Terrestrial surface stabilisation by modern analogues of the earliest land plants: A multi‐dimensional imaging study

The evolution of the first plant‐based terrestrial ecosystems in the early Palaeozoic had a profound effect on the development of soils, the architecture of sedimentary systems, and shifts in global biogeochemical cycles. In part, this was due to the evolution of complex below‐ground (root‐like) anchorage systems in plants, which expanded and promoted plant–mineral interactions, weathering, and resulting surface sediment stabilisation. However, little is understood about how these micro‐scale processes occurred, because of a lack of in situ plant fossils in sedimentary rocks/palaeosols that exhibit these interactions. Some modern plants (e.g., liverworts, mosses, lycophytes) share key features with the earliest land plants; these include uni‐ or multicellular rhizoid‐like anchorage systems or simple roots, and the ability to develop below‐ground networks through prostrate axes, and intimate associations with fungi, making them suitable analogues. Here, we investigated cryptogamic ground covers in Iceland and New Zealand to better understand these interactions, and how they initiate the sediment stabilisation process. We employed multi‐dimensional and multi‐scale imaging, including scanning electron microscopy (SEM) and X‐ray Computed Tomography (μCT) of non‐vascular liverworts (Haplomitriopsida and complex thalloids) and mosses, with additional imaging of vascular lycopods. We find that plants interact with their substrate in multiple ways, including: (1) through the development of extensive surface coverings as mats; (2) entrapment of sediment grains within and between networks of rhizoids; (3) grain entwining and adherence by rhizoids, through mucilage secretions, biofilm‐like envelopment of thalli on surface grains; and (4) through grain entrapment within upright ‘leafy’ structures. Significantly, μCT imaging allows us to ascertain that rhizoids are the main method for entrapment and stabilisation of soil grains in the thalloid liverworts. This information provides us with details of how the earliest land plants may have significantly influenced early Palaeozoic sedimentary system architectures, promoted in situ weathering and proto‐soil development, and how these interactions diversified over time with the evolution of new plant organ systems. Further, this study highlights the importance of cryptogamic organisms in the early stages of sediment stabilisation and soil formation today.

The earliest land plants were non-vascular and small statured (Edwards et al., 2021a;Strullu-Derrien et al., 2018), and progressively evolved to more extensive coverings as trees and forests by the middle Devonian (~385 Ma; Stein et al., 2020). The pre-vascular and earliest plant-based biotas are mostly comparable to modern cryptogamic ground covers (CGCs) (Kenrick et al., 2012;Mitchell et al., 2016Mitchell et al., , 2021a, which are variable communities composed of bryophytes (liverworts, hornworts, mosses), lichens, algae, fungi, and bacteria. Although uncertainties remain on how evolutionary ancient some of these lineages are, some, e.g., the liverworts have a lengthy fossil record (Hernick et al., 2008) and the relevance of modern bryophytes has been supported by recent phylogenetic and molecular clock analyses (de Sousa et al., 2019;Harris et al., 2020;Leebens-Mack et al., 2019;Morris et al., 2018;Puttick et al., 2018) (Figure 1). Crucially, modern liverworts (and hornworts) share morphological features with the early land plants, as evidenced from fossils in the exceptionally preserved 407 Ma Rhynie chert; these features include unicellular rhizoids Jones & Dolan, 2012;) and symbiotic associations with microbes ( Figure 1b). Specifically, some Rhynie chert plants and many modern liverworts and hornworts form mutualistic symbiotic associations with soil fungi (Desiro et al., 2013;Duckett et al., 2006;Field et al., 2015;Humphreys et al., 2010;Rimington et al., 2018;Strullu-Derrien et al., 2014, 2018. Cyanobacterial endophytes are ubiquitous in hornworts while in liverworts they are restricted to the Blasiales (Adams & Duggan, 2008), however they are extremely rare in the Rhynie chert (Strullu-Derrien, 2018). While early land plants (cryptophytes; Edwards et al., 2015Edwards et al., , 2021a were mostly axial/erect, leafless, and rhizomatous (Edwards et al., 2014;Strullu-Derrien et al., 2018), there are some fossilised remains suggestive of thalloid plants from the Silurian and Early Devonian (Edwards et al., 2021a;Tomescu & Rothwell, 2006). Moreover, both early-(Blasia and Lunularia, Galloway et al., 2017) and later diverging (Marchantia, Crandall-Stotler et al., 2009) genera secrete the polysaccharide xyloglucan from their rhizoids. Xyloglucan released from plant rhizoids and roots has been shown to be an effective soil particle aggregator and, given its occurrence in extant liverworts, it has been suggested that xyloglucan released from the rhizoids/rhizoid-like structures of the earliest land plants may have had a similar role, aiding the formation of primeval soils (Galloway et al., 2017). Consequently, the exudates and anchorage structures of modern bryophytes make them highly suitable analogues for understanding how early land plants contributed to the stabilisation of sedimentary surfaces at the scale of micrometres to centimetres.
