Advertisement

Euphytica

, Volume 207, Issue 2, pp 353–365 | Cite as

Genotype by environment interactions and agronomic performance of doubled haploids testcross maize (Zea mays L.) hybrids

  • Julius Pyton Sserumaga
  • Sylvester O. Oikeh
  • Stephen Mugo
  • Godfrey Asea
  • Michael Otim
  • Yoseph Beyene
  • Grace Abalo
  • Joseph Kikafunda
Open Access
Article

Abstract

In vivo production of maternal haploid plants and advancement in chromosome doubling technology has led to rapid production of doubled haploid homozygous lines. These in turn have boosted rapid advancement in most breeding programs. This has resulted in production of a large number of maize hybrids which need testing across production environments to select the most suitable hybrids for release and cultivation. The objective of this study was to assess the genotype × environment interactions (GE) for grain yield and other agronomic traits and evaluate the performance of 44 recently developed doubled haploids (DH) testcross hybrids along with six checks across five locations in Uganda. Significant mean squares for environment (E), genotype (G) and GE were observed for all studied traits. Environment explained 46.5 % of the total variance, while G and GE contributed 13.2 and 7.2 %, respectively. Genetic correlations among locations were high (0.999), suggesting little GE among environments. The 10 best testcross hybrids had a 49.2 % average grain yield advantage over the six checks at all locations. DH hybrids CKHDHH0887, CKDHH0878, CKDHH0859, WM1210, CKDHH0858, and WM1214 were the most stable, across locations. The DH testcross hybrids produced higher grain yield and possessed acceptable agronomic traits compared to the commercial hybrids developed earlier. Use of the best DH testcross hybrids, well targeted to the production environments, could boost maize production among farmers.

Keywords

Doubled haploids East Africa Genotype × environment Grain yield Maize Stability 

Introduction

Doubled haploids (DH) technology has paved the way to rapidly generate large number of inbred lines. The technology involves in vivo haploid induction by specific inducers that lead to production of haploid seeds from the maternal plants (Beyene et al. 2011). These haploid maternal plants then get their chromosomes doubled through use of colchicine. DH lines are highly efficient tools in genetic research and practical maize breeding (Thomas et al. 2003; Bordes et al. 2007; Beyene et al. 2011). Major advantages of DH lines compared to pedigree lines include (i) maximum genetic variance among lines for per se and testcross performance from the first generation; (ii) reduced length of breeding cycle; (iii) perfect fulfillment of distinctness, uniformity, and stability (DUS) for satisfying varietal status; (iv) reduced costs in maintenance breeding; (v) simplified logistics; and (vi) increased efficiency in marker-assisted selection, gene introgression, and gene stacking.

Since 2008, there has been an effort to enhance rapid development of elite lines for tolerance to drought and other stresses by the International Maize and Wheat Improvement Center (CIMMYT), the National Agricultural Research Systems (NARS) partners, and Monsanto Company through the Water Efficient Maize for Africa (WEMA) project. The partnership has developed DH lines from several drought tolerant maize source populations. Several DH hybrids have also been developed and tested for their performance in different drought-stress and non-drought stress environments in Kenya (Beyene et al. 2011; 2013). Results showed that the use of DH testcrosses performed much better than the commercial hybrids in both stressed and non-stressed environments (Beyene et al. 2011). These DH hybrids needed to be evaluated in different environments to assess their performance and adaptability, and to identify the major basis of the genotype adaptation.

The performance of a genotype can vary from one environment to another and genotypes that are superior in one environment may not be superior in other environments due to genotype-by-environment interactions (GE) (Makumbi et al. 2015). The presence of GE results in the failure of genotypes to achieve the same relative performance in different environments (Baker 1988; Beyene et al. 2011). This reduces the correlation between phenotypes and genotypes, complicating breeding and selection of superior cultivars (Kang 1993; Makumbi et al. 2015).

The WEMA project team formed WEMA-Wide trials (WWT) through which common maize trials are grown in different environments across five countries (Kenya, Mozambique, South Africa, Tanzania and Uganda) in eastern and southern Africa to identify high-yielding and adapted varieties for release and cultivation in the respective countries. WWT are grown tested in multi-environment that sample drought and non-stress locations.

Significant gains in grain yield (GY) performance have been reported in these CIMMYT Maize Regional Trials (Beyene et al. 2011). Earlier studies have suggested that by considering GE, superior genotypes were selected for commercial release to farmers in Africa (Pixley and Bjarnason 2002; Beyene et al. 2011). Genotype × environment interactions have been investigated through the use of statistical tools such as the additive main effects and multiplicative interaction (AMMI) analysis for grain yield and grain micronutrients’ concentrations and stability (Zobel et al. 1988; Gauch 2006; Kassa et al. 2013); and genotype main effect plus genotype × environment interaction (GGE) for the analyses of grain yield and stability in tropical maize (Yan et al. 2000; Makumbi et al. 2015). These analytical methods provide an insight into the extent of GE present in a given study. Genetic correlations can be used to quantify the importance of GE (Falconer 1952); and have been used in GE studies (Cooper and DeLacy 1994). However, although DH maize hybrids developed by CIMMYT and its partners have been evaluated in eastern and southern Africa, the level of their GE has not been assessed. There is limited information available in the literature on the level of GE on grain yield performance and stability of maize testcrosses developed from DH lines. Therefore, the objectives of this study were to (i) assess GE for GY and other agronomic traits; and (ii) evaluate the performance and stability of 44 DH maize testcrosses across five locations in Uganda.

