The copy numbers for 16S and ITS1 rRNA, and the sequencing depth for all samples are presented in Supplementary File 3 (qPCR data, Sequencing Rarefaction Curves). An average of 14,265.25 reads per housefly sample for the V4 16SrRNA and 16,149.4 reads per housefly sample for the ITS1 were retained after quality filtering. After quality filtering of the egg-laying substrate samples, an average of 10,371.75 reads were retained per sample for the V4 16SrRNA, and an average of 25,479.75 reads were retained per sample for the ITS1 region. The extracted DNA from newly emerged adult houseflies of the Spanish laboratory strain (12 samples in total, newly emerged adults, three replicates from four generations, strain SP100) returned a low copy number for the fungal ITS1 (qPCR data, Supplementary File 3) and a low number of acquired sequencing reads; they were therefore omitted from any further analysis of the fungal microbiota. In addition, the mitochondrial COI phylogeny showed that the Dutch wild-caught strain and the Dutch laboratory strain, which were sampled from the same locality at different times, are in close proximity and form a separate clade from the Spanish lab strain phylotypes (Supplementary File 2).
The housefly microbiota alpha-diversity is determined by sampling environment
Absolute richness (number of ASVs), Shannon index, and Phylogenetic diversity for all housefly strains and developmental stages are shown in Fig. 1. The highest bacterial alpha diversity was observed for the wild-caught housefly population GK0. Strain was an important factor for separating Shannon biodiversity levels both for newly emerged (F = 4.37, P < 0.05) and 3-day-old adults (F = 14.70, P < 0.001), as shown by single-factor ANOVA. Pairwise comparisons between strains showed that GK0 was significantly different from the Dutch laboratory strain GK50 (newly emerged adults: F = 4.83, P < 0.05, 3-day-old adults: F = 12.64, P < 0.01) and the Spanish laboratory strain SP100 (newly emerged adults: F = 5.79, P < 0.05, 3-day-old adults: F = 18.49, P < 0.001), in terms of bacterial alpha diversity. However, the adults of the two laboratory strains (GK50 and SP100) showed no significant difference (P > 0.05). Comparisons for fungal alpha diversity between all strains found similar results with the respective comparisons for the bacterial alpha diversity for the newly emerged adults (F = 4.55, P < 0.05) and the 3-day-old adults (F = 29.11, P < 0.001). The GK0 strain harbored a more diverse fungal community than the GK50 strain (F = 6.43, P < 0.05 and F = 65.00, P < 0.001, for the newly emerged and the 3-day-old adults, respectively) and the Spanish laboratory strain SP100 (F = 12.23, P < 0.05, for the 3-day-old adults).
Bacterial and fungal alpha diversity of the microbial communities from developmental stages of three Musca domestica strains, estimated with three diversity metrics: the number of Amplicon Sequence Variants, Shannon index, and Faith’s Phylogenetic Diversity. The line in each boxplot represents the median of all values, with the upper limit of the box being the first and the lower limit of the box being the third quartile. Each boxplot represents all samples of one housefly strain (GK0, GK50, or SP100) per developmental stage (3 day old adults, eggs, larvae, pupae, 1 day old adults), i.e. three biological replicates for four generations. For the GK0 strain, one separate boxplot represents the wild-caught adults immediately after collection (adults from the farm). GK0: field population, the Netherlands; GK50: laboratory population originally collected in the Netherlands; SP100: laboratory population originally collected in Spain.
In general, adult age was a significant factor when comparing the bacterial alpha diversity of newly emerged and 3-day old adults within the Dutch laboratory strain (F = 19.90, P < 0.001) and the Spanish laboratory strain (F = 97.18, P < 0.001), with newly emerged adults showing more diverse communities. On the contrary, bacterial alpha diversity did not differ between newly emerged and 3-day old adults for the wild-caught strain. An opposite pattern was found for the fungal alpha diversity, where only the wild-caught strain (GK0) showed a significant difference between 1-day old and 3-day old adults (F = 16.00, P < 0.001).
