Comparison and suitability of genotype by environment analysis methods for yield-related traits of pearl millet

Abstract

Pearl millet (Pennisetum glaucum (L.) R. Br.) is an important food security and income crop for households living in semi-arid zones in Uganda. However, the genotype by environment interaction, in addition to the several methods used for its assessment, complicates selection of varieties adapted to such semi-arid areas. The objective of this study, therefore, was to compare common methods used to assess stability and adaptability of improved genotypes. Seventy six genotypes were planted in four environments in an alpha experimental design with two replications. Results showed that genotype by environment interactions were significant at p<0.05 for grain yield, days to 50% flowering and 50% physiological maturity, percentage of productive tillers and panicle area. Results further showed inconsistency in ranking of genotypes between methods; although Cultivar Superiority, REML, Yield Stability Index and GGE biplot were consistently correlated and identified high yielding and stable genotypes.
Keywords: GGE biplot, grain yield, pearl millet, stability analysis, Uganda

Full text

Uganda Journal of Agricultural Sciences, 2016, 17 (1): 51 - 66 ISSN 1026-0919 (Print)
ISSN 2410-6909 (Online)
Printed in Uganda. All rights reserved © 2016, National Agricultural Research Organisation
Uganda Journal of Agricultural Sciences by National Agricultural Research Organisation
is licensed under a Creative Commons Attribution 4.0 International License.
Based on a work at www.ajol.info
http://dx.doi.org/10.4314/ujas.v17i1.6
Comparison and suitability of genotype by environment analysis
methods for yield-related traits of pearl millet
G. Lubadde1, J. Ebiyau1, B. Akello2 and M.A. Ugen1
1National Semi-Arid Resources Research Institute, P. O. Box 56, Soroti, Uganda
2Mukono Zonal Agricultural Research and Development Institute, P. O. Box 164, Mukono, Uganda
Author for correspondence: glubadde@gmail.com
Abstract
Pearl millet (Pennisetum glaucum (L.) R. Br.) is an important food security and income crop for
households living in semi-arid zones in Uganda. However, the genotype by environment interaction,
in addition to the several methods used for its assessment, complicates selection of varieties
adapted to such semi-arid areas. The objective of this study, therefore, was to compare common
methods used to assess stability and adaptability of improved genotypes. Seventy six genotypes
were planted in four environments in an alpha experimental design with two replications. Results
showed that genotype by environment interactions were significant at p<0.05 for grain yield, days
to 50% flowering and 50% physiological maturity, percentage of productive tillers and panicle
area. Results further showed inconsistency in ranking of genotypes between methods; although
Cultivar Superiority, REML, Yield Stability Index and GGE biplot were consistently correlated
and identified high yielding and stable genotypes.
Key words: GGE biplot, grain yield, pearl millet, stability analysis, Uganda
Introduction
Pearl millet (Pennisetum glaucum (L.)
R. Br.) is one of the widely grown millets
with several food and non-food uses
(IFAD, 1999). The crop responds
positively to adverse environments that are
extremely variable and often associated
with erratic and low annual rainfall (Bashir
et al., 2014). Despite the adaptability,
average productivity of about 600 kg ha-1
(Rai et al., 1999) from farmers’ fields is
low much as relatively high yielding
genotypes adapted to low-input and
drought-prone environments have been
developed (Serraj et al., 2003; Vadez et
al., 2012). This is partly because the
potential performance of the high-yielding
genotypes under marginalised conditions
is always obscured by the multiplicative
effect of genotype by environment
interaction (GEI) (Yan and Racjan, 2002).

52 G. Lubadde et al.
Accordingly, this causes inconsistent
performance of genotypes (Alberts, 2004),
and thus leading to false selection (Crossa,
1990; Falconer, 1990).
It is in response to these challenges
that it is necessary to assess genotypes
for adaptability and stability (Becker and
Léon, 1988). Equally important is the need
to develop appropriate statistical models
that have the rigor and accuracy to support
selection decisions in case significant GEI
exists, and hence identification of a
reliable method is important (Yau, 1991).
Several statistical analysis methods
have been developed to assess GEI,
notable of which are; analysis of variance
(ANOVA), environmental variance (S2i),
deviation from regression (S2
d), Restricted
Maximum Likelihood (REML) (Bartlett,
1937), regression coefficient (bi) (Finlay
and Wilkinson, 1963), Wricke’s ecovalence
(Wi), Eberhart and Russell (1966), Best
Linear Unbiased Predictions (BLUP)
(Patterson and Thompson, 1971), Tai’s
(1971) approach, Shukla stability variance
(σi2) (Shukla, 1972), coefficient of
determination (ri
2) (Pinthus, 1973),
coefficient of variation (CV) (Francis and
Kannenberg, 1978), cultivar superiority
(Pi) (Lin and Binns, 1988) and static
stability (Becker and Léon, 1988). Some
of the most frequently used methods
include; Additive Main Effects and
Multiplicative Interaction (AMMI)
(Gauch, 1988), yield stability index (YSi)
(Kang, 1993), AMMI stability value
(ASVi) (Purchase, 2000), Genotype and
Genotype by Environment (GGE) biplot
(Yan and Hunt, 2002) and harmonic mean
of the relative performance of genotypic
values (MHPRVG) (Resende, 2007).
However, most of the methods have
deficiencies.
The ANOVA identifies sources of
variation due to GEI effect and allows for
estimation of variance components used
to calculate trait heritability. However, it
does not explore the underlying structure
within the GEI; making it difficult to
establish the true performance of
genotypes across environments (Crossa,
1990). The regression approach is widely
used (Westcott, 1986; Freeman and
Perkins, 1971) but limited in functionality
because genotype response to
environments is largely under multivariate
control; yet regression transforms it into
a univariate variable (Lin et al., 1986).
Crossa (1990) also noted that parameters
of regression (mean, slope, and deviation)
also make it difficult to identify superior
genotypes for particular environments.
The YSi has a weakness of weighing
strongly on yield, yet the trait is influenced
by many factors (Farshadfar et al., 2011).
Wricke’s partition of the interaction is non-
orthogonal yet the test is parametric
(Freeman and Perkins, 1971). The AMMI
models (Gauch, 2006; Gauch et al., 2008)
combine the ANOVA for the genotype
and environment main effects with
principal components analysis which helps
to obtain further insight into the nature and
extent of complex GEI (Alberts, 2004;
Gruneberg et al., 2005). However, there
is difficulty in interpretation of the
interaction when there is limited variability
accounted for by the first principal
component, which could indicate false
statistical stability of the genotypes and
environments (Lavoranti et al., 2007). The
AMMI and the GGE biplot combine
genotype (G) and genotype by
environment (GE) in mega environment
evaluation, but the GGE biplot is superior
to the AMMI in graphical analysis because
it better explains G+GE (Yan et al., 2007).
The inadequacy and contrasting argument
about the best stability and adaptability
analysis methods of GEI shows that most