Materials and methods

Field evaluation and experimental design

The genotypes used for the study comprised 44 DH hybrids developed from the WEMA project and 6 checks that included 1 internal CIMMYT hybrid, 3 commercial hybrids and 2 Ugandan local hybrids. These were grown in trials sown at five locations: Namulonge, Serere, Bulindi, Ngetta and Kasese under different environmental conditions in Uganda in 2012 (Tables 1, 2). The experiment design was a 5 × 10 Alpha lattice with two replications at each location. Each entry was planted in a two-row plot of 5 m long and 0.75 m apart with the hills spaced 0.25 m apart. Two seeds were initially planted per hill but were subsequently thinned to one plant per hill at 4 weeks after emergence to give a plant population of 53,333 plants per hectare. In all the experiments, standard cultural practices including weeding control throughout the growing season were followed. Fertilizer application at each location consisted of 125 kg N ha−1, 60 kg P2O5 ha−1.
Table 1

List of maize DH testcross hybrid genotypes used in the study and their sources

Genotype

Pedigree

Source

G1

WM1206

Monsanto

G2

WM1207

Monsanto

G3

WM1208

Monsanto

G4

WM1209

Monsanto

G5

WM1210

Monsanto

G6

WM1211

Monsanto

G7

WM1212

Monsanto

G8

WM1213

Monsanto

G9

WM1214

Monsanto

G10

WM1215

Monsanto

G11

WM1216

Monsanto

G12

WM1219

Monsanto

G13

WM1220

Monsanto

G14

WM1221

Monsanto

G15

WM1222

Monsanto

G16

CKDHH0858

CIMMYT

G17

CKDHH0888

CIMMYT

G18

CKDHH0873

CIMMYT

G19

CKDHH0859

CIMMYT

G20

CKDHH0889

CIMMYT

G21

CKDHH0874

CIMMYT

G22

CKDHH0860

CIMMYT

G23

CKDHH0890

CIMMYT

G24

CKDHH0875

CIMMYT

G25

CKDHH0861

CIMMYT

G26

CKDHH0891

CIMMYT

G27

CKDHH0876

CIMMYT

G28

CKDHH0862

CIMMYT

G29

CKDHH0892

CIMMYT

G30

CKDHH0877

CIMMYT

G31

CKDHH0863

CIMMYT

G32

CKDHH0893

CIMMYT

G33

CKDHH0878

CIMMYT

G34

CKDHH0864

CIMMYT

G35

CKDHH0894

CIMMYT

G36

CKDHH0879

CIMMYT

G37

CKDHH0865

CIMMYT

G38

CKDHH0895

CIMMYT

G39

CKDHH0880

CIMMYT

G40

CKDHH0866

CIMMYT

G41

CKDHH0896

CIMMYT

G42

CKDHH0881

CIMMYT

G43

CKDHH0872

CIMMYT

G44

CZH0616

CIMMYT

G45

CKDHH0887

CIMMYT

G46

H513

Seed company

G47

PH 3253

Seed company

G48

DK 8053

Seed company

G49

Local Check 1

NARO

G50

Local Check 2

NARO

Table 2

Agro-climatic description of the locations where maize DH testcross hybrids were evaluated during the study period in 2012

Site

Latitude

Longitude

Elevation (masl)

Mean rainfall (mm)

Temperature (°C)

Soil type

Min

Max

Namulonge

0.5297

32.6025

1150

1270

16

28

Sandy clay loam

Serere

1.5176

33.4579

1080

1419

19

31

Sandy clay loams and black clays

Ngetta

53.6947

22.9297

1300

1483

19

29

Sandy loam

Bulindi

−0.777

29.1402

1020

1338

19

29

Sandy loam

Kasese

0.1833

30.0833

960

1200

18

31

Peaty sands and clay

masl metres above sea level

Data collection

The data recorded from each plot included: days to anthesis (AD), i.e. days from planting to when 50 % of the plants shed pollen, and days to silking (SD), i.e. days from planting to when 50 % of the plants had extruded silks. Anthesis–silking interval (ASI) was determined as the difference between SD and AD. Plant height (PH) measured in centimeters as the distance from the base of the plant to the height of the first tassel branch, number of ears per plant (EPP), determined by dividing the total number of ears per plot by the number of plants harvested per plot, husk cover (HC), obtained by dividing the number of ears with poor husk cover by the number of plants harvested per plot; and expressed as percentage; lower value indicates best husk cover), ear aspect (EA), rated on a scale of 1–5, where 1 = nice uniform ears with the preferred texture; and 5 = cobs with the undesirable texture), plant aspect (PA) (1–5) 1 = short plant with uniform and short ear placement; 5 = tall plants with high ear placement ear position or height (EP): the ear height is determined by measuring a representative plant from the ground to the insertion of the top ear and grain moisture.

All ears harvested from each plot at all locations were weighed and randomly selected representative samples of ears were shelled and weighed. Grain moisture was determined using a Dickey Jones moisture meter. Grain yield in tons per hectare (t ha−1) was determined based grain moisture content of 12.5 %.

Statistical analysis

Analysis of variance (ANOVA) for all traits was done separately for each location, and combined across locations using PROC MIXED Model procedure from SAS (SAS Institute 2008). Genotypes were considered as fixed effects, and replications and blocks within replications as random effects. For the combined analysis, variances were partitioned into the relevant sources of variation to test for differences among genotypes and the presence of GE. Broad-sense heritability (H) was calculated as the proportion of genetic variance over the total phenotypic variance. Heritability estimates refer to entry means across environments and replicates (Hallauer and Miranda 1981). For comparing entries evaluated in different locations, the entry means were expressed as a percentage of the average performance of the best check hybrid in the respective locations.

Estimates of genotypic (\(\sigma_{G}^{2}\)), location (\(\sigma_{L}^{2}\)), genotype × location (\(\sigma_{G \times L}^{2}\)), and error variance (\(\sigma_{E}^{2}\)) were calculated using the PROC MIXED (option = REML) of SAS (SAS Institute 2008). Across environments, ANOVA for each trait was conducted using PROC GLM of SAS (SAS Institute 2008). In the across-environment ANOVA, genotype effects were tested for significance using the corresponding interaction with the environment as the error term, while the GE was tested using the pooled error.

