Reduced Chemical Fertilizer Combined with Microbial Inoculants: Implications for Soil Fertility and Profitability in Mediterranean Vegetable Production

Abstract

A three-year field experiment (2021–2023) in southeast Spain evaluated whether reduced mineral fertilization, with or without plant-growth-promoting microorganisms, could maintain crop productivity and modify selected soil indicators in a Mediterranean vegetable rotation. Four treatments were compared: conventional fertilization (T1), reduced fertilization (T2; −30% or −50%), reduced fertilization plus bacterial inoculants (T3), and reduced fertilization plus bacterial–fungal inoculants (T4). Crop yields were not significantly affected by fertilization strategy. Potato yields ranged from 55,661 to 60,741 kg ha⁻¹, those of broccoli from 14,928 to 16,797 kg ha⁻¹, and those of melon from 30,815 to 33,423 kg ha⁻¹. Inoculated treatments were associated with some quality responses, including higher potato tuber firmness in T4 (16.0 vs. 13.2 kg cm⁻² in T1), whereas melon soluble solids tended to be slightly lower. Soil analyses showed changes in some nutrient-related indicators, including a 217% increase in NH₄⁺ in T4 and a 0.75% decrease in pH in T3. Reduced fertilization lowered production costs by about 9%. Under the conditions of this field trial, reduced fertilization maintained yield and gross margin relative to conventional fertilization, and inoculated treatments under reduced fertilization showed differences in selected soil indicators.

1. Introduction

Vegetable production in semi-arid Mediterranean regions often relies on intensive mineral fertilization, which can increase production costs and the risk of nutrient losses without necessarily improving yield under site-specific conditions [1,2]. Central to this development has been the widespread use of synthetic inorganic fertilizers, especially those rich in N, P, and K. While these inputs have contributed to short-term yield gains, their intensive use has also generated important environmental and agronomic challenges [3,4]. In vegetable production systems, N fertilizers account for more than 78% of total greenhouse gas emissions [5,6], and soil denitrification in intensive greenhouse systems may result in losses of up to 50% of applied N as N₂ and N₂O [7]. In addition, nitrate leaching can be extremely high, reaching 117.9–884.2 kg ha⁻¹ N per season in conventional greenhouse systems cultivating Brassicaceae and Solanaceae [8].

Excess phosphorus accumulation is also a widespread issue. Large-scale assessments have shown that P inputs from mineral fertilizers and organic amendments often exceed crop removal, resulting in persistent surpluses in agricultural soils [9,10]. This long-term accumulation is environmentally relevant because surplus P can be mobilized through leaching and runoff, intensifying nutrient enrichment of water bodies and contributing to eutrophication, harmful algal blooms, and broader ecological and human health impacts [11]. In Europe alone, nitrogen pollution is estimated to cause economic losses of €70–320 billion per year [12], while in the United States, the total annual costs of eutrophication exceed USD 2.2 billion [12].

In response, international and European policy frameworks have called for substantial reductions in synthetic fertilizer use [13]. Achieving these targets, however, requires agronomic strategies capable of lowering inputs while maintaining crop yield and quality. In this context, crop rotation and plant growth-promoting microorganisms (PGPM), here intended as beneficial bacteria and fungi associated with the rhizosphere or endosphere, represent promising complementary tools [14].

Crop rotation is a well-established sustainable practice that improves soil fertility, structure, and microbial diversity [15,16]. By alternating crops with different nutrient demands, rooting patterns, and symbiotic interactions, rotations can disrupt pest and disease cycles, enhance nutrient cycling, and reduce soil degradation [17,18]. They may also improve long-term profitability by lowering external input requirements and stabilizing yields [19].

Similarly, PGPM may offer a low-input strategy to support nutrient availability and crop performance. These microorganisms, including N-fixing, phosphate- and potassium-solubilizing bacteria, as well as biocontrol agents such as Trichoderma, Pseudomonas, and Bacillus, can influence nutrient availability, phytohormone balance, and plant tolerance to abiotic and biotic stress [20,21]. Previous studies have shown that PGPM-assisted reductions in mineral fertilization can maintain yields comparable to full-fertilizer controls in tomato, rice, and potato systems [22,23,24,25] while also reducing input costs and environmental impact [26,27].

Within this broader context, the present study evaluated the effects of reduced mineral fertilization, applied either alone or in combination with microbial inoculants, within a crop rotation, on crop yield and quality, soil indicators, and economic performance, including production costs and gross margins, in a Mediterranean horticultural system.

2. Materials and Methods

2.1. Experimental Site and Crop Management

The field experiment was conducted from 2021 to 2023 at the Tomás Ferro Experimental Farm of the Universidad Politécnica de Cartagena, in southeast Spain (37°41′16.6″ N, 0°56′55.6″ W), and consisted of a four-crop rotation of potato (Solanum tuberosum var. Spunta), broccoli (Brassica oleracea var. italica Plenk), melon (Cucumis melo var. Piel de Sapo cv. F1 Paredes), and potato again. This rotation sequence was selected to reflect the crops most relevant under local commercial and agronomic conditions in Mediterranean irrigated horticulture.

The soil was classified as a Haplic Calcisol (loamy, hypercalcic) according to IUSS WRB (2022), with 25% CaCO₃ and a pH of 9. The area is characterized by a semiarid Mediterranean climate, with approximately 300 mm of annual precipitation, a mean annual temperature of 18 °C, and annual potential evapotranspiration exceeding 900 mm.

The experimental plots were kept under fallow during the year preceding the first potato crop, after three consecutive years of floricultural production. This study followed a randomized complete block design with four replicates per treatment, resulting in a total of 16 plots (4 treatments × 4 replicates). Experimental plot sizes were 740 m² for all treatments, where rotations and fertilization strategies were performed. Planting densities were 50,000 plants ha⁻¹ for potato and broccoli (100 × 20 cm spacing) and 33,333 plants ha⁻¹ for melon (200 × 150 cm spacing).

Soils were tilled to a depth of 25 cm before planting every crop. Irrigation was applied weekly via automated drip fertigation, scheduled according to climatic conditions, crop coefficients (Kc), and reference evapotranspiration (ET₀). The system ensured precise and uniform water distribution. Irrigation volumes for each cultivation cycle are detailed in Supplementary Table S1.

2.2. Fertilization Treatments and Inoculant Application

In each crop cycle, four fertilization treatments were compared:

• T1, conventional fertilization, based on the specific nutrient requirements of each crop.
• T2, reduced fertilization, with a 30% reduction relative to T1 in the first potato crop (2021) and a 50% reduction in the subsequent crops (broccoli, melon, and second potato crop, 2022–2023).
• T3, the same reduced fertilization regime as T2, combined with a commercial inoculant containing N-fixing and nutrient-solubilizing bacteria (Azospirillum, Pseudomonas, and Bacillus).
• T4, the same reduced fertilization regime as T2, combined with a commercial inoculant containing N-fixing and nutrient-solubilizing bacteria (Bacillus, Azotobacter) plus non-mycorrhizal fungi.

