| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a National Institute of Agricultural Biotechnology, Suwon 441-707, Korea
b current address: Dep. of Crop Science, Chungbuk National Univ., Chongju 361-763, Korea
c Honam Agricultural Research Institute, Iksan 570-080, Korea
d Youngnam Agricultural Research Institute, Milyang 627-130, Korea;. H. Gauch, Crop and Soil Sciences, Cornell Univ., Ithaca, NY 14853-1901
e Dep. of Plant Breeding, Cornell Univ., Ithaca, NY 14853-1901. This research was supported by grants from the Rural Development Administration and the Rockefeller Foundation's Biotechnology Career Fellowship (RF95001, No. 342) for Y.G. Cho's work at Cornell
* Corresponding author (SRM4{at}cornell.edu).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Abbreviations: AMMI, additive main effects and multiplicative interaction • ANOVA, analysis of variance • BR, brown/rough grain ratio • CIM, composite interval mapping • CL, culm length • DTH, days to heading • G x E, genotype x environment • IM, interval mapping • IPC, interactive principal component • LOD, log of the odds • LR, linear regression • PCA, principal component analysis • PL, panicle length • PPP, panicles per plant • PRG, percent ripened grain • QTL, quantitative trait locus • RIL, recombinant inbred line • SPP, spikelets per panicle • TGW, 1000-grain weight • YLD, yield
a National Institute of Agricultural Biotechnology, Suwon 441-707, Korea
b current address: Dep. of Crop Science, Chungbuk National Univ., Chongju 361-763, Korea
c Honam Agricultural Research Institute, Iksan 570-080, Korea
d Youngnam Agricultural Research Institute, Milyang 627-130, Korea;. H. Gauch, Crop and Soil Sciences, Cornell Univ., Ithaca, NY 14853-1901
e Dep. of Plant Breeding, Cornell Univ., Ithaca, NY 14853-1901. This research was supported by grants from the Rural Development Administration and the Rockefeller Foundation's Biotechnology Career Fellowship (RF95001, No. 342) for Y.G. Cho's work at Cornell
* Corresponding author (SRM4{at}cornell.edu).
A population of 164 recombinant inbred lines (RILs) of rice (Oryza sativa L.) derived from a cross between Milyang23 and Gihobyeo was evaluated for nine phenotypic characters over three years and two regions in Korea. The population had been previously mapped using 414 molecular markers. Genotype x environment (G x E) interaction was analyzed for six grain yield-related traits and three agronomic traits across years and locations using the AMMI model. The quantitative trait loci (QTLs) were detected by interval mapping and composite interval mapping. A total of 75 QTLs were identified for the nine traits across five environments and they were categorized as (i) 29 QTLs with main effect, (ii) 18 QTLs with minor effect, (iii) 13 QTLs with G x E interaction effect, (iv) six QTLs with both main effect and G x E interaction effect, and (v) nine potential QTLs with low log of the odds (LOD) scores. The AMMI model explained from 68.6% to 84.7% of the interaction effect and 19 QTLs were significantly associated with G x E interaction. Culm length had the least G x E, while the maximum G x E interaction was exhibited for spikelets per panicle (39.7%) and percent ripened grain (35.3%). Markers closely linked to main effect QTLs will be most useful for marker-assisted breeding.
Abbreviations: AMMI, additive main effects and multiplicative interaction • ANOVA, analysis of variance • BR, brown/rough grain ratio • CIM, composite interval mapping • CL, culm length • DTH, days to heading • G x E, genotype x environment • IM, interval mapping • IPC, interactive principal component • LOD, log of the odds • LR, linear regression • PCA, principal component analysis • PL, panicle length • PPP, panicles per plant • PRG, percent ripened grain • QTL, quantitative trait locus • RIL, recombinant inbred line • SPP, spikelets per panicle • TGW, 1000-grain weight • YLD, yield
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The QTL studies that are conducted over several years and locations provide information about which regions of the genome are consistently identified with target traits. In rice (Oryza sativa L.), approximately 8,000 QTLs have been identified as of December 2006 (www.gramene.org) and a comparison of QTL positions across populations and environments allows researchers to develop testable hypotheses about the behavior of genetic factors underlying the putative QTL.
