Integrated water–energy–nitrogen performance under nitrogen and biochar treatments in shallow groundwater tomato production

Efficient management of water, energy, and nitrogen is essential for sustaining irrigated crop production in arid regions. However, under hydrologically buffered conditions, similar crop yields may conceal substantial differences in input demand and resource productivity. This study evaluated the integrated water-energy-nitrogen performance of tomato cultivated in clay loam soil under shallow groundwater influence and drip irrigation managed at a uniform soil moisture level. A two-season field experiment compared three N-rates: 431, 345, and 215 kg N ha -1, corresponding to 100%, 80%, and 50% of the full N-rate, with and without biochar application. Fresh yield did not differ significantly among treatments (p>0.05) and ranged from 81.4 to 108.6 t ha -1 in pooled treatment means. Water productivity showed a significant Season × N-rate interaction (F 2,8.52 = 5.50, p=0.029), although no Tukey-adjusted pairwise comparison among N-rates was significant within either season. Irrigation energy ranged from 1.27 to 1.83 GJ ha -1 and was not significantly affected by season, biochar, N-rate, or their interactions (p>0.146). Fertilizer-associated N energy was 26.14, 20.92, and 13.08 GJ ha -1 under the 100%, 80%, and 50% N-rates, respectively. This corresponds to reductions of ~20% and 50% relative to the full N-rate. Fruit quality responses were governed mainly by season, with management effects restricted to specific variables and treatment contexts. The 80% N-rate included the highest pooled-yield treatment, W8 at 108.6 t ha -1, while reducing fertilizer-associated energy by approximately 20%. The 50% N-rate produced the lowest total accounted energy, ranging from 14.43 to 14.91 GJ ha -1, and the highest numerical input-productivity values; B5 achieved a water productivity of 39.59 kg m -3, an energy productivity of 6.87 kg MJ -1, and partial factor productivity of N of 459.09 kg yield kg -1 N. These findings show that limited yield differentiation did not imply equivalent management performance. Under the shallow-groundwater conditions evaluated, the 80% N-rate represented a candidate production-input compromise, whereas the 50% N-rate prioritized lower energy demand and higher numerical input productivity. Integrated indicators therefore provided management information that was not evident from yield alone, without identifying a statistically established universal optimum.

Water scarcity is a major constraint to agricultural production in arid and semi-arid regions. High temperatures, low and variable rainfall, and strong evaporative demand reduce water availability and increase the risk of crop losses, while competition among agricultural, industrial, and domestic sectors further intensifies pressure on limited supplies. These regions also face groundwater depletion, soil degradation, and increasing climatic variability, reinforcing the need for management strategies that maintain crop production while reducing water, energy, and nutrient inputs (Karimov et al., 2018; Haghighi et al., 2020; Priyan, 2021; Patel, 2023; Tanwar et al., 2023; Tan et al., 2024; Zhong et al., 2025). Tomato is a high-value horticultural crop commonly cultivated under drip irrigation and is sensitive to both water and nutrient management. Its relatively high water and N requirements make it suitable for evaluating interactions among irrigation performance, energy demand, and nutrient-use efficiency (Agbna et al., 2017; Guo et al., 2022; Karamian et al., 2023; Sibanda and Bacela, 2024; Amani et al., 2025).

Crop yield is commonly used as the principal criterion for evaluating irrigation and fertilizer management, but it may provide limited discrimination under hydrologically buffered conditions. In medium- and fine-textured soils, high soil-water storage can reduce differences in plant water availability among treatments, particularly when sensor-based irrigation maintains root-zone moisture near a defined threshold (Li et al., 2015; Zaccaria and Bali, 2024). Shallow groundwater may provide an additional source of hydrological buffering when the water table lies within the capillary-influence zone. Under such conditions, differences in surface water or nutrient management may alter water distribution and input demand without producing a proportional yield response. Previous work at the same experimental setting showed that biochar and N management altered vertical soil-water distribution and downward water flux under shallow-groundwater conditions, whereas yield responses remained limited (Elbana et al., 2025). These findings are consistent with a buffered root-zone environment, although the upward groundwater contribution was not quantified directly. Consequently, a nonsignificant yield response should not necessarily be interpreted as evidence that management programs are equivalent in water use, energy demand, or input productivity.

Biochar has been widely investigated as a soil amendment capable of modifying soil structure, water retention, hydraulic behavior, and nutrient availability. These effects may improve water and nutrient management under arid conditions, but field responses vary with soil texture, biochar characteristics, application rate, and interactions with fertilizer management (Haghighi et al., 2020; Han et al., 2024; Elbana et al., 2025). Field evidence from fine-textured soils affected by shallow groundwater remains comparatively limited. In such soils, biochar may influence water redistribution or drainage without necessarily increasing yield when crop water supply is already buffered. Evaluating biochar solely through yield may therefore overlook changes in water use or resource demand.

