Species can conserve particular functions across their own members while differing from other species in the breadth of those functions. In the metal-tolerance analyses, this layered model explains how relatively conserved metal-tolerance functions can coexist with species-level differences in genome-derived scores and environmental distributions. 1 The Pan-Bacterial Metal Fitness Atlas refines this model by separating broadly required core cellular functions from specialized accessory resistance functions: across 22 organisms and 14 metals, metal-important genes were 87.4% core versus 76.9% for baseline genes (OR=2.08, p=4.3e-162). 2
Evidence from Metal-Tolerance Scores
The BacDive Isolation Environment × Metal Tolerance Prediction tested whether isolation environments provide ecological validation for genome-based metal-tolerance scores by linking bacdive strains to gtdb pangenome species. 1 Among 97,334 BacDive strains, 42,227 (43.4%) matched 6,426 GTDB species, and 25,089 matched strains had isolation-source metadata. 1
Heavy-metal contamination isolates had a median metal score of 0.240 and a mean of 0.236, compared with an environmental-baseline median of 0.187 and mean of 0.195. 1 Their median delta was +0.053, with Cohen’s d = +1.00 and Mann-Whitney p=0.006 for n=10 isolates. 1 The contamination gradient was heavy metal d=+1.00, waste/sludge d=+0.57, all contamination d=+0.43, and industrial environments d=+0.20, with n=10, n=305, n=176, and n=796 isolates, respectively. 1 These environment-associated differences support between-species divergence in aggregate metal-tolerance capacity, even when individual functions may be conserved within species. 1
The new MicrobeAtlas analysis supports this between-species component at broader ecological scale: among 606 bacterial genera with AMR data, metal-type diversity was positively associated with inferred niche breadth in phylogenetic generalized least-squares (PGLS), with β = +0.021, SE = 0.0056, and p = 1.5×10⁻⁴, while total AMR cluster count and core AMR fraction were not significant. 3 This refines the score-based interpretation by indicating that breadth across metal types, rather than simply more resistance genes, distinguishes broad-niche genera, although the genus-level association is correlational and does not establish causality. 3
The score was normalized as metal clusters divided by annotated clusters, which controls for genome size but does not eliminate the possibility that metal-tolerance functions correlate with total metabolic complexity. 1 This denominator is therefore not necessarily neutral: differences in annotation breadth can alter both the apparent functional numerator and the reference set used for normalization. 1 Organisms with larger genomes or more broadly annotated gene clusters may receive different normalized scores because genome size, annotation coverage, and functional complexity are correlated. 1
The MicrobeAtlas result supports an association beyond genome size but does not remove the same interpretive concern: after controlling for genome size, metal-type diversity remained associated with niche breadth at β = +0.0218, SE = 0.0061, p = 3.6×10⁻⁴, while log(genome size) also had a positive association of β = +0.0263, SE = 0.0069, p = 1.5×10⁻⁴. 3 Thus, both studies require separation of metal-function effects from broader genomic and annotation complexity rather than treating a normalized or diversity score as a standalone causal measure. 1,3
The metal-fitness-atlas supports the existence of between-species functional breadth while sharpening this caveat: its atlas combined 559 experiments across 31 organisms and 16 metals, but metal coverage was uneven and concentrations were not normalized by dose or tolerance threshold. 2 Its genome-size-normalized repertoire score did not predict metal fitness, and the report therefore favors regulatory or expression-based models over simple presence/absence comparisons. 2
Conserved and Specialized Functional Layers
The gene-resolution Gene-Resolution Metal Cross-Resistance Across Diverse Bacteria supports this layered interpretation. Across 8,162 metal-important genes, general-stress, metal-shared, and metal-specific tiers had mean pangenome core fractions of 92.0%, 91.0%, and 89.8%, respectively; the percentages fully core at at least 95% were 57.2%, 50.4%, and 45.7%. 4 Thus, broadly conserved functions coexist with a less-conserved layer of specialized metal functions that can differentiate species. 4
