This work is a contribution to the ANR COCORISCO (French National Research Agency program for coastal risks assessment and management, ANR 2010-CEPL-001-01, Pôle-mer Bretagne). ANR COCORISCO, Domaines Océaniques laboratory (UMR CNRS 6538), GEOMER laboratory (UMR CNRS 6554) funded equipment, fieldwork, sample analysis and datings, together with PHILTRE and ARTEMIS programs. Town councils and the Conservatoire du Littoral provided full access to the study sites. Special thanks to Luigi Ardito (BRGM) for his drilling expertise. Finally, we are extremely grateful to the two anonymous reviewers whose constructive comments significantly helped to improve this paper.
1Recent years have seen a revival of interest in the use of relative sea-level (RSL) curves for a wide range of applications (Leorri et al., 2012) from the prediction of future sea-level changes in coastal risk assessment issues to the fine-tuning of geophysical Earth rheology models. The behaviour of coastal systems in reaction to Holocene sea-level rise is a complex process. To derive RSL (taken here as the relative altitude of past sea-level relative to a present reference level) changes from observations is hence a complex challenge. Though primarily determined by eustasy, RSL changes result from the combination of many interdependent processes. As clearly stated by I. Shennan et al. (2012), sea-level changes for a precise geographical location and epoch result from the sum of the following factors: (A) Eustatic sea-level, which varies in time following the melting history of ice-caps (“ice-equivalent” eustasy), changes in the ocean-basin volume (“tectono-eustasy”), and changes in the ocean mass-water distribution. (B) Isostatic processes affecting the vertical behaviour of the lithosphere during the post-glacial period of study, dependent on the geographical position relative to the past extension of ice-caps (glacio-isostasy) and of the loading by the water-charge (hydro-isostasy) during transgression. The sum of (A) and (B) produces the global trend in the RSL rise. Along with those major factors must be added: (C) Tectonic activity expressed by vertical displacements that could have occurred during the considered period. These latter displacements can explain rapid jumps in the RSL signal and must be looked for in regions known as tectonically active during the Holocene (Fouache, 2006; Morhange et al., 2006; Gandouin et al., 2007). In addition to these factors are superimposed more local effects, which bear direct consequences on the way the sedimentary sequences record the RSL changes: (D) Local effects are likely to control the sedimentary record and, as a consequence, the preservation and the present-day restitution of Holocene RSL. In fact, locally, sedimentary sequences undergo syn- and/or post-deposit processes: major local effects are traditionally considered as being the compaction of the deposits (effect of post-depositional consolidation), lowering the original altitude of a considered unit, and thus the RSL corresponding points), and the changes in tidal regime (leading to an alteration of the sea-level indicative meaning, which is derived from the present day tidal regimes). (E) Finally, I. Shennan et al. (2012) consider an additional factor regrouping other unspecified, random and hard-to-quantify factors influencing the RSL reconstruction at a given site. We propose here that erosion truncations of sedimentary sequences must be considered as one of the major components of this additional factor of the RSL equation, considering the importance of its probable impacts on both the indicative meaning of the studied sedimentary sequences and on the compaction processes. Several recent works have attempted to sort out the relative importance and the timing of local parameters (D) and (E), considering them as keys to a better understanding of the response of coastal systems to past sea-level rises (Shennan and Horton, 2002; Edwards, 2006; Long et al., 2006; Massey et al., 2006b; Behre, 2007; Vink et al., 2007; Horton and Shennan, 2009; Bungenstock and Weerts, 2010; Baeteman et al., 2011; Brain et al., 2011).
2Northwestern European coasts have been the subject of several works dealing with Holocene RSL variations (e.g., among many others: Fairbridge, 1961; Ters, 1973; Morzadec-Kerfourn, 1974; Ters, 1986; Van de Plassche, 1991; Denys and Baeteman, 1995; Long et al., 1996; Lambeck, 1997; Shennan and Horton, 2002; Massey et al., 2006a; Edwards and Horton, 2007; Massey et al., 2008; Bungenstock and Schäfer, 2009; Stephan, 2011; Leorri et al., 2012). In comparison with British coasts, French Western Atlantic and Channel coasts have been the subjects of fewer studies (Delibrias and Guillier, 1971; Ters, 1973; Morzadec-Kerfourn, 1974; Van de Plassche, 1991; Morzadec-Kerfourn, 1995; Lambeck, 1997; Clavé et al., 2001; Stephan, 2011; Leorri et al., 2012). In the Finistère region, only two authors really focused on this subject (Morzadec-Kerfourn, 1974, 1995; Stephan, 2011). Both authors proposed that plurimetric oscillations in RSL curves could have taken place ca. 3000 cal. BP. However, these works were conducted in two areas of relatively limited extension and produced curves based on a limited number of sea-level index points (a total of around fifteen dates for each of the two studies). Such limitations make inter-comparisons between sites difficult and hence can impair the deciphering of local processes in the RSL signal. Finally, the period ca. 3000 cal. BP was generally identified around northwestern European coasts as a period of increased storminess, which has caused strong perturbations and large hiatuses in the sedimentary sequences (Lamb, 1977; Long and Hughes, 1995; Van Geel et al., 1996; Moura et al., 2007). Hence, we consider that obtaining a reliable and high-frequency sea-level curve for Western-Brittany is of prime interest, as it will contribute both to the understanding of the coastal processes, which may be the consequences of the sea-level variations and, at a broader scale, the comprehension of post-glacial isostatic history of North-Western Europe.
3Within the framework of a coastal risk assessment research program (ANR COCORISCO, French National Research Agency), an extensive fieldwork has been conducted during the last three years all around the Finistère peninsula (Western part of Brittany, France). Tens of cores and drillings have been carried out in sites of various sedimentary environments and expositions. Whilst the principal objective of this fieldwork was to obtain data that would allow the study of the palaeostorminess during the Holocene, the sampled sedimentary sequences could also be fruitfully interpreted to derive RSL data. This allowed us to densify significantly the Holocene RSL data available for this region and also to re-assess and update the results of M.-T. Morzadec-Kerfourn (1974) and P. Stephan (2011). This paper is the first step of an effort in updating Western Brittany RSL data. The first part will briefly outline the geological context of the study area, considering possible internal geodynamic controls on vertical crustal movements that could have occurred during the period of investigation. The methodology we used to reconstruct RSL changes will be described in a second part, with particular attention paid to possible errors that may occur at various stages of the reconstruction process. In order to allow comparisons between previously published interpretations and our own results, derived from an extended new data set, the same methodology has been applied to both to new and previously published data. Finally, the ongoing need for a better deciphering of both local and regional effects for the fine-tuning of the Holocene RSL history is exemplified and the possible behaviour of coastal systems in reaction to the proposed Holocene RSL rise is briefly evaluated.
4This paper presents new stratigraphic data and RSL data collected along the coast of the Finistère peninsula (Western Brittany, France). As stated above, the rationale for the choice of site locations primarily resulted from the objectives of the COCORISCO program, which privileged sites showing very different exposures to major forcing agents (large tidal ranges, storms swells), resulting in diverse sedimentation conditions (in terms of sediment supply, accumulation and preservation). This largely explains why the selected sites are not always the most favourable to derive a long and continuous RSL signal. Nevertheless, we believe that these widespread and diverse site locations and types can be profitable to improve the regional representation of reconstructed RSL. Moreover, such a choice of the sites may, to a certain extent, avoid peculiar RSL biases that could result from working on a too local and geographically restricted dataset. New stratigraphic data were collected on eight sites distributed all along the Finistère peninsula: Kerlouan-Guissény and Landeda in the North Finistère, Porsmilin in the central part of the peninsula, and Treffiagat, Kermor-Tudy and Guidel in the southern part (fig. 1).
