Wheat–Pea Intercropping Responds to Nitrogen Fertilization and Maintains Yield Under Agroforestry in Central Italy

Abstract

Agriculture must shift to sustainable practices that support mitigation and adaptation, with crop diversification highlighted as a key adaptive practice. A field experiment was conducted in central Italy to study forage and grain production of wheat (Triticum aestivum L.) and pea (Pisum sativum L.) intercropping (IC) comparing three nitrogen fertilizations (NF) (i.e., 0, 70, and 140 kg ha⁻¹) and two cropping systems (CS) (i.e., arable (AR) and silvo-arable (SIAR)) for two consecutive cropping seasons (2023–2024 and 2024–2025). Forage and grain production were determined at flowering and at maturity to identify temporal trends in resource use. Overall, the results demonstrated that poplar presence did not significantly impact IC productivity, as forage biomass at flowering was 5.00 t ha⁻¹ in AR and 4.77 t ha⁻¹ in SIAR in 2024, and 3.20 t ha⁻¹ in both cropping systems in 2025. Moreover, NF was the main driver of both forage and grain production, without significant interaction with the CS, and a moderate N rate (i.e., 70 kg ha⁻¹) was the most efficient, ensuring both wheat and pea productivity. The absence of a yield penalty in the IC in the SIAR supports the agroecological value of integrating annual intercrops with tree components.

1. Introduction

The transition toward more sustainable and resilient agricultural systems by adopting management practices that help mitigate or adjust to changing conditions is essential to address today’s challenges posed by climate change and resource scarcity. Among the strategies proposed to enhance agroecosystem sustainability, crop diversification, i.e., growing different crops through rotation or multiple cropping—particularly through intercropping (IC) and agroforestry (AF)—has received growing attention due to its potential to improve resource-use efficiency, stabilize yields, and reduce external inputs [1].

IC annual legumes with cereals is an effective strategy to enhance productivity and nitrogen (N) use efficiency and is widely adopted in Mediterranean forage production [2,3,4,5]. Here, wheat (Triticum aestivum L.) and pea (Pisum sativum L.) are autumn-winter crops used for forage, silage, or grain to improve feed nutrition, and their IC benefits from complementary resource use and growth patterns [3], as wheat relies on soil and N fertilizers, while pea adds N through N₂ fixation. This process has advantages for both plants, potentially reducing the need for mineral fertilizers in crop rotation. However, biological nitrogen fixation rarely fully replaces fertilizer N inputs [6], so applying an appropriate rate of N fertilizer is still necessary in Mediterranean legume–cereal IC systems to ensure adequate yield and quality [5,6,7].

Therefore, the performance of wheat–pea IC systems is strongly influenced by N availability. While increasing mineral N inputs typically enhances wheat growth and yield, it may reduce pea competitiveness and inhibit symbiotic N₂ fixation, thereby altering species balance and potentially reducing system-level complementarity.

Finding N strategies that maintain high wheat productivity without introducing negative impacts on legume performance and functional complementarity as well as system-level efficiency may still be a major issue for targeted N management in cereal–legume systems because the advantages of IC are highly dependent on N supply which significantly influences the competitive balance, complementarity effects, and total resource use efficiency within the mixture [8,9,10].

In parallel, AF systems (i.e., the intentional integration of trees with crops) offer additional opportunities to increase the agroecosystem diversity and therefore its sustainability [11]. By increasing structural and functional diversity, AF can enhance resource capture, regulate microclimatic conditions, and contribute to soil fertility and carbon sequestration.

Therefore, the integration of AF with cereal–legume IC further increases structural and functional biodiversity by combining trees and annual intercrops and creating vertical and horizontal stratification of canopies and rooting zones, which modifies light distribution, soil moisture, and temperature regimes, thereby regulating microbial activity and N transformation processes [12,13,14].

The presence of trees further extends rooting depth and spatial nutrient capture, enabling N uptake from deeper soil layers and potentially reducing leaching. At the same time, increased plant diversity and litter inputs enhance soil organic matter turnover and microbial biomass, contributing to N cycling and improved N-use efficiency [15].

