The objective of our research was to evaluate distance between the stigma and the stamens, opening diameter and stamens distribution on the flower. We evaluated these variables on 7 flowers of each one of the 15 plant accessions from the Germplasm Bank. The flowers were classified according to distance between the stigma and the stamens into three classes: 0.2 (C1), 0.7 (C2) and 1,2 (C3) cm, in which 2, 9, and 4 accessions were included, respectively. The radial and random stamens distributions were verified in 6 and 9 accessions, respectively. These two variables behaved independently. Thereafter one plant were randomly taken from each class and marked 150 flowers, being 50 for each of the following treatments: T1 -- natural pollination; T2 -- screen which excluded only the birds from reaching the flowers; and T3 -- screen which excluded both birds and insects. It was evaluated the fruit set in all three treatments. We verified 47% of fruit set in T1 of C2 and 22% in T1 of C3. This difference can be due to genotypic differences and/or alternation of yield. In T2, fruit set was 31% in C2, 15.4% in C1, and 3% in C3. The superiority in fruit set of T1 over T2 can be credited to bird pollination. The fruit set in T1 was 7 times larger in C3 and 2 times larger in C2, than in T2. This suggests that C3 is more dependent on birds for pollination. The observed fruit set in T2 can be partially credited to insect pollination and was smaller in C3, as expected due to the larger pistils. Even in T3 we observed fruit set in C1 (7%) and in C2 (15%), which suggests that either wind plays a role in the pollination or that selfing may occur. Since there was no fruit set in T3 on C3, presence of autoincompatibility cannot be discarded.
Myrtaceae; Acca sellowiana; pineapple-guava; floral morphology