Nitrogen management presents a related challenge. Nitrogen is essential for crop growth, yield formation, and fruit quality, but excessive application contributes to nitrate leaching, soil acidification, greenhouse-gas emissions, and unnecessary embodied-energy demand (Hakeem et al., 2017; Zhao et al., 2019; Losacco et al., 2021; Chen et al., 2025; El-Emam, 2026). Moderate reduction in N-input may maintain crop production while improving input productivity and reducing environmental burdens, although the response depends on water supply, soil properties, and seasonal conditions (Zhao et al., 2019; Bhatt et al., 2025; Wang et al., 2025). Fertilizer-associated energy is particularly relevant because mineral-N production is energy intensive. Thus, two N-rates that produce similar yields may differ substantially in total accounted energy and partial factor productivity of N (Mandić et al., 2025; Fang et al., 2026).

Sensor-based irrigation scheduling can improve the timing and precision of water application and stabilize root-zone moisture. However, relatively few field studies have combined soil-moisture controlled irrigation, biochar, N management, fruit quality, and energy accounting within a unified analytical framework, particularly under shallow groundwater conditions (Hamidov and Helming, 2020; Taguta et al., 2022). Recent studies have consequently emphasized the need for joint evaluation of complementary indicators using consistent calculation procedures rather than relying on yield alone (Kyriakopoulos, 2023; Lin et al., 2025).

The present study therefore evaluated the combined effects of biochar application and three N-rates under soil-moisture-controlled drip irrigation over two growing seasons. Marketable fresh yield and fruit-quality attributes were assessed together with irrigation water applied, water productivity, irrigation energy, fertilizer-associated N energy, total accounted energy, energy productivity, and partial factor productivity of N. All treatments were managed under the same soil moisture controlled drip irrigation protocol; irrigation scheduling was therefore a common experimental condition rather than an experimental factor. The manipulated factors were N-rate and biochar application. The objective was not to identify a universally superior treatment based solely on yield, but to determine whether management programs with limited yield differentiation exhibited contrasting water-energy-nitrogen performance profiles.

Two inferential null hypotheses were evaluated: H 0Y, that N-rate and its interactions with season and biochar would not significantly affect marketable fresh yield; and H 0WP, that N-rate and its interactions with season and biochar would not significantly affect water productivity.

Fertilizer associated energy was determined directly from the assigned N-input, while energy productivity and partial factor productivity of N were derived indicators summarized descriptively. They were therefore not included as formal inferential null hypotheses. Instead, the study evaluated whether reduced N-rates lowered accounted energy and increased numerical input productivity while maintaining crop yield and fruit quality under shallow-groundwater conditions.

Study area and experiment design

The field experiment was conducted at the experimental farm of the Arid Lands Agricultural Studies and Research Institute, Ain Shams University, Egypt. The soil is classified as clay loam, with 34.4% clay, 35.9% silt, and 29.7% sand. Tomato is grown during Spring and Autumn seasons in this region. The climate is arid with negligible rainfall. Mean air temperatures ranged from 20 to 34 °C during the Spring season and 15 to 26 °C during the Autumn season.

Although based on the same two-season field experiment reported by (Elbana et al., 2025), this study addresses different research questions. The previous work examined soil-water distribution and downward water flux under shallow groundwater conditions, whereas the present study evaluates marketable fresh yield, fruit quality, irrigation water use, partial energy accounting, and nitrogen-use productivity. Soil-water profile and drainage data from the earlier study are cited solely for hydrological context and were not reanalyzed. The shallow water table was considered a boundary condition affecting irrigation scheduling and crop response, without repeating hydraulic modeling. Tomato (Solanum lycopersicum L.), cv. Rich Zoan K186 was grown during two seasons: Spring 2023 and Autumn 2023-2024.

Experimental design and treatments

The experiment consisted of two biochar levels, 0 or 35 t ha -1, combined factorially with three N-rates corresponding to 100%, 80%, and 50% of the local farmer practice. The reference rate was 431 kg N ha -1; consequently, the three nitrogen treatments received 431, 345, and 215 kg N ha -1, respectively. The six treatment combinations and their codes are presented in Table 1.