53
Genotype by environment analysis methods for yield-related traits of pearl millet
probably no stand-alone method exists
(Kaya et al., 2006). Thus the objective
was to assess stability analysis methods
for correlation and consistency using traits
of improved pearl millet genotypes.
Materials and methods
Test environments and materials
The study was conducted for two rainy
seasons which coincided with the second
rains of 2012 and first rains of 2013. The
evaluation was done in two locations
(Kitgum and Serere) and this resulted in
four environments. The Kitgum
environments (E1 and E2) are located at
03°132 N, 032°472 E, 969 m.a.s.l while
the Serere (E3 and E4) environments are
located at 01°32’N, 033°27’E, 1140
m.a.s.l. E1 received 391 mm of rainfall in
2012; while E2 received 817 mm of rainfall
in 2013. E3 received 499.3 mm of rainfall
in 2012; while E4 received 589 mm of
rainfall in 2013). The environments were
characterised as hot spots for rust disease
(Lubadde et al., 2014), sandy soils and
being semi-arid.
The 76 improved pearl millet genotypes
evaluated were replicated twice in a 4 x
19 alpha experimental design. The
materials were planted in 8 m x 5 m plots
at a spacing of 60 cm x 30 cm. A soil
fertility regime recommended for seed
production under rain fed conditions was
adopted and standard agronomic practices
for crop management were used
(Khairwal et al., 2007).
Data collection and analysis
Data were collected on at least 36
randomly selected plants per plot, using
the ‘Descriptors of Pearl Millet’ (IBPGR
and ICRISAT, 1993). The panicle area
(PAR) was calculated as 3.14 x L x W;
where L and W were panicle length and
width, respectively. Data were also
collected on: grain yield (GY in kg ha-1) at
50% physiological maturity after threshing,
days to 50% flowering (FLO50) at plot
level when 50% of the plants have
developed stigmas, days to 50%
physiological maturity (PSM50) and
percentage of productive tillers (PRO) at
plot level. Data analysis was conducted
using the Integrated Breeding Platform for
Breeding Management System version
3.0.8 (IBP-BMS, 2014) and GenStat 15th
Edition (Payne et al., 2012). The
performance and ranking of genotypes
was used to compare the consistency of
the GEI methods. The models and
computations for ANOVA, REML and
AMMI indices for calculating ASVi were
computed using GenStat 15 while the YSi,
Wricke’s ecovalence, Finlay and
Wilkinson, static stability, cultivar
superiority and were computed using IBP-
BMS 3.0.8.
Results
Assessing GEI effect using stability indices
The ANOVA showed that the main effects
of environments were significant (p<
0.05) on GY and PSM50 and highly
significant (p<0.001) for FLO50, PAR and
PRO. The main effects of the genotypes
were also significantly (p<0.05) important
for the yield-related traits except PAR.
In addition, (GEI) was significant (p<0.05)
for all the test traits.
Results for stability and GEI
assessment for twenty most stable
genotypes are shown in Tables 1- 8.
Generally, Cultivar superiority, REML,
Yield stability index (YSi) and GGE biplot
identified highly performing genotypes, as
being stable with a significant positive
correlation observed for most traits (Table
1) and among the methods (Table 2). A

54 G. Lubadde et al.
Table 1. Correlation between highly correlated stability methods and traits
Traits Pi+REML Pi+GGEbiplot Pi+YSi REML+YSi
GY 0.9** 0.5* 0.5* 0.5*
FLO50 -0.8** 0.5* -0.5* 0.5*
PSM50 -0.9** 0.5* -0.6* 0.6**
PRO 0.9** -0.0ns 0.8** 0.7**
PAR 0.6* -0.0ns 0.1ns -0.1ns
Traits: GY = Grain yield, FLO50 = Days to 50% flowering, PSM50 = Days to 50% physiological
maturity, PRO = Percentage of productive tillers, PAR = Panicle area
Methods: Pi = Cultivar superiority, REML = Restricted maximum likelihood, YSi = Yield stability
index
Table 2. Correlation among stability analysis methods for grain yield
Methods Wi Static Pi REML ASVi GGE YSi
stability biplot
bi -0.1 0.3 -0.1 -0.0 -0.2 0.2 -0.2
Wi 1.0 -0.0 0.0 0.1 -0.2 0.5* -0.2
Static stability 1.0 -0.5* -0.6* 0.3 -0.4 -0.6*
Pi 1.0 0.9** 0.1 0.5* 0.5*
REML 1.0 -0.0 0.5* 0.5*
ASVi1.0 0.0 -0.3
GGE biplot 1.0 0.1
Methods: bi = Finlay and Wilkinson, Wi = Wricke’s ecovalence, Pi = Cultivar superiority, REML
= Restricted maximum likelihood, ASVi = Ammi stability value, YSi = Yield stability index
high correlation was observed between
Cultivar superiority and REML, Cultivar
superiority and GGE biplot, Cultivar
Superiority and YSi, REML and YSi and
Finley and Wilkenson and Static stability
for all the traits. However, significant
negative correlation was observed
between Finley and Wilkenson and Static
stability for most traits except grain yield.
Some consistency in genotype ranking
was observed between Finley and
Wilkinson and Static stability then Wricke’s
ecovalence, static stability and ASVi for
all the traits while a similar pattern was
observed between Cultivar superiority and
REML for grain yield, panicle area and
percentage of productive tillers. Similarity
was also observed between Wricke’s
ecovalence and GGE biplot for days to
50% physiological maturity and
percentage of productive tillers.
Grain yield (GY)
Results of ranking of the twenty most
stable genotypes for grain yield are shown
in Table 3. Generally, differences in the
ranking of the genotypes existed for all
the seven stability analysis methods with
Finley and Wilkinson, Wrike’s ecovalence,
static stability and ASVi identifying low

