Tables

Table 1 Hurdle model for seed set and seed number for the full dataset including pollen type and self-compatibility of recipient species as predictor variables (SI: self-incompatible). The “zero-inflated” model is binomial with a logit link function, while the “conditional model” for the non-zero counts is a truncated negative binomial model with a log link function (“nbinom2” in glmmTMB ). The variance for the binomial model is given by where is the number of trials and the probability of a zero, while the variance for the negative binomial model is given by , where the dispersion parameter allows for a larger variance. By default, the zero-inflation model coefficients give the likelihood of a zero.