Regardless, environmental variability may condition these interactions, and the inter-annual differences in rainfall distribution, thermal regimes, and soil water status common in Mediterranean areas can strongly affect N uptake dynamics in cereals and the initiation and efficiency of N₂ fixation in legumes. Such variation can alter competitive outcomes and actual benefits of IC across years, as demonstrated in cereal–pea systems under temperate conditions [16].

Despite the advantages of both IC and AF, studies integrating these two approaches remain limited, particularly for annual cereal–legume systems. Most existing research has focused either on intercrops in open-field conditions or on sole crops under tree-based systems, while experimental evidence on cereal–legume IC within silvo-arable AF systems is still scarce [12,13].

To fill this gap, this study aims to evaluate the agronomic performance of wheat–pea IC within both arable and silvo-arable agroforestry systems under different NF. We carried out a dual-purpose evaluation of the IC for forage and grain production and nutrient accumulation by measuring and analyzing the biomass both at flowering and at maturity.

We hypothesize that: (i) Increasing NF rate may reduce interspecific competition, thereby supporting IC nutrient resource uptake and increase total production; (ii) interactions between intercropped species may vary across the growing seasons due to differences in nutrient demand and N availability; and (iii) the presence of poplar trees under AF conditions would not result in yield decline but may influence resource allocation and growth dynamics.

2. Materials and Methods

2.1. Experimental Site

The experiment was conducted over two cropping seasons (2023–2024 and 2024–2025) at the Centre for Agri-environmental Research “E. Avanzi”, University of Pisa, Italy (43°41′6.71″ N, 10°20′33.12″ E; 1 m a.s.l.), located about 4 km from the coast of central Italy, at a site with a Mediterranean climate (Csa, Köppen classification) characterized by mild, wet winters and hot, dry summers. The experimental field was ploughed to a depth of 40 cm, and treatments were re-randomized in the second year. Soil parameters did not differ between years, and average values are summarized in

Table 1. Physical and chemical properties of the experimental soil in Pisa, central Italy. Values are averaged over the two years.

Parameter	Unit	Value	Method	Reference
Sand (2000–2 µm)	g kg−1	361	Modified pipette	Indorante et al., 1990 [18]
Silt (2–50 µm)	g kg−1	524
Clay (50–2000 µm)	g kg−1	115
Organic matter	g 100 g−1	1.7	Walkley and Black	Walkley and Black, 1934 [19]
pH	-	8.1	Potentiometric	ASTM, 1995 [20]
Total nitrogen	g kg−1	1.2	Kjeldahl	Bremner, 1960 [21]
Available phosphorus	mg kg−1	31	Olsen	Olsen et al., 1954 [22]

. The soil is calcareous, xeric, and classified as Typic Xerofluvents [17], with medium loam texture, moderate organic matter content (1.7%), and adequate nutrient levels (total N = 1.2 g kg⁻¹).

2.2. Experimental Design

The trial followed a two-way randomized complete block design with three replications, testing three nitrogen fertilizations (NF) and two cropping systems (CS). NF were: (i) N0: control without nitrogen fertilization (0 kg N ha⁻¹); (ii) N70: optimal nitrogen fertilization rate for the intercrop (70 kg N ha⁻¹); and (iii) N140: double the optimal nitrogen fertilization rate (140 kg N ha⁻¹). The optimal rate (70 kg ha⁻¹) was based on recommendations for cereal–legume forage intercrops in central Italy [5], while the doubled rate (140 kg ha⁻¹) was expected to influence both forage and grain production. Nitrogen was applied as urea (46% N) in two equal topdressing splits at wheat tillering and stem elongation (i.e., 0 + 0, 35 + 35, and 70 + 70 kg N ha⁻¹ respectively, in N0, 70, and 140). CS were: (i) arable (AR): herbaceous intercrop of wheat and pea without trees, and (ii) silvo-arable (SIAR): herbaceous intercrop of wheat and pea combined with a single row of poplar trees (Populus × euramericana with five trees per plot, 2 m spacing along the row, planted along field margins (Figure 1)). Thus, each year, six treatments (two CS × three NF) were tested across 18 plots (six treatments × three replications), each measuring approximately 84 m² (8 m width × 10.5 m length), with each plot receiving a specific N treatment and either the presence or absence of poplar trees.

