major storm periods and climate forcing in the western

lable at ScienceDirect

Quaternary Science Reviews 129 (2015) 37e56

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Major storm periods and climate forcing in the WesternMediterranean during the Late Holocene

Jean-Philippe Degeai a, *, Benoît Devillers a, Laurent Dezileau b, Hamza Oueslati a,Gu�ena€elle Bony a

a CNRS UMR5140 ASM, Universit�e Montpellier III, MCC, 34970 Lattes, Franceb Universit�e de Montpellier, G�eosciences UMR5243, 34095 Montpellier, France

a r t i c l e i n f o

Article history:Received 5 June 2015Received in revised form4 September 2015Accepted 1 October 2015Available online xxx

Keywords:Storminess activityLagoonal sequenceClimate forcingSolar activityLate HoloceneWestern Mediterranean

* Corresponding author.E-mail address: [email protected] (J.-P.

http://dx.doi.org/10.1016/j.quascirev.2015.10.0090277-3791/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Big storm events represent a major risk for populations and infrastructures settled on coastal lowlands.In the Western Mediterranean, where human societies colonized and occupied the coastal areas since theAncient times, the variability of storm activity for the past three millennia was investigated with a multi-proxy sedimentological and geochemical study from a lagoonal sequence. Mappings of the geochemistryand magnetic susceptibility of detrital sources in the watershed of the lagoon and from the coastalbarriers were undertaken in order to track the terrestrial or coastal/marine origin of sediments depositedinto the lagoon. The multi-proxy analysis shows that coarser material, low magnetic susceptibility, andhigh strontium content characterize the sedimentological signature of the paleostorm levels identified inthe lagoonal sequence. A comparison with North Atlantic and Western Mediterranean paleoclimateproxies shows that the phases of high storm activity occurred during cold periods, suggesting aclimatically-controlled mechanism for the occurrence of these storm periods. Besides, an in-phase stormactivity pattern is found between the Western Mediterranean and Northern Europe. Spectral analysesperformed on the Sr content revealed a new 270-year solar-driven pattern of storm cyclicity. For the last3000 years, this 270-year cycle defines a succession of ten major storm periods (SP) with a meanduration of 96 ± 54 yr. Periods of higher storm activity are recorded from >680 to 560 cal yr BC (SP10,end of the Iron Age Cold Period), from 140 to 820 cal yr AD (SP7 to SP5) with a climax of storminessbetween 400 and 800 cal yr AD (Dark Ages Cold Period), and from 1230 to >1800 cal yr AD (SP3 to SP1,Little Ice Age). Periods of low storm activity occurred from 560 cal yr BC to 140 cal yr AD (SP9 and SP8,Roman Warm Period) and from 820 to 1230 cal yr AD (SP4, Medieval Warm Period).

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Cyclonic activity represents a major natural hazard to pop-ulations inhabiting coastal areas, and especially in highly populatedregions, as demonstrated by the catastrophic super-typhoonHaiyan in Philippines (Lagmay et al., 2015). Such dramatic eventshighlight the debate on the possible influence of climate changeand global warming on the increasing intensity and frequency ofcyclones (Gaertner et al., 2009; Mann et al., 2009; Knutson et al.,2010; Mendelsohn et al., 2012; Nissen et al., 2014). Some regionsin the world are particularly vulnerable to this cyclonic risk, like inthe Gulf of Mexico or in Southeastern Asia with respectively

Degeai).

hurricane and typhoon events. In the Mediterranean, wherecoastlines have generally been densely inhabited since the Ancienttimes, the most powerful cyclogenesis corresponds to explosivecyclones (Kouroutzoglou et al., 2014, 2015), tropical-like storms(Fita et al., 2007), or, more rarely, to hurricane-like events, the so-called medicanes (Emanuel, 2005; Flaounas et al., 2015). Numeri-cal models used to predict the occurrence and recurrence of thesestorm episodes are based upon past records. Instrumental datasetson the last decades or historic documentations on the last centuriesallow the study of the annual to multi-decadal storminess fre-quency (e.g. Bartholy et al., 2009), but to assess the centennial tomillennial cycles, it is necessary to use information from sedi-mentary sequences which have recorded the paleostorm events.This field of research, the palaeotempestology, was successfullyapplied to evidence the millennial frequency of cyclones during themid-to late-Holocene in the western North Atlantic (Donnelly and

Delta:1_given name

Delta:1_surname

Delta:1_given name

Delta:1_surname

Delta:1_given name

mailto:[email protected]

http://crossmark.crossref.org/dialog/?doi=10.1016/j.quascirev.2015.10.009&domain=pdf

www.sciencedirect.com/science/journal/02773791

http://www.elsevier.com/locate/quascirev

http://dx.doi.org/10.1016/j.quascirev.2015.10.009



J.-P. Degeai et al. / Quaternary Science Reviews 129 (2015) 37e5638

Woodruff, 2007), Northwest Florida (Liu and Fearn, 2000; Laneet al., 2011; Das et al., 2013), the Northeastern United States(Parris et al., 2010), the Central Pacific (Toomey et al., 2013),Southern Japan (Woodruff et al., 2009), Western Australia (Nott,2011), Northeastern New Zealand (Page et al., 2010), NorthernEurope (Sorrel et al., 2009, 2012), or the Northwestern Mediterra-nean (Dezileau et al., 2011; Sabatier et al., 2012).

The new storminess proxies established in these works weretentatively correlated with ice-rafted debris (IRD) indices and re-cords of sea-surface temperature (SST) or solar irradiance in orderto study the relationship with climate dynamics (Donnelly andWoodruff, 2007; Sabatier et al., 2012; Sorrel et al., 2012; VanVliet-Lano€e et al., 2014). For example, in the Northwestern Medi-terranean, the millennial cyclic evolution of the stormy episodeswas tentatively proposed to be explained by the variability of theNorth Atlantic Oscillation (NAO) during Holocene Cold Events(HCE), although relationships and models are complex and still nottotally satisfactory (Sabatier et al., 2012). Indeed, the periods of highstorminess activity in the North Atlantic and western Europe dur-ing the last millennium have been reported either to positive ornegative NAO phases (Trouet et al., 2012), while the NAO indicesreconstructed for the last millennia show some discrepancies ac-cording to the paleoclimate proxies used in the reconstructions(Trouet et al., 2009; Olsen et al., 2012; Ortega et al., 2015). Besides,less information is available on the frequency of storm events on ashorter centennial timescale (Muscheler, 2012; Sorrel et al., 2012).

