Fine Mapping and Characterization of Linked Quantitative Trait Loci Involved in the Transition of the Maize Apical Meristem From Vegetative to Generative Structures

Cristian Vlduu

Corresponding author: Ronald L. Phillips, Department of Agronomy and Plant Genetics, University of Minnesota, 411 Borlaug Hall, 1991 Upper Buford Circle, St. Paul, MN 55108., phill005{at}tc.umn.edu (E-mail)

Communicating editor: W. F. SHERIDAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Quantitative trait locus (QTL) mapping has detected two linked QTL in the 8L chromosome arm segment introgressed from Gaspé Flint (a Northern Flint open-pollinated population) into the background of N28 (a Corn Belt Dent inbred line). Homozygous recombinant lines, with a variable length of the introgressed segment, confirmed the presence of the two previously identified, linked QTL. In the N28 background, Gaspé Flint QTL alleles at both loci induce a reduction in node number, height, and days to anthesis (pollen shed). Given the determinate growth pattern of maize, the phenotypic effects indicate that the two QTL are involved in the transition of the apical meristem from vegetative to generative structures. Relative to the effects of the two QTL in the background of N28, we distinguish two general developmental factors affecting the timing of pollen shed. The primary factor is the timing of the transition of the apical meristem. The second, derivative factor is the global extent of internode elongation. Having separated the two linked QTL, we have laid the foundation for the positional cloning of the QTL with a larger effect.

THE formal distinction between qualitative (discontinuous) and quantitative (continuous) types of trait variation lies in the magnitude of the phenotypic effect of particular alleles in given genetic backgrounds and environments. Depending on the genetic constitution of the parental material used to generate the segregating population and the environmental conditions, virtually any trait can display a range of distributions, from discontinuous with clear-cut phenotypic classes to continuous unimodal. On the one hand, every major gene can be involved in the quantitative type of trait variation, i.e., it may behave as a quantitative trait locus (QTL; ROBERTSON 1985 ). On the other hand, there is a putative class of genes—those that are functionally redundant—for which extreme phenotypes cannot be identified in heterogeneous genetic backgrounds. Genes belonging to the latter class would always behave as QTL in heterogeneous backgrounds and consequently, increased efforts are required for their genetic analysis.

Following the example of WEHRHAHN and ALLARD 1965 , attempts to genetically dissect the quantitative behavior of trait variation have utilized backcross-derived lines (BDLs). In principle, the absence of other segregating QTL in crosses between the BDL and the recurrent parent allows for enhanced resolution of the phenotypic effect of the introgressed QTL. Following a whole-genome QTL analysis, the molecular marker-assisted generation of BDLs and a subsequent allelism test led to the identification of teosinte branched1 as one of the genes responsible for the drastic morphological differences between maize (Zea mays ssp. mays) and its probable progenitor, teosinte (Z. mays ssp. parviglumis; DOEBLEY and STEC 1993 ; DOEBLEY et al. 1995 ). The potential of the BDL method has been exploited on a broader scale in tomato through the development of an introgression library comprising 50 BDLs, each carrying a single chromosomal segment from Lycopersicon pennellii, a green-fruited wild tomato species, within the background of L. esculentum, the cultivated tomato (ESHED and ZAMIR 1994 ). The 50 BDLs with L. pennellii overlapping chromosomal segments spanning the entire tomato genome provide a valuable, continual tool for the fine genetic dissection of various traits (ESHED and ZAMIR 1994 ). One of the BDLs generated by ESHED and ZAMIR 1994 , ESHED and ZAMIR 1995 has been used for the construction of a high-resolution map around a fruit mass QTL (ALPERT et al. 1995 ; ALPERT and TANKSLEY 1996 ).

Employing the BDL method, in this study we have detected two linked QTL in maize, each affecting node number, height, and maturity (days to pollen shed). Given the determinate growth pattern of maize, the phenotypic effects indicate that the two QTL are involved in the transition of the apical meristem from vegetative to generative structures. The two QTL, of unequal effect, are located 12–20 cM apart on the chromosome 8L segment introgressed from Gaspé Flint (GF), the donor parent of "early maturity" genes, into the background of N28, the recurrent parent. We have confined the major QTL within a 5-cM interval.

Relative to the effects of the two QTL in the background of N28, we distinguish two developmental factors affecting the timing of anthesis (pollen shed). The primary factor, clearly affected by the two QTL, is the timing of the transition of the apical meristem. The second factor is the extent of stem elongation, which depends on the total number of phytomer primordia formed until the transition of the apical meristem from vegetative to generative growth. The trait values of the four homozygous genotypes at the two linked QTL were subjected to additive x additive contrasts. In the background of N28, the two linked QTL, in homozygous condition, behave in an additive manner with respect to node number (ND), suggesting the possibility of their being involved in different pathways during tassel initiation. The significant deviations from additivity for height (HT) and days to pollen shed (DPS) may suggest (1) differences in the systemic coordination of internode growth when the total number of stem units (phytomers) is altered; (2) that the two QTL interact during the post-transitional growth; or (3) introgression of other, linked QTL with a small effect, affecting shoot development. Finally, we discuss the feasibility of positional cloning of the QTL with a larger effect.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant materials:
GF, the donor parent of the "early maturity" genes, is a Northern Flint open-pollinated population collected from the Gaspé Peninsula of Quebec, Canada, and homogenized for early maturity through mass selection (BRAWN 1968 ). N28, the recurrent parent, is a Corn Belt Dent inbred line derived from Nebraska Stiff Stalk Synthetic (SSS), which, through Iowa SSS, traces back to Reid Yellow Dent, a Corn Belt Dent open-pollinated population (GERDES et al. 1993 ). E20, the early maturing derivative of N28, is the result of 20 uninterrupted backcrosses following the cross between GF and N28 (D. SHAVER, personal communication). Within each backcross generation, several of the earliest plants were backcrossed again to N28. After the 20th backcross within each backcross progeny, the earliest plants were sib-mated twice and thereafter selfed. Pooled kernels from one of the resulting selfed ears were the source of E20 (D. SHAVER, personal communication). Kernels of GF, N28, and E20 were provided by D. Shaver (Cornnuts Inc., Salinas, CA). E20 sheds pollen earlier, is shorter, and has fewer nodes than N28. An initial survey of N28 and E20 with 95 genomic clones, covering all 10 chromosomes of maize, identified restriction fragment length polymorphism (RFLP) for UMC12a on chromosome arm 8L and UMC84a on 1L (KIM 1992 ). Only UMC12a showed significant association with the variation of days to pollen shed, days to silking, height, and node number in E20 x N28 F₂ and F₃ populations, each of 90 entries (KIM 1992 ).

Screening the E20 material more extensively with DNA clones, we detected heterogeneity within E20 relative to the chromosomal segments retained from GF (Fig 1). This indicates that the source of E20 was heterozygous/heterogeneous. The three E20 genotypes identified so far, E20-A, E20-B, and E20-C (Fig 1), are phenotypically similar. Except for the chromosomal segments retained from GF in 1L, 6L, and 8L (Fig 1), the genome of the E20 versions is assumed to be that of N28. The use of E20 plants in some experiments preceded the identification of their exact genotype.

View larger version (15K): In this window In a new window Download PPT slide   Figure 1. RFLP heterogeneity within E20 and linkage maps of chromosome arms 1L, 6L and 8L. , homozygous N28; , homozygous GF. The maps (in Haldane centimorgans) were constructed using 88 E20-A x N28 F2 plants. The marker order and approximate genetic distances for 6L are based on the UMC and BNL 1995 maps (COE et al. 1995 ; MATZ et al. 1995 ). The relative position of UMC236 in 8L is based on the UMC 1995 map (COE et al. 1995 ) and on the mapping data obtained from a GF x N28 F2 population (VLADUTU 1998 ). 1L and 6L are not scaled relative to 8L. For each chromosomal segment, the left and right ends indicate the proximal and distal ends, respectively.