Fig. 1 – Location maps of the study area.
Fig. 1 – Cartes de localisation de la zone d’étude.
Numbers refer to sites presented in the text. 1: Kerlouan-Guissény; 2: Landeda-Tariec; 3: Porsmilin; 4: Treffiagat; 5: Kermor-Tudy; 6 : Guidel. Letters A to D refer to sites from which sedimentary sequences were re-assessed from M.-T. Morzadec-Kerfourn (1974). A: Plouescat-Porsguen. B: Ploudalmezeau-Treompan. C: Argenton. D: Trézien. Letters H to J refer to sites studied by P. Stephan (2011) in the Bay of Brest. H: Porzguen marsh. I: Troaon marsh. J: Arun marsh.
La numérotation renvoie aux sites tels que présentés dans le texte. 1 : Kerlouan- Guissény ; 2 : Landeda-Tariec ; 3 : Porsmilin ; 4 : Treffiagat ; 5 : Kermor-Tudy ; 6 : Guidel. Les lettres A à D désignent les sites pour lesquels les séquences étudiées par M.-T. Morzadec-Kerfourn (1974) ont été réactualisées. A : Plouescat-Porsguen ; B : Ploudalmezeau-Treompan ; C : Argenton ; D : Trézien. Les lettres H à J désignent les sites étudiés en rade de Brest par P. Stephan (2011). H : Marais de Porsguen. I : Marais de Troaon. J : Marais d’Arun.
5Geological and geodynamic factors, which control the possible vertical movements of any particular crustal area, are of prime importance in RSL reconstitutions and can be considered as the primary producers of RSL histories. They can be divided as follows: (i) Long-term geological evolutions, typically spanning over tens of thousands to millions years, (ii) Holocene glacio- and hydro-isostatic processes (from a few hundreds to a few thousands of years), and (iii) short-term tectonic events (a few days). As geological processes are concerned, the following will be considered as the basis of this study and be used for the comparison of our data with southern-UK works. On a long timescale, Brittany is considered to subside step-wise since the opening of the Bay of Biscay during Early Cretaceous (Evans, 1990), as a passive margin and as a flank of the Western Channel graben. Globally, long-term subsidence has been evaluated between 0.02 mm/a and 0.04 mm/a since the Eocene for the northern coasts and western/southern coasts of Brittany, respectively (Bonnet et al., 2000). The regional subsidence can be considered as negligible at the timescale of the Holocene (~ 40 mm for the last 10,000 years), with respect to the amplitude of the RSL changes we consider.
6As underlined in the introduction, estimations of the amplitude of potential isostatic adjustments undergone by Western Brittany, as a consequence of the melting of the North-western Europe and North hemisphere ice caps still lack precision, awaiting specific numerical studies and geophysical models, well-constrained by local high-quality and dense RSL information. During the Last Glacial Maximum, Brittany occupied a peripheral position with respect to the British Ice Sheet (BIS). The maximum extent of the BIS reached the Scillies (Scourse et al., 1990; Scourse 1991) and had already retracted before 17 ka as evidenced from the dating of loess along the coasts of Wales (Wintle, 1981). Recent studies, conducted in South-Western UK (Massey et al., 2008; Bradley et al., 2011; Shennan et al., 2012), consider that this latter region must have been under the joint influence of both the BIS forebulge and long wavelength isostatic responses related to Fennoscandia ice-sheet. E. Leorri et al. (2012) performed modelling work from RSL datasets around the Bay of Biscay and conclude that a north-south increasing RSL gradient exists between Brittany and Portugal. As primarily proposed by K. Lambeck (1997), this would confirm that vertical land-motions induced by the subsidence of a peripheral bulge, associated to the melting of Eurasian Ice-Sheets, have had (and maybe still have) noticeable effects at least as far south as the southern Bay of Biscay. While both (i) the position, size and migration timing of the forebulge apex remain uncertain and (ii) the role of the deep faults bordering the Channel and Western Approaches domains during the Holocene is still questionable, it appears that Brittany could have also been concerned by such glacio-isostatic vertical movements. Nonetheless, modelling work performed by E. Leorri et al. (2012) has to be taken with caution as Brittany RSL reconstruction on which are based these modelling are rather old and did not received proper error re-assessment. Furthermore, this study got RSL data from different regions confusedly mixed up from their original locations. In fact, a careful inspection of the datasets reveals that a site named “Saint-Marc” is taken by the authors to be in south-Brittany, while original data from M. Ters (1973) mention “Côtes-du-Nord” which is situated in north-Brittany. Hence, in absence of more precise work regarding this problem, the question of the Holocene vertical behaviour of the Brittany region towards the glacio-isostatic effects remains suspended.
7The hydro-isostatic vertical movement induced by the loading of the Atlantic continental shelf and the Channel platform during the Holocene transgression can be considered as of comparable amplitude between South-Western UK and Brittany, considering the similar ages and, therefore, the comparable rigidities of the crust of the two regions. Finally, neo-tectonics is generally considered to have been inactive in Brittany over the study period.
8Due to the direct opening of the region towards the Atlantic Ocean, the Finistère peninsula is the part of Brittany most exposed to dominant storm winds and swells. These exposure conditions result in (i) a majority of sand-dominated sedimentary environments and (ii) a rather limited number of protected low-energy environments susceptible to have recorded continuously the RSL history. As revealed by the several erosion surfaces observed within our sequences, impacts of storminess during the Holocene have heavily disturbed and discontinued mid- to late-Holocene sedimentary sequences along the Finistère coasts. As this study principally focuses on basal deposits, this will not bear consequences on the results presented in this paper. The possible role played by storminess and sedimentary hiatuses will be discussed when comparing our results to other studies, which use non-basal deposits to derive RSLs.
9Tidal settings are of major importance in RSL studies. In fact, most RSL reconstructions are based on the study of the altitudinal zonation of intertidal indicators (whether biological or sedimentological) with respect to a reference tidal level (such as the Mean High Water Spring Tide, MHWST, or the Mean Sea Level, MSL). Hence, by determining the vertical range in which an indicator could have been emplaced (the “indicative range”), the tidal range of a studied site will partially determinate the accuracy of past RSLs. While micro-tidal regions, such as the Mediterranean Sea, allow well-constrained and precise indicative ranges (pluri-centimetres to decimetre scale), macro-tidal conditions, such as the ones encountered along the coast of Brittany, can introduce larger errors in RSL reconstitutions (pluri-decimetres to metre-scale error terms). Moreover, potential changes in the tidal ranges during the Holocene could have resulted in supplementary errors in past RSL reconstitutions, by over- or under- estimating the indicative meaning. Indeed, as stated by W.R. Gehrels et al. (1995), many RSL studies are based on high-tide indicators. Thus, changes in tidal ranges through time may cause the reconstructed RSL curve to differ from the “true” MSL curve (as termed in Gehrels, 1999). The Finistère peninsula is concerned by macro- to meso-tidal conditions, with tidal ranges globally decreasing from north to south (tab. 1). Variability in the tidal ranges between the different study sites will be taken into account in the determination of the indicative ranges error terms (see methodology). Evaluation of possible changes in the tidal ranges and regimes during the study period is beyond the scope of this paper. Possible implications of these changes will be discussed in the quantification of the RSL reconstruction error-terms.
