There are two major mechanisms for the formation of karstic caves. Classical epigenic cave systems form as water flows through the soil and produces karst networks by seepages, absorbing CO₂ and forming a dilute carbonate solution. In epigenic caves, exchanges occur with subaerial processes (such as water movement and mixture), enhancing dissolution capacity in the direction of flow (Sendra et al., 2014). In contrast, hypogenic cave systems form as water recharges the soluble bedrocks from below. The formation of hypogenic caves is fueled by hydrostatic pressure or other energy sources rather than recharge from the overlying or adjacent surface (Ford, 2006). Due to the deep origin of rising water, hypogenic cave systems are not directly influenced by seepages. Among explanations for hypogenic cave formation, sulfuric acid speleogenesis (the formation process of a cave) is a major one, resulting in some of the largest cave systems including Carlsbad Caverns and Lechuguilla Cave in New Mexico. Hydrogen sulfide leaked upward along with fractures from hydrocarbon deposits, and form sulfuric acid upon reaching oxygenated meteoric groundwater (Jagnow et al., 2000). Microbial communities dominated by sulfur-oxidizing acidophilic bacteria Acidithiobacillus formed dense biofilms in these caves, accelerating mineral dissolution and cave enlargement (Jones et al., 2014). Geobiochemical differences occur between epigenic and hypogenic cave systems, because caves formed from the two mechanisms would have different environmental conditions. For example, hypogenic caves such as sulfuric acid speleogenetic caves are sulfur-rich with relatively low pH, as a result, these caves support microbial communities that are acidophilic and actively involved in sulfur transformation. On the other hand, epigenic caves are pH neutral or slightly alkaline, and the microbial communities would be more diverse. Studies concerning snottites from sulfuric caves revealed very low biodiversity with Chao1 index ranged from 1 to 10 (Hose et al., 2000; Vlasceanu et al., 2000; Macalady et al., 2007), while more than ten thousand OTUs were detected from epigenic cave samples with Chao 1 index reached more than 1,000 (Zhu et al., 2019). Lava caves are also important subterrain environment, although karstic caves accounted for a larger proportion (Jones and Macalady, 2016). Lava caves formed from the heat volcanic flows and are mainly composed of basalt (Gabriel and Northup, 2013). These substantial differences from karstic caves granted lava caves unique microbial communities, which are mainly affected by geographical location and the availability of organic carbon, nitrogen and copper in the lava rock (Hathaway et al., 2014).

Multiple studies have suggested direct or indirect microbial involvement in the formation of carbonate minerals. Microbial calcite precipitation in calcium-rich environments was assumed to be the result of a detoxification process, through which growing cells actively export the excess calcium ions to maintain cellular metal homeostasis (Banks et al., 2010). Microbial biofilms also served as the initial crystal nucleation sites that contributed to the formation of secondary carbonate deposits within caves (Tisato et al., 2015). Cañaveras et al. (2006) proposed a model of moonmilk formation based on extensive observations in Altamira Cave (Spain) that microbial filaments provided a template for the precipitation of calcite fiber crystals in the early stages of moonmilk deposition. Moreover, the growth of microbes can increase the environmental pH, which would in turn increase the saturation index of carbonate and drive precipitation. For example, nitrogen metabolic pathways including ammonification of amino acids, dissimilatory reduction of nitrate and degradation of urea or uric acid induced formation of carbonate and bicarbonate ions, as the metabolic end product ammonia increased local pH (Castanier et al., 2000). Maciejewska et al. (2017) proved that Streptomyces promoted calcification in moonmilk through ammonification and, less importantly, ureolysis.

Biologically induced carbonate deposition was initially attributed mainly to fungi, studies reported that nano-fibers presented in the crystalline structure of moonmilk were related to biomineralized fungal hyphae (Bindschedler et al., 2010, 2014). However, recent studies indicated that bacteria played a major role in the induction of cave carbonate precipitation (Cañaveras et al., 2006; Portillo et al., 2009). γ-Proteobacteria and Actinobacteria were the major groups detected in white colonizations that were able to raise pH through metabolism in Altamira Cave (Portillo and Gonzalez, 2011); isolates belonging to α-Proteobacteria, β-Proteobacteria, γ-Proteobacteria, Firmicutes and Actinobacteria from cave speleothem were also confirmed to perform calcification (Banks et al., 2010). The presence of Archaea was identified in moonmilk deposits, yet their role in the formation of moonmilk remains to be discovered (Gonzalez et al., 2006; Reitschuler et al., 2016).

