Root system architecture modulates nitrogen uptake and use efficiency, influencing grain yield in rice

Abstract

Nitrogen (N) is essential for rice productivity, yet excessive fertilization increases costs and environmental risks, making improvement of nitrogen use efficiency (NUE) a key breeding target. We evaluated 99 rice genotypes to investigate relationships between root system architecture (RSA) and NUE using three complementary experiments. In hydroponics under full (N100) and half N (N50), ImageJ-SmartRoot analysis revealed substantial variation in RSA. Principal component analysis identified 12 contrasting genotypes with vigorous and weak RSA, which were further evaluated under pot and field conditions. Under pot conditions, DJ123 and KSK-133 maintained the highest grain yields under reduced nitrogen, whereas IR30 and TKM6 showed the lowest yields. Thousand grain weight (TGW) under the same treatment was also highest in KSK-133 (15.8 g) and DJ123 (15.3 g), compared with IR30 (6.6 g) and TKM6 (7.6 g). Field results confirmed this trend, with TGW ranging from 23.9 g in KSK-133 to only 11.4 g in TKM6 under N50. Ammonium nutrition improved SPAD, biomass, and yield by up to 7.65%, 25.3%, and 27.2%, respectively, compared with nitrate. Efficient genotypes maintained superior performance under reduced N, with DJ123 and KSK-133 showing up to 10.3% and 43.2% increases in nitrate reductase activity. Overall, these genotypes represent promising donor parents and can serve as valuable genetic resources for breeding rice with enhanced NUE and stable yield under low-N conditions.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-50828-0.

Introduction

Ensuring global food security is increasingly challenging as the world’s population is projected to reach 9.7 billion by 2050. This demand is further intensified by land degradation, urban expansion, and water scarcity. As a primary staple for over half the global population, rice (Oryza sativa L.) plays a pivotal role in achieving food security¹. Nitrogen (N) fertilizers are often applied excessively to maximize yield while N use efficiency (NUE) remains low, contributing to economic and environmental concerns². In flooded rice systems, N is primarily available as ammonium, which is preferable for uptake, while nitrate offers complementary benefits by enhancing N transport, activating aquaporins, and buffering cytosolic pH^3,4. Root development is intricately influenced by ammonium and nitrate. Ammonium promotes lateral root (LR) branching, while nitrates stimulate LR elongation^5,6. Together, they optimize root morphology, enabling efficient nutrient acquisition. Branched LRs capture immobile ammonium in shallow soil, whereas elongated LRs access mobile nitrate in deeper layers. Simulating the seasonal flux of ammonium under N starvation in the nutrient solution showed that varieties better suited to low N conditions exhibit greater root morphological plasticity, which aids in improving N uptake⁷. Although ammonium is the preferred form of N over nitrate for rice, root system architecture (RSA) is strengthened by partial nitrate nutrition⁸. Root morphological plasticity enables adaptation to fluctuating N availability, and early-stage RSA traits may serve as predictive indicators of NUE^9–11. Despite the critical role of RSA in determining N efficiency, its potential remains underutilized in breeding programs. Genetic variation in NUE among rice germplasms presents an opportunity to breed high-efficiency cultivars. Genes such as NPF6.3 and AMT1;3 function as both transporters and sensors, modulating root growth in response to external N cues^5,12. Key enzymes involved in nitrate assimilation include nitrate reductase (NR), nitrite reductase, glutamine synthetase (GS), and glutamate synthase (Glutamine oxoglutarate aminotransferase; GOGAT). NR catalyzes the reduction of nitrate to nitrite, regulated by environmental factors like N availability and light¹³. Efficient nitrate assimilation underpins sustainable agriculture by improving NUE and reducing fertilizer dependence. Glutamine synthetase (GS) is central to this process, assimilating ammonium into glutamine via two isoforms: GS1 (cytosolic, critical for growth and grain filling) and GS2 (chloroplastic, which recycles ammonia from photorespiration and photosynthesis). While most plants have a single GS2 gene, species like soybean and alfalfa possess multiple copies¹⁴. While previous studies have explored the relationship between RSA and NUE in rice with a smaller number of genotypes^15,16, a comprehensive assessment across large, genetically diverse panels remains limited. To address this gap, the present study aimed to (i) screen a diverse panel of 99 rice genotypes for associations between RSA and NUE based on seedling-stage root traits, chlorophyll content, and biomass; (ii) validate the performance of contrasting genotypes under pot and field conditions; and (iii) elucidate the physiological and biochemical mechanisms underlying NUE by assessing N metabolism related enzyme activities under contrasting N sources and levels. We hypothesized that genotypes with vigorous RSA would exhibit enhanced N uptake and biomass accumulation and that early RSA traits screening could serve as reliable indicators for identifying NUE-efficient genotypes, considering ammonium and nitrate nitrogenous fertilizers.

Materials and methods

Study 1: Screening germplasm for root system architecture using a hydroponic experimental setup

The study was designed to elaborate on the association of rice root system architecture with enhanced NUE. Given the challenges of studying root traits in soil environments, utilizing a hydroponic system has proven helpful for this research. For these 99 rice genotypes, were screened using variable sources and levels of N. The hydroponic experiment was conducted in the growth room under a 12-hour light/dark cycle with an average temperature of 30 ℃±2. A total of 104 rice genotypes were evaluated, including 7 from the Bengal Asam Aus Panel (BAAP) and 97 from Pakistan (Table S3; Supplementary dataset). Seeds for the BAAP genotypes were sourced from the International Rice Research Institute in the Philippines, while seeds for the Pakistani genotypes were obtained from the Rice Research Institute in Kala Shah Kaku. Out of 104 genotypes, five did not germinate and were excluded, resulting in a final analysis of 99 genotypes. Plastic rectangular-shaped boxes with a 10 L capacity for nutrient solutions were used in this study. Uniform ten-day-old seedlings were transplanted into individual holes in a lid positioned over pots. The nutrient solution was prepared following¹⁷ except for some modifications to achieve the planned treatments. The experiment consisted of four treatments arranged in a completely randomized design (CRD) with a two-factor factorial arrangement and four replications. Each replication corresponded to an individual hydroponic container (box), resulting in a total of 16 boxes (4 treatments × 4 replications). Genotypes (n = 20) were grown within each box (Photo-S1). The plants were grown for 12 days under two N levels and forms: 100% and 50% of the recommended dose, using two N sources i.e., calcium nitrate [Ca (NO₃)₂.4H₂O, Source; Sigma Aldrich] and urea [NH₂CONH₂, Source; Sigma Aldrich]. Nr-N100 and Nr-N50 represent 100% and 50% of the recommended nitrate-N source, whereas Am-N100 and Am-N50 represent 100% and 50% of the recommended ammonium-N source. The N concentrations for the treatment’s sources (calcium nitrate and urea) were maintained at 5 mM for 100% and 2.5 mM for 50% of the recommended level (Table S8). The experimental design was a CRD with a two-factor factorial arrangement. Images of the roots were captured immediately after harvest using a camera (Nikon Coa-PTX L840) within a customized imaging setup. The system comprised a 30 cm × 30 cm enclosed chamber with the camera mounted overhead at a height of 60 cm. For imaging, plants were positioned on a calibrated rigid sheet placed at the base of the setup. RSAIN was calculated by standardizing each root trait using a constant value of 20. For each trait, 20 was divided by its maximum observed value to obtain a scaling factor, which was then multiplied by all trait values. The scaled values were summed to obtain RSAIN for each treatment (Table S9). This method was adopted from^18,19. The images were processed on ImageJ-based SmartRoot software. Scale calibration in ImageJ was performed by dividing the pixel value (displayed at the top of the image) by the maximum value on the reference scale. The resulting value was set as the pixel distance corresponding to 1 cm, with pixel aspect ratio fixed at 1.0 and units defined in centimeters²⁰. The root traits studied were primary root length (PRL), lateral root length (LRL), lateral root number (LRN), total lateral root length (TLRL), total root length (TRL), and lateral root density (LRD) according to²¹. For total plant biomass, the root and shoot samples were dried in an oven for 72 h, and biomass was measured using a digital weighing balance (PA4102, OHAUS corporation, USA). To estimate the relative chlorophyll content, a chlorophyll meter SPAD-502 Konica Minolta was used.

