Dual Production of Full-Fat Soy and Expanded Soybean Cake from Non-GMO Soybeans: Agronomic and Nutritional Insights Under Semi-Organic Cultivation

Featured Application

The developed dual production model allows for the scalable manufacturing of protein-rich, expanded soybean cake (ESC) from non-GMO soybeans under semi-organic cultivation. The ESC product can be applied in clean-label, extruded foods, functional nutrition, hybrid protein formulations, and specialized feed systems aligned with EU Green Deal objectives.

Abstract

The diversification of plant protein sources is a strategic priority for European food systems, particularly under the EU Green Deal and Farm to Fork strategies. In this study, dual production of full-fat soy (FFS) and expanded soybean cake (ESC) was evaluated using non-GMO soybeans cultivated under semi-organic conditions in Central Poland. Two agronomic systems—post-emergence mechanical weeding with rotary harrow weed control (P1) and conventional herbicide-based control (P2)—were compared over a four-year period. The P1 system produced consistently higher yields (e.g., 35.6 dt/ha in 2024 vs. 33.4 dt/ha in P2) and larger seed size (TSW: up to 223 g). Barothermal and press-assisted processing yielded FFS with protein content of 32.4–34.5% and oil content of 20.8–22.4%, while ESC exhibited enhanced characteristics: higher protein (37.4–39.0%), lower oil (11.6–13.3%), and elevated dietary fiber (15.8–16.3%). ESC also showed reduced anti-nutritional factors (e.g., trypsin inhibitors and phytic acid) and remained microbiologically and oxidatively stable over six months. The semi-organic P1 system offers a scalable, low-input approach to local soy production, while the dual-product model supports circular, zero-waste protein systems aligned with EU sustainability targets.

1. Introduction

The diversification of plant protein sources in the European Union (EU) has become a strategic imperative in response to sustainability goals and the need for food system resilience. Replacing 50% of imported soybean meal would require approximately 6.6 million hectares of arable land to be reallocated from other crops, with Poland alone projected to expand soybean cultivation by 0.5 million hectares [1,2]. In the context of the European Green Deal and the Farm to Fork Strategy [3], the European Union promotes crop diversification—including soybean—as a key tool to enhance biodiversity and soil health, and to reduce dependency on synthetic pesticides and imported feedstocks. These policy frameworks emphasize the strategic role of protein crops in transitioning toward resilient and sustainable food systems. As noted in the 2024 European Parliament report [4] on input dependency on organic soybean imports, the EU imported approximately 192,000 tons of organic soybeans in 2022, primarily from non-EU countries such as Ukraine, Togo, and China. This dependency poses risks to supply stability, especially under global trade disruptions and geopolitical crises. The Institute for European Environmental Policy further highlights that soybeans and other protein crops currently cover less than 3% of EU arable land. Expanding domestic soybean cultivation is therefore crucial to reduce Europe’s reliance on imported protein sources and to align agricultural production with agroecological and climate objectives under the Green Deal [4,5].

Soybean (Glycine max), characterized by its high protein yield per hectare and favorable nutritional profile, is increasingly viewed as a cornerstone crop for achieving European protein self-sufficiency. Central and Eastern European regions, including Poland, offer significant potential for soybean cultivation, provided that agro-climatic constraints are carefully addressed. According to Debaeke et al. [3], understanding the local agro-climatic constraints and designing location-specific cropping systems is crucial for developing resilient soybean production systems in new European zones.

Integrating soybean into locally tailored cropping systems offers a dual advantage: decreased dependence on imported feedstocks and improved environmental sustainability [6]. In this context, the promotion of non-GMO and ecologically suited soybean cultivars is gaining momentum, addressing both consumer demand and policy ambitions. However, as Purnhagen et al. [7] emphasize, the coexistence of strict biotechnology regulations and sustainability goals poses regulatory and agronomic challenges. Field-based approaches, exemplified by this study, provide actionable strategies for reconciling these policy objectives.

Soybean cultivation is constrained by agro-climatic factors such as shortened growing seasons, summer droughts, and regional photothermal limitations. These conditions necessitate the adoption of location-specific agronomic practices, early-maturing cultivars, and non-GMO varieties suited to temperate environments. Expanding soybean cultivation northward into temperate regions requires refinement of agronomic practices [8]. Our study addresses this premise by evaluating non-GMO, full-fat soybean cultivation under contrasting cropping systems in Central Poland. Aligning sowing dates and cultivar maturity with regional photothermal profiles remains essential for yield optimization, especially under summer drought and shortened vegetative seasons.

Klaiss et al. [9] highlight that such demand brings agronomic challenges, particularly in weed control and nitrogen management within organic and low-input systems. Our research addresses these challenges through a semi-organic cropping system that eliminates herbicides and incorporates mechanical weed control.

Tataridas et al. [10] advocate integrating mechanical tools, crop rotation, and cover crops to suppress weed populations in accordance with Green Deal principles. Similarly, Winkler et al. [11] report that conservation tillage and biodiversity-driven management can shift weed composition toward more controllable species. Recent glyphosate-alternative reports [12] and smallholder studies [13] further confirm the viability of non-chemical weed suppression strategies.

The aim of this study was to evaluate the technical feasibility and nutritional value of producing two soybean-based ingredients—full-fat soy (FFS) and expanded soybean cake (ESC)—using a continuous barothermal processing system applied to non-GMO soybeans cultivated under semi-organic conditions in Central Poland.

The specific objectives of this study were to

(i)

Compare the agronomic performance and seed composition under semi-organic (P1) and conventional (P2) weed control systems.

(ii)

Evaluate the nutritional, functional, and anti-nutritional properties of full-fat soy (FFS) and expanded soybean cake (ESC) produced via barothermal processing.

(iii)

Assess the storage stability, oxidative safety, and microbiological quality of both products during shelf-life under ambient conditions.

It was hypothesized that barothermal-pressing combined with semi-organic cultivation would produce stable, nutritionally valuable soy ingredients suitable for circular food and feed systems aligned with EU Green Deal priorities.

2. Materials and Methods

2.1. Field Experiment and Sample Collection

Soybeans (non-GMO) were cultivated in experimental plots located in Central Poland, under semi-organic conditions aligned with Green Deal principles. After harvest, mature soybean seeds were cleaned and stored under controlled temperature and humidity prior to processing and analysis.

2.1.1. Design and Location

Field experiments were conducted from 2021 to 2024 in Brodnica County, Kuyavian-Pomeranian Voivodeship, Poland (GPS: 53.1912 N, 19.4337 E). The soil was classified as quality class IV, with a granulometric composition of >1.0 mm: 5%, 1.0–0.1 mm: 52%, 0.1–0.02 mm: 22%, and <0.02 mm: 21%. Soil pH (1 M KCl) ranged from 6.4 to 6.6. Available nutrients included P₂O₅: 27.3–42.0 mg/100 g, K₂O: 13.6–20.2 mg/100 g, and Mg: 4.1–6.2 mg/100 g. Organic matter content remained between 1.42% and 1.51%.

Five randomized blocks (each 0.172 ha) were established, with 2 × 2 m microplots replicated five times per treatment. The total cultivated area was 4 ha.

