Experimental setup to study the long-term evolution of L. plantarum in Drosophila simulans populations.

Pangenome of L. plantarum genomes isolated from the Florida experiment (Table S1), sorted by phylogeny.

Left panel: Maximum likelihood tree built based on the core genome alignment. Tree leaves correspond to the name of the isolate and are coloured based on isolation origin: ancestral population (grey), hot-evolved population (red) or cold-evolved population (blue). Nodes supported by bootstrap values ≥ 90% based on 1000 replicates are highlighted in yellow. Middle panel: Pattern of genes presence (black)/absence (white) in each genome. Right panel: pairwise average nucleotide identity between the isolates. The colour ranges between 98.5% (purple) and 100% (yellow).

Maximum likelihood tree of 92 L. plantarum genomes.

Tree leaves are coloured by source of isolation. The genomes from our lab are coloured according to experimental treatment: unevolved in grey, hot-evolved in red, and cold-evolved in blue. Nodes supported by bootstrap values based on 1000 replicates are highlighted in orange (≥ 70% bootstrap support) or yellow (≥ 90% bootstrap support).

Clade composition over time in each replicate population from both temperature regimes.

C in light blue, clade H in red and clade U in grey. Clade relative abundance was inferred by mapping competitively short reads against the three clades’ reference sequences.

Clade-specific growth dynamics at hot fluctuating 28/23 °C.

(A) Overlapped curves of all replicates. Lines are coloured by clade. Background colour represents the growth temperature at a given time. (B-D) Boxplots depicting the carrying capacity, growth rate, and inflection time of each clade. We measured the growth of four isolates from clade U, nine isolates from clade C and sixteen isolates from clade H (Supplementary table S1). Each isolate was grown three time. Each dot corresponds to a technical replicate. Data is grouped and coloured by clade. Statistical significance was determined using Dunn’s test with Holm-adjusted p-values. Only significant comparisons are indicated. **** p < 0.0001; *** p < 0.001; ** p < 0.01; * p < 0.05

Fitness effect of L. plantarum inoculation in axenic D. melanogaster and conventionally reared D. simulans in the first transfer (See Materials and methods).

(A, B) Total number of F1 flies eclosed normalized by day and female under the cold (A) and hot (B) regime. (C, D) Developmental time, estimated as the number of days it takes 50% of the offspring to eclose. Measurements were grouped by inoculation treatment and Drosophila species. Each dot corresponds to a biological replicate (n = 10 for D. melanogaster and n = 9 for D. simulans). Statistical significance was determined using Dunn’s test with Holm-adjusted p-values. Only significant comparisons are indicated. *** p < 0.001; ** p < 0.01; * p < 0.05.

Pangenome of all L. plantarum genomes (Table S1), sorted by phylogeny.

Left panel: Maximum likelihood tree built based on the core genome alignment. Tree leaves correspond to the name of the isolate and are coloured based on isolation origin: unevolved populations (grey), hot-evolved populations (red) or cold-evolved populations (blue). Tree tips are coloured based on the experiment from which the isolate was obtained: Florida (white), Portugal (pink) or South Africa (green). Middle panel: Pattern of genes presence (black)/absence (white) in each genome. Right panel: heat map depicting the pairwise average nucleotide identity between the isolates. The colour ranges between 98.5% (purple) and 100% (yellow).

Hierarchical clustering of 92 L. plantarum genomes based on the presence-absence of accessory genes.

Tree leaves are coloured by source of isolation. The genomes from our lab are coloured according to experimental treatment: unevolved in grey, hot-evolved in red, and cold-evolved in blue. Red values in each node indicate the p-value for each cluster’s robustness based on 1000 bootstrap iterations.

Relative abundance of each L. plantarum clade across different reads depths in simulated reads sets with known clade composition.

Dashed vertical lines show the values of 100, 500, and 1000 reads. Dashed horizontal lines show the true simulated relative abundances of each clade in the reads set. The top row titles reference to the ratio of reads from each clade following descending order: C:H:U. The bottom plots depict scenarios in which all the reads belong to a single taxon. Diamonds at the bottom of each panel show the sampled number of reads. The x-axis was log10-scaled to show better the sampling distribution.

Clade-specific growth dynamics at constant 20 °C.