Eighty-eight E20-A x N28 F₃ families were used for QTL mapping in the background of N28. Because two linked QTL were detected (see RESULTS), we selected homozygous recombinant F₃ and F₄ plants with a variable length of the 8L segment introgressed from GF in the otherwise homogeneous background of N28. The recombinant plants with an "A" designation were selected from E20-A x N28 F_2:4 families whereas those with a "C" designation were selected from E20-C x N28 F₃ families (see RESULTS).

In each experiment (see below), the experimental unit consisted of a row 6.7 m in length. Thirty-five kernels were planted in each row; later the stands were thinned to 30 plants/row. The distance between rows was 76 cm, resulting in 58,917 plants/ha.

Fifteen to 25 14-day-old seedlings, grown in pots in the greenhouse, from F₃, F_3:4, and F_4:5 families, were pooled for DNA extraction to infer the molecular marker genotype of their progenitor plant.

RFLP analysis:
DNA isolation, Southern blotting, and ³²P-labeled hybridization were performed according to established procedures (FEINBERG and VOGELSTEIN 1983 ; SAGHAI-MAROOF et al. 1984 ; SAMBROOK et al. 1989 ). The DNA clones used as probes were provided by the University of Missouri at Columbia (UMC and CSU), Plant Gene Expression Center, Albany, CA (Ec2-11), University of California at Berkeley (UCBanp1), CSIRO, Division of Plant Industry, Australia (PIC6), University of Bologna, Italy (PG7), Kansas State University (DGG9), Native Plants Inc. (NPI), Brookhaven National Laboratory (BNL), and Universität zu Köln, Germany (ZmHox1).

Trait measurement, experimental design, and statistical analysis:
The three traits measured in each experiment were DPS, ND, and HT. DPS was invariably measured as the number of days from planting until 50% of the plants within a row had the first healthy anther extruded. DPS was recorded on a daily basis. Within a given experiment, ND and HT (in centimeters) were measured (see below) a few days after the latest entries shed pollen.

QTL mapping in the background of N28: The 88 E20-A x N28 F₃ families were grown within a randomized complete block design (RCBD) in two locations situated ~60 km apart, St. Paul and Rosemount, Minnesota, in 1994. Two replications were planted in each location, May 10 in St. Paul and May 11 in Rosemount. HT was measured from ground to the level of the ligule of the top leaf and ND was counted from the first node above the ground to the node corresponding to the top leaf. For each row, a single, average plant was measured for ND and HT.

Within each location, the phenotypic data were first subjected to an analysis of variance with the (random effect) model,

where Y is the trait value of a row, µ is the mean of the F₃ population, B is the effect of the replication, G is the effect of the F₃ family, and is the error. This model was used to estimate the genetic variances and broad-sense heritabilities within each location. For each trait, an F-test was used to check the homogeneity of the error variances between the two locations. The F-test was not significant (P > 0.05) for DPS but significant (P < 0.05) for ND and HT. Nonetheless, considering the small differences between the error variances for ND (0.67 for St. Paul vs. 0.45 for Rosemount) and HT (32.9 for St. Paul vs. 22.4 for Rosemount), we further subjected each trait and molecular marker data to an analysis of variance across the two locations. The following (mixed, unbalanced) model was used for the analysis of variance across locations,

where Y is the trait value of a row, µ is the mean of the F₃ population, L is the effect of the location, B(L) is the effect of the replication nested in location, M is the effect of the molecular marker, G(M) is the effect of the F₃ family nested in molecular marker, M * L is the effect of the interaction between molecular marker and location, G(M * L) is the effect of the interaction between the F₃ family nested in molecular marker and location, and is the error. All factors were considered random except the molecular marker, which was considered fixed. The above model is essentially the one proposed by KNAPP 1994 ; the slight modification consisted in nesting the replication in location. The above model was used for each pairwise combination of trait and molecular marker. Because significant interactions were found between molecular markers and location, the data were further analyzed separately for each location.

Composite interval mapping (CIM) by multiple regression using selected markers as cofactors (JANSEN 1993 ; ZENG 1994 ) was performed with the PLABQTL software (UTZ and MELCHINGER 1996 ). The inclusion of appropriate markers as cofactors in multiple regression models removes the confounding effects of QTL placed outside the intervals being tested (JANSEN 1993 ; ZENG 1994 ). Compared to simple interval mapping (SIM; without cofactors), CIM provides an increased power to detect linked QTL (ZENG 1994 ). Empirical threshold levels for the LOD score (DOERGE and CHURCHILL 1996 ) were established for each chromosomal arm (1L and 8L) and trait. The empirical threshold levels ( = 0.05) were based on 1000 permutations of the phenotypic data using model 3 (simple interval mapping) of the QTL CARTOGRAPHER software (BASTEN et al. 1995 –1996).

SIM was the first step performed with PLABQTL. SIM could not distinguish between a single or linked QTL on 8L. Next, the 8L marker with the highest LOD score detected by SIM (either PG7 or DGG9, depending on the trait and location) was used as cofactor. A secondary, linked region marked by ZmHox1a was identified for DPS but not for ND and HT. The inclusion of both DGG9 and ZmHox1a as cofactors in the regression models for DPS evidenced a significant QTL on 1L for the St. Paul data. For each trait and location, the final regression model, which used the exact coordinates of significant QTL peaks, was the model that minimized the Aikaike's information criterion (JANSEN 1993 ).

Confirmation of the two linked QTL on chromosome arm 8L: The selfed progenies of the homozygous recombinant F₃ and F₄ plants (see Plant materials), along with several other lines (Fig 5), were planted within a RCBD with six replications in St. Paul, May 10, 1997. In each row, seven random plants were measured for ND and several other traits (see below). For each row, the mean value of the seven plants was used in the statistical analyses. In this experiment ND represents the total (leaf) ND and it was counted as follows. The sixth leaf (excluding the coleoptile) was marked when the first leaf was still visible (June 19). The 12th leaf was marked when the 6th, previously marked leaf was still visible (July 5). The 12th leaves marked were still attached to plants when the total number of leaves was counted (August 5). On June 19 and July 5, the number of the last leaf emerging from the whorl was recorded. On July 5 we also counted the leaf on which the transition from juvenile to adult traits (marked by the occurrence of epidermal hairs) was first apparent. In addition, we counted the number of nodes situated above the insertion of the top ear.

View larger version (31K): In this window In a new window Download PPT slide   Figure 2. Histograms displaying the frequency distributions of the three traits, DPS, ND, and HT, within the 88 E20-A x N28 F3 families grown in two locations, St. Paul and Rosemount, in 1994. For each location, the trait values represent the means of two replications. In each graph, the left arrow indicates the mean of E20 (the exact genotype not known) and the right arrow indicates the mean of N28. The absolute frequencies are shown along the y-axis. The trait values are shown on the x-axis. See MATERIALS AND METHODS for trait measurement.

View larger version (17K): In this window In a new window Download PPT slide   Figure 3. Graphic display of the associations between the DNA markers of chromosome arm 8L and the variation of the three traits (HT, DPS, and ND) within the E20-A x N28 F3 population as established by the analysis of variance across the two locations (St. Paul and Rosemount, 1994). The lines uniting the sum of squares were used only to help visualize the data; they do not indicate the values between the markers.

View larger version (21K): In this window In a new window Download PPT slide   Figure 4. QTL likelihood maps for DPS, HT, and ND on chromosome arm 8L in the E20-A x N28 F3 population. The QTL maps, corresponding to the final regression models computed by PLABQTL, are shown separately for each location, St. Paul and Rosemount, in 1994. For each location, the trait values were the means of two replications. The horizontal dashed line indicates the threshold value (also shown above the dashed line).

