Statistical Analysis

The data from the pathogenicity assay were processed with primary data validation performed with a logistic regression (logit link function) to identify whether the inoculum position was a determining factor for necrotic lesion expansion (event 1) or not (event 0). This strategy was adopted to verify if putative differences in the ripeness stages between the inoculation points could affect the disease progress. The linear equation used in the logistic fit included a full model containing elapsed time, Phytophthora isolates, cacao clones, inoculum position and all interactions among these variables.

Given the multifactorial nature of the experimental design growth kinetics, data from the cultural characterization were fitted through linear mixed-effects models. Two competing models were tested: the null model using the elapsed time, culture media and Phytophthora species as fixed effects, and the isolate level as a random effect (Average mycelial growth = time: Phytophthora species: cult. media) + (1 | Phytophthora isolate); and the alternative model, which differed from the null model because it included a Phytophthora isolate as a fixed effect instead of an effect at the species level. Both models were compared to verify whether the inclusion of isolate level variation (alternative model) produced a less erroneous model. Pairwise comparisons were performed with adjusted Tukey’s HSD multiplicity test. The statistical analyses were performed in R v 3.5 (Pinheiro et al., 2020) and data is publicly available on GitHub⁴.