2.3. Crop Management

The trial took place under rainfed conditions, using consistent agronomic practices typically followed by local farmers across all plots, except for the experimental treatments (NF and CS). In both CS, the intercrop consisted of common wheat (Triticum aestivum L., cv. Ludwig) and pea (Pisum sativum L., cv. Paprika). Ludwig is a strategic variety for dairy cow forage (hay or silage) and was selected for its medium-late maturity and suitability for forage production. Audit variety is for animal feed and was selected for its high productivity and protein content. In the SIAR, five 1-year-old stems of poplars (clone I-214, Populus × euramericana (Dode) Guinier) were manually planted, north-south oriented, in each plot on 23 February 2021, 2 m apart along one plot edge, 2 m away from drainage ditches, and 1 m apart from the beginning of the wheat/pea vegetation, whereas in AR the intercropped crops were grown without trees. Fields were ploughed to ~40 cm depth in autumn (18 October 2023 and 27 September 2024, respectively) and fertilized basally with 50 kg ha⁻¹ P₂O₅ (triple superphosphate, 0-46-0) and 40 kg ha⁻¹ K₂O (potassium sulphate) before sowing (20 November 2023 and 14 October 2024). Wheat and pea were sown with a mechanical sower on 2 February 2024 and 13 November 2024 at rates of 180 kg ha⁻¹ for wheat and 240 kg ha⁻¹ for pea. Wheat rows were spaced 36 cm apart, with one pea row between each pair (18 cm).

N fertilization was applied at three rates (0, 70, and 140 kg ha⁻¹) according to the experimental design described above. In the first year, fertilizers were applied on 7 March 2024 and 17 April 2024; in the second year, on 10 January 2025 and 21 March 2025, following the same scheme, with application timing adjusted according to crop development. No herbicides or pesticides were applied as weed and disease pressure remained below economic thresholds.

2.4. Crop Sampling, Analysis, and Calculation

Crop phenological phases were tracked in both years using the BBCH scale [23] to accurately determine growth stages for scheduling inorganic nitrogen applications and harvests. Plant density was measured at crop emergence by counting plants along 1 m of the row and converting to plants m⁻².

Aboveground biomass was sampled at flowering (BBCH 65) and maturity (BBCH 97), corresponding to forage and grain harvest stages, respectively, using two 25 × 25 cm quadrats per plot. Samples were separated into wheat, pea, weeds, and residues, weighed fresh, oven-dried to constant weight, and expressed on an area basis (Mg ha⁻¹).

At maturity, grain yield was determined from harvested plant material collected within the quadrats, with wheat and pea separated into reproductive and vegetative organs. Grain dry weight was used to calculate crop yield (Mg ha⁻¹). Yield components were recorded separately, including spike number per unit area (n m⁻²), spike biomass (Mg ha⁻¹), and mean spike weight (g) for wheat, and pod biomass (Mg ha⁻¹), pod number per unit area (n m⁻²), and mean pod weight (g) for pea. Harvest index (HI) was calculated as the ratio of grain dry weight to total aboveground dry biomass.

For each experimental year and for both sampling stages, subsamples of the harvested plant material were analyzed for elemental composition. N and carbon (C) concentrations were determined by dry combustion using an elemental analyzer (FlashSmart™ Elemental Analyzer, Thermo Scientific, Waltham, MA, USA). Phosphorus (P) concentration was determined using a colorimetric method following acid digestion, measured with a UV–visible spectrophotometer (UV-1900i, Shimadzu, Kyoto, Japan) according to Wieczorek et al. [24]. Nutrient accumulation (N, P, and C) was calculated by multiplying the dry biomass of each plant component by its respective nutrient concentration and expressed on an area basis (kg ha⁻¹).

Leaf area index (LAI) was measured non-destructively using an I-FishI hemispherical camera coupled with an LAI-2000 Plant Canopy Analyzer (LI-COR Biosciences, Lincoln, NE, USA). Measurements were conducted under diffuse light conditions at sunset, with eight readings per plot collected on multiple dates during the growing season in each experimental year.

