The process of vermicomposting pre-treated BSG was carried out following the methodology described by Bianco et al. (2022). A single area measuring 1 × 1 m was filled with a mixture of 10 kg of pre–treated BSG and 90 kg of cow manure, supplemented with 5 kg of earthworms (Eisenia fetida, Savigny, 1826). Upon completion of the worm-cast production phase (90 days) and the subsequent curing period (30 days), fully matured vermicompost suitable for direct soil application was obtained. The composting process of BSG was carried out as described by Assandri et al. (2021). Enriched compost was obtained by incorporating a commercial microbial consortium, in the form of a wettable powder with a concentration of 3 g L⁻¹. Biochars were produced from pre-dried BSG at 45°C for 48 h. Subsequently, the dried BSG was subjected to pyrolysis inside a carbolite furnace, Gallenkamp Muffle Furnace model, at temperatures of 350 and 550°C (Rodríguez-Vila et al., 2018). The BSG was loaded into steel containers (diameter 10.00 cm and height 18.20 cm) and covered with a perforated lid to prevent overpressure. The containers were heated in the carbolite furnace to the specified temperature and then held for 1 h before being allowed to cool overnight. The biochar was then removed (Rodríguez-Vila et al., 2022). During pyrolysis, there was no automatic air control, and low oxygen/anoxic conditions were maintained by loading feedstock into the steel containers and saturating them with feedstock materials. Moreover, the containers are closed then air cannot get inside the steel containers after closing them. BSG-based substrates used in the seedbed experiment and the soil collected from the surface horizon (0–30 cm) for the pot experiment were analyzed according to the Italian Official Methods of Soil Analysis (Ministero delle Politiche Agricole e Forestali, 1999). Microbiological analyses were performed on the BSG-based substrates following the method described by Bianco et al. (2022). The physicochemical characteristics and microbial count are shown in Tables 1, 2.

Enumeration of Escherichia coli was carried out according to the ISO 16649-2 (2001). The incidence of Salmonella spp. was determined as described in ISO 6579-1 (2017). Before carrying out the pathogenicity tests with S. sclerotiorum, the soil used in the tests was analyzed to exclude the presence of any pathogens. A “soil dilution plate technique” was used (Burgess et al., 1994), where 1 g of soil was suspended in a flask containing 99 ml of 0.05% water agar. Furthermore, 1 mL of the soil suspension was spread onto Potato Dextrose Agar (PDA) plate, and three replicates were performed. The plates were incubated at 25°C with a 12 h photoperiod. After 7 days, the colonies were transferred to new PDA plates, and monosporic or monohyphal cultures were performed. The mycelium (~100 mg) from the pure cultures obtained was used the DNA extraction. Amplification of the internal transcribed spacer (ITS) region was carried out (Chen et al., 2011).

The germination test was carried out on cress (Lepidium sativum, L. purchased from Fratelli Ingegnoli, Milano, Italy) and lettuce (Lactuca sativa L. var. capitata “Regina dei ghiacci”) seeds (Pagano Domenico & figli, Salerno, Italy) according to the methodology described by Zucconi et al. (1981). Vermicompost, compost, and biochar were mixed with distilled water in a 1:10 (w/v) ratio and shaken at 150 rpm for 1 h. They were then centrifuged at 4,000 rpm for 20 min and filtered through 0.45 μm Millipore membrane filters. Dilutions were prepared from each sample using distilled water: 50% (50% extract, 50% water), 75% (75% extract, 25% water), and control (only distilled water). Whatman sterilized filter paper was moistened with 5 ml of extract and placed inside a Petri dish of 9 cm of diameter. Ten seeds of L. sativum and L. sativa were placed on the filter paper, and the Petri dishes were incubated in darkness at 25 ± 2°C. After 3 days of incubation, the percentages of germination and root length were recorded. The germination index (GI) was calculated with the mean value from three experiments as follows: GI (%) = [(Gt %)/(Gc %)] × [Lt/Lc] × 100; where Gt% and Gc% are the mean percentages of germination with the extract treatment and the control, respectively, and Lt and Lc are the mean root lengths with the extract treatment and the control, respectively (Mohamed et al., 2020).

