IMPROVING GRAIN YIELD AND YIELD COMPONENTS VIA BACKCROSS PROCEDURE

ABSTRACT - Transferring a quantitative trait from a donormaize (Zea may L.) line into a local elite line is an impor-tant way to improve local maize line. A backcross proce-dure is widely used method for this purpose. However,little knowledge exists on how effective a backcross pro-cedure is for improving one and/or multiple quantitativetraits of maize. The objectives of this study were to 1) de-termine the backcross generations at which a quantitativetrait was recovered to a similar homozygous level of re-current parent; 2) investigate if multiple quantitative traitscan be transferred from a donor line into a local elite linein a consecutive backcross procedure. 3) evaluate if theparental lines’ mean, GCA, and SCA effects impact on thechanges in the studied traits at different backcross genera-tions. The results had shown that four consecutive back-cross generations are necessary to make a new line recov-er to similar homozygous level of recurrent parent in thestudied quantitative traits. Multiple quantitative traits fromexotic donor line have been successfully transferred intothe new line by the consecutive backcross procedure in-corporated with key agronomic trait selection. A maizeline with high means from recurrent parent lines (RPL)and high positive GCA effects from both parent lines ispreferred to SCA effect in a cross for improving quantita-tive traits of a recipient line because fixation of additivegenes can occur in consecutive backcross generations.

KEY WORDS: Maize (Zea may L.) breeding; Consecutivebackcross procedure with key agronomic trait selection;General combining ability (GCA); Specific combining abil-ity (SCA); Quantitative trait improvement; Grain yield andyield components.

INTRODUCTION

Introducing exotic maize germplasm from otherregions/countries to improve and broaden a localmaize genetic base is a widely used method acrossthe world (ALBRECHT and DUDLEY, 1987; VASAL et al.,1992a,b; RON PARRA and HALLAUER, 1997; GOODMAN,1999; ABADASSI and HERVÉ, 2000; LI et al., 2001). Grainyield is the most important trait in maize breedingprograms and it is highly determined by yield com-ponents (FAN et al., 2008). Improvement of yieldcomponents of a local line from an exotic line is aneffective way to improve grain yield of the localmaize inbred lines (GODSHALK and KAUFFMANN, 1995;AUSTIN and LEE, 1998; FAN et al., 2002a,b, 2008).

To utilize exotic elite maize germplasm, an exot-ic line with a target trait is usually used as donorand crossed with a local elite maize line. The tradi-tional backcross procedure has been a majormethod used for transferring favorable alleles froma donor genotype to a recipient elite genotype (AL-LARD, 1960; HALLAUER and MIRANDA, 1988). It hadbeen successfully used to transfer favorable allelesfor monogenetic traits, e.g. transferring insect resist-ant genes (CRAIG et al., 2000; MUGO et al., 2005) andfor high-heritability polygenic traits, e.g. transferringgenes for controlling early flowering (RINKE andSENTZ, 1961; SHAVER, 1976).

A quantitative trait, such as maize grain yield, earlength (EL), ear diameter (ED), row number per ear(RE), kernel number per row (KR) and 1000- or 100-kernel weight (KW) is key quantitative trait in anymaize breeding program. These quantitative traits aregenerally controlled by multiple genes (FU et al.,2010). Maize breeders usually use various recurrentselection procedures to obtain an improved inbredline (DUVICK, 1974; HALLAUER and MIRANDA 1988;Pandey and Gardner 1992). Recurrent selectionbreeding procedure usually takes more time than a

Maydica 55 (2010): 145-153

IMPROVING GRAIN YIELD AND YIELD COMPONENTSVIA BACKCROSS PROCEDURE

H.M. Chen1, Y.D. Zhang1, W. Chen1, M.S. Kang2, J. Tan1, Y.F. Wang1, J.Y. Yang1, X.M. Fan1,*

1 Institute of Food Crops, Yunnan Academy of Agricultural Sciences, Kunming 650205, Yunnan Province, China2 Dep. of Agronomy & Environ. Management, Louisiana State Univ. Agric. Center, Baton Rouge, LA 70803-2110, USA

Re ceived June 1, 2010

* This research was funded by Yunnan advanced talent in-troduction project foundation (20080A006) and Major State BasicResearch Development Program of China (2009CB126003).

