Environment and diet shape the geography-specificDrosophila melanogastermicrobiota composition

latitude, temperature, photoperiod, neutral theory, lactic acid bacteria, local 13 adaptation, life history, rapid adaptation 14 15 Running Head: The geography-specific D. melanogaster microbiota 16 17 # Address correspondence to John M. Chaston, [email protected], 801-422-4553, 701 E 18 University Pkwy, 4105 LSB, Brigham Young University, Provo, Utah, USA, 84602 19 20 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint

21 Geographic and environmental variation in the animal microbiota can be directly linked to the 22 evolution and wild fitness of their hosts but often appears to be disordered. Here, we sought to 23 better understand patterns that underlie wild variation in the microbiota composition of 24 Drosophila melanogaster. First, environmental temperature predicted geographic variation in fly 25 microbial communities better than latitude did. The microbiota also differed between wild flies 26 and their diets, supporting previous conclusions that the fly microbiota is not merely a reflection 27 of diet. Flies feeding on different diets varied significantly in their microbiota composition, and 28 flies sampled from individual apples were exceptionally depauperate for the Lactic Acid Bacteria 29 (LAB), a major bacterial group in wild and laboratory flies. However, flies bore significantly 30 more LAB when sampled from other fruits or compost piles. Follow-up analyses revealed that 31 LAB abundance in the flies uniquely responds to fruit decomposition, whereas other microbiota 32 members better indicate temporal seasonal progression. Finally, we show that diet-dependent 33 variation in the fly microbiota is associated with phenotypic differentiation of fly lines collected 34 in a single orchard. These last findings link covariation between the flies’ dietary history, 35 microbiota composition, and genetic variation across relatively small (single-orchard) 36 landscapes, reinforcing the critical role that environment-dependent variation in microbiota 37 composition can play in local adaptation and genomic differentiation of a model animal host. 38 39 SIGNIFICANCE STATEMENT 40 The microbial communities of animals influence their hosts’ evolution and wild fitness, but it is 41 hard to predict and explain how the microbiota varies in wild animals. Here, we describe that the 42 microbiota composition of wild Drosophila melanogaster can be ordered by temperature, 43 humidity, geographic distance, diet decomposition, and diet type. We show how these 44 determinants of microbiota variation can help explain lactic acid bacteria (LAB) abundance in 45 the flies, including the rarity of LAB in some previous studies. Finally, we show that wild fly 46 phenotypes segregate with the flies’ diet and microbiota composition, illuminating links between 47 the microbiota and host evolution. Together, these findings help explain how variation in 48 microbiota compositions can shape an animal’s life history. 49 50 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint

51 Animal-associated microorganisms (‘microbiota’) can profoundly impact the behavior, 52 physiology, and evolution of their hosts. The types and abundances of microorganisms in wild 53 animals, especially animals colonized by horizontally-acquired poly-species communities, often 54 vary dramatically in response to numerous factors, including diet, time, and space. Because the 55 types and abundances of the microorganisms can determine specific host traits, these changes in 56 microbiota composition can influence adaptation of their hosts (1). Here, we sought to better 57 understand the causes of variation in the microbiota composition of animals in the wild by 58 studying the microbiota of the fruit fly Drosophila melanogaster, a host with a relatively low- 59 diversity and low-abundance microbial community that is a model for understanding patterns of 60 microbial community assembly (2), microbe-microbe interactions (3, 4), and host-microbe 61 interactions, including in a wild setting (1, 5-7). 62 63 As in many other animals with a horizontally-acquired gut microbiota, the D. melanogaster 64 microbiota composition is driven by the diet, the host, and other members of the community. In 65 the wild or the laboratory, a single fly typically bears several hundred thousand bacterial cells 66 from fewer than 100 species, most of which are consolidated in fewer than 10 highly abundant 67 species (8-17). Flies are typically dominated by Acetic Acid Bacteria (AAB), Lactic Acid 68 Bacteria (LAB), or bacteria from the order Enterobacteriaceae (5, 6, 8-10, 12, 15, 18-20), and the 69 types and abundances of these microorganisms are typically influenced by the same factors as 70 influence the gut microbiota in mammals: diet (15, 16, 20), host genotype (21), and individuality, 71 including vial effects for flies reared in the same containers (16, 22). In the laboratory, members 72 of the Acetobacter, Lentilactobacillus, and Lactoplantibacillus genera have been commonly used 73 as representative isolates in the flies (23). Studies with bacteria from these and other genera have 74 revealed that some, but not all bacterial strains persistently colonize the flies for longer than the 75 bulk passage of diet through the gut and that there are specific foregut niches for some of these 76 persistently colonizing bacteria (22, 24-28). Also, the types and abundances of the microbes in 77 the flies are distinct from those in the diet (16), dramatically influenced by host genetic selection 78 and microbe-microbe interactions (3, 21, 23, 29, 30), and typically best defined by neutral 79 community assembly rules, suggesting strong roles for ecological drift and passive dispersal (6, 80 12). Thus, a substantial body of work has established key drivers of microbial community 81 structure. 82 83 The fruit fly D. melanogaster is an established model for studying geographic variation and the 84 microbiota. The geographic life history of D. melanogaster may be the most extensively studied 85 of any animal on the planet, with hundreds of studies documenting latitudinal clines in allele 86 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint frequencies at candidate genes (e.g. (31-35)) and fitness-associated traits (e.g. (36-40)), and in 87 patterns of genomic differentiation (e.g. (41-43)). Populations of D. melanogaster in the eastern 88 United States show clear genomic and phenotypic differentiation between different latitudes and 89 seasons. The microbiota composition of D. melanogaster also varies across these geographic and 90 seasonal clines, and can profoundly influence the life history traits and evolution of its host (1, 5, 91 6). Two recent studies of the microbiota of wild flies from the eastern USA have each reported 92 substantial geographic variation in microbiota composition, with some conflicting findings 93 between the studies. In one of these studies, we reported substantial numbers of LAB in many of 94 the sampled flies, a latitudinal gradient in the AAB:LAB ratio, and suggested that this pattern 95 revealed congruence between microbial abundance, the influence of those microorganisms on 96 host traits, and the host traits naturally adopted in the sampled locations (5). Another study 97 sampled deeply at more locations and found essentially no LAB in any of the samples and little 98 evidence for a latitudinal cline within- or between- host variation in microbiota composition (6). 99 The authors concluded that variation in microbiota composition was determined primarily by 100 neutral processes and strict host filtering. Also, following evidence that LAB are rare in some (6, 101 15) but not other (5, 12, 18-20) samplings of wild flies, the authors suggested that LAB may be 102 more likely to colonize laboratory than wild flies. These disparate findings show gaps in our 103 understanding of what determines the wild fly microbiota composition. 104 105 To better understand the relationship between sampling location and D. melanogaster microbiota 106 composition, we asked four major questions: 1) Can previously observed latitudinal patterns in 107 microbiota composition be observed in fresh samplings of wild flies? 2) What is the relationship 108 between the microbiota of wild flies and their wild diet? 3) Why are LAB readily recovered in 109 some, but not all wild D. melanogaster samplings? 4) How is variation in the microbiota 110 composition of wild flies related to their life history? We addressed these questions by 111 comparing the sequencing results of previous and new collections of wild D. melanogaster, 112 measuring the microbiota composition of wild and laboratory fly populations reared on distinct 113 environmental conditions or on different diets, and experimentally dissecting the contributions of 114 time and diet decomposition to microbiota composition. We also measured a key life history trait 115 in wild-caught, laboratory-reared fly populations. Together, these results evidence that specific 116 environmental conditions predict patterns in microbiota composition better than latitude, that diet 117 influences but does not necessarily seed the wild fly microbiota, and that fly phenotypes can 118 segregate with the diet of flies in the wild. 119 120 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint

