Crop Science

Henning, John A.; Teuber, Larry R.

doi: 10.2135/cropsci1996.0011183X003600010001xpmid: N/A

Breeders need efficient multiple trait selection schemes that allow selection for each trait within its respective optimum environment. This report describes such a breeding procedure entitled, “modified convergent improvement,” and compares theoretical predictions of genetic gain per cycle (Δq/cycle) for the following schemes: modified convergent improvement, tandem selection in two dissimilar populations followed by a population cross (strain crossing), recurrent selection for single traits in dissimilar populations followed by a multiple population cross (multiple strain cross), and independent culling. Estimates of Δq/cycle were calculated for both additive and dominant gene action when selecting for three to six traits. Modified convergent improvement (Δq/cycle = 0.13) and independent culling (Δq/cycle 0.11) exhibited highest average Δq/cycle for traits controlled by recessive genes (P≤ 0.05). Independent culling exhibited the highest average Δq/cycle when selecting for traits controlled by dominant genes (Δq/cycle = 0.13) or by additive gene action (Δq/cycle = 0.20). Modified convergent improvement proved superior to independent culling when selecting for traits controlled by additive gene action or by recessive genes when gene frequency was low (q = 0.04−0.12). Theoretical Δq/cycle values for multiple strain cross and strain crossing were consistently lower than those obtained for either independent culling or modified convergent improvement.We propose that modified convergent improvement is a superior method for multiple trait selection when traits must be evaluated in dissimilar environments or during the initial stages of germplasm development. This procedure confers additional benefits including reducing inbreeding potential and/or disequilibrium over that observed in single population breeding procedures.

Mazzucato, Andrea; Nijs, Anton P. M.; Falcinelli, Mario

doi: 10.2135/cropsci1996.0011183X003600010002xpmid: N/A

To estimate the frequency of apomixis in Kentucky bluegrass (Poapratensis L.) by use of traditional embryo sac classification and progeny testing methodologies is difficult and time consuming. This study was conducted to test the applicability of the auxin test, a recently developed technique that allows the asexual potential to be estimated quickly as the frequency of auxin‐induced, endospermless grains containing one or more parthenogenetic embryos. Frequency of parthenogenesis was estimated by the auxin test for eight clones in two seasons and two locations. The 2,4‐D (2,4‐dichlorophenoxyacetic acid) induction of endospermless grains ranged from 50 to 100%. Two clones were 100% sterile, while the others presented sterility ranging from 2 to 63%. Expression of parthenogenesis was fairly stable within and between environments, suggesting a strong genetic control. Differences in parthenogenesis frequency between clones were evident. Statistical analysis based on pooled data discriminated highly (87–89%), intermediate (78%), and moderate (69–73%) parthenogenetic clones. Estimates of parthenogenesis frequency obtained by embryo sac analysis were similar to or lower than those from the auxin‐induction test. Estimates of apomictic reproduction by progeny tests were usually higher than both estimates of parthenogenesis. Auxin‐induced grains of one clone were analyzed by histological sections. Most of the embryos in the parthenogenetic ovules and grains (80%) seemed normal with regard to position, shape, and cell pattern. It is therefore concluded that the auxin test is a workable technique for determining the percentage of parthenogenesis in breeding programs.

