A pot experiment was conducted in a custom-made growth chamber (about 5.5 m³, Piardi Tecnologie del Freddo s.r.l., Castenedolo, BS, Italy), under controlled conditions of temperature, humidity, and photoperiod (Giuliani et al., 2014) to evaluate the response of chickpea treated with zeolite, PGPB, and their combination under water deficit. Growing conditions are reported in Figure 1, with day length and temperature adjusted according to crop development, Rh at 70% and a photon flux density of 500 μmol/m²/s. A complete randomized block design, with three replicates, was adopted with a factorial combination of two genotypes and four agronomic treatments consisting of the application of zeolite (Z), a PGPB strain, their combination (Z + PGPB), and an untreated control (CTR), under two irrigation levels, for a total of 48 pots. The two water regimes included a full and a deficit level, at 100 and 50% of the irrigation requirement, respectively. In particular, through gravimetric measurements of the pots, the soil was maintained at field capacity with periodic water replenishments at the full irrigation level and at 50% of the full irrigation level in the deficit level. Genetic material consisted of two genotypes: a large seed cultivar (Pascià) and a small seed one (Sultano), previously characterized in field trials and showing a good adaptability to water deficit (Ruggeri et al., 2017; De Santis et al., 2021, 2022).

The soil was classified as loam consisting of 32.5% sand, 42.3% silt, 25.2% clay, 32.0% field capacity and 14.5% wilting point, 1.18 g/cm³ bulk density, 7.9 pH, 361 mS/cm EC, 0.191% total N, 67.2 ppm available P, 585 ppm exchangeable K, and 1.91% organic C. Treated soil was amended with 2% of synthetic zeolite (on dry weight basis) previously characterized (Belviso et al., 2022). Hydraulic parameters, including field capacity, wilting point, and available water content of the soil before and after the addition of zeolite, are reported in Table 1. The choice to add zeolite at a rate of 2% was guided by the results of previous experiments which demonstrated that larger amounts of this synthetic material (also depending on soil type) affect the porous and hydrophysical parameters of soil, resulting in a lack of water for plants and, consequently, negative effects on crop growth (Belviso et al., 2022); moreover, the previous characterization of this zeolitic material by leaching tests (Belviso et al., 2015) indicated a very low mobility of trace elements (As, Cd, Co, Cr, Cs, Cu, Mn, Ni, Pb, Se, V, and Zn), thus permitting the use in this new type of tests. A commercial PGPB was adopted (B. subtilis QST 713, Serenade ASO, Bayer Crop Science, Italy) through plant–soil spray inoculations at a rate of 10⁸ colony-forming unit (CFU)/m², 2 times at 16 and 30 days after sowing (das). Sowing was carried out on 30 December 2023 with a final density of two plants per pot. Final mean water use (WU) was 622 and 356 L/m² for full and deficit irrigation, respectively. Nitrogen and phosphorus were supplied at a rate of 2.5 and 5.0 g/m² as ammonium nitrate and triple superphosphate, respectively.

Crop growth was assessed by plant height (PH) throughout the crop cycle, with measurements taken from early development to harvest. Flowering was first recorded in both genotypes at 55 das (~1,000 °C d), according to previous observations (De Santis et al., 2021, 2022). At maturity (142 das), plant biomass was harvested, separated into root, straw, and grains, and oven dried (60 °C; 72 h). Dry weight biomass and mineral accumulation were determined in each plot and expressed as g/m². Harvest index (HI) was calculated as the ratio of grain yield (GY) to total above-ground biomass. Grain weight (GW) was also determined. Grains and straw were milled (Tecator, 1,093 Cyclotec; Foss Italia, Padova, Italy) using 1.0-mm sieves for analysis. N concentration was determined (elemental analyzer LECO) in grains and straw, and multiplied by their biomass to determine N uptake. The nitrogen harvest index (NHI) was determined by dividing grain N uptake by total plant uptake. Grain protein content (PC) was determined by multiplying grain N concentration × 6.25. Water-use efficiency (WUE) was determined as the ratio of GY to WU and expressed as g/m²/l.

