Microbiologically induced aesthetic and structural changes to dimension stone

Dimension stone is natural rock prepared for building use. It is rapidly colonised by microorganisms that cause discoloration (mainly cyanobacteria, algae and fungi) and structural damage. Microbial mobilisation of ions leads to new superficial or internal deposits, weakening the structure. Cyanobacteria and fungi may penetrate, filling pores or creating new spaces. Lichens, fungus/phototroph associations, colonise surfaces and damage stone through ingrowing rhizines and acid production. Initial degradation produces conditions suitable for germination of seeds of higher plants and further destruction. Emerging techniques to elucidate stone-cell interactions and control of initial biofilm formation that eventuates in stone disintegration are discussed.


INTRODUCTION
Dimension stone is any type of natural rock employed for building; the main rocks used are calcareous: limestone, marble, travertine; and siliceous: granite, quartzite, sandstone, soapstone, brownstone, basalt and slate. They can be used for building facades, monuments, paving and even furniture, such as kitchen worktops. The cutting and polishing of the rock removes any surface imperfections and previous incrustations that may have formed during its lifetime of natural exposure. However, once installed in its new position, the polished stone immediately becomes subject to new colonisation, initially by microorganisms 1 . Since the environment has changed, it is obvious that the new microbial flora will be different from that on the original rock and this has been shown by Vlasov et al. 2 , who investigated the microorganisms colonising Rapakivi granite on St. Petersburg monuments and the original rock quarries. Differences were found especially in the fungal communities, with dark-coloured genera like Cladosporium and Alternaria being dominant in the urban, built environment. These granites are unusual, however, and their microbial populations may differ from other types. In particular, they do not have high quartz content, which may be the main granite component conferring relative resistance to microbial colonization because of its non-porous structure 3,4 ; this may limit entry of, especially, filamentous microorganisms, actinomycetes, cyanobacteria and fungi 5 .
The primary colonisers of the polished stone are the phototrophs and chemotrophic bacteria, which require no organic food materials [6][7][8] . These pave the way for heterotrophic and more invasive higher organisms, as well as having their own disfiguring and destructive effects. The chemical and physical interactions between the stone and the colonising organisms (bacteria, algae, fungi and, eventually, higher plants) can cause both aesthetic and structural, or mechanical, damage. The first colonisers of the surface produce thin coating layers, known as biofilms, that contain not only the living organisms, but also their metabolic products, including acids, oxidising/reducing agents, osmolytes and extracellular polymeric substances (EPS), all of which can affect the structure of the stone [9][10][11] . This review critically examines some of the many publications in this area, synthesising the diverse information available, and considers not only how the most recent microbiological techniques may alter our understanding of the phenomenon, but also, briefly, the methods available for controlling stone colonisation.

AESTHETIC CHANGES
An undesired alteration in the appearance of the stone, which does not accompany any apparent weakening of its structure, is termed Biodeterioration without Biodegradation. In the case of dimension stone, this is a sufficiently severe change to give impetus to the development of non-damaging cleaning techniques and protective treatments, which will be discussed later.
Many microbial biofilms simply produce a grey/black discolouration on the stone surface. This may be due to black or brown pigments produced by the cells (see Table 1), or may be a mixture of variously coloured substances that have aged and become oxidised.
The cyanobacteria, for example, can produce a wide variety of pigments, but they are often seen as a black discolouration on the dried surfaces of stone (Fig. 1); when rehydrated, the biofilm can regain its basic green colour, as the cells begin to multiply once more.