We investigated CGCs growing on loose regolith sediments in a variety of settings from Iceland and New Zealand. We applied 2D and 3D imaging through scanning electron microscopy (SEM) and X-ray computed tomography (μCT), respectively, to determine how sediment grains become stabilised by CGC organisms. By understanding these processes in modern CGC plants, our aim was to infer how the earliest embryophytes stabilised loose surface sediments, leading to promotion of plant-mineral interactions, in situ weathering, proto-soil development, and the likely impact on the architecture (and mudrock content) of fluvial sedimentary systems in deep time.
2 | MATERIAL S AND ME THODS 2.1 | Fieldwork, organisms, and biological sample preparation CGC proto-soils were collected in various forms (as cores, clumps, and sections) from various sites in Iceland and New Zealand, including glacial moraine, volcaniclastic regolith, and lava flows (see Table S1 for further information). Cores measuring 25 × 80 mm at their largest were extracted using a cork borer and placed in plastic vials, and clumps were collected at random to obtain larger soil surface areas. Plants were fixed with 10% formalin, which was added for preservation and to prevent desiccation. The liverworts studied include Treubia lacunosa, Haplomitrium gibbsiae, and Lunularia cruciata from New Zealand, and Blasia pusilla from Iceland. Mosses studied include Polytrichum alpinium and Racomitrium lanuginosum, both from Iceland. We also collected some Lycopodium specimens from Iceland. Sites were selected based on (a) the presence of early diverging land plants (Figure 1), and (b) substrate, including loose volcaniclastic regolith which represents virgin 'primeval' terrains, being devoid of vascular plants, and whose formation can be accurately K E Y W O R D S bryophytes, cryptogamic ground covers, Palaeoenvironments, plant evolution, sediment stabilisation, soil, X-ray computed tomography dated from geologically recent effusive eruptions. Sample preparation and analyses were carried out in the Imaging and Analysis Centre (IAC) at The Natural History Museum, London (UK). Some cores and soil clumps were imaged as they were collected, but in some cases plants were removed to enable clearer imaging of their ventral surface using scanning electron microscopy (SEM).
The removed plants were put through an alcohol series and critically point dried. Critical point drying (CPD) dehydrates biological tissue and replaces the water with liquid CO 2 . When the temperature is raised above 35°C, the liquid CO 2 becomes vapour, keeping the biological morphology and ultrastructure intact. This allows biological tissues to be studied in an SEM (or other imaging instrumentation under vacuum) without desiccation and collapse, which would likely occur with simple air drying. The plant is first put through an alcohol (ethanol) series of varying concentration percentages (between 30% and 100%) to gradually remove water from the plant. Following this, the plant is placed inside the critical point dryer; ethanol is added to the chamber, followed by liquid CO 2 , and after 30 s the mixture is vented from the chamber. This cycle is repeated six times. Lastly, the sample is heated to 39°C and the pressure set to the critical point of 75 BAR. Following this, the sample may be removed.

