Preference of granivorous rodents for seeds of 12 temperate tree and shrub species used in direct sowing

Direct seeding is potentially a cost-effective method for creating woodland and is particularly suited to establishing mixtures of species. Seed predation can cause poor establishment, and information on granivore preference for seed of different species is essential to evaluate relative risk and the need for control measures. We investigated consistencies in the preference of small mammals for removing seeds of 12 tree and shrub species at three sites in southern England using freely available seeds presented as single or mixed species plots. Seed preferences were modelled as preference for removal of seeds of a particular species in relation to others present in a given time period. Across-site comparison of seed preference ranks showed significant agreement, especially at each extreme of the ranks. Quercus robur and Fagus sylvatica were always ranked first or second, followed by Corylus avellana and Acer pseudoplatanus, while Crataegus monogyna and Fraxinus excelsior were ranked last; the relative position of middle ranking species varied between tests. Removal of seeds of highly preferred species was usually absolute, whereas seeds of lower ranked species were more likely to be partially removed. For most species, preference coefficients were positively related to two explanatory variates: the logarithms of whole seed and coat dry mass. The exception was Fraxinus excelsior, where preference was lower than predicted, possibly because of chemical defences. Presentation of seeds as either single or mixed species plots had no overall effect on preference ranking. The results suggest that species that produce large nuts, are more vulnerable to predation-loss and might require additional measures, such as treatment with repellents, to reduce predation risk, or might, as in the case of Q. robur, be so attractive to rodents that it is unrealistic to direct sow them. The majority of the other species with smaller seed size, however, were much less susceptible to removal by predators. Any losses that do occur with these species are likely to be relatively small and tolerable, and localised.

► We modelled preferences of rodents for removing seeds of 12 tree and shrub species.
► Preference ranks showed significant agreement across sites.
► Preference coefficients were positively related to whole seed and coat dry mass.