The cross-resistance analysis identified 318 ortholog groups that were metal-shared in at least 2 organisms, with conserved families spanning cell-envelope, energy-metabolism, DNA-repair, and ion-homeostasis functions. 4 It therefore refines the score-based interpretation by separating general-stress, metal-shared, and metal-specific architectures rather than treating all metal-associated functions as equivalent. 4
The metal-fitness atlas supports the general-stress component but refines the expected conservation pattern. Metal-important genes were more core than baseline in 21 of 22 organisms, while essential-metal tolerance genes for Fe, Mo, W, Se, and Mn had a mean core-fraction delta of +0.148 versus +0.081 for toxic metals (one-sided Mann-Whitney U: U=39, p=0.015). 2 The strongest reported enrichments included manganese (+0.198, all 30 important genes core), zinc (+0.151), molybdenum (+0.148), tungsten (+0.145), and iron (+0.116), whereas cadmium had delta=-0.010 (p=0.92) and uranium had delta=+0.035 (p=0.34). 2 This supports conserved cellular robustness, but the uneven coverage of rare metals means that some cross-species patterns largely reflect DvH and psRCH2 biology. 2
The MicrobeAtlas analysis supports this distinction at the genus level: metal-type diversity had intermediate phylogenetic signal, Pagel’s λ = 0.335, compared with λ = 0.441 for core AMR fraction and λ = 0.260 for total AMR cluster count, whereas niche breadth had λ = 0.787. 3 This pattern is consistent with conserved ecological and functional structure coexisting with more labile resistance repertoires, but the report attributes the architecture to a mixture of vertical inheritance and horizontal gene transfer rather than directly measuring within-species conservation. 3
The metal-fitness atlas further supports the proposed two-tier model: its interpretation assigns most observed metal-important genes to conserved core functions such as envelope processes, DNA repair, central metabolism, protein quality control, and general stress response, while accessory efflux, sequestration, and detoxification functions form a specialized layer. 2 It identified 149 conserved candidates without full functional annotation—89 truly unknown, 43 DUF/UPF, and 17 with partial hints—so the proposed species-divergent layer remains partly unresolved functionally. 2
The BacDive report attributes an 88% core estimate for metal-tolerance genes to the separate Metal-Specific vs General Stress Genes project. Within this report the figure is secondhand evidence, not a new measurement. 1 The report proposes that core enrichment describes the within-species pattern, in which most metal genes are present in most genomes of a species. Under this proposal, between-species variation in the total number of metal-tolerance genes creates the environmental signal. 1 This reconciliation is an interpretation, not a separate test. The project did not directly test causality between core-genome size, metal-tolerance score, and contamination occupancy. 1
Genome Size, Annotation Breadth, and Ecological Interpretation
The genome-size mechanism is especially important for interpreting the smaller host-associated effect. Host-associated isolates had a metal-tolerance effect size of d=+0.14 with p<0.0001, whereas contamination categories ranged from d=+0.43 to d=+1.00. 1 The report attributes the host-associated elevation as likely reflecting genome-size confounding because host-associated BacDive records were dominated by pseudomonas, klebsiella, and acinetobacter baumannii pathogens, which have larger genomes and more KEGG-annotated gene clusters. 1 Genome-size normalization reduced but did not eliminate this possible effect. 1 The smaller host-associated association therefore does not establish the same ecological interpretation as the stronger contamination-associated signals. 1
The contamination results support an ecological metal-adaptation signal, but they do not show that it is independent of genome size, annotation breadth, or correlated metabolic complexity. 1 The report proposes that species with larger core genomes encoding more metal-tolerance functions could score higher and be more likely to occur in contaminated environments. 1 This remains a hypothesis rather than a direct causal test: normalized scores may combine metal adaptation with genome-scale and annotation-scale differences. 1
