非正态分布状态对林木遗传力估算的影响

Effect of non normal distributions on estimating forest tree heritabilities.

针对林木遗传力估算过于粗放的情况 ,该文以对数正态分布和Γ 分布为例 ,应用模拟数据阐述了非正态分布对林木遗传力估算的影响 .结果表明 :对不满足假定前提的数据直接应用方差分析法估算遗传力 ,会导致估值出现很大偏差 (特别是单株遗传力 ) ,甚至产生遗传力估值为负或大于 1的情况 ;数据转换则是完全必要和可行的 .

In view of the existing situation that forest tree heritabilities is estimated too extensively, the paper elaborated the effect of non normal distributions (citing the normal logarithmic distribution and Γ distribution as instances) on estimating heritabilities with simulated data. The results show that applying analysis of variance to estimate heritabilities with data non satisfied with presumption, the estimation deviates from its value greatly (especially the single heritability). It is likely that the estimation is negative or bigger than 1. Data transformation is completely necessary and workable.