2.1.2. Climatic Conditions

Meteorological data were recorded using an automated weather station throughout the April–October growing seasons. Rainfall ranged from 525.5 mm to 561.5 mm, with notable variability in May precipitation (Figure 1). Mean temperatures varied from 6.2 °C in April to over 21 °C in July/August. The driest year was 2024 (8 mm rainfall in May), potentially influencing crop development and yield.

2.1.3. Plant Material and Agronomic Practices

The early-maturing ‘Abaca’ soybean variety (Saatbau) was used. This cultivar was selected due to its short vegetation cycle, high fat content (~22.7%), and moderate protein levels (~36.2%), which make it particularly suitable for temperate Central European agro-climatic conditions. Additionally, ‘Abaca’ has demonstrated reliable performance under drought-prone summer periods and offers favorable processing characteristics for both full-fat soy and oil-extracted products. Soybeans were sown at a density of 70 seeds/m² using a disc seeder (Agro Masz SN300, Strzelce Małe, Poland) (row spacing: 24 cm; depth: 2.5–4.5 cm). Fertilization (P + K + Ca, Mg, S) was applied uniformly (350 kg/ha annually). Herbicide regimes varied annually, with increasing reliance on mechanical weeding (P1) versus conventional chemical protection (P2).

• P1 (semi-organic): post-emergence weed control was carried out mechanically twice per growing season using a rotary harrow. Herbicides were used only in pre-emergence stages and limited post-emergence applications, aligned with semi-organic practice standards.
• P2 (conventional): Weed control relied exclusively on chemical herbicides, with both pre-emergence and multiple post-emergence applications. No mechanical interventions were applied in this system.

In both systems, metribuzin and S-metolachlor were used for pre-emergence control (replaced in 2024 by petoxamide + clomazone due to regulatory changes). Post-emergence treatments included bentazon + imazamox as needed (

Table 1. Summary of agronomic practices for soybean cv. ‘Abaca’ cultivation from 2021 to 2024, including herbicide regimes, mechanical weeding, and key dates.

Year	Treatment	Pre-Emergence Herbicides (g ha−1)	Post-Emergence Herbicides (g ha−1)	Mechanical Weeding	Sowing Date	Harvest Date
2021	P1	Metribuzin 200 + S-metolachlor 960	None	2×	11 May	30 October
	P2	Metribuzin 200 + S-metolachlor 960	None	None	11 May	30 October
2022	P1	Metribuzin 200 + S-metolachlor 960	1× Bentazon 600 + Imazamox 28	2×	05 May	1 October
	P2	Metribuzin 200 + S-metolachlor 960	1× Bentazon 600 + Imazamox 28	None	05 May	1 October
2023	P1	Metribuzin 200 + S-metolachlor 960	2× Bentazon 300 + Imazamox 14	2×	22 April	22 September
	P2	Metribuzin 200 + S-metolachlor 960	2× Bentazon 300 + Imazamox 14	None	22 April	22 September
2024	P1	Metribuzin 240 + Clomazone 48 + Petoxamide 800	4× Bentazon 150 + Imazamox 7	2×	27 April	9 September
	P2	Metribuzin 240 + Clomazone 48 + Petoxamide 800	4× Bentazon 150 + Imazamox 7	None	27 April	9 September

).

2.1.4. Sampling and Biometric Data Collection

Post-emergence and pre-harvest plant density (plants per m²), plant height (cm), and pod number per plant were recorded. Ten plants per treatment were sampled for biometric traits. Yield was calculated at 13% moisture content. Thousand seed weight (TSW, g) was measured in triplicate.

2.2. Laboratory Analysis of Soybean Seeds

Seed samples were analyzed using near-infrared reflectance spectroscopy (NIRS) on a FOSS Infratec NOVA analyzer (Hillerød, Denmark) following the PN EN ISO 12099:2017 standard [14]. Total protein, oil content, moisture, and bulk density were assessed. Additionally, raw seeds were analyzed for trypsin inhibitor activity (TIA) and protein solubility in water and potassium hydroxide (KOH-SP). All analyses were conducted in accredited laboratories, ensuring methodological reliability and data integrity.

2.2.1. Processing of Soybean Seeds

The entire barothermal processing line, including optional oil pressing, is presented in Figure 2, with key physico-chemical changes summarized in

Table 2. Processing stages, durations, and temperature conditions in the barothermal and press-assisted modification system for soy products.

Stage	Duration (min)	Temperature (°C)	Physico-Chemical Transformation
Crushing on roller mills	0.1	15–20	Particle size reduction (4–6 fragments per seed)
Aspiration with rotary sieve	0.7	15–20	Husk removal
Conditioning (water + steam)	0.3	80–85	Moisture increase and mixing
Steam buffering (saturation)	13.3	95–98	Inactivation of trypsin inhibitors
Pressure expansion	0.5	120–135	Gelatinization, cell rupture
Drying	16	120 → 60	Moisture reduction to ~10%
Optional screw pressing	1.5	~80–120	Fat removal #
Cooling	8.0	60 → 35	Protein stabilization, microbiological control
Mechanical structuring	0.2	35 → 30	Compaction and fraction uniformity

^# Only applied in ESC production pathway.

. This continuous system ensured reproducible production of FFS and ESC under controlled time–temperature profiles.

The technological process described in Table 2 represents a simplified flow of a continuous, steam-assisted processing line. The longest and most critical phase involves soybean exposure to barothermal conditions in a hydrothermal reactor. Our internal optimization trials demonstrated that thermolabile anti-nutritional compounds—especially trypsin inhibitors (TIA), used as an inactivation marker—are most effectively reduced during the steam saturation phase (13.3 min at 95–98 °C), where water and steam create a buffered thermal environment. While elevated temperatures are known to cause some degree of protein denaturation, the presence of moisture in this system plays a protective role, preventing overheating and structural degradation. Under these controlled conditions, we achieved maximum PDI (Protein Dispersibility Index) values while significantly reducing TIA. Furthermore, optional screw pressing of the expanded FFS fraction contributed to further TIA reduction, albeit at the cost of moderately lowered protein solubility (PDI). This balance between TIA inactivation and protein preservation was a key design objective of the barothermal process.

To reduce the oiliness of the final product, modifications were introduced after preliminary drying, directing the soy expandant to screw pressing, followed by further drying and structuring. The soybean processing described in this study was conducted on a pilot-scale industrial line operated by Agrolok Sp. z o.o. (Poland), a private-sector partner in the national Industrial PhD Program (“doktorat wdrożeniowy”), of which this publication is a direct outcome. The company collaborated on the design, execution, and technical validation of the ESC and FFS production processes.

2.2.2. Chemical Composition and Analytical Procedures

The analytical procedures for chemical composition, fatty acid profile, dietary fiber, and anti-nutritional factors were conducted as described in our previous work (Ambroziak & Wenda-Piesik, 2025) [15], using accredited laboratory protocols compliant with PN-EN ISO and AOAC standards. For detailed methods, instrumentation, and calibration procedures, see Ambroziak & Wenda-Piesik [15]. The following determinations were performed:

• Crude Protein (CP): Kjeldahl method (PN-EN ISO 20483:2007).
• Oil Content (OC): Soxhlet extraction (PN-EN ISO 734-1:2007).
• Crude Fiber (CF): PN-EN ISO 6865.
• Moisture and Ash: Gravimetric methods.
• Fatty Acid Profile (SFA, MUFA, PUFA, n-3, n-6): PN-EN ISO 12966-1:2015 + AC:2015 + ISO 12966-2:2017.
• Total and Digestible Carbohydrates: Subtraction method according to Regulation (EU) No 1169/2011 and EC No 152/2009.

