Genetic analysis and genotype 3 environment (G 3 E)for grey leaf spot disease resistance in elite African maize(Zea mays L.) germplasm

Julia Sibiya • Pangirayi Tongoona •

John Derera • Neil van Rij

Received: 2 November 2010 / Accepted: 20 May 2011 / Published online: 1 June 2011

� Springer Science+Business Media B.V. 2011

Abstract Maize grey leaf spot (GLS) disease

remains an important foliar disease in sub-Saharan

Africa accounting for more than 25% yield losses in

maize. Information on inheritance of GLS resistance

of germplasm adapted to African environments is

required in new sources being identified. Therefore,

hybrids generated from a 10 9 10 half-diallel mating

of tropical advanced maize inbred lines were evalu-

ated in six environments to determine combining

ability, genotype 9 environment interaction (G 9 E)

and the impact of GLS disease on grain yield.

General combining ability effects were highly sig-

nificant and accounted for 72 and 68% of the

variation for GLS resistance and grain yield, respec-

tively. Significant specific combining ability effects

associated with reduced disease levels were observed

in some hybrids when one parent was resistant, and

these may be exploited in developing single cross

maize hybrids. Regression analysis showed a

260–320 kg ha-1 decrease in maize grain yield per

each increase in GLS disease severity score, and

significant associations (r = -0.31 to -0.60) were

observed between grain yield and GLS severity

scores. This showed the potential of GLS disease to

reduce yield in susceptible varieties grown under

favourable disease conditions, without control mea-

sures. Genotype and genotype 9 environment biplots

and correlation analysis indicated that the significant

G 9 E observed was not due to changes in hybrid

ranking, implying absence of a significant crossover

interaction. Therefore, predominance of additive gene

effects imply that breeding progress for GLS disease

resistance would be made through selection and this

could be achieved at a few hot-spot sites, such as

Baynesfield and Cedara locations in South Africa,

and still deploy the resistant germplasm to other

environments in which they are adapted.

Keywords Combining ability � Gene action �Genotype 9 environment interaction � Grey leaf spot

disease � Maize (Zea mays L.) � Yield loss

Introduction

Maize grey leaf spot (GLS) disease caused by

Cercospora spp. is currently one of the most impor-

tant foliar diseases in sub-Saharan Africa (SSA) and

is now endemic to the region (Menkir and Ayodele

J. Sibiya (&) � P. Tongoona � J. Derera

African Centre for Crop Improvement, School

of Agricultural Sciences and Agribusiness, University

of KwaZulu-Natal, P. Bag X01, Scottsville,

Pietermaritzburg 3209, South Africa

e-mail: jsibiya@excite.com; juliasibiya@hotmail.com

N. van Rij

Crop Protection, Cedara, KwaZulu-Natal Department

of Agriculture, Environmental Affairs and Rural

Development, P. Bag X9059, Pietermaritzburg 3200,

South Africa

123

Euphytica (2012) 185:349–362

DOI 10.1007/s10681-011-0466-2

2005; Vivek et al. 2010). More than one species of

Cercospora spp. has been associated with GLS

disease of maize (Zea mays L), namely C. zeae-

maydis Group I and II. Group I was reported to be

dominant in the USA but also occurs elsewhere in the

world, while Group II occurs in the USA, Africa and

possibly elsewhere (Dunkle and Levy 2000).

Recently, Group II, which is genetically and pheno-

typically distinct from Group I, has been demon-

strated to be a distinct species and renamed

Cercospora zeina (Crous et al. 2006). Analysis of

samples from South Africa, Zimbabwe and Zambia,

further confirmed the presence of C. zeina and

absence of C. zeae-maydis in commercial maize

plantations in southern Africa (Crous et al. 2006;

Meisel et al. 2009).

GLS disease of maize results in yield losses of

around 10–25% annually, but losses as much as 90%

due to severe deterioration of the leaves and stalk

lodging have also been recorded (Latterell and Rossi

1983; Ward et al. 1997a, b). This has serious

consequences on food security because maize is the

predominant staple crop for millions of people in the

SSA. Management of GLS disease has focused

mostly on the use of fungicides and genetic resistance

(Gordon et al. 2006), but most smallholder farmers in

the SSA have limited access to chemicals because of

their prohibitive cost (Ward et al. 1999). Conse-

quently, control of GLS disease in the smallholder

sector in Africa is difficult when susceptible varieties

are grown. Therefore, it would be prudent to breed

new maize varieties with acceptable levels of GLS

disease resistance to minimize grain yield losses.

Diverse sources and the genetic basis of the

resistance to GLS have been identified and reported

by different researchers (Thompson et al. 1987; Huff

et al. 1988; Donahue et al. 1991; Gevers et al. 1994;

Coates and White 1998; Menkir and Ayodele 2005;

Derera et al. 2008; Vivek et al. 2010). However, the

majority of them have been conducted in temperate

environments with temperate germplasm that cannot

be used directly in African’s tropical environments.

Only a few sources of resistance have been identified

from some tropical and subtropical African germ-

plasm up to now (Gevers et al. 1994; Menkir and

Ayodele 2005; Derera et al. 2008; Vivek et al. 2010).

Gevers et al. (1994) observed high resistance in the

South African maize cultivars (KO54W and SO507),

which contain the white modified opaque-2 gene and

belong to the (F) and (M) heterotic groups, respec-

tively. However, these subtropical quality protein

maize inbreds are low yielding, and may not be

adapted to tropical conditions in SSA. Although some

resistant sources have been reported for heterotic

groups used in southern African breeding pro-

grammes and other maize lines adapted to African

tropical conditions, the nature and magnitude of GLS

disease resistance requires further analysis in some of

these elite inbreds (Menkir and Ayodele 2005; Derera

et al. 2008; Vivek et al. 2010). In addition, given the

potential that GLS disease has to threaten food

security, more sources of GLS disease resistance

would be useful, especially those that can contribute

resistance to inbred lines that are susceptible to GLS;

but are high yielding and are widely used in the

hybrid maize production in Africa. The new sources

of GLS resistance should be well characterised and

the gene action and combining ability (CA) of the

lines established. Moreover, information is still

limited on the mode of inheritance of GLS disease

resistance for most of the germplasm that are adapted

to African environments (Derera et al. 2008). Such

information is required to optimise their breeding

strategies.

GLS disease resistance has been shown to be

controlled mainly by additive gene effects (Thomp-

son et al. 1987; Huff et al. 1988; Ulrich et al. 1990;

Gevers et al. 1994; Derera et al. 2008; Vivek et al.

2010). However, some studies have also indicated

dominant gene action to have a significant role in

conferring resistance in some maize hybrids (Elwin-

ger et al. 1990; Coates and White 1998; Derera et al.

2008; Vivek et al. 2010). On the other hand, Hohls

et al. (1995) reported that resistance was conditioned

by additive and complete dominance with minor

epistasis in a diallel comprising maize lines from

three divergent backgrounds in South Africa.

In addition, production environments in the SSA

are highly variable and stress-prone resulting in

complicated genotype 9 environment (G 9 E) inter-

actions (FAO and CIMMYT 1997). Therefore, dis-

eases including GLS disease are often difficult to

control, as their occurrence year after year is less

predictable because of their high dependence on

prevailing weather conditions (Vivek et al. 2010).

