The hydroponic experiment was conducted during the rice growth season in 2020 at the experimental station of Huazhong Agricultural University, Wuhan, China (30°29’ N, 114°22’ E), using two rice (Oryza sativa L.) varieties, Nonken 57 (NK57, Japonica rice) and Yangdao 6 (YD6, Indica rice). Compared with NK57, YD6 had significantly higher nitrogen uptake efficiency in the previous study (Zhu et al., 2015).

Rice seeds were disinfected in 10% H₂O₂ solution for 10 min and sown in plastic seeding trays with loam soil after breaking dormancy at 37 °C for 48 h. The seedlings with 3 visible leaves were transplanted to a 12 L plastic pot (25.5 cm in height, 24.4 cm in top diameter), eight seedlings were planted in each pot, which contained a hydroponic nutrient solution with two nitrogen levels i.e., high nitrogen concentration (40 mg L^-1) and low nitrogen concentration (5 mg L^-1) in the form of (NH₄)₂SO₄. The other nutrient composition of the solutions followed the description by Yang et al. (2012): P as 10 mg L^-1 NaH₂PO₄, K as 40 mg L^-1 K₂SO₄, Ca as 40 mg L^-1 CaCl₂, Mg as 40 mg L^-1 MgSO₄, Fe as 2.0 mg L^-1 Fe-EDTA, Mn as 0.5 mg L^-1 MnCl₂·4H₂O, Mo as 0.05 mg L^-1 (NH₄)₆Mo₇O₂₄·4H₂O, B as 0.2 mg L^-1 H₃BO₃, Zn as 0.01 mg L^-1 ZnSO₄·7H₂O, Cu as 0.01 mg L^-1 CuSO₄·5H₂O, and Si as 2.8 mg L^-1 Na₂SiO₃·9H₂O. Each nitrogen treatment and each variety were repeated three times, and nine pots were used for each of the four combinations across two varieties and two nitrogen treatments, six pots for each repetition. The nutrient solution was replaced every 4 days and the pH was adjusted to 5.50 ± 0.1 every day with 1 mol L^-1 HCl or 1 mol L^-1 NaOH. The uniform rice seedlings were sampled for measurements after 24 days following nitrogen treatment.

To investigate the relationships of root characteristics with rice grain yield formation and nitrogen accumulation at late growth stages, a soil pot experiment was conducted using the same varieties in the experiment 1. On the 24^th day after seed sowing, three uniform rice seedlings were transplanted to a 12 L plastic pot (25.5 cm in height, 24.4 cm in top diameter) filled with 10 kg sandy loam soil. The properties of the soil were as follows: 0.99 g total nitrogen kg^-1, 113.7 mg exchangeable potassium kg^-1, 8.11 mg Olsen phosphorous kg^-1, pH 6.60, and 10.50 g organic matter kg⁻¹. Two nitrogen application rates were used, high nitrogen (2.16 g nitrogen each pot) and low nitrogen (0.27 g nitrogen each pot) in the form of urea, which were spit-applied with three times, 40% of total nitrogen as basal fertilizer (before transplanting), 30% at tilling stage (12 days after transplanting) and 30% at panicle initiation stage. A total of 1.5 g phosphorous and 1.5 g potassium were applied for each pot as basal fertilizers in the form of calcium superphosphate and potassium chloride, respectively. The basal fertilizers in each pot were thoroughly mixed with the sandy loam soil before transplanting, the topdressing nitrogen fertilizers were dissolved in water and applied by an irrigating manner. Each nitrogen treatment was repeated thrice. During entire growth period, the rice plants were well watered with a shallow water layer of 1-2 cm, the weeds, pests and diseases were well controlled.

As described by Fu et al. (2014), root lignin histological staining and observation were performed. A razor blade was used to cut thin root slices from 12.5%, 25%, and 50% lengths of the newly formed adventitious roots (8-12 cm) from the root tip in experiment 1. These slices were soaked in 10 g L^-1 phloroglucinol solution in 95% ethanol for 5 min and then rinsed with distilled water. The root slice was then acidified in 30% (w/w) hydrochloride solution for 1 min, and rinsed with distilled water again. After sealing by 50% (w/w) glycerol, the root slice was placed on a glass slide with a drop of water for microscopic observation and photographing using Nikon Ti inverted fluorescence microscope (Nikon-Ti-SR, Japan) at 4× magnification. The obtained images were used to observe the lignin histological distribution.

