BacDive Isolation Environment × Metal Tolerance Prediction

The Metal Fitness Atlas scored 27,702 pangenome species for metal tolerance using a genome-based functional signature. This project validates those predictions against BacDive's isolation source metadata for 97K strains. By linking BacDive genome accessions to pangenome species, we test whether organisms from heavy-metal contamination sites, industrial environments, and waste/sludge sites have higher predicted metal tolerance than organisms from host-associated or uncontaminated environments.

Bridge Summary

Environment Comparisons

Phylum-Stratified Results

Genome Content Predicts Environmental Metal Tolerance

The central finding — that heavy metal contamination isolates score d=+1.0 above environmental baseline — validates the Metal Fitness Atlas prediction method. The genome-based metal tolerance score, derived from cross-species RB-TnSeq fitness data and projected onto 27,702 pangenome species via KEGG functional annotations, captures real ecological signal. Bacteria that have been isolated from metal-contaminated environments genuinely carry more metal tolerance gene content than bacteria from uncontaminated environments.

The dose-response relationship (heavy metal > waste > contamination > industrial) further supports this interpretation: environments with higher expected metal concentrations produce isolates with higher predicted metal tolerance.

Reconciling With the Core Genome Robustness Model

The metal_specificity project found that metal tolerance genes are 88% core — predominantly conserved across species. If metal tolerance is core, how can it also vary across species enough to predict isolation environment? The answer is that core enrichment describes the within-species pattern (most metal genes are present in most genomes of a species), while the between-species variation in the total number of metal tolerance genes creates the environmental signal. Species with larger core genomes encoding more metal tolerance functions score higher — and are more likely to be found in contaminated environments.

The Lab_Field_Ecology Connection

The lab_field_ecology project found a suggestive but non-significant correlation between lab metal tolerance (Fitness Browser) and field abundance at Oak Ridge (rho=0.50, p=0.095, n=11 genera). The present analysis provides stronger evidence for the same hypothesis at much larger scale (n=10 for heavy metal, n=176 for all contamination, d=+0.43-1.00). The difference in significance reflects the power gain from BacDive's larger sample size compared to the Oak Ridge site-specific analysis.

Phylogenetic Confounding Is Partial

The phylum-stratified analysis shows the signal is real within Pseudomonadota and Actinomycetota but absent in Bacillota and Bacteroidota. Two explanations:
1. Real biology: Pseudomonadota and Actinomycetota include the major soil bacteria that encounter metals in natural environments, while Bacillota and Bacteroidota are primarily gut-associated and rarely encounter metal contamination.
2. Statistical power: Contamination isolates from Bacillota and Bacteroidota are rare in BacDive (culture collection bias toward Pseudomonadota), limiting power.

Both explanations are consistent with the data and are not mutually exclusive.

Host-Associated Bacteria Score Higher Than Expected

H1d predicted that host-associated bacteria would have lower metal scores than environmental bacteria, reflecting reduced metal exposure in host niches. The data show the opposite: host-associated bacteria score slightly but significantly higher (d=+0.14, p<0.0001). This likely reflects genome size confounding: host-associated bacteria in BacDive are dominated by Pseudomonadota pathogens (e.g., Pseudomonas, Klebsiella, Acinetobacter) with large genomes and correspondingly more KEGG-annotated gene clusters, inflating the normalized metal score. The genome-size normalization (metal clusters / annotated clusters) reduces but does not eliminate this effect because metal tolerance functions correlate with total metabolic complexity. This finding does not undermine H1a-H1c because the contamination signal (d=+0.43-1.00) is much larger than the host-associated signal (d=+0.14).

Limitations

• Small heavy metal group: n=10 after matching is at the detection limit for d=1.0. The effect size estimate is imprecise.
• Culture collection bias: BacDive represents culturable, described strains — not the full diversity of environmental bacteria. Metal-tolerant extremophiles may be under-represented.
• Species-level matching is lossy: 56.6% of BacDive strains didn't match any GTDB species, primarily because GTDB uses different species boundaries than LPSN/DSMZ. Genome accession-based matching (GCA→pangenome genome_id) would improve coverage but requires a Spark query.
• Metal tolerance score is genome-size-normalized: The metal_score_norm controls for genome size, which is important because Pseudomonadota tend to have larger genomes. Without normalization, the environmental signal could partly reflect genome size rather than metal gene content.
• Metal utilization phenotypes inconclusive: Only 24 records matched — far too few for reliable statistics.

1. GCA accession-based matching: Query kbase_ke_pangenome.genome to match BacDive GCA accessions directly to pangenome genome_ids, potentially recovering more of the 56.6% unmatched strains.
2. ENIGMA CORAL integration: Cross-reference with ENIGMA community data at the Oak Ridge metal-contaminated site for a complementary field validation.
3. Per-metal environment analysis: For metals with sufficient BacDive representation (iron, manganese), test whether specific metal tolerance gene families predict specific metal contamination environments.
4. Expand BacDive metal phenotypes: The BacDive API may contain additional metal tolerance test data not captured in the current metabolite_utilization table (e.g., MIC data, growth inhibition).

Generated Data

Figures

• Reimer LC et al. (2022). "BacDive in 2022: the knowledge base for standardized bacterial and archaeal data." Nucleic Acids Research 50:D741-D746.
• Carlson HK et al. (2019). "The selective pressures on the microbial community in a metal-contaminated aquifer." ISME J 13:937-949.
• Metal Fitness Atlas (this observatory) — projects/metal_fitness_atlas/REPORT.md
• Metal Specificity (this observatory) — projects/metal_specificity/REPORT.md
• Lab Field Ecology (this observatory) — projects/lab_field_ecology/REPORT.md