* For correspondence (fax: 86-871-5894923; e.mail: [email protected]).

backcross procedure to obtain an improved stablemaize inbred line that can be used in a hybridbreeding program (HALLAUER and MIRANDA, 1988; HAL-LAUER 1992; PANDEY and GARDNER, 1992). A traditionalbackcross procedure had been reported to be em-ployed by MEREDITH (1977) transferring cotton fiberstrength from a donor parent cotton variety FTA 263-20 to a recurrent parent DPL 16 in three backcrosscycles. The fiber strength of the plants at third back-crossed generations was significantly better than thatof its recurrent parent DPL 16 (MEREDITH, 1977). Theresult was achieved without selection for other traits,such as yield and lint percentage, being practiced.MEREDITH (1977) suggested that if undesired plantsand progenies were eliminated in the backcrossingprocess, the backcross procedure might be more ef-ficient in breeding program. DUVICK (1974) had suc-cessfully increased the prolificacy of C103 maize lineby crossing it to a highly prolific popcorn, backcross-ing four times to the inbred with selection in eachbackcross population for signs of the incompletelyrecessive prolific trait, and then selfing to homozy-gosity while selecting for prolificacy. These two stud-ies suggested that continuous backcrossing can beused for rapid transfer of a quantitative trait from ex-otic to adapted lines or varieties.

However, little is known about whether a favor-able maize quantitative trait, such as grain yield andfive yield components of EL, ED, RE, KR and KW,can be transferred from an exotic elite maize lineinto local elite maize lines, and if more than onefrom these quantitative traits can be transferred si-multaneously with a backcross procedure. Andmany basic questions need to be answered beforeplant breeders can apply the backcross procedureto transfer one or more maize yield components intheir maize breeding programs. The objectives ofthis study were to 1) determine the backcross gen-erations at which a quantitative trait was recoveredto a similar homozygous level of recurrent parent;2) investigate if one or multiple quantitative traitscan be transferred from donor maize line into a lo-cal elite line.3) evaluate if the parental lines’ mean,GCA, and SCA effects impact on the changes in thestudied traits at different backcross generations.

MATERIALS AND METHODS

Three tropical, elite maize inbred lines (CML161, CML166,CML171) were used as non-recurrent parent lines (NRPL), andthree temperate maize inbred lines (Ye107, K22, and Y1218)were used as recurrent parent lines (RPL). The selected NRPLs

have one or more quantitative traits being better than those ofRPLs and are target traits for transferring. Nine backcross groupswere formed. The three NRPLs performed well in most of keyagronomic traits in Yunnan environment after several years of ex-perimentation for adaptation. The three RPLs are among the mostpopular parental lines used in hybrids planted in northern China.They possess high combining ability and many favorable agro-nomic traits. Each backcross group consisted of seven popula-tions: an NRPL, an RPL, F1 of the cross between NRPL and RPL,and four consecutive backcross generations called BC1F1, BC2F1,BC3F1, and BC4F1. Information about NRPL and RPL is listed inTable 1.

The NRPLs and RPLs were planted in 2003. The three NRPLswere planted in 30 rows, each of 3 meter length, with a widerow spacing of 0.8 meter and narrow row space of 0.4 meters.Each row had seven hills with two plants per hill. The three RPLswere planted in 10 rows with the same plot size as used for NR-PL. During the flowering stage, mixed pollen from at least 100plants of each RPL was collected and used to pollinate at least100 plants of each NRPL, which made up 9 F1 populations. Theoffspring of each of the 9 F1 populations were planted and polli-nated with the mixed pollen collected from at least 100 RPLplants of Ye107, K22, and Y1218. This yielded nine BC1F1 popu-lations. The plants from each of the nine BC1F1 populations werepollinated with mixed pollen from at least 100 plants of the threeRPL lines again in next season to obtain nine BC2F1 populations.This backcross procedure was continued until nine BC3F1 andnine BC4F1 populations were obtained.

No selection was carried out on RPL plants for pollen collec-tion. However, in each backcross generation, the pollination wasonly applied to the selected plants that showed vigorous growth,resistance to lodging, and resistance to disease, and ideal agro-nomic traits. At harvest, 50 healthy ears with good seeds were se-lected; equal numbers of seeds from each of the 50 ears weremixed and used for the next generation of backcrossing. The restof the seeds were kept for heterosis testing and final field trial.

In 2006, seeds from seven generations, including NRPL, RPL,F1, BC1F1, BC2F1, BC3F1, and BC4F1, from each of the nine back-cross groups with a total of 63 maize entries were planted atKunming, Yunnan. The experiment used a split-plot design, withthe nine backcross groups being as main factor and the sevengenerations as sub-factor. Randomization, with three replications,was employed for both main factor and sub-factor. Each plotconsisted of 3 rows, each 4 meter long, with a spacing of 0.8 me-ters. Each row had nine hills with two plants per hill. A total of

146 H.M. CHEN, Y.D. ZHANG, W. CHEN, M.S. KANG, J. TAN, Y.F. WANG, J.Y. YANG, X.M. FAN

TABLE 1 - Information of recurrent (RPL) and non-recurrent (NR-PL) parental lines used for this study.––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Inbred lines Source Ecology adaptation––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––NRPL CML171 Pool25QPM Tropical

CML161 Pool25QPM Tropical

CML166 Pop66QPM Tropical––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––RPL Ye107 Foreign Hybrid XL80 Temperate

K22 K11 × Ye478 Temperate

Y1218 Variation of K12 Temperate(Huangzaosi×Huaichun)

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

56000 plants per hectare were used. Data for plant yield, EL, ED,RE, KR and KW were collected after harvest.

