Introduction

Modern reefs adopt a wide array of morphological forms and develop in various bathymetrical zones, ranging from very shallow to mesophotic. Bioconstructions develo** in various settings and submitted to various biological, physical and chemical environmental factors show different responses, which is reflected in their shapes, sizes and internal structures. Geomorphology and internal structure of modern offshore areas, shelfs and reef complexes has been thoroughly studied and described in numerous areas of the world (e.g. Wood 1999; Woodroffe 2002; Hopley et al. 2007; Sherman et al. 2019). In the case of fossil reef deposits, the insight into a three-dimensional structure of a vast and internally complex bioconstruction is often hindered by limited exposures. For this reason among others, in the case of fossil reefs, especially those from the Palaeozoic era (541–251 Ma), particular sub-environments and ecosystems, such as shallow-water habitats and deeper mesophotic settings, have been generally described separately, without proper understanding and description of the connections between them. In this context, all exposures providing insight into the mutual relations between various reef sub-environments, which can be interpreted in terms of palaeoecology of particular groups of benthic fauna, their bathymetric distribution and environmental constrains, are of special value, as they can shed light on the overall structure of the whole framework composed of biostructures and adjacent areas of non-reef deposition.

The mid-Palaeozoic peak of reef development (444–375 Ma) resulted in strongly diversified reef ecosystems (Copper 2002). Silurian bioconstructions, particularly those dominated by stromatoporoids, are commonly described as level bottom communities and represented by biostromes devoid of a rigid framework (Kershaw 1984; Łuczyński 2020). Therefore, the term “reef” is used here in the broadest possible meaning (Wood 1999). One of the classical areas of Silurian reefs studies is the Swedish island of Gotland (Fig. 1), on which large, shallow-water reefs are known since the nineteenth century (Munthe 1902; Manten 1971). Excellent exposures, a large part of which are on long sea cliffs (Laufeld 1974), combined with a general lack of visible tectonic disturbances and of late diagenetic alternations, allowed thorough sedimentological and palaeontological studies (Calner et al. 2004a; Calner 2005). Over one hundred years of research has led to discoveries of a wide array of reef ecosystems (Zapalski and Berkowski 2019). The coral communities of the Lower Visby Beds (Fig. 2) have been identified as typically mesophotic (Zapalski and Berkowski 2019), and they represent the bottom part of a shallowing upward succession. The succession exposed in a beach cliff in Ygne, south of Visby town (Calner et al. 2004a, b), is dominated by coral- and stromatoporoid bearing facies, and in its upper part it is cut by a large erosional channel, which is the main focus of this study.

Fig. 1
figure 1

Distribution of the Silurian facies on the margin of Baltica 1—littoral and lagoonal facies, 2—barriers and shoals, 3—open-marine carbonate facies, 4—deep muddy shelf. Modified after Einasto et al. (1986)

Full size image
Fig. 2
figure 2

Location of the studied profile. a Location of the Rövar Liljas Häla pseudoraukar on a sketch geological map of northern Gotland. Based upon Calner et al. (2004a). b Beach cliffs north of Ygne. Arrows point to a small ravine behind the pseudoraukar, in which the described channel is exposed

Full size image

Channels, furrows and gutters of various dimensions, depths and shapes are common in modern reefs (Harris et al. 2005; Hopley et al. 2007). Most typically, however, they are developed in very shallow settings, while sets of spurs and grooves of a combined accretionary and erosional origin, or more widely spaced accretional deep buttresses are characteristic for the deeper parts of reef fronts (Sherman et al 2010, 2019; Duce et al. 2014, 2016). The shallow-water channels are connected with the main passages of recurrent rapid water movement driven by various factors causing its landward or seaward flow through the shallows that are colonised by benthic fauna forming more or less rigid structures. The most common factor regulating the circulation pattern are reversal tidal currents, but storm surges and associated backwash flows are also of significant importance (e.g. Goreau and Goreau 1959; Tucker and Wright 1990; Flügel 2010; Harris et al. 2005; Hopley et al. 2007). A number of Mesozoic and Cenozoic scleractinian reefs had developed spurs and grooves on their slopes (e.g. Esteban 1996; Leinfelder 2001).

