Assessment of genotype x environment interaction and pod yield evaluation of groundnut ( Arachis hypogaea L . ) genotypes in Zimbabwe

Groundnut is an important component of the diet of both rural and urban populations in the SubSaharan Africa. The national average pod yield is 0.25 t/ha which is far less than the global average. The diverse environmental conditions of Zimbabwe make selection and release of stable groundnut genotypes a challenge, mainly due to genotype x environment interaction (GEI). Twenty-five groundnut genotypes were evaluated to examine the level and type of GEI on pod yield. The genotypes were evaluated under multi-environmental yield trial conducted in (2013/14 summer season) at five environments. The objectives of the experiment were to determine the presence of GEI on pod yield stability of groundnut genotypes, to identify genotypes that are specifically or widely adapted. General combined analysis of variance (GenStat Version 14) at 5% significance level indicated that genotypes (G) were not significant (p = 0.153), environments (E) and genotype x environment interactions (GEI) were highly significant (P < 0.05) on pod yield. The environment influenced yield of the groundnut genotypes. Environment and genotype explained 58.8 and 6.1% respectively of the total treatment variance, whilst the genotype by environment interaction accounted for 35.1%, indicating that environment influenced a lot on the performance of the genotypes. High significant level of GEI indicates that some genotypes may be released for specific environments. Basing on the mean pod yield value from the combined (ANOVA) analysis of variance results, groundnut genotype G24 (3.34 t/ha, check variety) was the highest yielder, followed by G7 (3.31t/ha) and then G2, G14 and G11 (3.29, 3.25 and 3.02 t/ha respectively). The results indicate that the experimental genotypes have great potential to be released and grown on large scale production. Stability analysis based on one multivariate or various uni-variate parameters to extract more information on the GEI on pod yield stability of groundnut is recommended.


INTRODUCTION
Zimbabwean environment is so diverse and so sophisticated (Nyamapfene, 1991;Rukuni et al., 2006) and that leads to very high levels of genotype x environment interactions (GEI).The heterogeneous nature of the Zimbabwean environments (agro-ecological regions) results in the performance of different groundnut genotypes to differ both within and across environments.Higher genotype x environment interaction is usually expected to be as a result of large environmental differences as in Zimbabwe.In most cases, this kind of interaction may lead one genotype in having the highest yield in some environments and may be lowest in others, whilst the second genotype may excels in other environments in which the first one might have failed (Gauch and Zobel, 1996).For that reason, it is important to know and understand the level of the interactions in the selection of genotypes across several environments rather than only calculating the average performance of the genotypes under evaluation (Fehr, 1991;Gauch and Zobel, 1997).
It has been noted that genotypes tested in different locations or years often have significant fluctuation in yield due to the response of genotypes to environmental factors such as climate, soil fertility, pests and disease pathogens (Kang, 2004).These variations in yield are the ones that are usually referred to as genotype x environment interaction (GEI) and they are so frequent whenever experiments are conducted.Genotype x environment interactions has been studied in many crops by many different researchers.One of the major complications in all breeding programs is genotype x environment (G x E) interaction.It has been noted that a proper understanding of the environmental and genetic factors that causes the interaction as well as an assessment of their importance is likely to have a great impact on the development, evaluation and selection of superior germplasm (Magari and Kang, 1993;Basford and Cooper, 1998).
The phenotypic expression of an organism (plants/crops included) is due to its genotype (G), the surrounding environment (E) as well as the interaction of the two (G x E).The presence of significant GxE interactions complicates the process of selecting only genotypes with higher performance (high yielding) since the genotype is going to yield differently in different environments (this leads to change in rank order).For that reason, multi-environment trials are now broadly used to assess the suitability of genotypes for target environments (DeLacy et al., 1996).
When experiments are conducted under varying environments, genotypes that always give high average yields with minimum G x E interaction have been gaining importance over increased yields (Ceccarelli, 1989;Gauch and Zobel;1997, Kang, 1998)).The analysis of G x E interaction is closely related with the quantitative estimation of phenotypic stability of genotypes over different environments (Kang, 1998).When significant G Savemore et al. 55 x E interaction is observed, the effects of genotypes and environments are statistically non-additive; this implies that the differences between genotypes are due to the environment and not genotypes themselves.G x E interactions may, but not all the time, lead to different rank orders of genotypes in different environments.The presence of G x E interaction in multi environment trials leads to a need for the analysis of genotype stability (usually yield stability).Many authors have described yield stability in many different ways over the years and there have also been different concepts of stability tests (Lin et al., 1986).According to Becker and Leon (1988), many researchers use the terms adaptation, phenotypic stability and yield stability in different ways.Chahal and Gosal (2002) noted that stability indicates consistency in performance that would mean minimum variation among environments for a particular genotype.
The prime reason for researchers to perform multienvironmental genotype evaluation is to estimate and or predict how the genotype is likely to perform in future years and future environments, basing on the performance data of the past, and to develop or recommend superior ones.In almost all multi-location trials, there exists interaction and noise (Purchase, 1997).Selecting for high pod yielding and genotypes that have wide adaptation is the ideal situation that breeders would want to concentrate on rather than concentrating on genotypes that might give the highest pod yield in only one environment.Alternatively, in the case that there are no ideal genotypes, then selecting for specifically adapted genotypes would be the next option.That means the genotypes will be released and recommended for specific areas and not for the growing areas.This chapter focuses on the level and nature of G x E interaction based on the general combined analysis of variance techniques as a primary tool to differentiate these groundnut genotypes according to their pod yield performance.
This study was designed to (i) examine the level and type of Genotype x Environment Interaction for pod yield (ii) identify groundnut genotypes with high yielding varieties, and (iii) determine the need of doing pod yield stability analysis.