This short-term pot experiment took place during the winter/spring season (from February to May 2022) in a Mediterranean greenhouse at the Ottava experimental station of the University of Sassari (40°46′ 47" N, 8°29′ 45" E). The greenhouse was unheated, and the average minimum air temperature during the trial was 5°C, while the average maximum air temperature was 16°C. The average heliophany was ~421.45 min day⁻¹.

The trial was performed using BSG-based substrates, especially vermicompost (V), compost (CP), compost enriched with the microbial consortium (CPE), and biochar obtained from 350 and 550°C pyrolysis temperatures (B350 and B550), mixed with peat at two different percentages, 2 and 4%, and compared with two control treatments, C1 and C2, consisting of 100 % peat and peat fertilized with a ternary chemical fertilizer (NPK 20–20–20), respectively (Figure 1, Table 3). In the control treatment C2, a nitrogen amount equivalent to 35 kg ha⁻¹ was provided.

The seedbed phase extended for 43 days from seeding to transplanting, while the crop cycle lasted 58 days from transplanting to harvesting the lettuce heads. Overall, the entire experiment lasted for 101 days. In both growth phases, a completely randomized block design was adopted.

In the seedbed phase, 12 replicates per treatment were performed with five lettuce seeds for each phytocell (approximately volume 60 ml plug⁻¹) per 2 plastic seedbeds (18 × 13.5 × 5 cm; 12 phytocells). Upon reaching the stage of three true leaves, 43 days after seeding, six plants (six replicates) per treatment were randomly selected for the evaluation of quantitative parameters (Table 5). The remaining six plants were transplanted into pots (20 × 9.5 cm), each filled with 500 g of soil, for a total of 72 pots. The phenological stage of the three true leaves and the harvesting time were assessed using the BBCH scale (Biologische Bundesantalt, Bundessortenamt, and CHemische Industrie, Berlin, Germany).

The plants were transplanted into pots without further inputs of chemical fertilizers in the control treatment and without the addition of further BSG-based substrates in the organic treatments.

During the crop cycle phase, the pots were irrigated twice a week with 500 mL of distilled water. The plants were inoculated with the pathogen (S. sclerotiorum) 10 days before transplanting, dividing the total of 72 pots into two groups: 36 pots with the pathogen (Y) and 36 pots without the pathogen (N). The inoculum was added as 3 plugs of fungal mycelium (1 cm in diameter) per pot of a 10-day-old culture of S. sclerotiorum 10 days before transplanting. Agar plugs without mycelium were used in the control pots.

After 58 days of plant growtht, fresh root-adhering soil samples were collected from all pots receiving the same treatment, bulked together, sieved to <2 mm, and used for the analysis dehydrogenase (DHG) and urease activity (URE). The DHG activity was determined by the amount of triphenyltetrazolium red developed from soil samples treated with triphenyltetrazolium chloride at 37°C for 24 h of incubation. The URE activity was determined by quantifying the ammonia released from soil samples treated with urea during 2 h of incubation at 37°C (Alef and Nannipieri, 1995).

The analyses were conducted using R statistic software version 4.2.3 (2022–04–22 ucrt) with the utilization of the “car,” “emmens,” “multicomp,” “factoextra,” and “multicompView” packages. The data were analyzed through a generalized linear model (GLM) with a Gaussian distribution. Assumptions of homogeneity of variances, normality, and independence of the residuals were graphically checked. The two phases (seedbed and harvest) were analyzed separately because, in the first one, only the treatment was considered as a factor, while in the second experiment, both the treatment and the presence of the pathogen were considered as factors. In this case, the assessment of the significance of fixed and interacting factors in linear models were carried out using the “join_tests” of the “emmeans” package. In both experiments, the significance of the differences between mean values was assessed using the Sidak test at p < 0.05. The principal component analysis was carried out using normalized phenotypic and biochemical data collected at the end of the second experiment. Clusters represented in the PCA biplot were determined by the k–means algorithm, as implemented in the “stats” package of R software.