Analysis of total phenolic content (TPC) was carried out on grains according to (Abou Chehade et al., 2024), with minor modifications. Briefly, 100 mg of chickpea flour, previously obtained after laboratory milling, was suspended in 1 mL of 80% (v/v) methanol. The mixture was sonicated for 30 min and centrifuged (2,000 g for 10 min). Total phenolic content (TPC) was determined in the extracts using the Folin–Ciocalteu method and expressed as gallic acid equivalents (mg/g).

Soil samples were collected at half of the crop cycle (71 das), during stem elongation, and at harvest maturity (142 das), and soil microbial analysis was carried out according to the protocol reported in (De Santis et al., 2023). Briefly, 25 g of each sample were diluted with 225 mL of sterile saline solution (0.9% NaCl) and mixed on an orbital shaker at 300 rpm for 30 min; after that, decimal serial dilutions were carried out in saline solution and plated onto appropriate medium to select and count Mesophilic bacteria (Plate Count Agar supplemented with cycloheximide, 0.17 g L⁻¹; 30 °C for 48 h); pseudomonads (Pseudomonas Agar Base supplemented with Pseudomonas Selective Supplement; 25 °C for 48–72 h); spore-formers (Plate Count Agar, after heat-treating the dilutions at 80 °C for 10 min; the plates were incubated at 30 °C for 24 h); actinobacteria (Bacteriological Peptone, 10 g L⁻¹; Beef Extract, 5 g L⁻¹; NaCl, 5 g L⁻¹; Glycerol, 10 g L⁻¹; Agar Technical n. 3, 20 g L⁻¹; pH 7.00–7.20; 22–24 °C for 7–14 days); Enterobacteriaceae (Violet Red Bile Agar, incubated at 37 °C for 18–24 h); total soil microorganisms (PCA, 22 °C for 3–4 days); yeasts (Yeast Peptone Glucose Agar, supplemented with chloramphenicol, 0.1 g L⁻¹; 25 °C for 3–4 days); molds (Potato Dextrose Agar, PDA; 25 °C for 5 days); and nitrogen-fixing bacteria on Brown medium, incubated at 25 °C for 5 days. All media and supplements were from Oxoid (Milan, Italy). A random selection of colonies on plates, microscope examination, Gram staining, a test for spore production, catalase, and oxidase tests confirmed microbiological counts.

For each irrigation level, analysis of variance (ANOVA) was conducted on the factorial combination of genotype and agronomic treatments, with Tukey’s test used as the post hoc test. Mean values between the two irrigation levels were also compared by t-test. Multiple regression analysis (Pearson) and principal component analysis (PCA) were performed on the correlation matrix of the traits investigated. Microbiological data were initially analyzed using a t-test to assess significant differences between the baseline (day 0) and the end of the trial. Subsequently, a preliminary standardization was performed, and the results were expressed as increases or decreases relative to the beginning of the experiment. When the t-test indicated a non-significant difference, a value of 0 was assigned as the standardized result. Standardized data were then analyzed using a multiple-correlation approach, with a two-way joining (heatmap) method employed for visualization. Raw data were also analyzed using two-way ANOVA, with the agronomic treatment (control, zeolite, PGPB, and their combination) as categorical predictors and the two water conditions (deficit or full) as the other factor. The decomposition of the statistical hypothesis was used as the output.

