Sampling of shared-bicycles and surrounding air
We collected samples (n = 130) from several shared bicycles and air in five selected locations (central, eastern, western, southern and northern areas) of a metropolis (Chengdu, China) for fungal community analysis. The central location in Chengdu was near general hospitals, and the eastern location was in a residential area; in contrast, the western location was near busy commercial streets. The southern and northern locations were close to a senior high school and metro station, respectively. A total of 110 qualified DNA samples were obtained, including samples from shared bicycles (n = 98) and the air around them (A group, n = 12). Shared-bicycle samples were collected from the handles (H group, n = 50) and saddles (S group, n = 48).
Eligible DNA extracted from these samples was subjected to ITS sequencing analysis. After stringent-quality sequencing and filtering, we obtained a total of 1,321,604 clean reads (82% of the total 1,608,556 raw reads) from the 110 samples. Then, 3724 unique OTUs were clustered, and 717 genera and 49 classes from 14 fungal phyla were identified by the UNITE database; 1·1% of the OTUs could not be matched to taxa in the database.
Fungal community composition
The fungal community constituents of the three groups were observed at the phylum level (Fig. 1A). The relative abundances of Ascomycota and Basidiomycota were markedly dominant, at 75·7% and 15·9% on handles, 78·4% and 13·8% on saddles, and 60·1% and 19·0% in air samples, respectively. An unclassified fungal phylum with a relative abundance of 20·0% was identified in the air group. The similarity between the saddles and handle samples was reconfirmed by the weighted UniFrac UPGMA distance matrix.
Aspergillaceae had the highest abundance among the top ten families in the shared-bicycle samples (handles and saddles) (Fig. 1B), followed by Aureobasidiaceae in handle samples and unclassified fungi in saddle samples. These results differed from those of the air samples, which were characterized by a relatively high abundance of unclassified fungi and Pleosporaceae, which accounted for approximately 19·7% and 12·2%, respectively.
The relative abundances of the top 30 fungal constituents at the genus level in the three groups also varied (Fig. 1C, see Additional Fig. 1). Excluding unclassified fungi, five genera, namely, Aspergillus, Candida, Alternaria, Cladosporium and Erythrobasidium, were enriched in the handle samples. Aspergillus and Cladosporium also dominated the mycobiota in the saddle samples. The fungal communities in air samples were more homogeneous; Alternaria and Cladosporium had the highest abundances, followed by Irpex and Erythrobasidium.
At the species level, the absolute abundance of Aspergillus pseudoglaucus was overrepresented in the handle and saddle samples. Alternaria rhadina was the most abundant species in air, followed by Cladosporium tenuissimum and Erythrobasidium hasegawianum, which were highly abundant in all three groups. In addition, Aureobasidium melanogenum and Filobasidium magnum were the main species on handles. Cladosporium tenuissimum, Alternaria rhadina, and Kazachstania pintolopesii were the main species on saddles. Malassezia sp., Candida tropicalis, Malassezia globose and Trichomeriaceae sp. were also included in the top 30 most abundant species (Fig. 1D).
Fungal alpha diversity
The highest community richness was found on saddles, and the lowest richness was found in air (Fig. 2A; see Additional Figs. 2 and 3). Species were more similar between handles and saddles, with three and five times the number of endemic OTUs than air, respectively. The proportions of OTUs shared by the air and bicycle surfaces were 19·9% (615/3109), 26·5% (529/1999) and 23·3% (541/2320) for the combined surfaces, handles, and saddles, respectively. Combined with richness and evenness, the Shannon index (Fig. 2B) indicated a more complex fungal community diversity in air than on shared bicycles. However, there were no significant differences in the species richness and Shannon diversity indices among the three groups.
Fungal beta diversity
In the PCoA based on weighted UniFrac distances (Fig. 3A), the OTUs from handle and saddle samples tended to cluster together and were more distant from those from air samples. This indicates that the fungal community structures on handles and saddles are highly similar to each other and different from that in air. In addition to confirming the above conclusion, ANOSIM analysis (Fig. 3B-D) demonstrated a significant differences in the fungal community structure between the bicycle surfaces (R > 0).
