Principal components in the study of soil and plant properties in precision coffee farming
¹Federal University of Lavras – UFLA, Departament of Agricultural Engineering, University Campus, BR37200-000 Lavras-MG, Brazil
²Rural Federal University of Rio de Janeiro – UFRRJ, BR-465, Km 7, BR 23.897-000 Seropédica- RJ, Brazil
³Department of Agriculture, Food, Environment and Forestry (DAGRI), Università degli Studi di Firenze, Via San Bonaventura, 13, IT50145 Firenze, Itália
*Correspondence: gabriel.ferraz@ufla.br
Abstract:
In this work, a principal component analysis was performed to evaluate the possibility of discarding obsolete soil and plant variables in a coffee field to eliminate redundant and difficult-to-measure information in precision coffee farming. This work was conducted at Brejão Farm in Três Pontas, Minas Gerais, Brazil, in a coffee field planted with 22 ha of Topázio cultivar. The evaluated variables were the yield, plant height, crown diameter, fruit maturation index, degree of fruit maturation, leafing, soil pH, available phosphorus (P), remaining phosphorus (Prem), available potassium (K), exchangeable calcium (Ca2+), exchangeable magnesium (Mg2+), exchangeable acidity (Al3+), potential acidity (H + Al), aluminium saturation (N(Al)), potential CEC (CECp), actual CEC (CECa), sum of bases (SB), base saturation (BS) and organic matter (OM). The data were evaluated by a principal component analysis, which generated 20 components. Of these, 7 representing 88.98% of the data variation were chosen. The variables were discarded based on the preservation of the variables with the greatest coefficients in absolute values corresponding to the first component, followed by the variable with the second highest absolute value corresponding to the second principal component. Based on the results, the variables V, OM, fruit maturity index, plant height, yield, leafing and P were selected. The other variables were discarded.
Key words:
coffee plant, fertility, management, Multivariate analysis, precision agriculture