Abstract
East African banana (Musa sp.) breeding efforts have focused mainly on enhancing ‘Matooke’ productivity through the development of high-yielding, pathogen-resistant cultivars with adequate stability to contribute to regional food security. Before a breeding program can recommend promising cultivars for release, they must pass the sensory screens; be evaluated in the target population environments; and the data analyzed for yield, adaptability, and stability. Twenty-four primary and secondary triploid hybrids [NARITA (N)] derived from ‘Matooke’ bananas, six triploid local ‘Matooke’ cultivars, and one exotic cultivar were evaluated for their yield, adaptability, and stability across the East African region at three highland sites in Uganda’s western and central regions, as well as at three sites in Tanzania’s northeastern and southern highlands regions, from 2016–19. A randomized complete block design with four replicates was used for multisite trials. The mixed-model restricted maximum likelihood/best linear unbiased prediction approach, along with additive main effect multiplicative interaction model biplots, were used to dissect and visualize genotype-by-environment patterns. Following the likelihood ratio test, both genotype and interaction effects were highly significant, confirming the influence of genotype and site heterogeneity for selecting specific and broadly adapted cultivars. N23 had the greatest yield across all sites associated with adaptability and stability, outperforming the overall mean yield of all genotypes by 34.2%. In Tanzania, N27 (second), N7 (third), N18 (fourth), N4 (fifth), N12 (sixth), and N13 (seventh); and in Uganda, N17 (second), N18 (third), N2 (fourth), N8 (fifth), N13 (sixth), N12 (seventh), N4 (eighth), and N24 (ninth) demonstrated good adaptability and stability, as well as high yield. Furthermore, the fungal pathogen Pseudocercospora fijiensis had no significant effect (P > 0.05) on yield, stability, and adaptability of the hybrids. As a result, they can be introduced into areas where black leaf streak constrains banana production significantly and threatens farmers’ livelihoods. The average site yield potential ranged from 9.7 to 24.3 t⋅ha–1 per year. The best discriminating sites for testing breeding clones were Lyamungo in Tanzania and Sendusu in Uganda. Hence, these testing sites are recommended as ideal examples of locations for selecting superior genotypes.
Bananas (Musa sp.) are among the 10 most important food crops worldwide. They include dessert, beer, and cooking cultivars (Food and Agriculture Organization of the United Nations 2014; Ortiz and Swennen 2014). They are popular in more than 150 nations because of their year-round production and high demand (Uma et al. 2011). With a global production of 145 million tonnes, worth $26.3 billion, the crop feeds millions of people worldwide (Brown et al. 2017; Lescot 2018). Most banana and plantain cultivars grown are intraspecific or interspecific triploid (2n = 3× = 33) hybrids derived from the diploid (2n = 2× = 22) Musa acuminata and Musa balbisiana, respectively (Ortiz and Vuylsteke 1994b). Wild species are diploid, whereas cultivars are diploid, triploid, and tetraploid (2n = 4× = 44) after natural or artificial hybridization (Robinson 1996).
The AAA triploids of the ‘Mutika’ subgroup, originally named ‘Mutika-Lujugira’ by Shepherd (1957), and often referred to as East African Highland bananas (EAHBs), are the edible derivatives of the wild species M. acuminata ssp. zebrina and ssp. banksii (Karamura and Pickersgill 1999; Kitavi et al. 2016; Li et al. 2013). They are farmer-selected cultivars that dominate the East African Great Lakes region (Karamura 1998; Perrier et al. 2019; Pillay et al. 2001). Their fruit provide between 3% to 22% of daily caloric intake, estimated to be 147 kilocal/person, and generate more than $4.3 billion per year, or roughly 5% of the region’s gross domestic product (Food and Agriculture Organization of the United Nations 2014; Kalyebara et al. 2007; Tinzaara et al. 2018).
Primary and secondary triploid EAHB hybrids, as well as their parental landraces and exotic cooking bananas, developed by Uganda’s National Agriculture Research Organization and the International Institute of Tropical Agriculture, have been evaluated across Uganda’s and Tanzania’s diverse agroecozones (Tushemereirwe et al. 2015). The goal of this set of multilocational trials was to find genotypes with resistance to the leaf spot disease black leaf streak (BLS) caused by the airborne fungal pathogen Pseudocercospora fijiensis, as well as genotypes that demonstrate a consistent high yield and other desirable characteristics for farmers and consumers. The impetus was a significant drop in EAHB productivity resulting from several biotic constraints including BLS (Ortiz and Vuylsteke 1994a; Swennen and Vuylsteke 1993; Swennen et al. 1989, 2013; Tushemereirwe et al. 2015; Vuylsteke et al. 1993). As a result, female fertile triploid EAHB cultivars were crossed with a BLS-resistant male wild diploid banana [‘Calcutta 4’, AA (Tushemereirwe et al. 2015)]. The hybrid progeny produced ranged in ploidy level, with the vast majority being tetraploids. Because these primary tetraploids were more fertile than their triploid parents, they were crossed with improved diploids to produce the BLS-resistant triploid hybrids known as NARITA (N) (Batte 2019; Tushemereirwe et al. 2015).
In any crop, researchers and farmers aspire for more stable and high-yielding cultivars. Similarly, for banana, a breeder generally desires to develop a cultivar that thrives adequately in different environments. As a result, targeting cultivar selection onto its growing environment is the prime interest of any plant breeding program and a prerequisite for the recommendation of novel selections for large-scale production (Annicchiarico 2002). To achieve these goals, breeding programs usually undertake a rigorous evaluation of the performance of a set of diverse genotypes across locations and over years, mostly during the final stage of the cultivar development process. Multi-environment trials (METs) allow for the assessment of genotypes’ relative performance and stability for yield and yield-related traits (Annicchiarico 2002; Kang 2004; Vaezi et al. 2019).
Yield is a complex trait that is influenced by genotype, environment (E), and genotype-by-environment interactions (GEIs). For breeders, the GEIs effect is important because it reflects yield variation not explained by individual genotypic and environmental effects (Ebdon and Gauch 2002; Yan and Hunt 2001). Although GEIs cause inconsistency in performance across environments and complicate cultivar selection, they can provide useful information to breeders (Busey 1983; Kang 1998; Magari and Kang 1993). For example, they justify the need for additional wide-based testing in different environments and predict the variability expected among testing sites (Busey 1983). The heritability and phenotypic expression of yield and other quantitative traits also vary as a result of genotypic differences, environmental influences, and GEIs (Bradshaw 1965; Crossa et al. 1990). The magnitudes of these variations are important when designing a breeding strategy and improving selection responses. Several numerical and graphical stability analyses are available that determine GEIs to recommend better performing and higher yielding genotypes across different environments (Ortiz and Ekanayake 2000).
Mixed models’ restricted maximum likelihood and best linear unbiased prediction (BLUP) have been shown to be effective in evaluating genotypic performance and stability (Henderson 1975; Patterson and Thompson 1971). They allow for more accurate and reliable estimation of genetic and environmental parameters, as well as nonbiased genotypic value prediction (Searle et al. 1992; Smith et al. 2005). The analyses treat genetic values as random effects, resulting in a more accurate prediction of the candidates’ genetic value, which is both unbiased and has a low prediction error variance (Henderson 1985; Piepho et al. 2008; Robinson 1991). Furthermore, mixed-model approaches reduce the noise caused by unbalanced designs and non-additive traits, both of which are common problems with MET data (Hu 2015; Piepho 1994). Predicted genotypic values, on the other hand, can be used to calculate the harmonic mean of relative performance of genotypic values (HMRPGV). This method has been used to evaluate the adaptability and genotypic stability of crops such as winter oilseed rape [Brassica napus (Bocianowski and Liersch 2021)], sugarcane [Saccharum officinarum (Bajpai and Kumar 2005)], and wheat [Triticum aestivum (Mohammadi and Amri 2008)]. It allows for simultaneous selection of stability, adaptability, and mean performance, which are expressed as a unique value that can be multiplied by the general mean (μ) to produce genotypic values for each genotype (HMRPGV × μ) that are penalized for instability and capitalized for GEIs.