View larger version (27K): In this window In a new window Download PPT slide   Figure 5. RFLP composition of selected lines. The lines are listed in the same order as in Table 3 and Table 4. , homozygous N28; , homozygous GF; , heterozygous; , heterozygous or homozygous GF; , not determined. The A lines were derived from E20-A x N28 crosses; the C lines were derived from an E20-C x N28 cross. The dotted lines delimit the position of the two 8L-QTL as detected by QTL mapping (Fig 4) and confirmed by the data of Table 3 and Table 4. Assuming that each QTL harbors a single gene we named the respective genes Vegetative to generative transition1 and -2 (Vgt1 and Vgt2). Chromosome arms 1L and 6L are not scaled relative to 8L. The scale bar refers only to 8L.

HT was measured from ground to the tip of the tassel. A single, average plant was measured for HT in each row.

The homogeneity of variances was checked and validated (at = 0.05 for DPS and ND, and at = 0.01 for HT) by the Hartley test (HARTLEY 1950 ). Multiple comparisons between lines were carried out only for DPS, ND, and HT. Pairwise comparisons were performed using t-tests with a comparisonwise type I error set by the Bonferroni inequality (MILLER 1981 ) at = 0.00064 (this ensures an experimentwise < 0.05). Four contrasts (STEEL and TORRIE 1980 ) were preplanned. Three contrasts were intended to discriminate between groups of lines. The fourth, additive x additive (A x A), contrast tested the null hypothesis, H₀, NN + GG = NG + GN, vs. the alternative hypothesis, H_a, NN + GG NG + GN, where N designates homozygosity for the N28 alleles, G designates homozygosity for the GF alleles, and for each genotype, the first letter represents the major QTL and the second letter represents the minor QTL of chromosome arm 8L. Using the Bonferroni inequality for each contrast, the comparisonwise type I error was set at = 0.0125.

The raw phenotypic data were used in all statistical analyses mentioned above. The analyses of variance and multiple comparisons were performed using the GLM procedure of SAS version 6.3 (SAS INSTITUTE 1994). The segregation ratios for the molecular markers were analyzed using LINKAGE-1 version 3.5 (SUITER et al. 1983 ). The linkage maps (in Haldane centimorgans) were assembled using MAPMAKER/EXP version 3.0 (LINCOLN et al. 1992 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

RFLP composition of E20 and linkage analysis in the background of N28:
RFLP screening detected heterogeneity within E20 relative to chromosome arms 1L, 6L, and 8L (Fig 1). The only common chromosomal segment retained from GF by the three E20 versions lies on 8L between UMC236 and UMC12a. The phenotypic similarity of the three E20 genotypes suggests that the common GF segment of 8L largely accounts for the phenotypic differences between the E20 genotypes and N28. The 6L segment retained from GF in E20-B shares duplicated DNA sequences (such as ZmHox1) with the 8L segment retained from GF by all three E20 variants.

The interspersed RFLP composition of the 8L segments is likely the result of recombination (after the completion of the backcross program) between two hypothetical original versions: one with an uninterrupted GF segment between UMC236 and UMC12a and the other with an uninterrupted GF segment encompassing UMC124a and CSU31a.

In the E20-A x N28 F₂ population, the P values associated with the goodness-of-fit (²) tests for 1:2:1 or 3:1 segregation ratios ranged from 0.09 for CSU66c to 0.97 for UMC89a.

After 20 backcrosses, ~5 cM are expected to be retained on each side of the selected gene (NAVEIRA and BARBADILLA 1992 ). Ignoring the interspersed configuration of the 8L segments, all chromosomal fragments retained from GF exceed the expected linkage drag (Fig 1). This discrepancy suggests selection for linked genes and/or reflects sampling variability during selection in each backcross generation.

QTL mapping in the background of N28:
Fig 2 shows the distribution of the three traits (DPS, ND, and HT) within the E20-A x N28 F₃ population in the two locations. The trimodal (in St. Paul) and bimodal (in Rosemount) distributions of DPS suggest segregation of a QTL with a large effect(s). The genetic variances for each trait were highly significant (P < 0.001) in each location.

In the analyses of variance across locations, all markers (the M term in the linear model) of chromosome arm 8L were significantly associated with the variation of DPS (P < 0.05 for UMC31a and P < 0.01 or P < 0.001 for the rest of the markers) and HT (P < 0.01 for UMC31a and P < 0.001 for the rest of the markers). For ND, the P values ranged from 0.064 for UCBanp1 to 0.001 for UMC12a. Neither marker of chromosome arm 1L showed significant association (P > 0.1) with the variation of any of the three traits. Fig 3 shows the type III sums of squares plotted against the corresponding markers of 8L. The similar distribution patterns of sums of squares for the three traits indicate that the same chromosomal region (PG7–UMC89a) accounts for most of the variation of the traits. Significant (from P < 0.05 to P < 0.001) marker x location interaction (the M * L term in the linear model) was detected for DPS and ND but not for HT. The significance of the M * L terms for DPS and ND is the statistical expression of differences in the relative effect(s) of the QTL, linked to the respective markers, in the two locations.

The QTL likelihood maps of chromosome arm 8L, corresponding to the final regression models computed by PLABQTL, are shown in Fig 4. In accordance with the distributions of sums of squares (Fig 3), for each trait and location, the major QTL of 8L is placed approximately in the same interval between PG7 and UMC89a. In addition, a linked, minor QTL, placed between CSU66b and UCBanp1, was also detected for DPS in both locations (Fig 4). For chromosome arm 1L, the LOD score (2.45) exceeded the threshold level (1.89) only for DPS in St. Paul (not shown). The detected QTL explain most of the genetic variance for each trait and location (Table 1). Compared with the analyses of variance (Fig 3) and SIM (not shown), which could not resolve the two linked QTL of 8L, these results (Fig 4 and Table 1) exemplify the increased power of CIM to detect linked QTL.

The negative value of the additive effect (a) for the QTL detected on 1L (DPS, St. Paul; Table 1) would indicate that the GF allele increases the trait value. The differences between the a values at the major 8L-QTL for DPS and ND in the two locations (Table 1) alone could explain the significance of the M * L terms in the analyses of variances across locations. Across traits and locations, the d/a ratios (which can be computed using the d and a values from Table 1) are in the range of additivity-semidominance (toward the N28 allele) at the major 8L-QTL and additivity at the minor 8L-QTL and 1L-QTL.

Given the determinate growth pattern of maize, the fact that the major 8L-QTL is approximately placed at the same chromosomal position for each of the three traits (Fig 3 and Fig 4) indicates that the high correlations between the traits (Table 2) are likely the result of pleiotropy.

Confirmation of the two linked QTL of chromosome arm 8L:
To confirm and separate the two linked QTL of 8L, we selected recombinant F₃ and F₄ plants derived from crosses between E20 variants and N28 (Fig 5). The 8L-recombinant plants (A44-21, -18, -10, -8; C22-6, -4, -1; and C24-1, -2) are homozygous in the intervals likely containing the QTL (Fig 4), and are devoid of GF segments in 1L and 6L (Fig 5). The interspersed RFLP composition of 8L in the recombinant plants (Fig 5) partly originates in the parental material (E20-A, -C; Fig 1), and partly is the result of single- and double-crossover events that occurred in successive meioses.