2.5. Statistical Analysis

Statistical analyses were carried out using JMP software (SAS Institute Inc., Cary, NC, USA, Student Edition 19). The effect of NF and CS was assessed on grain yield and all the yield components for wheat and pea, respectively. The effects of NF and CS, and their interaction, were analyzed using a linear model, followed by a two-way ANOVA. To investigate the effect of NF and CS and their interaction on leaf area index (LAI), which was measured repeatedly at different growth stages, a mixed model with repeated measures over time was applied within each year, with NF, CS, and time (T) as fixed factors, and block as a random factor; a three-way ANOVA was then carried out. Initially, a combined analysis of the two years was performed for all the parameters tested. As the year and its interactions with the other factors tested were significant for most measured traits, data were subsequently analyzed separately for each year. Treatment means were compared using Tukey’s Honest Significant Difference (HSD) test at p < 0.05, and are presented with letters indicating significant differences.

3. Results

3.1. Weather Conditions

Meteorological data were obtained from the nearest public weather station of the Tuscany Region (43°40′51.60″ N, 10°16′48.00″ E; http://www.sir.toscana.it (accessed on 5 March 2026)). Long-term (1993–2023) averages indicate annual precipitation of 890 mm and mean minimum and maximum temperatures of 12.7 °C and 20.5 °C, respectively. During the two experimental seasons (October 2023–June 2025), temperature trends followed the regional pattern, with mild winters (minimum temperatures of 2–6 °C) and hot summers (up to 30 °C). Overall, temperatures were similar across the two years and remained within the normal range. Although total precipitation from October to July was similar in the two growing seasons (925 and 934 mm, respectively), its distribution was highly irregular, with distinct wet and dry phases (Figure 2). Notable rainfall spikes occurred in February 2024 (210 mm), October 2024 (272 mm), and a weaker one in March 2025 (163.2). Conversely, there were long dry periods, particularly in early summer in June/July 2024 and 2025, where no rainfall was observed or was very minimal. Generally, precipitation displayed an irregular distribution, involving lengthy dry spells followed by heavy rainfall sessions.

3.2. Leaf Area Index

In 2024, LAI was significantly affected by T, CS, and NF (Table S1). Higher LAI values were observed during the earlier T sampling (3.24 in April) than during the later ones (Figure S1). AR generally exhibited higher LAI values than SIAR, with a more pronounced decline over time in SIAR. NF increased LAI at all sampling dates: (i) from 3.00 to 3.40 at T1; (ii) from 2.73 to 3.00 at T2; and (iii) from 2.52 to 3.19 at T3 (+12%, +10% and +27%, for T1, T2, and T3 respectively). In 2025, LAI was significantly influenced by T and N rate but not by CS (Table S1). The highest LAI values were recorded in April and mid-May (3.82 and 3.72, respectively), followed by a gradual decline towards maturity. The N140 treatment maintained the highest LAI values throughout the season, while N70 showed intermediate values (3.45 averaged over T dates), suggesting a partial saturation of the N response during peak vegetative growth (Figure S1).

3.3. Measurements at Flowering Stage

3.3.1. Aboveground Biomass

NF significantly affected wheat aboveground biomass (p < 0.001) at flowering in both years whereas pea and weed biomass were largely unaffected (Table S2). In 2024, wheat biomass increased progressively with N supply, reaching a maximum of 1.63 Mg ha⁻¹ at 140 kg N ha⁻¹, which was significantly higher than at 70 kg N ha⁻¹ (0.92 Mg ha⁻¹) and in the unfertilized control (0.28 Mg ha⁻¹). A similar response was observed in 2025, with wheat biomass increasing from 1.72 Mg ha⁻¹ at 0 kg N ha⁻¹, to 2.93 Mg ha⁻¹ at 140 kg N ha⁻¹ (Figure 3a).