During a storm event, a sedimentary layer consisting of detritalmaterial reworked from the coastal area or the inner shelf, the so-called tempestite, can be deposited in the terrestrial topographicdepressions along the shoreline (Budillon et al., 2005; Sabatieret al., 2008; Horton et al., 2009; Hawkes and Horton, 2012).Consequently, the coastal lakes and wetlands such as lagoons ormaritime marshes at the back of coastal barriers provide a relevantgeomorphic setting to track the paleostorm activity (Sabatier et al.,2008, 2010a; Woodruff et al., 2009; Dezileau et al., 2011; Lane et al.,2011; Otvos, 2011). Different methods have been used to detectthese tempestites in a sedimentary sequence: biological indicatorssuch as foraminifera (Collins et al., 1999; Hippensteel and Martin,1999; Hawkes and Horton, 2012; Pilarczyk et al., 2014), diatom(Parsons, 1998; Page et al., 2010), molluscs (Jelgersma et al., 1995;Sabatier et al., 2008, 2012), or pollen (Liu et al., 2008); sedimen-tological characteristics such as grain-size (Liu and Fearn, 2000;Sabatier et al., 2008, 2012; Horton et al., 2009; Parris et al., 2010;Dezileau et al., 2011; Toomey et al., 2013), mineralogy (Sabatieret al., 2010a, 2012), or microtextural features of quartz grains(Costa et al., 2012); and elemental or isotopic geochemistry(Lambert et al., 2008; Woodruff et al., 2009; Page et al., 2010;Sabatier et al., 2010a, 2012; Dezileau et al., 2011; Das et al., 2013).

This paper focuses on the study of paleostorms from high-resolution geochemical and sedimentogical analyses of a lagoonalsequence in the Northwestern Mediterranean. The main objectivesare: (1) find a combination of proxies detecting the palaeostormevents in the sedimentary sequence; (2) analyse the time ofrecurrence of these events during the last millennia at differenttimescales (centennial to millennial); (3) compare our results withother regional and global climate proxies in order to understandthe causes of the variability of storm periods.

2. Geological setting

The study area is located in the Languedoc region, along thecontinental shelf of the Gulf of Lions in the Northwestern Medi-terranean (Fig. 1A). The many lagoons in this coastal plain give anexcellent opportunity to find sedimentary sequences recording thepalaeostorm events. The Languedoc plain is surrounded by the

Mesozoic tabular karstic plateau of the Larzac to the northwest andthe Hercynian crystalline to metamorphic basement of the Cev-ennes to the northeast (Fig. 1B). A phase of subsidence occurred inthe Languedoc plain during the Oligo-Miocene period when therifting that led to the opening and oceanization of the westernMediterranean affected the passive margin of the Gulf of Lions(Barruol and Granet, 2002; D�ezes et al., 2004; Bache et al., 2010).

During the Quaternary sea-level falls, valleys incised the innercontinental shelf and the coastal lowlands (Tesson et al., 2005;Larue, 2008; Raynal et al., 2009; Labaune et al., 2010). The lastpost-glacial sea-level rise on the continental shelf began ca. 18 kyrago (Lambeck and Bard, 2000; Rabineau et al., 2006). From themid-Holocene, the decelerating eustatic sea-level rise along the shore-line of the Gulf of Lions induced the construction of sandy coastalbarriers at the back of which formed lagoons (Barusseau et al.,1996; Certain et al., 2005; Tesson et al., 2005; Raynal et al., 2009;Court-Picon et al., 2010). Amongst these ones, the lagoon of theBagnas is located in the southern termination of the Thau lagoon,between the cities of Agde and Marseillan (Fig. 1C). This is a 2 km-long on 1.5 km-width semi-elliptical brackish and shallow (lessthan 1 m-depth) body of water almost completely filled with sed-iments, transforming from a lagoon into a maritime marsh. Thewatershed basin (ca. 10 km2) extends mainly to the northwest ofthe Bagnas pond and is drained by the 4-km long Bragues Riverflowing in a NNW-SSE direction. The maximal elevation on thewatershed peaks at 114 m at the summit of the scoria cone of theMont Saint-Loup in the southern corner of the catchment area. Theelevation does not exceed 40 m in the most part of the watershedbasin. This is covered essentially with Pleistocene alluvial terraces(Fig. 1D), which are composed of quartz and basalt gravels andpebblesmixed in a brownish to reddish silty clayeymatrix. Volcanicterrains outcrop on the northern flank of the Mont Saint-Loup witha basaltic lavaflow fossilizing phreatomagmatic deposits super-imposed by scoriaceous material.

The Bagnas pond is partly surrounded with marshy depositsprogressively filling the basin. To the Southeast, beaches and dunesin the littoral zone form a sandy coastal barrier with a maximalwidth of 500m (Fig.1D). The geomorphic analysis of the subsurfacecoastal topography and field surveys reveal the presence of wash-over fans associated with storm events at the back of the littoralsandbar southeast to the Bagnas pond. All these coastal landformsbelong to the Holocene highstand sea-level transgressive systemtracts which started to extend in this area between 6500 and4500 cal yr BP (Labaune et al., 2008).

3. Material and methods

3.1. Sampling

A 9 m long core (B1, Fig. 2) was taken on a levee in the Bagnaspond using a drillingmachine equippedwith a hydraulic piston anda 1 m-long cylindrical corer with a 80 mm-diameter cutting shoe.Supplementary cores were taken in the floodplain of the H�eraultRiver with a percussion corer to document the sedimentologicalproperties of the alluvial deposits from the H�erault River (Fig. 1C).Additionally, 62 samples of sediments were collected from thewatershed of the Bagnas lagoon and from the coastal beach anddune barriers in order to study the composition of the detritalmaterial transported then deposited in the Bagnas depression. Thisstudy is focused on sedimentological and geochemical analysesowing to the low abundance of micro-fauna in some parts of the B1sequence, preventing to obtain consistent high-resolution contin-uous records on the whole sequence with micro-palaeontologicalproxies.