The severity of most diseases thus tends to vary with

locations or environments resulting in significant

G 9 E (Levy and Pataky 1992; Egesi et al. 2009).

350 Euphytica (2012) 185:349–362

123

This G 9 E complicates breeding and selection for

GLS disease resistance as the phenotypic expression

of the host may vary across the different environ-

ments (location and season), because of pathogen-

sensitivity to environmental changes (Carson et al.

2002). Vivek et al. (2010) reported significant G 9 E

for a number of important maize foliar diseases,

which included GLS disease, northern corn leaf

blight (NCLB) disease, common rust disease, and

Phaeosphaeria leaf spot (PLS) disease, which was

due to differences in disease pressure influenced by

the prevailing weather conditions in the different

environments.

This study was therefore conducted to (i) determine

the CA and types and magnitude of gene action for

resistance to the GLS disease among selected inbred

maize lines using a diallel analysis, (ii) investigate

G 9 E for the GLS disease resistance in different

environments as location and season and, finally (iii)

determine the impact of GLS disease on yield of the

hybrids developed from a diallel analysis of selected

elite germplasm.

Materials and methods

Germplasm

Maize inbred lines from heterotic groups adapted

mostly to sub-tropical and tropical, mid-altitude

environments were sourced from the International

Maize and Wheat Improvement Center (CIMMYT)

programme in Harare, Zimbabwe and the Crop

Breeding Institute in Zimbabwe (Table 1). The impor-

tant heterotic groups included the broad CIMMYT

(A), (B) and (AB) classification (CIMMYT 2001), the

SC (Southern Cross), N (Northern cross) and the

(P) heterotic group (Potchefstroom Pearl), and these

are widely used in eastern and southern Africa (Gevers

and Whythe 1987; Mickelson et al. 2001). The

CIMMYT heterotic group (A) is mainly derived from

populations and lines which include Tuxpeno, Kitale,

BSSS (Iowa Stiff Stalk Synthetic), Salisbury white and

B73, whereas the CIMMYT heterotic group (B) com-

prises derivatives from populations and lines related to

ETO, Ecuador 573, Lancaster, Southern cross and

Mo17 (CIMMYT 2001). Maize lines with white grain

colour only were used and selected based on their

reactions to the GLS disease.

Experimental design and management

Ten advanced maize inbred lines (Table 1) were

crossed in a half diallel-mating scheme in 2006/2007

season at the University of KwaZulu-Natal (UKZN),

Pietermaritzburg, South Africa. The resulting 45

single cross, F1 hybrids plus 9 control varieties were

evaluated in 2007/2008 and 2008/2009 seasons

across six environments. The locations included two

subtropical environments; Cedara (CED), South

Africa (30�160 E, 29�320 S, 1076 m asl) and Baynes-

field (BF) Estate, South Africa (30�210 E, 29�460 S,

758 m asl); and two tropical environments; Rattray

Arnold Research Station (RARS), Zimbabwe

(31�140 E, 17�400 S, 1300 m asl), and Mpongwe,

Zambia (ZAMB) (28�80 E, 13�310 S, 1219 m asl).

Plantings at Cedara, were done in November 2007

(CED108), January 2008 (CED208) and November

2008 (CED09), while at RARS plantings were in

December 2007 (RARS08), Mpongwe, January 2008

(ZAMB08), and Baynesfield Estate in December

2008 (BF09) giving a total of six environments. The

parents plus four inbred controls were evaluated in

trials adjacent to the hybrid trials, but only in three

environments (CED108, CED09 and BF09).

The F1 hybrids and control hybrid varieties were

laid out in the field in two replications using a 9 9 6

alpha lattice design in each locality. The inbred parents

and their controls were planted in a randomized

complete block design with two replications. Plot sizes

for the hybrids and inbred parental lines in each

locality were two rows, 3–5 m long, with 0.75 m inter-

row spacing and 0.3 m intra-row spacing. Two seeds

per hill were planted and later thinned to one resulting

in plant population densities of approximately 44,000

per hectare in all the six environments. Two to four

rows each of a susceptible maize hybrid and inbred

line were used as borders to enhance disease infection

and give an indication of the disease severity.

Standard agronomic practices were followed

according to the recommendations. The rates of

fertilizer (kg ha-1) applied at the sites were as follows:

Cedara and Baynesfield, 120 kg N, 33 kg P, 44 kg K;

RARS, 176 kg N, 70 kg P, 35 kg K; and Mpongwe

208 kg N, 35 kg P, 21 kg K. Fields were left to natural

incidence of the disease without artificial inoculations.

The total rainfall amounts received at each of the

environments during the growing season were;

729 mm at CED108, 301 mm at CED 208 and

Euphytica (2012) 185:349–362 351

123

603 mm at CED09 (Agricultural Research Council-

ISCW AgroMet Potchefstroom 2009), 844 mm at

BF09 (Baynesfield Estate 2009), 806 mm at RARS

and 950 mm at ZAMB08 (Seed Co. Zimbabwe and

Zambia 2009, respectively).

GLS disease severity was assessed twice at mid-

silking (GLS1) and at hard dough stages (GLS2), based

on visual assessment of the whole plot using a slight

modification of the 1–9 rating scale of Munkvold et al.

(2001) as follows; 1 = 0%, 2 = \1%, 3 = 1–3%,

4 = 4–6%, 5 = 7–12%, 6 = 13–25%, 7 = 26–50%,

8 = 51–75% and 9 = 75–100% leaf area showing

disease symptoms. The scores were further classified

into the following disease reaction types; 1.0 = symp-

tomless, 2.0–4.0 = resistant, 4.1–5.0 = moderately

resistant, 5.1–6.0 = moderately susceptible, 6.1–9.0 =

susceptible (Derera 2005). A Pearson correlation anal-

ysis was performed using PROC CORR module in the

SAS statistical software between GLS1 and GLS2 scores

and there was a strong positive correlation between the

two scores (r = 0.94, P B 0.001). However, the score

recorded at the hard-dough stage was used for the

statistical analysis as it reflected the total amount of

disease at the end of the season.

Other records were taken following the standard

practice used at CIMMYT (CIMMYT 1985) and these

included; the number of days to anthesis (number of

days to 50% pollen shed), and grain yield (kg). For grain

yield, ears were harvested on a whole plot basis and the

fresh ear weight (kg) determined. Five ears from each

plot were shelled to determine moisture content (%) at

harvest and shelling percentages. The field weight was

then used to estimate the grain yield (t ha-1) adjusted to

12.5% grain moisture content (CIMMYT 1985).

