Optimizing legume green manure applications for enhanced forage sorghum-sudangrass performance and soil property improvements

Site description and climate

The experiment was established at the Aydın Adnan Menderes University Research and Application Farm, located in Koçarlı district, Aydın province, Türkiye (37°44′N, 27°45′E, at an elevation of 25–65 m above sea level). The field is situated in the Büyük Menderes Basin, where the Mediterranean climate prevails. Before the establishment of the experiment, the land had been utilized for rotational cultivation of various leguminous and grass forage crops until 2022. The study was carried out from October 2022 to September 2024 to assess the impact of green manure applications on forage crop performance and associated soil properties. To evaluate the effects of different green manure treatments, three forage legume species, common vetch (Vicia sativa L.), narbon vetch (Vicia narbonensis L.), and forage pea (Pisum sativum subsp. arvense (L.) Asch. & Graebn.) were tested alongside fertilized and unfertilized control plots. The soils at the experimental site are young alluvial deposits with moderate lime content, classified as Inceptisols under the Soil Taxonomy. These soils fall under the Xerept suborder, which exhibits relatively weak horizon development, and their organic matter content and total nitrogen levels range from low to moderate. Their susceptibility to waterlogging, particularly in winter, and the presence of stoniness are significant factors that limit their agricultural use. According to the FAO/WRB classification, these soils are identified as Hypocalcic Calcisol, where high lime content may limit the availability of plant nutrients, particularly phosphorus (Mert Akça & Atatanır, 2020). Soil samples taken before green manure incorporation revealed the following properties: a pH of 8.09, available phosphorus of 5.11 ppm, exchangeable potassium of 197 ppm, exchangeable calcium (Ca) of 4,301 ppm, iron (Fe) of 14.94 ppm, zinc (Zn) of 0.35 ppm, and a sandy loam texture.

Temperature data from 2022 to 2024 indicated notable deviations from long-term averages. While temperatures in January and February remained below long-term averages (8.2 °C and 9.4 °C, respectively), March recorded 12.09 °C, exceeding the long-term mean of 11.8 °C. In terms of precipitation, January recorded 151.2 mm, surpassing the long-term average of 119.1 mm, whereas February precipitation dropped sharply to 14.8 mm, significantly below the average of 91.3 mm. The combination of low winter temperatures and reduced precipitation suggests an increased risk of drought, while relatively high spring temperatures reflect the regional impacts of seasonal variability (Fig. 1).

Notably, maximum summer temperatures frequently ranged between 35°C and 40°C, surpassing long-term averages. This trend reflects a rising frequency of extreme heat events, thereby increasing the region’s susceptibility to drought stress (Fig. 2).

Experimental design and agronomic management

Treatments for the field experiment were arranged in a randomized complete block design with three replicates. The study included five winter treatments, consisting of various legume-based green manure strategies and controls:

• (1)

  Winter fallow – unfertilized sorghum × sudangrass (cv. Tonka),

• (2)

  Winter fallow – fertilized sorghum × sudangrass,

• (3)

  Common vetch (cv. Alper) + sorghum × sudangrass,

• (4)

  Narbon vetch (cv. Bozdag) + sorghum × sudangrass,

• (5)

  Forage pea (cv. Taskent) + sorghum × sudangrass.

Each green manure treatment was further combined with three phenological incorporation stages: pre-flowering, 10% flowering, and 100% flowering. The incorporation stage was selected based on previous findings that demonstrate its influence on nutrient release and subsequent crop uptake (Morelli et al., 2022).

No fertilization was applied prior to the sowing of green manure crops. Rhizobium inoculation was not performed because the experimental field had been previously cultivated for 13 consecutive years with various leguminous forage species, during which effective nodulation had been consistently observed. As a result, it was assumed that sufficient populations of native Rhizobium were already present in the soil, making additional inoculation unnecessary.