Fig. 1. Location maps and geological setting. (A) Location of the studied area in the Northwestern Mediterranean. (B) Location of the main lagoons in the coastal plain of Languedoc.(C) Topography of the Bagnas watershed and its surroundings. (D) Geology of the Bagnas watershed and its surroundings.

J.-P. Degeai et al. / Quaternary Science Reviews 129 (2015) 37e56 39

Fig. 2. Stratigraphic log, AMS 14C age-depth model, and age reservoir R(t) vs. atmospheric 14C age of the Bagnas B1 sequence. Coloured parts of the calibrated age probabilitydistributions correspond to the 2-s confidence level. The parts of the calibrated age probability distributions in red correspond to the upper parts of the 2-s range not considered inthe age-depth model (see explanation in the text). The vertical grey band on the right of the age-depth model represents the period during which the site was emerged. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)


3.2. Sedimentology and geochemistry

Elemental geochemical analyses by energy-dispersive X-rayfluorescence (ED-XRF) spectrometry were undertaken with a DeltaInnovX spectrometer and the workstation A-020-D. Measurementswere realized with the 3-beam soil analytical mode on the water-shed samples and each 2 cm on average for the B1 sequence. Theparameters of voltage, amperage, and filter for this analytical modewere set as following: 40 kV,100 mA and filter #3 for the first beam;40 kV, 100 mA and filter #2 for the second beam; and 10 kV, 200 mAand filter #5 for the third beam. Counting times were respectively15 s for the two first beams and 20 s for the third beam. Elementsabove the limit of detection are Cl, K, Ca, Ti, Cr, Mn, Fe, Co, Zn, As, Rb,Sr, Zr, Ba, Pb. International powder standards (NIST2702 andNIST2781) were used to assess the analytical error and accuracy ofmeasurement, which are lower than 5% for Ti, Cr, Fe, Zn, Pb, be-tween 5 and 15% for Ca, Mn, As, Rb, Sr, and between ca. 15 and 25%for K and Co. Tests of repeatability show a RSD error on the preci-sion of measurement lower than 1%.

Analyses of magnetic susceptibility (MS) were undertaken witha BartingtonMS2meter and either the MS2B dual frequency sensorfor the watershed samples or the MS2F probe for the B1 coresequence. Measurements were achieved each 2 cm on average forthe B1 sequence. Volume magnetic susceptibility (k) was measuredin SI unit at an operating frequency of 4.65 kHz and a measurementperiod of 15 s SI (range 0.1) with the MS2B sensor, and at anoperating frequency of 0.58 kHz and a measurement period of 11 sSI (range 0.1) with the MS2F probe. Precision of measurement wasassessed by tests of repeatability: RSD errors are lower than 1% forsamples with MS higher than 4.10�5 SI and between 1 and 5% forsamples with MS lower than 4.10�5 SI. Magnetic susceptibility andgeochemical analyses of the B1 core have been measured on the

bulk sediment covered with an ultrafine polyethylene film.Laser grain-size analyses were achieved with a Beckmann-

Coulter LS13320 Particle Size Analyser (Geosciences Montpellier).Grain-size analyses were performed on the B1 sequence with anaverage interval of 10 cm. For each sample, a small homogeneousamount of sediment was mixed in deionized water then sieved at1.5 mm diameter before pouring in the Fluid Module of the ParticleSizer until to obtain an optimal obscuration rate between 7 and 12%in the Fraunhofer optical cell. The time of background and samplemeasurement was set to 90 s and sonicationwas applied during themeasurement of the sample in order to improve the dispersion offine particles in the fluid. Each sample was measured twice and thegood repeatability of measurement was verified according to thestatistics from the international standard ISO 13320-1. The RSDerror was lower than 3% for the median and lower than 5% for thedifferences between the quantiles q10-q30 and q70-q90.

3.3. Chronological framework

The chronology of the B1 core was established using AcceleratorMass Spectrometry (AMS) 14C radiocarbon dates. Twelve 14C dateson in-situ mollusc shells and terrestrial material (wood and char-coal) have been used to establish the age-depth model (Table 1).The dated shells were monospecific bivalve lagoonal species(Pavicardium Exiguum) sampled in fine sediments and with theirtwo valves connected in order to minimize the risk of samplingreworked shells. Radiocarbon ages on shells were calibrated bytaking into account the 14C reservoir age R(t) of the Bagnas lagoon.This reservoir age, which includes the local reservoir age DR(Stuiver and Braziunas, 1993; Siani et al., 2000; Hughen et al.,2004), was modelled using five chronological control points (CP1to CP5, Fig. 2), making the B1 core one of the best-dated sequences

Table 1AMS 14C dates and reservoir ages of the Bagnas B1 sequence. (*) 2-s range not taken into account in the age-depth model (see explanation in the text).

Depth (cm) Laboratory code Material 14C age (BP) Reservoir age R(t) in yr BP 2-s ranges of calibrated age

60 Lyon-11181 Pavicardium Exiguum 1620 ± 30 1483 ± 49 1669e1781, 1797e1894, 1905e1946* cal AD75 Lyon-11180 Pavicardium Exiguum 1720 ± 40 1580 ± 56 1668e1782, 1797e1893, 1906e1948* cal AD235 Lyon-11169 wood 120 ± 30 0 1679e1764, 1775e1775*, 1801e1939* cal AD252.5 Lyon-11168 wood 95 ± 30 0 1683e1735, 1806e1930* cal AD265 Lyon-11167 wood 170 ± 35 0 1656e1706, 1719e1819*, 1824e1825*, 1832e1883*,

1914e1949* cal AD275 Lyon-11179 Pavicardium Exiguum 1370 ± 35 1162 ± 49 1641e1693, 1727e1812*, 1919e1949* cal AD365 Lyon-11166 charcoal 555 ± 35 0 1306e1363, 1385e1433 cal AD382.5 Lyon-11178 Pavicardium Exiguum 1350 ± 35 728 ± 49 1290e1401 cal AD480 Lyon-11177 Pavicardium Exiguum 1995 ± 30 690 ± 49 640e781, 787e877 cal AD690 Lyon-11173 Pavicardium Exiguum 2240 ± 30 676 ± 49 385e615 cal AD805 Lyon-11172 Pavicardium Exiguum 2480 ± 35 662 ± 49 67e348, 370e377 cal AD847.5 Lyon-11171 Pavicardium Exiguum 2780 ± 30 644 ± 49 362e40 cal BC


for the determination of the evolution of the Holocene age reservoirin the Mediterranean area.