The biochar of 35 t ha -1 was selected before establishment of the experiment as an intermediate field application rate within the range of 25–50 t ha -1 reported in previous studies for tomato production and for soil amendment under arid or reduced-nitrogen conditions (Li et al., 2018; Zhang et al., 2020; Guo et al., 2021; Diedhiou et al., 2024). The objective was to compare the presence and absence of biochar across nitrogen rates rather than to determine the optimum biochar application rate. Each plot measured 5.5 × 2.3 m. Plots were hydraulically isolated using trenches (0.7 m deep × 0.3 m wide), to minimize lateral water movement. The experiment was arranged in a randomized complete block design (RCBD) with three replicates.

The biochar (corn-cob-derived) was incorporated into the top 0.3-m soil layer one week before transplanting. The biochar characterization and soil preparation procedures are described in detail in an earlier publication (Elbana et al., 2025). Nitrogen was supplied as ammonium sulfate and ammonium nitrate in four split applications at 15, 50, 70, and 90 days after transplanting.

Irrigation system and scheduling

A pressurized drip irrigation system was used to apply water to each plot through a dedicated valve to allow independent scheduling. Soil water content (SWC) in the effective root zone (0 – 0.4 m) was monitored daily using a PR2/6 time-domain reflectometry (TDR) probe (Delta-T Devices Ltd., UK). Calibration procedures followed the established protocol for the site (Elbana et al., 2025).

Irrigation events for each plot were triggered when root-zone SWC approached 50% of available water. Application volumes were determined to restore the soil profile to field capacity. Irrigation scheduling was maintained as a constant experimental condition and was not considered an experimental factor. This ensured consistent soil moisture control across treatments, enabling a comparison of water, energy, and nitrogen performance rather than induced water-stress responses.

System performance metrics and analytical framework

In this study, integrated assessment refers to the joint examination of original-scale crop-, water-, energy-, and N-input indicators, supported by normalized descriptive visualizations. It does not constitute a new composite index, an indicator-weighting procedure, or formal multi-criteria optimization. System performance was evaluated using indicators of water use, operational energy input, nitrogen use, crop productivity, and fruit quality. The analysis focused on seasonal, field-scale management performance and included irrigation water applied, irrigation pumping energy, embodied energy associated with N-fertilizer production, marketable fresh yield, water productivity, energy productivity, and partial factor productivity of N. Energy associated with irrigation infrastructure, biochar production and application, other farm inputs, and post-harvest operations was excluded. The resulting energy balance therefore represents a partial operational-energy accounting boundary rather than a complete life cycle assessment.

Water use indicators

Seasonal irrigation water applied (I) was measured using flowmeter (ISO 4064, Military Production Co., Egypt) installed on the sub-main line. Water productivity (WP) was calculated as:

where Y is marketable fresh yield (kg ha -1) and I is in m 3 ha -1.

Energy accounting

Irrigation energy

Irrigation water was pumped using an electric centrifugal pump (CALPEDA INN 25/180 AE; 1.5kW, 2900 rpm, power factor 0.81). Seasonal irrigation energy consumption (E irr, MJ ha -1) was calculated as:

where P is pumping power (kW), t is the operating time per hectare (h ha -1), and 3.6 converts kWh to MJ. Because all plots were irrigated using the same pumping and distribution system configuration, pump power and distribution conditions were considered uniform among treatments.

Nitrogen fertilizer energy

Nitrogen fertilizer was manually applied using ammonium sulfate containing 21% N and ammonium nitrate containing 34% N. Fertilizer quantities recorded at each application event in g plot -1 were converted to a hectare basis using the corresponding experimental plot area, which averaged approximately 12 m 2. Seasonal elemental-N-input was calculated by summing the contribution of all fertilizer application events:

where F j is the fertilizer product application rate at event j, expressed on a hectare basis, and f N,j is the elemental N fraction of that fertilizer.

The embodied energy associated with N-fertilizer production (E N, MJ ha -1) was calculated following (Iqbal et al., 2024)

where EC N is the embodied energy coefficient for N-fertilizer production, taken as 60.6 MJ Kg -1 N. Because E N was calculated directly from the assigned N-input and a fixed conversion coefficient, it was treated as a deterministic accounting indicator and summarized descriptively rather than subjected to inferential statistical testing.

Energy associated with biochar production and application was excluded because biochar was applied once before the experiment and its effects were expected to extend beyond the two seasons evaluated.

Total accounted energy and energy productivity

Total accounted seasonal energy input (E tot, MJ ha -1) was calculated as:

Energy productivity (EP, kg MJ -1) was calculated as:

where Y is the marketable fresh yield in kg ha -1.

