Partitioning of Phenotypic Variance in Life-history Traits in

A characteristic of all individual-based population studies is the large variation between individuals within fitness traits. Partitioning the sources of this variation between individuals is a fundamental aim of such studies. Using a newly developed statistical approach (Bayesian cross-classified multivariate models), we partition the variance present in life-history traits (clutch-size, lay-date, nestling mass, successful recruitment per nest and adult survival) of known pairs of great tits Parus major, into that attributable to year and nest-box effects, separate male and female parental effects, and the unidentified variance remaining between individual observations. Our conclusions are: 1) Patterns in the allocation of the phenotypic variance differ between traits: e.g. between-female differences account for a large proportion of the variance in clutch-size, whilst between-year effects dominate the variance in lay-date; (2) With the exception of nestling mass, the influence of between-male variance is generally small within traits; (3) There is little evidence that local, nest-specific environmental effects (i.e. ‘territory ’ effects) are important; (4) A large proportion of the variation seems to be due to effects that may represent the stochastic nature of environmental variation. We suggest some further applications of this technique to similar long-term data sets