The −30% reduction adopted in the first potato cycle was selected as a conservative starting level to reduce inputs while limiting the risk of substantial yield penalties. After the first cycle showed comparable yields, the reduction level was increased to −50% in the subsequent cycles to test a more ambitious reduction and strengthen the contrast among fertilization strategies across the rotation.

Because inoculants were tested only under reduced fertilization, the experimental design does not allow their effects to be distinguished from those of reduced fertilization itself. Thus, T3 and T4 represent inoculated reduced-fertilization strategies rather than inoculant effects alone. This choice was aligned with the study objective (testing inoculation as a management option to support fertilizer reduction under field conditions) and with practical constraints of a multi-year field-scale rotation. Including a conventional-fertilization + inoculation treatment would have doubled the number of experimental units and operational complexity.

The microbial products, supplied by Fertilizantes y Nutrientes Ecológicos, S.L., were developed from endemic strains of plant growth-promoting microorganisms. According to the manufacturer, these strains were selected for their potential to solubilize nutrients, suppress pathogens, and persist under the semi-arid, low-rainfall conditions of the study area. The manufacturer also indicates that the use of endemic strains may reduce potential incompatibility with native soil microbial communities. The exact strain-level composition and microbial concentration (e.g., viable counts expressed as CFU per volume/weight) were requested from the manufacturer. Still, they were not provided as they were considered proprietary information. Therefore, beyond the functional description reported here, microbial concentration could not be included. Both products were commercial liquid formulations applied via fertigation at 6 L ha⁻¹ per application for T3 and 30 L ha⁻¹ per application for T4, following the schedule reported in

Table 1. Crop Cycles and Farm Management: Planting, Fertilization, Irrigation, and Pest Control.

Crop Cycle	Planting/Harvest	Fertilization Reduction in T2, T3 and T4	Inoculant Treatments (Doses, Dates)	Irrigation (m3 ha−1, Weeks)	Pesticides (Date, Product, Doses)
Potato	Planted 22 December 2020; harvested 31 May–5 June 2021 (160–164 DAS)	−30%	T3, 6 L ha−1, 2 apps (25 February, 11 March); T4, 30 L ha−1, 3 apps (25 February, 11 March, 25 March)	2400 m3 ha−1 over 23 weeks	8 February 2021: Cypermethrin 10%, 0.225 kg ha−1;13 April 2021: Mandipropamid 25% + Difenoconazole 25%, 0.175 kg ha−1
Broccoli	Transplanted 5 October 2021; harvested 5–10 January 2022 (92–97 DAT)	−50%	T3, 6 L ha−1, 2 apps (30 November, 9 December); T4, 30 L ha−1, 3 apps (30 November, 9 December, 14 December)	1100 m3 ha−1 over 14 weeks	2 November 2022: Lambda-cyhalothrin 1.5%, 1.2 kg ha−1; Azoxystrobin 25% 1.2 kg ha−1
Melon	Transplanted 29 March 2022; harvested 5–14 July 2022 (98–107 DAT)	−50%	T3, 6 L ha−1, 2 apps (13 April, 25 April); T4, 30 L ha−1, 3 apps (13 April, 25 April, 6 May)	3400 m3 ha−1 over 11 weeks	8 April 2022: Sulfur 98.5%, 25 kg ha−1; Jun: Abamectin 1.8% 24, 1.45 kg ha−1; 15 June: Boscalid 20% + Kresoxim-methyl 10%, 0.73 kg ha−1
Potato	Planted 16 December 2022; harvested 16 May 2023 (151 DAS)	−50%	T3, 6 L ha−1, 2 apps (1 March, 15 March); T4, 30 L ha−1, 3 apps (1 March, 15 March, 29 March)	2000 m3 ha−1 over 11 weeks	14 April 2023: Chlorantraniliprole 20%, 0.6 kg ha−1; Mandipropamid 25% + Difenoconazole 25%, 0.175 kg ha−1

T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi. DAS: days after seeding; DAT: days after transplanting; apps: applications. Full information details are available in the Zenodo repository: https://zenodo.org/records/18554641 (accessed on 9 February 2026).

. The formulations were produced to ensure high microbial activity at the time of application.

The product used in T3 was applied twice per crop cycle via fertigation at 15-day intervals, whereas the product used in T4 was applied three times per cycle, also via fertigation and at 15-day intervals (Table 1). Mineral fertilization was supplied weekly through fertigation (Table 1 and Supplementary Table S1). Preventive pesticide and fungicide treatments were applied according to local agricultural practice (Table 1). Crop nutrient requirements were determined using standard references for P and K [28,29] and the official nitrogen calculator of the Murcia region [30]. Weekly N, P, and K application rates are reported in Supplementary Tables S2–S5.

2.3. Weather Data Recording

Meteorological data were collected using an automatic weather station located at the experimental farm and integrated within the SIAR agrometeorological network. The station operates in accordance with the UNE 176.101:2010 standard [31], recording average or accumulated values, as well as maximum and minimum values, of air temperature, relative humidity, and precipitation.

2.4. Soil Sampling and Analysis

Bulk soil samples (0–25 cm) were first collected on 22 April 2020 to characterize baseline soil conditions before the start of the field trial (

Table 2. Soil properties from the initial sampling of the 2020 experiment. Values mean ± standard error (n = 4).

