The experiment was carried out during 2023–2024 at a test site in Ningxia Central Station of China Irrigation Experiment (106°10′E, 38°45′N, 1116 m altitude), Yinchuan City, Ningxia Province, China. The test area has an arid and semi-arid temperate zone climate. Average annual precipitation and evaporation are 198 mm and 1580 mm, respectively. Precipitation is mainly concentrated from July to September (accounting for 60% of the annual precipitation). Average annual temperature and average annual sunshine are 8.5 °C and 2736 h, respectively. The experiment was conducted in the station’s substrate-based soilless cultivation system described in Section 2.3.

The field experiment was designed as a completely randomized design. During the entire growth period, a complete nutrient solution was prepared using standard Hoagland formulation (Fertilizer A and Fertilizer B). N concentration was used as the control variable during nutrient solution preparation, with four treatment levels established: 250 mg L^-1 (T1), 300 mg L^-1 (T2), 350 mg L^-1 (T3), and 400 mg L^-1 (T4). The ratio schemes of fertilizer A and fertilizer B in different treatments are provided in Table 1. Each treatment included three replicates. All other nutrients were kept consistent across treatments, and crop management practices were identical to minimize confounding effects.

The experiment was conducted in a purpose-built substrate-based soilless cultivation system at the station, designed for quantifying nutrient demand and operating a closed-loop nutrient-solution recirculation. The facility included two planting troughs and a basement utility room housing independent nutrient-solution supply units. Each planting trough contained cultivation buckets filled with washed perlite as the growth substrate. The troughs measured 16 m × 0.7 m × 0.7 m (L × W × H), and the stainless-steel buckets were 60 cm in diameter and 60 cm in height. The baseline nutrient contents of the perlite were low (total N: 8.42 mg L^-1; total P: 2.75 mg L^-1; total K: 0.82 mg L^-1).

A basement room adjacent to the troughs accommodated 12 independent nutrient-solution supply and recirculation units, each equipped with a low-power submersible pump. Each unit supplied two cultivation buckets. Drainage outlets at the bottom of each bucket were connected to return pipes so that unused nutrient solution flowed back to the corresponding storage tank, forming a closed recirculating loop for each unit.

The wolfberry (Lycium barbarum L.) variety in this study was Ningqi-7, four years of age. Drip fertigation was applied using pressure-compensating emitters. Each bucket was fitted with one rectangular and one circular drip ring. Emitters had a flow rate of 4 L h^-1, with 12 cm spacing between emitters. Irrigation was automatically controlled by a timer, operating every 2 h from 08:00 to 18:00, with each event lasting 10 min. To minimize evaporation and prevent external water inputs, each bucket was sealed with plastic film and secured with bricks; the surface of the troughs was covered with reflective aluminum foil and a black shading net. No other water sources were allowed to enter the system to ensure accurate control of water and nutrient supply.

In order to comprehensively evaluate the effects of different treatments on crop growth and yield, with yield, 100 grain weight, number of berries per 50 g and element absorption (N, P, K) as evaluation indexes, Entropy Weight-TOPSIS (He et al., 2024; Li et al., 2023) was used to comprehensively evaluate different treatments. The detailed equations for entropy weight determination are as follows (Equations 4–7):

where, W_m represents the entropy weight; E_m represents the entropy value for the m^th; P_mn represents the proportion of the n^th evaluation object under the m^th index; y_mn represents the dimensionless value of the i^th evaluation object under j^th index; r_mn represents the value of n^th evaluation object of the m^th index.

The relative proximity C was used as the score index of each evaluation scheme. The detailed equations for deciding the relative proximity are as follows (Equations 8–11):

Data are presented as mean ± standard error (SE) with three replicates. Treatment effects were tested using one-way analysis of variance (ANOVA) separately for each year. When significant differences were detected, means were compared using the least significant difference (LSD) test at P< 0.05. All analyses were conducted using appropriate statistical software.

Wolfberry growth parameters differed significantly among N treatments (Figure 1). Growth parameters initially increased and then decreased with increasing N concentration. The highest plant height value appeared in T3, at 111.47 cm and 122.79 cm in 2023 and 2024, respectively. The minimum values for the two experimental years were recorded in T1, which were 106.90 cm and 117.84 cm, respectively. Plant height under T3 treatment was 4.24% greater compared to T1 treatment. Basal stem diameter also reached its maximum in T3, being 14.16%, 11.15%, and 4.31% higher than T1, T2, and T4, respectively (Figure 1).

