Phenotypic diversity for fruit quality traits and bioactive compounds in red-fleshed guava: Insights from multivariate analyses and machine learning algorithms

D.S. Mishra,M.K. Berwal,Anshuman Singh,A.K. Singh,V.V. Appa Rao,Vikas Yadav,B.D. Sharma
DOI: https://doi.org/10.1016/j.sajb.2022.06.043
IF: 3.111
2022-09-01
South African Journal of Botany
Abstract:Precise characterization of genetic resources is critically important for developing high yielding and nutri-rich guava cultivars. This study aimed at delineating phenotypic diversity for fruit yield, quality and bioactive compounds in the newly developed red-fleshed guava selections and hybrids. Analysis of Variance (ANOVA) and the associated effect-size estimates revealed a high degree of phenotypic diversity for most of tree growth and fruit quality traits, minerals and bioactive compounds in the guava accessions analyzed. Principal Component Analysis efficiently discerned key trends in data with first and second Principal Components representing nutritional and bioactive compounds, and fruit yield attributes, respectively. Grouping of guava accessions into separate k-means clusters evinced a fairly high genetic divergence for the traits examined. Correlation and network analyses revealed that total antioxidant activity of fruits was strongly and positively regulated by total phenols (TP), β-carotene (β-car) and flavonoids (Flav). Gradient Boosting and Random Forest regression models also confirmed an overarching influence of TP, β-car and Flav in determining the total antioxidant activity. Our findings are expected to assist the guava breeders in selecting suitable parental combinations for future hybridization programs. The promising guava accessions identified by us meet the fresh consumption and processing standards, and may thus be considered for commercial planting.
plant sciences
What problem does this paper attempt to address?