An experimental trial was conducted under two water conditions, with cumulative water supplies of 8.6 and 14.1 L/pot (depth 0.2 m), respectively 384 (water deficit) and 627 L/m² (full water supply). Crop growth duration was 142 days from sowing to maturity (Figure 1), with no imposed heat stress during terminal growth stages, allowing us to focus on the effect of water deficit. The growth trend, expressed as plant height (PH), under the two different water regimes is shown in Figure 2. Under both conditions, the two genotypes showed a comparable PH up to 62 days, after that the growth rate of Sultano was higher up to maturity (Figure 2), with a generally higher rate under full water supply. In contrast, the use of soil amendments and microbial biostimulants did not affect growth rate and PH under both water supplies. Mean root biomass was also higher under full irrigation. As regards productive and quality traits, analysis of variance showed a general significance of the effect of the genotype, while the effect of the agronomic treatments was more frequent under water-deficit conditions (Supplementary Table S1); the interaction between genotype and treatment was generally not significant. On average, water deficit led to a significant reduction in GY (−50%), N uptake and water-use efficiency (WUE); in fact, GY was about 2 times higher under full irrigation, in a range comparable to other observations in open field trials (from 73 to 254 g/m²); in contrast, a higher total phenolic content in grains (TPC) was observed under water deficit (Table 2).

Figure 3a shows the viable counts for five representative soil group at the end of the trial. Microbial abundance was generally higher under well-watered conditions, for most groups, except spore-forming bacteria. However, for a better understanding of microbial dynamics, data were further analyzed through a multivariate analysis after standardization and reported an increase/decrease compared to the beginning of the trial (Figure 3b). The color gradient reflects the degree of variation from the initial state, with green indicating increased abundance, red indicating a reduction, and white indicating a non-significant variation.

Under deficit irrigation (Figure 3b), the response of the microbial groups exhibited a clear treatment-dependent pattern. Pseudomonads were among the most negatively affected taxa in the control sample, as indicated by a strong reduction in the viable count (approximately −1.5 log CFU/g), consistent with their known sensitivity to osmotic stress and reduced soil moisture. This decline was considerably attenuated when PGPB were applied (reduction in the range 0.5–1 log CFU/g). In addition, the use of zeolite, either alone or combined with PGPB, counteracted the negative effect of water deficit, as evidenced by values close to zero.

Actinobacteria also showed a decrease under deficit irrigation in the control (≈ − 0.5 log CFU/g), confirming the vulnerability of several non-sporulating taxa to low soil water levels. In contrast, the treatments, including zeolite, with or without the application of PGPB, increased in this microbial group (0.5–1 log CFU/g). This suggests that zeolite may enhance aeration and water-holding capacity in the root zone, providing microenvironments suitable for Actinobacteria.

Under full irrigation, the microbial community exhibited greater overall stability than under deficit conditions. Both mesophilic and spore-forming bacteria did not show great changes; instead, they remained largely stable in the control samples, with differences near zero. In several treatments (particularly PGPB and Z + PGPB), slight decreases (0.5 log CFU/g) were observed, suggesting that optimal moisture availability alone may not be sufficient to stimulate these groups beyond baseline levels.

Actinobacteria displayed a more distinct treatment response under full irrigation. In the control condition, this group slightly decreased (0.5 log CFU/g), possibly reflecting competitive exclusion by faster-growing taxa under high-moisture conditions. However, the presence of zeolite resulted in a marked increase (2 log CFU/g), the highest recorded among all combinations, while the joint application of zeolite and PGPB also promoted a substantial rise (1 log CFU/g).

Multiple regression analysis was performed among the investigated traits, as reported in Supplementary Table 2. In general, WU showed a significant negative correlation with TPC and a positive correlation with GY, N uptake, PH, and the group of N-fixing bacteria (Nfb); in fact, Nfb showed a good correlation with GY; furthermore, WUE was significantly correlated with HI, NHI, and the soil Pseudomonas bacteria (Pse) content at maturity.

Principal component analysis, performed on the correlation matrix, identified two major components that together explained 74.2% of the observed variability (Figure 4). Component 1 was associated with the variations in yield components and WUE due to the investigated agronomic treatments and referred to “effect of agronomic treatments on productivity.” In particular, samples treated with both zeolite and PGPB were distributed in the right part of the biplot, associated with an increase in both HI and NHI, and then GY, especially in cultivar Sultano, characterized by lower GW. In contrast, component 2 was mainly associated with variations in TPC and, secondarily, PC, due to water use (WU) and was referred to as “effect of water supply on quality.” In this framework, the co-application of the two investigated treatments emerged, under the current growing conditions, as generally more impactful on productivity traits rather than quality, regardless of water supply, which influenced quality in both genotypes.