Many agronomic traits of importance are quantitatively expressed. Quantitative traits are also influenced by environment and tend to show varied degrees of genotype x environment (G x E) interactions (Zhuang et al., 1997; Austin and Lee, 1998; Jiang et al., 1999; Crossa et al., 1999; Hayes et al., 1993; Lu et al., 1996). G x E interaction occurs when genotypes perform differently in various environments. Significant G x E interaction has been reported by comparing QTLs detected in multiple environments (Stuber et al., 1992; Zhuang et al., 1997). In these studies, the appearance of QTLs detected in one environment but not in another was considered to be an indication of G x E interaction. However it has also been shown that QTLs that are stable and consistently detected across environments may still have significant G x E effects (Yan et al., 1998).
It is not always clear whether inconsistent QTL detection is due to the type-II error arising from the use of single thresholds or to true differential trait expression across environments. Using composite interval mapping, Tinker et al. (1996) were able to detect considerable QTL x environment interaction for seven agronomic traits in two barley crosses, even though many of the detected QTLs were highly consistent across environments. Li et al. (2003) reported significant G x E interaction associated with main-effect QTLs for plant height and heading date traits with a rice doubled haploid (DH) population in nine different environments of Asia and quantified main-effect QTLs and epistatic QTLs.
Statistical procedures such as analysis of variance (ANOVA), principal component analysis (PCA), and linear regression (LR) analysis can be used to evaluate genotype and environmental main-effects and G x E interaction. Their limitations have been discussed (Gollob, 1968; Mandel, 1971; Bradu and Gabriel, 1978; Kempton, 1984). The additive main effects and multiplicative interaction (AMMI) model is useful in understanding both main effects and G x E interaction in multi-location variety trials (Zobel et al., 1988; Gauch, 1992; Romagosa et al., 1996; Gauch and Zobel, 1997; Hittalmani et al., 2003; Gauch, 2006a). The AMMI model combines ANOVA for genotype and environment main effects with PCA of the G x E interaction into a single model with additive and multiplicative parameters. It has proven useful for understanding complex G x E interaction (Kang, 1996; Ebdon and Gauch, 2002; Gauch, 2006b). Results can be graphed in a very informative biplot that shows both main and interaction effects for both genotypes and environments.
The AMMI model is particularly useful in understanding G x E interaction and summarizing patterns and relationships of genotypes and environments (Crossa, 1990). During the initial ANOVA the total variation is partitioned into three orthogonal sources, genotypes (G), environments (E) and G x E interaction. Romagosa and Fox (1993) observed that "in most yield trials, the proportion of sum of squares due to differences among sites ranged from 80 to 90% and variation due to G x E interaction was usually larger than genotypic variation." In AMMI analysis, even just the first interactive principle component (IPC1) sum of squares alone is often larger than for G. As genotypes and environments become more diverse, G x E interaction tends to increase and may reach 40 to 60% of total variation. The environmental main effect, which sometimes contributes up to 90% of the total variation, is of interest to soil scientists but only G and G x E interaction are relevant for plant breeders and their selection procedures. The AMMI model can produce graphs (biplots) that focus on the data structure relevant to selection, in other words on the G and G x E interaction (Romagosa and Fox, 1993; Gauch and Zobel, 1997). The PCA portion of the AMMI model partitions G x E interaction into several orthogonal components, so a choice must be made regarding how many components to include in the model, particularly because an ideal choice can gain accuracy (Gauch, 2006a; Ebdon and Gauch, 2002). Gauch and Zobel (1996) reported that the AMMI model, with one or two interaction components, is often most accurate. Gauch (1992) and Cornelius (1993) presented several statistical tests for guiding this choice for each individual data set.