A robust interpretation should distinguish three quantities: the absolute number of metal-tolerance clusters, the total number of annotated clusters used as the denominator, and the residual association with isolation environment after accounting for genome size and annotation breadth. 1 This refines the use of the metal-fitness-atlas score from a standalone ecological proxy toward a quantity requiring covariate-aware validation. 1
The MicrobeAtlas study supports the need for this separation: the metal-type association persisted after adding pangenome coverage and in a three-covariate model including genome size and the number of sampled species per genus, but pangenome coverage itself was uneven and correlated with inferred metal-type diversity at r = 0.35 with p = 2.6×10⁻¹⁹. 3 Its strict 5% prevalence analysis retained a positive coefficient but was not significant (β = +0.0166, SE = 0.0099, p = 0.092), showing that detection and sampling choices can affect the apparent ecological breadth. 3
Phylum-Stratified Support and Limits
The contamination signal persisted within pseudomonadota, where contamination isolates had metal scores delta=+0.040 above environmental isolates (p<0.001; n=85 contamination and n=5,655 environmental), and within actinobacteria, where the delta was +0.035 (p<0.001; n=62 and n=772, respectively). 1 These results support functional divergence beyond broad phylum composition in the two most-sampled phyla, but they do not by themselves remove genome-size or annotation-breadth confounding. 1
The signal was not significant within bacillota-b, where delta=-0.012 and p=0.285 (n=19 contamination and n=492 environmental), or bacteroidota, where delta=-0.008 and p=0.456 (n=5 and n=156, respectively). 1 These null results cannot distinguish biological differences from limited statistical power because contamination-isolate sample sizes were small. 1
Between-Species Divergence Beyond Metal Tolerance
Carbon-catabolism evidence from pseudomonas extends the layered model beyond metals. Pseudomonas sensu stricto, the pseudomonas-aeruginosa group, shows near-complete loss of plant-derived sugar catabolism compared with pseudomonas e, the P. fluorescens/putida group. 5 Of the 62 gapmind carbon pathways tested, 43 differed significantly between the two subgenera. The test was Mann-Whitney U at BH-FDR < 0.05, where FDR is the false discovery rate. 5 The Carbon Source Utilization Predicts Ecology and Lifestyle in Pseudomonas states that pathway-completeness differences between these host-associated and free-living subgenera exceeded 50 percentage points for 7 pathways and 10 percentage points for 15 pathways. 5
The tested comparison set the 7 Pseudomonas s.s. species against the 189 Pseudomonas_E species, each with >=5 genomes. 5 The central discoveries digest describes the wider analysis as covering 433 Pseudomonas species (12,732 genomes). Those numbers give the size of the analyzed dataset, not the denominators of the subgenus comparison. The digest also reports 43 of 62 pathways differing at FDR < 0.05. 6 The largest effects were in xylose (+74 pp), ribose (+64 pp), arabinose (+63 pp), galacturonate (+60 pp), and myo-inositol (+59 pp). Amino acid catabolism remained near-universal (>99.5%) in both groups. 6 This supports the two-layer model, with conserved amino-acid catabolism alongside divergent sugar catabolism. The divergence is measured between subgenus-level groups rather than between individual species pairs. 6,5
Within P. aeruginosa (PA), the Rational Design of Protective Microbiome Formulations for Competitive Exclusion of *Pseudomonas aeruginosa* in Cystic Fibrosis Airways supports the within-species conservation component. Across 1,796 lung PA genomes, amino acid catabolic pathways had 97.4% mean completeness. Proline, PA14’s top substrate, was complete in 97% of lung isolates. 7 The main variation among lung PA lay in carbon-source pathways such as sorbitol, mannitol, and gluconate. The first principal component, which captures this variation, is reported as PC1 = 79%. 7 CF versus non-CF PA showed zero statistically significant amino acid pathway differences. This is a null result, not demonstrated exact identity, and equivalent formulation efficacy across PA variants remains untested. 7
The same report contrasts its >95% metabolic conservation finding with Shao et al. (2026), which it cites as finding strong strain-level functional divergence in Bifidobacterium. Within-species conservation therefore appears taxon-dependent rather than universal. This contrast rests on a cited external study rather than a BERIL analysis. 7