(A) Overlapped curves of all replicates. Lines are coloured by clade. (B-D) Boxplots depicting the carrying capacity, growth rate, and inflection time of each clade. We measured the growth of four isolates from each clade (Supplementary table S1). Each isolate was grown six times, at two different starting points (See Materials and Methods). Each dot corresponds to a technical replicate. Data is grouped and coloured by clade. Statistical significance was determined using Dunn’s test with Holm-adjusted p-values. Only significant comparisons are indicated. **** p < 0.0001; *** p < 0.001; ** p < 0.01; * p < 0.05.

Fitness effect of L. plantarum inoculation in axenic D. melanogaster and conventionally reared D. simulans in the second transfer (See Materials and methods).

(A, B) Total number of F1 flies eclosed normalized by day and female under the cold (A) and hot (B) regime. (C, D) Developmental time, estimated as the number of days it takes 50% of the offspring to eclose. Measurements were grouped by inoculation treatment and Drosophila species. Each dot corresponds to a biological replicate (n = 10 for D. melanogaster and n = 9 for D. simulans). Statistical significance was determined using Dunn’s test with Holm-adjusted p-values. Only significant comparisons are indicated. ** p < 0.01; * p < 0.05.

Bacterial load of D. simulans individual flies inoculated with L. plantarum.

Four individual flies were allowed to feed in the corresponding inoculum for 24 h, and then moved to axenic vials for 72 h before crunching and plating their content. Measurements were grouped by inoculation treatment. Each dot corresponds to an individual fly (n = 4). A pseudo-count of 0.1 was added to fit zeros in the log-scaled y-axis. Statistical significance was determined using Dunn’s test with Holm-adjusted p-values. Only significant comparisons are indicated. ** p < 0.01; * p < 0.05.

Venn diagram depicting the overlap in KEGG Orthologs between the three clades.

Segments were coloured by number of KOs.

Graphical representation of the KEGG pathway map02060 (Phosphotransferase system).

Each rectangle represents a KEGG Ortholog (KO). Orthologs that are enriched in each clade are coloured in grey (present in clade U), blue (present in clade C) and/or red (present in clade H). The three clades differ in the set of sugar-related PTS transporters. Clades C and H have the capacity to internalize sorbitol and galactitol, whereas clades H and U have the ability to import galactosamine.

Graphical representation of the KEGG pathway map00740 (Riboflavin metabolism).

Each rectangle represents a KEGG Ortholog (KO). Orthologs that are enriched in each clade are coloured in grey (present in clade U), blue (present in clade C) and/or red (present in clade H). Clade U lacks the capacity to produce de novo FAD from GTP, since it is lacking several genes from the pathway.

Graphical representation of the KEGG pathway map00790 (Folate biosynthesis).

Each rectangle represents a KEGG Ortholog (KO). Orthologs that are enriched in each clade are coloured in grey (present in clade U), blue (present in clade C) and/or red (present in clade H). Clade U has the unique capacity to synthesize de novo guanylyl molybdenum cofactor.

Subset of the KEGG pathways map02020 (Two-component system; A), map00910 (Nitrogen metabolism; B), and map00920 (Sulfur metabolism; C).

Each rectangle represents a KEGG Ortholog (KO). Orthologs that are enriched in each clade are coloured in grey (present in clade U), blue (present in clade C) and/or red (present in clade H). Clade U encodes a unique alternative respiratory system. The operons nreABC and narGHIJ encode for genes that sense anoxic conditions and use nitrate as terminal electron donor instead of oxygen (A, B). The enzyme sulfur oxidoreductase (Nsr) is highly enriched in the clade C, which suggests that it can potentially use sulfur as terminal electron acceptor in absence of oxygen (C).

We quantified bacterial growth speed in solid fly food in an agar-drop assay (See Materials and Methods).

Briefly, six plates were cultivated with a 3x4 grid of 10 µl drops from each clade (rows) at four concentrations (columns); from the original suspension (0) to dilution 1:106 (6). Three plates were incubated at either hot (top panel) or cold (bottom panel) temperature conditions. We recorded growth progression after incubation days three and six for the hot, and six and ten for the cold. Regardless of the temperature regime, we observed delayed growth of clade U relative to clades C and H. This was seen as overall dimmer drops of this clade in dilution 2 in the first time point (top sub-panels). In the second time point, most drops had reached bacterial saturation (bottom sub-panels).