Resistance islands are recurring groups of antimicrobial-resistance (AMR) genes that co-occur within strains, providing a way to study whether resistance mechanisms are inherited as linked modules rather than independently. In the analyzed pangenome resource, co-occurrence indicates genomic linkage or shared lineage history, but it does not by itself establish co-selection, functional synergy, or physical location on a particular mobile element. 1
Evidence from Within-Species AMR Variation
The study detected 1,517 resistance islands across 705 species, representing 54% of the species analyzed. 1 Islands had a mean size of 6.2 genes, a median size of 4 genes, a maximum size of 43 genes, and a mean phi coefficient of 0.827. 1 The cross-project discoveries digest reports the same result as 1,517 resistance islands in 705/1,305 species, with mean phi coefficient = 0.827, 88% containing multiple resistance mechanisms, and a maximum island of 43 genes. 2 The phi coefficient measures pairwise co-occurrence, so the mean value indicates very tight co-inheritance among genes assigned to these islands. 1
Of the 1,517 islands, 1,343 (88%) contained genes from multiple resistance mechanisms. 1 Efflux pumps occurred in 954 islands and enzymatic inactivation occurred in 698 islands, making them the most common components reported in the analysis. 1 The complete mechanism counts were 1,026 for Other/Unclassified, 954 for Efflux, 698 for Enzymatic inactivation, 694 for Oxidoreductase, 502 for Regulatory, 341 for Beta-lactamase, 293 for Target modification, and 137 for Cell wall modification. 1
The report interprets the prevalence of multi-mechanism islands as suggesting that resistance islands provide coordinated defense against multiple drug classes simultaneously; this is an interpretation of island composition, not a directly demonstrated functional effect. 1 This evidence supports cofitness-network-architecture by showing that strong AMR gene co-occurrence can form a structured network of tightly linked modules rather than only isolated pairwise associations. 1 The result remains an association: genes may be linked on the same mobile genetic element or maintained by lineage history without providing functional synergy or being jointly selected. 1
The broader pangenome atlas refines this interpretation by showing that AMR is generally more accessory than the surrounding genome: only 30.3% of AMR genes were core versus 46.8% for the pangenome baseline, while the auxiliary genome was 2.2x enriched for AMR (33.6% versus 15.3%). 3 In a paired test of 4,252 species with at least 5 AMR clusters, 63.7% had AMR less core than their species baseline (Wilcoxon p=1.1e-130, mean difference -0.102). 3 These results support the possibility that some resistance islands represent mobile or variably inherited modules, but they do not establish physical linkage for the statistically defined islands. 3
The prophage-AMR analysis refines this distinction by providing partial genomic-context evidence without demonstrating mechanistic mobilization. Across 1,953 genomes sampled from 100 species, 20,073 of 36,041 AMR gene instances (55.7%) were on contigs carrying strict prophage markers, but only 1,991 (5.5%) were within 5 genes of a marker; among the AMR genes co-localized on prophage-marker contigs (not the full AMR population analyzed), the median distance to the nearest prophage marker was 34 genes. 4 AMR genes within 10 genes were 67.6% accessory versus 65.5% for distal genes; the odds ratio was 1.10 with one-sided p=0.005 and bootstrap 95% confidence interval [1.024, 1.185], a statistically significant but modest 2.1 percentage point difference. 4 The threshold-dependent odds ratios—0.78 at 3 genes, 0.92 at 5 genes, 1.10 at 10 genes, 1.19 at 15 genes, and 1.28 at 50 genes—together with a median species-level odds ratio of 0.85 show heterogeneous proximity effects rather than uniform physical co-inheritance. 4 This supports the existing interpretation that accessory status and statistical co-occurrence are insufficient to establish resistance-island structure or phage-mediated transfer.