Statistical analysis

Analyses of variance were performed for GLS

disease severity and grain yield (t ha-1) data of the

hybrids and the hybrid controls using PROC GLM

procedure in SAS 9.1 (SAS Institute Inc 2002). Data

for disease severity (score 1–9) and grain yield

(t ha-1) were also analysed for the CA (Griffing’s

Method IV) using the Diallel SAS05 module in the

SAS statistical software (Zhang et al. 2005). The F1

hybrids and the inbred parents were treated as fixed

effects in the statistical analysis and environments

(both spatial and temporal environments) as random

effects. Only the 45 F1 experimental hybrids were

used in the analysis for CA effects. To estimate the

general combining ability (GCA) and specific com-

bining ability (SCA) effects; Griffing’s diallel anal-

yses, Model 1 (fixed genotype effects), Method 4

(crosses only) were used according to the linear

model for analysis of variance across environments:

Yijkl ¼ lþ vij þ rk þ rvð Þijkþeijkl

where Yijkl is the observed measurement for the ijth

cross in the kth replication, and lth environment; l is

the overall mean, vij is the genotype effect which is

Table 1 Pedigrees and heterotic groups for the parent inbred lines used to generate hybrids in the diallel mating scheme

Designation Pedigree or population (OPVs) Heterotic groupinga

A1220-4 [(CML395/CML444)-B-4-1-3-1-B/CML395//SC/

ZM605#b-19-2-X]-1-2-X-1-1-BBBBBB]-7-1-3-2-BBB

B/SC

A16 S89500F2-2-2-1-1-B*5-BBB A

CML445 [[TUXPSEQ]C1F2/P49-SR]F2-45-7-5-1-BBB AB

CML488 DTPWC8F31-4-2-1-5-BBB AB

CZL00001 INTA-191-2-1-2-BBBB A

CZL00009 INTA-F2-192-2-1-1-1-BBBBB A

MP18 [[[NAW5867/P30SR]-111-2/[NAW5867/P30SR]-25-1]-9-2-3-

B-2-B/CML388]-B- 35-2-B-1-#-1-BB

A/P

MP59 [AC8342/IKENNE{1}8149SR//PL9A]C1F1-500-4-X-1-1-BB-1-BB AB

MP82 [EMSR]#B#bF101sr-2-1-sr-3-2-4-b-b B

N3-2-3-3 Salisbury White N

a SC, N and P are principal heterotic groups for the national programme in Zimbabwe and lines in these groups were derived from

landraces Southern cross, Salisbury white (Northern cross) and Natal Potchefstroom Pearl (Mickelson et al. 2001)

352 Euphytica (2012) 185:349–362

123

equal to gi ? gj ? sij, where gi and gj, are GCA effects

for the ith and jth parents, respectively; sij is the SCA

effect for the ijth cross; rk the replication effect; (rv)ijk

the interaction between the ijth cross within the kth

replication and eijkl is the error term for the Yijkl

observation. The assumption was that the eijkl was

normally and independently distributed with

mean = 0 and variance = re2. The G 9 E interaction

terms were used to test for the significance of the

corresponding genetic effects (Zhang and Kang 1997).

The environments and replications within environ-

ments were considered random and therefore tested

against the residual error term. The relative importance

of GCA and SCA in predicting progeny performance

was determined by the proportion that each contributed

to the total sum of squares of the hybrids (Cisar et al.

1982) and by calculating the GCA/SCA ratio of the

mean squares (Kasuga and Inoue 2001).

Pearson correlation coefficients for the GLS disease

severity scores with grain yield were computed using

the PROC CORR SAS 9.1 (SAS Institute Inc 2002). In

addition, a simple linear regression analysis was also

used to determine the impact of GLS disease severity

scores on grain yield of the experimental hybrids for all

the environments using PROC REG SAS 9.1(SAS

Institute Inc 2002). The genotypes in each environment

were also ranked based on their reaction to the GLS

disease using the PROC RANK in SAS 9.1. The

Spearman rank order correlation coefficients were then

calculated on the genotype rank scores module in SAS

9.1 statistical software.

Genotype 9 environment analysis

The variation due to genotypes and G 9 E for the

GLS disease was explained using the genotype and

genotype 9 environment (GGE) biplot based on the

principal component analysis (PCA) of environment-

centred data (Yan et al. 2000). The GGE biplot

provided a visual relation among the hybrids and test

environments and was performed with Genstat 12

software (Payne et al. 2010) using the model based on

singular value decomposition (SVD) of the first two

principal components (Yan 2002) as follows:

Yij � l� bj ¼ k1ni1gj1 þ k2ni2gj2 þ eij

where: Yij is the yield mean of ith hybrid in jth

environment, l is the grand mean, bj is the main

effect of environment j, l ? bj is the mean yield

across all hybrids in environment j, k1 and k2 are the

singular values (SV) for the first and second principal

component (PC1 and PC2), respectively, ni1 and ni2

are the eigen vectors of hybrid i for PC1 and PC2,

respectively, g1j and g2j are the eigen vectors of

environment j for PC1 and PC2, respectively and eij is

the residual associated with hybrid i in environment j.

The ‘ideal’ environment for screening the geno-

types should have a high PC1 value, that is, it

provides better disease discrimination and PC2 values

near zero, which is a closer representative of the

environment mean (Yan et al. 2000). To visualize the

performance of the genotypes in terms of their

reaction to GLS, in each environment and groups of

environments, a polygon view (Yan 2002) was drawn

by connecting hybrids that were furthest from the

biplot origin such that all hybrids were enclosed

within the polygon. Perpendicular lines were then

drawn to each side of the polygon starting from the

biplot origin (Yan 2002).

The biplot was also used to explore the interrela-

tionships among environments by constructing lines

(environment vectors) from the biplot origin to

markers for the environments. An angle of zero

indicated a correlation of ?1, while an angle of 90�or -90�, a correlation of zero, and an angle of 180�, a

correlation of -1 (Yan 2002). The length of the

vectors was also used to determine the discriminating

ability of each of the test environments, with a shorter

vector implying that the environment was not well

represented by PC1 and PC2 (Yan et al. 2007).

Results and discussion

Hybrids and controls presented in Table 2 were

selected based on anthesis days and high yield

(t ha-1, that is, the top one-third high-yielding

hybrids presented for the late maturing group). There

was significant variation (P B 0.05) for the GLS

disease scores amongst the F1 hybrids. Most of the

hybrids exhibited a resistant reaction (scores 2.0–4.0)

across the six environments. Baynesfield location

(BF09) had the highest scores ranging from (1.0–9.0)

for GLS disease severity, whereas RARS08 and

CED208 had the lowest scores ranging from

(1.0–6.0) and (1.0–5.0), respectively (data not

shown). In addition, means for the GLS disease

Euphytica (2012) 185:349–362 353

123

scores of the susceptible hybrid and inbred line used

as border rows at the different environments ranged

from 5.0 to 9.0 depending on the environment (data

not shown). This showed that the disease pressure

was relatively high in all the localities. There is

continuous maize cultivation at all the locations and

high humidity, high temperatures and high summer

rainfall; that is, conditions favourable for the GLS

disease development (Rupe et al. 1982; de Nazareno

et al. 1993; Bhatia and Munkvold 2002).