55
Genotype by environment analysis methods for yield-related traits of pearl millet
Table 3. Genotype by environment analysis for grain yield (kg ha-1)
Rank Finley and Wricke’s Static stability Cultivar REML ASVi GGE biplot Yield stability
Wilkenson ecovalence superiority index
Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means
type type type type type type type type
11x8 1820 21812 2x12 1482 6x10 2506 6x10 2324 21812 5x12 2322 3x11 2413
21x16 1585 82005 6x16 1306 3x11 2413 3x11 2258 6x16 1306 6x8 2387 3x12 2257
31x9 1977 62054 1x16 1585 4x16 2344 6x8 2210 1x11 1427 1x14 2355 62054
44x12 1712 92027 4x12 1712 6x8 2387 5x12 2183 3x11 2413 5x8 2187 82005
56x16 1306 12 1878 1x11 1427 3x12 2257 1x14 2173 82005 6x7 2149 92027
62x12 1482 3x9 1797 1x12 1518 1x14 2355 6x9 2172 62054 4x11 2100 6x10 2506
72x15 2169 4x7 1903 16 1799 5x15 2230 4x16 2171 2x12 1482 4x14 2054 41952
84x13 2026 41952 5x16 1621 5x12 2322 3x12 2154 3x9 1797 5x13 2210 4x16 2344
91x7 1671 3x7 1784 11787 6x9 2371 5x13 2102 4x10 1680 6x11 2030 6x8 2387
10 1x13 1906 4x10 1680 3x14 1642 6x7 2149 5x8 2076 12 1878 2x11 1971 6x7 2149
11 2x7 1723 13 1907 4x7 1903 2x15 2169 1x15 2071 4x7 1903 4x8 2003 4x7 1903
12 3x16 1923 16 1799 4x10 1680 4x11 2100 6x12 2057 3x12 2257 6x12 2229 21812
13 6x14 2003 2x9 1822 71869 5x8 2187 5x15 2046 92027 3x11 2413 12 1878
14 3x14 1642 31864 41952 62054 4x11 2041 41952 1x16 1585 51993
15 4x15 1821 3x12 2257 4x15 1821 92027 6x7 2023 3x7 1784 6x10 2506 1x15 2027
16 5x15 2230 11787 21812 6x14 2003 2x15 2011 3x13 1572 5x10 1938 15 1965
17 1x12 1518 10 1855 14 1922 82005 5x9 2002 16 1799 11 1929 13 1907
18 6x13 1914 71869 3x13 1572 4x14 2054 61992 3x10 1463 4x9 1904 5x12 2322
19 1x11 1427 14 1922 3x10 1463 15 1965 4x14 1988 1x12 1518 51984 4x11 2100
20 5x16 1621 15 1965 3x9 1797 5x13 2210 4x8 1976 11787 14 1917 14 1922
1 = ICMV3771, 2 = Manganara, 3 = Okashana2, 4 = ITMV8001, 5 = SDMV94001, 6 = Shibe, 7 = Exbornu, 8 = CIVT9206, 9 = GGB8735, 10 = ICMV221, 11 =
ICMV221white, 12 = KatPM1, 13 = Okoa, 14 = SDMV96053, 15 = Sosank, 16 = Okollo

56 G. Lubadde et al.
Table 4. Genotype by environment analysis for days to 50% flowering
Rank Finley and Wricke’s Static stability Cultivar REML ASVi GGE biplot Yield stability
Wilkenson ecovalence superiority index
Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means
type type type type type type type type
15x13 57.5 12 56.4 4x14 55.9 5x7 62.8 2x11 53.1 2x14 54.6 6x8 59.9 2x14 54.6
24x14 55.9 10 57.7 13 57.3 4x15 61.4 1x9 53.6 2x16 56.5 4x13 60.3 11 55.8
31x13 58.0 11 55.8 5x10 58.3 3x16 60.4 1x11 54.6 12 56.4 4x16 59.9 2x10 55.6
42x12 55.5 1x7 56.3 4x12 59.4 4x8 60.9 3x12 54.6 1x7 56.3 4x10 60.6 1x7 56.3
54x10 60.6 657.5 6x14 55.6 6x8 59.9 2x12 54.9 4x11 57.5 4x8 60.9 1x10 54.8
63x10 56.3 458.6 2x7 58.6 4x10 60.6 2x14 54.9 10 57.7 1x16 60.0 12 56.4
75x9 55.8 4x11 57.5 10 57.7 4x13 60.3 6x13 55.0 13 57.3 1x14 58.4 2x16 56.5
84x12 59.4 2x16 56.5 5x9 55.8 2x15 59.1 1x10 55.1 857.5 6x10 58.4 6x14 55.6
93x11 56.1 758.2 12 56.4 1x16 60.0 2x9 55.6 11 55.8 5x16 57.9 5x12 55.9
10 5x10 58.3 6x16 57.6 3x7 56.5 6x15 59.8 3x8 55.7 458.6 11 55.3 6x12 56.6
11 3x13 55.9 157.8 3x14 58.9 3x14 58.9 2x10 55.8 6x12 56.6 1x15 58.0 13 57.3
12 5x7 62.8 857.5 1x10 54.8 1x12 58.6 6x14 55.8 657.5 2x15 57.5 4x11 57.5
13 1x12 58.6 16 58.5 11 55.8 4x16 59.9 5x8 56.0 1x16 60.0 1x12 58.6 2x9 54.8
14 13 57.3 2x10 55.6 957.2 4x12 59.4 6x11 56.0 2x10 55.6 5x13 57.5 356.3
15 5x11 56.1 1x10 54.8 157.8 3x9 57.9 11 56.0 16 58.5 6x9 57.6 3x12 54.1
16 4x13 60.3 13 57.3 5x11 56.1 2x7 58.6 4x14 56.0 6x16 57.6 16 58.7 5x8 55.4
17 4x15 61.4 5x12 55.9 6x9 57.6 6x10 58.4 5x12 56.1 2x7 58.6 3x14 58.9 657.5
18 6x13 56.4 6x14 55.6 3x13 55.9 4x7 60.9 3x7 56.2 758.2 14 56.6 857.5
19 3x7 56.5 6x11 56.5 6x16 57.6 16 58.5 1x13 56.2 157.8 1x8 57.4 4x14 55.9
20 6x14 55.6 1x15 58.0 3x11 56.1 15 58.3 1x7 56.2 6x14 55.6 2x11 51.6 4x9 56.1
1 = ICMV3771, 2 = Manganara, 3 = Okashana2, 4 = ITMV8001, 5 = SDMV94001, 6 = Shibe, 7 = Exbornu, 8 = CIVT9206, 9 = GGB8735, 10 = ICMV221,
11 = ICMV221white, 12 = KatPM1, 13 = Okoa, 14 = SDMV96053, 15 = Sosank, 16 = Okollo

