“…The Julian day of occurrence of the first attacks was evaluated by means of linear mixed-effects models ( Burnham and Anderson, 2002 ; Rondoni et al., 2012 ), excluding from the final analysis the data from 2017, due to the limited number of infestation outbreaks. For both types of models, the relevance of the fixed and random structure was evaluated by means of the likelihood ratio test (LRT) and Akaike information criteria (AIC) ( Pinheiro and Bates, 2006 ; Jacobs et al., 2022 ; Tidau et al., 2023 ). The best-fitted models, i.e., retaining the minimum number of explanatory variables, were selected using LRT ( Burnham and Anderson, 2002 ; Ferracini et al., 2023 ).…”