121 A geography-specific D. melanogaster microbiota composition is associated with 122 environmental temperature 123 We reanalyzed two previously published samplings of flies in the eastern USA and compared 124 these to the results of freshly collected samples, to better understand the reasons for their 125 different outcomes. The previous studies included samples collected in 2009 (FIG. 1B, from 126 apples and peaches) and 2018 (FIG. 1C (from grapes)-D (from apples)). We added two additional 127 samplings, one in the eastern USA (FIG. 1E, 2021, from apples) and one in the state of Utah, 128 USA (FIG. S1, 2020, from peaches). Each of the eastern USA collections included at least some 129 sites shared with the other studies. The most abundant genera in the new samplings generally 130 mirrored the previous studies, including that the flies were dominated especially by AAB, with 131 substantial numbers of Enterobacteria or LAB in certain samples. Mantel tests of the relationship 132 between microbiota composition and latitude recapitulated previous conclusions by showing that 133 the 2009, but not the 2018, microbiota composition significantly covaried with latitudinal 134 distance (FIG. 1F-H, FIG. S2). As in 2009, the microbiota composition of the two new samplings 135 covaried with latitude (FIG. 1I, FIG. S1C). We sought to reconcile these different outcomes by 136 testing if environmental conditions could help expain the variation in microbiota composition 137 from different locations. Of 41 variables we tested, the microbiota in all five locations 138 significantly covaried with just one factor - the daily maximum temperature (TABLE S3) - 139 suggesting temperature could be a major determinant of wild fly microbiota composition (FIG. 140 1J-M, FIG. S1D). 141 142 We then tested if these findings applied to flies sampled in other areas and at various times by 143 comparing the same environmental and geographic metrics with the microbiota composition of 144 flies collected as part of the Drosophila Evolution over Space and Time (DEST) data set, which 145 were collected across North America and western Europe mostly between 2014 and 2016 (FIG. 146 1N-Q, (44)). As before, the geographic and microbiota distances covaried significantly in one , 147 but not both sampling areas (FIG. 1R,T). However, the difference in maximum temperature on the 148 day of sampling was significantly correlated with the distance in microbiota composition in 149 samples from both North America and Europe (FIG. 1S,U, TABLE S3). We also confirmed 150 experimentally that variation in temperature significantly determines gnotobiotic microbiota 151 composition (FIG. 1V-W, TABLE S4, FIG. S3). Beyond temperature, UV irradiance and humidity 152 also covaried with microbiota composition in many samplings (TABLE S3). We investigated 153 microbiota effects of humidity previously (45), and report here that photoperiod is also 154 associated with significant changes in microbiota composition, though the changes are less 155 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint marked than those caused by temperature (FIG. S4, TABLE S4). From these findings temperature 156 emerges as a primary, but not the only, driver of variation in wild D. melanogaster microbiota 157 composition. 158 159 Figure 1. Latitude- and temperature-dependent variation in microbiota composition of D. melanogaster from 160 the eastern USA in multiple years and samplings. Drosophila melanogaster from (A) the same general area in the 161 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint eastern USA were sampled and sequenced across portions of the 16S rRNA V4 regions by two laboratories in B) 162 2009 (apples and peaches, N=15), C) 2018 (grapes, N=71); D) 2018 (apples, N=79), and E) 2021 (apples, N=60). 163 Taxon plots are an average of samples which are pools of flies (B, rarefaction threshold (rt) = 475) or individual flies 164 (C-E, rarefaction threshold= 500). Mantel tests reporting the relationship of the Bray Curtis and environmental 165 distances between each sample were also performed, with environmental distance calculated from distance matrices 166 based on F-I) latitude or J-M) latitude plus maximum temperature and minimum relative humidity (RH) on the day 167 of sampling. Letters over F-M report the Mantel test correlation coefficient  and the p-value results when the results 168 are (blue) or are not (black) significant. P-Q) A similar analysis as for eastern USA flies, conducted using sequences 169 extracted from whole genome sequencing of flies collected with the DEST dataset. N-O) Maps, P-Q) taxon plots, 170 and R-U) Mantel test results for DEST flies from N,P,R,S) North American and O,Q,T,U ) Europe are reported. 171 Samplings from Guadeloupe, a department of France in the Caribbean, is not shown on the map. V-W) Common 172 garden populations of individual isofemale lines from each location in FIG. 2 were reared under 6-species 173 gnotobiotic conditions in the laboratory until adult flies were 3 days old, transferred to test conditions for 3 days, 174 then the microbiota composition of pools of 2 surface-sterilized adults was analyzed by homogenization and 175 dilution plating (2 pools of 2 females and 2 males per vial, 3 vials per common garden population in each 176 experiment, 3 separate experiments in time). The relative V) and absolute W) abundances of AAB (red) and LAB 177 (blue) colony forming units (CFUs) in flies reared at varying temperatures. Relative abundances are shown as the 178 mean of AAB counts divided by the mean of LAB counts, with the fraction of LAB shown as a white point and the 179 overlayed violin plot showing the distribution of fractional LAB abundance. Significant differences in relative 180 abundances of LAB were determined by PERMANOVA (TABLE S4). Absolute CFU abundances are shown as the 181 mean and standard error of the mean of all replicates. Significant differences in AAB and LAB abundance were 182 determined by a Kruskal-Wallis test with a post-hoc Dunn test, and different letters over (AAB) or under (LAB) the 183 bars report significant differences in their abundance 184 185 The D. melanogaster microbiota is not a reflection of the microbiota in its diet 186 To further define the contributions of a wild environment to wild flies’ microbiota, we compared 187 community composition in flies bearing a total (resident + transient) microbial community, flies 188 bearing only a resident microbiota, and neighboring fruits and soils. We report on comparisons 189 between individual fruits, wild flies captured from those fruits, and adjacent soil samples. Low 190 prevalence of D. melanogaster in some sites meant that we did not obtain samples of flies 191 bearing a total and resident-only microbial community from all fruit samplings, but among the 192 fruit sites that did have D. melanogaster we sequenced their microbiota and compared it to the 193 corresponding fruit and sample locations (FIG. 2A-B). These samples’ microbiota composition 194 varied significantly with both the sample type (resident fly microbiota, fruit, etc, F3,218 = 20.08, 195 R2 = 0.20, p < 10-4, TABLE S5) and the orchard they were sampled from (F7,218 = 2.28, R2 = 0.05, p 196 < 10-4, TABLE S5). No bacterial genera varied significantly in abundance between the total and 197 resident portion of the microbiota, but there were significant differences in the abundances of 198 bacteria between the diet and wild flies, consistent with the expectation of strict filtering by the 199 flies (6). More than 85% of total reads were assigned to the < 20% of ASVs that were shared 200 between fruit and flies (FIG. 2C, TABLE 1), and among these Commensalibacter, Lactobacillus, 201 Enterococcus and reads that could not be assigned below the Pasteurellales level were all 202 significantly more abundant in flies than diets (TABLE 1). Other bacterial strains were more 203 abundant in the diet than the flies (TABLE 1), together identifying differences between the 204 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint microbiota of flies and their diets and classifying bacteria from the different genera as fly- or 205 diet-preferred. Across both types of fly samples, neutral modeling reported lower AIC scores 206 when microbiota was sampled from itself rather than diet, further confirming that the patterns in 207 microbiota composition follow neutral patterns ((6), FIG. 2D). Also, the observation that the beta-208 diversity distances between fruits and the total or resident microbiota were similar tends to 209 confirm the conclusion that the colonization niche of the bacteria does not influence similarity to 210 the diet (FIG. 2E). 211 212 Separately, the resident microbiota of flies collected from compost piles at three of these sites 213 was significantly different from the composition in flies from fallen apples (F1, 31 = 2.11, R2 = 214 0.05, p = 0.01, TABLE S6, FIG. 2F; these were the only sites with compost piles, samplings from 215 compost and apples were done at the same time, and the apple bars include data shown in FIG. 216 2B). This finding is somewhat surprising because we assumed microbiota composition would be 217 relatively homogenous across an orchard as Drosophila can readily migrate miles overnight (46-218 48). Two genera differed in abundance in flies sampled from apples and compost piles, and both 219 were members of the LAB: Leuconostoc and Weissella (FIG. 2G-H). These findings confirm that 220 there is diet-dependent variation in microbiota composition of flies within the same orchard. The 221 data also reveal that flies sampled from apples are less likely to bear LAB abundantly, and may 222 help to explain why two recent collections of flies from individual fallen apples reported low 223 levels of LAB (FIG. 1C-E). However, because the compost piles had many different types of 224 decomposing material and were not just apples, it is not clear if the differences in fly microbiota 225 composition when collected from individual fruits or composted material are due to differences 226 in the substrate or to its decomposition state. 227 228 Thus, while diet contributes to variation in microbiota composition, both the resident and 229 transient portions of the microbiota can be distinct from their proximal dietary sources at the 230 time of collection. The differences may be due to filtering, as has been asserted previously (6), 231 but the very low abundance and prevalence of the fly-preferred microbes in the flies’ diets also 232 suggests transmission fidelity between flies. Alternatively, the flies may have obtained their 233 microorganisms from diets distinct from those that they were feeding on at time of capture, 234 although the finding that the microbiota differs between flies feeding on different diets in the 235 same orchard is inconsistent with this expectation. 