Statistical analysis
Plant traits. For leaf thickness and herbivory rate, we performed a generalized linear model (GLM) with Poisson distribution and link: log due to our data not meeting the assumption of a normal distribution even after transformation. In the case of the leaf water content, we performed a GLM with normal distribution. In the three cases, the effect of sex (male and female), time, and their interaction were included in the model as fixed factors. Finally, for sex-ratio, frequencies ofB. cordata were compared to a null model (1:1 male-female ratio) by using a 𝛘2 test.
Arthropod traits. Abundance was calculated by tree and by time; however, our data did not meet the assumptions of normality, then we performed a GLM with a Poisson distribution and a log-link function. We treated the effect of sex (male and female), time, and their interaction as fixed factors.
For the diversity parameters, we first estimated sampling completeness estimates using the iNext website (Hsieh et al. 2016). We evaluated the effect of plant sex on the arthropod community. To evaluate the richness, diversity, and evenness of the community of adult arthropods between plant sex (male or female), we used the Hill number (0D , 1D , and2D ). The diversity of order 0 (0D ) correspond to the richness of observed species, order 1 (1D ) corresponds to the exponential of the Shannon-Wiener entropy index (H ’), and order 2 (2D ) is the diversity of the most dominant species, which is equivalent to the inverse of the Simpson dominance index (Jost 2006, Sanjuan-Trejo et al. 2021). Similarity among plants and sex was estimated by using Jaccard’s coefficient indices. We assigned functional groups by categorizing the species into guilds (herbivores, carnivores, and detritivores) based on literature and field observations. We used a t -test to compare the richness of species in the guilds per plant sex per collection time, but in case our data did not meet the assumptions of normal distribution we performed a Wilcoxon rank test.