Phenotypic variation and correlations among traits under P treatments
Data for all nine traits were normally distributed across the three P treatments (Fig. S1). Under low P condition, Zhongmai 895 showed higher RL, RV, ROSA, RDW, TDW and RRS, but lower RD, RTN, and SDW than Yangmai 16. Except for RD and RRS in the high P treatment, Yangmai 16 had higher RL, RV, RTN, ROSA, SDW, RDW and TDW than Zhongmai 895 (Fig. S2).
Phenotypic variances among the DH lines were significant (P <0.0001). Transgressive segregations across P treatments were observed for most of traits (Fig. S2; Table 1). The average values of the DH lines for RL, RV, RTN, and RRS were higher than parents in the low P treatment, indicating positive effects for root vigor from both parents (Fig. 1,S2). Broad-sense heritabilities of nine traits were ranged from 0.76 to 0.91 (Table 1).
RSA and RBT traits were correlated significantly (r = 0.59 to 0.98 at all three P levels). However, RD was negatively correlated with RL, RTN and ROSA (r = –0.23 to –0.69) (Fig. 2).
QTLs for root system architecture traits
Nineteen QTLs from the three P treatments were identified for RSA traits RL, RV, RTN and ROSA on chromosomes 1BL, 2BL, 2DL, 3DL, 4BS, 6AL, 6BL (2), 6DS, 7AS (5), 7AL (3) and 7BL (2) (Table 2). Among them 6 QTLs were detected for the zero control, and 7 in the low and 6 in the high P treatments with the phenotypic variances explained of 45.1, 48.0, 53.7%, respectively. Seventeen QTLs on chromosomes 2BL, 2DL, 3DL (2), 4BS (3), 6AL, 6BL (3), 6DS (2), 7AS (2), 7AL (2) conferred positive additive effects contributed by Zhongmai 895 (Table 2). In the low P treatment, three QTLs were identified for each of RL and RTN, explaining 20.7% and 21.7% of phenotypic variances, respectively. RL and RTN were co-located on chromosomes 6BL, 7AL and 7BL. In addition, Zhongmai 895 possessed positive alleles on 6BL and 7AL for increased phenotypic values (Table 2).
QTLs for root biomass-related traits
Fifteen QTLs for RBT traits were identified on chromosomes 3AS (3), 3DL, 4BS (5), 4DS (4), 6BL and 6DS (Table 2). Two, 4 and 9 QTLs were identified in the zero, low and high P treatments, and explained 34.0, 42.8 and 67.6% of the phenotypic variances, respectively. A stable QTL (QRDW.caas–4BS) was identified under both low and high P conditions, explaining 8.1 to 17.7% of the phenotypic variances for RDW. A pleiotropic QTL on chromosome 4DS in interval AX–109816583 - AX–109478820 (16.64 - 30.66 Mb) detected in the low P treatment for RDW co-located with 3 QTLs in all three P treatments for RRS explained 7.1 to 20.4% of the phenotypic variances (Table 2).
QTL clusters
Fifteen QTLs in seven clusters (C1 to C7) for different traits in all three P treatments were identified in the same or close marker intervals (Fig. 3). QTL clusters were located on 3DL, 4BS, 4DS, 6BL, 7AS, 7AL and 7BS chromosomes, respectively. Of these, C1 was on 3DL for SDW and RTN under high P treatment. C2 for RV and RRS on 4BS was detected in a 0.31 Mb interval between SNPs AX–109491270 (21.79 Mb) and AX–108815849 (21.42 Mb). C3 on 4DS comprised QTLs for RDW and RRS. QTL in clusters on 4B (16.64 - 30.66 Mb) and 4D (32.42 - 37.86 Mb) were at or very near to reduced plant height gene loci Rht-B1 (30.8Mb) and Rht-D1 (18.9Mb), respectively. C4, C6 and C7 involved the same traits (RL and RTN) on chromosomes 6BL, 7AL and 7BL. Joint QTL analysis reveal that, QTLs presented in four clusters. i.e. C3 (RDW and RRS), C4 (RL and RTN), C5 (RL, RTN, ROSA) and C7 (RL and RTN) were identified as pleiotropic QTL under different P levels (Fig. 3). C5 on 7AS (AX–109955164-AX–109445593) affected RL, RTN and ROSA, with the favorable allele contributed by Zhongmai 895 (Fig. 3).