Four cucumber (Cucumus sativus L.) inbred lines were intermated then bulked maternally to create four base populations denoted as cycle 0 (i.e., Pop.1 C0, Pop.2 C0, Pop.3 C0, Pop.4 C0). Each of these populations underwent phenotypic selection (PHE; open-field evaluations), selection by marker (MAS; genotyping at 20 marker loci), and random mating (RAN; no selection) for three cycles. The four traits under selection, multiple lateral branching (MLB), gynoecious sex expression (GYN), earliness (EAR), and fruit length to diameter ratio (L:D), are quantitatively inherited, controlled by relatively few (two to six) QTL per trait and are directly related to yield. Using the same C0 populations and selection scheme allowed a direct comparison of the effectiveness of MAS and PHE. Because each C0 population varied for any given trait, the response to MAS and PHE was not the same for each population. In general, C0 populations that were inferior for a trait either responded favorably to selection or remained constant, while those with superior trait values either did not change or decreased. Both MAS and PHE provided improvements in all traits under selection in at least one population, with the exception of MAS for EAR. MAS and PHE were equally effective at improving MLB and L:D, but PHE was generally more effective than MAS for GYN and EAR. When considering all traits, responses to PHE were superior in three of the four populations. The population for which MAS was superior, however, showed the only increase in yield (fruit per plant), which was not under direct selection. These results indicate that both MAS and PHE are useful for multi-trait improvement in cucumber, but their effectiveness depends on the traits and populations under selection.
Matthew D. Robbins and Jack E. Staub
Matthew D. Robbins, Jack E. Staub, and Zhicheng Fan
To increase yield in cucumber (Cucumussativus L.), we designed a recurrent selection program utilizing phenotypic (PHE) and marker-assisted (MAS) selection for the development of multiple lateral branching (MLB; branches per plant), gynoecious, early genotypes possessing high fruit length to diameter ratio (L:D). These yield components are under genetic control of few quantitative trait loci (QTL; 2-6 per trait), which have been placed on a moderately saturated molecular linkage map. Four inbred lines, complementary for the target traits, were intermated and the resulting population underwent MAS and PHE, as well as random mating (RAN), for three cycles. Selections by PHE were visually made for all four traits at the whole plant level. Selections based on MAS contained the highest number of desired marker genotypes from 20 marker loci (SSR, RAPD, SCAR, SNP). Using the same selection scheme and intensity allowed a direct comparison of MAS to PHE. Selection was equally effective for MLB and L:D by MAS (3.5 and 3.0) and PHE (3.6 and 3.0), which were both superior to RAN (2.8 and 2.8). For earliness (days to anthesis) and gynoecy (percent female), MAS (41.8 and 26.6) was less effective than PHE (40.5 and 81.8) and RAN (41.0 and 80.9), which were equal. For yield (fruit per plant), RAN (1.90) and MAS (1.88) were equal, but less than PHE (2.15). After three cycles of PHE, further selection by MAS identified superior genotypes, which were intermated. Superior hybrids were selected by MAS and underwent one backcross generation. In some backcrosses, gains were made in every trait compared to the PHE Cycle 3 mean, while in other backcrosses, gains were made only in some traits. Improvement by MAS was very effective during line extraction for these yield components.
Kimberly J. Walters, George L. Hosfield, Mark A. Uebersax, and James D. Kelly
Three populations of navy bean (Phaseolus vulgaris L.), consisting of recombinant inbred lines, were grown at two locations for 2 years and were used to study canning quality. The traits measured included visual appeal (VIS), texture (TXT), and washed drained mass (WDM). Genotype mean squares were significant for all three traits across populations, although location and year mean squares were higher. We found a positive correlation (r = 0.19 to 0.66) between VIS and TXT and a negative correlation (r = -0.26 to -0.66) between VIS and WDM and between TXT and WDM (r = -0.53 to -0.83) in all three populations. Heritability estimates were calculated for VIS, TXT, and WDM, and these values were moderate to high (0.48 to 0.78). Random amplified polymorphic DNA markers associated with quantitative trait loci (QTL) for the same canning quality traits were identified and studied in each population. Marker-QTL associations were established using the general linear models procedure with significance set at P=0.05. Location and population specificity was common among the marker-QTL associations identified. Coefficient of determination (R2) values for groups of markers used in multiple regression analyses ranged from 0.2 to 0.52 for VIS, 0.11 to 0.38 for TXT, and 0.25 to 0.38 for WDM. Markers were identified that were associated with multiple traits and those associations supported correlations between phenotypic traits. MAS would offer no advantage over phenotypic selection for the improvement of negatively associated traits.