Additional compounds:

• Dietary Fiber (DF): AOAC 985.29.
• Non-Starch Polysaccharides (NSP): GC per Englyst & Cummings, AOAC 994.13.
• Klason Lignin (KL): AACC 32-25.
• Uronic Acid (UA): Colorimetric method.
• Raffinose Family Oligosaccharides (RFO): GC method.
• Phytic Acid (PA): Haug & Lantzsch method.
• Total Phenolic Content (TPC): Folin–Ciocalteu assay.

The barothermal-pressing system operated under carefully controlled conditions to ensure both microbial safety and nutrient preservation. During the steam-buffering and expansion phases, material was exposed to temperatures ranging from 95 to 135 °C, with transient internal pressure buildup reaching approximately 0.8–1.2 MPa. These thermal-pressure conditions were sufficient to inactivate anti-nutritional factors such as trypsin inhibitors while preserving the solubility and functional integrity of protein fractions. Additionally, the moderate residence time (~14 min cumulative) minimized nutrient degradation and supported the retention of dietary fiber and polyunsaturated fatty acids. Post-expansion drying and optional pressing further concentrated the protein and fiber content, particularly in the ESC fraction. For detailed procedures and instrument calibration, see Ambroziak & Wenda-Piesik [15].

2.2.3. Storage Stability Assessment

FFS, ESC, and oil were stored for 1, 3, and 6 months under ambient, light-protected conditions. Evaluations included:

Lipid oxidation indices: Peroxide Value (PV), Anisidine Value (AV), Totox Index (2PV + AV).

Microbiological safety: Mesophilic Aerobic Microorganisms (MAM), Coliforms, Enterobacteriaceae, Staphylococcus aureus (CPS), Salmonella spp., yeasts, and molds.

All analyses followed PN-EN ISO standards.

2.3. Statistical Analysis

Data were tested for normal distribution (Shapiro–Wilk test). Square-root transformations were used where applicable. One-way or two-way ANOVA (p = 0.05) was applied, followed by Tukey’s HSD post hoc test (p = 0.05). Statistical analyses were performed using Statistica 13.1.

3. Results and Discussion

3.1. Agronomic Performance Under Semi-Organic and Conventional Systems

Soybeans cultivated under the P1 system consistently demonstrated higher plant height and pod number, likely due to enhanced vegetative development supported by mechanical weeding and optimal sowing timing (

Table 3. Agronomic parameters (mean ± SE) for soybean cv. ‘Abaca’ cultivated from 2021 to 2024.

Year	Treatment	Post-Emergence Density (Plants Per m2)	Pre Harvest Density (Plants Per m2)	Number of Pods Per Plant	Plant Height at Maturity Stage(cm)	Seed Yield Converted to 13% Moisture (dt ha−1)	TSW (g)
2021	P1 z	55.10 ± 2.20 b,*	50.20 ± 2.00 a	29.90 ± 3.05 a	74.90 ± 11.30 a	30.20 ± 0.20 a	223.20 ± 5.12 a
	P2	62.30 ± 2.49 a	48.80 ± 1.96 a	31.00 ± 3.20 a	70.50 ± 12.02 b	28.60 ± 0.20 b	201.30 ± 5.40 b
2022	P1	49.60 ± 1.98 a	48.60 ± 1.94 a	30.30 ± 2.85 a	60.90 ± 8.80 a	21.20 ± 0.14 a	192.60 ± 4.80 a
	P2	50.20 ± 2.00 a	43.80 ± 1.75 b	27.10 ± 3.10 b	60.30 ± 7.60 a	20.40 ± 0.15 a	180.70 ± 4.30 b
2023	P1	52.40 ± 2.09 a	50.20 ± 2.00 a	30.50 ± 4.12 a	74.90 ± 7.19 a	30.40 ± 0.22 a	215.10 ± 4.80 a
	P2	53.60 ± 2.14 a	49.30 ± 1.97 a	30.60 ± 4.00 a	71.90 ± 8.30 b	27.20 ± 0.21 b	205.30 ± 4.60 b
2024	P1	58.20 ± 2.32 b	57.40 ± 2.30 a	36.50 ± 3.60 a	87.70 ± 12.40 a	35.60 ± 0.20 a	220.70 ± 5.30 a
	P2	62.40 ± 2.49 a	57.70 ± 2.30 a	34.40 ± 3.18 b	81.90 ± 13.10 b	33.40 ± 0.20 b	215.40 ± 5.10 a

^z P1—post-emergence mechanical weeding with rotary harrow, P2—conventional herbicide-based weed control without mechanical weeding. *—small superscript letters (a, b) indicate significant differences between P1 and P2 within each year according to Tukey’s HSD test (p = 0.05). TSW—thousand seed weight.

). These findings align with Karges et al. [8], who emphasized the need for aligning sowing date and cultivar maturity with regional photothermal conditions. In our study, the mechanical disturbance associated with rotary weeding may have promoted compensatory pod development, consistent with reports by Toleikiene et al. [16].

Although P2 plots initially showed higher post-emergence plant density in certain years (2021, 2024), P1 maintained superior pre-harvest density and yield across most seasons. In 2022 and 2024, low rainfall in May (36.5 mm and 8.0 mm, respectively) and higher temperatures disrupted early development. This likely shortened the R1–R7 phase, limiting assimilate accumulation and pod filling. Conversely, cooler, wetter conditions in 2021 and 2023 supported prolonged phenological stages, aligning with higher yields, especially under the P1 cultivation system. The lowest yields were observed in 2022, coinciding with delayed sowing and drought during flowering. Drought and elevated temperatures negatively impact nodulation and biological nitrogen fixation [17,18]. Our results corroborate this, showing yield suppression during critical phenophases. Otherwise, year 2023 and 2024 showed yield recovery due to earlier sowing and better rainfall distribution [19,20]. Yield variability from 2021 to 2024 reflects strong climatic influences on soybean development. Setiyono et al. [20] emphasized temperature and daylength as key drivers of phenology, with elevated temperatures accelerating development and reducing yield potential if critical phases like flowering or seed filling are shortened. Matthews et al. [21] observed similar outcomes in temperate zones under fluctuating moisture regimes. In 2021 and 2024, the yield gap between P1 and P2 reached 1.5 to 2.0 dt/ha (Table 3). The use of mechanical weed control also contributed to higher thousand-seed weight (TSW), especially in dry years like 2024. The P1 system, combining mechanical weeding and increased sowing precision, effectively suppressed weed infestation without post-emergence herbicide use. This is in line with the findings of Tataridas et al., [10] and Winkler et al., [11] who supported mechanical solutions as glyphosate alternatives under EU Green Deal constraints. Advances in AI-powered weed detection offer future opportunities for automation of mechanical weeding [22].