The R(t) of the control points CP1 and CP2 at 0.6 and 0.75 mdepth (i.e. respectively �0.1 and �0.25 m asl) were obtained bycomparing the conventional 14C age of shells in the uppermost partof the sequence (Lyon-11181 and Lyon-11180, Table 1) with the 14Cage of land emergence above 0 m asl at the drilling site, whichemerged from the lagoon between the years of 1774 and 1821 ADfrom the respective historical archives of themap of the Royal Canalof Languedoc and the Napoleonic cadastre, i.e. between 168 ± 7 and102 ± 6 14C yr BP from the IntCal13 calibration curve (Reimer et al.,2013). The R(t) of the control points CP3 and CP4 was obtained bycomparing the conventional 14C ages of shells with the atmospheric14C ages of terrestrial material vertically close to the depths of CP3and CP4 (respectively the two pairs of datings Lyon-11179/Lyon-11167 and Lyon-11178/Lyon-11166, Table 1). The reservoir age of600 ± 49 yr BP at the 14C age of 2935 ± 35 BP found for the adjacentThau lagoon from Court-Picon et al. (2010) was used as an addi-tional control point (CP5). The reservoir age R(t) of the fourlowermost shells was calculated using a linear regression betweenthe conventional 14C age and the R(t) of the two control points CP4and CP5 (Fig. 2).

The 14C reservoir age R(t) of the Bagnas lagoon increases from600 ± 49 yr BP in the lower part of the sequence to 728 ± 49 yr BP atca. 3.5 m-depth, and then highly spreads until 1580 ± 56 yr BP nearthe top of the sequence (Fig. 2). This increase of R(t) is probably dueto the relative isolation of the lagoon from marine inputs in theupper part of the sequence, as it was already observed in otherlagoons of the Western Mediterranean (Zoppi et al., 2001; Sabatieret al., 2010b). The subtraction of R(t) from the conventional 14C ageof shells yields the atmospheric 14C age of shells (Stuiver andBraziunas, 1993; Reimer and McCormac, 2002; Sabatier et al.,2010b). These ages as well as radiocarbon dates on terrestrial ma-terial were then calibrated using Calib 7.0.4 and the IntCal13 cali-bration curve (Reimer et al., 2013). The upper ranges of the 2-sconfidence level for the youngest calibrated ages Lyon-11167, �11168, �11169, �11179, �11180, and �11181 were nottaken into account in the age-depth model owing to the chrono-logical and sedimentological inconsistencies between the upperpart of the 2-s range of these datings on materials sampled inlagoonal sediments and the above-mentioned historical archivesshowing that the B1 drilling site was out of water from the late1700s/early 1800s AD (Fig. 2).

The age scale of sedimentological proxies was established withAnalySeries v2.0.8 (Paillard et al., 1996).

3.4. Spectral analysis

Spectral analyses were performed in order to study the

periodicity of storm events. Given the advantages and disadvan-tages of each methods of statistical analysis of time series, it isrecommended to use several methods and to compare their results(Berger et al., 1991; Yiou et al., 1996). This procedure allows todiscard spurious results due to biases of one particular method byevaluating the variance and stability of the results from thedifferent methods (Desprat et al., 2003).

Two power spectrum estimation methods were applied in ourstudy: the multi-taper method (MTM) and the maximum entropymethod (MEM). The MTM is a non-parametric method that at-tempts to reduce the variance of spectral estimates from a combi-nation of multiple orthogonal windows (Thomson, 1982; Percivaland Walden, 1993). A set of multiple windows (data tapers) isapplied to the data in the time domain before Fourier transforming(Thomson,1990). This method is independent of the spectral power(Loulergue et al., 2008). For climate data, the MTM has theadvantage of providing a narrowband F-test for the presence andsignificance of periodic components (Thomson, 1990). The MEM isbased on the selection of the spectrum with the highest entropy,which is the least biased estimate possible on the given informa-tion, i.e. the maximally noncommittal with regard to missing in-formation (Jaynes, 1957; Harremoes and Topsoe, 2001). Thismethod is able to fit sharp spectral features and improves thequality of the spectrum, which presents an excellent frequencyresolution (Berger et al., 1991; Dubar, 2006; Pardo-Iguzquiza andRodriguez-Tovar, 2006).

4. Results

4.1. Characterization of detrital sources

A principal component analysis (PCA) was performed on thesediment sampled around the Bagnas pond in order to characterizethe geochemistry and magnetic susceptibility of the potentialsources of sediments deposited in the lagoon (Fig. 3). Four differentsources were distinguished: (1) the coastal beach and dune bar-riers, (2) the Pleistocene alluvial terraces in the watershed of theBagnas, (3) the volcanic products of the Mont Saint Loup, and (4)the Holocene floodplain of the H�erault River. For the fourth source,elemental content and magnetic susceptibility are represented by amean value of 9e15 measurements made on the red to brown siltyto silty clayey fluvial sediments drilled in the floodplain sequence.From our micropalaeontological analyses, these alluvial sedimentsare characterized by a faunal assemblage of terrestrial or freshwatermolluscs (Theba pisana) and ostracods (Candona neglecta, Hetero-cypris salina), and lay on dark grey to blackishmudwith lagoonal ormarine molluscs (Hydrobia ventrosa, Cerastoderma glaucum, Scro-bicularia plana, Rissoa ventrosa) and ostracods (Ilyocypris bradyi,Cyprideis torosa, Loxoconcha elliptica, Loxoconcha tamarindus,

Fig. 3. Principal component analysis of geochemical and magnetic susceptibility data from the Bagnas watershed and the H�erault River floodplain.