Yield, fruit quality, and nitrogen-use indicators

Marketable fresh yield was determined from the central 1 m 2 of each plot and converted to a hectare-equivalent basis. Fruit hardness, (kg cm -2), titratable acidity (%), nitrogen content (%), ammonium-N (ppm), and nitrate-N (ppm) were measured using standard laboratory procedures. Partial factor productivity of nitrogen (PFP N, kg yield kg -1 N applied) was calculated as:

where N applied was 431, 345, or 215 kg N ha -1, depending on the assigned N program. Plant N uptake was not calculated because plant-tissue N accumulation data were unavailable.

Integration of water, energy, and nitrogen indicators

Yield, WP, E tot, EP, and PFP N were integrated using radar plots, whereas E irr and E N were displayed using a standardized heatmap. Indicator values were first summarized within each Season × Treatment combination and then averaged across seasons, thereby giving equal weight to the two growing seasons. Treatments were arranged in matched biochar and non-biochar pairs – B1/W1, B8/W8, and B5/W5 – to facilitate descriptive comparison within each N program.

Because the indicators were expressed in different units, min-max normalization was applied. For the radar plots, indicators were directionally normalized so that higher standardized values consistently represented more favorable performance. Yield, WP, EP, and PFP N were normalized directly:

Total accounted energy was reverse-normalized because lower energy input was considered preferable:

where X i is the pooled treatment mean and X min and X max are the minimum and maximum pooled means for the corresponding indicator. Indicators lacking between-treatment variation were assigned a normalized value of 0.5 to avoid division by zero. Consequently, larger radial values represented more favorable directional performance. Nevertheless, radar-polygon area was not used as a formal composite score because the indicators were equally weighted and the polygon geometry depends on axis order and scaling.

For the heatmap E irr and E N were independently normalized using direct min-max scaling. Thus, color intensity represented the relative magnitude of each treatment within a given energy indicator but did not preserve absolute magnitude differences between the two energy components.

The radar plots and heatmap were used exclusively as descriptive visualization tools to illustrate relative patterns and management trade-offs. They were not used as inferential tests, and differences in polygon shape or heatmap color intensity were not interpreted as evidence of statistically significant treatment effects.

Reproducibility and statistical analysis

All data processing, statistical analysis, and graphical visualizations were performed in R version 4.5.1 (The R Core Team, 2025). Data import, cleaning, transformation, and summarization were conducted using the readxl, dplyr, tidyr, and broader tidyverse packages (Wickham et al., 2019). Bar charts and heatmaps were generated using ggplot2 (Wickham, 2016), whereas radar or spider plots were produced using the fmsb package (Nakazawa, 2010).

The effects of season, biochar application, nitrogen fertilization rate, and their interactions on marketable fresh yield, fruit-quality variables, water productivity, and irrigation energy were evaluated using separate linear mixed-effects models. Because the same physical plots were evaluated in both growing seasons, plot identity, defined by the replicate × biochar × nitrogen-rate combination, was included as a random intercept to account for repeated observations from the same experimental unit. Season, biochar, nitrogen rate and all two- and three-way interactions were included as fixed effects. Models were fitted using the lme4 (Bates et al., 2015) and lmerTest (Kuznetsova et al., 2017) packages, and fixed effects were evaluated using Type III F-tests with Kenward-Roger denominator degrees of freedom, as implemented through the pbkrtest framework (Halekoh and Højsgaard, 2014).

Model assumptions were assessed using residual-versus-fitted plots, normal Q-Q plots, and Shapiro-Wilk tests of model residuals. Model convergence, random-effect variances and singularity were also examined. When the estimated plot-level variance approached zero and a model produced a singular fit, a fixed-effects sensitivity model retaining replicate as a blocking factor was fitted to assess the robustness of the conclusions.

Estimated marginal means and model-based standard errors were generated using the emmeans package (Lenth and Piaskowski, 2017). When a significant interaction involving nitrogen rate was detected at p<0.05, Tukey-adjusted pairwise comparisons among nitrogen rates were conducted within each Season × Biochar combination. When no interaction involving N-rates was significant, pairwise comparisons among N-rates were performed only when the N-rates main effect was significant.

Fertilizer-associated N energy (E N) was not subjected to inferential testing because it was calculated deterministically from the assigned N-input and a fixed embodied-energy coefficient. Total accounted energy, energy productivity, and partial factor productivity of N were summarized descriptively. Values used in the heatmap, and radar plots were first summarized within each Season × Treatment combination and then averaged across seasons. The visualizations were used only to illustrate relative treatment patterns and were not interpreted as statistical tests or formal composite rankings. The dataset and complete R scripts are available through the Mendeley Data repository cited in the ‘Data Availability Statement.’

