Correct estimation of preferential chromosome pairing in autotetraploids

In recent work, a statistical model was proposed for the purpose of estimating parameters associated with quantitative trait locus (QTL) mapping and preferential pairing within a polyploidy framework. The statistical model contained several parameters that, when estimated from experimental data, supplied information about QTL, including a preferential pairing factor. Among the results reported were estimates of preferential pairing, many of which indicated high levels of preferential pairing (p = 0.60) that were inconsistent with biological expectations. By using the biological inconsistencies as our motivation, we present a reformulated statistical method for estimating preferential pairing, and use this method to reanalyze the same autotetraploid alfalfa data and to conduct a simulation study. Our results directly contradict the current findings of significant preferential pairing and affirm the traditional view of random chromosome segregation in alfalfa.