Fruit set evaluation
The evaluation of the pollen effect on fruit set was carried out according to Socias i Company et al. (2004) and Kodad and Socias i Company (2008) during three consecutive seasons (2006–2008). For set estimates, several branches around the trees were selected and assigned randomly to the treatments. The control was open pollination and the treatments applied were self-pollination, and cross-pollination. Control branches were left intact for open pollination. On the branches used for artificial pollinations, open flowers and closed buds were removed, leaving only the flowers at stage D, which were later emasculated and self- and cross-pollinated with different cross compatible pollen. The experimental unit was the branch and the total number of branches per treatment…show more content… Fruits were counted in June, after fruit fall, and fruit set on control branches (open pollination) was determined as the number of fruit in relation to the initial bud number, and in the case of branches used for artificial pollinations fruit set was determined as the number of fruit in relation to the number of pollinated pistils. Additionally, branches were bagged to ascertain the autogamy level of each selection.
Statistical analysis inference
All the statistical analysis was performed with the SAS programme (SAS Institute, 2000). The analysis of variance with the PROC GLM procedure was applied. The mean separation was done with the LSD test at a probability of 0.05. The coefficients of phenotypical correlation between the flower bud density and fruit set from control were determined with the PROC CORR procedure.
Identification of S-allele
The consensus and specific primers allowed the determination of the S allele constitution of the studied selections (Table 1, Fig. 1), confirming the presence in all selections of the Sf allele associated with