57
Genotype by environment analysis methods for yield-related traits of pearl millet
Table 5. Genotype by environment analysis for days to 50% physiological maturity
Rank Finley and Wricke’s Static stability Cultivar REML ASVi GGE biplot Yield stability
Wilkenson ecovalence superiority index
Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means
type type type type type type type type
11x13 84.3 12 86.2 4x14 86.3 4x16 95.3 2x11 82.0 13 87.7 4x10 91.3 2x9 81.6
25x13 88.6 4x11 87.1 3x7 87.4 4x7 94.5 2x9 82.2 12 86.2 490.9 1x10 83.5
35x10 89.5 687.5 6x15 89.6 4x15 92.8 3x12 82.8 687.5 387.2 6x14 83.4
43x14 89.8 11 85.2 1x10 83.5 3x16 91.5 1x9 83.0 4x11 87.1 2x12 84.4 11 85.2
54x14 86.3 5x11 83.5 5x10 89.5 5x7 93.0 1x13 83.0 11 85.2 4x12 91.0 12 86.2
66x16 87.4 15 88.2 6x14 83.4 4x12 91.0 6x14 83.2 3x8 86.6 5x12 84.9 5x11 83.5
76x9 88.0 16 90.1 6x16 87.4 4x8 91.9 1x11 83.3 888.0 12 86.8 2x11 81.5
83x7 87.4 187.7 6x9 88.0 4x10 91.3 5x11 83.4 1x10 83.5 11 84.3 3x8 86.6
96x15 89.6 10 87.9 10 87.9 6x8 91.5 1x10 83.5 1x16 90.4 3x12 83.0 2x10 85.0
10 1x10 83.5 4x12 91.0 13 87.7 4x13 91.5 2x12 83.5 2x7 88.4 3x14 89.8 4x11 87.1
11 5x7 93.0 2x16 88.3 3x11 84.3 16 90.1 3x11 84.0 6x14 83.4 16 90.3 4x14 86.3
12 2x12 84.4 6x12 86.5 988.0 1x16 90.4 5x8 84.3 2x9 81.6 4x11 87.1 687.5
13 5x9 89.5 490.0 1x13 84.3 490.0 2x10 84.4 1x15 87.6 687.6 2x14 84.1
14 6x14 83.4 6x10 88.4 2x12 84.4 6x15 89.6 2x14 84.7 1x14 89.6 4x9 87.3 13 87.7
15 1x14 89.6 13 87.69 5x11 83.5 1x12 90.3 11 84.9 10 87.9 3x15 88.4 6x12 86.5
16 4x15 92.8 789.0 187.7 789.0 5x12 84.9 5x11 83.5 986.4 3x11 84.3
17 4x16 95.3 5x12 84.9 3x13 85.1 5x10 89.5 3x13 85.0 15 88.2 15 89.0 5x12 84.9
18 3x11 84.3 6x14 83.4 687.5 1x8 90.3 4x14 85.8 4x14 86.3 5x11 83.5 2x8 87.0
19 13 87.7 386.5 3x14 89.8 3x14 89.8 6x7 85.9 2x10 85.0 5x8 83.6 888.0
20 1x12 90.3 286.7 14 87.5 5x9 89.5 2x8 85.9 2x8 87.0 2x11 81.5 5x15 87.0
1 = ICMV3771, 2 = Manganara, 3 = Okashana2, 4 = ITMV8001, 5 = SDMV94001, 6 = Shibe, 7 = Exbornu, 8 = CIVT9206, 9 = GGB8735, 10 = ICMV221,
11 = ICMV221white, 12 = KatPM1, 13 = Okoa, 14 = SDMV96053, 15 = Sosank, 16 = Okollo

58 G. Lubadde et al.
Table 6. Genotype by environment analysis for percentage of productive tillers
Rank Finley and Wricke’s Static stability Cultivar REML ASVi GGE biplot Yield stability
Wilkenson ecovalence superiority index
Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means
type type type type type type type type
12x15 71.04 182.35 1x9 92.49 1x9 92.49 1x9 92.16 11 85.61 382.22 1x9 92.49
21x14 68.5 14 82.68 5x8 91.24 5x12 91.92 6x7 91.28 182.35 4x13 86.76 1x13 89.46
36x10 86.24 583.53 5x12 91.92 2x11 91.23 5x8 91.16 3x12 86.35 282.15 4x7 89.94
46x11 88.51 281.69 2x11 91.23 6x7 92.17 5x12 90.98 14 82.68 684.21 4x10 90.56
55x9 78.96 13 87.24 5x10 81.55 5x8 91.24 2x11 90.84 6x15 73.77 15 76.61 4x9 88.91
66x12 84.27 682.65 982.26 4x9 88.91 4x10 90.8 583.53 1x10 85.93 4x14 89.79
75x10 81.55 11 85.61 682.65 4x7 89.94 1x13 89.72 785.19 10 84.64 11 85.61
81x9 92.49 10 82.99 6x15 73.77 4x14 89.79 4x11 89.36 281.69 5x13 84.82 785.19
93x13 86.57 485.5 2x16 82.53 13 87.24 5x7 89.3 10 82.99 3x13 86.58 1x10 85.93
10 5x12 91.92 982.26 583.53 6x14 87.72 4x7 89.24 13 87.24 2x15 81.48 4x11 89.19
11 1x16 86.13 4x16 78.88 182.35 1x13 89.46 4x14 88.46 5x8 91.24 5x16 76.6 1x16 86.13
12 5x8 91.24 2x11 91.23 12 82.5 2x13 87.98 4x9 88.41 682.65 2x16 83.39 4x13 86.75
13 4x14 89.79 2x16 82.53 3x12 86.35 3x12 86.35 2x14 87.54 4x7 89.94 12 82.1 10 82.99
14 4x15 74.8 785.19 5x9 78.96 6x11 88.52 3x11 87.53 1x11 84.23 6x12 84.27 2x11 91.23
15 3x12 86.35 4x9 88.91 4x9 88.91 11 85.61 6x8 87.38 485.5 985.03 485.5
16 6x14 87.72 12 82.5 10 82.99 4x11 89.19 5x14 87.18 982.26 2x8 81.38 2x13 87.98
17 2x13 87.98 1x13 89.46 14 82.68 1x16 86.13 13 87.15 2x11 91.23 5x15 84.62 982.26
18 2x12 75.51 2x9 81.66 6x14 87.72 4x13 86.75 6x11 87.08 4x16 78.88 6x16 77.71 1x11 84.23
19 6x15 73.77 382.97 6x7 92.17 485.5 4x13 87.08 4x9 88.91 3x10 81.81 182.35
20 5x11 75.14 6x7 92.17 6x11 88.52 2x14 86.97 6x14 86.85 2x16 82.53 2x9 81.66 2x14 86.97
1= ICMV3771, 2 = Manganara, 3 = Okashana2, 4 = ITMV8001, 5 = SDMV94001, 6 = Shibe, 7 = Exbornu, 8 = CIVT9206, 9 = GGB8735, 10 = ICMV221,
11 = ICMV221white, 12 = KatPM1, 13 = Okoa, 14 = SDMV96053, 15 = Sosank, 16 = Okollo