Based on principal component analysis (PCA) (Fig. 3), data from experiment 1, six genotypes from each category (vigorous and weak RSA), common to all treatments, were selected, resulting in a total of twelve genotypes (Table 1).

Fig. 3.

Principal component analysis for the evaluation of rice genotypes (n = 99) based on root system architecture index (RSAIN), chlorophyll content (Chl), and total plant biomass (TPB). The points on the biplot depict the individual response of genotypes for RSAIN, TPB, and Chl under variable N treatments. The variables under different treatments are indicated by black vectors. The treatments are abbreviated as Am-N100: recommended ammonium-Nas urea, Am-N50: half of the recommended ammonium-Nas urea, Nr-N100: recommended nitrate-N as calcium nitrate, and Nr-N50: half of the recommended nitrate-N as calcium nitrate. The biplot differentiates genotypes into two groups: WRSA (Weak Root System Architecture) and VRSA (Vigorous Root System Architecture).

Table 1.

Rice genotypes (n = 12) selected from the hydroponic experiment for the pot experiment.

Selection criteria	Genotypes
Genotypes with vigorous root system architecture (RSA), higher total biomass (TPB), and higher chlorophyll content (SPAD value	Shua 92, DJ123, KSK-133, 98PP-7, BAS 5854, PK BB1
Genotypes with weak root system architecture (RSA), lower total plant biomass (TPB), and lower chlorophyll content (SPAD value).	IR 30, DR-82, TKM6, Hansraj-1, PGRI_NARC_Rice_7692, RC82

Study 2: Comprehensive evaluation of selected genotypes for morphological traits, yield, and nitrogen use efficiency under pot and field conditions

The first experiment presents a detailed evaluation of these twelve selected genotypes under a pot trial, focusing on yield performance and NUE-related traits. The experiment was conducted at the old Botanic Garden, University of Agriculture, during the kharif season of 2021. The plants were grown in earthen pots with 20 kg of soil. Physicochemical analysis of the soil is provided in Table S7 of the supplementary dataset. Urea and calcium nitrate were applied as variable N sources at the recommended rate of 170 kg ha^− 1 (2.55 g N per pot). N was applied in three equal splits: as a basal dose before sowing, at the tillering stage, and at the panicle initiation stage. The recommended rates for phosphorus (P) and potassium (K) were 100 and 80 kg ha^− 1. The sources for P and K were single superphosphate and potassium sulfate. The combinations of N source and levels were represented by four treatments: Am-N100-recommended ammonium-Nas urea, Am-N50-half of recommended ammonium-Nas urea, Nr-N100-recommended nitrate-N as calcium nitrate, and Nr-N50-half of recommended nitrate-N as calcium nitrate. The experiment was arranged in a completely randomized design (CRD) with a two-factor factorial arrangement, where genotypes and treatments were considered as factors. In total, 192 experimental units were used, comprising 12 genotypes, 4 treatments, and 4 replicates. Ten seeds were sown directly in each pot, and after the plant establishment, thinning was done, keeping four plants in each pot. The plants were grown up to maturity, and all standard management practices were maintained for irrigation, insecticide, and pesticide application (Photo-S2).

In the second experiment, the evaluation of these selected genotypes under field conditions. The experiment was conducted at the Research Area, University of Agriculture, Faisalabad, during Kharif 2022. The GPS coordinates of the experimental field are 31.4396° N, 73.0768° E. The treatments were the same as explained in the pot experiment. The experiment was conducted using a randomized complete block design (RCBD) with four replications (Photo-S3). In each block, the four N treatments were randomly assigned to plots. Each experimental plot consisted of 12 rows representing 12 genotypes, with 12 plants per row. Plant-to-plant spacing was maintained at 15 cm, while row-to-row spacing was 20 cm. The plot size was approximately 3.63 m² with a 0.61 m path between each plot to facilitate field operations and data recording. Standard management practices were performed in terms of irrigation to maintain a continuous flooded cultivation method. The treatments were replicated four times. Twelve rice varieties were used, and the nursery was sown on 6th June 2022 and transplanted on 18th July 2022. Before transplantation, plants were dipped in zinc oxide solution to reduce the transplantation shock, and Riffit 500EC herbicide was applied after 10 days of transplantation to control weeds. The plants were harvested on the 17th of October 2022. Weather data of the growing season of rice was collected from the meteorological department of the University of Agriculture, Faisalabad, Pakistan. From May 2022 to October 2022, the monthly mean data of rainfall (mm), Minimum and maximum temperature (ºC) are represented graphically (Figure S1; Supplementary dataset). Physicochemical analysis of soil has also been provided in Supplementary Table S7. For both experiments, the following parameters were taken, which include the number of tillers per pot and the number of tillers per six plants for the field, panicle length (cm), grain yield per pot and per six plants for the field (g), harvest index, and thousand grain weight (g) were measured at the time of harvesting. Harvest index was estimated by using the following formula

.