Water use, water productivity, and yield

Applied irrigation water and fresh marketable yield showed limited variation among the biochar × N-rate combinations, and no significant treatment effects were detected for either variable (p>0.05; Figure 1). Water productivity showed a significant Season × N-rate interaction (F 2,8.52 = 5.50, p = 0.029). This indicates that the response of WP to N-rate differed between the two growing seasons. However, none of the Tukey-adjusted pairwise comparisons among N-treatments within either season was significant. Thus, the treatment means shown in Figure 1 represent numerical patterns rather than statistically established differences among individual N-rate treatments.

Across the pooled treatment means, biochar-amended treatments generally showed numerically lower irrigation-water use and, in some cases, higher WP than the corresponding non-biochar treatments. These differences were not statistically significant.

Fruit quality responses

Fruit quality responses were influenced primarily by season, with fewer effects attributable to biochar or N-rate application (Table 2; Supplementary Table 1). Fruit acidity was significantly affected by season (F 1,12 = 119.91, p<0.001) and biochar application (F 1,12 = 8.96, p = 0.011). Significant Season × Biochar (F 1,12 = 5.25, p = 0.041) and Biochar × N-rate interactions (F 2,12 = 8.00, p = 0.006) were also detected. Tukey-adjusted comparisons showed that, in spring under biochar application, fruit acidity was significantly higher under the full N-rate treatment B1 than under the 80% N-rate treatment B8. Treatment B5 was intermediate and did not differ significantly from either B1 or B8. No other comparison among N-rates within a Season × Biochar combination was significant.

Fruit N concentration differed significantly between seasons (F 1,12 = 18.04, p = 0.001) and showed a significant Season × Biochar interaction (F 1,12 = 7.15, p = 0.020). Fruit N concentration was numerically lower under biochar application in spring but slightly higher under biochar in Autumn. Neither the N-rate main effect nor any interaction involving N-rate was significant in the primary mixed model. Consequently, no N-rate superscript letters were assigned for fruit N in Supplementary Table 1.

Fruit hardness was affected only by season (F 1,12 = 6.90, p = 0.022). Biochar, N-rate, and all interaction terms were nonsignificant (p>0.05), and no treatment-level post hoc comparisons were warranted. Fruit NH 4 -N concentration showed a significant Biochar × N-rate interaction (F 2,12 = 4.71, p = 0.031). However, none of the Tukey-adjusted comparisons among N-rates within the individual Season × Biochar combinations were significant. Superscript letters were therefore omitted from the NH 4 -N column in the Supplementary Table 1. Fruit NO 3 -N concentration was affected only by season (F 1,12 = 12.22, p = 0.004). Biochar, N-rate, and all interactions were nonsignificant, and no N-rate post hoc comparisons were therefore conducted.

The Type III F-tests in Table 2 evaluate overall main effects and interactions, whereas the superscript letters in Supplementary Table 1 represent specific Tukey-adjusted comparisons among N-rates. Accordingly, a significant overall interaction did not necessarily result in significant separation among individual treatment means. Different superscript letters were assigned only for acidity, for which a significant adjusted pairwise difference were detected.

Energy use and energy indicators

The mixed model analysis indicated that irrigation energy consumption (E irr) was not significantly affected by season, biochar application, N-rate, or their interactions (p > 0.146). Nevertheless, numerical variations occurred among the pooled treatment means.

Pooled E irr ranged from 1,270.5 MJ ha -1 under B1 to 1,831.5 MJ ha -1 under W5. Within each N-rate, the biochar treatment had numerically lower irrigation energy than the corresponding treatment without biochar: 1,270.5 versus 1,359.4 MJ ha -1 for B1 and W1, 1,405.1 versus 1,666.1 MJ ha -1 for B8 and W8, and 1,351.5 versus 1,831.5 MJ ha -1 for B5 and W5, respectively. Because these differences were not statistically significant, they should be interpreted as descriptive patterns rather than evidence of a consistent biochar effect on irrigation-energy demand.

Fertilizer-associated N energy (E N) was determined directly by the amount of N applied. The 100%, 80%, and 50% N-rates supplied 431, 345, and 215 kg N ha -1, corresponding to E N values of 26,143.4, 20,924.6, and 13,078.1 MJ ha -1, respectively. Relative to the full N-rate, the 80% and 50% rates reduced N-input and its associated embodied energy by approximately 20% and 50%, respectively. Because these differences arose directly from the assigned fertilizer inputs, E N was not subjected to inferential statistical testing. Across treatments, E N was approximately 7.1-20.6 times greater than E irr, making fertilizer associate N energy the dominant component of the accounted energy input. Consequently, total accounted energy (E tot) largely followed the N-input pattern.