Heritability and genetic correlations

Broad-sense heritability (H) for individual trials was estimated according to Hallauer et al. (2010):
$$H = \frac{{\sigma_{G}^{2} }}{{\left[ {\sigma_{G}^{2} + \left\{ {\frac{{\sigma_{E}^{2} }}{r}} \right\}} \right]}}$$
where \(\sigma_{G}^{2}\) is the genotypic variance, \(\sigma_{E}^{2}\) is the error variance, and r the number of replications.
H for traits across environments was estimated using the variance components according to Hallauer et al. (2010) as:
$$H = \frac{{\sigma_{G}^{2} }}{{\left[ {\sigma_{G}^{2} + \frac{{\sigma_{GxL}^{2} }}{E} + \frac{{\sigma_{E}^{2} }}{ER}} \right]}}$$
where \(\sigma_{G}^{2}\), \(\sigma_{G \times L}^{2}\) and \(\sigma_{E}^{2}\) are genotypic, genotype × location, and residual variance components, respectively: E is the number of environments, and R is the number of replications. Genotypic correlations (r) between locations were estimated according to Cooper et al. (1996) as:
$$r_{g = } \frac{{r_{{p^{{\left( {12} \right)}} }} }}{{\left( {H_{1} \times H_{2} } \right)^{1/2} }}$$
where \(r_{{p^{{\left( {12} \right)}} }}\) is the phenotypic correlation between the traits measured in locations 1 and 2, H1 and H2 are the broad-sense heritabilities for the traits measured in locations 1 and 2, respectively.

Cluster analysis using Ward’s minimum variance method (Ward 1963; Makumbi et al. 2015) was performed to group environments based on genetic correlations among the environments. The SAS procedure PROC CLUSTER was used for cluster analysis. The PROC TREE procedure of SAS was used to generate the dendrograms.

Genotype main effect and genotype × environment (GGE) biplot analysis

Adjusted GY from ANOVA was subjected to GGE biplot analysis to decompose the GE of each experiment (Yan et al. 2000; Yan 2001) to compare genotype stability in performance across the various environments.
$${\text{Yijl }} = {\text{m}} + {\text{ Gi }} + {\text{ Ej }} + \Big(\sum\uplambda {\rm k} \upalpha {\rm ik} \upgamma {\rm jk}\Big) + {\text{ dij }} + {\text{ eijl}},$$
where λk = kth eigenvalue, αik = principal component score for the ith genotype for the kth principal component axis, γjk = principal component score for the jth environment for the kth principal component axis, dij = residual G × E not explained by model.

The GGE biplot shows the first two principal components (PC1 and PC2) derived from subjecting the environment-centered yield data (the yield variation due to GGE) to singular value decomposition (Yan et al. 2000). GGE biplots were constructed using R package named GGEBiplotGUI (Frutos et al. 2014).

Results

Analysis of variance

Combined analysis of variance across the five locations revealed that G, E, and GE were significant for all the traits except GE for SD, ASI, and HC (Table 3). There were, therefore differences in the performance of the test materials at the different locations. GE was highly significant (P < 0.001) for PA and EP and significant (P < 0.05) for GY, AD, grain moisture (MOI), ear per plant (EPP), and ear aspect (EA). Overall on the grain yield, the effect of environment explained 46.5 % of the total variance while genotype contributed 13.2 %, and GE contributed the least (7.2 %) to the total variation hence the environment influences a lot on the performance of different germplasm (Table 4).
Table 3

Mean squares and degrees of freedom from ANOVA for grain yield and agronomic traits for 44 maize doubled haploids (DH) testcross hybrids, and six checks across five locations in Uganda in 2012

 

Mean square

Grain yield (GY)

Days to anthesis (AD)

Days to silking (SD)

Anthesis silking interval (ASI)

Grain moisture (MOI)

Ear per plant (EPP)

Husk cover (HC)

Ear aspect (EA)

Plant aspect (PA)

Ear position (EP)

Source

df

(t ha−1)

(days)

(days)

(days)

(%)

(#)

(%)

(1–5)

(1–5)

(cm)

Environment (E)

4

1128.6***

12552***

15518***

426.3***

1375.8***

1.891***

99,528***

30.9***

164.87***

106,8136***

Genotype (G)

49

520.7***

1244***

1622***

219.6***

127.1*

1.891***

27,256***

14.32***

14.34***

70,263***

GE

196

590.5*

728*

896

428.7

459.9*

4.548*

43,597

24.8*

51.08***

185,958***

Error

245

557.7

685

962

463.6

459.7

4.477

55,511

24.58

35.92

148,432

Significance level: ns not significant, P > 0.05; * P < 0.05; ** <0.001; *** P < 0.0001

Table 4

Variance decomposition, heritability of grain yield and agronomic traits 44 of maize doubled haploids (DH) testcross hybrids, and 6 checks across 5 locations in Uganda in 2012

Statistic

Grain yield (GY)

Days to anthesis (AD)

Days to silking (SD)

Anthesis silking interval (ASI)

Grain moisture (MOI)

Ear per plant (EPP)

Husk cover (HC)

Ear aspect (EA)

Plant aspect (PA)

Ear position (EP)

Genotypic variance

0.74

2.11

2.91

0.23

0.02

0.00

3.00

0.02

0.00

0.00

Location variance

2.60

31.07

38.48

1.03

3.39

0.00

0.47

0.07

0.41

0.00

Gen × env variance

0.40

0.42

0.25

0.15

0.23

0.00

9.42

0.01

0.06

0.00

Residual variance

1.85

2.24

3.02

1.89

1.60

0.02

5.87

0.09

0.12

0.00

Grand mean

8.01

60.54

61.37

0.84

15.30

0.99

5.17

2.80

3.20

0.52

LSD0.05

2.67

2.93

3.41

2.70

2.48

0.26

4.75

0.58

0.69

0.06

Heritability

0.74

0.87

0.89

0.51

0.09

0.40

0.55

0.59

0.16

0.72

Genotypes performance at individual location

Grain yield of the test hybrids varied with locations. The lowest yield of 4.2–7.1 t ha−1 was obtained at Namulonge, and the highest yield of 7.2–12.1 t ha−1 was recorded at Serere (Table 5, Supplementary 1). The highest yielding testcross hybrids at Kasese, Namulonge, Serere, Bulindi, and Ngetta were 51, 48, 30, 17 and 5 % above the best commercial hybrid, DK 8053, respectively. At each location, the best testcross hybrid out-yielded the CIMMYT internal check, CZH0616, by 9–34 %.There were no significant yield advantage between the commercial check and the CIMMYT internal check at all locations. The environments could be ranked in terms of their grain yielding potential: Serere > Kasese > Ngetta > Bulindi > Namulonge; and the ranking of the environments in terms of heritability followed a similar pattern as grain yield.
Table 5