Root lignin content was measured using ultraviolet spectrophotometry, as described by Ren et al. (2015) with minor modifications. All roots of three rice seedlings were collected and oven-dried in experiment 1, followed by grinding to a fine powder using a ball mill (MM301, Retsch GmbH, Haan, Germany). The lignin was extracted from 10 mg of powder sample with 10 ml of 80% ethanol (v/v) under 50°C for 20 min, followed by centrifugation at 4024 g for 10 min, and removal of the supernatant. The extraction was repeated two times. The solid residue in the tube was collected and oven-dried to a constant weight at 80°C.

All of the oven-dried residues were transferred to a 2 ml plastic tube filled with 0.6 ml of 25% (v/v) acetyl bromide in acetic acid and 24 μl perchloric acid, which was then homogeneously mixed and incubated in an 80°C water bath for 60 min. The reaction mixture was added 0.6 ml of 2 mol L^-1 NaOH to terminate the reaction and centrifuged at 8049 g for 10 min. Then 50 μl of supernatant was sampled and adjusted with acetic acid to 1.2 ml. The absorbance of the resulting solution was measured at 280 nm using an ultraviolet spectrophotometer (U-3900 UV-VIS spectrophotometer, Hitachi, Tokyo, Japan). Six dosages (from 1-6 mg) of pure lignin (Yuanye Bio-Technology Company, Shanghai, China) were sampled and performed in the same procedure to obtain a standard curve. The root lignin amount (mg) of the test powder sample was computed from the standard curve of lignin. The root lignin content (%) was expressed as the ratio of the lignin amount to the weight of the test sample.

Full-length fresh roots (~8-12 cm) of five rice seedlings were used to extract total RNA in experiment 1, using Trizol Reagent (Sigma, USA) according to the manufacturer’s instructions. The concentration and purity were determined using Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). After removing genomic DNA from RNA extract with DNase I (Invitrogen, USA), the first strand cDNA was synthesized with M-MLV reverse transcriptase (Invitrogen, USA) according to the manufacturer’s protocol. After a 10-fold dilution of the synthesized cDNA solution, quantitative real-time PCR was performed on ABI ViiATM7 real-time PCR system (Applied Biosystems, USA), using a Roche fast start universal SYBR Green master (Rox) kit (Roche, Switzerland). The procedure was 95°C for 2 min, followed by 45 cycles at 95°C for 10 s, 60°C for 10 s, and 72°C for 20 s. The housekeeping rice gene Actin was used as the internal control, and the expression analyses of ammonium transporter genes were carried out using the gene-specific primers listed in Table S5 (Gaur et al., 2012) The relative quantification of gene expression was calculated using the 2^−ΔΔCT method (Livak and Schmittgen, 2001), and the average expression level of OsAMT1;1 gene in YD6 across three biological replicates under LN was set as 1.

The values for the investigated traits were expressed as means of three replicates, and were compared between the two nitrogen treatments for an identical variety and between the two varieties under an identical nitrogen treatment based on the least significant difference (LSD) test at a 5% probability level, using Statistix 9 software package (Analytical Software, Tallahassee, FL, USA). The Sigmaplot 10.0 software package (SPSS Inc., Chicago, IL, USA) was utilized for plotting and linear correlation analyses. In the study, the three traits (total nitrogen accumulation, net NH₄ ⁺ influx, and NH₄ ⁺ uptake via apoplasmic pathway) were considered as nitrogen absorption traits.

As compared to HN, LN significantly decreased nitrogen uptake in both varieties ( Table 1 ). In experiment 1, total nitrogen accumulation, net NH₄ ⁺ influx on the root surface, and NH₄ ⁺ uptake via apoplasmic pathway decreased by 75.2%, 73.1% and 90.5% in NK57, 71.2%, 66.3% and 84.3% in YD6 under LN in comparison to HN, respectively ( Table 1 ; Figure S1 ). YD6 had higher performances in total nitrogen accumulation and net NH₄ ⁺ influxes than NK57 under both HN and LN. The two varieties had similar NH₄ ⁺ uptake via the apoplasmic pathway under HN, however, YD6 had significantly higher value than NK57 under LN ( Table 1 ).