For the length of new branch, T3 resulted in the greatest values, reaching 65.04 cm in 2023 and 65.56 cm in 2024; the lowest values in both 2023 and 2024 were observed in T1, measuring 60.35 cm and 61.52 cm, respectively (Figure 1). Under T3 treatment, the stem diameter of new branch reached the highest value, exceeding those of T1, T2, and T4 by 0.61 mm, 0.28 mm, and 0.55 mm, respectively (Figure 1). In terms of north-south canopy width, the treatments ranked in the following order: T3, T4, T2, and T1. T3 showed the greatest east-west canopy, with measurements of 130.34 cm and 125.06 cm in two growing seasons. (Figure 1).

There were significant differences (p<0.05, Table 4) in NUE, PUE, and KUE under different treatments. NUE, PUE, and KUE initially decreased and then increased with increasing N concentration. The highest NUE occurred in treatment T1 in two growing seasons, with value of 4.26 kg kg^-1 in 2023 and 4.73 kg kg^-1in 2024. The lowest NUE was observed in T3, with the value of 3.51 kg kg^-1 in 2023 and 3.91 kg kg^-1 in 2024. In 2023, the PUE followed the order T1 > T4 > T2 > T3 while, in 2024, the sequence was T1 > T2 > T4 > T3. In both growing seasons, the highest KUE was observed under the T4 treatment, with values of 7.85 kg kg^-1 in 2023 and 8.47 kg kg^-1 in 2024 (Table 4).

The comprehensive yield and resource-use efficiency scores in 2023 and 2024, calculated by using Entropy Weight-TOPSIS, are shown in Table 5. Entropy Weight-TOPSIS showed that T3, T1, and T2 ranked first, second and third, respectively, whereas T4 ranked lasted in 2023. Similarly, T3, T2, and T1 were ranked first, second and third, whereas T4 ranked lasted in 2024. T3 showed the highest comprehensive score under soilless cultivation, based on traits such as yield, 100-berry weight, number of berries per 50 g and NUE, PUE, and KUE.

Plant morphological indicators can provide an intuitive reflection of plant growth status (Watanabe et al., 2017; Zheng et al., 2024). The present study showed that plant height, basal stem diameter, length of new branch, stem diameter of new branch, north-south canopy and east-west canopy increased initially with increasing irrigation quota and irrigation frequency, then subsequently decreased, indicating that moderate N concentration optimized wolfberry growth, as reflected by these morphological indicators. The values of growth parameters were the highest under T3 (with a N concentration of 350 mg L^-1). This treatment led to the highest plant height and stem diameter, and it promoted the growth of new branches (Figure 1). This aligns with the results of other crop studies. For instance, Xiao et al. (2023) demonstrated that N concentration significantly influences tomato growth, particularly in terms of plant height and leaf area, with suitable N concentrations enhancing the growth rate. According to Zhu et al. (2016), maize plant height increased with rising N levels within a specific range, while excessive N concentrations suppressed maize growth. This occurred because excessive N fertilization restricted root development and impaired the roots’ ability to absorb water (Feng et al., 2016), ultimately limiting crop growth. N nutrition was closely linked to the level of cytokinins. Insufficient nutrient supply resulted in a decrease in cytokinin content in plant leaves, while N nutrition promoted the accumulation of cytokinins (Samuelson and Larsson, 1993). Therefore, under low N conditions, the reduced cytokinin content inhibited plant growth and development, leading to stunting and wilting.

Elements such as N, P, and K are the most essential nutrients for plant growth and development (Ma et al., 2022; Guo et al., 2019). Macronutrients alone are insufficient to meet the growth requirements of crops; secondary macronutrients are also necessary to meet demand (Grzebisz, Zielewicz, & Przygocka-Cyna 2023; Bekele and Birhan, 2021). This study found that with an increasing N concentration, the uptake of different elements by wolfberry first increased and then decreased. Under T3, the uptake of N, P, K, Ca, and Mg reached their maximum values, indicating that an appropriate N concentration promotes the absorption of various elements by wolfberry. The highest proportion of uptake for both macronutrients and secondary macronutrients occurred during the summer fruiting period. This period is crucial for fruit development in wolfberry and it serves as a key stage for balancing vegetative and reproductive growth. Therefore, meeting the nutrient requirements of wolfberry during the summer fruiting stage is critical to improve both yield and quality.