The seeding rates were determined as 100 kg ha⁻¹ for common vetch and 150 kg ha⁻¹ for narbon vetch and forage pea. No pesticides were applied during the green manure growth period. The study was conducted with 11 different treatments, each replicated three times. Each plot measured 28 m² (8 × 3.5 m), and this plot configuration was consistently used for both the green manure and subsequent sorghum × sudangrass phases. The row spacing was set at 25 cm for green manure and 50 cm for sorghum × sudangrass. The aboveground biomass of the green manure crops was first mowed using a mechanical mower. Subsequently, the residues were incorporated into the soil using a plot-type rototiller (Yurduser Agr., İzmir, Türkiye) to a depth of approximately 15–25 cm. To ensure thorough and uniform incorporation, the tillage process was performed twice. Twenty days after incorporation at the 100% flowering stage, sorghum × sudangrass was sown using a seed drill (Maschio Gaspardo, Italy). Following the green manure and control treatments, no fertilization was applied to sorghum × sudangrass plots. However, in the fertilized sorghum × sudangrass treatment, a two-stage fertilization strategy was implemented. Basal fertilization consisted of 50 kg ha⁻¹ of nitrogen and 50 kg ha⁻¹ of phosphorus, applied using a compound fertilizer (20-20-0, N-P₂O₅-K₂O). Topdressing fertilization included 50 kg ha⁻¹ of nitrogen, applied as ammonium sulfate ((NH₄)₂SO₄). Irrigation was applied uniformly across all plots using a drip irrigation system. Water applications were scheduled based on calculated soil field capacity, ensuring that moisture levels remained within 70–80% of field capacity to avoid both drought stress and excess leaching. Soil moisture was monitored regularly to determine irrigation timing and amounts, and adjustments were made to maintain optimal conditions for both green manure and sorghum × sudangrass growth. Sorghum × sudangrass was harvested uniformly by cutting at the milk-dough stage in both measurement years to an average stubble height of 10 cm. To minimize edge effects, samples for dry matter content and chemical analyses were taken from the three central rows after mowing, while 50 cm was left unharvested at both the front and rear ends of each plot, and the outermost rows on both sides were excluded from sampling.

Measurements

Green manure biomass was measured using 1 × 1 m quadrats immediately before incorporation. Total fresh forage yield (t ha⁻¹) was recorded at the harvest of sorghum × sudangrass in each plot. A subsample was then oven-dried at 78 °C for 48 h to determine the hay yield (t ha⁻¹). Samples were then ground to pass through a 2-mm screen with a grinder and a subsample was obtained for further analysis. The measurements of plant height (cm) and leaf number per plant were conducted by extending the uppermost fully unfolded leaves using a tape measure in 15 randomly selected plants in each plot.

The crude protein content (%) was calculated by multiplying the total nitrogen (N) concentration by 6.25 using the Kjeldahl method (Association of Analytical Communities (AOAC), 2003). Following an automated fiber analyzer (ANKOM 2000 Fiber Analyser, ANKOM Technology, Macedon, NY, USA), the neutral detergent fiber (NDF; %) and acid detergent fiber (ADF; %) contents were determined following the techniques outlined by Van Soest, Robertson & Lewis (1991). Briefly, 0.5 g of ground sample was put into the fiber filter bag (F57, ANKOM Technology, Macedon, NY, USA) for NDF analysis. The samples were then digested using diluted neutral detergent solution with triethylene glycol (FND20C). For determining ADF, the residues left behind after determining NDF can be directly digested using a solution of acid detergent concentrate and dry CTAB powder (FAD20C, ANKOM Technology, Macedon, NY, USA). Crude protein yield (t ha⁻¹) and relative feed value were calculated based on the collected data and the method of Horrocks & Vallentine (1999).

Chlorophyll content (SPAD) was measured using the SPAD-502 device (Konica-Minolta Camera Co. Ltd., Tokyo, Japan) at three phenological stages of sorghum × sudangrass; boot (SPADa), flowering (SPADb), and soft dough (SPADc) stages (Hoel, 1998). Measurements were taken from the topmost fully expanded leaf during vegetative stage, and from the flag leaf during reproductive stages (Rabin & Snapp, 2024). In each plot, SPAD values were recorded from ten randomly selected plants, and the average was used for analysis.

The normalized difference vegetation index (NDVI) values were assessed using a GreenSeeker™ handheld crop sensor (HCS 100, Trimble Ltd., Sunnyvale, CA, USA). Five measurements per plot were taken by holding the sensor 1 m above the crop canopy over a 1 m² area, and the mean value was used for further analysis (Tagarakis et al., 2017).

Leaf area measurements were conducted using a portable leaf area meter (model LI-3000C, LI-COR, Lincoln, NE, USA). For each plot, ten representative plants were sampled at the flowering stage to determine average leaf area. The leaf area index (LAI) was subsequently calculated based on the total leaf area per unit ground area following the standard equation provided by Allamine et al. (2023).