Tab. 1 – Present-day tidal ranges on studied sites.
Tab. 1 – Marnages actuels sur les sites d’étude.
no.
|
Location
|
Latitude
|
Longitude
|
Tidal range
(in m)
|
HAT (in m NGF)
|
MHWST (in m NGF)
|
MHWNT (in m NGF)
|
MTL (in m NGF)
|
MLWST (in m NGF)
|
A
|
Plouescat
|
48°39' N
|
4°12' W
|
7.43
|
5.10
|
4.26
|
2.49
|
0.67
|
-3.17
|
1
|
Guisseny
|
48°38'14 N
|
4°23'56 W
|
7.52
|
4.74
|
4.06
|
2.36
|
0.62
|
-3.46
|
2
|
Landeda-Tariec
|
48°36'10 N
|
4°36'30 W
|
6.75
|
4.57
|
3.78
|
2.13
|
0.46
|
-2.97
|
B
|
Ploudal.-Treompan
|
48°34' N
|
4°40' W
|
6.6
|
4.54
|
3.69
|
2.11
|
0.46
|
-2.91
|
C
|
Argenton
|
48°31' N
|
4°45' W
|
6.47
|
4.48
|
3.68
|
2.08
|
0.47
|
-2.79
|
D
|
Trezien
|
48°25' N
|
4°47' W
|
6.08
|
4.32
|
3.51
|
1.96
|
0.48
|
-2.57
|
3
|
Porsmilin
|
48°21'19 N
|
4°40'49 W
|
5.55
|
3.99
|
3.17
|
1.67
|
0.39
|
-2.38
|
4
|
Treffiagat
|
47°47'29 N
|
4°15'54 W
|
4.20
|
3.16
|
2.66
|
1.56
|
0.50
|
-1.54
|
5
|
Kermor-Tudy
|
47°51'54 N
|
4°07'37 W
|
4.20
|
3.09
|
2.45
|
1.35
|
0.41
|
-1.75
|
6
|
Guidel
|
47°45'44 N
|
3°30'10 W
|
4.15
|
2.78
|
2.23
|
1.18
|
0.17
|
-1.92
|
HAT (Highest Astronomical Tide), MHWST (Mean High Water Spring Tides), MHWNT (Mean High Water Neap Tides), MTL (Mean Tide Level), MLWST (Mean Low Water Spring Tides). Letters and Numbers in the first column refer to fig. 1.
HAT (plus hautes mers astronomiques), MHWST (pleines mers de vive-eau moyenne), MHWNT (pleines mers de morte-eau moyenne), MTL (marée moyenne), MLWST (basses mers de vive-eau). Les lettres et numéros figurant dans la première colonne renvoient aux sites tels que présentés dans la fig. 1.
10RSL reconstitutions generally rely on the acquisition of Sea-Level Index Points (SLIPs), or Sea-level Indexes (SLIs), which characterise the altitudinal position of a chosen tide level (e.g., the Mean High Water Spring Tide, which will be the reference throughout this work) relative to its modern altitude, considered as the zero reference level. As stated by W.R. Gehrels (1999), the vertical determination of a SLIP at time (t) can be expressed as:
11SLIP (t) = H – D – I (t) + T (t) - L+ A (t) (1)
12where H is the height of the coring station from which the indicator sample is taken relative to the modern tidal altitudinal reference, D is the depth of the sample in the core, I(t) is the “indicative meaning”, i.e. the relation between the altitude of the original deposition environment of a given sample and the palaeo-tidal reference level (e.g., MHWST) at the time of deposition (Shennan, 1986; Van de Plassche, 1986; Gehrels et al., 1996), T(t) represents the potential changes in tidal amplitude between time (t) and present, L stands for the compaction induced by the coring-process as evaluated in the field and A(t) is the auto-compaction undergone by the sample after deposition. If an “indicative meaning” cannot be properly assessed for a deposit, then a status of “limiting point” is generally attributed to the sample. Limiting points can help to constraint former RSL positions, and are defined as follow: “High-Limiting” (HL) or “Low-Limiting” (LL) whether the deposit is identified to have formed at an unquantified altitude above or below the reference tidal level, respectively. To reach some reliability and precision, RSL reconstructions must rely on a combination of methods, as no single method is considered to be unequivocal. The methods used in this study, or in re-assessed previous works, are detailed below in some details. For a general view on RSL reconstruction methods, readers can refer to I. Shennan (2007), R.J. Edwards (2007), and to the publications of the UNESCO-IGCP projects (e.g., projects no. 61, 274, 588).
13Basal peats have been extensively used in Holocene sea-level studies, especially in northwestern Europe (e.g., Jelgersma, 1961; Van de Plassche, 1982; Törnqvist, 1993; Denys and Baeteman, 1995; Van de Plassche, 1995; Beets and Van Der Spek, 2000; Shennan et al., 2000; Shennan and Horton, 2002; Makaske et al., 2003; Berendsen et al., 2007), and also along other oceanic facades (e.g., Redfield and Rubin, 1962; Pirazzoli and Pluet, 1991; Gehrels and Belknap, 1993; Gehrels et al., 1996; Long et al., 1998; Törnqvist et al., 2004, 2006). As underlined by T.E. Törnqvist et al. (2004), this approach is particularly suitable for reconstructing RSL history along submerging coastlines and has widely proven its accuracy to provide consistent RSL records. Indeed, the submergence of the Pleistocene glacial plains by the coastal Ground Water Level (GWL) wedge during post LGM sea-level rise provoked a decrease of the drainage and subsequently the formation of freshwater to brackish marsh environments belts environments, in which peat accumulated (Baeteman et al., 1999; Vink et al., 2007). A landward and upward progression of the peat layers marks the emplacement of the rising water level. Moreover, basal peats directly overlay the geological substrate or, more often, Pleistocene silty-loess of fluvial deposits, traditionally considered as having experienced negligible compaction during the Holocene. The altitude of the basal peats hence can be used as a reliable marker of RSL rise.
14Nonetheless, the representativeness of basal peats as RSL indicators has been regularly discussed. Several limitation problems have been identified in their use as sea-level indicators. The unclear relations between the GWL and sea level may cause errors in RSL reconstructions. As identified by O. Van de Plassche (1982, 1995), distortions of the tidal wave in large deltaic plain environments (“flood-basin effect”) or in estuarine settings (Fairbridge, 1961; Vink et al., 2007), can lead to under- or over-estimation of the RSL. The importance of local continental hydrological and drainage conditions can also be invoked. In order to avoid these errors, some authors recommended that only basal peats should be used as an upper RSL limit (Edwards, 2007; Vink et al., 2007). Others (Gehrels et al., 1996; Gehrels, 1999; Törnqvist et al., 2004) advocated that basal peats remain usable as SLIPs if a clear relationship between their original deposit environment and sea level can be established (i.e., from the presence of indicators undoubtedly attesting that peat formed in an environment submitted to brackish to saline conditions). Only in this case, basal peats can be considered to have formed between MSL and HWST and can be used as SLIPs (Van de Plassche, 1982; Törnqvist et al., 2004).
