Repeatability coefficient for the determination of the optimal number of harvests for the selection of zucchini hybrids

Abstract The objective of this work was to estimate the optimal number of harvests for the reliable selection of zucchini (Cucurbita pepo) hybrids through the repeatability coefficient. The experimental design was randomized complete blocks with 33 treatments (31 experimental hybrids and 2 commercial ones) and four replicates, with six plants per plot. Fifteen harvests were carried out. Seven morphoagronomic fruit characteristics were evaluated, and repeatability coefficients were estimated using four statistical methods. The repeatability coefficients ranged from low to moderate, regardless of the studied characteristic. For a high-precision selection (R2≥90%), a high number of evaluated harvests was required, especially for traits related to fruit yield, as follows: 30 to 54 harvests for selection based on total yield; and 43 to 83 harvests for commercial yield, which varied according to the statistical estimation method. The principal component analysis based on the covariance matrix required the least number of harvests for a satisfactory selection precision. Fifteen harvests are sufficient for a satisfactory selection of all evaluated characteristics, with a precision above 70%.摘要 本研究旨在通过重复性系数（repeatability coefficient）估算西葫芦（Cucurbita pepo）杂交种可靠选育所需的最优采收次数。本试验采用完全随机区组设计，共设置33个处理（31个试验杂交种与2个商业杂交种），设置4次重复，每小区种植6株植株，共开展15次采收。研究对7个果实形态农艺性状进行了评价，并采用4种统计方法估算重复性系数。结果显示，无论所研究的性状类型如何，重复性系数均处于低至中等水平。对于高精度选育（R²≥90%），需要较多的采收次数，尤其针对果实产量相关性状，具体要求如下：基于总产量的选育需30至54次采收；基于商品产量的选育需43至83次采收，具体数值因统计估算方法而异。基于协方差矩阵的主成分分析所需采收次数最少，即可获得令人满意的选育精度。若要对所有被测性状实现精度高于70%的可靠选育，15次采收已足够。