(1) $$\mathrm{L}\mathrm{A}\mathrm{I}=(\mathrm{n}\times \mathrm{L}\mathrm{A})/\mathrm{A}$$where LAI is the leaf area index (cm² cm²), n is the number of leaves, LA is the mean leaf area (cm²), and A is the sampled ground area (cm²).

Soil samples

In order to establish a baseline for soil properties characterization, soil samples were collected from the 0–30 cm layer at three time points: prior to cultivation, after the incorporation of green manures, and following the harvest of sorghum × sudangrass. Samples were taken prior to any treatment application at each phenological stage, allowing a reliable assessment of baseline site conditions. The effect of green manure on soil organic matter content was determined using the Walkley–Black method (Walkley & Black, 1934). Total nitrogen content was analyzed using the Kjeldahl method (Bremner, 1960).

Statistical analyses

All collected data were subjected to analysis of variance (ANOVA) to determine the effects of treatments on the measured parameters. The experiment included 11 treatments, comprising nine combinations of three forage legume species (common vetch, narbon vetch, and forage pea) and three incorporation stages (pre-flowering, 10% flowering, and 100% flowering), in addition to one fertilized and one unfertilized control. Each treatment was considered a distinct level in a one-way classification design. The experiment followed a randomized complete block design (RCBD) with three replications.

Significant differences among treatment means were assessed using Fisher’s least significant difference (LSD) test at a significance level of p < 0.01. LSD values were used to assign letters to treatment groups, where treatments sharing the same letter were not significantly different. These groupings were indicated above error bars in the figures and annotated in all relevant tables.

All analyses were conducted in R software (v.4.2.2; R Core Team, 2022). ANOVA and mean comparisons were performed using the agricolae package (De Mendiburu & De Mendiburu, 2019). Pearson correlation analysis was conducted using metan package (Olivoto & Lúcio, 2020) to explore relationships among traits. For graphical visualizations, Python programming language was employed with the Matplotlib library (Hunter, 2007).

Greeen manure yield

Among the legume species, narbon vetch consistently exhibited the highest green manure biomass across all incorporation stages, reaching 50.60 t ha⁻¹ at full flowering. This was followed by forage pea and common vetch, with maximum yields of 44.56 t ha⁻¹ and 42.62 t ha⁻¹, respectively, also recorded at the full flowering stage. Biomass accumulation increased progressively from the pre-flowering to the full flowering stage in all species, reflecting the typical growth trajectory of maturing legumes. At the pre-flowering stage, yields remained relatively low, ranging between 30.64 and 32.82 t ha⁻¹, while intermediate values were observed at 10% flowering, such as 39.40 t ha⁻¹ in forage pea and 42.06 t ha⁻¹ in narbon vetch. These findings demonstrate that narbon vetch outperformed the other species at all incorporation stages, emphasizing the critical importance of timing in green manure incorporation to optimize biomass production (Table 2).

Biomass production

The incorporation of different legume species at various phenological stages had a statistically significant impact on the fresh forage yield (FFY) of sorghum × sudangrass. Based on the 2-year average, the highest FFY was observed in the fertilized control treatment, which reached 86.61 t ha⁻¹, representing a substantial increase compared to the unfertilized control (32.41 t ha⁻¹). Among the green manure treatments, the highest FFY was recorded in common vetch incorporated at the 10% flowering stage (82.03 t ha⁻¹), followed closely by narbon vetch incorporated at the pre-flowering stage (78.53 t ha⁻¹) and common vetch incorporated at the same stage (77.33 t ha⁻¹). Forage pea treatments also exhibited a notable positive influence on FFY, particularly when incorporated at the pre-flowering stage, reaching 76.17 t ha⁻¹, markedly outperforming the unfertilized control (Table 3).

The interaction between year and green manure treatments further highlighted the role of climatic variation. In the second year, the fertilized control outperformed the unfertilized control by 182%, while common vetch treatments showed yield increases of up to 169%. In contrast, narbon vetch at full flowering resulted in the lowest relative increase, with yields approximately 74% higher than the unfertilized control (Fig. 3).

