Air-mass back trajectories and the potential origin of the dust
The sampling date, air-mass sources and PM10 concentrations are presented in Table 1. The air-mass back trajectories are also shown in Table S1. The air masses were classified into four groups considering the back trajectory analysis: NW, NE, SW and SE.
Air-mass origins, according to the calculated back trajectories, that coincided with high concentrations of suspended particulate matter in the dust column maps were assigned as the potential dust source. Air-mass sources classified as NE, SW and SE were associated with the three different main dust sources in the Eastern Mediterranean: Iraq and eastern Syria (NE), the Sahara Desert (SW) and the Arabian Peninsula (SE) 68. Two of these sources, the Sahara Desert and the Arabian Peninsula, are among the world’s largest dust sources, accounting for more than 50% of global dust emissions 69-71. The air-mass source classified as NW was associated with low PM10 air masses with little to no contribution from the neighboring dust sources and thus represents mostly local airborne microorganisms.
Community-level multivariate comparison
Differences between the air-mass source, particle size class and PM10 concentration
Initially, we explored which of the different environmental parameters imposed a significant effect on the composition of the sampled airborne communities (both DNA and RNA). According to the PERMANOVA test, the dust source accounted for the highest ratio of variation for both DNA and RNA communities. When compared between the two, it was more pronounced in the RNA community (R2 = 0.39, p < 0.001) than in the DNA community (R2 = 0.24, p < 0.001). The PM10 concentrations for DNA and RNA had values of R2 = 0.09 and 0.07 (p < 0.001), respectively. Finally, the particle size classes for DNA and RNA had values of R2 = 0.06 and 0.04 (p < 0.001), respectively.
Differences between DNA and RNA within each air-mass source
To evaluate whether DNA and RNA communities significantly differed from each other as well as to assess the relative effect of particle size on the variability of bacterial communities, we separated communities according to the air-mass source and then performed a PERMANOVA significance test between the two communities, accounting for the particle size class. The results are presented in Table 2. According to the results, there was a significant difference between the DNA and RNA communities originating from the NE, SW and SE but not in those from the NW, possibly due to the high variance within these communities, along with the low number of samples from the NW. The effect of particle size was significant in the NE, SW and SE communities with similar ratios of variance (R2 = 0.05-0.10, p = 0.025, 0.007 and 0.003) but not in those from the NW (p = 0.062), suggesting a great within-group similarity for different particle size classes in the NW communities.
Diversity and richness of DNA and RNA communities
Effect of the air-mass source
Alpha-diversity analyses, based on richness (observed number of ASVs) and Shannon–Wiener index of diversity, were conducted on DNA and RNA communities by air-mass source. The results are presented in Figure 1.
Similar patterns of richness and diversity between DNA and RNA of different air-mass sources were observed. According to a Kruskal–Wallis test, the source of the air mass significantly affected the diversity and richness of both the DNA and RNA communities (Kruskal–Wallis, p < 0.001). Specifically, according to a Wilcoxon signed-rank test, diversity and richness were significantly higher in air-mass samples that were associated with the dust sources (i.e., NE, SW and SE) than those with NW trajectories. Both the richness and diversity of the RNA and DNA communities of the SE air masses were significantly higher than those of the SW air masses (Table 3).
We also compared the diversity and richness of the DNA and RNA communities of the same air-mass source. According to the Kruskal–Wallis test results, the diversity of DNA differed significantly from that of RNA (p = 0.009), but the observed richness did not (p = 0.429). Comparing the diversity of DNA vs. that of RNA in each air-mass source revealed no significant differences between the DNA and RNA diversity in the NE, SW and SE samples (Wilcoxon signed-rank test, p > 0.05). In the NW samples, the RNA community diversity was significantly lower than the DNA community diversity (p = 0.039), possibly due to the high variance within the sources. All p values are presented in Table S2.
Effect of the particle-size class
When all the samples (i.e., DNA and RNA) were divided by the particulate matter size classes (i.e., fine, intermediate, and coarse), the diversity and richness differed significantly (Kruskal–Wallis, p = 0.037 and 0.003), as shown in Figure 1. Specifically, the richness of the coarse particle size class was significantly higher than that of the intermediate and fine particle size classes (Wilcoxon signed-rank test, p = 0.038 and p = 0.007). All p values are presented in Table S3.
Quantitative PCR
The qPCR results describe the number of 16S ribosomal RNA gene copies (DNA) and transcripts (RNA) per sampled m3, as shown in Figure 2. Overall, the RNA concentrations were significantly higher than the DNA concentrations in all sources except those from the NW (NE p = 0.024; SW p < 0.001; SE p = 0.001; NW p = 0.472). In addition, the RNA concentrations were significantly higher than the DNA concentrations in the intermediate and coarse particle size classes (p = 0.029 and 0.040, respectively) but not in the fine size class (p = 0.195).
