How a river submerges into the sea: a geological record of changing a fluvial to a marine paleoenvironment during early Holocene sea level rise

Coastal seas, and in particular estuarine systems, were significantly affected by Quaternary sea level changes. Furthermore, the dynamics of shelf and coastal evolution have had a strong impact on coastal landscapes inhabited by humans. The postglacial evolution of the vast North Sea shelf with its huge drainage systems, e.g. the Elbe Paleovalley and its tributary system, is an excellent research target to understand how coastal shelf environments change in response to sea level rise. In this study, we investigate infill sediments of the Paleo‐Ems valley – a drowned extension of the modern Ems River and part of the Elbe Paleovalley drainage system. Radiocarbon‐dated transgression sequences provide several new observations regarding the mode and rate of the river system submerging due to the Holocene transgression. Thus, the Paleo‐Ems valley submerged within a short time span of~200 years since the river was not able to adjust its gradient to the rapid rising sea level. The fate of the Paleo‐Ems is exemplary for the rapid change of a former fluvial landscape into a coastal landscape and finally into a submarine seascape. © 2019 The Authors. Journal of Quaternary Science Published by John Wiley & Sons Ltd.


Introduction
Changes in sea level driven by climate variability control the dynamic development of coastal seas, which has had fundamental impacts on human settling strategies in coastal regions, especially in the youngest geological history (Nicholls and Cazenave, 2010;Harff et al., 2016). Since the Last Glacial Maximum (LGM), the paleolandscape, which now forms the North Sea seafloor, has been drastically reshaped by glacial ice sheet coverage and periglacial processes and finally drowned by the postglacial sea level rise (Vink et al., 2007;Peeters and Cohen, 2014;Phillips et al., 2017). The rising sea level converted wide parts of coastal lowlands, partly known as Doggerland ( Fig. 1; Ehlers, 2011), to shelf and coastal seas. Especially the Elbe Paleovalley (EPV) played a significant role in the postglacial North Sea setting. The EPV is a large valley structure, which drained vast coastal lowlands throughout the Weichselian until it was drowned by the advancing North Sea during the Holocene transgression (Hallik, 1962;Figge, 1980;Coles, 2000;Clarke, 2009;Özmaral, 2017).
Early studies on the development of the EPV since the LGM (Figge, 1980;Streif et al., 1983) have been recently extended by a comprehensive study about the postglacial transgression into the EPV using sediment echosounder and high-resolution seismic records as well as sediment cores (Özmaral, 2017). Recent studies indicate that the EPV might have been part of a vast ice-dammed lake, which developed during the LGM (Hjelstuen et al., 2017;Özmaral, 2017;Phillips et al., 2018). At the end of the last glaciation, the EPV drained meltwater from the east, from the Fennoscandian Ice Shield along its southern margin, and the German low mountain range in the south across the periglacial paleolandscape towards the North Sea (Ehlers, 2011;Özmaral, 2017). Özmaral (2017) demonstrated that pre-transgressional sediment deposits within the EPV were almost eroded, with the exception of infill sediments from a channel network, which veined the valley plain from south to north. With rising sea level, the EPV was subject to at least three infill phases, in response to different sea level highstands and changes in the current regime (Özmaral, 2017). Based on relative sea level index points from the German Bight and adjacent areas as well as from Northwest German coastal areas, Vink et al. (2007) demonstrated that the postglacial paleolandscape in this area started to submerge at about 10 600 cal a BP and the entire landscape including the EPV was largely drowned at about 4400 cal a BP.
During the LGM sea level lowstand, the EPV was fed by various larger and smaller rivers as indicated by the paleovalleys of the modern Weser, Ems, and Eider Rivers ( Fig. 1) as well as by now-lost rivers or channel networks which drained the Dogger Bank (Warnke et al., 2014;Hepp et al., 2017). The Elbe and Weser Rivers are thought to be the main feeders of the EPV (Figge, 1980). The Elbe River reflects the onshore part of the EPV system (Ehlers, 2011). The Weser River is a relict of the chronologically older Weser Paleovalley, which had drained the meltwaters from the Elsterian and Saalian ice shields toward the west and since the Middle Saalian, towards the EPV (Meyer, 1983). Unfortunately, due to massive Holocene sediment remobilisation in the Elbe and Weser estuaries, a direct linkage of the Weser Paleovalley to the EPV cannot be confirmed.
Based on geophysical records, Tietze (1983) identified a 2-3-km-wide paleo river valley, which discharged Weichselian meltwaters from the Jutland Peninsula toward the eastern flank of the EPV. Landward, this valley structure can be linked to the Weichselian and present-day Eider River (Fig. 1). Along the western flank of the EPV, two tributary systems are known: the Entenschnabel river system in the northern part of the EPV, and the Paleo-Ems in the southern part (Hepp et al., 2012;Warnke et al., 2014;Hepp et al., 2017;Fig. 1). The Entenschnabel river system drained the Dogger Bank to the southeast. The system was influenced by postglacial sea level rise and developed from a single deep valley, which may initially have originated during the last glacial, to a network of shallower rivers and tributaries (Hepp et al., 2017). The morphology of this network is similar to river structures known from the western part of the Dogger Bank (Fitch et al., 2005;Phillips et al., 2018). The Paleo-Ems is the drowned extension of the modern Ems River ( Fig. 2(a)) that discharged into the southern part of the EPV during late Weichselian and the early Holocene (Warnke et al., 2014), together with the Elbe, Weser and Eider Rivers. In addition to these five documented tributaries of the EPV, it is expected that many more fluvial systems drained the late glacial and postglacial Doggerland during sea level lowstands.
While the pathways of some tributaries of the EPV are well documented in the geophysical record, only a little is known about the impact of the Holocene sea level rise on the depositional regime of the tributaries, particularly the rates of siltation and salinisation and not least the time which was needed to inundate a tributary valley to become a shallow shelf environment. The impact of the Holocene transgression Figure 1. Synthetic map of the present-day German Bight in the North Sea with a land-sea distribution reflecting the situation at~11 000 cal a BP with the paleocoastline being based on the -50 m bathymetric contour line (modified from Warnke et al., 2014). Present day coastlines are indicated by the thin dark lines. The supposed extent of large valley structures, like the EPV, is marked in light blue. The EPV and its tributaries (Elbe, Weser, Eider and Ems Rivers) were draining the then exposed coastal lowlands also known as Doggerland (Gaffney et al., 2009). The Entenschnabel river system south of the eastern Dogger Bank is marked in red. The red square marks the study area along the lower reaches of the Ems paleovalley (see on the depositional regime of the Paleo-Ems valley and its interconnection to the EPV is an ideal case study to understand how coastal lowland rivers respond to Holocene sea level rise. Based on sediment cores taken from the formerly mapped Paleo-Ems valley (Hepp et al., 2012;Warnke et al., 2014;Hepp et al., 2017) and the EPV (Özmaral, 2017), we discuss the topographical and chronological relationship between the developments of the Paleo-Ems tributary, the EPV, and the surrounding landscape. Radiocarbon ages coupled with grain size distribution, organic carbon content and micropaleontological records from Core GeoB17721-1 provide insights into (i) the timing and duration of the inundation of the Paleo-Ems, into (ii) changes in the transport energy regime during the Holocene transgression as well as into (iii) changes in the paleoenvironmental setting shifting from fluvial to marine conditions, with the latter being reflected by changes in the diatom assemblage over time providing an insight on grade and rate of salinisation in the estuarine environment. Since these processes are not only phenomena of the past, this study also aims to contribute to the understanding of the implications of sea level rise on modern coastal lowland rivers. For example, the modern Ems River is presently affected by enhanced mud aggradation induced by sea level rise, changes in tidal dynamics, and human interference in the river system. The installation of tide locks and intensive dredging are presently the only temporary successful measures to prevent the estuarine river system from siltation caused by a disequilibrium of ebb and flood tide currents, channel depth decrease and upstream shift of the channel mouth (de Jonge et al., 2014).