		T1	T2	T3	T4
pH		9.46 ± 0.05	9.43 ± 0.03	9.38 ± 0.04	9.35 ± 0.03
EC	dS m−1	0.14 ± 0	0.14 ± 0	0.15 ± 0.01	0.13 ± 0.02
TOC	g kg−1	10.46 ± 0.88	11.04 ± 0.57	11.07 ± 0.29	10.14 ± 0.24
POC	g kg−1	2 ± 0.18	2.03 ± 0.06	2.42 ± 0.22	1.65 ± 0.17
Nt	g kg−1	0.86 ± 0.07	0.91 ± 0.05	0.93 ± 0.04	0.88 ± 0.02
CEC	cmol kg−1	17.87 ± 0.57	18.61 ± 0.71	17.42 ± 0.43	19.05 ± 0.43
Caex	cmol kg−1	11.76 ± 0.45	12.08 ± 0.47	11.31 ± 0.33	12.41 ± 0.24
Mgex	cmol kg−1	4.66 ± 0.11	4.88 ± 0.18	4.66 ± 0.08	4.96 ± 0.11
Kex	cmol kg−1	0.87 ± 0.02	0.96 ± 0.05	0.9 ± 0.04	0.95 ± 0.03
Naex	cmol kg−1	0.57 ± 0.03	0.69 ± 0.07	0.56 ± 0.03	0.74 ± 0.07
Bs	cmol kg−1	17.87 ± 0.57	18.61 ± 0.71	17.42 ± 0.43	19.05 ± 0.43
CaCO3	%	26.27 ± 0.49	26.12 ± 0.63	25.92 ± 0.36	26.18 ± 0.6
NH4+	mg kg−1	13.45 ± 0.21	14.89 ± 0.76	14.33 ± 0.71	15.69 ± 1.08
NO3−	mg kg−1	2.76 ± 0.43	3.06 ± 0.62	2.6 ± 0.23	3.5 ± 0.17
Pav	mg kg−1	37.33 ± 0.34	38.58 ± 0.66	37.75 ± 1.6	41.12 ± 2.16
Feba	mg kg−1	98.62 ± 4.02	93.08 ± 6.19	98.78 ± 3.82	92.03 ± 4.02
Cuba	mg kg−1	3.12 ± 0.12	3.04 ± 0.08	3.15 ± 0.33	2.94 ± 0.08
Znba	mg kg−1	2.66 ± 0.08	2.57 ± 0.1	2.53 ± 0.12	2.66 ± 0.09
Mnba	mg kg−1	7.6 ± 0.49	6.81 ± 0.22	6.82 ± 0.2	7.09 ± 0.27
Moba	mg kg−1	5.9 ± 0.45	5.06 ± 0.2	5.28 ± 0.29	5.59 ± 0.2
Bba	mg kg−1	0.02 ± 0	0.02 ± 0	0.02 ± 0	0.02 ± 0

EC (electrical conductivity), TOC (total organic carbon), POC (particulate organic carbon), Ca_ex (exchangeable calcium), Mg_ex (exchangeable magnesium), K_ex (exchangeable potassium), Na_ex (exchangeable sodium), Bs (sum of bases), CaCO₃ (carbonate), Nt (total nitrogen), CEC (cation exchange capacity), P_av (available phosphorus), Fe_ba (bioavailable iron), Cu_ba (bioavailable copper), Zn_ba (bioavailable zinc), Mn_ba (bioavailable manganese), Mo_ba (bioavailable molybdenum), B_ba (bioavailable boron). T1, conventional mineral fertilization; T2, reduced mineral fertilization; T3, reduced mineral fertilization plus bacterial inoculant; T4, reduced mineral fertilization plus bacterial and fungal inoculant. Full analytical details are available in the Zenodo repository: https://zenodo.org/records/18554641 (accessed on 9 February 2026).

). In subsequent years, soil sampling was consistently conducted between April and May to ensure comparable climatic conditions. The sampling dates for potato (13 May 2021) and melon (15 May 2022) were chosen to coincide with the flowering stage of each crop. Broccoli was not sampled as a separate soil campaign because this study aimed to compare soil indicators under comparable seasonal conditions across years. To reduce confounding due to seasonal variability, sampling was kept within a fixed April–May window and aligned with the flowering stage of the crop present at that time. At each campaign, soil was collected from the same experimental plots (treatment-replicate units), enabling repeated measurements over time. The final sampling was carried out during potato flowering on 14 April 2023.

Sampling was performed in the planting rows, between two plants. Four composite samples were obtained per plot. Each composite sample consisted of five subsamples randomly collected within the plot using a hand auger. Soil samples were air-dried and sieved to <2 mm before analysis [32].

Soil pH and electrical conductivity (EC) were measured in a 1:5 soil-to-water ratio (w/v). CaCO₃ content was determined by calcination at 450 °C for 4 h [33]. Total nitrogen (TN), total organic carbon (TOC), and particulate organic carbon (POC) were measured using a CHN elemental analyzer (EA-1108, Carlo Erba, Barcelona, Spain). Exchangeable cations and cation exchange capacity (CEC) were determined with BaCl₂ according to ISO 13536 (Quality, 1995) [34]. Ammonium (NH₄⁺) was extracted with 2M KCl (1:10 ratio) and quantified spectrophotometrically [35,36], while nitrate (NO₃⁻) was extracted with deionized water (1:10) and analyzed via ion chromatography (Metrohm 861) [36,37]. Available Fe, Mn, Cu, and Zn were extracted using DTPA [38]. Available phosphorus (P) was assessed using the Olsen method [39], and the concentrations of all extractable nutrients were determined by inductively coupled plasma mass spectrometry (ICP-MS, 7500CE, Agilent Technologies, Santa Clara, CA, USA). Available boron was extracted using an adapted method from Nable et al. (1997) [40] and quantified by AAS. Full methodological details are available in the Zenodo repository: https://zenodo.org/records/18554641 (accessed on 9 February 2026).

2.5. Total Yield and Quality Evaluations

At the time of harvesting, the yield was analyzed across the entire experimental plot. Potato tubers were collected and weighed, corresponding to the total yield. The quality of the tubers (firmness, density, % starch, weight, and size) was analysed in a sample of 10 potatoes for each plot. The firmness analysis was carried out employing a penetrometer to determine the flesh firmness [41]. Tuber weight, dimensions (V), and density (ρ) were determined with a laboratory balance and a Vernier caliper according to Glasbey et al. (1988) [42]. The starch content was determined by indirect methods, calculated through specific weight [43]. Broccoli heads were harvested progressively as they reached market maturity. Collections were carried out from defined areas within each plot of every treatment, for a total sampled area of 198 m². The number and total weight of inflorescences were recorded to determine yield, and head circumference and stem diameter were measured. Melons were harvested from defined areas within each plot, for a total sampled area of 540 m². Fruits were collected progressively as they reached optimal maturity stage, characterized by a transition in rind color from green to yellowish-green and a slight yielding of the flesh to gentle pressure. The harvested melons were counted and weighed to determine yield. A subsample of 40 individual melons per plot was then selected for quality assessment, specifically measuring ºBrix using a refractometer (model Pocket PAL-1, Atago; Tokyo, Japan).

2.6. Economic Analysis

The economic assessment combined partial budgeting analysis with data obtained from the field experiment. Partial crop budgeting evaluates how changes in the production process affect profitability, such as those arising from modified management practices or the adoption of new technologies. It is one of the most widely used approaches for assessing the economic performance of alternative agricultural practices and technologies [44], and it is frequently supported by experimental field data [45,46,47,48,49].

