Microbiological analysis of the indigenous cultivated microbiota was performed at different composting stages (0, 4, 9, 21, 52, and 120 days). After mixing, 1 g of substrate was taken and suspended in 10 ml of sterilized physiological water (9 g NaCl/L distilled water). The suspension was homogenized by first shaking and then treatment for 10 to 15 min by sonication (El Fels et al., 2015; Bouhia et al., 2021). Afterward, the suspensions were serially diluted to 10^–9. The pH of each co-composting sampled mixture was adjusted to be the same.

For Actinobacteria enumeration, actinomycete isolation agar (AIA) (Sigma) was supplemented with 50 μg/ml cycloheximide and 10 μg/ml nalidixic acid to inhibit fungi and gram-negative bacteria without affecting the growth of Actinobacteria. Bacterial community enumeration was performed using standard medium (nutrient agar) with the addition of cycloheximide (Sigma). For estimation of the fungal community, suspension calibration was performed in potato dextrose agar (PDA) (Panreac) by the addition of 5 μg/ml chloramphenicol. Colony forming units (CFU) were counted in triplicate, and the results are expressed as CFU g^–1 dry weight. The pH of each medium was adjusted to 5.6; 5.7; 5.9; 6.5; 7.4; and 8, respectively, for 0, 4, 9, 21, 52, and 120 days to be similar to that of each composting phases. Samples were incubated at both mesophilic (35°C) and thermophilic (45°C) temperatures.

To study the variation in the total microbiota (bacteria and fungi) during the co-composting phases (mesophilic, thermophilic and maturation phases), we performed comprehensive metagenomic analysis. DNA from the samples (≈ 1 g of compost from each phase) was extracted by using a DNeasy PowerSoil Pro Kit (Qiagen Inc.) following the manufacturer’s instructions. Then, the quantity and quality of the extracted DNA were evaluated spectrophotometrically using a NanoDrop 2000c before being processed for 16S and ITS amplicon sequencing. Sequence libraries were generated using the complete primers ITS1 (5′CTTGGTCATTTAGAGGAAGTAA 3′) and ITS2 (5′ GCTGCGTTCTTCATCGATGC 3′) for fungi. For bacteria, the 16S rRNA gene V4 variable region PCR primers 515/806 were used in a 30–35 PCR cycle using HotStarTaq Plus Master Mix Kit (Qiagen Inc.). PCR was performed under the following conditions: 95°C for 5 min, then 30–35 cycles at 95°C for 30 s, followed by 53°C for 40 s and 72°C for 1 min. Finally, a final elongation was performed at 72°C for 10 min. Afterward, samples were multiplexed using distinctive dual indices and equally pooled together based on their molecular weight and DNA concentration. Then, they were purified using calibrated Ampure XP beads and used to prepare an Illumina DNA library. Sequencing was performed with MR DNA (Shallowater, TX, United States)¹ on a MiSeq following the manufacturer’s guidelines, and sequence data were processed using MR DNA analysis pipeline (MR DNA, Shallowater, TX, United States). Briefly, sequences were joined, and sequences < 150 bp or with ambiguous base calls were removed. Sequences were then quality filtered (error threshold of 1.0) and dereplicated. The dereplicated or unique sequences were denoised, and unique sequences identified with sequencing and/or PCR point errors were removed, followed by chimera removal. thereby providing a denoised sequence or zOTU. Final zOTUs were taxonomically classified using BLASTn against a curated database from the NCBI.²

The results are the mean of 3 values, and measurement data are shown as the mean ± standard deviation (SD). Statistical analysis was carried out using IBM SPSS Win software version 20. Principal component analysis and Pearson’s correlation were executed using Excel Analyze-it 1.56.

The evolution of the total mesophilic and thermophilic microbiota during the composting process (Figure 2) was correlated with temperature variation. The total mesophilic microbiota showed a maximum increase at day 4 from 3.53 × 10⁷ to 26.61 × 10⁷. Then, a gradual decrease was observed, and a value of 2.77 × 10⁷ was reached after 4 months of co-composting. For the thermophilic microbiota, the optimal increase in enumerated microorganisms (from a value of 6.54 to 29.7 × 10⁶) was determined at day 9, corresponding to the thermophilic phase. Later, a decrease to 3.9 × 10⁶ by the end of the co-composting process was recorded.

