The field experiment was conducted over a period of 2 years (2022 and 2023) at the research farm of the ICAR-Indian Agricultural Research Institute (IARI) in New Delhi (Figure 1). The location of the farm is at 28°38′15´´ N latitude, 77°09′10″ E longitude, and an elevation of 228.6 meters above mean sea level. Throughout the experimental years (2022 and 2023), meteorological data regarding daily temperature and rainfall were recorded. The monthly weather data for the experimental period (2022 and 2023) was sourced from the ICAR-IARI meteorological observatory, situated in close proximity to the experimental site (Figure 2). The soil at the site is described as well-drained, non-saline, slightly alkaline in nature (pH 7.3), and possessing a sandy loam texture. The experimental site is situated in a semi-arid region with an average annual precipitation of 708 mm. The detailed soil characteristics are given in the Table 1.

The experiment was conducted using a randomized block design (RBD). The layout of the different fertilizer doses in RBD is shown in Figure 3. This design was selected to minimize the effects of variability and confounding factors by randomly allocating the different fertilizer doses to separate blocks within the study area (Kushwah et al., 2023). Pusa Basmati (PB)-1509 rice variety was taken as test crop. Each cultivation method (transplanted and direct seeded) was considered as a separate block, ensuring that any variations observed in the results could be attributed to the fertilizer doses rather than other factors.

For transplanted rice, the seedbed preparation involved two passes of a disc harrow, followed by one pass of a cultivator with a wooden plank, and finally the puddling operation. On the other hand, for direct-seeded rice, the field preparation only included two passes of a disc harrow and one pass of a cultivator with a wooden plank. After 21 days of growth, the rice seedlings were manually transplanted in the case of transplanted rice, while direct seeding was done using a tractor operated 9-row DSR planter immediately after field preparation in the case of direct-seeded rice (Kushwah et al., 2024a). The rice seeds were subjected to a treatment with Carbendazim 50 WP at a concentration of 2 g per kilogram of seeds prior to sowing. This treatment was applied to manage fungal diseases, tip burn, black spot, and collar spot (Chowdhury et al., 2023c). The treated PB-1509 rice seeds were sown at a seeding rate of 20 kg per hectare.

The N fertilizer doses varied from 0 to 240 kg ha⁻¹ with the interval of 40 kg ha⁻¹ for both transplanted and direct seeded rice were applied. The range of N fertilizer doses were chosen for both transplanted and direct-seeded rice in order to evaluate the effects of different levels of N fertilizer application on the crop. By varying the doses within this range, the experiment was aimed to assess the response of the rice plants to increasing amounts of fertilizer and determine the optimal dosage that would maximize yield and overall plant health. This range allows for a comprehensive analysis of the impact of fertilizer on both cultivation methods and provides valuable insights for agricultural practices and crop management. Phosphorus (P), potassium (K) and zinc (Zn) were applied as basal at the rate of 60 (P₂O₅), 60 (K₂O), 25 (zinc sulfate) kg ha⁻¹, respectively. P and K were applied through single super phosphate (SSP) and Muriate of Potash (MOP), respectively. The treatment details are given in Table 2.

Spectral data, crucial for understanding crop development, were meticulously collected at significant growth stages for both TR and DSR. The spectral signature was recorded at three growth stages; 20 days after transplanting (DAT), 40 DAT and 60 DAT for TR. Whereas, the growth stages for DSR were taken 40 days after sowing (DAS), 60 DAS and 80 DAS to capture the spectral reflectance of the rice canopy. Employing a hyperspectral sensor, the Spectroradiometer (Spectral Evolution, PSR-1100f), data acquisition covered the extensive wavelength range of 320 nm to 1,100 nm, as shown in Plate 1A. Calibration using a white reference panel was undertaken to ensure precision in reflectance measurements. Maintaining strict consistency, the sensor was positioned at a height of 60 cm, capturing readings across experimental plots. The data acquisition process was synchronized, conducted at noon, when the sun is at zenith, to mitigate variations due to diurnal lighting changes (Bhagat et al., 2023). The captured spectral data were then meticulously analyzed using DARWin SP data acquisition software, focusing on bands pertinent to vegetation monitoring and index computation.

