INTRODUCTION
New potato cultivars generally need to be evaluated at different environments
for several years before being released. To achieve this goal, Multi Environmental
Trails (MET) from the core of varietals testing program should be conducted.
Also, MET are important for testing general and specific cultivar adaptation.
A cultivar grown in different environments will frequently show significant
fluctuation in yield performance relative to other cultivars.
The effect of GxE interactions in breeding programs is to reduce correlations
between phenotypic and genotype resulting in invalid or biased conclusions about
genetic variance (Collins et al., 1987).
Genotypexenvironment interaction refers to as differential response of genotypes
or cultivars to across a range of environments. Specific adaptations of genotypes
to subsets of environments is a fundamental issue to be studied in plant breeding
because one genotype may perform well under specific environmental conditions
and may give a poor performance under other conditions (Yan
et al., 2001).
Potato (Solanum tuberosum) are grown around the world in diverse environments
and ranks the world’s fourth most important food crop after wheat, rice
and maize (Manrique and Hermann, 2000). Many traits of
interest to potato breeders have been shown to be sensitive to environmental
changes as shown by previous GxE studies on several traits (Tai,
1971; Yildirim and Caliskan, 1985). It has been
observed that the magnitude of the GxE interactions is a linear function of
the environmental effects. Thus, differences in response by individual cultivars
to a wide range of environments often, follow an orderly pattern which can be
measured as differences between coefficients of linear regression of individual
cultivars in the environments (Hill, 1975).
The regression technique for testing the genotypexenvironment interaction was
first suggested by Yates and Cochran (1938). This technique
was used and modified by Finlay and Wilkinson (1963)
to analyze the adaptation of number of barley cultivars, grown at different
environments. Eberhart and Russell (1966) proposed the
use of two statistical parameters, a regression coefficient (b_{i})
and the deviation from regression ,
to estimate the stability of numeral cultivars. They defined a stable cultivar
as one having a regression coefficient of unity (bi = 1) and the minimum deviation
from regression .
So, joint regression in which cultivars response in regressed on an environmental
index is an important supplementary approach for elucidating the response of
the individual cultivar to a particular environment.
The present study was carried out to achieve the following goals:
• 
Determining the magnitude of GxE interactions variation in
potato regarding economical and quality characters 
• 
Determining adaptability and stability parameters of some potato cultivars
under different environments of Egypt 
• 
Estimation of correlation coefficient among the different studied environments 
MATERIALS AND METHODS
Five new potato cultivars namely Lady Rosetta, Lady Belfour, Valour, Saxon
and Bambino were planted under eight environments presented combinations of
2 locationsx2 seasonsx2 years Table 1 and 2.
Table 1: 
Description of the studied environments 

Table 2: 
Description of the studied locations 

Table 3: 
Some monthly meteorological data of the experimental locations
during the two years (2008 and 2009) of study 

T: Temperature, RH: Relative humidity, W: Winds speed, R:
Rate of rains 
A Randomized Complete Block Design (RCBD), with three replications was used.
The experimental unit at Abbies farm consisted of 10 rows 0.75 m apart and 4
m long with a surface irrigation method. At Sadat city, the experimental unit
consisted of 3 rows, 0.75 m apart and 10 m long under a drip irrigation system.
Spacing between plants within rows was at 25 cm at the both locations.
The description of the eight experimental environments and their planting date
are presented in Table 1 while, the description of the studied
locations are presented in Table 2. Some monthly meteorological
data of the experimental locations, during the four growing seasons at each
location are listed in Table 3. The physical and chemical
analysis of the soil of the two used experimental cities is presented in Table
4. All the agricultural practices used for potato production were carried
out in all experiments in accordance with locally recommended practices.
Measured characters: Ten whole plant samples per plot were randomly
used, 70 days after planting for the determination of the vegetative growth
(plant height (m), number of branches, plant fresh weight (kg), tubers weight/plant
(kg), tubers number/plant, average tuber weight (g), tuber diameter (cm)). Data
were recorded for total yield/ plot at the end of the second season and was
converted to ton/feddan (Feddan = 0.4 ha).
Table 4: 
Physical properties and chemical analyses of the experimental
soils 

