<正>利用１５个试点的马尾松种源区试林测定材料，估算不同产地的遗传稳定性，探究其地理变异模式和生态学基础。结果表明，６年生树高和胸径存在着显著的产地×地点互作效应，但难以用环境效应的线性回归进行预测。几种单变量和多变量分析技术的对比研究发现，不能用线性回归模型和主分量分析技术确定产地间稳定性差异，而Ｗｒｉｃｋｅ提出的生态价和Ｈｕｈｎ的Ｓｌｉ是两种较有效方法。产地稳定性呈明显的纬向地理变异模式，产地温度是造成这种变异模式的主要环境因子。云开大山和南岭山脉的产地产量高、稳定性低，可选用较好的立地建立高产人工林；武夷山脉和大娄山地的产地生长较迅速，且具有较好的稳定性和适应性，值得推广；北方产地产量低稳定性高，可运用于较差的立地。研究还发现，立地质量高的地点，不仅树木生长量大，而且对参试产地鉴别能力强。

The set of 6-year height and dbh data from 15-locations provenances test of masson pine was used to evaluate stability of different seed sources and determine its geographic variation pattern and ecological basis. The results showed that there existed significant seed source by site interaction,but its pattern was largely unpredictable.Comparisons of several univariate and multivariate methods suggested that stability differences between seed sources couldn’t be determined by using joint regression and principal component analysis.While Wricke’s ecovalence and Huhn’s Sli are two effective methods to discriminate stability of seed sources.The stability appeared to be related to the latitude of the seed source,but not to longitude and altitude of seed source.The temperature of seed source was found to be a main reason for this geographic pattern.The seed sources from Yudkai and Nanling mountain areas were of high growth and low stability and better adapted to good sites.The seed sources from Wuyi and Dalou mountain areas are worth the extension and use because they had higher growth rate and average stability and adaptability.Those seed sources from the Northern were of low growth rate and high stability and better adapted to poor sites. The results suggest that the efficient discrimination of seed source differences be achieved through intensive field testing at good sites.