Ontogeny of a subtidal point bar in the microtidal Venice Lagoon (Italy) revealed by three‐dimensional architectural analyses

Sedimentological and architectural features of meandering subtidal channels are relatively unexplored, and their deposits are commonly investigated based on facies models set up for intertidal meandering channels. The Venice Lagoon (northern Adriatic Sea, Italy) is affected by a micro‐tidal regime and hosts a dense network of active and buried tidal channels. It represents an excellent natural laboratory to improve the current knowledge on subtidal meander morphodynamics and related deposits. In this study, the integration of high‐resolution geophysical images and core data allows reconstruction of the architectural three‐dimensional model of a meandering subtidal palaeochannel, which is buried below a modern subtidal flat. The study palaeochannel was 35 m wide and 3 m deep, and formed three adjacent meander bends and related point bars. A detailed three‐dimensional architectural reconstruction was carried out for deposits associated with one of these meander bends, that was crossed by a minor, low‐sinuosity channel with two minor bank‐attached bars. This reconstruction highlights that the study point bar has a horseshoe shape, which arose from the onset of bar accretion from an already‐sinuous channel. Reconstructed growth stages of the studied bends show that point‐bar accretion can follow different styles of planform transformation, also experiencing simultaneously landward (or seaward) deposition according to the dominant flow direction (i.e. local tidal asymmetry). The analyses show that planform transformations occurred in parallel with elevation changes of the related channel thalweg, which shaped pools with geometry varying with the radius of curvature of the bend. The present study highlights the relevance of high‐resolution three‐dimensional reconstructions to link palaeomorphodynamic processes with related sedimentary products.

Knowledge gaps are even more relevant for subtidal channels, which have been investigated by few studies (e.g. La Croix & Dashtgard, 2015;Ghinassi et al., 2019b) despite their widespread presence in coastal environments [for example, the Geum River estuary (Korea) and Hammocks Beach State Park, North Carolina (USA)]. Subtidal meandering channels can be found in the most seaward portion of tidal basins (Rieu et al., 2005), where current velocities are higher because of the inlets, and they are permanently submerged. They usually develop on top of tidal flood deltas and their hydrodynamics is complex (Rieu et al., 2005;Ghinassi et al., 2019b) since the flow within the channels can be influenced by the currents flowing on the surrounding subtidal platforms and by the wind-induced wave winnowing (Carniello & Defina, 2012;Tommasini et al., 2019). Subtidal meanders can migrate laterally accumulating point bars that thereby do not experience any subaerial exposure. Therefore, defining the internal architecture and the sediment distribution within subtidal point-bar deposits is a critical step to deepen current knowledge on the genesis and evolution of subtidal meanders.
The present study aims at improving current understanding on the morphodynamic evolution and internal architecture of subtidal point-bar deposits, focusing on a Holocene tidal meandering channel cutting through a subtidal mudflat of the northern Venice Lagoon (Adriatic Sea, Italy; Fig. 1). With its complex network of tidal channels developed in a microtidal regime, the Venice Lagoon represents a key site to study facies distribution and architectural features of a microtidal system, where effects of tidal currents are amplified by meteorological forcings. By integrating geophysical and sedimentological data with morphological characteristics of a previouslydetected Holocene palaeochannel (Madricardo et al., 2007), this work reconstructs the detailed 3D architectural model of a subtidal meander and investigates: (i) the onset of tidal meanders and following planform transformations; and (ii) the related changes in bend morphometry and channel bathymetry.

The Venice Lagoon
Covering an area of about 550 km 2 , the Venice Lagoon is the largest brackish water body of the Mediterranean Sea; with its north-east/south-west oriented oblong and arched shape, the lagoon is 50 km long and on average 10 km wide (Zecchin et al., 2008), and exchanges water with the Adriatic Sea through the LidoÀSan Nicol o, Malamocco and Chioggia inlets (Fig. 1A). The lagoon is affected by a semidiurnal microtidal range, with a mean tidal range and maximum astronomical oscillation of 1.0 m and 0.75 m around mean sealevel (MSL), respectively (D'Alpaos & Carniello, 2013), which can be amplified by local meteorological conditions. Low (high) atmospheric pressure can produce considerably high (low) tides. Surge events strongly contribute to enhance the astronomical tides, especially those associated with the Sirocco wind (Mel & Carniello, 2019). The Venice Lagoon is part of the Venetian foreland basin (north-east Italy), which developed between the Apennine and the South-Alpine chains (Massari et al., 1986;Zecchin et al., 2008). Since the Early Pleistocene, the Venice area was characterized by a shallowing-upward depositional trend from deep-water to deltaic and shoreface sediments (Massari et al., 2004). These deposits were overlain by a cyclical alternation of continental and shallow-marine facies (Kent et al., 2002). The Venice Lagoon developed over the last 7500 years during Holocene transgression (McClennen & Housley, 2006;Zecchin et al., 2008;Rizzetto et al., 2009) and consists of tidal flats, salt marshes, and subtidal and intertidal channel deposits. Sediment ranges in grain size between mud and fine sand, although mud is largely dominant (Brivio et al., 2016;Ghinassi et al., 2018a;Ghinassi et al., 2019b). The channel network comprises tidal channels with different cross-sections, from almost 400 m to <1 m wide as the distance from the inlets rises up. Mean cross-sectional velocities within the channels can vary widely depending on the distance from the inlets and their channel width. At ca 6 km from the Lido inlet (Fig. 1B), the Gaggian channel and its two major tributaries (100 m and 30 to 40 m wide, respectively) show mean cross-sectional velocities of AE0.6 and AE0.4 m s À1 , respectively .
During the last centuries, human interventions contributed to modify the local tidal regime and morphologies within the lagoon (D'Alpaos, 2010;Ferrarin et al., 2015). The building of massive jetties at the inlets led to their deepening and the modification of the hydrodynamic regime. The diversion of all major rivers debouching into the northern part of the Lagoon dramatically decreased the fluvial sediment input (XV to XVIII century; Tosi et al., 2009;D'Alpaos, 2010;Bondesan & Furlanetto, 2012). Recently, the activation of the mobile gates (Project Mo.S.E.), designed to protect the lagoon settlements from extensive flooding, has changed the lagoon hydrodynamics (Mel et al., 2021) and, consequently, the sediment transport, contributing to the deepening of the tidal flats and reducing salt-marsh sedimentation (Tognin et al., 2021).

