Assessment of Tuber Yield Stability and Adaptability of Some Elite Yam Genotypes in the Guinea Savannah Ecology of Northern Ghana

Studies were conducted to determine tuber yield stability and adaptability of some elite yam (Dioscorea sp.) genotypes in northern Ghana. Ten elite exotic yam genotypes alongside one locally cultivated farmer-preferred variety, Laribako, were grown in five environments between 2010 and 2012. These 11 genotypes were arranged in a randomised complete block design with three replications and assessed for tuber yield and yield components. Analysis of variance indicated significant 
 
 
 
 p
 <
 0.05
 
 
 
 genotypic variation for tuber yield and the yield components studied. Genotype × environment interaction effect was significant 
 
 
 
 p
 <
 0.05
 
 
 
 for tuber yield and mean tuber weight but not significant 
 
 
 
 p
 >
 0.05
 
 
 
 for number of tubers per mound. Apart from genotype 95/18922, all the exotic genotypes had significantly 
 
 
 
 p
 <
 0.05
 
 
 
 higher tuber yields than the local check, Laribako. The highest tuber yield (16.03 t ha−1) across environments was obtained from 96/19158 followed by 95/00594 (14.9 t ha−1). According to the additive main effect multiplicative interaction (AMMI) analysis, genotype (G), environment (E), and GxE interaction, respectively, explained 39.71%, 36.03%, and 24.26% of the total sum of squares for tuber yield. For number of tubers per plant, GxE effect explained the greatest percentage (60.46%) of the total sum of squares compared to genotype effect (22.00%) and environment effect (17.54%). The local variety, Laribako, was more stable across all environments though low yielding compared to the exotic genotypes. Three genotypes, 95/19158, 95/19177, and 96/02025, were more stable across environments than the other exotic genotypes. Genotype 95/18544 was the most sensitive and for that matter responded positively in the favorable environments. The study identified genotypes with specific and general adaptation potential across different environments for tuber yield that can be further tested in on-farm trials for possible release.


Introduction
Yam (Dioscorea sp.) is among the oldest recorded food crops believed to have originated in tropical areas [1]. ough six economically important species (D. rotundata, D. alata, D. caryensis, D. esculenta, D. dometorum, and D. bulbifera) are grown as staple foods in Africa [2], D. rotundata is the most widely cultivated within the yam zones of West Africa. Sub-Saharan Africa accounts for 70% of world yam production [3], and in Ghana, in particular, the crop is cultivated in 34 out of the 43 agriculturally important districts in the Guinea savannah zone in the north, savannah transition Northern Region [7]. Ghana earned over US $1.8 million in 2016 from the export of yam as a result of increased production of over seven million metric tonnes of yam. is made yam the number one foreign exchange earner for the country among all arable crop commodities [6].
Despite these important roles played by yam in ensuring food security and income generation, few improved varieties exist for cultivation by farmers. Most farmers in northern Ghana still prefer and cultivate their local varieties which though low yielding and susceptible to pests and diseases, still meet their food requirements [8]. Kusi et al. [9] observed that pests and diseases occurrences in yam cultivation are prevalent at earlier stages of growth and hence proposed both preplanting and postplanting protective treatments to ensure successful production of yam. However, several agronomic interventions on enhancing tuber yield have been made in yam research, and aspects of the genetic improvement of the crop are still at rudimentary stage with only few varieties released after introduction and evaluation of exotic genotypes in Ghana (the first official release was done in 2005) [10]. However, these have failed to adapt to all ecologies in Ghana. Most often, these introduced genotypes also fail to adapt to the local environments due to genotype × environment interaction which occurs as a result of differential responses of genotypes to different environmental conditions [4].
Earlier reports [11][12][13] have attributed variation in yield performance of yam to inherent genotypic characteristics, as well as preferences for different environmental conditions [14,15]. erefore, careful evaluation is critical in identifying the suitable genotypes that will give the highest possible yield in different environments [16]. High yield and stability of genotypes across different environments are very important attributes desired by plant breeders. As a result, breeding materials are tested in diverse environments to assess the consistency in genotypic performance in order to identify superior ones for wider or specific adaptation [17,18]. Genotypes whose G × E interaction effect remains insignificant from one environment to the other and across years are considered as stable [14,19].
ough several methods have been used to assess genotype × environment interaction and stability for yield in crops, the additive main effect and multiplication interaction (AMMI) and genotype main effect and genotype × environment (GGE) biplot analyses which combine principal component analysis (PCA) and a graphical explanation of GEI have been very useful in recent times. e AMMI model integrates analysis of variance (ANOVA) and PCA for the analysis of stability of genotypes in multienvironment trials [20] and also estimates the contribution of each genotype and environment to the total GEI variation [21]. De Oliveira et al. [22] assessed the adaptability and stability of yellow passion fruit varieties in Brazil and suggested that the use of the AMMI stability value and yield stability in a single nonparametric index was useful for discriminating genotypes with superior and stable fruit yield. e GGE biplot on the other hand provides graphical representation of genotypes' adaptation to particular environments and facilitates the identification of representative environments, detects the ability of test environments to discriminate and identifies stable genotypes in MET [20,23]. Otoo et al. [15] observed differential performances of 12 yam genotypes in different ecologies using the GGE biplot. Ali et al. [17] also identified specifically adapted upland cotton genotypes for yield and fiber traits; likewise, Jahanzaib et al. [24] also estimated genotype × environment interaction for seed yield in groundnut and adaptability to specific environments in Iran. ese findings suggest that the GGE biplot is an appropriate tool for assessing yam genotypes for yield stability and adaptability to specific yam growing areas. e objective of this study was to assess different yam genotypes for tuber yield stability and identification of elite yam genotypes adapted to specific environments within the Guinea savannah ecology of northern Ghana.