Leptocythere lacertosa, Leptocythere fabaerformis, Cytherois fisheri).The PCA dataset consisted of sixteen input variables: the fifteen

elemental contents and the value of magnetic susceptibility. Thedata of each variable were transformed into a standardized andmean normalized distribution. The two first factorial factors plottedon the PCA diagram explain respectively 54.5 and 21.4% of thevariance of the dataset. The first factor shows a good correlationwith strontium and chlorine elements, and an anti-correlationwithmore terrigenous elements as Ti, Co, Fe, Cr, Mn, Zr, Zn, Ba, Pb, Rb, K,and As. This first axis establishes a very good discrimination be-tween the coastal sandbars, which are characterized by highercontents in Sr and Cl, and the terrestrial sediments sampled in thewatershed of the Bagnas lagoon or in the Holocene floodplain of theH�erault River. The second axis distinguishes between, on the onehand, material with high values of magnetic susceptibility andtransition metals as Fe, Ti, Cr, and Co, and, on the other hand,alkaline or alkaline-earth metals as K, Rb, Ca, Ba. On the factorialplane, this second axis splits two groups of samples: a ferromag-netic group of samples collected from the volcanic deposits of theMont Saint-Loup scoria cone, and an alkaline group of samplestaken from the Pleistocene alluvial terraces or Holocene floodplaindeposits.

Mapping of strontium content and values of magnetic suscep-tibility confirms the spatial discrimination of detrital sourcesaround the Bagnas lagoon (Fig. 4). High Sr contents (>350 ppm)characterize the sandbars of the coastal beaches and dunes, and areon average three times higher than their terrestrial counterparts,for which the values of strontium are generally lower than 150 ppm(Fig. 4A). The map of magnetic susceptibility shows diamagnetic orvery low paramagnetic material in the coastal barriers with valuesless than 5� 10�5 SI (Fig. 4B). There are intermediary values in thewatershed northwest to the Bagnas lagoon and in the Holocenefloodplain of the H�erault River, while high values exceeding

100� 10�5 SI are found on the northeastern flank of the scoria coneof the Mont Saint-Loup, where outcrop volcanic deposits charac-terized by their ferromagnetic component. The map of calciumshows intermediate to high Ca contents in the coastal sandbars aswell as in the Pleistocene alluvial terraces from the watershed(Fig. 4C), which are mainly due to the respective presence of shelldebris and detrital calcite. The low spatial discrimination betweenthe watershed and coastal detrital sources from this ubiquistelement is evidenced on the PCA graph by the intermediary posi-tion of the Ca variable between the groups of coastal sandbars andPleistocene alluvial terraces (Fig. 3).

Besides, in the Pierre-Blanche lagoon to the north of the studiedarea (Fig. 1B), the zirconium was reported as issuing from detritalreworking of the coastal sandbar (Sabatier et al., 2010a; Dezileauet al., 2011), while it was supplied preferentially from the water-shed in the case of the Bagnas lagoon (Fig. 4D). These regionaldiscrepancies show the influence of the local lithology and the roleof the longshore drift on the lagoonal sedimentation, and highlightthe need to perform a systematic sedimentological study of detritalsources in order to find reliable mineralogical or geochemicalproxies for storm deposits.

4.2. The B1 sequence

4.2.1. Stratigraphy and sedimentation ratesTwo main phases can be distinguished in the evolution of the

lagoonal sedimentation of the B1 core (Fig. 2). The lower part of thesequence from 9 to ca. 6 m-depth (ca. 2650e1350 cal yr BP) isdominated by greenish grey clay and light grey sand with generallyhigher chlorine content representing probably a lagoon with ma-rine influences. Above 6 m depth, the reddish-brown clay and siltyclay in the middle and upper parts of the sequence evidence aclosed lagoon with increasing terrestrial inputs. The recent

Fig. 4. Maps of strontium, calcium, zirconium contents and magnetic susceptibility values around the Bagnas pond.


terrigenous top-soil above the current sea level in the first half-meter of the sequence was not taken into account to draw thesedimentological proxies.

The sedimentation rates show four main steps in the evolutionof the pace of the lagoonal sedimentary filling. From ca. 700 cal yrBC to 200 cal yr AD, the accumulation rate is lowwith a mean valueof 1 mm.yr�1. The accumulation rate then increases at about6 mm.yr�1 from 200 to ca. 750 cal yr AD. Next, there is a decrease at2.3 mm.yr�1 from 750 to ca. 1700 cal yr AD. The highest sedimen-tation rates of the sequence (>18 mm.yr�1) are then found in theupper 3m during the 18th century AD before the land emergence ofthe drilling site, contemporaneously with an increase in the fre-quency of flood events in the H�erault River (Blanchemanche, 2009).

4.2.2. Grain-sizeStorm surge deposits are characterized by coarser particle size

indicating an increased flow velocity of the wave climate (Hawkesand Horton, 2012). Storm surge deposits consist generally of sandlayers related to a washover fan formed by reworking of particlesfrom the coastal sandbars into the back barrier lagoons where amuddy fine-grained sedimentation accumulates (Liu and Fearn,2000; Donnelly et al., 2004; Sabatier et al., 2008, 2012; Dezileauet al., 2011; Raji et al., 2015). Eroded sand from coastal dunes can

be transported inland as a sandy sheet that thins landwards (Nott,2004). Coarse layers interbedded within the fine-grained lagoonalsediments result from marine flooding events overtopping orbreaching the lido and transporting the barrier and nearshoresediments into the lagoon (Dezileau et al., 2011).

A method based on the calculus of the standard deviation foreach grain size class of the particle size distribution (PSD) was usedin order to identify the storm events (Fig. 5). This method detectsthe part of the PSD with the highest variability and determines thegrain size classes that have the most significant variation throughtime (Sabatier et al., 2008, 2012; Raji et al., 2015). It evidences forthe B1 sequence two main groups of particle population withhigher variability between 4 and 15 mm and between 50 and300 mm, i.e. respectively for fine silt and fine sand (Fig. 5). Thepercentage of particles in the latter group of fine sand was used asproxy to detect the coarser detrital inputs into the lagoon.