For the N analysis, the plant samples were separated into three parts, i.e., root, shoot, and grain, while for the field, only shoot and grain N were estimated. Dehusked grains were used. The samples were oven-dried at 65 ℃ for 72 h and ground into fine powder. N content was determined following the standard Kjeldahl method adapted from the ICARDA manual. The 0.5 g ground plant sample was digested with 5 mL concentrated sulphuric acid at 300 ℃ using a hot plate until a clear solution was obtained. No catalyst was used during digestion. After cooling, the digest was diluted to a final volume of 50 mL with distilled water. For distillation, a 5 mL aliquot of the digested sample was mixed with 5 mL NaOH and distilled using a Kjeldahl distillation unit (VELP Scientifica, UDK 126 D). The released ammonia was trapped in 10 mL boric acid solution and subsequently titrated with 0.1 N H₂SO₄. The blank whole procedure of digestion to titration was performed without plant material. Nitrogen use efficiency was estimated based on biomass production. All N use efficiency-related traits have been calculated according to the methods proposed by^22,23.

Nitrogen uptake efficiency (NUpE) was calculated only for the pot trial by using the following formula

.

Study 3: Evaluation of four contrasting genotypes for N assimilation-related enzyme activity

Nitrogen metabolism-related enzyme activity in leaves of four selected genotypes was examined using two sources and doses of N. The rice genotypes are referred to here as IR30 and TKM6 (N use inefficient) and DJ123 and KSK-133 (N use efficient). In this study, N treatments were adjusted to a wider contrast (10 mM vs. 2.5 mM, corresponding to 100% and 25% N) to better capture physiological responses associated with N metabolism in selected contrasting genotypes. The combinations of N source and availability are (i) 20 mM nitrate, (ii) 5 mM nitrate, (iii) 20 mM ammonium, (iv) 5 mM ammonium, abbreviated as Nr-N100, Nr-N25, Am-N100, and Am-N25. The details for the treatment application are provided in Supplementary Table S8. The experiment was laid under hydroponic conditions in a growth room with a completely randomized design with a two-factor factorial arrangement at the National Institute of Biotechnology and Genetic Engineering (NIBGE), Faisalabad, with four replicates per treatment. For nitrate and ammonium sources, calcium nitrate and urea were used, respectively.

Estimation of glutamine synthetase enzyme activity

Glutamine synthetase activity (GSA) was tested in rice leaves (15 days old) using²⁴ technique. The solution containing 50 mM Tris–HCl (pH 7.6), 1 mM MgCl₂, 1 mM EDTA, 4 mM leupeptine, and 14.3 mM β-mercaptoethanol was used to extract GSA from 0.25 g of leaf sample. The crude extract was centrifuged for 10 min at 16,000 × g, and 1mL of crude enzyme extract (supernatant) was examined for enzymatic activity with 0.5 M MgSO₄, 0.3 M glutamic natrium, 0.03 M ATP-Na, 0.25 M imidazole HCl, and 1 M hydroxylamine HCl; pH 7.0, after incubation at 25 °C for 20 min. One mL of stop solution containing 0.122 M FeCl₃, 0.5 M trichloroacetic acid, and 2 N HCl was added to halt the process. After centrifuging the samples for 10 min at 10,000 × g, the absorbance of the supernatant at 540 nm was used to track the synthesis of g-glutamyl hydroxymate^24,25. L-glutamine (Sigma-Aldrich) was used to prepare a series of standard solutions for constructing a calibration curve.

Estimation of nitrate reductase enzyme activity

For nitrate reductase activity (NRA), the fresh leaf samples (15 days old) were frozen at -20 °C, and activity was determined following²⁶. Leaf samples (0.25 g) were ground with digestion buffer containing 1 mM EDTA, 1–25 mM cysteine, and 0.1 M HCl to make an enzyme extract. The extract was filtered and centrifuged for 10 min at 10,000 × g to obtain supernatant. After centrifugation, the enzyme extract was allowed to react with 0.1 M potassium nitrate (dissolved in phosphate buffer, pH 7.5), and 2 mM NADH solution, yielding a final concentration of 0.21 mM. After incubation at 30 °C for 15 min, the reaction was terminated with the addition of 1% w/v sulphanilamide, followed by 0.02% naphthyl ethylenediamine reagent, and kept aside for 2 min. Then, the absorbance was measured at 540 nm on a spectrophotometer (Cecil CE 7400).

Data analysis

RStudio (R software) was used for statistical analysis and visualization of the entire dataset²⁷. The R packages ‘ggplot2’, ‘ggtext’, ‘dplyr’, ‘tidyr’, ‘agricolae’, ‘patchwork’, and ‘GGally’ were used for graphical data visualization. For both pot and field experiments, two-way analysis of variance techniques were employed to check the effect of treatments on genotypes and vice versa. The statistical model used was Trait ~ Genotype + Treatment + Genotype: Treatment, where genotype and treatment were considered fixed factors. Mean comparisons among genotype × treatment combinations were performed at 5% level of probability using Fisher’s least significant difference (LSD) test implemented in the agricolae package. Principal component analysis (PCA) was performed for the grouping of genotypes as efficient and inefficient, associated with vigorous and weak RSA. Correlation between the trait values was performed using the Pearson correlation method. Broad-sense heritability (H²) of each trait was calculated based on the equation described by²⁸.

Results

Study 1: Screening germplasm for root system architecture using a hydroponic experimental setup

Genetic variation in root-related traits

Statistical analysis revealed a significant (p < 0.001) difference between genotypes, treatments, and their interaction for LRD, LRN, SLRL, PRL, and TRL. The genotype × treatment interaction was also significant for these traits (p < 0.01 − 0.001), indicating differential genotype responses under varying N levels. All traits had strong heritability estimates (H² > 0.95), indicating that genetic factors contributed to phenotypic variance (Table S1; Supplementary dataset). Overall, the histograms for all traits across all treatments display a unimodal distribution. Each histogram shows a single peak with a positively skewed, non-symmetrical shape, indicating that the distribution of genotypes is skewed towards lower values with a tail extending to higher values (Fig. 1; Figure S4; Table S2; Supplementary dataset).

Fig. 1.

Histogram showing genotypic distribution for root-related traits (n = 99) under different N treatments. The treatments included: Am-N100-recommended ammonium-Nas urea, Am-N50-half of recommended ammonium-Nas urea, Nr-N100-recommended nitrate-N as calcium nitrate, and Nr-N50-half of recommended nitrate-N as calcium nitrate. The Figure includes (a) Lateral root density (No. of roots per cm of primary root), (b) Lateral root number, (c) Total root length(cm), (d) Total lateral root length (cm), (e) Primary root length(cm), (f) Chlorophyll content (SPAD value), and (g) Total plant biomass.

