TURF QUALITY OF COOL SEASON GRASSES AT LOW INPUTS: RELIABILITY ACROSS YEARS, SEASONS AND SITES OF EVALUATION

Quality is a complex character of fundamental importance in turfgrass evaluation. It takes into account aesthetic and functional aspects, and depends on a number of individual components, such as uniformity, density, texture, growth habit, smoothness and colour, which may vary with time. The use of a synthetic, simple quality score is necessary when evaluating large numbers of varieties. The quality of 110 turf varieties belonging to 4 species (Lolium perenne L., Poa pratensis L., Festuca arundinacea Schreb. and F. rubra L.) was assessed across seasons of the second and third year of the crop cycle at three Italian locations, i.e., Lodi (Po Valley, continental climate), Perugia (central Italy, sub-Mediterranean climate) and Foggia (southern Italy, Mediterranean climate), using a visual score ranging from 9 (outstanding turf) to 1 (very poor turf). The experimental design was a Group Balanced Design with 3 replicates, grouping varieties within species to facilitate an optimal management of cutting procedures for each species. On average, F. rubra had the lowest quality in summer, whereas the other species had the lowest quality in winter and the highest quality in summer and autumn. Genotypic and genotype x environment (GE) interaction components of variance were estimated for each species. GE effects were about five times larger for P. pratensis, three times larger for F. rubra, two times larger for F. arundinacea and only about 0.75 times for L. perenne relative to genotypic effects. Relatively large GE effects were observed for variety x locations interaction in tall and fine fescues, variety x season interactions in L. perenne, F. rubra and P. pratensis, and higher-order interactions in P. pratensis. Stability of turf quality across sites, seasons and years was measured for each cultivar in terms of environmental variance. Mean and stability values of varieties were integrated into an index of reliability, estimating the lowest score value expected in 80% of the cases, thus identifying for each species, a small subset of highly reliable cultivars.