This approach makes it possible to construct detailed budgets that incorporate experimentally observed changes in both inputs and outputs and to compare them across treatments. First, a detailed technical characterization of the production process of each crop under each experimental treatment was developed using information collected during the field trial, including crop yield, input use, machinery use, and labour requirements. These data were subsequently validated with local farmers and technical advisors.

Second, economic data, including crop sale prices, input prices, and labour costs, were collected from input suppliers, farmers, technical advisors, and official statistical sources. Fertilizer and other chemical input prices were obtained from commercial suppliers, excluding VAT. Electricity costs were calculated as the three-year average price paid under the 3.0TD tariff, the most common electricity tariff among agricultural enterprises, using data from the Spanish electricity system operator. Fuel costs were estimated from the average agricultural diesel price over the previous three years, based on official national data. Irrigation water costs were calculated from the average tariff applied by the irrigation district where the experimental plots were located, which showed little variation over time. Unit labour costs were based on the most recent regional collective labour agreement for the agricultural sector. Market prices for the crops included in the analysis were estimated as detrended average annual prices over the 2014–2023 period using official agricultural price databases from the Murcia Region.

Third, the technical and economic information was integrated to define specific cost structures and budgets for each crop cycle and experimental treatment. The cost structure was established following the crop production cost assessment methodology of the Spanish Ministry of Agriculture [50], in line with the framework of the European Farm Accountancy Data Network. Because the tested management practices were not expected to affect indirect production costs, profitability was assessed using gross margin, defined as the difference between revenue (production value plus subsidies) and the variable costs directly attributable to each crop [51]. Crop revenue and gross margin were calculated for each plot replicate, treatment, and crop, whereas production costs were calculated at the treatment level.

2.7. Statistical Analysis

The normality of data distribution for each measured variable was assessed using the Kolmogorov–Smirnov test, and homogeneity of variances was evaluated using Levene’s test. Treatment effects on soil physicochemical parameters, crop yield, and quality traits were analysed by one-way analysis of variance (ANOVA). When significant treatment effects were detected, means were separated using Duncan’s multiple range test (DMRT) at the 5% significance level. DMRT was selected as a post hoc test because it is widely used in agricultural studies for the interpretation of treatment-related differences. For variables that did not meet the assumption of normality, the Kruskal–Wallis test was applied, followed by Dunn’s test for pairwise comparisons. The same statistical approach was used for plot-level economic variables, including revenue and gross margin. In contrast, direct costs and the aggregated economic results of the complete four-cycle crop rotation were calculated at the treatment level and were therefore evaluated descriptively rather than statistically. Relationships among soil chemical variables were assessed using Spearman’s rank correlation.

Because the experiment included four crop cycles over three years (potato 2021; broccoli 2022; melon 2022; potato 2023), treatment comparisons for yield and crop-quality variables were performed within each crop cycle. Yields were not statistically compared across different crop species and seasons. For soil variables, temporal changes in the main physicochemical properties were analysed using linear mixed models, with sampling year, treatment, and their interaction as fixed effects, and plot as a random effect. When significant effects were detected, pairwise comparisons were performed using estimated marginal means. In addition, one-way ANOVA followed by DMRT was applied within each sampling year.

To assess temporal changes in the measured soil properties during the crop rotation, relative variation was calculated for each variable between the initial sampling of the experimental phase (13 May 2021) and the final sampling (12 April 2023), following the equation of Soto-Gómez et al. (2025) [52]:

Characteristic evolution with time Δ (%) = [(Final value − Initial value)/Initial value] × 100

A principal component analysis (PCA) was then used as an exploratory tool to summarize multivariate co-variation among Δ soil indicators. Statistical analyses were performed using IBM SPSS Statistics, version 24, and Stata, version 15. Figures were generated in R version 4.4.2 using the packages ggplot2 and corrplot within RStudio 2026.01.1+403 (Posit Software, PBC, Boston, MA, USA).

3. Results

3.1. Weather Conditions During the Trial

The first potato growing season (December 2020–May 2021) was characterized by a mean temperature of 14 °C, total precipitation of 309 mm, and a mean monthly ET₀ of 70 mm. In the subsequent cropping cycles, the broccoli season (October 2021–January 2022) and the melon season (March 2022–July 2022) recorded mean temperatures of 14 °C and 19 °C, respectively. For broccoli, total precipitation was 70.8 mm and the mean monthly ET₀ was 38.5 mm, whereas the melon season received 184.4 mm of rainfall and had a mean monthly ET₀ of 88.4 mm. Finally, during the second potato-growing season (December 2022–May 2023), the mean temperature was 19.5 °C, total precipitation reached 103.8 mm, and the mean monthly ET₀ was 72.3 mm.

3.2. Total Yield and Agronomical Evaluation

No significant differences in crop yield were detected among fertilization treatments in any crop (Figure 1). Potato yield in 2021 ranged from 55,661 ± 2117 to 60,741 ± 882 kg ha⁻¹, while broccoli yield in 2022 varied between 14,928 ± 627 and 16,797 ± 460 kg ha⁻¹. In the 2022 melon crop, yields ranged from 30,815 ± 1553 to 33,423 ± 1486 kg ha⁻¹, with microbial inoculant treatments under reduced fertilization (T3–T4) showing a slight, non-significant increase compared with T2. A similar trend was observed in the final potato crop (2023), where T4 showed the highest yield (39,146 ± 416 kg ha⁻¹).

Regarding tuber quality, in 2021, the T3 treatment showed a significantly lower tuber weight compared to the other treatments, and a smaller size than T2 and T4. However, tuber firmness in T3 was significantly higher than in T1 and T4. In contrast, in 2023, T4 produced tubers with significantly greater weight than T1 and T3, while T1 showed significantly lower firmness than the other treatments. No significant differences in tuber starch content were observed among any treatments in any season (

Table 3. Quality assessment of potato in cropping seasons 2021 and 2023.

Potato		T1	T2	T3	T4	F ANOVA
Weight (g)	2021	181.0 ± 28.3 a	206.8 ± 19.8 a	137.3 ± 11.0 b	186.1 ± 58.0 a	2.92
Estimated tuber volume (cm3)		161.1 ± 27.1 ab	184.4 ± 21.6 a	119.1 ± 15.4 b	175.9 ± 56.3 a	2.91
Firmness (kg cm−2)		14.6 ± 0.4 c	15.4 ± 0.7 ab	15.5 ± 0.3 a	14.8 ± 0.3 bc	4.39
Starch content (%)		23.3 ± 6.1	24.4 ± 1.0	23.9 ± 2.5	22.5 ± 4.8	0.18 ns
Weight (g)	2023	158.5 ± 6.7 b	205.8 ± 17.4 ab	171.3 ± 4.5 b	244.3 ± 24.4 a	6.14
Estimated tuber volume (cm3)		143.0 ± 12.8	179.0 ± 20.8	162.8 ± 32.8	224.5 ± 39.8	1.48 ns
Firmness (kg cm−2)		13.2 ± 0.1 b	14.8 ± 0.4 a	15.5 ± 0.4 a	16.0 ± 0.3 a	14.23
Starch content (%)		21.0 ± 1.1	20.3 ± 0.6	20.9 ± 7.9	21.4 ± 1.9	0.05 ns

Different letters indicate significant differences among treatments within the same year based on DMRT (α < 0.05). ns, no significant differences. T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi.

