Microbial community dynamics in phyto-thermotherapy baths viewed through next generation sequencing and metabolomics approach

Temperature and pH

Figure 1 shows the temperature trend during the July, August and October PTBs. In the first 35 h, all the PTBs showed similar temperatures rising from 26 to 35 °C – 39 °C (Fig. 1). In the following hours, the temperature showed different trends according to the month. In July, the temperature was stable for the first four days (96 h) and then rose, reaching the max. temperature of 51 °C. In August, the temperature was stable for the first 55 h and then rose reaching the max. temperature of 61 °C. Finally, in October, after the first 35 h, the temperature rose immediately reaching 50 °C, and then was stable reaching the max. of 64.8 °C.

The pH (Table 1), in the first two days, was in a range between 6.2 and 6.5 with the exception of October when pH sowed a mean value of 7.4. After d5, the pH increased significantly and stabilized in a range between 7.4 and 7.8.

Microbial counts in herbs during PT process

The counts of viable total aerobic, mesophilic and thermophilic anaerobic bacteria, enterobacteria, yeasts and moulds in July, August and October at day 0, d2, d3, d5 and d7 are shown in Table 1. The plate counts showed no significant difference for depth of sampling (p value > 0.05; data not shown). The aerobic bacteria were always high at day 0 (6.7, 7.7 and 9.1 Log CFU g⁻¹ in July, August and October respectively). In July and August, they significantly rose until d7, reaching similar amounts (8.7 and 8.8 Log CFU g⁻¹ respectively). The mesophilic and thermophilic lactic acid bacteria (LAB) were very low in July and August at day 0 when they were present in traces or not detected (Table 1). Mesophilic LAB rose from d2 to d7 with a similar trend to aerobic bacteria reaching 5.9 Log CFU g⁻¹. By contrast, in October they showed different trend: as aerobic bacteria, mesophilic LAB counts were very high at day 0 and stable without significant differences until d3 and then significantly decreased until d7 reaching 4.9 Log CFU g⁻¹. Counts of thermophilic LAB showed similar trends in all the three months: they rose until d7 reaching similar amount in July, August and October (5.2, 5.7 and 5.1 Log CFU g⁻¹ respectively). Enterobacteria counts were lower in herbs at day 0 (3.8, 4.3 and 5.0 Log CFU g⁻¹ in July, August and October respectively) and then significantly rose until d7 reaching 6.2, 5.7 and 6.7 Log CFU g⁻¹ in July, August and October respectively. In July, yeasts and moulds in the first two days were not detected or present in traces, then reached their highest value at d3 and significantly decreased until d7. By contrast, their counts were very high in both August and October; in particular, moulds counts trends were similar: they significantly decreased until d7 to 3.2 and 6.4 Log CFU g⁻¹ in August and October respectively.

Characteristics of the sequencing data

The DNA extracted from the 90 PTB samples had been all successfully amplified. After merging and quality trimming the raw data, 2,980,511 reads for bacteria and 1,227,092 reads for fungi remained for subsequent analysis (Table S1). After alignment, the remaining Operational Taxonomy Units (OTU) had been clustered at a 3% of distance.

Bacteria and fungi: alpha diversity

The number of OTUs and the Shannon diversity index were determined using QIIME2 at 97% similarity levels (Table 2), in order to analyse the bacterial and fungi community richness in samples obtained during the PT process. Regarding the sample position (5 and 40 cm depth), there was no significant difference in both observed OTUs number and Shannon diversity index for bacteria and fungi communities (p value > 0.05). It is worth noting that the degree of bacterial diversity was significantly higher in July and August than in October samples, by contrast, the degree of fungal diversity was significantly higher in July and October than in August samples (Shannon diversity index in Table 2).

The variation in number of OTUs and Shannon diversity index over time indicates highest microbial diversity at day 0 and d2, and a lower microbial diversity at d5 and d7.

Bacteria and fungi: beta diversity

The PCoA of UniFrac and Jaccard metric indicated clear clustering of both bacterial and fungi communities according to the different PTB days (Fig. 2a,b).