During the biodegradation phase (Figure 2A), the mesophilic microbiota was initially present. Once the temperature surpassed 40°C, thermophilic microorganisms became increasingly dynamic and progressively replaced the mesophilic microorganisms. From days 0 to 9 of the composting process, mesophilic Actinobacteria showed a decrease in total abundance from 16.56 to 9.15%, while thermophilic Actinobacteria showed an increase from 6.88 to 15.72%. In the bacterial community, the total abundance was less than 7% for both categories, with a successive increase from 4.63 to 6.46% and decrease from 5.5 to 4.47% for mesophilic and thermophilic bacteria. However, the fungal abundance presented the highest values during the entire process of composting, with variations of 78.8 to 84.37% and 87.61 to 79.79% for mesophilic and thermophilic fungi, respectively, after 9 days of composting.

By the end of the co-composting process, the abundance of mesophilic and thermophilic fungi decreased from 78.80 and 87.61% to 48.73 and 60%, respectively. Nevertheless, a decrease in Actinobacteria abundancy from 30.69 and 25% to 20.58 and 15% for mesophilic and thermophilic bacteria, respectively, was recorded.

Metagenomic analysis of samples representing the key phases of the composting process (mesophilic, thermophilic and maturation) revealed that bacterial taxonomic diversity initially existing in raw OMWS was mostly dominated by Gammaproteobacteria (52.4%), Bacilli (18%) and Actinobacteria (17.8%). At the species level, bacteria were prevalent, namely, Zymobacter palmae (20%), Pseudomonas sp. (19%), and Lactobacillus acidipiscis (6%). Moreover, fecal bacteria were also identified, such as Staphylococcus sciuri (0.003%), Enterococcus faecalis (0.1%), Enterococcus gallinarum (0.1%), and Enterococcus raffinosus (0.003%), which should be considered potential infectious agents. For fungi, diversity analysis revealed that yeasts were the most abundant genus in raw OMWS (Figure 3). These were mostly represented by the Pichia genus, with almost 53% of the total identified population. The remaining genera were Tortispora (16%), Dipodascus (12%), Rhizophydium (11%), Candida (4.4%), Galactomyces (2.4%), and Scedosporium (0.4%).

In the initial mixture (Figure 3C), the fungal microbiome was dominated by four species (> 5%), namely, Dipodascus australiensis (33.18%), Penicillium roqueforti (27.27%), Pichia sp. (8.17%) and Cryptococcus sp. (7%). Similarly, the bacterial microbiome (Figure 3D) was dominated by Psychrobacter aquaticus (30.3%), Corynebacterium variabile (28.7%), Carnobacterium maltaromaticum (11.5%), and Psychrobacter pulmonis (9.7%).

By the end of thermophilic phase, major changes were observed in taxa abundancy (Figure 3C), as the compost was dominated by only two fungal and three bacterial species, namely, Candida freyschussii (87.3%), Sporopachydermia lactativora (10.6%), Pseudomonas syringae (47.8%), Actinobacter sp. (30.3%) and Corynebacterium variabile (6.1%). After 120 days of compost maturation (Figure 3D), the fungal community was mainly represented by four species, Penicillium crustosum (48%), Nectria mauritiicola (20.7%), Acremonium sp. (6.7%) and Guehomyces pullulans (5.9%). The minor representatives (>1%) were dominated by Penicillium sp. (2.9%), Penicillium roqueforti (2.9%) and Penicillium hirsutum (1.6%). On a related note, the bacterial diversity significantly shifted at the end of the maturation phase compared to the initial mixture. Indeed, fecal bacteria were not detected, and the percentage of each bacterial class decreased except for Actinobacteria, for which a threefold increase (60%) was noted compared to raw OMWS. At the species level, the most dominant bacteria were Promicromonospora sukumoe (13.9%), Pseudomonas stutzeri (7.3%), Pedobacter sp. (7.1%), Brevibacterium linens (7%) and Arthrobacter protophormiae (5.3%), followed by Isoptericola sp. (4.6%), Carnobacterium maltaromaticum (4%), Psychrobacter aquaticus (2.3%) and Promicromonospora vindobonensis (2%). Moreover, alpha diversity was determined for important composting phases by calculating the number of OTUs, Shannon index and Faith’s PD (Figure 4). Higher microbial diversity index was recorded for raw OMWS and samples of the compost maturation phase, while the lower diversity was recorded for samples of the thermophilic phase.

