Detection of quantitative trait loci for grain yield and yield components in maize across generations in stress and nonstress environments.
Reductions in grain yield of maize have also been associated with excess moisture stress (Mukhtar et al., 1990; Kanwar et al., 1988; Joshi and Dastane, 1966). Evaluations of [F.sub.2:3] progeny have identified QTL controlling morphological (Veldboom and Lee, 1996b) and grain yield traits (Veldboom and Lee, 1996a) in nonstress and stress environments. For the morphological and grain yield traits, about 50% of the QTL were detected in both environments. Austin (1997), evaluating the same population at the [F.sub.6:7] generation in more stressful conditions, detected 35 % of the QTL for morphological traits in both the stress and nonstress environments. These studies suggest that environmental factors can greatly affect the perception of QTL.
Genotype x environment interactions (G x E) are commonly observed for quantitative traits. One possible explanation for G x E would be that different QTL or alleles at the same QTL are responsible for genetic variation under diverse environmental conditions. Few QTL were consistently detected across environments for morphological traits in tomato, Lycopersicon esculentum Mill. (Paterson et al., 1991), for response to the disease gray leaf spot (Cercospora zeae-maydis Tehon & Daniels) in maize (Bubeck et al., 1993), for plant height and lodging in soybean, Glycine max (L.) Merr. (Lee et al., 1996), and agronomic traits in rice, Oryza sativa L. (Lu et al., 1997). In contrast, studies in maize have shown relatively consistent detection of QTL across diverse environments for grain yield (Stuber et al., 1992), morphological traits (Sch6n et al., 1994), second generation European corn borer resistance (Sch6n et al., 1993), and northern corn leaf blight [Setosphaeria turcica (Luttrell) K.J. Leonard & E.G. Suggs] resistance (Freymark et al., 1993; Dingerdissen et al., 1996). These differences in consistency may be attributable to the biology of the traits, progeny and environment sampling variation, QTL detection methods, and threshold levels (Beavis, 1994).
In a marker-assisted breeding program, desirable genotypes are those with consistent performance across a set of environments representative of the target environment. Identification and selection for QTL with consistent relative effects would hopefully help to attain such stable performance. Also, QTL with inconsistent effects across environments might be the genetic factors responsible for G x E. These QTL could then serve as targets for future studies to increase our understanding of the genetic and physiological basis for interaction of the phenotype with the environment (Beavis and Keim, 1996).
In the present study, grain yield and four yield components were investigated in a [F.sub.6:7] population derived from a cross between inbreds Mol7 and H99. The first objective of our study was to compare QTL detection in two climatically diverse (stress and nonstress) environments. The stress environment was defined by 128% above average rainfall during the growing season (MaySeptember) and a 56% reduction in grain yield relative to the nonstress environment. QTL were identified individually in each environment to determine differences in environmental response of genetic factors controlling trait variation. The second objective was to compare QTL detection in the [F.sub.6:7] with results from the [F.sub.2:3] generation from the same population grown at the same location but in different years. Recombinant inbreds (RIs) represent a permanent mapping population with advantages over other mapping populations for the detection of QTL (See Austin and Lee, 1996a, for review). Previous studies in maize (Austin and Lee, 1996a,b) and tomato (Goldman et al., 1995) have demonstrated the effectiveness of RIs in detecting a greater number of QTL than [F.sub.2] derived populations.
MATERIALS AND METHODS
The formation of the mapping population, collection of the restriction fragment length polymorphisms (RFLP) and simple-sequence repeats (SSR) data, field design, and procedures for field data analysis have been described in detail (Austin and Lee, 1998). Herein, field trials were conducted in 1993 and 1994 to evaluate 185 [F.sub.6:7] lines derived from a cross between maize inbreds Mol7 and H99. The lines and parental checks were evaluated in two replications of a 14 by 14 lattice design of one-row plots at the Agronomy and Agricultural Engineering Research Center near Ames, IA, in 1993 and 1994. Row length and width were 5.5 m and 0.76 m, respectively. Plots were machine-planted on 15 May 1993 and on 3 May 1994 at a density of 76 540 kernels [ha.sup.-1] and thinned to 57 400 plants [ha.sup.-1] (24 plants plot-1) at the six-eight leaf stage. Grain yield and yield components were measured on a plot basis as follows: grain yield (GY) is the total weight (g) of shelled grain converted to Mg [ha.sup.-1]. Kernel weight (KWT) is the weight (g) of a 300-kernel sample taken from shelled grain of the plot's total. Ear number per plant (ENP) is the total number of ears harvested from the plot divided by the number of plants in the plot. Ear length (EL) is the average length (cm) of 10 primary (top) ears. Ear diameter (ED) is the average diameter (cm) of 10 primary ears. Plots were harvested by hand. Abbreviations will be used in the proceeding text and tables. Dry conditions in May of 1994 resulted in poor emergence and stand establishment of some entries in both replications. Covariance analysis for GY with stand as a covariate was not effective at reducing the error variance because the variation was due to entries rather than field variation. Therefore, the GY of a plot in 1994 was calculated by dividing the grain weight of the plot by the number of plants in the plot, then multiplying by 24 to give values on a full stand basis.