236 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 237 Figure 2. The total and resident D. melanogaster microbiota composition is distinct from the diets of wild flies. 238 A-B) The 16S rRNA V4 region was sequenced in samples of individual wild Drosophila melanogaster, their diets, 239 and nearby soil locations from multiple locations in the eastern USA. Flies were immediately frozen after collection 240 (‘Flies (total)’) or starved in empty vials for > 2 h (‘Flies (resident)’) after transient microorganisms had passed 241 through the fly gut with the bulk flow of diet. C-D) Reads assigned to the genus Commensalibacter are shown in 242 different samples. * = significant differences in abundance by ANCOM. E) Venn diagrams showing the average 243 fractional abundance of ASVs ± s.e.m. that were unique to or shared between sampled fruit and the total or resident 244 fly microbiota. F) AIC values for neutral models calculated with the group indicated in the ‘Reference’ row was 245 compared to itself (‘Reference’, gray squares) or the group in the ‘Comparison’ (black squares) row. G) Weighted 246 Unifrac distances between samples assigned to the groups in the Reference and Comparison rows, including the 247 mean distance between samples (blue bar). Different letters over the clusters of points represent statistically 248 significant differences in distance between the comparison groups as determined by a Kruskal-Wallis test with a 249 post-hoc Dunn test. F-G) Microbiota composition was measured In individual apples or compost piles at multiple 250 locations in the eastern USA, Bars are the averages of mulitiple samples (N = mean 6.5 ± sem 1.2 , min = 3, max = 251 10 samples per bar condition) rarefied to 845 reads each. Reads assigned to the genera E) Leuconostoc and F) 252 Weissella differed significantly in abundance between locations as determined by ANCOM. 253 254 Wild diets determine differences in the D. melanogaster microbiota composition 255 To follow up on the role of diet in shaping the fly microbiota composition, especially the 256 abundance of LAB, we compared the microbiota composition of flies sampled from different 257 wild diets in a single orchard. The flies’ microbiota significantly varied when sampled 258 individually from varieties of apple, peach, and pear in a single orchard (F2, 31 = 2.04, R2 = 0.10, 259 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint p < 10-4, TABLE S7). However, no bacterial groups at any taxonomic level varied significantly 260 with the type of fruit the flies were sampled from. We adopted a more controlled approach by 261 showing that microbiota composition in laboratory flies reared under gnotobiotic conditions with 262 12 different bacterial species varied significantly with the type of fruit the flies were reared on 263 (F9, 52 = 2.42, R2 = 0.34, p < 10-4, TABLE S8). Four of the species varied significantly in abundance 264 in flies reared on the different fruits: L. brevis, W. paramesenteroides, G. cerinus, and Pantoea 265 sp. JGM106 (ANCOM W score = 11 for each species). L. brevis was particularly prevalent in 266 peaches, pears, and oranges, and LAB also tended to have higher average abundance in the non-267 apple samples from our wild samplings (FIG. 3A). Additionally, apples had far greater average 268 AAB abundance than any other samples, consistent with observations that flies sampled from 269 apples were dominated by AAB (FIG. 1C-E), and that our samplings in Utah, which were from 270 peaches, frequently bore abundant LAB (FIG. S1). These findings reveal that the D. melanogaster 271 microbiota varies with diet in both the wild and laboratory, including in flies reared with the 272 same starting set of microorganisms. 273 274 275 Figure 3. The microbiota composition of wild and laboratory D. melanogaster varies with diet type. We 276 sequenced the 16S rRNA V4 region of the D. melanogaster microbiota collected from different fruits in the wild and 277 the laboratory. A) Composition of the resident microbiota in wild male D. melanogaster collected from fallen fruit at 278 Lyman Orchards in Middlefield, CT. Bars are the averages of mulitiple samples each rarefied to 120 reads (N = 279 mean 3.2 ± sem 0.25 , min = 2, max = 4 samples per bar). B) Composition of the total microbiota in gnotobiotic 12-280 sp D. melanogaster CantonS male flies. Bars are the averages of mulitiple samples each rarefied to 200 reads (N = 281 mean 5.3 ± sem 0.54 , min = 3, max = 7 samples per bar). 282 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 283 Population dynamics of the wild Drosophila microbiota are distinct with time and 284 decomposition of wild fly diets 285 We next investigated how time and diet decomposition shape the microbiota composition of wild 286 flies, especially their LAB abundance, by capturing flies from fruit piles in different 287 decomposition states. In a single fall season, we established separate piles of peaches and apples 288 every two weeks and sequenced the resident microbiota of flies sampled twice weekly from the 289 various piles. The microbiota of the flies varied with the type of fruit they were sampled from 290 (F1, 183 = 11.79, R2 = 0.05, p < 10-4, TABLE S9) and over time (F1, 183 = 10.5, R2 = 0.05, p < 10-4, 291 TABLE S9) (FIG. 4A). Also, the overall microbiota composition covaried with time (FIG. 4B, FIG. 292 S5, s = 0.12, p < 10-4). Gluconacetobacter, Leuconostoc, and reads that could not be assigned 293 below the family Enterobacteriaceae significantly varied over time, and each was significantly 294 correlated with time (FIG. 4D-E). No genera varied significantly in flies sampled only from 295 apples, but Gluconacetobacter and Lactobacillus reads varied significantly with time in flies 296 sampled from peaches (FIG. 4F-G). 297 298 Then, we tested if diet decomposition shaped microbiota composition by comparing the 299 microbiota between these same samples when time was defined relative to the ‘establishment 300 time’ of the pile instead of calendar date. There was a significant effect of establishment time 301 (FIG. 4H, F1, 183 = 4.06, R2 = 0.02, p = 0.01) on microbiota composition, but microbiota 302 composition did not covary with establishment time (FIG. 4I, FIG. S5, s = - 0.005; p = 0.50). 303 Further, the Gluconacetobacter (all samples, FIG. 4I; peaches only, FIG. 4K) and 304 Enterobacteriaceae (FIG. 4J) reads that covaried with time did not covary with establishment 305 time of the piles. Conversely, Leuconostoc abundance in peaches and apples (FIG. 4L) and 306 Lactobacillus abundance in peaches only (FIG. 4N) varied more significantly with establishment 307 time than calendar date. Together, these findings suggest that the relative abundance of most of 308 the genera that vary in abundance in the flies, covary with time, but that the populations of the 309 two LAB that varied across the different samplings responded more to establishment time of the 310 piles, or their decomposition, than calendar date. The absence of these patterns when apples were 311 considered alone is consistent with our previous observations that LAB are generally 312 depauperate in apple samples and shows some diets are incompatible with the seasonal patterns 313 we document here. Together, these experiments reveal that variation in the abundance of LAB in 314 the flies is complex and determined at least in part by interactions between time, diet 315 decomposition, and microbial compatibility with the fly diet. 316 317 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 318 Figure 4. The microbiota composition of wild D. melanogaster varies with time and diet decomposition. We 319 sequenced the 16S rRNA V4 region of the resident microbiota in individual wild Drosophila collected from 1-320 bushel piles of fruit established at different times in an experimental orchard in Provo, UT. The same data are shown 321 on two timescales, either relative to A-G) calendar date (time = 0 when the first piles were established), or H-N) pile 322 establishment time (time = 0 for each pile when it was established). A,H) Taxon plot, plus a timeline showing the 323 establishment and sampling times (S) from each of four apple (A) and three peach (P) piles. B,I) Mantel test 324 showing the relationship of the weighted Unifrac and calendar date (B) or pile establishment time (I) distances 325 between each sample, including the slope of the trendline (blue). Plots showing the abundances of specific members 326 of the microbiota that varied in C-E,J-L) apples and peaches, or only in peaches F-G, M-N) in one or both 327 timescales. The Spearman’s rank correlation coefficient (S) and p-value (p), plus a trendline for change in 328 abundance over time (black line) are shown for each. 329 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 330 Diet determines microbiota-dependent genetic stratification of D. melanogaster in a single 331 orchard 332 Finally, we tested if diet-dependent variation in microbiota composition was associated with 333 variation in the life history of flies in a wild setting. We reasoned that since the microbiota is an 334 agent of selection that can drive adaptation of its host (1), flies from diets with distinct 335 microbiota composition might segregate phenotypically. When we measured the development 336 time to adulthood for wild-caught isofemale lines reared in the laboratory under gnotobiotic and 337 bacteria-free conditions, there were significant, non-interactive effects of the dietary source (Z1, 338 4173 = 8.04, p < 10-15) and rearing condition (Z1, 4173 = 7.81, p < 10-14) of the flies (FIG. 5A-B, 339 nonsignificant interaction was Z3, 4173 = -1.54, p = 0.12). The flies sampled from compost piles 340 (), which support greater levels of LAB in the flies (FIG. 2G), developed to reproductive 341 maturity more quickly than flies sampled from apples in multiple locations throughout the 342 orchard (,∆). This pattern, where the flies richer in LAB develop to reproductive maturity 343 more quickly than flies depauperate for LAB, mirrors the pattern observed in flies that were 344 evolved in the wild while being fed L. brevis or A. tropicalis, then reared bacteria-free in the 345 laboratory (FIG. 5C, Z3, 1810 = 3.73, p = 0.00019). Together , these findings support the conclusion 346 that diet-dependent microbiota composition is associated with the genotypic differentiation in 347 two distinct fly populations sampled from the wild. 348 349 Figure 5. D. melanogaster phenotypes segregate with the diet they are collected from . A) Wild D. melanogaster 350 were sampled from compost piles () or individual fallen apples (,∆) throughout a single orchard, shown in a 351 screenshot from Apple maps taken Summer 2024. B) Isofemale lines derived from the wild collections were reared 352 in the laboratory under gnotobiotic or bacteria-free conditions, and their development time to adulthood (eclosion) 353 was measured. C) Development time to adulthood was measured in flies reported previously (1) when reared 354 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint bacteria-free in the laboratory at the conclusion of selection. Significant differences between groups were 355 determined by a Cox mixed-effect model. 356 357 358