N management played a crucial role in regulating wolfberry yield components, including total yield, 100-berry weight, and number of berries per 50 g, as well as determining fruit quality attributes. Across two growing seasons, the results clearly indicated that moderate N application (T3, 350 mg L^-1) significantly enhanced wolfberry yield, while excessive or insufficient N levels led to suboptimal outcomes. Yield performance under different N treatments followed a typical parabolic response pattern, consistent with findings in tomatoes (Li et al., 2021) under controlled environments. The highest yields recorded in T3 (2759.65 kg ha^-1 in 2023 and 2930.93 kg ha^-1 in 2024) reflect the benefits of sufficient N availability for maximizing reproductive growth and dry matter translocation toward fruits. In contrast, the lowest yields observed in T1 (250 mg L^-1) suggest that N deficiency limits canopy expansion, photosynthetic capacity, and assimilate supply for fruit development, a trend similarly reported by Wang et al. (2019) in cucumber under soilless conditions. In terms of fruit weight, 100-berry weight under T3 was significantly higher than in other treatments, reaching 19.07 g in 2023 and 19.24 g in 2024. A higher 100-berry weight is indicative of larger fruit size and better market classification, which directly influences commercial value. These findings agree with previous studies showing that moderate N improves cell expansion and water accumulation in fruit tissues (Zhang et al., 2023). Notably, although T1 exhibited a higher number of berries per 50 g, suggesting smaller fruit size, it did not translate into better yield performance, underscoring the importance of balancing fruit number and fruit weight for overall productivity.

Regarding fruit quality, N concentration also exerted significant effects on key biochemical attributes. Sweet beet alkaloid (betaine) content showed a declining trend with increasing N levels, possibly due to the dilution effect caused by accelerated vegetative growth under high N supply. This phenomenon was similarly observed in fertilization experiments on other Lycium species (Zhou et al., 2021). The highest β-carotene concentration under T3 demonstrates that moderate N promotes carotenoid biosynthesis, which is essential for enhancing fruit nutritional value and antioxidant properties. Protein content and total content of EAAs were also maximized under T3. Adequate N availability likely stimulated protein biosynthesis and amino acid accumulation, improving the nutritional profile of wolfberry fruits. The higher EAA content under moderate N supply aligns with the findings of Xu et al., 2012 in other horticultural crops, where N optimization enhanced both yield and quality traits simultaneously.

NUE, PUE, and KUE are key determinants of sustainable agricultural productivity (Zhang et al., 2019), especially under soilless cultivation systems where precise nutrient management is critical. In this study, NUE, PUE, and KUE were significantly influenced by N concentration applied through drip fertigation. Moderate N concentration (350 mg L^-1, T3) optimized both crop productivity and nutrient efficiency, whereas higher or lower N inputs reduced overall nutrient use efficiencies. High NUE is associated with improved N assimilation and reduced N loss, contributing to both yield improvement and environmental sustainability (Xu et al., 2012). Estimates suggest that a 1% increase in crop NUE could result in an annual saving of approximately 1.1 billion USD (Kant et al., 2011). Excessive N application, as observed under T4 treatment, can lead to luxury consumption, nitrate accumulation, and lower NUE due to imbalance between vegetative and reproductive growth (Zhang et al., 2023). These findings align with earlier studies in greenhouse-grown crops under soilless cultivation systems (Rouphael and Colla, 2005), where moderate N supply enhanced biomass allocation towards economically valuable plant parts, improving yield and nutrient use efficiency. Similarly, PUE exhibited a decreasing trend with increased N concentration beyond the optimum. Efficient P uptake is crucial for root development and energy transfer processes (Vance et al., 2003). Inadequate management may lead to P fixation or leaching even under closed soilless systems, thereby reducing effective PUE. KUE is intimately linked with water regulation and stress tolerance mechanisms (Wang et al., 2013). The observed improvements in KUE under moderate N conditions suggest better coordination between water uptake, osmotic adjustment, and carbohydrate translocation, ultimately supporting fruit development and quality.

The comprehensive performance evaluation using the entropy weight-TOPSIS method effectively integrated multiple agronomic and physiological indicators, including yield, 100-berry weight, berry size, and NUE, PUE, and KUE, to assess the overall impact of different N treatments on wolfberry under soilless cultivation. This section is included because these indicators may not change in the same direction among treatments; therefore, a single-trait comparison may be insufficient to identify an overall optimal N level. This approach provided an objective framework to rank treatments by simultaneously considering production output and resource use sustainability. The results showed that the T3 treatment (350 mg L^-1) consistently ranked highest over two growing seasons, confirming that moderate N supply optimized both yield and nutrient use efficiency. Excessive (T4) or insufficient (T1) N treatments, although advantageous in isolated traits, failed to achieve comparable overall performance. This finding agrees with the principles outlined by Zhou et al. (2006), who emphasized the importance of applying multi-criteria decision analysis to balance competing agricultural objectives. Furthermore, the entropy weighting approach provided an objective method to assign indicator importance based on data variability, minimizing subjective bias in the evaluation process (Zardari et al., 2015).