15The salinity regime of the original deposit environments can be assessed by several methods. Microfossils, such as diatoms or foraminifera, are considered to be the most reliable method available to achieve such an aim. When present, microfossils attest at least a minimal influence of marine conditions (Gehrels, 1999). The importance of this influence can be deduced by the determination of species assemblages. Unfortunately, foraminifera were absent or very scarce in the samples we studied, probably due to their limited potential of conservation or to unfavourable living conditions (sand dominated deposits). Micro- and macro-faunal indicators can also provide useful information. Plant communities obey a vertical zonation with respect to ecological factors (among which the frequency of inundation is dominant) and thus can be used as indicators to determine the salinity regimes. Pollen assemblages have been widely used in Holocene palaeo-environmental studies, as a proxy for reconstructing past landscapes and anthropogenic activity. It also revealed capacities to decipher between different sedimentary environments, especially in intertidal mud-dominated environments (such as mudflats and salt-marshes), where strong environmental gradients exist. Transitions between marine and freshwater dominated environments can be identified following the succession of dominating plant species.
16Stable carbon isotope ratios (13C/12C) have been less often used to recognise salt-marsh environments within organic sedimentary sequences (Chmura and Aharon, 1995; Gehrels et al., 1996; Törnqvist et al., 2004; Wilson et al., 2005; Lamb et al., 2006, 2007; Kemp et al., 2010). The stable carbon isotope ratio 13C/12C is noted δ13C, and expressed in parts per thousand PDB (‰ PDB) relative to the reference 13C value (standard determined from a Cretaceous marine fossil named Pee Dee Belemnite, “PDB”). Mudflat/low-marsh and high-marsh/supratidal specific plant species were found to follow different photosynthetic pathways (termed “C3” and “C4”), inducing very different δ13C imprints in the surface sediment on which those latter grow. Unfortunately, values in δ13C sometimes largely overlap and therefore preclude using δ13C directly as a ‘true’ RSL proxy (Wilson et al., 2005). This ratio has nevertheless proved its ability to clearly decipher between low-marsh and high-marsh environments. Thus, the δ13C stable carbon isotope ratio can help to provide qualitative data about the depositional environment of radiocarbon-dated samples (Wilson et al., 2005; Lamb et al., 2006, 2007; Kemp et al., 2010), especially if associated with other indicators (foraminifera, diatoms, C/N ratio). T.E. Törnqvist et al. (2004) used δ13C, together with the study of plant remains, to determine the salinity regime of basal peats found in the Mississippi delta, and consequently to determine if these basal peats could be used as reliable markers of past RSL values. The same approach has been followed in this study.
17Field data were collected by three techniques: corings (using an Eijkelkamp 60 mm Ø, hand-held and motor-driven percussion corer), screw drillings (using BRGM vehicle, French Research Agency for Geology and Mines) and hand-auguring (3-cm diameter/1-m long Eijkelkamp gouge). Cores positions were surveyed using a Trimble Differential Global Positioning System (DGPS). All locations were tied to geo-referenced IGN (French National Geographic Institute) benchmarks and levelled with respect to the NGF datum (French levelling datum, tied to the present MSL in Marseille, France). Possible deformations of the sequences during the coring stage (e.g., compaction induced by percussion corings) were measured in the field and taken into account in the vertical positioning of the samples. Percussion cores (recovered in 1-m long plastic tubes) were cut in half cylinders and sealed by plastic film. Screw-drillings and auguring samples were kept in sealed plastic bags. All were stored at +4 °C to slow down desiccation and oxidation of the deposits.
18Deposits were rapidly described by a litho-microstratigraphic study, in terms of grain size, colour (Munsell color chart), organic content, and characterisation of macrofossils. Particular attention was paid to identify possible reworked levels or erosion surfaces, as these define morphogenic active periods and can give hints on subsequent potential problems in sample dating. Sampling was performed with a 5-cm mean spacing and adapted to the complexity of the sequences. Grain size was measured after destruction of organic material with a solution of hydrogen peroxide in each sample for the <1-mm fraction with the help of a Malvern Mastersizer 2000 laser analyser. The size of the grains coarser than 1 mm was analysed by sieving.
19AMS radiocarbon dates were performed at the “Laboratoire de Mesure du Carbone 14” (Gif-sur-Yvette) and at the Poznan Radiocarbon Laboratory (Poland). Half of the new dates presented here were dated from bulk deposits (Guisseny cores). Other dated material was sampled with care in order to minimise sources of carbon pollution: in situ material was always preferred to potentially reworked material, such as charcoals or drifted wood. When present, plant remains were preferred as their fragility reduces the potential of re-deposition (Gehrels et al., 1996; Törnqvist et al., 1998; Gehrels, 1999). Stable carbon isotope (δ13C) was routinely measured for all the samples dated at the Gif-sur-Yvette laboratory. Calibration was performed with the Calib 6.0 software, using IntCal09 calibration curve (Reimer et al., 2009). All dates in this study are presented in calibrated age BP, with a 2-σ (95%) probability confidence interval (tab. 3 and tab. 4).
Tab. 2 – Errors terms used in RSL reconstruction.
Tab. 2 – Marges d’erreurs utilisées dans la reconstitution du niveau marin relatif.
|
This study
|
Re-assessment of
M.-T. Morzadec-Kerfourn (1974)
|
Percussion coring
|
Ø
|
Ø
|
Screw-drilling
|
± 20 cm
|
Ø
|
Hand-auguring
|
-2 cm/m depth
|
Ø
|
Levelling
|
±2 cm/m depth
|
±50 cm
|
Present tidal frames
|
±5 cm when estimated from distant tidal references
|
Indicative ranges
|
(HAT-MHWNT)/2
|
(HAT-MHWNT)/2
|
Tab. 3 – New radiocarbon dates and Sea-Level Index Points (SLIPs) and limiting dates.
Tab. 3 – Nouvelles datations radiocarbones, Sea-Level Index Points (SLIPs) et dates limitantes.
“HL” notation states for “High-Limiting”.
HL correspond à « High-limiting ».
Tab. 4 – Data re-assessed from M.-T. Morzadec-Kerfourn (1974).
Tab. 4 – Données réévaluées depuis M.-T. Morzadec-Kerfourn (1974).
“HL” notation states for “High-Limiting”.
La notation HL correspond à « High-limiting ».
20Sampled basal peats, lying at the basis of the Holocene sedimentary sequences we sampled, were generally found lying on a thin Pleistocene silt layer resting either directly on the weathered basement or, more often, on a mixed silt and gravel formation. This layer could either consist of a periglacial loess formation (Monnier, 1973; Hallégouët and Van Vliet, 1986) or of an inter-bedded lacustrine mud layer formed by the hampered drainage during the RSL rise. Those units generally progressively evolve towards an organic rich silt layer, which is interpreted as the response to the progressive flooding of coastal lowlands by the water table, as a primary response to sea-level rise. It generally ends up in a pluri-decametric to 1-m thick brown to black peat bed, often cut at its top by an erosional surface (fig. 2).
Fig. 2 – Stratigraphic logs of studied cores and drillings.
