Stability assessment for selection of elite sugarcane clones across multi-environment based on AMMI and GGE-biplot models

Seven field experiments were conducted at three experiment stations representing major sugarcane producing regions in Egypt. Each experiment comprised a randomized complete block design with three replications. Fourteen elite breeding lines typical of those routinely generated in the three final selection stages of sugarcane breeding programs in Egypt, along with one check variety (GT54-9) were evaluated for cane and sugar yield in this study during the 2018/2019, 2019/2020 and 2020/2021 seasons. Stability parameters including cultivar stability rank and superiority index were determined. The data was also investigated using GGE-biplots, the additive main effects and multiplicative interaction model (AMMI), and the AMMI stability value (ASV). The genotype main effect was used to visualize the G x E interaction. The results of these trials are of significance in guiding the selection and recommendation of superior sugarcane varieties and more stable in sugarcane production zones. The clone G.2016–129 had a mean sugar yield and cultivar superiority index for sugar yield exceeding that of GT54-9, and hence was recommended for commercial planting. Because of local conditions in Egypt, an elite sugarcane variety would have high and stable yield and would adapt to a wide range of environments. In the present study, only one clone G.2016–129 fit that definition by producing higher and more stable sugar yield than the commercial variety GT 54–9.. At the side of multivariate analyses, the ASV (AMMI stability value) supports selection of stable varieties in the AMMI Method. Varieties with lowest ASV are stable. Therefore, the results of this study exposed that G.2016–95, F-150 and G.2016–129 with lowest ASV for cane yield by contrast, G.2009–11, G.2016–128, F-150 and G.2016–95 with lowest ASV for sugar yield, were stable clones for cane and sugar yields, respectively.


Introduction
Sugarcane (Saccharum spp. hybrids) is a major cash and industrial crop in Egypt. It has been planted in Egypt since 641 AD, following the introduction of the crop by the Arabs. Sugarcane is presently planted along the Nile River, from Shandaweel station (Sohag governorate) ((latitude of 26 33° N, longitude of 31 41° E and altitude of 69 m above sea level) to Kom Ombo station (Aswan governorate) (24.46_ N and longitude of 32.93_ E). The last decade has seen significant advancements in the local breeding and selection programs in Egypt with Abstract Seven field experiments were conducted at three experiment stations representing major sugarcane producing regions in Egypt. Each experiment comprised a randomized complete block design with three replications. Fourteen elite breeding lines typical of those routinely generated in the three final selection stages of sugarcane breeding programs in Egypt, along with one check variety (GT54-9) were evaluated for cane and sugar yield in this study during the 2018/2019, 2019/2020 and 2020/2021 seasons. Stability parameters including cultivar stability rank and superiority index were determined. The data was also investigated using GGE-biplots, the additive main effects and multiplicative interaction model (AMMI), and the AMMI stability value (ASV). The genotype main effect was used to visualize the G x E interaction. The results of these trials are of significance in guiding the selection and recommendation of superior sugarcane varieties and more stable in sugarcane production zones. The clone G.2016-129 had a mean sugar yield and cultivar superiority index for sugar yield exceeding that of GT54-9, and hence was recommended for commercial planting. Because of local conditions in Egypt, an elite sugarcane variety would have high and stable yield and current commercial cultivars originating from local programs. Local programs now conduct the whole range of activities necessary for variety development including controlled hybridization, early clonal selection and testing through to the final varietal testing stage in multiple locations (Mehareb et al. 2021a).
Sugarcane varietal selection from a series of different environments trials can be considered as a multi-characteristic selection in which the yields in multi-environments are synonymous with characters. As such an analysis of the data combined over various environments should be conducted so as to form selection index (Smith et al. 2007).
Stability and adaptability are main criteria for selection of genotypes in any breeding program (Wolde et al. 2018). Adapted and stable varieties with high yield potential are identified throughout multilocation trials. Nevertheless, breeders invariably encounter genotype × environment interactions (GEI) when evaluating varieties over several of environments, which complicates response of the selection. It is important to quantify GEI in order to design selection methods which can accurately recognize superior cultivars in the final selection stages and predict their potential performance in numerous environments. The cultivar superiority index and stability measurements (Huehn 1990;Lin and Binns 1988), AMMI (Additive main effects and multiplicative interaction) (Gauch et al. 2006;Gauch et al. 2008) and GGE (genotype and genotype by environment) biplot analysis (Yan et al. 2000;Yan and Tinker 2006a) techniques are the most commonly used multivariate approaches to analyse the data collected from multi-environment trials.
The GGE-biplot and AMMI models can be used to support the stability and superiority indices in recognizing varieties with both specific and broad adaptation (Kaya 2006). Multi-environment trials are important for proper ranking of candidate cultivars and to recognize representative selection or production environments (Yan et al. 2007). This could accelerate breeding efficiency (Yan and Holland 2010) and strengthen the competitiveness of yield production (Gauch and Zobel 1997).
Cultivar superiority index is an alternative tool for assessing mean performance and stability simultaneously (Lin and Binns 1988). The cultivar superiority index defined as the mean square difference between a cultivar's observed value and the best variety within a given environment. This index can be a powerful tool for breeders to selection new sugarcane colons.
Thus, the main objectives of this study were to.
1. Determine the cane and sugar yield stability of studied genotypes using different stability measures and compare those stability statistics from trials conducted across seven test environments. 2. Identify new sugarcane genotypes with stable and high cane and sugar yield for release and registration through selection indices and cultivar superiority index by the AMMI and GGE-biplot models.