Under the Nr-N100 treatment, LRD ranged from a maximum of 3.8 to a minimum of 0.3 roots per cm. The genotype BAS 5854 exhibited the highest LRD, while the lowest was observed in AUS Khushi. Under Nr-N50, LRD varied between 0.4 and 2.6 roots per cm, with the highest value in PGRI_NARC_Rice_7782 and the lowest in IR-30. For the Am-N100 treatment, LRD ranged from 0.4 to 2.9 roots per cm, with the maximum observed in PGRI_NARC_Rice_7731 and the minimum in 98 PP-17. In the Am-N50 treatment, LRD varied from 0.4 to 3.1 roots per cm, showing a non-symmetric distribution (Fig. 1a; Table S2; Supplementary dataset).

Under the Nr-N100 treatment, LRN ranged from 3.2 to 24.7. The highest LRN was observed in Shua 92, while the lowest was found in DR-82. Under Nr-N50, LRN ranged from 3.2 to 19, with the maximum in DJ123 and the minimum in IR-30. For the Am-N100 treatment, LRN varied between 3.7 and 18.2, with BAS 5854 showing the highest LRN and PGRI_NARC_Rice_7709 the lowest. Under Am-N50, LRN ranged from 4 to 21.2, with KSK-133 displaying the highest value and PGRI_NARC_Rice_7709 the lowest (Fig. 1b; Table S2).

Under the Nr-N100 treatment, TRL ranged from 12.5 cm to 67.4 cm, with DJ123 exhibiting the maximum TRL and Madhukar the minimum. Under Nr-N50, TRL ranged from 9.8 cm to 63.3 cm, with DJ123 again showing the highest TRL and PGRI_NARC_RICE_7709 the lowest (Fig. 1c). For the Am-N100 treatment, TRL varied between 10.1 cm and 48.9 cm, with KSK-133 showing the highest TRL and IR-30 the lowest. Under Am-N50, TRL ranged from 10.7 cm in IR-30 to 57.6 cm in KSK-133.

Under the Nr-N100 treatment, TLRL ranged from 4.8 cm to 54.8 cm, with DJ123 showing the highest TLRL and Hansraj-1 the lowest. Under Nr-N50, TLRL varied from 5.1 cm to 50.2 cm, with DJ123 again exhibiting the highest value, while PGRI_NARC_Rice_7709 showed the lowest. For the Am-N100 treatment, TLRL ranged from 3.8 cm in IR-30 to 39.8 cm in KSK-133 (Fig. 1d). Under Am-N50, TLRL varied between 4.02 cm in IR-30 and 44.4 cm in KSK-133.

Under the Nr-N100 treatment, PRL ranged from 5.04 cm to 13 cm, with KSK-133 showing the highest PRL and RC-82 the lowest. Under Nr-N50, PRL varied from 4.2 cm in PGRI_NARC_Rice_7782 to 15 cm in KSK-133. For the Am-N100 treatment, PRL ranged from 4.9 cm to 13.4 cm, with PK 9966-10-1 exhibiting the highest PRL (Fig. 1e; Table S2). In the Am-N50 treatment, PRL ranged from 4.1 cm to 14 cm, with DR-92 showing the highest PRL and Minghui 63 the lowest.

Physiological performance and correlation among root-related traits

Relative chlorophyll content was measured using a SPAD meter. Under the Nr-N100 treatment, chlorophyll content ranged from 23.6 to 42.2, while under Nr-N50, it varied between 23.2 and 44. The histogram indicates a normal distribution of chlorophyll content among genotypes (Fig. 1f; Table S2; Supplementary dataset). The highest SPAD values across all treatments were observed in genotypes KSK-133, PK BB1, Shua 92, BAS 5854, 98 PP-7, and DJ 123, while the lowest values were recorded for genotypes PGRI_NARC_Rice_7709, Hansraj-1, RC-82, TKM 6, IR-30, and DR-82 (Table S2; Supplementary dataset).

Under the Nr-N100 treatment, TPB ranged from 13.22 mg to 31.57 mg, while under Nr-N50, it varied between 13 mg and 33.7 mg. In the Am-N100 and Am-N50 treatments, TPB ranged from 11 mg to 28 mg. The histogram indicates a normal distribution of TPB among genotypes (Fig. 1g; Table S2; Supplementary dataset). The genotypes DJ 123, PK BB 1, Shua 92, KSK-133, 98 PP-7, and BAS 5854 exhibited the highest TPB across all N treatments, while PGRI_NARC_Rice_7709, Hansraj-1, RC-82, TKM 6, IR-30, and DR-82 had the lowest TPB (Table S2; Supplementary dataset). The Pearson correlation matrix revealed a spectrum of associations, ranging from strong to weak and both positive and negative, among root morphological traits, chlorophyll content, and total plant biomass (Fig. 2). Lateral root traits were strongly correlated, with higher LRD and LRN consistently associated with greater TRL, enhanced chlorophyll content, and increased TPB. Primary root length also contributed positively to these traits, though its association with LRD was negative. Overall, coordinated development of root architecture was closely linked with improved chlorophyll status and biomass accumulation.

Fig. 2.

Pearson correlation matrix depicting the linkage between root system-related traits, chlorophyll content (SPAD value), and total plant biomass(n = 99). The diagonal depicts the distribution of the data; however, the lower half represents the scatter plots, and the upper half depicts the correlation coefficient R². Where, Chl: chlorophyll content, TPB: total plant biomass, TRL: total root length, PRL: primary root length, TLRL: total lateral root length, LRN: lateral root number, and LRD: lateral root density. Color scheme: Am-N100 (Coral), Am-N50 (Green), Nr-N100 (Violet), and Nr-N50 (Gray); applied consistently across scatter and density plots to distinguish N treatments.

Principal component analysis for the screening of twelve rice genotypes based on root system architecture index, chlorophyll content, and total plant biomass

A Principal Component Analysis (PCA) was performed to assess the variation in RSA indices, chlorophyll content, and TPB across four N treatments involving 99 rice genotypes. The analysis aimed to explore the contrasting responses of genotypes to different N forms and their potential associations with chlorophyll content and biomass production. Details of these genotypes are provided in Table S3, Supplementary dataset. From this analysis, twelve genotypes were selected, representing extremes in RSA vigor and their associated chlorophyll content and biomass levels. Specifically, these genotypes included those with significantly weak root system architecture (WRSA) and lower chlorophyll content and biomass, as well as those with vigorous root system architecture (VRSA) and higher chlorophyll content and biomass (Figure 3) Principal component analysis revealed that PC1 and PC2 explained 55.9% and 8.1% of the total variance, respectively, together accounting for 64.0% of the overall variability among traits. The PCA biplot (Figure 3) revealed distinct clustering patterns of genotypes under variable N treatments. Genotypes with significantly weak RSA predominantly clustered along Dim2, while those with vigorous RSA were positioned along the opposite dimension.

