Desert breath—How fog promotes a novel type of soil biocenosis, forming the coastal Atacama Desert’s living skin

The Atacama Desert is the driest non‐polar desert on Earth, presenting precarious conditions for biological activity. In the arid coastal belt, life is restricted to areas with fog events that cause almost daily wet–dry cycles. In such an area, we discovered a hitherto unknown and unique ground covering biocenosis dominated by lichens, fungi, and algae attached to grit‐sized (~6 mm) quartz and granitoid stones. Comparable biocenosis forming a kind of a layer on top of soil and rock surfaces in general is summarized as cryptogamic ground covers (CGC) in literature. In contrast to known CGC from arid environments to which frequent cyclic wetting events are lethal, in the Atacama Desert every fog event is answered by photosynthetic activity of the soil community and thus considered as the desert's breath. Photosynthesis of the new CGC type is activated by the lowest amount of water known for such a community worldwide thus enabling the unique biocenosis to fulfill a variety of ecosystem services. In a considerable portion of the coastal Atacama Desert, it protects the soil from sporadically occurring splash erosion and contributes to the accumulation of soil carbon and nitrogen as well as soil formation through bio‐weathering. The structure and function of the new CGC type are discussed, and we suggest the name grit–crust. We conclude that this type of CGC can be expected in all non‐polar fog deserts of the world and may resemble the cryptogam communities that shaped ancient Earth. It may thus represent a relevant player in current and ancient biogeochemical cycling.

. Precipitation levels in these dry inland zones allow the establishment mainly of cyanobacteria that were found, for example, under translucent quartz stones and are known as lithic communities (stone-inhabiting communities including hypoliths and endoliths; Warren-Rhodes et al., 2006). Different types of cryptogamic organizations such as biological soil crusts (hereafter referred to as biocrusts), which are an edaphic biocenosis (community of organisms that interact with each other in a soil-like habitat) of cyanobacteria, algae, lichens, bryophytes, heterotrophic bacteria, archaea, and fungi living in the top soil layers (Belnap & Gardner, 1993), have only recently been reported to occur very locally in the Coastal Range of the Atacama Desert, in areas that are affected by fog (Baumann et al., 2018;Bernhard et al., 2018;Lehnert, Jung, Obermeier, Büdel, & Bendix, 2018a;Wang, Michalski, Luo, & Caffee, 2017).
Such cryptogamic ground covers (CGC; Elbert et al., 2012) are estimated to occur on about 12% of the Earth´s terrestrial surface, thus they are one of the most abundant and most productive microbial communities (Rodriguez-Caballero et al., 2018;Weber, Büdel, & Belnap, 2016). As "ecosystem-engineers" (Bowker, Mau, Maestre, Escolar, & Castillo-Monroy, 2011), CGCs often form surface layers that provide important ecosystem services that can be categorized as supporting services, for example, primary production and soil formation via bio-weathering or regulating services such as erosion prevention or influences on biogeochemical cycles. For example, they have an estimated contribution of nearly half of the global terrestrial biological N fixation and 7% of the net primary production of terrestrial vegetation (Elbert et al., 2012). They may also contribute importantly to global C cycles. Estimates of global C sources and sinks are still imbalanced (Quéré et al., 2018) with suspected sink functions of drylands (Evans et al., 2014;Ma, Liu, Tang, Lan, & Li, 2014) which are, however, not well quantified due to a significant lack of studies (Maestre et al., 2013). As the occurrence of CGCs is predominantly expressed in drylands (Weber et al., 2016), the ecosystem services of these areas may be consistently underestimated. In the Namib Desert, for example, biological activity of a CGC dominated by lichens is known to follow wet-dry cycles which are caused by fog, dew, and seasonal rainfall, therewith providing 16 g C m −2 yr −1 to the soil on a large scale (Lange, Meyer, Zellner, & Heber, 1994). Despite significant advances in our understanding of the role of CGCs in deserts across the world, the questions about the extent of these biocoenoses or how they influence ecosystems remain unanswered. During explorations in the National Park Pan de Azúcar of the southern Atacama Desert, we discovered extended blackish patterns paving the ground which unlike the inorganic mineral crusts called desert varnish that are known from deserts worldwide (Selby, 1977), turned out to be caused by an hitherto unknown CGC type occurring at a large landscape scale.
In this article, we describe this new and unique type of CGC in the fog oasis of the Atacama Desert and address several hypotheses concerning its structure and function. We hypothesized that (a) these patterns of blackish and whitish soil areas were caused by variation in the density and composition of the cryptogamic cover, (b) this cryptogamic cover was dominated by green-algal lichens (chlorolichens), free-living green algae, and cyanobacteria, because water availability is limited and thus only less water demanding cryptogam groups can thrive there, and (c) that this variation in biological cover was correlated with variation in geochemical soil properties.
Additionally, we discuss the ecosystem services provided by the CGC organisms and relate this to the potential role of similar cryptogamic covers in ancient Earth and present-day biogeochemical processes.