The present study was conducted with a population of 164 recombinant inbred lines (RILs) of rice derived from a cross between Milyang23 and Gihobyeo. The population was evaluated for nine traits over 3 yr in Honam Agricultural Research Institute (HARI) and 2 yr in Youngnam Agricultural Research Institute (YARI) in Korea to determine both main effects and G x E interaction effects of QTLs using the AMMI model. Detecting QTL in regions of the world where rice cultivation is most intensively practiced is essential for developing new varieties based on marker-assisted breeding. Breeders are likely to be more interested in using information about QTLs of agronomic importance when they are detected within the relatively small regions that they are targeting for new varieties. Despite the restricted size of a target environment, environmental variation over years and across locations must be factored into the performance evaluation of a new variety. In this study, we aimed to identify main-effect QTLs and distinguish them from QTLs showing large G x E variation. This allows plant breeders to more efficiently target marker-assisted strategies for plant improvement.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Environments
The MG RI population was evaluated in two locations in Korea: Honam Agricultural Research Institute (HARI, abbreviated "H") for 3 yr (1995–1997) and Youngnam Agricultural Research Institute (YARI, abbreviated "Y") for 2 yr (1996–1997). HARI is located at Iksan (36°N, 126.5°E), which is in the lowland region of Southwestern Korea with an average rainfall of 1,058 mm yr–1, an average summer (May–September) temperature of 22.4°C and a relative humidity of 74.4% during the rice growing season. YARI is located at Milyang (35 °N, 128.5 °E), in the southern part of the Korean peninsula with an average rainfall of 1,103 mm yr–1 and a summer temperature of 22.0°C and a relative humidity of 73.6%
Field Trials and Trait Evaluation
The MG RI lines and their parents, Milyang 23 and Gihobyeo, were cultivated in 10 rows with two replications at both HARI and YARI. The seeds of the RI lines and their parents were planted in the seedling nursery in both locations in mid-May and the 30-d-old seedlings were transplanted in mid-June. Each row consisted of 30 plants with a spacing of 15 cm between the plants and 30 cm between rows. Fertilizer (N-P2O5–K2O) was applied at a rate of 110, 70, and 80 kg ha–1, respectively (Kang et al., 1998).
At maturity, 10 plants for each RI line and both parents were evaluated for panicles per plant (PPP), spikelets per panicle (SPP), 1000-grain weight (TGW), percent ripened grain (PRG), brown/rough grain ratio (BR), culm length (CL), panicle length (PL), and days to heading (DTH). Evaluation was similar to that described in Kang et al. (1998). Panicles per plant were the average number of panicles. Spikelets per panicle were measured as the average number of filled spikelets per panicle. 1000-grain weight was the average weight of 1000 filled spikelets, measured in grams, averaged over three samples taken from bulk harvested grain. Percent ripened grain was the number of filled spikelets divided by the total number of spikelets per panicle. Days to heading were evaluated as the average number of days from seeding until 10% of the panicles had headed. Culm length was measured as the average length in centimeters from the soil surface to the panicle tip of the main tiller. Panicle length was measured as the average number of centimeters from the panicle neck to the panicle tip (excluding the awn). Yield per plant was the average weight per plant of bulked harvested grain measured in grams for 25 plants and calculated on a per area basis (0.1 ha).
Additive Main Effects and Multiplicative Interaction Analysis for Main and Interaction Effects
The AMMI model is a powerful hybrid statistical model that analyzes both main and interaction effects for a two-way data structure (Gauch, 1992; Romagosa et al., 1996). From the ANOVA, the AMMI model first accounts for the additive main effects and then applies PCA to the interaction (residual) portion from the ANOVA to extract a new set of co-ordinate axes, that summarize the interaction patterns. The AMMI analysis was performed by MATMODEL Version 2.0 (Gauch, 2002). The AMMI model is:
 |
g = mean deviation of the genotype g; ße = mean deviation of the environment mean e; 
n = the singular value for IPC axis n; 
gn = the genotype g eigenvector value for IPC axis n; 
en = the environment e eigenvector value for IPC axis n; 
ge = the residual for phenotype g in environment e; and 
ger = the error for genotype g in environment e for replicate r. G x E interaction was analyzed using AMMI analysis (Crossa, 1990; Gauch, 1992) to assess similarity and dissimilarity among five environments and interaction patterns between genotypes and environments.