359 We sought here to define patterns in the microbiota composition of wild flies and reconcile 360 differences between conflicting previous analyses. We found that latitudinal differentiation in 361 microbiota composition could be better explained by considering temperature than latitude. 362 Followup experiments in the laboratory confirmed that temperature and, to a lesser extent, 363 photoperiod, influence microbiota composition (a role for humidity had been previously 364 demonstrated (45)). Additionally, we provide evidence that even though the flies’ microbiota is 365 not a direct reflection of their diet, the type and decomposition of fly diet influences distinct 366 members of the fly microbiota. Whereas the AAB were prevalent and abundant in a variety of 367 diets, LAB were most abundant in diets that were decomposing and were especially depauperate 368 in flies sampled from apples. Together, these factors can help explain the varied differences in 369 the microbiota composition of wild-caught flies, and show that numerous factors, most of which 370 vary asynchronously with time and space, help determine variation in microbiota composition. 371 372 Diet can help explain some of the patterns in microbiota abundance observed across four 373 different fly samplings (FIG. 1, FIG. S1). LAB were virtually absent in flies sampled from 374 individual apples (FIG. 1D-E) and grapes (FIG. 1C), but were abundant when sampled from rotten 375 peaches (FIG. S1) or not explicitly from individual fruits (FIG. 1B). We did not analyze the 376 microbiota of gnotobiotic flies reared on grapes, but gnotobiotic flies reared on apples bore the 377 highest fraction of AAB of any fruit we sampled, and, conversely, flies reared on peaches bore 378 relatively high LAB loads. The wild samplings suggest that grapes, like apples, support relatively 379 high AAB loads, and that the absence of LAB from the more recent east coast fly samples may 380 have been driven by sampling from fruits that naturally support high levels of AAB (FIG. 1C-E). 381 If samplings from individual apples or grapes, qualitatively assessed to be rotten, are less 382 decomposed than a compost pile, then decomposition state may also contribute to these 383 differences (but it is not possible to ascertain post-hoc). Diet type alone cannot explain 384 differences in microbiota composition between experiments because FIG. 1B flies were sampled 385 from apples (, , •, ) and peaches (,), and the peach-sampled flies had intermediate LAB 386 abundance relative to the other apple samples. However, if the higher latitude samplings were 387 from more decomposed fruits (not recorded at sampling), it could help explain why there were 388 more LAB in flies collected from apples at high-latitudes than in peaches at lower latitudes. 389 Regardless, these findings show that controlling for diet decomposition state and diet type in 390 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint fruit flies provides context for variation in microbiota composition, and our approaches highlight 391 one way that key drivers of microbiota variation can be identified in wild sampled animals. 392 393 Even though diet and its decomposition state were profoundly associated with variation in wild 394 fly microbiota composition, the microbiota composition of wild flies and their proximal diets at 395 time of collection were generally incongruent. The most abundant taxa in the flies were usually 396 detected in the diet, and vice versa, but multiple approaches – comparison of ASV identities, 397 modelling if fruit was the source of the fly microbiota, and directly comparing composition 398 between sample types - all failed to link the abundance of those ASVs between the two sample 399 types. The most conspicuous difference between the microbiota of flies and neighboring fruits 400 was that Commensalibacter was the most abundant ASV in flies and essentially absent from the 401 diet. Lactobacillus and Enterococcus were also abundant in flies but not their diets. These 402 findings are consistent with the substantial body of work showing that laboratory and wild flies 403 can be colonized persistently by specific sets of microorganisms (22, 24, 27, 30, 49). It remains 404 unclear what the reservoir of the fly- or diet- specific microorganisms is and how each is 405 transferred between flies or to fresh diets, but our data make it clear that abundant fly microbiota 406 ASVs are not necessarily sampled commonly from their diets. While there is ample evidence that 407 the fly microbiota is assembled neutrally (6, 12, 50) and has priority effects (27), this suggests 408 that either the flies are colonized early on with microorganisms that are not abundant in their 409 diet; or that certain microorganisms are adept at invading established communities. Together, 410 these findings are consistent with previous demonstrations that the host and community 411 interactions drive the microbiota composition in the gut and the external fly environment (16, 29, 412 30, 45). 413 414 Along with host genotype and diet, abiotic environmental factors shape the Drosophila 415 microbiota composition. We found that whereas no latitude was an inconsistent predictor of 416 microbiota composition, the microbiota of flies sampled from different diets or years was 417 individually correlated with environmental factors that influence microbiota composition. Of 418 these, temperature was the strongest determinant. Elevated temperatures increased absolute and 419 relative abundance of LAB in flies, consistent with their higher maximum growth temperatures 420 than AAB. The tolerance of LAB to heat may also explain their elevated abundance in compost 421 piles, which can reach internal temperatures above 60°C, relative to individual fallen fruits. 422 However, high frequencies of LAB at warmer temperatures are inconsistent with our previously 423 published expectation to detect LAB more abundantly in flies at high latitudes (5), but may 424 possibly be explained by differences in the decomposition state of the diet. Also, a previous 425 laboratory analysis of conventionally reared flies reported LAB were more abundant in flies 426 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint reared at low temperatures (51). Therefore, one or more of our uses of gnotobiotic flies, which 427 controls for the starting exposure to different microorganisms, or our use of different diets, host 428 genotypes, or microorganisms may have contributed to the different trends. Regardless, 429 temperature was a strong predictor of microbiota composition in the flies. 430 431 The various influences of temperature, photoperiod, humidity, diet type, and diet decomposition 432 state, each of which can vary asynchronously with season in different geographic areas, raises 433 questions about the role of seasonal variation in microbiota composition. Currently, the most 434 comprehensive longitudinal analysis of the fruit fly microbiota comes from flies fed laboratory 435 diets in outdoor mesocosms over a summer-to-fall season in the eastern USA (6). The flies’ 436 microbiota clearly shifted with seasonal progression, including a summer-to-fall shift from 437 Acetobacter to Commensalibacter dominance, and a mid-summer peak in the abundance of 438 Wautersiella, a genus from the Flavobacteria that was not abundant in the flies we report on here 439 (6). These seasonal patterns in the microbiota composition of flies reared outdoors were 440 independent of variation in the diet since the flies were fed a laboratory diet throughout the 441 experiment. Each of the samplings we report or reanalyze here were conducted over a relatively 442 narrow ~ 3 week window. We cannot speak to the other authors’ motivation, (FIG. 1B, E), but our 443 intention in narrow sampling times was to control for time. With the retrospective perspective 444 from the longitudinal analysis, our design actually increases variation by adding location-specific 445 seasonal progression as an unreplicated confounding variable. It remains unclear exactly how 446 seasonal progression should be defined, but our findings here suggest that temperature, time of 447 fruit onset, and type of fruit available, plus humidity, day length, and UV irradiance as key 448 factors to consider. If specific patterns in seasonal microbiota variation are common across 449 multiple seasons, then such variation may also be a useful tool. For example, the relative 450 abundances of Acetobacter and Commensalibacter in the FIG. 1 datasets are sometimes 451 positively, negatively, or not significantly correlated with each other (FIG. S6). One way of 452 interpreting these varied outcomes is that the datasets where Acetobacter and Commensalibacter 453 read counts are inversely related are asynchronous for seasonal progression; and datasets where 454 these abundances are not significantly negatively correlated are more seasonally-synchronized. 