In addition, most of the fields at Cedara (CED108

and CED09) and Baynesfield (BF09) localities that

had high disease pressure were under reduced tillage

suggesting high initial inoculum levels in the plant

debris and soil, which could have contributed to the

high disease levels. Besides, these two sites are

Table 2 Results of selected F1 hybrids and controls tested across six environments between 2007 and 2009 for the GLS disease

severity scores (1–9) (Munkvold et al. 2001), grain yield (t ha-1) and their rankings and days to 50% anthesis

Entry Cross Yield (t ha-1) Yield ranka Anthesis GLS score (1–9) GLS rankb

Hybrids with anthesis between 67 and 74 days

12 A1220-4 9 CZL00009 8.3 6 74 1.0 51

22 CZL00009 9 N3-2-3-3 8.0 9 73 1.8 33

26 CZL00009 9 MP59 7.7 21 74 1.3 39

21 CZL00009 9 CZL00001 7.3 28 73 1.2 46

27 CZL00009 9 CML488 6.6 42 72 2.1 29

23 CZL00009 9 MP82 6.5 44 72 1.3 39

25 CZL00009 9 MP18 6.2 51 72 1.3 39

Controls

57 PAN6227 6.8 38 74 1.3 39

62 SC513 6.6 41 73 1.5 38

56 PAN6017 6.2 50 72 2.2 28

Hybrids with anthesis between 75 and 82 days

16 A1220-4 9 A16 9.0 1 79 3.5 14

15 A1220-4 9 MP82 8.7 2 77 1.6 36

24 CZL00009 9 A16 8.7 3 75 1.3 39

37 N3-2-3-3 9 A16 8.5 4 79 7.1 1

47 A16 9 MP18 8.3 5 77 3.3 15

13 A1220-4 9 CZL00001 8.1 7 77 1.1 49

48 A16 9 MP59 8.0 10 80 1.7 34

39 N3-2-3-3 9 MP59 7.9 12 77 4.3 10

14 A1220-4 9 N3-2-3-3 7.8 13 78 5.6 5

31 CZL00001 9 A16 7.8 15 76 1.2 46

2 CML445 9 A1220-4 7.7 17 79 2.6 23

5 CML445 9 N3-2-3-3 7.7 18 77 6.2 2

3 CML445 9 CZL00009 7.7 19 75 1.1 49

9 CML445 9 MP59 7.7 20 78 2.6 23

Controls

60 SC633 8.1 7 75 3.8 12

61 SC719 8.0 11 83 2.1 29

63 PAN67 7.8 13 79 5.5 6

Entry mean 7.3 77 2.7

LSD (0.05) 1.0 1.3 0.5

a For grain yield; the lower the yield ranking, the higher the yield for the hybridb For mean disease severity scores; the lower the GLS disease ranking, the more susceptible the hybrid to GLS disease infection

354 Euphytica (2012) 185:349–362

123

located within the KwaZulu-Natal mist belt (Fair-

banks and Benn 2000; Derera et al. 2008) resulting in

conditions of prolonged fog or dew that are condu-

cive for the GLS disease development. However, the

late planting at Cedara (CED208) had a much lower

disease pressure. In this environment, a mid-season

drought was experienced from mid January 2008 to

end of March 2008 resulting in unfavourable condi-

tions for GLS disease development, which could have

contributed to the low disease score. At Rattray

Arnold Research Station, Zimbabwe and Mpongwe,

Zambia, deep ploughing is practiced (Derera et al.

2008), which could lead to a reduction in inoculum at

the beginning of the season, thus contributing to the

relatively lower disease levels observed compared to

Cedara and Baynesfield.

Combining ability (CA) estimates, gene

action(s) and sources of resistance

The combined analysis across environments showed

highly significant (P B 0.001) environment, entry,

GCA and SCA main effects and all the interactions

for both GLS disease severity scores and grain yield

(Table 3). This implied that both additive and non-

additive gene effects were important for the resis-

tance to GLS disease and grain yield in the maize

inbred lines used. The GCA effects accounted for a

greater proportion of the sum of squares than the SCA

effects [GCA for GLS disease severity (72%) and

grain yield (68%), SCA for GLS disease severity

(28%) and for grain yield (t ha-1) (32%)]. The

GCA:SCA ratio based on the mean squares (Table 3)

also indicated the relative importance of additive

effects vs. the non-additive gene effects (Kasuga and

Inoue 2001) for GLS disease severity and grain yield

in this study. Studies by other researchers using

different maize populations have indicated similar

findings where additive gene action was more

important than non-additive gene action (Derera

et al. 2008; Vivek et al. 2010). In contrast, studies

conducted in the USA using temperate germplasm

reported 100% GCA contribution to the variation for

GLS resistance (Thompson et al. 1987; Ulrich et al.

1990), implying that it would be possible to deter-

mine progeny performance for GLS resistance for

those materials based on GCA alone. However, since

the results in each case apply to the specific reference

populations used, the variations observed amongst

different researchers are therefore a result of the

different maize lines, environments used and possibly

GLS isolates.

For disease resistance, negative GCA and SCA

effects are desirable (Bookmyer et al. 2009). The

GCA effects, GLS mean scores and mean grain yield

(t ha-1) for the ten parents are presented in Table 4.

Table 3 Analysis of variance for GLS disease scores of 45 F1 hybrids, tested over six environments between 2007 and 2009 and the

percentage contribution of the different genetic effects to the total entry sum of squares

Source df MS (GLS 1–9) MS (yield t ha-1)

Environment 5 36.73*** 367.78***

Rep. environment 6 0.38 ns 7.17***

Hybrid 44 29.58*** 6.72***

GCA 9 103.39*** 22.35***

SCA 35 10.60*** 2.70*

Environment 9 Hybrid 220 2.05*** 3.41***

GCA 9 Environment 45 4.93*** 6.07***

SCA 9 Environment 175 1.31*** 2.73**

Pooled error 264 0.45 1.87

GCA (%) 71.50 68.04

SCA (%) 28.49 32.01

GCA:SCAa 9.75 8.27

ns non-significant (P [ 0.05)

*, **, *** Significant at P B 0.05, P B 0.01 and P B 0.001, respectivelya GCA:SCA based on the mean squares (MS) (Kasuga and Inoue 2001)

Euphytica (2012) 185:349–362 355

123

Ta

ble

4E

stim

ate

of

gen

eral

com

bin

ing

abil

ity

(GC

A)

effe

cts

for

the

GL

Sd

isea

sesc

ore

san

dg

rain

yie

ld(t

ha-

1)

mea

ns

fro

msi

xen

vir

on

men

tsb

etw

een

20

07

and

20

09

Par

ent

GL

SG

CA

per

env

iro

nm

ent

Acr

oss

env

iro

nm

ents

CE

D1

08

CE

D2

08

RA

RS

08

ZA

MB

08

CE

D0

9B

F0

9G

LS

Mea

nS

core

(1–

9)

Dis

ease

reac

tio

n

GL

S

GC

A

Mea

ng

rain

yie

ld

(th

a-1)