EC: Electrical conductivity, OM: Organic matter 
Statistical analysis: Analysis of variance at 0.05 and 0.01 test of
significance for the data from each individual environment were analyzed, using
proc ANOVA (SAS Institute, 2002) for the studied characters,
according to Steel and Torrie (1980). Data were subjected
to combined statistical analysis across environment using proc IML and proc
Mixed (Littell et al., 1996).
Stability parameters: Stability analysis for studied characters was
performed according to the following model of Eberhart and
Russell (1966):
Y_{ij} = u_{i}+B_{i}I_{j}+S_{ij} 
Where: 
Y_{ij} 
= 
The mean of ith cultivar at the jth environment 
u_{i} 
= 
Mean of the ith cultivar over all environment 
B_{i} 
= 
Regression coefficient for the response of the ith cultivar to varying
environments 
I_{j} 
= 
Environmental index obtained as the mean of all cultivars at the environment
min as the grand mean 
S_{ij} 
= 
The deviation from regression of ith cultivar and jth environment 
Simple linear regression analysis, using Proc REG (SAS Institute,
2002) was used to estimate Eberhart and Russell (1966)
parameters for genotypic stability of cultivars across environments.
From the regression analysis, the following four estimates of stability parameters
were calculated linear regression coefficient, i.e., b, S^{2}_{d}
(S^{2}_{y.x}S^{2}_{e}) = Mean square of deviation
from regression, R^{2} (determination coefficient) and the CV for each
cultivar as a fourth measure of stability. The four estimates of stability in
addition to the mean of the studied characters were included to determine the
most stable cultivar. Finally, Spearman’s
rank correlation coefficient between the different studied environments was
calculated.
RESULTS AND DISCUSSION
Mean performance of potato cultivars over the different environments:
The combined analysis of variance presented in Table 5 showed
that all studied morphological and yield characteristics showed highly significant
genotypic differences indicated that the evaluated cultivars differed in their
genetic potentials concerning these characters.
Table 5: 
Combined analysis of variance for the studied morphological
characters of potato 

*^{,}**Significant at 0.05 and 0.01 level of significant,
respectively. Rep; replicates/season (error a), Feddan = 0.4 hectare 
Most of the studied traits reflected clear significant effect for the environmental
factors (locations and seasons), except plant height and average tuber weight,
which indicated that there were some obvious fluctuations in the environmental
conditions throughout the different experiments of the present study, variability
among locations and years could mainly be related to differences in soil type,
temperature and soil moisture conditions, during the various growing seasons
which was also suggested by Ngeve (1991, 1993).
The presence of highly significant of locationxseasons, suggested that climate
was a significant factor in location differences affecting all studied characters
from year to year. Similar explanation was also reached by Harris
(1974), Gruneberg et al. (2005) and Claiskan
et al. (2007). The first order interaction (LxV) and (SxV) appeared
to be highly significant for most of the studied traits except plant height
and number of branches (Table 5), indicating that the cultivars
tended to rank differently when grown at different locations or at different
years, as mentioned by Yildirim and Caliskan (1985)
on potato and the experimental trials should be repeated over locations and
seasons.
The second order interaction (VxLxS), which was considered as the genotypexenvironment
interaction, reflected highly significant effect on all studied characters which
mean that the evaluated cultivars showed different response when grown under
variable environments and should be measured over multiple locations and seasons
to separate cultivarxenvironment interaction component from total genotypic
variance as stated also by Moussa et al. (2011)
on sweet potato. Thus, the findings of the present study seem to fulfill the
basic requirements for stability analysis, for the various studied characters.
Table 6: 
Mean performance of morphological yield and yield components
characters of the five potato cultivars calculated over the environments 