In contrast, pea biomass remained stable across N treatments in both years. In 2025, pea biomass was lower and remained unresponsive to N fertilization (Figure 3b). In 2024, total intercrop (IC) biomass did not respond significantly to N fertilization, although numerically higher values were observed under fertilized treatments (5.11–5.21 Mg ha⁻¹) compared with the unfertilized control (4.34 Mg ha⁻¹). In 2025, IC biomass increased significantly with NF rate (p < 0.05), rising from 2.84 Mg ha⁻¹ at N0 to 3.69 Mg ha⁻¹ at N140 (Figure 3c). Cropping system (CS) and CS × NF interaction did not significantly affect biomass production of any component in either year.

3.3.2. Nutrient Content

In both years, NF rate significantly affected total N content in wheat aboveground biomass (p < 0.001 both years), with increasing accumulation at higher fertilization levels (Figure 3d). In 2024, total N accumulation increased from 8.25 kg ha⁻¹ in the unfertilized control to 29.76 kg ha⁻¹ at 70 kg N ha⁻¹ and reached 53.06 kg ha⁻¹ at 140 kg N ha⁻¹, while in 2025, wheat N accumulation increased from 36.05 to 72.72 kg ha⁻¹ across the same N range (Figure 3d).

In 2025, NF rate also significantly affected carbon (C) and phosphorus (P) accumulation in wheat biomass (p < 0.001 and p < 0.05, respectively). At 140 kg N ha⁻¹, wheat C and P accumulation reached 1246.15 and 12.32 kg ha⁻¹, respectively (Figure 3e,f). In pea biomass, only C concentration differed significantly (p < 0.05), increasing from 37.24% in the control to 44.25% at N140 (Figure 3g), while N and P content remained unaffected.

Cropping system did not significantly influence N, C, or P content in either wheat or pea biomass in either year (Table S2).

3.4. Measurements at Maturity

3.4.1. Wheat Biomass and Harvest Index

At maturity in 2024, wheat grain (p < 0.05), vegetative (p ≤ 0.001), and total aboveground biomass (p ≤ 0.01) increased significantly with increasing N rate, while HI (p ≤ 0.01) decreased (Table S3; Figure 4a–d). Grain biomass increased from 0.45 Mg ha⁻¹ at N0 to 1.04 Mg ha⁻¹ at N140 (Figure 4a), whereas vegetative biomass increased from 0.90 to 3.44 Mg ha⁻¹ (Figure 4b), resulting in a total aboveground biomass increase from 1.35 to 4.48 Mg ha⁻¹ (Figure 4c). Conversely, HI declined from 0.44 to 0.23 across the same N range (Figure 4d). The CS had no significant effect on wheat biomass or HI.

In 2025, wheat biomass components were not significantly affected by N rate or cropping system (Figure 4a–c). However, HI was significantly influenced by NF rate (p < 0.01) and CS (p < 0.05), decreasing from 0.36 at N0 to 0.31 at N140 (Figure 4d), and being slightly higher in AR than SIAR (0.36 and 0.33, respectively).

3.4.2. Wheat Spike Traits

Across both years, spike number per unit area was unaffected by NF rate or CS (Table S3). In 2024, spike biomass and spike mean weight increased significantly with NF rate (p < 0.05 and p < 0.01, respectively). Spike biomass increased from 0.67 to 1.49 Mgha⁻¹, while spike mean weight increased from 0.37 to 0.78 g between N0 and N140 (Figure 4e,f). In 2025, none of the spike traits responded significantly to NF or CS (Figure 4e,f).

3.4.3. Pea Grain and Harvest Index

In 2024, pea grain biomass and HI (p < 0.01) were significantly changed by NF, while vegetative and total biomass were unaffected (Table S3). Grain biomass increased from 0.83 to 2.46 Mg ha⁻¹ and HI from 0.21 to 0.54 between N0 and N140 (Figure 5a,b). In 2025, no significant effects of NF rate or CS were observed on pea biomass components or HI (Figure 5a,b).

3.4.4. Pea Pod Traits

In 2024, NF significantly affected pod mean weight (p < 0.05; Table S3), increasing from 0.28 g in the unfertilized control to 0.56 g at N140 (Figure 5c). Pod biomass and pod number per unit area were not significantly affected, although pod biomass increased numerically with NF rate (1.38 Mg ha⁻¹ at N0 to 2.99 Mg ha⁻¹ at N140) (Figure 5d,e).