The level of humidity of the surface can make a considerable difference, not only to the colour of the biofilm, but also to its constituents. Some microorganisms require a much more humid surface than others for growth. This was demonstrated by the contrasting biofilms on two areas of a fort in Niteroi, Rio de Janeiro, Brazil. The dark green/brown biofilm below a leaking pipe contained Chloroflexi, a phototrophic filamentous bacterial group, as the major organism, while the grey/pale green dry biofilm was mainly non-photosynthetic Proteobacteria 12 . These climatic and positional factors are exemplified also by the biofilms seen on the Rio Bec style Mayan buildings in Campeche state, Mexico. Red coloration produced by overgrowth of the carotenoid-producing alga Trentepohlia was seen on North-and East-facing walls or on other sites protected by tree canopies or architectural elements, while on more sun-exposed sites the typical grey/black growth of dehydration-and UV-resistant cyanobacteria (mainly Gloeocapsa and Chroococcidiopsis) was present 13 . There was no evidence that stone degradation was occurring beneath the biofilms. Nevertheless, the mere presence of dark-coloured areas can result in stone degradation by differential expansion and contraction of the surface layers 14 . Figure 1 shows some of the colorations that can develop on stone surfaces following colonisation by various microorganisms.

STRUCTURAL AND MECHANICAL CHANGES
It can be difficult to distinguish non-biological weathering (degradation) of stone from that associated with microbiological growth. However, there can be no doubt that microorganisms are able to cause stone degradation. This has been demonstrated in laboratory simulation experiments that measure cation leaching 15 , physical changes such as alteration in weight, porosity and surface hardness 16,17 , or simple colonisation and concomitant biodeterioration 18,19 .
In urban areas, especially, atmospheric pollution is a major agent leading to degradation of various types of rocks in building facades, and can be linked to chemical, physical and biological processes. Rocks from building facades in urban environments tend to react with and fix pollutants, while microbial cells colonizing a stone surface can intensify the degrading effects of pollution. The deposited pollutants may be used by the cells for growth and pigment production, as well as the production of corrosive metabolites such as acids 20,21 .
Non-biological stone degradation may often be distinguished in polluted environments at those sites which are sheltered from direct rainwash, which may remove pollutants but not the more strongly adherent cells. Such stone areas will be exposed to more concentrated pollution attack in the form of occult (e.g. dew and frost), dry gaseous, and particulate deposition. This leads to the formation of a characteristic black damage layer, which usually gives rise to more severe decay that includes loss of stone material. These non-biological black crusts typically comprise interlocking tabular crystals of gypsum that entrap atmospheric particulates 22 and form a reactive surface on which other precipitates can form 23 . These ancillary components include inorganic airborne particulates (e.g. soil particles, dust, fly ash), organic airborne particulates (e.g. plant remains, pollen), inorganic precipitates, and organic growth in or on the crust surface (e.g. bacteria, fungi) 24 . Carbonaceous particles (such as flyash), derived from oil and coal combustion, are also frequently present, and these can act as active catalysts for the transformation of calcium carbonate in the stone to gypsum 25 . Another type of black crust also exists, however; this is the thin crust found on buildings in unpolluted environments, which have been shown to be composed entirely of cyanobacteria. Such microbially produced crusts have been found in the Mexican State of Campeche 26 , as well as in Laos, the Brazilian State of Minas Gerais 27 and Guatemala 28 .
In the polluted environment of the modern city of Rio de Janeiro the majority of the historically important buildings are composed mainly of augen gneiss, an extremely thick siliceous rock, with a large amount of ovoid and sub-rectangular k-feldspar megacrystals, varying in size from 2 to 10 cm. There is a smaller amount of megaplagioclase which, together with the k-feldspars, makes up 50 to 90% of the total rock volume. The matrix is quartzplagioclase of varying proportions and the main mafic mineral (Mg and Fe-rich silicate) is biotite, at about 10%. There are small quantities of garnet (<2%) and trace amounts of hypersthene. Among the main accessories are zircon and apatite. The augen gneiss in the area has a high degree of metamorphism. Our previously unpublished geochemical analyses of stone from exposed augen gneiss facades of central Rio churches show high concentrations of Ca, Na, Cl and SO 4 2− ( The presence of gypsum was confirmed by scanning electron microscopy (SEM) and petrographic thin section analyses (Figs. 2 and 3). The SEM analyses of the black crusts show high concentrations of gypsum occurring as needlelike crystals, with a lower concentration of halite. Gypsum crystals are accumulated in the inter-and intra-crystalline fractures of quartz and feldspar crystals and bridging gaps between open cleavage planes in micas, causing deformation and breaking (Fig. 2). In a previous analysis 21 , neogypsum deposits were seen clustering around endolithic microbial cells, indicating the importance of microbial metabolism in deposition of new minerals. This is associated with the ability of microorganisms to solubilise deposits from the rock and transfer them to alternative locations, as discussed in subsequent sections.