| Study site
Pan de Azúcar National Park is located between 25°53′ to 26°15′S and 70°29′ to 70°40′W along the coast of Chile and is characterized by an arid climate (Baumann et al., 2018;Lehnert, Thies, et al., 2018b). A narrow pediment close to the coast is followed by a steep escarpment reaching elevations up to 850 m a. s. l. After the ridge of the escarpment, the terrain descends slightly to elevations between 400 and 700 m a. s. l. toward inland. This study was conducted within a strip of 2.5 km on top of the escarpment along the Pacific coast (Lat: 25.96636111°S, Long: 70.61521111°W). In this zone, quartz-and granitoid grit-sized (0.6 mm) stones paved the desert surface and abundant blackish patterns were observed.

| Remote sensing
Along the coastline, five transects were selected from the first ridge toward the hinterland. Areal images of each transect were acquired using drones equipped with standard RGB-cameras. Position and elevation of ground reference points were recorded prior to the flights using a differential GPS in post-processing mode. Spatial resolutions of the images varied between 1.8 cm and 3.5 cm depending on the flight height above the terrain. The total area covered by aerial images was 206.5 ha. RGB images were processed using AgiSoft PhotoScan (Riano, Chuvieco, Salas, & Aguado, 2003). To derive mosaics covering the entire flight path, the single images were stitched together, and digital surface models were calculated.
Training areas in each image for classes "vegetation," "biocrust," and "bare soil" were defined using field observation data and the highresolution drone images. Afterward, maximum likelihood classifications were performed using the red, green, and blue bands of the topographically corrected mosaics.
To provide estimates of CGCs over larger areas, a cloud-free satellite scene from Landsat 8 was used. The level-2 data were radiometrically, atmospherically, and topographically corrected to ground albedo values. For the radiometric correction, the standard correction values provided with the metadata were applied. For the atmospheric correction, an enhanced version of the 6S-code was applied which proved to perform superior to the original one in areas with rugged terrain (Curatola Fernández et al., 2015) . The topographic correction method according to Minnaert was applied to the atmospherically corrected ground albedo values using the Aster digital elevation model.
The maximum likelihood classifications of the aerial images were used as reference for a cover classification of the Landsat 8 scene. For each Landsat 8 pixel, the percentage of the collocated aerial image pixels covered by the biocrust was calculated giving estimates on cover of the new biocrust. These estimates were used as response variables in random forest regression models.
As predictors, the atmospherically and topographically corrected Landsat 8 albedo values were used. Since the occurrence of the new CGC strongly depends on fog water fluxes, the mean values of liquid water path of low stratus clouds and fog as derived by a novel fog retrieval with 30 m spatial resolution were used as an additional predictor in the random forest models (Lehnert, Thies, & Bendix (unpublished data). Tuning and validation were performed based on a 10-fold cross-validation repeated five times. The rootmean-square error of the CGC estimates was 4.00% with an R 2 value of 0.88. The final model was applied to the entire area of the Pan de Azúcar National Park and its surrounding desert. The occurrence (presence/absence) of the grit-crust outside of the drone transects was successfully validated at ten additional sites shown as red dots in Figure 1d.