The study area
The study area is located in the northern part of the Venice Lagoon ( Fig. 1B) and it is relatively close to the present-day coastline (i.e. 4 km inland). The area is right behind the S. Erasmo island, which represented the littoral strip until the end of the XIX century (Bonardi & Canal, 1997;Balletti, 2006;Busato, 2007). Specifically, the study area, named Palude Santa Caterina, is sited between the Burano, Mazzorbo and San Francesco del Deserto islands (Fig. 1C); it is about 1.15 km 2 and shows water depths ranging between 3.0 m and 0.5 m. The area is currently crossed by the Canale San Francesco, a 35 m wide navigable channel that exhibits three meander bends with radii of curvature ranging between 210 m and 250 m (yellow lines in Fig. 1C).
Previous geophysical and sedimentological investigations (see Madricardo et al., 2007;Madricardo & Tegowski, 2012, for more details) reveal that in the study area the lagoonal Holocene succession is up to 8 m thick and covers a pedogenized alluvial substrate accumulated during the Last Glacial Maximum (Fig. 1D). The alluvial deposits consist of massive clay and fine sand layers that grade upward into overconsolidated silt, with pedogenetic nodules and roots. The lagoon deposits mainly consist of tidal-flat mud with isolated sandy to silty channelized bodies, which range in thickness between 0.5 m and 4.0 m.
Three buried palaeomeanders forming a northwest/south-east oriented belt (bends 1 to 3 in Fig. 1C) were documented in the study area (Madricardo et al., 2007(Madricardo et al., , 2012. These palaeomeanders were framed into a lagoonal depositional environment and associated with three sandy to silty, clinostratified point bars, which were buried by and cut into organic-rich lagoonal mud. Specifically, point-bar deposits consist of a shell-rich lag covered by an almost 3 m thick alternation of silty-sand and sandy-silt deposits. Shells belong to lagoonal faunas, including gastropods (for example, Bittium scabrum) and bivalves (for example, Acanthocardia tubercolata and Scrobicularia plana) (Cosma et al., 2019). The foraminiferal assemblage of bar deposits reveals that they accumulated in a subtidal setting of intermediate to inner-lagoon environment (Madricardo et al., 2007(Madricardo et al., , 2012, in the absence of significant riverine input. Bar-top deposits are muddy and do not show diagnostic sedimentary features of subaerial exposure in a salt marsh environment (for example, root traces and oxidation). Accordingly, their foraminiferal assemblage indicated a permanently submerged setting. Radiocarbon ages from channel lag shells date ca 4500 years BP (Madricardo et al., 2007(Madricardo et al., , 2012.

Geophysical data
The acoustic data were acquired using a single beam echosounder, modified to improve its resolution in extreme shallow waters (down to 0.5 m depth in some portions). The echosounder emits sound waves in short pulses (s = 0.15 ms) with a frequency fixed on 30 kHz, which propagate in the water down to the sea-bottom interface where the sound waves are partly reflected and partly penetrate the bottom sediments. The echosounder detects the intensity of the acoustic signals reflected from the interfaces between sediments with different properties. Despite generating a low-lateral single beam resolution in shallow water conditions, the transducer used for acquisitions provided useful data for qualitative studies thanks to its high vertical resolution, of the order of a decimetre (see Madricardo et al., 2007, for more technical details). A DGPS (differential global positioning system; Trimble DSM12 â ; Trimble Inc., Sunnyvale, CA, USA) was associated with the echosounder, to allow a positioning accuracy of the acoustic lines below 1 m. Line spacing was set to 2.5 m and, a total of 980 sections (i.e. about 500 km of acoustic profiles in 256-level greyscale) were acquired between 2003 and 2004, allowing for investigation of the upper 6 m of the Holocene lagoonal succession. In the frame of this work, 201 sections, 2.5 to 10 m apart from one another, have been selected to reconstruct a 3D model of the subsurface succession ( Fig. 2A). The Move 2018.2 TM software was used to visualize and correlate different sections in a georeferenced space. The main acoustic surfaces (for example, horizons) separating different depositional elements (for example, channel-fills, channel-bars and overbank deposits) were traced for each section. Three-dimensional surfaces were created with the Ordinary Kriging method, by interpolating categories of horizons representing the same depositional element. This approach allowed the reconstruction of different stages of channel evolution and related planform configurations. Meander morphometric properties were investigated using an automatic procedure based on the computation of the channel-axis signed curvature (see Marani et al., 2002;Finotello et al., 2020b, for a detailed description of the methodology). Inflection points of the channel centreline, defined as curvature zero-crossings, were employed to delimit individual meander bends and allowed to determine the relevant morphometric parameters. Specifically, this study derived measures of: (i) meander Cartesian length (Lxy), defined as the Euclidean distance between meander endpoints; (ii) meander intrinsic length (Ls), i.e. computed along the channel centreline; (iii) meander sinuosity, defined as the ratio between Ls and Lxy; (iv) meander radius of curvature (R) derived from the meander best-fitting circle (see Heo et al., 2009;Finotello et al., 2018); and (v) meander amplitude (A) defined as the maximum distance between the channel axis and the line through the two inflection points (Fig. 2B). Changes through time of the above parameters were analysed in relation to bend 2, since the available acoustic lines did not always allow to reconstruct the complete planform of bends 1 and 3. In addition to meander planform features, thalweg depths at different bar growth stages were also marked from the acoustic sections.