Also in 2006, elite maize inbred line CML171 was planted andcrossed with mixed pollen from each of RPL, F1, BC1F1, BC2F1,BC3F1, and BC4F1 from 3 RPL x 3 NRPL (i.e. 48 testcrosses in to-tal) to test if any of the test hybrids had same performance levelas the original F1 did. Data for grain yield and five yield compo-nents (EL, ED, RE, KR, and KW) were collected on all the hybrids.Seeds from the 45 testcrosses and 3 original hybrids (CML171x 3RPLs) were planted at Kunming, Yunnan, in 2007. A randomizedblock design was used with three replications. The plot consistedof three rows, each 4-meter long, with a row spacing of 0.8 me-ter. Each row had nine hills, with two plants per hill. Data collect-ed were the same as described above. Information on all RPLs,NRPLs, backcrosses and testcrosses is given in Table 2.

The GLM, mean, and correlation procedures of SAS (SAS ver-sion 9.1; SAS INSTITUTE, 2005) were used to execute the generallinear model for both the split-plot experiment and the random-

ized block experiment. General and specific combining abilityanalysis for F1 was conduced using the previous method (FAN etal., 2008).

RESULTS AND DISCUSSIONS

Generations needed for a transferredquantitative trait being similarhomozygosis as RPL

According to the heterosis theory, the purer thetwo inbred lines are and the higher the heterosis fortheir hybrid. Thus, when a backcross line (BCL) inthe consecutive backcross process reaches similarhomozygous level as its RPL, then grain yield of thetestcross between this BCL and corresponding NRPL

147IMPROVE MAIZE QUANTITATIVE TRAITS VIA BACKCROSS

Non recurrent Recurrent Crosses Generationparent lines parent lines Names†––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––CML161 Ye107 CML161xYe107 F1

F1xYe107 BC1F1

BC1F1xYe107 BC2F1

BC2F1xYe107 BC3F1

BC3F1xYe107 BC4F1

CML166 Ye107 CML166xYe107 F1

F1xYe107 BC1F1

BC1F1xYe107 BC2F1

BC2F1xYe107 BC3F1

BC3F1xYe107 BC4F1

CML171 Ye107 CML171xYe107 F1

F1xYe107 BC1F1

BC1F1xYe107 BC2F1

BC2F1xYe107 BC3F1

BC3F1xYe107 BC4F1

CML171xF1 PF1

CML171xBC1F1 PBC1F1

CML171xBC2F1 PBC2F1

CML171xBC3F1 PBC3F1

CML171xBC4F1 PBC4F1

CML161 K22 CML161xK22 F1

F1xK22 BC1F1

BC1F1xK22 BC2F1

BC2F1xK22 BC3F1

BC3F1xK22 BC4F1


F1xK22 BC1F1

BC1F1xK22 BC2F1

BC2F1xK22 BC3F1

BC3F1xK22 BC4F1

Non recurrent Recurrent Crosses Generationparent lines parent lines Names†––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––


F1xK22 BC1F1

BC1F1xK22 BC2F1

BC2F1xK22 BC3F1

BC3F1xK22 BC4F1

CML171xF1 PF1

CML171xBC1F1 PBC1F1

CML171xBC2F1 PBC2F1

CML171xBC3F1 PBC3F1

CML171xBC4F1 PBC4F1

CML161 Y1218 CML161xY1218 F1

F1xY1218 BC1F1

BC1F1xY1218 BC2F1

BC2F1xY1218 BC3F1

BC3F1xY1218 BC4F1

CML166 Y1218 CML166xY1218 F1

F1xY1218 BC1F1

BC1F1xY1218 BC2F1

BC2F1xY1218 BC3F1

BC3F1xY1218 BC4F1

CML171 Y1218 CML171xY1218 F1

F1xY1218 BC1F1

BC1F1xY1218 BC2F1

BC2F1xY1218 BC3F1

BC3F1xY1218 BC4F1

CML171xF1 PF1

CML171xBC1F1 PBC1F1

CML171xBC2F1 PBC2F1

CML171xBC3F1 PBC3F1

CML171xBC4F1 PBC4F1

TABLE 2 - Parent lines and their backcross and test generations.–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––† All crosses with P at beginning are the test crosses between CML171 and related backcrossed generations.

should be similar to that of original F1 between RPLand NRPL. The same theory should apply to the fiveyield components.