In experiment 2, total nitrogen accumulations at heading and maturity stages decreased by 53.9% and 52.8% in NK57, 44.6% and 38.7% in YD6 under LN compared to HN ( Table 1 ). No significant differences in total nitrogen accumulations were observed between the two varieties at heading and maturity stages under HN, however, YD6 exhibited higher total nitrogen accumulations than NK57 at the two stages under LN ( Table 1 ).

Under both LN and HN in experiment 1, YD6 exhibited higher dry weights of roots, stems, leaves, and total dry weight than NK57 ( Tables 2 ; S1 ; Figure S2 ), while no significant difference was observed in root, stem, and leaf nitrogen concentrations between YD6 and NK57 under HN and LN, respectively ( Table S2 ).

In experiment 2, root dry weight and total dry weight at heading stage decreased by 36.6% and 27.7% in NK57, 30.4% and 22.3% in YD6 under LN, compared to HN ( Table 2 ). In addition, root dry weight, total dry weight and grain yield at maturity stage decreased by 16.5%, 34.4% and 30.0% in NK57, 11.8%, 24.7% and 23.5% in YD6 under LN in comparison with HN, respectively ( Table 2 ). At heading stage, YD6 had higher root dry weight and total dry weight than NK57 under both LN and HN ( Table 2 ). At maturity stage, YD6 exhibited the similar grain yield, total dry weight and grain nitrogen concentration with NK57 under HN, however, these three parameters were significantly higher in YD6 than those in NK57 under LN ( Table 2 ; Table S2 ).

In NK57 and YD6, total root length, root surface area, root volume, and root number decreased at seedling stage ( Figures 1A–D ) and heading stage ( Figures 1F–I ) under LN, as compared to those under HN. Contrarily, LN significantly increased single root length at the two stages ( Figures 1E, J ; S2 ). YD6 exhibited larger phenotypic values in the five root traits at the two stages under LN, in comparison with NK57 ( Figure 1 ). Additionally, in experiment 1, total root length, root surface area, root volume, and single root length of YD6 were 17.8%, 19.5%, 7.3%, 37.5% higher than those of NK57 under HN, respectively; however, the four parameters of YD6 were 32.0%, 34.8%, 27.2%, and 49.4% higher than those of NK57 under LN ( Figures 1A–C, E ). In experiment 2, no significant difference was observed in total root length, root surface area, root volume between NK57 and YD6 under HN, however, these three parameters were significantly higher in YD6 than those in NK57 under LN ( Figures 1F–H ). The single root length in YD6 was 13.6% and 21.1% higher than that in NK57 under HN and LN, respectively ( Figure 1J ). In both experiments 1 and 2, YD6 had lower root number than NK57 under HN and LN ( Figures 1D, I ).

In NK57 and YD6, there was no significant difference in stele area, root diameter, stele diameter, and number of xylem vessels between LN and HN in experiment 1, respectively ( Table 3 ; Figure S3 ). In experiment 2, compared to HN, LN significantly decreased root diameter and stele diameter in NK57, but had no effects on the two parameters in YD6 ( Table 3 ). In both experiments 1 and 2, living cortical area under LN was similar to that under HN in YD6, however, LN significantly decreased the living cortical area in NK57. The proportion of RCA increased in NK57 under LN but not significant in YD6 in experiment 1, and increased in the two varieties in experiment 2 ( Table 3 ). YD6 had higher living cortical area, stele area, root diameter, stele diameter, and number of xylem vessels than NK57 under both HN and LN in experiments 1 and 2 ( Table 3 ; Figure S3 ). Most of these differences were significant.

As compared to HN, enhanced lignin depositions were observed in root sclerenchyma, endodermis cells, and xylem in both the varieties under LN ( Figure S4A ). However, there was no visible difference between YD6 and NK57 under LN and HN ( Figure S4A ). Root lignin content increased by 24.7% in NK57 and 20.5% in YD6 under LN in comparison with those under HN ( Figure S4B ); however, the values showed no significant difference between two varieties under both LN and HN.