Simultaneously, crop growth and vigor were monitored at specific growth stages using the GreenSeeker® handheld device, equipped with light-emitting diodes emitting red (656 nm) and near-infrared (NIR) (774 nm) light, as shown in Plate 1B. Normalized differential vegetation index (NDVI) was employed for assessing plots with varying fertilizer doses. To cross-verify measurements, comparisons were drawn between NDVIs obtained from GreenSeeker® and Spectroradiometer data. Beyond NDVI, numerous vegetation indices (VIs) were computed using specific wavelengths and formulas, tabulated in Table 3. This comprehensive approach ensured data accuracy, providing valuable insights into vegetative health and nitrogen status within the rice plants.

Various growth characteristics, including plant height and leaf area index (LAI), as well as yield-related factors like number of effective tillers per hill, panicles per hill, grains per panicle, panicle length, thousand-grain weight, grain and straw yield, and harvest index were measured and assessed. This analysis aimed to gauge the impact of different levels of nitrogen fertilizer application and cultivation techniques to gather insights that could guide strategic choices for improving rice production. These attributes provide insights into the performance, productivity, and quality of the rice crop. Plant height was measured at different growth stages of both transplanted and direct seeded rice. To capture variations within each plot, multiple plant samples were randomly selected, and plant height was measured from the base to the tip of the longest leaf using a measuring tape. Harvesting was conducted when the rice plants reached maturity, characterized by physiological maturity and grain moisture content within the recommended range. The rice plants from each plot were harvested, threshed to separate the grains from the straw, and cleaned to remove impurities. Grain yield was determined by weighing the harvested grains after drying to a constant moisture content. Biomass yield was calculated by weighing the straw collected from each plot. Harvest index (HI) was calculated using the following formula (Equation 1);

In order to determine the leaf area in transplanted and direct seeded rice plants, a Portable Leaf Area Meter (LI-COR 3000) was utilized. This device allows for accurate and efficient measurement of leaf area. The experiment involved selecting five representative plants from each plot and tagging them for identification. Leaf area measurements were taken periodically at 20 DAT, 40 DAT, 60 DAT and 90 DAT for transplanted rice, and 40 DAS, 60 DAS, 80 DAS and 110 DAS for direct seeded rice to monitor leaf growth and development. LAI was calculated using the following formula (Patra et al., 2023) (Equation 2);

The economic analysis involved evaluating the gross and net returns, as well as the net benefit–cost ratio (Chowdhury et al., 2023b; Kushwah et al., 2024b). The gross return (GR) was calculated by multiplying the economic outputs, which included grain and straw yields, by their respective prices. The calculation of GR involved utilizing the market minimum support price (MSP) for rice grain. On the other hand, the net returns (NR) were obtained from the Equation 3. Benefit–cost ratio was calculated by utilizing Equation 4. All economic analyses were conducted in Indian rupees (₹), which were subsequently converted into US dollars ($).

The collected data, including vegetation indices, growth and yield attributes, underwent rigorous statistical analysis to derive meaningful insights using statistical packages for social sciences (SPSS) software (BM Corp, 2012). Analysis of variance (ANOVA) was performed to assess the significance of differences among the N fertilizer doses and cultivation methods (transplanted and direct seeded rice). In order to determine the significance of the observed results, a p-value threshold of ≤0.05 was adopted. In cases where statistically significant differences were detected, Duncan’s Multiple Range Test (DMRT) was conducted to identify the specific differences among the treatments under investigation. Additionally, standard error (SE) was used to quantify the extent of variability among individual data values.