Chickpea is generally considered a drought-tolerant pulse crop (Abderemane et al., 2024); however, the response to water supply can affect yield and quality traits, which are favorably welcomed by consumers, such as grain size. The levels of grain productivity observed in the current experiment, even if pot scale, were in a range comparable to the productivity level of chickpea under field conditions, that is, 0.7–2.5 t/ha (López-Bellido et al., 2004; Oweis et al., 2004; Singh et al., 2014), with water deficit causing severe yield loss (Choudhary et al., 2024). Previous observations indicated good adaptability in the two investigated genotypes, although they differed in grain dimensions (De Santis et al., 2021, 2022). However, water deficit limited the genotypic differences in productivity, which resulted in clearer differences under non-limiting water conditions (Samarah et al., 2009; Choudhary et al., 2024). Under the current experimental conditions, without high temperature stress—since the post-anthesis period was set at around 25 °C—genotypic differences in grain productivity were observed only under optimal water supply. This is in accordance with other literature data, which confirms the productive genotype differences between the Pascià and Sultano genotypes, under good growing conditions (Ruggeri et al., 2017). The differences in yield were, in general, correlated with other morphological and crop physiological traits, in particular plant height and harvest index, while grain weight is generally more influenced by genotype effect (Kashiwagi et al., 2006; Keerthi Sree et al., 2023; Tiwari et al., 2023; Pappula-Reddy et al., 2024).

The interest in the use of zeolite from coal fly ash is also to promote a circular economy by converting a problematic industrial waste into a valuable product, reducing landfill dependency and resource depletion (Querol et al., 2002; Mlonka-Mędrala, 2023). Results obtained in the current study indicated a good efficacy of synthetic zeolite to modify the coarse-textured soil and improve the hydraulic properties, also in loam soil, despite a well-documented ability, especially in sandy soils (Lima et al., 2021; Comegna et al., 2023; Satriani et al., 2024). The trait that benefited the most from zeolite application was the harvest index, suggesting a good mitigation of water deficit, especially in the critical stages defining yield (Pappula-Reddy et al., 2024). A comparable behavior was preliminary observed in other horticulture crops, in which the application of zeolite did not lead to an increase in fresh and dry biomass in the vegetative parts (Castronuovo et al., 2023). The hypothesis by which the use of coal ash fly zeolite might show the best performance under water-deficit conditions seems confirmed in the current experiment. In contrast, under well-watered conditions, the addition of the soil amendment was associated with an increase in N uptake. In terms of soil microbial dynamics, in those conditions, the increase in Actinobacteria, generally linked to nitrate assimilation, might have contributed to promoting the increase in crop N uptake (Zhang et al., 2024). In contrast, zeolite applications in the field have recently been reported to improve marketable yield in tomato, possibly indicating a higher benefit on crop reproductive parts (Conversa et al., 2024).

The effects on water-use efficiency or crop productivity are, however, more complex, with the biofertilization resulting more promising, both under different water conditions. In fact, the inoculation of PGPB, in particular bacilli, has been recently reported to improve crop productivity in chickpea (Elkoca et al., 2007; Yadav and Verma, 2014; Khan et al., 2019, 2021), and in our study, this resulted in higher harvest index and water-use efficiency. From a microbial ecology standpoint, the application of B. subtilis QST 713 significantly modulated soil microbial communities, as demonstrated by multivariate analyses. The use of PGPB resulted in an increase of spore-forming bacilli, mainly with zeolite, across both irrigation regimes, indicating successful colonization and persistence, which are basic requisites for a biofertilizer (Etesami et al., 2023). A result of concern was found for deficit irrigation conditions, as in this case, the use in combination of zeolite and PGPB exerted a sort of “stabilizing effect” on microbial communities, mainly on actinobacteria and Pseudomonas; these taxa are often associated with stress mitigation and root health (Martins et al., 2023; Chen et al., 2024; Mikiciuk et al., 2024). In addition, zeolite appeared to counteract declines in some groups, suggesting that it could act as a physical buffer by improving water retention and microhabitat stability, and, as a consequence, could protect and/or enhance the ecological niches of beneficial microbes. This evidence confirms the suitability of zeolite for soil health (Hernández et al., 2024).