Similar to the housefly adults, the larvae from the GK0 strain harbored more diverse bacterial communities, with the housefly strain being an important factor for the overall comparison of all larvae (F = 8.70, P < 0.001). The larval bacterial alpha diversity from this strain was significantly higher than GK50 and SP100 (F = 11.02, P < 0.01 and F = 20.26, P < 0.001, respectively). In contrast, the two laboratory strains did not differ from each other (P > 0.05). The same pattern was found for the fungal alpha diversity for larvae (all larvae: F = 9.22, P < 0.001, GK0 with GK50: F = 49.20, P < 0.001, GK0 with SP100: F = 12.18, P < 0.05 and GK50 with SP100: P > 0.05) and pupae (all pupae: F = 13.98, P < 0.001, GK0 with GK50: F = 32.24, P < 0.001, GK0 with SP100: F = 22.99, P < 0.001 and GK50 with SP100: P > 0.05). However, we did not find similar results for the bacterial diversity of the pupae, for which the Dutch wild and the laboratory strains showed similar levels (P > 0.05).
Bacterial but not fungal alpha diversity is reduced after introduction into the laboratory
Bacterial and fungal alpha diversities were estimated separately for samples from each of the four generations (Fig. 2). For the GK0 strain, the absolute numbers of ASVs, Shannon Diversity, and Phylogenetic Diversity decreased for the bacterial communities derived from older adults throughout the four sampled generations (Fig. 2). The first generation of the strain consisted of the adults caught in the cattle farm, whereas the following generations were reared under laboratory conditions. Linear model testing for all three metrics returned significant results for this sample subset (ASVs: R2 = 0.46, P < 0.05; Shannon: R2 = 0.32, P < 0.05; Phylogenetic diversity: R2 = 0.39, P < 0.05), indicating a decline in alpha diversity over successive generations. No other combination of strain and developmental stage returned statistically significant results for the three alpha diversity metrics, indicating alpha diversity remained similar across generations. For the fungal communities, fungal alpha diversity did not decrease with the transfer of the GK0 adults into the laboratory; it remained similar for all strains and life stages across the four generations.
Bacterial and fungal alpha diversity from developmental stages of three Musca domestica strains over four generations, estimated with three diversity metrics: richness represented by the number of different Amplicon Sequence Variants, Shannon index, and Faith’s Phylogenetic Diversity. Each subplot represents a combination of one housefly strain (GK0, GK50, or SP100) and one developmental stage (3 day-old adults, eggs, larvae, pupae, 1-day old adults). Lines show the linear regression for each housefly strain alpha diversity values across generations. The shaded field around the lines stands for the standard error of the linear regression. Alpha diversity decreased significantly only for the bacterial communities derived from older adults of the GK0 strain, as the red asterisk suggests.
Beta diversity of housefly associated microbial communities is more similar between laboratory strains
Whether the housefly strain had been kept in the laboratory or was wild-caught served as a more significant explanatory factor for the shaping of the bacterial and fungal microbiota composition than the country of origin (Netherlands/Spain), for all developmental stages (Permutational multivariate analysis of variance, Table 1). Permutation test for homogeneity of dispersions showed that the bacterial communities from all developmental stages had similar beta dispersion values (average distance to the group centroid) for each of the three housefly strains (P > 0.05). The results for the fungal communities were similar, with the exception of pupae. More specifically, the Spanish laboratory strain showed a significantly lower beta-dispersion than the respective Dutch laboratory strain (P < 0.01**), thus lowering the significance of the respective PERMAVONA result (Table 1). To further investigate these factors potentially explaining the observed microbial beta diversity, we tested them as predictors for assigning samples to groups. Again, the sampling environment was a more accurate factor in predicting community structure for bacterial and fungal microbiota, since the assignment errors were lower than those for assignment according to the country of origin. Especially for the fungal communities, both larvae and adults were more accurately assigned to the correct sampling environment than to the correct country of origin (Random forest classification, Table 2). Furthermore, for eggs, larvae, and pupae, the two laboratory strains were similar, whereas the wild-caught farm strain was distinct from the remainder of the samples (NMDS ordination, Fig. 3). Adults from the two laboratory strains were also more similar; however, the clustering patterns were less clear. Indeed, only a small proportion of the total adult microbiota variance could be explained by rearing history in the laboratory (Table 1).
(a) Bacterial communities for each developmental stage of three housefly strains based on weighted Unifrac. NMDS stresses: 9.05e-05 (eggs), 0.09 (larvae), 0.08 (adults),0.06 (pupae) (b) Fungal communities for each developmental stage based on weighted Unifrac. NMDS stresses: 9.91e-05 (eggs), 0.07 (larvae), 0.06 (pupae), 0.15 (adults).