Study 2: Assessment of morphophysiological traits and yield components

Across both pot and field experiments, ammonium-based treatments (Am-N100 and Am-N50) consistently outperformed nitrate-based treatments (Nr-N100 and Nr-N50) for chlorophyll content, biomass production, and yield components. SPAD readings increased by 7.65% (Am-N100) and 4.50% (Am-N50) relative to nitrate counterparts (Fig. 4a). Genotypes BAS 5854, KSK-133, and DJ123 showed the highest SPAD values under Am-N50 (up to 45.4), although their percent reductions relative to N100 varied among genotypes. BAS 5854 showed comparatively lower reductions (2.97%) while KSK-133 and DJ123 showed greater declines of 26.8% and 30.7%, respectively, yet remained superior in absolute performance.

Fig. 4.

Response of twelve rice genotypes to four nitrogen (N) treatments, evaluated for (a) chlorophyll content (SPAD value), (b) total plant biomass c) grain yield and d) thousand grain weight. The treatments included: Am-N100-recommended ammonium-Nas urea, Am-N50-half of recommended ammonium-Nas urea, Nr-N100-recommended nitrate-N as calcium nitrate, and Nr-N50-half of recommended nitrate-N as calcium nitrate. The columns represent the mean value, and bars on columns indicate the standard error of the mean calculated (n = 4). Error bars annotated with different letters are significantly different at p < 0.05 according to the least significant difference (LSD) test. Whereas VRSA: genotypes having vigorous root system architecture and WRSA: genotypes having weak root system architecture.

TPB was 25.3% and 16.4% higher under Am- N100 and Am-N50, respectively, compared with nitrate treatments (Fig. 4b). Biomass accumulation was consistently highest in KSK-133 and DJ123 across both pot and field trials, reaching 27.3 g and 26.8 g under Am-N50 in pots and 227.4 g (Am-N100) and 179.2 g (Am-N50) in the field, while IR30 remained the lowest under all conditions. Shoot biomass followed similar patterns, with DJ123, KSK-133, and Shua 92 outperforming other genotypes. Grain yield was significantly influenced by genotype × treatment interactions (Fig. 4c), with Am-N100 and Am-N50 showing increases of 27.2% and 18.7% over nitrate treatments.

In the pots, DJ123 (11.5 g per pot) and KSK-133 (11.8 g) maintained the highest yields under Am-N50, despite reductions of 35.4% and 20.7%, respectively, compared to N100. In contrast, IR30 (2.1 g) and TKM6 (1.8 g) exhibited the lowest yield under reduced N. In the field, yields ranged from 18.7 to 27.5 g in IR30 to 145.6 g in KSK-133 under Am-N100, with reductions more pronounced under nitrate supply. Harvest index ranged from ~ 42% (TKM6) to ~ 71% (KSK-133), with consistent patterns across N forms. Thousand-grain weight (TGW) varied substantially across treatments (Fig. 4d). In the pots, KSK-133 (15.8 g) and DJ123 (15.3 g) recorded the highest TGW under Am-N50, whereas IR30 (6.6 g) and TKM6 (7.6 g) had the lowest. In the field, TGW ranged from 13.3 g (TKM6, Am-N100) to 26.0 g (DJ123, Am-N100), while under Am-N50 it spanned from 11.4 g (TKM6) to 23.9 g (KSK-133). Under nitrate nutrition, similar but generally lower trends were observed. Collectively, these results demonstrate that ammonium nutrition, particularly in genotypes KSK-133 and DJ123, enhanced chlorophyll content, biomass accumulation, and yield stability under both optimal and reduced N supply, while IR30 and TKM6 were most sensitive to N limitation.

Nitrogen concentration in grain, shoot, and root concentrations

Genotypic response to N treatments revealed substantial variability in N uptake and partitioning among plant organs (Table S4; Supplementary dataset). Grain N accumulation was markedly higher under ammonium than nitrate nutrition, with increases of 33% (Am-N100) and 29% (Am-N50) over corresponding nitrate treatments. Under Am-N50, KSK-133 (32.9 mg g⁻¹) and DJ123 (29.6 mg g⁻¹) exhibited the highest grain N concentrations, while IR30 (6.4 mg g⁻¹) and TKM6 (5.8 mg g⁻¹) remained the lowest. Field evaluations confirmed this trend, with grain N peaking at 36.4 mg g⁻¹ in KSK-133 and DJ123 under ammonium supply. Shoot N concentration also increased substantially, being 44% and 42% higher under Am-N100 and Am-N50 than nitrate equivalents. DJ123 and KSK-133 consistently maintained the highest shoot N (~ 22 mg g⁻¹), whereas IR30 and TKM6 recorded the lowest values (< 8 mg g⁻¹). Field results mirrored these patterns, with maximum shoot N of 20.7 mg g⁻¹ under Am-N100, declining to as low as 2.2 mg g⁻¹ in IR30 and TKM6 under reduced N. Root N concentration followed a similar trend, increasing by 7.3% (Am-N100) and 44.3% (Am-N50) over nitrate treatments (Table S5; Supplementary dataset). KSK-133 (15.4 mg g⁻¹) and DJ123 (15.2 mg g⁻¹) accumulated the most root N under Am-N50, while TKM6 and IR30 showed minimal values (~ 3.7–3.8 mg g⁻¹). Collectively, these results highlight strong genotype × N form interactions, with KSK-133 and DJ123 exhibiting superior N uptake and partitioning across tissues, in contrast to IR30 and TKM6, which remained highly sensitive to reduced N supply.

In the pot trial, genotypes exhibited marked variability in NUpE, which was 63% higher under Am-N50 compared with Nr-N50 (Fig. 5c). The highest NUpE was observed in KSK-133 (63.3%) and DJ123 (59.6%) under Am-N50, while IR30 (7.6%) and TKM6 (8.3%) showed the lowest values. By contrast, under Nr-N50, KSK-133 (30.2%) and DJ123 (34.1%) still outperformed other genotypes, whereas IR30 (6.4%) and TKM6 (5.6%) remained the least efficient. NHI varied significantly among genotypes but was unaffected by N source or dose (p > 0.05), mirroring the trends observed for HI (Fig. 5a, b). Together, these results highlight strong genotypic differences in NUE components, with KSK-133 and DJ123 demonstrating superior N uptake and partitioning efficiency, while IR30 and TKM6 remained highly inefficient.

Fig. 5.

Response of twelve rice genotypes to four nitrogen (N) treatments, evaluated for (a) harvest index, (b) nitrogen harvest index, and c) nitrogen uptake efficiency. The treatments included: Am-N100-recommended ammonium-N as urea, Am-N50-half of recommended ammonium-N as urea, Nr-N100-recommended nitrate-N as calcium nitrate, and Nr-N50-half of recommended nitrate-N as calcium nitrate. The columns represent the mean value, and bars on columns indicate the standard error of the mean calculated (n = 4). Error bars annotated with different letters are significantly different at p < 0.05 according to the least significant difference (LSD) test. Whereas VRSA: genotypes having vigorous root system architecture, and WRSA: genotypes having weak root system architecture.