455 The strong influence of temperature alone suggests that timing a latitudinal sampling scheme 456 based on predicted temperatures might be sufficient to account for this variation; although daily 457 or site-specific deviations from the norm might it prohibitively difficult to synchronize. If the 458 conditions are more varied, then such interpretations might require identifying additional 459 seasonal trends that span years and locations. Regardless, the established patterns in seasonal 460 evolution of D. melanogaster and the ability to track how variation in microbiota composition 461 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint influences the fly suggest an ideal framework for future investigation of the relationship between 462 seasonal variation, local adaptation of a host, and variation in microbiota composition. 463 464 The data presented here suggest the flies’ evolutionary history segregates with their diet and with 465 their diet-dependent microbiota composition. We interpret these findings to mean that these 466 factors together contribute phenotypic structure to wild populations even in the presence of 467 expected gene flow. The distinct phenotypes of axenic flies from the different sampling locations 468 confirms phenotypic differentiation between the flies collected from different diets at different 469 geographic distances from each other, and that host genotype, as well as the microbiota, 470 contributes to these differences. Beyond this, we observed wider differentiation of the 471 development time trait between the flies at these sites when the microbiota was present. These 472 latter findings are consistent with previous demonstrations that the microbiota can enhance 473 phenotypic variance in wild fly populations (5), and likely underrepresent the potential variance 474 because the flies were reared on the same diets and on a standardized microbiota composition. 475 This is because the constraints on diet and microbiota are both likely to be lower in the wild than 476 in the laboratory. It is possible that the diet-dependent segregation of host phenotypes is 477 mediated directly by host genotype based on variation in e.g., their dispersal, dietary or feeding 478 preference traits (52-58). However, we note that there were no differences in host genetic 479 selection on the microbiota in flies collected from compost or apples (FIG. S8), suggesting that 480 the phenotypic differences observed in the wild flies are driven by diet or are sampling-specific, 481 not from host genetic control of the microbiota (as in (21)). Further, the parallel phenotypic 482 outcomes of flies reared in dispersal-limited mescosms with LAB, and wild in conditions that 483 promote LAB abundance, are striking. When fly migration was restricted, flies feeding on L. 484 brevis in their diets adopted a ’faster’ life history strategy, weighing less, reaching larger 485 population sizes, and developing to reproductive maturity more quickly than their A. tropicalis-486 fed counterparts (FIG. 5C, (1)). Flies feeding on compost in the wild, which promotes LAB 487 abundance (FIG. 2F), mirrored these findings by displaying a shorter developmental period than 488 flies collected from apples (FIG. 5B), a diet that restricts LAB prevalence and abundance (FIG. 489 2F). Together, the parallel outcomes of the two different lines of inquiry, one of which 490 established the microbiota as a causal link, are consistent with the expectation that conditions 491 that promote different fly microbiomes can shape their host’s evolutionary trajectory. 492 493 The analysis we present here does not focus on intracellular bacterium Wolbachia, a reproductive 494 manipulator of many insects that is prevalent in wild Drosophila populations and can profoundly 495 influence the flies’ life history and physiology (59-67). The major reason for this omission is that 496 most of our datasets were poorly balanced for the presence and absence of Wolbachia; in some 497 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint locations many flies bore Wolbachia, whereas Wolbachia were rare in other samplings. We 498 generally discarded Wolbachia reads, then determined differences in microbiota composition 499 between samples. However, we often reported Wolbachia colonization status (±) as a covariate 500 in our analyses when it was significant, and performed unpublished comparisons of the 501 Wolbachia-colonized and Wolbachia-free flies throughout this work. Most of the major 502 statistical trends were observed when we analyzed the flies mixed together or separately 503 according to their Wolbachia colonization status. Exceptions to this rule were generally 504 confounded by small or unbalanced sample size and would require more sampling to make a 505

either way, which is why we do not formally report on them here. As an exception, 506 one dataset was well-balanced for Wolbachia colonization status: the analysis of flies collected 507 on different fruits in a Middlefield, CT orchard (TABLE S7). Among these flies there were no 508 genus-level differences in bacterial relative abundance between Wolbachia positive and -509 negative flies, but differences were detected at different taxonomic levels: Wolbachia positive 510 flies bore more family-level Acetobacteraceae reads and, at the ASV level, more of a 511 Commensalibacter ASV (FIG. S7). An association between Wolbachia and Commensalibacter 512 was reported previously (6) but is somewhat surprising because a sister-genus, Acetobacter, has 513 been reported by us and others to be negatively associated with Wolbachia abundance (5, 63, 514 68). These findings highlight the importance Wolbachia plays as a part of the Drosophila 515 microbiota, and the recent use of outdoor enclosures to study Wolbachia-Drosophila-microbiota 516 interactions highlights an elegant way to do so in a native setting while also maintaining a 517 balanced statistical design. 518 519 In summary, we present evidence that many factors contribute to geographic variation in the D. 520 melanogaster microbiota composition. We also show that identifying and considering such 521 factors makes it clear that there are ordered patterns in the types and abundances of 522 microorganisms in wild flies. In particular, several of our findings can help explain why LAB, 523 which are among the most dominant taxa in flies reared on certain laboratory diets, are nearly 524 absent from some recent surveys of flies in a wild setting. They also provide context supporting 525 that members of the Enterobacteriaceae are common and abundant in flies in a wild setting. 526 Further analysis of these and other factors will improve our ability to explain and predict the 527 many varied patterns of microbiota composition in these relatively simple communities of host-528 associated microorganisms, help to model the interactions in more complex communities of 529 partners, and establish the role that microbial partners play in shaping animal life histories. 530 531

532 General Rearing and Culture Conditions 533 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint We reared flies in general culture at 25°C on a 12-hour light:dark cycle. Flies from wild 534 collections were reared on a molasses diet (5). Some laboratory experiments were performed 535 using a CantonS fly line obtained from Mariana Wolfner that is Wolbachia-free and was reared 536 on a yeast-glucose (Y-G) diet (23). 537 538 Wild fly collections 539 We collected samples from eight orchards in the eastern USA in Fall of 2021 and five orchards 540 in Utah, USA in Fall of 2020 (TABLE S1). At each location we collected samples from ten to 541 twenty individual fallen fruit sites: the fruit, neighboring soil, a male D. melanogaster fly bearing 542 a total microbiota and a male D. melanogaster fly bearing its resident microbiota. Flies were 543 captured from an aerial insect net and divided into two empty, sterile vials; one immediately 544 frozen on dry ice, the other left for > 2 hours for defecation of non-resident microorganisms and 545 then frozen on dry ice. We aimed to collect 5-10 flies per vial at each fruit site. From each site, 546 females were also collected and reared individually on molasses diet to establish isofemale lines 547 that were kept in culture throughout the study. Additional fly populations were collected in 2023 548 following this method as well (TABLE S2). Collection and propagation of isofemale lines was as 549 previously described, including morphological and molecular analysis to retain only D. 550 melanogaster (45). Fruit and soil were sampled to ~ 1/2-inch depth using a 1/4-inch diameter 551 fruit corer that was briefly pre-sterilized in 10% bleach then pre-rinsed in sterile double-distilled 552 H2O. Each were immediately stored on dry ice. We also performed aerial collections of flies at 553 compost piles in three locations (TABLE S1), and single-fruit collections for the resident fly 554 microbiota from many different apple, peach, and pear varieties in the orchard at Middlefield, 555 CT. All samples were stored on dry ice until they could be permanently stored in a -80°C freezer. 556 557 Locale-specific fly populations were established from individual isofemale lines by 3 generations 558 of common-garden mixing for 2-3 generations as described previously (45). Eggs laid by F2 or 559 F3 flies were collected, dechorionated, and reared as gnotobiotic flies in association with the six 560 bacterial species Acetobacter tropicalis DmCS_006, Acetobacter sp. DsW_54, Acetobacter sp. 561 DmW_125, Lactiplantibacillus plantarum DmCS_001, Weissella paramesenteroides 562 DmW_115, and Leuconostoc suionicum DmW_098. Two days after eclosion of adult flies, vials 563 were transferred to new conditions to measure the influence of a 3-day perturbation in 564 photoperiod (left at a 12h light:dark cycle or incubated at 1h light:23h dark or 23h light:1h dark) 565 or temperature (left at 25°C or moved to 15 °C or 32 °C) on the adult fly microbiota. Three days 566 later we measured the adult fly microbiota. Flies caught from compost and apples in the same 567 orchard were reared under axenic and gnotobiotic conditions as individual isofemale lines, not in 568 a common garden. 