QTL Analysis
The QTL analysis was performed by interval mapping (IM) and composite interval mapping (CIM) with automatic cofactor selections by a forward/backward regression (forward P < 0.01, backward P < 0.01) using QTL Cartographer v2.0 (Basten et al., 1997). To determine the empirical significance thresholds (Churchill and Doerge, 1994) for declaring putative QTLs, we used QGene 3.06y and QTL Cartographer 2.0 to calculate the empirical log of the odds (LOD) and p = 1 significance levels. The odds ratio is the likelihood that two markers are linked divided by the likelihood that they are not linked. The QTLs that were at or above the 0.05 significance threshold for one or more environments are reported as putative QTLs and those that were at or above the 0.1 significance threshold for two or more environments are put in parenthesis as potential QTLs. Correlation analyses were performed using QGene 3.06y (Nelson, 1997). The QTLs were categorized as (i) QTLs with main effect that were identified over two or more environments, (ii) QTLs with minor effect that were detected in only one environment, (iii) QTLs with G x E interaction effect that were detected from IPC1 & 2, (iv) QTLs with both main effect and G x E interaction effect, or (v) potential QTLs with LOD scores just above the 0.05 significance threshold.
QTL Nomenclature
Nomenclature for QTLs was as described in (McCouch et al., 1997) where a two or three letter abbreviation is followed by the number of the chromosome on which the QTL is found and a terminal suffix, separated by a period, providing a unique identifier to distinguish multiple QTL on a single chromosome.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
Two of the six QTLs detected for the grand mean, ppp1.1 and ppp12.1, were identified across years in the same location (Honam), indicating a stable but narrow adaptation to this location. These QTLs might be useful for a breeding program that targeted the specific region around Honam. In particular, ppp1.1 has a high LOD value (5.24) and high R-square value (16.97%), so it would lend itself to selection using linked markers (i.e., RG140 and RM243) in a breeding program. ppp9.1 is likely to be less useful because it was not consistently detected across years or locations. When comparing QTL results across the Oryza genus, nine QTLs out of 12 were mapped onto similar locations in different studies (Table 5 ), indicating that genes in similar locations along the rice chromosomes may affect the same traits in different genetic populations and different environments.
|
spp1.2 (LOD 7.47; 15.87% variation) and spp1.4 (LOD 5.64; 9.73% variation) could be useful in the Honam region, while spp1.1 (LOD 5.95; 13.13% variation) would be widely adapted for both environments. All of these QTLs were located on similar positions of chromosomes with other results (Table 5), especially spp1.1, spp1.4, spp2.1, spp4.2 and spp8.1, which are reported in two or more previous studies, showing strong agreements of putative QTLs (Brondani et al., 2002; Hittalmani et al., 2003; Mei et al., 2003; Thomson et al., 2003; Xiao et al., 1995,1998; Xiong et al., 1999; Zhuang et al., 1997, 2002).
1000-Grain Weight
Eleven QTLs on eight chromosomes were associated with 1000-grain weight (TGW) (Table 4, Fig. 2 and Fig. 3). Of them six QTLs were detected in more than two environments showing main effect, especially tgw2.1, tgw2.2, tgw6.1, and tgw8.1 were identified in four or more environments with high LOD scores of 5.67–9.60. Two QTLs, tgw1.1 and tgw4.1 on chromosome 1 and 4, were detected for IPC1 as QTLs with G x E interaction effect. Overall, 10 putative QTLs were identified and one potential QTL was detected for 1000-grain weight. The phenotypic variation explained by individual QTLs ranged from 5.81 to 18.85%.
Interestingly, 1000-grain weight was increased by the Japonica type parent, Gihobyeo, at 9 QTLs, while three QTLs, tgw8, tgw12.1, and tgw12.2 have increased effects from the maternal parent, Milyang23. tgw2.1, tgw2.2, tgw6.1, and tgw8.1 were identified in four or more environments by both IM and CIM methods and have high LOD scores and R2 values, thus providing high confidence for MAS in breeding programs. The QTL mapping information provides a starting point to clone genes underlying specific QTL. For example, the grain-weight QTL, Xie et al. (2006) fine mapped a grain weight QTL on chromosome 8 and narrowed the region containing the gene(s) to 306 kb (
1.2 cM) and Li et al. (2004) fine mapped a grain weight QTL on chromosome 3 to 94 kb and the gene underlying this QTL was later cloned by Fan et al. (2006).