Mean performance of 10 highest yielding from among 44 maize Doubled haploids (DH) testcross hybrids, and six checks across five locations in Uganda in 2012 combined across five locations in Uganda

 

Namulonge

 

Serere

 

Bulindi

 

Ngetta

 

Kasese

 

Entry

Grain yield (GY)

 

Entry

Grain yield (GY)

 

Entry

Grain yield (GY)

 

Entry

Grain yield (GY)

 

Entry

Grain yield (GY)

  

(t ha−1)

  

(t ha−1)

  

(t ha−1)

  

(t ha−1)

  

(t ha−1)

26

CKDHH0891

7.1

9

WM1214

12.1

11

WM1216

9.0

11

WM1216

8.5

17

CKDHH0888

11.9

41

CKDHH0896

7.0

4

WM1209

12.0

30

CKDHH0877

8.9

32

CKDHH0893

8.4

25

CKDHH0861

10.6

5

WM1210

6.6

10

WM1215

11.8

16

CKDHH0858

8.6

42

CKDHH0881

8.4

16

CKDHH0858

10.4

19

CKDHH0859

6.5

22

CKDHH0860

11.7

4

WM1209

8.5

10

WM1215

8.4

36

CKDHH0879

10.3

36

CKDHH0879

6.5

16

CKDHH0858

11.6

45

CKDHH0887

8.5

4

WM1209

8.3

22

CKDHH0860

10.3

16

CKDHH0858

6.2

17

CKDHH0888

11.5

8

WM1213

8.4

5

WM1210

8.3

18

CKDHH0873

10.3

11

WM1216

6.2

11

WM1216

11.2

22

CKDHH0860

8.4

21

CKDHH0874

8.3

9

WM1214

10.2

25

CKDHH0861

6.0

42

CKDHH0881

11.1

10

WM1215

8.3

34

CKDHH0864

8.3

5

WM1210

10.2

35

CKDHH0894

5.9

5

WM1210

11.0

26

CKDHH0891

8.3

29

CKDHH0892

8.3

7

WM1212

10.1

6

WM1211

5.9

6

WM1211

10.9

23

CKDHH0890

8.3

16

CKDHH0858

8.3

39

CKDHH0880

10.1

44

CZH0616

5.5

47

PH 3253

10.3

44

CZH0616

7.9

46

H513

8.1

44

CZH0616

9.2

50

Local check 2

5.2

46

H513

10.1

46

H513

7.7

50

Local check 2

8.1

49

Local check 1

8.5

49

Local Check 1

5.0

48

DK 8053

9.3

48

DK 8053

7.7

48

DK 8053

8.1

46

H513

8.3

48

DK 8053

4.8

44

CZH0616

9.0

50

Local check 2

7.4

49

Local check 1

7.9

48

DK 8053

7.9

47

PH 3253

4.6

49

Local check 1

8.3

47

PH 3253

7.2

44

CZH0616

7.8

50

Local check 2

7.9

46

H513

4.2

50

Local check 2

7.2

49

Local check 1

7.1

47

PH 3253

7.7

47

PH 3253

5.9

 

Grand mean

5.4

  

9.8

  

7.8

  

8.1

  

9

 

LSD0.05

1.8

  

2.3

  

ns

  

ns

  

2

 

CV(%)

16.9

  

11.8

  

26.1

  

16.9

  

11.5

 

Heritability

0.7

  

0.7

  

0.3

  

0.2

  

0.8

Average genotypes performance across locations

The combined analysis for average performance across the five locations for the top 10 testcross hybrids showed 23 % yield advantage over the average of the six checks included in the trial (Table 6, Supplementary 2). The best performing hybrids (CKDHH0858 and WM1216) across five locations out-yielded the commercial check (DK8053) by 25 %.
Table 6

Mean Performance of the 10 highest yielding among 44 maize Doubled haploids (DH) testcross hybrids, and 6 checks across five locations in Uganda in 2012 combined across five locations in Uganda

Genotype no.

Entry

Grain yield (GY)

Days to anthesis (AD)

Days to silking (SD)

Anthesis silking interval (ASI)

Grain moisture (MOI)

Ear per plant (EPP)

Husk cover (HC)

Ear aspect (EA)

Plant aspect (PA)

(t ha−1)

(days)

(days)

(days)

(%)

(#)

(%)

(1–5)

(1–5)

16

CKDHH0858

9.4

60.5

61.8

1.1

15.4

1.0

4.4

2.7

3.2

11

WM1216

9.4

61.0

62.1

1.0

15.3

1.0

5.3

2.8

3.2

17

CKDHH0888

9.2

59.9

60.1

0.6

15.3

1.0

4.2

2.8

3.2

4

WM1209

9.1

60.7

60.6

0.4

15.3

1.0

5.9

2.7

3.2

5

WM1210

9.1

59.2

60.2

1.1

15.3

1.0

5.2

2.6

3.2

26

CKDHH0891

9.0

59.4

59.1

0.3

15.3

1.0

5.3

2.6

3.2

22

CKDHH0860

8.9

59.7

60.5

0.8

15.3

1.0

5.1

2.7

3.2

9

WM1214

8.7

60.8

62.7

1.5

15.3

1.0

5.3

2.8

3.2

42

CKDHH0881

8.7

60.6

61.6

1.0

15.3

1.0

4.0

2.8

3.2

30

CKDHH0877

8.6

62.1

63.3

1.0

15.4

1.0

4.3

2.7

3.2

44

CZH0616

7.9

58.5

59.2

0.8

15.3

1.0

4.9

2.8

3.2

46

H513

7.7

59.1

60.5

1.2

15.3

1.0

7.0

2.8

3.3

48

DK 8053

7.5

59.4

60.6

1.2

15.3

1.0

8.7

2.7

3.2

49

Local Check 1

7.0

62.3

64.1

1.3

15.3

1.1

5.0

2.8

3.2

50

Local Check 2

6.9

60.7

62.3

1.2

15.3

1.0

4.8

2.7

3.2

47

PH 3253

6.7

59.0

59.8

0.9

15.3

1.0

6.0

2.8

3.2

 