| Scanning electron microscopy (SEM)
Plant specimens were affixed to carbon sticky pads and mounted upside down (with ventral surface facing upwards) on standard stubs for SEM imaging, without coating, using a FEI Quanta 650 FEG SEM and a Zeiss Leo 1455 variable pressure SEM. Variable conditions were used depending on the sample in both secondary electron (SE) F I G U R E 1 (a) Summary chart highlighting the evolution of different CGC elements from contrasting molecular, phylogenetic and fossil dating methods. (b) Schematic land plant phylogeny of modern terrestrial organisms, focussing on the bryophytes (and specific liverwort genera) studied in this work. Key features (and lack of features) summarised graphically in different plant groups and liverwort genera. Groups correspond to different groups, based on morphology and other features, that are assigned in this work. (a) adapted from Mitchell et al., 2021a, (b) adapted from  and backscatter detector (BSD) modes. Low pressure BSD mode (on the Leo) was used for those samples that had not been critically point dried, and high pressure SE mode (on the Quanta) was used for those specimens that had been critically point dried (see Table S2 for further information relating to specific imaging configuration on both instruments). We also studied thin sections, which were prepared via a standard method of vacuum impregnation and cut to ~30 μm thickness. Thin sections were fixed to sample holders with copper tape to prevent charging, were without cover slips, and were uncoated.

| X-ray computed tomography (μ CT)
μCT was employed to visualise the 2D and 3D structure of proto-soil cores, clumps and plants non-destructively using a Nikon Metrology HMX ST 225 μCT scanner with a tungsten reflection target. Variable scanning conditions were used for each sample (summarised in Table S2). Voxel (3D pixel) sizes ranged between 15 and 38 μm. A copper filter was sometimes used to pre-harden the beam and remove unwanted lower energy X-rays. Other parameters, including the number of projections collected (3142), the μA (180), the exposure time (708 ms), and the frames per projection (1), were consistent for all scans. Average scan time was ~35 min. Scans were reconstructed into 3D tomographic datasets as .tiff stacks using CT Pro Software (Nikon Metrology) and were rendered using Drishti v2.5 (Limaye, 2012) and ORS Dragonfly software to reveal 3D and 2D (X, Y, Z axes) views (see Table S2). In some cases, liverwort thalli were digitally segmented, 3D thickness surface meshes were applied, and imaging analysis was undertaken using ORS Dragonfly software v 2020.1. No staining agents (e.g., iodine) were used, and the fixation of soil and plant material with 10% formalin, together with short scan times, aided in prevention of plant desiccation (and movement) during scanning.

| Methods of stabilisation
Our results show that there are many macro-to-nanoscale (cmμm) ways in which cryptogamic plants interact with sediment/soil surfaces leading to stabilisation, and that this varies depending on the morphology and features associated with the different plants; these are summarised in  However, the growth form of the plants appears to influence the type of interactions and therefore on soil development, as well as the potential interactions (Mitchell et al., 2021a). Thalloid plants tend to grow laterally, encrusting surfaces, whereas some forms of erect and axial plants that grow vertically can result in deeper, organic rich profiles (Mitchell, Strullu-Derrien, et al., 2021). Combining the observations F I G U R E 3 Examples of sediment/soil surface stabilisation by plant rhizoids from 3D imaging (μCT) in plants bearing a thallus, rhizoids, and with xyloglucan secretion from rhizoids (group 1) or non-xyloglucan exudates from the thallus (group 2). (a-e) Lunularia (thalloid liverwort; group 1), (f-j) Blasia (thalloid liverwort; group 1), (k-o) Treubia (liverwort; group 2). (a-c) 3D thickness maps of thalli; visualising the thalli by this method enables clear variations in thallus thickness, with the thickest region usually found along the midrib (yellow colour). Soil grains (white) adhering to specific locations of the thalli ventral surface, mostly concentrated along the thicker midrib. Rhizoids visible as purple-coloured thin strands. (g-j) Various soil, mineral and grain material stuck to the underside of a thalli. Some of these may be held in place by secretions/mucilage from pores. Examples of numerous other soil-dwelling organisms are indicated including fungal hyphae (g, i), diatoms (g, i), bacterial filaments/chains (h, j), and bacterial rods (h, green circle). White arrow = rhizoids, blue arrow = soil/ sediment material held in place by rhizoids, yellow arrow = grains stuck to rhizoids, green arrow and circle = other organisms stuck to/inhabiting plant surface.
are characterised by rhizoidless underground rhizomatous axes (Renzaglia et al., 2018). All rhizoids consist of elongate, tip-growing cells that function in water acquisition and also play a role in anchorage and in sediment stabilisation. However, there are considerable structural differences between the rhizoids of mosses and liverworts. There are structural differences even within liverworts, where some complex thalloids develop two types of rhizoids: larger smooth rhizoids grow at right angles to the thallus and provide anchorage to the sediment surface, and thinner pegged rhizoids run parallel to the thallus and function as an external water conducting system (Duckett et al., 2014; e.g., Lunularia, see Figure 3l).
