Utility of Ugandan genomic selection cassava breeding populations for prediction of cassava viral disease resistance and yield in West African clones.

Cassava (Manihot esculenta Crantz) is a staple crop for ~800 million people in sub-Saharan Africa. Its production and productivity are being heavily affected by the two viral diseases: cassava brown streak disease (CBSD) and cassava mosaic disease (CMD), impacting greatly on edible root yield. CBSD is currently endemic to central, eastern and southern Africa, if not contained could spread to West Africa the largest cassava producer and consumer in the continent. Genomic selection (GS) has been implemented in Ugandan cassava breeding for accelerated development of virus resistant and high yielding clones. This study leveraged available GS training data in Uganda for pre-emptive CBSD breeding in W. Africa alongside CMD and fresh root yield (FRW). First, we tracked genetic gain through the current three cycles of GS in Uganda. The mean genomic estimated breeding values (GEBVs), indicated general progress from initial cycle zero (C0) to cycle one (C1) and cycle two (C2) for CBSD traits and yield except for CMD. Secondly, we used foliar data of both CBSD and CMD, as well as harvest root necrosis and yield data to perform cross-validation predictions. Cross-validation prediction accuracies of five GS models were tested for each of the three GS cycles and West African (WA) germplasm as a test set. In all cases, cross-validation prediction accuracies were low to moderate, ranging from -0.16 to 0.68 for CBSD traits, -0.27 to 0.57 for CMD and -0.22 to 0.41 for fresh root weight (FRW). Overall, the highest prediction accuracies were recorded in C0 for all traits tested across models and the best performing model in cross-validation was G-BLUP. Lastly, we tested the predictive ability of the Ugandan training sets to predict CBSD in W. African clones. In general, the Ugandan training sets had low prediction accuracies for all traits across models in West African germplasm, varying from -0.18 to 0.1. Based on the findings of this study, the cassava breeding program in Uganda has made progress through application of GS for most target traits, but the utility of the training population for pre-emptive breeding in WA is limiting. In this case, efforts should be devoted to sharing Ugandan germplasm that possess resistance with the W. African breeding programs for hybridization to fully enable deployment of genomic selection as a pre-emptive CBSD breeding strategy in W. Africa.

Outlook of Cassava Brown Streak Disease Assessment: Perspectives of the Screening Methods of Breeders and Pathologists.

Cassava production and productivity in Eastern, Central, and Southern Africa are ravaged by cassava brown streak disease (CBSD), causing yield losses of up to 100% when susceptible varieties are grown. Efforts to develop CBSD-resistant clones are underway. However, the methods for screening CBSD resistance currently vary between breeders and pathologists, with the limited empirical data to support their choices. In this study, we used the empirical CBSD foliar and root necrosis data from two breeding populations, termed cycle zero (C0) and cycle one (C1), to assess and compare the effectiveness of the CBSD screening methods of breeders vs. pathologists. On the one hand, the estimates of broad-sense heritability (H 2) for the CBSD root necrosis assessment of breeder ranged from 0.15 to 0.87, while for the assessment method of pathologists, H 2 varied from 0.00 to 0.71 in C0 clones. On the other hand, the marker-based heritability estimates (h 2) for C0 ranged from 0.00 to 0.70 for the assessment method of breeders and from 0.00 to 0.63 for the assessment method of pathologists. For cycle one (C1) population, where both foliar and root necrosis data were analyzed for clones assessed at clonal evaluation trials (CETs) and advanced yield trials (AYTs), H 2 varied from 0.10 to 0.59 for the assessment method of breeders, while the H 2 values ranged from 0.09 to 0.35 for the CBSD computation method of pathologists. In general, higher correlations were recorded for foliar severity from the assessment method of breeders (r = 0.4, p ≤ 0.01 for CBSD3s and r = 0.37, p ≤ 0.01 for CBSD6s) in C1 clones evaluated at both clonal and advanced breeding stages than from the approach of pathologists. Ranking of top 10 C1 clones by their indexed best linear unbiased predictors (BLUPs) for CBSD foliar and root necrosis showed four overlapping clones between clonal and advanced selection stages for the method of breeders; meanwhile, only a clone featured in both clonal and advanced selection stages from the CBSD assessment method of pathologists. Overall, the CBSD assessment method of breeders was more effective than the assessment method of pathologists, and thus, it justifies its continued use in CBSD resistance breeding.

Prospects for Genomic Selection in Cassava Breeding.

Cassava ( Crantz) is a clonally propagated staple food crop in the tropics. Genomic selection (GS) has been implemented at three breeding institutions in Africa to reduce cycle times. Initial studies provided promising estimates of predictive abilities. Here, we expand on previous analyses by assessing the accuracy of seven prediction models for seven traits in three prediction scenarios: cross-validation within populations, cross-population prediction and cross-generation prediction. We also evaluated the impact of increasing the training population (TP) size by phenotyping progenies selected either at random or with a genetic algorithm. Cross-validation results were mostly consistent across programs, with nonadditive models predicting of 10% better on average. Cross-population accuracy was generally low (mean = 0.18) but prediction of cassava mosaic disease increased up to 57% in one Nigerian population when data from another related population were combined. Accuracy across generations was poorer than within-generation accuracy, as expected, but accuracy for dry matter content and mosaic disease severity should be sufficient for rapid-cycling GS. Selection of a prediction model made some difference across generations, but increasing TP size was more important. With a genetic algorithm, selection of one-third of progeny could achieve an accuracy equivalent to phenotyping all progeny. We are in the early stages of GS for this crop but the results are promising for some traits. General guidelines that are emerging are that TPs need to continue to grow but phenotyping can be done on a cleverly selected subset of individuals, reducing the overall phenotyping burden.