Sweet potato (Ipomoea batatas [L.] Lam), is a hexaploid (2n = 6x = 90) and is one of the most important food security crops globally. It belongs to the family Convolvulaceae, genus Ipomoea, and, according to Vaeasey et al. (2008), the genus has over 600 species, of which batatas is the only one with economic value. In many developing countries, sweet potato is reported to be the fifth most important food crop after rice, wheat, maize, and cassava (Aina et al., 2012). Over 110 million metric tons of sweet potato was produced in 2018, with China producing 53.01 million metric tons representing 65.6% of the world sweet potato production (FAOSTAT, 2018). Africa was responsible for 20.7 million tons which represents about 25.4% of the world production. Nigeria is the second highest sweet potato producer in Africa and the third highest producer in the world, with production of 4.03 million metric tons, which is 5.0% of the world's production (FAOSTAT, 2018). Over the years, sweet potato production has been on the increase in Nigeria. In the last two decades, production has increased 10-fold. However, the increased production can be attributed to the expansion of land under sweet potato cultivation rather than increased yield per unit area, as yield remains abysmally low at an average of <3.0 tons/ha. This low yield is mainly due to the widespread use of obsolete production inputs and methods, chief of which are the use old, unimproved cultivars and the practice of mixed cropping with incompatible crops. Therefore, one important way of mitigating against poor root yield in farmers' fields is to develop and release new sweet potato varieties with stable and high root yield potential into the farming system.

Understanding the differential response of crop genotypes to changing environmental conditions is of key importance in plant breeding. One major step toward the development of improved crop genotypes is the assessment of the nature of interactions that exist between genotypes and the production environment for a particular triat (Sabri et al., 2020). When genotypes are evaluated across a range of different locations and/or years, their yield performances could differ significantly. The existence of large G × E interaction usually causes serious confounding effects in comparing and recommending good genotypes for wide adaptation (Moussa et al., 2011). Previous G × E studies on several traits have demonstrated that sweet potato is sensitive to environmental changes. According to Madawal et al. (2015), Gurmu et al. (2017), and Ngailo et al. (2019), changes in environmental conditions have been reported to affect sweet potato storage root yield and yield components. This makes the analysis for G × E interaction crucial for genotype selection, cultivar release, and identification of suitable production environments for optimum yield. Therefore, having a basic understanding on G × E interactions, stability parameters, and genetic correlations for root yield and yield components are considered necessary for sweet potato breeders in making an informed choice concerning which locations and input systems should be used in their breeding efforts (Gruneberg et al., 2005).

Statistical tools such as the Additive Main Effect and Multiplicative Interaction (AMMI) (Gauch, 1992) and genotype and genotype-by-environment interaction (GGE) biplot analyses (Yan and Kang, 2003; Yan and Tinker, 2006) have been reported as appropriate for use in GEI analyses. These statistical tools have then been extensively used in several sweet potato improvement programs by authors such as Caliskan et al. (2007) (AMMI model analysis for GEI and stability analysis of sweet potato genotypes across different environments in Turkey) and Laurie and Booyse (2015) (GGE biplots used in and identifying suitable sweet potato genotypes and representative environments in South Africa). The objective of this study, therefore, was to determine the magnitude of GEI for storage root yield, yield-related traits, and sweet potato weevil (Cylas spp.) damage among candidate sweet potato genotypes, as well as to assess the adaptability and stability of 41 improved sweet potato genotypes in two sweet potato representative and contrasting production environments.

The mean squares of the analyses of variance (ANOVA) of the agronomic traits evaluated in two different locations (Abakaliki and Iresi) revealed that there were significant (p < 0.01) variations among the genotypes for all the traits in each location and across both locations (Tables 3–5). The combined analysis (Table 6) also showed that all the agronomic traits varied with respect to genotype, the location, and Genotype-by-Location Interaction. Correlation coefficients for the agronomic traits evaluated in the two locations and 2 years are presented in Table 7. Generally, all the traits except root Cylas spp. severity exhibited a positive and significant (p < 0.01) correlation with root yield. Most of the traits also exhibited significant and positive association among themselves, except root Cylas spp. severity at harvest which did not correlate with all other traits.