Fe and Mn oxides were found in karst caves as sedimentary fills, walls, ceiling and floor coatings/crusts, and sometimes as their own speleothems (Hill and Forti, 1997; Palmer, 2007). The importance of biological Fe oxidation in caves has long been recognized (Caumartin, 1963). Early studies on the bacterial role in Fe and Mn deposits most relied on microscopy: Peck first reported the presence of Fe-precipitating Leptothrix and Gallionella species in enrichment cultures inoculated with mud from cave pools and sumps (Peck, 1986). Later, more and more studies hinted the participance of microorganisms in cave Fe oxidation. For instance, freshly Fe oxide precipitates in Pautler Cave showed consistency with biomineralization structures of the microbial genera Gallionella and Leptothrix (Frierdich et al., 2011), and the iron mats in Borra caves appeared to be related to a community of mostly Leptothrix-like iron-oxidizing bacteria (Baskar et al., 2008, 2012). Abiotic Fe oxidation is rapid and prevalent at circumneutral pH yet is inhibited at low pH and microoxic/anoxic environments, where biological Fe oxidation more often occurs (Jones and Northup, 2021). The formation of Fe oxides and the identification of Fe-oxidizing bacteria do not necessarily indicate biological Fe oxidation.

Although Fe oxide minerals generate via both biological and abiotic processes, the presence of secondary Mn oxides is usually related to Mn-oxidizing microbes, because abiotic Mn oxidation is kinetically inhibited even at anoxic conditions (Luther, 2010; Johnson et al., 2013). Microbes could increase the rate of Mn oxidation by five orders of magnitude (Emerson, 2000; Tebo et al., 2005). Microbes related to Mn oxide production have long been reported from cave systems, and bacterial species belonging to Proteobacteria, Firmicutes, Actinobacteria and Bacteroidetes were reported to oxidize Mn (Carmichael and Bräuer, 2015). For example, Mn-oxidizing Pseudomonas, Leptothrix, Flavobacterium and Janthinobacterium species were isolated from cave ferromanganese deposits (Carmichael et al., 2013a); the well-known Mn-oxidizing Pedomicrobium and Caulobacter species were also observed in cave stromatolites (Lozano and Rossi, 2012). Most of the Mn oxidizers described to date are heterotrophic, they are supposed to oxidize Mn indirectly through producing superoxide during growth (Learman et al., 2011). However, Mn oxidation has long been regarded to be a potential energy-yielding reaction, and chemolithoautotrophic Mn oxidation was recently documented by Yu and Leadbetter (2020). A co-culture of two microbial species was obtained and the co-culture possibly coupled extracellular manganese oxidation to aerobic energy conservation and autotrophic carbon fixation (Yu and Leadbetter, 2020). Mn oxidation is largely influenced by exogenous carbon input, dilute sewage into the cave lead to massive bloom of a microbiome-driven and Mn-oxidizing biofilm (Carmichael et al., 2013b). Although bacteria attracted more attention in cave biogeochemical cycling, Mn-oxidizing fungi are also identified in caves. Acremonium nepalense was found responsible for black Mn oxides on clayey sediments of Lascaux Cave (Saiz-Jimenez et al., 2012); Mn-oxidizing members belonging to Ascomycota were obtained in southern Appalachian cave systems, and the results suggested that anthropogenic carbon sources stimulated fungi-driven Mn oxidation (Carmichael et al., 2015).

