Effects of plot type on detection of quantitative‐trait‐locus effects in barley (Hordeum vulgare L.)

Abstract

Hill plots offer advantages in testing large germplasm arrays with limited area and seed. However, the relationship of trait expression between hill vs. row plots remains in question. The development of quantitativetrait-locus (QTL) analysis offers alternative methods of testing hypotheses regarding levels of coincident gene effects in divergent plots. The objective of this study was to compare hill and row plots in terms of QTL detection, number of common QTLs and plot-type-specific QTLs for various characters in barley. Assessments were based on sets of 100 F,-derived doubled haploid progeny from two environments of hillplot data and two sets of multiple-environment row-plot data. Common large-effect QTLs for height, heading date, and 1000-kernel weight were detected in both plot types. Fewer QTLs were detected in the hill-plot data, with only one QTL detected for grain yield. There were no hillplot-specific QTL effects. Hill plots appear to be best suited to highly heritable characters such as height, heading date, and 1000-kernel weight. Evaluation of grain yield is best deferred until larger experimental units can be employed.Key words: Hordeum vulgare -quantitative trait locusdoubled haploids -hill plots Hill plots are an attractive phenotype assessment option for small-grain germplasm when there are large germplasm arrays and limited seed and/or land (McFerson and Frey 1990). The underlying issue, of course, is the level of coincident trait expression between hill and row plots. In general, there is a consensus that hill plots provide effective discrimination for highly heritable traits, such as maturity and plant height, but there is considerable disagreement as to their effectiveness for traits such as grain yield. The issue is not easily addressed from an experimental standpoint. An advantage of hill plots is that their limited spatial requirements minimize field heterogeneity when large numbers of lines are evaluated. However, in an empirical comparison of row vs. hill plots, the number of entries must be minimized in order not to unduly bias the performance of the row plots.Despite the variety of test statistics employed in comparing data frotn adjacent hill and row plot experiments -the nurnber of lines in common at given selection intensities (Tragoonrung et al. 1990), correlations of various sorts (St. Martin et al. 1990) and coefficients of variation (Pfelffer and Pilcher, 1989) ^ there is still no consensus as to the relative merits of the two plot types. The recent development of medium-density genome-mapping techniques has allowed for estimation of the chromosome location of genes underlying quantitative phenotypes. Quantitative-trait loci (QTLs) for a range of agronomic and maltingquahty traits in barley were described by Hayes et al. (1993), based on the medium-density map reported by Kleinhofs et al. (1993). Hayes et al. (1993) employed conservative criteria for identifying QTL peaks and support intervals for average performance in multiple environments, based on the rationale that chromoso...

Effects of selection and opportunities for recombination in doubled‐haploid populations of barley (Hordeum vulgare L.)

“…Kernels per spike and 1000-kernel weight were not among the QTL traits analysed by Hayes et al (1993), but, as discussed by Iyamabo and Hayes (1995), 1000-kernel weight QTLs were coincident with grain-yield QTLs and, in all cases, favourable alleles were contributed by 'Steptoe'. Plant height, heading date, and grainyield QTLs were mapped by Hayes et al (1993) in this population.…”

“…There is a general consensus that hill-plot phenotypes are reliable predictors of row-plot performance for heading date, plant height, and yield components (Tragoonrung et al 1990), but the utihty of hill plots for grain yield selection is still debatable (Pfeiffer andPilcher 1989, St.-Martin et al 1990). In a companion paper (Iyamabo and Hayes 1995), it is reported that hill plots are nearly as effective as row plots in detecting QTLs for plant height, heading date, and 1000-kernel weight, but not as effective in detecting yield QTLs. Whether this is attributable to higher experimental error or to altered competitive relationships is not known.…”