Materials and Methods
e study involved ten elite exotic yam genotypes (Table 1) obtained from the International Institute of Tropical Agriculture and one elite local farmer-preferred variety, Laribako. e local variety used is widely cultivated in most of the yam growing communities in the Guinea savannah ecology. e trials were carried out at five locations in the Guinea savannah agroecological zone (each being a representative of a wider location) between 2010 and 2012. ese locations were Mbowura, Lantinkpa, Yendi, and Nyankpala all in the Northern Region. e location (spatial) and year (temporal) combinations gave a total of five (5) environments (Table 2). e study areas have a monomodal rainfall pattern with an annual average of 1200 mm. At each environment, the land was slashed and ploughed after which the plots were laid out in a randomised complete block design with three replications. Mounds were manually raised using a spacing of 1.2 × 1.2 m. Each genotype was assigned to a plot which consisted of 20 mounds. Staking was done at two months after planting using stakes measuring about 1.5 m high. Weeding was done as and when necessary. Harvesting was done at eight months after planting. At harvest, data were collected on tuber weight (kilograms) and number of tubers per mound. Tuber yield (t ha −1 ) was then calculated as weight of tubers from the net plots and extrapolated to one hectare. Mean tuber weight (g) was also estimated as the weight of tubers harvested per plot divided by the number of tubers.

Data Analysis.
e data were subjected to combined analysis of variance using Genstat 12.1 [25]. Mean comparison among genotypic performances at different locations was done using the least significant difference test (LSD at 5%).