569 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 570 Experimentally-evolved flies were reared in outdoor mesocosms for 6.5 weeks as described 571 previously (1), then flies from each mesocosm were reared in common garden populations for 2-572 3 generations. Bacteria-free embryos were derived from each population and their development 573 time to adulthood was measured as described below. 574 575 Gnobiotic fly rearing and enumeration of bacterial colonization 576 Embryos laid by D. melanogaster were collected, dechorionated in 10% bleach for two 150s 577 washes, rinsed three times with sterile water, and 30-60 embryos were transferred to 7.5 ml 578 sterile molasses diet in a 50 ml microcentrifuge tube. Separately, bacterial strains were cultured 579 overnight in de Man-Rogosa-Sharpe (MRS) medium (Criterion C5932) for 1-2 days at 30°, 580 normalized to OD600 = 0.1, and mixed in equal ratios. Then, we inoculated 50 l of the bacterial 581 mixture to the sterile eggs and allowed the eggs to develop to adulthood. When the adult flies 582 were 4-7 days old, the adult fly microbiota was measured by collecting flies surface-sterilized in 583 ethanol under light CO2 anesthesia, performing whole-body homogenization, and dilution plating 584 the homogenate for CFU counting as described previously (27). From each fly vial we collected 585 4 pools of 2 male flies and 4 pools of 2 female flies. Fly homogenates were serially diluted onto 586 MRS (to culture AAB and LAB) and MRS plus 10 g/ml chloramphenicol and 10 g/ml 587 erythromycin (to culture only AAB). All dilution and plating was performed using an EpMotion 588 96. Plates were incubated at 30 °C for 2-3 days, and the antibiotic-free plates were cultures in 589 sealed CO2-flooded containers. Colonies of LAB and AAB were then manually or automatically 590 counted (27) to compare the abundances of each bacterial order in the fly microbiota. Each 591 experiment was repeated three times with triplicate vials for each population and condition. CFU 592 count data were analyzed as described previously (45). Significant influences of treatments on 593 microbiota composition were determined by PERMANOVA of a Bray-Curtis beta-diversity 594 distance matrix constructed from the LAB and AAB CFU counts, rarefied to 10000 595 (temperature) or 4000 (photoperiod) counts, using custom scripts, QIIME2, and the R package 596 Vegan (69). PERMANOVA was always performed with 1,000 permutations. Variation in the 597 absolute abundances of the LAB or AAB was tested for significance using a Kruskal-Wallis test 598 with a post-hoc Dunn test and the R packages dunn.test (70), rcompanion (71), and multcomp 599 (72). 600 601 Sample preparation and 16S rRNA gene sequencing 602 In the laboratory, DNA was extracted from each sample using the Zymo Quick-DNA™ 603 Fecal/Soil Microbe 96 Kit (D6011) following the manufacturer’s instructions except that DNA 604 was eluted in 50 l elution buffer. Flies were individually examined under a microscope to 605 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint sequence only male D. melanogaster, as determined by the presence of a distinct genital arch. 606 We also extracted DNA from 0.02 g of soil and 0.05 g of fruit. At the time DNA was extracted 607 we loaded a microbiome cell standard onto a single well of each 96-well plate, and left three 608 empty wells for a reagent-only control and two PCR amplification controls. We assured that 609 negative PCR controls did not yield visible bands on gel electrophoreses. 610 611 We performed 16S rRNA marker gene sequencing using a previously described dual-barcoding 612 approach (73). We normalized the reactions using the Just-a-Plate normalization kit (Charm 613 Biotech, JN-120-10), and combined equivalent volumes of 96 samples into single pools. The 614 pools were concentrated using a Zymo gDNA Clean & Concentrator 11-302C kit, fragment size 615 distributions were evaluated on an Agilent FemtoPulse (Agilent Technologies, Santa Clara, CA, 616 USA), and fragments in the 250-450 bp range were selected on a Sage Science Blue Pippin 617 (Sage Science, Beverly, MA, USA). The final molarity of the pool was estimated via qPCR at 618 the BYU DNA sequencing center. Then, we combined the pools and sequenced them on an 619 Illumina MiSeq using 500 cycle v2 chemistry as described previously (73). 620 621 Sequence analysis 622 Demultiplexed sequence reads were analyzed using QIIME2 (74) and R 4.1.2 (75). Previously 623 published datasets were accessed from our own personal archives (FIG. 1B (5)), though the data 624 are published at PRJNA589709, or from BioProject PRJNA873107. Using DADA2 (76), reads 625 were denoised, dereplicated, and amplicon sequence variants (ASVs) were called using trimming 626 lengths that maximize quality scores of the reads. Taxonomic assignments to ASV were made 627 using the GreenGenes classifier 13_8_99 (77), and samples were assigned as Wolbachia positive 628 if 20% or more of the total reads were assigned to Wolbachia. Then, reads that could be assigned 629 to Archaea, Chloroplast, Mitochondria, or Wolbachia were discarded. Operational taxonomic 630 unit (OTU) tables were filtered to various thresholds per sample, reported in each corresponding 631 figure legend. Significant differences between groups were determined by PERMANOVA (78) 632 of beta-diversity distance metrics (79, 80). Phylogenetic trees, supporting the use of the Unifrac 633 distance metrics, were built with fasttree2 (81) based on mafft alignment (82). Analysis of 634 Composition of Microbiomes (ANCOM) was used to define significant differences in the 635 abundances of individual microbes between samples (83). ANCOM was performed on reads 636 clustered at each taxonomic level and is generally reported at the genus or ASV level only, 637 though trends across levels were usually consistent. Beta-diversity distance analysis, calculation 638 of Venn diagrams (84), Spearman’s rank correlation tests, and PERMANOVA (78) were all 639 performed in R. Application of neutral models to sequencing data were performed in R using 640 default parameters as described previously (85, 86), and Akaike information criterion (AIC) 641 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint values for each test were recorded from each separate model. To perform Mantel tests, in R, 642 environmental metadata for each location were downloaded from the USA & Americas (1998-643 2022) database at https://nsrdb.nrel.gov/data-viewer in 30- or 60-minute intervals or from the 644 Meteostate Prime Meridian: Africa and Europe database in 60 minute intervals. Maximum, 645 average, or minimum values of each character were calculated on the day of sampling, a distance 646 matrix was constructed based on individual or multiple values, and the relationship between the 647 sampling location of each individual sample and the environmental metadata were calculated 648 using a Mantel test based on Spearman’s rank correlations. Sequences from the DEST dataset 649 were downloaded from the SRA using fasterq-dump v.2.10.8 from the SRA toolkit 650 (https://github.com/ncbi/sra-tools), and microbial profiling was performed using MetaPhlan3 and 651 the ChocoPhlan v30 database (87). The OTU table obtained from all outputs was then analyzed 652 using Bray Curtis distance metrics as described above. 653 654 Rearing gnotobiotic flies on fruit diets 655 To measure the impact of distinct diets on the microbiota composition of D. melanogaster, we 656 reared gnotobiotic D. melanogaster CantonS on ten types of fruit, obtained from grocery stores 657 in Provo and Orem, UT. We diced, froze, and transferred 5- 10 g of each fruit type to a 50 ml 658 centrifuge tube, then autoclaved the tubes. Then, we transferred 30-60 bleach-sterilized, 659 dechorionated fly embyos to the diets as described above, and inoculated each vial with a 660 mixture of separately cultured and OD600=0.1-normalized bacterial community composed of 661 Acetobacter pomorum DmCS_004, Acetobacter tropicalis DmCS_006, Bacillus subtilis 168, 662 Escherichia coli K-12 MG1655, Gluconobacter cerinus Dm-58, Komagataeibacter 663 medellinensis NBRC 3288, Lactiplantibacillus plantarum DmCS_001, Levilactobacillus brevis, 664 Pantoea sp. JGM49, Pantoea dispersa JGM106, DmCS_003, Providencia rettgeri JGM232, 665 Weissella paramesenteroides DmW_115. When the flies were 5-7 day old adults, one pool of 5 666 male flies was collected from each vial, frozen at -80 °C, and the V4 region of the 16S rRNA 667 gene was sequenced and analyzed as described above. We performed four separate experiments, 668 each with triplicate vials of flies, to target collection of 3 pools of flies in each of three 669 experiments. However, recovery of adult flies was often challenging because some diets poorly 670 supported fly growth or were wet and adult flies drowned. We collected and sequenced as many 671 samples as we could recover from the four experiments. 672 673 Rearing wild flies on decomposing diets 674 To dissect the separate influences of seasonal progression and diet decomposition on microbiota 675 composition, we established fresh fruit piles in a field site in Provo, UT in August 2022. Each 676 pile was established by dropping one bushel of fruit (Allred’s Orchards, Payson, UT), from a 677 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint height of 4 feet onto a grass lawn. We established piles from apples and peaches every two 678 weeks for six weeks, and a fourth apple pile two weeks later (fresh peaches were no longer 679 available). Twice a week up to 30 fruit flies were sampled from each pile with a hand insect 680 vacuum into empty vials, starved for 2 hours, then frozen in a -80 °C freezer. Sampling 681 continued until hard freeze onset, November 2022, and the 16S rRNA V4 region on individual 682 flies was sequenced as described above. Only 7 of 146 male flies were assigned as D. simulans 683 by morphological examination, so we analyzed male D. melanogaster and female Drosophila 684 together in this experiment. 