In 2025, NF significantly affected pod biomass and pod number (p < 0.05), both decreasing with increasing N supply (Figure 5d,e). Pod biomass declined from 0.98 to 0.45 Mg ha⁻¹, and pod number from 292.7 to 135.3 pods m⁻² between N0 and N140 (Figure 5d,e). CS did not significantly affect pod traits in either year.

3.4.5. Intercrop Total Aboveground Biomass and Weed Biomass

In 2024, total intercrop aboveground biomass increased significantly with NF rate (p < 0.05; Table S3), being approximately 78% higher at N140 than in the unfertilized control (Figure 5f). In contrast, intercrop biomass in 2025 was not significantly affected by NF rate or CS (Figure 5f). In both years, weed biomass at maturity remained low and was not significantly influenced by NF rate, CS, or their interaction, averaging 0.31 g m⁻² in 2024 and almost null (0.02 g m⁻²) in 2025.

3.4.6. Wheat Nutrient Concentration and Accumulation

In 2024, nitrogen (N), carbon (C), and phosphorus (P) concentrations in both vegetative biomass and wheat grain were not significantly affected by NF rate, CS or their interaction (Table S4). However, in 2025, it was observed that the concentration of nitrogen in wheat vegetative biomass was significantly affected by NF rate (p < 0.001), increasing from 0.33% in N0 to 0.61% in N140 (Figure 6a). In contrast, C and P concentrations in vegetative biomass, as well as N, C, and P concentrations in wheat grain, were not significantly affected by NF rate or CS.

However, in 2024, NF significantly increased vegetative biomass and N, C, and P accumulation (Table S4). For vegetative biomass, N accumulation increased from 9.31 kg ha⁻¹ under N0 to 31.76 kg ha⁻¹ under N140 (p < 0.05) (Figure 6b); C accumulation increased from 349.95 to 1312.54 kg ha⁻¹ (p < 0.001) (Figure 6c); and P accumulation increased from 2.56 to 6.91 kg ha⁻¹ (p < 0.01) (Figure 6d).

Grain N and C accumulation also increased significantly with NF rate, with grain N increasing from 9.05 to 21.91 kg ha⁻¹ and grain C from 184.63 to 429.92 kg ha⁻¹ (p < 0.05) (Figure 6e,f), whereas grain P accumulation was not significantly affected. CS had no significant effect on nutrient accumulation in either vegetative biomass or grain.

In 2025, N accumulation in wheat vegetative biomass increased significantly with N rate (p < 0.01), rising from 24.38 kg ha⁻¹ at N0 to 58.48 kg ha⁻¹ at N140 (Figure 6b). Vegetative C and P accumulation showed no significant response to NF rate (Figure 6c,d). Grain nutrient accumulation (N, C, and P) was not significantly influenced by NF rate or CS in 2025 (Figure 6e,f).

3.4.7. Pea Nutrient Concentration and Accumulation

In 2024, N rate significantly affected phosphorus (P) concentration in pea vegetative biomass (p < 0.05), which decreased from 0.14% at N0 to 0.08% at N140 (Figure 7a), while CS had no significant effect (Table S4). A significant CS × NF rate interaction was observed for grain N concentration (p < 0.05); however, grain N concentration ranged narrowly between 2.94 and 3.16%, with no clear or consistent differences between CS across NF rates (Figure 7). Vegetative P accumulation declined significantly with increasing NF rate, decreasing from 3.94 kg ha⁻¹ at N0 to 1.83 kg ha⁻¹ at N140 (p < 0.05; Figure 7b). In contrast, N accumulation in pea grain increased by increasing NF rate from 24.60 kg ha⁻¹ at N0 to 74.58 kg ha⁻¹ at N140 (p < 0.05) (Figure 7c). Carbon accumulation in both vegetative biomass and grain was not significantly affected by CS or NF rate.