The grouping of lines based on the Bonferroni t-tests (hereafter termed Bonferroni grouping) is presented in Table 3. The Bonferroni grouping for ND (see MATERIALS AND METHODS) clearly distinguishes four different groups of lines relative to the RFLP composition of 8L (Table 3 and Fig 5). The first group is represented by N28, the second by A44-21, -10, -8, the third by C22-6, -4, -1, and the fourth by C24-1, -2, A44-18, and A27. The Bonferroni grouping for DPS resembles the one for ND; the difference is that the two bottom groups (as defined by the Bonferroni grouping for ND) overlap. For HT, the three bottom groups (as defined by the Bonferroni grouping for ND) overlap. However, for both DPS and HT there is no change in rank between lines belonging to different groups (as defined by the Bonferroni grouping for ND). To discriminate between groups of lines not resolved by the Bonferroni grouping for DPS and HT, three contrasts (C1 for DPS, C2 and C3 for HT) were tested (Table 4). The Bonferroni groupings (Table 3) and the statistical significance of the three contrasts (Table 4) together with the RFLP composition of the selected lines (Fig 5) provide compelling evidence for the presence of two linked QTL on 8L. These results indicate that the minor 8L-QTL, detected by QTL mapping only for DPS (Fig 4), actually is involved, like the major 8L-QTL, in the vegetative-to-generative transition.

The four groups of lines defined by the Bonferroni grouping for ND, relative to the RFLP composition of 8L, correspond to the four homozygous genotypes at the two linked QTL. Assuming that each QTL harbors a single gene, we named the respective genes Vegetative to generative transition1 (Vgt1, at the major QTL) and -2 (Vgt2, at the minor QTL; Fig 5 and Table 4).

Because line A116 was significantly different than N28 for each trait (Table 3), the Bonferroni groupings also confirmed the presence of a QTL, with a small effect, placed on 1L. The 1L-QTL, like the 8L-QTL, seems to be involved in the transition of the apical meristem to generative structures. The phenotypic means of E20-B (Table 3) indicate that the combined effect of the GF alleles at Vgt1, Vgt2, 1L-QTL, and the putative 6L-QTL is not stronger than the effect of the GF alleles at Vgt1 and Vgt2 alone.

A x A contrasts—partial tests of interaction between the two linked genes:
The A x A contrast comparing the homozygous genotypes at the two linked genes (Vgt1 and Vgt2) was significant for DPS and HT but not significant for ND (Table 4). The graphic representation of the statistical significance of the A x A contrasts is provided in Fig 6. The morphological and phenological differences among the four homozygous genotypes are illustrated in Fig 7.

View larger version (16K): In this window In a new window Download PPT slide   Figure 6. Graphic representation of the mean trait values for the four homozygous genotypes at the two linked genes (Vgt1 and Vgt2) in the background of N28 (see Table 5). G indicates homozygosity for the GF alleles; N indicates homozygosity for the N28 alleles. Parallel lines (for ND) indicate additivity; nonparallel lines (for DPS and HT) indicate deviations from additivity (see also A x A contrasts in Table 4).

View larger version (125K): In this window In a new window Download PPT slide   Figure 7. Morphological and phenological (as indicated by the development of the topmost ear) differences among and within the four homozygous genotypes at Vgt1 and Vgt2 in the background of N28. The leaves and lower ears were removed and the brace roots trimmed to illustrate the aboveground internodes. The total number of nodes (phytomers) and the genotype (see Table 5) are indicated below each plant. The numbers on the left side of each plant indicate the phytomer bearing the topmost ear (upper number) and the first phytomer above the soil level (lower number). The tick marks are placed at 5-cm intervals. The plants, grown in the summer of 1998 in St. Paul, were photographed after N28 (NN) had shed pollen. In comparison with 1997 (Table 5), in 1998 the total number of nodes was similar for the NG and NN genotypes but averaged about one less node for the GG and GN genotypes.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

E20 as a product of backcrossing:
Map distances: It appears that the discrepancy between the expected genetic lengths of the introgressed segments (NAVEIRA and BARBADILLA 1992 ) and the genetic distances estimated in the E20-A x N28 F₂ population (Fig 1), as previously suggested, can be explained by at least two factors. The first factor, as established for the introgressed segment of chromosome arm 8L, is selection for linked genes (Fig 4 and Fig 5). The second factor, suggested again by the case of the 8L segment, is likely represented by sampling variability during the backcross program.

Also, the gradual homogenization of the genetic background during the backcross program may induce gradual or sudden changes in recombination frequencies in successive backcross generations. Therefore, the genetic distances estimated in the E20-A x N28 F₂ population would reflect the recombinogenic activity, in the respective chromosomal regions, attained only at the end of the backcross program.

QTL and phenotypic selection: Although both analysis of variance and regression are robust to deviations from normality (STEEL and TORRIE 1980 ), the nonnormality of the trait distributions (Fig 2) could have led to biased estimates. Nonetheless, the consistency of the statistical results suggests that overall, the genetic parameters were estimated accurately. Also, with the exception of one trait in one location (ND in St. Paul; P < 0.05), the distributions of the residuals from the final regression models did not significantly deviate from normality (P > 0.2) as judged by the Shapiro-Wilk test (SHAPIRO and WILK 1965 ). Furthermore, the presence and position of the three QTL detected by QTL interval mapping have been unambiguously confirmed by the analysis of recombinant lines.

In the background of N28 the GF alleles at both Vgt1 and Vgt2, as expected, reduce the values of the three correlated traits relative to the N28 alleles (the positive values of a_8M and a_8m in Table 1; Table 5). However, QTL mapping in the E20-A x N28 F₃ population somewhat unexpectedly indicated that at the 1L-QTL with a small effect, the GF allele delays the timing of anthesis relative to the N28 allele (the negative value of a₁ in Table 1). In contrast, the phenotypic means of line A116, which presumably is homozygous for the GF allele at the 1L-QTL and homozygous for the N28 alleles at Vgt1 and Vgt2 (Fig 5), were significantly smaller than the phenotypic means of N28 (Table 3). This apparent contradiction can be readily explained by interaction of 1L-QTL with either one or both genes of 8L. The fact that the combined effect of the GF alleles at Vgt1, Vgt2, 1L-QTL, and the putative 6L-QTL, present in E20-B (Fig 5), was not significantly different for DPS and HT and actually significantly less for ND than the combined effect of the GF alleles at Vgt1 and Vgt2 alone (Table 3), is also suggestive of gene interaction.

The relatively large magnitude of transgressive segregation apparent for ND and HT in Fig 2 is likely an artifact due to inadequate sampling of the parental genotypes. However, the potential interaction(s) between 1L-QTL and either one or both genes of 8L, although with a presumably small effect (see GVE values in Table 1), could have induced a certain amount of transgressive segregation for DPS, ND, and HT in the E20-A x N28 F₃ population.

Although, when homozygous, the GF alleles at 1L-QTL and the putative 6L-QTL do not lower the phenotypic values in the presence of homozygous GF alleles at Vgt1 and Vgt2 (Table 3), it is unlikely that during 20 backcrosses they had been retained by chance. The retention of the GF alleles at 1L-QTL and the putative 6L-QTL may suggest that the respective alleles have a stronger phenotypic effect in the environments where the backcrosses were conducted (Brookhaven, NY and Salinas, CA; D. SHAVER, personal communication) relative to the St. Paul environment. But given the potential interactions with Vgt1 and Vgt2, in our opinion, the most probable explanation for the retention of the GF alleles at 1L-QTL and the putative 6L-QTL could relate to the backcross procedure itself. Because the 20 backcrosses were never interrupted by selfing (D. SHAVER, personal communication), the phenotypic selection always discriminated between heterozygotes and homozygotes for the N28 alleles. Therefore, it is likely that in a heterozygous condition, the 1L-QTL and the putative 6L-QTL would visibly lower DPS in the presence of heterozygosity at Vgt1 and Vgt2. The reason for no significant interaction between the introgressed QTL, as detected by two-factor analyses of variance in the E20-A x N28 mapping population (not shown), presumably was the small size of the population.