Temperature evolution during the composting process is related to microbial community changes during the biodegradation of composted wastes, which plays an important role during maturation and stabilization (Ma et al., 2018). The initial increase in pH is attributed to the organic matter richness of the mixture, especially the lipid and phenolic compounds that microorganisms metabolize more quickly, in addition to the low molecular weight organic acids that are volatilized due to the higher temperature of the thermophilic phase (Tsai and Chang, 2019). The final pH value was 8, which is a direct result of the high decomposition rate of organic matter, which exceeded 62% after 120 days of composting, and the elimination of large amounts of secondary metabolites produced by microorganisms during biodegradation, such as acetic acid and butyric acid (Wei et al., 2016). Furthermore, the %TOC decrease during the composting process was attributed to microbial activity, which reduced the organic content in the composted mixture. According to Huang et al. (2019), this effect is related to active thermophilic microorganisms with a great ability to degrade organic carbon and eliminate various organic compounds initially present in the mixture. The enhancement in nitrification might explain the decrease in the NH₄^–/NO₃^– ratio to 0.91, indicating that compost reached maturity based on the standard maturation index reported by Barje et al. (2012).

After 4 months of composting, the total lipids and polyphenols decreased, which is in alignment with the findings of Hachicha et al. (2009), who reported similar results for OMWS/sesame bark cocomposting. Those authors attributed this reduction to the use of those compounds as a carbon source by indigenous microorganisms through the conversion of phenols to simpler structures or quinones through lignolytic enzymes such as laccase and peroxidase, which are supplied during humic substance polymerization (Ait Baddi et al., 2009). Lipid degradation could be explained by the ability of microbes to breakdown several forms of lipids, such as fatty acids and triglycerides.

The use of the germination index (GI) to evaluate compost maturity by studying phytotoxicity risks for plants is acceptable. Compost is exempt from any potential toxicity to plants if the GI values are greater than 80% (Zhang et al., 2020). Several factors have been previously reported to be responsible for inducing phytotoxicity effect, during the OMWS compost evaluation, this effect has been attributed to the presence of several compounds such as organic acids and phenol compounds, impacting negatively the seed germination (Hachicha et al., 2009; Filippi et al., 2013). According to Piotrowska et al. (2006), many seed species are sensitives to phenol presence such as tomato, which are related to their own genotype. In additions to the low pH value, the high concentrations of lipids in OMWS, can negatively influence the growth and biological activities of seeds, which was negatively correlated with GI reduction (Paulino et al., 2006; Hachicha et al., 2009). In agreement with our results, Said-Pullicino et al. (2007), concluded that a value of GI superior to 60%, the toxicity could be considered as well treated. Our results revealed that phytotoxic effects remained even after the thermophilic phase, as the PP and Lp contents were still relevant. However, at the end of the maturation phases, the phytotoxic effects were completely alleviated, as the GI% reached 110 and 82% for Cress and Turnip, respectively. Hence, the occurrence of plausible biostimulatory effects was demonstrated.