Values followed by the alphabetical in common, within a particular
group of means in each character, do not significantly differ, using Revised
LSD test at 0.05 level of probability 
Data presented in Table 6 indicate clearly that all of studied
morphological, yield and its component characters showed significant differences
among the evaluated cultivars and reflect a large amount of variability. The
cultivar Lady Belfouer produced the highest average fresh weight, number of
branches, tuber weight/plant accompanied with the highest tuber number/plant,
followed by the cultivar Valour and Saxon with significant difference. The cultivar
Lady Rosetta possessed the highest average tuber weight while, Bambino showed
the lowest value for average tuber weight and tuber diameter. Concerning tuber
yield/fad, the highest average were obtained from Lady Belfouer followed by
Valour with insignificant difference 11.52 and 10.86 ton fad.^{1},
respectively while lady Rosetta gave the lowest yield 6.54 ton fad.^{1}.
As for the average effect of the different environment on average tuber weight,
average tuber number and total yield, data in Table 7 declare
that the studied characters were highly affected by the different environment
and appeared to be sensitive for their changes.
Total yield of cultivars ranged from 24.33 ton fad.^{1} for Lady Belfouer
at summer season of 2008 at Sadat, to 0.81 ton fad.^{1} for Bambino
at summer season of 2007 at Abbies, on the other hand. It was noticed also,
that the summer season under Sadat location gave the highest total yield and
tuber weight/plant E_{1} and E_{3}, while, under Abbies location
the winter season E_{6} and E_{8} gave a yield more than the
summer season. The studied cultivars responded differently across environments.
Therefore, their ranks within environments indicated their specific adaptation
which reflect the highly magnitude of genotypexenvironment interaction.
For tuber number/plant, the values ranged from 7.33 for E_{3} to 4.63
E_{2} which reflect the highly effect of season summer and winter under
Sadat location. While, under Abbies location (E_{5}E_{8}) the
fluctuations for total tuber number was stable to a great extent. Lady Belfouer
at E_{3} gave the highest tuber number 10.4 while Lady Rosetta at E_{2}
environment gave the lowest value being 2.83.
Stability parameters: Data of analyses of variance for estimating stability
parameters of the studied yield characters are shown in Table
8.
Table 7: 
Mean performance of the five studied cultivars under different
environment for total yield, tuber weight and tuber number 

The method described by Eberhart and Russell (1966)
was used for estimating stability of the individual genotypes for the studied
characters which already, exhibited significant difference for genotypexenvironment
interaction. The differences among the evaluated cultivars and for the parameters
environment (linear) were found high enough to reach the used significance level
with respect to all studied characters.
The partitioning of mean square (environment+genotypexenvironment) showed that
environments (linear) differed significantly and were quite diverse with respect
to their effects on the performance of genotypes for potato yield and all yield
components. Further, the higher magnitude of mean squares due to environment
(linear) as compared to genotypexenvironment (linear) exhibited that linear
response of environments accounted for the major part of total variation for
all characters studied. Moussa et al. (2011)
also, reported similar results and stated that the mean differences between
seasonal effects and the effect of seasons on total yield of sweet potato were
quite real in nature. The significance of mean squares due to genotypexenvironment
(linear) component against pooled deviation for all studied characters subjected
that the genotypes were diverse for their regression response to change with
the environmental fluctuations. Similarly, the significant mean square due to
pooled deviation observed for all the characters studied suggested that the
deviation from linear regression also, contributed substantially towards the
differences in stability of genotypes. Thus, both linear (predictable) and nonlinear
(unpredictable) components significantly contributed to genotypexenvironment
interaction observed for all studied character. This suggested that predictable
as well as unpredictable were involved in the differential response of stability
but the major component of stability was due to the linear regression and not
to the deviation from the linear function as illustrated by Claiskan
et al. (2007).
Table 8: 
Analysis of variance for the estimated stability parameters
for the morphological and yield component characters, calculated from the
data averaged over all replications 

*^{,}**Significant at 0.05 and 0.01 level of significant,
respectively 
The hybrid is considered to be stable if its among environment variance is
small, or if its response to environment is parallel to the mean response of
all genotypes with in trial and the residual MS from the regression model on
the environmental index (S^{2}_{y.x}) is small.
The modified model of Eberhart and Russell (1966) was
used by several investigators.
A stable preferred hybrid would have approximately b = 1, S2d = 0, in addition
to its superiority in yield. The regression coefficients (b_{i}’s)
were calculated and significantly tested Table 9.
To select certain tested cultivars based on its desirability, the following
criteria were considered: The first criterion was the distance of each cross
from the overall mean, using revised LSD value. The second criterion was the
regression coefficients for each genotype (bi's). Variance of deviation from
regression on environmental index (S^{2}_{d}) was the third
criterion values insignificantly different from the pooled error, suggesting
that they were considered stable. The fourth criterion was the Coefficient of
Variation (CV) for each genotypes. The cultivars with the lowest CV value (<10.0)
would be considered as stable. The CV% values can not statistically be compared,
because it is a ratio between standard deviation and the mean. The genotypes
with high values of R^{2} were also considered desirable. Lin
et al. (1986) indicated that the b value was an indicator of the
response of variety for predictable or macroenvironmental features, while the
S^{2}_{d} was an indicated for micro changes.
Table 9: 
Stability parameters of morphological and yield component
characters of the five evaluated potato cultivars 

