more virulent than the Paecilomyces isolates. The CLL model was found to be very useful in describing grasshopper mortality as a function of time and dose. This approach combined with model and fungal isolate assessment statistics will be helpful for determining which pathogen isolates have the greatest potential for controlling grasshoppers and other pests in the future.