Experimental materials
The sugarcane genotypes including 14 elite breeding lines and one check variety GT54-9 at the three final selection stages (Main Experiment, Final Experiment and Demonstration) were used in this study. The genotypes were developed by Sugar Crop Research Institute, Agriculture Research Center (ARC), Egypt during 2018/2019, 2019/2020 and 2020/2021 seasons and could be considered as a representative of clones typically evaluated in the Egyptian sugarcane breeding program (Table 1).

Experimental design and growing conditions
A randomized complete block design with three replications was used in each of the seven experiments. Each experimental plot was 42 m 2 including 6 rows of 7 m in length and 1.0 m row spacing.
All field experiments were conducted at three stations for selection trials, extending from southern to middle Egypt, namely Kom Ombo station (Aswan governorate), Shandaweel station (Sohag governorate) and Mattana Station (Luxor governorate. These stations cover the range of the sugarcane production area ( Table 2). The experiments were planted during March and harvested after 12 months, in all seasons. Irrigation was withheld for 1 month before harvesting. Other agricultural practices were done following recommended commercial practice for the three regions. The plants were harvested at 12 months after planting for all seasons.

Statistical analysis
Analyses of variance were separately carried out as a randomized complete block design (RCBD) for each of the seven environments for the collected data as per (Gomez and Gomez 1984). The Means were compared by the Tukey's test (p < 0.05) using the Minitab 14 software.

Stability analysis and genotype by interaction (G × E)
The following stability measurements were performed on cane and sugar yields under the seven environments (two locations × 2 year plus one location × 3 years), superiority measure (Lin and Binns 1988), and mean absolute rank difference of varieties on the environment (Akcura and Kaya 2008). Additionally, the additive main effects and multiplicative interaction model (AMMI) (Romagosa and Fox 1993) was applied on the cane and sugar yields. Then the genotype main effect (Akcura and Kaya 2008) was used to visualize the G × E interaction.
The AMMI stability value ASV is the distance from the coordinate point to the origin in a twodimensional plot of scores for interaction the first principal component analysis IPCA1 and the second principal component analysis IPCA2 scores in the additive means effect and multiplicative.
interaction model (Purchase 1997). As the first principal component analysis IPCA1 score gives more to the genotype by environment interaction sum of squares, a weighted value is chosen. This was estimated for each genotype and each environment giving to the relative contribution of IPCA1 and the second principal component analysis IPCA2 as per (Purchase et al. 2000;Naroui Rad et al. 2013).
The GGE-biplot (genotype main effect plus G xE interaction) (Tollo et al. 2020) was used to visualize the GEI. All stability and G × E analysis was performed using GeneStat-18 software program.