*^{,}**Significant at 0.05 and 0.01 level of significant,
respectively 
The results in Table 9 show that mean value for tuber weight/plant
ranged from 0.40 for Bambino to 0.76 kg/plant for Lady Belfouer. The cultivar
Lady Rosetta showed in stability for this trait because its S^{2}_{d}
value were found significant and CV% value were high. The cultivars Lady Belfouer
and Valour appeared to be desirable and stable cultivars because there high
detected mean performance, b_{i} was close to unity and insignificant
value for S^{2}_{d}, while bambino was the least favorable cultivar
because of its lower mean performance.
For tuber number/plant, the high yielding cultivar Lady Belfouer had a high
bi value (2.02) indicating that this cultivar might reflect a higher response
potential for this character under the more favorable environmental conditions,
as illustrated by Habliza et al. (2009) on maize,
by ElSharkawy (2006) on summer squash and Naskar
and Singh (1992) on sweet potato. The cultivar Valour proved to be the most
desirable and stable cultivar followed by the cultivar Saxon.
The most stable cultivar for average tuber weight seemed to be Valour since
it didn't posses any significant difference for the linear regression or the
deviation from regression, the lowest value of CV% 8.31 and the higher value
for R^{2} (0.93). On the other hand, Bambino appeared to be undesirable
because of its lower mean performance (67.13).
Regarding total yield ton/fad., values of bi ranged from 0.84 to 1.13 with
insignificant differences among the evaluated cultivars. The high yielding genotypes
Valour produced mean yield 10.86 ton/fad., over all environments had regression
coefficient close to unity and deviation from regression close to zero. This
indicating its high yielding performance over all environments (and its high
adaptability). Cultivar Lady Belfouer also produced high total yield over range
of environments showed regression coefficient not significant from unity (1.13)
and higher deviation from regression (2.03) indicating specific adaptability
for this genotype at unfavorable environment or poor yielding environments.
The cultivar Saxon considered unstable (S^{2}_{d} value is significant),
while, Lady Rosetta seemed to be undesirable because of its lower production,
S^{2}_{d} significant and low value of R^{2}.
Data in Table 10 show the rank correlation among the studied
eight environments for potato total yield of the five studied cultivars in order
to exclude the highly correlated environment.
Table 10: 
Correlation coefficient among different studied eight environments
for total yield character 

*^{,}**Significant at 0.05 and 0.01 level of probability,
respectively 
The results clearly illustrate the presence of highly positive correlation
between E_{5} and E_{7} (0.98) which represent the summer season
of 2007 and 2008 under Abbies location. The same trend was noticed between E_{6}
and E_{8} (0.94) which represented the winter season of 2007 and 2008
under Abbies location. Meanwhile, this trend was not found under Sadat location.
On the other hand, a highly positive correlation was found between E_{3}
and E_{7} which represented the summer season of 2008 under Abbies and
ELSadat respectively. The results suggested that these environments provide
essentially the same information and consideration could be given to the elimination
or relocation of one of these environments. These results are in agreement with
those of Lynch and Kozub (1988).
CONCLUSION
In the light of results presented here, it can be concluded that the high yielding
genotypes Valour had regression coefficient close to unity and deviation from
regression close to zero this indicating its high adaptability over the studied
environment. Lady Belfouer potato cultivar produced high total yield over range
of environments since regression coefficient was not significant from unity
(1.13) and higher deviation from regression (2.03) indicating specific adaptability
for this genotype at unfavorable environment or poor yielding environment. Saxon
and Bambino cultivars gave low yield and thereby these cultivars deserve further
investigation to maximize their yield under the most proper environmental conditions.