Ignoring the complexity that may arise due to gene interaction (epistasis), the fact that the backcrosses were not interrupted by selfing would also impose constraints on the type of relationship between the alleles of the donor and recurrent parents at the introgressed loci. For QTL with moderate or large individual (nonepistatic) effects, the allelic relationship is expected to range from overdominance toward the phenotype of the donor parent to semidominance toward the phenotype of the recurrent parent. The allelic relationship at Vgt1 and Vgt2 estimated in both E20-A x N28 and GF x N28 F₃ populations (Table 1; VLADUTU 1998 ) conforms to this expectation. Also, because Vgt2 and the putative 6L-QTL map within duplicated regions (Fig 5), they might be structurally related.

At least three of the loci (Vgt1, Vgt2, and 1L-QTL) introgressed from GF into the background of N28 appear to be involved in the transition of the apical meristem from vegetative to generative structures. These findings are in agreement with previous observations (BRAWN 1968 ), which indicated that the extreme early maturity of GF is due to the very early timing of tassel initiation rather than a fast rate of growth.

In a GF x N28 population of 91 F₃ families, the effects of the 1L-QTL and the putative 6L-QTL were not detectable and the cumulative effect of Vgt1 and Vgt2 accounted for only 13% of the genetic variance for ND (VLADUTU 1998 ). These results indicate that other loci are also responsible for the 12–16 node difference between GF and N28.

The effects of Vgt1 and Vgt2:
Tassel initiation, stem elongation, and anthesis: Five to six incipient leaves (excluding the coleoptile) are usually formed in the maize embryo before it becomes dormant. The growth of the incipient leaves formed in the embryo and the initiation of leaf primordia are resumed soon after the onset of germination. ABBE and PHINNEY 1951 reported that under field conditions, the rate of successive postembryonic leaf primordia initiation increased exponentially. Given that in controlled environments, under constant temperature, the rate of leaf initiation is relatively constant (GREYSON et al. 1982 ; KINIRY and BONHOMME 1991 ; LEJEUNE and BERNIER 1996 ), the exponential increase in the rate of leaf initiation noted by ABBE and PHINNEY 1951 likely reflects the increase in the ambient temperature. ASKENASY 1880 (cited by ABBE and PHINNEY 1951 ) defined the plastochron as the time interval between the initiation of two successive stem units. In maize, the initiation of a stem unit (phytomer, as defined by GALINAT 1959 ) coincides with the initiation of its leaf component (SHARMAN 1942 ; ABBE and PHINNEY 1951 ). The production of leaves is terminated when the apical meristem switches to generative structures (KIESSELBACH 1949 ; LENG 1951 ). Depending on the genotype and environmental conditions, six to nine leaves are commonly visible at the time of tassel initiation (LENG 1951 ; LEJEUNE and BERNIER 1996 ; COLASANTI et al. 1998 ).

Most of the internode growth takes place after tassel initiation (KIESSELBACH 1949 ; LENG 1951 ). Both cell division occurring in the intercalary meristem situated at the base of the internode and cell elongation contribute to the final length of the internode (SACHS 1965 ). Within a phytomer, the elongation of the internode commences after its associated leaf has elongated (SHARMAN 1942 ). MORRISON et al. 1994 showed that in the Mo17 x B73 F₁ plants, the stem elongates in a hierarchical, stepwise manner; a given internode does not start visibly elongating until an older internode, placed several stem units below, stops elongating. At any time, several successive internodes are elongating, although at different rates (MORRISON et al. 1994 ). Similar observations were made by SHARMAN 1942 using a different maize genotype (Sutton's White Horse Tooth), indicating that the stepwise elongation of the stem is a common feature across different backgrounds. The tassel, which tips the maize stem, likely conforms to the same developmental rule, that is, its elongation is held in check (at variable degrees depending on the genotype) by the elongation of the internodes below. Anthesis usually occurs only after the tassel has emerged from the whorl.

While GF and its F₁ hybrids with a number of Corn Belt Dent lines (including N28) continue to elongate the internodes and tassel peduncle after the tassel starts shedding pollen, in the N28 background, irrespective of the alleles present at Vgt1 and Vgt2, the increase in height occurring after the first anthers shed pollen is none or negligible (up to 6 cm) in the St. Paul environment (C.VLDUU, personal observation).

Given the general features of shoot development summarized above, the interval from planting to anthesis can be formally divided into two phases. The first phase, consisting of the initiation of phytomers, encompasses the interval from planting to the transition of the apical meristem. The second phase, characterized by stem elongation, represents the interval from tassel initiation to anthesis. Based on a comparative study of inbred lines and F₁ hybrids, LENG 1951 suggested that the two phases of shoot development may be under the control of different sets of genes.

The LN-I and LN-II values (the trait acronyms are defined in Table 5) are similar among the four homozygous genotypes at Vgt1 and Vgt2 in the N28 background (Table 5). Also, when LN-I and LN-II were counted, the plants appeared uniform among the four homozygous genotypes (not shown). Considering the homogeneity of the genetic background, these observations suggest that the rate of phytomer initiation and the rate of post-transitional growth are similar for the four genotypes.

Therefore, the small average difference (0.9) in DPS, corresponding to one added node (phytomer) between the GN and GG genotypes (Table 6), can be taken as a rough, average estimate of the plastochrons 20 and 21 for GN, 20–22 for NG, and 20–24 for NN. This average estimate of the last plastochrons for the three homozygous genotypes (GN, NG, and NN) is likely biased upward for two reasons. First, the average difference in DPS, corresponding to one added phytomer between the GN and GG genotypes, should include a component representing the delay in DPS caused by the elongation of the extra phytomer. Second, considering the results of ABBE and PHINNEY 1951 , successive plastochrons become shorter under field conditions. This average estimate (0.9 days) of the last plastochrons can be used to compute approximately the differences in the timing of tassel initiation and the relative contribution of internode elongation to the delay in the timing of pollen shed between the homozygous genotypes at Vgt1 and Vgt2. For example, the average delay in the timing of pollen shed between the NN and GG genotypes is 10.2 days, of which at least 6.4 days are presumably due to additional stem elongation caused by the extra (4.2) internodes formed in the NN genotype relative to the GG genotype (Table 6). The difference of <3.8 days in the timing of tassel initiation between the NN and GG genotypes is augmented by the growth of the additional phytomers (formed in the NN genotype relative to the GG genotype) to a difference in the timing of anthesis of 10.2 days.

Thus, relative to the effects of the two linked loci (Vgt1 and Vgt2) in the background of N28, we distinguish two developmental factors affecting the timing of anthesis. The primordial factor, clearly affected by Vgt1 and Vgt2, is the timing of the transition of the apical meristem from vegetative to generative growth. The second factor is the global extent of internode elongation, which depends on the total number of phytomer primordia formed until tassel initiation. We consider that these inferences, although approximate, would remain valid in principle even if one or both genes (Vgt1 and Vgt2) slightly affect the rate of leaf initiation and/or the rate of post-transitional growth.

The differences in the DPS and HT values (Table 3) among lines belonging to a given group (as defined by the Bonferroni grouping for ND) could be caused by the individual or combined effect of several factors: (1) introgressed gene(s), other than Vgt1 and Vgt2, with a small effect, affecting shoot development; (2) slight differences in the epigenetic state of genes involved in post-transitional growth; and (3) experimental errors, such as uneven distribution of fertilizers within the plot.

A x A contrasts: As shown in Fig 7, the maize stem gradually tapers from bottom to top. Along the stem, the internodes vary in length and thickness, paralleling the variation in shape and size of leaves (GREYSON et al. 1982 ). MORRISON et al. 1994 found that the position of the internode in the stem and its function in the plant influence its growth rate and elongation pattern. The same authors noted that for the aboveground internodes, the duration of elongation increased from the basal internodes to the internodes with subtending ears; then it decreased for the internodes above. For internodes without subtending ears, the longer the elongation period, the longer was the final length; the internodes with subtending ears had the longest period of elongation but their final length was shorter than that of adjacent internodes without subtending ears.

