Estimating the growth rate in desert biological rock crusts by integrating archaeological and geological records

Biological rock crusts (BRCs) are ubiquitous features of rock surfaces in drylands composed of slow-growing microbial assemblages. BRC presence is often correlated with rock weathering, soiling effect, or with mitigating geomorphic processes. However, their development rate has not been quantified. In this work, we characterised and dated BRCs in an arid environment, under natural conditions, by integrating archaeological, microbiological and geological methods. To this end, we sampled rocks from a well-documented Byzantine archaeological site, and the surrounding area located in the Central 5 Negev Desert, Israel. The archaeological, which is dated to the 4th-7th centuries CE, was constructed from two lithologies, limestone and chalk. BRC started developing on the rocks after being carved, and its age should match that of the site. The BRC samples showed mild differences in the microbial community assemblages between the site and its surrounding, irrespective of lithology, and were dominated by Actinobacteria, Cyanobacteria and Proteobacteria. We further measured the BRC thickness, valued at 0.1-0.6 mm thick BRC on the surface of 1700 years old building stone block of about 0.1 square metres. Therefore, a 10 BRC growth rate was estimated, for the first time, to be 0.06-0.35 mm 1000 yr-1. We propose that BRC growth rates could be used as an affordable yet robust dating tool in archaeological sites in arid environments.


Introduction
In arid and hyper-arid environments where abiotic processes are considered as the primary contributor to landform formation, barren rock surfaces, free of vegetation, are a ubiquitous feature (Owen et al., 2011). These surfaces are exposed to multiple Microscopic examinations of thin sections (30 µm thick) that were prepared from the limestone and chalk blocks showed that, as expected, the BRCs were restricted to the atmospherically-exposed parts of the rocks (Wieler et al., 2019). The BRCs were characterised by a hardpan-laminated structure composed of masses of micritic to microsparitic carbonate layers interbedded with microbial coatings covering the lime and chalk host-rocks. BRC thickness significantly differed between those found on limestone compared to chalk building blocks, even when located on a single wall in the archaeological site (Table 1, Fig.   70 2). Limestone BRC thickness at both the Byzantine site and along the rock outcrops located in the limestone quarries at the adjacent slopes, ranged between 1.4-4.3 mm (Fig. 2D, 2E). In contrast, chalk BRC thickness differed between the blocks in the Byzantine site and the natural chalk slopes. The chalk BRC thickness at the Byzantine site ranged between 0.1-0.6 mm ( Fig.   2B), while at the adjacent slopes, it ranged between 0.5-1.8 mm (Fig. 2C). The observed 0.1-0.6 mm BRC thickness upon the chalk building blocks at the well-dated Byzantine site (dated between the 4 th -7 th centuries, CE), located under long stable arid conditions and free of anthropogenic effects, suggest a growth rate of 0.06-0.35 mm 1000 yr -1 .

Isotopic composition
The biogenic nature of the crusts was confirmed using a cross-section analysis of the stable carbon and oxygen isotope ratios in the crust and host rock. For the limestone sample, a shift was found between the δ 13 C values for the BRC (0-2 mm) and the host rock (2-5 mm) layers, with BRC values ranging between -4‰ and -5‰ and host-rock values between 0‰ and 1‰ 80 (Fig. 3A). However, for the chalk, δ 13 C values for the BRC ranged between -0.2‰ and -1.9 ‰, and between 0.1‰ and-2.7‰ in the host rock. These results were consistent for both the slope and archaeological site samples. The limestone δ 13 C values are typical indicators of carbon isotope exchange of primary marine CaCO 3 (abundant in the bedrock) with CO 2 released by microbial respiration (i.e. of carbon originating from photosynthesis) with the subsequent precipitation of pedogenic calcrete (Brlek and Glumac, 2014). The differences in δ 13 C values between the chalk and limestone are suggested to result from the 85 BRC thickness. Thicker BRCs, as observed in the limestone samples, may hold more biogenic activity compared to the chalk.  To elucidate the identity of the bacterial communities on the BRCs, we performed multiplexed barcoded amplicon sequencing of the small subunit ribosomal RNA gene (SSU rRNA gene). As expected, we found simple BRC communities (i.e., lowrichness and low-diversity) in all BRC samples. Comparing the BRCs collected from the Byzantine site to the nearby slopes from the two primary lithologies (the limestone and chalk) showed no statistically significant differences in the observed num-95 ber of OTUs or their inverse Simpson or Shannon indices, with all samples averaging 312 ± 11 OTUs (Fig 4. A, Table S1). The city samples did, however, show a slightly higher dominance of the most abundant OTU compared to the slope samples (BP values of 0.24 ± 0.03 vs 0.14 ± 0.01; p=0.0017). The bacterial community composition of the BRCs was very similar in both lithologies and sample sources and was heavily dominated by members of the phylum Actinobacteria, followed by Cyanobacteria, Proteobacteria (mainly Alphaproteobacteria from the orders Sphingomonadales and Rhodospirillales) and Bacteroidetes 100 (Fig. 4B). A variance partitioning analysis, did however, show a small difference between the samples from the city and the ones from the slopes, explaining 8.1% of the variance in the Morisita-Horn distance matrix (p=0.023). In contrast, neither the lithology nor the interaction between the lithology and sample source correlated with differences in bacterial community composition (Fig. 4C, Table S2). These differences were not detected when comparing the dominance each phylum between locations using Aligned Rank Transformed ANOVA test ( Fig 4D). Instead, 64 individual OTUs were differentially more abun-105 dant in the city samples while 58 were differentially more abundant in the slope samples (from a total of 732 OTUs; Fig S1).
These OTUs came from all the dominant phyla, with no discernible taxonomic pattern.  confidence intervals and statistical significance (based on ART-ANOVA) on the phylum level between slope and city samples (only phyla that account for more than 5% of the relative abundance are shown).