Microbial diversity analysis showed that raw OMWS was mainly dominated by fungi, which represented nearly 54% of all identified kingdoms (including bacteria). Similar results were recently reported by Martínez-Gallardo et al. (2021), who found that in OMWS sampled from several ponds, the fungal community systematically outnumbered its bacterial counterpart. Furthermore, the same authors revealed that the bacterial community structure of OMWS is highly dependent on physicochemical parameters, notably nutrient content, pH and moisture content. Indeed, Proteobacteria were prevalent in OMWS with neutral pH and high moisture. Actinobacteria were dominant in dry/alkaline OMWS, which is in alignment with our results. Notably, physicochemical traits are not the only factors that influence the microbial diversity and functionality of OMWS. According to Tsiamis and Tzagkaraki (2012), the latter are significantly affected by olive cultivars. Those authors investigated the bacterial profile of OMWS generated from various olive varieties and found only 15% similarity in terms of the identified OTUs. Moreover, the cultivation and harvesting practices had a large influence on the microbial community distribution in OMWS. For example, Carlozzi et al. (2015) showed that fermentative bacteria in OMWS resulting from the early collection and harvesting “hand collection” of the Olea europaea variety were restricted to the two families Peptococcaceae and Sporolactobacillaceae (Kavroulakis and Ntougias, 2011). On another note, the very few studies that thoroughly investigated the fungal diversity in OMWS reported very contrasting results. For example, Martínez-Gallardo et al. (2021) revealed that the OMWS fungal community was dominated by genera such as Fusarium, Aspergillus, Scopulariopsis, Tritirachium, Scedosporium, and Microascus, depending on the pond characteristics. Slama et al. (2021) identified four dominant genera, namely, Nakazawaea, Saccharomyces, Lachancea, and Candida. Our findings were very different from those previously reported, as the dominant fungal genus was Pichia, which constituted half of the identified fungal population, followed by Tortispora, Dipodascus, and Candida. The prevalence of Pichia sp. was not surprising, as several studies have reported a strong correlation between the occurrence of Pichia species and the detoxification of OMWS (Morillo et al., 2008; Sinigaglia et al., 2010; Arous et al., 2016). Moreover, to our knowledge, this is the first time that the Diplodocus genus has been identified in OMWS. This genus was represented solely by the cactophilic Dipodascus australiensis, which is not surprising knowing that the pond from where the sample was taken was surrounded by Opuntia spp. Similarly, Tortispora and Pichia are usually associated with decaying cactus tissue (Lachance et al., 1988). These observations suggest that the community structure of OMWS is highly dependent on the faunal and floral specificities of the sampling sites.

Adding green waste to OMWS significantly affected the composition of the microbial communities, as the diversity of Actinobacteria increased twofold compared to raw OMWS, which is plausible due to the actinobacterial richness of green waste. However, members of the Proteobacteria family (primarily represented by Moraxellaceae) were still the most dominant. Similarly, for fungi, the previously dominant Pichiacea were greatly outnumbered by Microascaceae, which represented more than 70% of the fungal diversity of the initial mixture. At the end of the thermophilic phase, half of the polyphenols and nearly 60% of total lipids were degraded, which was accompanied by a significant change in both fungal and bacterial diversity, suggesting the existence of a significant correlation between microbial changes and pollutant dynamics during the composting process. Principal component and Pearson correlation analyses (Figure 5 and Supplementary Material 1) revealed that overall pH, temperature and EC were the factors that affected most community structures. Moreover, the pace at which those parameters changed during the composting process was paramount, which was demonstrated by the occurrence of fast and slow phases of polyphenol and lipid degradation. Temperature is particularly important, as this is the first factor to undergo significant changes, which combined with PP content, initially shape the compost microbial community. Our study showed that most of the microbial species were negatively affected by temperature increases, thus demonstrating the prevalence of the mesophilic microbiota. For fungi, only three families were positively correlated with temperature, namely, Saccharomycetaceae, Dipodascaceae, and Pichiaceae (Figure 5A).