Yie

ld

GC

A

A1

22

0-4

-0

.23

ns

0.0

5n

s-

0.2

8n

s-

0.3

9*

*0

.49

**

*-

0.3

0*

2.0

R-

0.1

1n

s3

.10

.69

**

A1

60

.40

**

0.8

0*

**

0.9

8*

**

1.0

5*

**

0.6

8*

**

1.2

0*

**

6.3

S0

.85

**

*1

.70

.78

**

*

CM

L4

45

1.0

3*

**

-0

.08

ns

0.3

5n

s0

.43

**

0.8

7*

**

0.8

9*

**

5.0

S0

.58

**

*3

.80

.12

ns

CM

L4

88

0.0

9n

s0

.68

**

*0

.10

ns

0.4

6*

*0

.36

*1

.95

**

*4

.8M

S0

.60

**

*2

.7-

0.6

0*

*

CZ

L0

00

01

-1

.79

**

*-

0.8

9*

**

-0

.90

**

*-

1.2

0*

**

-1

.26

**

*-

2.1

8*

**

1.3

R-

1.3

7*

**

2.5

-0

.24

ns

CZ

L0

00

09

-1

.54

**

*-

0.7

6*

**

-0

.59

**

-1

.45

**

*-

1.4

5*

**

-2

.55

**

*1

.0R

-1

.39

**

*4

.40

.13

ns

MP

18

0.2

8n

s0

.11

ns

0.7

3*

**

0.2

4n

s-

0.8

3*

**

-0

.49

**

*5

.0M

S0

.01

ns

2.4

-0

.60

**

MP

59

0.2

8n

s-

0.3

3*

-0

.90

**

*-

0.7

0*

**

-0

.89

**

*-

0.4

3*

**

1.5

R-

0.4

9*

**

2.4

0.0

1n

s

MP

82

-1

.10

**

*-

0.6

4*

**

-0

.46

*-

0.5

1*

**

-0

.26

ns

-0

.80

**

*2

.0R

-0

.63

**

*2

.7-

0.4

4n

s

N3

-2-3

-32

.59

**

*1

.05

**

*0

.98

**

*2

.11

**

*2

.30

**

*2

.70

**

*8

.8S

1.9

5*

**

3.8

0.1

4n

s

SE

(gi-

gj)

0.1

50

.15

0.2

00

.16

0.1

60

.13

0.1

20

.24

LS

D (0.0

5)

1.0

1.1

ns,

no

n-s

ign

ifica

nt

(P[

0.0

5),

Rre

sist

ant,

Ssu

scep

tib

le,

MS

mo

der

atel

ysu

scep

tib

le

*,

**

,*

**

Sig

nifi

can

tat

PB

0.0

5,

PB

0.0

1an

dP

B0

.00

1,

resp

ecti

vel

y

356 Euphytica (2012) 185:349–362

123

The GCA effects were negative and highly significant

(P B 0.001) for CZL00009, CZL00001, MP82 and

MP59 across environments, indicating that they were

good general combiners for GLS disease resistance.

These same inbreds had resistant scores (Table 4) and

contributed to resistant hybrids and are therefore,

potential sources of resistance to GLS in breeding

programmes. Another potential source of resistance

was A1220-4 (Table 4). However, the line had a non-

significant negative GCA effect for GLS resistance,

but a significant positive GCA effect for grain yield.

This suggested that it contributed towards higher

yields in most of the hybrids.

The SCA estimates for the 45 F1 hybrids are

shown in Table 5. Across the environments, ten

hybrids had significant (P B 0.05), negative SCA

effects. Most of these hybrids were crosses between

parents with varying levels of disease resistance,

particularly resistant (R) 9 susceptible (S) lines,

especially those crossed to susceptible lines N3-2-3-

3, A16 and CML445. Some examples of these

hybrids include CZL00001 (R) 9 N3-2-3-3 (S);

CZL00009 (R) 9 N3-2-3-3 (S), MP82 (R) 9 N3-2-

3-3 (S), CZL00001 (R) 9 A16 (S), CZL00009

(R) 9 A16 (S), MP59 (R) 9 A16 (S), MP82

(R) 9 MP18 (S), CML445 (S) 9 CZL00001 (R),

CML445 (S) 9 CZL00009 (R). In general, the good

performance of the hybrids based on the SCA effects

corresponded to at least one of the parental lines

having a good GCA effect for disease resistance.

These results showed that susceptible parents could

be used in combination with resistant parents to

produce resistant hybrids. Therefore, the significant

SCA effects that were observed towards reduced

disease imply that non-additive gene effects can be

utilized in hybrid development. Similar results were

reported by Cromley et al. (2002) on temperate maize

germplasm and Menkir and Ayodele (2005) on some

tropical maize lines adapted to mid-altitude environ-

ments, between crosses of resistant and susceptible

parents. Therefore, breeders would select the other

parent based on some other criterion such as high

grain yield. For example, N3-2-3-3, which was

amongst the most susceptible parents to GLS, is

arguably the most productive inbred in eastern and

southern Africa. Therefore, inbreds with dominant

genes for the GLS disease resistance could be

combined with N3-2-3-3 to obtain highly productive

hybrids with acceptable levels of the GLS disease

resistance.

The positive GCA effects across environments for

the disease scores were observed for inbred lines

A16, CML445, CML488 and N3-2-3-3. These same

lines had high disease scores for the GLS disease

severity. The most susceptible hybrids were crosses

containing one of these parents; N3-2-3-3, A16,

CML445 and CML488. In addition, A16 had a

significant (P B 0.001), positive GCA effect for grain

yield (Table 4), implying that it can contribute to

high yields. This line (A16) was derived from

CML312 which, historically, has been amongst

CIMMYT’s most successful lines (CIMMYT 2001).

Table 5 Mean estimates of specific combining ability (SCA) effects (below diagonal) and mean disease severity scores (above

diagonal) for the GLS disease in six environments for the set of diallel crosses among ten maize inbred lines

A1220-4 A16 CML445 CML488 CZL00001 CZL00009 MP18 MP59 MP82 N3-2-3-3

A1220-4 3.5 2.6 3.2 1.1 1.0 2.3 1.7 1.6 5.6

A16 0.17 4.3 3.1 1.2 1.3 3.3 1.7 4.8 7.1

CML445 -0.47 0.23 3.7 1.0 1.1 3.3 2.6 3.3 6.2

CML488 0.09 -0.95*** -0.10 1.9 2.1 3.0 3.2 2.3 5.7

CZL00001 -0.03 -0.90*** -0.80** 0.10 1.2 1.3 1.0 1.6 2.1

CZL00009 -0.09 -0.80** -0.69* 0.29 1.34*** 1.3 1.3 1.3 1.8

MP18 -0.23 -0.11 0.16 -0.19 0.11 0.05 2.7 1.2 5.0

MP59 -0.32 -1.28*** -0.09 0.47 0.28 0.63* 0. 57 1.0 4.3

MP82 -0.27 1.94*** 0.71* -0.22 0.99*** 0.68* -0.80** -0.46 1.3

N3-2-3-3 1.15*** 1.69*** 1.05*** 0.53 -1.09*** -1.40*** 0.45 0.20 -2.58***

LSD(0.05) for the GLS mean disease severity scores = 0.54

*, **, *** Significant at P B 0.05, P B 0.01 and P B 0.001, respectively

Euphytica (2012) 185:349–362 357

123

The line CML312 is widely used by CIMMYT in

their breeding programmes as a tester. Dhliwayo

et al. (2009) also indicated that CIMMYT line

CML312 could contribute significantly to the breed-

ing programme of the International Institute of

Tropical Agriculture (IITA) based on the multi-

location evaluations of yield and CA they conducted.