Such differences demonstrate the complex nature of some anchorage systems and lead to different mechanisms for sediment grains entrapment.
We observed that dense rhizoid mats and tufts radiating from the ventral surface of liverwort thalli interact in complex ways with the sediment surface. Rhizoid tufts can hold sediment grains of variable sizes within their structure and in close proximity to the ventral surfaces of thalli, predominantly along the thallus midrib (particularly larger grains), from which the rhizoid tufts extend (e.g., see Lunularia in Figures 2-4). Grains can also become trapped by adherence to individual rhizoids through physically entwining at (h-j) Three views of the same moss core showing the complete soil profile in 3D, highlighting the different components; green = plant material, brown = finer grained soil minerals, red = high density soil grains. White arrow = rhizoids, blue arrow = soil/sediment material held in place by rhizoids, yellow arrow = grains stuck to rhizoids, red arrow = filamentous structures (likely fungi) holding sediment/ soil material in place. the rhizoid tip, through xyloglucan secretion (Galloway et al., 2017), and potentially by electrostatic forces. Moss rhizoids are significantly different, being multicellular structures that typically arise from the aerial axis where they can form greatly entangled tufts that bind sediment grains and organic materials. They are generally smooth but can have papillose surfaces, and are much narrower than the unicellular rhizoids of liverworts, typically measuring 20 μm in diameter. Moss rhizoid systems growing through soils branch extensively, with the sometimes thigmotropic (Pressel & Duckett, 2009) furthest ramifications only 3-5 μm diameter, which is a similar size to soil-dwelling fungal hyphae, indicating that moss rhizoids may also play a role in nutrient acquisition and potentially explaining the lack of fungal symbiosis in them (Field et al., 2015).
It has been suggested that the release of a non-cellulosic polysaccharide adhesive by rhizoid tips may also contribute to their attachment to the substrate (Jones & Dolan, 2012;Odu, 1989 and literature therein).
Rhizoid 'stickiness' is crucial for their adhesion to solid substrates, and there are variations in function even between different mosses. For example, Racomitrium lanuginosum does not produce new shoots from deep within the substrate; it has relatively few rhizoids, and because of that soil grains become trapped mainly in the upright leafy stems. On the other hand, Polytrichum sp. produces underground rhizomes with numerous rhizoids, increasing the probability that often-forming rope-like structures which increase the likelihood of grains adhering to them. Grains also become trapped in both the living above ground upright leafy stems and in the buried portions. We have also shown here that moss rhizoids, in Polytrichum at least, are capable of growing through grains, which will inevitably contribute to mineral destruction and soil development (Figure 8). This physical de-struction is in addition to chemical dissolution through from the release of organic acids, shown to be an effective weathering facilitator (Lenton et al., 2012). Therefore, in addition to aggregating F I G U R E 8 (a, b) 3D reconstruction of Polytrichum moss rhizoids growing through a soil grain, and (d-f) 3D reconstruction of Lycopodium and its subterranean rhizoids. (c) 2D digital section view of Polytrichum stem growing within the soil grain crack. (e) Soil digitally removed showing only high density soil grains, while in (f) all soil has been digitally removed. White arrow = rhizoids, green arrows = plants and surface organisms. 3D videos as Videos S8 and S9 complement this figure. sediments, moss rhizoids can potentially aid in soil development through mineral grain destruction.
