English French




In Ghana, sorghum (Sorghum bicolor L. Moench) ranks third after maize and rice, mainly as a dietary staple and is crucial for ensuring food security in the three northern regions. However, yields in farmers’ fields are extremely low (less than 2.0 tonnes ha-1) compared to the 4.5 to 5.0 tonnes ha-1 achievable in developed countries. A series of investigations were therefore carried out to identify farmers’ production constraints to facilitate development of strategies towards finding sustainable, long-term solutions within the context of plant breeding. A participatory rural appraisal was therefore conducted among smallholder sorghum farmers in four communities in the Talensi-Nabdam and Binduri districts of the Upper East Region of Ghana. The survey also sought to determine farmers’ perceptions of soil fertility as well as their current soil management strategies. Data were collected through focus group discussions, questionnaires and direct field observations. Within the communities, drought, high cost of farm input (labour and ploughing services) and declining soil fertility were the most important constraints to sorghum production. In spite of widespread low soil fertility, soil management methods such as fertiliser application, cover cropping and retention of crop residues were seldom practiced within the communities. Drought tolerance, high grain yield, earliness, grain quality (suitability for local foods and beverages) as well as low fertiliser requirement were the most preferred sorghum traits. Considering resources available to farmers and researchers in Ghana, mutation induction was considered a workable strategy in a breeding programme to expand the genetic diversity in the crop towards selection of phosphorus (P) use efficient genotypes. Mutation induction was achieved through chemical treatment of an inbred sorghum line BTx623 using ethyl methanesulphonate. A subset of 547 mutagenised lines at M3 generation were subjected to phenotypic assessment for expression of key morpho-agronomic traits of sorghum, both quantitative and qualitative. Significant differences (p < 0.05) were observed among the M3 families for all the traits studied as well as between the families and BTx623. High genotypic coefficients of variation (GCV) established the existence of considerable variability. Broad sense heritability was also high ranging from 0.68 for stem girth to 0.99 for number of leaves. Positive and significant (p < 0.05) correlations were detected between grain yield and number of leaves, panicle width and stem girth. Principal component analysis revealed three principal components that accounted for 64% of the total variability. Grain yield (0.364), panicle width (0.436) and panicle length (0.411) were the most contributing traits. Hierarchical agglomerative clustering based on Unweighted Pair Group Method with Arithmetic mean (UPGMA), resolved the population into three divergent clusters at 52% level of dissimilarity. Two hundred and fifty three M3 lines were grouped into clusters separate from the wildtype BTx623, which indicates considerable variability between the mutagenised lines and the control. Close inspection for novel or distinct phenotypes revealed putative mutants for brown midrib, erect leaf and bloomless traits. Considerable genetic variability exists within the population which can be exploited for further breeding work. Further, genetic variation for P-use efficiency among the mutagenised lines was evaluated alongside selected cultivated varieties. Two hundred lines, comprising 170 mutagenised lines (agronomically-superior) and 30 cultivated varieties were evaluated under screen-house and field conditions and in two contrasting soil-P environments using a randomized complete block design and an augmented (incomplete block) design respectively. Measured traits included seedling vigour, days to anthesis, yield (grain and stover) as well as P concentration in the grain and stover. Analysis of variance for each P environment revealed significant (p <0.01) differences among the genotypes for all measured traits. Combined analysis of variance also showed significant (p <0.01) effects between genotype and P environment for all traits except plant height and stem girth, indicating differences in genotypic response under contrasting P environments. Overall, grain and stover yield decreased by 49 and 27% respectively with the omission of P. Based on Pearson’s correlation analysis, grain yield was positively (p < 0.05) correlated with stover yield and P utilisation for grain production, but negatively correlated with P utilisation for stover production in both P environments, respectively. On the other hand, the relationship between stover yield and harvest index was significant (p < 0.05) in low-P soils but not significant in optimum-P soils. The contribution of each trait to observed genetic variation was determined by principal component analysis. The first three principal components (PC1, PC2 and PC3) explained 77% and 79% of the total variation in the low-P and optimum-P environments respectively. Traits that contributed the most variability in both P-environments were grain and stover yield. Subsequently, cluster analysis using UPGMA categorised the 200 genotypes into three divergent clusters at 49% dissimilarity. Cluster I, II and III contained 65, 106 and 29 genotypes respectively. Cluster I was characterised by reduced height and relatively late anthesis, Cluster II comprised the genotypes such as Grinkan, MR732, KuyumaWSV387 and CE151262A1 which produced the highest grain (mean = 56 g/plant) and stover yield (mean = 164 g/plant) whereas the genotypes in Cluster III were the tallest (mean height = 106 cm) in the population. The lines were then classified into four categories as P-uptake efficient, P-utilisation efficient, low-P stress tolerant or low-P stress sensitive, based on relative performance of the genotypes in both P environments. Kadaga West and Mut3412 were the most tolerant to low-P stress. Grinkan and ICSV1049 were the most P-uptake efficient whereas TxARG1 and Mut3708 were the most P-utilisation efficient. A comparison between the top-ranked mutant lines and cultivated genotypes indicated that the cultivated genotypes were more efficient at P-uptake whereas the putative mutants were more efficient at P-utilisation. There is potential to exploit the genotypic variability observed within the population for further breeding work. The identified P-use efficient mutagenised lines may either be released directly as mutant varieties following multi-locational trials or used as parents in future recombinant breeding programmes.