Genetic variability is essential for selection (Sarif et al., 2020). Except for Cylas weevil damage severity (in all environments) and weight of unmarketable roots (in some environments), significant variation exists among the genotypes evaluated for all the traits, an occurrence that has also been reported by Vimala et al. (2011), Madawal et al. (2015), and Afuape et al. (2019). In this study, year as a factor exerted a reducing influence on the agronomic traits. Cylas weevil damage severity is largely controlled by the environment as genetic resistance to Cylas weevil is still unavailable. By its epidemiology, Cylas spp. incidence and severity occurs in the dry season when soil moisture level is very low (Stathers et al., 2003). This explains the observed large environmental variance (0.43) compared to the genotypic variance (0.00), and the extremely low broad-sense heritability estimate (0.02) recorded in this study for the trait. The very low genotypic variance (0.00) and heritability estimate (H_B = 0.00) for weight of unmarketable roots (roots <100 g) also portend that environment is largely responsible for the formation of small sweet potato roots. This is true as plant density, drought spell, and poor weed management among other factors could lead to the formation of small storage roots.

The genetic study reveals that environmental variance (Ve) was higher than genotypic variance (Vg), leading to moderate broad sense heritability (H_B) estimates in both locations and in the combined form. As H_B estimate is a reflection of the genetic component of an observed phenotype expression of a trait, the moderate H_B in the combined genotype-location-year interaction analysis depicts that genetic gain could be achieved in such important traits as total root weight (H_B = 0.5), weight of marketable roots (0.52), root yield (0.46), and number of marketable roots (0.34). Moderate heritability is often deemed acceptable for quantitative traits (Tumwegamire et al., 2011) and has been variously reported for most root yield and yield components by Tumwegamire et al. (2011) and Afuape et al. (2019). The observed differences in the H_B estimates for each trait in both locations shows that location affects trait heritability estimates, with poor environment lowering the value by confounding the true genotypic value with non-experiment wise errors which could lead to selection error in the breeding cycle.

Correlation coefficient is a measure of the extent and direction of the relationship between any two traits (variables). The positive and strong relationships between root yield and number of marketable roots (r = 0.786), total number of roots (r = 0.438), marketable roots (r = 0.861), and total root weight (r = 0.861) suggests that these traits are important root yield components, and that their simultaneous selection will be a good approach to increasing root yield. This same relationship had been observed by Afuape et al. (2011) and Yahaya et al. (2015). As the severity of Cylas spp. is a yield-reducing affect, the negative correlation observed between it and root yield and yield components is expected as severely infested fields often exhibit reduced plant density at harvest, severe reduction in root weight, significant reduction in good roots, and impaired plant physiological processes due to severed food translocation organs, as infested plants often have the vine cortex and pith eaten up by the Cylas weevil larvae developing inside the plant.

The stability analysis aims at helping the breeder to identify which genotypes have specific and/or general adaptability to various production environments. It also helps in the analyses of the test environments for prudent decision making for future evaluations. As expected, the two locations (Iresi and Abakaliki) clustered in different quadrants, depicting that both locations are truly different. The Iresi environments (IRE 1 and IRE 2) were higher performing environments for root yield and Cylas damage suppression compared to the Abakaliki environments (ABA 1 and ABA 2). The differences between these locations were so large that year effect could not nullify it, meaning that testing data from one location cannot represent the performance of same materials in the other location, irrespective of the number of years used in the testing process. As different genotypes performed differently in each of these locations, we are able to identify genotypes that are specifically adapted to each environment, knowledge that will help breeders to adequately advise farmers on what cultivar to use where, provided the various cultivars possess acceptable end-user quality preferences. Conducting a stability analyses has thus helped in identifying specific genotypes for both locations, as well as a stable genotype that can be cultivated across all the locations tested and locations that share similar attributes to the test locations.

Dry matter content according to Mok et al. (1997) and Cervantes-Flores et al. (2011) is among the main attributes that further improve the quality and acceptability of sweet potato for consumption and processing in sub-Sahara Africa. The dry matter content ranged from 43.85% (F2M1/35) to 24.07% (F2M6/17) and varied significantly (p < 0.0001) between the sweet potato genotypes. Findings from this study are in accord with records reported by Okunade et al. (2019) in Osun (21.35–42.10%) and Kanu et al. (2018) in Umudike, Abia State (27.30–41.56%), but disagrees with the result (19.4–22.6%) reported by Laurie et al. (2013) for some OFSPs. The low dry matter content recorded in F2M6/17 could be a result of the presence of high moisture in the tuber. Factors such as variety, crop season, age of the plant, and location could affect dry matter content in sweet potato. According to Eleazu and Ironua (2013), high dry matter content improves storability, texture, and product yield. It also has the potential of being utilized for industrial purposes and for flour production in confectioneries. Proteins are essential nutrients for structural and functional performers of different biomolecules in the human body, and they provide the essential amino acids required for metabolism. The protein content recorded in this study varied significantly (p < 0.0001) across the forty sweet potato genotypes. Results from this study revealed that protein was highest in TIS 87/0087/01 (2.67%), followed by UMUSP 2/02 (2.53%), and lowest in SOLO-1/88 (1.54%). This result correlated well with reports credited to Rakesh et al. (2017) in some sweet potato cultivars. Studies show that protein content in sweet potato, especially orange fresh sweet potato (OFSP), is comparable to cassava (1.6%) and yam (1.5) (Woolfe, 2008).