59
Genotype by environment analysis methods for yield-related traits of pearl millet
Table 7. Genotype by environment analysis for percentage of productive tillers
Rank Finley and Wricke’s Static stability Cultivar REML ASVi GGE biplot Yield stability
Wilkenson ecovalence superiority index
Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means
type type type type type type type type
12x15 71.0 182.4 1x9 92.5 1x9 92.5 1x9 92.2 11 85.6 382.2 1x9 92.5
21x14 68.5 14 82.7 5x8 91.2 5x12 91.9 6x7 91.3 182.4 4x13 86.8 1x13 89.5
36x10 86.2 583.5 5x12 91.9 2x11 91.2 5x8 91.2 3x12 86.4 282.2 4x7 89.9
46x11 88.5 281.7 2x11 91.2 6x7 92.2 5x12 91.0 14 82.7 684.2 4x10 90.6
55x9 79.0 13 87.2 5x10 81.6 5x8 91.2 2x11 90.8 6x15 73.8 15 76.6 4x9 88.9
66x12 84.3 682.7 982.3 4x9 88.9 4x10 90.8 583.5 1x10 85.9 4x14 89.8
75x10 81.6 11 85.6 682.7 4x7 89.9 1x13 89.7 785.2 10 84.6 11 85.6
81x9 92.5 10 83.0 6x15 73.8 4x14 89.8 4x11 89.4 281.7 5x13 84.8 785.2
93x13 86.8 485.5 2x16 82.5 13 87.2 5x7 89.3 10 83.0 3x13 86.6 1x10 85.9
10 5x12 91.9 982.3 583.5 6x14 87.7 4x7 89.2 13 87.2 2x15 81.5 4x11 89.2
11 1x16 86.1 4x16 78.9 182.4 1x13 89.5 4x14 88.5 5x8 91.2 5x16 76.6 1x16 86.1
12 5x8 91.2 2x11 91.2 12 82.5 2x13 88.0 4x9 88.4 682.7 2x16 83.4 4x13 86.8
13 4x14 89.8 2x16 82.5 3x12 86.4 3x12 86.4 2x14 87.5 4x7 89.9 12 82.1 10 83.0
14 4x15 74.8 785.2 5x9 79.0 6x11 88.5 3x11 87.5 1x11 84.2 6x12 84.3 2x11 91.2
15 3x12 86.4 4x9 88.9 4x9 88.9 11 85.6 6x8 87.4 485.5 985.0 485.5
16 6x14 87.7 12 82.5 10 83.0 4x11 89.2 5x14 87.2 982.3 2x8 81.4 2x13 88.0
17 2x13 88.0 1x13 89.5 14 82.7 1x16 86.1 13 87.2 2x11 91.2 5x15 84.6 982.3
18 2x12 75.5 2x9 81.7 6x14 87.7 4x13 86.8 6x11 87.1 4x16 78.9 6x16 77.7 1x11 84.3
19 6x15 73.8 383.0 6x7 92.2 485.5 4x13 87.1 4x9 88.9 3x10 81.8 182.4
20 5x11 75.1 6x7 92.2 6x11 88.5 2x14 87.0 6x14 86.9 2x16 82.5 2x9 81.7 2x14 876.0
1 = ICMV3771, 2 = Manganara, 3 = Okashana2, 4 = ITMV8001, 5 = SDMV94001, 6 = Shibe, 7 = Exbornu, 8 = CIVT9206, 9 = GGB8735, 10 = ICMV221,
11 = ICMV221white, 12 = KatPM1, 13 = Okoa, 14 = SDMV96053, 15 = Sosank, 16 = Okollo

60 G. Lubadde et al.
Table 8. Genotype by environment analysis for panicle area
Rank Finley and Wricke’s Static stability Cultivar REML ASVi GGE biplot Yield stability
Wilkenson ecovalence superiority index
Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means Geno- Means
type type type type type type type type
14x12 759.8 6572.3 4x7 406.2 3x15 1065.3 2x15 1103.5 4x9 516.1 4536.5 1x16 663.7
23x15 1065.3 12 608.2 4x11 379.8 2x8 754.3 4x15 1093.9 6572.3 5x12 770.2 1x13 654.7
34x7 406.2 5x14 430.0 4x16 408.0 3x10 794.4 6x15 956.0 5x14 430.0 2x7 642.2 4x12 759.8
42x8 754.3 4x9 516.1 4533.2 6x8 718.5 6x10 942.7 12 608.2 4x9 516.1 10 600.0
56x8 718.5 3x9 434.9 2x9 654.7 4x12 759.8 5x12 835.1 5x11 485.8 9499.8 9597.3
62x9 654.7 1x10 437.1 6x14 547.2 15 655.5 6x16 759.4 5x10 418.1 14 515.9 4x15 749.5
74x16 408.0 3x12 390.4 2x11 513.1 6x15 809.2 9757.9 3x9 434.9 6x16 598.3 1x11 635.2
84x11 379.8 5x10 418.1 5x11 485.8 3603.3 6x12 744.5 6x11 362.6 4x11 379.8 8563.9
96x14 547.2 2598.4 6x11 362.6 8563.9 2x9 729.5 1x10 437.1 4x16 408.0 2x8 754.3
10 2x11 513.1 6x11 362.6 2x12 472.6 6x10 812.9 6x8 716.3 10 600.0 6x12 634.5 4x13 643.2
11 4533.2 6x9 436.1 11 477.5 12 608.2 1x11 710.5 3x12 390.4 4x14 468.7 7551.6
12 2x15 728.3 6x13 508.6 6x7 562.7 10 600.0 5x11 674.1 2598.4 15 734.6 1x12 579.1
13 6x7 562.7 1x15 446.2 6x12 634.5 1x16 663.7 3x15 673.8 6x9 436.1 1x7 547.3 2x10 656.1
14 4x15 749.5 10 600.0 6x8 718.5 4x8 526 16 633.0 6x13 508.6 2x15 749.3 2x15 728.3
15 5x11 485.8 11 477.5 6x16 598.3 4x15 749.5 12 610.2 11 477.5 3x11 591.0 3x10 794.4
16 2x12 472.6 3x7 491.2 12 608.2 2x7 642.2 2x7 602.3 1x15 446.2 1x11 635.2 2x9 654.7
17 6x16 598.3 14 538.8 9597.3 2598.4 15 601.7 6x16 598.3 2x16 595.9 1x7 547.3
18 6x11 362.6 3x16 452.6 2598.4 6x12 634.5 1x13 593.0 16 562.5 6x15 809.2 2x7 642.2
19 5x7 623.9 3x8 470.4 6x13 508.6 5x12 770.2 1x16 591.4 3x8 470.4 13 576.1 1537.7
20 9597.3 5526.9 6572.3 2x15 728.3 2590.1 3x7 491.2 2x13 483.9 3x15 1065.3
1 = ICMV3771, 2 = Manganara, 3 = Okashana2, 4 = ITMV8001, 5 = SDMV94001, 6 = Shibe, 7 = Exbornu, 8 = CIVT9206, 9 = GGB8735, 10 = ICMV221,
11 = ICMV221white, 12 = KatPM1, 13 = Okoa, 14 = SDMV96053, 15 = Sosank, 16 = Okollo