Quality traits showed greater sensitivity to growing conditions, particularly water availability. Under the current experimental conditions, water supply and, secondarily, genotype, showed an impact on the investigated quality traits, with the agronomic treatment showing a major effect on productivity. Protein content in chickpea is generally a stable trait, whereas most of the changes occur in protein composition (De Santis et al., 2021). Regarding bioactive compounds, it has been reported that drought-tolerant chickpea genotypes accumulate higher total phenolic content in leaves under water-limited conditions (Kasbi et al., 2024). In fact, secondary metabolites, including phenolic compounds, are generally produced by plants in response to biotic or abiotic stress (Jameel et al., 2021; Quiroz-Figueroa et al., 2022). Hyperosmotic stresses are generally responsible for the increase in phenolic compound accumulation, as also observed in the current study (Asati et al., 2024), while heat temperature may lead to a reduction of phenolic compounds (Jameel et al., 2021); heat and drought might occur especially in Mediterranean regions, increasing the complexity of the response of quality traits to environmental stress (Awasthi et al., 2024; Chimphango et al., 2025). Furthermore, the genotypic variation in phenolic acid content might be explained by the reported higher accumulation of phenolic compounds in small grain chickpea genotypes (Quiroz-Figueroa et al., 2022).

The effect on the soil microbiota was linked to agronomic outputs and to the positive modulation of some parameters; these outcomes could result from a combination of factors, including a more favorable rhizosphere environment for plants. It is well known that B. subtilis can stimulate root exudation and auxin production, enhancing root architecture and water uptake efficiency (Hakim et al., 2021). Moreover, the positive effect on Pseudomonas and actinobacteria could, in turn, promote a higher synthesis of siderophores or stress-related metabolites (Chen et al., 2024).

However, when focusing on the influence of agronomic inputs on the soil microbiota, it is worth noting that the changes resulted from a combination of factors. In particular, the heatmap reveals clear taxon-specific responses driven by moisture availability and amendment type. Under deficit irrigation, the decrease of pseudomonads and Actinobacteria in the control condition is in line with literature reports on the sensitivity of non-sporulating bacterial groups to reduced water potential and osmotic constraints (Naylor and Coleman-Derr, 2018). However, the application of zeolite—either alone or in combination with plant growth-promoting bacteria (PGPB)—substantially attenuated these negative effects, yielding profiles that approached stability or even positive deviations from the baseline. This pattern could be due to the ability of zeolite to enhance water retention, regulate ion exchange, and buffer fluctuations in soil microenvironments (Belviso, 2025). In addition, the mitigation of the negative effects on soil microorganisms by PGPB could be due to microbial network resilience, resulting from stimulation of rhizosphere activity, although their effects were less pronounced than those of zeolite. Under full irrigation, the overall stability of mesophilic and spore-forming bacteria is in line with the hypothesis that water deficit could be a major driver of microbiota in soil; thus, in well-watered conditions, microbiota tends to be more stable (Manzoni et al., 2012). However, differences were found for Actinobacteria, which experienced an increase with zeolite and PGPB also under full irrigation. This result suggests that some amendments modifying soil structure and aeration could generally favor some taxa (Lehmann and Kleber, 2015), as in the case of Actinobacteria, which are, to the authors’ knowledge, slow-growing taxa. The additive effect of PGPB in the zeolite-amended treatment further confirms the importance of multi-component soil amendments in shaping microbial communities.