Fossils and sedimentary structures provide direct evidence that rhizoid-based rooting systems evolved in early plants before the evolution of roots (Hillier et al., 2008;Jones & Dolan, 2012;. The best fossil examples come from the exceptionally preserved 407 Ma Rhynie chert. These plants had erect and axial rather than thalloid growth, and in all but one species, non-septate (i.e., unicellular) rhizoids developed in dense tufts from bulbous or creeping prostrate axes in contact with thin sediment/ proto-soil surfaces (Edwards, 2004). Nothia aphylla had a branched rhizome with a distinct ventral ridge of rhizoids (Kerp et al., 2001), analogous to the thallus midrib of complex thalloid liverworts.
Nothia is thought to have been a geophyte, with a persistent rhizomatous system growing within a shallow sandy sediment substrate containing plant remains. The aerial parts might have been shortlived, and the rhizomatous pattern of growth enabled large colonies of Nothia to develop (Daviero-Gomez et al., 2005). The rhizomes of two other Rhynie species (Ventarura lyonii and Trichopherophyton teuchansii) are also thought to have grown within the substrate sediment layer because rhizoids developed on all of their surfaces (Edwards, 2004). By contrast, in the plant Aglaophyton majus, rhizoids developed locally where prostrate axes made contact with the substrate surface. The upper surface of these axes have stomata, further supporting the idea that they were surficial rather than subterranean (Edwards, 2004). Rhizoids developed on one side of the axes interpreted as rhizomes in Rhynia gwynne-vaughanii as well as ectopically on aerial axes where they arose from small mounds of tissue, the so-called hemispherical projections. Horneophyton lignieri differs from these other species in the tuberous shape of the axis, a corm, in contact with the substrate sediment. Numerous rhizoids developed from its lower surface .
In the Rhynie chert plants, rhizoids typically measure 20-30 μm in diameter but they vary greatly in length, being shortest in Trichopherophyton (<250 μm) and Ventarura (<450 μm), and very elongate in Horneophyton (>2000 μm). The earliest fossil rhizoids therefore resemble the simple, robust, non-septate rhizoids of modern liverworts rather than the septate rhizoids with ramification of decreasing diameters of mosses. It is likely that functionally, and in the ways that they interacted physically with the substrate, they were closer too.
The earliest evidence of rooting systems comes from stem-group lycopods in the Drepanophycales. Their development and anatomy were recently characterised in detail in exceptional well-preserved specimens of Asteroxylon mackiei (Hetherington et al., 2021;Hetherington & Dolan, 2018), which is the only plant in the Rhynie chert known to possess root-like organs. Although these early roots differed in some key anatomical and developmental respects to those of modern lycopods they represent a step change in the manner in which plants interact with their substrates. The evolution of specialised geotropic axes opened up a new means of anchorage and interaction with the substrate, but came later than the early divergent rhizoid-bearing plants.

| Symbiosis
We did not specifically investigate fungal symbioses here, because fungal hyphae fall below the limit of resolution of our 3D imaging methods and are likely obstructed from view in our 2D SEM imaging. However, considerations of these are pertinent in the context of early divergent land plant interactions with their substrate.
Today, mutualistic symbioses (mycorrhizae) develop between the roots of most plants and soil fungi in the Dykaria (Ascomycota and Basidiomycota) and Mucoromycota (Mucoromycotina and Glomeromycotina); these plant-fungal mutualisms play crucial roles in the acquisition of key elements required in the host plant metabolism (e.g., nitrogen, phosphorus) (Smith & Read, 2008). Some extant, early divergent bryophytes (thalloid liverworts, hornworts) and spore-producing vascular plants (lycopods, ferns) (Figure 1 (Field et al., 2016;Rimington et al., 2015Rimington et al., , 2020, a relationship that extends back more than 400 million years (Berbee et al., 2020;Strullu-Derrien et al., 2014, 2018. With the possible exception of Nothia aphylla (Krings et al., 2007a(Krings et al., , 2007b, rhizoids are not known to be involved in the mycorrhizal-like associations of the Rhynie chert plants and in the earliest diverging Haplomitriopsida liverworts (i.e., the fungusfree rhizoids in Treubia and the absence of rhizoids in Haplomitrium), whilst they are the conduits for fungal entry in all other liverwortfungal associations (Duckett et al., 2006;Read et al., 2000), suggesting that Haplomitriopsida fungal relationships were established before the evolution of rhizoids. The notion that mycorrhizal-like symbioses evolved during the early colonisation of the land by plants is consistent with their global abundance in the soil today, their broad phylogenetic distribution, and recent evidence acquired from plant and fungal genomes (Berbee et al., 2020). There is also direct evidence of mycorrhizal-like associations in the fossil record.