Forty-eight testcrosses by using CML171 as atester and the each of forty five lines (viz. F1, BC1F1,BC2F1, BC3F1, and BC4F1 from three NRPLs x threeRPL) and three RPLs as line were evaluated. Analysisof variance (ANOVA) for grain yield and other fiveyield components traits was conducted and variancesfrom the generation were significantly for all traitsbut KR (data not shown). Since ANOVA were signifi-cant for grain yield, EL, ED, RE, and KW, the genera-tion means of these five traits were compared on thesix generations (Table 3). Least significant difference(LSD) test was employed for all mean comparisons.The results from table 3 showed that only the meansof testcrosses between CML171 and BC4F1 consistent-ly were statistically the same as or better than that ofthe F1 from CML171 with 3 RPL (F1) for all five traitsand other testcrosses (viz. CML171* F1, BC1F1, BC2F1,and BC3F1) had one or more means were significant-ly less than those of the F1. These results stronglysuggested that the plants of BC4F1 had reached simi-lar homozygous levels as their RPL plants had. Inother words, four-consecutive backcross is necessaryor good enough for recovering a quantitative back toits recurrent parent’s genetic structure.

Changes in grain yield and five yieldcomponents in four backcross generations

Genetic theory tells us that any cross betweentwo maize inbred lines can show some heterosis(HALLAUER and MIRANDA, 1988), a trait value of F1 isusually higher than values of both parental inbredlines. With the consecutive backcross procedureemploying key agronomic trait selection, the quanti-tative trait from RPL not only should recovered intoRPL level but also it should be improved due to in-corporation of more favorable genes from NRPL. Wehad classified 45 means at BC4F1 for the five traitsfrom nine backcrosses into three groups: means be-ing between RPL and NRPL, being higher than high-er parent, and being lower than lower parent. Thenumbers of means in the three groups, then, werecounted and listed in table 4 for two differentgroups of recurrent parents, one being higher thanNRPL and another being lower than NRPL. Chisquare test indicated that whatever RPLs were loweror higher than NRPL, there were significant highernumber (p<0.01) of means at BC4F1 either beinghigher than its RPL or in between RPL and NRPL.These results have indicated that the favorablequantitative traits not only from RPL but also fromNRPL have been integrated into a new developedline.


TABLE 3 - Means of grain yield testcrosses between CML171 and 3 recurrent parent lines (RPL) and their F1, BC1F1, BC2F1, BC3F1, and BC4F1.–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––Generations† Grain Mean per plant (g) Ear Length (cm) Ear Diameter (cm) Row of kernel per ear 100 kernel weight (g)–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––PRPL 172.222a‡ 17.5111a 5.3889b 14.889a 34.278a

PBC4F1 174.444a 17.2778ab 5.6222a 15.5556a 32.778ab

PBC3F1 138.889c 15.6667c 5.0333cd 13.7778bc 29.2c

PBC2F1 153.333b 16.5bc 5.0444cd 13.5556c 30.367bc

PBC1F1 134.444c 15.944c 5.1556c 15.3333a 29.622c

PF1 140bc 16.3889bc 4.8667d 14.6667ab 29.767c

LSD0.05 14.6 1.0091 0.2208 1.104 2.5036–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––† PRPL = CML171*Recurrent parent line; PBC4F1 = CML171*BC4F1; P BC3F1 = CML171*BC3F1; P BC2F1 = CML171* BC2F1; P BC1F1. =CML171*BC1F1; PF1. = CML171*F1.‡ Values followed by the same letter in the same column are not significantly different (P < 0.05).

TABLE 4 - Counts of testcrosses with trait values being higher than higher parent, lower than lower parent, and fell in between two parentsafter four consecutive back crossing†.–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––Recurrent parent Between parents Higher than higher parent Lower than lower parent–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––Higher 3 5 2

Lower 24 6 5

Sum 27 11 7–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––† Total 45 testcrosses based on 5 generations of F1, BC1F1, BC2F1, BC3F1, and BC4F1 from three testcrosses between three recurrentparental lines and three non-recurrent parental lines.


FIGURE 1- Means of grain yield (1a.1 and 1a.2) and the five yield components of ear length (1b.1 and 1b2), ear diameter (1c.1 and 1c.2),row of ear (1d.1 and 1d.2), kernel number per row (1e.1 and 1e.2), and 100-kernel weight (1f.1 and 1f.2) between three non-recurrentparent lines (NRPL) of CML161(blue), CML166(red), and CML171 (yellow) and two recurrent parent lines of Ye107(left panel) and K22(right panel), respectively. The aqua lines are the means from the three crosses between the three NRPLs and either of Ye107 or K22. P1 isRPL and P2 is NRPL.