In 2025, grain N concentration increased significantly with NF rate (p < 0.01), rising from 4.15% at N0 to 5.33% at N140 (Figure 7d). In contrast, N, C, and P concentrations in vegetative biomass were not significantly affected by CS or NF rate, with mean vegetative N concentration ranging from 1.41 to 1.58%. Nutrient accumulation in both vegetative biomass and grain was also unaffected by treatments.

3.4.8. Nitrogen Harvest Index (NHI)

NF rate significantly affected wheat NHI in both years (p < 0.05 in 2024 and p < 0.001 in 2025; Table S5). In wheat, NHI progressively decreased with increasing NF, indicating a reduced efficiency of N translocation to the grain at higher N rates (

Table 2. Mean comparison of the nitrogen harvest index (NHI) in wheat and pea at maturity during the 2023–2024 and 2024–2025 growing seasons under different cropping systems and N fertilization rates. For each trait within each factor, means followed by different letters are significantly different according to Tukey’s HSD test (p < 0.05).

	NHI Wheat		NHI Pea
	2023–2024	2024–2025	2023–2024	2024–2025
Cropping System (CS)
AR	0.489 a	0.699 a	0.576 a	0.555 a
SIAR	0.514 a	0.671 a	0.541 a	0.466 a
N Fertilization (NF)
0	0.627 a	0.743 a	0.353 b	0.581 a
70	0.468 ab	0.731 a	0.580 a	0.496 a
140	0.410 b	0.581 b	0.742 a	0.454 a

). Conversely, pea NHI increased with increasing N rate in 2024 (p < 0.01) but was unaffected in 2025, showing improved N allocation to the grain from the unfertilized control to the highest N treatment (Table 2).

4. Discussion

This research aimed to assess the physiological and agronomic performances of a wheat and pea intercrop cultivated with three rates of nitrogen fertilization and within an arable cropping system and a silvo-arable agroforestry system. We conducted observations at both the flowering and maturity stages to assess the impact of NF and CS on growth dynamics over two cropping seasons.

Overall, varying climate conditions across the two experimental years had significant and distinct impacts on yield and its components in the two companion crops.

In contrast, the effect of the CS was negligible for most measured traits, and no significant differences were observed in the IC performance between the AR and SIAR systems. The CS × NF rate interaction was generally not significant, except for pea grain N concentration in the first year.

NF was the main driver of forage and grain yield and N accumulation.

4.1. Effect of the Year

Large differences between the two years of the experiment confirmed the major effect of climatic factors on crop growth and nitrogen use efficiency in Mediterranean IC [3,4,5]. Rainfall patterns and planting dates had major impacts on crop phenology, biomass production, and grain yield in crops. In the first year 2023/2024, late planting of crops occurred due to heavy rainfall and led to reduced cycle duration, reduced tillering in wheat, reduced solar radiation interception, and greater reliance on mineral N nutrition; thus, addition of N fertilizer led to greater vegetative biomass but reduced harvest index and N harvest index because of the suppression associated primarily with dominant vegetative growth in a N-driven system under heat and photoperiod stress [25,26]. On the contrary, in experimental year 2024/2025, timely planting allowed complete phenological growth of crops, hence more tillering and biomass production in wheat; therefore, N fertilizer effect was less pronounced. Under these conditions, nitrogen was not the primary limiting factor, and crop growth was mainly driven by canopy development and radiation interception, indicating a radiation-driven system rather than a N-limited one [27,28,29]. Pea responses were likewise context-dependent, with greater reliance on mineral N under shortened cycles in the first year and increased dependence on symbiotic N₂ fixation under optimal conditions in the second [6].