The RNA and DNA concentrations differed significantly between the different air-mass sources as well as between the particle size classes. According to Wilcoxon signed-rank tests, the DNA and RNA concentrations were higher in the NE (p < 0.001 and p < 0.001), SW (p < 0.001 and p < 0.001) and SE (p < 0.001 and p < 0.001) dust sources than those in the clear air masses from the NW; however, significant differences in the DNA and RNA concentrations were observed only among the airborne communities between SW and SE (p = 0.020 and p = 0.005, respectively), the latter with higher concentrations.
Bacterial taxa overrepresented in the RNA community
To identify potentially viable bacterial taxa that are significantly more abundant in the RNA community than in the DNA community, we applied a linear mixed model (MaAsLin2), as described in the methods section. The results are presented in Figure 3. Each chart represents clr-transformed mean values of DNA and RNA. Each dot represents a specific ASV. Significant results are colored according to phylum. The black line represents a slope of 1 and visually separates ASVs with higher (dots over the line) and lower (dots under the line) mean RNA abundance than the mean DNA abundance. We assumed that the ASVs that were overrepresented in the RNA community (i.e., significantly high RNA abundance compared to DNA abundance) represented viable bacteria with a current protein synthesis potential 34; hence, these taxa are more likely to ensure a rapid response in a new environment 72, whereas the ASVs overrepresented in the DNA community (i.e., significantly high DNA abundance compared to RNA abundance) likely represented dead bacteria or relic DNA. The taxa that were significantly more abundant in the DNA and RNA communities were the source of variation between the two communities in each air-mass source. The ASVs that did not significantly differ between the RNA and DNA communities (i.e., denoted as “not significant” in Figure 3) represented taxa that were likely viable but, due to sampling and large variability in the population, were not identified as statistically significant in the model.
The number of ASVs that were overrepresented in the RNA community (MaAslin2, Benjamini–Hochberg adjusted p < 0.2) out of the total number of unique ASVs per air-mass source differed among the four air-mass sources: NW (152 of 1 249 ASVs), NE (307 of 3 388 ASVs), SW (2 438 of 4 164 ASVs) and SE (183 of 3 812 ASVs).
Proteobacteria and Firmicutes were the two dominant viable bacterial phyla (i.e., with significantly high RNA abundance) despite evident differences between the different air-mass sources. While Proteobacteria was the most dominant phylum in the NW and NE samples, Firmicutes was the dominant phylum in the SE samples, and in the SW samples, Proteobacteria and Firmicutes were equally dominant. Actinobacteriota and Bacteroidota were the other two dominant phyla in all the air-mass sources but to a lesser extent.
At the family level, taxonomic differences were observed between the different sources; thus, Sphingomonadaceae, Rhodobacteraceae and Pseudomonadaceae were more common in the NW samples, whereas Sphingomonadaceae, Rhodobacteraceae, Hymenobacteraceae and Beijerinckiaceae dominated in the NE samples; Lachnospiraceae, Sphingomonadaceae, Ruminococcaceae, Oscillospiraceae and Rhodobacteraceae dominated in the SW samples; and Lachnospiraceae, Ruminococcaceae, Rhodobacteraceae and Lactobacillaceae dominated in the SE samples. We did not find specific families dominated by dead ASVs (i.e., with significantly high DNA abundance). All the significant ASVs and their taxonomic classifications are presented in Table S4.
Viable bacterial taxa associated with particle size classes
Furthermore, we evaluated the associations between viable bacterial taxa (i.e., with significantly high RNA abundance) and the particle size classes. The results are presented in Figure 4. According to the results, 1 433 viable bacterial ASVs out of a total of 3 080 were associated with at least one of the particle size classes in all air-mass sources (MaAslin2, Benjamini–Hochberg adjusted p < 0.2). Many of these viable bacterial ASVs were associated with only the coarse particle size class (961 ASVs), followed by ASVs associated with both the intermediate and coarse size classes (369 ASVs) and then by the intermediate (89 ASVs) and fine (14 ASVs) size classes. Among the four different air-mass sources, more ASVs in the SW samples were found to be significantly associated with at least one of the particle size classes (1261 ASVs), followed by the NE (87 ASVs), NW (54 ASVs) and SE (31 ASVs) samples.
The viable bacterial families Bacteroidaceae, Hymenobacteraceae, Lachnospiraceae, Lactobacillaceae, Oscillospiraceae, Rhodobacteraceae, Ruminococcaceae and Sphingomonadaceae were associated with the coarse particle size class, while Acetobacteraceae, Micrococcaceae and Streptomycetaceae were associated with the fine particle size class.