RESUMO
Variability in climatic conditions of low-latitude tropical grass cultivation can affect forage production dynamics. Pasture ecosystems are complex and preferably studied from a multifactorial point of view through multivariate approaches. Therefore, in this study, we characterized different growing conditions for grasses of the Megathyrsus genus through studies conducted in tropical humid and semi-arid conditions. We applied principal component, canonical correlation, and discriminant function analyses to the measurements of agronomic and agrometeorological variables in six studies with Guinea and Massai grasses. The principal component analysis, through the climatic characterization by the first principal component, reflects the contrast between water availability and nitrogen variables and energy supply. Agronomic characterization occurred through the distinction between the density of tillers, forage accumulation, and increase in height, versus the accumulation of stems and dead material. The canonical correlation analysis generated a correlation coefficient of 0.84 between the agronomic and agrometeorological variables. There was a contrast between the dead material accumulation and the other agronomic variables, while the agrometeorological variables showed characteristics similar to the first principal component. Discriminant function 1, with 70.36% separation power, distinguished the cultivation conditions based on the study locations. Grass cultivars were differentiated by discriminant function 2, with a 19.20% separation power. From a multivariate variability analysis, despite the similarities of radiation and temperature in the regions studied, the availability of water and nutrients and measurements of agronomic variables can aid in future modeling studies on forage production.