Genotypic diversity in NUE under contrasting growth environments (pot vs. field)

Across both pot and field experiments, significant genotypic variation in NUE was evident, with treatment effects and genotype × treatment interactions contributing to the observed differences (Fig. 6). In pots, Am-N50 increased NUE by 16% compared with Nr-N50, with KSK-133 (21.4 g g⁻¹) and DJ123 (21.1 g g⁻¹) achieving the highest values, and IR 30 (9.3 g g⁻¹) and TKM6 (10.3 g g⁻¹) the lowest. Under Nr-N50, DJ123 (18.01 g g⁻¹) and KSK-133 (16.8 g g⁻¹) again ranked highest, while TKM6 (8.8 g g⁻¹), Hansraj-1 (9.3 g g⁻¹), and IR 30 (9.5 g g⁻¹) remained lowest. Field trials similarly showed higher NUE under low-N treatments (Am-N50 and Nr-N50), with DJ123 and KSK-133 consistently outperforming other genotypes, and TKM6 and IR 30 recording the lowest values. When ranked by cumulative NUE across all treatments and environments, the order from highest to lowest was DJ123 > KSK-133 > Shua 92 > 98PP-7 > PK BB1 > BAS 5854 > Hansraj-1 > PGRI_NARC_Rice_7692 > DR-82 > RC-82 > TKM6 > IR 30, highlighting the stability of high-NUE performers across contrasting growth conditions.

Fig. 6.

Response of four rice genotypes to four nitrogen (N) treatments, evaluated for (a) nitrogen use efficiency in pot trial and (b) nitrogen use efficiency in field. The treatments included: Am-N100-recommended ammonium-N as urea, Am-N50-half of recommended ammonium-N as urea, Nr-N100-recommended nitrate-N as calcium nitrate, and Nr-N50-half of recommended nitrate-N as calcium nitrate. The columns represent the mean value, and bars on columns indicate the standard error of the mean calculated (n = 4). Error bars annotated with different letters are significantly different at p < 0.05 according to the least significant difference (LSD) test. Whereas VRSA: genotypes having vigorous root system architecture, and WRSA: genotypes having weak root system architecture.

Correlation among key morpho-physiological, NUE, and yield-related traits

Pearson correlation analysis revealed consistently positive associations among key morpho-physiological, NUE, and yield-related traits (Fig. 7). SPAD values were strongly associated with PL, NOT, NUpE, NUE, NHI, HI, GY, and TGW. Similarly, TPB showed strong positive correlations with PL, NOT, NUpE, NUE, and NHI. Grain yield was particularly well-aligned with PL, NOT, NUpE, NUE, NHI, HI, and TGW, highlighting the close integration of yield potential with N efficiency traits. NUE itself correlated positively with PL, NOT, and NUpE, while NUpE displayed robust associations with PL and NOT. Both NHI and HI exhibited strong positive relationships with PL, NOT, NUpE, NUE, and GY, underscoring their central role in yield determination.

Fig. 7.

Pearson correlation matrix depicting the linkage between shoot N concentration (SN), root N concentration (RN), grain N concentration (GN), chlorophyll content (SPAD), thousand grain weight (TGW), grain yield (GY), harvest index (HI), total plant biomass (TPB), nitrogen harvest index (NHI), nitrogen use efficiency (NUE), nitrogen uptake efficiency (NUpE), number of tillers (NOT) and panicle length (PL), (n = 4). Diagonal depicts the distribution of the data; however, the lower half represents the scatter plots, and the upper half depicts the correlation coefficient (r). Color scheme: Am-N100 (Coral), Am-N50 (Green), Nr-N100 (Violet), and Nr-N50 (Gray); applied consistently across scatter and density plots to distinguish N treatments.

Grain and shoot N concentrations displayed highly similar correlation patterns, both being positively associated with PL, NOT, NUpE, NUE, NHI, HI, GY, TGW, and SPAD, as well as with each other. Root N concentration also showed moderate-to-strong positive associations with most yield-related traits, except NUE, where the relationship was weak. Importantly, under field conditions, a broadly similar correlation structure was observed (Figure S2; Supplementary dataset), with SPAD values positively associated with plant height, NOT, PL, GY, HI, SN, and GN. Both SN and GN maintained strong positive associations with PL, NOT, GY, HI, and each other, further reinforcing the pivotal role of N metabolism in supporting yield formation.

Study 3: Evaluation of four contrasting genotypes for N assimilation enzyme activity

Two-way ANOVA indicated that genotype, treatment, and interaction significantly (p < 0.05%) affected all traits measured, including root length, root N concentration, shoot N concentration, nitrate reductase activity, and glutamine synthetase activity (Table S6; Supplementary dataset). Across traits, efficient genotypes (KSK-133 and DJ123) consistently outperformed inefficient ones (IR30 and TKM6) under low-N conditions. KSK-133 maintained the highest shoot dry weight under both low ammonium (54.5 mg, − 25.2%) and low nitrate (56.7 mg, − 19.4%), while the inefficient TKM6 recorded the lowest (26.8–27.0 mg, -27-36%). DJ123 produced the greatest root biomass (6.9 mg in Am-N25; 4.4 mg in Nr-N25) and retained root length advantage under low ammonium (13.1 cm), whereas KSK-133 excelled under low nitrate (14.1 cm); TKM6 remained the shortest (5.5–6.1 cm). For shoot length, DJ123 (14.5–15.2 cm) showed moderate reductions, while TKM6 showed a sharp 30% decline under low ammonium. Root N concentration peaked in DJ123 (6.2 mg g^− 1) and KSK-133 (5.4 mg g^− 1) under ammonium and nitrate-limited conditions, respectively, while IR30 consistently exhibited the lowest (< 2.3 mg g^− 1). Shoot N concentration followed a similar trend, with DJ123 maintaining the highest and IR30 the lowest, corresponding to the largest proportional declines. The efficient genotype KSK-133 recorded the highest NRA under low ammonium (Am-N25), with 0.2 (µmol NO₂⁻ g⁻¹ h⁻¹ FW), an increase of 43.2% compared to the control. Under low nitrate (Nr-N25), the efficient genotype DJ123 had the highest NRA (0.4 µmol NO₂⁻ g⁻¹ h⁻¹ FW), showing a 10.3% increase over the control (Fig. 8a). In contrast, the inefficient genotype IR30 showed the lowest NRA, measuring 0.2 µmol NO₂⁻ g⁻¹ h⁻¹ FW in Am-N25 (− 11% vs. control) and 0.4 (µmol NO₂⁻ g⁻¹ h⁻¹ FW) in Nr-N25 (+ 32% vs. control). For GSA, KSK-133 again outperformed other genotypes, with 80.0 (µmol ϒ-GDH m^− 1 g^− 1 FW) under Am-N25 and 79.6 (µmol ϒ-GDH m^− 1 g^− 1 FW) under Nr-N25, showing only a marginal change (− 0.9% and + 1.2%, respectively) compared to their controls (Am-N100 and Nr-N100). Conversely, IR30 had the lowest GSA, 73.9 µmol ϒ-GDH m^− 1 g^− 1 FW in Am-N25 and 72.9 µmol ϒ-GDH m^− 1 g^− 1 FW in Nr-N25, representing decreases of 1.01% and 1.5%, respectively, relative to controls (Fig. 8b).