Min

6.7

59.0

59.8

0.9

15.3

1.0

6.0

2.8

3.2

 

Max

9.4

60.5

61.8

1.1

15.4

1.0

4.4

2.7

3.2

 

Grand mean

8.0

60.5

61.4

0.8

15.3

1.0

5.2

2.8

3.2

 

LSD0.05

2.7

2.9

3.4

2.7

2.5

0.3

4.7

0.6

0.7

 

CV (%)

17.0

2.5

2.8

164.5

8.3

13.6

46.8

10.5

11.0

 

Heritability

0.7

0.9

0.9

0.5

0.1

0.4

0.5

0.6

0.2

All the genotypes had comparable maturity with SD ranging from 59.1 to 64.1 days except the testcross CKDHH0891 that significantly flowered 4–5 days earlier than the testcross CKDHH0877 and the Local Check 1 (Table 6). Therefore, the testcross CKDHH0891 could be categorized as an early maturing genotype that also had the least anthesis-silking interval (ASI) of 0.3 days (Table 6). This is an indication that this testcross hybrid and testcross hybrid, WM1209 with a similar low ASI of 0.4 days might be the most drought-tolerant hybrids among the genotypes evaluated. All the test materials had similar rating of husk cover (4–5.9 %) (Table 6). But the testcross CKDHH0881 had significantly better husk cover rating (4 %) than the commercial check DK8053 (8.7 %).

Medium to high heritability estimates were found in different traits except for grain moisture and plant aspect with heritability of 0.1–0.4. The highest heritability of 0.9 was recorded for AD and SD, and the lowest was for MO (h2 = 0.1).

Genetic and phenotypic correlation among different test environments

Based on grain yield, genetic correlations among locations ranged from −0.2191 (between Ngetta and Kasese) to 0.999 between Bulindi and Serere, and between Ngetta and Bulindi locations (Table 7). The phenotypic correlations among locations for grain yield varied from −0.0909 between Ngetta and Kasese to 0.6106 between Bulindi and Serere locations (Table 8). Since genotypic correlations between Bulindi and Serere; and between Ngetta and Bulindi were high (r = 0.999), it implies a similar ranking of the genotypes in these pairs of locations.
Table 7

Genetic correlations for grain yield of 44 Doubled haploids (DH) testcross hybrids, 1 internal CIMMYT hybrid checks, 3 commercial hybrids and 2 local hybrid checks across five locations in Uganda

 

Namulonge

Serere

Bulindi

Ngetta

Kasese

Namulonge

1

    

Serere

0.4588

1

   

Bulindi

0.9123

0.9999

1

  

Ngetta

0.4314

0.5682

0.9999

1

 

Kasese

0.708

0.6276

0.8396

−0.2191

1

Table 8

Phenotypic correlations for grain yield of 44 doubled haploids (DH) testcross hybrids, 1 internal CIMMYT hybrid checks, 3 commercial hybrids and 2 local hybrid checks across five locations in Uganda

 

Namulonge

Serere

Bulindi

Ngetta

Kasese

Namulonge

1

    

Serere

0.3227

1

   

Bulindi

0.4013

0.6106

1

  

Ngetta

0.167

0.2331

0.2945

1

 

Kasese

0.5034

0.4728

0.3956

−0.0909

1

The genetic correlation between locations for grain yield was used for cluster analysis to classify the environments. Clustering based on genetic correlation for grain yield revealed two clusters at 0.75 (Fig. 1). Cluster 1 consisted of the two locations (Namulonge and Serere) that were separated into individual environment and Cluster II consisted of one individual environment and two sub-clusters (Fig. 1).
Fig. 1

Dendrogram of five environments based on grain yield, 2012

Genotype main effect plus genotype × environment interaction biplot analysis of performance and stability

The GGE biplot analysis was used to identify the best entries at each location and assess the stability of the entries. The bi-plot analysis gave a good visual assessment of GE based on grain yield which explained 73 % (PC1 = 54.4 and PC2 = 18.6 %) of the total variation across the test environments (Fig. 2). The environmental vector bi-plot identified Kasese, Serere and Bulindi as highly discriminating for the genotypes tested, as evidenced by the large environment vectors (Fig. 2). Along environment vector represents a good discriminating ability for a given environment. Discriminant test environment accurately resolve genotype differences, thereby providing the necessary information for selection by a breeder (Tukamuhabwa et al. 2012). Namulonge was the least discriminating of the five environments, as evidenced by the short environment vector. It’s advisable to evaluate genotypes in environments which are most representative and high discriminating capabilities verses the environments with low discriminating capability and lack of representativeness which might give misleading results.
Fig. 2

The Environment vector bi-plot showing environmental differences in discriminating the 50 genotypes for grain yield at the five test environments during 2012 season in Uganda

Based on the five locations used in this study, the results revealed five sectors with two mega environments with different “winning” genotypes identified using a scatter plot with polygon bisectors (Fig. 3). Mega environments are test environments with different winning genotypes located at the vertex of the polygon. Locations within mega environment 1 were Kasese, Namulonge, and Serere. Mega environment II comprised of Ngetta and Bulindi.
Fig. 3

An environment focused bi-plot showing “winning” genotypes for the two different mega environments for grain yield at the five environments during 2012 seasons in Uganda

The vertex genotypes were G17 (CKDHH0888), G18 (CKDHH0873), G43 (CKDHH0872), G13 (WM1220), and G11 (WM1216) (Fig. 3). The vertex genotype in each sector represents the highest yielding genotype in the location that fell within that particular sector (Yan et al. 2000; Makumbi et al. 2015). Genotype G11 (WM1216) was the vertex entry in the sector where three (Serere, Bulindi, and Ngetta) of the five locations fell, indicating that this genotype was the highest yielding entry in these locations. Genotypes G18 (CKDHH0873), G43 (CKDHH0872) and G13 (WM1220), did not have any location falling in the sectors where they were located, suggesting that these entries were low yielding in some or all of the locations.