MATERIALS AND METHODS
A total of 25 genotypes (4 commercially released varieties and 21 intermediate experimental lines) were tested in 2013/14 summer season.All the check varieties and the intermediated experimental lines were obtained from Crop Breeding Institute (C.B.I).Ilanda and Tern are the highest yielding short season groundnut varieties and for that reason they were included as check varieties.More details on genotypes and the information on their breeding status are highlighted in Table 1.
Author(s) agree that this article remains permanently open access under the terms of the Creative Commons Attribution License 4.0 International License

Management
The seeding rate that was used is 100 kg/ha for all environments.
Compound D was applied at planting at a general recommendedrate of 300 kg/ha.Gypsum was also applied during first flowering (7 to 8 weeks after planting) at a general recommended rate of 300 kg/ha.Harvesting was done manually, were 2.4 m (0.3 m from either sides of the row) of the 3 m rows were harvested as net plot by way of hand pulling as well as hand plucking.Pod yield was then recorded after drying the groundnut pods to 12.5% moisture content by exposing the pods to the sun and moisture content was measured using the moisture meter.All other recommended groundnut production practices such as weed, pest and disease management were followed and practiced.

Experimental design
The trials were laid in a Complete Randomized Block Design  (CRBD) at all the sites (Figures 1, 2 and 3).Each of the twenty-five treatments with 3 replicates and that translated to seventy-five plots in total.The plot sizes were 5.4 m 2 with 5 rows of 3 m long with spacing of 0.45 m between rows.The net plot size was 2.16 m 2 , 1 row from both sides and 0.3 m from either side was discarded.

Records taken
Records that were taken include, days to flowering, days to maturity, diseases scores, insect pest scores, pod size, seed size, shelling percentage and pod yield.For the sake of this study, only pod yield was considered for statistical analysis.Pod yield was recorded on the net plot basis.After drying and cleaning, the weights of the pods per plot were recorded and converted to t/ha using a formula.