The nitrogen levels exerted a substantial and statistically significant (p < 0.05) influence on plant height of both transplanted and direct seeded rice. Significant variations in plant height were observed across different nitrogen doses at 20, 40, 60, and 90 days after transplanting (DAT). Significantly higher plant heights of 40.68, 55.28, 70.85 and 92.70 cm were observed at 20, 40, 60, and 90 DAT respectively, from the 160 kg N ha⁻¹ (treatment N₄) applied plot (Figure 4A). Comparing to the control treatment (0 kg N ha⁻¹), the increase in plant height at harvest stage (90 DAT) amounted to 2.73, 7.83, 15.09, 15.15 14.91, and 13.35% for nitrogen doses of 40 kg, 80 kg, 120 kg, 160 kg, 200 kg and 240 kg N ha⁻¹, respectively. Plant height increased significantly up to the recommended dose of nitrogen (120 kg N ha⁻¹), above this, no significant differences were observed in transplanted rice.

Comparative analysis with TR revealed that DSR consistently exhibited greater plant height at equivalent growth stages, with DSR presenting a 6.09% increase in plant height at the time of harvest. Plant heights exhibited a significant increase in the plot treated with 240 kg N ha⁻¹ (treatment N₆), measuring 51.75 cm, 58.90 cm, 78.85 cm, and 98.35 cm at 40, 60, 80, and 110 days after sowing (DAS), respectively (Figure 4B). In comparison to the control treatment (0 kg N ha⁻¹), the growth in plant height at the harvest stage (110 DAS) showed increments of 3.16, 7.29, 11.59, 11.71, 12.88, and 12.91% for nitrogen doses of 40 kg, 80 kg, 120 kg, 160 kg, 200 kg, and 240 kg N ha⁻¹, respectively.

The two-year mean LAI observed at 20, 40, 60, and 90 DAT for transplanted rice and at 40, 60, 80, and 110 DAS for direct seeded rice, exhibited significant variations among the different N levels, as shown in Figure 5. Across all treatments, the LAI displayed a consistent pattern of increase until reaching a plateau phase, occurring at approximately 40–55 DAT for TR and 60–75 DAS for DSR. Specifically, at the early stages of the experiment (20 DAT for TR and 40 DAS for DSR), the LAI values did not exhibit significant differences among the various nitrogen-fertilized treatments. During these initial measurements, the LAI values fell within a range of 1.12 to 1.38 (for TR) and 0.95 to 1.3 (for DSR), indicating a relatively modest to low level of foliage coverage. Subsequently, as the rice plants progressed through their growth cycle, a substantial increase in LAI became evident. The LAI values witnessed notable growth, ranging from 2.85 to 3.62 (at 40 DAT) and 2.95 to 3.71 (at 60 DAS), suggesting a significant expansion of canopy coverage and leaf area. The maximum LAI values (4.32 and 4.25) were observed at 60 DAT for TR and 80 DAS for DSR at 160 and 200 kg N ha⁻¹ respectively, indicating further development of the canopy and an enhanced capacity to harness solar radiation for photosynthesis. Toward the harvesting stage, the LAI exhibited a reduction for both TR and DSR. During the active growth phase, 160 kg N ha⁻¹ consistently displayed the highest LAI values, followed by 200 kg N ha⁻¹, 240 kg N ha⁻¹, and 120 kg N ha⁻¹, respectively, for TR. In the case of DSR, 200 kg N ha⁻¹demonstrated the highest LAI, followed by 160 kg N ha⁻1, 240 kg N ha⁻¹, and 120 kg N ha⁻¹, respectively. Notably, the LAI values in Treatments N₃ (120 kg N ha⁻¹), N₄ (160 kg N ha⁻¹), N₅ (200 kg N ha⁻¹), and N₆ (240 kg N ha⁻¹) were comparable between transplanted and direct seeded rice.