As a potent greenhouse gas, the concentration of atmospheric methane is increasing (Nisbet et al., 2016; Prather and Holmes, 2017). Estimation showed that human activities and natural sources produce about 680 Tg year⁻¹ of methane to the atmosphere, while reactions with hydroxyl and chlorine radicals in the troposphere and stratosphere remove about 600 Tg year⁻¹ of methane (Kirschke et al., 2013). Methanotrophic microorganisms in forests, grasslands, paddy and other unsaturated soils play an essential role in mediating carbon cycle, and are believed to filter 30 Tg year⁻¹ of methane (Kirschke et al., 2013). However, there is still gap in the overall methane budget balances. Recent studies suggested that caves and related karst terrains may be an essential yet overlooked sink for atmospheric methane.

Fernandez-Cortes et al. (2015) first monitored the concentration of methane and carbon dioxide in seven caves located in Spain, and the results proved that subterranean environments acted as sinks for atmospheric methane on seasonal and daily scales. They also detected methane-oxidizing bacteria in some cave sediments where methane concentrations were near to the atmospheric background, yet no such microbes were detected in sample sites with minimal methane concentrations. Thus, Fernandez-Cortes et al. assumed that cave methane oxidation was mainly induced by oxidative capacity from high density of ions, and was not significantly intervened by methanotrophic bacteria. However, through controlled laboratory experiment, Nguyễn-Thuỳ et al. (2017) showed that the radiolysis hypothesis is kinetically constrained and is unlikely to lead to rapid methane loss. Instead, by performing a set of mesocosm experiments with rock samples from two Vietnamese caves, they revealed that the depleted concentrations of methane in caves were most likely associated with microbial activity rather than radiolysis. Following these pioneering works, more and more evidence were reported from various caves to support microbes involved cave methane oxidation. For example, stable carbon and hydrogen isotope compositions of methane from 33 epigenic caves in the United States and 3 in New Zealand all supported that microbial methanotrophy within caves was the main methane consumption mechanism (Webster et al., 2018). Ojeda et al. (2019) also noticed methanotrophic activity of γ- and α-Proteobacteria in Nerja cave in Spain.

Methanotrophs have been reported since early twentieth century, and were detected at various habitats such as mud, rivers, rice paddies, sediments and sewage sludge (Whittenbury et al., 1970; Hanson and Hanson, 1996). Although atmospheric methane oxidation rates can remain steady for more than 4 months at 1.7 ppmv of methane (Schnell and King, 1995), calculations based on the kinetic constants of isolated methanotrophic bacteria could not support such extended survival (Conrad, 1984). Studies with Methylosinus trichosporium and Methylobacter albus revealed that atmospheric methane oxidation did not supply sufficient cellular maintenance energy and reduced power for the methane monooxygenase (Roslev and King, 1994; Schnell and King, 1995). The organisms responsible for atmospheric methane uptake were unknown until Holmes et al. (1999) reported a novel group of bacteria belonging to α-Proteobacteria, which is distantly related to existing methane-oxidizing strains and is believed to consume atmospheric methane. Later, Knief et al. (2003) identified another novel atmospheric methane-oxidizing group belonging to γ-Proteobacteria through comparative sequence analysis of the pmoA gene, and named these two groups as “upland soil cluster α” (USC α) and “upland soil cluster γ” (USC γ), respectively. Culture-independent studies suggested that USC α is adapted to the low concentration of methane in neutral and acidic soils (Kolb et al., 2005; Martineau et al., 2014) while USC γ prefers neutral to alkalic soils (Kolb, 2009). Tveit et al. (2019) reported the first pure culture Methylocapsa gorgona MG08 that grows on air at atmospheric methane, and proved that the strain is closely related to uncultured members of USC α. However, the cultivation of members related to USC γ remains challenging.

