This study addresses the modeling of two sets of longitudinal biological data with complementary statistical challenges. The first study, which we will call Study I, analyzes the temporal dynamics of the vaginal microbiota in pregnant and non-pregnant women using negative binomial mixed models (NBMMs) and their zero-inflated extension (ZINBMMs), structures that allow for the simultaneous handling of overdispersion, zero inflation, and within-subject correlation. The second study, which we will call Study II, evaluates the growth of Quillaja saponaria in a nursery based on seven successive neck diameter (CD) measurements, using real data obtained in collaboration with the academic team of the School of Environmental Engineering at the University of Valparaíso. These data were collected within the framework of an applied research project at the National Botanical Garden of Viña del Mar. Various hierarchical linear mixed models (LMMs) with random effects and accumulated environmental covariates are compared, with the aim of accurately describing the interindividual variability in seedling growth. In both cases, a reproducible workflow was followed, including data cleaning, graphical exploration, fitting in R, and selection using AIC, BIC, and likelihood ratio tests. The results confirm that NBMM/ZINBMM models adequately capture bacterial abundance patterns associated with gestational stage, while LMM models with random slopes and cumulative temperature more accurately describe inter-plant variation. This work provides (i) a unified methodological guide for selecting mixed models based on the nature of the response, (ii) reproducible R code that integrates specialized packages, and (iii) recommendations on temporal design and zero treatment for future studies of plant microbiota and growth.