Genetics of Plant Architecture and Its Effect on Yield in Cassava (Manihot esculenta Crantz)

Abstract

One of the major biotic stresses affecting maize production in SSA is the parasitic weed Striga hermonthica (Del.) Benth. Breeding for Striga resistance is decades old and some progress has been made despite the complex nature of its inheritance. Progress in developing high-yield Striga resistant germplasm has been hampered by the narrow genetic base, lack of breeder- ready molecular markers for selection and limited phenotyping capacity for Striga resistance breeding programs in SSA. The objectives of this study were to: i) investigate gene action for resistance to Striga hermonthica in mid-altitude adapted tropical maize; ii) iii) identify genomic regions associated with Striga resistance among early maturing mid-altitude maize inbred lines predict performance of tested and untested doubled haploid (DH) lines under artificial Striga infested conditions. Three independent studies were conducted to gain understanding of the genetics of Striga resistance in mid-altitude tropical maize germplasm. In the first experiment, 132 hybrids were generated from 12 inbred lines through a diallel mating design. The inbred lines and the hybrids were phenotyped under artificial Striga infestation at Kibos and Alupe for two seasons and under optimal conditions in 2022 and 2023. An alpha lattice experimental design was used for hybrid evaluation, with two replicates at four trial locations under optimal conditions and two trial locations under artificial Striga infestation. Data were collected on grain yield, agronomic traits and Striga resistance parameters. Analysis of variance, variance components estimates, and heritability were carried using AsREML in R. Combining ability analysis of the hybrids and lines following Griffing’s Method 3, Model 1 was carried out using AGD-R software. Results showed that the inbred lines and the test hybrids varied significantly with broad sense heritability >73% for grain yield, agronomic and Striga resistance parameters under the two management schemes. Fifteen (15) hybrids outperformed the checks by 32.1% in grain yield under Striga infestation. The study revealed that additive and non-additive gene effects influenced grain yield, agronomic traits and Striga resistance parameters. Based on Baker’s ratio, additive gene effects were more important than non-additive gene effects indicating that the traits are genetically controlled. Maternal effects were not significant for most traits. Five parental lines namely: DL171342, DL17535, DL17611, DL17933 and TZISTR1163 had positive GCAs effects for grain yield and negative GCA for Striga resistance under Striga infestation. In the second experiment, 163 F6 early maturing inbred lines were used in a genome-wide association study. The 163 lines were testcrossed to generate 459 hybrids that were phenotyped under artificial Striga infested conditions at two locations (Kibos and Alupe) and natural Striga infestation at two locations (Madeya and Teso). The inbred lines were genotyped with 955,670 single nucleotide polymorphic (SNP) markers. After quality control 155 inbred lines and 151, 670 high quality SNP markers were retained for association analysis. Analysis of variance, variance components estimates, adjusted means and heritability were computed using AsREML in R. The FarmCPU model was used to identify significant SNPs associated with grain yield, Striga resistance parameters and agronomic traits using GAPIT package in R. The putative candidate genes linked to significant SNPs were obtained from the Maize Genetics and Genomics Database (MaizeGDB; http://maizegdb.org) using BedTools. Gene functions were obtained from MaizeMine. Analysis of variance revealed that hybrids varied significantly showing moderate broad sense heritability for all Striga resistance parameters except Striga damage rating whose heritability was high (62−64%). High significant negative correlations were observed between grain yield and Striga damage rating (-0.47***−-0.56***). A total of 42 significant SNPs, which accounted for 0.1–38.9% of the phenotypic variation for Striga resistance parameters, were identified. Eight loci in chromosomes 2 (S2_44331849, S2_87827811), 3 (S3_175540577, S3_8219084) and 6 (S6_159470193, S6_107754561, S6_96337848 and S6_109282273) accounted for 11.9−38.9% phenotypic variance in Striga resistance parameters. The significant SNPs were near several putative genes that coded for proteins, transcription factors, metabolic enzymes among other factors involved in plant growth and defense against pathogens. For the third objective, 606 doubled haploid (DH) lines were used for genomic prediction. A training population of 116 lines was phenotyped in hybrid combination by crossing the lines with two testers to generate 232 test cross hybrids. The testcrosses were phenotyped at Kibos, Alupe and Siaya under artificial Striga infestation in 2020. The 606 DH lines were genotyped with 8,439 rAmpSeq markers. After quality control, 5,380 high quality rAmpSeq markers were retained and used for genomic prediction. Analysis of variance was carried out to estimate components of variance and heritability. AsReml was used in generating the best linear unbiased estimates (BLUEs), best linear unbiased predictions (BLUPs), estimates of variance components and heritability. The BLUEs of the training population and the 5,380 rAmpSeq markers were used in genomic prediction of the 606 inbred lines. A genomic relationship matrix was computed and incorporated in the genomic prediction using the reaction norm model. Three cross validation schemes (CV0, CV1 and CV2) were used to estimate the prediction accuracy of the model. The model was then used to predict the performance of the tested and untested DH lines. Genomic estimated breeding values (GEBVs) were computed for various Striga resistance parameters. Genetic variance was larger than the G×E variance. Heritability was low to moderate (38‒65%) for Striga resistance parameters while that of grain yield was 54%. Prediction accuracy based on cross validation (CV) was low to moderate (0.24 to 0.53) for CV0 and CV2 (0.20 to 0.37). For GY, the prediction accuracies were 0.59 and 0.56 for CV0 and CV2, respectively. Using the reaction norm model, 300 DH lines with desirable GEBVs for reduced number of emerged Striga plants (STR) at 8, 10, and 12 weeks after planting were identified. The GEBVs of DH lines for Striga resistance associated traits in the training and testing sets were similar in magnitude. These results highlight the potential application of genomic selection for Striga resistance breeding in maize. In conclusion, the study findings indicate that additive gene effects were preponderant based on Baker’s ratio. Five inbred lines that can potentially be used as valuable sources of favourable alleles in Striga resistant hybrid breeding programs were identified. Additionally, the study identified 10 significant SNPs that can be considered for fine mapping and development breeder-ready markers for selection in Striga breeding programs. The application of GS where a large set is genotyped and phenotyping only a limited subset helps to reduce phenotyping costs due to limited land and the need for artificial Striga infestation. The integration of genomic-assisted strategies and DH technology for line development coupled with forward breeding for major adaptive traits will enhance genetic gains in breeding for Striga resistance in maize. The estimation of the GEBVs should be part of the standard operation procedures for Striga resistance breeding programs.