The mean versus stability view biplot (Fig. 4) was used to assess stability of the 50 genotypes across the five locations. This biplot accounted for 73 % of the variation in grain yield. In this biplot, the axis of the average environment coordinate (AEC) abscissa, or average environment axis, is the single-arrowed line that passes through the biplot origin and the average environment, which is at the center of the small circle. The axis of the AEC ordinate is the double-arrowed line that passes through the biplot origin and is perpendicular to the AEC abscissa (Yan et al. 2007; Makumbi et al. 2015). The cultivars were ranked along the average environment axis, with the arrow pointing to a greater value based on mean performance across all locations. The seven top ranking entries according to their projections onto the average environment axis were: G11 (WM1216), G16 (CKDHH0858), G17 (CKDHH0888), G22 (CKDHH0860), G4 (WM1209), G9 (WM1214) and G5 (WM1210) (Fig. 4). The stability of the cultivars was measured by their projections onto the AEC ordinates. Six entries G45 (CKHDHH0887), G33 (CKDHH0878), G19 (CKDHH0859, G5 (WM1210), G16 (CKDHH0858), and G9 (WM1214) were the most stable because their short projection onto the AEC ordinate. Among the most stable genotypes, testcross G16 (CKDHH0858) was the highest yielding across all locations. The second most high yielding G11 (WM1216), was not among the most stable, suggesting that this variety may have specific adaptation to some of the environments. The commercial check, G50 (Local Check 2) was among the lowest yielding genotype in this study but very stable in the test environments.
Fig. 4

The mean versus stability view of the genotype main effect plus genotype × environment interaction biplot based on yield data of 50 genotypes grown in five test environments in Uganda

Discussion

In this study we tested the agronomic performance and GE of recently developed DH hybrids in different environments of Uganda, East Africa. The study revealed that genotype, environment and genotype × environment interaction were significant for all the traits except GE for days to silking, Anthesis silking interval (ASI) and Husk cover (HC) suggesting differential responses of the genotypes across environments. This could be attributed to variations in terms of climatic and edaphic factors in the test environments. Similar observations were reported by Butron et al. (2002) in which they indicated that G × E effects for grain yield in maize were mainly due to environmental yield-limiting factors such as the mean minimum temperature and relative humidity. Also, Makumbi et al. (2015) reported similar significant genotypic differences for grain yield, emerged Striga plants and other agronomic traits except ASI across Striga-infested locations. GE effects for grain yield were found to be lower than the genotype effect as earlier reported by Van Eeuwijk et al. (1995) who found that variation due to the G × E interaction was smaller than the genotypic variation for silage dry matter content of 18 Dutch maize varieties. Also, Beyene et al. (2011) reported that variation due to the G × E interaction was smaller than the genotypic variation while testing for the performance of double haploid maize lines from tropical adapted backcross population. But our results are in contrast to earlier studies were G × E effects were higher than the genotypic effect in a study of early-maturity maize variety trials in France (Epinat-Le et al. 2001).

The DH hybrids used in this study exhibited a broad range of variation in grain yield and other agronomic traits under contrasting environments. Similar observations were reported by Munyiri et al. (2010) who characterized Kenyan maize landraces for drought tolerance; and Odiyo et al. (2014) who examined the performance and adaptability of DH maize testcross hybrids under drought stress and non-stress conditions in East Africa.

According to the results in this study, DH hybrids outperformed the commercial hybrids for grain yield and other agronomic traits assessed. Similar to the present study, Beyene et al. (2011) and Odiyo et al. (2014) reported superiority in performance by DH hybrids over the commercial checks in their studies. The best DH hybrid G16 (CKDHH0858) in our study produced 22 % over the best commercial check G46 (H513) across the five locations. This implied that DH lines were superior in performance over the commercially available hybrids that farmers use. Therefore, the performance of the DH testcross hybrids indicated that the DH lines used in creating them offered potential new sources for rapidly producing high yielding and drought tolerant maize hybrids.

Broad-sense heritability is defined as an estimate of the upper boundary of narrow-sense heritability (Robinson 1963). The moderate broad-sense heritability (0.51–0.59) for Anthesis-Silking interval (ASI), husk cover (HC) and ear aspect (EA) in this study suggested that the actual heritability estimates might be lower (Falconer and Mackay 1996), which may lead to low genetic gain from selection for these traits in the five test environments. Conversely, the broad-sense heritability estimate for grain yield was 0.74, suggesting that actual heritability estimates might be high (Falconer and Mackay 1996), which might lead to high genetic gain when selecting for this trait. For a trait measured from the same genotype in different environments, indirect selection can be applied given information on the heritability and the genetic correlation for the trait in the two environments (Makumbi et al. 2015).

In this study, the majority of the genetic correlations among locations were positive and highly significant for grain yield. There were some low genetic correlations between some pairs of locations suggesting that these environments are very different (Falconer 1952; Malla et al. 2010; Makumbi et al. 2015). This also indicates that GE has a strong influence (Falconer 1952; Cooper and DeLacy 1994) and hence different systems operate in the two environment (Falconer 1952; Eisen and Saxton 1983). Genotypes × environment interactions are of importance where there are environmental extremes that induce stress conditions (Eisen and Saxton 1983). The Ngetta location had low genetic correlations with other locations in this study. Burdon (1977) pointed out that locations with low genetic correlations between them should be treated separately thus, based on our results Ngetta should be considered a unique environment for evaluations of genotypes for high yield potential.