Database and methods
This study mainly focuses on the 5.8-m-long sediment core GeoB17721-1 (located at 54°03.031' N, 06°58.585' E), which was recovered from the infill of the Paleo-Ems valley near its confluence with the EPV (Fig. 1 and Fig. 2(a)). Grain size distributions, organic carbon content and diatom assemblages were used to reveal the sedimentary and environmental response of the Paleo-Ems estuary to deglacial sea level rise. Six additional sediment cores (cores 09-1 to 09-3, and 09-5 to 09-7) were taken further upstream in the river valley ( Fig. 2(b)) and the sedimentology documented. From all the cores, radiocarbon ages obtained on basal peat sequences of the infill of the Paleo-Ems valley were used to specify the timing and development of the submergence of the Paleo-Ems and the chronology of the transgression phase of the EPV (Table 1). In order to connect the results from the Paleo-Ems with the infill of the valley incision at the base of the EPV, we used the radiocarbon-dated sediment record from core GeoB17718-3 provided by Özmaral (2017).

Sediment samples
Sediment core GeoB17721-1 was obtained in a water depth of 30.5 m, during RV Heincke expedition HE405 in 2013 using a Geo-Corer 6000 Vibrocoring System with 28 Hz vibromotor and a barrel length of 5.80 m. On board, the core was split, photographed, and the lithological and sedimentary characteristics were documented. Additionally, samples were taken from basal peat intervals from sediment cores 09-x, located approximately 8.5-20.4 km southwest of core GeoB17721-1 and in the upstream direction of the Ems paleovalley ( Fig. 2(b)). The exact coordinates cannot be given here due to commercial restriction. These cores were recovered in 2009 from water depths between 28.7 and 31.1 m. All cores provide sample material from the infill of the Paleo-Ems valley (see Hepp et al., 2017).