The objectives of our research were to identify high-yielding banana genotypes and estimate variance components as well as broad sense heritability for yield, and to select cultivars with specific and wide adaptation potential across the East African region. Our results should assist banana breeders in East Africa and other similar environments in planning large-scale evaluation trials of promising cultivars or breeding lines before their official release to target environments.
Materials and Methods
Twenty-four ‘Matooke’ primary and secondary triploid NARITA hybrids, six ‘Matooke’ triploid local cultivars, and one exotic cultivar were evaluated for yield and other related traits across six sites in Uganda’s western and central regions, as well as in Tanzania’s northeastern and southern highlands regions (namely, Kilimanjaro, Kagera, and Mbeya) spanning a 3-year period (2016–19). The selected areas are the main banana-producing zones in both countries and were Kawanda, Mbarara, and Sendusu in Uganda; and Maruku, Mitalula, and Lyamungo in Tanzania. Tables 1 and 2 provide detailed information on the genotypes tested, the six sites, their rainfall, and other environmental characteristics. Twelve plants of each genotype were raised in each block in a randomized complete block design with four replications. Tanzanian and Ugandan sites were planted in April and May 2016, respectively. Farmers’ site-specific landraces, as well as the widely grown ‘Mbwazirume’ (used as a common local cultivar check), were planted alongside a reference banana Cavendish ‘Williams’ (Musa AAA), which are all BLS-susceptible cultivars. The selected local checks are a good representation of what farmers are currently growing.
Code, name, and origin of 30 banana genotypes evaluated for yield potential and stability in six Tanzanian and Ugandan sites between 2016 and 2019, as well as their use, cultivar type, and ploidy level.
Description of agroclimatic characteristics (altitude, rainfall, temperature, soil type, and sites’ global position), site mean yield potential (YLD), and broad sense heritability (H2) of six testing sites in Tanzania and Uganda used to evaluate 30 banana genotypes for yield potential and stability.
The plants were spaced 3 m apart, yielding a plant density of 1152 plants/ha. The planting hole was 100 cm in diameter. Some plants died after planting because of a variety of factors, including drought, and were replaced with suckers from surviving mats of the same cultivar in the trial. To reduce competition for food and water, three plants were kept per mat (i.e., mother plant, daughter, and granddaughter). Farmyard manure was placed in the holes before planting at a rate of 10 kg/plant. Weeding was done every 2 to 3 months. Dead leaves were removed on a regular basis. Mulching was performed at the start of each dry season in the Ugandan sites and in Maruku, whereas furrow and basket irrigation were used in Lyamungo and Mitalula, respectively. Staking was done to keep the plants upright. The other trial management practices were consistent with good crop husbandry practices undertaken by farmers. Similar crop husbandry practices were used at Sendusu, where plant density was 1667 plants/ha. When at least one fruit finger on a bunch began to ripen, the bunch was harvested, and the bunch weight was measured in kilograms per plant.
Variance components and genetic parameters.
This methodology is an optimal procedure for unbalanced data.
Stability and adaptability analysis.
We used the singular value decomposition of the matrix of BLUPs for the GEIs effects generated by a linear mixed model to evaluate genotypic stability by additive main effects multiplicative interaction (AMMI) biplots and visualized the relationships among selection sites and the performance of candidate genotypes (Gauch 2013; Gauch and Zobel 1988; Olivoto et al. 2019a). The biplots were generated by subjecting a BLUP-based mixed model’s shrunken GEIs effects matrix to an AMMI-like analysis using the singular value decomposition method. The interaction principal components were obtained by fitting the singular value decomposition to the double centered BLUP interaction effects matrix produced from a linear mixed model with symmetric singular value partitioning (α = 1/2).
Results and Discussion
One of the first decisions farmers must make is which cultivar to grow in a field based on its anticipated economic and social benefits, which are typically defined in terms of the greatest yield potential and performance stability. This critical decision determines how long their banana production can be sustained. However, determining the best cultivars across a diverse set of environments exposed to intricate biotic and abiotic patterns and interactions that frequently result in significant variations in cultivar rank is far from trivial. As a result, one of the primary goals of plant breeding programs is to identify the ability of advanced bred germplasm to adapt to different agroecological settings.
Combined ANOVA.
Table 3 gives the results of the combined ANOVA for a yield of 30 banana genotypes studied across six locations in Tanzania and Uganda. Genotype and environment, as well as their GEIs, were statistically significant (P < 0.001). The environment effect accounted for 41.7% of variation in yield, whereas the genotype and GEIs effects accounted for 28.7% and 11.2%, respectively. The significance of GEIs highlighted the importance of studying phenotypic stability by revealing differences in genotypic responses to agroecological differences in years and locations. This result suggests that some genotypes or groups of genotypes have specific adaptation to sites, whereas others may show broad adaptation, thereby confirming the importance of multilocational testing of cultivars before release. METs are critical for identifying cultivars that perform consistently year after year (with little temporal variation), as well as cultivars that perform consistently from location to location (small spatial variability). Farmers value and benefit from temporal stability, whereas breeders and seed producers value and benefit from spatial stability (Crossa et al. 1990; Kang 1990; Kang and Gauch 1996).
Combined analysis of variance of 30 banana genotypes evaluated for yield potential and stability across six sites in Tanzania and Uganda.
The mean site yield potential varied greatly, ranging from 9.7 t⋅ha–1 per year in Maruku, Tanzania, to 24.3 t⋅ha–1 per year in Sendusu, Uganda (Table 2). The wide range of yield potential confirmed, among other factors, the impact of different environments on genotype performance. Farmers in the developing world often have limited inputs and grow bananas in harsh and unpredictable environments, so a diverse set of conditions is required to conduct an accurate evaluation of yield stability.
Estimates of heritability, variance components, and genetic parameters.
Estimates of the degree of phenotypic variation and heritability of yield must be reliable and accurate to optimize banana breeding selection efficiency. The components of hybrid phenotypic variation, as well as trait heritability and other important genetic parameters, are listed in Table 4. The likelihood ratio test revealed highly significant effects (P < 0.001) for both genotype and GEI effects (Table 4). Genotypic variation accounted for 40.1% of the phenotypic variation in yield across sites, whereas GEIs accounted for 7.8%. In Tanzania, genotypic variation accounted for 39.2% of phenotypic yield variation, whereas GEI variation accounted for 2.2%. Genotypic variation accounted for 33% of total phenotypic variation in Uganda, with GEIs accounting for a greater proportion (22.2%). The residual variance represented 58.6% of the phenotypic yield variation in Tanzania, 44.9% in Uganda, and 51.8% across sites.
Likelihood ratio test, estimated variance components, and genetic parameters for yield potential (tonnes/hectare/year) of 30 banana genotypes evaluated for yield and stability across six sites in Tanzania and Uganda.
Genetic variation provides the grounds for selection in banana breeding. The median genotypic variance obtained in our study, as well as the considerable residual variation, underlined the complexity of the genetic architecture of yield in banana, resulting from its multigenic inheritance, and phenotypic plasticity. As a consequence, quantitative traits are more vulnerable than qualitative traits to alteration by the variation in environmental conditions to which plants in the population are subjected (Acquaah 2012). Tenkouano (2001) reported that the multiploidy and heterogenomic structure of breeding populations result in unpredictable variation in genome size and structure across and within generations. Usually this complicates phenotypic selection for most yield and growth-related traits (Ortiz and Vuylsteke 1996). Breeders would gain in efficiency if they could assign segregating offspring to ploidy and genome classes putatively predictive of their prospective use before field evaluation (Tenkouano 2001).