Fig. 8.

Response of four rice genotypes to four nitrogen (N) treatments, evaluated for (a) glutamine synthetase activity (µmol ϒ-GDH m^− 1 g^− 1 FW) and (b) nitrate reductase activity (µmol NO₂⁻ g^− 1 h^− 1 FW). The treatments included: Am-N100-recommended ammonium-N as urea, Am-N25-25% recommended ammonium-N as urea, Nr-N100-recommended nitrate-N as calcium nitrate, and Nr-N25-25% of recommended nitrate-N as calcium nitrate. The columns represent the mean value, and bars on columns indicate the standard error of the mean calculated (n = 4). Error bars annotated with different letters are significantly different at p < 0.05 according to the least significant difference (LSD) test. Genotypes have been grouped as N use efficient (selected from VRSA group), and N use inefficient (selected from WRSA group).

The Pearson correlation analysis highlighted a strong functional linkage between plant growth, N accumulation, and GSA (Figure S3; Supplementary dataset). Morphological traits such as shoot dry weight, root dry weight, root length, and shoot length were all strongly interrelated and strongly associated with both shoot and root N concentrations. GSA exhibited similarly high positive correlations with these traits, suggesting that enhanced ammonium assimilation via GS may be a key determinant of biomass accumulation and N status. In contrast, NRA displayed only weak positive associations with growth, N concentration, and GSA, indicating a comparatively limited role in driving variation under the tested conditions. Collectively, these patterns suggest that biomass production and N accumulation in the evaluated genotypes are more tightly coupled to GS-mediated ammonium assimilation than to NR-dependent nitrate reduction.

Discussion

Root system architecture (RSA) defines the spatial arrangement of roots and influences the capabilities of plants for soil exploration²⁹. This study evaluated the association of root-related traits with chlorophyll content and total plant biomass (TPB) to infer the response of genotypes in relation to NUE. Root architecture is critical for understanding plant nutrient acquisition and adaptation to environmental conditions³⁰. In the current study, through analysis of root traits under different N sources, it is revealed that ammonium treatments (Am-N100 and Am-N50) generally enhance root development compared to nitrate treatments (Nr-N100 and Nr-N50). Specifically, LRD and LRN were consistently higher under ammonium conditions, with genotypes such as KSK-133 and Shua 92 showing pronounced increases (Fig. 1). While some research has shown that nitrate can stimulate root branching and lateral root growth in certain species or under specific conditions³¹, other studies, particularly in the model grass Brachypodium distachyon, have reported that ammonium promotes the development of thin, highly branched roots, with increased lateral root length and number at higher concentrations³². Furthermore, a recent study in cotton suggests that optimized RSA provides valuable potential for breeding climate-resilient crops under stress conditions³³. The superior effectiveness of ammonium in supporting robust root development can be attributed to its role in acidifying the root zone, which enhances nutrient uptake and root growth^34,35. Earlier⁷, indicated that some rice genotypes respond to N deprivation followed by resupply with ammonium by developing a more branched root system, improving nutrient uptake. The observed variation in root traits under ammonium may also be explained by the underlying physiological regulation of root growth. Enhanced arginosuccinate lyase (ASL) activity could promote auxin biosynthesis in the root meristem, supporting cell division and elongation under ammonium nutrition. Additionally, hormones such as abscisic acid (ABA) and strigolactones may help maintain meristem activity by modulating ROS balance, thereby reducing ammonium-induced stress. These mechanisms likely contribute to the improved root performance observed in certain genotypes^36,37. The strong positive correlations of LRD with LRN, TLRL, TRL, and TPB suggest that increased lateral root development enhances overall root system efficiency, improving nutrient uptake and biomass accumulation. In contrast, its negative correlation with PRL indicates a possible trade-off between lateral and primary root growth. Similarly, positive associations of TRL and TLRL with chlorophyll content and biomass further highlight the role of an extensive root system in supporting plant growth^38–40. Overall, Ammonium enhanced root traits (e.g., LRD and LRN in KSK-133 and Shua-92) relative to nitrate, promoting RSA, potentially via rhizosphere acidification. Lateral roots were positively associated with chlorophyll content and TPB but inversely related to primary root growth.

In pot and field experiments, ammonium N consistently increased tissue N compared to that for nitrate. Grain N accumulation rose by 33% (Am-N100) and 29% (Am-N50) over nitrate counterparts, with peak grain N of 36.4 mg g⁻¹ in the field under ammonium. Shoot N was likewise elevated to 44% (Am-N100) and 42% (Am-N50), and root N increased by 7.3% (Am-N100) and 44.3% (Am-N50). These patterns align with the view that ammonium can reduce the energetic cost of N assimilation relative to nitrate and favor rapid incorporation into amino acids, thereby boosting whole-plant N status under constrained supply⁴¹. Importantly, the magnitude of response was N-form dependent and strongest at the low dose (Am-N50), indicating an opportunity to realize N savings without compromising plant N status. Genotypic contrasts were also pronounced. The data demonstrate that (i) ammonium supply amplifies uptake gains, and (ii) intrinsic genotypic capacity to acquire and partition N is a dominant driver of NUE components. This pattern accords with reports that, at low N, yield is primarily limited by NUpE, whereas at high N it shifts toward NUtE^42,43. N fertilization is also known to enhance macronutrient uptake, but the magnitude of this response varies among genotypes due to inherent physiological and biochemical adaptations. N-efficient genotypes have been shown to exhibit superior morphophysiological traits that optimize nutrient acquisition and utilization across varying N regimes. Their enhanced performance is attributed to an active root system, the release of organic root exudates, the ability to modify rhizosphere pH, and facilitate greater nutrient solubilization and acquisition from the soil^44,45. Our results identify KSK-133 and DJ123 as high-NUpE genotypes under reduced N, a desirable profile for low-input systems and for breeding programs targeting stable performance under N scarcity^46,47.