Matzk, Fritz

doi: 10.2135/cropsci1996.0011183X003600010003xpmid: N/A

Plant breeders are interested in enlarging the genetic variability within a species of interest and in improving breeding strategies. A wide hybridization program was conducted from 1990 to 1994 in Germany to determine the prospects for sexual gene transfer or haploid production in oat (Avena sativa L.). Five cultivars of oat were pollinated with pearl millet [Pennisetum americanum (L.) Leeke], eastern gamagrass [Tripsacum dactyloides (L.) L.], and maize (Zea mays L.; 2x and 4x) in the greenhouse. Exogenous auxin application, embryo rescue, and early colchicine‐mediated doubling of chromosomes were used to overcome postzygotic barriers. The recovered plants were analyzed cytologically and phenotypically under greenhouse and field conditions. Depending on the pollinator species, the embryo frequencies varied from 0.4% (Z. mays 4x) to 9.8% (pearl millet). In embryo rescue, initial growth occurred frequently, but many plantlets died. Chromosomes of the pollinator species were lost late in embryo and plant development. One to four chromosomes of the pollinator species were found at the tillering stage in root tip cells. Overall, four viable plants were produced, among them hybrids with pearl millet and eastern gamagrass for the first time. The efficiency of haploid production (<0.1%) was too low in all combinations for application in plant breeding programs; however, the transfer of genes or chromosomes appears promising in crosses of oat with maize and pearl millet.

Meredith, W. R.; Pettigrew, W. T.; Heitholt, J. J.

doi: 10.2135/cropsci1996.0011183X003600010004xpmid: N/A

Combining all useful traits into a single genetic background does not always result in a genotype whose performance could be predicted by individual trait effects. Our objective was to evaluate the effects of three cotton (Gossypium hirsutum L.) traits, and their interactions with one another and environments. Sub‐okra leaf, semi‐smooth leaf, and nectariless were backcrossed into ‘DES 119’ from MD65‐11S. All eight combinations of the three traits were evaluated as BC4F4 and BC4F5 near isogenic populations in 1992 and 1993. They were grown at four locations near Stoneville, MS, in a randomized complete block design with six replications for yield and three for all other performance characteristics. No significant total yield response due to any trait was detected; however, sub‐okra leaf types produced significantly higher (35 kg ha−1 or 4%) first harvest yield than normal leaf types. Significant differences due to the main effects were detected for all other characteristics. Lint percentages for hairy and semi‐smooth types were 37.3 and 36.8, respectively. Fiber strengths for nectaried and nectariless isolines averaged 196 and 192 kN m kg‐1, respectively. Normal leaf had heavier bolls and seed, longer fiber, and lower fiber elongation than sub‐okra types. Seed of semi‐smooth isolines were heavier than those of hirsute cottons. Nectaried populations averaged larger bolls and seed, longer 2.5% span length, stronger fiber and higher elongation than that for nectariless populations. Interactions were smaller than the 1 to 4% differences detected for main effects. No major negative interactions among traits or with other environments were encountered. Incorporating these traits into one cultivar with similar genetic background to DES 119 should result in no major breeding problems.

Fountain, M. O.; Hallauer, A. R.

doi: 10.2135/cropsci1996.0011183X003600010005xpmid: N/A

Proper choice of germplasm determines the ultimate success of selection for genetic improvement. Germplasm included in maize (Zea mays L.) breeding has evolved from the original landrace cultivars to F2 populations of elite line crosses. Objectives of our study were to determine the mean and genetic variability (σ2G) of yield and other traits within three F2 populations (B73 × B84, B90 × Mo17, and B73 × Mo17), within three genetically narrow‐base synthetics (NBS) [BS14, Mol7‐Syn., and H73‐Syn.], and within three genetically broadbase synthetics (BBS) [BSll(FR)C9, BSCBI(R)C9, and BSSS(R)C9]. Fifty S1 progenies for each of the nine populations were evaluated at three locations for 2 yr. Differences among S1 progenies within each population were significant for most traits. Estimates of σ2G were determined from the S1 component of variance, and average estimates of σ2G for grain yield were similar for F2 and BBS populations, whereas the average estimate of σ2G for NBS populations was 40% smaller. Within population types, the related line cross (B73 × B84)F2 had 50% less σ2G for grain yield than the other two F2 populations. BS14, H73‐Syn., and BSCBI(R)C9 had smaller estimates of σ2G for grain yield than the other synthetic populations. Estimates of σ2G suggest selected F2 and BBS populations have greater potential for improved gains than the NBS populations. The related line (B73 × B84)F2 had 50% less 62G than either (B90 × Mo17)F2 or (B73 × Mo17)F2. (B90× Mo17)F2 and (B73 × Mo17)F2 had similar estimates of σ2G, although B90 and Mo17 are in the same heterotic group, whereas B73 and Mo17 are in different heterotic groups [Reid Yellow Dent (B73) and Lancaster Sure Crop (Mo17)]. Selection for improvement of yield would be more effective within the (B90 × Mo17)F2 than within (B73 × B84)F2 and more useful than within (B73 × Mo17)F2.

