My personal experience suggests to me that there is a high degree of heritability for somatotypes. This study is interesting in that it tries to figure out just how heritable they are.
Primary objective: To examine familial resemblance in the Heath-Carter anthropometric somatotype in a sample of 328 participants from 103 nuclear families in Northern Ontario (Canada). Methods and procedures: The three somatotype components (endomorphy, mesomorphy, ectomorphy) were subjected to principal components analysis and the resulting first principal component (PC1) was used as an additional index of physique. The four phenotypes were adjusted for age, sex and generation effects, while each of the three somatotype components was further adjusted for the effects of the other two components using regression procedures. A familial correlation model was fit to the data and used to estimate the degree of familial resemblance in somatotype. Main outcome and results: For all somatotype variables, the most parsimonious model was one in which there was no spouse resemblance and no sex or generation effects in the familial correlations. Maximal heritabilities were 56%, 68%, 56% and 64% for endomorphy, mesomorphy, ectomorphy and PC1, respectively, indicating significant familial resemblance for the Heath-Carter anthropometric somatotype. Further, the pattern of familial correlations suggests the role of genetic factors in explaining variation in human physique. Conclusions: In general, a pattern of no spouse but significant parent-child correlations implicates the role of genes on human physique, provided that mating is random with regard to these traits.