To determine whether the culturing of the strains in the laboratory or the transfer from field to laboratory lead to substantial community shifts across generations, we evaluated the microbial turnover over generations, for each strain and stage (Fig. 4). Beta-diversity patterns showed no significant shifts between housefly generations for the GK0 strain, similarly to the other two laboratory strains.
Weighted UniFrac pairwise distances for all housefly developmental stages of three strains over four consecutive generations as a metric for beta diversity turnover for bacterial and fungal communities. Beta-diversity patterns showed no significant shifts between housefly generations. Lines show the linear regression for each housefly strain beta diversity values across generations. The shaded field around the lines stands for the standard error of the linear regression. Beta diversity did not shift significantly across generations for any housefly strain.
Beta diversity analysis of the housefly-associated microbial communities showed that the fungal communities were less consistent than the bacterial communities within the same developmental stage, with the revealed bacterial communities being more homogeneous within the same stage. More specifically, the beta dispersion value (average distance to the group centroid) for the bacterial communities retrieved from housefly eggs was 0.20 compared to 0.41 for fungal communities. Microbial communities of larvae and pupae were also more homogeneous in bacterial composition (beta dispersion: 0.15 and 0.10, respectively) than in terms of fungal composition (beta dispersion: 0.29 and 0.22, respectively). Finally, fungal communities varied more in the case of newly emerged adults (beta dispersion: 0.40) and 3-day old adults (beta dispersion: 0.36) than bacterial communities from newly emerged adults (beta dispersion: 0.19) and 3-day old adults (beta dispersion: 0.20).
Upon correlating the distance matrices of both the bacterial and the fungal dataset, we detected a weak co-variance of the bacterial and fungal communities within the whole dataset (r = 0.27, P < 0.001). To investigate whether this significant correlation was an effect of strain or life cycle stage specificity, we ran the same tests within subsets of all samples. We found statistically significant correlations, both when we investigated each strain and each stage of the housefly life cycle, separately (Table 3). When we conducted the same tests within each developmental stage of the same housefly strain, we detected significant correlations for all subsets except for the microbial communities associated with eggs from all strains and pupae from the SP100 strain (Table 3).
Microbial composition of the laboratory strains is simpler, and a set of recurring microbial phylotypes can be distinguished
Two bacterial genera, Providencia and Myroides, were detected in all samples from larvae and pupae of all housefly strains and generations (Fig. 5). These same taxa were not detected in the samples from the provided substrate (Microbial composition, Supplementary File 3). Furthermore, the genus Morganella was detected mainly from samples of the wild strain (GK0) larvae and pupae. In contrast, the genus Proteus was detected in larvae and pupae of the laboratory strains (Fig. 5). Again, none of these bacterial taxa were found in the fresh substrate we provided to the flies. Finally, the bacterial family of Enterobacteriaceae was more dominant in the adult fly guts of the two laboratory housefly strains (Fig. 5), along with the genera Pseudomonas and Acinetobacter. Both laboratory strains showed similar patterns of bacterial composition succession during the housefly life cycle, across the four generations. In contrast, the GK0 strain showed a more complex microbiota structure in line with its revealed higher alpha-diversity.
Microbial composition of developmental stages of three housefly strains over four generations. Each separate plot stands for all samples (three replicates for every developmental stage) of the same housefly strain and the same generation. Each circle represents the housefly life cycle from the egg stage to the adult stage. Lines stand for the relative abundance of bacterial (top) or fungal (down) taxa. Bacterial and fungal taxa are included if the relative abundance of the phylotypes assigned to them are detected at a percentage of at least 5% in the whole dataset.
Two yeasts assigned to the fungal taxa Trichosporon sp. and Candida tropicalis, were prevalent in all larvae and pupae of our dataset (Fig. 5). The same taxa were detected in all four batches of the feed, which was provided as the egg-laying substrate (Microbial composition, Supplementary File 3), indicating that the flies acquire them from their environment. Sequences assigned to the fungal order Saccharomycetales were detected in all batches of the provided substrate as well. However, it was mainly found in flies from the wild-caught strain and was not dominant in flies from the laboratory strains (Microbial composition, Supplementary File 3).
Source: Ecology - nature.com
