PacBio HiFi sequencing datasets of culture-enriched airborne microbial cave communities from dolomitic Sudwala Caves, South Africa.

We present a dataset integrating physico-chemical air quality measurements with long-read PacBio HiFi shotgun metagenomic sequences from culture-enriched airborne samples collected in Sudwala Caves, one of the oldest known cave systems in South Africa. This resource provides baseline characterization of airborne microbial communities and associated environmental parameters within a subterranean karst ecosystem. A total of 106 air samples were collected across six different cave compartments and three external reference sites spanning two seasonal periods, the winter-spring transition (September-October 2024) and the summer-autumn window (February-March 2025). Environmental metadata include temperature, relative humidity, particulate matter (PM₁.₀, PM₂.₅, PM₁₀), and formaldehyde (HCHO) concentrations, enabling direct linkage between microbial composition and air quality dynamics. Post-quality control of eighteen (18) culture-enriched metagenome datasets yielded 7.7 × 10⁴ to 7.8 × 10⁵ HiFi reads per sample corresponding to 0.63-6.71 Gb of high-accuracy sequence data per sample. Kaiju classification assigned 65.1-83.4% of assembled sequences to reference taxa. Domain-level profiles were dominated by Bacteria (98.7-99.9% of classified sequences), with minor representation of Eukaryota (0.06-0.15%) and extremely low abundances of Archaea (0.002-0.009%) and Viruses (0.000-0.001%). At the phylum level, airborne bacterial communities were consistently dominated by Bacillota (mean relative abundance: 46.92%), Pseudomonadota (34.28%), and Actinomycetota (15.71%) across all sampling sites and seasons, with Pseudomonadota and Actinomycetota exhibiting proportionally higher representation within cave interior environments relative to outdoor reference sites. At the genus level, Staphylococcus, Bacillus, Microbacterium, Arthrobacter, and Pseudomonas were among the most consistently detected and abundant airborne genera within cave compartments, whilst outdoor aerobiome communities were characterised by greater relative abundances of Planococcus, Sphingomonas, Stenotrophomonas, and Arthrobacter. Functional annotation using the DRAM pipeline identified 1205,651 predicted genes, with 579,682 KEGG orthologs (KO), 62,261 MEROPs peptidases, 904,193 Pfam domains, and 21,859 CAZy genes annotated. This dataset supports investigations of culturable airborne microbial composition, functional capacity, bioaerosol dynamics, and environmental health indicators in dolomitic subterranean karst systems, providing a reference framework for comparative studies of low-biomass atmospheric environments.