Performance of sugarcane genotypes across seven environments
The performance of sugarcane genotypes across seven environments (Tables 3 and 4) shows, that the genotypes varied significantly in cane and sugar yield, respectively. G2 (G.2016-168) produced the higher value of cane yield of 68.04 and 67.68 ton.acre-1,in E1 and E2 which was 8.9% and 6.2% higher than the check variety GT. 54-9 (Table 5). Additionally, G.2016-168 produced a sugar yield of 8.45 and 8.04 ton.acre −1 , in E1 and E2, respectively, which was 30% and 20.5% higher than the check variety GT. 54-9.  5.84 cd 7.50bc 6.23bcd 7.05a 6.46bcd 6.21ab G. 2000-5 (G14) 8.01ab 5.18e 4.47e 5.91d 6.71ab 5.91def 4.81f G.T.54-9 (G15) 6.51f 6.67b 9.60a 6.79ab 5.23 cd 6.98b 5.48cde Additive main effect and multiplicative interaction (AMMI) The combined analysis of variance for cane and sugar yield showed that genotypes (G) contributed 12% and 17.8% respectively to the total sum of squares while environments (E) contributed 21.4% and 8.4% respectively. GEI contributed to 66.3% and 73.4% to the total sum of squares of cane and sugar yield respectively (Table 5). The first and second interaction principal component axis (IPCA1 and IPCA2) were highly significant and accounted for 56.11and 20.81% of the sums of squares for cane yield and 44.06 and 28.52% for sugar yield of the total GEI variation, respectively. The GEI was highly significant implying differential response of genotypes to environments (Elbasyoni 2018). An ideal environment (Al-Naggar et al. 2020) and genotype are the one which is on the central circle ( Figs. 1 and 2). Thus, Fig. 1 and 2 shows the comparison plot for clones, and a model clone is one which is near or at the middle of the concentric circle. Based on this assumption, G7 (G2016-129) was the most ideal clone, with high mean cane, sugar yields and high stability.
The analysis AMMI stability values (ASV) exposed that some sugarcane genotypes have high adaptation; though, most of genotypes have specific adaptability ( Table 6). The ASV values showed variations in cane yield stability among the fifteen sugarcane clones (Table 6). According to Purchase et al. (2000), a stable cultivar is defined as one with AMMI stability values (ASV) value close to zero. The larger the ASV value, either negative or positive, the more specifically adapted a genotype was to certain environments. A smaller ASV value indicated a more stable genotype across environments (Purchase 1997). Consequently, the genotypes G.2016-95 and F-150 with ASVof 0.54 and 1.48, respectively, in addition to G.2016-129 with ASVof 1.64 were the most stable, while the genotypes such as G.99-80, G.2016-158 and G.2016-62 were the least stable (Table 6). Other than ASVexposed differences in sugar yield stability among the 15 sugarcane genotypes. On the other hand the genotypes G.2009-11, G.2016-128 and G.2016-129 were the most stable for sugar yield.

