以辽宁、吉林和黑龙江3个省份13个种源的47株东北杏〔Armeniaca mandshurica （Maxim.） Skv.〕成熟果实为研究对象，采用多重比较、变异分析、巢式方差分析、主成分分析和聚类分析方法，对各种源的16个果实表型性状进行统计和分析。结果表明：各种源果实表型性状的差异较大；16个果实表型性状的Shannon-Wiener指数为1.92~2.04，其中，果核壳厚度和单果仁质量的Shannon-Wiener指数相等且最大。果实表型性状的种源间变异系数为9.40%~34.77%，种源内变异系数为5.35%~19.67%，且果实表型性状的种源间变异系数均值（17.38%）大于种源内变异系数均值（11.20%）。除单果仁质量、果仁纵径和果仁侧径外，其他果实表型性状在种源间和种源内大多存在显著或极显著差异；供试所有果实表型性状的种源间表型分化系数均值（52.19%）大于种源内表型分化系数均值（47.81%）。主成分分析结果表明：前4个主成分的累计贡献率达79.79%，能基本反映东北杏果实表型性状的大部分信息。其中，第1主成分的贡献率为39.83%，主要反映果实性状特征；第2主成分的贡献率为18.37%，主要反映果核性状特征；第3和第4主成分的贡献率分别为12.18%和9.41%，二者主要反映果仁性状特征。聚类分析结果表明：在欧氏距离17.5处，供试的13个种源被分成A和B类，并在欧氏距离7.5处各分成4和2个亚类。分析结果表明：A类包含辽宁桓仁、辽宁清原、辽宁东陵、吉林磐石、吉林龙潭、黑龙江宁安、黑龙江麻山和黑龙江阿城8个种源，这些种源的果实大、出仁率高，适用于选育大果或仁用类型；B类包含辽宁凤城、辽宁本溪、辽宁新宾、吉林抚松和吉林敦化5个种源，这些种源的果核大、果核壳厚、出核率高，适用于选育核壳用类型。研究结果显示：东北杏果实性状变异丰富，且变异主要来源于种源间，宜根据育种目标选择不同类群种源作为育种材料。

Taking ripe fruits of 47 individuals of Armeniaca mandshurica （Maxim.） Skv. from 13 provenances in 3 provinces of Liaoning， Jilin， and Heilongjiang as research objects， statistics and analysis on 16 fruit phenotypic characters of each provenance were carried out by methods of multiple comparison， variation analysis， nested analysis of variance， principal component analysis， and cluster analysis. The results show that the differences in fruit phenotypic characters of each provenance are relatively great； Shannon-Wiener indexes of 16 fruit phenotypic characters are 1.92-2.04， in which， those of nutlet shell thickness and single kernel mass are equivalent and the largest. The coefficients of variation of fruit phenotypic characters among provenances are 9.40%-34.77%， while those within provenance are 5.35%-19.67%， meanwhile， the average of coefficient of variation of fruit phenotypic characters among provenances （17.38%） is greater than that within provenance （11.20%）. Except single kernel mass， kernel vertical diameter， and kernel side diameter， other fruit phenotypic characters among provenances and within provenance most have significant or extremely significant differences； the average of phenotypic differentiation coefficient of all test fruit phenotypic characters among provenances （52.19%） is greater than that within provenance （47.81%）. The principal component analysis result shows that the cumulative contribution rate of the first four principal components reaches 79.79%， which can basically represent most informations of fruit phenotypic characters of A. mandshurica. In which, the contribution rate of the first principal component is 39.83%, which mainly represents fruit phenotypic characters; the contribution rate of the second principal component is 18.37%, which mainly represents nutlet phenotypic characters; the contribution rates of the third and fourth principal components are 12.18% and 9.41%, respectively, both of them mainly represent kernel phenotypic characters. The cluster analysis result shows that at Euclidean distance of 17.5， 13 test provenances are divided into group A and B, and each is further divided into 4 and 2 subgroups at Euclidean distance of 7.5. The analysis result shows that group A contains 8 provenances from Huanren of Liaoning, Qingyuan of Liaoning, Dongling of Liaoning, Panshi of Jilin, Longtan of Jilin, Ning'an of Heilongjiang, Mashan of Heilongjiang, and Acheng of Heilongjiang, these provenances have big fruit and high kernel rate， which are suitable for breeding big fruit or kernel apricot type； group B contains 5 provenances from Fengcheng of Liaoning, Benxi of Liaoning, Xinbin of Liaoning, Fusong of Jilin, and Dunhua of Jilin, these provenances have big nutlet， thick nutlet shell， and high nutlet rate， which are suitable for breeding nutlet shell type. It is suggested that the variations of fruit phenotypic characters of A. mandshurica are abundant， and the variations mainly originate from provenances， therefore， provenances of different groups should be selected as breeding materials according to breeding objectives.