Statistical analysis of trait data was performed as described by Veldboom et al. (1994). Adjusted entry means were obtained by correcting for incomplete block effects in the lattice design within the 1993 and 1994 trials. The adjusted entry means of the two trials were averaged to give trait values for the mean environment used in the QTL analysis. From the combined analysis of variance across environments, heritabilities ([h.sup.2]) were estimated according to Hallauer and Miranda (1988) with exact 90% confidence intervals calculated according to Knapp et al. (1985). Adjusted entry means were used to calculate phenotypic and genotypic correlations between traits according to Mode and Robinson (1959). Standard errors of the genotypic correlations were calculated as described by Falconer (1989). Phenotypic correlations between the [F.sub.2:3] and the [F.sub.6:7] were estimated by the values from the mean environments from the 143 lines evaluated in both generations.
QTL were determined on the adjusted entry means of the 1993 and 1994 environments and the mean environment (the average of the 2 yr) by the composite interval mapping method (Jansen and Stam, 1994; Zeng, 1994). All computations for this method were performed with the software package PLABQTL (Utz and Melchinger, 1996), which employs interval mapping by the regression approach (Haley and Knott, 1992) with selected markers as cofactors. Details of the underlying model have been previously presented (Bohn et al., 1996; Lubberstedt et al., 1997). Because [F.sub.6:7] lines were evaluated, few heterozygotes were present and an additive model was used for QTL detection. Cofactors were selected by stepwise regression and final selection was for the model that minimized Akaike's information criterion with penalty = 3.0 (Jansen, 1993). The primary interest of this study was to compare QTL main effects across environments and generations, so epistatic interactions were not included in the analysis. To enable comparisons with previous reports based on this population (Veldboom and Lee, 1996a; Veldboom and Lee, 1996b), a LOD threshold of 2.0 was selected for QTL detection. For each QTL, a one-LOD support interval was constructed as described by Lander and Botstein (1989). QTL with nonoverlapping one-LOD support intervals (SI) were considered as different regions.
QTL analyses were also performed on the [F.sub.2:3] generation for GY, KWT, ENP, EL, and ED with mean performance across the 1989 and 1990 environments. These data have previously been evaluated by the simple interval mapping method (Veldboom and Lee, 1996a), but have been reevaluated by composite interval mapping with an additive model to allow direct comparison with the [F.sub.6:7] QTL data. To allow comparison of QTL positions across generations, [F.sub.2:3] QTL positions were adjusted to correspond to the [F.sub.6:7] linkage map on the basis of relative position to the 88 (87 RFLP; 1 morphological, P1) loci common to both generations. One hundred forty-three of the [F.sub.6:7] lines are descendants of the [F.sub.2:3] lines previously evaluated (Veldboom and Lee, 1996a,b).
Phenotypic variance estimates for individual QTL were obtained by the square of the partial correlation coefficient between the QTL and the trait data, keeping all other QTL effects fixed. Estimates of QTL additive effects as well as the total LOD score and phenotypic variation explained by all QTL were obtained by simultaneously fitting a model including all QTL detected for the trait. The proportion of the genotypic variation explained by the final model was estimated by dividing the percentage of phenotypic variation explained by the trait h2 (Ottaviano et al., 1991; Sch6n et al., 1994).
The 1993 and 1994 growing seasons were different in terms of precipitation, accumulation of growing degree days (GDD), and sunlight (Austin and Lee, 1998). Conditions during the 1994 growing season (May-September) were near normal with precipitation 7% below average and GDD accumulation 1% above average. The 1993 growing season (May-September) was the wettest on record with GDD accumulation 7% below average. The period of July to August was especially stressful for late vegetative growth, pollination, and early kernel development. During this period, precipitation totals were three and a half times the normal levels. Low GDD accumulation during grain filling in September (28% below average GDD) also occurred in 1993. The grain yield average in Iowa for 1993 was the lowest since 1974, whereas the 1994 average was the highest on record (National Agricultural Statistics Service, Washington, DC, http://www.nass.usda.gov/). Similar, albeit less severe, stress conditions prevailed in 1989 (nonstress) and 1990 (stress) when the [F.sub.2:3] generation was evaluated (Veldboom and Lee, 1996a). Herein, the [F.sub.6:7] stress environment was defined by precipitation 128% above the average and a 56% reduction in grain yield, whereas the [F.sub.2:3] stress environment had precipitation 55% above the average and a 17% reduction in grain yield relative to the respective nonstress environments and generations.
RESULTS AND DISCUSSION
Analysis of Field Data
The means of the parental inbreds and [F.sub.6:7] lines in 1993, 1994, and the mean environment are shown in Table 1. All trait values of the lines were significantly (P [is less than or equal to] 0.001) lower in 1993. Average GY of the [F.sub.6:7] lines was 1.41 Mg [ha.sup.-1] in 1993 and 3.18 in 1994. Mean values for GY, ENP, KWT, EL, and ED were 56, 30, 25, 21, and 10% lower in 1993. Although excess moisture was recorded during each month of the growing season (May-September), the majority (50% of the total excess) came during July, which corresponded to the pretassel stage because the majority of the lines flowered (pollen shed and silk emergence) during late July and early August. Average precipitation in July is 88 mm, whereas 416 mm were recorded at Ames in 1993. Excess moisture can severely reduce grain yield in maize (Mukhtar et al., 1990; Kanwar et al., 1988; Joshi and Dastane, 1966), and the greatest reductions in grain yield occur when stress due to excess water occurs during the late vegetative, pre-tassel stage (Evans et al., 1990). The morphological traits were much less affected by the stress environment with plant height being nearly equal in the two environments, whereas anthesis and silk emergence were both delayed by 5% in 1993 relative to 1994 (Austin, 1997). This is likely due to the timing of the stress periods, which occurred mostly just before and after pollination.