Grain size analyses
A total of 25 samples from core GeoB17721-1 with a sampling space of about 20 cm were analysed for their grain size distribution using a Beckman Coulter Laser Diffraction Particle Size Analyzer LS 13320. The measurements were taken at the Particle Size Laboratory at MARUM, University of Bremen. Prior to the measurements, the terrigenous sediment fractions were isolated by removing organic carbon, calcium carbonate, and biogenic opal. The obtained results provide the particle size distribution of a sample from 0.04 to 2000 μm divided into 116 size classes and is given in volume percentage (vol%). The average standard deviation integrated over all size classes is better than ± 4 vol%. The grain size frequencies were processed using the GRADISTAT 8.0 software to obtain particle size summary parameters (Blott and Pye, 2001).

Organic carbon and sulfur contents
Carbon and sulfur content was determined using the automated C-S analyser Multi EA 4000 (Analytik Jena AG). Fiftyone samples from core GeoB17721-1 were freeze-dried and pulverised. To obtain the total organic carbon (TOC) content, 10% HCl was added to 300 mg of sample material to digest the total inorganic carbon. The analysis of the decalcified material in the C-S analyser provided TOC and total sulfur (TS) content in weight percentage (wt%).

Radiocarbon ages and age conversion
Hand-picked plant macrofossils (peat or plant remains) or marine calcareous fossils (mainly foraminifera and shell valves or their fragments) from 22 samples, as well as one wellpreserved but possibly reworked shell valve were dated using accelerator mass spectrometry 14 C dating. Radiocarbon ages of samples from core GeoB17721-1 were determined at the Poznań Radiocarbon Laboratory (Poznań, Poland), whereas radiocarbon ages from peat layers or layers of decayed organic matter in cores 09-x were determined at the W. M. Keck Carbon Cycle Accelerator Mass Spectrometry Laboratory (University of California, Irvine) (see Table 1 for details). All radiocarbon dates were converted to calendar ages using the CALIB 7.0 program (Stuiver and Reimer, 1993; http://calib. org). The reservoir effect of the marine samples was corrected with ΔR values provided by Scourse et al. (2012) for the German Bight (mean ΔR = 83), with the restriction that the ΔR values were determined on post-industrial samples. For comparison, the external age dates provided by Özmaral (2017; see GeoB17718-3) and Vink et al. (2007) used in this study were recalibrated accordingly. All calibrated radiocarbon ages are given as cal a BP. The median of the probability distribution of the σ2 range is used as a reliable estimation of the sample's calendar age.

Diatom assemblages
The study of the diatom assemblage was carried out on 15 samples, spaced at every 20 cm of core GeoB17721-1. Samples were prepared according to the standard randomly distributed microfossil method of Schrader and Gersonde (1978). Qualitative and quantitative analyses were carried out on permanent slides of acid-cleaned material (Naphrax® mounting medium) at 1000x magnification using a Zeiss® Axioscop (MARUM, University Bremen, Germany). We used the counting methodology of valves proposed by Schrader and Table 1. Accelerator mass spectrometry (AMS) radiocarbon dates and calibrated ages of sites used in this study. Radiocarbon dates of GeoB17718-3 were recalibrated from dates provided by Özmaral (2017  , the peat and shell layers were not analysed; (d) total organic carbon content (TOC, red line) and total sulfur content (TS, black line); (e) relative diatom abundance (%) of three main groups: into freshwater, brackish and coastal planktonic (marine) species, as well as total diatom concentration (valves/gr; red line); (f) calibrated radiocarbon ages (see Table 1). The colours refer to the type of the dated material (yellow = marine calcareous fossils or shells, red = plant remains, black = peat); (g) calculated linear sedimentation rates (LSR  Gersonde (1978). Several traverses across each cover slip were examined, depending on diatom abundances (between 250 and 400 valves were counted for each cover slip). Two cover slips per sample were scanned in this way.