Core data
A total of 21 sediment cores was used to validate geophysical data and define the spatial distribution of sedimentary facies ( Fig. 2A). A hydraulic probe and a wire-line corer placed on a floating platform were used to recover 15 cores between 2004 and 2005 (Madricardo et al., 2007(Madricardo et al., , 2012. These cores are 10 cm in diameter and their depth spans from 8.0 to 9.1 m. The grain size was defined in situ, whereas lithology, sedimentary structures and Munsell colour were successively detected through laboratory analyses, where cores were sampled for micropalaeontological and radiocarbon analyses (see Madricardo et al., 2007Madricardo et al., , 2012. Six additional sedimentary cores were recovered in 2018 using an Eijkelkamp hand auger (Eijkelkamp Soil & Water, Giesbeek, The Netherlands), through a gouge sampler with a length of 1 m and a diameter of 3 cm. Coring sites were planned after preliminary results of the 3D model to analyse point-bar deposits. Coring depth spans from 4.0 to 5.8 m. Collected cores were kept humid in PVC liners and successively cut longitudinally, photographed and logged. Core logging was carried out following the basic principles of facies analysis (Walker, 1992), emphasizing sediment grain size, presence of sedimentary structures and bioturbation, vertical grain-size trends, and occurrence of plant debris and shell fragments. To verify the along-bend grain-size trend, grain-size analyses were carried out through a Mastersizer 2000 (Version 5.40, Malvern Instruments, Malvern, UK).
The terminology used in the present work is recalled in Fig. 2B and is analogous to that used for fluvial point bars and related deposits, although some modifications are required to adjust to the tidal environment (cf. Brivio et al., 2016).

Acoustic data and architectural elements
Interpretation of acoustic profiles allowed to identify distinct acoustic facies that can be linked to three main architectural elements: channel fill, channel bar and overbank. The acoustic reflectors are interpreted to represent primary accretion and boundary surfaces (cf. Skelly & Bristow, 2003;Sambrook Smith et al., 2006).

Channel bar
The channel bars are characterized by sets of packages of inclined reflectors, defined by a poorly detectable planar top and a well-marked concave-up base (Fig. 3A). They range in thickness between 2.0 m and 3.0 m starting from À1.0 m under the lagoon surface, and the inclined reflectors dip between 6°and 22°showing lateral extent up to 390 m. Channel bars show three main stratal geometries: (i) sets of packages with no internal dip changes, with cross-layered and/or oblique configuration (sensu Roksandic, 1978) Roksandic, 1978); and (iii) mounded package geometry with a concave-downward shape of the inclined reflections (Fig. 3C).   The clinostratified surfaces document processes of point-bar accretion related to lateral channel shift. The three stratal geometries are related to different orientations of the acoustic sections with respect to the main accretion/ channel migration direction. The first geometry shows the real dip angles of the inclined reflections, in cases where the direction of accretion remained parallel to that of the acoustic section. Changes in inclination angle of acoustic reflectors document variations of the direction of accretion relative to the acoustic section. Bar accretion parallel and transverse to acoustic section generated inclined (i.e. true dip angle) and sub-horizontal reflectors, respectively. Mounded reflectors occur in sections cutting through the whole point bar and highlight its positive topographic relief.

Channel fill
Channel fills show concave-upward reflectors (Fig. 3E) and range in thickness between 1.7 m and 3.0 m, with the top surface located at approximately 1.0 m below the ground. Dip angles vary from 5°to 20°. Channel-fill reflectors can show two main geometries: (i) the asymmetrical configuration (i.e. semi-elliptical trend), which is the more common one and is always found between sets of bar reflection and overbank deposits (Fig. 3D); and (ii) the symmetrical configuration (i.e. semi-cylindrical or concentric) that is found between sets of bar reflection dipping in opposite directions (Fig. 3E).
Concave-upward reflectors testify to the progressive infill of the abandoned channel. The two geometries result from the combination between the channel orientation and that of the acoustic sections. Asymmetrical shapes are indicative of sections that are slightly oblique to the axis of the abandoned channel, whereas symmetrical ones appear in sections that are transverse to the main channel. The occurrence of symmetrical shapes between adjacent point bars is a coincidence (Fig. 3E).

Overbank
Overbanks are characterized by parallel subhorizontal reflectors with moderate to good continuity (Fig. 3F). Localized lensoidal geometries with a basal concave upward surface can occur. These sub-horizontal reflectors represent tidal flat deposits and include localized minor channels that are too small to be imaged distinctly.

