The design was completely random for soil variables;
for growth, flowering and fruits yield variables was
used a split-plot in completely random, being the mango
cultivars the large plot, and the fertilizer sources the
small plot, with three repetitions, considering one tree
as an experimental unit.