Abstract: 【Objective】The objective of this research is to study the relationships between root morphology, nitrogen (N) absorption and utilization and grain yield in rice with different N use efficiencies, clarify the root morphology characters of high yielding, and put forward the corresponding optimized water-N management practice. 【Method】 In 2013, high N use efficiency (HN) variety Chuannongyou 498 and low N use efficiency (LN) variety Chuanyou 6203 were used as trial materials to conduct a split-plot experiment with different panicle fertilizer ratio. In 2014, a split-split plot experiment was performed with the same varieties, where seedling number per hole was assigned to the split plot, and the proportion of spikelet preserving fertilizer and spikelet promoting fertilizer was assigned to the split-split plot. In 2015, a split-split plot experiment was carried out with water management practice as split plot and N-management practice as split-split plot, where HN variety Dexiang 4103 and LN variety Yixiang 3724 were used as trial materials. Grain yield, root morphology characters, and N absorption and utilization characters, obtained from the above 3 field experiments, were performed multiple collinearity diagnostics through calculating the variance inflation factor. After confirming that there were serious multiple collinearity problems, ridge regression analysis was used as the main data analysis method to investigate the relationships between root morphology characters, N absorption and utilization and grain yield. Based on the recurrence of these relationships from 2013 to 2015, common regularities hidden among these relationships were located. 【Result】 The results were as follows: (1) For HN varieties, the determination coefficients of ridge regression equation between grain yield and N accumulation (NA), N dry matter production efficiency (NDMPE), and dry matter weight (DW) at elongation stage, heading stage, and maturity stage, were in the range of 0.0219-0.3961, 0.0452-0.1379, and 0.0914-0.6694, respectively. For LN varieties, the determination coefficients were in the range of 0.0084-0.6190, 0.1224-0.4341, and 0.0818-0.4881, respectively. The relationship between grain yield and N uptake and utilization was less repeatable during the year, and there was no obvious common regularity. Ridge regression analysis was performed with 11 root morphology characters of total root dry weight, adventitious root number, length, surface area, volume, coarse lateral root length, surface area, volume, and fine lateral root length, surface area, volume as independent variables and NA, NDMPE, and DW as dependent variables, respectively. For HN varieties, the determination coefficients were in the range of 0.0527-0.2728, 0.0653-0.3139, and 0.0714-0.3158, respectively. For LN varieties, the determination coefficients were in the range of 0.0607-0.5040, 0.0555-0.4411, and 0.0724-0.5449, respectively. There were significant differences in the relationships between N absorption and utilization and root morphology among years. From 2013 to 2015, the determination coefficients of ridge regression equation between grain yield and root morphology characters of both HN varieties and LN varieties all exceeded 0.8 (P＜0.001), which showed an common regularity that root morphology at heading stage had a significant effect on grain yield. (2) In the ridge regression equation with root morphology characters as independent variables, and grain yield as dependent variable, the coarse lateral root length of HN varieties obtained the largest standard coefficient, and had the greatest impact on grain yield. However, for the LN varieties, the fine lateral root surface area played the most important role in increasing grain yield. (3) With N rate of 150 kg·hm^-2, the optimal N management practice (panicle fertilizer ratio was 40% and the ratio of spikelet preserving fertilizer to panicle fertilizer reached or exceeded 50%) combined with dry-wet alternate irrigation was beneficial to optimize the root morphology. For the SPAD-diagnosis N management practice, N rate reduced from 150 kg·hm^-2 to 120 kg·hm^-2, conventional irrigation mode was more favorable to root morphology optimization than dry-wet alternate irrigation.【Conclusion】Rice root morphology at heading stage had close relationship with grain yield, and appropriate water-N management practice could optimize rice morphology to increase grain yield. SPAD-diagnosis N management practice coupled with conventional irrigation or optimal N management combined with dry-wet alternate irrigation mode could extend the coarse lateral root length of HN varieties as well as the fine lateral root surface area of LN varieties at the heading stage to increase their grain yields, respectively.

Key words: N use efficiency, rice, root morphology, grain yield, ridge regression analysis