4.2.3. Magnetic susceptibilityThe values of magnetic susceptibility (MS) range from slightly

negative values (diamagnetic material) to ca. 50 � 10�5 SI (para-magnetic material) (Fig. 6). The lower part of the sequence until6.5 m depth shows low values of magnetic susceptibility (ca. �1 to5 � 10�5 SI) with two higher MS levels (10e15 � 10�5 SI)

Fig. 5. Grain size of the Bagnas B1 sequence. (A) Standard deviation of the particle sizedistribution (PSD) with white dots representing the part of the PSD with the highestvariability between 4 and 15 mm and between 50 and 300 mm. (B) Root-square fre-quency of the PSD along the B1 core.


intercalated at 8.1 and 8.7 m depth. These generally low values ofMS suggest a low sedimentary supply from the watershed. Themiddle part of the sequence between 6.5 and 4.7 m depth shows agreater MS variability with two main cycles exhibiting the highestamplitude of MS variation of the series, with an alternation be-tween very low MS rates (470e500 and 540e600 cm depth) andhigher values (500e540 and 600e650 cm depth). The highest valueof magnetic susceptibility of the sequence is found at 610 cm depth(47.2 � 10�5 SI). The phases of increasing MS evidence probablysedimentary input from the watershed during run-off or floodingepisodes. Above 4.7 m depth, the sequence displays a general trendof high MS values (20e40 � 10�5 SI) probably due to more intensealluvial inputs. However, several 5e20 cm thickness levels forwhich MS decreases at ca. 10 � 10�5 SI or lower are intercalated inthe upper part of the sequence between 350 and 400 cm depth,around 300 cm depth, and above ca. 1 m depth.

4.2.4. GeochemistryAs seen above, the strontium may represent a tracer element of

detrital flux from the coastal barriers in the case of the Bagnaslagoon. For the B1 core, we can observe a relatively stable base levelof strontium around 150e200 ppm in the lower and upper parts ofthe sequence, i.e. respectively below 7.4 and above 3.7 m depth(Fig. 6). Between 3.7 and 7.4 m depth, the Sr base level ranges from200 to 250 ppm. Many peaks of Sr disconnect from this base levelall along the sequence. These high values of Sr vary from ca. 300to 1224 ppm. The maximum value is found at 560 cm depth.

At the base of the sequence, between 8 and 9 m depth, thesummit of Sr peaks ranges from about 300 to 600 ppm. Between 6and 7.5 m depth, the Sr peaks are slightly more numerous thanbelow and their summit varies between ca. 300 and 550 ppm. Thehighest andmost abundant Sr peaks are found in themiddle part ofthe sequence between 4.4 and 6m depth, with a Sr curve showing astrong variability and high amplitudes between minimum andmaximum values of Sr. In the upper part of sequence, above 4 mdepth, the Sr peaks are scarcer and isolated from one another bylarge sections of low Sr levels.

4.3. Multi-proxy analysis

4.3.1. Storm event signatureStrontium content measured by XRF core scanner was already

used as geochemical proxy to evidence the paleo-typhoonsdeposited into a lagoon from the southern Japan (Woodruff et al.,2009). In this study, the paleo-typhoons were reconstructed fromlayers exhibiting an increase in grain-size and elevated Sr con-centrations (Woodruff et al., 2009). Given the different geochemicalcomposition of sediments from the coastal barrier and from thewatershed and small tributaries around this Japanese lagoon,detrital coarse levels with high Sr concentrations in the lagoonalsequence were supposed to be issued from reworking of sand andshell or algal materials derived from the coastal barrier, rather thancoarse sediment carried into the lagoon from the watershed duringhigh runoff events (Woodruff et al., 2009). Based on a similaranalytical approach, Raji et al. (2015) identified extreme-sea events(storm or tsunami) in a lagoonal sequence in the northeast ofMorocco from sedimentary layers with high Sr/Fe ratios andcoarser grain size indicating marine inputs into the lagoon.Although these works seem to indicate that more elevated Srconcentrations are related to the presence of shell debris in sedi-ments, it is also possible that high strontium content may be partlydue to inputs of Sr-rich marine saltwater, given that the prolongedflooding associated with storm surges can drive saltwater morethan 15 km inland from the shore (Morton et al., 2007).

In another study on paleostorms, high strontium concentrationswere related to either carbonated shells in sandbarriers or shellabundance in lagoonal sediments due to the biogenic content(Sabatier et al., 2010a). Consequently, in order to use the Sr contentas a consistent proxy to detect the paleostorm event in a lagoonalsedimentary sequence, it is important to decipher the shell prov-enance, either from reworking of clasts from coastal and near-shorebarriers, or from a higher in-situ lagoonal biogenic productivity. Toaddress this problem, a crossover study of strontium content andgrain size of sediments can supply additional information. One cansuppose that authigenous biogenic Sr peaks will be preferentiallyassociated with mollusc shells from a lagoonal malacofaunadeveloped in finer clayey deposits, which are typical of decantationprocesses in lagoonal sedimentation (Liu and Fearn, 2000;Donnelly et al., 2004). By contrast, allochthonous shell debrisreworked and hydrodynamically transported from extra-lagoonalsediments will be more probably associated with coarser levels inthe intra-lagoonal sequence, as proposed byWoodruff et al. (2009).

Fig. 6. Sedimentological proxies of the Bagnas B1 sequence with the strontium content, volume magnetic susceptibility (k), and percentage of fine sand (50e300 mm). Twenty-seven storm episodes (S1 to S27) characterized by high strontium content, low value of magnetic susceptibility and coarser grain-size were distinguished. The red arrowsrepresent Sr peaks associated with finer grain-size and possibly originating from lagoonal biogenic productivity. (For interpretation of the references to colour in this figure legend,the reader is referred to the web version of this article.)


In addition, the provenance of shell debris may be also deducedby coupling the Sr and grain-size records with the magnetic sus-ceptibility proxy. Considering the MS of sediments around theBagnas lagoon, we can expect for higherMS values due to terrestrialpollution during continental runoff or flood events, and, on thecontrary, lower MS values derived from wave surge with a marineand coastal origin during storm event. This sedimentologicalfeature should be more particularly significant in the upper half ofthe sequence where the MS decreases can be well identified insidethe generally high MS levels.

