Effect of environment on multi-element grain composition of pigeonpea cultivars under farmers' conditions

Pigeonpea (Cajanus cajan L. Millsp.) is often intercropped with maize (Zea mays L.) in eastern and southern Africa. The studies aimed at determining how different genotypes of pigeonpea responded in terms of grain element composition under farmers' cropping conditions. Approx. 78 farmers participated. They came from four study sites in Tanzania (Babati and Gairo) and Malawi (Nyambi and Ntonda) that differed in terms of tradition for using pigeonpea as well as in environmental conditions. The individual grain weight of the pigeonpea crops from Malawi were 21% (P < 0.05) higher than those from Tanzania. However, only B, Cu, Mo, N, Ni, P and S were affected by grain weight (P < 0.05). Weak (r(2) < 0.10) negative correlations existed between grain yield and the grains' proportion of Ca, Mg, P, and Zn. The proportion of every element, with the exception of Cr, in the grain differed between sites (P < 0.05) but not between varieties (P > 0.05). The amounts of K, Mg, S and Fe accumulated per grain were slightly lower (P < 0.11) in ICEAP00040 compared to the more traditional varieties. Variations in DTPA-exchangeable Zn and Fe in the soil were not reflected in grain concentrations but grain P had a curvi-linear relation (r(2) = 0.44) to the soil NaHCO3--exchangeable P indicating P deficiency for several soils. The P and Zn content were correlated (r(2) > 0.41) as the only two grain elements. Unique fingerprinting by multivariate statistics was possible for each site when using the element proportion of the grain dry matter with or without soil characteristics. In all cases, different elements contributed with varying weight to the discrimination between the sites. However, it was not possible to distinguish between the varieties when considering all four environments. Reducing the models to include Fe, K, Mg, P, S and Zn only, did however allow some distinction between the two genotypes, which indicates that genotypic variability is expressed in a fairly limited number of elements. In the cases of Gairo and Nyambi, it was possible to distinguish between varieties. In the case of Gairo, the models distinguished between ICEAP00068 and the others, i.e. ICEAP00040 and Babati White where ICEAP00068 was associated with a higher proportion of Fe, P, S, and Zn in the grain. In the case of Nyambi, the models distinguished between ICEAP00040 and ICP9145 where ICEAP00040 was associated with a lower proportion of Ca, Cu, Fe, Mn, Ni, P, S, and Zn and a higher proportion of Cr and Na in the grain. It is thus possible in some cases to separate varieties based on multi-element grain content across a relatively narrow environmental gradient but not generally across all environments. These findings should be included in breeding programmes focusing on the improvement of the nutritional value of our food crops.