Percent Ripened Grain
Six QTLs on three chromosomes were identified for percent ripened grain (PRG) (Table 4, Fig. 2 and Fig. 3). Three were detected in more than two environments showing main effect, but prg2.2 was detected for three environments, grand mean, and IPC1 as QTLs with main and G x E interaction effect. Overall, five putative QTLs were identified and one potential QTL was detected for PRG. The phenotypic variation explained by individual QTLs ranged from 6.90 to 18.48%. prg1.2 (8.58), prg2.2 (5.59), and prg9.1 (5.19) had very high LOD scores with high levels of percent variation ranging from 12.16 to 18.48.
Yield
Five QTLs on five chromosomes were associated with grain yield (YLD) (Table 4, Fig. 2 and 3). Two were detected in more than two environments as QTLs with main effect. Locus, yld1.1 between RM5 and RG462 showed a good agreement with results from a wild rice relative, Oryza rufipogon (Xiao et al., 1998). In our present results, Milyang23 alleles were associated with YLD increases at four of these loci, while the Gihobyeo allele, yld6.1, was linked with a YLD decrease. Locus, yld7.1 was identified for IPC1 showing G x E interaction effect. Two potential QTLs, yld6.1 and yld8.1, were detected for YLD. The positions of four QTLs on chromosome 1, 6, 7, and 9, yld1.1, yld6.1, yld7.1 and yld9.1, coincide with other reports (Brondani et al., 2002; Hittalmani et al., 2003; Li et al., 2000; Septiningsih et al., 2003; Thomson et al., 2003; Xiao et al., 1998; Xiao et al., 1995) (Table 5). The phenotypic variation explained by individual QTLs ranged from 8.41 to 15.94%.
A QTL on the short arm of rice chromosome 1 that increases grain productivity in rice was recently cloned by Ashikari et al. (2005) and shown to be cytokinin oxidase/dehydrogenase (OsCKX2), an enzyme that degrades the phytohormone cytokinin. This gene does not map to any of the yield QTL identified in the present study.
Brown/Rough Grain Ratio
Six QTLs on five chromosomes were identified for brown/rough grain ratio (BR) (Table 4, Fig. 2 and 3). br4.1 was detected in two environments as a QTL with main effect and br1.1 was identified for IPC1 showing G x E interaction effect. Two potential QTLs, br6.1 and br6.2, were detected for BR. The phenotypic variation explained by individual QTLs ranged from 5.73to 18.74%.
Days to Heading
Nine QTLs on seven chromosomes showed significant association with days to heading (DTH) (Table 4, Fig. 2 and 3). Five QTLs, dth1.1, dth3.1, dth6.1, dth7.1, and dth8.1, were identified in two or more environments, especially dth1.1, dth3.1, dth6.1, and dth7.1 were identified in four or more environments with high LOD scores of 10.37 to 14.63 (except dth1.1– 3.42). But dth3.1, dth6.1, and dth7.1 were also detected for grand mean and IPC1 as QTLs with main effect and G x E interaction effect, having relatively high phenotypic variations ranged from 17.16 to 30.20%. The QTLs, dth3.1, dth6.1, and dth7.1, were mapped in a similar position in several different studies (Hittalmani et al., 2003; Lin et al., 1998, 2000; Maheswaran et al., 2000; Moncada et al., 2001; Septiningsih et al., 2003; Thomson et al., 2003; Xiao et al., 1998; Xiong et al., 1999; Yano et al., 1997) (Table 5), thus showing wide adaptation across various populations and environments. The markers bordering the QTLs are likely to be useful for selecting favorable lines by MAS. The overall phenotypic variation explained by individual QTLs ranged from 6.42 to 30.20%. Gihobyeo alleles dth6.1 and dth9.1 were associated with decreases in days to heading by 5.5 and 1.8 d, respectively.