Fig. 2 – Logs stratigraphiques des carottes et forages étudiés.
1: clay; 2: silt; 3: mud; 4: high organic content; 5: humifying level; 6: peat; 7: fine-grained sand; 8: medium-grained sand; 9: coarse-grained sand; 10: gravels; 11: pebbles; 12: fluvial gravels and pebbles; 13: geological substrate/weathered basement (granite/shale); 14: soil; 15: charcoals; 16: roots in living position; 17: wood fragment; 18: wood in living position/tree stem remains; 19: shell fragments; 20: shell in living position; 21: erosion surface.
1 : argiles ; 2 : limons ; 3 : vase ; 4 : importante fraction organique ; 5 : niveau humifère ; 6 : tourbe ; 7 : sables fins ; 8 : sables moyens ; 9 : sables grossiers ; 10 : graviers ; 11 : galets ; 12 : graviers et galets fluviatiles ; 13 : substrat géologique/régolithe (granite/schiste) ; 14 : sol ; 15 : charbons ; 16 : racines en position de vie ; 17 : fragment de bois ; 18 : bois en position de vie/restes de tronc ; 19 : fragments coquillers ; 20 : coquille en position de vie ; 21 : surface d’érosion.
21The determination of the salinity regime in which basal peats formed is of prime importance. This determination states whether a peat level can be used as a SLIP (this is the case if the peat level formed under saline or brackish influences) or must be only considered as a high-limiting point (if the peat formed in freshwater environments, without any saline influences). In the absence of micro-faunal indicators in our basal deposits, the determination of the salinity regime was achieved using both stable carbon isotope ratios and the observation of macro-plants remains. As local δ13C study is still in progress on present-day salt-marsh environments of our study zone, δ13C ratios obtained by T.E. Törnqvist et al. (2004), G.P. Wilson et al. (2005), H.H. Lamb et al. (2007), A.C. Kemp et al. (2010) and S. Engelhart et al. (2013) were used as references for discriminating between mud-flats, low-saltmarshes and high-marshes environments. The results of G.P. Wilson et al. (2005) and S. Engelhart et al. (2013) proved especially useful, as the environments they describe are quite similar to those we observed. This led us to consider average δ13C values over -16‰, between -17‰ and -27‰ and under -28‰ as limits for saline, brackish to intermediate and strictly freshwater environments, respectively. Nonetheless, the correlation between δ13C values and foraminifera assemblages obtained by P. Stephan (2011), led us to slightly reconsider the -28‰ δ13C value, fixed by these studies as the boundary between brackish and freshwater environments. In fact, δ13C values as small as -28,8‰ were obtained by P. Stephan (2011) from deposits identified as high saltmarsh by foraminifera analyses (dominated by Jadammina macrescens and Trochamina inflata species). Nevertheless, taking into account the difficulties to unambiguously discriminate between high-marsh/supratidal environments solely on the basis of δ13C values alone (a clearer distinction would require C/N measurements as noted by G.P. Wilson et al., 2005 and H.H. Lamb et al., 2007), this indicator will be only used only to confirm brackish conditions and not a RSL proxy. Additional information was retrieved from examination of macro-plant remains if those were easily identifiable. Palynological data were also used when available on the studied deposits.
22Accordingly to O. Van de Plassche (1982), when salt-water-influenced basal peats can be clearly identified, they can be considered as having formed at an altitude between MSL and MHWST. We consider that this indicative range must be adapted on the basis of the δ13C data available for our study zone. Indeed, a MSL deposition environment would have led to a very brackish environment characterised by very halophilous plant species, and hence to heavy δ13C values. Stable carbon isotope measurements we obtained show that basal peat would rather have formed around the high-marsh/freshwater supratidal limit. Hence, we chose to change the indicative meaning associated to brackish peats and set it between high astronomical tides (HAT) and mean high-water neap tides (MHWNT). Freshwater deposits are considered as high-limiting deposits and form a very maximum for former MHWST levels. Tidal levels were obtained for each site from the data of the nearest tidal-gauge (or tidal reference) station.
23The evaluation of the error terms is of particular importance in RSL reconstructions. These error terms must be assessed before any RSL changes can be reliably proposed, and before any comparison between different studies can be attempted (Heyworth and Kidson, 1982; Shennan, 1982, 1986). Possible error sources are the following: (i) inaccurate altitude level of the borehole, (ii) imprecise position of the sampled horizon within the core, (iii) indicative meaning of the sample, (iv) contemporaneous differences of tidal ranges between surveyed site and reference tide-gauge, (v) potential tidal changes over the studied period, (vi) post-depositional compaction, (vii) effects induced by exceptional tides and storm surges, (viii) interferences induced by sedimentary processes (e.g., accumulation rates, peat growth independent from water level, hiatuses), and (ix) age errors inherent to the radiocarbon dating process and potentially induced by the presence of allochtonous carbon in the sampled deposit. All these errors terms, considered for both our own and re-assessed dates, are detailed below and listed in table 2.
24Both altitudinal and age error terms were quantified in this study. Altitudinal errors consist in the combination of errors associated to the fieldwork (coring, sampling, surveying) and of errors linked to the determination of the indicative meanings and indicative ranges (Shennan, 1986; Van de Plassche, 1986; Gehrels et al., 1996; Brook and Edwards, 2006). Hand-augurings are known to be prone to a potential underestimation of the altitude of sampled layers due to the flexion of the rods. Hence, an error of -2 cm/m depth was applied to samples obtained by hand-augurings (Törnqvist et al., 2004). As it was consistently checked in the field, verticality of the percussion and screw-drillings was not taken into account as a possible source of altitudinal error. An error of ±20 cm was nonetheless assumed for the “twirl” effect potentially induced by the screw-drilling method. A total altitudinal error of ±2 cm was applied for the GPS-surveys (Suanez et al., 2008). In order to straddle the whole vertical range in which brackish basal peat may have formed, an error term (indicative range) was added to their altitude. Considering the indicative meaning we chose to apply to brackish-influenced basal deposits, this error was fixed to (HAT-MHWNT)/2 and at each site determined from the contemporaneous local tidal range. When tidal levels were obtained by interpolation between tide-gauges stations, an additional error of ±5 cm was assumed to account for a possible interpolation inaccuracy. We assume the error associated with potential changes in the tidal ranges during the study period to be negligible over the last 7000 years cal. BP, even if no evidence unambiguously supports such an assumption. Modelling work by K. Uheara et al. (2006), though conducted at the scale of northeast Atlantic and thus ignoring a possible influence of changes in local sedimentary environments, predicted that such changes must have been smaller than 0.5 m over the study period. Age errors were directly obtained from the radiocarbon dating and calibration processes (2-σ confidence interval).
25As only basal deposits are addressed in this study, only simplified stratigraphic sequences will be presented (fig. 2). Work on producing fully detailed sequences is still underway and will be presented in a forthcoming paper. Study sites are presented from the north to the south of the study area (fig. 1). The coring sites coordinates are listed in table 1 as well as the tidal setting for each site.