The highest values occurred under the full N-treatments, reaching 27,413.9 MJ ha -1 for B1 and 27,502.8 MJ ha -1 for W1. Intermediate values were recorded under B8 and W8, at 22,329.7 MJ ha -1 and 22,590.7 MJ ha -1, respectively. The lowest values occurred under B5 and W5, at 14,429.6 and 14,909.6 MJ ha -1, respectively. Irrigation energy intensity remained constant at 0.54 MJ m -3 reflecting the common pumping-energy requirement per unit of irrigation water across treatments. The standardized heatmap illustrates these relative energy patterns (Figure 2). The full N-rates had the highest standardized E N values. The 80% N-rates occupied an intermediate position, and the 50% N-rates had the lowest values. For irrigation energy, B1 had the lowest relative value and W5 the highest. Within each N-rate, the paired biochar and non-biochar treatments had identical E N, whereas their heatmap differences resulted from numerical variations in E irr.

Integrated performance trade-offs (water-energy-nitrogen)

The directionally normalized radar plots integrated marketable fresh yield, WP, energy saving based on reverse-normalized E tot, EP, and PFP N (Figure 3). All radar axes originated so that values closer to one represented more favorable directional performance. The profiles were used descriptively and did not constitute statistically tested composite rankings.

The full N-treatments B1 and W1 had the highest accounted energy inputs and the lowest numerical EP and PFP N. B1 produced a pooled marketable yield of 81,428 kg ha -1, with a WP of 34.27 kg m -3, an EP of 2.96 Kg MJ -1, and a PFP N of 188.75 kg yield kg -1 N. The corresponding values for W1 were 86,810 kg ha -1, 35.30 kg m -3, 3.16 kg MJ -1, and 201.22 kg yield kg -1 N, respectively. Their low energy-saving scores reflected the high E tot associated with the full N treatments.

The 80% N-rate had intermediate energy-saving and resource-productivity profiles. Treatment W8 had the highest pooled numerical yield among all treatments, at 108,583 kg ha -1, compared with 95,902 kg ha -1 under B8. In contrast, B8 had a higher pooled WP than W8, at 37.13 versus 35.51 kg m -3. Energy productivity was 4.30 kg MJ -1 for B8 and 4.81 Kg MJ -1 for W8, whereas PFP N was 277.74 and 314.47 kg yield kg -1 N, respectively.

The 50% N-rate had the highest energy-saving scores because they received the lowest N-input and had the lowest E tot. Treatment B5 produced 99,077 kg ha -1 and had the highest pooled WP, EP, and PFP N, at 39.59 kg m -3, 6.87 kg MJ -1, and 459.09 kg yield kg -1 N, respectively. Treatment W5 produced 97,238 kg ha -1, and had high EP and PFP N, at 6.51 kg MJ -1 and 450.57 kg yield kg -1 N, respectively. However, its pooled WP of 28.39 kg m -3 was the lowest among the treatments.

Collectively, the original-scale and normalized indicators showed that limited differentiation in marketable fresh yield did not imply equivalent resource performance. Treatment W8 had the highest pooled numerical yield, whereas treatment B5 had the highest pooled WP, EP, and PFP N and the greatest energy-saving score. The full N-rate had substantially greater accounted energy inputs without a corresponding statistical yield advantage. Thus, treatments that were weakly differentiated by yield exhibited contrasting profiles of N-input, energy demand, and input productivity.

Crop production and water productivity responses

The limited differences in marketable fresh yield among treatments indicate that tomato production was maintained across the tested biochar and N-rates under soil moisture-controlled drip irrigation. The clay loam soil and sensor-based irrigation scheduling likely reduced the development of substantial differences in root-zone water availability, thereby limiting treatment effects on yield. Comparable yield stability has been reported under well-managed drip irrigation in fine-textured soils, where soil-water storage can buffer differences in irrigation frequency or nutrient supply (Li et al., 2015; Zaccaria and Bali, 2024; Elbana et al., 2025). Under such conditions, yield alone may be insufficient to distinguish management performance, and complementary indicators such as WP, energy productivity, and N-input productivity become more informative.

The significant Season × N-rate interaction for WP shows that the response to N-input was not consistent across the two growing seasons. However, the absence of significant Tukey-adjusted pairwise differences within either season prevents attribution of statistically superior WP to a specific N-rate. Likewise, the numerically higher WP observed in some biochar treatments should not be interpreted as a generalized biochar effect because neither the biochar main effect nor relevant pairwise contrasts demonstrate consistent statistical separation. The results therefore indicate that WP was season-dependent, whereas marketable yield remained comparatively stable under the imposed soil-moisture-controlled irrigation regime.