For all traits, genetic and genotype x environment variance components were highly significant (P [is less than or equal to] 0.001), and estimates of genetic variance were less in the stress environment (Table 1). This observation was also made in the [F.sub.2:3] (Veldboom and Lee, 1996a) and is consistent with theoretical studies (Rosielle and Hamblin, 1981). Values for [h.sup.2] (from combined analysis across environments) were lowest for ENP (0.19) and GY (0.39) and highest for KWT (0.79) and ED (0.80; Table 2). Heritability estimates from the combined analysis of the [F.sub.2:3] generation were greater for all traits, with GY (0.83) and ENP (0.71) having the greatest differences relative to the [F.sub.6:7] generation (Veldboom and Lee, 1996a). Heritabilities in the [F.sub.6:7] (data not shown for each environment) were greater for all traits in the nonstress environment. GY and ENP were the only traits with variance component estimates for [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] larger than [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] indicating a large environmental component of the trait variation (Table 1). Phenotypic correlations among the traits were similar for the two environments; therefore, only the phenotypic correlations across environments are presented (Table 3). Phenotypic correlations between grain yield and each of the four yield components were significant and positive. Similar to previous evaluations of the [F.sub.2:3] generation (Veldboom and Lee, 1996a), GY was most strongly correlated with EL. Genotypic correlations were similar to the phenotypic correlations.
[TABULAR DATA 1 to 3 NOT REPRODUCIBLE IN ASCII]
In the combined analysis, the parental checks (Mo17 and H99) were significantly (P [is less than or equal to] 0.001) different for KWT and EL with Mo17 having greater values (Table 1). Mo17 also had values greater (non-significant) than H99 for GY, ENP, and ED. The non-significant differences between the parents in the combined analysis for GY and ED are likely due to the low differences observed in 1993 (nonsignificant), whereas in 1994 the differences were highly significant (P [is less than or equal to] 0.001) and consistent with previous observations (Veldboom and Lee, 1996a). The parents were not significantly different for ENP in either of the individual environments. In the combined analysis, transgressive segregation exceeding high and low parental values was observed for all traits. The population means were near the mid-parent values for all traits.
The additive effects and parental contributions for each [F.sub.6:7] QTL detected in the 1993 (low-yield), 1994, (high-yield), and mean environments are presented in Table 4. A total of 67 QTL were detected (map region x trait combinations) for the five traits over each environment and the mean environment. Of the 67 QTL, 60% (40) were detected in the mean environment, 54% (36) were detected in the stress environment, and 49% (33) were detected in the nonstress environment. Nine QTL (13%) were detected across the stress, nonstress, and mean environments. In contrast, 36 of 120 (30%) QTL for the morphological traits were observed in both environments and the mean (Austin, 1997). In general, the yield-related traits had lower [h.sup.2] values, a larger ratio of [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]/ [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], and the population means were more affected in the stress environment than the morphological traits. These factors are likely responsible for the lower consistency of QTL detection across the stress and nonstress environments for grain yield and yield components.
[TABULAR DATA 4 NOT REPRODUCIBLE IN ASCII]
Because of the reduced standard error of trait values, the mean environment should be the most efficient for the identification of QTL (Knapp and Bridges, 1990). Previous studies in this population at the [F.sub.2:3] (Veldboom and Lee, 1996a; Veldboom and Lee, 1996b) and [F.sub.6:7] generations (Austin, 1997) have shown the mean environment to be the most efficient for the detection of QTL. Herein, 40 QTL were detected in the mean environment, slightly more than either the stress or nonstress environments (Table 2). The mean environment also should allow the detection of QTL effects too small to be detected in the individual environments. This was observed with eight QTL being detected in the mean environment that were not detected in either the stress or nonstress environments. These included two QTL for GY (3L and 5S), two QTL for KWT (2S and 3S), two QTL for ENP (1L and 6L), and two QTL for EL (1L and 9S). Of the eight QTL detected only in the mean environment, six were among the QTL with the smallest additive effects for their respective traits. The two exceptions, KWT QTL on 2S and 3S, had the eighth and fifth largest effects of 13 QTL detected in the mean environment for KWT. When a QTL was detected in the stress and nonstress environments, it was nearly always detected in the mean environment. Of the 10 QTL detected in both years, nine were also detected in the mean environment. Similar effectiveness of the mean environment was observed for the morphological traits where 92% of QTL detected across the stress and nonstress environments were also detected in the mean environment (Austin, 1997). QTL unique to one of the two environments were also represented in the mean environment. Eleven QTL detected in 1993 but not 1994, and 12 QTL detected in 1994 but not 1993 were detected in the mean environment.
Over all live traits, 59 QTL were detected in at least one of the two environments. The number of QTL detected in both environments ranged from zero for EL to four for KWT (Table 2). This suggests that effects of the majority (83%) of the QTL could be unique to their environment. Similar results were also observed for morphological traits with 65% (71 of 110) of the QTL being detected in only one of the two environments (Austin, 1997). However, the possibility of sampling variation cannot be dismissed, and such differences between environments could be the result of QTL detected in one environment being a subset of the total for the trait and population (Beavis, 1994). For the 10 yield-related QTL detected in both environments, the parental sources of the additive effects were the same. For eight of those 10 QTL, the additive effects were smaller in the stress environment, corresponding to the lower trait means observed in 1993. Such consistent parental contributions were also observed in this population at the [F.sub.6:7] generation for morphological traits (Austin and Lee, 1998), at the [F.sub.2:3] generation for morphological and grain yield traits (Veldboom and Lee, 1996a, 1996b), and in studies evaluating nonstress and drought stress environments for flowering traits (Ribaut et al., 1996) and grain yield (Beavis and Keim, 1996). Observations of stable QTL are promising for the potential application of marker-assisted selection.