Abdalla, Osman S.; Crossa, José; Autrique, Enrique; DeLacy, Ian H.

doi: 10.2135/cropsci1996.0011183X003600010006xpmid: N/A

Knowledge of the relationships among international test sites is valuable information for effective targeting of germplasm exchange. Five years of grain yield data from the Elite Durum Yield Trial (EDYT), consisting of 132 trials from 32 locations in 22 countries, were studied. Pattern analysis, the combination of ordination and cluster analysis, was used to identify groupings of international test sites that represent similar selection environments and compare location association with the mega‐environment designations of the International Maize and Wheat Improvement Center (CIMMYT) breeding program. Two main environmental groups and six subgroups were identified for durum wheat (Triticum turgidum L. var. durum). The major determinants of the groupings were latitude and moisture supply. Biotic and abiotic stresses influenced further delineation of the clusters. Discrepancies between mega‐environment designation and pattern analysis results warrant further investigations of the underlying causes. The relationships among test sites documented in this study should provide a framework for effectively targeting germplasm and information exchange between comparable programs.

Goblirsch, C. A.; Horsley, R. D.; Schwarz, P. B.

doi: 10.2135/cropsci1996.0011183X003600010007xpmid: N/A

Grain protein content is an important malt quality trait in barley (Hordeum vulgare L.). A source of low‐protein used by the North Dakota six‐rowed barley breeding program has been the cultivar Karl. No cultivars with Karl in their pedigree have been released from this program because all low‐protein lines have had unacceptably dark kernel color, low diastatic power (DP), or both. The objective of this study was to determine why low‐protein cultivars with acceptable kernel color, DP, and grain protein content have not been developed and to suggest a strategy to breed such cultivars. To fulfill these objectives, the following grain protein content, kernel color, and DP data were collected from three crosses: broad‐sense heritability estimates, expected and observed gains from selection, genetic correlations between the three traits, and the correlated response of a trait when selecting for another trait. Low to medium heritability estimates for all traits, low genetic correlation between traits, and the small expected correlated changes in DP when selecting for decreased grain protein content or a brighter kernel color suggest it should be possible to identify low‐protein lines with acceptable kernel color and DP. Using independent culling, DP of selected lines in all crosses was similar when either grain protein content or kernel color was the first trait selected; therefore, either trait could be used as the primary selection trait. A strategy for developing low‐protein cultivars with acceptable kernel color and DP would be to use independent culling in F2:3 or F3:4 families in which grain protein content or kernel color would be the first trait and DP would be the last trait selected. The population size needed to select such lines needs to be greater than the I00 to 200 F2:3 or F3:4 families currently used by the North Dakota six‐rowed barley breeding program.

Milligan, S. B.; Martin, F. A.; Gravois, K. A.

doi: 10.2135/cropsci1996.0011183X003600010008xpmid: N/A

The inheritance of ratooning ability and the relationship of traits among crops in sugarcane (Saccharum spp. hyb.) has not been well examined. Ratooning ability (RA) was defined as the second ratoon(SR) crop yield percent of the plant cane yield. A replicated 4‐yr test at four locations of 37 genotypes was studied for two three‐crop cycles. Broad‐sense single‐plot heritabilities for RA were low (H ≤ 17%), while the genetic coefficient of variation of RA was largest for sucrose yield and cane yield (GCV ≈ 14.5%), and smallest for stalk diameter (GCV = 1.5%). Cane and sucrose yield RA demonstrated the largest potential for gain, while stalk weight, stalk diameter, and stalk length showed the least. Except for sucrose and cane yield and stalk number, other traits were highly correlated between plant cane and SR crops (r ≥ 0.78). Stalk number in the younger crop was the only trait significantly correlated to ratoon crop cane yield (r = 0.56), suggesting that selection for stalk number in the younger crops would enhance older crop yields. The results indicate that SR crop yields could be predicted by first ratoon crop yields. However, the best improvement of SR yields would be realized by selection in the SR.

