Difference in frequency of traits across beach levels
We evaluated the differences in frequency of five traits related and one
trait unrelated to hydrodynamics across beach levels. For each trait, we
fitted a generalized linear mixed effect model (GLMM) using the
proportion of species that present each trait at each sample as a
response variable, the beach level as predictor, and the identity of
each beach as a random effect. We selected a binomial(logit) family
model to account for proportion data, using the function ‘cbind’. The
GLMM were run with the function ‘glmer’ in the R package ‘stats’ of the
R core team. We tested significance of models for pairwise differences
across habitats by comparing marginal means between the three beach
levels using a Tukey HSD test as implemented in the functions ‘glht’ in
the R package ‘multcomp’ v. 1.8.1-1 (Hothorn et al., 2008). The results
were then represented using box plots drawn using the R package ggplot2,
and Type II ANOVA tables from the R package car to produce output tables
from GLMM with both categorical and continuous predictors.
A reproducibility checklist for the study (Palacio et al., 2022)
including detailed information on sample size, number of traits, and
statistical procedures is available in Supplementary material
(Supplementary Table S4).