The earliest geological evidence again comes from the Rhynie chert.
Both the sporophyte and gametophyte of Aglaophyton majus formed associations with a fungus attributed to the Glomeromycota (Remy et al., 1994;Strullu-derrien et al., 2014;Taylor et al., 2005) and, like in Treubia, its rhizoids were fungus free indicating an alternate route for colonisation. Mycorrhizal-like endophytes were documented in Nothia aphylla (Krings et al., 2007a(Krings et al., , 2007b, and here rhizoids are thought to have been the route of infection. Recently, a mycorrhizallike association with Mucoromycotina was documented in the plant Horneophyton lignieri . In both Nothia and Horneophyton, fungal colonisation consisted of intra-and intercellular phases, as typical of Treubia (Pressel et al., 2010). Mycorrhizal associations were therefore present in early terrestrial ecosystems where they probably played a crucial role in plant colonisation during primary succession, which often begins under conditions of nutrient impoverishment (Smith & Read, 2008). The fungal structures observed in the fossil plants represent the end points of a much larger mycelial network that extended into the soil far beyond the rhizomes and rhizoids of the host plant (Smith & Read, 2008). This relationship likely made a profound contribution to plant-soil interactions and the formation of proto-soils leading to aggregation and stabilisation of sediment surfaces and the promotion of in situ biologically mediated weathering (Mitchell et al., 2019(Mitchell et al., , 2021b. Fungal hyphae and their interactions with plant cells fall below the limit of resolution of the 3D imaging methods that we employed here, but it might be possible in the future to resolve structures of this scale using a correlative microscopy approach utilising nanotomography and FIB-SEM (e.g., Mitchell et al., 2021b). In the results presented here, the appressed nature of the liverwort thalli means sediment grains can be stabilised by the surface envelopment of thalli, entrapment between cells, and by mucopolysaccharide secretions (i.e., mucilage), in addition to rhizoid interactions. Some bryophytes have the ability to adhere to solid objects through the secretion from their rhizoids of a sticky and viscous sulphated mucopolysaccharide (Odu, 1989), while some liverworts (Haplomitrium), devoid of a thallus and instead having an erect structure, must rely on a subterranean axes and mucilage secretions for anchorage because of a lack of rhizoids. In Lunularia and Treubia, while most grains are entrapped by rhizoids where the thalli (midribs) touch the soil surface, we were able to visualise a smaller number of other grains that are held in place in other areas of the thalli, probably by mucilage secretions. This will vary between xyloglucans secreted from rhizoids for Lunularia and Blasia, and thalli mucilage for Treubia. Mucilages are carbohydrates that are involved in the absorption and retention of water. Most mosses and liverworts produce mucilage from slime papillae (Renzaglia et al., 2000); in rare cases, very large quantities of mucilage are produced from clefts in the thallus (Treubia) or by underground axes (Haplomitrium) (Carafa et al., 2003;Duckett et al., 2006;Renzaglia et al., 2007). Mucilage produced in this way, in addition to conferring a measure of protection against desiccation, could have contributed towards the aggregation of sediment grains in early soils. The basal thalloid liverworts Blasia and Lunularia secrete the polysaccharide xyloglucan which acts as an efficient soil particle aggregator at the modern day (Galloway et al., 2017), likely also making it an important method of stabilisation in the geologic past. Ligrone et al. (2012) inferred that the last common ancestor of present-day land plants was leafless, had an axial/erect growth form, and bore unicellular rhizoids and mucilage papillae. All these features have been observed in wellpreserved early fossil land plants, with the exception of mucilage papillae. Documentation of mucilage papillae or mucilage production within the tissue systems of fossils is challenging but could be envisaged under exceptional conditions of preservation. Papillate epidermal cells have been documented in several plants from the Rhynie chert (Edwards, 2004;Lyon and Edwards, 1991), and these might have had a secretory function. Thus, secretions and mucilage were likely an important method of substrate stabilisation in early land plants contributing to the formation of thin proto-soils.