The fat content among the forty sweet potato varieties ranged between 0.17 to 0.31% and showed significant (p < 0.0001) differences among the genotypes. The fat content was high in solo-1/21 (0.31%) while fat was least in the MD/12 genotype with 0.17%. The results of fat content from this study were in agreement with Ishida et al. (2000) and Mohammad et al. (2016), who reported 0.17–0.30% and 0.2–0.33%, respectively. Sweet potato, like other known roots and tubers, is well-known for its low fat content. The fat concentration of some sweet potato, especially the orange fresh sweet potato, are still little better than other roots and tubers such as cassava (0.28%) and yam (0.17%) (USDA, 2018). Crude fiber, which is also an important indicator of a healthy food material that plays a major role in reducing the incidences of colon cancer, diabetes, heart disease, and certain digestive diseases, were observed in an appreciable amount. The crude fiber content of the sweet potato ranged from 0.33% (in SOLO-1/21) to 0.54% (in TIS 87/0087/01) and was significantly different among the genotypes. This result is in line with 0.35% reported by Endrias et al. (2016) for different varieties of OFSP.

Ash is an inorganic residue in any food substance which directly denotes the mineral content. The sweet potato has a reasonable amount of ash content, an indication of rich mineral constituents. The ash value varied from 1.14% (in F2M6/1 and SOLO-1/21) to 1.32% (in F2M1/31). The ash values obtained from this study were in line with the results reported by Mohammad et al. (2016), who recorded 1.17–1.31% ash content. Carbohydrate was among the most abundant nutritive constituent of the sweet potato. This is consistent with previous reports credited to Adepoju and Adejumo (2015), Amha and Baruch (2016), and Endrias et al. (2016). The carbohydrate content obtained from this study ranged from 19.95% in F2M6/17 to 39.75% in F5M1/3 and varied significantly among the genotypes, indicating that they are a good source of energy. This study is within FAO Corporate Document Repository carbohydrate range but appreciably higher than 21 and 25% for nine orange-fleshed sweet potato varieties grown in Bangladesh by Mohammad et al. (2016).

There was high variation among the genotypes with respect to total carotene content. The orange fleshed sweet potato recorded among the genotypes had the highest amount of total carotene content. The amount of total carotenoid content of the roots ranged from 0.18 to 5.55 mg. Highest total carotene content was observed in Mother Delight with 5.55 mg/100 g (fresh weight basis), followed by F2M6/17 (5.31 mg/100 g) which are orange fleshed, and was lower in genotype Ex- Oyunga/17 (0.48 mg/100 g), Solo-1/165 (0.47 mg/100 g), and Solo-1/88 (0.18 mg/100 g). This result agrees with the findings of Adepoju and Adejumo (2015), Mohammad et al. (2016), and Islam et al. (2016) but inconsistent with Donado-Pestana et al. (2012) who reported 0.390 to 8.823 mg/100 g.

The current study determined the magnitude of genotype-by-environment interaction and stability for storage root yield and root Cylas severity. Since most sweet potato breeding programs are often tailored toward the development of high yielding, biotic and abiotic resistance and/or tolerance, this work has identified G14 as a high yielding and Cylas spp. tolerant genotype that is stable for both traits. These combined attributes will make it possible for its deployment across many environments. Genotypes such as G11 and G13 could be cultivated in environments with low Cylas spp. pressure as they are highly productive but exhibit low Cylas spp. tolerance. Genotypes G24, G37, and G31, which had average yield but high and stable tolerance to Cylas damage, could be used as elite germplasm toward the development of new varieties that combine high yield and Cylas tolerance. In general, the knowledge and extent of the genotype-by-environment interaction (GEI) provided by this study would assist breeders in deploying limited resources in the right varietal development cycle. This study suggests that deploying resources into conducting genotype testing in many locations and years would be a worthwhile investment.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.