Lines such as CML488 are amongst the most

promising CIMMYT lines in terms of their contribu-

tions to yield under drought stress conditions (CI-

MMYT 2001). Although these lines (A16, CML445,

N3-2-3-3 and CML488) can be crossed with resistant

lines to obtain resistant hybrids, it would also be a

good strategy to improve them for the GLS disease

resistance.

Impact of the GLS disease on maize grain yield

Yields of some selected maize hybrids (selected

based on high yield and anthesis days) and their

rankings based on the mean grain yields (t ha-1)

across the six environments are presented in Table 2.

Yields were relatively high in BF09, CED108,

CED09 and ZAMB08 environments, with the envi-

ronment means ranging from about 7.6–9.4 t ha-1

compared with CED208 (3.4 t ha-1) and RARS08

(6.8 t ha-1) (data not shown). Hybrids at the anthesis

stage between 67 and 74 days had the resistant inbred

line CZL00009 in common and yields ranging from

6.1 to 8.3 t ha-1. Some of the susceptible hybrids

(top 10 for the GLS disease severity susceptibility

ranking) such as N3-2-3-3 9 A16, A1220-4 9 N3-2-

3-3, CML445 9 N3-2-3-3, and N3-2-3-3 9 MP59

were apparently amongst the top 20 high yielding

hybrids (Table 2). All these hybrids had N3-2-3-3

inbred parent in common. This finding confirms the

contribution of N3-2-3-3 to high yields and should,

therefore, only be used in crosses with GLS disease

resistant lines in hybrid production or alternatively,

the GLS disease resistance be introduced in the line.

A simple regression analysis of the grain yield on

the GLS disease severity scores, was significant

(P B 0.001) for three environments (CED108,

CED09 and BF09; Table 6), with the exception of

CED208, RARS08 and ZAMB08. The slope of the

regression was negative (-0.26 to -0.32) in all the

three environments, indicating suppression of grain

yields of maize by 260–320 kg ha-1 per each

increase in GLS disease severity score. However,

the R2 values ranged from 22 to 36% implying that

the regression models in all the four environments

accounted for less than 36% of the total variation

through their linear relationship. This indicated that

the GLS disease severity was not the only factor

affecting grain yield in these environments, which

also supports the observation that some susceptible

hybrids were among the highest yielding in all the

environments. Significant and negative correlations

(-0.45 to -0.60, P \ 0.001) between grain yield

(t ha-1) and GLS scores were also observed in the

Table 6 Linear regression and Pearson’s correlation analysis of grain yield (t ha-1) on the GLS disease scores in three environments

that had significant regression models

Environment Regression parameters Regression equation Correlation

coefficient (r)b

Intercept

(a) ± se

Slope

(b) ± seaCoefficient of determination

(R2, %)

CED108 8.62 ± 0.82 -0.26 ± 0.06 25 Yield = 8.62 - 0.26GLS

severity score

-0.50***

CED09 10.60 ± 0.25 -0.32 ± 0.08 22 Yield = 10.60 - 0.32GLS

severity score

-0.45***

BF09 9.74 ± 0.23 -0.31 ± 0.05 36 Yield = 9.74 - 0.31GLS

severity score

-0.60***

CED108 Cedara November 2007 planting, CED09 Cedara November 2008 planting, BF09 Baynesfield December 2008 planting,

se standard error

*** Significant at P B 0.001a Linear regression model is given as y = a - bxb Pearson correlation coefficients for 45 hybrids

358 Euphytica (2012) 185:349–362

123

same environments (Table 6). This showed a moder-

ate to strong negative correlation between the GLS

disease severity scores and grain yield.

These results are in agreement with reports made by

other researchers on grain yield losses especially

during severe GLS disease epidemics. For example,

Ward et al. (1997a, b) in their studies on grain yield

loss potential of GLS on maize in South Africa

observed reductions of 41.7 to 43.3 kg ha-1 for each

1% increase in disease severity. However, in seasons

less conducive to the GLS disease development, Ward

et al. (1996) reported yield losses of around 38% (for

susceptible hybrids) and 20% (for moderately resistant

hybrids). In the current study, only three environments

had significant correlations and linear regression

models between the GLS disease severity scores and

grain yield, confirming the observation that the GLS

disease is less predictable in different seasons and

locations as it is highly dependent on favourable

weather conditions for disease development and

severity (Ward et al. 1999; Vivek et al. 2010). These

results also show the potential that the GLS disease

still has to suppress grain yields especially when

susceptible hybrids are grown under high disease

pressure and favourable weather conditions, stressing

the need for resistant germplasm.

Genotype 9 environment and screening sites

The response of the hybrids to the GLS disease

severity was different in the six environments as

depicted by the highly significant (Hybrid 9 Envi-

ronment) observed (Table 3). Means for the GLS

disease severity scores of the susceptible hybrid and

inbred line used as border rows at the different

environments ranged from 4.0 to 9.0 depending on

the environment.

However, although the environments appeared

different, there were significant (P B 0.001) and

positive Spearman’s rank order correlation coeffi-

cients for the GLS disease severity scores between

pairs of test environments (Table 7). This implied

that the reaction of the hybrids across the environ-

ments was consistently similar suggesting similar

ranking of the hybrids for the GLS disease resistance

in the different environments. Thus any of these

environments would be sufficient for selection of

the GLS disease resistant germplasm. This type of

interaction, which does not involve changes in

ranking, does not create any serious problems as

breeding for specific adaptation is not required. Other

researchers have reported similar interactions where

hybrid ranking remained the same and only the GLS

disease severity at the different locations and years

changed (Lipps 1998; Carson et al. 2002; Menkir and

Ayodele 2005; Derera et al. 2008; Vivek et al. 2010).

This implies that, in order to reduce costs in regional

breeding programmes, selection for GLS resistance

can be done at one reliable site and still deploy

resistant lines or hybrids to other environments in

which they are adapted.

From the GGE biplots, the first two principal

components (PCs) explained 87.88% (PC1 =

81.80% and PC2 = 6.08%) of the total GGE variation

for GLS disease severity. Results of the polygon view

of the GGE biplot is presented in Fig. 1. This biplot

indicated the most susceptible hybrid(s) for each

environment and the groups of environments. The

Table 7 Spearman’s rank order correlation coefficients of the GLS disease scores between pairs of test environments

CED108 CED208 RARS08 ZAMB08 CED09 BF09

CED108 1

CED208 0.68*** 1

RARS08 0.61*** 0.65*** 1

ZAMB08 0.75*** 0.72*** 0.76*** 1

CED09 0.71*** 0.68*** 0.67*** 0.80*** 1

BF09 0.82*** 0.75*** 0.64*** 0.87*** 0.81*** 1

CED108 Cedara November 2007 planting, CED208 Cedara January 2008 planting, CED09 Cedara November 2008 planting, BF09Baynesfield December 2008 planting, RARS08 Rattray Arnold Research station December 2007 planting and ZAMB08 Mpongwe,

Zambia January 2008 planting

*** Significant at P B 0.001

Euphytica (2012) 185:349–362 359

123

rays of the biplot divided the plot into nine sections,

with the six environments all appearing in one sector.