The channel exposed in Ygne distinctly differs from shallow-water tidal channels, both modern and those described from other Silurian exposures of the stromatoporoid-coral successions of the Baltica shelf (e.g. Skompski et al. 2008). The differences relate to the shape and dimensions, but most importantly and profoundly are associated with the bathymetrical position of the phenomenon, reflected by facies development and benthic assemblages of the succession, in which the channel is developed. This is the first ever described Palaeozoic erosional channel originating in shallow waters that continues down deep enough to reach the mesophotic ecosystem. It contains material derived from hypothetical adjacent coeval shallow-water reefal settings, which indicates that it served as an interconnecting zone between shallow- and deep-water benthic communities.

A proper interpretation of the channel exposed in Ygne – of its position within the Silurian shelf with its numerous bioconstructions, of its bathymetry, and of its development and infilling by deposits, is crucial in order to understand the palaeogeography of the area and of the processes governing sedimentation and fauna distribution. It would also improve the consistency between absolute depth estimates of Silurian sedimentary environments based on benthic assemblages and on sedimentary structures (Brett et al. 1993). The present paper is an attempt of such an interpretation.

Geological setting

The age of the Silurian strata exposed on Gotland ranges from the Telychian (latest Llandovery, ~ 435 Ma) to the Ludfordian (late Ludlow, ~ 423 Ma) (Calner et al. 2004a; Cohen et al. 2013). They represent an erosional remnant of an extensive system of carbonate platforms of the Baltic Basin running through Gotland and Baltic States and extending south to western Ukraine (Fig. 1). This system was a part of a vast carbonate shelf belonging to a marginal sea that rimmed the Baltica continent from the south (Silurian orientation). Over its whole length of almost 2000 km, the belt of carbonate sedimentation had a generally constant facies pattern, in which the central position was occupied by stromatoporoid shoals and barriers (Einasto et al. 1986; Samtleben et al. 2000; Calner et al. 2004a; Łuczyński et al. 2009). The shoal areas separated open-marine basinal and slope environments from the back-reef and lagoonal areas. During the Silurian, the position of particular zones shifted, but the general pattern remained unchanged.

The present study focused on the beach cliff exposure at Ygne, often referred to as a Rövar Liljas Häla pseudoraukar (Munthe 1921), located at 57˚35ˊ37˝N, 18˚11ˊ28˝E (Fig. 2a). In a continuous section exposed are the Lower and Upper Visby Beds (Calner et al. 2004b), representing the oldest Silurian deposits on Gotland (Fig. 3). The Lower Visby Beds (latest Llandovery to Lower Wenlock, ~ 435–433 Ma; Calner et al. 2004a, b; Cohen et al. 2013) are developed as alternating limestone-marl beds with some scattered small bioherms and are interpreted as relatively deep environments (Samtleben et al. 1996) with mesophotic fauna (Zapalski and Berkowski 2019). The Upper Visby Beds (Lower Wenlock, ~ 433 Ma) represent somewhat shallower environments, however still within the mesophotic zone. The limestone-marl ratio increases upwards (Munnecke and Samtleben 1996; Munnecke 1997) and detritic limestones and reef mounds (Axelsro-type patch reefs—incipient bioconstructions built by tabulate corals and stromatoporoids on crinoid gravel; Watts and Riding 2000) start to appear (Calner et al. 2004b). The shallowing tendency continues in the overlying Högklint Formation, represented by stromatoporoid-coral facies with patch reefs and inter-reef facies. The deposition of the Visby Beds corresponds to the Ireviken Event; the oldest of the Silurian events (relatively brief periods of unstable oceanic and climatic conditions) that are recorded as major geochemical excursions combined with fauna extinction (Jeppsson 1990, 1993; Bickert et al. 1997; Munnecke et al. 2003).

Fig. 3
figure 3

General stratigraphy of the investigated section, and the position of the investigated erosional channel. Based upon Calner et al. (2004a)

Full size image

Exposures of the Visby Beds occur on beach cliffs along the north-eastern shores of Gotland at a distance of 70 km (Calner et al. 2004a, b), which, due to a lack of tectonic disturbances, reflects the distances in the Silurian. Between Ygne and Ireviken, at a distance of approximately 41 km, the exposures are almost continuous, and thanks to the occurrence of some characteristic horizons, they can be correlated almost on a bed-by-bed basis. This provides an insight into the lateral variability of the formation, especially in such aspects as the abundance and character of benthic fauna, and the occurrence of bioconstructions, such as bioherms (Berkowski and Zapalski 2018; Zapalski and Berkowski 2019).

