• bears;
  • fecundity;
  • life history;
  • natural selection;
  • Oncorhynchus nerka ;
  • predation;
  • salmon;
  • spawning success;
  • Ursus arctos


  1. Selection drives evolutionary changes but is often difficult to quantify and tightly link to phenotypic trait distributions in wild populations.
  2. Sampling a single population of sockeye salmon (Oncorhynchus nerka) for over a decade, we calculated the expected spawning success for females resulting from: (i) increased fecundity with body length, (ii) the mortality cost of prolonged marine residency necessary to achieve large size, (iii) size-selective natural mortality on the spawning grounds from biotic (bear and gull predation) and abiotic (stranding) processes and (iv) exploitation by a size-selective commercial fishery. We quantified the size-specific probability of different modes of death and the resultant potential for successful spawning, and then modelled the theoretical relationship between female length and fitness (spawning success) in the population.
  3. This optimal distribution closely matched the observed length distribution of the focal population, when removals by the size-selective commercial fishery were included. We then used a likelihood-based approach to compare competing model predictions to length distributions from other populations in the watershed with different levels of size-selective freshwater mortality, as determined by the physical characteristics of the spawning grounds.
  4. This study provides a quantitative framework for assessing female spawning success in wild populations, as represented by the expected number of eggs deposited per spawning female. These results advance previous analyses of natural selection in that predictions for phenotypic distributions were generated and then compared to those observed in situ, rather than assuming the adaptive nature of observed distributions.