Surprisingly, those fungal families were not strongly correlated with polyphenol and lipid removal when taking into consideration the whole composting process. However, this does not exclude a plausible involvement in the degradation of those organic compounds at early stages of composting, notably during the fast degradation phase observed in the first several days (Figure 1B). Indeed, the works of Ben Sassi et al. (2006) and Sinigaglia et al. (2010) previously reported the ability of Pichia and Candida species (Candida diddensiae, Candida ernobii, Pichia holstii, and Pichia membranifaciens) isolated from Moroccan OMWS to degrade PP, such as p-coumaric, caffeic and vanillic acids. Additionally, many other authors have reported that OMWS detoxification can be directly attributed to species of the Pichia genus (Morillo et al., 2008; Arous et al., 2016). Similar trends were observed for bacteria, as the only bacterial classes (Clostridia and Gammaproteobacteria) that were positively correlated with temperature had limited effects on PP and Lp removal, as shown by PCA. More importantly, even if Actinobacteria were negatively affected by temperature, some thermotolerant representatives (Supplementary Material) were likely to be involved in PP and Lp degradation. For example, at the end of the thermophilic phase, the actinobacterial composition of the mixture was dominated by Corynebacterium sp., which have been reported as potent PP degraders (Pradeep et al., 2015). Moreover, the highly sensitive Pseudomonas syringae constituted more than 50% of the identified bacterial species, which could be explained by the favorable physicochemical conditions, including the decrease in PP content following microbial removal by fungal and actinobacterial species. In fact, Pseudomonadaceae was strongly correlated with Actinomycetaceae (0.995), which demonstrates the importance of microbial species interactions during the composting process. Regarding fungi, the end of the thermophilic phase was marked by the prevalence of two species, namely, Candida freyschussii, an oleaginous yeast knowing for its ability to produce lipids from glycerol (Amaretti et al., 2012), and Sporopachydermia, which was previously reported to be a lipolytic microorganism (Fickers et al., 2005; Agnolucci et al., 2013). Following the thermophilic phase, pH and EC were the most determinant factors influencing microbial community structure, and similar trends were observed for fungi and bacteria, as phyla (Actinobacteria and Ascomycota) that were favored by pH and EC increases were positively correlated with WSPP and TLEP. On a related note, the results of the culture-dependent analysis revealed that fungi were probably more active than bacteria during the thermophilic phase, as they represented more than 70% of all cultivated species. Usually, bacteria constitute the majority of the microorganisms during composting, with a greater number than Actinobacteria and fungi. Our findings revealed opposite results, which could be attributed to the antimicrobial effect of OMWS (richness in monomeric and polymeric phenolic compounds). Additionally, fungi are largely known for being directly involved in biodegradation through the production of polyphenol oxidase during composting (Carraro et al., 2014; Rigane et al., 2015; Steinmetz et al., 2019). Overall, few studies have investigated shifts in microbial communities during the cocomposting of OMWS, and most of them could not identify relevant causative factors (Milanović et al., 2019). In our case, even if relative abundancy lacks the functionality aspect, both correlation analysis and culture-dependent assays clearly demonstrated that pH, temperature, EC, PP content and microbial competition are the main factors affecting microbial succession.

At the species level, Pichia manshurica seems to specialize in lipid degradation, and Candida sp. and Galactomyces sp. are likely to be strongly involved in phenol biodegradation (Supplementary Material 2), which agrees with several previous works (Pinedo-Rivilla et al., 2009; Alves et al., 2014; Karimi and Hassanshahian, 2016; Oliveira et al., 2018). The identified bacterial species revealed that Lactobacillus acidipiscis was the agent most correlated with phenolic and lipid biodegradation (Supplementary Material 2), which was previously identified in a work recently done by Uylaşer and Yildiz (2014) and has been identified as a halophilic lactic acid bacterium able to tolerate up to 8% NaCl from black and green olive samples, representing almost 28% of the total identified genus. The same authors showed important lipolytic and pectolytic activity up to 1.09 and 5.29 U ml^–1, respectively, as well as a positive decarboxylase activity. To the best of our knowledge, this species has never been identified as a direct agent of phenol degradation, despite its high potential, which will be very useful in the detoxification of organic pollutants in olive mill waste. L. acidipiscis, first described by Tanasupawat et al. (2000), was found in isolates from fermented fish. It has also been isolated from soy sauce mash (Tanasupawat et al., 2002).