Relationship to Phylogeny and Acquisition
AMR repertoire similarity was significantly associated with phylogenetic similarity in 701 of 1,261 species (55.6%) tested using Mantel tests at FDR < 0.05, where FDR is the false-discovery rate. 1 The median Mantel r for all AMR genes was 0.247, and 87.8% of species showed positive correlation between ANI distance and AMR Jaccard distance. 1 These results indicate that resistance-island distributions can reflect stable lineage-associated inheritance in addition to recent horizontal acquisition. 1
Non-core, putatively acquired AMR genes had a median Mantel r of 0.222, compared with 0.117 for core, intrinsic genes; the paired t-test gave t = -8.35, p = 7.0e-16, and n = 489. 1 This finding supports the hypothesis that acquired resistance elements can become stably maintained and vertically transmitted within lineages. 1 However, the comparison is partly constrained by the low Jaccard-distance variance of near-universal core genes, so the stronger non-core signal cannot be interpreted as a purely biological measure of island inheritance. 1
The atlas supports the intrinsic-acquired distinction at larger scale: beta-lactamases were 54.9% core, whereas regulatory genes were 6.5% core; blaTEM, tet(C), and ant(2”)-Ia were 0% core, while intrinsic efflux pumps such as emhABC were >95% core. 3 This pattern is consistent with vertically inherited intrinsic resistance being more stable than many acquired genes, but the atlas also cautions that AMRFinderPlus includes stress-response genes alongside classical antibiotic-resistance genes. 3
The environmental-resistome analysis refines this lineage-focused interpretation: clinical species had a median of 5 AMR gene clusters versus 2 for soil, aquatic, and host-associated species, while accessory AMR comprised 68% in clinical species versus 43% in soil species. 5 Environment effects also remained significant within 5 of 6 major phyla and in 20 of 141 testable families, supporting an ecological contribution beyond phylogeny while leaving most families insufficiently resolved. 5 These are species-level ecological associations, not evidence that the observed resistance islands are physically linked or co-selected.
The atlas supports this ecological qualification: human/Clinical species carried 10.6 AMR clusters per species, compared with 4.6 for Soil/Terrestrial, 3.9 for Aquatic, and 3.0 for Animal, with a significant Kruskal-Wallis test (H=440, p=7.0e-93). 3 Clinical AMR was less core (30.8%) than soil AMR (58.1%) or plant AMR (63.1%), consistent with greater representation of acquired resistance in clinical settings, although the comparison remains observational. 3 Among 2,684 species with at least 3 genomes and AlphaEarth embeddings, environmental diversity predicted AMR count (Spearman rho=0.466, p=1.6e-144), while environmental diversity and AMR core fraction were negatively correlated (rho=-0.173, p=1.8e-19); these associations do not identify island-level co-selection. 3
At the species level, the prophage-AMR analysis supports a broader mobile-element or acquisition-context explanation while retaining the phylogenetic caveat: prophage marker density correlated with AMR repertoire breadth (Spearman rho=0.572, p<10^-300, n=4,770 species), with partial Spearman rho=0.464 after controlling for genome count (p=1.0×10^-253). 4 The association was significant across Pseudomonadota (rho=0.54), Bacillota_A (rho=0.55), Bacillota (rho=0.40), Bacteroidota (rho=0.59), and Actinomycetota (rho=0.29). 4 This cross-phylogenetic pattern is consistent with prophage-mediated transfer or shared recombination potential, but it does not distinguish those mechanisms from species with open pangenomes independently accumulating both prophages and AMR genes. 4
Mechanistic Interpretation and Limits
The 88% prevalence of multi-mechanism islands suggests the hypothesis that resistance modules may provide broad protection against multiple drug classes or may be co-maintained because their genes share genomic context. 1 The environmental analysis supports ecological co-exposure as one plausible contributor to that maintenance: among AMR clusters classified to a mechanism (unclassified clusters were excluded from these fractions), metal resistance represented 45.0% in aquatic species and 44.0% in soil species, whereas target modification represented 43.6% in human-gut species. 5 However, that analysis is observational and does not establish island-level co-selection; the available co-occurrence analysis likewise does not distinguish among physical linkage, ecological co-exposure, clonal expansion, co-selection, and direct functional interaction. 1 Consequently, the evidence supports mechanistic linkage as a testable model, but does not establish that the linked genes act synergistically. 1
The prophage analysis supports this caution: prophage density tracks AMR breadth at species scale, but proximity to prophage markers has only a modest and heterogeneous relationship with AMR accessory status. 4 Prophage markers were identified by keyword and Pfam matching in bakta_annotations rather than dedicated prophage-prediction tools, and gene distances were parsed from ordinal gene positions rather than base-pair coordinates; false-positive markers and inaccurate distances therefore limit mechanistic inference. 4 The 100-species proximity analysis also sampled 20 genomes per species rather than all 293K genomes. 4 The report’s attempted fitness comparison could not be tested because the KBase Data Lakehouse fitness browser contains RB-TnSeq (random barcode transposon sequencing) data for only 48 model organisms with poor overlap to the GTDB species analyzed. 4