Upper Visby Beds exposed in a small hidden valley behind a pseudoraukar (a lone-standing rock formed by mass movements) in the upper part of the cliff (Fig. 2b) host a large erosional channel, filled with detritic limestones and deeply incised within the Upper Visby Beds. The Visby Beds have been a subject of several investigations, including studies of facies development, sediment accumulation and erosion (Samtleben et al. 1996; Munnecke 1997; Calner et al. 2004a; Adomat et al. 2016), stratigraphy (Jeppsson 1997; Calner et al. 2004a, b) and faunistic content (Stel 1978; Berkowski and Zapalski 2018; Zapalski and Berkowski 2019; Zatoń et al. 2020).

Materials and methods

The shoreline of Gotland, together with its sea stacks; the raukars and beach cliffs, is under strict protection (Nature 2000 area), and therefore the use of hammers and direct collection of rocks and fossils is strictly prohibited. For this reason, the only collected rock samples were tiny fragments representing characteristic facies handpicked from weathered parts of the exposure or from the scree. Thin plates for microfacies analysis were prepared from these samples. A detailed bed-by-bed sedimentological profile of the Visby Beds and of the deposits infilling the channel has been drawn and was supplemented by standard facies and macrofauna description. Special attention was paid to the changes of the shapes of tabulate corals and stromatoporoids in a vertical profile, which were interpreted in terms of sedimentary environment and bathymetry. Morphology of tabulate corals was studied according to the methodology adopted by Zapalski et al. (2017) and Zapalski and Berkowski (2019). The stromatoporoid shapes were ascribed to particular shapes according to the parameterisation method introduced by Kershaw and Riding (1978) and improved by Kershaw (1984) and Łuczyński (2005, 2006, 2020). However, due to the impossibility of extracting the specimens from the rocks, only general morphometrical observations were performed for both groups, based mainly on the basal dimension/vertical height ratio. Thin plates and rock samples are housed at the Museum of the Faculty of Geology, University of Warsaw.

Results

Facies profile of the Visby Beds

The uppermost 2.5 m of the Lower Visby Beds (set I on Fig. 4), outcrop** above the present level of the beach, are developed as an alternation of argillaceous limestones (containing a considerable amount of clayey matter), detrital limestones, and marls with a nodular texture (Adomat et al. 2016; Fig. 5a). The marls distinctly prevail over the detrital beds. Benthic macrofauna is relatively rare (as compared with the upper parts of the succession) and consists mainly of high-domical stromatoporoids with envelo** latilaminae arrangements and platy tabulates (mostly heliolitids) loosely scattered in the sediments (Fig. 6). All specimens, irrespective of their shape profile, are in growth position. Stromatoporoids occur only in the uppermost one metre of the Lower Visby Beds and are absent in their lower parts also in other outcrops. Fragmented crinoids are common in the detrital beds, which microfacially are represented by bioclastic wackstones with fragmented and scattered rugose corals, crinoids, Beyrichidae ostracods and trilobites (Fig. 7a).

Fig. 4
figure 4

Detailed sedimentological section exposed on the Rövar Liljas Häla pseudoraukar

Full size image
Fig. 5
figure 5

Field photographs of the profile and the channel. a Lower Visby Beds; argillaceous limestone/marl alternation; set I. b Erosional bed of detrital limestones within the Upper Visby Beds; set III. c General view of the erosional channel. Arrow points to the place, in which the boundary bounds upwards to become almost vertical. d Enlarged area marked by the rectangle on picture c with the places of microfacies samplings. e, f Lowermost part of the channel. Arrows point to places, in which erosion of the underlying beds is visible

Full size image
Fig. 6
figure 6

Macrophotographs of fauna from the Visby Formation. a, b High-domical stromatoporoid with smooth surface and envelo** latilaminae, in situ in the wall (a) and its bottom surface with concentric growth bands (b); set II. c, d Platy tabulates (c) and laminar stromatoporoids (d); set IV. e Small vertical bioherm within the Upper Visby Beds; set IV. f Large high-domical stromatoporoid with envelo** latilaminae; set VI. g Redeposited massive tabulate; set VI. h Shell pavement (coquina); set VIII