For GY, six QTL were detected in at least one of the two environments (two QTL detected in the mean environment only). Four QTL were detected in 1993 and three in 1994 with one common between the two environments. The common QTL (near npi280 on 6L) had the largest effect in the nonstress environment (0.48 Mg [ha.sub.-1], Mo17) and the third largest effect (0.09 Mg [ha.sub.-1], Mo17) in the stress environment. The additive effect was 81% smaller in the stress environment, corresponding to the 56% smaller population mean. The alleles with the largest effects in the stress environment (near umc21 on 6L) were derived from H99 and were not detected in the nonstress environment. This contribution of H99 was surprising because Mo17 had a significantly higher GY than H99 in the stress environment. H99, however, seems to be less responsive to the stressful conditions. The GY of Mo17 was 51% lower in the stress environment than in the nonstress environment, similar to the 56% decrease in the population mean. In contrast, H99 GY was 37% lower in the stress environment.
Environment also seems to play a role in the ability to detect and in the relative magnitude of specific QTL for the four yield components. It is expected that a reduction in the genetic variance would limit the number of QTL that could be detected (Lande and Thompson, 1990). All yield components had lower genetic variances in the stress environment and more QTL were detected for ENP and EL in the nonstress environment. In contrast, more QTL were detected for KWT and ED in the stress environment. For ENP, QTL with the largest effects in both the stress and nonstress environments were detected on 3L near isu1. However, another ENP QTL detected in the stress environment on 6L had the same size effect and was not detected in the nonstress environment. For KWT, the QTL with the largest effect in 1993 (near bnl9. 11 on 8C) had the fourth largest effect in 1994. The KWT QTL with the largest effect in 1994 located on 2S (near umc5S) was not detected in 1993. For both EL and ED, the QTL with the largest effects in the stress and nonstress environments were not detected in the other environment. The LOD threshold (2.0) utilized herein was fairly low, and the presence of QTL in only one environment could indicate false positives. However, with one exception, QTL with the largest effect in one environment that were not detected in the other environment had LOD scores greater than 3.0 (data not shown) indicating that they are likely not false positives. In nearly all instances, the QTL with the largest effect in the stress or nonstress environment was observed in the mean environment. The only exception was the KWT QTL in 1994 on 2S. A QTL was detected in the mean environment in this region, but the one-LOD SI did not overlap.
On the basis of the results from this population for grain yield and yield components, for morphological traits (Austin, 1997), and from an earlier generation (Veldboom and Lee, 1996a,b), detection of QTL in one environment was not representative of QTL in the other environment. Thus, the mean environment seems to provide the best representation of QTL with larger and/ or consistent effects.
Because the majority of the QTL detected herein are observed in the mean environment, these QTL and their effects have been presented in Table 5 and Fig. 1. GY and ENP had the fewest QTL (five and four, respectively) in the mean environment and the lowest phenotypic variation associated with DNA marker loci, 27% for both traits (Table 2). However, the two traits had the highest percentage of genotypic variation explained by their low heritabilities (% genotypic explained = % phenotypic explained/[h.sup.2]). KWT had the most QTL detected (13), whereas ED had the largest percentage of the phenotypic variation explained (49%). Mol 7 alleles contributed a positive effect at 26 of the 40 (65%) QTL detected in the mean environment (Table 5). For GY and KWT, the alleles with the largest effects were derived from Mo17. For ENP, EL, and ED, H99 alleles had the largest effects. For EL and ED, the QTL with the largest effects in the mean environment were also detected in the stress environment but not in the non-stress environment. As mentioned previously, H99 is less responsive and perhaps more tolerant to the stress environment for GY. H99 has also been observed to be tolerant of stress related to water deficits (W.A. Russell, 1988, pets. comm.). Because ED and EL are the two components with the highest phenotypic correlations with GY (Table 3), it is not surprising to observe H99 alleles with major effects in the stress and mean environments.
[TABULAR DATA 5 NOT REPRODUCIBLE IN ASCII]
[Figure 1 ILLUSTRATION OMITTED]
The contribution of positive alleles from both parents likely contributes to the transgressive segregation observed for traits in this population (deVincente and Tanksley, 1993). Supporting evidence is provided by evaluating the parental composition at loci associated with GY QTL for the five lines with the highest and lowest GY in the mean environment (data not shown). Mol 7 alleles had the largest additive effects in the mean environment (6L near npi280). The five highest yielding lines were homozygous Mo17 at npi280, whereas all five of the lowest yielding lines were homozygous H99. Alleles from H99 had the third largest effect (IL near npi429). The five highest yielding lines were homozygous for the H99 allele at npi429, whereas the lowest yielding lines were homozygous for the Mo17 allele. Summarized over all five GY QTL, the highest yielding lines had four or five of the favorable genotypes whereas the lowest yielding lines had only none to two of the favorable genotypes.