Another problem can arise for distinguishing in a sedimentarysequence between storm and tsunami deposits for coastal domainssubject to tsunami surges (Goff et al., 2004; Morton et al., 2007), asit is possible in the Mediterranean (Soloviev et al., 2000; Bony et al.,2012). However, the sedimentary facies derived from storm ortsunami surges show often similar characteristics (Kortekaas andDawson, 2007). Nevertheless, tsunami surges did probably notinduced major geomorphic impact on the French Mediterraneancoast of Languedoc on the considered timescale. On the one hand,submarine landslide phenomenon susceptible to trigger tsunamiwaves has never been reported in the Languedoc area where theshelf is wide and the slope far offshore, conditions that are notpropitious to the generation of large submarine landslides(Dezileau and Castaings, 2014). On the other hand, it is very un-likely that the seismic-generated tsunamis in the Western Medi-terranean propagated onto the Languedoc area with a highamplitude, considering the distance of earthquake epicentre andgeneral features of these events (Dezileau and Castaings, 2014).Moreover, numerical simulations showed that tsunamiwave height

is negligible inside the French coast (Pelinovsky et al., 2002; Tintiet al., 2005).

In brief, the various lines of evidence presented in this studytend to indicate that the coastal landscape of Languedoc has beenaffected by a succession of exceptional wave impacts. Consideringthe available data, tsunami events do not seem to be at the origin ofthe different Sr-rich coarse-sized layers in the Bagnas pond. Incontrast, there is clear evidence that these Sr-rich coarse layers arecompatible with large storm waves.

4.3.2. Paleostorms in the Bagnas lagoonConsequently to the above-mentioned considerations about the

storm event signature, a multi-proxy analysis of the strontiumcontent, themagnetic susceptibility and the percentage of fine sand(50e300 mm) was undertaken in order to identify the storm de-posits (Fig. 6). Taking into account the geochemical and magneticfeatures of the sedimentary sources in the watershed and thecoastal barriers around the Bagnas pond, on one side, and theincreasing grain size of storm surge levels in the lagoonal sedi-mentary filling, on the other side, the sedimentological signature ofa storm event in the B1 sequence will be characterized by highstrontium content, low value of magnetic susceptibility, andcoarser particle size.

More specifically, the Sr peak at 812 cm depth corresponds to avery thin layer related to a decreasing MS in a main level of higherMS values (Fig. 6). It is considered to be a storm surge layerdeposited during a single weather event combining a stormepisode with fluvial flooding and run-off on the hillslope of thewatershed. In general, coarser material associated to high MS

Fig. 7. Periodicity of storm activity from the strontium data of the Bagnas B1 sequence.(A) Spectral analyses of the strontium content were performed with AnalySeries v.2.0.8(Paillard et al., 1996) by using the multi-taper method (MTM) (linear trend removed,width.ndata product: 1.3, number of windows: 2) and the maximum entropy method(linear trend removed, % of series: 40, number of lags: 198). (B) Strontium content inthe Bagnas B1 sequence and Gaussian filter on strontium data for the 270 yr periodicity(frequency: 0.0037, bandwidth: 0.0005, mean value kept). (C) Number of storm epi-sodes per storm period in the Bagnas B1 sequence from the 7th century BC to the 18thcentury AD (red rectangles) with the addition of major regional storm events in Lan-guedoc during the 19th century AD from Sabatier et al. (2008, 2010a) and Dezileauet al. (2011) (light orange rectangles). (For interpretation of the references to colourin this figure legend, the reader is referred to the web version of this article.)


values and low Sr contents was supposed to be representative ofinland flood events reworking sediments from the watershed.Given the sensitivity of the grain size and MS proxies to detect bothcoastal- or terrestrial-originated flooding events, the Sr record waspreferentially used in order to quantitatively track the seaestormactivity. A storm episode is detected when the relative deviation(RD) of the Sr value of a single peak from the minimum Sr valuepreceding directly this peak exceeds 50%. This procedure yields aconsistent identification of the storm episodes and ensures thereproducibility of their selection. A total of twenty-seven stormepisodes (S1 to S27) were so identified (Fig. 6). These storm epi-sodes are systematically associated with coarser grain size anddecreasing or low values of MS.

Several Sr peaks at 464e467, 525e528, 545e546, 670e672,740e742, 818e824 cm-depth are not correlated with coarser grain-size (red arrows on Fig. 6). These Sr-enriched levels could be due toa stronger development of an in-situ lagoonal micro-fauna, asbiogenic content variations are evident in lagoonal sediments byshell abundances with high concentrations of CaO and Sr (Sabatieret al., 2010a). As an assumption, the fine sedimentation associatedwith these Sr-peaks could testify for low energy environmentpossibly favourable to a higher biogenic productivity in the lagoon.These six abnormally Sr-rich fine-sized layers were not consideredas tempestites deposited from storm events.

4.4. Recurrence and duration of storm periods

Spectral analyses were processed on the time series of stron-tium concentration in order to highlight the periodic patterns ofstorminess activity (Fig. 7A). The six above-mentioned biogenic Srpeaks were not taken into account in the dataset, representing 18points removed on 452 measurements, i.e. 4% of the dataset.However, frequency analyses processed with or without with-drawal of these six Sr peaks yielded similar results. These onesshow a major period with high spectral power density and signif-icant F-test value at ca. 270 yr. This period shows a double-toppedpeak on the MTM spectrum with maximal frequencies at 3.52 and3.91 � 10�3 yr�1. In this range of frequency, the MEM spectrumexhibits a single peak at 3.9 � 10�3 yr�1. The maximal F-test value(0.856) is found at the frequency of 3.71 � 10�3 yr�1, i.e. a peri-odicity of 270 yr. A range of periodicity of 270 ± 30 yr can beestimated by taking into account the full width at half maximum(FWHM) of the 270 yr peak on the MTM spectrum.