In NK57 and YD6, the expression level of OsAMT1;1, OsAMT1;2 genes were significantly downregulated under LN ( Figures 2A, B ). Compared to NK57, YD6 exhibited significantly higher expression levels of OsAMT1;1, OsAMT1;2, OsAMT1;3 genes under LN ( Figure 2A ). However, no significant difference was observed between NK57 and YD6 under HN ( Figure 2B ).

At seedling stage in experiment 1, total root length, root surface area, root volume and single root length were positively and significantly correlated with total nitrogen accumulation and net NH₄ ⁺ influx under both LN and HN, however, root volume was not significantly correlated with the two nitrogen absorption traits under HN ( Table 4 ). The four root traits were positively and significantly correlated with NH₄ ⁺ uptake via apoplasmic pathway under LN. Root number were negatively and significantly correlated with the three nitrogen absorption traits but not with NH₄ ⁺ uptake via apoplasmic pathway under LN. Similarly, the five root traits mentioned above were significantly correlated with total nitrogen accumulation at heading and maturity stages only under LN in experiment 2 ( Table 4 ).

In experiment 2, grain yield was positively and significantly correlated with total root length, root surface area, root volume, and single root length at heading and maturity stages under both HN and LN but not with total root length and single root length under LN, and negatively and significantly with root number under LN ( Table 4 ).

In experiment 1, significant and positive correlations were observed among the three nitrogen absorption traits under LN and HN but correlations of NH₄ ⁺ uptake via apoplasmic pathway with total nitrogen accumulation and net NH₄ ⁺ influx under HN ( Table 5 ). Living cortical area, stele area, root diameter, stele diameter, and xylem vessel number were positively correlated with the three nitrogen absorption traits under LN and HN, of which most correlations were significant ( Table 5 ). The proportion of root aerenchyma was significantly and negatively correlated with the three nitrogen absorption traits under LN and HN but not with NH₄ ⁺ uptake via apoplasmic pathway under HN ( Table 5 ).

In experiment 2, total nitrogen accumulation at heading and maturity stages was significantly and positively correlated with the six root anatomical traits under LN, no correlations were observed under HN ( Table 5 ). Generally, grain yield was positively correlated with total nitrogen accumulation at heading and maturity stage, with the six root anatomical traits but proportion of root aerenchyma (negative correlation) under LN, similar correlations were found under HN but not significance.

The expression of OsAMT1;1, OsAMT1;2, and OsAMT1;3 genes, which were significantly different between the two varieties, were significantly and positively correlated with the three nitrogen absorption traits under LN ( Figures 3A–I ), however, the correlation between the expression of OsAMT1;1 and total nitrogen accumulation was insignificant ( Figure 3A ). Nevertheless, the expression of the three genes were unrelated to the three nitrogen absorption traits under HN ( Figures 3A–I ).

In the study, YD6 had larger root system than NK57 at seedling and heading stages ( Figures 1 ; S2 ; Table 2 ); accordingly, the variety had higher nitrogen absorption capacity under LN and HN, and a smaller reduction in total nitrogen accumulation (71.6% in YD6 vs.75.2% in NK57) and influx (66.3% vs.73.1%) under LN ( Table 1 ). Previous investigations reported that rice varieties with high nitrogen uptake often exhibit large root system, reflected by high root dry weight, total root length, root surface area, and single root length (Ju et al., 2015; Wei et al., 2018; Chen et al., 2020). These results suggest that high root uptake in YD6 was attributed to the large root system, supported by the positive correlation of total nitrogen accumulation with most root traits, especially under LN ( Table 4 ). On the other hand, we observed the tight positive association of four root morphological traits with grain yield, especially under LN ( Table 4 ). Similarly, Liu et al. (2014) found that genetic improvement significantly increased root dry weight, root length and root surface area, which benefited shoot growth and grain yield in rice varieties. Therefore, for green rice production with reduced nitrogen application, the development of varieties with efficient nitrogen uptake should be focused on improving root system.