The above phenomenon was also observed fromFigure 1 that plotted the means of grain yield andthe 5 yield components from the backcrosses be-tween three NRPLs of CML161, CML166, CML171and two RPLs of Ye107 and K22 (the data fromcrosses between three NRPLs and Y1218 are not in-cluded to reduce the number of graphs in the fig-ure). For example, the values of the five yield com-ponents were approaching its RPLs (P1) in general(Fig, 1b.1, 1b.2, 1c.1, 1c.2. 1d.1, 1d.2, 1e.1, 1e.2,1f.2). Interestingly, when RPLs had higher valuessuch as on grain yield (Fig. 1a.2), ED (Fig. 1c.2),and RE (Fig. 1d.2), the means at BC4F1 for thesetraits were all higher than or close to RPL means.While when NRPL means were higher than those ofRPLs, most of the means at BC4F1 were betweenRPL and NRPL (Fig. 1), with strong influence of NR-PL were observed on EL, (Fig. 1b.2), and KR (Fig.1e.1), KW (Fig. 1f.2). These results were consistentwith table 4 results and had strongly suggested thatthis special consecutive backcross procedure withselections on key agronomic traits were able totransfer a quantitative trait from donor line (i.e. NR-PL) into another line.

Factors influencing trait performancein consecutive backcross generations

As shown in Fig. 1, F1 values and correspondingtrait values in different backcross generation werequite different in the different crosses. A new statis-tic called value decrease rate (VDR) was calculatedby formula of (F1-BC4F1)/F1. To understand the un-

derlying reasons for this difference, we calculatedeight new variables viz. means of RPL and NRPL,difference of RPL minus NRPL, general combiningability (GCA) of RPL and NRPL, specific combiningability (SCA), total General Combining ability (TG-CA = GCA of RPL + GCA of NRPL), and total com-bining ability (TCA) (TCA=GCA from RPL + GCAfrom NRPL+ their SCA). The correlation coefficientsbetween F1, VDR, BC4 and the eight variables werecomputed and given in Table 5.

Several interesting points were found from Table5 and Fig. 1 and 2. First, all three selected NRPLshave more than one traits being better that RPLsand they were used as donor in this experiment.From Fig. 2, we have seen that in most of the cross-es, all traits at BC4F1 were improved due to highervalues from NRPLs than those from RPLs. For exam-ples, the traits EL (Fig. 1b.2), ED (Fig. 1c.2), RE (Fig.1d.2), and KW (Fig. 1f.2) from local elite line of K22had been greatly improved by three NRPLs ofCML161, CML166, and CML171. This result furtherconfirmed previous conclusion on the effectivenessof the special consecutive backcross procedure em-ployed by this study for transfering quantitativetraits from donor line into recurrent lines.

Second, the VDRs of grain yield, ED and REwere significantly negatively correlated with RPLmean and the difference between RPL and NRPL.That means the higher the RPL is, the slower VDRdecreases. The results from Table 5 also showedthat the means of ED and RE from RPL were signifi-cantly correlated with the corresponding trait means


TABLE 5 - Correlation coefficients between trait value decreasing rate (VDR), F1 (F1), BC4F1(BC4) and eight derived variables.–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Derived VariableED EL GY KR KW RE

VDR F1 BC4 VDR F1 BC4 VDR F1 BC4 VDR F1 BC4 VDR F1 BC4 VDR F1 BC4–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––RPL_NRPL -0.87** 0.3 0.92** -0.15 0.40 0.57 -0.70* 0.015 0.84** -0.32 -0.12 0.33 -0.12 -0.09 0.05 -0.90** 0.46 0.96**

RPL_Mean -0.85** 0.35 0.92** -0.36 0.34 0.76* -0.70* 0.01 0.84** -0.26 -0.10 0.27 -0.30 -0.09 0.25 -0.85** 0.56 0.97**

NRPL_Mean 0.37 0.66* -0.02 -0.27 -0.22 0.12 -0.08 -0.20 -0.06 0.30 0.11 -0.32 -0.23 0.02 0.28 0.41 0.70* 0.01

