Grid and quadrants size selection to determine crop density and spatial variability are still very controversial, either when accessing isolated sampling factors, either, combining both. Thus, this work aimed to evaluate the ideal sampling grid and quadrant size to determine the crop density and spatial variability of African pumpkin (Momordica balsamina. L.). The trial was conducted in Pambara, Vilankulo district, Mozambique, based on the factorial scheme, on a completely randomized design, comprising four levels of each factor: grid: 10 x 10 m 2, 20 x 20 m 2, 30 x 30 m 2, 40 x 40 m 2; quadrant: 1 x 1 m 2, 0.75 x 0.75 m 2, 0.5 x 0.5 m 2, 0.25 x 0.25 m 2, comprising 16 treatments and 928 georeferenced samples. To evaluate the trial, crop density data were collected, and then submitted to Normality test, Analysis of Variance, mean test at 5% significance, Geostatistics and multivariate. The results showed that the combination of the grid 20 x 20 m 2 with the quadrant 0.25 x 0.25 m 2 was ideal for determining the crop density. The principal component analysis showed that only two components contain 66,5% of the total information.

Keywords: Precision agriculture; cluster analysis; Geostatistics.