Genetic correlations can also be used to evaluate similarities among locations. In this study, cluster analysis using genetic correlations based on GY revealed different groups of locations. Similar grouping pattern among locations were reported by Makumbi et al. (2015) while examining the agronomic performance and genotype × environment interaction of Imidazolinone-resistant (IR) open-pollinated maize varieties (OPVs) under Striga-infested and Striga-free conditions in East Africa. But in contrast, Malla et al. (2010) reported that locations used to evaluate wheat germplasm were not clustered according to geographical location. Results showed that Ngetta and Kasese with different altitudes were distinct from the rest locations. The presence of locations (Ngetta and Kasese) that clustered separately suggested the presence of GE and the effect of different crop management practices (Makumbi et al. 2015). These results provided further support that the presence of GE was due to the low genetic correlations between some locations.

Utilization of GGE biplot analysis gave us good visual information on variety performance and stability. An ideal genotype should have both high mean grain yield and high stability within a mega environment (Yan and Tinker 2006; Makumbi et al. 2015). The most stable test genotypes and the check in this study were: G45 (CKHDHH0887), G33 (CKDHH0878), G19 (CKDHH0859, G5 (WM1210), G16 (CKDHH0858), and G9 (WM1214) because their short projection onto the AEC ordinate. Among the most stable genotypes testcross G16 (CKDHH0858) had the highest yield across all the five locations. The second most high yielding G11 (WM1216) was not among the most stable, suggesting that this variety may have specific adaptation to some of the environments as previously reported by Badu-Apraku et al. (2012) who identified high yielding but unstable varieties in West Africa. The commercial check, G50 (Local Check 2) was among the lowest yielding genotype in this study but very stable in the test environments (Badu-Apraku et al. 2012).

A number of DH testcross hybrids showed superiority in GY and other agronomic traits compared to all the checks used by the farmers and in the breeding programs. This suggested there would be increase in production and productivity if these hybrids are eventually released and adopted by farmers. Also, the GGE biplot approach used in this study could help breeders to make better decisions on what genotypes should be recommended for release in the region based on adaptation and stability. Experimental hybrids with more than 20 % yield advantage over the commercial check (e.g. CKDHH0858) and stable across environments should be recommended for release in Uganda and other similar environments in East Africa for adoption.

Notes

Acknowledgments

This research was supported by the Bill and Melinda Gates and the Howard G. Buffet Foundations, and the United States Agency for International Development through the Water Efficient Maize for Africa project. We appreciate all the Zonal Agricultural Research Development Institutes (ZARDI) and Mobuku Irrigation Scheme for making their facilities available for this study. We thank Dr. Dan Makumbi for helpful comments and suggestions on the manuscript. We also appreciate the constructive comments of anonymous reviewers, who helped to improve the manuscript. Also the authors would like to thank Ochen Stephen, Solomon Kaboyo, Annet Nakayima, Majid Walusimbi, Moses Ebellu, Late Stephen Okanya, Fred Ssemazzi and Jane Alupo for data collection at the various experimental sites.

Supplementary material

10681_2015_1549_MOESM1_ESM.xls (123 kb)
Supplementary material 1 (xls 123 kb)
10681_2015_1549_MOESM2_ESM.xls (45 kb)
Supplementary material 2 (xls 45 kb)