Grain yield is influenced by a plethora of genetic and physiological networks which respond to environmental cues throughout the growing season. Also, differences in traits such as maturity, disease and insect resistance, and stress tolerance all can greatly affect final grain yield. Herein, four yield components had positive correlations with grain yield (Table 3). In the mean environment, yield component QTL, most often EL and ENP, were detected in the same regions as four of the five GY QTL locations (Fig. 1). GY QTL located on 1L (EL, ED), 3L (ENP, EL), 6L (ENP, EL), and 8L (ENP, ED) all had SI overlapping yield component QTL with the same parental contributions. Of the yield components, KWT had the lowest correlation ([r.sub.p]: 0.19), and no QTL locations were common with GY. Phenotypic correlations with GY for the remaining three yield components ranged from 0.43 to 0.59, and each component shared two QTL locations with GY. For KWT, EL, and ED, more QTL were detected per trait in the mean environment than for GY. Each component may have an effect on GY, but they may be too small to be distinguished from experimental error in this type of study. GY also had highly significant (P [is less than or equal to] 0.001) phenotypic correlations with morphological traits evaluated in this population (Austin and Lee, 1998) including plant height ([r.sub.p]: = 0.28), anthesis ([r.sub.p] = -0.35), silk emergence ([r.sub.p] = -0.46), and anthesis-silk interval ([r.sub.p] = -0.40). Plant height and GY shared QTL locations on 3L, 6L, and 8L with the same parental allele having a positive effect on both traits, corresponding to their positive correlation. Anthesis (2L, 5S, 8L), silk emergence (1L, 5S, 6L), and anthesis-silk interval (3L, 5S, 6L) each had three regions in common with GY QTL. At each region, the parental allele increased or decreased trait values in agreement with their negative correlations with GY. Genetic explanations for multiple trait correlations include pleiotropic effects or the presence of linked QTL controlling different traits, and these two possibilities cannot be resolved in this type of study.
Comparison of QTL Detection--[F.sub.2:3] and [F.sub.6:7]
Table 5 presents a comparison of QTL detected in the mean environment for the [F.sub.6:7] (1993 and 1994) and [F.sub.2:3] (1989 and 11990) generations of this population. Typical of most breeding programs, the generations were evaluated at the same location but in different years. Phenotypic correlations between the [F.sub.2:3] and [F.sub.6:7] generations were significant (P [is less than or equal to] 0.01) for all traits (Table 3). Grain yield had a correlation of 0.26, and the four yield components had correlations greater than 0.38. The presence of significant correlations between the trait values in the two generations indicates some level of consistency in the relative performance of the lines.
Over all five traits, the same number (40) of QTL were detected in the [F.sub.2:3] and [F.sub.6:7] generations. More QTL were detected in the [F.sub.2:3] for GY and ENP, whereas more QTL were detected in the [F.sub.6:7] for KWT, EL, and ED. The [F.sub.6:7] generation should be more efficient and powerful for QTL detection because of increased homozygosity, homogeneity, and increased recombination for separation of linked QTL (for review see Austin and Lee, 1996a). For morphological traits in this population, 33% more QTL were detected in the [F.sub.6:7] than the [F.sub.2:3] generation (Austin, 1997; Austin and Lee, 1998). For the morphological traits, [h.sup.2] values were of similar magnitude across the generations. Herein, [h.sup.2] values for GY (0.39) and ENP (0.19) in the [F.sub.6:7] evaluations were much lower than those observed in the [F.sub.2:3] (0.83 and 0.71, respectively; Veldboom and Lee, 1996a). The lower [h.sup.2] values may be attributed to the higher degree of stress of encountered during the [F.sub.6:7] evaluations (56% lower GY in 1993 than 1994) and the greater response of highly inbred maize genotypes to environmental variation. Higher [h.sup.2] values, expected to enhance detection of QTL associated with larger portions of the genetic variance (Lande and Thompson, 1990), could explain the greater number of GY and ENP QTL in the [F.sub.2:3] despite the advantages of recombinant inbred ([F.sub.6:7]) lines for identification of QTL.
Three of the eight GY QTL detected in the [F.sub.2:3] generation were detected in the [F.sub.6:7] generation on the basis of overlapping SI: on 6L (near npi280, from Mo17) with the largest additive effect in both generations, on 8L (umc48-npi268, from Mo17) with the second largest effect in the [F.sub.6:7] and the fourth largest effect in the [F.sub.2:3] generation, and on 3L (near bnl3.18, from H99). The QTL on 6L explained 31 and 13% of the phenotypic variation in the [F.sub.2:3] and [F.sub.6:7] generations, respectively. The additive effect in the [F.sub.6:7] generation (0.24 Mg ha-1) was 76% smaller than in the [F.sub.2:3] (0.99 Mg ha-1). Some of this difference is likely due to the lower GY average (48% lower) observed for the F6:7] generation (2.31 Mg ha-1) than the [F.sub.2:3] generation (4.43 Mg ha-1). Overestimation of the effects in the [F.sub.2:3] generation and/or the effects of additional rounds of recombination observed with recombinant inbreds could also contribute to this difference (Lee, 1995; Darvasi and Soller, 1995, Burr et al., 1988). In both generations, the QTL on 6L had smaller effects in the stress environments. In the [F.sub.2:3], the QTL had a 34% smaller effect in the stress (1990) environment (Veldboom, 1996a). In the [F.sub.6:7], the effect of that QTL on GY in the stress environment (1993) was 81% smaller than in the nonstress environment (1994; Table 4). These reductions were greater than the reductions in the population mean in the stress environment for the [F.sub.6:7] (56%) and [F.sub.2:3] (17%) indicating that the QTL effects could be interacting with the environment. In this population, 6L seems to have a major effect on GY and was detected across generations and environments. Previous studies in maize have also reported major effects for GY on 6L (Edwards et al., 1992; Stuber et al., 1992). In contrast, Beavis et al. (1994) did not detect any GY QTL in this region with inbred progeny of a population derived from inbred lines B73 and Mo17. Although direct comparisons are complicated by different marker loci and QTL detection methods, it would seem that the significance of this QTL is limited to certain populations.