All the tested BSG-based substrates showed a positive effect on germination rates of L. sativum and L. sativa seeds when used up to a concentration of 50% (Table 4), exceeding the minimum GI required as per Annex 2 of Legislative Decree 75 of 29 April 2010.¹ The GI values recorded at 75% show a reduction in the germination of seeds.

Salmonella and E. coli were not detected in any of the tested BSG-based substrates, in compliance with Legislative Decree No. 75/2010. The molecular identification of Ottava soil confirmed the absence of lettuce pathogens.

At the transplant time (Table 5), significant differences (p < 0.05) among treatments were found in terms of HPt, PFWt, and RLt.

The HPt ranged from 0.6 ± 0.05 cm (B350 2%) to 2.6 ± 0.55 cm (C2), with the latter proving to be the most effective treatment, statistically different from all others. On the other hand, the B350 2% treatment, despite being less effective, was only statistically different from the CP 4%, CPE 4%, and C2 treatments. The lowest RLt was observed in the B350 2% treatment, measuring 1.3 ± 0.75 cm, which was statistically different from the CP 4%, CPE 4%, and V 4% treatments. The highest RLt values were recorded at 9.0 ± 1.27 cm for the CP 4% treatment and 8.5 ± 2.43 cm for the CPE 4% treatment, both of which were statistically different from all other treatments (p < 0.05). Both control treatments (C1 and C2) showed average root length (RLt) without significant differences compared to all other BSG-based substrate treatments (p > 0.05). Similar to HPt, the highest average value for PFWt was observed under the C2 treatment at 0.08 ± 0.03 g, which was statistically different from all other treatments (p < 0.05). However, the lowest value of 0.01 g was recorded under B350 2%, B550 2%, CP 2%, and V 2% treatments. No significant differences (p > 0.05) emerged for root fresh weight (RFWt) that ranged from 0.003 ± 0.00 g in the B350 2% treatment to 0.03 ± 0.06 g under the C1 control treatment.

Considering that the edible part of L. sativa corresponds to the fresh plant material, the relative dry weight of plants and roots were also determined. In both cases, the best performing treatment was CPE 4%, while the least effective was B550 4% (Supplementary material). This analysis revealed a similar trend across the parameters evaluated.

Overall, in the absence of the pathogen, at the time of harvest (Table 6), the control treatment C2 met the expectations, proving to be the most productive treatment, with the highest fresh weight of plants (PFW), leaves (LFW), number of leaves (LN), and leaf area (LA/LN), with average values of 36.0 ± 10.43 g, 30.5 ± 9.95 g, 16.3 ± 0.58 leaves, and 390 ± 181 cm², respectively (Figure 2). While the lowest PFW and LFW were observed in the B350 2% treatment, with 15.7 ± 0.58 g and 12.3 ± 1.15 g, respectively. Considering the parameters, such as number of leaves (NL) and LA/LN, the lowest values, 9.5 ± 1.50 and 226 ± 120 cm², respectively, were found in the control treatment C1.

Even in the presence of biotic stress, the control treatment C2 showed the highest values in all parameters except for LA/LN. The highest leaves area (LA/LN) was recorded in the B 550 4% treatment, with an average value of 268 ± 159 cm². On the contrary, the treatment with the lowest values for PFW, LN, and LA/LN was the control treatment C1, with average values of 8.8 ± 0.68 g, 9.0 ± 1.00 number of leaves and 126 ± 10.0 cm², respectively. The lowest LFW was observed in the B350 2% treatment with 6.6 ± 5.39 g.