K.P. Baiyeri, B.N. Mbah, and A. Tenkouano
The additive main effects and multiplicative interaction (AMMI) model was used to evaluate the stability patterns of 36 Musa genotypes in four cropping environments for bunch weight, pulp weight, and dry matter content. Alleycropping generally induced higher means for all traits than did sole cropping. The triploid plantains produced smaller bunch weights and were less stable than dessert and cooking bananas. In this ploidy group, bunch weight was highest for the cooking bananas `Cardaba' and `Fougamou', but only `Fougamou' was stable across environments. Among the hybrids, only `FHIA23' (dessert banana) expressed high and stable bunch weights, while other high-yielding hybrids displayed specific adaptation to alleycropping. Pulp weight was lower but more stable in plantains than in other triploid genotypes. Among the hybrids, pulp weight was high and stable for one cooking banana (`FHIA3'), one dessert banana (`FHIA1'), and three plantains (`PITA1', `PITA2', and `PITA7'). Dry matter content was highest in plantains and lowest in dessert bananas at both triploid and tetraploid levels, and was also more stable than the other traits. Thus, the adaptation patterns of genotypes across environments varied according to the trait studied. When rank changes were not observed across traits for a given genotype, differences were still noted in the relative magnitude of the IPCA1 score. Hence, both farm gate traits and postharvest processing traits should be considered in selecting for broad or specific adaptation. Determination of the genetic relationships between processing traits and farm gate traits could allow Musa breeders to construct selection indices that would facilitate multiple trait selection and enhance breeding efficiency, with respect to cultivar stability and adaptation across environments.
Matthew D. Kleinhenz and Annette Wszelaki
Yield and relationships among head traits were recorded in order to better understand the effects of planting date and cultivar selection on crop quality characteristics and to help increase the efficiency of cultivar development, evaluation, and selection. A total of seven cultivars of fresh market-type cabbage (Brassica oleracea L., Capitata Group) were planted in May and June of 1999 and 2000 at the OARDC Vegetable Crops Research Branch in Fremont, Ohio. Total and marketable yield, head traits (e.g., size, weight, density), and core dimensions were recorded at harvest. Main effects of year (Y), planting date (PD), and cultivar (C) and the Y × C interaction significantly affected seven to 10 of 10 head and core traits. However, the PD × C interaction was significant for head density, the ratio of head polar and equatorial diameter, and core base width. The Y × PD interaction was significant for six of 10 head and core traits. May planting tended to result in greater yield and larger, heavier heads with greater polar/equatorial diameter values relative to June planting. However, head density was unaffected by planting date. The number of head and core traits affected by planting date differed among cultivars. For example, six of 10 head and core traits were significantly affected by planting date in `Cheers' and `DPSX315' while one trait was affected by planting date in `SuperElite Hybrid'. The weight of numerous, individual, market-ready, trimmed heads showed a strong (avg. R 2 value = 0.92) quadratic relationship to average head diameter. These data suggest that large-scale germplasm evaluations may benefit by including multiple plantings, as head weight, volume, diameter, and shape were affected by planting date, possibly due to variation in temperature and rainfall patterns. The data also suggest that routine measurement of numerous head traits in the same evaluations may be unnecessary, as selected traits (e.g., diameter and weight, head volume, and core volume) were strongly related.
Valdomiro A.B. de Souza, David H. Byrne, and Jeremy F. Taylor
Seedlings of 108 families from crosses among 42 peach [Prunus persica (L.) Batsch] cultivars and selections were evaluated for six plant characteristics in 1993, 1994, and 1995. The data were analyzed by using a mixed linear model, with years treated as fixed and additive genotypes as random factors. Best linear unbiased prediction (BLUP) was used to estimate fixed effects. Restricted maximum likelihood (REML) was used to estimate variance components, and a multiple trait model was used to estimate genetic and phenotypic covariances among traits. The narrow-sense heritability estimates were 0.41, 0.29, 0.48, 0.47, 0.43, and 0.23 for flower density, flowers per node, node density, fruit density, fruit set, and blind node propensity, respectively. Most genetic correlations among pairs of traits were ≥0.30 and were, in general, much higher than the corresponding phenotypic correlations. Flower density and flowers per node (ra = 0.95), fruit density and fruit set (ra = 0.84) and flower density and fruit density (ra = 0.71) were the combinations of traits that had the highest genetic correlation estimates. Direct selection practiced solely for flower density (either direction) is expected to have a greater effect on fruit density than direct selection for fruit density.
