Evaluation of herbage yield in a forage grass breeding program: comparison of visual rating versus measurement in single-row plots or swards

To refine selection methods for a perennial ryegrass (Lolium perenne L.) breeding program, half-sib families and commercial cultivars were evaluated for 3 years with treatments sown as both single-drill rows or swards. Dry matter yield of the perennial ryegrass treatments was evaluated several times in each year as a visual score which was subsequently calibrated against a regression determined by cutting a subset of plots or by cutting all plots. Thus, the experiment evaluated 2 aspects of herbage-yield determination in a perennial ryegrass breeding program: (i) the use of visual estimates of herbage yield to reduce the time spent cutting plots, and (ii) the use of single-row plots compared with swards. The correlation (either as Pearsons correlation coefficient, or Spearmans rank correlation coefficient) between visual estimates of herbage yield was always significant (P<0.01), with the exception of the rank correlation for sward plots in the summer 1995 (r = 0.4; P<0.05). However, the extent of the correlation varied (r = 0.4–0.9), and at some harvests calibrated visual ratings only explained a small proportion of the variance observed in harvested dry matter yields. These data suggest that visual ratings of herbage yield would be accurate enough to be used to detect large differences between families, breeding lines, cultivars or accessions of perennial ryegrass. However, when differences between lines are likely to be small, then harvesting all plots would give a more accurate estimate of the yield of perennial ryegrass lines. Likewise, the herbage yield of perennial ryegrass in single-row plots was significantly correlated with the herbage yield of perennial ryegrass sown as swards (P<0.01 or P<0.05). However, the correlation was again variable leading to the conclusion that evaluation of perennial ryegrass as single-row plots was not always an accurate indicator of sward yield. For those 4 (of 13) harvests over 3 years where the interaction between sward yield and row yield of the perennial ryegrass lines was significant (P<0.05), this interaction was shown not to be due to significant rank changes but rather to an increase in the differences of yield in swards or yield in single-row plots. We conclude that the harvesting of swards was the most reliable method of estimating the dry matter yield of perennial ryegrass cultivars. However, significant correlations between visual rating of treatments, or yield in single-row plots and measured yield as swards illustrated that these methods (visual ratings and single-plot yields) could be used to reduce the cost of evaluating differences in the herbage yield potential of perennial ryegrass, especially when these differences were likely to be large or when seed is limited, such as during the evaluation of accessions.