Regression tree for prediction the yield of fresh matter of teosinte aerial part in function of meteorological variables

Abstract

The objective of this work was to verify if it is possible to predict the yield of fresh matter of teosinte aerial part in function of meteorological variables. An experiment was conducted with nine sowing dates (10/08/2021 to 01/01/2022). Sowings were carried out in a 5 m long row, spaced 0.80 m between rows and 0.20 m between plants in the row. In each row, five plants were randomly selected, totaling 45 plants. On 03/28/2022, in the first five sowing dates, and on 04/22/2022, in the last four sowing dates, the fresh matter of the aerial part of the plant (FMAP, in g) was determined. The cumulative global solar radiation and the thermal sum of the subperiods from sowing to male flowering (vegetative stage) and male flowering to harvest (reproductive stage) were calculated. The regression tree analysis algorithm was applied to forecast the FMAP as a function of meteorological variables. The regression tree was performed with data from all plants and sowing dates (n=45). The cumulative global solar radiation from male flowering to harvest was the main dividing point. In the second hierarchical node, the division criteria were thermal sum from sowing to male flowering and cumulative global solar radiation from sowing to male flowering. The highest productive performance (FMAP = 1162 g plant^-1) was observed in plants with cumulative global solar radiation in the reproductive stage lower than 494 MJ m^-2 and global accumulated solar radiation in the vegetative stage lower than 3257 MJ m^-2 (33% of observations).