Hay yield (HY) varied markedly depending on the legume species and the phenological stage at which green manure was incorporated into the soil. The highest HY was observed in the fertilized control, reaching 16.81 t ha⁻¹, while the lowest yield was recorded in the unfertilized control (7.40 t ha⁻¹). Among green manure treatments, common vetch incorporated at the 10% flowering stage produced the highest HY (15.55 t ha⁻¹), closely followed by forage pea incorporated at the pre-flowering stage, which achieved 15.39 t ha⁻¹. These findings highlight the superior biomass contribution of certain species when incorporated at optimal developmental stages. In general, green manure applications led to statistically significant improvements in HY compared to the unfertilized control. However, delayed incorporation (e.g., at full flowering) tended to be less effective, suggesting that early-stage incorporation is more favorable for maximizing sorghum × sudangrass productivity. A trend consistent with fresh forage yield was also observed across species, reinforcing the role of phenological timing. Notably, a strong interaction between year and treatment was identified: common vetch incorporated at 10% flowering led to a remarkable 124% increase in HY over the unfertilized control in the second year. Nevertheless, the greatest overall enhancement was recorded in the fertilized control, which yielded a 149% increase relative to the baseline (Table 3; Fig. 3).

Plant height was significantly influenced by green manure applications, with notable differences observed across legume species and their respective incorporation stages. Based on the 2-year average, the tallest plants were recorded under the fertilized control treatment (268.66 cm), closely followed by common vetch incorporated at the pre-flowering (265.00 cm) and 10% flowering (264.83 cm) stages, both of which were statistically comparable to the fertilized control. The shortest plants were observed in the unfertilized control, where sorghum × sudangrass reached only 176.33 cm. Among green manure treatments, narbon vetch resulted in the lowest plant height improvements relative to the unfertilized control. Notably, the interaction between year and treatment revealed that plant heights under the fertilized control and under common vetch applied at both the pre-flowering and 10% flowering stages were comparable. In the second year of the experiment, the fertilized control exhibited the greatest relative increase in plant height over the unfertilized control at 54%. This was closely followed by common vetch applied at the pre-flowering and 10% flowering stages, which achieved 50% and 49% increases, respectively. These findings highlight the effectiveness of early-stage common vetch incorporation in promoting vegetative growth in sorghum × sudangrass, nearly matching the performance of mineral fertilization (Table 3; Fig. 3).

Leaf number per plant was significantly influenced by the interaction between green manure treatments and experimental years. Although slight year-to-year variations were observed, leaf counts were generally higher in 2023 compared to 2022, suggesting more favorable climatic conditions during the second year. Based on the 2-year average, the highest leaf number (12.91 leaves per plant) was recorded under the fertilized control, reaffirming the positive contribution of mineral fertilization to vegetative development. Among the green manure treatments, the incorporation of all three legume species at the pre-flowering stage and common vetch at the 10% flowering stage yielded the greatest leaf numbers in sorghum × sudangrass. In contrast, the lowest leaf count (6.73 leaves per plant) was observed in the unfertilized control, reflecting the limiting effect of nutrient deficiency on foliar development. These findings are consistent with previously reported trends in fresh and dry forage yield and plant height, underscoring once again the critical role of appropriate legume species selection and incorporation timing in optimizing sorghum × sudangrass performance within green manure-based nutrient management systems (Table 3; Fig. 3).

Forage quality

ADF content was significantly influenced by year, green manure treatments, and their interaction (p < 0.01). Among all treatments, the highest ADF value was recorded in the fertilized plots (39.08%), representing an increase compared to the unfertilized control (36.56%). Among the legume-based green manure treatments, forage pea incorporated at the pre-flowering stage exhibited the highest ADF content (38.96%), which was higher than the control and closely approximated the value observed in the fertilized treatment. Conversely, the lowest ADF level was observed in the common vetch treatment incorporated at the pre-flowering stage (34.58%), indicating a notable reduction relative to the unfertilized control. Incorporation timing exerted a clear influence; particularly, pre-flowering applications of common vetch and full-flowering applications of narbon vetch resulted in lower ADF concentrations, suggesting enhanced forage digestibility. According to the interaction effect, ADF contents generally increased in the second year; however, this trend diverged in the case of common vetch, indicating a species-specific response to interannual variability (Table 4).