685 686 Measuring fly development time to adulthood 687 Fly development time to reproductive maturity was measured at 1, 6, and 11 hours into the daily 688 light cycle each day as the time to eclosion for each individual pupa on the side of a fly vial until 689 all flies in a vial had eclosed or there were two consecutive time points where no flies eclosed, 690 whichever came first. Significant differences in fly development time were determined using a 691 Cox mixed-effects survival model with the source diet (apples or compost) and rearing treatment 692 (axenic or gnotobiotic) as interative terms and experimental replicate (four fly vials per treatment 693 were reared in each of three distinct experiments in time) as a random effect. The analyses were 694 performed using survival-model specific R packages (88, 89). 695 696 Data analysis 697 Some R packages we used are not cited elsewhere (90-103). Maps were made in R using (104). 698 A screenshot from Apple Maps taken in summer 2024 was used in FIG. 5A. Raw data and scripts 699 for these analysis are available at https://github.com/johnchaston/Gale2024, most prominently 700 the script file Gale_script.Rmd and the knitted HTML file from that script, Gale_script.html.zip. 701 702 Data availability 703 Sequences from this study will be deposited and made publically available in the SRA at 704 accession XXXXX when the article is accepted. 705 706

707 Research reported in this publication was supported in part by the National Institute of General 708 Medical Sciences of the National Institutes of Health under Award Number R15GM140388. The 709 content is solely the responsibility of the authors and does not necessarily represent the official 710 views of the National Institutes of Health. 711 712 713 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint

714 1. S. M. Rudman et al., Microbiome composition shapes rapid genomic adaptation of Drosophila 715 melanogaster. Proc Natl Acad Sci U S A 116, 20025-20032 (2019). 716 2. W. B. Ludington, The importance of host physical niches for the stability of gut microbiome 717 composition. Philosophical transactions of the Royal Society of London. Series B, Biological 718 sciences 379, 20230066 (2024). 719 3. A. J. Barron et al., Microbiome-derived acidity protects against microbial invasion in Drosophila. 720 Cell reports 43, 114087 (2024). 721 4. J. Consuegra et al., Metabolic Cooperation among Commensal Bacteria Supports Drosophila 722 Juvenile Growth under Nutritional Stress. iScience 23, 101232 (2020). 723 5. A. W. Walters et al., The microbiota influences the Drosophila melanogaster life history strategy. 724 Mol Ecol 29, 639-653 (2020). 725 6. L. P. Henry, J. F. Ayroles, Drosophila melanogaster microbiome is shaped by strict filtering and 726 neutrality along a latitudinal cline. Mol Ecol 31, 5861-5871 (2022). 727 7. L. P. Henry, M. Fernandez, S. Wolf, V. Abhyankar, J. F. Ayroles, Wolbachia impacts microbiome 728 diversity and fitness-associated traits for Drosophila melanogaster in a seasonally fluctuating 729 environment. Ecology and evolution 14, e70004 (2024). 730 8. G. Storelli et al., Lactobacillus plantarum promotes Drosophila systemic growth by modulating 731 hormonal signals through TOR-dependent nutrient sensing. Cell Metab 14, 403-414 (2011). 732 9. J. H. Ryu et al., Innate immune homeostasis by the homeobox gene caudal and commensal-gut 733 mutualism in Drosophila. Science 319, 777-782 (2008). 734 10. S. C. Shin et al., Drosophila microbiome modulates host developmental and metabolic 735 homeostasis via insulin signaling. Science 334, 670-674 (2011). 736 11. N. A. Broderick, N. Buchon, B. Lemaitre, Microbiota-induced changes in drosophila melanogaster 737 host gene expression and gut morphology. mBio 5, e01117-01114 (2014). 738 12. K. L. Adair, M. Wilson, A. Bost, A. E. Douglas, Microbial community assembly in wild populations 739 of the fruit fly Drosophila melanogaster. The ISME journal 12, 959-972 (2018). 740 13. V. Corby-Harris et al., Geographical distribution and diversity of bacteria associated with natural 741 populations of Drosophila melanogaster. Appl Environ Microbiol 73, 3470-3479 (2007). 742 14. C. Ren, P. Webster, S. E. Finkel, J. Tower, Increased internal and external bacterial load during 743 Drosophila aging without life-span trade-off. Cell Metab 6, 144-152 (2007). 744 15. F. Staubach, J. F. Baines, S. Kunzel, E. M. Bik, D. A. Petrov, Host species and environmental 745 effects on bacterial communities associated with Drosophila in the laboratory and in the natural 746 environment. PLoS One 8, e70749 (2013). 747 16. A. C. Wong et al., The Host as the Driver of the Microbiota in the Gut and External Environment 748 of Drosophila melanogaster. Appl Environ Microbiol 81, 6232-6240 (2015). 749 17. C. N. Wong, P. Ng, A. E. Douglas, Low-diversity bacterial community in the gut of the fruitfly 750 Drosophila melanogaster. Environ Microbiol 13, 1889-1900 (2011). 751 18. A. Bost et al., Functional variation in the gut microbiome of wild Drosophila populations. Mol 752 Ecol 27, 2834-2845 (2018). 753 19. V. G. Martinson, A. E. Douglas, J. Jaenike, Community structure of the gut microbiota in 754 sympatric species of wild Drosophila. Ecology letters 20, 629-639 (2017). 755 20. J. A. Chandler, J. M. Lang, S. Bhatnagar, J. A. Eisen, A. Kopp, Bacterial communities of diverse 756 Drosophila species: ecological context of a host-microbe model system. Plos Genet 7, e1002272 757 (2011). 758 21. A. J. Dobson et al., Host genetic determinants of microbiota-dependent nutrition revealed by 759 genome-wide analysis of Drosophila melanogaster. Nature communications 6, 6312 (2015). 760 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 22. I. S. Pais, R. S. Valente, M. Sporniak, L. Teixeira, Drosophila melanogaster establishes a species-761 specific mutualistic interaction with stable gut-colonizing bacteria. PLoS biology 16, e2005710 762 (2018). 763 23. P. D. Newell, A. E. Douglas, Interspecies interactions determine the impact of the gut microbiota 764 on nutrient allocation in Drosophila melanogaster. Appl Environ Microbiol 80, 788-796 (2014). 765 24. S. J. Morgan, J. M. Chaston, Flagellar Genes Are Associated with the Colonization Persistence 766 Phenotype of the Drosophila melanogaster Microbiota. Microbiol Spectr 11, e0458522 (2023). 767 25. B. Obadia et al., Probabilistic Invasion Underlies Natural Gut Microbiome Stability. Curr Biol 27, 768 1999-2006 e1998 (2017). 769 26. N. J. Winans et al., A genomic investigation of ecological differentiation between free-living and 770 Drosophila-associated bacteria. Mol Ecol 26, 4536-4550 (2017). 771 27. R. Dodge et al., A symbiotic physical niche in Drosophila melanogaster regulates stable 772 association of a multi-species gut microbiota. Nature communications 14, 1557 (2023). 773 28. K. Gutierrez-Garcia et al., A conserved genetic basis for commensal-host specificity through live 774 imaging of colonization dynamics. BioRXiv (2024). 775 29. C. N. Fischer et al., Metabolite exchange between microbiome members produces compounds 776 that influence Drosophila behavior. eLife 6, e18855 (2017). 777 30. A. L. Gould et al., Microbiome interactions shape host fitness. Proc Natl Acad Sci U S A 115, 778 E11951-E11960 (2018). 779 31. J. G. Oakeshott, G. K. Chambers, J. B. Gibson, D. A. Willcocks, Latitudinal relationships of 780 esterase-6 and phosphoglucomutase gene frequencies in Drosophila melanogaster. Heredity 47, 781 385-396 (1981). 782 32. P. A. Umina, A. R. Weeks, M. R. Kearney, S. W. McKechnie, A. A. Hoffmann, A rapid shift in a 783 classic clinal pattern in Drosophila reflecting climate change. Science 308, 691-693 (2005). 784 33. P. S. Schmidt et al., Ecological genetics in the North Atlantic: environmental gradients and 785 adaptation at specific loci. Ecology 89, S91-107 (2008). 786 34. J. Overgaard, T. N. Kristensen, K. A. Mitchell, A. A. Hoffmann, Thermal tolerance in widespread 787 and tropical Drosophila species: does phenotypic plasticity increase with latitude? The American 788 naturalist 178 Suppl 1, S80-96 (2011). 789 35. A. B. Paaby, A. O. Bergland, E. L. Behrman, P. S. Schmidt, A highly pleiotropic amino acid 790 polymorphism in the Drosophila insulin receptor contributes to life-history adaptation. Evolution 791 68, 3395-3409 (2014). 792 36. P. S. Schmidt, A. B. Paaby, M. S. Heschel, Genetic variance for diapause expression and 793 associated life histories in Drosophila melanogaster. Evolution 59, 2616-2625 (2005). 794 37. R. Parkash, S. Rajpurohit, S. Ramniwas, Changes in body melanisation and desiccation resistance 795 in highland vs. lowland populations of D. melanogaster. J Insect Physiol 54, 1050-1056 (2008). 796 38. C. M. Sgro et al., A comprehensive assessment of geographic variation in heat tolerance and 797 hardening capacity in populations of Drosophila melanogaster from eastern Australia. J Evol Biol 798 23, 2484-2493 (2010). 799 39. S. F. Lee et al., Molecular basis of adaptive shift in body size in Drosophila melanogaster: 800 functional and sequence analyses of the Dca gene. Molecular biology and evolution 28, 2393-801 2402 (2011). 802 40. L. M. Travers, F. Garcia-Gonzalez, L. W. Simmons, Live fast die young life history in females: 803 evolutionary trade-off between early life mating and lifespan in female Drosophila 804 melanogaster. Sci Rep 5, 15469 (2015). 805 41. B. Kolaczkowski, A. D. Kern, A. K. Holloway, D. J. Begun, Genomic differentiation between 806 temperate and tropical Australian populations of Drosophila melanogaster. Genetics 187, 245-807 260 (2011). 808 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 42. A. O. Bergland, E. L. Behrman, K. R. O'Brien, P. S. Schmidt, D. A. Petrov, Genomic evidence of 809 rapid and stable adaptive oscillations over seasonal time scales in Drosophila. Plos Genet 10, 810 e1004775 (2014). 811 43. A. O. Bergland, R. Tobler, J. Gonzalez, P. Schmidt, D. Petrov, Secondary contact and local 812 adaptation contribute to genome-wide patterns of clinal variation in Drosophila melanogaster. 