In the years 2022 and 2023, noteworthy distinctions were observed in the performance of transplanted rice across various vegetation indices, including NDVI, RVI, NDRE, and GNDVI among different nitrogen fertilizer treatments, as delineated in Table 4. Analysis of NDVI, a key indicator of vegetation health, revealed conspicuous variations among the nitrogen fertilizer treatments, with mean values spanning from 0.45 (0 kg N ha⁻¹) at 20 DAT to 0.83 (160 kg N ha⁻¹) at 60 DAT in 2022 and 0.43 (0 kg N ha⁻¹) at 20 DAT to 0.85 (0 kg N ha⁻¹) at 60 DAT in 2023. Notably, treatment N₄ consistently exhibited the highest NDVI values at various growth stages. In comparison to the control treatment (0 kg N ha⁻¹), treatment N₄ (160 kg N ha⁻¹) exhibited a remarkable increment in NDVI values, registering a 28.89 and 39.53% increase at 20 DAT in 2022 and 2023 respectively, 16.13 and 23.73% increase at 40 DAT in 2022 and 2023, and a 12.16 and 14.86% increase at 60 DAT in 2022 and 2023, respectively. A similar trend was observed for RVI, where significant differences were evident among the N fertilizer treatments, with mean values ranging from 5.65 and 5.78 (0 kg N ha⁻¹) in 2022 and 2023 at 20 DAT to 14.85 and 15.41 (200 kg N ha⁻¹) in 2022 and 2023 at 60 DAT. In comparison to the control treatment, treatment N₅ (200 kg N ha⁻¹) exhibited notable increases in RVI values, with a 29.20% (2022) and 29.24% (2023) rise at 20 DAT, a 25.55% (2022) and 23.98% (2023) rise at 40 DAT, and an 18.23% (2022) and 17.99% (2023) rise at 60 DAT. Furthermore, the NDRE values ranged from 0.18 and 0.19 (0 kg N ha⁻¹) at 20 DAT to 0.37 (under 120 kg N ha⁻¹ and 200 kg N ha⁻¹) and 0.38 (under 160 kg N ha⁻¹) at 60 DAT in 2022 and 2023, respectively. In comparison to the control treatment (0 kg N ha⁻¹), there were substantial increases in NDRE values, including a 44.44% (2022) and 47.37% (2023) rise at 20 DAT with N₅ (200 kg N ha⁻¹) and N₆ (240 kg N ha⁻¹) treatments, 50% (2022) and 47.83% (2023) rise at 40 DAT with N₅ treatment (200 kg N ha⁻¹), and 42.31% (2022) and 40.74% (2023) rise at 60 DAT with 120 kg N ha⁻¹, 160 kg N ha⁻1 and 240 kg N ha⁻¹. Additionally, the mean values of GNDVI ranged from 0.28 and 0.29 (under 0 kg N ha⁻¹) at 20 DAT to 0.49 and 0.52 (under 160 kg N ha⁻¹) at 60 DAT in 2022 and 2023, with treatment N₄ (160 kg N ha⁻¹) exhibiting the highest GNDVI values throughout the growth stages of transplanted rice. Compared to the control treatment, there were significant increases in GNDVI values, with a 39.28% (2022) and 37.93% (2023) rise at 20 DAT, 35.29% (2022) and 37.14% (2023) rise, and 28.95 and 30% rise at 60 DAT with N₄ treatment (160 kg N ha⁻¹) in 2022 and 2023, respectively.