According to Yan et al. (2007), when different

environments fall into different sectors, it implies that

they have different high yielding cultivars (susceptible

or resistant) for those sectors and it shows crossover

G 9 E, suggesting that the test environments could be

divided into mega-environments. However, in this

study, all the environments fell within one sector,

indicating one mega-environment and absence of

significant crossover interaction. The vertex families

for each quadrant represented the hybrids with the

highest mean disease severity score and for these

environments, the vertex hybrid was entry 37 (N3-2-3-

3 9 A16). The other hybrids were located within the

polygon and most of them near the plot origin.

According to Yan et al. (2000), genotypes within the

polygon, especially those located near the plot origin,

were less responsive than the vertex genotypes.

Environment vectors were drawn from the biplot

origin to connect the environments (Fig. 2). All the

locations had positive PC1 scores indicating high

severity scores and good GLS disease discriminative

ability. According to Yan (2002), the cosine of the

angle between the two vectors of two environments

approximated the correlation coefficient between

them. The angles between all the six environments in

this study were less that 90�, indicating high correla-

tions amongst them. This result corroborated the

observation from the Spearman’s rank order correla-

tion coefficients (Table 7), which showed relatively

strong positive associations among test environments,

indicating that the hybrids were ranked similarly in the

different environments. Environments with the longest

vectors from the biplot origin were the most discrim-

inating of the hybrids. Of these six environments,

BF09 supported higher disease expression and

discrimination (high positive PC1 scores, longest

vector) than the others. The environments could be

ranked as follows in terms of disease expression and

discrimination of the hybrids; BF09 [ CED108 [CED09 = ZAMB08 [ RARS08 [ CED208. Envi-

ronment CED208 had PC2 scores close to zero

confirming that it provided little or no information

(less discriminating) about the genotypes. It appears,

therefore, that at Cedara site, early planting was more

reliable than late planting for GLS resistance screen-

ing. However, because of the high correlation among

the environments and the absence of a significant

crossover interaction, evaluation of the inbreds or

hybrids for GLS resistance should be possible at one

reliable hot spot in southern Africa.

54

52

2526

327

5

29

730 93112

32

14

3316

34

18

36

21

37

23

38

239

6

40

10

42

15

4319

44

24

45

8

4717

413 22

48

49

51

CED108

CED09

BF09

RARS08

CED208

ZAMB08

-0.8

0.0

-0.6

0.4

-0.4

0.8

-0.2

1.2

0.0

0.2

0.4

0.6

0.8

0.2 1.00.6-0.2

PC1 - 81.80%

Genotype scoresEnvironment scoresConvex hullSectors of convex hull

Fig. 1 Polygon view of the GGE biplot based on the GLS

disease scores (1–9) for six environments. The environments

are CED108 = Cedara, 2007/2008, first planting; CED208 =

Cedara, 2007/2008, second planting; CED09 = Cedara,

2008/2009; BF09 = Baynesfield 2008/2009; RARS08 =

Rattray Arnold Research Station 2007/2008; ZAMB08 =

Mpongwe 2007/2008

54

52

51

49

48 47

45

444342

4039

38

37

3634

33

32

3130

29

27

262524

232221

19

18

17 16

1514

1312

10

9

8

76 5

43

2

ZAMB08

CED208CED09

CED108RARS08

BF09

-0.2 0.2 0.6 1.0-0.8

-0.6

-0.4

-0.2

0.4

0.0

1.2

0.2

8.00.0

0.8

0.4

0.6

PC1 - 81.80%

Genotype scoresEnvironment scoresVectors

Fig. 2 GGE biplot based on GLS disease scores (1–9) for six

environments showing the relationship among the environ-

ments. Description of the environments is provided in Fig. 1

360 Euphytica (2012) 185:349–362

123

Conclusion

The most resistant inbred lines CZL00009, CZL00001,

MP82, and MP59 displayed good GCA for the GLS

disease resistance and contributed negative SCA effects

in their respective crosses. Both additive and non-

additive effects conditioned GLS disease resistance and

grain yield in the inbred lines, but the additive effects

were more predominant than the non-additive. Predom-

inance of additive effects suggests that GLS disease

resistance could be enhanced by selection in some of the

populations. This study also revealed that the use of one

parent with resistance would provide adequate GLS

disease resistance in single cross hybrids. Therefore,

non-additive effects towards reduced disease severity

levels may be exploited in developing single cross

hybrids for deployment to environments that are highly

conducive to the GLS disease. The results also showed

the potential that the GLS disease still has in suppressing

grain yield, under high disease pressure, emphasizing

the need for resistant varieties to be grown in the region.

Although G 9 E was observed for the GLS disease, all

the six environments fell within the same quadrant of the

polygon view, implying that the environments were

similar in terms of the GLS disease discriminative

ability and were highly correlated suggesting absence of

the crossover G 9 E; therefore selection for GLS

resistance in the region can be done in one hot spot

area to reduce costs for regional breeding programmes.

The recommended sites from this study would be

Baynesfied and Cedara (normal growing season) as they

provided the highest GLS disease discriminative ability.

Acknowledgments The researchers would like to thank the

Rockefeller Foundation, New York for funding this work

through the African Centre for Crop Improvement (ACCI) in

South Africa. We are grateful to the International Maize and

Wheat Improvement Centre (CIMMYT), Zimbabwe for

providing germplasm, and the assistance of Dr Cosmos

Magorokosho in generating some of the crosses in Harare.

We also express our appreciation to the staff from the Crop

Protection unit, Cedara, South Africa; Mr Walter Chivasa

(SeedCo, Rattray Arnold Research Station, Zimbabwe); Mr

Herbert Masole (SeedCo, Mpongwe, Zambia) and the ACCI

support staff for the assistance in running the trials.

References

Bhatia A, Munkvold GP (2002) Relationships of environmental

and cultural factors with severity of grey leaf spot in

maize. Plant Dis 86:1127–1133

Bookmyer JM, Bonos SA, Meyer WA (2009) Inheritance

characteristics of brown patch resistance in tall fescue.

Crop Sci 49:2302–2308

Carson ML, Goodman MM, Williamson SM (2002) Variation

in aggressiveness among isolates of Cercospora from

maize as a potential cause of genotype-environment

interaction in gray leaf spot trials. Plant Dis 86:1089–1093

CIMMYT (1985) Managing trials and reporting data for CI-

MMYT International Maize Testing Program. CIMMYT,

Mexico

CIMMYT (2001) Maize inbred lines released by CIMMYT: a

compilation of 454 CIMMYT maize lines (CMLs),

CML1–CML454. August 2001. Second Draft. CIMMYT,

Mexico

Cisar G, Brown CM, Jedlinski H (1982) Diallel analyses for

tolerance in winter wheat to the barley yellow dwarf virus.

Crop Sci 22:328–333

Coates ST, White DG (1998) Inheritance of resistance to gray

leaf spot in crosses involving selected resistant inbred

lines of corn. Phytopathology 88:972–982

Cromley JMD, Hallauer AR, Martinson CA (2002) Inheritance

of gray leaf spot resistance in corn. J Iowa Acad Sci

109:25–29

Crous PW, Groenewald JZ, Groenewald M, Caldwell P, Braun

U, Harrington TC (2006) Species of Cercospora associ-

ated with grey leaf spot of maize. Stud Mycol 55:189–197

de Nazareno NRX, Lipps PE, Madden LV (1993) Effects of

levels of corn residue on epidemiology of grey leaf spot of

corn in Ohio. Plant Dis 77:67–70

Derera J (2005) Genetic effects and associations between grain

yield potential, stress tolerance and yield stability in

southern African maize (Zea mays L.) base germplasm.