| Climatic data
In order to determine climate conditions in the area of the CGC (Lat: 25.96636111°S, Long: 70.61521111°W; 2.5 km distance from the Pacific coast), an automatic weather station (AWS grit) was installed which has delivered data between July 2017 and March 2018 in 5min intervals. The AWS is equipped with standard sensors measuring wind speed and direction, air and soil temperature, relative humidity, and a surface temperature sensor pointing at the biocrust. Fog and dew water flues are determined using a cylindrical ("harp"-type) fog collector and a dew balance on the ground, with the biocrust pieces glued on. Additionally, rainfall is measured at another AWS (AWS main, Lat: 25.984372°S, Lon: 70.61528°W) directly at the Pacific coast. The setup, calibration, and first results are described in Lehnert, Thies, et al., 2018b.

| Sampling
To characterize the blackish and whitish areas chemically, physiologically, and taxonomically, five substrate replicates of the blackish and whitish patterns each were taken from random positions in close vicinity to the AWS. For the soil analysis and chlorophyll determination, five replicates were used and for ecophysiological experiments and species identification, ten replicates were used. Samples were taken from the top first cm by pressing a 9 cm diameter sterile petri dish into the substrate. All samples were stored in a dry state and kept in the dark at room temperature for one week until the investigations.

| Substrate parameters and chlorophyll contents
From five replicates, each of whitish and blackish pattern substrate, pH (H 2 O, 1:2.5 w:v) and electrical conductivity (EC, 1:5 w:v) were determined by standard procedures (Blume, Stahr, & Leinweber, 2010). For total element analyses, air-dry substrate was ground to <0.5 mm. Total C, N, and S (C t , N t , S t ) were measured using a Vario EL elemental analyzer (Elementar Analysensysteme). Inorganic carbon (C inorg ) was analyzed by the Scheibler method (DIN ISO 10693), and organic carbon (C org ) was calculated by difference (C t -C inorg ). Total Al, Ca, Fe, K, Mg, Mn, P, and Zn were extracted from 0.5 g substrate by microwave-assisted digestion with aqua regia solution (3:1 hydrochloric acid:nitric acid) (Chen & Ma, 2001 ISO standard 11466) and subsequent determination by ion-coupled plasma-optical emission spectroscopy (ICP-OES).
Chlorophyll was extracted from five samples of the whitish pattern as well as five samples of the blackish one using a DMSO method as described in (Ronen & Galun, 1984) and calculated to area.

| Pyrolysis field ionization mass spectrometry (Py-FIMS)
To gain in depth information on the organic matter potentially provided by the CGC during soil formation, pyrolysis field ionization mass spectrometry (Py-FIMS) was applied to three replicates containing 0.5 mg of the CGC community which were scraped off the grit stones. Samples were degraded by pyrolysis in the ion source (emitter: 4.7 kV, counter electrode −5.5 kV) of a double-focusing Finnigan MAT 95. The samples were heated in a vacuum of 10 -4 Pa from 50°C to 650°C, in temperature steps of 10°C over a time period of 15 min. Between magnetic scans, the emitter was flash heated to avoid residues of pyrolysis products. Sixty spectra were recorded for the mass range 15-900 m/z. At each scan, the mass range of m/z 15-900 was recorded and the absolute and relative ion intensities of nine classes of chemical compounds were calculated by summation of the ion intensities of indicator signals to obtain thermograms of their volatilization (Schulten & Leinweber, 1999). The compound classes were: carbohydrates with pentose and hexose subunits (CHYDR), phenols and lignin monomers (PHLM), lignin dimers (LDIM), lipids, alkanes, alkenes, bound fatty acids, alkyl monoesters and sterols (LIPSTERO), alkylaromatics (ALKYL), mainly heterocyclic N-containing compounds (NCOMP), peptides (PEPTI), suberin (SUBR), and free fatty acids (FATTY). Single magnetic scans were combined to obtain one thermogram of total ion intensity (TII) and a summed Py-FI mass spectrum. All Py-FIMS data were normalized per mg sample, and an average value for the CGC community was calculated from the three field replicates. Thermostability of the sample was calculated by dividing the sum of all ion intensities from 50 to 400°C by the sum of all ion intensities from 50 to 650°C.