Stability measurements and cultivar superiority index
The sugarcane genotypes were ranked according to their superiority indices, cane and sugar yield and stability measures over the seven environments (Tables 7  and 8). Sugarcane genotypes exhibited variances in their means superiority and stability values. Cultivar superiority index varied from 69 to 378.10. GT54-9 showed the highest mean cane yield, and was regarded as greatest superior and stable genotype with a superiority index of 69 and stability values of 123.7. By contrast, G.2016-129 genotype recorded the highest mean sugar yield, and was regarded as maximum superior and stable genotype with a superiority index of 0.66 and stability values of 0.46. GT54-9 ranked second in terms of superiority index in sugar yield however its cane yield was the highest mean. GGE biplot: genotypic discriminating ability and representativeness of the studied environments The GGE -biplot method showed that the PC1 and PC2 could explain 65.98% and 64.11% of total GEI variation for cane and sugar yields, respectively (Figs. 3, 4). Stable varieties and environments with small IPCA-1 and IPCA-2 scores are close the origin of the GGE biplot graph (Yan and Tinker 2005). Which-won-where GGE-biplot charts are separated by an equality line into sectors in which dissimilar mega environments can be noticed Tinker 2005, 2006b). In this experiment, the equality line divided the studied environments into six mega-environments for cane yield and four mega-environments for sugar yields (Figs. 2, 3, 4). For cane yields, the first mega environment consisted of E1 and E4 (Fig. 3), while the first mega  (Fig. 4). Figure 3 and 4 help visualize the distance between each environment and the perfect environment, ''model studied environment'', which is at the middle of the concentric circles. Therefore, E1 and E4 for cane yield and E1, E2 and E5 for sugar yield, respectively were the greatest representative environment and had the maximum ability for discriminating varieties with respect to cane and sugar yields. The vertex clone connected by the polygon in Fig. 5 were G14 and G7 for cane yield (Fig. 5a). Additionally, G7 (G2016-129) was the vertex clone for sugar yield (Fig. 5b). The vertex clones were those extreme from the origin of the GGE biplot, signifying that these were either the best performers in terms of sugar yield in all or some of the environments based on their direction from origin (Tollo et al. 2020).

Discussion
Fifteen sugarcane genotypes were evaluated for cane and sugar yield potential and stability in this study. These involved fourteen elite breeding lines at the final selection stages of sugarcane clones were used in this study developed by Sugar Crop Research and one check variety GT54-9. These clones are representative products of the sugarcane breeding efforts in Egypt. Seven pilot test environments were selected from the all sugarcane production zones for evaluation and demonstration purposes. The results of these trials are of significance in guiding the selection and recommendation of superior sugarcane varieties and more stable in Egypt. Five clones, namely, G7 (G.2016-129), G15 (G.T.54-9), G2 (G.2016-168), G13 (G.2005-47) and G1 (F-150) produced both high sugar and cane yields as were predicted by the GGE and AMMI biplots models.  One clone, G7 (G.2016-129) produced relatively high sugar yield, but lower cane yield compared to the commercial variety GT54-9.
Three clones G6 (G.2016-99), G10 (G.2009-11) and G14(G.2000-5) gave higher Cultivar Superiority index for relatively high cane yields, but lower Cultivar Superiority index for sugar yields. The clone G7 (G.2016-129) had better sugar yield and Cultivar Superiority index for sugar yield then that of GT54-9, and hence was recommended for planting using a reasonable distribution of sugarcane varieties based on local conditions. Because of local conditions in Egypt, an elite sugarcane variety should have high and stable yield in order to adapt to a wide range of environments. In the present study, only one clone G7 (G.2016-129), fits that definition by producing higher and more stable sugar yield than the commercial variety GT 54-9.
GGE-biplot and AMMI methods are able to estimates the weight of the environments, the genotypes and G × E using a value that estimate variety stability in all environments taking into account the cane and sugar yield. Based on the observation and analysis of G × E in different-environment, cane and sugar yield trials are very important for selection, evaluation and recommendation of crop varieties. Stability measurements, AMMI; and GGEbiplot method were beneficial to determine the ideal genotypes for multi -environments. The AMMI methods are commonly used in breeding program of the most popular crops such as rice ( (Bassiony et al. 2020;Mehareb et al. 2021b).
Based on the multivariate analyses, the ASV (AMMI stability value) supports selection of stable varieties in the AMMI Method. Varieties with lowest ASV are stable. Therefore, the results of this study revealed that G9 (G.2016-95), G1 (F-150) and G7 (G.2016-129) with lowest ASV for cane yield by contrast, G10, G8 (G.2016-128), G1 and G9 (G.2016-95) with lowest ASV for sugar yield, were stable clones for cane and sugar yields, respectively. Similar results were found in a study on sugar beet (Hassani et al. 2018).
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). The authors have not disclosed any funding.

Conflict of interest
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.