Bacterial and fungi communities were more phylogenetically dissimilar between successional days (day 0, d2, d3, d5 and d7) or months (July, August and October) than between position of sampling (5 and 40 cm). The bacterial Weighted UniFrac PCoA (Fig. 2a, total variation explained: 52.09%) and the fungal Jaccards PCoA (Fig. 2b, total variation explained: 32.83%) revealed a clearer picture of the similarities across different days. The PCoA plots emphasized the similarities of the bacterial and fungi communities in the PTB at d2 and d3 when compared with d5 and d7.

These results were supported by the PERMANOVA statistical analysis (Table 3). The differences between position of sampling (5 and 40 cm) were not significant both for bacteria and fungi communities as the p values were 0.19 for bacteria and 0.27 for fungi community respectively. The differences among months of sampling were significant comparing July and August with October for both bacteria and fungi communities as well as the differences among microbial communities through time. The pairwise comparison (Table 3) clearly showed two significantly different stages during the PTB after day 0: the first stage including d2 and d3 (1st stage), and the second stage including d5 and d7 (2nd stage). The time effect on microbial composition (pseudo-F value) was showing a growing trend with the proceeding of the sampling days (Table 3). The bacterial pseudo-F values were smaller and smaller with progression of the days; by contrast, fungi pseudo-F values were higher. This means that the bacteria became more similar and the fungi more different in the samples from the three piles studied as the process progressed.

Bacterial community structure

Eleven prokaryotic phyla were found in the 90 samples accounting for the total bacterial community. The predominant phylum across all bacterial communities was the Proteobacteria, accounting for 33%–83% of the OTUs in each time and month of the PTB (Fig. 3). The trend of Proteobacteria was similar in all the months with different amount of presence: at day 0 and d2 Proteobacteria abundance was between 61% and 85%; at d3, showed a slight decrease (between 58% and 72%) and reached the lower values at d5 and d7 (between 33% and 65%). Twenty-six bacterial phylotypes were found dominant across all samples, accounting for the 90% of the total bacterial community. Of these 26 phylotypes (in Fig. 4), 10 belonged to Proteobacteria. Erwinia was the most abundant genus in this phylum and reached the higher presence at d2 and d3. Other frequently sequenced genera included Acinetobacter, Pseudomonas, mainly found at day 0, and unclassified genera belonging to Xanthomonadaceae family.

Of the Alphaproteobacteria, the genera Agrobacterium, Methylobacterium, Sphingomonas and unclassified genera of the family Beijerinckiaceae were the most abundant and dominant at day 0. All these phylotypes showed a lowering trend in time decreasing from day 0 to d7 with the only exceptions of the Beijerinckiaceae that were higher at d7 than at day 0.

Of the Betaproteobacteria, unclassified genera belonging to Alcaligenaceae and Oxalobacteraceae were the most abundant families. Their relative abundances were constant along the time and showed higher differences among the months than the sampling days (average values of relative abundance were 2.4%, 4.5% and 0.5% in July, August and October respectively).

Firmicutes accounted for 12%–38% of the OTUs from d2 to d7 of the PT process. At day 0, Firmicutes were present only in October’s samples at 7%, then increased at d2 and remained stable until the end of the process (Fig. 3). Overall, there were 10 Firmicutes in the 26 most abundant phylotypes found during the PTBs (Fig. 4). Most of these were thermophilic bacteria such as Bacillus, Thermoactinomycetaceae, Brevibacillus and Paenibacillus⁶. These themophilic phylotypes had never been detected at day 0, appeared at d2 and reached their higher values at d5 when their presence accounted for 29.9%, 19.4% and 8.5% in July, August and October respectively.

The Bacteroidetes constituted another dominant phylum detected in all the samples (Fig. 3). Bacteroidetes abundance at d0 was in the range of 1.4%–17.0%, decreased at d2, d3 and d5 to a range of 0.7%–8.1% and reached the highest values at d7 (between 8.1% and 21.2%). The most abundant genera belonging to this phylum were Flavobacterium and Sphingobacterium.