GCA_RPL -0.33 0.64 0.58 -0.21 0.41 0.67* 0.09 0.23 0.07 0.36 0.48 -0.08 0.67* 0.80** -0.06 -0.86** 0.56 0.97**

GCA_NRPL 0.40 0.66* -0.04 0.23 0.40 0.17 0.19 0.25 -0.10 0.13 0.11 -0.10 0.10 0.33 0.19 0.40 0.80** 0.06

TGCA 0.06 0.92** 0.37 0.01 0.58 0.60 0.20 0.34 -0.02 0.38 0.49 -0.10 0.66* 0.87** 0.02 -0.15 0.98** 0.60

SCA -0.13 0.4 0.29 0.70* 0.82** 0.00 0.44 0.94** 0.02 0.52 0.87** -0.03 -0.03 0.50 0.53 0.18 0.21 -0.01

TCA 0.01 1 0.45 0.58 1 0.35 0.48 1 0.01 0.64 1 -0.07 0.56 1 0.28 -0.11 1 0.59–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––1 Values followed by *, ** are significant at 0.05 and 0.01 levels, respectively.2 ED=ear diameter; EL=ear length; GY=grain yield; KR=kernel row per ear; KW=1000-kernel weight; RE=row number per ear.3 RPL_NRPL=recurrent parent line – non-recurrent parent line; RPL_mean=mean of recurrent parent line; NRPL_mean=mean of non-recur-rent parent line; GCA_RPL=general combining effect of recurrent parent line; GCA_NRPL=general combining effect of non-recurrent parentline; TGCA=total general combining ability= (GCA_RPL+GCA_NRPL); SCA=specific combining ability; TCA=total combining ability =(GCA_RPL+GCA_NRPL+SCA).

at BC4F1 and no means at BC4F1 showed significantcorrelation with means from NRPL. These results in-dicated that selecting a line with higher trait meanbeing used as RPL might have better chance to ob-tain a stable new line having high value for the traitwhen the line gets stabilized at BC4F1.

Third, the GCA effects of RPL for EL and RE werealso significantly correlated with the means of thecorresponding traits at BC4F1 and no traits at BC4F1showed significant correlation with both GCA effectsof NRPL and SCA effects (Table 5). These results im-plied that GCA effects from RPL were most impor-tant than NRPL’s GCA and the SCA effects of thecrosses between RPL and NRPL for quantitative traitperformace at BC4F1. Interestingly, from Figs. 1 and2, we further found that high positive SCA effects ongrain yield (Fig. 2a.1), on EL (Fig. 2b.1), and on KR(Fig. 2e.1) for the cross of CML161xYe107 and on EL(Fig. 2b.2) and on KR (Fig. 2e.2) for the cross ofCML171xK22 were main contributors to their high F1values. These high trait values had decreased quick-er (Fig. 1a.1, 1b.1, 1e.1, 1b.2, 1e.2) compared withother traits. It seemed that the high SCA effects inthese crosses were not transferred and/or carriedover to the next backcross generations. In contrast,the high F1 values on ED (Fig. 2c.1, 2c.2) in thecrosses of CML161xYe107 and CML161xK22, respec-tively, on RE (Fig. 2d.1) for the cross CML166xYe107,on EL (Fig. 2b.2) for the cross CML166xK22, and onKW (Fig. 2f.2) for the cross CML161xK22, weremainly caused by large positive GCA effects fromNRPLs. The high values of these traits decreased atmuch slower rate in the corresponding backcrossgenerations (Fig. 1c.1, 1c.2, 1d.1, 1b.2, 1f.2, and1d.2). It seemed that when there were high positiveGCA effects in one or both parent lines, the trait val-ue in the backcross generation decreased muchslower than in the crosses without a parent line hav-ing positive GCA effect. This result indicated thatGCA effects seemed to be fixed in the consecutivebackcross generations whereas the SCA effects werecontinuously reduced in the consecutive backcrossgenerations. This result definitely confirmed the ba-sic quantitative genetic theory that GCA effect is fix-able and SCA effect is not fixable. Thus, when weplan to improve an inbred line with a specific weakyield component, we should select another maizeline with high positive GCA effect for that specifictrait.

With the rapid development of molecular-markertechnology, more reliable techniques for identifyingQTL for maize grain yield and yield components are

expected to be developed. Thus, Marker assistedbackcross (MAB) (BENCHIMOL et al., 2005; RIBAUT andRAGOT, 2007), Marker assisted recurrent selection(MARS) (MOREAU et al., 2004), and other new mark-er-assisted selection technologies should be applica-ble to maize breeding programs aimed at transfer-ring quantitative trait from one line to another lineor improving local maize lines in one or two yieldcomponents. However, even at that time, based onthis study, the consecutive backcross procedurecould still be an easier, reliable, cost-effective wayof improving yield components of local lines. Thereason for this may be that most of the yield com-ponent traits were controlled by genes located atcommon chromosome segments as revealed bycommon QTL found by several researchers (VELD-BOOM and LEE, 1996; RIBAUT et al., 1997; AUSTIN andLEE, 1998). Another reason for this consecutivebackcross procedure’s success may be that the se-lection in each backcross generations on vigorousgrowth, resistance to lodging, and resistance to dis-ease, and ideal agronomic traits had effectively ac-cumulated favorable genes or alleles for multipletarget quantitative traits.