813 Mol Ecol 25, 1157-1174 (2016). 814 44. M. Kapun et al., Drosophila Evolution over Space and Time (DEST): A New Population Genomics 815 Resource. Molecular biology and evolution 38, 5782-5805 (2021). 816 45. C. Massey et al., Humidity determines penetrance of a latitudinal gradient in genetic selection 817 on the microbiota by Drosophila melanogaster. bioRxiv 10.1101/2024.05.02.591907 (2024). 818 46. J. A. Coyne, B. Milstead, Long-Distance Migration of Drosophila. 3. Dispersal of D. melanogaster 819 Alleles from a Maryland Orchard. The American naturalist 130, 70-82 (1987). 820 47. J. A. Coyne et al., Long-Distance Migration of Drosophila. American Naturalist 119, 589-595 821 (1982). 822 48. K. J. Leitch, F. V. Ponce, W. B. Dickson, F. van Breugel, M. H. Dickinson, The long-distance flight 823 behavior of Drosophila supports an agent-based model for wind-assisted dispersal in insects. 824 Proc Natl Acad Sci U S A 118, e2013342118 (2021). 825 49. H. Zhu, W. B. Ludington, A. C. Spradling, Cellular and molecular organization of the Drosophila 826 foregut. Proc Natl Acad Sci U S A 121, e2318760121 (2024). 827 50. H. Inamine et al., Spatiotemporally Heterogeneous Population Dynamics of Gut Bacteria Inferred 828 from Fecal Time Series Data. mBio 9, e01453-01417 (2018). 829 51. N. N. Moghadam et al., Strong responses of Drosophila melanogaster microbiota to 830 developmental temperature. Fly (Austin) 12, 1-12 (2018). 831 52. A. C. Wong et al., Gut Microbiota Modifies Olfactory-Guided Microbial Preferences and Foraging 832 Decisions in Drosophila. Curr Biol 27, 2397-2404 e2394 (2017). 833 53. T. B. Call, E. K. Davis, J. D. Bean, S. G. Lemmon, J. M. Chaston, Bacterial Metabolism and 834 Transport Genes Are Associated with the Preference of Drosophila melanogaster for Dietary 835 Yeast. Appl Environ Microbiol 88, e0072022 (2022). 836 54. R. Leitao-Goncalves et al., Commensal bacteria and essential amino acids control food choice 837 behavior and reproduction. PLoS biology 15, e2000862 (2017). 838 55. S. F. Henriques et al., Metabolic cross-feeding in imbalanced diets allows gut microbes to 839 improve reproduction and alter host behaviour. Nature communications 11, 4236 (2020). 840 56. M. F. Camus, K. Fowler, M. W. D. Piper, M. Reuter, Sex and genotype effects on nutrient-841 dependent fitness landscapes in Drosophila melanogaster. Proc Biol Sci 284 (2017). 842 57. A. J. Reddiex, T. P. Gosden, R. Bonduriansky, S. F. Chenoweth, Sex-specific fitness consequences 843 of nutrient intake and the evolvability of diet preferences. The American naturalist 182, 91-102 844 (2013). 845 58. C. Ribeiro, B. J. Dickson, Sex peptide receptor and neuronal TOR/S6K signaling modulate nutrient 846 balancing in Drosophila. Curr Biol 20, 1000-1005 (2010). 847 59. A. Strunov, C. Schoenherr, M. Kapun, Wolbachia has subtle effects on thermal preference in 848 highly inbred Drosophila melanogaster which vary with life stage and environmental conditions. 849 Sci Rep 13, 13792 (2023). 850 60. A. Strunov, S. Lerch, W. U. Blanckenhorn, W. J. Miller, M. Kapun, Complex effects of 851 environment and Wolbachia infections on the life history of Drosophila melanogaster hosts. J 852 Evol Biol 35, 788-802 (2022). 853 61. L. J. Cao, W. Jiang, A. A. Hoffmann, Life History Effects Linked to an Advantage for wAu 854 Wolbachia in Drosophila. Insects 10, 126 (2019). 855 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 62. M. F. Richardson et al., Population genomics of the Wolbachia endosymbiont in Drosophila 856 melanogaster. Plos Genet 8, e1003129 (2012). 857 63. M. Detcharoen, F. M. Jiggins, B. C. Schlick-Steiner, F. M. Steiner, Wolbachia endosymbiotic 858 bacteria alter the gut microbiome in the fly Drosophila nigrosparsa. J Invertebr Pathol 198, 859 107915 (2023). 860 64. M. Martin, S. Lopez-Madrigal, I. L. G. Newton, The Wolbachia WalE1 effector alters Drosophila 861 endocytosis. PLoS Pathog 20, e1011245 (2024). 862 65. A. R. Lindsey et al., The intracellular symbiont Wolbachia alters Drosophila development and 863 metabolism to buffer against nutritional stress. bioRxiv 10.1101/2023.01.20.524972 (2024). 864 66. L. B. Nevalainen, E. M. Layton, I. L. G. Newton, Wolbachia Promotes Its Own Uptake by Host 865 Cells. Infect Immun 91, e0055722 (2023). 866 67. K. N. Bryant, I. L. G. Newton, The Intracellular Symbiont Wolbachia pipientis Enhances 867 Recombination in a Dose-Dependent Manner. Insects 11, 284 (2020). 868 68. R. K. Simhadri et al., The Gut Commensal Microbiome of Drosophila melanogaster Is Modified by 869 the Endosymbiont Wolbachia. mSphere 2, e00287-00217 (2017). 870 69. J. Oksanen et al. (2018) vegan: Community Ecology Package. 871 70. A. Dinno (2017) dunn.test: Dunn's Test of Multiple Comparisons Using Rank Sums. 872 71. S. Mangiafico (2022) rcompanion: Functions to Support Extension Education Program 873 Evaluation. 874 72. T. Hothorn, F. Bretz, P. Westfall, Simultaneous inference in general parametric models. Biom J 875 50, 346-363 (2008). 876 73. J. J. Kozich, S. L. Westcott, N. T. Baxter, S. K. Highlander, P. D. Schloss, Development of a dual-877 index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the 878 MiSeq Illumina sequencing platform. Appl Environ Microbiol 79, 5112-5120 (2013). 879 74. E. Bolyen et al., Reproducible, interactive, scalable and extensible microbiome data science 880 using QIIME 2. Nature biotechnology 37, 852-857 (2019). 881 75. R. Core Team (2021) R: A Language and Environment for Statistical Computing. (R Foundation 882 for Statistical Computing, Vienna, Austria). 883 76. B. J. Callahan et al., DADA2: High-resolution sample inference from Illumina amplicon data. 884 Nature methods 13, 581-583 (2016). 885 77. D. McDonald et al., An improved Greengenes taxonomy with explicit ranks for ecological and 886 evolutionary analyses of bacteria and archaea. The ISME journal 6, 610-618 (2012). 887 78. J. Oksanen et al. (2020) vegan: Community Ecology Package. 888 79. C. Lozupone, R. Knight, UniFrac: a new phylogenetic method for comparing microbial 889 communities. Appl Environ Microbiol 71, 8228-8235 (2005). 890 80. C. A. Lozupone, R. Knight, Global patterns in bacterial diversity. Proc Natl Acad Sci U S A 104, 891 11436-11440 (2007). 892 81. M. N. Price, P. S. Dehal, A. P. Arkin, FastTree 2--approximately maximum-likelihood trees for 893 large alignments. PLoS One 5, e9490 (2010). 894 82. K. Katoh, K. Misawa, K. Kuma, T. Miyata, MAFFT: a novel method for rapid multiple sequence 895 alignment based on fast Fourier transform. Nucleic Acids Res 30, 3059-3066 (2002). 896 83. S. Mandal et al., Analysis of composition of microbiomes: a novel method for studying microbial 897 composition. Microbial ecology in health and disease 26, 27663 (2015). 898 84. H. Chen (2021) VennDiagram: Generate High-Resolution Venn and Euler Plots. 899 85. A. R. Burns et al., Contribution of neutral processes to the assembly of gut microbial 900 communities in the zebrafish over host development. The ISME journal 10, 655-664 (2016). 901 86. W. T. Sloan et al., Quantifying the roles of immigration and chance in shaping prokaryote 902 community structure. Environ Microbiol 8, 732-740 (2006). 903 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint 87. A. Blanco-Miguez et al., Extending and improving metagenomic taxonomic profiling with 904 uncharacterized species using MetaPhlAn 4. Nature biotechnology 41, 1633-1644 (2023). 905 88. T. M. Therneau (2020) coxme: Mixed Effects Cox Models. 906 89. T. Therneau (2022) A Package for Survival Analysis in R. 907 90. H. Wickham, L. Henry (2020) tidyr: Tidy Messy Data. 908 91. H. Wickham, Reshaping data with the reshape package. Journal of Statistical Software 21, 1-20 909 (2007). 910 92. H. Wickham (2016) ggplot2: Elegant Graphics for Data Analysis. (Springer-Verlag New York). 911 93. A. Baptiste (2017) gridExtra: Miscellaneous Functions for "Grid" Graphics 912 94. H. Wickham, J. Bryan (2019) readxl: Read Excel Files. 913 95. H. Wickham, R. François, L. Henry, K. Müller (2022) dplyr: A Grammar of Data Manipulation. 914 96. C. O. Wilke (2020) cowplot: Streamlined Plot Theme and Plot Annotations for 'ggplot2'. 915 97. C. O. Wilke (2020) ggtext: Improved Text Rendering Support for 'ggplot2'. 916 98. C. O. Wilke, B. M. Wiernik (2022) gridtext: Improved Text Rendering Support for 'Grid' Graphics. 917 99. T. v. d. Brand (2024) ggh4x: Hacks for 'ggplot2'. 918 100. T. V. Elzhov, K. M. Mullen, A.-N. Spiess, B. Bolker (2022) minpack.lm: R Interface to the 919 Levenberg-Marquardt Nonlinear Least-Squares Algorithm Found in MINPACK, Plus Support for 920 Bounds. 921 101. F. E. H. Jr. (2021) Hmisc: Harrell Miscellaneous. 922 102. A. Kassambara (2020) ggpubr: 'ggplot2' Based Publication Ready Plots. 923 103. H. Wickham (2019) stringr: Simple, Consistent Wrappers for Common String Operations. 924 104. R. A. Becker, A. R. Wilks, R. Brownrigg, T. P. Minka, A. Deckmyn (2021) maps: Draw Geographical 925 Maps. 926 927 928 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint Table 1. Read counts of bacterial genera in flies relative to the fruits they were feeding on 929 930 Flies (total microbiota) Flies (resident microbiota) 931 Genus Flies Fruit % in fly p-value Flies Fruit % in fly chisq_p Commensalibacter 6113 0 100 10-99 6675 23 100 10-99 Lactobacillus n.t. 255 2 99 10-54 Enterococcus n.t. 416 18 96 10-79 Unassigned Enterobacteriaceae 2009 1187 63 10-43 1474 1683 47 0.0004 Unassigned Pasteurellales 641 393 62 10-13 550 290 66 10-17 Pseudomonas 92 100 48 0.62 61 189 24 10-14 Gluconobacter 3484 3878 47 0.00004 3839 3680 51 0.10 Acetobacter 2418 4565 35 10-121 2342 3923 37 10-74 Gluconacetobacter 1120 2574 30 10-113 489 2932 14 10-99 Unassigned Xanthomonadaceae 62 193 24 10-15 91 213 30 10-11 Unassigned Bacteria 88 1420 6 10-246 59 1991 3 10-99 Lactococcus 5 274 2 10-56 9 356 2 10-71 Unassigned Proteobacteria 1 393 0 10-84 n.t. 932 a n.t. = not tested, < 1% of total reads collected 933 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 8, 2024. ; https://doi.org/10.1101/2024.10.07.617096doi: bioRxiv preprint