YD6 had thicker roots compared to NK57, accompanied by higher nitrogen accumulation and a smaller reduction in total nitrogen accumulation at seedling stage ( Tables 3 ; 1 ). Similarly, wheat and maize plants with large root diameter had high nitrogen uptake, which accounted for better shoot growth under normal and low nitrogen applications (Yang et al., 2019; Zhang et al., 2020; Schneider et al., 2021). These observations were supported by the positive correlations of root diameter with nitrogen uptake ( Table 5 ). Additionally, root diameter-related anatomical characteristics (such as root cortical area, stele area, stele diameter and the number of xylem vessels) were also higher in YD6 than those in NK57 ( Table 3 ), and were generally positively correlated with the three nitrogen absorption traits ( Table 5 ). In addition, the study observed that tight positive correlations of grain yield with root diameter and stele diameter, especially under LN ( Table 5 ). Similarly, compared with old varieties, the modern super rice varieties had higher root diameter, which contributed to their better shoot growth and higher grain yield under same nitrogen application (Liu et al., 2014). Thus, these results suggest that thick roots may be conductive to nitrogen uptake in rice, increasing root diameter should be considered for breeding elite rice varieties suitable for green rice production.

Compared with NK57, YD6 had larger root stele area and more xylem vessels ( Table 3 ; Figure S3 ) and smaller reductions in total nitrogen accumulation and NH₄ ⁺ influx, and the two traits were positively correlated with the three nitrogen absorption traits, especially under LN ( Table 5 ). Similarly, Kadam et al. (2015) and Kong et al. (2017) reported that plant roots with large stele and xylem areas had high water and nutrient transport capacity. Most current maize cultivars have more xylem vessels and are more resistant to low nitrogen stress than earlier variants (York et al., 2015). These results strongly suggest that large root stele area and xylem vessel number may be beneficial for nitrogen uptake in rice, especially under LN. Additionally, our study observed that tight positive correlations of grain yield with these root anatomical characteristics, especially under LN ( Table 5 ). Similarly, rice genotypes with larger root stele and root xylem vessels exhibited better growth and higher grain yield (Jeong et al., 2013). Thus, selecting rice varieties with large root stele area and xylem vessel number may also have great significance for green rice production with reduced nitrogen application.

It’s noticeable that compared to HN, LN significantly repressed the root growth in two varieties ( Figures 1A–I ; Table 2 ); however, single root length significantly increased in both varieties under LN ( Figures 1E, J ). Therefore, the reduction of total root length, root surface area, and root volume under LN was mainly attributed to the reduction in root number. Similar results were reported in previous studies (Saengwilai et al., 2014; Luo et al., 2022). Noticeably, root number reduced and negatively associated with nitrogen absorption, dry weight and grain yield, especially under LN ( Figure 1 ; Tables 4 ; S6 ). Under nitrogen-deficient conditions, less adventitious roots may reduce the competition with each other for nutrients and decrease root metabolic cost, which was beneficial to maintaining growth and development of the whole plant (Giehl and von Wirén, 2014; Saengwilai et al., 2014; Gao et al., 2015). Moreover, genotypes with small root number had more accessible resources to develop longer and deeper roots, and also had higher nitrogen accumulation and grain yield under LN (Saengwilai et al., 2014; Schneider et al., 2021). In line with these observations, the single root length increased under LN, and exhibited favorable association with the three nitrogen absorption traits and grain yield ( Figure 1 ; Table 4 ). Previous studies also reported that long and deep root system made it easier to obtain more nitrogen and was conducive to nitrogen accumulation and grain yield formation in rice under LN (Arai-Sanoh et al., 2015; Ju et al., 2015; Chu et al., 2018; Chen et al., 2020). Therefore, these results together suggest that less root number and enhanced root elongation in responses to LN are favorable for nitrogen absorption, developing rice varieties with long and deep roots could promote the coordination of high yield and low nitrogen input for green rice production.