Though data from this experiment showed thatin BC4F1, most of traits reached a similar homozy-gous level as its RPL, authors think it would be agood idea to advance the backcross generations toBC5F1 or BC6F1 for affirming our observations.

CONCLUSIONS

This study suggested that the consecutive back-cross procedure with selection on vigorous growth,resistance to lodging, and resistance to disease, andideal agronomic traits in each backcross generationis an effective way to transfer maize grain yield andyield components from a donor line into a recipientline. The data from this experiment has shown thatfour consecutive backcrosses were necessary to re-cover a quantitative trait back to reach at the samehomozygous level as in the local RPL as well as tosuccessfully transfer multiple quantitative traits fromdonor maize line into local elite lines. Moreover,our result strongly suggested that a line with highmeans from RPL and high positive GCA effects fromboth parent lines is preferred to a line with highpositive SCA effect in across for improving grainyield or any of the five yield components of a recip-ient line because fixation of additive genes can oc-cur in consecutive backcross generations.



FIGURE 2 - Histograms of the GCA effects from the three non-recurrent parent lines (NRPL) (purple) of CML161, CML166, and CML171 andtwo recurrent parent lines (RPL) (blue) of Ye107 and K22, the SCA effect (yellow) for grain yield (2a.1 and 2a.) and the five yield compo-nents of ear length (2b.1 and 2b.2), ear diameter (2c.1 and 2c.2), row of ear (2d.1 and 2d.2), kernel number per row (2e.1 and 2e.2), and100-kernel weight (2f.1 and 2f.2) in the crosses between three NRPLs and Ye107 (left) and K22 (right).

REFERENCES

ALLARD R.W., 1960 Principles of plant breeding. John Wiley &Sons, New York.

ABADASSI J., Y. HERVÉ, 2000 Introgression of temperategermplasm to improve an elite tropical maize population.Euphytica 113: 125-133.

ALBRECHT B., J.W. DUDLE, 1987 Evaluation of 4 maize popula-tions containing di erent proportions of exotic germplasm.Crop Sci. 27: 480-486.

AUSTIN D.F., M. LEE, 1998 Detection of quantitative trait loci forgrain yield and yield components in maize across genera-tions in stress and nonstress environments. Crop Sci. 38:1296-1308.

BAYLES M.B., L.M. VERHALEN, L.L. MCCALL, W.M. JOHNSON, B.R.BARNES, 2005 Recovery of recurrent parent traits when back-crossing in cotton. Crop Sci. 45: 2087-2095.

BENCHIMOL L.L., C.L. SOUZA, A.P. SOUZA, 2005 Microsatellite-assist-ed backcross selection in maize. Genet. Mol. Biol. 28: 789-797.

BRIGGS F.N., R.W. ALLARD, 1953 The current status of the back-cross method of plant breeding. Agron. J. 45: 131-138.

CRAIG A., R.L. WILSON, B.R. WISEMAN, W.H. WHITE, F.M. DAVIS, 2000Conventional resistance of experimental maize lines to cornearworm (Lepidoptera: Noctuidae), fall armyworm (Lepi-doptera: Noctuidae), southwestern corn borer (Lepidoptera:Crambidae), and sugarcane borer (Lepidoptera: Crambidae).J. Econ. Entomol. 93: 982-988.

DUVICK D.N., 1974 Continuous backcrossing to transfer prolifi-cacy to a single-eared inbred line of maize. Crop Sci. 14: 69-71.

FAN X.M., J. TAN, Z.L. CHEN, J.Y. YANG, 2002a Combining abilityand heterotic grouping of ten temperature, tropical, and sub-tropical quality protein maize inbreds. pp. 10-18. In: G. Srini-vasan et al. (Eds.), Proc. 8th Asian Regional Maize Workshop:New Technologies for the New Millennium, Bangkok, Thai-land, 5-8 Aug. 2002. CIMMYT, Mexico.

FAN X.M., J. TAN, J.Y. YANG, F. LIU, B.H. HUANG, Y.X. HUANG, 2002bStudy on combining ability for yield and genetic relationshipbetween exotic tropical, subtropical maize inbreeds and do-mestic temperate maize inbreeds. Sci. Agric. Sin. 35: 743-749.

FAN X.M., H.M. CHEN, J. TAN, C.X. XU, Y.D. ZHANG, Y.X. HUANG,M.S. KANG, L.M. LUO, 2008 Combining Abilities for Yieldand Yield Components in Maize. Maydica 53: 39-46.