Stability Analysis.
Additive Main effect and Multiplicative Interaction (AMMI) analysis of variance was further carried out for tuber weight of the 11 yam genotypes obtained per plot across environments using the Genstat software [25]. e AMMI model combines the analysis of variance and principal component analysis in a single analysis [26]. e principal component analysis decomposes the GEI component of the overall phenotype. AMMI stability value and yield stability index (YSI) were computed according to Adjebeng-Danquah et al. [27]. e AMMI model according to Zobel et al. [28] is presented as where Y ij is the response of i th genotype in the j th environment, µ is the grand mean, g i is the mean of the i th genotype minus the grand mean, e j is the mean of the j th environment minus the grand mean, λ k is the square root of the eigenvalue of the PCA axis k, α ik , and c jk are the principal component scores for PCA axis k of the i th genotype and the j th environment, respectively, and e ij is the residual. e environment and genotypic PCA scores are expressed as unit vector times the square root of λ k , i.e., environment PCA score � λ k 0.5 c jk , 0.5, genotype PCA score � λ k 0.5 α ik . AMMI stability value (ASV) was calculated for each genotype according to the relative contributions of the principal component axis scores (IPCA1 and IPCA2) to the interaction sum of squares. e AMMI stability value (ASV) as described by Purchase et al. [29] was calculated as follows: where IPCA1 Sum of squares /IPCA2 Sum of squares � the weight given to the IPCA1-value by dividing the IPCA1 sum of squares (from the AMMI analysis of variance table) by the IPCA2 sum of squares. e larger the IPCA score, either negative or positive, the more adapted a genotype is to a certain environment. Smaller ASV scores indicate a more stable genotype across environments [30]. e yield stability index was also calculated using the sum of the ranking based on yield and ranking based on the AMMI stability value: where RASV � the rank of the genotypes based on the AMMI stability value and RY � the rank of the genotypes based on yield across environments (RY). YSI incorporates both mean yield and stability in a single criterion. Low values of both parameters show desirable genotypes with high mean yield and stability [31,32].
Subsequently, the GGE biplot method as proposed by Yan [33] was used to visualise the genotype main effect and genotype × environment interaction patterns in the data in a biplot.
e "which-won-where" pattern was used to visually identify the high performing and adapted as well as those suitable for the different test environments. is was carried out using the R software package [34].  International Journal of Agronomy 3

Results
Analysis of variance showed significant (p < 0.05) interaction and highly significant (p < 0.001) main effects for tuber yield (Table 3). Genotype effect explained 39.71% of the total sum of squares due to treatment followed by environment (36.03%) and interaction effect (24.26%). Only the first interaction principal component axis (IPCA 1) was significant and accounted for 59.74% of the interaction sum of squares with the IPCA2 accounting for 24.36%.

Mean Performance of Genotypes in Five
Environments. e performance of the 11 elite yam genotypes in each of the five environments is presented in Table 4

Ranking of Genotypes Based on Yield and Stability.
Ranking of genotypes based on tuber yield, AMMI stability value (ASV), and yield stability index (YSI) is presented in Table 5. Genotype 95/18922 had the lowest ASV making it the most stable or nonresponsive. On the other hand, genotype 95/19158 with the highest ASV score was the most unstable or responsive genotype. In terms of yield, the reverse was the case. Genotype 95/19158 with the highest mean tuber yield across environments was ranked as number one. Based on the yield stability index (ranking of ASV + ranking of tuber yield), genotype 96/00594 was ranked as the best genotype combining high yield with high stability. e highest yielding genotype, 95/19158, was ranked 5 th overall due to its low ranking according to the ASV. Two genotypes, 96/00594 and 96/02610, combined high yield with high stability, whereas 95/19158 can be considered as high yielding and moderately stable. On the other hand, 95/19177 can be considered as high yielding and unstable whilst the lowest yielding genotypes with high stability can be considered unresponsive to the environment.

Discriminatory Ability and Representativeness of the Different Environments.
Average-environment view of the 11 genotypes evaluated in five environments is presented in Figure 1. e environment vector with the smallest angle with the AEA is more representative of the other test environments. erefore, E4, E2, and E3 were more representative than the other environments. E1 and E5 were the least representative, whereas E2 and E3 were identified as good for selecting generally adapted genotypes. e length of the environment vector indicates its discriminatory ability. erefore, E5 was the most discriminating environment and as such more suitable for initial selection of genotypes for tuber yield.

Relationships between Test Environments.
e environment vector view of the biplot for tuber yield of 11 genotypes in five environments indicates the correlation between the different environments ( Figure 2). e angle between two environments indicates their correlation. Environments with acute angles are highly correlated whereas environments with obtuse angle represent negative correlation. In this case, E4 (Yendi 2012) and E5 (Nyankpala 2012) were highly and positively correlated, whereas environments E1 (Mbowura 2010), E2 (Lantinkpa 2011), and E3 (Lantinkpa 2012) were also highly positively correlated.
ere was a weak correlation between E1 and E5.