Cell-envelope evidence from the Regulatory and proteomic architecture of Δfur-permitted lipid A loss in *Caulobacter crescentus* gives a gene-presence example of between-species divergence. The report ran a comparative PaperBLAST presence/absence screen across caulobacter-crescentus, acinetobacter baumannii, neisseria meningitidis, and moraxella catarrhalis. 8 In that screen, sphingolipid biosynthesis genes were found in Caulobacter but were absent in the other three species. These genes were spt (serine palmitoyltransferase), bcerS (bacterial ceramide synthase), and sphingosine kinase. 8 The chvg–chvi envelope-stress regulatory circuit was also Caulobacter-restricted, consistent with its published alphaproteobacterial restriction. 8 NCBI annotation reported pattern 1000 for spt, cerR (ceramide reductase), ChvG, and ChvI, meaning each was found only in Caulobacter among the four species. 8 This is a four-species comparison and does not measure within-species conservation. 8
The Caulobacter report also refines the model: divergent gene content can still lead to analogous outcomes. 8 A. baumannii Δlpxc survival via PBP1A loss and LdtJ/LdtK is described as mechanistically distinct from, but biologically analogous to, Caulobacter Δlpxc engagement of SdpA/PleA and Pal-Tol peptidoglycan remodeling. Both organisms reorganize peptidoglycan when LPS/LOS is removed, but through different enzyme-family specifics. 8 The report’s revised reading of tol pal system upregulation, which it labels reframed, is narrower than structural remodeling alone. Citing Tan & Chng 2025, it treats the primary Tol-Pal function as retrograde phospholipid transport that maintains outer-membrane lipid homeostasis. Upregulation in Δlpxc is described as consistent with increased retrograde transport, which is an inference rather than a direct measurement. 8 The underlying outer-membrane proteome is single-replicate, and the report states that replicated proteomics are required. 8 Between-species divergence may therefore lie in the genes used rather than the function achieved. This hypothesis is discussed further under outer-membrane-lipid-homeostasis. 8
Tensions
There is a validation-scale tension between the positive BacDive association and the cross-resistance study’s null species-scale test. BacDive found Cohen’s d = +1.00 for heavy-metal contamination in n=10 isolates compared with approximately 5,000 environmental-baseline strains. 1 The cross-resistance report describes that prior project as a pangenome-scale analysis (42K strains, Cohen’s d = +1.0). 4 The 42K figure reflects the scale of the matched-strain bridge, not the denominator of the effect size. It is therefore not a separate validation result. 1,4 In contrast, the cross-resistance study found no correlation between multi-metal tolerance scores and BacDive metal-environment isolation at Fitness Browser species scale (Spearman rho approximately -0.02, p > 0.8); after matching and collapsing strains, it retained 20 independent species and judged the test underpowered. 4 These results do not establish whether the discrepancy reflects scale, matching, phenotype definition, or sampling. 4
The metal-fitness atlas supports the conservation side of this tension through its 87.4% core fraction, but it also refines the ecological interpretation: the broad metal-important definition (fit < -1 OR n_sick ≥ 1) captures general stress functions as well as metal-specific resistance, and the simple repertoire score failed to predict fitness. 2 Thus, core conservation does not by itself validate normalized environmental scores or imply that gene presence predicts metal tolerance. 2
The MicrobeAtlas result supports the positive ecological-association side but does not resolve this tension: metal-type diversity predicted broader inferred niche breadth after phylogenetic adjustment, yet the analysis used genus-level occurrence proxies rather than direct metal-exposure phenotypes. 3 Its groundwater validation was also mixed: metal-type diversity correlated with groundwater prevalence (Spearman ρ = +0.112, p = 0.0019), but groundwater-specific fold-enrichment was null (ρ = +0.042, p = 0.242). 3 These findings therefore strengthen the broader-range association without demonstrating a uniquely metal-contamination-driven mechanism. 3
The metal-cross-resistance study supports the conservation component through broadly positive gene-level responses across 28 organisms, but its organisms were not phylogenetically independent. 4 Its gene-level conservation analysis does not directly measure within-species environmental occupancy, so it strengthens the proposed architecture without establishing the causal link left open by the BacDive study. 4 PGLS (phylogenetic generalized least squares) or independent contrasts would strengthen the conservation claim. 4