In the years 2022 and 2023, we observed significant and systematic disparities in the performance of direct seeded rice concerning a spectrum of vegetation indices. These differences were particularly notable in response to varying levels of nitrogen treatments, as comprehensively tabulated in Table 5. NDVI demonstrated a conspicuous trend of enhancement as N fertilizer levels incrementally increased from N₀ (0 kg N ha⁻¹) to N₆ (240 kg N ha⁻¹) across all growth stages of the rice crop. In comparison to the control treatment, treatment N₆ (240 kg N ha⁻¹) exhibited remarkable increments in NDVI values, with a 40% increase in 2022 and a 44.12% increase in 2023 at 40 DAS, a 34.69% increase in 2022 and a 34.04% increase in 2023 at 60 DAS, and a 16.67% increase in 2022 and a 30.65% increase in 2023 at 80 DAS. Similarly, RVI displayed a coherent elevation in response to the progressive escalation of N fertilizer levels (0 kg N ha⁻¹ to 240 kg N ha⁻¹). Once again, treatment N₆ (0 kg N ha⁻¹) emerged with the highest RVI values, underscoring the beneficial influence of elevated N levels on vegetation vigor. In comparison to the control treatment (0 kg N ha⁻¹), treatment N₆ exhibited remarkable increments in RVI values, with a 53.19% increase in 2022 and a 58.04% increase in 2023 at 40 DAS, a 34.09% increase in 2022 and a 26.04% increase in 2023 at 60 DAS, and a 21.94% increase in 2022 and a 22.75% increase in 2023 at 80 DAS. The NDRE values for both years exhibited a conspicuous trend of improvement alongside increasing N levels. In comparison to the control treatment (0 kg N ha⁻¹), treatment N₆ (240 kg N ha⁻¹) registered remarkable increments in NDRE values, with a 50% increase in 2022 and a 58.82% increase in 2023 at 40 DAS, a 52.63% increase in 2022 and a 54.54% increase in 2023 at 60 DAS, and a 42.31% increase in 2022 and a 50% increase in 2023 at 80 DAS. Furthermore, GNDVI values mirrored the overarching trend of improvement in conjunction with escalating N fertilizer levels. Treatment N₆ (240 kg N ha⁻¹) exhibited notable increments in GNDVI values compared to the control treatment, with a 52% increase in 2022 and a 48.15% increase in 2023 at 40 DAS, a 37.5% increase in 2022 and a 46.87% increase in 2023 at 60 DAS, and a 38.89% increase in 2022 and a 35.14% increase in 2023 at 80 DAS.

The results indicated that different nitrogen treatments had a significant (p < 0.05) impact on the yield attributes and grain yield for transplanted rice over the 2 years of the study (Table 6). In terms of the number of effective tillers per hill, displayed significant variations among the treatments. Treatment N₄ (160 kg N ha⁻¹) exhibited the highest values (17.54 and 18.80) in both 2022 and 2023 cropping season. The number of panicles per hill showed significant variations among treatments in both 2022 and 2023. Notably, Treatment N₄ (160 kg N ha⁻¹) consistently resulted in the highest number of panicles m⁻² (348.72 in 2022 and 335.28 in 2023) in both the years. Figure 6A depicts the linear relation relationship between the number of panicles and grain yield. Treatment N₄ (160 kg N ha⁻¹) consistently exhibited the highest number of filled grains per panicle (71.76 in 2022 and 74.55 in 2023) in both years, suggesting their positive influence on grain filling and yield potential (Figure 6B). Treatments N₄ (160 kg N ha⁻¹) and N₅ (200 kg N ha⁻¹) consistently led to longer panicles in both years (27.70 cm in 2022 and 27.94 cm in 2023). Figure 6C shows the direct relationship between panicle length and grain yield. Treatment N₄ (160 kg N ha⁻¹) displayed the highest 1,000-grain weight (26.86 g and 26.92 g) in 2022 and 2023, highlighting the influence of nitrogen on grain size and density. Figure 6D depicts the linearity between 1,000-grain weight and grain yield. As the N levels increased the grain yield also increased up to N₄ treatment (160 kg N ha⁻¹), after that the grain yield reduced in both the years, as shown in Figure 7A. Treatment N₄ (160 kg N ha⁻¹) consistently resulted in the highest grain yield m⁻² (481.60 g and 495.65 g) in both 2022 and 2023, emphasizing its effectiveness in maximizing grain production. Treatment N₆ (240 kg N ha⁻¹) often had the highest straw yield m⁻² (765.04 g in 2022 and 785.61 g in 2023) in both the year. Treatment N₃ (120 kg N ha⁻¹) consistently displayed the highest harvest indices (40.82% in 2022 and 39.37% in 2023) in both years. Using the obtained quadratic relationship (y = −0.0846x² + 29.416x + 2178.8) between the rates of applied nitrogen (x) and the resulting grain yield (y) for TR, we determined the optimal N levels, for maximizing economic profit. we considered maximizing profit by finding the N level that maximizes the ratio of price per unit nitrogen (P_x, ₹11.65) to price per unit grain yield (P_y, ₹21.83). Solving the quadratic equation, we found that applying 170.69 kg N hectare⁻¹ would generate the maximum profit, leading to a calculated rice production of 4733.97 kg hectare⁻¹ at this N level.

