Ecoinformatics Can Infer Causal Effects of Crop Variety on Insect Attack by Capitalizing on 'Pseudoexperiments'...

Capturing the complementary strengths of observational and experimental research methods usually requires the researcher to gather separate experimental and observational data sets. In some cases, however, commercial agricultural practices produce the spatial and temporal mixing of 'treatments' independently of other possibly covarying factors that is normally achieved only with formal experimentation. The resulting 'pseudoexperiments' can provide strong evidence for causal relationships. Here, we analyze a large observational data set that creates a series of such pseudoexperiments to assess the effect of different commercial varieties of almond, Prunus dulcis (Mill.) on the impact of two key lepidopteran pests, the navel orangeworm Amyelois transitella (Walker) (Lepidoptera: Pyralidae), and the peach twig borer Anarsia lineatella Zeller (Lepidoptera: Gelechiidae). Almonds are universally planted as polycultures of different varieties to obtain efficient cross-pollination. We find substantial differences across almond varieties in the rates of infestation of almond hulls and nutmeats by the two pests. We find no support for the hypothesis that earlier-maturing varieties sustain higher attack; for A. transitella, later-maturing varieties instead had more frequent infestation. On many almond varieties, A. lineatella reaches high infestation levels by feeding almost exclusively on the hulls, rather than nutmeats. Given the importance of these pests in directly destroying almond nuts and in promoting aflatoxin-producing Aspergillus sp. fungal infections of almonds, further work exploring the impact of these pests is warranted. Because many crops requiring cross-pollination are planted as mixtures of different varieties, commercial agricultural production data hold great potential for studying within-crop variation in susceptibility to insect attack.