The additive behavior of the four homozygous genotypes with respect to node number (Table 4 and Fig 6) suggests the possibility that Vgt1 and Vgt2 are involved in different pathways during tassel initiation. However, a full test of epistasis, including the heterozygous genotypes, is needed for more conclusive results in this regard.

The significant deviations from additivity for HT and timing of anthesis (DPS) can be interpreted in three broad ways. First, note that by only altering the timing of tassel initiation, and thus the total number of phytomers, particular phytomers will change their position in the stem relative to the tassel and topmost ear. Consequently, a change in the total number of phytomers would bring about changes in thickness, final length, and duration of elongation for particular internodes.

Taking GG as the reference genotype, the average difference in HT and DPS, corresponding to one added node, increases as the difference in the total ND increases (Table 6). For each genotype, seven to eight very short internodes remain underground (Fig 7). Although somewhat fewer for the GG genotype, the phytomers placed between the tassel and the topmost ear are relatively constant in number, length, and thickness among the four genotypes (Table 5 and Fig 7). It is the phytomers placed between the topmost ear and soil level that vary considerably in number, final length, and thickness among the four genotypes (Fig 7). From the GG to the NN genotypes there is an increase in number, length, and thickness of the phytomers placed below the topmost ear (Fig 7). The progressive increase in thickness of the basal internodes can be viewed as a systemic adjustment to sustain the additional vegetative growth induced by the formation of additional phytomers. Also, more nodes generate brace roots in NN relative to GG (Fig 7). Therefore, one possible interpretation of the significance of the A x A contrasts for HT and DPS (Table 4 and Fig 6) is that, through systemic regulation of post-transitional shoot development, alterations in the total number of phytomers could induce nonadditive changes in HT and DPS. This interpretation does not require any assumption regarding the state of activity of Vgt1 and Vgt2 during post-transitional growth.

The second possible explanation is that the two genes interact during post-transitional growth. However, the 8L-recombinant lines are heterogeneous in respect to GF segments in 8L outside the intervals harboring Vgt1 and Vgt2 (Fig 5). Therefore, a third possible explanation for the nonadditivity of DPS and HT would invoke introgression of linked genes, other than Vgt1 and Vgt2, with a small effect, affecting shoot development. Regarding this third hypothesis, note that for each genotype, the relatively long internodes placed immediately below the topmost ear are considerably thicker than the internodes above it (Fig 7). These thick internodes would require a longer time to attain their final length in comparison with the thinner internodes placed above the topmost ear. Therefore, within a given background, the placement of the topmost ear would affect, to a certain extent, the timing of anthesis. Conceivably, the lower the placement of the topmost ear (the fewer the number of thick internodes), the earlier the timing of anthesis. However, with the same total number of phytomers, the lines A44-18 and A27 shed pollen consistently earlier than the lines C24-1 and C24-2 (Table 3). This fact could be attributed to differences in the placement of the topmost ear, which on average was ~0.5 phytomers lower in A44-18 and A27 than in C24-1 and C24-2 (not shown). Given the differences in RFLP composition between A44-18, A27 and C24-1, C24-2 (Fig 5), a gene(s) controlling the position of the topmost axillary bud (ear) relative to the tassel could reside either around UMC124a or UMC12a. Isolation of additional N28 derivatives from the existing 8L-recombinant lines (Fig 5), with even smaller, unique GF fragments will help clarify the phenotypic effects of other genes introgressed along with Vgt1 and Vgt2.

Potential functions: With the data in hand we cannot distinguish between the two ultimate, potential functions of Vgt1 and Vgt2. They may act in pathways that either promote or repress the transition from vegetative to generative growth. Both GF and N28 are considered insensitive to photoperiod (RUSSELL and STUBER 1983 ), a fact that suggests that both genes are components of endogenously controlled pathway(s) of tassel initiation. Because the florally determined state is labile (IRISH and NELSON 1991 ; COLASANTI et al. 1998 ), Vgt1 and Vgt2 could be (again) expected to be expressed post-transitionally irrespective of the pathway in which they act. The similarity of juvenile-adult transition (JAT) values among the four homozygous genotypes (Table 5) suggests that the two genes are probably not involved in the transition from juvenile to adult phases of vegetative development.

From this and other QTL experiments (ABLER et al. 1991 ; KIM 1992 ; ZEHR et al. 1992 ; KOESTER et al. 1993 ; KOZUMPLIK et al. 1996 ; VLADUTU 1998 ) it appears that allelic differences at Vgt1 and Vgt2 play a significant role in the variation of maturity across diverse genetic backgrounds. Therefore, Vgt1 and Vgt2 are probably among the important loci in maize subjected to selection in geographic regions with a short growing season.

Feasibility of positional cloning of Vgt1:
To our knowledge, no maize mutant has been identified that, on the basis of map position and phenotype, could represent a strong candidate gene for either Vgt1 or Vgt2. Knock-out mutations may not have been identified by chance or because they are lethal, or, due to functional redundancy, the loss of function for either gene does not produce a striking phenotype in heterogeneous backgrounds. Because, in principle, it is difficult to accommodate the directed transposon-tagging approach for isolating genes with relatively small effects in heterogeneous backgrounds, a potentially feasible option for isolating Vgt1 and Vgt2 is positional cloning.

To date, there has been no gene isolated through positional cloning in a maize YAC library. The map-based approach for cloning maize genes has been discouraged by several features of the maize genome: large average ratio of physical/genetic distance (~1.5 Mb/cM; CIVARDI et al. 1994 ), large amount (60–80%) of medium and highly repetitive DNA sequences (HAKE and WALBOT 1980 ; SPRINGER et al. 1994 ), and the relative ease of cloning maize genes by transposon tagging. One of the strategies proposed to overcome the limitations imposed by the maize genome to positional cloning is chromosome landing (TANKSLEY et al. 1995 ). It requires flanking markers at a physical distance from the target gene that is less than the average insert size of the genomic library being used (TANKSLEY et al. 1995 ).

Having separated the GF alleles at the two linked loci, we have laid the foundation for the positional cloning of Vgt1. The additive effects of the GF allele relative to the N28 allele at Vgt-1, in the background of N28, are relatively large (compare the ND, DPS, and HT values of the GN and NN genotypes in Table 5). This allows for unambiguous inference of the allelic constitution at Vgt1 based on the phenotypic values of individuals homozygous at DNA markers closely flanking the gene.

For simplicity of discussion, thus far we have assumed that each of the two linked QTL of 8L harbors a single gene (Vgt1 and Vgt2, respectively). Although the short interval (~5 cM) delimiting the position of the major 8L-QTL (Fig 4 and Fig 5) suggests the possibility of a single gene (Vgt1) being present at the QTL, it does not guarantee it. For example, five genes encoding enzymes of the biosynthetic pathway of cyclic hydroxamic acids in maize are clustered within 6 cM on chromosome arm 4S (FREY et al. 1997 ). The presence of two (or more) linked genes within the major 8L-QTL would complicate their identification.

Generally, the choice of the parental material for QTL studies is based upon morphological and physiological differences resulting from repeated cycles of natural or human selection. Thus, the presence of closely linked genes underlying the phenotypic differences would always constitute a potential problem for isolating genes using the BDL approach. Despite the potential presence of closely linked genes affecting the traits investigated and the extensive amount of time and labor required to develop the BDLs, the construction of introgression libraries, such as the one generated by ESHED and ZAMIR 1994 in tomato, would be a useful component of structural and functional genomics in any crop species.

	ACKNOWLEDGMENTS

We thank D. Shaver for providing plant material and detailed information regarding the backcrossing procedure; B. Burr, B. Gill, S. Hake, T. Musket, T. Pryor, R. Tuberosa, J. Vogel, and W. Werr for providing DNA clones; S. Livingston, J. Suresh, R. Schierman, and C. Castell for technical help; S. Salvi for technical help and discussions; J. Doebley, R. Shaw, M. Olsen, and two anonymous reviewers for comments or discussions. This research was supported in part by the U.S. Department of Agriculture–National Research Initiative grant number 92-37300-7524 and Northrup King Inc. This article is publication number 98-1-13-0090 of the Minnesota Agricultural Research Station.