).

The harvested broccoli showed substantial uniformity, with no significant differences observed between treatments for the quality trait evaluated (

Table 4. Agronomic evaluation in broccoli and melon 2022.

	T1	T2	T3	T4	F ANOVA
Broccoli
Number of heads m−2	5.0 ± 0.1	4.8 ± 0.5	4.7 ± 0.2	4.8 ± 0.2	0.33 ns
Weight (g head−1)	322.4 ± 5.4	316.5 ± 12.2	334.3 ± 12.1	309.8 ± 3.6	0.59 ns
Head circumference (cm)	40.9 ± 11.0	41.6 ± 1.2	41.1 ± 1.4	41.3 ± 0.8	0.67 ns
Stem diameter (cm)	3.5 ± 0.2	3.8 ± 0.3	3.9 ± 0.2	3.7 ± 0.2	0.44 ns
Melon
Number of fruits m−2	0.9 ± 0.4	0.9 ± 0.43	0.9 ± 0.43	0.9 ± 0.43	0.17 ns
Weight (kg fruit−1)	3.7 ± 0.1	3.6 ± 0.22	3.8 ± 0.1	3.7 ± 0.1	0.75 ns
ºBrix	13.5 ± 0.3 a	13.0 ± 0.3 ab	13.0 ± 0.23 ab	12.1 ± 0.66 b	3.4

Different letters indicate significant differences among treatments within the same year based on DMRT (α < 0.05). ns, no significant differences. T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi.

).

Melon showed no significant differences in harvested fruit weight among treatments (Table 4). However, a notable trend emerged regarding fruit sugar content, where melons treated with microbial inoculants generally displayed lower sugar levels than the untreated melons. Specifically, the T4 treatment resulted in a significantly lower value compared to T1 (Table 4).

3.3. Evolution of Soil Physicochemical Properties Throughout the Rotation

The changes in soil physicochemical properties (Δdata) between the beginning (2021) and the end (2023) of the experiment indicated generally modest treatment effects, as shown in

Table 5. Percentage change (Δ) in general physicochemical soil properties between the beginning and the end of the cultivation cycle for the different treatments. Values mean ± standard error (n = 4).

Treatment	ΔpH	ΔEC	ΔTOC	ΔPOC	ΔCaCO3
T1	0.19 ± 0.07 a	32.71 ± 2.97 b	25.49 ± 18.13 a	−36.07 ± 4.33 a	−9.17 ± 14.45 a
T2	−0.69 ± 0.36 b	57.63 ± 11.96 ab	6.4 ± 2.09 a	−45.2 ± 9.93 a	−4.36 ± 8.77 a
T3	−0.75 ± 0.24 b	64.94 ± 11.12 a	4.89 ± 9.88 a	−39.49 ± 8.36 a	−3.69 ± 4.91 a
T4	−0.14 ± 0.31 ab	57.64 ± 7.37 ab	7.68 ± 3.49 a	−35.23 ± 4.32 a	−10.75 ± 5.55 a
P ANOVA	0.080 ns	0.117 ns	0.497 ns	0.757 ns	0.932 ns

ΔEC (electrical conductivity), ΔTOC (total organic carbon), ΔPOC (particulate organic carbon). Different letters indicate significant differences among the treatments (DMRT, α < 0.05). ns, no significant differences. T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi.

for general physicochemical properties,

Table 6. Percentage change (Δ) in nutrient content and exchangeable cations in soil from the start to the end of the cultivation cycle for the different treatments. Values mean ± standard error (n = 4).

Treatment	ΔNt	ΔNH4+	ΔNO3−	ΔCEC	ΔCaex	ΔMgex	ΔKex	ΔNaex
T1	17.87 ± 2.52 a	79.81 ± 33.3 b	125.82 ± 56.52 a	29.31 ± 2.47 a	22 ± 0.88 a	105.09 ± 4.28 a	−22.09 ± 12.21 a	−2.8 ± 3.7 a
T2	14.96 ± 4.16 a	85.68 ± 4.41 b	189.68 ± 60.49 a	23.34 ± 2.04 ab	18.21 ± 2.32 ab	98.7 ± 4.6 a	−44.87 ± 2.48 a	14.09 ± 7.58 a
T3	10.64 ± 1.04 a	81.36 ± 10.78 b	376.48 ± 110.68 a	26 ± 2.24 ab	20.32 ± 1.06 a	103.58 ± 3.72 a	−38.07 ± 5.99 a	4.6 ± 6.35 a
T4	17.55 ± 1.86 a	216.98 ± 50.78 a	497.94 ± 391.24 a	21.24 ± 2.86 b	14.15 ± 2.67 b	91.92 ± 4.62 a	−35.88 ± 4.13 a	−1.68 ± 3.89 a
P ANOVA	0.244 ns	0.021	0.691 ns	0.158 ns	0.063 ns	0.186 ns	0.209 ns	0.184 ns

ΔNt (total nitrogen), ΔCEC (cation exchange capacity), ΔCa_ex (exchangeable calcium), ΔMg_ex (exchangeable magnesium), ΔK_ex (exchangeable potassium), ΔNa_ex (exchangeable sodium). Different letters indicate significant differences between different treatments (DMRT, α < 0.05). ns, no significant differences. T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi.

for nutrient content and exchangeable cations, and

Table 7. Percentage change (Δ) in bioavailable micronutrients and phosphorus in soil between the beginning and end of the cultivation cycle for the different treatments. Values mean ± standard error (n = 4).