In comparison with HN, LN treatment increased lignin deposition ( Figure S4 ). Ren et al. (2015) also reported similar observation, and many genes involved in lignin biosynthesis were upregulated in rice seedlings under nitrogen deficiency (Shao et al., 2020). The three nitrogen absorption characteristics sharply decreased under LN, and were significant correlations with lignin content across the two nitrogen treatments ( Figure S6 ). Yang et al. (2022) found that the increased root lignification under a high NH₄ ⁺/NO₃ ^- ratio resulted in poor nitrogen uptake in wheat. For the possible cause, the enhanced root lignin content may incur the formation of root apoplasmic barriers, and inevitably impedes water and nitrogen uptake via the apoplasmic pathway (Ren et al., 2015; Huang et al., 2019). Therefore, the elevated root lignin content induced by LN should partially account for the decreases nitrogen uptake in the two varieties.

In the study, LN promoted RCA formation and reduced the living cortical area in both varieties ( Table 3 ; Figure S3 ), the observation was consistent with Abiko and Obara (2014). Additionally, RCA was negatively correlated with nitrogen absorption, dry weight, and grain yield, especially under LN ( Table 5 ; Table S7 ). The RCA formation reduced the number and area of living cortical cells and seemingly limited nutrient transport via the cell-to-cell pathway (Schneider et al., 2017). Additionally, the enhanced RCA formation resulted in long and tortuous routes and fewer intercellular routes, thereby erecting greater effects on the apoplasmic pathway than on the cell-to-cell pathway in maize roots (Hu et al., 2014). Previous study showed that genotypes with high nitrogen use efficiency and yield had less RCA formation compared with ones with low efficiency and yield under LN, and relieved the stress of nitrogen deficit (Yang et al., 2019). These findings imply that the enhanced RCA induced by LN may hinder nitrogen transport via the apoplasmic pathway, and may be also responsible for low nitrogen accumulation and low dry matter accumulation and grain yield in the study.

In both varieties, LN significantly decreased NH₄ ⁺ uptake via the apoplasmic pathway ( Table 1 ), and also decreased the expression of OsAMT1;1, OsAMT1;2 genes ( Figure 2A ). Similarly, Sonada et al. (2003) and Li et al. (2016) also reported the down-regulation of the expression of the two genes in rice root epidermis and stele under LN. Yuan et al. (2007) showed that AMT1;2 mediated NH₄ ⁺ uptake via apoplasmic pathway. On the other hand, YD6 showed high nitrogen accumulation, high NH₄ ⁺ uptake via apoplasmic pathway, and high expression of OsAMT1;1, OsAMT1;2, OsAMT1;3 genes under LN ( Table 1 ; Figure 2A ), and the expression of the three genes were positively correlated with NH₄ ⁺ uptake via apoplasmic pathway ( Figures 3C, F, I ). These findings strongly imply that the expression of OsAMT genes (especially OsAMT1;1, OsAMT1;2 genes) is tightly associated with NH₄ ⁺ uptake transport in the root apoplasmic pathway under LN. Under LN, high expression of OsAMT genes enhanced root nitrogen uptake, which further increased leaf photosynthetic rate and promoted grain yield formation in rice (Ranathunge et al., 2014; Ferreira et al., 2015). Therefore, it should be possible approach to develop rice variety with high expression of ammonium transporter for green rice production with nitrogen reducing application.

As compared to NK57, YD6 has higher nitrogen uptake under LN and smaller decrease in total nitrogen accumulation under LN at the three stages ( Table 1 ), companying higher net NH₄ ⁺ influx and significant positive correlations among the three nitrogen absorption traits at seedling stage, especially under LN ( Tables 1 ; 5 ; Figure S1 ). Previous research observed the strong linkage between high NH₄ ⁺ uptake via the apoplasmic pathway and high nitrogen accumulation in plants (Yuan et al., 2007). These results suggest that high NH₄ ⁺ uptake via the apoplasmic pathway may also partially contribute to high nitrogen uptake in YD6 under LN ( Figure 4 ).

