Selection for persistence under grazing in winter-active populations of the perennial grass, Phalaris aquatica L. (phalaris)

Abstract

Winter-active cultivars of phalaris (Phalaris aquatica L.) have the potential to be more productive than semi-winter-dormant cultivars but may be less persistent under the stress of heavy grazing. The feasibility of improving persistence of winter-active phalaris was examined by subjecting plots of half-sib families from several winter-active populations to 2 cycles of heavy grazing pressure by sheep. Because variation among families in initial density influenced final density score in both cycles, persistence was assessed by 2 methods, final density with initial density as a covariate and the ratio of final over initial density.

In Cycle 1, significant family variance was observed in a broadly based population after 3 years of continuous grazing. Changes in persistence were smaller but still significant among the families comprising cv. Holdfast after 2 years of rotational grazing on limed but not on unlimed acid soil. In Cycle 2, 4 populations selected in Cycle 1 and elsewhere were screened under continuous grazing at 2 sites, one lower in soil pH and phosphorus than the other. Family by site interaction for final density was significant (P < 0.05) for one population only and this arose before grazing commenced. The selected populations displayed higher density and persistence than control winter-active cultivars. Estimates of predicted heritability were generally high but not always significant at P < 0.05. Overall they were considered to indicate that differences in persistence between families were present and heritable. This was supported by the observation that generations of 2 populations selected in Cycle 1 were higher in final density and persistence than the base generations, although differences in persistence as distinct from final density were not always at the P = 0.05 significance level.

The ratio of final over initial density scores was considered less appropriate than the covariate method of estimating persistence in this cycle because plots that established at very low density and increased slightly thereafter resulted in high error variance and biased selection towards families that established at low density.