3.2. Nutritional Composition of Soybean Seeds

The proximate composition of raw soybean seeds harvested from both agronomic systems during 2021–2024 is presented in

Table 4. Composition (mean ± SE) of raw soybean seeds (cv. ‘Abaca’) under two weed control systems from 2021 to 2024.

Year	Treatment	Crude Protein (%)	Oil Content (%)	Water (%)	Bulk Density (kg hL−1)	Admixture (%)
2021	P1 z	33.14 ± 1.12 a,*	18.97 ± 0.67 a	13.23 ± 0.42 a	69.70 ± 4.70 a	1.01 ± 0.02 a
	P2	33.32 ± 1.30 a	18.65 ± 0.72 a	13.37 ± 0.37 a	65.52 ± 5.10 b	0.66 ± 0.04 b
2022	P1	32.00 ± 0.96 a	20.11 ± 0.55 a	13.31 ± 0.34 a	69.85 ± 6.11 a	1.15 ± 0.04 a
	P2	32.64 ± 1.06 a	19.50 ± 0.60 a	13.05 ± 0.27 a	69.52 ± 5.80 a	1.12 ± 0.03 a
2023	P1	33.51 ± 1.07 a	19.28 ± 0.64 a	12.97 ± 0.30 a	70.86 ± 6.95 a	1.10 ± 0.02 a
	P2	34.12 ± 1.32 a	18.90 ± 0.69 a	12.60 ± 0.32 a	70.48 ± 7.14 a	1.13 ± 0.02 a
2024	P1	32.52 ± 1.14 a	20.02 ± 0.58 a	12.46 ± 0.33 a	71.13 ± 8.02 a	1.09 ± 0.04 a
	P2	32.53 ± 1.17 a	20.05 ± 0.62 a	11.78 ± 0.33 b	70.03 ± 7.80 a	1.06 ± 0.04 a

^z P1—post-emergence mechanical weeding with rotary harrow, P2—conventional herbicide-based weed control without mechanical weeding. *—small superscript letters (a, b) indicate significant differences between P1 and P2 within each year according to Tukey’s HSD test (p < 0.05).

.

Crude protein (CP) levels remained relatively stable, ranging from 32.00% to 34.12%, with no statistically significant differences observed between the P1 (semi-organic) and P2 (conventional) treatments (p > 0.05). These results are consistent with prior multi-site evaluations of non-GMO cultivars under Central Polish conditions, where the protein content for ‘Abaca’ ranged from 39.6% to 42.8% depending on seasonal hydrothermal conditions [23]. In the present study, lower protein levels may reflect partial drought exposure during key vegetative phases, especially in 2022.

Oil content (CO) exhibited minor inter-annual variation, ranging from 18.65% to 20.11%. While oil levels tended to be slightly higher under P1 management in certain years (e.g., 2022 and 2024), differences were generally non-significant. These oil concentrations confirm the suitability of ‘Abaca’ for both full-fat soy processing and oil extraction, as also supported by prior cultivar screenings [23]. Interestingly, our earlier research highlighted that oil content in soybean can be sensitive to hydrothermal variability, with drier seasons sometimes yielding elevated oil concentrations due to altered seed filling dynamics.

Moisture content at harvest ranged from 11.78% to 13.37%, reflecting proper harvest timing and consistent post-harvest drying. A small but statistically significant moisture reduction was observed in P2 during 2024 (p < 0.05), potentially linked to field microclimate variability and plant maturity differences.

Bulk density values ranged between 65.52 and 71.13 kg hL⁻¹, with slightly higher values under P1 conditions, suggesting improved kernel development in mechanically weeded plots. Bulk density is an important indicator for storage stability and processing efficiency, with values above 68 kg hL⁻¹ generally meeting industrial processing standards [24,25].

Seed admixture content ranged from 0.66% to 1.15%. The slightly elevated admixture levels observed under P1 conditions may reflect minor mechanical seed coat damage induced by rotary hoeing and the absence of chemical desiccation prior to harvest [10]. Despite these differences, all admixture levels remained well within commercial acceptance limits for food-grade soybeans [26].

Taken together, these findings confirm that both cultivation systems maintained high compositional stability of soybean seeds. Importantly, the semi-organic P1 approach preserved seed quality while offering potential agroecological benefits, in line with EU Green Deal and Farm to Fork policy objectives. The convergence of results between the current experiment and our earlier field phenotyping study [23] reinforces the suitability of ‘Abaca’ as a robust cultivar for Central European non-GMO soybean production.

The chemical composition of the processed soy products, full-fat soy (FFS), and expanded soybean cake (ESC), obtained from the barothermal and press-assisted processing system, is summarized in

Table 5. Nutritional composition (mean ± SE) of full-fat soy (FFS) and expanded soybean cake (ESC) produced via barothermal processing, 2022–2024.

Characteristic	2022		2023		2024
	FFS	ESC	FFS	ESC	FFS	ESC
Oil content (%)	20.79 ± 0.35 a,*	11.56 ± 0.56 b	21.35 ± 0.25 a	13.33 ± 0.51 b	22.40 ± 0.54 a	12.05 ± 0.25 b
Crude protein (%)	34.48 ± 0.58 b	38.98 ± 0.68 a	34.30 ± 0.90 b	37.57 ± 0.22 a	32.40 ± 0.67 b	37.40 ± 0.40 a
Ash (%)	4.82 ± 0.10 a	4.99 ± 0.11 a	5.05 ± 0.45 a	5.50 ± 0.13 a	5.10 ± 0.35 a	5.05 ± 0.15 a
Sugar (%)	9.23 ± 0.24 b	10.98 ± 0.15 a	8.75 ± 0.05 b	10.83 ± 0.23 a	7.80 ± 0.16 b	10.75 ± 0.25 a
Crude fiber (%)	5.57 ± 0.42 b	6.49 ± 0.11 a	5.65 ± 0.35 a	5.88 ± 0.28 a	5.40 ± 0.31 b	6.10 ± 0.50 a
Dietary fiber (%)	13.63 ± 1.36 b	16.19 ± 0.37 a	14.30 ± 0.30 b	15.84 ± 0.54 a	14.00 ± 0.65 b	16.30 ± 0.90 a
TOT Carb (%)	43.70 ± 1.08 a	31.48 ± 0.91 b	37.80 ± 0.20 a	31.64 ± 0.90 b	43.00 ± 0.67 a	31.10 ± 2.10 b
Dig Carb (%)	12.81 ± 0.25 b	22.84 ± 1.90 a	13.35 ± 0.25 b	19.03 ± 1.67 a	14.20 ± 0.23 b	19.35 ± 0.95 a
EV (kcal 100 g−1)	410.2 ± 5.41 a	379.4 ± 5.02 b	416.5 ± 13.50 a	380.5 ± 6.70 b	432.0 ± 9.04 a	389.5 ± 0.50 b
Water (%)	11.93 ± 0.42 a	12.13 ± 0.16 a	12.35 ± 0.05 a	10.70 ± 0.40 a	12.50 ± 0.05 a	10.20 ± 0.10 a

*—small superscript letters (a, b) indicate significant differences between FFS—full-fat soya and ESC—expanded soybean cake within each year according to Tukey’s HSD test (p < 0.05). TOT Carb—total carbohydrates, Dig Carb—digestible carbohydrates, EV—energy value.

