We mapped and characterized quantitative trait loci (QTL) affecting testcross performance of important forage maize (Zea mays L.) traits and investigated their consistency across different populations. Four independent populations comprising 380 F3 lines (A✕Bc), 113 F5 lines (A✕BV), 131 F4 lines (A✕CV), and 140 F4 lines (C✕DV) were produced from four elite flint inbreds (A, B, C, D) and genotyped at 89, 151, 104, and 122 RFLP marker loci, respectively. All lines were testcrossed to the same dent inbred tester and evaluated in field trials with two replications in five environments. Six traits were analyzed: dry matter yield (DMY), dry matter concentration (DMC), plant height (PHT), in vitro digestible organic matter (IVDOM), starch concentration (STC), and protein concentration (CPC) of forage. Genotypic variances (σ̂2g) were mostly significant for these traits in all populations and heritabilities generally exceeded 0.50. Between 2 and 14 QTL were detected in individual populations by composite interval mapping, which explained between 17.6 and 65.7% of the total phenotypic variance (σ̂2p). Few of the detected QTL displayed significant digenic epistatic or QTL ✕ environment interactions. Consistency of QTL positions across all four populations was greater for DMC, STC, and PHT than for DMY, IVDOM, and CPC. Across all traits' QTL, results were largely consistent among A✕BC and A✕Bv, but not among A✕Bc and A✕Cv or A✕Bc and C✕Dv. When QTL positions and substitution effects for DMY obtained in A✕BC, explaining 43% of σ̂2p, were employed for prediction of QTL positions and substitution effects in the other three populations they explained 24% of σ̂2p in A✕BV, 11% σ̂2p in A✕Cv, and 0% of σ̂2p in C✕DV. QTL results were poorly consistent among crosses within the flint heterotic pool, suggesting that prior to marker‐assisted selection, QTL mapping must be performed separately for each population.