| Ecophysiology
CO 2 exchange measurements were conducted under controlled laboratory conditions using a minicuvette system (CMS400, Walz Company, Effeltrich, Germany). The response of gross photosynthesis (net photosynthesis (NP) + dark respiration (DR)) related to water content was determined for five samples. Complete desiccation cycles (from water-saturated thalli to air-dry) were measured at 800 μmol photons/m 2 s −1 , ambient CO 2 , and at temperatures of 5°C, 10°C, 15°C, and 20°C. Samples were weighed before each measurement and the water content was calculated as mm precipitation equivalent, following the determination of the samples dry weights after a three-day drying period at 65°C in a drying oven (Heraeus Instruments T6P, Thermo Fisher Scientific Inc.). To obtain the NP response to light, fully hydrated samples were exposed to stepwise increasing photosynthetically active radiation (PAR) from 0 to 2,000 μmol photons/m 2 s −1 at optimal temperature (15°C) and ambient CO 2 concentration. The light cycle (about 45 min duration) was repeated until the samples were completely dry (after 5 hr). Light saturation point (LSP) was defined as the photosynthetic photon flux density at 90% of maximum NP. The CO 2 exchange of the samples was related to surface area.

| Microscopy and cultures
To study the taxonomic composition of the CGC, we used several methods, depending on the taxa. Direct microscopy of the blackish and whitish samples was applied to address the morphological diversity of pro-and eukaryotic algae. An aliquot of a CGC sample, prepared by washing loose organisms off the grit stones, was collected in a drop of water and examined under a stereoscope binocular (2000-C, Zeiss, Germany). The samples were investigated with a light microscope (Axioskop, Zeiss, Germany with DIC optics) under 630 magnification and oil immersion. Additionally, a culturebased approach was applied to all biocrust samples as described in (Baumann et al., 2018).
After at least four weeks, green algae were transferred to BBM agar plates (Bischoff & Bold, 1963) and cyanobacteria to BG11 agar plates (Stanier, Kunisawa, Mandel, & Cohen-Bazire, 1971), using a sterile needle. This was repeated until unialgal cultures were established. Morphology of the phototrophs was studied by light microscopy as explained above. Based on the relevant morphological characters, eukaryotic algae and cyanobacteria were identified to the genus, and if possible, to the species level based on the most recent literature as described by Baumann et al. (2018).
Lichens were identified by carefully removing them from a single grit in a hydrated state with a sterile needle and transferred to a freezing microtome. Thin sections of 25 µm thickness were prepared and transferred to a drop of water on an object slide. Fully hydrated lichens covering grit stones were visualized with the same stereoscope binocular. For electron microscopy, a low-temperature scanning electron microscope (Supra 55VP; Carl Zeiss, Oberkochen, Germany) was used to study fully hydrated Lichenothelia sp. The samples were frozen in liquid nitrogen slush (K1250X Cryogenic preparation system, Quorum technologies) and mounted on special brass trays. After sublimation for 30 min at −80°C, samples were sputter-coated with gold-palladium and viewed at a temperature of −130°C and 5 kV accelerator voltage.
Various colonized grit stones were embedded in waterfree resin and cut into blocks after drying and hardening. The blocks were carefully ground down to a thickness of 30 µm and mounted between a glass objective slide and a cover slide. These thin sections were visualized with a epifluorescence microscope (Axioskop; HBO 50; Zeiss) as described above by exciting the autofluorescence of the mycobiont and the photobionts (blue-violet 395-440 nm excitatory filter; 460 nm chromatic beam splitter; 470 nm barrier filter).