Manuscript received November 13, 1998; Accepted for publication June 25, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ABBE, E. C. and B. O. PHINNEY, 1951  The growth of the shoot apex in maize: external features. Am. J. Bot. 38:737-744.

ABLER, B. S. B., M. D. EDWARDS, and C. W. STUBER, 1991  Isoenzymatic identification of quantitative trait loci in crosses of elite maize inbreds. Crop Sci. 31:267-274[Abstract/Free Full Text].

ALPERT, K. B. and S. D. TANKSLEY, 1996  High-resolution mapping and isolation of a yeast artificial chromosome contig containing fw2.2: a major fruit weight quantitative trait locus in tomato. Proc. Natl. Acad. Sci. USA 93:15503-15507[Abstract/Free Full Text].

ALPERT, K. B., S. GRANDILLO, and S. D. TANKSLEY, 1995  fw 2.2: a major QTL controlling fruit weight is common to both red- and green-fruited tomato species. Theor. Appl. Genet. 91:994-1000.

ASKENASY, E., 1880  Ueber eine neue Methode, um die Vertheilung der Wachstumsintensität in wachsenden Theilen zu bestimmen. Verhandl. Naturhist.-Medic. Ver. Heidelberg. N.F. 2:70-153.

BASTEN, C. J., B. S. WEIR and Z. B. ZENG, 1995–1996 QTL CARTOGRAPHER: A Reference Manual and Tutorial for QTL Mapping. Department of Statistics, North Carolina State University, Raleigh, NC.

BRAWN, R. I., 1968  Breeding corn for earliness. Proc. Annu. Corn Sorghum Res. Conf. 23:59-66.

CIVARDI, L., Y. XIA, K. J. EDWARDS, P. S. SCHNABLE, and B. J. NIKOLAU, 1994  The relationship between genetic and physical distances in the cloned a1-sh2 interval of the Zea mays L. genome. Proc. Natl. Acad. Sci. USA 91:8268-8272[Abstract/Free Full Text].

COE, E. H., G. DAVIS, M. MCMULLEN, and M. POLACCO, 1995  RFLP and genetic maps. Maize Genet. Coop. Newsl. 69:247-256.

COLASANTI, J., Z. YUAN, and V. SUNDARESAN, 1998  The indeterminate gene encodes a zinc finger protein and regulates a leaf-generated signal required for the transition to flowering in maize. Cell 93:593-603[Medline].

DOEBLEY, J. and A. STEC, 1993  Inheritance of the morphological differences between maize and teosinte: comparisons of results for two F₂ populations. Genetics 134:559-570[Abstract].

DOEBLEY, J., A. STEC, and C. GUSTUS, 1995  teosinte branched1 and the origin of maize: evidence for epistasis and the evolution of dominance. Genetics 141:333-346[Abstract].

DOERGE, R. W. and G. A. CHURCHILL, 1996  Permutation tests for multiple loci affecting a quantitative character. Genetics 142:285-294[Abstract].

ESHED, Y. and D. ZAMIR, 1994  A genomic library of Lycopersicon pennelli in L. esculentum: a tool for fine mapping of genes. Euphytica 79:175-179.

ESHED, Y. and D. ZAMIR, 1995  An introgression line population of Lycopersicon pennelli in the cultivated tomato enables the identification and fine mapping of yield-associated QTL. Genetics 141:1147-1162[Abstract].

FEINBERG, A. P. and B. VOGELSTEIN, 1983  A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity. Anal Biochem. 132:6-13[Medline].

FREY, M., P. CHOMET, E. GLAWISCHING, C. STETTNER, and S. GRÜN et al., 1997  Analysis of a chemical plant defense mechanism in grasses. Science 277:696-699[Abstract/Free Full Text].

GALINAT, W. C., 1959  The phytomer in relation to the floral homologies in the American Maydeae. Bot. Mus. Leaflets, Harvard U. 19:1-32.

GERDES, J. T., C. F. BEHR, J. G. COORS and W. F. TRACY, 1993 Compilation of North American Maize Breeding Germplasm. Crop Science Society of America Inc., Madison, WI.

GREYSON, R. I., D. B. WALDEN, and W. J. SMITH, 1982  Leaf and stem heteroblasty in Zea.. Bot. Gaz. 143:73-78.

HAKE, S. and V. WALBOT, 1980  The genome of Zea mays, its organization and homology to related grasses. Chromosoma 79:251-270.

HALLAUER, A. R., and J. B. MIRANDA, 1988 Quantitative Genetics in Maize Breeding, Ed. 2. Iowa University Press, Ames, IA.

HARTLEY, H. O., 1950  The maximum F-ratio as a short-cut test for heterogeneity of variance. Biometrika 37:308-312[Free Full Text].

IRISH, E. E. and T. M. NELSON, 1991  Identification of multiple stages in the conversion of maize meristems from vegetative to floral development. Development 112:891-898[Abstract].

JANSEN, R. C., 1993  Interval mapping of multiple quantitative trait loci. Genetics 135:205-211[Abstract].

KIESSELBACH, T. A., 1949 The structure and reproduction of corn. University of Nebraska Press, Lincoln, NE.

KIM, T. S., 1992 Identification of genomic regions controlling maturity in maize (Zea mays L.). Ph.D. Thesis, University of Minnesota, St. Paul, MN.

KINIRY, J. R., and R. BONHOMME, 1991 Predicting maize phenology, pp. 115–131 in Predicting Crop Phenology, edited by T. HODGES. CRC Press, Boca Raton, FL.

KNAPP, S. J., 1994 Mapping quantitative trait loci, pp. 58–96 in DNA-Based Markers in Plants, edited by R. L. PHILLIPS and I. K. VASIL. Kluwer Academic Publishers, Dordrecht, The Netherlands.

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[Abstract/Free Full Text].

KOESTER, R. P., P. H. SISCO, and C. W. STUBER, 1993  Identification of quantitative trait loci controlling days to flowering and plant height in two near isogenic lines of maize. Crop Sci. 33:1209-1216[Abstract/Free Full Text].

KOZUMPLIK, V., I. PEJIC, L. SENIOR, R. PAVILNA, and G. GRAHAM et al., 1996  Use of molecular markers for QTL detection in segregating maize populations derived from exotic germplasm. Maydica 41:211-217.

LEJEUNE, P. and G. BERNIER, 1996  Effect of environment on the early steps of ear initiation in maize (Zea mays L.). Plant Cell Environ. 19:217-224.

LENG, E. R., 1951  Time relationships in tassel development in inbred and hybrid corn. Agron. J. 43:445-449[Free Full Text].

LINCOLN, S., M. DALY and E. LANDER, 1992 Constructing Genetic Maps with MAPMAKER/EXP 3.0, Ed. 3. Whitehead Institute Technical Report, Cambridge, MA.

MATZ, E. C., F. A. BURR, and B. BURR, 1995  Molecular map based on T x CM and Co x Tx recombinant inbred families. Maize Genet. Coop. Newsl. 69:257-267.

MILLER, R. G. Jr., 1981 Simultaneous Statistical Inference. Springer-Verlag, New York.

MORRISON, T. A., J. R. KESSLER, and D. R. BUXTON, 1994  Maize internode elongation patterns. Crop Sci. 34:1055-1060[Abstract/Free Full Text].

NAVEIRA, H. and A. BARBADILLA, 1992  The theoretical distribution of lengths of intact chromosome segments around a locus held heterozygous with backcrossing in a diploid species. Genetics 130:205-209[Abstract].