The heritability of a trait broadly expresses the proportion of phenotypic variance within a population that can be attributed to heritable genetic factors. Its estimation is critical because it shows how much of a trait is genetically based and allows the best improvement approach to maximize the selection response (Falconer and Mackay 1996). The estimate of broad sense heritability was 0.33 for Uganda, 0.39 for Tanzania, and 0.4 across all sites, implying that genetic differences accounted for 40% of the variance in mean yield observed among the 30 genotypes tested (Table 4). The heritability of the genotypic mean, commonly estimated when means are used as selection criteria, was 0.87 and 0.74 for Tanzania and Uganda, respectively, whereas 0.92 was achieved across all sites.
The development of new banana hybrid cultivars should optimize breeding initiatives that strike a balance between population size and selection intensity (Xu et al. 2017). Improved estimation of trait heritability remains critical for increasing the rate of genetic gain. Because yield is a multigenic trait with continuous phenotypic variation (Ortiz and Tenkouano 2011), analyzing the effects of the underlying genes and determining their inheritance mode is difficult. As a result, their estimated heritability is low, necessitating indirect selection based on highly heritable component traits to aid in the accumulation of desirable genes. According to Batte et al. (2021), the main traits contributing to bunch weight (which is a proxy for edible yield) in EAHBs are fruit length, number of fruit in a bunch, fruit circumference, and number of hands in a bunch. As a result, when selecting parents for use in breeding for yield in EAHBs, these traits should be evaluated to guarantee they are passed on to the new hybrids targeted for release to farmers.
Experimental designs could contribute to improving heritability estimates. Increasing the number of replications and locations has been reported to contribute toward an increase in heritability estimates (Schmidt 2019; Xu et al. 2017). Because the heritability estimates do not respond linearly to an increase in replications, increasing the number of target locations for evaluations is considered a better option to increase heritability estimates (Cobb et al. 2019; Weikai 2014). Usually, this results in additional costs. However, with additional testing environments, a breeder–agronomist can identify cultivars with specific adaptation as well as those with a broad adaptation, which would not be possible from testing in a single environment. In this context, the adoption of partially replicated trials or nonreplicated designs may be beneficial in cases in which spatial adjustments can be done properly (Cullis et al. 2006; Schmidt 2019; Williams et al. 2011).
Ssali et al. (2016) and Ortiz (1997a) reported broad sense heritability of bunch weight of secondary triploid banana ‘Matooke’ (Musa sp., AAA-EA) in Uganda and Musa germplasm in Nigeria to be 47.8% and 66%, respectively. Batte (2019) observed a high heritability of 84% for EAHB yield and 76% for bunch weight; however, his study, although using multigeneration trials, was conducted in a single site in Uganda (Sendusu), so the estimated heritability could be overestimated because of a lack of GEIs—in other words, across site variation. This is further supported by the investigation of broad sense heritability for individual sites in our study, the values of which are more than 60% because single-site heritability estimates do not account for GEIs (Table 2).
In addition to genetic gain, recent emphasis in plant breeding has also been on the genotypes with premium value and quality to satisfy consumer preferences (Akankwasa et al. 2020; Thiele et al. 2021). For example, superior banana cultivars should achieve genetic gain linked with fruit quality attributes and sensory perception (Cobb et al. 2019; Nowakunda and Tushemereirwe 2004). The concept of genetic gain may also be extended to cover the gain farmers can achieve in their income with unit cost or input—a trait that is also linked to the environment in banana (Meya 2021). Decentralized selection has been conceptualized more systematically during the past two decades, with the goal of increasing selection gains for marginal, low-input farming systems. Ceccarelli (1996) concluded that cultivar selection and testing for marginal production circumstances and resource-limited farmers should be conducted more intensively in farmers’ field target environments. By determining the optimal genotypes for each target environment, it is possible to exploit favorably the interaction between plant populations and specific environmental conditions (i.e., GEI).
The genotypic correlation of genotype performance across sites was 0.2, thereby indicating the presence of GEIs that change genotype ranking across environments (Table 4). The result emphasizes the importance of assessing genotype adaptability and stability to provide accurate recommendations to farmers and breeders in various target regions (Yan and Tinker 2006). Inconsistency in genotype performance across locations or years provides additional information for breeders and suggests that, along with justifying the need for more broad-based testing in different environments, the degree of inconsistency could help predict the variability expected among different fields (Busey 1983). The genotypic variation coefficients across locations were greater than 10%, indicating the presence of genetic variability and the possibility of effective clonal selection (Table 4). Ssali et al. (2016) obtained similar results using the same method and mixed models for 11 secondary triploid banana ‘Matooke’ hybrids.
The high cross-site selection accuracy (0.9) indicates that the experimental design was effective in reducing potentially disruptive effects. It also shows that the predicted and true genotypic values are highly correlated, implying a high precision in the identification and the possibility of success in the selection of individuals with specific or broad adaptation. Resende and Duarte (2007) recommended accuracy values greater than 0.7 for intermediate stages of the breeding program and greater than 0.9 for cultivar recommendations. A medium to high CV was observed (CVe = 25.5), but CVe estimates alone cannot judge experimental quality. Instead, the CVr (= CVg/CVe) must be estimated, with magnitudes close to or greater than one being preferred (Olivoto et al. 2017). Yield had CVr ≥ 0.9 across sites, thus indicating the possibility of achieving selection gain.
Genotype yield, stability, and adaptability.
Breeding programs must test hybrids in target environments and analyze data for yield, adaptability, and stability to develop cultivars that are well adapted to growing regions. Tables 5, 6, and 7 summarize the three BLUP-based indices HMGV, RPGV, and HMRPGV for selecting genotypes with high mean performance, stability, and adaptability. Coincidence in genotype ranking was observed for all indices, indicating the possibility of making reliable genetic value predictions using a single selection criterion that encompasses yield, stability, and adaptability. N23 had the greatest yield associated with adaptability and stability across all sites (Table 5). This hybrid outperformed the overall mean of all genotypes tested in Tanzania and Uganda by 34.2%, resulting in an HMRPGV × μ value of 22.5 t⋅ha–1 per year. Despite producing the greatest yield (37.8 t⋅ha–1) during 3 years of advanced yield trials in Uganda, this hybrid was not advanced for release as a new cultivar (Tushemereirwe et al. 2015). It was instead reserved for multilocation participatory trials in Tanzania and Uganda to find clones that combine BLS resistance with stable high yield and other desirable quality traits by farmers (Kubiriba et al. 2016; Lorenzen et al. 2010; Tushemereirwe et al. 2015). N23 was developed through a series of interploidy crosses between the female fertile EAHB ‘Kazirakwe’ and the ‘Matooke’ improved diploid ‘7197-2’. As a result, its exceptional performance is most likely a result of its parent ‘Kazirakwe’, which is known for its high yield and adaptability.
Best linear unbiased predictions (BLUPs) for yield potential [YLD (tonnes/hectare/year)] of 30 banana genotypes evaluated for yield and stability in six Tanzanian and Ugandan sites, along with the stability of their genotypic values [harmonic mean of the genotypic values (HMGV)], adaptability of genotypic values [relative performance of the genotypic values (RPGV)], genotypic values capitalized by the interaction (RPGV × μ), stability and adaptability of genotypic values [harmonic mean of relative performance of genotypic values (HMRPGV)], and mean genotypic values (HMRPGV × μ).