Treatment	ΔPav	ΔFeba	ΔCuba	ΔZnba	ΔMnba	ΔMoba	ΔBba
T1	82.25 ± 6.22 a	−66.43 ± 1.68 a	−70.23 ± 1.17 a	−75.15 ± 1.3 a	−67.29 ± 2.27 a	−59.89 ± 20.07 b	−4.3 ± 8.56 ab
T2	62.8 ± 1.31 b	−61.76 ± 3.67 a	−71.03 ± 1.61 a	−77.05 ± 0.85 a	−59.81 ± 3.06 a	−72.22 ± 2.78 b	9.06 ± 5.65 a
T3	57.25 ± 4.96 b	−61.96 ± 3.89 a	−71.56 ± 2.3 a	−76.87 ± 1.91 a	−60.66 ± 4.59 a	35.71 ± 41.85 a	−4.86 ± 6.92 ab
T4	46.99 ± 8.13 b	−64.67 ± 4.04 a	−69.7 ± 2.33 a	−72.92 ± 2.87 a	−67.43 ± 4.42 a	−39.34 ± 14.34 b	−23.86 ± 10.69 b
P ANOVA	0.007	0.739 ns	0.903 ns	0.414 ns	0.339 ns	0.040	0.089 ns

ΔP_av (available phosphorus), ΔFe_ba (bioavailable iron), ΔCu_ba (bioavailable copper), ΔZn_ba (bioavailable zinc), ΔMn_ba (bioavailable manganese), ΔMo_ba (bioavailable molybdenum), ΔB_ba (bioavailable boron). Different letters indicate significant differences between different treatments (DMRT, α < 0.05). ns, no significant differences. T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi.

for bioavailable micronutrients and available phosphorus.

Statistically significant differences among treatments were detected only for ΔNH₄⁺, ΔPav, and ΔMoba. In particular, ΔNH₄⁺ increased most markedly in T4 (+217%), whereas available phosphorus (ΔPav) showed the greatest increase in T1 (+82%). Bioavailable Mo (ΔMoba) also differed significantly among treatments, with T3 being the only treatment showing a positive change (+36%), while the other treatments showed declines over time.

For the remaining variables, no significant treatment effects were detected, although some treatment-related patterns were apparent. Soil pH tended to decrease more in T2 and T3, whereas a slight increase was observed in T1. Similarly, ΔCaex tended to be higher in T1 and T3, while ΔBba showed considerable variability among treatments, with T2 showing a slight increase and T4 the greatest decrease. By contrast, ΔEC, ΔTOC, ΔPOC, ΔNt, ΔNO₃⁻, ΔCEC, ΔMgex, ΔKex, ΔNaex, and ΔCaCO₃ did not differ significantly among treatments. Overall, nitrogen-related variables, cation exchange capacity, and exchangeable Ca and Mg generally increased over time, whereas particulate organic carbon and most micronutrients tended to decline across treatments.

Spearman’s correlation analysis (Figure 2) revealed clear relationships among soil chemical and nutrient variables. The strongest positive correlations were observed among exchangeable base cations and cation exchange capacity, particularly between CEC and Caex (ρ = 0.98, p < 0.001), CEC and Mgex (ρ = 0.95, p < 0.001), and Caex and Mgex (ρ = 0.93, p < 0.001). Nutrient availability variables were also positively associated with this cluster, with significant correlations between CEC and NO₃⁻ (ρ = 0.66, p < 0.001), CEC and Pav (ρ = 0.63, p < 0.001), and Mgex and Pav (ρ = 0.68, p < 0.001). Electrical conductivity showed a strong positive relationship with total nitrogen (Nt; ρ = 0.70, p < 0.001), indicating a link between soil salinity and nitrogen availability. In contrast, several negative correlations were observed, particularly between pH and EC (ρ = −0.70, p < 0.001), POC and Nt (ρ = −0.61, p < 0.001), and Kex and Pav (ρ = −0.59, p < 0.001), suggesting contrasting patterns between carbon fractions, potassium dynamics, and nutrient availability.

Temporal variation was the main source of change in soil physicochemical properties during the 2021–2023 period. Repeated-measures analysis showed a significant effect of sampling year on pH, EC, Nt, CEC, NH₄⁺, NO₃⁻, and available P (Pav), whereas CaCO₃ did not vary significantly over time. Treatment effects were generally limited and were significant only for NH₄⁺, for which a significant treatment × year interaction was also detected (Figure S1).

Soil pH showed only minor temporal variation, with significant differences across years observed only in T2 and T3, where values were higher in 2023 than in 2022. Electrical conductivity decreased significantly from 2021 to 2022 in all treatments and remained lower in 2023 than in 2021. Total N also declined over time, particularly in 2023, although the pattern differed slightly among treatments. CEC showed a consistent decrease from 2021 to the subsequent sampling years across all treatments.

Among all variables, NH₄⁺ showed the clearest treatment-related response. Concentrations were low and similar across treatments in 2021, increased markedly in 2022, and declined again in 2023. However, in the final sampling year, T4 maintained the highest NH₄⁺ concentration (4.45 mg kg⁻¹), significantly exceeding T1, T2, and T3. NO₃⁻ showed a significant overall year effect, although pairwise differences within treatments were limited. Available P declined markedly over time in all treatments, with the strongest decreases generally observed between 2021 and 2023.

The PCA analysis showed a separation among treatments based on the percentage changes (Δ) in soil physicochemical properties during the 2021–2023 period (Figure 3). T1 was clearly separated from the other treatments and was associated mainly with higher ΔCEC, ΔCa_ex, ΔMg_ex, ΔTOC, and ΔP_av. In contrast, T3 and T4 clustered more closely together and were associated with ΔEC, ΔNO₃⁻, ΔNH₄⁺, and ΔMoba. Among the variables contributing most to treatment separation, ΔMg_ex, ΔCa_ex, ΔCEC, and ΔMn_ba showed the highest positive loadings on PC1 (0.888, 0.835, 0.808, and 0.856, respectively), whereas ΔP_av and ΔpH showed the highest positive loadings on PC2 (0.756 and 0.708, respectively). By contrast, ΔEC showed a strong negative loading on PC2 (−0.714).

3.4. Economic Assessment

Table 8. Revenue, direct costs, and gross margin by crop cycle and experimental treatment.

Crop Cycle	Experimental Treatment	Average Revenue (€ ha−1)	Direct Costs (€ ha−1)	Average Gross Margin (€ ha−1)
I (potato 2021)	T1	27,005	7481	19,524
	T2	28,548	7281	21,268
	T3	26,161	7292	18,869
	T4	27,907	7394	20,514
	F ANOVA	0.96 ns	-	0.99 ns
II (broccoli 2022)	T1	10,078	6778	3300
	T2	9373	6056	3317
	T3	8957	6148	2809
	T4	9049	6136	2913
	F ANOVA	0.97 ns	-	0.26 ns
III (melon 2022)	T1	12,891	9280	3611
	T2	12,068	8372	3696
	T3	13,082	8633	4449
	T4	12,767	8572	4195
	F ANOVA	0.68 ns	-	0.56 ns
IV (potato 2023)	T1	17,331	8109	9222
	T2	17,179	7172	10,007
	T3	17,237	7462	9775
	T4	18,399	7446	10,953
	F ANOVA	0.85 ns	-	1.51 ns
Full crop rotation	T1	67,306	31,648	35,658
	T2	67,168	28,880	38,288
	T3	65,436	29,535	35,901
	T4	68,122	29,547	38,575
	F ANOVA	0.53 ns	-	0.99 ns