Full size image
Fig. 7
figure 7

Photographs of thin plates. a Bioclastic wackstone; detrital beds within the Lower Visby Formation (set I). b Crinoidal-bryozoan packstone; detrital beds of the Upper Visby Formation (set IX). c Bioclastic/lithoclastic packstone; channel (sample 2 on Fig. 5d). d Wackstone lithoclasts and gastropods filled with wackstones; channel (sample 2 on Fig. 5d). e Wetheredella incrustations; channel (sample 6 on Fig. 5d). f Tabulate coral with bioclastic wackstone in interstitial spaces; channel (sample 3 on Fig. 5d). g Tabulate boundstone with lithoclasts; channel (sample 1 on Fig. 5d), h. Bioclastic packstone with Halysites and Favosites, channel (sample 4 on Fig. 5d)

Full size image

More abundant benthic fauna starts to occur directly below the horizon with Phaulactis angusta, marking the boundary between the Lower and Upper Visby Beds (Samtleben et al. 1996; Munnecke et al. 2003; Adomat et al. 2016). The lowest part of the Upper Visby Beds (sets II and IV on Fig. 4) is developed as an alternation of argillaceous/detrital limestones and marls, generally similar to those below, however with a different proportion between the two lithologies, with limestones dominating. The limestone beds are thinner, more detrital (including biodetritus; mainly reworked brachiopod shells) and less nodular. This succession is interrupted by laterally limited lenses of detrital beds of variable thickness and erosional lower boundaries (sets III and V on Figs. 4; 5b).

The limestone/marl alternations (sets II and IV) abound in tabulates (Halysites, Favosites, Heliolites) and stromatoporoids of various shapes, accompanied by solitary rugose corals. Platy tabulates and laminar stromatoporoids still occur and even form distinct horizons, in which they are arranged in piles, one over another (Fig. 6c, d). Stromatoporoids are represented mainly by domical forms with envelo** latilaminae arrangements (Fig. 6a, b). Small bioherms occur, built of platy tabulates and laminar stromatoporoids encrusting each other (Fig. 6e). The detrital beds (sets III and V) abound in redeposited and often fragmented fauna of platy and massive tabulates, stromatoporoids and single solitary rugose corals (Nannophyllum?).

The upper part of the profile, including the big erosional channel discussed here, is exposed on the eastern wall the Rövar Liljas Häla stack, in a small ravine (Fig. 2b). This part of the succession can be divided into two main parts. Lower part—set VI is developed as nodular, mainly marly beds with faint bedding. The overlying interval (sets VII to X) is represented by a package of argillaceous limestones and detrital beds.

The abundant tabulates and stromatoporoids occurring in set VI are both in situ and redeposited (Fig. 6g). They also adopt a wide array of shapes and dimensions and are accompanied by colonial and solitary rugose corals (Nannophyllum). Tabulates are massive, but also thin, platy and encrusting. Stromatoporoids are represented mainly by large (ø up to 15 cm) high-domical specimens with envelo** latilaminae (Fig. 6f). The uppermost part of the set VI yields flat and small domical stromatoporoids with mamelons and small domical heliolids.

Sets VII to X, in which the channel is incised, with the combined thickness accessible for studies of 2.5 m, are mainly composed of detrital limestone beds alternating with subordinate argillaceous marly layers (Fig. 4). The detrital beds are of variable thickness, with erosional bottom surfaces and often with gradational tops. Microfacially they are represented mainly by fine-grained crinoidal-bryozoan packstones with bioturbation channels filled by spar (Fig. 7b). Each such layer is packed by diverse fauna, often unsorted and occurring in lenses. Lower parts of some beds are developed as shell pavements (coquinas) composed mainly of brachiopods, crinoids (Fig. 6h), and subordinate debris of gastropods, rugose corals and bryozoans. The stromatoporoid and tabulate fauna occurs in a great variety of forms. In situ platy tabulates are common and reach large dimensions (ø exceeding 70 cm). Most commonly they occur in the upper parts of particular beds. Redeposited benthic fauna consists mainly of dendroidal colonial and solitary rugose corals (Nannophyllum), dendroidal tabulates, small massive tabulates (Favosites, Alveolites) and massive and branching heliolitids. They are accompanied by overturned extended domical stromatoporoids with envelo** latilaminae.