61
Genotype by environment analysis methods for yield-related traits of pearl millet
yielding (<2000 kg ha-1) genotypes as being
the most stable across environments; while
Cultivar superiority, REML, GGE biplot
and YSi identified high yielding genotypes
as being the most stable. A significant
positive correlation was also observed
between Cultivar superiority, REML, GGE
biplot and YSi although the correlation was
stronger between Cultivar superiority and
REML where both methods identified 16
genotypes as being stable but with a slight
difference in ranking. The Wricke’s
ecovalence, static stability and ASVi
identified 11 out 20 genotypes as being
stable although ranked differently.
Days to 50% flowering (FLO50)
The ranking of the genotypes by the
methods was different for the trait, with
similarity existing only in number of
genotypes identified by each method
(Table 4). The Finley and Wilkinson and
Static stability had 10 genotypes in
common, 6 with Cultivar superiority and
REML while Wricke’s ecovalence and
ASVi had 16 in common, 8 with static
stability and 7 with REML. Cultivar
superiority also had 9 genotypes in
common with GGEbiplot and no genotype
in common with REML.
Days to 50% physiological maturity
(PSM50)
Variation in genotypes and ranking was
also observed across the methods for days
to 50% physiological maturity. In addition,
the similarity level in number of genotypes
commonly identified also varied (Table 5).
The Finley and Wilkinson and Static
stability methods had the highest number
(13) of genotypes in common but ranked
differently. This was followed by Wricke’s
ecovalence and GGE biplot (11), then
Wricke’s ecovalence and ASVi (9). The
cultivar superiority had no genotype in
common with REML while it had only one
with ASVi.
Percentage of productive tillers (PRO)
Differences in genotypes and ranking by
the stability methods were observed for
productive tillers (Table 7). The Cultivar
superiority and REML identified 15 of the
20 genotypes in common and 9 out of 20
with Finley and Wilkinson’s while Wricke’s
ecovalence and ASVi had 14 of 20 most
stable genotypes in common but
differences existed in ranking. Using static
stability, 6 out of 20 genotypes were
common with REML while for GGE biplot,
6 genotypes ranked in common with
Wricke’s ecovalence method. The ranking
for all the genotypes was different in all
the stability methods tested irrespective
of the commonality observed.
Panicle area (PAR)
Variation in ranking of the most stable
genotypes by the tested stability methods
was also observed for panicle area
although some similarities among the
methods existed (Table 8). The Finley and
Wilkinson and static stability had 14
genotypes in common of the 20 most
stable; while Cultivar stability and REML
methods identified 12 genotypes in
common. In addition, Wricke’s ecovalence
and ASVi identified 17 common genotypes
out of 20 most stable genotypes across
environments. The GGE biplot identified
6 common genotypes as Cultivar
superiority and REML while 5 common
genotypes were identified by Finley and
Wilkinson and Static stability.
Discussion
Across the evaluation sites, yield ranged
between 1427 kg ha-1 to 2506 kg ha-1. The
ANOVA indicated significant variation

62 G. Lubadde et al.
among the genotypes tested and the GEI,
showing that the multiplicative interaction
of the genotypes and environments
affected the performance of the test
materials as also reported by Subi et al.
(2013). However, as noted by Crossa
(1990), ANOVA does not explore the
underlying structure within the GEI and
thus other methods were adapted.
Significant correlation among the Cultivar
superiority with REML, YSi and GGE
biplot shows that a prediction of
comparable results can be revealed when
any of the methods is used independently
with minimal variation in the ranking of
the genotypes.
Significant correlation was also
observed elsewhere between Cultivar
superiority and YSi in cotton (Blanche Sr.,
2005) and Faba bean (Temesgena et al.,
2015) studies. These correlated methods
aid in simultaneously selecting stable and
high yielding genotypes unlike the Finlay
and Wilkinson, Wricke’s ecovalence, ASVi
and Static stability which, in this study,
identified mostly low yielding genotypes
as being the most stable. Except ASVi,
similar observations were made by
Mohammadi and Amri (2008) in studies
on wheat. Wrike’s method has also been
reported to identify low yielding genotypes
in sugar cane (Mendes de Paula et al.,
2014) and field pea (Fikere et al., 2014)
as also observed in this study.
The various analysis methods ranked
genotypes differently for the same traits
across the test environments. Similar
observations were also made by Pabale
and Pandya (2010) when they compared
Eberhart and Russell (1966), Perkins and
Jinks (1968) and Freeman and Perkins
(1971) models in ranking of pearl millet
genotypes basing on grain yield. Mustapha
and Bakari (2014) reported no similarity
between static and cultivar superiority;
while cultivar superiority and GGE biplot
identified the same genotypes as being
stable, but ranked them differently in pearl
millet. In this study, Cultivar superiority
and GGE biplot were significantly
correlated for grain yield, days to 50%
flowering and days to 50% physiological
maturity; with a difference in ranking of
genotypes. Variation in ranking of
genotypes was also reported by Parmar
et al. (2012) when they compared
nonparametric tests in rice; Mosleh et al.
(2012) when they compared Wricke’s
ecovalence, Shukla stability variance, rank
test, and Eberhart and Russell; and
Namorato et al. (2009) when they
compared AMMI and Eberhart and
Russell methods in maize. The
inconsistency in ranking was also reported
by Alberts (2004) and Khosa (2012) when
cultivar superiority, Finlay and Wilkinson,
Wricke’s ecovalence and ASVi were
compared in maize. In addition, Dehghani
et al. (2008) also observed variation in
ranking Lentil genotypes, although they
observed similarity between Shukla and
Wricke’s, cultivar superiority and Wricke’s
ecovalence, Finlay and Wilkinson and
cultivar superiority. However, in the
present study the methods had no
significant correlation. The lack of
significant association and differential
ranking of genotypes by ASVi and GGE
biplot was also observed in wheat studies
by Naroui Rad et al. (2013). Results
showed no significant association between
cultivar superiority and Finlay and
Wilkinson’s methods as also reported by
Purchase et al. (2000). On the contrary,
Purchase et al. (2000) reported a
significant correlation between ASVi and
Wricke’s ecovalence as also noted by
Alberts (2004). This implies that results
from the comparisons may greatly depend
on the method, types of genotypes and