| Statistics
Shapiro-Wilk test and F test revealed that the distribution of substrate and chlorophyll data was not always normal, and variances were not always homogeneous. Hence, Wilcoxon-Mann-Whitney U test was used to determine significant differences of the substrate parameters between whitish and blackish pattern. All statistical analyses were performed using R version 3.4.3 (R Core Team, 2017). Unless otherwise noted significant differences refer to p ≤ .05.

| Description of the microhabitat and biocenosis
The habitat of the newly detected CGC was located 2.5 km off the Pacific coast in the local fog oasis "Las Lomitas" which is situated in the National Park Pan de Azúcar in the Southern Atacama Desert. Here, quartz-and granitoid grit-sized (6 mm) stones paved the desert surface. Blackish and whitish patterns could frequently be observed on the ground in the landscape that were caused by varying colonization rates of the grit stones (Figure 1a). The whitish spots were colonized by <10% of different small lichen thalli whereas blackish appearing grits were 90% coated by these lichen communities resulting in blackish pattern in the landscape.
Wherever this saxicolous (living on stones) biocenosis was present, terrain surfaces were concatenated by organisms occurring on and between the grits and smaller-sized mineral particles ( Figure 1b). Amounts of chlorophyll a+b were more than five times higher in blackish compared with the whitish grit parts (Table 1).
Remote sensing techniques revealed that the area covered by both patterns in the Pan de Azúcar National Park was as large as 350 km 2 with coverages between 20% and 80% (Figure 1c). The biocenosis (Figure 2a

| Ecophysiology and environmental conditions
Climate records showed that dew occurred frequently, predominately during night-time providing between 0.025 and 0.088 mm of liquid water per day (Table 1, Figure 3a). In contrast to dew, fog usually occurred during daytime and provided higher water fluxes delivering 0.38-1.25 mm per day (Table 1). As such, the majority of the total water deposition (92%) was from fog, which mainly was deposited from morning till noon (Figure 3a). High irradiation, with maxima of more than 1,500 µmol photons/m 2 s -1 , was recorded during cloudfree periods (Table 1, Figure 3b). Mean annual temperatures (MAT) ranged between 12°C (air) and 15°C (ground level) (Table 1).
During gas exchange measurements in the laboratory, the optimum water content for photosynthesis of the CGC was found at 0.25 mm (Figure 3c). Besides that, we measured a high light compensation point (LCP; the light level where photosynthesis equals respiration), as well as a high light saturation point for photosynthesis (LSP, 90% of max. net photosynthesis (NP)) of 460 ± 29 and 1625 ± 112 µmol photons/m 2 s -1 , respectively ( Table 2).
The maximum of NP of the CGC was reached at 10°C while an increase to 15 and 20°C had no effect on NP (Figure 3d).
However, dark respiration (DR) strongly increased with raising temperatures, indicating that increasing gross photosynthesis did compensate for additional respirational CO 2 release. Thus, optimal temperature regimes for net primary productivity are those of daytime temperatures above 10°C and night temperatures below 10°C (Figure 3d).

| Ecosystem services provided by the new biocrust type
Cross-sections of colonized grit stones revealed mycobiont hyphae penetrating the outer and inner grit-sized stones and photobionts affiliated by mycobiont hyphae inside the grit stones (Figure 4a). Thin sections of the lichens showed mineral fragments incorporated into the lichen's thalli (Figure 4b).
Elemental analyses on C, N, and P revealed that the blackish pattern contained over three times more organic C and N than the whitish pattern (Table 2). We extrapolated C, N, and P stocks

