The Influence of Missing Hills on Potato Yield and Evaluation of Correction Methods

Abstract Missing hills affect the yield of a single plant in experimental plots, as the remaining plants have more space and resources compared to plants in plots without missing hills. However, up to date, no universal way on how to correct for missing hills exists in the context of potato breeding programs. The aim of our study was to (i) compare different approaches to adjust the yield when hills are missing in a plot, (ii) propose a universal approach for correcting missing hills in potato plots, and (iii) test if the estimated parameters are transferable from a training population to an independent validation population. We evaluated in a set of 1066 different clones approaches to adjust single plant yield for the proportion of missing hills in a plot. These approaches were a grid search, a linear mixed effects model, or approaches previously proposed in the literature. We compared these approaches based on the heritability of single plant yield. Our results showed that all advanced correction approaches improved the heritability and reduced the residuals compared to the standard approaches, where the yield of the plot was just divided by the actual plant number or the intended plant number. The most promising results were observed for the four-parameter Weibull type 2 regression of the effects for each proportion of missing hills derived by a linear mixed model analysis with the proportion of missing hills used as a cofactor. The approach derived here is transferable across populations and increases heritability of yield estimates without new parameter adaptation. Therefore, our approach provides an easily applicable procedure to improve the comparison of single plant yield between clones in potato breeding programs.