Fruit-quality responses to season and management

Season was the principal source of variation in fruit hardness, acidity, fruit-N concentration, and NO 3 -N, indicating that environmental differences between the growing periods exerted stronger effects on fruit composition than the imposed N and biochar treatments. Acidity showed the clearest management-related response, including significant biochar and Biochar × N-rate effects. Nevertheless, the only significant adjusted pairwise difference occurred in spring under biochar, where acidity was higher under B1 than under B8, with B5 intermediate. The response was therefore specific to one season and one biochar condition rather than consistent across the experiment. This agrees with (Kasampalis et al., 2020) who reported that growing season significantly affected multiple quality characteristics of tomato fruits, including acidity.

The significant Biochar × N-rate interaction for NH 4 -N was not accompanied by significant Tukey-adjusted comparisons among individual N-rate means. This indicates a difference in the overall response pattern across biochar levels but does not identify a clearly separated treatment pair. Similarly, fruit-N showed a Season × Biochar interaction, whereas hardness and NO 3 -N responded only to season.

Overall, reducing the N-rate did not produce a generalized deterioration in fruit-quality attributes. This agrees with reports that tomato yield and quality can remain relatively stable under efficient irrigation and moderate adjustment of nutrient inputs (Li et al., 2015; Zaccaria and Bali, 2024; Elbana et al., 2025). Previous studies have also shown that biochar, alone or combined with mineral or organic fertilizer, can maintain tomato performance under appropriately managed irrigation, although responses vary with soil, biochar properties, and application conditions (Agbna et al., 2017; Wacal et al., 2023; Kebalo et al., 2025).

The present results therefore support a cautious interpretation: management-related fruit-quality effects occurred, but they were generally variable-specific, season-dependent, or insufficiently separated in the adjusted pairwise comparisons.

Energy structure and the role of nitrogen input

The energy analysis showed that fertilizer-associated N energy was substantially greater than irrigation pumping energy across all treatments. Consequently, the assigned N-rate, rather than variation in pumping demand, was the main determinant of total accounted energy.

The approximately 20% and 50% reductions in E N under the 80% and 50% programs were direct consequence of the corresponding reductions in N-input. They should therefore be interpreted as deterministic accounting outcomes rather than statistically estimated treatment effects. The absence of significant treatment effects on E irr is also important. Although biochar treatments had numerically lower pooled irrigation energy than their non-biochar counterparts at all three N-rates, the mixed model did not support a consistent biochar effect on pumping-energy demand. The heatmap facilitates visual comparison of these numerical patterns but does not provide inferential evidence.

The dominance of fertilizer-associated energy is consistent with studies showing that mineral N production can account for a large proportion of agricultural energy demand and that changes in fertilizer rate may alter the energy structure of crop-production systems more strongly than moderate changes in irrigation energy (Li et al., 2022; Rathore et al., 2026).

Within the accounting boundary adopted here under shallow groundwater table conditions, improved energy performance therefore depended primarily on reducing N-input while maintaining crop output. However, the analysis did not include the production, transport, or application energy associated with biochar or other farm operations, and the reported E tot should not be interpreted as complete life-cycle energy use.

Integrated management trade-offs

The integrated performance profiles revealed an objective-dependent distinction between production-oriented and input-efficiency-oriented management. The 80% N-rate, corresponding to 345 kg N ha -1, numerically reduced N-input and fertilizer-associated energy by approximately 20% relative to the full N-rate. Within this level, W8 produced the highest pooled numerical yield, whereas B8 had higher numerical WP. On a descriptive basis, the 80% N-rate may therefore be regarded as a candidate management compromise because it included the highest-yielding treatment while reducing N-associated energy input. The 50% N-rate provided the greatest energy saving and the highest numerical EP and PFP N. B5 combined comparatively high yield with the highest pooled WP, EP, and PFP N, whereas W5 retained high EP and PFP N but had the lowest pooled WP. These patterns indicate that the lowest N-input favored input-productivity ratios but did not produce a uniformly superior profile across biochar conditions.

The high PFP N and EP values under the 50% N-rate should also be interpreted with care. Both are ratio-based indicators, and their values increase partly because of N-input and total accounted energy in the denominators. They therefore quantify input productivity but do not independently demonstrate agronomic superiority.

From a broader environmental perspective, reducing unnecessary N-inputs is relevant because excessive N use can contribute to soil acidification, nitrate leaching, and greenhouse gas emissions (Do Nascimento et al., 2019; Griffiths et al., 2024). Fertilizer production and use have also been reported to constitute a substantial share of greenhouse gas emissions in horticultural systems (Gu et al., 2020; Wang et al., 2023). The present results consequently support differentiated management priorities among the tested N-rates. The 80% N-rate (345 kg N ha -1) provided a plausible descriptive balance between high production and reduced N-associated energy. The 50% N-rate (215 kg N ha -1) prioritizes lower accounted energy and greater numerical input productivity, whereas the full N-rate (431 kg N ha -1) increased energy demand without providing a consistent yield or quality advantage.