Three-dimensional reconstruction
Three-dimensional surfaces delimiting channel fills and channel bars were created to reconstruct the morphologies and orientation of the study sedimentary bodies (Fig. 4A). Minor channelized bodies (i.e. <2 m thick), though locally detected, were not modelled due to their discontinuous lateral documentation. Two channelized elements are included in the proposed 3D reconstruction and are named here as 'major' and 'minor' channels, respectively.
The major channel is that identified by Madricardo et al.   PB1 is located in the western portion (i.e. landward sector) of the study area, but it is only partially visible and its complete geometry cannot be characterized (Figs 4A and 5A). Bar axis trends approximately north-east/south-west and deposits forming the observed part of this bar are ca 132 000 m 3 . PB1 is on average 2.6 m thick, and its thickest (3.6 m) and thinnest (1.7 m) parts occur towards the innermost and outermost part of the bar, respectively (Fig. 5A). These changes in thickness are associated with a progressive upward-shift of the bar basal surface along the direction of bar accretion, with the channel becoming shallower through time (Fig. 5A). Inclined beds forming PB1 are organized into three major bedsets (Fig. 5B) and document an overall south-west accretion of at least 200 m.
PB2 is associated with a meander bend showing an amplitude of ca 400 m and a radius of curvature of about 225 m. The bend axis trends north-east/south-west. The cartesian length and the intrinsic length of meander bend 2 measure ca 360 m and 1065 m, respectively, defining a sinuosity of 2.9. Bar deposits occupy only a limited area within the meander loop and form a Ushaped body surrounding overbank tabular beds (Fig. 4A). PB2 is on average 3 m thick, and its thickest (3.9 m) and thinnest (2.2 m) parts occur in correspondence with its major depressions and ridges, respectively (Figs 5C and 6). As for PB1, these differences in thickness are associated with an overall progressive upward-shift of the bar basal surface (Fig. 5A). The volume of sediment stored in this bar is ca 316 000 m 3 . Inclined beds of this bar are organized to form four major bedsets (BS1 to BS4 in Fig. 5D), which create arcuate bodies with a variable lateral extent on different sides of the bar. BS1 and BS2 occur along the eastward side of the bar. BS1 covers only the eastward side of the bend, whereas BS2 occurs also in the bend apex zone but shows its maximum downdip extent along the eastward side. BS3 is sited along the western side of the bar, whereas BS4 surrounds almost the whole bar, reaching its maximum downdip extent in the bar apex zone (Fig. 5D). Reconstructed morphologies of the bar basal surface show the occurrence of alternating ridges and swales, which follow the overall curvature of the associated bar and related bedsets (Fig. 5C). The major depression defined by this surface occurs below BS2, whereas its maximum elevation is associated with BS4 (Fig. 6).
PB3 is located in the easternmost portion (i.e. seaward sector) of the study area and, as for PB1, it is only partially visible (Figs 4A and 5A). Its axis trends ca ENE-SSW. The volume of sediments forming the observed part of PB3 is ca 107 000 m 3 . PB3 is, on average, 3 m thick, and its thickest (3.5 m) and thinnest (2.0 m) parts occur towards the innermost and outermost parts of the bar, respectively (Fig. 5E). A progressive upward-shift of the bar basal surface along the direction of accretion occurs also in PB3 (Fig. 5E). The overall direction of bar accretion is towards the west, and produced two laterally offset bedsets (Fig. 5F).
The minor channel is slightly sinuous and trends north-west/south-east connecting meanders 1 and 3 of the major channel (Fig. 4B). The minor channel is 20 m wide and the channel fill ranges in depth between 1.7 m and 2.0 m. This channel contains two bank-attached bars, which are sited along its southern (BB1) and northern (BB2) flank, respectively (Fig. 4B). This bar typology can be found in low-sinuosity channels that do not migrate laterally enough to develop point bars, but accumulate sediment along the banks. BB1 and BB2 are characterized by north-dipping and southdipping internal clinoforms, respectively. BB1 extends for 300 m along the entire channel (Fig. 5G). The mean thickness is 2.2 m, and it decreases to <1 m moving northward (Fig. 5H). BB2 is the smallest bar of the study area, and is found at the confluence between the minor and major channels (Fig. 4A), extending laterally for ca 100 m. BB2 is 3 m thick and thins to 1.7 m thick southward, where it overlaps BB1. The reconstructed volumes of BB1 and BB2 are ca 27 000 m 3 and 6700 m 3 , respectively (Fig. 5H).

Core data
Sedimentary cores recovered in the area delimited by meander bend 2 allow depicting sedimentary features of PB2 and related channel-fill, along with those of the surrounding overbank deposits (Fig. 4A).
Channel-bar deposits are floored by a massive channel lag consisting of mud-free, fine sand characterized by a high content in shell fragments, with plant debris and sporadic mudclasts (Fig. 7A). Bar deposits are essentially made of silty-rich mud with subordinate 1 to 5 cm thick intervals of mud-rich very fine sand. Beds are thinly laminated with dominance of planeparallel stratification (Fig. 7C) and local evidence of ripple cross-lamination in the sandyrich layers (Fig. 7B). These deposits also appear locally deformed, possibly by slump processes (Fig. 7D). Scattered centimetric mudclasts are locally concentrated in sandy intervals (Fig. 7E). Plant debris is common and locally forms 1 to 5 mm thick layers (Fig. 7F). The upper part of bar deposits consists of silty-rich mud with a moderate content of plant debris (Fig. 7G) and burrows (Fig. 7H). Bar deposits are overlaid by massive tidal flat mud that is characterized by a high content of mollusc shells (Cerastoderma glaucum and Bittium reticulatum), and both bioturbations and plant debris are common (Fig. 7I).
Sedimentary cores recovered at different sites along the bar (Fig. 4A) highlight that the whole bar body does not show any relevant vertical grain-size change. Grain-size analyses from cores recovered from BS4 in the landward and seaward part of the bar do not reveal any relevant vertical or lateral (i.e. seaward or landward) change in grain size (i.e. cores SC1 to SC4 in Fig. 8).
Channel-fill deposits cover a basal silty lag and grade upward to structureless mud. Channel-fill deposits contain sporadic plant remains and mollusc shells (B. reticulatum and C. glaucum) (Fig. 9B). Overbank deposits are mostly made of structureless mud with a high content of mollusc shells (B. reticulatum and C. glaucum) (Fig. 9C) and plant debris. These deposits do not show any evidence of subaerial exposure (for example, oxidation and root traces).