T1 inorganic fertilizer applied at crop nutritional requirements; T2, reduced T1 fertilization by 30% (potato 2021) or 50%; T3, 30% or 50% fertilization reduction plus a microbial inoculant containing Azospirillum, Pseudomonas, and Bacillus; T4, 30% or 50% fertilization reduction plus a microbial inoculant containing Bacillus, Azotobacter, and non-mycorrhizal fungi. ns, no significant differences. Production costs were calculated for each experimental treatment, but not for each repetition, and are therefore not subject to statistical analysis.

shows the average values of crop revenue and gross margin and the calculated direct production costs for each crop cycle and for the full four-cycle crop rotation by experimental treatment. Production costs were calculated by experimental treatment but not per repetition. Firstly, differences in crop revenue for the same crop cycle were explained by differences in crop yield for each experimental treatment. The results from Duncan’s multiple range test showed that there were no significant differences in crop revenue by experimental treatment, nor for any of the crop cycles, or for the whole crop rotation. This is explained by the relatively large variability in the observed yields for each repetition and experimental treatment.

Secondly, observable differences in direct production costs among experimental treatments (Table 8) resulted from the differences in the application of fertilizers and microbial inoculants in each treatment, which also implied some differences in the use of labour, irrigation water, and electricity. Production costs were lower for treatments where fertilization is reduced (T2, T4, and T3, respectively), with respect to the control treatment (T1). On the other hand, Table 8 also shows very similar cost increases with respect to T2 for T4 and T3, where microbial inoculants are also applied.

Lastly, the observed differences in the average gross margin (Table 8) suggest that reducing fertilization increased gross margin in all crops. However, the differences are minimal in some cases (broccoli and melon) and greater in others (potato). The application of T4 reduced the average gross margin in the first two crop cycles with respect to the T2 treatment (reduced fertilization only) but increased average gross margin in the third and fourth crop cycles. On the contrary, the application of T3 resulted in a reduced average profitability with respect to the application of T4 in the potato and broccoli cycles and for the whole crop rotation. However, these average values hide a significant variability in the calculated gross margin for some treatments and crop cycles, including the full crop rotation, as illustrated in Figure 4. In fact, there were no significant differences in gross margin by experimental treatment for any of the crop cycles, nor for the whole crop rotation, as shown in Table 8.

4. Discussion

4.1. Quality Parameters Are More Responsive than Yield to Reduced Fertilization and Microbial Inoculation

In the potato–broccoli–melon–potato rotation (2021–2023), reduced fertilization did not negatively affect crop yields, and microbial inoculation did not result in significant yield differences compared with the non-inoculated treatments. In the melon and final potato cycles, mean yields in the reduced-fertilization + inoculant treatments (T3 and T4) were numerically higher than those observed under both reduced fertilization alone (T2) and full-rate fertilization (T1), although these differences were not statistically significant. This pattern is consistent with previous reports indicating that consortia of phosphate- and potassium-solubilizing bacteria can enhance nutrient uptake efficiency, particularly under reduced fertilization [23,53,54], and may maintain or increase potato yields under reduced NPK inputs [55]. It may also be considered in the context of gradual changes in soil nutrient availability and fertility dynamics, which generally require long-term management to become evident [56,57].

Interannual variability in potato yield likely reflects seasonal environmental conditions, including temperature, precipitation, and radiation, which are known to influence crop growth and productivity [24,58,59]. In particular, the lower yields observed in 2023 likely reflect the substantially warmer and drier conditions compared with 2021. Although irrigation was managed according to crop requirements, it is possible that water supply was applied too conservatively under those conditions. A moderate Rhizoctonia infection may also have contributed to the yield reduction. Microbial inoculant treatments also affected crop quality, although responses were crop-specific. In potato, T4 was associated with higher tuber firmness and greater average tuber weight, indicating a potential improvement in tuber quality. Although previous studies have suggested that microbial inoculants may affect nutrient uptake and related quality traits [53,60,61], the physiological and nutritional mechanisms underlying these responses were not assessed in the present study, and this interpretation should therefore be considered only a plausible hypothesis. Broccoli quality was generally less responsive to treatments. One possible explanation is that crop-specific rhizosphere conditions, potentially including compounds released by Brassicaceae, may have limited inoculant establishment or activity [62,63]. However, beneficial effects of biofertilizers have also been reported in Brassicaceae, including improved nutrient uptake and growth in broccoli [64], as well as higher yield and quality-related traits in other species such as kale [65]. These findings suggest that the response to PGPMs within this crop family may be species-, genotype-, and context-dependent, rather than uniformly positive. In contrast, melon soluble solids showed small differences among treatments, particularly in sugar content. Although all melons exceeded the minimum °Brix threshold of 8 (Commission Implementing Regulation (EU) No 543/2011), fruits from microbial treatments exhibited slightly lower sugar concentrations, likely due to increased fruit weight and associated water accumulation, causing sugar dilution [66]. Variability in melon response to microbial inoculants is also cultivar-dependent [67]. In Cucurbitaceae, PGPMs have been associated with improved nutrient uptake, soil fertility, stress tolerance, and plant vigor under different environmental conditions [68,69,70], although effects on fruit quality are not always consistent across crops and management contexts. These results contrast with studies reporting increased °Brix following microbial inoculation in melons [71], tomatoes [72,73], peppers [74], and beans [75], highlighting the complex interplay between plant genotype, microbial consortia, and environmental conditions.

4.2. Reduced Fertilization with Microbial Inoculants Was Associated with Changes in Soil Indicators Related to N Dynamics

An overarching contextual factor in this study is the strongly alkaline, calcareous soil (pH ≈ 9; high CaCO₃). At high pH, the availability of several nutrients, particularly P and micronutrients such as Fe, Zn and Mn, can be constrained by precipitation/sorption reactions and reduced solubility, and plant uptake responses may differ from those in neutral or acidic soils [76]. Accordingly, the observed soil-chemical responses were likely shaped by this specific pedological context. In this framework, soil pH decreased in all reduced-fertilization treatments (T2–T4), whereas it significantly increased in T1. These differences should be interpreted cautiously, as bulk soil pH is expected to be highly buffered. The decreases observed under T2–T4 may reflect the combined effect of lower mineral fertilizer inputs and rhizosphere-related processes, including the possible release of organic acids [77,78,79], although these mechanisms were not directly assessed in the present study. As previously reported, simultaneous biochemical and geochemical processes may obscure the net effect of microbial P turnover, meaning that increased microbial activity is not necessarily reflected in higher measurable soil-available P [80]. By the same logic, the observed responses may also be consistent with differences in organic matter mineralization and N cycling, including NH₄⁺ and NO₃⁻ availability, although these mechanisms were not directly assessed here and should therefore be regarded only as plausible hypotheses [81,82,83].