. The pressing stage significantly altered the nutrient profile of the products, yielding two distinct matrices suitable for different food and feed applications.

Crude protein (CP) content was significantly higher in ESC, ranging from 37.40% to 38.98%, compared to 32.40% to 34.48% in FFS (p < 0.05). This increase reflects the protein concentration effect resulting from partial fat removal during the mechanical pressing stage. Similar protein concentration patterns have been reported in previous processing studies on plant-based protein concentrates [15,27].

Oil content (OC) was strongly reduced in ESC due to the press-assisted processing step, ranging from 11.56% to 13.33%, while FFS retained oil levels between 20.79% and 22.40% (p < 0.05). The oil reduction in ESC supports its applicability for protein-rich formulations where excessive lipid content is undesirable, such as extruded protein snacks or feed pellets.

Total sugar content was moderately higher in ESC (10.75–10.98%) compared to FFS (7.80–9.23%). This may reflect partial concentration of water-soluble carbohydrates during oil extraction and post-expansion drying. Crude fiber (CF) content also increased slightly in ESC (5.88–6.49%) relative to FFS (5.40–5.65%), contributing to improved dietary fiber fractions.

Dietary fiber (DF) content reached 15.84–16.30% in ESC, significantly higher than the 13.63–14.30% observed in FFS. This increase may offer functional benefits for both feed digestibility and food applications targeting glycemic control and satiety [28,29].

Total carbohydrate (TOT Carb) content was consistently higher in FFS (37.80–43.70%), while ESC exhibited lower totals (31.10–31.64%), again reflecting compositional shifts following lipid extraction. Conversely, digestible carbohydrates (Dig Carb) were significantly elevated in ESC (19.03–22.84%) compared to FFS (12.81–14.20%), suggesting potential advantages for energy modulation in nutritional formulations. These findings are consistent with previous reports indicating that oil removal concentrates structural carbohydrates, enhancing dietary fiber content [27,29].

The calculated energy value (EV) decreased significantly in ESC, ranging from 379.4 to 389.5 kcal/100 g, while FFS provided higher caloric density (410.2–432.0 kcal/100 g), primarily due to the higher residual oil content. The elevated fiber content may support glycemic regulation, microbiota modulation, and stool bulk, which are relevant for both feed and functional food applications [30]. Soluble and insoluble fiber fractions in soy-based matrices can also influence viscosity and bile acid binding, contributing to cholesterol-lowering potential [31,32]. The inverse relationship between fiber and digestible carbohydrates aligns with energy-reduction goals in high-fiber diet formulations [33]. Furthermore, the fiber-rich nature of ESC may enhance satiety, as shown in studies with other plant-derived protein-fiber blends [28]. These nutritional differences between FFS and ESC highlight the technical flexibility of the barothermal-pressing system in producing tailored soy ingredients with distinct compositional profiles. ESC offers particular advantages as a clean-label, high-protein, fiber-enriched product suitable for multiple functional food and specialized feed applications. Szulc et al. [34] and Schulp et al. [35] emphasized that regionally integrated production, processing, and consumption systems are essential for sustainable transitions. The dual-product approach using FFS and ESC illustrates the circular valorization of protein, oil, and fiber streams [34]. Environmental stressors, as noted by Szulc [34] and Hou et al. [36], can alter biomass allocation, contributing to observed yield variation. Ordoñez et al. [37] further demonstrated that shoot carbon-to-nitrogen ratios respond to soil fertility gradients and climatic drivers, influencing aboveground resource partitioning. Their findings suggest that stoichiometric flexibility in soybean may underlie adaptive strategies to environmental heterogeneity. Finally, breeding programs must continue integrating traits for stability and nutritional value [38]. Water content remained stable across processing years and treatments, ranging from 10.20% to 12.50%, indicating good process control and standardization.

3.3. Processing Outcomes and Product Differentiation (FFS vs. ESC)

The comparative assessment with commercial reference products (defatted soybean meal, soybean cake, and raw soybeans;

Table 6. Comparison of nutritional and anti-nutritional parameters of FFS and ESC versus commercial reference soy products.

Compound	Product
	FFS	ESC	DSMB x	SC y	RS z
CO (%)	21.51 ± 0.28 d,*	12.31 ± 0.27 c	1.78 ± 0.10 a	6.10 ± 0.10 b	20.68 ± 0.49 d
CP (%)	33.73 ± 0.44 a	37.98 ± 0.29 a,b	45.28 ± 1.44 b	42.05 ± 0.05 a,b	34.28 ± 0.64 a
Ash (%)	4.99 ± 0.10 a	5.18 ± 0.06 a,b	6.30 ± 0.13 b	5.95 ± 0.05 b	4.24 ± 0.08 a
CF (%)	5.54 ± 0.24 b	6.16 ± 0.12 c	4.20 ± 0.05 a	6.50 ± 0.05 c	5.58 ± 0.19 b
PDI (%)	35.01 ± 3.09 c	26.46 ± 2.34 b	14.31 ± 1.31 a	18.20 ± 0.05 a	85.30 ± 2.10 d
KOH-SP (% CP)	80.79 ± 4.11 a	90.98 ± 1.54 b	94.30 ± 1.58 b	78.20 ± 0.10 a	96.00 ± 0.00 b
DM (%)	87.74 ± 0.24 a	88.99 ± 0.20 a	88.53 ± 0.33 a	87.80 ± 0.50 a	88.24 ± 0.34 a
TIA (mg/g)	2.86 ± 0.17 a	3.33 ± 0.38 a	2.15 ± 0.38 a	8.60 ± 0.05 b	19.48 ± 0.29 c

^x—defatted soybean meal, ^y—soybean cake, ^z—raw soybean seeds. CO—crude oil, CP—crude protein, CF—crude fiber, PDI—Protein Dispersibility Index, KOH-SP—soluble protein in potassium hydroxide, DM—dry mass, TIA—trypsin inhibitor activity. *—small superscript letters (a, b, c, d) indicate significant differences between products within each parameter according to Tukey’s HSD test (p < 0.05).

) demonstrates that ESC achieves a balanced profile with superior protein concentration, enhanced protein solubility, moderate anti-nutritional load, and elevated dietary fiber content. Such characteristics highlight ESC as a promising clean-label ingredient for both functional foods and high-protein feed formulations, especially where controlled oil content and digestibility optimization are desired. The elevated dietary fiber content of ESC (15.8–16.3%) has practical relevance for both food and feed applications. In human nutrition, this level of fiber supports gut microbiota modulation, glycemic control, and increased satiety, making ESC particularly suitable for functional foods aligned with clean-label and high-fiber dietary trends. In animal feed, it may improve digestive health and nutrient utilization, especially in monogastric species. Moreover, the protein concentration in ESC (37.4–39.0%) exceeds that of conventional soybean cake (typically 30–35%), offering enhanced nutritional value and formulation efficiency in high-protein diets. This composition positions ESC as a superior alternative in the development of plant-based, protein-enriched, or hybrid protein products, responding to both sustainability goals and evolving consumer demands.