26The bay of Tressény is the northernmost site studied in this work. It is a N 315°-oriented narrow bay, terminated at its southernmost part by a relatively sheltered salt-mash cove, forming the mouth of the small “Quillimadec” River. The study of Kerlouan-Tressény site was conducted within the framework – and funded by – the Molène Archipelago archaeological research program of the Brittany Region Archaeology Service (Pailler et al., 2011). The present local tidal levels (table 1) were estimated from the two nearest tidal-references stations (Aber-Wrac’h and Brignogan, ~10 km west and east of the site, HAT=4.74 m NGF, MHWST=4.06 m NGF, MHWNT=2.36 m NGF).
27The base of the sedimentary record consists in 0.4 m of Pleistocene gravel rich clay of probable fluvial origin, progressively topped with ~0.6 m of a highly-organic stratified peaty-mud, containing very scarce foraminifera and reed fragments. This basal-deposit then evolves into a 0.15 m layer of black-peat ending in an abrupt erosive contact (-0.8 m NGF). Above this contact rests a 0.6 m thick sand layer, turning progressively finer from coarse to medium sand. Above, sedimentation alternates between more or less organic silty deposits. Some sand-enriched levels are observed that could be interpreted as either storm deposits for the shell-rich ones or as emplaced by the shifting of river channels. Above 2.95 m NGF, the sequence becomes highly organic again to the top of the core, the topmost part being composed of the contemporaneous salt-marsh soil (at 3.75 m NGF). The very base of the basal peat-mud layer (-1.52 m NGF) was dated 6351 ± 52 cal. BP. Used alone, the -28.8‰ δ13C value does not allow to clearly discriminate between brackish or freshwater supratidal environment but the presence of reeds leaves remains and foraminifera attests for brackish influences. Former MHWST on the site ca. 6351 ± 52 al. BP can hence be positioned at -5.06 ± 1.26 m below contemporaneous MHWST level (Index no. 2).
28Two hand-augurings were made at the bottom of the Vougot beach situated at the mouth of Tressény cove. The first core (Guis-S1) recovered 0.6 m of a brown reed peat resting on the Pleistocene loess layer. Overlying this peat deposit is ~1.6 m of white sand, progressively gaining in organic content towards its top, till it becomes a true sandy-peat layer at ~1.05 m NGF. The topmost 10 cm of the core correspond to the dark-massive brown peat layer today outcropping at the bottom of the beach. Base of the base peat layer (-1.15 m NGF) was dated to 4900±68 cal. B.P (UBA 15461). Reed fragments reveal a brackish deposit environment, confirmed by the δ13C value yielding -28‰. This deposit was considered to have formed between HAT and MHWNT and permitted to reconstruct a RSL position of -4.69±1.31 m below contemporaneous ones for the 4900±68 cal. BP period (Index no. 1). The second core (Guis-S2) permitted to retrieve 0.65 m of brown peat before reaching the underlying loess formation. Base of the peat was dated to 6872±80 cal. BP (UBA 15460) at -3.52 m NGF. Its brackish character is attested by a heavy δ13C value of -23.4‰. Hence, the palaeo-RSL at 6872±80 cal. BP can be positioned at -7.06±1.27 m under the present MHWST level (Index no. 3).
29An auguring was drilled on the foreshore of the Sainte Marguerite dunes, a N 315°-exposed site, located between the mouths of two large rivers (Aber-Benoît and Aber-Wrac’h). Local present tidal levels (table 1) were directly obtained from the Aber-Benoît tidal-reference situated ~3 km from the site (HAT=6.75 m NGF, MHWST=4.57 m NGF, MHWNT=2.13 m NGF). An approximately 0.6-m thick peat deposit was reached at -3 m NGF, under ~1 m of sand. This basal peat lies over a Pleistocene loessy silt level of unknown thickness. It consists of a highly fibrous peat, principally composed of embedded reed-leaves fragments, suggesting a brackish swamp sedimentation environment. Base of the peat layer yielded an age of 6845±69 cal. BP at -3.6 m NGF. This date is thus contemporaneous with that of the peat layer of the Curnic beach. As already revealed by the reeds remains embedded in the peat, the brackish conditions and a probable high-marsh environment of deposition are confirmed by a δ13C value of -26.9‰. Palaeo-RSL position on the site ~6845 cal. BP was -6.95±1.27 m under present MHWST (Index no. 4).
30Porsmilin is a cove situated in the Bertheaume Bay. The site comprises a small sandy beach limited landward by a small dune barrier which somewhat isolates a presently filled marsh. On the foreshore of the beach a large step of peat crops out, covered with fallen tree stems. The exceptionally good state of conservation of the deposit in such a situation (S to SW-oriented beach, highly exposed to storm swells), was an incentive to retain Porsmilin as a study site in the framework of the COCORISCO project. Five piston-cores were carried out from the foreshore to the bottom of the back-barrier but only the lowermost core (Pors-C5) reached Holocene basal deposits. Local present tidal levels (table 1) were determined according to the Trez-Hir tidal reference situated at ~2 km west of the site (HAT=3.99 m NGF, MHWST=3.17 m NGF, MHWNT=1.67 m NGF). Core PorsC5 was made from a peat-step surface cropping out at the bottom of the beach. The weathered basement was reached at an approximate depth of 3.25 m below the surface. On top lies 25 cm of Pleistocene sandy silt, which becomes progressively organic in its upper part before turning in a 0.15 m thick black peat deposit at around -3.5 m NGF. A complex high frequency layering of organic-mud and peat patterns makes up the rest of the sequence. The base of the basal peat deposit was dated on organic content to 6863±127 cal. BP at -2.45 m NGF. The high δ13C value (-21‰) seems to clearly attest for a brackish-influenced intertidal environment (e.g., lagoonal marsh). Poacea and Cyperacea pollen spores were found to dominate the plant palynological assemblages (A. Fernane, pers. comm.). Whilst species were not differentiated, this could correspond to brackish adapted plant species such as Spartina altinerflora, Phragmitus australis and Juncus roemarianus, respectively. If it appears that this deposit may have formed in brackish conditions, we consider that we lack sufficiently reliable indicators to use it as a SLIP. It was therefore taken as a high-limiting deposit, constraining the former MHWST level to lay at a maximum of 5.62 m below present one around 6863±127 cal. BP.
31Treffiagat is a flat south-oriented site situated in the south of our study zone. A large dune-barrier isolates a back-barrier swamp today still awash with water. On the foreshore just in front of the dune an important peat body is cropping out, eroded seaward by a micro-cliff. Two piston-cores were taken near the swamp in the back-barrier zone, respectively at 2.2 m NGF and 1.75 m NGF. Local present tidal levels (table 1) were determined according to the tidal reference of Le Guilvinec harbour, situated directly to the west of the site (HAT=4.20 m NGF, MHWST=3.16 m NGF, MHWNT=1.56 m NGF). The observed sequence begins at -1.25 m NGF with ~0.3 m of a probably Pleistocene silty-mud complex overlaying the weathered basement. This deposit contains many traces of oxidation and undated roots remains, attesting of its surface exposure. Above, lies a 30-cm thick organic silty-peat layer, progressively turning into a wooden black peat above ca. -0.7 m NGF. This peat layer is capped by ~1.80 m of sand-barrier medium to coarse sand. A date was obtained on organic contents at the transition between the Pleistocene silty-mud and the silty-peat (-0.95 m NGF), giving an age of 4909±69 cal. BP. No arguments exist to date in favour of a brackish or freshwater deposit environment. This deposit was hence considered as high-limiting, and led to assess that former mean MHWST level was at least -3.61 m below contemporaneous MHWST ca. 4900 cal. BP (Index no. 6).