PhD Thesis, University of KwaZulu-Natal

Derera J, Tongoona P, Pixley KV, Vivek B, Laing MD, van Rij

NC (2008) Gene action controlling gray leaf spot resis-

tance in southern African maize germplasm. Crop Sci

48:93–98

Dhliwayo T, Pixley K, Menkir A, Warburton M (2009)

Combining ability, genetic distances, and heterosis among

elite CIMMYT and IITA tropical maize inbred lines. Crop

Sci 49:1201–1210

Donahue PJ, Stromberg EL, Myers SL (1991) Inheritance of

reaction to gray leaf spot in a diallel cross of 14 maize

inbreds. Crop Sci 31:926–931

Dunkle LD, Levy M (2000) Genetic relatedness of African and

United States populations of Cercospora zeae-maydis.

Phytopathology 90:486–490

Egesi CN, Onyeka TJ, Asiedu R (2009) Environmental sta-

bility of resistance to anthracnose and virus diseases of

water yam (Dioscorea alata). Afr J Agric Res 4:113–118

Elwinger GF, Johnson MW, Hill RR Jr, Ayers JE (1990)

Inheritance of resistance to gray leaf spot of corn. Crop

Sci 30:350–358

Fairbanks DHK, Benn GA (2000) Identifying regional land-

scapes for conservation planning: a case study from

KwaZulu-Natal, South Africa. Landscape Urban Plan

50:237–257

FAO and CIMMYT (1997) White maize: a traditional food grain

in developing countries. Food and Agriculture website.

http://www.fao.org/docrep/W2698E/w2698e00.htm.

Accessed 10 Oct 2010

Euphytica (2012) 185:349–362 361

123

Gevers HO, Whythe IV (1987) Patterns of heterosis in South

African maize breeding material. In: Fourie AP, Du

Plessis JG (eds) Proceedings of the seventh South African

maize breeding symposium. Summer Grain Centre, Grain

Crops Research Institute, Potchefstroom, South Africa,

pp 21–26

Gevers HO, Lake JK, Hohls T (1994) Diallel cross analysis

of resistance to gray leaf spot in maize. Plant Dis

78:379–383

Gordon SG, Lipps PE, Pratt RC (2006) Heritability and com-

ponents of resistance to Cercospora zeae-maydis derived

from maize inbred VO613Y. Phytopathology 96:593–598

Hohls T, Shanahan PE, Clarke GPY, Gevers HO (1995)

Genetic analyses of grey leaf spot infection of maize in a

single location 12 9 12 diallel. S Afr J Plant Soil

12:133–139

Huff CA, Ayers JE, Hill RR Jr (1988) Inheritance of resistance

in corn (Zea mays) to gray leaf spot. Phytopathology

78:790–794

Kasuga S, Inoue N (2001) Diallel analysis of the resistance to

sheath blight (Rhizoctonia solani Kuhn) in sorghum.

Grassland Sci 47:45–49

Latterell FM, Rossi AE (1983) Gray leaf spot of maize: a

disease on the move. Plant Dis 67:842–847

Levy Y, Pataky JK (1992) Epidemiology of northern leaf blight

on sweet corn. Phytoparasitica 20:53–66

Lipps PE (1998) Gray leaf spot: a global threat to corn production.

APSnet Features. Online. doi:10.1094/APSnetFeature-1998-

0598. http://www.apsnet.org/publications/apsnetfeatures/

Pages/GrayLeafSpot.aspx. Accessed 18 Oct 2009

Meisel B, Korsman J, Kloppers FJ, Berger DK (2009) Cer-cospora zeina is the causal agent of grey leaf spot disease

of maize in southern Africa. Eur J Plant Pathol

124:577–583

Menkir A, Ayodele M (2005) Genetic analysis of resistance to

gray leaf spot of mid-altitude maize inbred lines. Crop Sci

45:163–170

Mickelson HR, Cordova H, Pixley KV, Bjarnason MS (2001)

Heterotic relationships among nine temperate and sub-

tropical maize populations. Crop Sci 41:1012–1020

Munkvold GP, Martinson CA, Shriver JM, Dixon PM (2001)

Probabilities for profitable fungicide use against gray leaf

spot in hybrid maize. Phytopathology 91:477–484

Payne RW, Murray DA, Harding SA, Baird DB, Soutar DM

(2010) GenStat for Windows (12th edition) introduction.

VSN International, Hemel Hempstead

Rupe JC, Siegel MR, Hartman JR (1982) Influence of envi-

ronment and plant maturity on gray leaf spot of corn

caused by Cercospora zeae-maydis. Phytopathology

72:1587–1591

SAS Institute Inc (2002) The SAS system for Windows, release

9.3. SAS Institute Inc., Cary

Thompson DL, Bergquist RR, Payne GA, Bowman DT,

Goodman MM (1987) Inheritance of resistance to gray

leaf spot in maize. Crop Sci 27:243–246

Ulrich JF, Hawk JA, Carroll RB (1990) Diallel analysis of

maize inbreds for resistance to gray leaf spot. Crop Sci

30:1198–1200

Vivek BS, Odongo O, Njuguna J, Imanywoha J, Bigirwa G,

Diallo A, Pixley K (2010) Diallel analysis of grain yield

and resistance to seven diseases of 12 African maize (Zeamays L.) inbred lines. Euphytica 172:329–340

Ward JMJ, Hohls T, Laing MD, Rijkenberg FHJ (1996) Fun-

gicide responses of maize hybrids and grey leaf spot. Eur J

Plant Pathol 102:765–771

Ward JMJ, Laing MD, Nowell DC (1997a) Chemical control of

maize grey leaf spot. Crop Prot 16:265–271

Ward JMJ, Laing MD, Rijkenberg FHJ (1997b) Frequency and

timing of fungicide applications for the control of grey

leaf spot in maize. Plant Dis 81:41–48

Ward JMJ, Stromberg EL, Nowell DC, Nutter FW Jr (1999)

Gray leaf spot: a disease of global importance in maize

production. Plant Dis 83:884–895

Yan W (2002) Singular-value partitioning in biplot analysis of

multienvironmental trial data. Agron J 94:990–996

Yan W, Hunt LA, Sheng Q, Szlavnics Z (2000) Cultivar

evaluation and mega-environment investigation based on

the GGE Biplot. Crop Sci 40:597–605

Yan W, Kang BM, Woods S, Cornelius PL (2007) GGE biplot

vs. AMMI analysis of genotype-by-genotype environment

data. Crop Sci 47:643–655

Zhang Y, Kang MS (1997) DIALLEL-SAS: a SAS program for

Griffing’s diallel analyses. Agron J 89:176–182

Zhang Y, Kang MS, Lamkey KR (2005) Diallel-SAS05: a

comprehensive program for Griffing’s and Gardner-

Eberhart analyses. Agron J 97:1097–1106

362 Euphytica (2012) 185:349–362

123