FROVA C., P. KRAJEWSKI, N.D. FONZO, M. VILLA, M.S. GORIA, 1999Genetic analysis of drought tolerance in maize by molecularmarkers I. Yield components. Theor. Appl. Genet. 99: 280-288.

GOODMAN M.M., 1999 Broadening the genetic diversity in maizebreeding by use of exotic germplasm. pp. 139-148. In: J.G.Coors, S. Pandey (Eds.), The genetic and exploitation of het-erosis in crops. ASA-CSSA-SSSA, Madison, WI.

GODSHALK E.B., K.D. KAUFFMANN, 1995 Performance of exotic xtemperate single cross maize hybrids. Crop Sci. 35: 1042-1045.

HALLAUER A.R., 1985 Compendium of recurrent selection meth-ods and their application. Crit. Rev. Plant Sci. 3: 1-33.

HALLAUER A.R., 1992 Recurrent selection in maize. Plant Breed.Rev. 9: 115-119.

HALLAUER A.R., J.B. MIRANDA, 1988 Quantitative genetics in maizebreeding. 2nd ed. Iowa State University Press. Ames, IA.

LI X.H., S.Z. XU, J.S. LI, J.L. LIU, 2001 Heterosis among CIMMYTpopulation and Chinese key inbred lines in maize. ActaAgron. Sin. 27: 575-581.

MEREDITH W.R. JR., 1977 Backcross breeding to increase fiberstrength of cotton. Crop Sci. 17: 172-175.

MOREAU L., A. CLARCOSSET, A. GALLIAIS, 2004 Experimental evalua-tion of several cycles of marke-assisted selection. Euphytica137: 111-118.

MUGO S., H.D. GROOTE, D. BERGVINSON, M. MULAA, J. SONGA, S.GICHUKI, 2005 Developing Bt maize for resource-poor farm-ers-Recent advances in the IRMA project. African J. Biotech-nol. 4: 1490-1504.

OPENSHAW S.J., S.G. JARBOE, W.D. BEAVIS, 1994 Marker-assistedselection in backcross breeding. pp. 41-43. In: Proc. Sympo-sium “Analysis of Molecular Marker Data”, Joint Plant Breed-ing Symposia Series, American Soc. Horticultural Science/Crop Science of America, Corvallis, Oregon.

PATERNIANI E., 1990 Maize breeding in the tropics. Crit. Rev.Plant Sci. 9: 125-154.

PANDEY S., C.O. GARDNER, 1992 Recurrent selection for popula-tion variety, and hybrid improvement in tropical maize. Adv.Agron. 48: 1-87.

RIBAUT J.M., C. JING, D. GONZALEZ-DE-LEON, G.O. EDMEADS, D.A.HOISINGTON, 1997 Identification of quantitative trait loci un-der drought conditions in tropical maize. 2. Yield compo-nents and marker-assisted selection strategies. Theor. Appl.Genet. 94: 887-896.

RIBAUT J.M., M. RAGOT, 2007 Marker-assistant selection to im-prove drought adaptation in maize: the backcross approach,perspectives, limitations, and alternatives, J. Exper. Botany58: 351-360.

RINKE E.H., J.C. SENTZ, 1961 Moving Corn Belt germplasm north-ward. pp. 53-56. In: W. Heckendorn, J. Sutherland (Eds.),Proc. 16th Annu. Corn and Sorghum Int. Res. Conf., Am. SeedTrade. Assoc., Washington, DC.

RON PARRA J., A.R. HALLAUER, 1997 Utilization of exotic maizegermplasm. Plant Breed. Rev. 14: 165-187.

SAS INSTITUTE, 2005 SAS user’s guide: statistics. SAS Institute,Cary, NC.

SHAVER D.L., 1976 Conversion for earliness in maize inbreds.Maize Genet. Coop. Newsletters 50: 20-23.

VASAL S.K., G. SRINIVASAN, J. CROSSA, D.L. BECK, 1992a Heterosisand combining ability of CIMMYT’s subtropical and temper-ate early-maturity maize germplasm. Crop Sci. 32: 884-890.

VASAL S.K., G. SRINIVASAN, F. GONZÁLEZ C., G.C. HAN, S. PANDEY,D.L. BECK, J. CROSSA, 1992b Heterosis and combining abilityof CIMMYT’s tropical x subtropical maize germplasm. CropSci. 32: 1483-1489.

VELDBOOM L.R., M. LEE, 1996 Genetic mapping of quantitativetrait loci in maize in stress and nonstress environments: I.Grain yield and yield components. Crop Sci. 36: 1310-1319.


Documents

IMPROVING GRAIN YIELD AND YIELD COMPONENTS VIA BACKCROSS PROCEDURE