Gypsum (CaSO 4 .2H 2 O) crystallises in the monoclinic system. Petrographic thin sections showed the formation of gypsum crystals in the rock microfractures (Fig. 3).
The fact that black gypsum crusts can develop over entire facades in a humid subtropical environment is testimony to the high levels of local pollution, especially particulate deposition. Reduced rainwash in the sheltered micro-environments of the narrow, canyon-like streets of central Rio de Janeiro overcomes the tendency for gypsum to be washed away. These observations further highlight that decay processes are primarily controlled by microclimatic conditions 23 . Gypsum is one of the most destructive of all salts 32 ; it has been widely identified in stone monuments 33 . Cardell et al. 33 consider that the physical stress resulting from salt (halite and gypsite) crystallization in the rock pores is the most important mechanism of the deterioration of ornamental stone. Both of these salts are involved in disruption of the stone structure. Gypsum occurs mainly in areas associated with the black crust, or in fractures within the rock. The halite detected in the stone samples mainly relates to interaction with marine aerosols from the sea in Guanabara Bay; the crystals are found within the stone along with microbial cells 21 and, indeed, halite rocks have been shown to have a close relationship with microorganisms, especially cyanobacteria 34 . These two salts play a very important role in weathering of stonework, mainly when associated with the biological activities demonstrated by various authors 6,26 . The growth and metabolic activity of individual or complex microbial communities, algae, bacteria, cyanobacteria, fungi and lichens, influences the complex interaction between the various types of materials present in stone, leading to physical and chemical damage. Indeed, the simple formation of a thin biofilm on the stone surface can lead to damaging water retention, differential heating/cooling and removal of surface stone flakes by the drying and contracting adhesive gel 1,14 . Endolithic microorganisms growing within the stone can result in spalling of surface layers and endolithic fungi and cyanobacteria have often been detected in historic stone monuments 21,[35][36][37] . Figure 4 shows green growth of mixed cyanobacteria discovered beneath a flake of sandstone removed from the surface of a church in Minas Gerais, Brazil.
More specific degradation comes from the metabolic activities of the attached organisms.
Acid production Acid attack was the first mechanism of microbial stone attack to be postulated; Munz (1890, cited in ref. 35 ) suggested that nitric acid produced by nitrifying bacteria was the cause of stone erosion. This type of attack is often associated with the production of pits. For instance, the pitting caused in the limestone monuments of the Mayan culture at Edzna, Mexico, has been shown to be associated with colonies of Trentepohlia 38 , a redpigment-producing alga whose growth is normally considered simply as an aesthetic problem, although spalling of sandstone apparently caused by this organism has been reported earlier 39 . Since no bacteria were seen alongside the algal colonies in Edzna, it seems likely that the organic acids known to be produced by algae 40 were responsible for the pits. It is possible, but unlikely because of the very localised attack, that cation chelation (q.v.) was responsible for the degradation. It has been suggested that the main mode of stone attack by fungi is through the production of organic acids 41 and certainly fungi isolated from deteriorated limestone at the Mayan site of Uxmal, Mexico, produced oxalic acid, which reacted with solubilized calcium from the stone to produce crystals of whewellite and weddellite 42 . However, it is unlikely that organic acids degrade siliceous stones by direct acid attack; it is more likely that in this case they are acting as chelating agents (q.v.).