YD6 had larger root system relative to NK57, especially under LN ( Figures 1A–J ; Table 2 ), as well as lower reduction degree in total root length, root surface area and root volume at seedling and heading stages) but higher degree increase in single root length. Previous studies also showed that rice varieties with higher nitrogen accumulation had greater the former three root traits under LN than varieties with lower nitrogen accumulation under LN (Ju et al., 2015). These root characteristics were positively correlated with the three nitrogen absorption traits, especially under LN ( Table 4 ). Thus, high nitrogen uptake in YD6 should be attributed to the large and deep root system due to the small degree of reduction in root traits under LN ( Figure 4 ). In addition, root physiological characteristics such as root vigorous activity (e.g. oxidation activity) and root hormones are closely related to nitrogen uptake. Nitrogen efficient rice varieties with large and deep root also had high root oxidation activity (Ju et al., 2015), vigorous roots produced high contents of cytokinins and indole-3-acetic acid under LN, which were advantageous to delaying plant senescence and maintaining plant growth (Arai-Sanoh et al., 2015; Wu et al., 2019). On the other hand, high contents of cytokinins and indole-3-acetic acid enhanced root nitrogen uptake by altering root system architecture and up-regulating the expression of nitrogen transporter genes under LN (Wu et al., 2019; Luo et al., 2022). Thus, the high nitrogen uptake in YD6 may also be attributed to its high root physiological characteristics, the mechanism remains to be further clarified.

In the study, YD6 had higher expression of root OsAMT1;1, OsAMT1;2, OsAMT1;3 genes than NK57 under LN ( Figure 2A ), and YD6 also had smaller degree reductions in the expression of OsAMT1;1 and OsAMT1;2 under LN, but larger increase in OsAMT1;3 expression ( Figure 2A ). Similarly, Shi et al. (2010) found that rice varieties with high nitrogen uptake exhibited upregulated expression of root OsAMT1;1 gene under LN, LN boosted the expression of OsAMT1;3 in low-nitrogen tolerant rice varieties but not in low nitrogen-intolerant rice varieties (Gaur et al., 2012), transgenic rice plants with over-expression of OsAMT1;3 showed stronger resistance to low nitrogen condition (Ferreira et al., 2015). Additionally, the expression of the three genes was tightly and favorably related with nitrogen uptake under LN ( Figures 3A–I ). The absence of gene AMT1;2 hindered NH₄ ⁺ uptake via the apoplasmic pathway in Arabidopsis roots (Yuan et al., 2007). Therefore, high expression of OsAMT genes should contribute to higher nitrogen uptake in YD6 under LN ( Figures 3A–I ; 4 ).

In comparison with NK57, YD6 exhibited lower RCA and high root living cortical area under LN and HN, and had smaller degree decrease in root living cortical area under LN ( Table 3 ; Figure S3 ). Additionally, RCA and root living cortical area adversely affected nitrogen uptake, especially under LN ( Table 4 ). Yang et al. (2019) reported that maize genotypes exhibiting high tolerance to low-nitrogen application had less RCA under LN, Hu et al. (2014) and Schneider et al. (2017) also reported that larger RCA and less living cortical area may diminish the intercellular routes for nutrient transport via the apoplasmic pathway in maize and wheat roots. These results together suggest that high living cortical area may partially explain high nitrogen accumulation in YD6 under LN ( Figure 4 ).

Compared to NK57, YD6 had smaller reduction degree in grain yield under LN ( Table 2 ), this was also accompanied by smaller reduction degree in total root length, root surface area, root volume ( Figures 1F–H ) and higher increase degree in single root length under LN ( Figure 1J ). Additionally, YD6 maintained higher root living cortical area, root diameter and stele diameter but root number and RCA area under LN compared with NK57 ( Figure 1 ; Table 3 ). Similarly, Ju et al. (2015) reported that nitrogen-efficient rice varieties maintained higher root biomass, total root length and deeper roots than nitrogen-inefficient rice varieties under LN, which counted for more grain yield under LN. For possible explanation, deep root system produced high cytokinin flux transported to aboveground plants, which promoted rice grain yield formation under LN (Arai-Sanoh et al., 2015), Wu et al. (2016) found that more root cytokinins from roots to aboveground plants contributed the small reduction of spikelets per panicle under heat stress. In addition, wheat varieties with high grain yield showed smaller reduction degree in root diameter at late growth stage than varieties with low grain yield under LN (Zhang et al., 2020). These results totally suggest that maintaining high performances of several root traits is also beneficial for promoting rice yield formation under LN ( Tables 2 , 3 ; Figures 1 , 4 ; Table S3 ), developing varieties with positive root responses to nitrogen reduction should be a considerable target for green rice production in the future.