Core description and grain size analysis
Sediment core GeoB17721-1 penetrated the base of the Ems paleovalley and comprises the whole infill of the valley and the overlying sediments. Based on colour, lithological contacts, sediment composition, and the grain size distributions, the core can be divided into six units ( Fig. 3 and Fig. 4). In the following, the units are described from bottom to top: Unit I (541-479 cm) is mainly composed of over-consolidated dark greyish brown medium sand (~78 vol%) with some fine sands (~18 vol%). The mean median grain size (D 50 ) is 991 µm. The sediment is moderately sorted, devoid of any shell fragments and contains tiny particles of dark organic matter. The skewness is fine skewed and the kurtosis mesokurtic (Fig. 4). The upper contact to Unit II is erosive.
Unit II (479-398 cm) is a very dark brown peat interval. The peat is quite stiff and shows internal stratification. Larger plant fragments were not observed, but plant fibres, seeds and pollen are well preserved. This unit was not sampled for particle size measurements. The upper contact to Unit III is erosive.
Unit III (398-146 cm) is composed of very dark greenishgrey, poorly-sorted clayey silt (average total silt content: 78 vol %; average total clay content: 20 vol%; mean D 50 = 24 µm). The sediment is stratified with very thin layers of dark organic matter derived from plant and root fragments. A shift from fine to coarse silt at about 250 cm core depth coincides with a prominent change in linear sedimentation rates (LSR; see section 3.3 below) and divides subunits IIIa and IIIb. For subunit IIIa the mean D 50 is 20 µm, the skewness is symmetrical to fine skewed and the kurtosis mesokurtic (Fig. 4). For Unit IIIb the mean D 50 is 29 µm, the skewness is fine skewed and the kurtosis platykurtic. The upper contact to Unit IV is erosive.
Unit IV (146-40 cm) is composed of dark greenish-grey, poorly-sorted silt to fine sands with shell fragments (average total coarse silt content: 49 vol%; average total fine sand content: 48 vol%; mean D 50 = 189 µm), with a slight coarsening upward. The skewness is fine skewed and the kurtosis mesokurtic to platykurtic (Fig. 4). The upper contact to Unit V is erosive.
Unit V (40-30 cm) describes an agglomeration of wellpreserved shell valves and shell fragments (<2 cm large) of the families Cardiidae and Myidae. Both are coastal species typical of the modern North Sea. The erosive lower contact and the formation as a thick layer suggests a deposition of reworked shells during a single high-energy event or over a short time period. The upper contact to Unit VI is gradational.
Unit VI (30-0 cm) is mainly composed of poorly-sorted fine sand (av. 70 vol%) to medium sand, interspersed with shell fragments. The mean D 50 is 369 µm, the skewness is symmetrical and the kurtosis is very leptokurtic (Fig. 4).
The additional cores 09-x were first presented by Hepp et al. (2017). They show a similar lithologic stratigraphy to GeoB17721-1 (Fig. 6): Medium sands are overlain by peat intervals and the transitions of the peat interval are erosive. In contrast to GeoB17721-1, cores 09-5 and 09-7 show additional peat layers, which are intercalated with fine sandy silts or silty fine sands. The peat sequence is often overlain by clayey silts (cores 09-1, 09-3, and 09-6), again overlain by fine sandy silts. The top of the cores is composed of fine sands (cores 09-2, 09-3, 09-6, and 09-7). Core 09-5 shows a similar shell layer to that described for Unit V in core GeoB17721-1.

Organic carbon and sulfur contents
The TOC content of core GeoB17721-1 is consistent with the established lithological units (Fig. 3(d)). Very high TOC content of 35-45 wt% characterises the peat (Unit II). In the clayey silts (Unit III), the TOC content decreases within the first ten centimetres above Unit II to a constant level of 3.3-5.0 wt%. However, the TOC seems to be more constant in Unit IIIb than in Unit IIIa. The lowest TOC content of 0.1-1.0 wt% was observed in sandy units (Unit I, IV, and VI) making up the base and the top of the core.
The TS content exhibits a similar pattern as the TOC content; however, with much lower values. Unit II has a TS content of 5.7-8.7 wt% and shows an upward-increasing trend, reaching a maximum of 12.8 wt% at the transition to Unit III. In the clayey silts (Unit III) the TS content varies between 2.8-5.9 wt %, whereas in the sands (Unit I, IV, and VI) the TS content is always <0.8 wt%.
The TOC/TS ratio is very high with values of 4.2-8.0 in the peat section (Unit II) and ranges between 0.7 and 1.3 in the silty clay section (Unit III), and most parts of the sand sections (Units I, IV and VI). Only in the uppermost 30 cm, the youngest sediments of Unit VI above the shell layer, the ratio increases up to 2.0-2.3. The TOC-TS scatter plot shows clear clusters for each unit (Fig. 5). The TOC/TS ratios of the sand and silty sand sections (Units I, IV, VI) fall mainly within the range of normal marine siliciclastic sediments (TOC/TS = 2.8 ± 0.8) as proposed by Morse and Berner (1995) for weight ratios of organic carbon to pyrite sulfur. The TOC/TS ratios of the silty clay sections (Unit III) point to more suboxic conditions. However, these finding have to be handled with some caution, because determining factors like rates of aerobic and anaerobic metabolism in the sediment or burial rates have not been considered so far. Unit VI Unit IV Unit IIIb Unit IIIa Unit II Unit I Figure 5. Relationship between total sulfur (TS) content and total organic carbon (TOC) content (according to Berner and Raiswell, 1984) for samples studied from core GeoB17721-1. The colors correspond to Units I-IV and VI (see text). [Color figure can be viewed at wileyonlinelibrary.com]