| Landscape evolution with the evolution of land plants
The ways in which modern biological soil crusts and cryptogamic ground covers enhance resistance of surface run-off and erosion, while also promoting dust capture, are well understood (Belnap, 2003;Gao et al., 2016;Williams et al., 2012). However, unequivocally recognising structures of these types in the rock record (see Davies et al., 2020), quantifying their influence on rates of weathering, distinguishing between the effect of rhizoid-like and root-like anchorage systems, and elucidating different influence of plant body plan variations, can be challenging. In many instances, the likely effects of plant influence can be identified in otherwise unfossiliferous rocks, by comparison between strata dating to before the evolution of land plants (Davies et al., 2020), but evidence of 'smoking guns' is rare, often because even in situ plants are often parautochtonous and there is a paucity of records of small standing fossil plants in situ. Where palaeosols developed under larger vascular plants, weathering effects similar to those in modern soils can be recognised as well as evidence of rooting systems (Driese et al., 2021).
For smaller statured plants, one potential way forward is to identify and characterise vegetation-induced sedimentary structures (VISS) in the rock record, where indirect effects of sediment accumulation and erosion can also be inferred Rygel et al., 2004). Another approach is to identify micro-to-nanoscale biologically mediated weathering features (BWFs) that are present in potential proto-soil like substrates (Mitchell et al., 2019). The recognition of such features in the sedimentary record could provide an indicator of the presence of proto-soil forming communities, and lends value to studying and characterising these features in modern analogues at various scales.
When plants first colonised the land they had a profound influence on fluvial environments, recorded as a frequency distribution shift from sedimentary facies suggestive of a global preponderance of braided fluvial systems to a record containing greater evidence for meandering deposition (Davies & Gibling, 2010). An upsurge in alluvial mudrock also occurs, with explanations for this trend considering that mud production (from weathering) and retention (from binding, baffling and organically induced flocculation and settling) would have been important drivers (Davies & McMahon, 2021;McMahon & Davies, 2018;Zeichner et al., 2021;McMahon et al., submitted). However, it is less clear how different types of terrestrial communities influenced the scale of this weathering and landscape evolution, particularly by smaller statured plants (Edwards et al., 2015;Quirk et al., 2015).
Results presented here provide an indication that micro-scale processes can contribute to sediment stabilisation and soil formation in diminutive communities; such phenomena would have been wholly novel Earth surface processes in the early Palaeozoic  Puttick et al., 2018). This precedes the earliest fossil evidence from cryptospores, which first appeared about 480 Ma (Strother & Foster, 2021) and diversified through the Ordovician and Silurian Wellman & Strother, 2015). The earliest unequivocal plant macrofossil remains are documented at about 430 Ma (Libertín et al., 2018). The discrepancy in timing seen among calibrated phylogenetic trees, spores, and macrofossil has been widely discussed (Strother & Foster, 2021), and may in part reflect a taphonomic bias favouring the larger vascular plants (Kenrick et al., 2012). The growth forms of the earliest land plants therefore remain somewhat speculative. There is some evidence of thalloid communities associated with cyanobacterial mats in a braided fluvial system at 440 Ma (Tomescu et al., 2008;Tomescu & Rothwell, 2006). Direct evidence of the cryptospore-producing plants comes from Late Silurian and Early Devonian fossils preserved as charcoal (Edwards et al., 2014;Edwards, Morris, Axe, Taylor, et al., 2021). These are predominantly axial fossils with rarer associated thalloid remains. The plants were diminutive, with axes measuring less than 1 mm in diameter and height probably not exceeding a centimetre or two. It seems likely that rhizoid based systems preceded the evolution of simple roots and therefore that the earliest plants interacted with and stabilised substrates in a similar manner to modern liverworts in groups 1-3, including the entrapment, sticking and entwining of sedimentary grains within rhizoid bundles, grain aggregation due to xyloglucan, mucilage, and other secretions, and the biofilm-like encrusting of thalli or axial rhizomatous systems over surfaces creating a protective layer that reduced erosion. While rhizoids are known to have existed in early land plants in the geologic past, it seems likely that some may have been devoid of rhizoids and had subterranean axes, akin to modern Haplomitrium, and may have relied on secretions for their main form of stabilisation. Thalloid, lichen-like associations are also known from rocks of the Lower Devonian Lücking & Nelsen, 2018), further reinforcing the potential importance of encrusting growth forms. These plants and the associated organisms would have formed relatively thin proto-soils, mostly devoid of organic material with thin, surface organic layers (Mitchell et al., 2021a).
