Morfofisiologia e métodos de predição de variáveis produtivas do capim-braquiária pastejado por ovinos

Abstract

Pasture is the main way of feeding herds that offer meat and milk to the world population. However, processes of monitoring this biomass have not been effective as there are no methodologies that can precisely diagnose the condition of the canopy. Because of this, there has been a lack of productivity in domestic herds and a perenniality of the utilized forage species. For that reason, this thesis aims to evaluate the physiological and structural characteristics of the species Brachiaria decumbens cv. Basilisk managed under two frequencies and two defoliation intensities. The study also applies different methodologies of pasture condition monitoring which generates and validates models that provide the described characteristics. This effort aims to achieve an application at a property level that contributes to the management process of production systems. In the first chapter of this research, physiological, structural, and productive responses of the signal grass species were characterized in four different procedures based on rates of 85% and 95% of Light Interception (LI), as well as on the rates of 1.3 and 1.8 of Residual Leaf Area Index (RLAI). A larger production of Total Forage Biomass (TFB) was observed when the canopy was managed in the lowest defoliation frequency (95% of LI), which indicated a larger Normalized Difference Vegetation Index (NDVI) of the canopy. Nevertheless, physiological variables (especially photosynthesis index) were negatively affected by the greater defoliation frequency (85% of LI). In the second chapter, different forms of quantifying the height of the canopy were used. Hence Pearson correlation analysis (association) and trail (cause/effect) with physiological, structural, and productive variables were analyzed in different (residual and pre-grazing) conditions of the canopy. Canopy height (CHeigh) methods such as Leaf Area Index (LAI), weighted height 1 (WH1), and weighted height 8 (WH8) were the most efficient in predicting TFB and leaf photosynthetic rates. Nonetheless, distinct angular coefficients and intercepts from the calibration linear models were found. In the third chapter, some calibrated, validated, simple, and multiple non-linear models were developed regarding the existential relation from the Spearman correlation and trail analysis among CH, LAI, and NDVI to predict TFB and Green Leaf Lamina Biomass (GLLB). Among the models that were used for predicting TFB and GLLB, the exponential one using NDVI as well as the potential model using LAI and height were the ones that better fitted. In the validation phase, models related to height demonstrated the greater performances: d-index of 0.9531 (TFB) and 0.9638 (GLLB); bias of -2.3 (TFB) and -7.20 (GLLB); R² of 0.8532 (TFB) and 0.8932 (GLLB). When the product system aims for a larger gain per animal, signal grass must be managed with a frequency of 85% of LI with an RLAI rate of 1.8 or NDVI rates of 0.49 and 0.85, respectively. On the other hand, if the system aims for a gain per area, the defoliation frequency must be 95% of LI associated with an RLAI rate of 1.3 or NDVI rates of 0.44 and 0.88, respectively. The CH and NDVI methods offered good efficiency in predicting TFB and GLLB. CH is recommended to smaller areas while NDVI is efficient in larger ones.