The atlas refines the mechanism evidence by showing that 18,448 of 83,008 AMRFinderPlus hits (22.2%) were Other/Unclassified and that mechanism labels were assigned by keyword matching against product descriptions rather than CARD ARO terms. 3 Thus, both island-level mechanism composition and any inferred cross-mechanism linkage should be treated as annotation-sensitive; the atlas also reports that 55.4% of sparsely annotated one-source clusters were singletons versus 34.6% of two-source clusters. 3
The report specifically identifies genomic-context mapping to plasmids, chromosomes, integron boundaries, and insertion sequences as a next step for resolving the mechanism of island formation and maintenance. 1 The report also proposes, but does not demonstrate, a PanKA-style predictive modeling approach that would use the resistance-island co-occurrence structure to predict which AMR genes are likely to be co-acquired in the future. 1 Separately, the pangenome atlas lists as a proposed analysis, not a completed one, whether AMR genes are clustered in genomic islands and whether they co-localize with mobile element markers such as IS elements and integrons. 3 Such analyses would connect the statistical islands to physical genomic structures and help distinguish horizontally transferred cassettes from chromosome-linked lineage signatures. 1
Relation to Other Concepts
This concept refines resistance-island-coinheritance by separating the observed statistical co-inheritance of resistance genes from the stronger, unproven claim of functional co-selection. 1 It also supports environmental-resistome because non-core AMR genes showed a stronger phylogenetic signal than core AMR genes, while retaining the report’s statistical caveat. 1 The findings connect to environmental-resistome because environmental or host-associated selection could contribute to island maintenance, but the present co-occurrence results do not establish that causal pathway. 1 The environmental-resistome study supports this ecological link at the broader species level while showing that mechanism composition varies by environment, without resolving island-level causation. 5 The pangenome atlas supports the same distinction by linking clinical AMR enrichment to lower core fractions while warning that taxonomic, annotation, and sampling biases limit causal interpretation. 3
The prophage-AMR study supports phage-defense-syndromes-and-arms-race by adding a cross-species association between prophage-marker density and AMR breadth, while its modest proximity effect and marker-identification caveats argue against treating prophage markers as proof of resistance-island mobilization. 4 It also refines pangenome-integration by linking AMR clusters, prophage markers, gene neighborhoods, species-level pangenomes, and taxonomy, while exposing the need for sequence-resolved validation. 4
The detailed source reports are Within-Species AMR Strain Variation, Environmental Resistome at Pangenome Scale, Pan-Bacterial AMR Gene Landscape, and Prophage-AMR Co-mobilization Atlas. 1 5 3 4
Open Directions
- Use the 1,517 island records together with genome assemblies and genomic-context mapping to plasmids, chromosomes, integron boundaries, and insertion sequences; determine which islands are physically linked and which are lineage-level co-occurrence patterns. 1
- Apply geNomad or PHASTER to the KBase Data Lakehouse genomes and use scaffold sequences for base-pair-resolution distances; test whether strict prophage proximity, rather than contig co-occurrence, predicts AMR accessory status. 4
- Combine island membership with curated exposure metadata and phylogeny-aware models; test whether multi-mechanism islands are co-selected under shared antimicrobial environments rather than merely inherited together. 1
- Integrate resistance-island genes with virulence-factor profiles and metabolic pathway variation; test whether island composition predicts host-associated phenotypes or fitness beyond lineage background. 1
- Apply perturbation or competition experiments to representative multi-mechanism islands; test whether linked resistance genes produce additive, synergistic, or independent protection across drug classes. 1
- Use the reported phi coefficients with fitness and genomic-context data in predictive models; test whether high co-inheritance predicts future AMR gene co-acquisition. 1
- Revisit fitness-cost comparisons as fitness-browser coverage expands, and separate prophage, plasmid, and integrative-conjugative-element mobilization in the same genomes. 4
- Stratify physical-island and co-selection analyses by the environmental categories and mechanism profiles reported in the 293K-genome resistome dataset; test whether clinical, soil, and aquatic contexts predict distinct island architectures after phylogenetic control. 5
- Reclassify island genes with CARD ARO mappings and test whether the clinical-versus-environmental core-fraction contrast persists after separating antibiotic resistance from metal and other stress-response genes. 3