Best linear unbiased predictions (BLUPs) for yield potential [YLD (tonnes/hectare/year)] of 24 banana genotypes evaluated for yield and stability in three Tanzania sites, along with the stability of their genotypic values [harmonic mean of the genotypic values (HMGV)], adaptability of genotypic values [relative performance of the genotypic values (RPGV)], genotypic values capitalized by the interaction (RPGV × μT), stability and adaptability of genotypic values [harmonic mean of relative performance of genotypic values (HMRPGV)], and mean genotypic values (HMRPGV × μT).
Best linear unbiased predictions (BLUPs) for yield potential [YLD (tonnes/hectare/year)] of 22 banana genotypes evaluated for yield and stability in three Uganda sites, along with the stability of their genotypic values [harmonic mean of the genotypic values (HMGV)], adaptability of genotypic values [relative performance of the genotypic values (RPGV)], genotypic values capitalized by the interaction (RPGV × μg), stability and adaptability of genotypic values [harmonic mean of relative performance of genotypic values (HMRPGV)], and mean genotypic values (HMRPGV × μg).
The hybrid N17 was ranked second overall, and similarly in Uganda, outperforming the overall average yield by 29.5% (Tables 5 and 7). It was also preferred by Ugandan farmers for its culinary qualities, and it won first place in sensory testing (data not shown). As a result, it is a candidate for release in Uganda. N27, N18, N13, and N4 were ranked third, fourth, fifth, and sixth across all sites, respectively, in terms of yield, stability, and adaptability, implying they are also the most stable and adapted. In Tanzania, the top five for yield, stability, and adaptability were N23, N27, N7, N18, and N4, whereas in Uganda, the top five were N23, N17, N18, N2, and N8. N23 and N18 genotypes are thus shared among the top five in both countries, whereas the rest are specific to Tanzania and Uganda. Others that performed well but did not make the top five were N12, N13, N8, N9, and N25 in Tanzania, and N13, N12, N4, N24, and N11 in Uganda. N27 was denied for advancement in Tanzania because of a lack of sensory characteristics, whereas N24 was among the top performers in Uganda in terms of sensory characteristics (Marimo et al. 2020). N8 and N13, on the other hand, performed well, but because they are of a juice banana type, they cannot be recommended to farmers for food use. However, one juice banana in Tanzania (N13) and two in Uganda (N13 and N8) tended to be among the best for yield, stability, and adaptability, thus indicating an additional source of income for farmers as a means of sustaining their livelihood.
Four of the top 10 genotypes have been released in the past 6 years: the N23, N18, and N4 genotypes in Tanzania, and the N7 genotype in both Tanzania and Uganda. These cultivar releases illustrate the current success that is being experienced by banana breeding in East Africa. ‘Mbwazirume’, a comparison local check, was in 23rd place out of 30 genotypes tested across sites, 20th out of 24 genotypes tested in Tanzania, and 16th out of 22 genotypes tested in Uganda. Despite being one of the best EAHBs in terms of farmers’ preferred sensory attributes, the poor yield performance of ‘Mbwazirume’ is unsurprising, as noted in previous research results (Erima et al. 2016; Ssali et al. 2010). Not only did the local checks perform poorly, also several hybrids fared poorly and were ranked among the last. N15 and N19, for example, were the two genotypes that ranked among the last across all sites. N15 and N19 were the last two in Tanzania, whereas N16 and ‘Kisansa’ were the last two in Uganda. These findings suggest that high-yielding, stable genotypes are not always the result of banana crossbreeding.
Relationship between yield and BLS resistance.
A variety of pathogens and pests damage crop plants, resulting in significant yield loss. Resistance has been defined as the “inherent capacity of a plant to prevent or restrict the entry or subsequent activities of a pathogenic agent when the plant is exposed, under suitable environmental conditions, to sufficient inoculum of a pathogen to cause disease” (Bhargava and Srivastava 2019). In addition, any resistance breeding effort attempts to develop superior high-yielding genotypes that are resistant for a long time (Craenen and Ortiz 2003; Tushemereirwe 1996). The regression analysis results revealed that INSL had no significant effect (P > 0.05) on hybrid yield, stability, and adaptability (Table 8). Indeed, INSL accounts for only 0.43% yield variance across sites, 0.07% in Tanzania, and 3.36% in Uganda.
Regression analysis for yield potential (tonnes/hectare/year) of 30 banana genotypes evaluated for yield and stability in six Tanzanian and Ugandan sites, with 24 genotypes evaluated in three Tanzanian sites and 22 genotypes evaluated in three Ugandan sites in relation to host plant resistance to the black leaf streak pathogen measured by the percentage index of nonspotted leaves.
Genotype, environment, and their interaction effects on yield.
The ability of a plant breeding program to provide farmers with genotypes with guaranteed superior performance for yield or quality across a range of environments is critical to its success. Understanding the factors that lead to a good phenotype is necessary to achieve this goal (Malosetti et al. 2013). The genotypes with high yield potential and stability, as well as the testing environment relationships were visualized using AMMI biplots. Principal component 1 (PC1) and principal component 2 (PC2) accounted for 100% of the variation in yield in Tanzania and 100% of the variation in Uganda (Fig. 1A–D). In Tanzania, PC1 explained 76.56% of the genotype and GEI (GGE) variance in yield, whereas PC2 explained 23.44%. PC1 accounted for 88% of the GGE variance for yield in Uganda, whereas PC2 accounted for 11%. Given that a model’s variability must be at least 70% to be deemed reasonably reliable (Gauch 2013), the combined variability of PC1 and PC2 was adequate. The AMMI1 model, with yield on the abscissa and PC1 scores for genotypes and environments on the ordinate, is depicted in Fig. 1A and C. The larger the IPCA scores, either negative or positive, the more specifically adapted a genotype is to a certain environment; the smaller the IPCA scores, the more stable the genotype is over all environments investigated (Crossa et al. 1990; Gauch 2006).
Biplots showing principal components (PCs) for 24, 22, and 30 banana genotypes (Gen) after their testing in six sites (Env) in Tanzania (A, B), Uganda (C, D), and across countries (E, F), respectively, evaluated for yield potential and stability. The scores were calculated by fitting the singular value decomposition of the double-centered best linear unbiased prediction interaction effects matrix derived from a linear mixed model with symmetric singular value decomposition (α = 12). The axes are equally scaled. Cultivar codes are given in Table 1.
Citation: Journal of the American Society for Horticultural Science 147, 6; 10.21273/JASHS05246-22
The best genotypes for wide adaptation are those with a high mean yield and stability. They are close to the biplot’s center and above the grand mean of main effect yield. As banana breeders are frequently drawn to genotypes that are high yielding and relatively more stable, N2, N4, N8, N12, N13, N18, N25, and N27 in Tanzania, and N2, N4, N8, N12, N18, N13, N24, and ‘Mpologama’ in Uganda were deemed the best. Stable genotypes, according to Yan and Kang (2003), ensure consistent yields with little variation year after year. With average stability, the top-yielding genotypes were N7 and N23 in Tanzania, and N17 and N23 in Uganda. These hybrids are also advantageous because they are more closely related to stable genotypes, and breeders routinely select genotypes with a high mean yield and moderate stability that perform well in specific environments for specific adaptation. Denis and Gower (1996) suggested that plant breeders should consider GEI to avoid missing a cultivar whose average performance was poor but performed well when grown in specific environments or selecting a cultivar whose average performance was good but performed poorly when grown in a specific environment. N19, N15, and ‘Ndizi Uganda’ in Tanzania, as well as ‘Kisansa’, ‘Nakitembe’, and ‘Mbwazirume’ in Uganda, had the lowest yield and stability. These results demonstrate that most farmers’ cultivars used as checks had low and unstable yield, which may be attributed to their limited ability to withstand unpredictable climatic conditions combined with greater pathogen impacts. The biplot also revealed that the greatest yielding sites were Lyamungo in Tanzania and Sendusu in Uganda, whereas Maruku in Tanzania and Kawanda in Uganda were the lowest yielding sites (Fig. 1A and C).