The conventional treatment (T1) clustered distinctly, being associated with increases in CEC, exchangeable Ca²⁺ and Mg²⁺, and TOC, in line with other studies [84]. This outcome may be related to the rapid hydrolysis and dissolution of inorganic mineral fertilizers, which release ions potentially favoring organo-mineral interactions and contributing to changes in TOC and CEC [85,86].

The closer association of T3 and T4 with ΔNO₃⁻, ΔNH₄⁺, ΔEC, and micronutrient changes may be consistent with differences in nutrient dynamics under the combined reduced-fertilization + inoculant strategy. In other systems, similar responses have been reported following PGPM application, where inoculation was associated with changes in mineral N availability, enzymatic activity, and ionic fluxes in the rhizosphere, sometimes resulting in temporary increases in EC and mineral N pools [21,79,87]. In particular, the significant increase in NH₄⁺ observed in T4 may indicate a stimulation of microbial N turnover under the combined bacterial–fungal inoculation. Previous studies have shown that PGPM inoculation can influence the abundance or expression of functional genes involved in N cycling, including nirK, nifH, ureC, and amoA [88,89,90]. In the same potato field experiment (2023 season), combined microbial inoculation was previously associated with increased abundance of key N-cycling genes, particularly nifH and nirK, in the rhizosphere after inoculation, suggesting an enhancement in microbial functional potential under reduced mineral fertilization [91]. Although those results were obtained in the rhizosphere rather than in bulk soil, they provide contextual support for the hypothesis that T4 may have influenced N transformation processes under field conditions. In this context, the higher NH₄⁺ content in T4 may be consistent with enhanced NH₄⁺ producing processes, such as organic N mineralization, ammonification, and urea hydrolysis, as reported in other biofertilized systems [92,93,94,95]. Likewise, the pronounced reduction in the C/N ratio may be compatible with a more intense mineralization pattern, since a declining C/N ratio has been associated with enhanced ammonification and greater NH₄⁺ release [96]. However, these mechanisms remain speculative in the present study because functional genes and process rates were not directly measured in bulk soil, and the incomplete characterization of the inoculants further limits mechanistic interpretation and reproducibility. Under this interpretative framework, a faster nutrient turnover could also be compatible with a greater depletion of labile carbon pools, in agreement with the positive correlation observed between POC and total N. This pattern may be consistent with the lack of TOC increase in T3–T4, unlike in T1, where lower mineralization intensity may have favored greater accumulation of soil organic C. At the same time, the decline in POC across treatments may also reflect the effect of conventional tillage, which can disrupt soil aggregates and expose labile carbon to oxidation, thereby enhancing POC mineralization [97,98,99]. Similarly, the overall decline in micronutrients may be related to crop uptake and land management, as conventional tillage is known to reduce micronutrient retention relative to no-till systems [100,101]. Finally, the marked accumulation of P_av in T1 may reflect sustained high P inputs rather than improved plant uptake efficiency, consistent with evidence that excessive P fertilization does not improve potato yield and may increase economic and environmental risks [2].

Overall, both univariate results and PCA indicated contrasting soil response patterns among treatments. Conventional fertilization (T1) was mainly associated with higher values of selected chemical indicators, particularly CEC, exchangeable Ca²⁺ and Mg²⁺, TOC, and available P, whereas reduced fertilization combined with inoculants was associated with a different set of variables, especially mineral N forms, EC, and some micronutrient changes [102,103]. These findings suggest that the different fertilization strategies shaped soil chemical status in different ways, although the magnitude and direction of these responses likely depend on soil type and management conditions [104].

4.3. Profitability Is Maintained Under Reduced Fertilization and Microbial Inoculation

The application of reduced fertilization combined with a bacterial–fungal inoculant (T4) was associated with a lower average gross margin than that of T2 (reduced fertilization only) during the first two crop cycles, whereas average profitability was higher from the second year onward. Under our experimental conditions, this pattern may suggest that any economic effect of microbial inoculants, if present, was not immediate and may have emerged only from the third crop in the rotation onward. However, gross margin did not differ significantly among the four treatments in any of the four crop cycles or across the entire rotation. Therefore, although reduced fertilization led to a decrease in direct production costs (approximately 9%), our results do not support either an increase or a decrease in profitability. As expected, reducing fertilizer use lowers production costs and may positively affect crop profitability, provided that yields are not adversely affected [105]. In addition, the resulting cost savings may compensate for slight yield reductions, thereby helping to preserve overall profitability [106]. Similarly, the use of microbial inoculants increased costs without increasing revenues, but this did not result in significantly lower gross margins. Although previous studies reported positive effects of microbial inoculation on crop profitability [107,108]. Our results do not allow us to conclude either a positive or a negative effect on profitability.

However, these benefits are not universally guaranteed and depend heavily on site-specific variables, including crop type and cultivar, soil texture, organic matter content, and climate conditions [109,110]. For instance, a meta-analysis of N reduction strategies in Chinese horticultural systems found highly variable outcomes based on management practices and agroecological contexts [111]. This suggests that a one-size-fits-all approach to fertilizer reduction is unlikely to succeed. Instead, adaptive strategies that integrate local knowledge with scientific insight are necessary. Profitability is a key driver for the successful adoption of more sustainable farming practices. Nevertheless, it is important to ensure that the cost of PGPM inoculation does not offset the savings achieved through reduced inputs [107].

5. Conclusions

The results of this study indicate that, under the specific conditions of this trial, reduced mineral fertilization within the crop rotation maintained crop yield and gross margin at levels comparable to those of conventional fertilization. Under reduced fertilization, inoculated treatments were associated with differences in selected soil indicators, although the experimental design did not allow inoculant effects to be clearly disentangled from those of the reduced fertilization background. An additional limitation is that the detailed composition of the commercial inoculants, including strain-level identity and viable microbial concentration, was not disclosed by the manufacturer. This constrains reproducibility and calls for caution when extrapolating the present findings to other inoculant formulations or production systems. Furthermore, nutrient use efficiency, environmental impact, and the microbiological and process-based mechanisms underlying soil responses were beyond the scope of the present study. Overall, this study supports the feasibility of fertilizer reduction in this rotation under local conditions. At the same time, it provides a detailed field-based assessment of agronomic, soil, and economic responses under a multi-year vegetable rotation, offering a useful applied basis for future research under comparable Mediterranean conditions.