References

  1. Badu-Apraku B, Oyekunle M, Obeng-Antwi K, Osuman A, Ado S, Coulibaly N, Yallou C, Abdulai M, Boakyewaa G, Didjeira A (2012) Performance of extra-early maize cultivars based on GGE biplot and AMMI analysis. J Agric Sci Camb 150:473–483CrossRefGoogle Scholar
  2. Baker RJ (1988) Tests for crossover genotype × environment interactions. Can J Plant Sci 68:405–410CrossRefGoogle Scholar
  3. Beyene Y, Mugo S, Pillay K, Tefera TSA, Njoka S, Karaya H, Gakunga J (2011) Testcross performance of doubled haploid maize lines derived from tropical adapted backcross populations. Maydica 56:351–358Google Scholar
  4. Beyene Y, Tarekegne A, Gakunga J, Mugo S, Tefera T, Karaya H, Semagn K, Gethi J, Chavangi A, Asea G, Kiula B, Trevisan W (2013) Genetic distance among doubled haploid maize lines and their testcross performance under drought stress and non-stress conditions. Euphytica 192:379–392CrossRefGoogle Scholar
  5. Bordes J, Charmet G, Dumas De Vaulx R, Pollacsek M, Beckert M, Gallais A, Lapierre A (2007) Doubled-haploid versus single-seed descent and S1-family variation for testcross performance in a maize population. Euphytica 154:41–51CrossRefGoogle Scholar
  6. Burdon RD (1977) Genetic correlation as a concept for studying genotype-environment interaction in forest tree breeding. Silvae Genet. 26:168–175Google Scholar
  7. Butron A, Widstrom N, Snook M, Wiseman B (2002) Recurrent selection for corn earworm (Lepidoptera: Noctuidae) resistance in three closely related corn southern synthetics. J Econ Entomol 95:458–462CrossRefGoogle Scholar
  8. Cooper M, Delacy IH (1994) Relationships among analytical methods used to study genotypic variation and genotype-by environment interaction in plant breeding multi-environment experiments. Theor Appl Genet 88:561–572CrossRefGoogle Scholar
  9. Cooper M, Delacy IH, Basford KE (1996) Relationships among analytical methods used to analyze genotypic adaptation in multi-environment trials. In: Cooper M, Hammer GL (eds) Plant adaptation and crop improvement. CAB Int, Wallingford, pp 193–224Google Scholar
  10. Eisen EJ, Saxton AM (1983) Genotype by environment interactions and genetic correlations involving two environmental factors. Theor Appl Genet 67:75–86CrossRefGoogle Scholar
  11. Epinat-Le S, Dousse S, Lorgeou J, Denis J, Bon-Homme R, Carolo P, Charcosset A (2001) Interpretation of genotype × environment interactions for early maize hybrids over 12 years. Crop Sci 41:663–669CrossRefGoogle Scholar
  12. Falconer D (1952) The problem of environment and selection. Am Nat 86:293–298CrossRefGoogle Scholar
  13. Falconer D, Mackay T (1996) Introduction to quantitative genetics. Longman, LondonGoogle Scholar
  14. Frutos E, Galindo MP, Leiva V (2014) An interactive biplot implementation in R for modeling genotype-by-environment interaction. Stoch Environ Res Risk Assess 28:1629–1641CrossRefGoogle Scholar
  15. Gauch HG (2006) Statistical analysis of yield trials by AMMI and GGE. Crop Sci 46:1488–1500CrossRefGoogle Scholar
  16. Hallauer A, Miranda J (1981) Quantitative genetics in maize breeding. Iowa State University Press, AmesGoogle Scholar
  17. Hallauer AR, Carena M, Miranda Filho JB (2010) Quantitative genetics in maize breeding, 3rd edn. Iowa State University Press, AmesGoogle Scholar
  18. Kang MS (1993) Simultaneous selection for yield and stability in crop performance: consequences for growers. Agron J 85:754–757CrossRefGoogle Scholar
  19. Kassa Y, Asea G, Demissew AK, Ligeyo D, Demewoz N, Saina E, Sserumaga JP, Twumais-Afriyie S, Opio F, Rwomushana I, Gelase N, Gudeta N, Wondimu F, Solomon A, Habtamu Z, Andualem WBA, Habte J, Muduruma Z (2013) Stability in performance of normal and nutritionally enhanced highland maize hybrid genotypes in Eastern Africa. Asian J Plant Sci 12:51–60CrossRefGoogle Scholar
  20. Makumbi D, Diallo A, Kanampiu K, Mugo S, Karaya H (2015) Agronomic performance and genotype x environment interaction of herbicide-resistant maize varieties in Eastern Africa. Crop Sci 55:540–555CrossRefGoogle Scholar
  21. Malla S, Ibrahim AMH, Little R, Kalsbeck S, Glover KD, Ren C (2010) Comparison of shifted multiplicative model, rank correlation, and biplot analysis for clustering winter wheat production environments. Euphytica 174:357–370CrossRefGoogle Scholar
  22. Munyiri S, Pathak R, Tabu I, Gemenet D (2010) Effects of moisture stress at flowering on phenotypic characters of selected local maize landraces in Kenya. J Anim Plant Sci 8:892–899Google Scholar
  23. Odiyo O, Njoroge K, Chemining’wa G, Beyene Y (2014) Performance and adaptability of doubled haploid maize testcross hybrids under drought stress and non-stress conditions. Int Res J Agric Sci Soil Sci 4:150–158CrossRefGoogle Scholar
  24. Oikeh SO, Menkir A, Maziya-Dixon B, Welch RM, Glahn RP, Gauch JRG (2004) Environmental stability of iron and zinc concentrations in grain of elite early-maturing tropical maize genotypes grown under field conditions. J Agric Sci 142:543–551CrossRefGoogle Scholar
  25. Pixley K, Bjarnason M (2002) Stability of grain yield, endosperm modification, and protein quality of hybrid and open-pollinated quality protein maize (QPM) cultivars. Crop Sci 42:1882–1890CrossRefGoogle Scholar
  26. Robinson P (1963) Heritability: a second look. In: Hanson WD, Robinson HF (eds) Statistical genetics and plant breeding. Publ. 982. National Academy of Science. National Research Council, Washington, DC, p 609–614Google Scholar
  27. SAS Institute (2008) SAS/STAT user’s guide. SAS Institute, CaryGoogle Scholar
  28. Thomas W, Forster B, Gertsson B (2003) Doubled haploids in breeding. In: Maluszynski M et al. (eds) Doubled haploid production in crop plants. A manual. Kluwer Academic Publishers, Dordrecht, pp 337–349CrossRefGoogle Scholar
  29. Tukamuhabwa P, Assiimwe M, Nabasirye M, Kabayi P, Maphosa M (2012) Genotype by environment interaction of advanced generation soybean lines for grain yield in Uganda. Afric Crop Sci J 20:107–115Google Scholar
  30. Van Eeuwijk F, Keizer L, Bekker J (1995) Linear and bilinear models for the analysis of multi-environment trials: II. An application to data from the Dutch Maize Variety Trials. Euphytica 84:9–22CrossRefGoogle Scholar
  31. Ward JH (1963) Hierarchical grouping to optimize an objective function. J Am Stat Assoc 58:236–244. doi: 10.1080/01621459.1963.10500845 CrossRefGoogle Scholar
  32. Yan W (2001) GGE biplot: A windows application for graphical analysis of multi-environment trial data and other types of two-way data. Agron J 93:1111–1118CrossRefGoogle Scholar
  33. Yan W, Tinker NA (2006) Biplot analysis of multi-environment trial data: principles and applications. Can J Plant Sci 86:623–645CrossRefGoogle Scholar
  34. Yan W, Hunt L, Sheng Q, Szlavnics Z (2000) Cultivar evaluation and mega-environment investigation based on the GGE biplot. Crop Sci 40:597–605CrossRefGoogle Scholar
  35. Yan W, Kang MS, Ma B, Woods S, Cornelius P (2007) GGE biplot vs. AMMI analysis of genotype-by-environment data. Crop Sci 47:643–655CrossRefGoogle Scholar
  36. Zobel RW, Wright MJ, Gauch HG (1988) Statistical analysis of yield trial. Agron J 80:388–393CrossRefGoogle Scholar

Copyright information

© The Author(s) 2015
corrected publication September 2018

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors and Affiliations

  • Julius Pyton Sserumaga
    • 1
  • Sylvester O. Oikeh
    • 2
  • Stephen Mugo
    • 3
  • Godfrey Asea
    • 1
  • Michael Otim
    • 1
  • Yoseph Beyene
    • 3
  • Grace Abalo
    • 1
  • Joseph Kikafunda
    • 1
  1. 1.Cereals ProgramNational Crops Resources Research Institute (NaCRRI), National Agricultural Research Organization (NARO)KampalaUganda
  2. 2.African Agricultural Technology Foundation (AATF)NairobiKenya
  3. 3.International Maize and Wheat Improvement Center (CIMMYT), ICRAF HouseNairobiKenya

Personalised recommendations