| Ecology of the novel biocenosis
The blackish patterns paving the ground on a large landscape scale in the fog oasis Las Lomitas in the Atacama Desert were caused by an intimate biocenosis made of fungi, pro-, and eukaryotic algae on grit-sized quartz and granitoid stones (locally called "maicillo"), that can be assigned to the CGCs as described by Elbert et al. (2012), in accordance with our first hypothesis. This CGC is a saxicolous biocenosis since the organisms almost exclusively colonize grit-sized stones. However, it far more resembles a biocrust (Belnap, Büdel, & Lange, 2001, 2003 due to its ability to concatenate the grit-sized stones and to thus form stabilized pavement on the ground. Further, the Regosols in the area were marked by the occurrence of a thin A horizon with very low soil organic matter contents (<0.4%) and a finer grain structure under the thin pavement of grit (Bernhard et al., 2018). This resembles an edaphic substrate rather than rock. For these reasons, we named this type of CGC "grit-crust" to imply the transitional form of this biocenosis between a soil crust and rock cover, due to its unique substrate. In contrast to known biocrust types, such as pinnacled or flat biocrusts (Belnap, 2003), this new biocrust grew around grit stones covering them on all sides and concatenating them.

Various epiphytic and saxicolous lichens covering a large area in
Pan de Azúcar were reported earlier (Follmann, 1965;Redon, 1973;Rundel, 1978;Rundel et al., 1991), while our results showed that the blackish pattern on the ground was caused by high colonization rates of the grit-crust that locally covered 80% of the desert surface as discovered by remote sensing techniques. With this, remote sensing respectively. In addition, this part of the Atacama Desert experiences infrequent rain events (Thompson, Palma, Knowles, & Holbrook, 2003) which are followed by the emergence of ephemeral herb cover (plants with a short life cycle). It will be interesting to study how this cover affects the performance and distribution of grit-crust organisms, as it temporarily changes the near-ground microclimate, reducing light levels and possibly dew formation, while increasing organic-matter inputs.
Similar blackish and whitish patterns in the landscape were reported from the lichen fields in the fog zone of Walvis Bay in the Namib Desert (Lange, Meyer, Zellner, et al., 1994) (Figure 1d).
However, from Walvis Bay, which has a similar water regime as Las Lomitas, a biocrust of the grit-crust type is unknown so far. Only biocrusts dominated by the terricolous chlorolichen Acarospora gypsidesertii are reported (Büdel et al., 2009;Lange, Meyer, Zellner, et al., 1994). The easily overlooked grit-crust of Las Lomitas has not been explicitly searched for in other fog-influenced desert zones worldwide so far, so that it remains uncertain to what extent grit-crust-like biocrusts occur.