The findings of the 2-years study revealed that nitrogen treatments exerted a statistically significant (p < 0.05) influence on the yield attributes and grain yield in direct seeded rice, as presented in Table 7. Notably, Treatment N₄ (160 kg N ha⁻¹) consistently exhibited the highest values during both the 2022 and 2023 cropping seasons, signifying its capacity to stimulate robust tillering. The number of panicles per hill demonstrated notable disparities across the treatments in both 2022 and 2023. Noteworthy was the consistent predominance of Treatment N₄ (160 kg N ha⁻¹), consistently yielding the highest number of panicles m⁻² (370.52 in 2022 and 376.85 in 2023). Figure 6A’ illustrates the linear relationship between the number of panicles and grain yield. Regarding the number of filled grains per panicle, treatment N₄ (160 kg N ha⁻¹) consistently showcased the highest values (60.05 in 2022 and 62.43 in 2023) in both years. This relationship is depicted in Figure 6B’. Treatment N₅ consistently led to longer panicles (26.16 cm in 2022 and 27.05 cm in 2023) during both years, implying its role in enhancing panicle elongation. Figure 6C’ visualizes the direct connection between panicle length and grain yield. Treatment N₄ exhibited the highest 1,000-grain weight in 2022 (26.30 g) and 2023 (26.82 g), underscoring the substantial influence of nitrogen on grain size and density. Figure 6D’ portrays the linear correlation between 1,000-grain weight and grain yield. As the nitrogen levels increased, grain yield also experienced a corresponding rise up to the N₄ treatment, after which grain yield reduced significantly, as depicted in Figure 7B. Treatment N₄ consistently yielded the highest grain output during both the 2022 (441.48 g m⁻²) and 2023 (456.23 g m⁻²) cropping seasons, highlighting its effectiveness in maximizing grain production in DSR. There was no statistically significant difference in DSR grain yield between the treatments N₃ (120 kg N ha⁻¹) and N₄ (160 kg N ha⁻¹) in both the years. Treatment N₆ (240 kg N ha⁻¹) frequently produced the highest straw yield (688.89 g m⁻² in 2022 and 726.97 g m⁻² in 2023) throughout both years. Treatment N₃ (120 kg N ha⁻¹) displayed the highest harvest index of 39.68% in 2022 and 40.33% in 2023.Utilizing the derived quadratic correlation (y = −0.0879x² + 29.246x + 2081.4) between varying nitrogen application rates (x) and resultant grain yield (y) in the context of DSR, we conducted calculations to ascertain the optimal N levels, in terms of maximizing economic profit using the same methods used in TR. Solving the quadratic equation, we determined that applying 163.32 kg N hectare⁻¹ would yield the maximum profit, resulting in a calculated rice production of 4513.26 kg hectare⁻¹ at this optimal economic level.

The 2-year mean experimental results highlighted the influence of nitrogen treatments on the cost economics of both transplanted and direct seeded rice (Table 8). In the case of TR and DSR, the cost of production varied with nitrogen treatments. In TR, the highest cost of production was observed for treatment N₆ (240 kg N ha⁻¹), reaching $519.72 per hectare, while the lowest was for treatment N₀ (0 kg N ha⁻¹) at $486.16 per hectare. The gross return ranged from $695.91 (at 0 kg N ha⁻¹) to $1370.01 (at 160 kg N ha⁻¹), resulting in net returns between $209.75 (at 0 kg N ha⁻¹) and $861.48 (at 160 kg N ha⁻¹). The results indicate a range of benefit–cost ratio (BCR) values, with each treatment combination yielding a different level of profitability. The BCR values ranged from 0.43 (at 0 kg N ha⁻¹) to 1.71 (at 120 kg N ha⁻¹).