Valdomiro A.B. de Souza, David H. Byrne, and Jeremy F. Taylor
Thirteen peach [Prunus persica (L.) Batsch] fruit characteristics were investigated for 3 years, 1993, 1994, and 1995, in College Station, Texas, to determine heritability, genetic and phenotypic correlations, and predicted response to selection. Seedlings of 108 families resulting from crosses among 42 peach cultivars and selections were used in the evaluations. A mixed linear model, with years treated as fixed and additive genotypes as random factors, was employed to analyze the data. Best linear unbiased prediction (BLUP) was used to estimate fixed effects. Restricted maximum likelihood (REML) was used to estimate variance components, and a multiple trait model was used to estimate genetic and phenotypic covariances between traits. Genetic and phenotypic correlations ≥0.65 and <0.30 were considered strong or very strong and weak, respectively. Date of ripening, fruit development period (FDP) and date of full bloom had the highest heritability (h2) estimates, 0.94, 0.91, and 0.78, respectively. Fruit cheek diameter and titratable acidity (h2 = 0.31) were the traits with the lowest estimates. Fruit development period, fruit blush, and date of ripening had the highest predicted selection responses, whereas fruit suture, fruit cheek, L/W12 (ratio fruit length to average fruit diameters), and fruit tip had the lowest values. Most genetic correlations were ≥0.30 and were, in general, much higher than the corresponding phenotypic correlations. All four measures of fruit size were genetically and phenotypically very strongly correlated. Important genetic correlation estimates were also observed for date of ripening with FDP (ra = 0.93), date of ripening and FDP with fruit blush (ra = -0.77, ra = -0.72), SS (percent soluble solids) (ra = 0.63, ra = 0.62) and TA (ra = 0.55, ra = 0.64), and SS with TA (ra = -0.56). Direct selection practiced solely for early ripening and short FDP is expected to have a greater effect on correlated traits than direct selection for early bloom and large fruit mass.
Cecil T. Pounders and G. Sam Foster
Analysis of clonal variation for two rooting traits of western hemlock [Tsuga heteterophylla (Raf.) Sarg.] clones indicated that realized genetic gain would be improved by more effective partitioning and reduction of environmental differences associated with clones. Interactions between clones and multiple propagation dates were significant, but clone rank changes between dates were minor. Number of meted cuttings per plot (RC) and number of main roots per cutting (MR) were more highly correlated genetically (1.06) than phenotypically (0.36). Broad-sense heritabilitiesHx 2 = 0.62 and Hx 2 = 0.79 for RC and MR, respectively) and predicted genetic gain from clonal selection were moderately high. Both the percentage of rooted cuttings and root system quality could be rapidly improved by the clonal selection procedures used.
Valdomiro A.B. de Souza, David H. Byrne, and Jeremy F. Taylor
Heritability estimates are useful to predict genetic progress among offspring when the parents are selected on their performance, but they also provide information about major changes in the amount and nature of genetic variability through generations. Genetic and phenotypic correlations, on the other hand, are useful for better planning of selection programs. In this research, seedlings of 39 families resulting from crosses among 27 peach [Prunus persica (L.) Batsch] cultivars and selections were evaluated for date of full bloom (DFB), date of ripening (DR), fruit period development (FDP), flower density (FD), node density (ND), fruit density (FRD), fruit weight (WT), soluble solids content (SS), apical protuberance (TIP), red skin color (BLUSH), and shape (SH) in 1993 and 1994. The data were analyzed using the mixed linear model. The best linear unbiased prediction (BLUP) was used to estimate fixed effects and predict breeding values (BV). Restricted maximum likelihood (REML) was used to estimate variance components, and a multiple-trait model to estimate genetic and phenotypic covariances between traits. The data indicates high heritability for DFB, DR, FDP, and BLUSH, intermediate heritability for WT, TIP, and SH, and low heritability for FD, ND, FRD, and SS. They also indicate year effect as a major environmental component affecting seedling performance. High correlation estimates were found between some traits, but further analysis is needed to determine their significance.
D.P. Coyne, J.R. Steadman, D.T. Lindgren, David Nuland, Durward Smith, J.R. Stavely, J. Reiser, and L. Sutton
Common bacterial blight (CBB), rust (RU), and white mold (WM) are serious diseases of great northern (GN) and pinto (P) beans in Nebraska and Colorado. The bacterial diseases halo blight (HB) and brown spot (BS) are sporadic. Severe Fe-induced leaf chlorosis (Fe ILC) occurs on calcareous sites. Separate inoculated disease nurseries are used to screen for resistance to the pathogens causing the above diseases. Yields and seed quality of lines are also determined in non-disease trials. Sources of exotic resistance to the above pathogens and to Fe ILD have been identified and their inheritance determined. A non-structured recurrent selection scheme has mainly been used, occasionally with a backcross program, to combine high levels of the desired traits. Selection for highly heritable traits such as seed size, shape and color, maturity, plant architecture, and RU resistance occurs in early generations while traits of low heritability, such as CBB resistance, WM avoidance, yield, seed coat cracking resistance, and canning quality, are evaluated in separate replicated tests over several years and finally for yield in on-farm-trials. A number of multiple disease resistant, high-yielding, well-adapted GN and P lines are or will be released; P `Chase' (on about 30,000 acres in 1996) and GN WM 3-94-9 (for possible release).