Unlike ADF, NDF content was not significantly affected by year. However, green manure species and incorporation timing caused notable variations. The highest NDF value was observed in the unfertilized control (59.40%), while lower values were recorded for common vetch at pre-flowering and 100% flowering, and for narbon vetch at 10% and 100% flowering stages. These reductions indicate improved forage quality due to lower fiber concentration. Among legume treatments, common vetch and narbon vetch generally resulted in lower NDF levels compared to other species. In contrast, forage pea maintained relatively high NDF contents across all incorporation stages, with the highest value (58.46%) observed at 100% flowering. According to the year × treatment interaction, forage pea incorporated at full flowering in the second year showed the highest NDF content among green manure treatments, closely resembling the unfertilized control. These results emphasize the importance of legume selection and incorporation timing in managing forage fiber composition (Table 4; Fig. 4).

The crude protein ratio and crude protein yield of sorghum × sudangrass were significantly influenced by green manure species, incorporation timing, and growing season (p < 0.01). Both parameters exhibited superior outcomes in 2023 compared to 2024, likely due to more favorable climatic conditions that enhanced nitrogen mineralization and uptake efficiency. Among all treatments, the highest CPR was recorded for common vetch incorporated at the pre-flowering stage (11.28%), followed by narbon vetch at 100% flowering (10.89%) and 10% flowering (10.57%). These values were substantially higher than that of the unfertilized control (7.32%), highlighting the protein-enhancing potential of early-stage legume incorporation. According to the year × treatment interaction, all green manure treatments, except forage pea, generally outperformed the fertilized control in CPR. The most pronounced CPR increase over the unfertilized control was observed under pre-flowering narbon vetch in the second year, reaching 59%, followed by 100% and pre-flowering narbon vetch in the first year (Table 4; Fig. 4).

A consistent response pattern was also evident in CPY, with the highest value achieved under the fertilized control (1.64 t ha⁻¹). Among legume-based treatments, narbon vetch incorporated at the pre-flowering stage yielded 1.55 t ha⁻¹, followed by common vetch at pre-flowering (1.51 t ha⁻¹) and 10% flowering (1.49 t ha⁻¹). Notably, the year × treatment interaction revealed that narbon vetch at the pre-flowering stage in the first year led to the most substantial improvement, producing a CPY 186% higher than the unfertilized control—surpassing even the fertilized treatment. In contrast, forage pea showed comparatively lower CPR and CPY values across all stages, particularly at full flowering. Nonetheless, even forage pea applications yielded 134% to 171% higher CPY than the unfertilized control, underscoring the general effectiveness of legume green manure strategies (Table 4; Fig. 4).

Physiological traits

The chlorophyll content of sorghum × sudangrass, assessed via SPAD readings at three phenological stages (SPADa—boot, SPADb—flowering, SPADc—soft dough), was significantly influenced by green manure species and incorporation timing (p < 0.01). Across all growth stages, early-stage green manure incorporation, particularly at the pre-flowering stage, consistently yielded the highest SPAD values. At the boot stage (SPADa), chlorophyll levels were markedly elevated in plots where narbon vetch (52.55) and forage pea (52.81) were incorporated at the pre-flowering stage. Similarly, at flowering (SPADb), early incorporation again led to superior SPAD values, with narbon vetch (58.91), forage pea (57.90), and common vetch (56.88) outperforming later treatments. During the soft dough stage (SPADc), pre-flowering incorporation maintained its advantage, with forage pea reaching 56.90 and common vetch 55.91 (Table 5; Fig. 5).

In contrast, incorporation at full flowering (100%) consistently produced the lowest chlorophyll readings for instance, forage pea dropped to 49.68. At all three measurement points, control plots exhibited the lowest chlorophyll content, with values progressively declining as the crop advanced through its phenological stages. Fertilized plots performed moderately better than the control, but generally failed to match the chlorophyll levels observed in early green manure applications. These findings underscore that green manure incorporation, especially at the pre-flowering stage, can be equally or even more effective than conventional fertilization in maintaining chlorophyll content and enhancing nitrogen availability during key stages of plant development. As such, early legume-based green manure application emerges as a viable and sustainable alternative to mineral fertilization strategies for optimizing sorghum × sudangrass physiological performance (Table 5; Fig. 5).

