2.4. Analysis

Nut damage by emerging C. sayi larvae was visualized using histograms and density distributions then modeled using generalized additive models.

Mortality of C. sayi larvae resulting from nematode infection was evaluated using logistic regression considering the interaction between nematode strain and days post inoculation as factors. Best fit models were chosen after consideration of all potential interactions, residual analysis, goodness of fit metrics, and likelihood ratio tests. Post-hoc treatment comparisons were conducted using the Dunnett method of contrasting treatments with the control.

The efficacy of C. sayi management strategies was evaluated using poisson regression models for trap catch and emergence from microcosms. Logistic regression was used to evaluate nut infestation from respective treatments. Best fit models were chosen after consideration of all potential interactions, residual analysis, goodness of fit metrics, and likelihood ratio tests. Post-hoc treatment comparisons were conducted using custom contrasts for desired comparisons and the Bonferroni correction for adjusting the family-wise error rate.

All analysis was conducted in R version 4.2.1 [20] using RStudio [21] as an integrated development environment (IDE). The following packages facilitated the analysis, modeling, and visualization of results: tidyverse for data preparation and plotting [22]; car and emmeans for model evaluation and post-hoc comparisons [23,24]; ggpubr, here, and cowplot for assistance in plotting and figure generation [25,26,27].