Quantitative trait loci that have high phenotypic main effects can be efficiently targeted based on fine mapping with nearly isogenic lines. Recently, QTLs related to flowering time have been targeted for cloning. Three genes, Hd6, Hd1, and Hd3a on chromosomes 3 and 6, that are involved in the control of flowering time of rice in response to changes in daylength have been identified by map-based cloning (Kojima et al., 2002; Takahashi et al., 2001; Yano, 2001).
Culm Length
Four QTLs on four chromosomes were identified for culm length (CL) (Table 4, Fig. 2 and 3). Of which three, cl1.1, cl7.1, and cl8.1, were detected in three or more environments as QTLs with main effect and cl4.1 was identified for IPC2 as a QTL with G x E interaction effect. The phenotypic variation explained by individual QTLs ranged from 4.58 to 61.07%. Two QTLs in particular, cl1.1 and cl6.1, were significant across five environments, indicating that closely linked markers are likely to be useful for MAS in breeding for plant height. cl1.1 with a LOD of 40.62 explained 61.07% the phenotypic variation, the highest for the 75 QTLs detected in this study, thus behaving almost like a single genic effect. The chromosomal position of this QTL coincides with the major semi-dwarf gene sd-1, associated with the green revolution (Cho et al., 1994; Ashikari et al., 2002; Monna et al., 2002; Sasaki et al., 2002; Spielmeyer et al., 2002) (Table 5).
Panicle Length
Twelve QTLs were significantly associated with panicle length (PL) (Table 4, Fig. 2 and 3). Of which only three, pl3.1, pl6.1, and pl8.2, were detected in two or more environments as QTLs with main effect, especially pl3.1 was identified in five environments and the grand mean, showing that the markers bordering it are likely to be useful for panicle length selection. pl7.2 and pl8.1 were identified for IPC1 and IPC2, respectively as QTLs with G x E interaction effect. However, pl12.1 was detected for both grand mean and IPC1 as a QTL with main effect and G x E interaction effect. The phenotypic variation explained by individual QTLs ranged from 6.91 to 15.67%.
Implications for Plant Breeders
The ability to use markers associated with QTLs of interest for MAS or map-based cloning requires that QTLs showing genotypic main effects can be discriminated from QTLs with a sizable G x E interaction. This study identified many QTLs that are stable across environments in the two different southern agricultural regions in Korea and evaluated the percent of variance explained by each so that markers closely associated with useful alleles could be used to trace the inheritance of specific chromosomal segments in a segregating population.
Nine QTLs, tgw2.1, tgw6.1, and tgw8.1 for TGW; dth1.1, dth3.1, and dth7.1 for DTH; cl1.1 and cl6.1 for CL; and pl3.1 for PL, were significant in all years in both environments as QTLs with main effect and were also detected for the grand mean. Markers delineating these QTLs are likely to be very useful for practical plant breeding using MAS.
The Korean peninsula is a temperate rice growing region. There are two major areas where rice is intensively cultivated, one in the southwestern part (Iksan, 36°N,126.5°E) and one in the southeastern part (Milyang, 35 °N,128.5 °E). Information about performance-enhancing QTL from these regions offers breeders a useful set of markers for developing new varieties that can be widely adapted in these agricultural areas. Other nations that share similar quality and agronomic concerns may also make use of this information to facilitate breeding of improved varieties. Extended analysis of the RILs used in this study may offer additional insights into the stability of QTLs in new environments and help increase the efficiency of variety development.
This information can also be used as a starting point for positional cloning aimed at isolating genes of interest. As the genes underlying the QTLs are identified, and the interactions between genes and their environments (G x E) as well as among genes within a genotype (G x G) are better understood, more accurate predictions about the genotype–phenotype relationship will enable plant breeders to make more informed decisions about useful parental combinations and more efficient selection among recombinants.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Received for publication August 2, 2006.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. G. Gauch Jr., H.-P. Piepho, and P. Annicchiarico Statistical Analysis of Yield Trials by AMMI and GGE: Further Considerations Crop Sci., May 1, 2008; 48(3): 866 - 889. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