In the case of DSR, the cost of production ranged from $399.78 (at 0 kg N ha⁻¹) to $433.34 (at 240 kg N ha⁻¹) per hectare. Gross returns varied between $658.06 (at 0 kg N ha⁻¹) and $1258.22 (at 240 kg N ha⁻¹), resulting in net returns ranging from $258.28 (at 0 kg N ha⁻¹) to $836.07 (at 240 kg N ha⁻¹). The BCR for DSR treatments ranged from 0.65 (at 0 kg N ha⁻¹) to 2.00 (at 120 kg N ha⁻¹), indicating favorable returns on investment from treatment N₂ to N₆. DSR cultivation resulted in higher BCR than transplanted rice at treatment N₃ (120 kg N ha⁻¹). While the cost of production, gross returns, and net returns varied across treatments, the benefit–cost ratios suggest the profitability and economic viability of certain nitrogen treatments. However, Treatment N₃ (120 kg N ha⁻¹) resulted in highest BCR in transplanted (1.71) and direct seeded rice (2.00), indicating the highest returns on investment which was at par with treatment N₄ (160 kg N ha⁻¹).

The effect of N levels on plant height and LAI in both TR and DSR is a crucial aspect of comprehending rice cultivation dynamics. Our study revealed that elevated N levels had a positive impact on plant height in both TR and DSR, a phenomenon rooted in fundamental biological processes. Nitrogen is an indispensable element for the synthesis of amino acids, proteins, and chlorophyll, pivotal components for rice plant growth. An ample supply of nitrogen stimulates the production of proteins and enzymes directly involved in cell elongation, thereby promoting increased plant height (Bao et al., 2015; Upadhyay et al., 2022).

Furthermore, at comparable growth stages, DSR exhibited superior plant growth compared to TR. Saha et al. (2020) similarly reported significantly greater plant height in direct seeded rice compared to transplanted rice and noted that transplant shock in TR delayed growth, reduced tillering, and foliage development. The enhanced plant height in DSR can be attributed to advantages such as seed-based root establishment, absence of transplant shock, reduced interplant competition due to wider row spacing, a prioritized early root development strategy through resource allocation, adaptation to local conditions, and potentially more favorable environmental factors. These combined factors contribute to the overall enhanced plant growth observed in DSR.

The impact of varying N levels on rice LAI was conspicuous, as escalating N levels consistently led to increased LAI values in both TR and DSR. This pattern aligns with the established role of nitrogen in bolstering chlorophyll synthesis and leaf expansion, thereby augmenting photosynthetic capacity (Peng et al., 2021). However, it is essential to emphasize that there existed an optimal N level for optimizing LAI in both DSR and TR, beyond which excessive N application precipitated nutrient uptake imbalances, environmental apprehensions, and heightened vulnerability to lodging (Luo et al., 2022) and diseases. At the initial stage, DSR exhibited a lower LAI in comparison to TR at equivalent growth stages and N levels; however, by 80 DAS, DSR surpassed TR in LAI. Treatment N₃ (120 kg N ha⁻¹), N₄ (160 kg N ha⁻¹), N₅ (200 kg N ha⁻¹), and N₆ (240 kg N ha⁻¹) demonstrated statistical parity in LAI under TR, whereas treatment N₄ (160 kg N ha⁻¹), N₅ (200 kg N ha⁻¹), and N₆ (240 kg N ha⁻¹) exhibited statistical equivalence in LAI under DSR. Similar findings were corroborated by Meena et al. (2023) and Pattanayak et al. (2023).

Vegetation indices (VIs) (NDVI, RVI, NDRE and GNDVI) were greatly influenced by applied N levels and cultivation methods (TR and DSR). VIs increased with increasing the N levels in both TR and DSR. Elevated nitrogen levels lead to a higher chlorophyll concentration in plant leaves, improving the plant’s capacity to absorb sunlight and convert it into energy (Peng et al., 2021). Radiation absorption is improved when chlorophyll concentration is higher in leaves. All plants transmit a significant portion of near infrared (NIR) radiation since these wavelengths are not absorbed by photoreceptor cells. However, DSR exhibited less values as compared to TR at equivalent growth stages and N doses.