Song et al. 4 carried out quantitative analyses in vitro to show that Bacillus subtilis produced pits in a granite surface. Plagioclase or albite (aluminosilicates) were found to be the most vulnerable minerals in the rock, suggesting that the quartz component could be resistant because of its hard, non-porous structure. Barker et al. 43 had previously suggested that the aluminosilicate feldspar components were more prone to microbial attack because they contain ions that react strongly with organic acids.

Cation mobilisation
Most minerals are sparingly soluble in pure water, with equilibrium levels being reached before significant lattice damage can occur. Organic chelating agents, such as acids and polysaccharides, enhance cation solubility 40 . They react with the ions Fig. 1 Microbial discolouration of stone. Colours developed on stone surfaces by microbial colonisation. a Red and green growth of the alga Trentepohlia on gneiss; the orange/red coloration on the curved area indicates intracellular carotenoid droplets, possibly protecting the cells from lack of moisture or overexposure to sunlight. b Grey and light brown discoloration caused by cyanobacteria and fungal growth on limestone doorway of a church in Rio de Janeiro, Brazil. c Thick grey cyanobacterial and fungal biofilms on Amber fort, Rajasthan, India. The pink stone is the original colour of the sandstone. d Grey cyanobacteria-dominated biofilms next to red/brown growth of the alga, Trentepohlia, on the base of the limestone tower of the Palace, Palenque, Chiapas, Mexico.
as they are released, producing soluble organic complexes that can move through the pores of the stone and be deposited when they come into contact with a suitable precipitant, which may simply be oxygen at the surface or near-surface.
Barrionuevo et al. 44 found evidence of iron and manganese mobilisation from within the sandstones of the ruins of the Argentine missions, with a surface layer enriched in these minerals compared to the interior. There was a complex biofilm on these buildings, many components of which could be involved in such mobilisation and redeposition activities through cation chelation and acid production. A wide range of chelating agents are produced by bacteria 45 , algae 46 , cyanobacteria 47 and fungi 48 . Those that are involved in solubilisation of iron, facilitating uptake by the cells, have been termed siderophores. As they are secreted by the cells that synthesise them, they may also be used by other organisms 49 as well as for dissolution of iron from within the stone structure resulting in weakening and potential transport and reprecipitation in other areas.

Alkaline dissolution of Si and Al
The majority of the literature emphasises the damage caused by acidic microbial metabolites on built stone and this is doubtless one of the main degrading activities of bacteria and fungi. However, Gaylarde and Gaylarde 35 consider the susceptibility of siliceous stone to alkaline degradation. In concrete, the microfractures and spalling that can be induced when the silicate content of the aggregate particles reacts with alkali hydroxides that may be present in the cement itself, are well recognised 50 and this reaction has been the subject of considerable research. The first step in alkaline silicate weathering seems to be the acidbase reaction between the silica and the alkali to produce siloxane. The siloxane bridges are then attacked by alkali to cause disintegration of the stone. Ichikawa and Koizumi 51 found that irradiation increased the susceptibility of concrete to the alkalisilicate reaction; both crystalline and amorphous quartz were    Fig. 2.
affected, the increased degradation being due to the formation of the more susceptible distorted amorphous quartz on the surface. Considering siliceous stone itself, rather than aggregate, the Si and Al content will be mobilised above pH 9.5, a pH value which is readily attained by phototrophic organisms during photosynthesis during the hours of sunlight 52,53 . Hence the alkaline degradation of siliceous stone by cyanobacterial and algal biofilms doubtless occurs. This may also be induced by cellular production of polyols. Polyols are produced by all types of living organisms. They are organic compounds containing multiple hydroxyl groups, for example, sugar alcohols like mannitol and sorbitols (low molecular weight polyols) and polysaccharides (high molecular weight polyols). Such chemicals bind to siloxane layers in siliceous stone, causing them to expand and thus weakening the structure 35 . The findings of Song et al. 4 on differential degradation by Bacillus subtilis of quartz and aluminosilicates in granite could be explained by the action of bacterial polyols, known to be produced by this species 54 . Many microorganisms, bacteria, fungi 55,56 , algae 57 and cyanobacteria 58 produce polyols. Vlasov et al. 2 found that polyols were higher in stone biofilms with a predominance of fungi. They found no important difference between quarry rocks and built monuments in St. Petersburg in terms of fungal/phototroph dominance, but the polyol content was higher in the urban samples; they concluded that this was a function of the taxonomic composition of the biofilms, but did not investigate any related biodegradation.