Planform evolution
Spatial distribution, orientation and curvature of different bedsets detected within the three study point bars (Fig. 5) provide the inferences which are commonly derived from the investigation of bar-top scroll pattern to reconstruct planform evolution of fluvial point bars (Ielpi & Ghinassi, 2014;Durkin et al., 2015;Ielpi et al., 2018;Mason & Mohrig, 2019). Bedsets are considered here as the subsurface expression of bar-top scroll-sets . Integration between architectural and sedimentological data show that accretion of point bars started from a sinuous channel and continued through four distinct stages (stages 1 to 4 in Fig. 10). The early preserved establishment stage of evolution (T0 in Fig. 11), which defined the sinuous initial morphology of the channel and was not depositional, was characterized by Ls and Lxy of meander bend 2 of 440 m and 260 m, respectively, defining a sinuosity of 1.69 (black line, Fig. 10). The meander radius of curvature and amplitude measured 134 m and 126 m, respectively, and the thalweg maximum depth value was about 1 m (Fig. 11).
During the first stage (light grey lines in Fig. 10A), meander bend 1 migrated towards the south-west with a gentle downward shift of its thalweg (Fig. 5A) and allowed the expansional accretion (sensu Jackson, 1976a) of the related bar with the accumulation of its first bedset package (BS1 in Fig. 5B). Simultaneously, meander bend 2 shifted eastward allowing seaward accretion (cf. Ghinassi et al., 2018b) of PB2, with deposition of its first bedset, and erosion of the landward side of meander bend 3. During this stage, both Ls (ca 459 m) and Lxy (ca 270 m) of meander bend 2 slightly increased relative to the initial bend morphology (T1 in Fig. 11A). In contrast, the radius of curvature and sinuosity of the bend remained nearly constant during this  Fig. 11A). During the second stage (light blue lines in Fig. 10B), meander bend 1 changed its growth pattern, experiencing an overall SSE shift that allowed PB1 to accrete accordingly and generate its second bedset package (BS2 in Fig. 5B). This stage was characterized by a marked rising of the channel thalweg (Fig. 5B). Consistently, bend 2 experienced a seaward translation, associated with a north-east migration of the bar apex, that resulted in the deposition of its second bedset package. During this second stage, meander bend 2 modified its radius of curvature from 133 m to almost 173 m and, at the end of the stage, the Ls and Lxy of meander bend 2 increased to 724 m and 360 m, respectively, producing an overall sinuosity of 2.01. Also, the meander-bend amplitude increased, doubling to 280 m (T2 in Fig. 11A). During the accumulation of BS2, the channel reached a maximum depth of 4.92 m from the lagoon bottom (T2 in Fig. 11A) and the bar basal topography generated by thalweg shifts during this stage produced two major pool depressions close to the bend-apex zone (Fig. 6). Following meander bend 2, meander bend 3 experienced a seaward migration during this stage. Activation of the minor channel possibly occurred at this stage.
During the third depositional stage (grey lines in Fig. 10C) meander bend 1 experienced a minor SSE shift, with a gentle rising of the thalweg that generated its third bedset package (BS3 in Fig. 5B). Meander bend 2 experienced an asymmetrical growth characterized by an overall NNW landward shift and consequent accretion of the related bar through deposition of BS3. This process was associated with a partial erosion of the seaward side of bend 1. The meander bend 2 sinuosity remained almost unchanged during stage 3, but the radius of curvature increased from 173 to 199 m (T3 in Fig. 11A) defining a semi-circular and more symmetrical morphology of the bend. During this stage, the Ls and Lxy increased up to 827 m and 385 m, respectively, while the meander-bend amplitude reached a value of 306 m. The maximum thalweg depth slightly rose to 4.37 m (T3 in Fig. 11A), though the available data do not offer a clear identification of the pool zone (Fig. 6). Meander bend 3 was affected by an asymmetrical growth that caused a landward accretion of the related bar, resulting in a westward shift of the channel and the consequent erosion of PB2 deposits. This growth accumulated the first preserved bedset of PB3 (BS1 in Fig. 5F). During stage 3, the minor channel migrated towards the north-east accumulating most of the BB1 deposits (pink lines, Fig. 10C).
During stage 4 (blue lines, Fig. 10D), meander bend 1 continued to grow southward. Meander bend 2 was characterized by an asymmetrical growth towards the NNE, and consequently the final radius of curvature and sinuosity of meander bend 2 are equal to 224 m and 2.94 m, respectively. This migration leads to the increase in the Ls up to 1064 m, and the decrease in the cartesian length up to 361 m, whereas the meander amplitude reached the final value of 392 m (T4 in Fig. 11A). During this stage, the thalweg shift  shaped two distinct pool depressions (Fig. 6), which are separated by a ridge sited in the bend apex zone. This erosional surface floors the fourth bedset forming BD4, during which the maximum depth reached by the channel was 4.44 m from the lagoon bottom (T4 in Fig. 11A). In parallel, meander bend 3 shifted landward, leading to the accretion of PB3 and causing the erosion of PB2 deposits and accumulating the second preserved bedset package of PB3 (BS2 in Fig. 5F). The fourth depositional stage is also characterized by a minimal northward shift of the western portion of the minor channel, accreting the related deposits of BB1 (brown lines, Fig. 10D), and an abrupt growth towards the south-west of the eastern portion of the channel, which implies the deposition of BB2. This south-west shift was characterized by a marked rise of the thalweg and allowed overlap between BB1 and BB2.
Variations of the main morphometric parameters of meander bend 2 during the four morphodynamic stages are summarized in Fig. 11B, along with the changes of maximum thalweg depth. Although an increase in the intrinsic length of the bend clearly appears, bend evolution was also associated with a progressive increase in its cartesian length, sinuosity, amplitude and radius of curvature. The maximum thalweg depth increased by ca 3.5 m (from À1.0 to À4.75 m from the bottom) during the first accretionary stage (T0-T1 in Fig. 11), reaching the maximum depth during stage 2 (T1-T2 in Fig. 11). It decreased during stage 3 (from À4.92 to À4.37 m below the lagoon bottom) and remained almost constant during stage 4 (T3-T4 in Fig. 11).