Radiocarbon ages and linear sedimentation rates
Eight radiocarbon ages obtained on core GeoB17721-1 provide information on the timing of valley infill and the deposition of the overlying sediments (Table 1 and Fig. 3(f)). Bases and tops of the peat section (Unit II), the clayey silt section (Unit III) and the overlying sands (Units IV-VI) were selected for radiocarbon dating. According to the resulting ages, peat growth started at about 11 440 cal a BP (477 cm core depth), which is equivalent to the beginning of the early Holocene. The youngest age obtained from the uppermost part of the peat section is 9690 cal a BP (403 cm core depth). Due to the assumed hiatus at the transition to Unit II, it is not possible to determine the exact termination of this section.
The clayey silt section (Unit III) started at about 9470 cal a BP (381 cm core depth). The youngest age obtained for this unit is 9290 cal a BP (151 cm core depth). Hence, the sediments of Unit III were deposited quickly with~1260 cm ka -1 over a very short time span of about 180 years. An additional dating of 9410 cal a BP (261 cm core depth) might point to a prominent drop in LSR within Unit III with~2000 cm ka -1 in Unit IIIa and 894 cm ka -1 in Unit IIIb. However, the short time span between the oldest and youngest ages of Unit III is subject to some uncertainty with respect to the error ranges of the dating (Table 1). Nevertheless, a shift in the grain size distribution at the same depth (Fig. 3(c)) implies that a change in the sedimentary regime probably also results in a shift in LSR. The remarkably high age of 19 990 cal a BP determined on plant remains found at 141.5 cm core depth does not fit the age model based on the remaining seven radiocarbon ages. The sample was taken directly above the prominent hiatus at the transition from Unit III to Unit IV (Fig. 3) and, thus, might be affected by reworked material. Consequently, this dating has not been considered further.
The sand Unit IV spans from somewhat after 9290 cal a BP (considering the hiatus) to 7480 cal a BP (61 cm core depth), . Lithological columns (a) from the Paleo-Ems valley with radiocarbon ages from peat intervals (cores 09-1 to 09-3, 09-5 to 09-7 and GeoB17721-1), and (b) from an incision at the base of the EPV with radiocarbon age from plant fragments (core GeoB17718-3; see Özmaral, 2017). All core depths are adjusted to the 1-m-bathymetric dataset provided by the project "Geopotenzial Deutsche Nordsee" (https://www.gpdn.de). Radiocarbon ages are given in which results in a minimum estimate for the LSR of 50 cm ka -1 . Based on the radiocarbon ages, the agglomeration of wellpreserved valves and shell fragments (Unit V) was deposited roughly between 7480 cal a BP and 7330 cal a BP (30 cm core depth). Assuming the modern sea floor was completely recovered in the sediment core, we calculate a deposition rate of about 4 cm ka -1 for the uppermost Unit VI. In addition, basal and intercalated peat layers as well as layers of highly decayed organic matter from the additional six cores 09-x were sampled for radiocarbon age dating (Table 1 and Fig. 6(a)). The radiocarbon ages at the base of the basal peat layers range between 11 700 cal a BP and 11 020 cal a BP, matching the beginning of peat growth documented for core GeoB17721-1. The youngest radiocarbon ages of the basal and intercalated peat sequences range between 10 050 cal a BP and 9350 cal a BP, also matching the data obtained on core GeoB17721-2.

Diatom content and composition of the assemblage
The total diatom concentration in core GeoB17721-1 ranges between 4.8 * 10 5 and 8.0 * 10 6 valves g -1 (average = 2.5 * 10 6 ± 1.7 * 10 7 valves g -1 ). Values are highest between 9690 and 9470 cal a BP and strongly decrease upward in the core (Fig. 3(e), Table 2). The diatom preservation is generally good. Valves show no significant enlargement of the areolae or evidence for any dissolution of the valve margin. This suggests rapid deposition and/or little reworking. In addition to marine diatoms, other siliceous components are silicoflagellates, radiolarians and land-derived freshwater diatoms or phytoliths (silica bodies of epidermic grass cells). In terms of number of individuals, diatoms dominate the siliceous fraction throughout core GeoB17721-1; their concentration is always one to four orders of magnitude higher than that of the abovementioned siliceous organisms encountered (not shown here).
The diatom assemblage is diverse and contains around 90 species with the vast majority being purely marine or brackish/intertidal, while 18 of them are fully freshwater diatoms (Table 3). The marine diatom community is dominated by coastal planktonic and brackish species. Coastal planktonic diatoms thrive in neritic, oligo-to-mesotrophic waters with high to moderate dissolved silica levels, becoming more abundant during intervals of weak turbulence and responding mainly to moderate upwelling (Romero and Armand, 2010;Crosta et al., 2012). Main components of the brackish group mostly live attached to a substratum as epipelic, epiphytic or epizoic diatoms. The main brackish groups are Nitzschia angulata, Diploneis smithii, Melosira nummuloides, Delphineis surirella and several species of Cymatosiraceae (Round et al., 1990). The freshwater community is dominated by species of Aulacoseira and Cocconeis, and Navicula cincta and Gyrosigma scalproides (Hartley, 1996).
In general, two main intervals are recognised in the speciesspecific assemblage of diatoms; between 9690 and 9470 cal a BP (beginning of Unit IIIa; Fig. 3(e) and (h)), freshwater and brackish water diatoms dominate (common contribution always exceeds 55%), whereas coastal planktonic components increase after 9470 cal a BP. This increase matches the strong decrease in the total diatom concentration (Fig. 3(e)). According to grain size, TOC and LSR data, the differentiation between Units IIIa and IIIb can also be observed in the rise in the relative abundance of coastal planktonic diatoms, more obvious in Unit IIIb than in Unit IIIa. The relative abundance of coastal planktonic diatoms increases steadily towards the top of the core and reaches its first maximum at about 210 cm core depth, which corresponds to an age of about 9380 cal a BP; based on the assumption of a constant sedimentation rate within Unit IIIb.