The barothermal-pressing process effectively reduced key anti-nutritional factors and improved protein digestibility in both full-fat soy (FFS) and expanded soybean cake (ESC). The trypsin inhibitor activity (TIA) was significantly lowered from 19.48 mg/g in raw soybeans to 2.86 mg/g in FFS and 3.33 mg/g in ESC (Table 6), fully complying with recommended safety thresholds for animal feed and human consumption [39,40,41].

Protein functionality parameters showed favorable modifications upon processing. The protein solubility in potassium hydroxide (KOH-SP) improved markedly in ESC, reaching 90.98% of crude protein compared to 80.79% in FFS (p < 0.05), indicating higher extractability of functional protein fractions after oil removal. Protein dispersibility index (PDI) values followed the opposite trend, with slightly lower values for ESC (26.46%) relative to FFS (35.01%), reflecting the thermal aggregation of certain protein fractions during pressure-expansion. Nonetheless, both values remain within acceptable ranges for most functional food and feed applications (Table 6).

Thermal-pressure treatment reduced TIA to 2.86 mg/g FFS and 3.33 mg/g ESC compared to raw soy 19.48 mg/g (Table 6), within safe feed limits [40,41]. Similarly, Bales and Lock [39] confirmed the benefit of heat treatment for deactivating trypsin inhibitors. Our products also showed reduced phytic acid (~0.25%) and RFO (~4%) relative to raw soy [42]. While RFOs support microbiota [43], their presence in feed may limit use for monogastric animals. Authors recommend further processing (fermentation, enzymatic treatment) to improve bioavailability [38,44].

3.4. Functional and Anti-Nutritional Properties of Soy Products

Total non-starch polysaccharides (T-NSPs), representing the indigestible carbohydrate fraction, were significantly enriched in ESC, ranging from 12.43% to 12.90%, compared to 10.34–10.89% in FFS (p < 0.05) (

Table 7. Anti-nutritional and dietary fiber components (% dry matter) in FFS and ESC, 2022–2024.

Compound Content (% Dry Matter)	2022		2023		2024
	FFS	ESC	FFS	ESC	FFS	ESC
CP	36.12 ± 0.39 b,*	41.99 ± 0.25 a	36.10 ± 0.95 b	39.82 ± 0.24 a	34.16 ± 0.71 b	39.42 ± 0.43 a
CO	24.13 ± 0.04 a	14.78 ± 0.05 b	24.90 ± 0.30 a	16.77 ± 0.64 b	26.16 ± 0.63 a	15.07 ± 0.31 b
Ash	6.31 ± 0.00 a	6.24 ± 0.00 a	6.64 ± 0.59 b	6.77 ± 0.16 a	6.73 ± 0.46 a	6.17 ± 0.19 b
T-NSP	10.34 ± 0.01 b	12.90 ± 0.08 a	10.89 ± 0.01 b	12.43 ± 0.08 a	10.69 ± 0.01 b	12.71 ± 0.08 a
UA	2.53 ± 0.01 a	2.52 ± 0.04 a	2.67 ± 0.01 a	2.43 ± 0.04 b	2.61 ± 0.01 a	2.48 ± 0.04 b
KL	1.20 ± 0.01 a	1.33 ± 0.03 a	1.26 ± 0.01 a	1.28 ± 0.02 a	1.24 ± 0.01 b	1.31 ± 0.02 a
RFO	3.93 ± 0.04 a	4.11 ± 0.03 a	4.14 ± 0.04 a	3.96 ± 0.04 b	4.06 ± 0.03 a	4.05 ± 0.04 b
DF	17.98 ± 0.00 b	20.99 ± 0.00 a	18.94 ± 0.39 b	20.22 ± 0.68 a	18.58 ± 0.86 b	20.68 ± 1.14 a
PA	1.40 ± 0.01 a	1.43 ± 0.01 a	1.47 ± 0.01 a	1.38 ± 0.01 b	1.45 ± 0.01 a	1.41 ± 0.01 b
TPC	2.49 ± 0.03 a	2.43 ± 0.02 a	2.62 ± 0.03 a	2.34 ± 0.02 b	2.57 ± 0.03 a	2.39 ± 0.03 b

*—small superscript letters (a, b) indicate significant differences between FFS and ESC within each year according to Tukey’s HSD test (p < 0.05). T-NSP—total non-starch polysaccharides, UA—uronic acid, KL—Klason lignin, RFO—raffinose family oligosaccharides, DF—dietary fiber, PA—phytic acid, TPC—total phenolic content.

).

This increase correlates with enhanced fiber concentration after oil extraction. Dietary fiber (DF) content also increased in ESC (20.22–20.99%) relative to FFS (17.98–18.94%), offering potential prebiotic and satiety-enhancing properties for specialized formulations [28,43].

Raffinose family oligosaccharides (RFO), commonly associated with flatulence in monogastric animals, remained relatively stable across processing and storage, with values ranging from 3.93% to 4.14% in FFS and 3.96% to 4.11% in ESC. While moderate RFO content may offer prebiotic benefits by modulating gut microbiota [41], further enzymatic or fermentative processing could optimize its application in monogastric nutrition [29,31].

Phytic acid (PA) levels remained between 1.38% and 1.47% across all samples, with ESC showing slightly lower values than FFS (p < 0.05). These concentrations are consistent with previously reported values for thermally treated soy products and fall within acceptable nutritional limits [39]. Nevertheless, phytic acid reduction strategies may further improve mineral bioavailability in target formulations.

Total phenolic content (TPC), measured as gallic acid equivalents (GAE), ranged from 2.34 to 2.62 mg GAE/g dry matter, showing minor differences between products and years (Table 7). The preservation of these bioactive compounds supports the antioxidant potential of both FFS and ESC in functional applications.

3.5. Fatty Acid Profile and Lipid Composition

The fatty acid (FA) composition of both full-fat soy (FFS) and expanded soybean cake (ESC) was analyzed across a 6-month storage period (

Table 8. Fatty acid composition (% total FA) of FFS and ESC during 1–6 months of storage.

Heat Treatment Method	Storage Duration (Months)	SFA (%)	MUFA (%)	PUFA (%)	n-3 (%)	n-6 (%)
FFS	1	34.55 ± 0.01 a,*	23.85 ± 0.46 a	65.25 ± 0.79 a	7.60 ± 0.23 a	49.30 ± 1.49 a
	3	34.40 ± 0.40 a	22.40 ± 0.60 a	64.00 ± 1.00 a	8.00 ± 0.40 a	52.60 ± 2.40 a
	6	33.70 ± 0.00 a	24.20 ± 0.00 a	62.30 ± 0.00 a	6.60 ± 0.00 a	47.00 ± 0.00 a
ESC	1	15.88 ± 0.74 c	22.93 ± 1.77 b	69.58 ± 5.14 b	7.93 ± 0.77 b	57.44 ± 4.90 b
	3	18.60 ± 1.83 b	23.18 ± 2.22 b	70.75 ± 6.23 b	8.83 ± 0.96 b	58.58 ± 6.20 b
	6	22.00 ± 0.00 a	28.00 ± 0.00 a	86.00 ± 1.00 a	11.00 ± 0.00 a	75.00 ± 1.00 a

*—small superscript letters (a, b, c) indicate significant differences between months within FFS and ESC according to Tukey’s HSD test (p < 0.05). SFA—saturated fatty acids, MUFA—monounsaturated fatty acids, PUFA—polyunsaturated fatty acids, n-3—omega3, n-6—omega6.

