Cultivar and Clonal Differences for Water-use Efficiency and Yield in Four Forage Grasses

Available water is a major factor limiting productivity of forage grasses grown in the northern Great Plains of the USA. Cultivars with greater water-use efficiency (WUE) would be beneficial to forage production and utilization in this semiarid region. Our objective was to assess variability in forage yield and WUE among genotypes of four common forage grasses. Four clones with contrasting phenotypes were selected from each of three cultivars or strains of crested wheatgrass [Agropyron desertorum (Fisch. ex Link) Shult.], western wheatgrass [Pascopyrum smithii Rybd. (Löve)], intermediate wheatgrass [Thinopyrum Intermedium (Host) Barkworth and Dewey], and reed canarygrass (Phalaris arundinacea L.). Soil water content was determined by neutron probe for 5 yr on clonal plots grown on a Parshall sandy loam soil (coarse-loamy, mixed Pachic Haploboroll) near Mandan, ND. A single forage harvest was taken each year when plants were at soft dough stage of seed development. Forage yield and WUE increased the first 2 yr of the study, then decreased for the next 2 yr. Increases and decreases in forage yield were associated with total available water (soil water plus growing season precipitation) prior to harvest and effects of stress in one year on subsequent years. There were no significant (P≥0.05) differences among entries for total water used before harvest. Environmental effects (years and replications) on yield and WUE in each species were large compared with genotypic effects. Genotypic mean squares were from 2 to over 35 times larger than experimental error mean squares for yield and WUE. Broad sense heritability estimates ranged from 71 to 95% for yield and from 69 to 94% for WUE over the four species, which indicated that gain from selection may be possible. However, because of the large year effects, extensive testing will be necessary to evaluate yield and WUE in the four grasses we studied. Development of more effective screening techniques will be necessary to reduce resource input required for evaluating WUE.