Channel morphology, facies and microfacies

The overall dimensions of the erosional channel are difficult to estimate. The structure is exposed in a cross-cut more or less perpendicular to its run (Fig. 5c). The visible height of the structure, up to the top of the rock, is around 5.5 m. The maximum width is around 14 m, but again, if the structure continued upwards, it may have attained even larger horizontal dimensions. The channel is incised as deep as set VI, and sets VII to X of the hosting succession can be traced only at its side. The lower erosional boundary (dashed lines on Fig. 5c-f) is sharp and distinctly cuts the underlying layers (Fig. 5f). On a distance of ca. 10 m it is close to horizontal and runs within set VI (Fig. 5e). The walls are steep, close to vertical (Fig. 5c).

The infilling of the channel consists of massive to faintly and irregularly bedded detrital limestones. The material is unsorted and strongly reworked. Fragmented bioclasts, which are most abundant in the lower part, are represented mainly by rugose corals, brachiopods, stromatoporoids (commonly with ragged margins), massive heliolitids and Favosites, with diameters not exceeding 10 cm; however, platy tabulates also occur, particularly in the bottom part. A variety of lithoclasts of various dimensions is also present.

Weathered fragments of rocks were handpicked from the bottommost part of the sediments infilling the channel (Fig. 5d). The microfacies represented by these samples are very diversified. Probably the most representative microfacies is a densely packed bioclastic/lithoclastic packstone with a debris of crinoids, tabulate fragments, bryozoans, gastropods, brachiopods, ostracods and trilobites (Fig. 7c; samples 2 and 6 on Fig. 5d). The lithoclasts are represented mainly by dark, fine-grained wackstones with scattered shell debris (Fig. 7d). The same material can be found also as infilling of gastropod shells (Fig. 7d). Some bioclasts are covered by Wetheredella incrustations (Fig. 7e; Jarochowska and Munnecke 2014). Larger bioclasts are commonly oriented vertically and the facies abounds in porous spaces (Fig. 7c). Other collected samples represent bioclasts present in the channel (samples 3, 5 on Fig. 5e), including tabulates (Fig. 7f) and stromatoporoids, but also lithoclasts (samples 1, 4 on Fig. 5e) represented by tabulate boundstones (Fig. 7g) and packstones (Fig. 7h).

Discussion

In spite of the fact that the mid-Palaeozoic was an interval of profound reef development throughout the world (Wood 1999; Copper 2002; Calner 2005) and that the Silurian and Devonian reefs have been a subject of long and extensive studies, the existence and importance of mesophotic ecosystems of that time has been recognised only recently (Zapalski et al. 2017; Zapalski and Berkowski 2019; Zatoń et al. 2020; Majchrzyk et al. 2022), and there is still much that is unknown. Shallow water and mesophotic ecosystem connectivity in modern environments has been intensively studied, and there is evidence demonstrating that mesophotic ecosystems are, to large extent, distinct from their shallow-water counterparts (Rocha et al. 2018; Stefanoudis et al. 2019). Lesser et al. (2018) in their review provided examples of diversified levels of connectivity between shallow and mesophotic reefs, depending on geography, local condition and taxa. As a result, vertical connectivity may strongly differ between species and their reproduction mode (brooding vs. broadcasting), as demonstrated by Bongaerts et al. 2017).

Mesophotic facies of the Visby Beds

The described succession of the Visby beds hosting a large erosional channel provides a unique opportunity to study the relation between shallow water and mesophotic ecosystems. The mesophotic environment is embodied by the Visby Beds, in which the channel is incised, while the shallow-water habitats are represented by the redeposited material infilling the channel (Fig. 8).

Fig. 8
figure 8

Palaeoenvironmental reconstruction presenting the position of the described erosional channel connecting mesophotic and shallow-water reef ecosystems

Full size image

The Visby Beds start the Silurian carbonate sequence of Gotland and represent the deepest facies of the succession (Calner et al. 2004a, b). The detailed lithological and facies development of the outcrop** beds presented here (Fig. 4) generally corresponds to the previous descriptions of the formation (Samtleben et al. 1996; Munnecke 1997) and illustrates a gradual increase over time of the influence of high-energetic factors on deposition.

