Multiple paternity and offspring variability in wild populations of the garter snake, Thamnophis sirtalis (Colubridae)

Information about the relationships among phenotypic and genetic traits is of increasing importance in biology. Analytical approaches to the study of evolutionary adaptations commonly use optimization techniques in order to make the mathematics tractable. The validity of these approaches depends on the assumption that the predicted response is genetically possible. Knowledge of underlying genetic correlations clarifies relevant tradeoffs for use in these theories. Quantitative genetic models for the evolution of phenotypic characters rely on the assumption that genetic and phenotypic covariance matrices remain relatively constant. Therefore, empirical data on covariance matrices can be used to test these assumptions.

The garter snake,Thamnophis sirtalis (Colubridae), was used to study the quantitative genetics of several behavioral, morphological and physiological traits. The assumption that members of a single litter are full-siblings was tested. Contrary to the earlier predictions, high levels (~50%) of multiple paternity were found in these natural populations of garter snakes. Data were also collected on scale counts, aggregative tendencies, antipredator behavior, responses to chemical prey extracts, and critical thermal minima to examine potential genetic correlations among the characters. Significant litter and population differences were found in nearly every trait. Most of the traits had significant heritabilities. Inclusion of multiple paternity litters in full-sibling analyses leads to underestimates of heritabilities. The hypothesis that multiple paternity might increase variability within litters was examined. Variability was greatly increased for antipredator behavior in multiple paternity litters, but overall variability was not greatly within these litters was not enhanced.

Comparisons of the genetic and phenotypic correlation matrices suggest that the phenotypic correlations give a good estimate of the underlying genetic correlation but that whenever possible the genetic correlations should be determined.