THE SPECIAL CASE OF LICHENS
Lichens are symbiotic associations of a filamentous fungus, the mycobiont, and one or more phototrophs, the photobiont. They take various physical forms, of which the crustose form is the most damaging to stone, being particularly strongly adherent to the surface. Other forms are foliose and fruticose. It is traditionally assumed that the phycobiont (an alga or cyanobacterium) donates organic carbon to the relationship, itself gaining minerals and protection from the fungus. Recently, however, it has become apparent that lichens may also contain yeasts 59 and nonphotosynthetic bacteria 60 that also contribute to the relationship. It is difficult, therefore, to determine the exact origin of the destructive activities of lichens on stone surfaces, but there is clear evidence of such destruction on many historic stone buildings [61][62][63] . By lichenic dissolution of mineral elements from the stone, degradation occurs; for example, lichen growth on basalt was shown to produce ferromanganese minerals, leaving a calcium-rich surface (Jones et al., 1980, cited in ref. 64 ), ready to react with sulfur compounds from polluted air, as previously described in this article.
The etching produced by lichens on rocks and stones (Fig. 5) was commonly thought to be produced by so-called "lichenic acid" 65 ; these authors also note the neoformation of calcium carbonate, calcium oxalates and gypsum under lichen action. Gadd 66 discussed the various acids produced by lichens that could degrade stone surfaces, the majority being organic acids such as oxalic.
This group of organisms also physically penetrate rocks with their rhizines, small, fungal 'roots', which probably require prior weakening of the stone surface for their entry. This effect gives the lichen a very strong physical hold on the surface. It has been suggested that it is inadvisable to remove such growths, not only because rough removal techniques can further damage the surface, but also because the actual physical presence of the lichen may protect against attack by other agents, including nonbiological climatic influences 67 . Nevertheless, lichens are  C.C. Gaylarde and J.A. Baptista-Neto inherently a danger to stone integrity; not only do crustose lichens actively penetrate the rock, they may also harbour and protect degradative microorganisms beneath their thalli. Endolithic cyanobacteria beneath lichen growths have been shown to disaggregate the rock of the Tomb of Cyrus, in Pasargadae, Iran 61 . Stone degradation by lichens is complex and yet to be completely elucidated.

POTENTIAL CONTROL
As already noted, stone biodeterioration begins with its colonisation by microorganisms. The prevention of this process, therefore, presents a method of stopping biodeterioration before it begins. Coating the surface with a suitable protective layer and inhibiting the attachment and/or growth of the initial colonisers are potential strategies.
The use of a hydrophobic coating is a possible solution. This protects the stone from absorption of water and materials carried in it; it should also be resistant to wear and abrasion and aggressive chemicals like those produced by microorganisms. However, it will not necessarily be resistant to the attachment of microorganisms. The development of antibiofilm coatings is extremely challenging 68 . Traditional coatings are acrylic polymers, siloxanes, fluoropolyethers and fluorinated acrylic polymers. These may suffer from poor adhesion to the underlying substrate, may lose their protective effect quite rapidly with time, and also may, themselves, be colonised by microorganisms 69 . New and developing technologies, involving nanomaterials, biomimetic approaches and packaging/carrier systems (nanocapsules) may revolutionise the available protective systems, if the increased costs can be overcome; for example Jin et al. 70 have produced halloysite nanotubes that offer controlled release of a fungicide that could be useful for long-term control of fungal growth on stone. The group of Ruggiero, in Rome, has been very active in this field recently [71][72][73] , developing antimicrobial and water-repellant nanocoatings for use on stone surfaces, while De Leo et al. 68 have developed surface-active ionic liquid-containing coatings that can both remove colonising organisms and prevent new biofilm formation. The latter is an important objective, as reapplication of removal treatments can be not only expensive, but also potentially damaging to the stone.