Mean Performance versus Stability.
e GGE biplots in Figure 3 present mean vs. stability display of the 11 yam genotypes in five environments. e single environmental line that passes through the biplot origin and the average tester (vertical line) is referred to as average tester coordinate axis. e vertical line separates genotypes with below average means (to the opposite in direction of the arrow) from those with above-average means, with lines parallel to the vertical line indicating genotypic stability. Based on this information,

Adaptability of Genotypes to Specific Environments.
e polygon view drawn on the farthest genotypes is presented in Figure 4. e vertex genotypes indicate the best or poorest in the environments within that sector. e biplot was divided into five sectors. Only sectors 1 and 5 had environments. Sector 1 had four environments; E2, E3, E4, and E5 whilst sector 5 had only one environment, E1. A genotype at the apex of a vertex represents the winning genotype in all the environments within that sector. In this case, 95/19158 (G6) was the winning genotype at Lantinkpa 2011 (E2), Lantinkpa 2012 (E3), Nyankpala 2012 (E4), and Yendi 2012 (E5). e proximity of a genotype to an environment indicates that the genotype or group of genotypes had their best performance or more adapted to that particular environment. is therefore means that genotype 96/2610 (G10) was better adapted to the prevailing conditions at Mbowura 2010, Lantinkpa 2011, and Lantinkpa 2012. Figure 5, G8 (9/00594) was closer to the centre of the concentric circles (ideal genotype) and therefore more desirable since it combines high tuber yield with high stability across environments.

