Genetic analysis of pasting properties of selected cassava genotypes in Tanzania for baking applications

Abstract

Partial substitution of wheat flour with high-quality cassava flour (HQCF) in baked products can increase cassava utilization, support cassava commercialization and reduce dependence on imported wheat in Tanzania. However, successful cassava-wheat flour substitution requires HQCF with appropriate pasting properties harvested at appropriate times. Information on the pasting properties of Tanzanian cassava genotypes and their appropriate harvesting time for producing HQCF with desirable properties remains limited. Furthermore, rapid development of cassava varieties with targeted pasting properties is constrained by the high cost and time consuming conventional phenotyping methods. Portable near-infrared spectroscopy (NIRS), such as the SCiO™ sensor, offers a potential high throughput rapid alternative tool for screening large breeding populations at early stages in the breeding cycle. However, its application for predicting cassava HQCF pasting properties has not been adequately evaluated. In addition, limited knowledge of genomic regions and candidate genes associated with pasting properties constrains marker-assisted improvement of cassava for specific end uses. This study aimed to: (i) identify suitable varities × harvesting time combinations for desirable pasting properties for baking applications; (ii) evaluate the potential of SCiO™ NIRS for rapid screening of cassava genotypes for pasting properties; and (iii) identify putative genomic regions and putative candidate genes associated with key HQCF pasting properties using genome-wide association studies (GWAS). For the first study, eleven cassava genotypes were evaluated at TARI-Ukiriguru during the 2020/2021 and 2021/2022 seasons and harvested at 10 and 12 months after planting (MAP). For the second study, 221 diverse genotypes were planted at TARI Ukiriguru and Chambezi experimental station during same seasons using an augmented design. For the third study, 236 diverse cassava genotypes were evaluated at TARI-Ukiriguru and Chambezi experimental stations during the same seasons using an augmented design. HQCF pasting properties were determined using a Rapid Visco Analyzer (RVA), while SCiO™ was used to acquire nearinfrared spectra. Genotype and harvest-time effects were analyzed using general linear models in R. For NIRS prediction, Gradient Boosting Machines (gbm), Linear Model (lm), and Lasso and Elastic-Net Regularized Generalized Linear Models (glmnet) were used for model calibration and validation. GWAS and putative candidate gene annotation was conducted using appropriate statistical packages in R. Kipusa recorded the lowest setback viscosity both at 10 and 12 MAP, indicating its potential for baked products requiring a softer texture and reduced firming after cooling. In contrast, Hombolo recorded the highest setback viscosity, indicating its potential for products requiring greater structural stability after cooling. For SCiO™ prediction, glmnet with first-derivative spectral pretreatment produced the best models for peak viscosity and breakdown, with validation coefficients of determination (R²) of 0.51 and 0.66, respectively. The corresponding calibration R² values were 0.56 and 0.66. Prediction of pasting temperature was best achieved using a combination of spectral pretreatments. These results demonstrate the potential of SCiO™ NIRS for screening cassava genotypes for selected pasting properties. GWAS identified 48 putative significant SNPs associated with the six evaluated pasting properties traits and 22 putative candidate genes within the associated genomic regions. These putative genomic regions and putative candidate genes provide potential targets for further validation and development of molecular markers to support breeding for cassava varieties with desirable pasting properties. Overall, the study provides complementary phenotypic, technological, and genetic approaches for improving cassava HQCF for specific end uses. Identification of suitable genotype × harvesttime combinations, demonstration of the potential of SCiO™ NIRS for rapid phenotyping, and identification of these putative genomic regions associated with pasting properties can contribute to more efficient cassava breeding and increased utilization of cassava in wheat-cassava baked products, thereby supporting cassava commercialization and reducing dependence on imported wheat in Tanzania.