Evaluating the impact of mislabeled accessions in a cacao progeny trial

The issue of offtypes (misidentified accessions) in germplasm collections of cacao (Theobroma cacao L.), presents a significant challenge to breeding and research efforts to improve generational gains in key traits like production and disease resistance. Breeding programs rely on the assumption that clonal parents have been correctly characterized and identified to develop the best families and select the best progenies, but it has been estimated that anywhere from 13-40% of global germplasm collections are contaminated with offtypes.  Molecular fingerprints from a progeny trial (PT08) at the Mars Center for Cocoa Science in Ilhéus, Brazil were used to verify the identities of 3,436 progeny from 69 crosses made according to a Circulant Factorial design and planted in 2009.  Three years of phenotypic data including yield, pod index, seed index, percent of healthy pods, and Witches Broom Disease (Moniliopthora perniciosa) was analyzed using linear mixed modeling in ASREML.  Analyses were made with and without the exclusion of offtypes to estimate their effect on selections, prediction of genetic gains, and the inferred genetic architecture of phenotypic traits of interest.  Different approaches to handling offtypes through data analysis are compared, such as use of a pedigree matrix.  Additional approaches to improving reliability and accuracy of trait estimations through modifications to the trial designs and analyses are discussed.