The occurrence of platy, low profile shapes of tabulate colonies was the main argument that allowed Zapalski and Berkowski (2019) to position the sedimentation of the argillaceous limestones and marls of the Lower Visby Beds (set I) within the mesophotic zone, thus describing the oldest MCE’s (mesophotic coral ecosystems) in the world yet found. Potential photosensitivity of stromatoporoids is still a subject of debate (Łuczyński 2009; Kershaw et al. 2018); however, their exclusion from deep-water settings in this profile (no stromatoporoids in the lowermost part of the outcrop** beds) can be used as an argument here. Envelo** latilaminae with concentric growth bands visible on the bottom surfaces (Fig. 6b), combined with high-domical shape profiles of unturned specimens point to a stable, low energy depositional environment with a low deposition rate (Łuczyński 2020), which is in accordance with the postulated mesophotic depths. It also advocates against a soft and muddy character of the sea bottom, which could potentially provide an alternative explanation for platy tabulates. In many cases they exhibit epizoans on their lower surfaces (Zatoń et al. 2020), thus attesting that these platy colonies were projecting into the water column and their lower surfaces were raised over the substratum (see Zapalski and Berkowski 2019). Moreover, no encrusting and intergrowth of skeletons one on another has been observed (in the cases of both tabulates and stromatoporoids), which is a common adaptation to muddy bottoms (Kershaw et al. 2006).

Earliest signs of depositional environment punctuation by high-energy sedimentary events can be found already at the Phaulactis angusta horizon marking the boundary between the Lower and Upper Visby Formation (Adomat et al. 2016). Above this level, lenses of detrital beds (sets III and V) interrupting the limestone-marl succession (sets II-VI), are filled by biodetritus and show erosional lower boundaries. A more energetic character of deposition is reflected by the benthic faunal assemblage, which becomes more abundant and variable in terms of taxa, morphologies and taphonomy, both in the detrital lenses and in the succession itself. Typical mesophotic forms found in situ, such as platy tabulates, cooccur with redeposited corals and stromatoporoids of various shapes, sizes and states of preservation. The nodular marly beds of set VI most probably represent a somewhat calmer episode of deposition (especially in its upper part), but even here redeposited benthic fauna is present. Predominance of high domical forms with envelo** latilaminae and with mamelons (small protrusions on the surface) among the stromatoporoids in both growth and redeposited positions point to an environment generally favourable for stromatoporoid growth in terms of low deposition rate and high nutrients supply, but exposed to occasional high-energy events causing exhumation and transport (Łuczynski et al. 2014; Łuczyński 2020, 2022).

The proportion of the calm conditions of deposition and benthic fauna growth in a mesophotic setting, and the high-energetic sedimentary events interrupting this environment, distinctly changed in the upper part of succession (sets VII-X). This part is predominated by beds of detrital limestones of various thickness and with erosional bottom surfaces. Each such layer represents a distinct depositional episode, connected with a high-energetic event. The abundant and diverse benthic fauna of these beds, often fragmented, unsorted and occurring in lenses, represents the original growth habitats, from which it has been redeposited. Probably, these settings were mostly relatively calm and favourable for growth, as can be inferred from the diversity of benthic fauna, as well as from their branching (corals) and massive high profile (stromatoporoids) forms. They must have been repeatedly punctuated by high-energy episodes, which caused redeposition and sedimentation of detrital beds in somewhat deeper settings. Redeposition from shallow-water habitats is reflected also by the occurrence of green algae indicative for the shallow photic zone in the detrital beds and their absence in the micritic beds. The occurrence of large platy in situ tabulates in the upper parts of particular beds probably represents colonisation of the sea bottom after such episodes. Their platy shapes, usually attributed to mesophotic depth intervals (Zapalski and Berkowski 2019), in this case were probably associated with high water turbidity and large amounts of fine sediments suspended in the bottom parts of the water column, which hindered light penetration (Zapalski et al. 2021). It is therefore difficult to access the absolute depth of such a setting. On the other hand, platy tabulates occur also in the subordinate argillaceous marly layers, in which they are accompanied by large high profile stromatoporoids with envelo** latilaminae. These layers represent calm episodes of deposition between high-energetic events, which suggests that the whole succession was deposited in mesophotic depth intervals.