Discussion
Age relationship between the Ems and Elbe paleovalleys with respect to models of relative sea level rise The core transect investigated covers the entire lower~20 km of the Ems paleovalley (Fig. 1). The 09-x cores show a lithological valley infill succession (Hepp et al., 2017) comparable to GeoB17721-1 (Fig. 6(a)): pre-river sandsbasal peat formationtransgressional clays and silts with intercalated peatssands with shell layers (possibly tempestites). Thus, basal peats or layers rich in organic matter were found in all cores investigated here. Radiocarbon ages obtained at the base of these sections define the beginning of peat growth between 11 700 and 11 020 cal a BP (Table 1 and Fig. 6). The youngest radiocarbon ages date between 9800 and 9350 cal a BP and define the termination of peat formation or the beginning of the transgressional units. Differences from core to core in unit thicknesses as well as the occurrence of intercalated peat layers ( Fig. 6(a)) are most likely a consequence of different geomorphological settings along the valley, e.g. on the valley centre, valley flanks, point bars, and cut banks ( Fig. 2(b)). Cores 09-5 and 09-7 exhibit intercalated peat sequences, which might be explained by common regional dislocations of the river bed in a meandering river setting (Hepp et al., 2017). The chronological relation of the Paleo-Ems valley and the EPV can be assessed with radiocarbon ages obtained from the stratigraphic oldest known infill sequences of both valleys. Özmaral (2017) provides one radiocarbon age obtained from the stratigraphically oldest known infill sequences at the base of the EPV. This infill is composed of alternating layers of dark greyish brown medium sand and an intercalated agglomeration of well-preserved and fragmented marine shells (Cardiidae) overlain by very dark greenish grey clayey silt, which coarsens upward to clayey fine sands with common shell, plant and wood fragments (Özmaral, 2017). The radiocarbon age was determined on plant remains taken from core GeoB17718-3 (Table 1 and Fig. 6(b)) and with 10 770 cal a BP it corresponds to the age range of the peats from the Ems paleovalley (Fig. 1). The corresponding age ranges could imply that the timing of peat growth in the Paleo-Ems valley and the drowning of the EPV are linked to the same marine flooding phase. However, the difference in elevation of more than 5 m between the bases of the EPV and the Paleo-Ems River valleys ( Fig. 6 and Fig. 7) seems to contradict this assumption. A comparison with the hydrogeological map provided by the Lower Saxony Soil Information System NIBIS® (https://nibis. lbeg.de/cardomap3/) shows that height gradients of about 5 m for the sea-level-coupled groundwater in the coastal area onshore over a distance of few kilometres indeed occur, although these are mostly linked to impermeable strata (clays or clayey soils) below the groundwater. Although the sediments below the peat unit in the Paleo-Ems valley are composed of overconsolidated sands, clayey sediments found along the western flank of the EPV (Özmaral, 2017) could have the potential to impound the ground water. Furthermore, the greater distance of~20 km between the Paleo-Ems valley and the EPV sites gives support to a sea level groundwater effect affecting both sites at the same time despite the observed 5 m height difference. An alternative scenario could be that the peat in or at the flanks of the Paleo-Ems valley developed on a local scale and the peat growth was independent from relative sea-level-coupled groundwater. Sea level independent postglacial peat bogs are a well described phenomenon in the southern North Sea (Jelgersma, 1961). A local paludification can occur when the surface of the Pleistocene subsoil runs almost horizontally and groundwater accumulates in shallow depressions (Lange and Menke, 1967). However, the very good spatio-temporal fit of the inundation of the peats in the Paleo-Ems valley to the regional sea level record (see below and Fig. 7) points to the link with sea level rise and also to the onset of local peat growth.
The radiocarbon ages from the Paleo-Ems valley infill can be integrated into the relative sea level model  Fig. 4) complemented by radiocarbon ages resulting from this study. Colour-coded filled circles mark basal peat index points from three distinct areas in the southern North Sea (including the related linear regression lines) as indicated in the legend (Vink et al., 2007): the 'EPV (Northern areas)' covers the area between the western and the eastern flanks of the EPV south of the Dogger Bank, the 'EPV (Southern areas)' is mainly congruent to the study area and extends between the southeastern flank of the EPV and the East Friesian Island Juist, and the 'Dutch North Sea (Western areas)' reaches from the Dogger Bank to the Oyster Grounds. Colour-coded filled triangles, diamonds and stars mark index points resulting from this study of the infill of the Paleo-Ems River valley (for details see legend). The horizontal error bars represent the σ2-calibrated age range of the calibrated or recalibrated radiocarbon ages; the vertical error bars correspond to the total vertical error in sample altitude. [Color figure can be viewed at wileyonlinelibrary.com] presented by Vink et al. (2007;Fig . 7). The age-depth distribution of age points from core GeoB17721-1 and cores 09-1 to 09-3 and 09-5 to 09-7 shows two clearly defined groups: (1) age points from the top of peat intervals (Fig. 7, green or grey triangles facing up) match the index points, based on radiocarbon ages from basal peats, either from the areas in the southern part of the EPV (Fig. 7, yellow circles) or from the western part of the Dutch North Sea (Fig. 7, blue circles); whereas (2) the age points from the bottom or the middle of the peat intervals are significantly older (green or grey triangles facing down and grey rhombs). Ages obtained from the shallow marine, brackish or intertidal interval of core GeoB17721-1 (green diamonds), overlying the basal peat, match the index points from the southern part of the EPV (yellow circles) and, thus, reflect the flooding of the fluvial system. The age point from the marine infill of the incision at the base of the EVP (Fig. 7, white star) shows that the EPV experienced the sea level rise around 1000 years earlier than peats from the same depth (see Fig. 7, red circles) in the northern EPV area but outside of the valley. One tentative explanation would be that the EPV was topographicaly better connected to the advancing North Sea than other morphological depressions.
In contrast to the valley infill recovered in core GeoB17721-1, sediment core VC 2068 taken from a slightly more elevated setting approximately 15 km northwest of the lower part of the Ems paleovalley comprises early Holocene coastal mire deposits probably filling an isolated local depression (Wolters et al., 2010) (Fig. 2(a)). Both cores reveal the transition from Pleistocene sands to peat formations at about 35.4 or 35.2 m water depths, which in core VC 2068 have been dated to have formed between 10 700 and 9350 cal a BP (Wolters et al., 2010). This means that the onset of the peat growth in the Paleo-Ems was 600-700 years earlier (between 11 480 and 11 300 cal a BP), possibliy linked to a sea-level-induced rise of the groundwater table compared with the onset of peat growth in a mire setting in the exposed landscape. Wolters et al. (2010) proposed that the peat growth (core VC 2068) started independent of sea level rise. However, it seems that both the valleys and the exposed landscape were inundated during the same flooding phase. The transgressional phase is documented by a gradual transition from solely freshwater-fed mire communities to brackish reed communities in the pollen diagram of VC 2068 dated to 9460-9230 cal a BP Wolters et al. (2010). A gradual transition from freshwater to marine species is observed in the composition of the diatom assemblage in core GeoB17721-1.