| Ecophysiology of the grit-crust community
The water regime at Las Lomitas facilitated mainly eukaryotic green algae and green-algal lichens to colonize the grit stones. Both types of organisms are known to be activated by liquid water sources (rainfall, fog, dew) and additionally by relative air humidity higher than 85% (Lange and Büdel 1994;Büdel, Vivas, & Lange, 2013). In contrast, cyanobacterial lichens (cyanolichens) and cyanobacteria can only be activated by liquid water (Büdel & Lange, 1991;Lange, Kilian, & Ziegler, 1986). Interestingly, in contrast to our second hypothesis, cyanobacteria were part of the grit-crust community as well, probably because the two main liquid water sources, fog and dew, were frequently available at Las Lomitas (Lehnert, Thies, et al., 2018b). We noticed that cyanobacterial presence in the arid and hyperarid zone of the Atacama Desert (e.g., Warren-Rhodes et al., 2006, Wierzchos, Ascaso, & McKay, 2006 is apparently directly related to microhabitats that provide a condensation surface large enough for dew and fog to precipitate. This seems to provide a sufficient amount of liquid water for cyanobacteria to thrive. To support and maintain its biological activity, the grit-crust community depends on achieving a positive net photosynthesis under strongly limiting moisture conditions. In the laboratory, the optimum water content for photosynthesis of the grit-crust community was 0.25 mm (Fig. 3c), which is a remarkably low amount of water, if compared to other biocrusts worldwide. In the fog zone of the Namib Desert, for example, biocrusts dominated by the lichen Acarospora schleicheri had their optimal water content at more than twice as much (6 mm) (Lange, Meyer, Zellner, et al., 1994), and other lichen-dominated biocrusts from the Sonora Desert had their optimal water content between 0.5 and 1 mm (Büdel et al., 2013). Thus, it can be deduced that the newly discovered gritcrust is well adapted to the arid conditions in the Atacama Desert, where water is solely available from fog and dew. The pronounced tolerance of the grit organisms against xeric stress has also been demonstrated for lichens such as Pleopsidium chlorophanum and Buellia species from Antarctica, which are reported to even acclimatize to Martian conditions (de Vera et al., 2014;Verseux et al., 2016), and which were both also found in the Las Lomitas gritcrust. Conditions in the Atacama Desert have indeed been compared to those on Mars and the very low water demands of the grit-crust community documented here supports the hypothesis (Verseux et al., 2016) about the extremophile character of these organisms, which may even be suitable candidates for extra-terrestrial colonization and may have played a role shaping soils and the atmosphere on the ancient Earth.
TA B L E 2 Chemical and ecophysiological characteristics of blackish and whitish patterns (0-0.9 cm depth) observed in the Coastal Atacama Desert of Chile Note: Given are means and standard deviations; n = 5; asterisks indicate significant differences of parameters between whitish and blackish pattern at p ≤ .05; Chl a+b = Chlorophyll a + b. Abbreviations: LCP = light compensation point of grit-crust photosynthesis; LSP = light saturation point.
Intensive solar radiation and high temperature in desert environments can also limit photosynthesis. During fog events radiation levels are low, but as soon as the fog clears intensive radiation might cause (i) a fast heating-up of the biocrust combined with a rapid desiccation that protects the photosynthetic apparatus from damages or (ii) photoinhibition during a slow heating process. As shown by its high LCP, however, the grit-crust was well adapted to high light intensities. Gas exchange measurements showed that the optimal temperatures for net primary productivity were above 10°C during the day, with no decrease in NP up to at least 20°C.
Such conditions are provided with the average daily temperature course at this site ( Figure 3b) (Baumann et al., 2018). Adaptation to fog-desert conditions requires an optimum photosynthetic activity during fog events where light can be quite low, temperatures are moderate, and sufficient liquid water is available but will quickly evaporate after the fog disappears. According to our measurements, optimal photosynthetic activity of the grit-crust community can be expected in the morning hours, especially between 9 and 11 a.m. when light availability is good and water supply is warranted by dew and fog (Figure 3a,b). The diurnal cycle of fog and dew water fluxes demonstrated that the most critical hours for photosynthesis of the grit-crust can be expected in the afternoon, especially after 4 p.m., when liquid water supply by fog and dew is hardly available. A similar conclusion is supported by the climate data recorded by Thompson et al. (2003) over the course of three years for the same location. The fluctuating water regime at Las Lomitas causes frequent wet-dry cycles, which would be lethal for some biocrust types from arid zones without fog. For those biocrusts, such slight and almost daily slender wetting would cost more in recovery respiration than would be won by photosynthetic gain (Belnap, Phillips, & Miller, 2004;Reed et al., 2012). Therefore, the newly discovered grit-crust is quite unique in a way that every fog event can be seen as the "desert's breath" which is followed by instant biological activity such as primary production and other fundamental ecosystem services which are of high significance in poor ecosystems such as the Atacama Desert.