Genesis and planform evolution of the bend
The overbank deposits nested within the horseshoe point bar allow to rule out that bar accretion started from an almost straight channel, but indicate that the track of the channel was quite already sinuous (black line in Fig. 10; T0 in Fig. 11) at time of deposition inception. Accretion of PB1, PB2 and PB3 started along the inner bank of a bend, allowing PB2 to grow preserving at its core a relic of overbank mud. A similar contrast between the external geometry of meanders and the architecture of related point bars has been highlighted for fluvial meanders (Russell et al., 2019) through the analysis of their complex scroll-bar patterns. Fluvial and tidal point bars are assumed to arise from a progressive increase in sinuosity of relatively straight channels (Leopold & Wolman, 1960;Daniel, 1971;Brice, 1974;Lewin, 1976;Nanson & Page, 1983;Knighton & Woodroffe, 1992;Hughes, 2012;Ghinassi et al., 2014), although the planform patterns of newly-formed channels are commonly irregular, being controlled by adaptation of the flow to local micro-reliefs of the newly-drained areas (Aslan & Blum, 1999;Taylor, 1999;Motta et al., 2012;Cassiani et al., 2020;Bellizia et al., 2021). Although adaptation of the Holocene lagoon mud to the underlying alluvial lowstand topography (Castiglioni & Favero, 1987;Amorosi et al., 2008) could have generated the micro-relief required to trigger wandering of a newly-formed channel, an avulsive relocation (Slingerland & Smith, 1998Mohrig et al., 2000) of the study channel is ruled out by the lack of a significant superelevation between sub-tidal channels and surrounding overbanks. Development of such a superelevation was hindered by removal of possible lev ee-deposits by the combined action of waves and tidal currents (Ghinassi et al., 2019b). The high density which typifies tidal networks (Marani et al., 2003;Passalacqua et al., 2013;Hoitink et al., 2017) promotes interactions between adjacent migrating bends, causing piracies and connection of adjacent branches (Cosma et al., 2020). This latter process allows the development of new creeks with a sinuous shape, which lead to a renewed circulation pattern over a pre-existing network. Where the newly-established formative discharge resulted increased by the network reorganization, an increase in the cross-sectional area of the creek (D'Alpaos et al., 2010) will occur.
Reconstructed planform evolution of the study channel reveals complex patterns, which are comparable to those characterizing fluvial meanders (Daniel, 1971;Brice, 1974;Jackson, 1976b;Ielpi & Ghinassi, 2014;Yan et al., 2017Yan et al., , 2019Willis & Sech, 2018b). Apart from confined meanders [for example, Beaver River, Wapiti River, Red Deer River and Fontas River (Canada)] that are almost forced to translate downstream as part of a continuum of planform evolution (Lewin & Brindle, 1977;Nicoll, 2008;Nicoll & Hickin, 2010), fluvial bends behave almost independently experiencing a combination of migration styles, according to local geomorphic and sedimentary constraints, which are mainly represented by floodplain heterogeneities (G€ uneralp & Rhoads, 2011;Bogoni & Putti, 2017). The study bends cut into a homogeneous, muddy overbank (Madricardo et al., 2007); this lack of overbank heterogeneities hindered the development of differential planform behaviours and promoted a uniform response of the study bends to the same forcings. The restricted range of discharge, and therefore of the flowing tidal prism, and the lack of any fluvial input did not prevent the style of planform evolution from changing in time. The alternation between phases of accretion along the seaward (for example, stage 2) and landward (for example, stage 3) side of the three adjacent bends could be related to the asymmetrical character of the tidal flow. Tambroni & Luchi (2017) demonstrated that point-bar patterns in tidal meandering channels form according to the local tidal asymmetry. Tidal asymmetries control the long-term net sediment transport in tidal environments (Hoitink & Hoekstra, 2003;Van Maren & Hoekstra, 2004), with ebb (flood) dominance leading to dominant accretion of the bar along the seaward (landward) side of the bend.
In the study case, phases dominated by flood currents (for example, stage 3) were associated with erosion on the seaward side of PB2 and accretion on its landward side. Conversely, phases of ebb dominance (i.e. stage 2) were associated with erosion on the landward side of PB2 and accretion on its seaward side (cf. Ghinassi et al., 2018b;Finotello et al., 2019). Alternation between landward and seaward shifts of the channel contributes to shaping the planform profile of the overbank relic deposits preserved within the bar (Figs 6 and 10). Recent studies on meanders of the Venice Lagoon Finotello et al., 2020b) corroborated the findings of changes in sedimentation patterns with the local tidal asymmetry, and also demonstrated that morphological adaptation of meander bends to changes in local tidal asymmetry occurs over relatively short timescales (i.e. decades) even in microtidal regimes. Changes in local tidal asymmetry within tidal channels can arise both from changes in the overall structure of the tidal channel network and from modifications of the overall basin morphology. In the former instance, network modifications can happen due to meander cut-offs (Brivio et al., 2016) or channel piracy events (Cosma et al., 2020;Finotello et al., 2020b). These processes lead to redistribution of the tidal prisms and adjustments of channel cross-section, with channel shallowing (for example, silting and infill) where the prism decreases and cross-section enlarging where the prism increases. Such dynamics are more likely within dense networks of small channels, like those cutting through salt marshes (Marani et al., 2003). Changes in local tidal asymmetry due to modifications of the basin morphology are more likely to affect larger tidal channels, especially if located in proximity of the inlets where tidal currents are more intense (Walton, 2002;Finotello et al., 2019). These modifications are related to the overall morphodynamic evolution of the basin, which depends on a number of features including, for example, the inlet position and morphology, the relative extent of subtidal versus intertidal areas, the depth of tidal flat surfaces, and the position of a given channel within the basin itself, as well as possible anthropogenic influences on morphodynamics (Townend, 2010;Ferrarin et al., 2015;Silvestri et al., 2018;Finotello et al., 2019;Guo et al., 2019). Although the lack of information concerning changes of the palaeo-drainage, the size of the study palaeochannel would suggest that it suffered from the effect of a basin reorganization following a change in the inlet position, especially considering its proximity to the ancient barrier.
During the last stage of bar growth, the reducing tidal asymmetry would be consistent with a more uniform accretion along the inner bank, which was also associated with the lack of relevant changes in grain size along the bar (Fig. 8) and the development of two major pool zones along the thalweg (Fig. 6). A similar pattern of grain-size distribution and basal bar morphology has been detected in other point bars of the Venice Lagoon (Brivio et al., 2016), and testifies that the seaward and landward sides of the bar alternatively experienced similar intensity of bed shear stresses over a tidal cycle, due to flood and ebb currents.