For the yield components, 10 of the 32 QTL detected in the [F.sub.2:3] generation were detected (on the basis of overlapping SI) in the [F.sub.6:7] with the same parental contributions. KWT had the most QTL detected in the mean environments of the [F.sub.2:3] (10) and [F.sub.6:7] (13) generations; however, only two QTL seem to be common across the generations. The KWT QTL with the largest effect in the [F.sub.2:3] had the second largest effect in the [F.sub.6:7] (5L, bnl10.06-bnl7.71). The QTL with the largest effect in the [F.sub.6:7] (near bnl9.11 on 8S), however, was not detected in the [F.sub.2:3]. ENP, with the lowest [h.sup.2] in both generations, had two QTL common to both generations. This included the QTL with the largest effect for both generations (3L, umc165-bn13.18). The positive alleles on 3L were derived from H99 and explained 27% ([F.sub.2:3]) and 17 % ([F.sub.6:7]) of the phenotypic variation with nearly identical additive effects of 0.11 ([F.sub.2:3]) and 0.10 ([F.sub.6:7]). In the [F.sub.6:7], this QTL had the largest effect in both the stress and nonstress environments (Table 4), whereas it had the largest effect in the stress and second largest effect in the nonstress environments of the [F.sub.2:3] generation (Veldboom and Lee, 1996a). Thus, the ENP QTL on 3L seems to be fairly stable across environments and generations. For EL, three QTL were common across generations (1S, 1L, 6L). On 6L (near npi280), H99 alleles had the largest effect in the [F.sub.2:3] and fourth largest effect in the [F.sub.6:7]. However, the EL QTL with the largest effect in the [F.sub.6:7] on 3L was not detected in the [F.sub.2:3]. On IS, two linked EL QTL (43 centimorgan [cM]) were detected in the [F.sub.2:3]. Alleles of the distal QTL were derived from H99 while those of the proximal QTL were from Mo17 (second largest effect). In the [F.sub.6:7], Mo17 alleles for EL were detected midway between those two QTL of the [F.sub.2:3]. The [F.sub.6:7] QTL SI overlapped that of the distal [F.sub.2:3] QTL by 2 cM. It is possible that this represents a cross-over type of interaction. The LOD peak of the [F.sub.6:7] QTL, however, is actually closer to the LOD peak of the proximal [F.sub.2:3] QTL of the same parental contribution (Mo17) and similar relative (second largest) effect. ED had the highest [h.sup.2] values of all traits and the most QTL detected in both generations (four). The QTL with the largest effects in the [F.sub.2:3] (2L) and [F.sub.6:7] (7L) were among the four common ED QTL.
In addition to sampling variation of the population (Beavis, 1994), the perception of QTL and their effects can be greatly affected by environmental factors. Herein, only 17 % of the QTL detected among the stress and nonstress environments were detected in both environments for grain yield and yield components. In contrast, Veldboom and Lee (1996a) reported 50% of the [F.sub.2:3] QTL, for the same traits, were detected in both stress and nonstress environments. Herein, the stress environment was defined by precipitation 128% above the average and a 56% reduction in grain yield, whereas the [F.sub.2:3] stress environment had precipitation 55% above the average and a 17% reduction in grain yield. The reduced consistency of QTL detection across environments in the [F.sub.6:7] generation may be attributed to the more broader environmental range and the more highly inbred progeny. For morphological traits in this population, 35 % of the [F.sub.6:7] QTL were detected in both environments (Austin, 1997). The morphological traits had higher [h.sup.2] values and were less affected by the stress conditions, which likely afforded the more consistent QTL detection. For grain yield and yield components, the results reported herein indicate that trait variation under stress conditions may be controlled by different sets of QTL, as suggested previously (Falconer, 1989).
Herein, evidence of cross-over type QTL interactions was not observed for grain yield and yield components. Several maize QTL mapping studies conducted under stress and nonstress conditions have reported similar observations (Ribaut et al., 1996; Beavis and Keim, 1996; Veldboom and Lee, 1996a, 1996b). QTL x environment interactions seem to be in the form of change in magnitude of effects. The mean environment provided the most complete representation of QTL controlling trait variation in this population. QTL detected in the mean environment included 32 of the 59 (54%) detected among the stress and nonstress environments. The increased precision afforded by the mean environment resulted in the detection of eight additional QTL not detected in either of the individual environments.
Thirteen of the 40 GY and yield component QTL detected in the mean environment of the [F.sub.2:3] generation were also detected in the mean environment of the [F.sub.6:7] generation. For GY and ENP, the QTL with the largest effects were consistent across generations. In nearly all instances, the QTL with the largest effect in the mean environment of one generation were also detected in the mean environment of the other generation. Parental contributions were consistent across generations and environments. The relative magnitudes of QTL effects across generations were not always consistent, indicating possible environmental interactions because the progeny were evaluated in different years.
The detection of QTL with consistent effects across generations and contrasting environments is promising for the application of marker-assisted selection. However, the prevalence of QTL which seem to be unique to their environments could limit the potential gains from marker-assisted selection. An effective breeding strategy may be to use the QTL with consistent effects in an early screening procedure. The selected genotypes should be adapted to diverse environmental conditions, and further field evaluations could be utilized to select for specific adaptability to the target environments. These conclusions are based on the evaluation of inbred progeny, and further evaluations are needed to determine if similar conclusions can be made for hybrid progeny.