4.2. Effect of the Cropping System

Although AR plots tended to show slightly higher LAI and early biomass under high NF rate, these differences did not translate into significant yield gains, and the presence of poplars in SIAR did not have significant effects on crop yield, biomass accumulation, and nutrient dynamics. Trees likely induced mild shading and belowground competition, but these effects were insufficient to constrain crop growth because of the autumn–winter cycle of wheat–pea IC. Indeed, as reported by several studies conducted in temperate agroforestry systems, light competition is more critical for summer crops than for winter cereals and legumes [30,31]. On the contrary, as reported by Piotto et al. [32], selecting deciduous tree species such as poplar in combination with C3 winter crops reduces the temporal overlap in growth, allowing winter crops, such as winter wheat, to reach at least stem elongation and flag-leaf formation before full poplar leaf expansion. Indeed, C3 species are well suited to these systems because of their lower light saturation point and the limited overlap with the leafing period of deciduous trees [33]. This results in reduced competition for resources such as light and water at flowering, which is a crucial phase for these crops, and yields are not significantly affected compared with systems without trees [32,34]. Similar findings were reported by Blanchet et al. [35] for winter pea in southern France within a 25 year hybrid walnut (Juglans regia × nigra, cv. NG23) AF system, although yield reductions increased with increasing shade levels. This is consistent with recent work in this study area by Tramacere et al. [13,36] on sulla, a winter legume crop, but generally is confirmed by studies conducted in temperate regions on legumes [37,38,39]. Considering the overall cereal–legume IC, in Mediterranean environments, evidence from silvo-arable systems with deciduous trees mainly concerns annual crops grown as sole crops (e.g., winter cereals or grain legumes), whereas studies explicitly testing cereal–legume IC under deciduous tree rows are scarce; the closest available evidence comes from grass–legume mixtures used as forage [40,41]. However, alley-cropping systems with olive (Olea europaea L.) provide initial indications that winter cereal–legume associations can still be advantageous under tree cover, and also when intercropped with evergreen trees. Indeed, in rainfed olive orchards, barley–pea and triticale–pea intercrops had increased productivity and improved resource-use efficiency, although responses varied across years [42,43].

4.3. Effect of the Nitrogen Rate

We found that N availability was the main driver of productivity and N uptake in wheat–pea IC. Across both years, the component crops had different responses; wheat showed significant increases in aboveground biomass and nutrient accumulation with higher N rates, especially during the shorter season (i.e., 2023–2024). However, higher NF rates reduced HI and NHI, indicating excessive vegetative growth at the expense of N remobilization to grain [44]. These results emphasize that maximizing N supply does not necessarily lead to higher N use efficiency or grain yield.

In contrast, pea showed inconsistent and context-dependent responses to N fertilization. Higher N availability increased pea grain yield, pod weight, HI, and NHI, suggesting that pea was more dependent on mineral N rather than biological N fixation under short and stressful cycles, like in the first year. In contrast, in the second year, under optimal growth conditions, increased N supply reduced pod number and pod biomass, possibly due to inhibition of rhizobial activity and reduced symbiotic efficiency, which has been widely reported for legumes exposed to high soil mineral N levels [6,9,10].

Overall, under stress or disturbed phenological conditions, legumes show positive responses to NF, while pea performance relies more on BNF under optimal conditions.

In both years, the intermediate NF rate (70 kg N ha⁻¹) represented the best agronomic compromise, allowing high productivity of both crops while preserving wheat nitrogen efficiency, maintaining legume function, and avoiding excessive vegetative growth. This NF rate also favoured good weed control by providing a rapid canopy closure without overstimulation of the spontaneous vegetation.

Even if in this study N dynamics have been based on crop yield and nutrient accumulation, and direct measurement studies of mineral N in soils and BNF have not been conducted, considering all the above observations together, we conclude that under shorter crop cycles, NF is a yield driver, while during optimal cycle duration, N regulates assimilative distribution and nutrient efficiency.

5. Conclusions

Although further research is needed to validate different herbaceous and tree crop combinations under different pedoclimatic conditions, our study underscores that wheat–pea IC is a sound agroecological strategy able to balance, under Mediterranean conditions, the requirements for productivity with those of nitrogen use efficiency and resilience. The IC displayed functional plasticity both for forage and grain production, while maintaining productivity and N efficiency even in challenging climatic and management conditions.

Moreover, we did not find any yield penalty in the silvo-arable system, supporting the agroecological value of integrating annual intercrops with tree components.

Finally, N management was identified as one of the major drivers affecting system performance, and 70 kg ha⁻¹ was proven to be the most sustainable strategy to obtain both forage and grain.