| Ecosystem services and the potential importance of cryptogams for the ancient Earth
Biocrusts have played an important role during the evolution of terrestrial ecosystems by mineral weathering and soil development since the Paleozoic (Mitchell et al., 2016). The occurrence of for example lichens as a symbiosis between algae and fungi dates back to the Precambrian 400 million years ago (Taylor & Osborn, 1996), while cyanobacteria are suspected to have played a significant role during early oxygen production and atmospheric enrichment with oxygen even before the Great Oxidation Event (GOE) 2.45-2.32 billion years ago (Lalonde & Konhauser, 2015;Schirrmeister, de Vos, Antonelli, & Bagheri, 2013). Until land plants appeared, about 70 million years ago (Heckman, Geiser, Eidell, Stauffer, Kardos, & Hedges, 2001), the terrestrial organic C pool of early soils as well as the atmospheric oxygen pool was fueled solely by the biological activity of cryptogams for a long geological period (Lalonde & Konhauser, 2015;Mergelov et al., 2018). Assuming that the arid to hyperarid conditions of the Atacama Desert have been stable since the middle Miocene (Sun, Bao, Reich, & Hemming, 2018), the grit-crust community found at the modern coastal Atacama could be a continuation of a very early community type. As a consequence, the grit-crust could act as a model community which fixes C during photosynthesis, conducts bio-weathering processes, and releases oxygen in the same way it may have since before the emergence of land plants (Makeev, 2015;Mergelov et al., 2018;Retallack, 2001).
For example cyanobacteria detected in the recent grit-crust, promote bio-weathering by causing alkaline conditions which can dissolve quartz. This has been demonstrated for the cyanobacterium Chroococcidiopsis (Büdel et al., 2004), which is one of the genera found in the grit-crust community. In addition to bio-weathering activity via alkalization, we also found mechanical bio-weathering by contraction and expansion of lichen thalli during the wet-dry cycles. This causes cracks within the stones (Figure 4a) which is known to contribute to bioweathering (Chen, Blume, & Beyer, 2000) and, finally, to initial soil formation. In addition to lichens and cyanobacteria, also the fungal genus Lichenothelia was part of the grit-crust community and is known to also grow on and within exposed rocks in the Sonoran and Mojave Desert contributing to bio-weathering in the same way as lichens (Muggia, Kocourková, & Knudsen, 2015). Besides mineral breakdown by the gritcrust organisms, the desert surface is transformed into initial soils by C and N input from the biocrust (Table 2) influencing geomorphological processes from the micro-to the landscape scale. As Py-FIMS analyses revealed, the grit-crust community and thus the potential input of organic C from the grit-crust consisted of 11.5% phenolic derivates followed by 9.3% alkyl aromatic compounds. These substance classes are not just relatively heat resistant but also relatively difficult to decompose and are known to accumulate, for example, in steppe-like environments (Leinweber et al., 2009). We conclude that accumulation processes lead to the formation of a terrestrial protopedon with a cryptic A horizon enriched in C and N, as suspected in our third hypothesis. Similar processes of soil formation were reported for Antarctica (Mergelov et al., 2018), where cyanobacteria and fungi contributed to the formation of soils. This could be a significant contribution to biogeochemical cycles in such organic matter poor desert environments and even to global cycles if we assume that large areas in fog-influenced deserts may be covered by similar but, to date, undiscovered biocrust types. Although probably slow growing and not being a large pool of C and N, the new grit-crust type is still the one and only transformer of the desert's mineral substrate into an early type of soil (protopedon) that we are aware of for a considerable part of the Atacama Desert.

| CON CLUS IONS
We conclude that the grit-crust is the primary supporter of the poor desert ecosystem with its few and highly scattered vascular plants and a hitherto unknown or only partly known contingent of heterotrophic bacteria, microfungi, protists, invertebrates, and the betterknown small vertebrates and finally larger mammals like foxes and guanacos. Additionally, this biocenosis might be highly comparable to communities shaping ancient Earth. For both of these reasons, this highly interesting and widespread grit-crust ecosystem certainly deserves more research to better understand its full taxonomic composition and the ecophysiological traits that guarantee activity and survival. Additional multidisciplinary surveys are recommended in the Atacama and other deserts globally, to extend our knowledge of the functioning of this newly discovered biocrust type and its contribution to these unique ecosystems.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.