Comparisons between groups
After the above diversity analysis, we focused on differential abundances in fungal communities among the three groups. The t-test and MetaStat analysis (Fig. 4A-C; see Additional Fig. 4) between groups indicated significant differences in community structure between not only the shared bicycles and air samples but also the handles and saddles at the genus and species levels, even though the diversity on the handles and saddles had greater similarity.
Compared with the air samples, the shared bicycle (handle and saddle) samples showed significant enrichment of the genera Aspergillus, Kazachstania, Candida and Malassezia (Fig. 4A-B, P < 0·05). It is worth noting that certain species, such as Aspergillus pseudoglaucus, Kazachstania pintolopesii, Candida tropicalis, and Malassezia globose, are likely responsible for the differences observed in abundances at the genus level (see Additional Fig. 4).
The abundances of Aureobasidium and Filobasidium were more enriched in the handle group than in both the saddle and air groups, while Cladosporium was more enriched on saddles (Fig. 4A-C, see Additional Fig. 4, P < 0·05). Similarly, the differences in these genera are likely primarily attributed to the high relative abundance of Aureobasidium melanogenum and Filobasidium magnum on handles and Cladosporium tenuissimum on saddles.
The evolutionary branches of the LEfSe analysis indicated that the fungal markers with significant differences among the three groups originated from 8 genera of 3 classes (Saccharomycetales, Eurotiomycetes, and Sordariomycetes) (Fig. 4D). The 8 genera were Kazachstania and Candida in the saddle samples; Aspergillus, Aureobasidium and Filobasidium in the handle samples; and Alternaria, Didymella and an unclassified Verrucariaceae in the air samples (Fig. 4D). At the species level, more complete identification is needed.
Fungal trophic modes and functional guilds
FUNGuild was used to analyze the trophic modes and functional guilds of the fungal communities. Saprotrophs had an influential relative abundance of approximately 50% in bicycle surface specimens and approximately 1/8 in air specimens (see Additional Fig. 5). The heatmap representing functional taxonomy shows a high abundance in air, which is mostly nonoverlapping with that on shared bicycles (Fig. 5A). In the handle and saddle samples, the abundance of animal pathogens was obviously higher than that in the air samples (Fig. 5A). Compared with those in the saddle samples, more animal pathogens and fewer endophyte-plant pathogens were found in the handle samples (P < 0·05, Fig. 5B-D).
Fungal co-occurrence networks
Correlation (ρ = 0·6) and P values (P = 0·01) were established. The fungal co-occurrence networks were obviously distinct between the shared-bicycle surfaces (Fig. 6A and B). Compared with the saddle specimens, the handle specimens had a larger clustering coefficient (CC: 50·5% vs. 48·8%) and a smaller average path length (APL: 2·18 vs. 3·03), indicating that the fungal communities on the handles were more closely connected than those on the saddles; the lower species richness was reconfirmed by the APL combined with a 0·2% smaller graph density.
Considering the statistical parameters, seven genera were enriched on handles. Two of the core genera, Aspergillus and Kazachstania, were mutually positively related, and both were associated with two subdominant genera (Candida and Thermomyces). Aspergillus was negatively correlated with four other enriched genera (unclassified fungi, Alternaria, Aureobasidium, and unclassified Trichomeriaceae) and two subdominant genera (Periconia and Didymella). Kazachstania had an inverse correlation with the dominant genera Alternaria and Cladosporium and unclassified fungi. However, lower relative abundances and greater mutual distances were present among genera of phylum Basidiomycota.
Eight genera were enriched on saddles. Among them, Aspergillus showed a negative relationship with Didymella, Cladosporium, and an unclassified fungi and a positive relationship with Kazachstania and the subdominant genus Thermomyces. Cladosporium was negatively correlated with Alternaria and an unclassified fungi. Erythrobasidium, as a highlighted genus of the phylum Basidiomycota, was positively correlated with Aureobasidium and negatively correlated with Thermomyces.