Comparison Biplot Showing Ideal Genotype for High and Stable Yield. From
ough G6 (95/ 19158) had the highest average tuber yield across locations, it was considered less desirable compared to G8 (9/00594). Genotypes on the right side of the biplot had below average performance. G11 (Laribako) and G3 (95/18922) were the poorest in terms of tuber yield and stability due to their consistently poor performance across environments as indicated by their distance from the centre of the concentric circles.
is phenomenon confounds selection of such genotypes for general adaptation. In the present study, genotype, environment, and genotype × environment interaction effects significantly affected tuber yield in the genotypes studied. Nduwumuremyi et al. [37] indicated that the presence of significant genotype × environment interaction effect for quantitative traits such as tuber yield can seriously limit the efforts made in selecting superior genotypes for different environments.
is is because such performance cannot be replicated in environments with different growing conditions. Different genotypes of yam possess inherent varietal characteristics and preferences for different environmental conditions [13][14][15] especially for foreign genotypes that have been introduced into new environments.
erefore, genotypes need to be evaluated across different environments in order to identify specific locations that best suit them and where such genotypes can attain their full yield potential [19]. e use of stability analysis enables breeders to quantify the extent of genotype × environment interaction and characterise genotypes as widely adapted (to several environments) or specifically adapted (to single or few environment) based on stability indices [38].
A genotype is considered to be stable if it has a low degree of fluctuation in its yielding ability when planted over diverse environments [39,40]. Osei et al. [35] and Yan and Tinker [23] also suggested that genotypes whose variances remain relatively constant from one environment to the other are considered stable. However, a consistently poor genotype can still be stable. erefore, stability should always be considered in addition to superior performance for that trait of interest. Purchase et al. [29] therefore proposed a yield stability index that combines ranking based on high yield and stability (based on the AMMI stability value) into a single index, the yield stability index (YSI). e AMMI analysis identifies specific genotypes which are stable across environments [20,21]. From this study, genotype 95/19158 ranked highest in terms of yield across environments, but low in terms of stability.
is implies that it responded positively under favorable environmental conditions but poorly under the less favorable environmental, thus suggesting specific adaptation. With good agricultural practices, such a genotype would be ideal genotype for high input agriculture of growth in favorable environmental conditions (agronomic stability).
ough the yield of locally adaptable genotypes tends to remain consistent across environments, they still fail to respond positively even in favorable environments. Such genotypes are classified as unresponsive to the environmental conditions (biologically stable) and remain the best insurance for farmers in harsh conditions. Other genotypes however tend to respond positively to favorable environments whilst their yields remain moderate under harsh conditions. Such genotypes are normally selected for specific environments where they could exhibit their full yield potential. In the present study, the yield of the farmer variety, Laribako, remained consistent, though low across the five environments. However, the tuber yield of genotype 95/ 19158 remained moderatively consistent across environments and was the highest ranked genotype in four out of the five environments. e genotype main effect and genotype × environment (GGE) biplot visualizes the main effect of genotype as well as genotype × environment interaction [41]. e "which-wonwhere" pattern of the polygon view-based interaction of GGE biplot is useful for the identification of elite genotypes  in single or multiple environments [42]. Genotypes at the apices of the polygon (vertex cultivars) represent the best performing genotypes of all environments within that sector. In the present study, the use of the GGE biplots revealed the genotypes that combined high mean performance with high stability as well as preferences and adaptation to some specific environments (Figures 3 and 4). is indicated that genotype 95/19177 was adapted to the environments, E4 (Yendi 2012) and E5 (Nyankpala 2012) whereas genotypes 95/18544 and 96/00594 had their best performances at E1 (Mbowura 2010), E2 (Lantinkpa 2011), and E3 (Lantinkpa 2012). However, genotype 95/19158 was adapted to four environments; E1 (Mbowura 2010), E2 (Lantinkpa 2011), E3 (Lantinkpa 2012), and E4 (Yendi 2012). is phenomenon of specific adaptation or environmental preferences by different yam genotypes has also been reported in earlier studies on genotype × environment analysis. Otoo and Asiedu [43] evaluated and identified specifically adapted Dioscorea rotundata "Tela" genotypes in Ghana. Otoo et al. [15] also in a study involving 12 Dioscorea rotundata genotypes identified specifically adapted cultivars in 16 environments using the GGE biplot, again confirming the environmental specificity of different yam genotypes.
According to Yan and Tinker [23], an ideal genotype is one that combines high mean yield with high level of stability. ough such an ideal genotype may not exist under natural conditions, it is used as a reference for genotype evaluation [44]. e centre of the concentric circles (Figure 5) represents the position of an ideal genotype, which is defined by a projection onto the mean-environment axis that equals the longest vector of the genotypes that had aboveaverage mean yield and by a zero projection onto the perpendicular line (zero variability across environments). A genotype is more desirable if it is closer to the ideal genotype. It allows the evaluation of genotypes according to their nearness or how far they are to this centre [44,45]. From Figure 5, G8 (9/00594) was the most desirable genotype which combined high yield with high stability across environments. ough G6 (95/19158) had the highest average tuber yield across locations, it was considered less desirable compared to G8 (9/00594). Genotypes on the right side of the biplot had below average performance. Laribako (G11) and 95/18922 (G3) were undesirable due to their consistently poor performances across environments as indicated by their distance from the centre of the concentric circles [46]. is study provides novel information on genotype × environment interaction effect on yam yield in the Guinea savannah ecology of Ghana and portrays the need to test yam genotypes for environmental specificity before their release for cultivation by farmers in the area.

Conclusion
Adaptation and stability of yield performance are very important in developing new yam varieties especially for introduced genotypes. e different genotypes showed adaptation to different environments. Genotype 95/19158 had the highest overall performance across environments with moderate stability. Genotypes 96/00594 and 96/02610 combined high yield with high stability according to the yield stability index (YSI). Genotype 95/19177 though relatively high yielding was less stable according to the AMMI stability value and yield stability index and therefore suited for specific adaptation. e farmer-preferred variety, Laribako, was consistently low yielding across environments. From the study, genotype 95/19158 had specific adaptation whilst 96/00594, 96/02610, 96/02025, and 95/18949 had general adaptation. e findings from this study provide important information on the need for multilocational testing of new yam genotypes prior to release.
Data Availability e data obtained in the experiment have been used in the writing of this manuscript.

Conflicts of Interest
e authors declare that they have no conflicts of interest regarding the publication of this paper.