The abundance of Actinobacteria was 1.0%–3.3% at day 0 and increased until the end of the process reaching a maximum of 7.5–36.6% relative abundances respectively (Fig. 3). The most abundant phylotypes in the Actinobacteria phylum were Microbacteriaceae and Streptomyces (Fig. 4); in particular, Streptomyces was one of the genera dominating bacterial community biodiversity at d5 and d7.

Cyanobacteria were found at relative abundances higher than 1% only in July and August samples at day 0 and d2, with a maximum at day 0 (16.9% and 4.5% relative abundances in June and July samples respectively). After d2, Cyanobacteria decreased until the end of the process (Fig. 3). The relative abundance of Verrucomicrobia was 0.35%–2.8% at d0 and remained constant during the whole PTB process (Fig. 3). Chtoniobacteraceae was the most abundant bacterial family of the Verrucomicrobia phylum.

Further bacterial phyla had always been found at very low relative abundances (never higher than 1.0%, Fig. 3).

Fungal community structure

Before the PTB started, the fungal community in the herbs (day 0 in Fig. 5), was dominated by Mycosphaerellaceaes (Mycosphaerella, Ramularia and Zymoseptoria genera), representing the 21.5%, 21.4% and 15.0% of the total in July, August and October respectively, and Bulleribasidiaceaes (Vishniacozyma, and Dioszegia genera) representing the 22.3%, 32.4% and 23.4% of the total in July, August and October respectively. Other fungal taxonomic groups, mainly belonging to the Ascomycota phylum, were detected in lower relative abundance (lower than 10%). After two days (d2), the Aspergillaceae family was emerging, mainly constituted by the Aspergillus genus with traces of Penicillium in 13 out of the 90 samples. Aspergillaceae dramatically increased from day 0 (always less than 1%) to d2 in July and August trials (10.6% and 24.9% respectively), and after d3 they became the most dominant fungi (26.6% and 83.2% respectively). By October, Bullerobasidiaceae was always the dominant fungal family at d2 and d3. After five days (d5 in Fig. 5), Bulleribasidiaceaes (lower than 10.6%) and Mycosphaerellaceae (lower than 6.5%) relative abundances decreased sharply. Aspergillaceae kept their rising trend, remaining the dominant fungal family in July and August trials (47.9% and 56.6% respectively), but they represented only the 11.4% of the fungal relative abundance in October. Other thermophilic species emerged in July and August: the Trichocomaceae mainly constituted by Thermomyces lanuginosus whose relative abundance was never higher than 2.1% in the first three days and then suddenly increased to 32.8% and 23.2% in July and August respectively.

After seven days (d7 in Fig. 5), the fungal community was totally dominated by Aspergillaceae (46.6%, 69.2% and 35.9% in July, August and October respectively) and Thermomyces lanuginosus (44.2%, 23.3% and 21.6% in July, August and October respectively). None of the OTUs was predominant throughout all samples.

Volatiles organic compounds (VOCs) released during the PTBs

After raw GCxGC–MS data deconvolution and pre-processing, the three dataset (July, August and October) consisted of 722, 1105 and 815 volatiles respectively. As already reported by Narduzzi et al.⁷, the majority of the VOCs are not in common among the months. The identified volatiles through all the three months’ datasets were matched using their InchiKey, and produced a table consisting of 295 VOCs present in all the months (Table S2). As shown in the top part of the Fig. 6, there is a cluster of 34 compounds that are the most representative of all PTB samples because contributing to the 85% of the total VOCs mass emissions. In the heatmap Fig. 6, the samples from the same month clustered together. Moreover, within each month, the samples split in two different clusters according to the stages already identified in the microbial analysis. The first cluster is composed by the samples of d1, d2 and d3 (1st stage) and the second cluster by the samples of d4, d5, and d6 (2nd stage). Looking at the differences in the days within each batch, the d1, d2 and d3 samples were richer in aliphatic hydrocarbons (heneicosane, hexadecane, tetradecane and 3-methyltridecane), alpha-terpineol, and estragole. By contrast, the d4, d5 and d6 samples were richer in 1-methylnaphthalene, nonanal, 2-nonanone, 3-octanol, m-xylene, 2,6-dimethylheptadecane and 2-ethyl-hexanol.

Source: Ecology - nature.com