Mitalula in Tanzania, as well as Kawanda and Mbarara in Uganda, contributed little to GEIs because of poor genotype discrimination ability (Fig. 1A and C). The vectors were significantly longer at the Tanzanian sites of Maruku and Lyamungo, as well as Sendusu in Uganda, thereby contributing significantly to the GEI. As a result, they provided the ideal environment for cultivar genetic differentiation. Frutos et al. (2014) emphasized that the ideal test environments are both discriminating (informative) and representative of the target environment, whereas Ortiz and de Cauwer (1998) suggested that the ideal environment for breeders’ selections would be one that maximizes phenotypic differences among genotypes—in other words, one in which breeders can do effective visual selection. The ability to discriminate among genotypes for yield performance was highly correlated at the Ugandan sites of Mbarara and Kawanda, particularly for N7. Sendusu forms an obtuse angle with Kawanda and Mbarara, thereby indicating that it is distinct from the others and may have influenced its high yield. N7 and N21 were adapted to Kawanda and Mbarara, respectively, whereas N17 and N23 were adapted solely to Sendusu. The vectors for Lyamungo and Maruku had a maximum angle of less than 90°. This suggests that genotypes are classified similarly at these two sites. N23 has been adapted to both Maruku and Lyamungo, whereas N4 has been adapted to Lyamungo, and N12 and N18 have been adapted to Maruku. Mitalula and Lyamungo form an obtuse angle, thus indicating that these two sites are distinct. N6, N15, and N19 were the best adapted to Mitalula (Fig. 1A and C).
The level of adaptation of the hybrids and local cultivars, as well as the effects of different environments on their yield, are shown in Fig. 1B and D. The biplots reveal that most of them were relatively close to the mean (stable), with the exception of the top-yielding hybrids, which retained their average stability and adaptability to a specific location. In Tanzania, for example, N23 continues to be adapted to the Lyamungo and Maruku sites, whereas N4 and ‘Mbwazirume’ are only adapted to the Lyamungo site. Most of the low-yielding hybrids, such as N6, N11, and N20, were adapted to Mitalula or Maruku, thereby confirming the findings revealed by Fig. 1A and C (i.e., these are low-yielding sites in Tanzania). In Uganda, N17 as well as N10 and N12 were the most unstable, contributing significantly to GEIs. N17 was adapted for the high-yielding site in Sendusu, whereas N10 and N12 were adapted to Kawanda and Mbarara, respectively. Similar to Tanzania, most genotypes in Uganda are near the center, thereby indicating that they are stable.
The biplots from six Tanzanian and Ugandan sites are shown in Fig. 1E and F. For the first two IPCAs, the cumulative variance was 88.5%. PC1 was responsible for 81.9% of the GGE variance in yield, whereas PC2 was responsible for 6.7% (Fig. 1E and F). The results of the biplots across sites confirm the findings of the individual country analysis, indicating that high-yielding genotypes such as N23 and N17 remain suited to the Lyamungo and Sendusu sites, respectively (Fig. 1E). The majority of genotypes that were stable in the individual country analysis remained stable, with N12, N8, N4, N18, N24, and ‘Mpologoma’ retaining a high level of stability (Fig. 1E). The genotypes with lower yields, such as N19, N15, ‘Ndizi Uganda’, ‘Enyoya’, ‘Kisansa’, N16, ‘Nakitembe’, and N14, have held their position as low-yielding genotypes in the cross-site analysis. Similarly, Mitalula, Maruku, and Kawanda have remained low-yielding sites. Figure 1F reveals the level of adaptation across sites, with N17 adapting to Sendusu in Uganda, N23 adapting to Mbarara and Sendusu in Uganda, and ‘Mbwazirume’, N21, and N2 adapting to Lyamungo in Tanzania. Most genotypes, including N18, N4, and N13, were near to the biplot origin and above the grand mean, thus indicating high yield and stability (Fig. 1E). Sendusu and Mbarara in Uganda, as well as Lyamungo in Tanzania, provide an optimal setting for cultivar genetic differentiation (i.e., discrimination ability) (Fig. 1F).
The hybrids N23, N7, N4, N27, and N18 in Tanzania, and N18, N4, N12, N24, N17, N2, and N23 in Uganda are recommended for cultivar release and ‘Matooke’ banana production in the target population of environments in the East African region from where the testing sites were drawn. These hybrids combine high yield, stability, and adaptability. The three BLUP-based indices ranked these hybrids similarly, thus confirming their unique performance. These hybrids have high host BLS resistance, as indicated by the nonsignificant effect of INSL scores on yield, stability, and adaptability. As a result, they are reliable to be introduced into areas where BLS has a severe impact and threatens farmers’ livelihoods.
Lyamungo in Tanzania and Sendusu in Uganda provide the greatest mean productivity combined with good discrimination ability, making them ideal for future breeding evaluation. Furthermore, the six sites were found to be diverse, and their clustering suggests individual groups that could be used as separate zones for cultivar evaluation and regional cultivar deployment. The findings of this study also reveal that, after more than 20 years of breeding, a reasonable genetic progress for yield trait was achieved. However, enhancing genetic gains in banana breeding programs and its realization in farmers’ fields calls for an integration of multiple aspects including germplasm resources, genomics, breeding, and agronomic practices together with improved seed delivery systems. Cultivar evaluation in the presence of unpredictable GEIs is a persistent problem in banana breeding. There appears to be no easier way to select superior banana cultivars than to test widely and select for both average yield and stability.