). The processing stages and subsequent oil removal significantly altered the FA profiles between the two products, while temporal storage effects remained limited.

Saturated fatty acids (SFA) were substantially lower in ESC compared to FFS. In FFS, SFA content ranged from 33.70% to 34.55% across storage time, while in ESC, values increased slightly from 15.88% at one month to 22.00% after six months. The consistently lower SFA fraction in ESC enhances its lipid quality profile, in alignment with nutritional recommendations favoring lower saturated fat intake [45].

Polyunsaturated fatty acids (PUFAs) represented the dominant lipid fraction in both products. ESC exhibited particularly favorable PUFA levels, ranging from 69.58% to 86.00%, whereas FFS contained PUFA from 62.30% to 65.25%. The PUFA/SFA ratio exceeded 2.5 in ESC throughout storage, which is well above the FAO/WHO minimum recommendation of 0.45 for cardiovascular health benefits [45].

The n-3 (omega-3) fatty acid fraction remained stable in both products but was slightly higher in ESC (7.93–11.00%) than in FFS (6.60–8.00%). Nonetheless, the n-6/n-3 ratio averaged 6.8:1 across both matrices, remaining above optimal nutritional targets. This indicates potential for further improvement via blending with omega-3 rich oils such as flaxseed, chia, or canola to better align with dietary recommendations [46,47].

The processing-induced lipid modifications observed in ESC are consistent with prior reports highlighting the impact of partial oil removal and thermal expansion on fatty acid partitioning within the protein-fiber matrix [27]. The relative enrichment of PUFA in ESC suggests potential applications in functional foods targeting anti-inflammatory and cardio-protective formulations, while its moderate residual lipid content supports favorable textural and sensory attributes in high-protein products.

Overall, both FFS and ESC demonstrated high oxidative stability of their lipid fractions across the 6-month storage period, with no substantial degradation or unfavorable shifts in fatty acid composition.

3.6. Oxidative Stability and Microbiological Safety During Storage

The oxidative stability of both full-fat soy (FFS) and expanded soybean cake (ESC) was monitored over a 6-month storage period under ambient, light-protected conditions (

Table 9. Oxidative stability parameters of full-fat soy (FFS), expanded soybean cake (ESC), and extracted soybean oil during 6-month storage.

Characteristic	FFS			ESC			Oil
	1	3	6	1	3	6	1	3	6
AV (mg g−1 KOH)	2.03 ± 0.01 a,b,*	3.10 ± 0.02 b	4.40 ± 0.03 c	1.37 ± 0.01 a	1.83 ± 0.01 a	3.25 ± 0.02 b	0.64 ± 0.00 a	2.10 ± 0.01 a	13.8 ± 0.90 b
FFA (%)	1.33 ± 0.01 a	2.67 ± 0.02 b	3.46 ± 0.02 c	0.89 ± 0.01 a	1.48 ± 0.01 b	2.01 ± 0.01 c	0.95 ± 0.01 a	1.15 ± 0.01 a,b	1.50 ± 0.01 b
PV (meq O2 kg−1)	2.62 ± 0.02 a,b	3.25 ± 0.02 b	5.70 ± 0.04 c	1.13 ± 0.01 a	1.48 ± 0.01 a	2.15 ± 0.01 a,b	4.20 ± 0.03 a	4.50 ± 0.03 a	4.80 ± 0.03 a
ANV	0.61 ± 0.00 a	0.95 ± 0.01 a,b	1.30 ± 0.01 b	0.38 ± 0.00 a	1.30 ± 0.01 b	3.61 ± 0.02 c	0.40 ± 0.00 a	0.60 ± 0.00 a	0.90 ± 0.01 a
Totox index	5.84 ± 0.04 a,b	7.45 ± 0.05 a,b	12.7 ± 08 b	2.65 ± 0.02 a	4.25 ± 0.03 a	7.91 ± 0.05 a,b	8.80 ± 0.06 a	9.60 ± 0.06 a	10.5 ± 0.07 a

*—small superscript letters (a, b, c) indicate significant differences between storage months for FFS, ESC, and oil, according to Tukey’s HSD test (p < 0.05). AV—acid value, FFA—free fatty acids, PV—peroxide value, ANV—anisidine value.

). ESC consistently demonstrated superior oxidative stability compared to FFS throughout storage.

In FFS, Totox index values progressively increased from 5.84 at month 1 to 12.70 after 6 months, indicating gradual lipid oxidation during storage. In contrast, ESC maintained low Totox values, ranging from 2.65 to 7.91, even after 6 months, confirming the stabilizing effect of partial fat removal. These results align with previous findings that reduced lipid content enhances oxidative stability in soy-based products [48,49].

Peroxide value (PV), anisidine value (ANV), and acid value (AV) followed similar trends, with ESC consistently showing lower values compared to FFS (p < 0.05). The reduction of free fatty acids (FFA) during processing further contributed to ESC’s superior oxidative performance.

Soybean oil extracted during ESC production exhibited moderate oxidative stability. While Totox values remained acceptable up to 3 months (8.80–9.60), they increased to 10.5 by month 6, suggesting limited storage stability without antioxidant supplementation. These results underscore the need for additional stabilization strategies for isolated soybean oil if extended shelf-life is required.

Microbiological quality remained within safe limits across all samples and storage periods (

Table 10. Microbiological counts (cfu g⁻¹) in FFS and ESC during storage for 1–6 months.

Soy Product	Month	MAM at 30 °C	Coli	Y & M	Enterobacteriaceae	CPS	Salmonella
FFS	1	<100	<10	<10	<10	<10	nd 1
	3	129	<10	<100	<10	<10	nd
	6	1624	<100	113	<10	<10	nd
ESC	1	<100	<10	<10	<10	<10	nd
	3	<100	<10	<100	<10	<10	nd
	6	1000	<10	<100	<10	<10	nd

¹—not detected. MAMs—mesophilic aerobic microorganisms, Y&M—yeast and molds, CPS—Staphylococcus aureus.

). Total mesophilic aerobic microorganisms (MAMs) remained below 1000 cfu/g in ESC at all timepoints. In FFS, MAM levels slightly increased after 6 months (1624 cfu/g) but stayed within acceptable food safety standards. Coliforms, Enterobacteriaceae, Staphylococcus aureus (CPS), yeasts and molds (Y&M), and Salmonella spp. were not detected in any sample throughout storage, confirming effective microbiological safety of both products.

The combination of high oxidative stability, controlled microbial quality, and stable lipid profile confirms the suitability of ESC for prolonged storage and extended distribution chains. These properties are particularly relevant for both feed and functional food sectors, where product stability under varying storage conditions is a critical requirement.

Additionally, the mild thermal conditions used during processing may contribute not only to oxidative stability but also to the reduction of allergenic protein isoforms, as previously reported for thermally processed soy products [50].

