Estimates of reciprocal effects and heterosis for root morphological traits in tropical maize under contrasting nitrogen levels

In maize, the nitrogen (N) use efficiency does not exceed 50% and one way to overcome this problem is to improve the N-use efficiency (NUE) through the genetic improvement of root morphological and architecture traits. However, little is known about the influence of reciprocal effects and heterosis on the expression of root traits on NUE. Here, our objective was to estimate the reciprocal effect and heterosis among NUE tropical maize lines for root morphological traits under contrasting N levels. We evaluated 90 maize crosses (45 F1 and 45 reciprocal F1 crosses), and their 10 parental inbred lines for root morphological traits in a greenhouse experiment under two treatments: low N (LN) and high N (HN) supply. A diallel analysis using a mixed model approach was carried out to estimate the effects of general and specific combining ability (GCA and SCA) and reciprocal (RGCA and RSCA). Furthermore, we estimated the average heteroses (AH) for each trait under the N level. Pearson's correlation coefficient was estimated between: AH of the F1 and the AH of the reciprocal F1, RGCA and the AH of the F1 hybrids, and between RGCA and the AH of the reciprocal F1. The GCA components of variance were significant (P<0.05) for almost all traits, except the root-shoot ratio (RSR) and axial root length (ARL) in both N levels. For SCA, there was significance in LN for the traits root dry weight (RDW), RSR, ARL, root average diameter, root volume (RV), and in HN for shoot dry weight (SDW), ARL, root surface area (RSA) and RV. For RGCA there was significance for the traits RDW, SDW, ARL, RSA, and RV at both levels of N. Under LN, we found that the estimates of correlations between the AH of F1 and the reciprocal F1 showed low magnitude for all traits. Whereas under HN, these estimates ranged from 0,57 (RV) to 0,88 (ARL). Regardless of N levels, the estimates of the correlation between RGCA and F1 effects showed low magnitude whereas RGCA presented moderate-to-strong and positive correlations for all traits with reciprocal F1. We concluded that the reciprocal effects have a high influence on the expression of AH, mainly under LN.