A simple model for pollen-parent fecundity distributions in bee-pollinated forage legume polycrosses

Authors: Riday, Heathcliffe; SMITH, MARK - S&w Seed Company; Peel, Michael

Submitted to: Journal of Theoretical and Applied Genetics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/5/2015
Publication Date: 6/24/2015
Publication URL: http://handle.nal.usda.gov/10113/62851
Citation: Riday, H., Smith, M.A., Peel, M. 2015. A simple model for pollen-parent fecundity distributions in bee-pollinated forage legume polycrosses. Journal of Theoretical and Applied Genetics. 128:1865–1879.

Interpretive Summary: Close mating leads to inbreeding in plants and animals, and breeders need to avoid too much inbreeding because it leads to decreased vigor in offspring. Breeders use quantitative genetic theory to predict the amount of expected inbreeding in groups of mating individuals based on various factors. The default assumption, known as panmixis or random mating, is that every individual is equally likely to mate with every other individual and that all individuals produce equal numbers of offspring. Geneticists have surmised for years that this is probably not true. This study measured mating patterns in forage legume crosses that were pollinated by bees in order to determine observed pollen-parent contribution frequencies to the progeny generation (i.e., pollen-parent fertility distributions). As expected, pollen-parent fertility distributions in forage legume plants were not in panmixis. Surprisingly, however, the distributions were very predictable. Using this data, a simple model was developed that only requires knowledge of the number of individuals in a group of mating individuals to predict the expected pollen-parent fertility distribution. This model is useful in more accurately predicting expected inbreeding in forage legume breeding programs. It may also be useful much more broadly in other plant species in predicting pollen-parent fertility distributions. And the model could potentially be incorporated into quantitative genetic models and simulations that currently utilize a panmixis assumption.

Technical Abstract: Random mating or panmixis is a fundamental assumption in quantitative genetic theory. Random mating is sometimes thought to occur in actual fact although a large body of empirical work shows that this is often not the case in nature. Models have been developed to model many non-random mating phenomena. This paper measured pollen-parent fecundity distributions among outbred perennial forage legume species (autotetraploid alfalfa [Medicago sativa L.], autohexaploid kura clover [Trifolium ambiguum M. Bieb.], and diploid red clover [Trifolium pratense L.]) in ten polycrosses ranging in size (N) from 9 to 94, pollinated with bee pollinators (Bumble Bees [Bombus impatiens Cr.] and leafcutter bees [Megachile rotundata F.]). A Weibull distribution best fit the observed pollen-parent fecundity distributions. After standardizing data among the 10 polycrosses, a single Weibull distribution-based model was obtained with an R-sq. of 0.978. The model is able to predict pollen-parent fecundity distributions based on polycross size alone. The model predicts that the effective polycross size will be approximately 9% smaller than under random mating (i.e., Ne/N ~ 0.91). The model is simple and can easily be incorporated into other models or simulations requiring a pollen-parent fecundity distribution. Further work is needed to determine how widely applicable the model is.