NDVI of sorghum × sudangrass was significantly affected by green manure species and their incorporation timing (p < 0.01), whereas the year effect was not statistically significant. The highest NDVI value (0.908) was recorded in the fertilized control, demonstrating the positive impact of mineral fertilization on canopy development. However, green manure treatments also markedly enhanced NDVI relative to the unfertilized control (0.476), underscoring their contribution to vegetative vigor. Among the legume-based treatments, forage pea incorporated at the pre-flowering (0.871) and 10% flowering (0.846) stages, as well as narbon vetch applied at pre-flowering (0.850) and 10% flowering (0.855), achieved the highest NDVI values across both years. In contrast, full-flowering incorporation generally resulted in diminished NDVI values, indicating reduced canopy performance when application timing was delayed.

Interaction effects revealed interannual variability in response to legume type and incorporation stage. Nevertheless, a consistent trend emerged: the contribution of green manure to NDVI tended to decline as phenological development progressed. The fertilized control remained superior in both years, showing 67% and 119% increases over the unfertilized control in the first and second years, respectively. Among green manure treatments, the most pronounced enhancement was observed with forage pea incorporated at the pre-flowering stage in the second year, exhibiting a 110% NDVI increase relative to the unfertilized control. Conversely, the smallest gain was recorded under common vetch incorporated at full flowering in the first year, which still provided a 32% improvement (Table 5; Fig. 6).

According to the results, the leaf area index (LAI) of sorghum × sudangrass was significantly influenced by the green manure species, incorporation timing, and year. Green manure applications conducted at early phenological stages, particularly with common vetch and forage pea, resulted in higher LAI values compared to later-stage applications and the unfertilized control. However, narbon vetch exhibited a distinct pattern, differing from the other species. Based on the 2-year average, common vetch incorporated at the pre-flowering stage produced the highest LAI (10.01 cm² cm⁻²), clearly outperforming all other green manure species and incorporation timings. This treatment enhanced LAI by 43% in the first year and 56% in the second year relative to the unfertilized control. Similarly, common vetch incorporated at 100% flowering (8.41 cm² cm⁻²), and forage pea applied at pre-flowering (8.31 cm² cm⁻²), also led to notable improvements over the control. In contrast, narbon vetch consistently produced lower LAI values across all incorporation timings. The lowest LAI (7.07 cm² cm⁻²) was recorded under narbon vetch applied at full flowering, with only modest increases of 5% and 6% over the control in the first and second years, respectively. While the fertilized control achieved a higher LAI (9.14 cm² cm⁻²) than the unfertilized control, it remained lower than the value obtained from early incorporation of common vetch These findings highlight the superior potential of early-incorporated common vetch in promoting leaf area development, demonstrating performance that exceeds not only the unfertilized control but also the mineral fertilization strategy (Table 6; Fig. 7).

Relative feed value

The relative feed value of sorghum × sudangrass was significantly influenced by both the green manure species and their incorporation timing. The highest RFV values were recorded in treatments involving narbon vetch and common vetch, with the latter consistently achieving statistically superior values across all incorporation stages except at 10% flowering. In these treatments, RFV ranged from 146.98 to 149.62, with the highest value observed in narbon vetch incorporated at the full flowering stage (149.62). In contrast, forage pea treatments resulted in comparatively lower RFV values, and no statistically significant differences were detected among its incorporation stages (ranging between 141.76 and 142.41). All green manure applications led to improvements in RFV compared to the unfertilized control plots. Notably, narbon vetch applied at full flowering enhanced RFV by approximately 9% in the first year. A particularly noteworthy finding is that the fertilized control failed to surpass the RFV levels attained by the most effective green manure treatments. These results underscore the potential of specific legume species—particularly narbon vetch and common vetch—to enhance forage quality when incorporated at appropriate phenological stages, offering a viable alternative to mineral fertilization (Table 7; Fig. 8).

Pearson correlation’s of phenological stages of green manures applied to sorghum-sudangrass

At the pre-flowering stage, several statistically significant correlations were identified among yield, physiological, and forage quality parameters. A particularly strong negative correlation was observed between relative feed value (RFV) and neutral detergent fiber (NDF) (r = −1.00), reinforcing the well-established inverse relationship between fiber content and forage quality. Likewise, acid detergent fiber (ADF) and NDF showed a strong negative correlation (r = −0.87), reflecting similar behavior in fiber fractions. Hay yield (HY) was negatively correlated with both the leaf area index (LAI; r = −0.72) and crude protein ratio (CPR; r = −0.65), suggesting that early-stage biomass accumulation may limit either canopy expansion or nitrogen assimilation (Fig. 9A).