Analysis of variance
General combined analysis of variance (ANOVA) for pod yield data was conducted using GenStat 14 th Edition software to determine  the G, E and GEI effects.The effects of the genotypes, environments as well as their interaction were determined from ANOVA analysis.

General combined ANOVA and mean yield performance
General analysis of variance at 5% significance level indicated that genotypes (G) were not significant (p = 0.153), but environments (E) and genotype x environment interactions (GEI) were highly significant both (P < 0.001) on pod yield of twenty-five groundnut genotypes and accounted for 4.12, 39.68 and 23.72% of the total sum of squares, respectively (Table 3).This indicates that the environment influenced the yielding ability of the groundnut genotypes.In this research, environment and genotype explained 58.8 and 6.1% of the total treatment variance, whilst the genotype by environment interaction accounted for 35.1%, this indicates that the environment had a lot of influence on the performance of the genotypes.Similar results that confirm that environment contributes a more genotype and environment interaction to the total treatment variance were obtained on wheat, were the genotypes, environments and their interactions were significant, with the environment contributing much of the variation (Gauch, 2006).In their research, the effects of environment and genotype explained 83.78 and 2.71% of total treatment variance respectively, whereas the interaction explained 10.08% of the total treatment variance.In Table 3 it is shown that the (large) total variance for environments was 76.02% indicating higher heterogeneity in the environmental conditions among the five locations used in the study, hence the groundnut genotype pod yield was largely influenced by the environments.This was also consistent with findings  Kang, 2003) which showed that environment is the dominant source of variation, while G and GE are relatively small in yield trials across locations.According to Zerihun (2011), in most cases under normal multienvironment yield trials, environment (E) accounts for 80% or higher of the total yield variation, while genotype (G) and genotype x environment interaction (GEI) each account for about 10%.The magnitude of genotype by environment interaction sum of squares (1.89%) was larger than of genotypes (1.32%), indicating that there were substantial differences in genotypic responses across environments (Table 3).The analysis of genotype by environment interaction pattern is highly important for scientists such as plant breeders, because it enables them to design the correct strategies (selecting for wide/general or specific adaptation) for new genotypes to be released for commercial production.

Genotype x environment interaction
There were inconsistencies in pod yield rankings of genotypes across environments as shown in Table 4.This gives rise to cross over type of GEI indicating that there was inconsistent genotype pod yield performance across environments.Table 4 indicates that the following genotypes had highest pod yield at different environments; G7 in E1 (Harare); G11 in E3 (Kadoma); G1 in E4 (Pamure); G7 in E5 (Save Valley); G2 in E2 (Gwebi VTC).The presence of cross over GEI shows the existence of different mega environments in which different winning genotypes can be selected (Crossa et al., 1991).The mean pod yields in Table 4 indicate that there were differences in rankings of pod yield performance among genotypes across environments (cross over GEI).This is shown by some genotypes which attained maximum yields in more than one environment, for instance, G7 in E1 (Harare) and E5 (Save Valley); (Table 4).According to (Crossa et al., 1991), it is common for a multi environment yield trial to constitute a mixture of cross over and non-cross over types of GEI.Crossover interactions result due to the G x E interaction, in which case there will be non-parallel response curves of genotypes (intersecting each other)

Figure 1 .
Figure 1.Field showing part of the trial under study at Gwebi VTC.

Figure 2 .
Figure 2. Field showing part of the trial under study at Harare.

Figure 3 .
Figure 3. Field showing part of the trial under study at Kadoma.

Table 1 .
Pedigree information and source of the planting materials.

Table 2 .
Description for the sites used on the multi-environmental groundnut yield trials in 2014.
Experimental Station).More details on the testing sites and the agro-ecological characteristics for all the locations used are shown in Table2.

Table 3 .
General combined analysis of variance for pod yield (t/ha) of twenty-five groundnut genotypes evaluated across five locations over a season.Coefficient of variation (%CV) = 18.1%.