1) Agronomy Department, Zhejiang Agricultural University, Hangzhou, 310029, China
2) Institute of Genetics, Academia Shinica, Beijing, 100101, China
Since most characters of economic importance are quantitatively
inherited in rice (Oryza sativa L.), mapping quantitative trait loci (QTLs)
is of central importance in rice genetics and breeding. The discovery of
uniquitous molecular genetic markers such as restriction fragment length
polymorphisms (RFLPs) provided geneticists and plant breeders with ideal
markers for mapping and following the individual genes underlying the
continuous variation. Although several statistical models have been proposed
for QTL mapping based on molecular markers, the approaches and the relevant
programs, including interval mapping described by Lander and Botstein
(1989), have been widely considered by animal/plant breeders as being
difficult to understand and this has hindered the efficient use of the
methods (Luo and Kearsey 1992). In this research, an F\2\ population was
derived from a cross between Indica variety Zhe-Ye-Qing 8 (ZYQ 8) and
Japonica variety Jing-Xi 17 (JX 17), and an RFLP framework map was
constructed. A series of computer programs suitable for use on a personal
computer with low memory capacity were designed. Four QTL mapping approaches
were used for ten quantitative traits including days to heading, plant height
and grain weight (Table 1).
Using Mapmaker computer software package (Lander et al. 1987), an RFLP map was constructed based on the segregation data of 59 RFLP markers and 76 F\2\ individuals, which encompassed 11 chromosomes except chromosome 9.
By mean analysis, phenotypic differences of quantitative traits among alternative marker genotypes were used to detect associations between markers and traits. 21% RFLP markers were identified to be significantly associated with at least one trait, and these markers were unevenly distributed on different chromosomes. Tightly linked markers usually showed associations with the same trait. One-way analysis of variance among alternative marker genotypes supported the results from mean analysis (Table 1).
Using maximum likelihood methods described by Luo and Kearsey (1989), the computer program was designed in this research. The estimator (r) of
Table 1. OTL mapping in ZYQ 8/JX 17 F\2\ based mean analysis (A/a), one-way
analysis (B/b), maximum likelihood estimation (C/c) and interval mapping
(D/d) for days to heading (1), tiller angle (2), 200-grain weight (3), plant
height (4), panicle number per plant (5), panicle length (6), spikelet number
per plant (7). Spikelet density (8), grain number per panicle (9) and
fertility (10). Capitalized and lower-case characters show
significance at the 1% and 5% levels, respectively
============================================================================= Chr Markers 1 2 3 4 5 6 7 8 9 10 ============================================================================= 1 RG1O1-RG220 ABCd Ab d Bd RG220-RG780 d d 3 RG104-RG227 a A b d a d RG227-RG369 A b a 4 RG620-RG776 a a b Ab Abc b RG776-RG449 a abCd d d d Abd Abd ab 5 RG556-RG360 a RG360-RG229 a d D RG229-RG573 ab b b b RG573-RG13 ab bd b b 6 Waxy-RG64A Ab ab ab D D RG64A-RG716 d RG716-RG172 b bcD bcD RG172-RG653 b bcD bcD RG653-RG244 Cd c 7 RG678-RG511 AC RG511-RG528 AC 8 RG29-RZ617 b D d d a RZ6]7-RG1034 ABD ABcd acd ad RG1034-RG28 ABD ABcd ab ABcd a RG28-RG1 ABD ABd ab AB a RGI-RG136 a a 10 RG134-RG241 b d 11 RG303-RG103 D D bd RG103-RG2 d D RG2-RG167 a A b RG167-RG211 a A b 12 RG436-RG323 C RG323-RG218 C RG361-RG457 c RG457-RG869 Ab Abc b RG869-RG81 Ab Ab b b RG81-RG235 ac a b b =============================================================================
Using the interval mapping procedure described by Lander and Botstein (1989) and made it attainable by van Ooijen (1992), a generalized computer program which is suitable for F\2\, DH, BC populations was designed in this research. QTL likelihood maps can be produced when the results are transferred into a system such as Quattro Pro 3. The LOD score was calculated for positions at each 1 cM between the two nearest flanking markers. When the maximum LOD score for a chromosome exceeds the predefined significance threshold, 3, a potential site of a QTL is determined. Table 1 gives the flanking marker-trait combinations whose LOD scores are larger than 2. Although identified QTLs were often located within 1 cM away from the marker locus, some of them existed between two flanking markers spacing more than 2 cM. For some specific quantitative traits, more than one QTL was revealed by one molecular marker. The correlated traits showed quite similar QTL likelihood maps (Fig. 1), which suggested that these traits might be governed by a set of QTLs or an identical polygenic system.
Most associations (linkages) between markers and traits were confirmed in this report by more than one method if they exist. QTLs controlling a trait were often associated with the markers on some specific chromosomes, and their distribution on chromosomes was uneven. An integration of the results from above four methods determined 17 QTLs including one for days to heading, two for tiller angle, two for plant height, five for spikelet number and seven for spikelet density.
In order to improve the power of QTL mapping, the following approaches can be exploited. (1) Selective RFLP genotyping of the only individuals (lines) from the high and low phenotypic tails of the entire sample population can increase the power of QTL mapping even with the same population size. (2) Using permanent segregating populations such as DH, RI and SSD to replicate phenotypic measurement can reduce the environmental variance. (3) The isogenic lines differing only in the region containing a QTL can be used to eliminate the majority of the genetic variance, and make it possible to dissect the remaining interval by examining various recombinants for flanking markers. (4)
Fig. 1. QTL likelihood map on chromosome 6 for spikelet number per panicle
(solid line) and spikelet density (dotted line) in ZYQ 8/JX 17 F\2\ population.
When an assumption holds that all alleles increasing the quantitative trait
on any given chromosome are fixed in one parent while all the decreasing
alleles are fixed in the other, the phenotypic difference between parents
becomes larger, and QTL mapping should have higher efficiency. In rice, we
have developed a DH population for molecular mapping. Using this population
in QTL mapping is under way.
References
Lander, E.S. and D. Botstein. 1989. mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121: 185-199.
Lander, E.S., P. Green, J. Abrahason, A. Barlow, M.J. Daly, S.E. Lincoln and L. Newburg. 1987. MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1: 174-181.
Luo, Z.W. and M.J. Kearsey. 1989. Maximum likelihood estimation of linkage between a marker gene and a quantitative locus. Heredity 63: 401-408.
Luo, Z.W. and M.J. Kearsey. 1992. Interval mapping of quantitative trait loci in an F\2\ population. Heredity 69: 236-242.
Miller, R.G. 1974. The jackknife-a review. Biometrika 61: 1-15.
van Ooijen, J.W. 1992. Accuracy of mapping quantitative trait loci in autogamous species. Theor. Appl. Genet. 84: 803-811.