The most suggestive and obvious interpretation of the growing influence of high-energy sedimentary events within the succession of the Visby Beds exposed in Ygne would be its gradual shallowing (Samtleben et al. 1996; Munnecke 1997; Calner et al. 2004a). The same may be inferred from the gradual appearance of Axelsro-type patch reefs in the Upper Visby Beds, followed by their abundance in the overlying Högklint Formation. This may partly be the case, however it is contradicted by the occurrence of in situ platy tabulates accompanied by extended domical stromatoporoids with envelo** latilaminae in the upper part of the succession, which points to growth in calm waters of mesophotic depths. All faunal elements characteristic for more shallow-water settings occur as redeposited and often fragmented components of detrital beds. Therefore, this intensification of turbulence and redeposition should probably be attributed instead to the growing proximity of the steep edges of shallow-water shoal areas than simply to eustatic changes. The whole Silurian succession of Gotland is composed of several stacked platform generations, separated by stratigraphic discontinuities, each evolving from ramps to rimmed shelves with steep edges (Bjerkéus and Erikson 2001; Calner et al. 2004a). The enhanced reef development resulted in progradation of stromatoporoid-coral complexes and shoal areas, which gradually became situated closer and closer to the area exposed in Ygne, and which is reflected by the growing occurrence of detrital beds with redeposited shallow-water fauna within the Visby Beds. At the onset of deposition of the overlying Högklint Formation, this progradation resulted in the replacement of deeper mesophotic environments by a reef complex composed of patch reefs and inter-reef areas (Riding and Watts 1991; Watts and Riding 2000; Calner et al. 2004a).

Erosional channel and its significance

The large erosional channel exposed on the Rövar Liljas Häla pseudoraukar in Ygne is incised in the succession of the Upper Visby Formation (sets VI-X), which (as indicated above) represents relatively deep water mesophotic environments located in the vicinity of a slope of a shallow-water reef complex. The recurrent erosional slope processes are usually very destructive, and therefore the chances of their preservation in the fossil record are very low. Only those structures that reached the foot of such slope can be preserved, which makes the described channel so special and unique. On the other hand, in modern reefs, erosional structures cutting the slopes of shallow-water complexes and reaching to mesophotic depths are not common. The reef fronts below the shelf eak are commonly covered by spurs and grooves, and the erosional channels tend to form on steeper reef fronts (30–70˚) and are accompanied by deep buttresses, which represent prime habitats of well-developed mesophotic coral ecosystems (Sherman et al. 2019). This may suggest that the described channel has developed on a particularly steep section of the slope and could explain why similar structures are not found regularly in other corresponding settings. While we have only one evident erosional channel, it seems likely that in other parts of the Visby Beds there are other similar channels (e.g. north of Visby, Jan Król, personal communication, 30 August 2022) that may be somewhat diachronic.

The channel is intraformational and developed simultaneously with the deposition of the sediments of the Upper Visby Formation. In terms of development of deposition in the sedimentary basin, they represent the same stage. It was active for a considerable time, which enabled its infilling by a thick pile of sediments. The deposits within the channel are indistinctly layered (Fig. 5d, e, f), which indicates that their sedimentation took place in several phases, probably associated with high-energy sedimentary events, such as storms. Unfortunately, the uppermost part of the channel is not preserved and therefore its contact with the overlying deposits cannot be observed. The erosion and formation of the channel took place in the substrate that already underwent considerable lithification, which is indicated by the steepness of its walls (Fig. 5c). Also the redeposited infilling the structure must have been at least partly lithified and cemented, as can be deduced from the occurrence of sharp-edged lithoclasts (Fig. 7d) and of bioclasts with primary pores occluded by cements (Fig. 7h). While the erosional channel obviously post-dates the adjacent mesophotic beds, the time between their deposition and the erosional processes cannot be easily estimated. Numerous studies show that the lithification of micritic muds is very rapid (e.g. Flügel 2010); therefore, the time gap between lithification and erosion did not have to be very long. On the other hand, precision of conodont stratigraphy does not allow any reliable estimation of that time.