The need for more environmentally acceptable antimicrobial treatments has led to a number of research groups working on socalled "natural biocides" or "green conservation", using substances extracted from plants or other forms of life that kill or inhibit microorganisms. These are generally cheap to produce and are considered safe not only for the environment but also for the stone structure and the people working with it 9,73-76 . Their efficacy is, however, not always proven by application to existing buildings 75,77 .
Once the biofilm (sometimes called a patina) has been formed it is necessary to use a removal technique that does not damage the underlying stone substrate; scrubbing and use of bleach, for example, are not recommended. Gabriele et al. 78 tested a hydrogel that incorporated hypochlorite into sodium alginate and found that it effectively removed the filamentous cyanobacteria and algae from calcarenite Lecce samples. Laser cleaning has some advantages over mechanical and chemical cleaning; it is gradual, selective, contactless and environmentally friendly. However, the wavelength must be carefully selected if damage is not to occur. Barreiro et al. 79 found that a wavelength of 532 nm produced the best biofilm removal from Vilachán granite, but was somewhat aggressive, producing greatest change in appearance through selective extraction of some weathering materials. Biotite melting occurred on all surfaces regardless of the wavelength.
While complete prevention of biofilm formation may not be possible at a reasonable financial and environmental cost, periodic inspection of stone buildings is essential. Ferreira et al. 80 carried out a study on 203 Portuguese buildings with natural stone claddings and determined that a system of cleaning and minor repair, when first signs of degradation were seen, promoted the durability of the structure. Although costs were increased, the strategy increased the predicted lifetime of the materials to a worthwhile extent. This, indeed, conforms to the 'good housekeeping' rationale promoted by the Biodeterioration community 81 .

CONTRIBUTION OF NEW METHODOLOGY TO MICROBIOLOGICAL STUDIES ON STONE
The new techniques of genetic and metabolic analyses will profoundly influence our understanding of microbial interactions with stone in the future. Although speculations on the possible microbial activities that result in stone degradation exist (for example, chelation of Mn, solubilisation of Fe, sulfur and silicon metabolism, production of organic acids and pigments), there is no direct evidence to show that the corresponding changes in degraded stone were actually caused by microbial cells. Only rather recently have the so-called 'next-generation sequencing' (NGS) techniques been applied to stone 21,82 . These have allowed the detection of a much greater number of microbial taxa than was previously recognised. Now, metagenomic analyses have begun to be applied to these ecological systems; these can lead to the identification of functional genes in the microbial population, allowing us to detect cells with the abilities to produce chelating molecules, to metabolise sulfur, manganese, nitrogen, and to produce relevant organic acids and pigments. Esposito et al. 83 used shotgun metagenomic sequencing to analyse the microbial population of a siliceous rock varnish in the Matsch Valley, Italy. Interestingly, they found a high number of Archaea in the varnish. Functional genes of interest detected within the genomes showed the potential for CO 2 fixation, carbohydrate and nitrogen metabolism, siderophore biosynthesis and genes associated with photosynthesis. The very young science of metabolomics has not yet been applied to rocks, although Gutarowska et al. 84 have used untargeted metabolomics with ultra-high performance liquid chromatography coupled to high-resolution mass spectrometry to detect the activated pathways of microbial cells in wood and brick. This group used the same method to detect putative activities in the degradation of a variety of building materials: sulfur metabolism, carbohydrate digestion, carotenoid biosynthesis and photosynthesis 85 . Sanmartin et al. 86 reviewed some of the metabolic profiling methods that had been used to analyse biodeterioration of various types of cultural heritage, including stone, up to that time. Together with older techniques, such as scanning electron microscopy, these may allow us to understand the real interactions between microorganisms and their stone substrate, which should lead to an increased ability to control the biodegradation of dimension stone in the future.

DATA AVAILABILITY
The datasets analysed during this study are included in this published article.