The heavy-metal estimate is also imprecise: n=10 matched isolates produced an observed d=1.00 that barely exceeded the approximately d=0.93 minimum detectable effect size reported for 80% power with n=10 heavy-metal isolates versus approximately 5,000 environmental-baseline strains. 1 The contamination gradient therefore supports the layered model, but the strongest estimate requires replication and explicit adjustment for genome size, annotated-cluster count, taxonomy, and sampling source. 1
The BacDive bridge creates an additional evidence limitation. Of 97,334 strains, 55,107 (56.6%) were unmatched, primarily because GTDB species boundaries differed from LPSN/DSMZ-based naming; the observed pattern may therefore not represent all described or environmental bacteria. 1 BacDive represents culturable, described strains rather than the full diversity of environmental bacteria and may under-represent metal-tolerant extremophiles. 1 This incomplete pangenome-integration bridge can also change which genome-size and annotation-breadth classes enter the analysis. 1
The interpretation connects to environmental-resistome because species-level conservation and divergence can be obscured when strain-level variation and environmental sampling are treated as interchangeable. 1
Open Directions
- Match BacDive GCA accessions directly to kbase-ke-pangenome genome IDs and test whether the environmental score gradient persists after recovering part of the 56.6% unmatched set; test whether matching failure is associated with genome size or isolation environment. 1
- Compare within-species strain variation and between-species score differences using accession-linked pangenome genomes, and test whether species-level divergence explains more variance than strain-level heterogeneity. 1
- Refit the BacDive environment comparisons with genome size, annotated-cluster count, taxonomic group, and culture-source category as covariates; separate absolute metal-tolerance cluster counts from denominator effects and test whether the contamination association remains. 1
- Compare absolute metal-tolerance cluster counts with normalized scores across the same matched strains and test whether ecological rankings change when annotation breadth is held constant. 1
- Expand matched metal phenotypes beyond the 24 existing utilization records and current metabolite-utilization records with MIC and growth-inhibition data; test whether measured phenotypes and specific metal-tolerance gene families track normalized scores independently of genome and annotation breadth. 1
- Integrate enigma-coral community data from the Oak Ridge metal-contaminated site and compare field distributions or abundance associations with genome-size-adjusted, species-level pangenome scores. 1
- Apply phylogenetic independent contrasts or PGLS to the 28-organism cross-resistance dataset and compare gene-tier conservation with species-level environmental scores. 4
- Map cross-resistance gene signatures with KEGG/PFAM across 27K species and test whether pangenome-scale validation resolves the species-scale null result. 4
- Reanalyze the MicrobeAtlas association with finer environmental categories, alternative phylogenetic trees, and genus-level rarefaction curves; test whether metal-type diversity predicts direct metal-exposure phenotypes rather than sequencing-effort proxies. 3
- Reanalyze the Metal Fitness Atlas after excluding genes important under non-metal stresses, normalize exposure by concentration relative to organism-specific tolerance thresholds, and test whether core conservation and species-level ecological associations remain. 2
- Functionally characterize the 149 conserved metal-fitness candidates using PaperBLAST, InterPro, and structural prediction, then test whether candidates explain within-species conservation or between-species metal-response differences. 2
- Quantify within-species GapMind pathway completeness in both Pseudomonas subgenera and in lung PA. Test whether the 43 carbon pathways that diverge between subgenera are also the most variable within species, as the dominance of carbon pathways in lung PA PC1 variation suggests. 5,7
- Extend the four-species presence/absence screen for spt, bcerS, cerR, and ChvG–ChvI to pangenomes of species that carry them. Test whether these Caulobacter-restricted envelope functions are core within those species. 8