32The site of Kermor/Tudy is situated in the Bénodet cove. The site is composed of a large dune barrier oriented N 135° that isolates a back-barrier lagoon nowadays polderized. Several drillings were taken in the polder in the back-barrier zone. Only one drilling taken at the eastern end of the site, near to the barrier root, permitted to reach basal peat deposits. Local contemporaneous tidal levels (table 1) were determined accordingly to the tidal reference for the harbour of Loctudy situated directly northwest of the site (HAT=3.09 m NGF, MHWST=2.45 m NGF, MHWNT=1.35 m NGF). Core K-X reached the weathered geological substrate at a depth of 4 m, the top of the core being situated at 1.3 m NGF. The base of the sequence is composed by a 0.1-m thick sandy silt layer, progressively evolving into a black peat until ca. -2.6 m NGF. Superimposed on this peat level is 0.8 m of very organic stratified silt horizons. On top of a sharp sedimentary hiatus appears a Hydrobia ulvae-dominated marine mud layer. This layer becomes more and more organic to the top (grey to brown mud, with less shell fragments, implying a probable salt-marsh facies) until ca. -1 m NGF. The uppermost part of the core is composed of 1.3 m of brown to green clay, on which rests a dune sand level. The basal peat was dated on organic content to 5536±70 cal. BP. at -2.49±0.22 m NGF, ~10 cm above the contact between the sandy silt and the peat. No clear argument could help us to determine the original deposit environment. Hence, this sample was considered as a high-limiting point. It constrains the former MHWST level to a lay at a maximum of -4.94±0.22 m under present one (Index no. 7). Due to the elevation uncertainties induced by the sampling method, it cannot be fully attested that this deposit is a “pure” compaction-free basal peat. It must so be kept in mind that it could have undergone an unknown compaction-induced lowering.
33Guidel-Loc’h is the southernmost site of this study. It consists in a south 225°-oriented ria, polderized since the 19th c. Two screw drillings were completed along a SW/NE transect across the Guidel-Loc’h pond: GLG-1 was drilled in the middle of the pond, while GLG-2 was drilled about 500 m seaward. Only GLG-1 reached a basal peat deposit. Local contemporaneous tidal levels were determined according to the tidal reference of Le Pouldu situated at ~2 km north of the site (HAT=2.78 m NGF, MHWST=2.23 m NGF, MHWNT=1.18 m NGF). GLG-1 is an 11-m deep drilling, reaching the micaschist, weathered basement at -8.3 m NGF. The base of the sequence consists of ca. 0.2 meter of lacustrine clay encroaching the weathered, micaschist basement. It may represent a lacustrine environment that was emplaced as a result of the hampered drainage in the RSL rise context or a Pleistocene floodplain deposit. The centre of this layer was dated to ca. 7900 cal. BP in GLG-2 drilling, but we also consider probable that this datation could have been rejuvenated by roots penetration. The clay evolves upwards into a 20 cm silty peat layer, overlaid in unconformity by 2.5 m of silty sand. The sequence continues with a succession of 7 m of a sand and muddy-silty layers interrupted by several organic-rich salt-marsh soils. The roof of the sequence is composed of 1 m of sandy marine deposits. The basal-peat and the underlying silt deposit was dated to 7523±64 cal. BP on organic fragments embedded in the peat matrix at -7.87±0.22 m NGF. In the absence of microbiological or geochemical indicators, this layer could only be considered as a high-limiting deposit. It led to assess a former maximum MHWST level of -10.1±0.22 m below the present one (Index no. 8). As it cannot be ruled out that the underlying clay deposit is totally compaction-free and because the peat layer was not sampled at its very base, a potential shortcoming exists in the representativeness of this deposit as a basal-deposit. Thus, again, it cannot be considered as a fully reliable upper-constraint of the RSL and must be only considered as indicative.
34In order to allow comparisons between our results and previously published ones, former RSL results had to be re-assessed. Such a re-assessment requested older data to be re-processed through the same method we used.
35M.-T. Morzadec-Kerfourn (1974) constructed a Holocene RSL curve for northwestern Brittany based on 15 radiocarbon dates. Results were presented in an age/altitude diagram. Though the question of the importance of altitude and age error terms was originally addressed in the text, these error terms were taken into account neither in the data analysis nor in the curve construction. An estimation of possible errors must therefore be re-addressed. The deposits were observed and sampled either on outcrops or by the use of a hand-drilling instrument. Altitudes were presented in m NGF, even if no details were given on the levelling method used. The deposit environments of the layers were originally determined by M.T. Morzadec-Kerfourn using palynological assemblages and absence/presence of dynocysts. The deposits were hence described as either fresh or brackish and related to a relative a sea-level position. Past RSL positions were determined and linked to actual HWST level on the study sites. Error terms on the indicative ranges of brackish peat deposits are lacking. In addition, archaeological megaliths were used in order to refine past HWST maxima. Radiocarbon datings were performed on wood fragments and bulk deposits and presented as conventional carbon ages.
36In the absence of information concerning the levelling methods used by M.-T. Morzadec-Kerfourn (1974) and the potential uncertainty of those used in the 1970s, a conservative ±0.5 m vertical error term was tied to the elevation of sampled deposits. Indicative meanings of M.-T. Morzadec-Kerfourn were re-assessed following the approach used in the present study. As error terms on the indicative ranges of brackish peat deposits were found lacking in the original paper, those were fixed, as above, as equal to (HAT-HWNT)/2, as can be measured on present-day tidal ranges on the sites. Above cited error terms are listed in table 2. All dates were recalibrated with respect to modern standards. After re-calibration, the dates close to the boundaries of the study period show a spectacular mean shift of 300 years, even reaching 840 years for GIF-766 (index-point number 12, 6610±323 cal. BP). All the dates retained from the work of M.-T. Morzadec-Kerfourn (1974) are presented in table 4, accompanied by their original summarised stratigraphic, palynological and micro-faunal data.
37Four new basal SLIPs, together with two others re-assessed from previously published work (Morzadec-Kerfourn, 1974) and six limiting dates are used to reconstruct long-term (millennial) trends of western Brittany RSL history from ca. 7600 and ca. 4000 cal. BP. Total RSL rise reaches ca. 10±0.2 m since ca. 7500 cal. BP (fig. 3). The RSL rise seems to slow down ca. 7000 cal. BP, to an average value of ca. 0.92 mm/a until ca. 4500 cal. BP. A gap in the data exists between ca. 6300 and 5000 cal. BP, period from which only one basal peat could be retrieved within the sequences we worked on: Limiting date no. 7 constrains the rising tendency and indicates a RSL being -4.7 m below present-day HWST around 5500 cal. BP but it cannot be fully attested that this point has not undergone some lowering due to compaction. No SLIPs were obtained from basal-peat data after ca. 4500 BP (fig. 3).
Fig. 3 – Former Holocene RSL positions derived from basal peat deposits dated ca. 7600-4000 cal. BP in the Finistère region.
Fig. 3 – Positions passées du niveau marin relatif holocène dans le Finistère obtenues à partir des tourbes de base pour la période entre environ 7600 et 4000 cal. BP.
A detailed legend is included in the figure. Numbers correspond to table 3 and table 4. Dotted lines indicates points for which potential compaction cannot be ruled out.