Channel morphology, thalweg trajectories and pools development
Changes of the main morphometric parameters of bend 2 during the four growth stages (Fig. 11A) are compared with changes in thalweg elevation (Fig. 11B). The overall descending, ascending and horizontal trends of the thalweg trajectory reflect adaptation of the channel to different growth stages and related flow configurations (cf. Cosma et al., 2019). The initial descending trend of the thalweg (i.e. stages 1 and 2) documents the channel adjustment to its newly-established formative discharge (Cosma et al., 2019). Deep scouring was also probably promoted by the development of a bend with the lowest radius of curvature, which enhanced the erosional effects of a secondary circulation in the bend apex zone (Finotello et al., 2020b). The subsequent gentle rise of the thalweg during stage 3 was probably related to the maintenance of its formative depth, which was reached at the end of stage 2. During this stage, sediments are accumulated in the channel to maintain a constant equilibrium depth in the frame of keeping dynamic equilibrium conditions (sensu Allen, 2000) with the surrounding aggradational tidal flats (Cosma et al., 2019(Cosma et al., , 2021. This process could have also been enhanced by the activation of the minor channel, which connected bends 1 and 3 causing a local drop in the strength of erosional forces within bend 2 (cf. Toonen & Kleinhans, 2012). The almost sub-horizontal trajectory of stage 4 indicates the achievement of a new tidal prism configuration, which allowed the channel to slightly increase its depth by keeping its thalweg at a constant depth during aggradation of the surrounding tidal flats.
During the different stages of evolution, meander bend 2 developed erosive pools that are identifiable on the basal surface of PB2 as major depressions. Although stage 1 pool is not clearly defined, stage 2 shows a deep pool zone that migrates south-east following the main seaward accretion of PB2 (Fig. 6). During stage 3, the pool zone moved northward and started to show two major depocentres, which became evident in stage 4 (Fig. 6). The switching from the one-pool configuration (stages 1 and 2) to the double-pool one (stage 4) can be explained by considering the increase in the radius of curvature in a tidal cyclicity scenario. The alternate turnover of the position of impingement zones is associated with curvature-induced secondary flows, rotating in opposite directions (Finotello et al., 2020b). A lower radius of curvature promotes the formation of a sharp bend where the impingement zones of flood and ebb flows against the outer bank almost coincide, causing a localized and intense scouring. In this frame, the more the radius increases, the fewer ebb and flood impingement zones against the outer bank coincide (Fig. 12). The increase in the radius of curvature caused the separation of the ebb and flood impingement zones, with the establishment of two separated zones affected by ebb and flood helical circulation, respectively (Finotello et al., 2020b). Similar findings have been recently documented in a meander bend of an intertidal channel of the Venice Lagoon, sited almost 3 km east from the study area (site SC in Finotello et al., 2020b), where flood and ebb flows impinge along channel banks downstream and upstream of the bend apex, respectively. This double configuration is similar to the one provided by box-shaped meanders (sensu Dalrymple et al., 2012), and is responsible for the formation of two distinct scour pools (Hughes, 2012;Finotello, 2017) separated by a localized high (Hughes, 2012;Brivio et al., 2016).

Aggradation of the minor channel
Acoustic profiles and relative 3D reconstruction also depict the stratigraphic relationship between adjacent bars in the minor channel. Here, BB2 overlaps BB1 below related channel-fill deposits (Fig. 5H), suggesting that the channel thalweg shifted southward while aggrading vertically. This vertical aggradation component of the channel trajectory (cf. Cosma et al., 2019Cosma et al., , 2020 allowed the preservation of the southern bar (BB1), which was not completely dismantled by the southward channel shift. This ascending trajectory of the channel thalweg highlights that the rate of thalweg rise was sufficiently high to impact the dynamics of lateral shift of the channel. Such a configuration could derive either from the channel tendency to preserve a constant equilibrium depth under intense aggradation of overbank areas (Cosma et al., 2019), or from a gradual decrease in tidal prism (Rieu et al., 2005). In both cases, the occurrence of rising channel trajectories critically increases bar connectivity, as demonstrated by numerical models (Willis & Tang, 2010) and outcrop studies  for fluvial meandering rivers.

