Efficient inference of paternity and sibship inference given known maternity via hierarchical clustering

Thomas James Ellis, David Luke Field, Nicholas H. Barton*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)


Pedigree and sibship reconstruction are important methods in quantifying relationships and fitness of individuals in natural populations. Current methods employ a Markov chain-based algorithm to explore plausible possible pedigrees iteratively. This provides accurate results, but is time-consuming. Here, we develop a method to infer sibship and paternity relationships from half-sibling arrays of known maternity using hierarchical clustering. Given 50 or more unlinked SNP markers and empirically derived error rates, the method performs as well as the widely used package Colony, but is faster by two orders of magnitude. Using simulations, we show that the method performs well across contrasting mating scenarios, even when samples are large. We then apply the method to open-pollinated arrays of the snapdragon Antirrhinum majus and find evidence for a high degree of multiple mating. Although we focus on diploid SNP data, the method does not depend on marker type and as such has broad applications in nonmodel systems.

Original languageEnglish
Pages (from-to)988-999
Number of pages12
JournalMolecular Ecology Resources
Issue number5
Early online date6 Apr 2018
Publication statusPublished - Sept 2018
Externally publishedYes


  • Antirrhinum
  • fractional assignment
  • paternity
  • pedigree
  • sibships


Dive into the research topics of 'Efficient inference of paternity and sibship inference given known maternity via hierarchical clustering'. Together they form a unique fingerprint.

Cite this