Sedimentation history
The reconstruction of the depositional history of the Paleo-Ems estuary is based on the combined interpretation of grain size distributions, organic carbon content, radiocarbon dating, diatom concentrations, and the composition of the diatom assemblage. In general, we distinguish between fluvial, transgressional and marine phases (see stratigraphic interpretation in Fig. 3(h)). The sediments of core GeoB17721-1 belong primarily to the Buried River Valley Formation and the Upper Marine Formation within the stratigraphic concept developed for the German North Sea sector by Coughlan et al. (2018).
Fluvial phase (Late Weichselian to 11 500 cal a BP) The Paleo-Ems River probably already has a long history; for example, in Westphalia alluvial terraces of the Ems River are documented from the late Saalian and Weichselian times (Hempel, 1963). Based on seismic data and sediment core studies, Hepp et al. (2017) propose that the Paleo-Ems incised into Pleistocene glaciofluvial sands at least since the LGM, thereby creating a meandering river (see also Fig. 2(b)). Also, the new core data reveals the incision of the river into Pleistocene sands. The age of the basal peat of core 09-7 (11 700 cal a BP) indicates that peat growth had already started at the end of the fluvial phase.
Perimarine phase (11 500 to 9500 cal a BP) In core GeoB17721-1 and most of the 09-x cores, 40-160 cmthick basal peat layers directly overlie the valley base, formed by earlier fluvial erosion processes. The absence of evidence of characteristic fluvial deposits suggests that the fluvial activity was restricted to a small area of the valley at this point, so that the peat could grow directly from the base. Combining all cores (except core 09-7) provides evidence for the onset of peat growth between 11 480 and 11 300 cal a BP and its termination at around 9500 cal a BP. Although there are some indications (see above) that the Paleo-Ems was coupled to groundwater via the EPV at the beginning of the perimarine phase; there is no in situ evidence. However, the influence of the sea level rise on the depositional regime in the Paleo-Ems valley increased and resulted in a direct control at the end of the phase.
Brackish phase (9500 to 9300 cal a BP) Peat-growing ceased at 9500 cal a BP. After this, clay-rich muds with fine sands and high organic carbon contents dominate the sediment composition of GeoB17721-1. Rising sea level had a fundamental impact on the dynamic of the tidal regime. High sedimentation rates of 894 to 2000 cm ka -1 demonstrate that the valley was silting up rapidly within a few hundred years, maybe due to an upstream shift of the coastline and/or a disequilibrium of ebb and flood tide currents ( Fig. 3(g)). An upward coarsening of the very poorly sorted clayey silts just above the peat points to a slight increase in the stream competence. This upward coarsening coincides with the rapid decrease in total diatom concentration followed by the rapid replacement of fluvial-transported freshwater diatoms by brackish and marine species. However, we suppose that the siltation and salinisation were caused by a decrease of the stream capacity, since the river was not able to adjust its gradient to the rapidly rising sea level. Presently, there is no certainty as to whether the stream capacity also had a changing effect on the tidal regime. A slight jump to coarser grain sizes is seen at around 9410 cal a BP, accompanied by a further increase in the relative proportion of coastal planktonic diatoms, which become the dominant group. The transition shown in both grain size distribution and the diatom assemblage are interpreted to represent the transgressive contact, when freshwater conditions changed to a marineinfluenced environment, thus resulting in brackish or coastal lagoonal conditions.