Subtidal versus intertidal point bars from the Venice Lagoon: differences and similarities
Findings of this work contribute to improve the current knowledge about subtidal meanders, by shedding some light on their morphodynamic evolution and through a comparison with their intertidal counterparts. This comparison can be carried in the frame of a similar depositional setting, that is offered here by the backbarrier basin forming the Venice Lagoon. The study case, along with a further subtidal point bar described by Ghinassi et al. (2019b) in the southern lagoon, provides insights on the comparison with several intertidal counterparts which were recently described (Brivio et al., 2016;Ghinassi et al., 2018a). In the Venice Lagoon, although all of the intertidal channels are flooded at high tides, only the smaller ones (i.e. shallower than 60-70 cm) can be completely dry at low tides, whereas the deeper ones are still partially to largely flooded. On the contrary, subtidal channels are permanently submerged also at the lowest tides. If the size of channels is similar, the grain size of point bars does not strongly differ from subtidal (Ghinassi et al., 2019b) and intertidal settings , and channel-lag deposits seem to be ubiquitously characterized by abundance of shells and shell fragments (Brivio et al., 2016;Ghinassi et al., 2018a;Cosma et al., 2019). Similarly, although bar deposits are laminated, no clear sedimentary structures can be detected, neither in subtidal nor in intertidal settings (Brivio et al., 2016;Ghinassi et al., 2018b;Cosma et al., 2019). Plant-debris laminae are also widespread in both types of point bars (Brivio et al., 2016;Ghinassi et al., 2018b;Cosma et al., 2019;Ghinassi et al., 2019b). Differences occur essentially in the upper part of the bars, which show contrasting architecture and sedimentary features in subtidal and intertidal settings, respectively. The upper part of intertidal point bars is colonized by halophytic vegetation and consists of plane-parallel laminated mud (Brivio et al., 2016;Ghinassi et al., 2018b;Cosma et al., 2019). Mud is oxidized as a consequence of repeated episodes of subaerial exposure, and bears abundant root traces (Brivio et al., 2016;Ghinassi et al., 2018b;Cosma et al., 2019). Laminations are generally well-preserved, and commonly highlighted by the occurrence of millimetric sandy laminae, which have been commonly related to activity of wind-waves during major storm events (Brivio et al., 2016;Ghinassi et al., 2018b;Cosma et al., 2019).
Although geophysical data imaging of the transition between inclined bar slope beds and overlying sub-horizontal bar top layers are missing, this transition is associated with a gradual decrease of dip angle (Brivio et al., 2016), suggesting a sigmoidal (sensu Gobo & Ghinassi, 2014;Gobo & Ghinassi, 2015) bar top geometry (Cosma et al., 2019). Bar top deposits in the subtidal counterparts range in grain size from mud (this study) to veryfine sand (Ghinassi et al., 2019b). Despite this difference, bar-top deposits are always massive due to the intense bioturbation, and lack any evidence of oxidation. They appear generally grey to dark grey and contain abundant shells and shell fragments. Bivalves can occur in life position (Ghinassi et al., 2019b). Seismic data from the southern Venice Lagoon (Ghinassi et al., 2019b) show that subhorizontal bar-top beds abruptly cover inclined bar slope beds. Such an abrupt truncation was ascribed to the erosive effect of wind storm waves, which causes erosion in the bar-top area and also triggers collapses of the bar-slope deposits. Similar features were not imaged in the study site, mainly because of the low resolution of the acoustic data in the bar-top zone. The dominance of muddy deposits at the study site would be consistent with a lower efficiency of wave activity on the bar-top zone, due to the reduced wind fetch characterizing the study site during the major storm events associated with the north-east blowing Bora wind (Carniello & Defina, 2009;Carniello et al., 2012;D'Alpaos et al., 2013). Large waves affect the southern Venice Lagoon during these storms (Carniello et al., 2009(Carniello et al., , 2012D'Alpaos et al., 2013), and account for the occurrence of erosive truncations and sandy deposits in the bar-top deposits.

CONCLUSIONS
Acoustic and sedimentary-core data allowed reconstruction of depositional patterns of deposits associated with a subtidal channel in the microtidal Venice Lagoon (Italy). The study palaeochannel was 35 m wide and 3 m deep and formed three adjacent meander bends and related point bars. A detailed 3D architectural reconstruction was carried out for deposits associated with one of these meander bends, that was crossed by a minor, low-sinuosity channel with two minor bank-attached bars.
The major insights from this 3D sedimentological reconstruction can be summarized as follows: • The study point bar has an arcuate shape, which preserves at its core tidal-flat mud. Such a peculiar geometry was generated by the onset of bar accretion from an already sinuous channel. This process was probably triggered by the establishment of a new circulation pattern over a pre-existing network, possibly following a piracy event that connected adjacent tidal branches.
• Planform evolution of the study bend occurred through several stages that reflect asymmetrical growth of the point bars, as commonly observed in fluvial systems. This shows that, although the tidal currents work within a restricted and monotonous range of discharge, the style of bend planform evolution can change in time, possibly under the effect of variation in tidal asymmetry. Results show that under asymmetrical tidal flows, subtidal meanders develop depositional patterns according to the dominant flow direction.
• Planform transformations of the investigated bend occurred in parallel with changes in elevation of the related channel thalweg. These changes reflect the tendency of the channel to reach and maintain a specific equilibrium depth. Activation of a subordinate channel could modify the effective discharge of the major channel, influencing the vertical shift of its thalweg.
• Progressive increase in the radius of curvature of the study bend caused the splitting of the related pool scour. While the bend maintained a small radius of curvature, recirculating flows triggered by ebb end flood currents excavated the channel almost in the same area, generating a single pool depression. A progressive increase in meander-bend radius caused flood and ebb currents to create recirculation flows in different areas, with consequent development of two separated pool depressions.
• Bank-attached bars associated with the minor channel developed across the study bend migrated under aggradational conditions. This allowed overlap of these bars, which created a single and well-connected sedimentary body.

ACKNOWLEDGEMENTS
This work was sponsored by "HYDROSEM: Fluvial and tidal meanders of the Venetian-Po plain: from hydrodynamics to stratigraphy" project (Progetto di Eccellenza CARIPARO 2017, PI Massimiliano Ghinassi) and University of Padova (SID2016 project, titled "From channels to rock record: morphodynamic evolution of tidal meanders and related sedimentary products", PI Massimiliano Ghinassi). Authors are thankful to the "Echos" project supported by the Ministero delle Infrastrutture e dei Trasporti-Provveditorato Interregionale per le Opere Pubbliche del Veneto -Trentino Alto Adige -Friuli Venezia Giulia tramite il concessionario Consorzio Venezia Nuova, to provide part of sediment cores and the acoustic data. This manuscript benefits from the constructive comments from Reviewers R. Dalrymple and V. Zuchuat, and from the Associate Editor C. Bristow. Authors are grateful to the Journal Office Manager E. Richardson and the Chief Editor P. Plink-Bj€ orklund, for their editorial assistance. Open Access Funding provided by Universita degli Studi di Padova within the CRUI-CARE Agreement.