Despite large differences in uptake, NHI did not respond to N source or dose (p > 0.05) and mirrored HI (Fig. 5a, b). This suggests that, within the tested range, genotypes maintain relatively stable N partitioning rules to the grain, and NUE gains are achieved mainly through increased acquisition rather than altered allocation. The literature offers several, not mutually exclusive, explanations for weak or inconsistent NHI responses: (i) dilution of N/protein at higher yields^48,49(ii) faster improvement in HI relative to NHI⁵⁰(iii) competition for assimilates between vegetative growth and grain filling⁵¹ and (iv) mismatches in the temporal dynamics of grain N and carbohydrate deposition⁵². Within this context, our observation of stable NHI alongside strong NUpE shifts indicates that, for these materials and environments, improving the “front end” (acquisition) outweighed changes at the “back end” (allocation). This does not necessarily limit further optimization of NUE; rather, it indicates that, under the present conditions, greater gains are more likely to be achieved by enhancing N acquisition and, potentially, by improving post-anthesis N remobilization efficiency under different environmental or management scenarios. Conversely, grain N production efficiency (NUEg) tends to be higher when vegetative N is lower, particularly at zero or reduced N inputs, whereas absolute grain N content increases with high N⁵³. Our data fit this framework: ammonium supply boosted tissue N widely, yet genotypes differed in how that N was translated into putative yield components. The superior performance of KSK-133 and DJ123-high NUpE, high tissue N, and high grain N, suggests a favorable balance of sink size/strength and post-anthesis assimilation, traits repeatedly linked to NUE under N scarcity^46,47. In contrast, IR30 and TKM6 combined low NUpE with low tissue N across organs, consistent with reduced sink strength and/or continued allocation to vegetative growth during grain filling as hypothesized by⁵¹. Collectively, these results demonstrate that efficient genotypes sustained greater biomass, morphological vigor, and assimilation efficiency under both ammonium and nitrate-limited conditions, whereas inefficient genotypes experienced substantial growth and N-uptake reductions.

Variability in N metabolism among rice genotypes reflects differences in uptake, assimilation, and utilization efficiencies. Our findings revealed significant genotypic variation in NRA and GSA across N treatments (Fig. 8). KSK-133 exhibited the highest NRA under Nr-N100 (0.523 µmol NO₂⁻ g⁻¹ h⁻¹ FW) but declined at Nr-N25, while TKM6, IR30, and DJ123 increased, suggesting distinct regulatory thresholds. The high NUE genotypes possess low NR threshold levels, enabling efficient nitrate assimilation⁵⁴. NR activity was consistently lower under ammonium than nitrate treatments, reaffirming its nitrate inducibility⁵⁵. NR, as a rate-limiting enzyme in N assimilation, depends on nitrate availability; higher activity under nitrate treatments supports efficient carbon metabolism and N use⁵⁶. Reversible phosphorylation, which is triggered by low carbon status or dark conditions, tightly regulates NR activity post-translationally by promoting 14-3-3 binding and inactivation. Previous research on rice reveals genotype-specific variations; high-NUE lines, such as those with altered OsNR, exhibit less phosphorylation-dependent inhibition, which is linked to increased NUE through enhanced assimilation^57,58. KSK-133 maintained the highest activity under Am-N100 but showed a 38.8% decline relative to Nr-N100, whereas DJ123 exhibited more stable activity under reduced N, highlighting adaptability to suboptimal conditions. These results corroborate reports that high NUE genotypes sustain NR activity across N levels, while inefficient genotypes respond primarily to higher N inputs^59,60. For GS, KSK-133 (high NUE) maintained higher activity under low nitrate but declined slightly under low ammonium, whereas IR30 (low NUE) showed the lowest activity under both N sources. The high NUE genotypes retain greater GSA due to efficient N re-translocation^61,62. also reported GSA doubling under low N, reflecting its role in scavenging ammonia from photorespiration. Hence, GSA may be increased to utilize this ammonia as a substrate. A positive correlation of leaf GSA with grain yield and kernel number under low N-input has been reported⁶³, which aligns with findings of the present study, where we observed a strong positive correlation of GSA with shoot and root dry weight and N concentrations, while a weak positive correlation with NR (Figure S3; Supplementary dataset). Collectively, these results highlight that high NUE genotypes such as KSK-133 sustain NR and GS activity under limiting N, underpinning their superior N metabolism and growth performance compared to inefficient genotypes. While the present study emphasizes structural root traits, future research should integrate functional aspects of root hydraulic architecture, including aquaporin activity and axial conductance, as shown by⁶⁴. Such a combined structural functional characterization of contrasting genotypes (e.g., DJ123 and KSK-133) would provide a more comprehensive understanding of their resource acquisition strategies under stress conditions. These superior root and NUE-associated traits can be strategically introgressed into elite cultivars to develop high-yielding rice varieties with stable performance under low-N environments.

For farmers and agricultural managers, the incorporation of these genotypes into locally adapted varieties may contribute to improved NUE, with the potential to reduce fertilizer inputs while maintaining yield levels. Such improvements could help lower production costs and mitigate environmental impacts, including nutrient losses and runoff.

Conclusion

Across complementary hydroponic, pot, and field evaluations, our study demonstrates that NUE in rice is primarily driven by RSA vigor, N form responsiveness, and the capacity to sustain N assimilation under suboptimal supply. Genotypes DJ123 and KSK-133 consistently exhibited superior performance, with enhanced root development, higher tissue N content, sustained GS and NR activities, and improved physiological traits (SPAD, harvest index, and NUpE), resulting in significantly greater biomass and grain yield under both optimal and reduced N conditions, particularly with ammonium nutrition. In contrast, IR30 and TKM6 showed poor RSA, unstable enzyme activity, and marked yield reductions across environments. Importantly, our results identify DJ123 and KSK-133 as elite donor genotypes with robust and resilient root systems and superior N assimilation capacity. These traits enable stable yield performance under reduced N inputs, making them strong candidates for incorporation into breeding programs aimed at improving N uptake efficiency and developing sustainable, low-input rice production systems.

Supplemental material

The supplementary Excel file contains detailed statistical analyses of the studied traits, including analysis of variance (ANOVA) tables and supporting datasets. It also includes supplementary figures related to trait performance, along with details of the genotypes used in this study.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Zunaira Bano (ZB), Aysha Kiran (AK), Asma Imran (AI), Abdul Wakeel (AW)Supervision: AK, AI, AW; conceptualization: AK, AI, AW; methodology: AK, AW, ZB; execution of the experimental work: ZB;1st draft preparation: ZB; draft reviewed and edited: AK, AI, AW, ZB. All the authors have reviewed the manuscript and agreed to submit it in its current form for publication in the Journal.

Funding

All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.

Data availability

All data generated or analyzed during this study and its supplementary information files available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.