References
Acquaah, G. 2012 Principles of plant genetics and breeding 2nd ed Blackwell Publishing Oxford, UK
Akankwasa, K., Marimo, P., Tumuhimbise, R., Asasira, M., Khakasa, E., Mpirirwe, I., Kleih, U., Forsythe, L., Fliedel, G., Dufour, D. & Nowakunda, K. 2020 The East African highland cooking bananas ‘Matooke’ preferences of farmers and traders: Implications for variety development Int. J. Food Sci. Technol. 56 3 1 11 https://doi.org/10.1111/ijfs.14813
Annicchiarico, P. 2002 Genotype × environment interaction: Challenges and opportunities for plant breeding and cultivar recommendations Plant Production and Protection Paper No. 174 Food and Agriculture Organization of the United Nations Rome, Italy
Bajpai, P.K. & Kumar, R. 2005 Comparison of methods for studying genotype × environment interaction in sugarcane Sugar Tech 7 4 129 135 https://doi.org/10.1007/BF02950597
Batte, M. 2019 Increasing efficiency of the breeding pipeline for East African highland bananas PhD Diss Swedish University of Agricultural Sciences Alnarp, Sweden
Batte, M., Swennen, R., Uwimana, B., Akech, V., Brown, A., Geleta, M. & Ortiz, R. 2021 Traits that define yield and genetic gain in East African highland banana breeding Euphytica 217 193 1 10 https://doi.org/10.1007/s10681-021-02920-x
Bhargava, A. & Srivastava, S. 2019 Participatory plant breeding: Concept and applications 1st ed 39 Springer Nature Singapore Pte Ltd Gateway East, Singapore https://doi.org/10.1007/978-981-13-7119-6
Bocianowski, J. & Liersch, A. 2021 Multi-environmental evaluation of winter oilseed rape genotypic performance using mixed models Euphytica 217 80 1 9 https://doi.org/10.1007/s10681-020-02760-1
Bradshaw, A.D. 1965 Evolutionary significance of phenotypic plasticity in plants Adv. Genet. 13 115 155 https://doi.org/10.1016/S0065-2660(08)60048-6
Brown, A., Tumuhimbise, R., Amah, D., Uwimana, B., Nyine, M., Mduma, H., Talengera, D., Karamura, D., Kubiriba, J. & Swennen, R. 2017 Bananas and plantains (Musa sp.) 219 240 Campos, H. & Caligari, P.D. Genetic improvement of tropical crops. Springer Basel, Switzerland https://doi.org/10.1007/978-3-319-59819-2_7
Busey, P. 1983 Management of crop breeding 31 54 Wood, D.R. Crop breeding. Crop Science Society of America Madison, WI, USA
Carlier, J., De Waele, D. & Escalant, J. 2003 Global evaluation of Musa germplasm for resistance to fusarium wilt, Mycosphaerella leaf spot diseases and nematodes: Performance evaluation technical guidelines Bioversity International Rome, Italy
Ceccarelli, S. 1996 Adaptation to low/high input cultivation Euphytica 92 203 214 https://doi.org/10.1007/BF00022846
Cobb, J.N., Juma, R.U., Biswas, P.S., Arbelaez, J.D., Rutkoski, J., Atlin, G., Hagen, T., Quinn, M. & Ng, E.H. 2019 Enhancing the rate of genetic gain in public-sector plant breeding programs: Lessons from the breeder’s equation Theor. Appl. Genet. 132 627 645 https://doi.org/10.1007/s00122-019-03317-0
Craenen, K. & Ortiz, R. 2003 Genetic improvement for a sustainable management of the resistance 181 198 Jacome, L., Lepoivre, P., Marin, D., Ortiz, R., Romero, R. & Escalant, J.V. Mycosphaerella leaf spot diseases of bananas: Present status and outlook. International Network for the Improvement of Banana and Plantain Montpellier, France
Crossa, J., Gauch, H.G. & Zobel, R.W. 1990 Additive main effects and multiplicative interaction analysis of two international maize cultivar trials Crop Sci. 30 3 493 500 https://doi.org/10.2135/cropsci1990.0011183X003000030003x
Cullis, B.R., Smith, A.B. & Coombes, N.E. 2006 On the design of early generation variety trials with correlated data J. Agric. Biol. Environ. Stat. 11 4 381 393 https://doi.org/10.1198/108571106X154443
Dempster, A.P., Laird, N.M. & Rubin, D.B. 1977 Maximum likelihood from incomplete data via the EM algorithm J. R. Statist. Soc. B (Methodol.) 39 1 1 38
Denis, J. & Gower, J.C. 1996 Asymptotic confidence regions for bi-additive models: Interpreting genotype-environment interactions J. R. Stat. Soc. C Appl. Stat. 45 4 479 493 https://doi.org/10.2307/2986069
Ebdon, J.S. & Gauch, H.G. Jr 2002 Additive main effect and multiplicative interaction analysis of national turfgrass performance trials: I. Interpretation of genotype × environment interaction Crop Sci. 42 2 489 496 https://doi.org/10.2135/cropsci2002.4890
Erima, R., Ssebuliba, J.M., Nankinga, C.M.K., Ragama, P., Barekye, A. & Tushemereirwe, W. 2016 Yield performance of East African highland banana hybrids on farmers’ fields under high disease pressure Afr. J. Agric. Res. 11 13 1141 1149 https://doi.org/10.5897/AJAR12.1610
Falconer, D.S. & Mackay, T.F.C. 1996 Introduction to quantitative genetics 4th ed Longmans Green Harlow, Essex, UK
Food and Agriculture Organization of the United Nations 2014 Food and agriculture statistics http://www.fao.org/statistics/databases/en/. [accessed 20 Feb 2019]
Frutos, E., Galindo, M.P. & Leiva, V. 2014 An interactive biplot implementation in R for modeling genotype-by-environment interaction Stochastic Environ. Res. Risk Assess. 28 1629 1641 https://doi.org/10.1007/s00477-013-0821-z
Gauch, H.G. Jr 2006 Statistical analysis of yield trials by AMMI and GGE Crop Sci. 46 4 1488 1500 https://doi.org/10.2135/cropsci2005.07-0193
Gauch, H.G. Jr 2013 A simple protocol for AMMI analysis of yield trials Crop Sci. 53 5 1860 1869 https://doi.org/10.2135/cropsci2013.04.0241
Gauch, H.G. Jr & Zobel, R.W. 1988 Predictive and postdictive success of statistical analyses of yield trials Theor. Appl. Genet. 76 1 10 https://doi.org/10.1007/BF00288824
Gauhl, F. 1994 Epidemiology and ecology of black Sigatoka Mycosphaerella fijiensis (Morelet) on plantain and banana (Musa sp.) in Costa Rica, Central America International Network for the Improvement of Banana and Plantain Montpellier, France
Henderson, C.R. 1975 Best linear unbiased estimation and prediction under a selection model Biometrics 31 2 423 447 https://doi.org/10.2307/2529430
Henderson, C.R. 1985 Best linear unbiased prediction of non-additive genetic merits in noninbred populations J. Anim. Sci. 60 1 111 117 https://doi.org/10.2527/jas1985.601111x
Hu, X. 2015 A comprehensive comparison between ANOVA and BLUP to valuate location specific genotype effects for rape cultivar trials with random locations Field Crops Res. 179 144 149 https://doi.org/10.1016/j.fcr.2015.04.023
Kalyebara, R., Wood, S. & Abodi, P.N. 2007 Assessing the potential impact of selected technologies on the banana industry in Uganda 141 154 Smale, M. & Tushemereirwe, W. An economic assessment of banana genetic improvement and innovation in the Lake Victoria Region of Uganda and Tanzania. International Food Policy Research Institute Washington, DC, USA
Kang, M.S. 1990 Genotype-by-environment interaction and plant breeding Louisiana State Univ. Agric. Center Baton Rouge, LA, USA
Kang, M.S. 1998 Using genotype-by-environment interaction for crop cultivar development Adv. Agron. 62 199 252 https://doi.org/10.1016/S0065-2113(08)60569-6