Comparison with possible modern analogues and final conclusions

Direct comparison with modern structures, especially with gutters cutting through rigid coral reefs, can be misleading, due to a different ecology of Mid-Palaeozoic benthic communities commonly described as level bottom communities and represented by biostromes devoid of a rigid framework (Kershaw 1981, 1984; Łuczyński 2020; Zatoń et al. 2020; Claussen et al. 2022). Erosion and development of profound incisions in such sediments must have been easier. Also the exhumation and redeposition of benthic fauna (mainly tabulate corals and stromatoporoids) took place in a different manner than today, most probably specimen by specimen, as they mainly grew in a solitary manner and were not incorporated into a rigid framework (Kershaw 1990; Kershaw and Keeling 1994). The character of the processes that caused the formation of the channel and later triggered the consecutive acts of its infilling by sediments derived from shallow-water areas is difficult to reconstruct. Most probably the structure must have been spatially connected with the systems of back and forth flow of water that developed in the shallow-water areas, such as seaward extensions of tidal channels, and possibly acted as a continuation of one of them onto the slope and down to deeper water settings.

The connectivity of modern shallow-water reefs and the adjacent mesophotic ecosystems results in a significant overlap between these communities (Laverick et al. 2018). This is, in case of corals, mostly a result of larvae migration. Vertical connectivity, both downwards and upwards, strongly relies on larval abilities to migrate, and taxa common both in shallow and mesophotic reefs can be both brooders and broadcasters, but they differ in the settlement dynamics (Holstein et al. 2016). The channel described in the present paper most probably played an important role in connectivity between shallow and mesophotic ecosystems. The redeposited material infilling the channel includes lithoclasts with a variety of microfacies characteristic of Silurian shallow-water reef environments, such as bioclastic wackstones and packstones (Fig. 7a, b, c, d, h) and tabulate boundstones (Fig. 7g). The lithoclasts are accompanied by reworked fauna of various dimensions and states of preservation, including branching rugose and tabulate corals, and stromatoporoids with ragged margins, which are typical for settings with high and changeable deposition rates (Kershaw 1984; Łuczyński 2020), such as shallow-water shoals and barriers. We can state that the downward transport, as evidenced by sedimentological features, obviously must have included also living organisms, as well as their larvae. The larvae during transport may have got into suspension and settled (e.g. on the channel rims), thus causing the enrichment of biodiversity. This ensured the ecosystem connectivity between the shallow and mesophotic reefs. In recent reefs, there is commonly a significant community overlap between shallow-water and mesophotic environments, but mesophotic communities are not homogenous and distinct biodiversity clusters often occur (Semmler et al. 2017). On the other hand, mesophotic communities are often characterised by unique taxa unknown from shallow-water communities (Lesser et al. 2009), and they are not homogenous through their vertical range. We may speculate that similar distribution occurred here, as the coral faunas underwent typical palaeoecological succession from quiet water stage to rough water stage. The taxonomic composition of tabulate corals in Hogklint Formation became much more diversified (Stel 1978). As direct sampling is prohibited on the Gotland shores, we could not investigate this interesting issue.

The Axelsro-type, high-relief patch reefs first occur in the Upper Visby Formation, roughly in the same position in the profiles as the channel described here (Riding and Watts 1991; Watts and Riding 2000; Calner et al. 2004a). This may be for a reason and may indicate that downslope redeposition of sessile fauna may have enhanced their development, as many of these patch reefs show erosional lower surfaces (Watts and Riding 2000). The transported material included larvae, but also fragments of still alive colonies that might give rise to their establishment on the channels upper surface and further dispersal in the nearest surroundings (Fig. 8). Downward transport of living coral fragments towards more muddy and deeper surroundings is an important mechanism of colonization in modern peri-reefal environments. Living fragments are transported into depths and when they successfully start growing they form further substrate for larvae, thus expanding colonisation (Hughes 1999). Probably this was the case also in the described structure, in which indistinctive bedding that is visible in its lower part, faints upwards.

One of the significant differences between shallower and deeper settings is different shapes and dimensions of tabulates and stromatoporoids (Zapalski et al. 2017; Zapalski and Berkowski 2019). In fact this was an important means of inhabiting the environments located in mesophotic depth intervals. This is especially important, when taking into account that the Visby Beds are the oldest example of a fossil mesophotic ecosystem described so far (Zapalski and Berkowski 2019). The example presented here probably illustrates the process of its establishment. If true, this indicates that the development of the earliest mesophotic ecosystems should not be attributed solely to eustatic sea level changes and/or gradual expansion of corals and stromatoporoids into greater depths, but likely took place also by means of redistribution of shallow-water taxa during high-energy sedimentary events. This can be treated as a working hypothesis, and potential further research could be directed towards its further testing.