The two predefined inferential null hypotheses had different outcomes. The null hypothesis for marketable fresh yield (H 0Y) was not rejected because neither the N-rate main effect nor the interactions involving N-rate significantly affected yield. By contrast, the null hypothesis for water productivity (H 0WP) was rejected because the significant Season × N-rate interaction indicated that WP response to N-input varied between growing seasons. Nevertheless, the absence of significant Tukey-adjusted comparisons among N-rates within either season prevents identification of a particular N-rate as statistically superior. Fertilizer-associated N energy, energy productivity, and PFP N were not included in the formal inferential hypotheses because they were deterministic or derived indicators. Their numerical patterns were therefore interpreted descriptively as evidence that treatments with limited yield differentiation could still differ in accounted energy demand and input productivity. This reflects the combined evidence from original-scale measurements, mixed-model inference, and descriptive integrated profiles.

Broader irrigation-management implications

Under sensor-based irrigation in clay loam soil, fresh yield was a relatively weak discriminator of management performance because yield was maintained across a broad range of N-inputs. In this context, WP, EP, PFP N, and E tot provided complementary information on how treatment differed in resource demand and input productivity. The findings do not demonstrate that biochar consistently improved WP or reduced irrigation energy. Instead, biochar-related differences varied with N-rate and season and were largely numerical. The practical value of the combined assessment lies in identifying such trade-offs without converting them into unsupported claims of treatment superiority. Maintaining production while reducing N-derived energy demand may support both economic and environmental objectives. The results indicate that substantial reductions in accounted energy were possible through lower N-input, although the choice between the 80% and 50% N-rates depends on whether the principal objective is maximizing production, reducing input use, or balancing both.

Joint evaluation of complementary indicators can therefore extend conventional yield analysis by identifying treatments that maintain crop output while differing in water use, N-input productivity, and energy demand.

Strengths and limitations

The major strength of this study is the integrated evaluation of fresh yield, fruit quality, irrigation water, N-input, and energy performance over two growing seasons under controlled drip irrigation and shallow groundwater conditions. The repeated-plot mixed-model framework accounted for evaluation of the same physical plots across seasons, while the standardized heatmap and directionally normalized radar plots provided transparent descriptive summaries of indicators expressed in different units. The graphical visualizations were not used to replace statistical inference. Rather, they were used to display numerical patterns and management scales. Several limitations should nevertheless be acknowledged.

First, the experiment was conducted at a single site with clay loam soil, which limits extrapolation to contrasting soil textures and hydrological conditions. Second, only one biochar rate was evaluated; the study can therefore assess the presence or absence of biochar under different N programs but cannot identify an optimum biochar application rate. Third, plant N uptake was not measured, preventing calculation of recovery-based N-use-efficiency indices. Fourth, the energy analysis included irrigation pumping and the embodied energy of N fertilizer but excluded biochar production and application, irrigation infrastructure, other field operations, and post-harvest activities. It therefore represents a partial accounting boundary rather than a complete life-cycle assessment. Finally, the radar plots applied equal directional scaling to all indicators. Although the reversed energy axis improved visual interpretation, polygon area depends on normalization, axis order, and implicit equal weighting and was not used as a formal composite performance score.

Future studies should evaluate multiple biochar and N-rates, quantify N uptake, include broader life-cycle energy boundaries, and test whether the observed management trade-offs persist over longer periods and in coarser-textured soils where irrigation effects may be more pronounced.

Under the fixed soil-moisture controlled drip irrigation regime and shallow groundwater conditions evaluated, biochar and N treatments produced limited statistical differentiation in marketable fresh yield and most fruit quality parameters. However, similar yields did not imply equivalent resource performance: lower N-inputs directly reduced fertilizer-associated energy, while descriptive EP and PFP N values revealed contrasting input-productivity profiles. The 80% N-rate included the highest pooled numerical yield while reducing N-input and associated energy by approximately 20%, whereas the 50% rate achieved the lowest accounted energy and the highest numerical input-productivity values.

The yield null hypothesis was not rejected, whereas the WP null hypothesis was rejected because the response to N-rate varied between seasons; however, no within-season comparison identified a consistently superior N-rate. Overall, the findings show that yield alone may not adequately characterize management performance in hydrologically buffered systems and should be interpreted alongside water, energy, and N-input indicators. Further multi-site studies directly quantifying capillary contribution, drainage, N uptake, and full life-cycle energy inputs are needed to assess the transferability of these findings.