Lab Fitness Signals Versus Field Ecology

Tension

The Atlas uses lab fitness as a strong evidence layer, but the field is not a larger version of the lab. Fitness effects can transfer to ecology when the trait and metadata align; they can also fail when site chemistry, community interactions, or metadata gaps dominate.

Review Brief

What changed: field-validation pages now draw more heavily on environmental metadata, geochemistry, and multi-omics interpretation. That makes this conflict the review checkpoint for any page that wants to move from lab signal to field claim.

Why review matters: lab fitness is one of BERIL's strongest evidence layers, but overgeneralizing it would weaken the Atlas. Reviewers should decide where it is a prior, where it is validated, and where it is insufficient.

Evidence to inspect:

• lab_field_ecology and field_vs_lab_fitness for transfer successes and limits.
• bacdive_metal_validation for phenotype/isolation-context validation.
• enigma_sso_asv_ecology for site-level ecology and missing geochemistry.
• Data completeness in Environment Harmonization.

Questions for reviewers:

• Which field claims should require measured geochemistry instead of environment labels?
• Are lab fitness scores being used as priors or as direct predictors?
• What metadata-quality threshold should be required before field validation is considered strong?
• Should failure cases become a separate derived benchmark rather than caveat text?

Current Interpretation

Treat lab fitness as a reusable prior, not a universal predictor. Field validation should preserve environment labels, geochemistry, sampling context, and missingness.

Resolving Analysis

The highest-value analysis is a matched benchmark: for each organism or gene family, compare lab-derived scores against field abundance or isolation context while explicitly modeling site chemistry and metadata quality.