Efficiency of combined selection over sequential selection in forest tree progeny trials

A comparison was made between two methods of forest breeding by recurrent s election. Combined selection (CS) based on an index that combines half-sib family and individual merits and sequential selection between and within ha lf-sib families (SS) were considered. Expected efficiency of CS relative to SS was computed as the ratio of the expected gains of each method (CS/SS). For analysis, data from the literature and from a specific open pollinated progeny trial of Eucalyptus grandis (HILL) MAIDEN were taken. An average e fficiency of 1.21 (SS = 1.00) was found for the published data, concerning growth characters. For the specific progeny trial efficiency was of 1.27 fo r H (plant height) and 1.31 for DBH (diameter at breast height) and V (volu me). A procedure that allows predicting minimum efficiency for a given tria l was worked out. Both methods were also compared under equal effective pop ulation sizes (N-e). It became evident that CS is superior to SS, and this superiority is more evident for traits with small narrow sense individual h eritabilities or trials with small plot sizes (or both). Efficiency of CS o ver SS is less evident for traits in general, when effective population siz e was maintained equal in both schemes. Combined selection is recommendable in selection programs, specially if effective population sizes do not matt er.