The investigation of cave methanotrophic bacteria is far earlier than the discovery that caves may be important methane sink. Hutchens et al. (2004) identified strains of Methylomonas, Methylococcus and Methylocystis/Methylosinus as major methanotrophs in Movile Cave through DNA-based stable isotope probing. However, the air of Movile Cave contains 1–2% (10,000–20,000 ppmv) methane (Sarbu and Kane, 1995), which is much higher than general 1.86 ppmv atmospheric level. Zhao et al. (2018) explored the presence and diversity of methane-oxidizing bacteria in Heshang Cave, where methane concentration decreases from 1.9 ppmv at the entrance to 0.65 ppmv inside the cave. Their results provided compelling evidence that methane-oxidizing bacteria accounted for up to 20% of the whole microbial communities with the high-affinity USC γ being the dominant group. According to sequencing analysis of pmoA and 16S rRNA genes of weathered rock samples from three karst cave in southwest China, Cheng et al. (2021) demonstrated that USC γ group dominated the atmospheric methane-oxidizing communities, and was identified as a keystone taxon in cooccurrence networks of both methane-oxidizing and the total bacterial communities. In addition to atmospheric methane-oxidizing groups, anaerobic methane-oxidizing bacteria were also detected from some cave samples. For example, members of the phylum Rokubacteria were found in Pindal Cave, these organisms perform anaerobic oxidation of methane coupled to nitrite reduction (Cuezva et al., 2020).

Infectious diseases have long been threatening human society, while sprouting antibiotic-resistant pathogens are posing heavier burdens to human public health. According to the World Health Organization (WHO), there is an urgent need for investment of new antibiotics to combat antibiotic-resistant infections (World Health Organization, 2015). With a combination of unique environmental conditions and rare human intervention, more and more studies turned to caves for microorganisms producing novel antibiotics.

It has been proved that Actinobacteria are prolific producers of promising bioactive compounds with wide application. Approximately 45% of identified bioactive compounds are produced by Actinobacteria, among which 80% are derived from the Streptomyces genus (Bérdy, 2005). Sequencing analysis revealed the dominance of Actinobacteria in plenty of cave samples, such as cave soils (Wiseschart et al., 2019), cave sediments (De Mandal et al., 2015a,b), cave rocks (Zhu et al., 2019) and colonies on cave Paleolithic paintings (Stomeo et al., 2008). The isolates of Streptomyces genus were also obtained from many caves, such as Sigangli Cave in China (Fang et al., 2017), Altamira Cave and Tito Bustillo Cave in Spain (Groth et al., 1999), and Iron Curtain Cave in Canada (Gosse et al., 2019). Caves are not only home for various known actinobacterial taxa, but also excellent reservoir for new species of Actinobacteria. Forty-seven species within 30 genera belonging to Actinobacteria were isolated from caves and cave-related habitats from 1999 to 2018, among which seven represented novel genera (Rangseekaew and Pathom-aree, 2019). It is assumed that the extreme conditions within caves stimulated the inhabitant microorganisms to mutate their genes, making it more likely to evolve new species and novel metabolites (Tiwari and Gupta, 2013).

Isolating cave microorganisms and subsequently checking their antimicrobial activity against pathogens is one of the main approaches to identify novel antibiotics. By adopting such strategy, Herold et al. (2005) reported cervimycins, which are highly active against some Gram-positive bacteria. These compounds are produced by Streptomyces tendae HKI 0179, a strain isolated from a rock wall of an ancient cave in Italy. Another example is the discovery of Huanglingmycins, which are produced by Streptomyces sp. CB09001 from cave soil of China (Jiang et al., 2018). Noticeably, Huanglongmycin A showed not only weak activity against some Gram-negative bacteria, but also moderate cytotoxicity against A549 lung cancer cell line. Although certain bioactive compounds were not identified, some rare cave actinobacterium also showed anticancer potential. For instance, Spirillospora albida isolated from Phanangkhoi Cave were active against NCI-H1870 (human small lung cancer cell; Nakaew et al., 2009a); Nonomurea roseola isolated from Phatup Cave showed activity against human oral cavity cancer and human small lung cancer cells (Nakaew et al., 2009b). In addition, antifungal compounds were also documented from cave environment. Antagonistic Streptomyces, Micromonospora, Streptosporangium, and Dactylosporangium isolated from five caves in Korea showed biocontrol activities against at least one of the rice pathogenic fungi (Nimaichand et al., 2015); Streptomyces sp. from five caves in New Mexico and the United States has been suggested to inhibit the growth of the causative fungus of white-nose syndrome in bats (Hamm et al., 2017).