Discussion
BRCs are a common and important feature of atmospherically-exposed rock surfaces in drylands around the globe, yet reliable growth estimations based on field data are rare. Only a handful of studies tried to estimate BRC growth rates, among them 110 Lange (1990) measured radial growth rates of epilithic lichens in the arid Negev Desert, Israel, and reported an average growth rate of 0.371 mm yr -1 . Another study conducted by Krumbein and Jens (1981) found that desert varnish, a typical rock crust in desert regions, show black fungi microcolonies after eight weeks isolation in the lab. Liu and Broecker (2000)  In this study, we observed 0.1-0.6 mm thick biological rock crust coverage on chalk building blocks of the Byzantine site (Fig.   2B) dated to the 4 th -7 th centuries CE (Tepper et al., 2018). The building blocks experienced long stable arid conditions and were free from direct anthropogenic influences. Thanks to the processing method of the bricks used for construction, leaving a BRC-free surfaces exposed to the elements, hence, the BRC is confined by the dating of the archaeological site (Tepper et al., 2018). This provides us with an estimated growth rate of 0.06-0.35 mm 1000 yr -1 for BRCs of primarily bacterial origin.

120
Besides, the presence of a 1.4-4.3 mm thick BRC in all limestone rocks (Fig. 2D, 2E) and 0.5-1.8 mm on the chalk rocks collected from the slopes (Fig. 2C), provide a plausible maximum for BRC growth under these conditions. From a microbiological perspective, all the samples studied here showed very similar microbial community composition, irrespective of lithology or BRC thickness. This similarity in composition demonstrates the indifference of microorganisms to the type of attachment surface in this case and that the community probably changes very little after establishing. However, the 125 differences in BRC thickness between the chalk and limestone sampled from the slopes could indicate that the latter can better support BRC growth.
Differences in the dominance of the most abundant OTU as well as some differences in the relative abundance of about 17% of the OTUs were observed between BRCs from the city and slope samples. However, these differences were relatively minor and lacked a clear taxonomic pattern. Overall, the results point to a very deterministic successional course of BRCs development 130 on rock surfaces. Moreover, the similarly in the microbial composition between rocks that were in contact with the ground to the those that were detached from the ground (chalk BRCs found in Shivta), indicate a major role to aeolian processes in determining the community composition of BRCs in deserts as was previously reported (Wieler et al., 2019).
Assisted recovery of lithobiontic communities has not been conducted in natural settings, and most research focused on the regeneration of soil biocrust (Velasco Ayuso et al., 2017). Artificial cultivation of soil biocrusts (Zhang et al., 2018) suggest 135 that inoculation of cyanobacteria and algal communities enhance the recovery of biocrusts. Testing the soiling impact on rock surfaces, suggest that bacterial colonisation play an important part in the development of fungal biofilms (Viles and Gorbushina, city (Fig. 1). BRCs are common on the atmospherically-exposed parts of many of the rocks. The site covers an area of 0.8 km 2 and contains the remains of three churches, a central watch tower and associated structures, two sizeable public water reservoirs, three wine-presses, a large inn of the Byzantine period and numerous private buildings. The main occupation of the residents in the Byzantine period appears to have been agricultural, mainly viticulture for the export of wine, and road services for Christian pilgrims travelling to and from Mt. Sinai (in modern-day Egypt). Thanks to these activities, the village appears to stone with a minimum use of wood, which would otherwise had to be imported from great distances. Therefore, the upper floors were supported with stone arches and wood was sparsely used for doorframes, doors and shelves. In lieu of wood, many installations, such as animal troughs, were carved from local stone. The walls of the houses are well-preserved, offering researchers a unique view of buildings constructed over 1700 years ago that stand to heights of two and even three stories high.
The lower courses of the walls on the ground floor were made using rough limestone blocks, taken from Nezer Formation, 195 while the upper walls were constructed of a lighter chalkstone, taken from Menuha Formation that is more easily worked into blocks. The spaces between the heavy stones of the lower courses were sealed with mortar, and the interior walls were often covered with a base of mud plaster covered with white lime plaster (Fig. 2).
Twelve rock samples were collected from the limey Nezer Formation including six samples from the archaeological site and six from the nearby natural slope. The same sampling procedure was applied for the chalky Menuha Formation. To avoid the slope aspect effect that may lead to different moisture regime, all samples were retrieved from a south-facing slopes/walls and were collected during January 2015.

Geological analysis
The geological methods used in this study are based on direct field observations and characterisation of the subjected lithologies (i.e., Limestone and Chalk) using thin sections, XRD analyses, total effective porosity and stable isotope analysis. Petrographic thin sections, 30 µm thick, were prepared for each lithology to test the main components in both the rock crust and the host rocks examined under a light microscope (Zeiss, Oberkochen, Germany). XRD analyses for bulk mineralogical components 210 (Sandler et al., 2015) were conducted separately on the rock crust and on the host rock, three replicates were collected from each lithology. Powdered samples were scanned by a PANALYTICAL X'Pert3 Powder diffractometer equipped with a PIXcel  (Scherer, 1999) was performed using a gas permeameter and porosimeter device (Core laboratories, Amsterdam, Netherlands) on 12 rock core cylinder samples (radius 18.5 mm and height 26.5 mm). The value of chalk total effective porosity was cited from Schütz,et 215 al., (2012) and reported in Table 1.
For the stable isotopes δ 13 C and δ 18 O analyses, 1-2 mg of rock surface powder was obtained using a microdrill (Dremel,