The earliest fossils with rooting systems that have been reconstructed in some detail come from the Early Devonian and these are either rhizoid-based systems in plants of axial growth form  or shallow root-like systems in small herbaceous lycopods (Hetherington & Dolan, 2018;Tomescu 2016, 2017). By the Middle Devonian, tree-like growth forms with wood and extensive rooting systems were evolving independently in several major clades of plants Morris et al., 2015;Stein et al., 2007Stein et al., , 2012Stein et al., , 2020Xu et al., 2017).
With the evolution of roots in the vascular plants the influence of vegetation on sedimentary systems starts to become apparent.
By 407 Ma, the average proportion of mudrock in alluvial successions globally is 15.4% of total stratigraphic thickness, compared with 1.3% in earlier units (McMahon & Davies, 2018). Additionally small meandering channels had become globally abundant (Gibling & Davies, 2012) likely not only initiated by more widespread land plants but also with an increased number of cohesive muddy substrates. Proto-soils would probably also have been more organicrich (Mitchell et al., 2021a), and more buried organic matter likely changed sediment properties leading to improved cohesiveness , with multicellular rhizoids able to stabilise and adhere to soil mineral grains, akin to group 4 in this study. With the evolution of vascular plants with 'true' root systems there was a marked shift in sedimentary architecture towards channelled sandbed rivers, meandering rivers, and muddy floodplains (the average proportion of mudrock in alluvial successions is 29.9% for formations deposited after the Early Devonian evolution of rooting; Gibling & Davies, 2012;McMahon & Davies, 2018).
The nature of early terrestrial communities therefore changed significantly from the Cambrian through to the Devonian, with the greatest changes happening in the plant morphology. Plants increased in size by several orders of magnitude and developed specialised organ systems including stems, leaves and roots. These changes would have influenced the ways that they interacted with their substrates and therefore their broader impacts on sedimentary systems. In terms of rock weathering, this might simply reflect a change in rate. However, it may be that a threshold in plant size, and the shifts in associated features (rhizoids, appressed vs upright, secretions), needed to be reached to influence the shape and flow of river systems.
F I G U R E 9 Summary diagram highlighting the above ground (gametophyte) and subterranean interactions with substrate mineral grains in groups 1-4 organisms.

| CON CLUS IONS
The organisms involved in proto-soil formation from the Cambrian through the Silurian were all small, ranging in size from fungal hyphae measuring micrometres in diameter to plants a few centimetres in size. The effective forces and mechanisms at play were proportional to the scale of the organisms. Based on our analysis of modern analogous systems, these include electrostatic forces, secretion of mucilage, entrapment of matter on the substrate by thalli and rhizoids, and the entrapment of grains by leaves. Together, these resulted in the development of thin, generally weakly structured proto-soils. The growth form of the plant also influences soil devel-

ACK N OWLED G M ENTS
Authors thank Dan Sykes for CT scanning assistance, Tomasz Goral for SEM assistance, and Alex Ball for general imaging preparation advice. Additional thanks go to William Rimington who provided excellent field assistance in New Zealand. We also wish to thank the Icelandic Institute of Natural History and New Zealand's Department of Conservation for assistance in acquiring sample permits.

FU N D I N G I N FO R M ATI O N
Funding for this work was provided by the Natural History Museum (London) Origins and Evolution Initiative. CS-D thanks the Foundation ARS Cuttoli-Paul Appell/Foundation de France for supporting her work on fossil fungi (grant no. 00103178).

CO N FLI C T O F I NTER E S T S TATEM ENT
There are no conflicts of interest from any of the authors.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.