以广西贵港覃塘、梧州苍梧、南宁宾阳、桂林雁山和桂林平乐的野生水松〔Glyptostrobus pensilis (Staunton ex D. Don) K. Koch〕为研究对象，采用方差分析、相关性分析、主成分分析、综合评价及聚类分析等方法对5个分布地水松球果、种子和叶的表型性状进行分析。结果表明：南宁宾阳水松的果长、果宽、单果鲜质量、单果出籽率、种子宽、种子厚、种翅宽、种子大小、种子千粒质量和叶形指数最大；桂林雁山水松的果长、果宽、单果鲜质量、单果籽粒数、果形指数、单果出籽率、种子长、种子宽、种子厚、种翅长、种翅宽、种形指数、种子大小和种翅长宽比最小，而其叶长、叶宽和叶面积最大。方差分析结果表明：分布地间球果、种子和叶的表型性状差异达到极显著水平(p<0.01)。供试表型性状的分布地间变异系数为7.43%~69.03%，其中，叶表型性状的分布地间变异系数均值最大(38.20%)。绝大多数表型性状间的相关性显著(p<0.05)或极显著，其中，果长、单果鲜质量、种子长、种翅长、种翅宽、种子大小和叶宽与其他表型性状显著或极显著相关，可作为水松优良种质筛选的重要指标。总体来看，球果和种子表型性状与经度、纬度、海拔、年均降水量和年均空气相对湿度呈显著或极显著负相关，与年均温呈极显著正相关；叶表型性状与经度、纬度、海拔、年均降水量和年均空气相对湿度呈极显著正相关，与年均温呈极显著负相关。主成分分析结果表明：前3个主成分的累计贡献率为95.890%，表明这3个主成分能够反映水松表型性状的绝大部分信息。从不同分布地水松表型性状的综合得分看，南宁宾阳最高(1.854)、贵港覃塘次之（1.187）、桂林雁山最低(-3.128)。聚类分析结果显示：贵港覃塘、梧州苍梧和南宁宾阳为一组，桂林雁山和桂林平乐分别单独为一组。综上所述，广西5个分布地野生水松的表型多样性较为丰富，尤其是叶表型性状；比较而言，南宁宾阳和贵港覃塘的水松表型性状较好，可作为广西水松优良单株筛选和良种选育的材料。经度、纬度、海拔、年均温、年均降水量和年均空气相对湿度是影响广西水松表型性状变化的重要地理-气候因子。

Taking wild Glyptostrobus pensilis (Staunton ex D. Don) K. Koch from Qintang of Guigang, Cangwu of Wuzhou, Binyang of Nanning, Yanshan of Guilin, and Pingle of Guilin in Guangxi as research objects, the phenotypic traits of cones, seeds, and leaves of G. pensilis from five distribution areas were analyzed by using the methods of variance analysis, correlation analysis, principal component analysis, comprehensive evaluation, cluster analysis, etc. The results show that the cone length, cone width, fresh mass per cone, seed yield per cone, seed width, seed thickness, seed wing width, seed size, 1 000-grain mass of seeds, and leaf shape index of G. pensilis from Binyang of Nanning are the largest; the cone length, cone width, fresh mass per cone, seed number per cone, cone shape index, seed yield per cone, seed length, seed width, seed thickness, seed wing length, seed wing width, seed shape index, seed size, and seed wing lengthwidth ratio of G. pensilis from Yanshan of Guilin are the smallest, while the leaf length, leaf width, and leaf area of G. pensilis from this distribution area are the largest. The variance analysis result shows that the differences in phenotypic traits of cones, seeds, and leaves among distribution areas reach an  extremely significant level (p<0.01). The coefficients of variation among distribution areas of test phenotypic traits are 7.43%-69.03%, in which, the mean of coefficient of variation among distribution areas of phenotypic traits of leaves is the largest (38.20%). There are significant (p<0.05) or extremely significant correlations between most phenotypic traits, in which, the cone length, fresh mass per cone, seed length, seed wing length, seed wing width, seed size, and leaf width show significant or extremely significant correlations with other phenotypic traits, so these phenotypic traits can be used as important indexes for screening excellent germplasm resources of G. pensilis. In general, the phenotypic traits of cones and seeds show significant or extremely significant negative correlations with longitude, latitude, altitude, mean annual precipitation, and mean annual air relative humidity, and show extremely significant positive correlations with mean annual temperature; while the phenotypic traits of leaves show extremely significant positive correlations with longitude, latitude, altitude, mean annual precipitation, and mean annual air relative humidity, and show extremely significant negative correlations with mean annual temperature. The principal component analysis result shows that the cumulative contribution rate of the first three principal components is 95.890%, indicating that the three principal components can reflect most information of the phenotype traits of G. pensilis. From the view of comprehensive score of phenotypic traits of G. pensilis from different distribution areas, Binyang of Nanning has the highest score (1.854), followed by Qintang of Guigang (1.187), while Yanshan of Guilin has the lowest score (-3.128). The cluster analysis result shows that Qintang of Guigang, Cangwu of Wuzhou, and Binyang of Nanning are clustered into one group, Yanshan of Guilin and Pingle of Guilin are in  separate groups. Taken together, the phenotypic diversity of wild G. pensilis from five distribution areas in Guangxi are relatively abundant, especially leaf phenotypic traits; in comparison, the phenotypic traits of G. pensilis from Binyang of Nanning and Qintang of Guigang are better, both of them can be used as the materials for screening excellent individuals and breeding improved varieties of G. pensilis in Guangxi. Longitude, latitude, altitude, mean annual temperature, mean annual precipitaiton, and mean annual air relative humidity are the important geographicalclimatic factors for affecting the variation of phenotypic traits of G. pensilis in Guangxi.