The V 4%, CPE 4%, and CPE 2% treatments exhibited statistically similar values (p > 0.05) across all four parameters, both in the absence and presence of the pathogen. Additionally, the CP (2 and 4%), B550 (2 and 4%), and B350 4% treatments did not show statistically significant differences in the PFW, LFW, and LA/LN parameters when compared to the C2 treatment.

Significant differences caused by the treatment–pathogen interaction in the carotenoid content for the CPE 4% treatment and in the flavonoid content for the majority of treatments were found (Table 7). The highest flavonoid content was recorded in the B350 2% parameter (0.84 ± 0.02 mg kg⁻¹), showing no statistically significant difference from all other pathogen-treated treatments. While the lowest flavonoid content, statistically different from all treatments, was observed in the treatment C1 (without pathogen) with an average value of 0.30 ± 0.03 mg kg⁻¹. The C2 treatment (0.71 ± 0.00 mg kg⁻¹) was determined to be statistically similar to all the remaining non-inoculated treatments.

The highest chlorophyll (α and β) and carotenoid contents were measured in the CPE 4% treatment (without pathogen) with average values of 17.9 ± 4.64 mg kg⁻¹, 5.7 ± 1.75 mg kg⁻¹, and 1428 ± 346 mg kg⁻¹, respectively. On the other hand, the lowest values of chlorophyll α and carotenoid contents were observed, in the presence of the pathogen, under the control treatment C2, with average values of 7.6 ± 1.85 mg kg⁻¹ and 658 ± 149 mg kg⁻¹, respectively.

In addition, the HPLC analysis of nitrates was conducted on lettuce leaves, both with and without the pathogen. The results showed, for all treatments, the nitrate content below the limits specified for greenhouse-grown lettuce harvested between April 1st and September 30th.² These findings were not shown in the data.

Principal component and cluster analyses (Figure 3) allowed the subdivision of the different treatments in five different groups. Treatments included in Group 3, consisting of the C1 treatment in the presence and absence of the pathogen, exhibited the lowest mean values of all the analyzed parameters. In contrast, Cluster 4 was characterized by treatments with the highest mean values, including B550 4%, CPE 4%, CP 4%, and C2. Of particular interest is the CPE 4% treatment, which exhibited higher levels of chlorophyll α, chlorophyll β, and carotenoids.

The DHG activity exhibited a significant increase across all pots containing BSG-based substrates, ranging from a minimum of ~1.40–fold for the V 4% treatments to a maximum of ~3.52–fold for the B350 2% treatment (Figure 4). For all BSG-based substrates (except CPE), 2% was more effective to increase DHG (between 1.61- and 2.42-fold) than 4%. On the other hand, the lowest DGH activity (214.80 μg TPF g⁻¹ h⁻¹) was recorded in the B550 4% treatment in the presence of the pathogen. Additionally, in this case, irrespective of the pathogen presence except for vermicompost, the lower rate (2%) of BSG-based substrates was more effective to increase DHG (between 1.22- and 3.18-fold) than the 4%.

As regards the N cycle, the URE activity increased for all BSG-based substrates at 2% except for CPE and B350 (Figure 5) more than those applied at 4% and when compared to the controls (C1 and C2). The most effective BSG-based substrate to enhance URE activity was B350 4% (52.21 μg NH₄-N g⁻¹ h⁻¹), while the less effective was B550 4% (35.03 μg NH₄-N g⁻¹ h⁻¹).

Generally, in the presence of the pathogen, the BSG-based substrates application was not effective or reduced the URE activity compared to the controls (C1 and C2). The only exception was vermicompost, especially V 2% that increased URE by 1.18-fold compared to C1 and C2. The lowest value of the URE activity was recorded for B550 4% (35.99 μg NH₄-N g⁻¹ h⁻¹) in the presence of the pathogen (Figure 5).