63
Genotype by environment analysis methods for yield-related traits of pearl millet
environments being evaluated as also
observed by Westcott (1986) and thus
more than one method should be used to
characterise and explore performance of
genotypes across environments as also
suggested by Lin and Binns (1988).
Acknowledgement
The National Semi Arid Resources
Research Institute for financial support
and the technical staff who helped with
trial management and data collection.
References
Alberts, M.J.A. 2004. A comparison of
statistical methods to describe
genotypexenvironment interaction and
yield stability in multi-location maize
trials. MSc. Thesis, University of Free
State, Bloemfontein, South Africa.
Bartlett, M.S. 1937. Properties of
Sufficiency and Statistical Tests.
Proceedings of the Royal Society A:
Mathematical, Physical and
Engineering Sciences 160 (901):268.
Bashir, E.M.A., Ali, A.M., Adam Ali, M.,
Ismail, M.I., Parzies, H.K. and
Haussmann, B.I.G. 2014. Patterns of
pearl millet genotype-by-environment
interaction for yield performance and
grain iron (Fe) and zinc (Zn)
concentrations in Sudan. Field Crops
Research 166:82-91.
Becker, H.C. and Léon, J. 1988. Stability
analysis in plant breeding. Plant
Breeding 101:1-23.
Blanche, Sr. S.B. 2005. New methods to
assess cotton varietal stability and
identify discriminating environments.
PhD Thesis, Louisiana State
University.
Crossa, J. 1990. Statistical analysis of
multilocational trials. Advances in
Agronomy 44:55-85.
Dehghani, H., Sabaghpour, S.H. and
Sabaghnia, N. 2008. Genotype x
environment interaction for grain yield
of some lentil genotypes and
relationship among univariate stability
statistics. Spanish Journal of
Agricultural Research 6(3):385-394.
Eberhart, S.A. and Russell, W.A. 1966.
Stability parameters for comparing
varieties. Crop Science 6:36-40.
Falconer, D.S. 1990. Selection in different
environments: effects on environmental
sensitivity (reaction norm) and on
mean performance. Genetics
Research Cambridge 56:57-70.
Farshadfar, E., Mahmodi, N. and
Yaghotipoor, A. 2011. AMMI stability
value and simultaneous estimation of
yield and yield stability in bread wheat
(Triticum aestivum L.). Australian
Journal of Crop Science 5(13):1837-
1844.
Francis, T.R. and Kannenberg, L.W.
1978. Yield stability studies in short-
season maize. 1. A descriptive method
for grouping genotypes. Canadian
Journal of Plant Science 58:1029-
1034.
Freeman, G.H. and Perkins, J.M. 1971.
Environmental and genotype-
environmental components of
variability VIII. Relations between
genotypes grown in different
environments and measures of these
environments. Heredity 27:15-23.
Fikere, M., Bing, D.J., Tadesse, T. and
Ayana, A. 2014. Comparison of
biometrical methods to describe yield
stability in field pea (Pisum sativum L.)
under southeastern Ethiopian

64 G. Lubadde et al.
conditions. African Journal for
Agriculture Research 9 (33):2574-
2583.
Finlay, K.W. and Wilkinson, G.N. 1963.
The analysis of adaptation in a plant
breeding programme. Australian
Journal of Agricultural Research
14:742-754.
Gauch, H.G. 1988. Model selection and
validation for yield trials with
interaction. Biometrics 44:705-715.
Gauch, H.G. 2006. Statistical analysis of
yield trials by AMMI and GGE. Crop
Science 46:1488-1500.
Gauch, H.G., Piepho, H.P. and
Annicchiarico, P. 2008. Statistical
analysis of yield trials by AMMI and
GGE: Further considerations. Crop
Science 48:866-889.
Gebre, W. 2014. Evaluation of pearl millet
(Pennisetum glaucum L.) genotypes
for yield and yield stability in South
Omo and West Hararghe. Journal of
Biology, Agriculture and Healthcare
4(8):99-121.
Gruneberg, W.J., Manrique, K., Zhang, D.
and Hermann, M. 2005. Genotype x
environment interactions for a diverse
set of sweet potato clones evaluated
across varying eco-graphic conditions
in Peru. Crop Science 45:2160-2171.
IBPGR and ICRISAT. 1993. Descriptors
for pearl millet [Pennisetum glaucum
(L.) R. Br.]. International Board for
Plant Genetic Resources, Rome, Italy;
International Crops Research Institute
for the Semi-Arid Tropics, Patancheru,
India.
IFAD. 1999. Farmer participatory testing
of technologies to increase sorghum
and pearl millet production in the Sahel.
www.ifad.org/grants.
Integrated Breeding Platform’s
Breeding Management System version
3.0.8 (IBP-BMS). 2014. Integrated
Breeding Platform, CIMMYT, Mexico-
Veracruz.
Ishaq, J. and Meseka, S. 2014. Genetic
Stability of Grain Yield and principal
component analysis in pearl millet
(Pennisetum glaucum L.). Greener
Journal of Plant Breeding and Crop
Science 2(4):88-92.
Kang, M.S. 1993. Simultaneous selection
for yield and stability in crop
performance trials: Consequences for
growers. Agronomy Journal 85:754-
757.
Kaya, Y., Aksura, M. and Taner, S. 2006.
GGE-Biplot analysis of
multienvironment yield trials in bread
wheat. Turkish Journal of
Agriculture and Forestry 30:325-337.
Khairwal, I.S., Yadav S.K., Rai K.N.,
Upadhyaya H.D., Kachhawa D.,
Nirwan B., Bhattacharjee R.,
Rajpurohit B.S., Dangaria, C.J. and
Srikant. 2007a. Evaluation and
identification of promising pearl millet
germplasm for grain and fodder traits.
Journal of SAT Agricultural
Research 5(1):1-5.
Khosa, S. 2012. Assessment of maize
germplasm lines for genetic diversity,
cultivar superiority and combining
ability. MSc Thesis, University of
KwaZulu Natal, South Africa.
Lavoranti, O.J., dos Santos Dias, C.T. and
Kraznowski, W.J. 2007. Phenotypic
stability via AMMI model with
bootstrap re-sampling. Pesquisa
Florestal Brasileira 54:45-52.
Lin, C.S. and Binns, M.R. 1988. A
superiority measure of cultivar
performance for cultivar x location
data. Canadian Journal of Plant
Science 68:193-198.
Mohammadi. R. and Amri, A. 2008.
Comparison of parametric and non-
parametric methods for selecting