Kang, M.S. 2004 Breeding: Genotype-by-environment interaction 218 221 Goodman, R.M. Encyclopedia of plant and crop science. Marcel-Dekker New York, NY, USA
Kang, M.S. & Gauch, H.G. Jr 1996 Genotype-by-environment interaction CRC Press Boca Raton, FL, USA
Karamura, D. 1998 Numerical taxonomic studies of the East African highland bananas (Musa AAA-East Africa) in Uganda PhD Diss University of Reading Reading, UK
Karamura, D. & Pickersgill, B. 1999 A classification of the clones of East African Highland bananas (Musa) found in Uganda Plant Gen. Res. 119 1 6
Kitavi, M., Downing, T., Lorenzen, J., Karamura, D., Onyango, M., Nyine, M., Ferguson, M. & Spillane, C. 2016 The triploid East African Highland Banana (EAHB) genepool is genetically uniform arising from a single ancestral clone that underwent population expansion by vegetative propagation Theor. Appl. Genet. 129 547 561 https://doi.org/10.1007/s00122-015-2647-1
Kubiriba, J., Ssali, R.T., Barekye, A., Akankwasa, K., Tushemereirwe, W.K., Batte, M., Karamura, E.B. & Karamura, D. 2016 The performance of East African highland bananas released in farmers’ fields and the need for their further improvement Acta Hortic. 1114 231 238 https://doi.org/10.17660/ActaHortic.2016.1114.32
Lescot, T. 2018 Banana genetic diversity French Agricultural Research Centre for International Development Montpellier, France
Li, L.F., Wang, H.Y., Zhang, C., Wang, X.F., Shi, F.X. & Chen, W.N. 2013 Origins and domestication of cultivated banana inferred from chloroplast and nuclear genes PLoS One 8 11 1 11 https://doi.org/10.1371/journal.pone.0080502
Lorenzen, J., Tenkouano, A., Bandyopadhyay, R., Vroh, B., Coyne, D. & Tripathi, L. 2010 Overview of banana and plantain (Musa sp.) improvement in Africa: Past and future Acta Hort. 879 595 603 https://doi.org/10.17660/ActaHortic.2010.879.66
Magari, R. & Kang, M.S. 1993 Genotype selection via a new yield-stability statistic in maize yield trials Euphytica 70 105 111 https://doi.org/10.1007/BF00029647
Malosetti, M., Ribaut, J. & van Eeuwijk, F.A. 2013 The statistical analysis of multi-environment data: Modeling genotype-by-environment interaction and its genetic basis Front. Physiol. 44 1 17 https://doi.org/10.3389/fphys.2013.00044
Marimo, P., Nowakunda, K., Aryamanya, W., Azath, H., Babley, H.F., Kazigye, F., Khakasa, E., Kibooga, C., Kibura, J.K., Kindimba, G., Kubiriba, J., Magohe, R.B., Magona, I., Masanja, S.R., Massawe, C., Mbongo, D.B., Mgenzi, A.R., Mubiru, D.M., Mugisha, J.A., Mugume, D., Mwabulambo, B.A., Namuddu, M.G., Ndagire, L., Ngabirano, W., Ngulinziva, L.D., Njau, M.C., Nsibirwa, L., Nyemenohi, S., Okurut, A.W., Shimwela, M.M., Swennen, R., Urio, P.A. & Van den Bergh, I. 2020 Report on consumer acceptability tests of ‘NARITA’ hybrids in Tanzania and Uganda The Alliance of Bioversity International and CIAT Kampala, Uganda
McCulloch, C.E. & Searle, S.R. 2001 Generalized, linear, and mixed models Wiley New York, NY, USA
Meya, I.A. 2021 Nutrient use efficiency in banana-bean intercropping systems in the upper Pangani catchment, Tanzania PhD Diss Katholieke Universiteit Leuven, Belgium
Mohammadi, R. & Amri, A. 2008 Comparison of parametric and non-parametric methods for selecting stable and adapted durum wheat genotypes in variable environments Euphytica 159 419 432 https://doi.org/10.1007/s10681-007-9600-6
Nowakunda, K. & Tushemereirwe, W. 2004 Farmer acceptance of the introduced banana genotypes in Uganda Afr. Crop Sci. J. 12 1 1 6 https://doi.org/10.4314/acsj.v12i1.27656
Olivoto, T. & Lúcio, A.D. 2020 Metan: An R package for multi-environment trial analysis Methods Ecol. Evol. 11 783 789 https://doi.org/10.1111/2041-210X.13384
Olivoto, T., Lúcio, A.D., da Silva, J.A.G., Marchioro, V.S., de Souza, V.Q. & Jost, E. 2019a Mean performance and stability in multi-environment trials I: Combining features of AMMI and BLUP techniques Agron. J. 111 6 2949 2960 https://doi.org/10.2134/agronj2019.03.0220
Olivoto, T., Lúcio, A.D., da Silva, J.A.G., Sari, B.G. & Diel, M.I. 2019b Mean performance and stability in multi environment trials II: Selection based on multiple traits Agron. J. 111 6 2961 2969 https://doi.org/10.2134/agronj2019.03.0221
Olivoto, T., Nardino, M., Carvalho, I.R., Follmann, D.N., Ferrari, M., Szareski, V.J., de Pelegrin, A.J. & de Souza, V.Q. 2017 REML/BLUP and sequential path analysis in estimating genotypic values and interrelationships among simple maize grain yield-related traits Genet. Mol. Res. 16 1 1 19 https://doi.org/10.4238/gmr16019525
Ortiz, R. 1997a Morphological variation in Musa germplasm Genet. Resourc. Crop Evol. 44 393 404 https://doi.org/10.1023/A:1008606411971
Ortiz, R. 1997b Secondary polyploids, heterosis and evolutionary crop breeding for further improvement of the plantain and banana (Musa sp. L) genome Theor. Appl. Genet. 94 1113 1120 https://doi.org/10.1007/s001220050524
Ortiz, R. & de Cauwer, I. 1998 Genotype-by-environment interaction and testing environments for plantain and banana (Musa sp. L.) breeding in West Africa Tropicultura 17 3 97 102
Ortiz, R. & Ekanayake, I.J. 2000 Assessment of genotype × environment interaction and role of physiological analyses for crop breeding 10 31 Ekanayake, I.J. & Ortiz, R. Genotype by environment interaction analysis of IITA mandate crops in sub-Saharan Africa. International Institute of Tropical Agriculture Ibadan, Nigeria
Ortiz, R. & Swennen, R. 2014 From crossbreeding to biotechnology: Facilitated improvement of banana and plantain Biotechnol. Adv. 32 1 158 169 https://doi.org/10.1016/j.biotechadv.2013.09.010
Ortiz, R. & Tenkouano, A. 2011 Genotype by environment interaction and Musa improvement 237 247 Pillay, M. & Tenkouano, A. Banana breeding: Progress and challenges. Taylor & Francis New York, NY, USA
Ortiz, R. & Vuylsteke, D. 1994a Inheritance of black Sigatoka disease resistance in plantain banana (Musa sp.) hybrids Theor. Appl. Genet. 89 146 152 https://doi.org/10.1007/BF00225134
Ortiz, R. & Vuylsteke, D. 1994b Plantain breeding at IITA: The improvement and testing of Musa 130 135 Jones, D. The Improvement and testing of Musa: A global partnership. Proceedings of the First Global Conference of the International Musa Testing Program International Network for the Improvement of Banana and Plantain Montpellier, France
Ortiz, R. & Vuylsteke, D. 1996 Recent advances in Musa genetics, breeding and biotechnology Plant Breed. Abstr. 66 1355 1363
Patterson, H.D. & Thompson, R. 1971 Recovery of inter-block information when block sizes are unequal Biometrika 58 3 545 554 https://doi.org/10.2307/2334389
Perrier, X., Jenny, C., Bakry, F., Karamura, D., Kitavi, M., Dubois, C., Hervouet, C., Philippson, G. & De Langhe, E. 2019 East African diploid and triploid bananas: A genetic complex transported from South-East Asia Ann. Bot. 123 1 19 36 https://doi.org/10.1093/aob/mcy156
Piepho, H.P. 1994 Best linear unbiased prediction (BLUP) for regional yield trials: A comparison to additive main effects and multiplicative interaction (AMMI) analysis Theor. Appl. Genet. 89 647 654 https://doi.org/10.1007/BF00222462
