Article Text

Cluster analysis of the genetic heterogeneity and disease distributions in purebred dog populations
  1. G. J. Ubbink, DVM1,
  2. J. van de Broek, PhD2,
  3. H. A. W. Hazewinkel, DVM, PhD1 and
  4. J. Rothuizen, DVM,PhD1
  1. 1 Department of Clinical Sciences of Companion Animals
  2. 2 Centre of Biostatistics, Faculty of Veterinary Medicine, University of Utrecht, PO Box 80 154, 3508 TD Utrecht, The Netherlands


Purebred dog populations have been subject to strong selection which has resulted in extreme differences between breeds and decreased heterogeneity within breeds. As a result, breed-specific inherited diseases have accumulated in many populations. The aim of this study was to analyse genetic heterogeneity in relation to the distribution of elbow dysplasia in labrador retrievers, portosystemic shunts in Irish wolfbounds, and hepatic copper toxicosis, in Bedlington terriers. Decreased heterogeneity was demonstrated in the multiple genetic interrelations in the three populations. In pedigrees containing seven generations of ancestors, the average number of common ancestors in all pair-wise combinations of dogs was five to six (range 0 to 18). These complex interrelationships were resolved by a cluster analysis on matrices of relatedness. This analysis gave clusters of highly related animals, the average relatedness of these clusters, and the average relatedness of the entire population, as expressions of its genetic variability. The mean relatedness was 0.032 for Irish wolihounds and Bedlington terriers, and 0-002 for labrador retrievers. The labrador retriever cohort was resolved into 31 clusters, and all cases of elbow dysplasia were concentrated in five highly related clusters with an overall incidence of 17 per cent. The Bedlington terrier cohort consisted of 12 clusters which all contained cases of copper toxicosis, with an overall incidence of 46 per cent. The Irish wolfhounds were divided into 14 clusters with a disease incidence of 4 per cent. Dogs with portosystemic shunts were found in four averagely related clusters. A genetic distribution became obvious only when relatedness due to common ancestors of the cases was used as a criterion, and the cases were then concentrated in five highly related clusters.

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.