Légende incluse dans la figure. La numérotation correspond au tableau 3 et au tableau 4. Les pointillées indiquent les points pour lesquels une potentielle compaction ne peut être exclue.
38P. Stephan (2011) published a Holocene RSL curve for the Bay of Brest, based on 15 SLIPs retrieved from three cores obtained in saltmarshes. Indicative meanings were originally determined by a transfer function, constructed from two present-day salt-mashes of the Bay of Brest. Log and reconstructed palaeo RSL are presented in fig. 4. Dates were recalibrated with respect to our standard to allow comparisons. As underlined by P. Stephan (2011), all dates he used were obtained on bulk deposits and hence may have suffered dating problems.
Fig. 4 – Sedimentary sequences studied in the Bay of Brest by P. Stephan (2011) and associated RSL positions.
Fig. 4 – Séquences sédimentaires étudiées par P. Stephan (2011) en Rade de Brest et positions associées du niveau marin relatifs.
Modified from P. Stephan (2011). Data as originally published. 1: peat; 2: silty-peat 3: organic mud; 4: marine mud; 5: shelly mud; 6: bed-rock (granite); 7: bed-rock (shale).
Modifié de P. Stephan (2011). Données telles qu’originalement publiées. 1 : tourbe ; 2 : tourbe argileuse ; 3 : vase organique ; 4 : vase marine ; 5 : vase coquillère ; 6 : substrat granitique ; 7 : substrat schisteux.
39Stephan’s SLIPs plot in relatively good accordance with our basal SLIPs (fig. 5). Vertical discrepancy between our data and P. Stephan’s one can be explained by the fact that 13 of the 15 dates published by this author were obtained from intercalated deposits particularly prone to compaction, while it could not have been the case for the basal peats we worked on. On the meantime, two of the 15 SLIPs published by P. Stephan (2011) are also basal deposits and it can be observed that these also plot quite low regarding our data. The ca. 2800 BP point was obtained from the base of a basal mud deposit full of fragmented shells. We believe that this point may have been reworked and deposited in a highly-morphogenic period, probably after an erosional event that artificially created an accommodation space and induced a lowering of the subsequent sedimentary infill. If this is the case, it casts doubts on its usefulness for RSL reconstruction purpose. We therefore considered this point as unreliable and chose to reject it for RSL reconstruction purpose. Indeed, ca. 3000 BP period is widely recognised in northwestern Europe (Lamb, 1977; Long and Hughes, 1995; Moura et al., 2007) as a period of highly increased storminess during which large perturbations on the sedimentary sequences may have occurred. The 5500 cal. BP point is more doubtful. As noted by P. Stephan (2011), the foraminifera assemblage obtained for this point had no modern analogues and thus reliability of the RSL position produce by the foraminifera transfer function may have been biased. Moreover, it was dated on bulk peat material, it cannot be ruled out this date was not rejuvenated by roots penetration.
40Large uncertainties remain on RSL changes during the ca. 6300-5000 cal. BP period. Indeed, considering the doubtful basal SLIP presented above, RSL evolution during the former period is only constrained by three SLIPs and one limiting date, which contribute to give to the RSL envelope a somewhat stabilising to character that cannot be at all ascertained, due to the lowering effect compaction surely had on these points. Further work is clearly needed in order to precise the RSL behaviour within this time-period.
41In spite of the above mentioned uncertainties and when the maximal extents of the SLIPs error boxes and limiting dates are taken into account, simplified RSL sub-trends can reliably be drawn between ca. 7500 and 7000 cal. BP and between ca. 7000 and 3000 cal. BP, corresponding to mean RSL rise rates of ca. 8 mm/a and 1.4 mm/a, respectively (fig. 5). After ca. 3400 cal. BP, P. Stephan data give an average rising rate of ca. 0.31 mm/a. We stress the fact that these average “sub-rates” must only be considered as indicative values, as they do not take quantitatively into account the errors tied to both the basal SLIPs and the high-limiting points and because of the uncertainties around the periods of RSL rise rates changes. RSL data still lack definition ca. 6000 cal. BP and ca. 3000 cal. BP to undoubtedly state on the RSL behaviour around these periods. Indeed, it cannot be ruled out that “stair-steps” of either stabilising or very-slowly rising RSL did not take place in between the more clearly defined transgressive periods. We speculate that barrier (and hence back-barrier) environments may have been particularly impacted during those decreasing RSL rising rates periods, surely bearing consequences on the RSL signal recorded as it was noted by A.C. Massey et al. (2008). Particularly, we speculate that compaction processes surely had major impacts on back-barriers lagoonal deposits during barriers landward shifts.
Fig. 5 – Comparison of long-term RSL reconstruction based on basal peats (this study) and data obtained by P. Stephan (2011) from foraminifera analyses.
Fig. 5 – Comparaison entre les données de niveau marin relatif obtenues dans cette étude à partir des tourbes de base et celles obtenues par P. Stephan (2011) à partir de l’étude des foraminifères.
Numbers correspond to table 3 and table 4. Dotted lines indicated that some potential compaction cannot be ruled out.
La numérotation correspond au tableau 3 et au tableau 4. Les pointillées indiquent les points pour lesquels une potentielle compaction ne peut être exclue.
42New data were obtained on Holocene RSL for the western Brittany region and allowed RSL history to be reconstructed on a longer period than was previously possible in the region. These new data are in good agreement with the data published in previous studies, after the latter have been re-assessed. Our work gives a new and more precise overall vision on RSL changes in western-Brittany during the Holocene.
43Basal peat deposits obtained from several sites around the Finistère peninsula show long-term millennial trends of RSL during the study period. RSL was situated around -10 m under present-day level ca. 7500 cal. BP, between -7 and -5 m ca. 7000 cal. BP, and around -1 m ca. 3200 cal. BP. The study of basal peat only allowed a long-term RSL reconstruction, but this reconstruction nevertheless reliably shows that the rate of the sea-level underwent a dramatic decrease ca. 7000 cal. BP and a further, if more minor, surely somewhere between 4000 and 3000 cal. BP. The comparison with foraminifera-based RSL data obtained in the region shows the good accordance between the datasets. Elevation discrepancies between both datasets can surely be partly attributed to compaction having lowered the elevation of non-basal Holocene deposits. Hence, it allows to gain in point density and to reach a better resolution for RSL reconstruction, as foraminifera-based RSL data available for the region obviously lacked basal compaction free-deposits while our data take advantage of the better-constrained elevation of the foraminifera-based SLIPs. It tends to confirm that RSL rise occurred at progressively decreasing rates before ca. 7000 cal. BP, between ca. 7000 and ca. 3000 cal. BP, and after ca. 4000-3000 BP. Doubts remain on the turning points of RSL rise changes because of pluri-centennial gaps existing between the dates around these periods. Dating-related problems are also suspected for some of the index-points defining those changes. It can be suspected that periods of diminishing rate in RSL rise each time provoked deep modifications in the functioning of the back-barrier sedimentary systems from where most of the data were retrieved.
44Further work is in progress to incorporate additional data obtained in several sites along the Brittany. When completed, this work will allow to better constrain the RSL history of western Brittany and will be of great importance to better understand the morphosedimentary dynamics that took place during the Holocene in reaction to the different changes in the RSL rise rates.