Piepho, H.P., Möhring, J., Melchinger, A.E. & Buchse, A. 2008 BLUP for phenotypic selection in plant breeding and variety testing Euphytica 161 209 228 https://doi.org/10.1007/s10681-007-9449-8
Pillay, M., Ogundiwin, E., Nwakanma, D., Ude, G. & Tenkouano, A. 2001 Analysis of genetic diversity and relationships in East African banana germplasm Theor. Appl. Genet. 102 965 970 https://doi.org/10.1007/s001220000500
Resende, M.D.V. 2004 Métodos estatísticos ótimos na análise de experimentos de campo Embrapa Florestas Colombo, Brazil
Resende, M.D.V. 2007a Matemática e estatística na análise de experimentos e no melhoramento genético Embrapa Florestas Colombo, Brazil
Resende, M.D.V. 2007b SELEGEN-REML/BLUP: Sistema estatístico e selecaö genética computadorizada via modelos lineares mistos Embrapa Florestas Colombo, Brazil
Resende, M.D.V. & Duarte, J.B. 2007 Precision and quality control in variety trials Pesqui. Agropecu. Trop. 37 3 182 194
Robinson, G.K. 1991 That BLUP is a good thing: The estimation of random effects Stat. Sci. 6 1 15 32 https://doi.org/10.1214/ss/1177011926
Robinson, J.C. 1996 Bananas and plantains CAB International Wallingford, UK
Schmidt, P. 2019 Estimating heritability in plant breeding programs PhD Diss University of Hohenheim Rostock, Germany
Searle, S.R., Casella, G. & McCulloch, C.E. 1992 Variance components Wiley Hoboken, NJ, USA
Shepherd, K. 1957 Banana cultivars in East Africa Trop. Agric. 34 277 286
Smith, A.B., Cullis, B.R. & Thompson, R. 2005 The analysis of crop cultivar breeding and evaluation trials: An overview of current mixed model approaches J. Agric. Sci. 143 6 449 462 https://doi.org/10.1017/S0021859605005587
Ssali, R.T., Barekye, A., Buregeya, H., Erima, R., Namanya, P. & Kubiriba, J. 2016 Genotypic variability estimates of agronomic traits in secondary triploid banana ‘Matooke’ (Musa sp., AAA-EA) hybrids Afr. J. Plant Sci. 10 4 84 88 https://doi.org/10.5897/AJPS2015.1343
Ssali, R.T., Nowankunda, K., Barekye, R.E., Batte, M. & Tushemereirwe, W.K. 2010 On-farm participatory evaluation of East African highland banana ‘Matooke’ hybrids (Musa sp.) Acta Hort. 879 585 591 https://doi.org/10.17660/ActaHortic.2010.879.65
Swennen, R., Blomme, G., Van Asten, P., Lepoint, P., Karamura, E., Njukwe, E., Tinzaara, W., Viljoen, A., Karangwa, P., Coyne, D. & Lorenzen, J. 2013 Mitigating the impact of biotic constraints to build resilient banana systems in central and eastern Africa 85 104 Van Lauwe, B., van Asten, P. & Blomme, G. Agro-ecological intensification of agricultural systems in the African Highlands. Taylor & Francis London, UK
Swennen, R. & De Langhe, E. 1985 Growth parameters of yield of plantain (Musa cv. AAB) Ann. Bot. 56 2 197 204
Swennen, R. & Vuylsteke, D. 1993 Breeding black Sigatoka-resistant plantains with a wild banana Trop. Agric. 70 1 74 77
Swennen, R., Vuylsteke, D. & Hahn, S.K. 1989 Combating the black Sigatoka threat to plantains IITA Research Briefs. 9 2 2 4
Tenkouano, A. 2001 Current issues and future directions for Musa genetic improvement research at the International Institute of Tropical Agriculture 11 23 Molina, A.B., Roa, V.N. & Maghuyop, M.A.G. Advancing banana and plantain R&D in Asia and the Pacific. International Network for the Improvement of Banana and Plantain Los Baños, Laguna, Philippines
Tenkouano, A., Lamien, N., Agogbua, J., Amah, D., Swennen, R., Traoré, S., Thiemele, D., Aby, N., Kobenan, K., Gnonhouri, G., Yao, N., Astin, G., Sawadogo-Kabore, S., Tarpaga, V., Issa, W., Lokossou, B., Adjanohoun, A., Amadji, G.L., Adangnitode, S., Igue, K.A.D. & Ortiz, R. 2019 Promising high-yielding tetraploid plantain-bred hybrids in West Africa Int. J. Agron. 2019 1 8 https://doi.org/10.1155/2019/3873198
Thiele, G., Dufour, D., Vernier, P., Mwanga, R.O.M., Parker, M.L., Geldermann, E.S., Teeken, B., Wossen, T., Gotor, E., Kikulwe, E., Tufan, H., Sinelle, S., Kouakou, A.M., Friedmann, M., Polar, V. & Hershey, C. 2021 A review of varietal change in roots, tubers and bananas: Consumer preferences and other drivers of adoption and implications for breeding Int. J. Food Sci. Technol. 56 3 1076 1092 https://doi.org/10.1111/ijfs.14684
Tinzaara, W., Stoian, D., Walter, O., Kikulwe, E., Otieno, G. & Blomme, G. 2018 Challenges and opportunities for smallholders in banana value chains 65 90 Kema, G. & Drenth, A. Achieving sustainable cultivation of bananas 1st ed Taylor & Francis London, UK https://doi.org/10.1201/9781351114622
Tushemereirwe, W.K. 1996 Factors influencing expression of leaf spot diseases of highland bananas in Uganda PhD Diss University of Reading Reading, UK
Tushemereirwe, W., Batte, M., Nyine, M., Tumuhimbise, R., Barekye, A., Ssali, T., Talengera, D., Kubiriba, J., Lorenzen, J., Swennen, R. & Uwimana, B. 2015 Performance of ‘NARITA’ banana hybrids in the preliminary yield trial for three cycles in Uganda National Agriculture Research Organization, International Institute of Tropical Agriculture Kampala, Uganda
Uma, S., Saraswathi, M.S. & Pillay, M. 2011 Evolution and genetics relationships in banana and plantains 21 34 Pillay, M. & Tenkouano, A. Banana breeding: Progress and challenges. Taylor & Francis New York, NY, USA
Vaezi, B., Pour-Aboughadareh, A., Mohammadi, R., Mehraban, A., Hossein-Pour, T., Koohkan, E., Ghasemi, S., Moradkhani, H. & Siddique, K.H.M. 2019 Integrating different stability models to investigate genotype × environment interactions and identify stable and high-yielding barley genotypes Euphytica 215 63 1 18 https://doi.org/10.1007/s10681-019-2386-5
Viljoen, A., Mahuku, G., Massawe, C., Ssali, R.T., Kimunye, J., Mostert, G., Ndayihanzamaso, P. & Coyne, D.L. 2017 Banana diseases and pests: Field guide for diagnostics and data collection International Institute of Tropical Agriculture Ibadan, Nigeria
Vuylsteke, D., Swennen, R. & Ortiz, R. 1993 Development and performance of black Sigatoka resistant tetraploid hybrids of plantain (Musa sp., AAB group) Euphytica 65 33 42 https://doi.org/10.1007/BF00022197
Weikai, Y. 2014 Crop variety trials: Data management and analysis Wiley West Sussex, UK
Williams, E., Piepho, H.P. & Whitaker, D. 2011 Augmented p-rep designs Biometr. J. 53 1 19 27 https://doi.org/10.1002/bimj.201000102
Xu, Y., Li, P., Zou, C., Lu, Y., Xie, C., Zhang, X., Prasanna, B.M. & Olsen, M.S. 2017 Enhancing genetic gain in the era of molecular breeding J. Expt. Bot. 68 11 2641 2666 https://doi.org/10.1093/jxb/erx135
Yan, W. & Hunt, L.A. 2001 Interpretation of genotype × environment interaction for winter wheat yield in Ontario Crop Sci. 41 1 19 25 https://doi.org/10.2135/cropsci2001.41119x
Yan, W. & Kang, M.S. 2003 GGE biplot analysis: A graphical tool for breeders, geneticists, and agronomists CRC Press Boca Raton, FL, USA
Yan, W. & Tinker, N.A. 2006 Biplot analysis of multi-environment trial data: Principles and applications Can. J. Plant Sci. 86 623 645 https://doi.org/10.4141/P05-169
Yang, R. 2007 Mixed-model analysis of crossover genotype–environment interactions Crop Sci. 47 3 1051 1062 https://doi.org/10.2135/cropsci2006.09.0611