Austin, D.F. 1997. Genetic analysis of quantitative trait loci with inbred and hybrid progeny of maize. Ph.D. diss. Iowa State University (Diss. Abstr. lSU1995A97).
Austin, D.F., and M. Lee. 1996a. Comparative mapping in [F.sub.2:3] and [F.sub.6:7] generations of quantitative trait loci for grain yield and yield components in maize. Theor. Appl. Genet. 92:817-826.
Austin, D.F., and M. Lee. 1996b. Genetic resolution and verification of quantitative trait loci for flowering and plant height with recombinant inbred lines of maize. Genome 39:957-968.
Beavis, W.D. 1994. The power and deceipt of QTL experiments: Lessons from comparative QTL studies, p. 250-266. in D.B. Wilkinson (ed.) Proc. Annu. Corn Sorghum Res. Conf., 49th, Chicago, IL. 7-8 Dec. 1994. Am. Seed Trade Assoc., Washington, DC.
Beavis, W.D., and P. Keim. 1996. Identification of QTL that are affected by the environment. In M.S. Kang, and H.G. Hugh (ed.) New perspectives on genotype-by-environment interactions. CRC Press, Inc., Boca Raton, FL.
Beavis, W.D., O.S. Smith, D. Grant, and R. Fincher. 1994. Identification of quantitative trait loci using a small sample of top-crossed and F4 progeny from maize. Crop Sci. 34:882-896.
Bohn, M., M.M. Khairallah, D. Gonzalez-de-Leon, D.A. Hoisington, H.F. Utz, J.A. Deutsch, D.C. Jewell, J.A. Mihm, and A.E. Melchinger. 1996. QTL mapping in tropical maize: I. Genomic regions affecting leaf feeding resistance to sugarcane borer and other traits. Crop Sci. 36:1352-1361.
Bubeck, D.M., M.M. Goodman, W.D. Beavis, and D. Grant. 1993. Quantitative trait loci controlling resistance to gray leaf spot in maize. Crop Sci. 33:838-847.
Burr, B., F.A. Burr, K.H. Thompson, M.C. Albertson, and C.W. Stuber. 1988. Gene mapping with recombinant inbreds in maize. Genetics 118:519-526.
Darvasi, A., and M. Soller. 1995. Advanced intercross lines, an experimental population for fine genetic mapping. Genetics 141:11991207.
deVincente, M.C., and S.D. Tanksley. 1993. QTL analysis of transgressive segregation in an interspecific tomato cross. Genetics 134:585-596.
Dingerdissen, A.L., H.H. Geiger, M. Lee, A. Schechert, and H.G. Welz. 1996. Interval mapping of genes for quantitative resistance of maize to Setosphaeria turcica, cause of northern leaf blight, in a tropical environment. Mol. Breed. 2:143-156.
Edwards, M.D., T. Helentjaris, S. Wright, and C.W. Stuber. 1992. Molecular marker facilitated investigations of quantitative trait loci in maize 4. Analysis based on genome saturation with isozyme and restriction fragment length polymorphism markers. Theor. Appl. Genet. 83:765-774.
Evans, R.O., R.W. Skaggs, and R.E. Sneed. 1990. Normalized crop susceptibility factors for corn and soybean to excess water stress. Trans ASAE 33:1153-1161.
Falconer, D.S. 1989. Introduction to quantitative genetics. 3rd ed. Longman Scientific and Technical, Essex, England.
Freymark, P.F., M. Lee, W.L. Woodman, and C.A. Martinson. 1993. Quantitative and qualitative trait Ioci affecting host plant response to Exserohilum turcicum in maize (Zea mays L.). Theor. Appl. Genet. 87:537-544.
Goldman, I.L., I. Paran, and D. Zamir. 1995. Quantitative trait locus analysis of a recombinant inbred line population derived from a Lycospersicon esculentum x Lycospersicon cheesmanii cross. Theor. Appl. Genet. 90:925-932.
Haley, C.S., and S.A. Knott. 1992. A simple regression method for mapping quantitative trait loci in line crosses using flanking markers. Heredity 69:315-324.
Hall, A.J., F. Vilella, N. Trapani, and C. Chimenti. 1982. The effects of water stress and genotype on the dynamics of pollen-shedding and silking in maize. Field Crops Res. 5:349-363.
Hallauer, A.R., and J.B. Miranda. 1988. Quantitative genetics in maize breeding. 2nd ed. Iowa Stare University Press, Ames, IA.
Jansen, R.C. 1993. Interval mapping of multiple quantitative trait loci. Genetics 135:205-211.
Jansen, R.C., and P. Stam. 1994. High resolution of quantitative traits into multiple loci via interval mapping. Genetics 136:1447-1455.
Joshi, M.S., and N.G. Dastane. 1966. Studies in excess water tolerance of crop plants. Indian J. Agron. 11:70-79.
Kanwar, R.S., J.L. Baker, and S. Mukhtar. 1988. Excessive soil water effects at various stages of development on the growth and yield of corn. Trans. ASAE 31:133-141.
Knapp, S.J., and W.C. Bridges. 1990. Using molecular markers to estimate quantitative trait locus parameters; power and genetic variances for unreplicated and replicated progeny. Genetics 126: 769-777.
Knapp, S.J., W.W. Stroup, and W.M. Ross. 1985. Exact confidence intervals for heritability on a progeny mean basis. Crop Sci. 25: 192-194.