Marine phase (after 9300 cal a BP)
The exact timing of the beginning of the marine phase is not clear because of a hiatus that is indicated by the sudden distinct shift in the grain size distribution at~150 cm core depth in GeoB17721-1. However, after 9290 cal a BP the depositional system changed to full marine conditions, as revealed by the almost complete absence of freshwater and brackish diatoms and the abrupt shift in the grain size distribution. This hiatus might be explained by the southward shift of the coastline due to the transgression and the onset of a regional dynamic current regime in the newly established North Sea area, as also corroborated by the sudden shift to sandy sediments above the hiatus (Fig. 3(c) and (h)). At this stage, the surrounding paleolandscape was probably flooded and the river valley rims were eroded, which might be reflected by a conspicuously high age of 19 990 cal a BP obtained on an apparently reworked shell. Unfortunately, no stratigraphic control is available for the time span covered by the hiatus, so that the earliest possible date for the establishment of full marine conditions is 9290 cal a BP. According to published sea level curves (Vink et al., 2007), the flooding of the paleolandscape within the study area was completed before 8200 cal a BP.
Above the uppermost sandy layer, 10 to 20 cm-thick layers of well-preserved or fragmented shells and coarser clastic sediments occur in cores GeoB17721-1 and 09-5 ( Fig. 3(b) and Fig. 6). These shell layers divide the most recent sediments from the underlying deposits of the first shallow marine phase and are a widespread phenomenon in the shallow North Sea, observed at similar core depths at many sites and are interpreted to be proximal storm layers (Aigner and Reineck, 1982;Uffenorde, 1982). The widespread distribution of such shell layers in the southern North Sea suggests a mega storm/ flood event or it might reflect a reorganisation of the current regime induced by the flooding of the Dogger Bank (Fitch et al., 2005).
The uppermost sand packages are commonly referred to as mobile sands that can be mobilised under the influence of recent currents, waves and tides (Zeiler et al., 2000). Coughlan et al. (2018) limited this interpretation to the uppermost part of the mobile sands overlying the well-preserved shell layer that obviously is not part of a mobile unit. Accordingly, only the uppermost 30 cm of core GeoB17721-1 above the shell layer are considered here to belong to the mobile sands.

Some implications for recent coastal lowland rivers
The presented study shows that a change in tidal dynamics due to sea level rise can result in a rapid increase of silt deposition and salinisation of coastal lowland rivers. Recent coastal systems in shallow shelf regions are strongly affected by ongoing sea level rise. Silt deposition in tide-influenced waterways, e.g. in the modern Ems River (van Maren et al., 2015) or sea-water intrusions into the river, which might threaten the drinking water supply of cities such as in the Yangtze River Delta (Ye, 2017), are current problems. The high rates of silt deposition and salinisation as demonstrated for the Paleo-Ems River show that such problems can become exacerbated within a few generations. While industrialised countries have the potential to react to changing environmental conditions within decades with proper coastal protection activities, the rapid silt deposition and salinisation of river mouths in coastal environments on century scales is still a persistent issue for developing or threshold countries.

Conclusion
Sediment cores from the infill of the submerged Paleo-Ems valley in the German North Sea reveal the geological record of the transgressional history in this area since the earliest Holocene. Radiocarbon-dated records of changes in sediment grain sizes and the diatom species composition provide new insights into the response of a coastal lowland river to the Holocene transgression and its relationship to the EPV and the surrounding landscape. The onset of the peat growth in the Paleo-Ems valley was most likely triggered by a sealevel-induced rise of the groundwater table through the link of the Paleo-Ems valley to the EPV. However, peat growth in the Paleo-Ems valley independent from relative sea-levelcoupled groundwater, simply as a subject of paludification within local topographic settings, cannot be excluded.
After a period of peat growth lasting for about 1900 years until 9500 cal a BP, the Paleo-Ems valley infill was controlled by brackish environmental conditions. The main finding of this study is that the river was not able to adjust its gradient to the rapidly rising sea level. Thus, the change from freshwater to brackish and finally to marine conditions occurred very rapidly within~200 years, corresponding to a~2.5 m sea level rise; well documented in the sedimentological and micropaleontological records. In terms of human timescales, this implies that the river valley silted up within a few generations.
The fate of the Ems paleovalley is exemplary for the evolution of the early Holocene coastal landscape in the German North Sea sector and the rapidly changing world of the Mesolithic hunters and gatherers living there. Humans and fauna directly experienced the drastic changes imposed by deglacial sea level rise. The former fluvial landscape changed rapidly into a coastal landscape with a prograding coastline swallowing the rivers and their levees. Thus, the human population was challenged to respond to these new environmental conditions within a few generations. A rapid increase of silt deposition and salinisation of river mouths in coastal environments on century scales due to sea level rise remains a persistent issue.