Improving initial trials in tree breeding using kinship and breeding values estimated in the wild: the case of Prosopis alba in Argentina

Most Argentinean forests have been lost by over exploitation and expansion of agricultural areas. The National breeding program for the native species Prosopis alba is still in its initial phase with only a few progeny trials installed from material collected in the wild and the first genetic studies are underway. Breeding value (BV) estimates based on pedigree data from a progeny trial were first compared to those derived by using microsatellite based kinship estimates to confirm the potential accuracy of g-best linear unbiased predictions (G-BLUP) when pedigree information is lacking. Afterwards, the possible genetic effects of alternative selection strategies to collect improved seeds from a wild population were evaluated. To achieve this goal the relationship among average genetic gain (predicted by G-BLUP), inbreeding and sampling size of the collected materials were weighed in a wild population consisting of trees of similar ages. The results obtained suggest that kinship estimates based on molecular data and breeding value predictions BV used for the selection of elite trees in wild populations may contribute to improve the genetic properties of the founder population. Controlling kinship allows reducing sampling size from 20 to 10 individuals per origin with no significant increase of inbreeding or loss of genetic gain. For a selected group of only ten top individuals per origin, the replacement of two strongly reduces the average group coancestry with minimal gain loss. Simultaneous selection for two traits by selection index might produce a gain of near 6 % in height and 2 % in diameter. The use of molecular marker information may contribute to reduce the time needed in a P. alba improvement program.