Bernardo, Rex

doi: 10.2135/cropsci1996.0011183X003600010009xpmid: N/A

In preliminary studies, best linear unbiased prediction (BLUP) has been found useful for identifying high‐yielding maize (Zea mays L.) single crosses prior to field evaluation. In this study, the effectiveness of BLUP for large‐scale prediction of yield, moisture, stalk lodging, and root lodging was investigated. Multilocation data from 1990 to 1994 were obtained from the hybrid testing program of Limagrain Genetics. For each of 16 heterotic patterns, the performance of m untested single crosses was predicted from the performance of n tested single crosses as YM = CMP Cpp‐1 YP, where YM = m × 1 vector of predicted performance of the untested single crosses; CMP = m × n matrix of genetic covariances between the untested single crosses and the tested single crosses; Cpp = n × n phenotypic covariance matrix among the tested single crosses; and YP = n × 1 vector of average performance of the tested single crosses, corrected for yield trial effects. Correlations between predicted and observed performance were obtained with a delete‐one cross‐validation procedure. For heterotic patterns with large (>100) numbers of tested single crosses, the correlations ranged from 0.426 to 0.762 for yield, 0.754 to 0.933 for moisture, 0.300 to 0.739 for stalk lodging, and 0.164 to 0.532 for root lodging. The correlations, especially for lodging traits, increased as larger numbers of tested single crosses were available. The results in this study were obtained from large and diverse data sets (600 inbreds, 15 183 data points, and 4099 tested single crosses across 16 heterotic patterns) and provide strong evidence that BLUP is useful for routine identification of superior single crosses prior to field testing.

Bekele, Frances; Bekele, Isaac

doi: 10.2135/cropsci1996.0011183X003600010010xpmid: N/A

The International Cocoa Gene bank, Trinidad, is an international cacao (Theobroma cacao L.) germplasm depository that conserves nearly 2500 accessions in its field collection. A portion of this germplasm was characterized for phenetic diversity with morphological descriptors from the International Board for Plant Genetic Resources cacao descriptor list. Data for 28 quantitative and 26 qualitative descriptors were obtained on 100 accessions representing 24 populations. Associations among the accessions were examined by hierarchical average linkage cluster analysis. Variances of the standardized values were computed for the quantitative descriptors. The diversity and evenness of the qualitative descriptors were assessed with the Shannon‐Weaver diversity index (SWDI). The variances and the Shannon‐Weaver diversity indices summarized the direct contributions of the quantitative and qualitative descriptors to the similarity measure. The variances of the standardized quantitative descriptors ranged from 0.03 for flower ligule length to 0.07 for fruit husk weight. About 75% of the fruit and bean descriptors had variances greater than 0.045, compared to 17% for the flower descriptors, indicating a relatively higher discriminative value of the former. Normalized SWDI values greater than 0.50 were obtained for 69% of the 26 qualitative descriptors. Eighty percent of the flower descriptors had SWDI values greater than 0.50, compared to 60% for those of the fruit and bean. Cluster analysis indicated rich phenetic diversity in this sample of germplasm. At the 75% level of similarity, the accessions were grouped in 11 clusters, each containing two or more accessions. Nine accessions were ungrouped. This diversity should prove useful for breeding programs. The observed link between geographic origin and accession grouping suggested that it is necessary to collect and conserve germplasm representing a broad geographic range.