The peach palm is a very useful plant for feeding Brazilians as fruit or palm heart producer. The interest for the peach palm besides being a perennial culture is: growth in full sun, precocity, rusticity, capacity to shoot, flavor and non-darkening of the palm heart after the cut. Estimates of genetic parameters in peach palm are scarce and constitute the most important tool to guide the improvement programs. The objective of this work was to study the genetic variability and estimate the individual genetic value as selection criterion, using the BLUP/REML procedure (Best linear unbiased prediction/restricted maximum likelihood). Two selection strategies for the palm heart production trait were adopted: a short term (CP - selection of the 9 families with 31 individuals of bigger genetic value) and a long term (LP - selection of the 15 families with 53 individuals). The progenies were evaluated in randomized block design with three replications, the plots were composed by rows of five plants, spaced in 2.0 m x 1.0 m and with a row around the experiment in the Experimental Field of Matapi, Porto Grande municipality, Amapa State, Brazil. The evaluation was accomplished to the 26 months after planting (2nd evaluation) being collected data of plant height (AP), diameter of the plant to the lap height (DPC), palm heart size (TP), palm heart diameter (DP), residual apical weight (PRA), basal weight (PRB) and of the liquid palm heart (PP) (exportation type). The data of AP, DPC, TP and DP corresponded to the clump of roots averages that presented more than a stem. However for the characters PA, PRB and PP corresponded the sum of the stems in the clump of roots. In general, the population presented low genetic variability. The narrow sense heritability at the individuals level was: AP (18.44%), DPC (3.16%), TP (42.47%), DP (10.54%), PP (5.70%), PRB (6.15%). The genetic gain estimated in relation to average of the population for PP were of 7.18% in the LP situation and 8.40% for CP, with effective size of 30.38 and 19.00, respectively.
linear models; effective size; prediction of random variables; perennial species; quantitative genetics