3.7. Integration into Circular Protein Systems

The soy-based products presented in this study—expanded soybean cake (ESC) and full-fat soy (FFS)—are derived from locally grown European soybeans, free from the environmental burdens associated with transatlantic transport. Our internal carbon footprint analysis of raw soybeans from field trials revealed a low emission level of 89 kg CO₂^−eq per ton of dry soybean matter. Moreover, the processing approach described here utilizes exclusively physical methods, such as mechanical treatment and steam conditioning, without the use of chemical solvents (e.g., hexane). The production of both FFS and ESC follows a zero-waste principle, as no by-products are generated. An additional systemic issue identified by the authors and an industrial partner is the lack of industrial-scale processing aimed at protein quality rather than maximum oil yield. In this context, protein should be prioritized as the main product, while oil should be considered a secondary fraction—reversing the conventional paradigm in soybean processing. The compositional and functional properties of expanded soybean cake (ESC) position it as a versatile ingredient for multiple food and feed applications. With a protein content ranging from 37.40% to 38.98%, ESC aligns with current formulation standards for high-protein extruded foods, nutritional supplements, and meat analogs. The moderate residual oil content (11.56–13.33%) provides favorable textural and sensory attributes, contributing to improved palatability and processing behavior, particularly in clean-label product development.

The digestibility profile and reduced anti-nutritional factors (low trypsin inhibitor activity and phytic acid levels) make ESC highly suitable for age-specific functional foods, including formulations targeting elderly populations, sports nutrition, and clinical nutrition [28,51]. Additionally, the elevated dietary fiber content (15.84–16.30%) supports glycemic regulation, promotes satiety, and contributes to favorable gut microbiota modulation [29,31,43].

The clean-label profile of ESC—free from synthetic binders or emulsifiers—addresses the growing consumer demand for minimally processed, transparent ingredients [52]. Its functional composition also facilitates the development of hybrid protein products that combine plant and dairy protein fractions to achieve optimal textural and nutritional balance [53].

ESC exhibits favorable rheological behavior during extrusion and thermal processing due to its unique matrix composition, which combines residual lipids, proteins, and fibers. The porous structure and partial oil retention enhance expansion properties, fiber integrity, and emulsification capacity, making ESC highly suitable for extruded snacks, protein-enriched bakery products, ready-to-drink beverages, and savory applications [54,55,56].

Furthermore, the presence of residual lipids improves lubrication during extrusion, reducing mechanical torque and energy consumption, while enhancing product uniformity and stability [57]. This processing efficiency adds economic value in industrial applications and supports the scalability of ESC-based formulations.

Collectively, the nutritional, functional, and technological characteristics of ESC position it as a promising ingredient for innovative plant-based protein systems aligned with current health, sustainability, and consumer trends.

The dual-product model developed in this study, yielding both expanded soybean cake (ESC) and soybean oil from non-GMO soybeans, exemplifies a circular processing approach that maximizes the valorization of local agricultural resources. By integrating both agronomic and bioprocessing innovations, the model addresses multiple strategic challenges facing European protein self-sufficiency, particularly under the EU Green Deal and Farm to Fork frameworks [34,35].

The ability to stabilize nutritional quality across variable growing seasons demonstrates that the combination of semi-organic cultivation (P1 system) with controlled barothermal processing can deliver protein-rich soy products that are resilient to climatic fluctuations. This resilience is especially relevant for Central and Eastern European regions where drought episodes and photothermal variability are becoming more frequent due to climate change [36,37].

The circularity of the model is further reinforced by its efficient partitioning of protein, oil, and fiber fractions without generating significant processing waste. ESC offers a high-protein, fiber-enriched matrix suitable for both food and feed chains, while the co-produced oil can be utilized in culinary applications, animal feed formulations, or bio-based industries, depending on market demands.

Moreover, the integration of such systems into regional supply chains reduces reliance on imported genetically modified soybean meals, supports local economies, and enhances traceability and food sovereignty. This aligns with growing consumer preferences for non-GMO, locally sourced, and sustainably produced plant proteins. ESC consistently demonstrated superior oxidative stability during storage compared to FFS. This can be attributed to two main factors: (i) the reduced residual lipid content, which limits the availability of substrates for lipid peroxidation, and (ii) the retention of natural antioxidants, such as phenolic compounds, during mild barothermal treatment. Together, these characteristics significantly mitigate oxidation and extend shelf-life, particularly under ambient conditions without synthetic stabilizers.

At the breeding level, the observed inter-annual variability in yield and composition also highlights the continuing need to select and develop soybean genotypes optimized for Central European agro-climatic conditions, with traits for both agronomic performance and compositional stability [23,38].

Collectively, the proposed circular processing platform provides a scalable and environmentally aligned alternative to conventional soybean supply chains, bridging primary production, functional processing, and health-oriented food innovation. Its implementation offers both economic and ecological benefits, contributing to resilient protein systems capable of addressing future global protein challenges.

The compositional and functional attributes of ESC—particularly its clean-label profile, high protein and fiber content, and oxidative stability—make it well-suited for scalable integration into sustainable food and feed chains. These characteristics directly support EU policy objectives for protein self-sufficiency, reduced reliance on GMO imports, and circular bioeconomy models under the Green Deal and Farm to Fork frameworks. ESC also aligns with growing industry demand for functional, plant-based ingredients that enable innovation in meat analogs, sports nutrition, and therapeutic formulations.

However, this study has certain limitations. The field trials were conducted using a single soybean cultivar (‘Abaca’), which may limit the generalizability of agronomic and compositional findings. Future studies should assess multiple genotypes under diverse agro-climatic conditions to validate the robustness and scalability of the proposed dual-product processing model.

4. Conclusions

This study confirms the technical, nutritional, and functional viability of producing two complementary soy-based ingredients—full-fat soy (FFS) and expanded soybean cake (ESC)—using a continuous, barothermal-pressing process specifically adapted for Central European agro-climatic conditions. The application of the P1 cultivation system, integrating mechanical weed control and optimized agronomic management, supported stable yields, improved plant performance, and consistent seed composition across variable weather conditions.

ESC showed distinct nutritional benefits, characterized by increased protein levels, decreased lipid content, and a higher dietary fiber fraction, and lowered anti-nutritional factors such as trypsin inhibitors and phytic acid. Its favorable oxidative stability and microbiological safety over a 6-month storage period further enhance its suitability for both food and feed applications. These attributes position ESC as a promising ingredient for clean-label, high-protein formulations, including extruded snacks, functional foods, and hybrid protein products.

From an agroecological perspective, the semi-organic P1 model offers a scalable, glyphosate-free alternative to conventional soybean cultivation, supporting EU Green Deal objectives for sustainable agriculture and regional protein self-sufficiency. The circular processing platform developed herein enables efficient partitioning of protein, lipid, and fiber fractions, contributing to resource-efficient and economically viable local supply chains.

Future research should focus on further valorization pathways for ESC in specialized food sectors, optimization of fatty acid profiles via oil blending to enhance omega-3 content, and comprehensive techno-economic assessments of full-scale implementation across diverse European agro-climatic zones.

5. Patents

[P] A. Wenda-Piesik., K. Ambroziak. ‘Method of producing protein-energy products based on oil seeds derived from soybean and hemp seeds’. The application was numbered: P.450940.