ROBERTSON, D. S., 1985  A possible technique for isolating genic DNA for quantitative traits in plants. J. Theor. Biol. 117:1-10.

RUSSELL, W. K. and C. W. STUBER, 1983  Effects of photoperiod and temperatures on the duration of vegetative growth in maize. Crop Sci. 23:847-850[Abstract/Free Full Text].

SACHS, R. M., 1965  Stem elongation. Annu. Rev. Plant Physiol. 16:73-97.

SAGHAI-MAROOF, M. A., K. M. SOLIMAN, R. JORGENSON, and R. A. ALLARD, 1984  Ribosomal DNA spacer length polymorphisms in barley: Mendelian inheritance, chromosomal location, and population dynamics. Proc. Natl. Acad. Sci. USA 81:8014-8018[Abstract/Free Full Text].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SAS INSTITUTE, 1994 Introductory Guide for Personal Computer, Version 6.01 Edition. SAS Institute, Cary, NC.

SHAPIRO, S. S. and M. B. WILK, 1965  An analysis of variance test for normality (complete samples). Biometrika 52:591-611[Free Full Text].

SHARMAN, B. C., 1942  Developmental anatomy of the shoot of Zea mays L. Ann. Bot. (N. S.) 6:245-282.

SPRINGER, P. S., K. J. EDWARDS, and J. L. BENNETZEN, 1994  DNA class organization on maize Adh1 yeast artificial chromosomes. Proc. Natl. Acad. Sci. USA. 91:863-867[Abstract/Free Full Text].

STEEL, R. G. D., and J. H. TORRIE, 1980 Principles and Procedures of Statistics: A Biometrical Approach, Ed. 2. McGraw-Hill, New York.

SUITER, K. A., J. F. WENDEL, and J. S. CASE, 1983  LINKAGE-1: a Pascal computer program for the detection and analysis of genetic linkage. J. Hered. 74:203-204[Abstract/Free Full Text].

TANKSLEY, S. D., M. W. GANAL, and G. B. MARTIN, 1995  Chromosome landing: a paradigm for map-based gene cloning in plants with large genomes. Trends Genet. 11:63-68[Medline].

UTZ, H. F., and A. E. MELCHINGER, 1996 PLABQTL: a program for composite interval mapping of QTL. J. Quant. Trait Loci, http://probe.nalusda.gov:8000/otherdocs/jqtl/jqtl1996-01/utz.html. (verified 10 Sept. 1999).

VLDUU, C. I., 1998 Fine mapping and characterization of linked QTL involved in the transition of the maize apical meristem from vegetative to generative structures. Ph.D. Thesis, University of Minnesota, St. Paul, MN.

WEHRHAHN, C. and R. W. ALLARD, 1965  The detection and measurement of the effects of individual genes involved in the inheritance of a quantitative trait in wheat. Genetics 51:109-119[Free Full Text].

ZEHR, B. E., J. W. DUDLEY, J. CHOJECKI, M. A. SAGHAI-MAROOF, and R. P. MOWERS, 1992  Use of RFLP markers to search for alleles in a maize population for improvement of an elite hybrid. Theor. Appl. Genet. 83:903-911.

ZENG, Z-B., 1994  Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc. Natl. Acad. Sci. USA. 90:10972-10976[Abstract/Free Full Text].

This article has been cited by other articles:

				N. Lauter, M. J. Moscou, J. Habiger, and S. P. Moose Quantitative Genetic Dissection of Shoot Architecture Traits in Maize: Towards a Functional Genomics Approach The Plant Genome, November 1, 2008; 1(2): 99 - 110. [Abstract] [Full Text] [PDF]

				N. C. Collins, F. Tardieu, and R. Tuberosa Quantitative Trait Loci and Crop Performance under Abiotic Stress: Where Do We Stand? Plant Physiology, June 1, 2008; 147(2): 469 - 486. [Full Text] [PDF]

				S. Ducrocq, D. Madur, J.-B. Veyrieras, L. Camus-Kulandaivelu, M. Kloiber-Maitz, T. Presterl, M. Ouzunova, D. Manicacci, and A. Charcosset Key Impact of Vgt1 on Flowering Time Adaptation in Maize: Evidence From Association Mapping and Ecogeographical Information Genetics, April 1, 2008; 178(4): 2433 - 2437. [Abstract] [Full Text] [PDF]

				O. N. Danilevskaya, X. Meng, Z. Hou, E. V. Ananiev, and C. R. Simmons A Genomic and Expression Compendium of the Expanded PEBP Gene Family from Maize Plant Physiology, January 1, 2008; 146(1): 250 - 264. [Abstract] [Full Text] [PDF]

				R. Tuberosa, S. Salvi, S. Giuliani, M. C. Sanguineti, M. Bellotti, S. Conti, and P. Landi Genome-wide Approaches to Investigate and Improve Maize Response to Drought Crop Sci., December 18, 2007; 47(Supplement_3): S-120 - S-141. [Abstract] [Full Text] [PDF]

				S. Salvi, G. Sponza, M. Morgante, D. Tomes, X. Niu, K. A. Fengler, R. Meeley, E. V. Ananiev, S. Svitashev, E. Bruggemann, et al. Conserved noncoding genomic sequences associated with a flowering-time quantitative trait locus in maize PNAS, July 3, 2007; 104(27): 11376 - 11381. [Abstract] [Full Text] [PDF]

				E. Frascaroli, M. A. Cane, P. Landi, G. Pea, L. Gianfranceschi, M. Villa, M. Morgante, and M. E. Pe Classical Genetic and Quantitative Trait Loci Analyses of Heterosis in a Maize Hybrid Between Two Elite Inbred Lines Genetics, May 1, 2007; 176(1): 625 - 644. [Abstract] [Full Text] [PDF]

				Y. Fu, T.-J. Wen, Y. I. Ronin, H. D. Chen, L. Guo, D. I. Mester, Y. Yang, M. Lee, A. B. Korol, D. A. Ashlock, et al. Genetic Dissection of Intermated Recombinant Inbred Lines Using a New Genetic Map of Maize Genetics, November 1, 2006; 174(3): 1671 - 1683. [Abstract] [Full Text] [PDF]

				A. F. Troyer Adaptedness and Heterosis in Corn and Mule Hybrids Crop Sci., February 1, 2006; 46(2): 528 - 543. [Abstract] [Full Text] [PDF]

				Y.-M. Zhang, Y. Mao, C. Xie, H. Smith, L. Luo, and S. Xu Mapping Quantitative Trait Loci Using Naturally Occurring Genetic Variance Among Commercial Inbred Lines of Maize (Zea mays L.) Genetics, April 1, 2005; 169(4): 2267 - 2275. [Abstract] [Full Text] [PDF]

				F. Chardon, B. Virlon, L. Moreau, M. Falque, J. Joets, L. Decousset, A. Murigneux, and A. Charcosset Genetic Architecture of Flowering Time in Maize As Inferred From Quantitative Trait Loci Meta-analysis and Synteny Conservation With the Rice Genome Genetics, December 1, 2004; 168(4): 2169 - 2185. [Abstract] [Full Text] [PDF]

				A. F. Troyer Background of U.S. Hybrid Corn II: Breeding, Climate, and Food Crop Sci., March 1, 2004; 44(2): 370 - 380. [Abstract] [Full Text] [PDF]

				A. Bouchez, F. Hospital, M. Causse, A. Gallais, and A. Charcosset Marker-Assisted Introgression of Favorable Alleles at Quantitative Trait Loci Between Maize Elite Lines Genetics, December 1, 2002; 162(4): 1945 - 1959. [Abstract] [Full Text] [PDF]

				S. H. Vega, M. Sauer, J. A. J. Orkwiszewski, and R. S. Poethig The early phase change Gene in Maize PLANT CELL, January 1, 2002; 14(1): 133 - 147. [Abstract] [Full Text] [PDF]