65
Genotype by environment analysis methods for yield-related traits of pearl millet
stable and adapted durum wheat
genotypes in variable environments.
Euphytica 159:419-432.
Mosleh, R.Z., Sepas, A.Y., Chaichi, M.
and Lalehloo, B.S. 2015. The Analysis
of Genotype-Environment Interactions:
Comparison of Parametric and Non-
Parametric Tests for Interactions in
Bread Wheat Genotypes in Cold
Regions of Iran. MAGNT Research
Report 3(3):103-109.
Mustapha, M. and Bakari, H.R. 2014.
Statistical Evaluation of Grain Yield in
Millet Trials Using Principal
Component Analysis. Current Trends
in Technology and Sciences journal
3(6):422-428.
Namorato, H., Miranda, G.V., Vagno de
Souza, T., Oliveira, L.R., DeLima,
R.O. and Mantovani, E.E. 2009.
Comparing Biplot Multivariate
Analyses with Eberhart and Russell’
method for genotype x environment
interaction. Crop Breeding and
Applied Biotechnology 9:299-307.
Naroui Rad, M.R., Abdul Kadir, M., Rafii,
M.Y., Jaafar, H.Z.E., Naghavi, M.R.
and Ahmadi F. 2013. Genotype ×
environment interaction by AMMI and
GGE biplot analysis in three
consecutive generations of wheat
(Triticum aestivum) under normal and
drought stress conditions. Australian
Journal of Crop Science 7(7):956-
961.
Pabale, S.S. and Pandya, H.R. 2010. A
comparison of different stability models
for genotype x environment interaction
in pearl millet. Electronic Journal of
Plant Breeding 1(5):1294-1298.
Parmar, D.J., Patel, J.S., Mehta, A.M.,
Makwana, M.G. and Patel S.R. 2012.
Non-Parametric methods for
interpreting genotype x environment
interaction of rice genotypes (Oryza
Sativa L.). Journal of Rice Research
5:1-2.
Patterson, H.D. and Thompson, R. 1971.
Recovery of inter-block in formation
when block sizes are unequal.
Biometrika 58:545-554.
Payne, R.W., Murray, D.A., Harding,
S.A., Baird, D.B. and Soutar, D.M.
2012 Genstat® for Windows™, 15th
Edition, Introduction. VSN
International, Oxford.
Perkins, J.M., and Jink J.L. 1968.
Environmental and genotype
environmental components of
variability. III. Multiple lines and
crosses. Heredity 23(2):39-256.
Pinthus, M.J. 1973. Estimate of genotypic
value: A proposed method. Euphytica
22:121-123.
Purchase, J.L., Hatting, H. and Van
Deventer, C.S. 2000. Genotype x
environment interaction of winter
wheat (T. aestivum) in South Africa:
Stability analysis of yield performance.
South African Journal of Plant and
Soil 17(3):101-107.
Rai, K.N., D.S. Murty, D.J. Andrews. and
Bramel-Cox P.J. 1999. Genetic
enhancement of pearl millet and
sorghum for the semi-arid tropics of
Asia and Africa. Genome 42:617-628.
Serraj, R., Bidinger, F.R., Chauhan, Y.S.,
Seetharama, N., Nigam S.N. and
Saxena, N.P. 2003. Management of
drought in ICRISAT cereal and legume
mandate crops. pp. 127-144. In: Kijne,
J.W., Barker R. and Molden, D. (eds.).
Water Productivity in Agriculture:
Limits and Opportunities for
Improvement. CAB International.
Shukla, G.K. 1972. Some statistical
aspects of partitioning genotype-
environmental components of
variability. Heredity 29:237-245.

66 G. Lubadde et al.
Subi, M.I.M. and Idris, A.E. 2013. Genetic
variability, heritability and genetic
advance in pearl millet (Pennisetum
glaucum [L.] R. Br.) genotypes.
British Biotechnology Journal
3(1):54-65.
Tai, G.C.C. 1971. Genotypic stability
analysis and application to potato
regional trials. Crop Science 11:184-
190.
Temesgena, T., Kenenib, G., Seferaa, T.
and Jarsob, M. 2015. Yield stability and
relationships among stability
parameters in faba bean (Vicia faba
L.) genotypes. The Crop Journal
3:258-268.
Vadez, V., Hash, T., Bidinger, F.R. and J.
Kholova. 2012. Phenotyping pearl
millet for adaptation to drought.
Frontiers of Physiology 3:1-12. doi:
10.3389/fphys.2012.00386.
Westcott, B. 1986. Some methods of
analysing genotype-environment
interaction. Heredity 56:243-253.
Yan, W. and Hunt L.A. 2002. Genetic and
environmental causes of GXE
interaction for winter wheat yield in
Ontario. Crop Science 41:19-25.
Yan, W., M.S. Kang, Ma, B., Woods, S.
and Cornelius, P.L. 2007. GGE Biplot
vs. AMMI Analysis of genotype-by-
environment data. Crop Science
47(2):643-653.
Yan, W. and Racjan, I. 2002. Biplot
analysis of test environments and trait
relations of soybean in Ontario. Crop
Science 42:11-20.
Yau, S.K. 1991. Need of scale
transformation in cluster analysis of
genotypes based on multi-location yield
data. Journal of Genetics and
Breeding 45:71-76.