Lande, R., and R. Thompson. 1990. Efficiency of marker-assisted selection in the improvement of quantitative traits. Genetics 124: 743-756.
Lander, E.S., and D. Botstein. 1989. Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185-199.
Lee, M. 1995. DNA markers and plant breeding programs. Adv. Agron. 55:265-344.
Lee, S.H., M.A. Bailey, M.A.R. Mian, T.E. Carter, D.A. Ashley, R.S. Hussey, W.A. Parrott, and H.R. Boerma. 1996. Molecular markers associated with soybean plant height, lodging, and maturity across locations. Crop. Sci. 36:728-735.
Lu, C., L. Shen, Z. Tan, Y. Xu, P. He, Y. Chen, and L. Zhu. 1997. Comparative mapping of QTLs for agronomic traits of rice across environments by using a doubled-haploid population. Theor. Appl. Genet. 94:145-150.
Lubberstedt, T., A.E. Melchinger, C.C. Schon, H.F. Utz, and D. Klein. 1997. QTL Mapping in testcrosses of European flint lines of maize: I. Comparison of different testers for forage yield traits. Crop Sci. 37:921-931.
Mode, C.J., and H.F. Robinson. 1959. Pleiotropism and the genetic variance and covariance. Biometrics 15:518-537.
Mukhtar, S., J.L. Baker, and R.S. Kanwar. 1990. Corn growth as affected by excess soil water. Trans. ASAE 33:437442.
Ottaviano, E., M.S. Gorla, E. P, and C. Frova. 1991. Molecular markers (RFLPs and HSPs) for the genetic dissection of thermotolerance in maize. Theor. Appl. Genet. 81:713-719.
Paterson, A.H., S. Damon, J.D. Hewitt, D. Zamir, H.D. Rabinowitch, S.E. Lincoln, E.S. Lander, and S.D. Tanksley. 1991. Mendelian factors underlying quantitative traits in tomato: Comparison across species, generations, and environments. Genetics 127:181-197.
Ribaut, J.-M., D.A. Hoisington, J.A. Duetsch, C. Jiang, and D. Gonzalez-de-Leon. 1996. Identification of quantitative trait loci under drought conditions in tropical maize. 1. Flowering parameters and the anthesis-silking interval. Theor. Appl. Genet. 92:905-914.
Rosielle, A.A., and J. Hamblin. 1981. Theoretical aspects of selection for yield in stress and non-stress environments. Crop. Sci. 21: 943-946.
Schon, C.C., M. Lee, A.E. Melchinger, W.D. Guthrie, and W.L. Woodman. 1993. Mapping and characterization of quantitative trait loci affecting resistance against second-generation European corn borer in maize with the aid of RFLPs. Heredity 70:648-659.
Schon, C.C., A.E. Melchinger, J. Boppenmaier, E. Brunklaus-Jung, R.G. Herrmann, and J.F. Seitzer. 1994. RFLP mapping in maize: Quantitative trait loci affecting testcross performance of elite European flint lines. Crop Sci. 34:378-389.
Stuber, C.W., S.E. Lincoln, D.W. Wolff, T. Helentjaris, and E.S. Lander. 1992. Identification of genetic factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecular markers. Genetics 132:823-839.
Tuinstra, M.R., E.M. Grote, P.B. Goldsrough, and G. Ejeta. 1996. Identification of quantitative trait loci associated with pre-flowering drought tolerance in sorghum. Crop Sci. 36:1337-1344.
Utz, H.F., and A.E. Melchinger. 1996. PLABQTL: A program for composite interval mapping of QTL. J. Quant. Trait Loci 2:1 (http:// probe.nalusda.gov:8000/otherdocs/jqtl/jqt11996-01/utz.html).
Veldboom, L.R., M. Lee, and W.L. Woodman. 1994. Molecular marker-facilitated studies in an elite maize population: I. Linkage analysis and determination of QTL for morphological traits. Theor. Appl. Genet. 88:7-16.
Veldboom, L.R., and M. Lee. 1996a. Genetic mapping of quantitative trait loci in maize in stress and nonstress environments: I. Grain yield and yield components. Crop Sci. 36:1310-1319.
Veldboom, L.R., and M. Lee. 1996b. Genetic mapping of quantitative trait loci in maize in stress and nonstress environments: II. Plant height and flowering. Crop Sci. 36:1320-1327.
Zeng, Z.B. 1994. Precision mapping of quantitative trait Ioci. Genetics 136:1457-1468.
Dep. of Agronomy, Iowa State Univ., Ames, IA 50011. Journal Paper no. J-17448 of the Iowa Agric. and Home Economics Exp. Stn. Project no. 3134. Received 13 June 1997. *Corresponding author (mlee@ iastate.edu).
Abbreviations: cM, centimorgan; ED, ear diameter; EL, ear length; ENP, ear number per plant; GDD, growing degree days; G x E, genotype x environment interaction; GY, grain yield; KWT, kernel weight; QTL, quantitative trait locus(i); RI, recombinant inbreds; RFLP, restriction fragment length polymorphisms; SI, support intervals; SSR, simple-sequence repeats.
|Printer friendly Cite/link Email Feedback|
|Author:||Austin, David F.; Lee, Micahel|
|Date:||Sep 1, 1998|
|Previous Article:||Comparison across populations for forage traits.|
|Next Article:||The NIa-proteinase of different plant potyviruses provides specific resistance to viral infection.|