Notably, 120 kg N ha⁻¹, 160 kg N ha⁻¹, 200 kg N ha⁻¹, and 240 kg N ha⁻¹ treatments showed no statistically significant differences in NDVI, RVI, NDRE, and GNDVI values across the various growth stages of transplanted and direct seeded rice in both the years, indicating their similar effects on rice vegetation indices. The consistent trends across 2 years affirm the influence of different N fertilizer treatments on vegetation indices, with specific treatments consistently yielding higher values. Similar findings were reported by Kumar et al. (2023) where higher values of NDVI were obtained at higher nitrogen dose and growth stage.

Nitrogen fertilization and the choice of rice cultivation methods, namely TR and DSR, exerted significant effects on multiple agronomic parameters. Notably, grain yield exhibited a significant quadratic response to varying N levels across both years and cultivation methods. This observation aligns with findings reported by Liu et al. (2016) and Jahan et al. (2022). Both the cultivation methods (TR and DSR) resulted in highest grain yield at treatment N₄ (160 kg N ha⁻¹) which was at par with treatment N₃ (120 kg N ha⁻¹). In comparison to treatment N₄, the grain yield exhibited reductions of 9.70 and 9.78% under TR at N₅ (200 kg N ha⁻¹) and N₆ (240 kg N ha⁻¹) treatments, and 7.39 and 9.09% under DSR at N₅ and N₆ treatments, respectively. The grain yield in both TR and DSR is contingent upon distinct yield components, including the number of panicles m⁻², the spikelets panicle⁻¹, thousand-grain weight, and the proportion of filled spikelets. The consensus in the literature underscores the pivotal role of panicle density in attaining higher yields, particularly in DSR (Gunawardena and Fukai, 2005; Tayefe et al., 2014). It is noteworthy that TR consistently outperformed DSR in terms of grain yield across various N levels, a finding that is congruent with the results obtained by Akhgari and Kaviani (2011) and Bastola et al. (2021). A primary contributing factor to the reduced yield in DSR as compared to puddled transplanting rice is the higher incidence of spikelet sterility, as elucidated by Xu et al. (2019) and Ishfaq et al. (2020). The number of panicles m⁻² increased with escalating N levels in both cultivation methods. Among these, treatments N₃, N₄, N₅, and N₆ exhibited statistically equivalent numbers of panicles per hill. This trend aligns with the findings reported by Artacho et al. (2009), which underscored a linear relationship between N fertilization and the number of panicles per square meter. Furthermore, it is worth noting that DSR, in comparison to TR, produced a higher number of panicles m⁻² but a lower number of spikelets panicle⁻¹, as corroborated by Huang et al. (2011). The increased number of panicles m⁻² in DSR primarily stemmed from a higher number of tillers m⁻², as the number of tillers hill⁻¹ remained modest in DSR plants. In terms of the thousand-grain weight, a pivotal parameter for achieving higher grain yield, TR consistently outperformed DSR across all N treatments. These outcomes align with prior research conducted by Naresh et al. (2013) and Bastola et al. (2021), which concurred that transplanted rice exhibited a higher thousand-grain weight in comparison to direct seeded rice.

Our study revealed that, in the context of DSR cultivation with varying levels of nitrogen treatments, the net returns were statistically at par to those of TR. However, the benefit–cost ratio was notably higher in DSR compared to TR. Specifically, in DSR with RDN treatment (N₃), the cost–benefit ratio surpassed that of TR by 16.96%. This observation aligns with the findings of Chakraborty et al. (2017), who also reported higher net economic returns for farmers in wet or dry DSR compared to TR, with a 13% increase in returns for wet DSR. In contrast to TR, the cost of production resulted in a decrease of US$86.38 per hectare in the case of DSR. In a broader context, an extensive analysis of 77 published studies across Asian countries conducted by Kumar and Ladha (2011) demonstrated that various DSR methods led to a reduction in production costs by a range of US$9 to US$125 hectare⁻¹ when compared to TR.