In contrast, RFV was positively correlated with CPR (r = 0.78), confirming the role of protein concentration in enhancing forage quality. A moderate positive correlation between LAI and CPR (r = 0.65) indicates that canopy development contributes to nitrogen uptake. Similarly, CPR and crude protein yield (CPY) were positively correlated (r = 0.65), reflecting synchronized improvements in both protein concentration and total nitrogen accumulation as a result of early green manure incorporation (Fig. 9A).

At the 10% flowering incorporation stage, the correlation patterns became more complex, reflecting physiological transitions in plant development. A strong positive correlation was recorded between leaf number (LN) and fresh forage yield (FFY) (r = 0.81), as well as between plant height (PH) and hay yield (HY) (r = 0.85), highlighting the importance of vegetative growth in biomass production. CPY also showed positive associations with HY (r = 0.66), and NDF was positively linked to PH (r = 0.61). Additionally, NDVI exhibited positive correlations with all SPAD readings, while CPR and RFV were positively associated with all SPAD indices except SPADa, emphasizing the relationship between chlorophyll content and nitrogen-based quality traits (Fig. 9B).

Nevertheless, several notable negative correlations emerged. SPADa showed a negative association with FFY (r = –0.63), while NDVI was also negatively correlated with FFY (r = −0.77). All SPAD readings were negatively correlated with both HY and PH, suggesting potential trade-offs between chlorophyll concentration and structural biomass. Additionally, CPR and HY (r = −0.61) and NDVI and HY (r = −0.69) showed inverse relationships. Strong negative correlations were also observed between CPR and ADF/NDF, and between RFV and NDF (r = −1.00), reaffirming the negative impact of fiber accumulation on forage quality (Fig. 9B).

Under the full flowering incorporation stage, correlation structures remained robust (Fig. 9C). A strong positive relationship was found between PH and FFY (r = 0.85), and a moderate positive correlation between LAI and FFY (r = 0.64), indicating that canopy size continues to influence biomass accumulation. ADF and LAI were positively correlated (r = 0.78), as were NDF and ADF (r = 0.81), suggesting coordinated increases in fiber components with plant maturity. Importantly, RFV was strongly and positively associated with both CPR (r = 0.86) and CPY (r = 0.76), further validating the relevance of protein content in determining feed quality.

Conversely, all SPAD indices were negatively correlated with FFY and HY, and SPADb showed a negative relationship with PH, indicating possible declines in chlorophyll content under delayed incorporation. LAI was negatively associated with SPADa, hinting at reduced canopy light penetration or altered leaf structure at later stages. Notably, CPR, CPY, and RFV all showed strong negative correlations with both ADF and NDF, emphasizing the adverse effects of fiber accumulation on forage digestibility and nutritional value. Furthermore, green manure biomass yield was negatively correlated with FFY, suggesting that higher green manure productivity may not always translate into enhanced subsequent crop yields under late incorporation (Fig. 9C).

Soil organic matter and total nitrogen

Soil organic matter (SOM) content exhibited a marked improvement following the incorporation of legume-based green manures. Initially present at low levels prior to incorporation, SOM increased substantially after application, frequently exceeding 1.50 g kg⁻¹, particularly under treatments with common vetch, narbon vetch, and forage pea. Although a slight decline was observed after the sorghum harvest, SOM levels remained notably higher than baseline values, indicating that green manure incorporation contributed to the buildup of soil organic carbon while partially offsetting the depletion caused by the subsequent cropping cycle. Moreover, SOM levels in 2024 were marginally higher than in 2023, suggesting a cumulative benefit of repeated green manure applications on long-term soil health. In terms of incorporation timing, applications made between 10% and 100% flowering stages were especially effective in enhancing SOM levels, underscoring the critical role of phenological stage in maximizing organic matter contribution (Fig. 10A). Total nitrogen (TN) content in the soil followed a similar trajectory. The lowest TN levels were recorded prior to green manure incorporation, followed by significant increases across legume-based treatments. In several instances, TN concentrations reached or surpassed 0.80 g kg⁻¹, highlighting the capacity of these species to enhance soil nitrogen status. Despite minor reductions following sorghum harvest, TN levels remained well above initial values. The slightly elevated TN content observed in 2024 further reinforces the sustained impact of green manures on nitrogen enrichment over time. Although there was some variability in the response across incorporation timings, applications made at approximately 10% flowering—particularly with common vetch—proved most effective in preserving soil nitrogen levels (Fig. 10B).