The raw Sr data were then filtered with a band-pass Gaussianfilter at 270 yr in order to smooth and regularize the Sr peaks ac-cording to a normal distribution (Fig. 7B). The high frequency os-cillations associated to the 270 yr period exhibit ten cycles withapex at 600, 336, 73 cal yr BC, and 188, 448, 713, 984, 1264,1544 cal yr AD. The apogee of the last cycle was not captured by thedataset due to the Sr time-series of the Bagnas sequence ending at1785 cal yr AD. However, the ascending phase of this last cycle wasprobably prolonged after 1785 cal yr AD, as major regional paleo-storms were identified during the XIXth century AD at 1848 and1893 AD from Sabatier et al. (2008, 2010a) and Dezileau et al.(2011).

The filtered curve matches very well the variations of the orig-inal Sr curve (Fig. 7B). The ten 270-yr cycles define ten storm pe-riods (SP), which are associated with either a single Sr peak or agroup of Sr peaks. The storm episodes S1 to S27 and the 270-yrfiltered Sr curve were used to chronologically demarcate thestorm periods. The boundaries of a SP correspond to the start andend of the respective first and last storm episodes taking placeduring the cycle of the considered SP. This cycle is comprised be-tween two consecutive minima on the 270-yr filtered Sr curve,while the duration and limits of a storm episode are defined by the

full width at half maximum (FWHM) of the corresponding Sr peak.Hence, the time ranges of the storm periods SP10 to SP2 are eval-uated respectively to 620e557, 373e295, 94e43 cal yr BC, and138e247, 404e549, 607e816, 960e983, 1229e1347,1543e1580 cal yr AD. The last storm period (SP1) started at1768 cal yr AD and probably ended to the year of 1893 AD, whichcorresponds to the last major paleostorm recorded in the Pierre-Blanche lagoon 35 km to the northeast from the Bagnas pond(Sabatier et al., 2008; Dezileau et al., 2011). Consequently, the meanduration of a storm periodmay be evaluated to 96 ± 54 yr, while theinter-SP duration is estimated at 173 ± 49 yr. Thus the mean totalduration of a SP/inter-SP cycle may be evaluated to 269 ± 73 yr.

Counting the number of storm episodes per storm periods givesan overview of the strength of storminess activity for each SP


(Fig. 7C). The number of storm episodes per SP is maximal (11) forSP5, which define a period of very high storminess activity betweenca. 600 and 800 cal yr AD. On the contrary, the storm periods SP2,SP4, SP8, and SP9 are characterized by a single episode of stormi-ness, while SP1, SP3, SP7, and SP10 totalize two storm episodeseach. However, this is a minimum value for SP1 given that its upperlimit is not observed in the Bagnas sequence owing to the termi-nation of the Sr time-series at 1785 cal yr AD. The SP1 frequencycould reach four storm episodes by adding the two major regionalstorm events evidenced by Sabatier et al. (2008, 2010a) andDezileau et al. (2011) at 1848 and 1893 AD.

4.5. Comparison with historical storm records

The reliability of our storm record was assessed by a comparisonwith the historical storm records available in the Western Medi-terranean for the last centuries or the last millennium fromwrittensources and archives (Fig. 8). The two longest records in the AdriaticSea and the Western Mediterranean off Barcelona show higherfrequencies of sea storm events at the end of the 1700s AD andrespectively during the first halves of the 1500s and 1600s AD(Camuffo et al., 2000). The first phase in the late 1700s is syn-chronous with the storm episodes S1 and S2 of the storm periodSP1 evidenced in the Bagnas B1 sequence, while the storm episodeS3 (SP2) occurred between the above-mentioned phases of highstorminess frequency in the early 1500s and 1600s. However, S3might as well be associated with the early 1500s phase by takinginto account the calibrated 14C 2-s chronological range of the stormepisodes in the Bagnas lagoon.

A more accurate comparative analysis can be carried outconsidering the FWHM time range of the Sr-peaks that characterizethe storm episodes in the B1 sequence. The S1 and S2 storm epi-sodes, which occurred respectively between 1768 and 1772 AD and

Fig. 8. Chronological comparison of the storm episodes in the Bagnas sequence with historic1: Languedoc, South of France (Sabatier et al., 2008), 2: Salerno Bay (Budillon et al., 2005(Camuffo et al., 2000).

between 1779 and 1782 AD, match chronologically the historicalstorm events that arose at 1771 AD in Languedoc (Sabatier et al.,2008), at 1770 and 1781 AD in the Salerno Bay (Budillon et al.,2005), at 1770 and 1782 AD in the Western Mediterranean offBarcelona (Camuffo et al., 2000), and at 1768, 1771, and 1779 AD inthe Adriatic Sea (Camuffo et al., 2000). The FWHM interval of thestorm episode S3 from 1543 to 1580 AD encompasses the historicalstorm events recorded at 1544 and 1550 AD respectively in theSalerno Bay (Budillon et al., 2005) and the Adriatic Sea (Camuffoet al., 2000). The storm episode S4 (FWHM range from 1330 to1347 AD) occurred synchronously with the sea storm eventsdocumented at 1343 and 1346 AD respectively in the Adriatic Seaand the Western Mediterranean off Barcelona (Camuffo et al.,2000). The storm episodes S5 and S6, with respective FWHMranges from 1229 to 1247 AD and from 960 to 983 AD, arecontemporaneous with the historical sea storm events recordedrespectively at 1240 and 963 AD in the Adriatic Sea (Camuffo et al.,2000).

5. Discussion

5.1. Site sensitivity to storminess activity

The impacts of environmental and geomorphic changes on theconfiguration of the Mediterranean lagoonal systems may influ-ence the record of the storm-induced deposition (Dezileau et al.,2011; Sabatier et al., 2012). For example, changes in sea-level,sedimentary supply, inlet, barrier position and height may poten-tially affect the intensity and frequency of storm events recorded ina lagoonal sequence (Donnelly and Webb, 2004; Dezileau et al.,2011; Otvos, 2011). Reconstructions of sea level changes on theFrench Mediterranean coast show no significant eustatic fluctua-tions for the last 3000 years, with a sea-level rise remaining either

al storm records fromwritten sources in the NWMediterranean for the last millennium.), 3: Western Mediterranean Sea off Barcelona (Camuffo et al., 2000), 4: Adriatic Sea


lower than 1 m (Laborel et al., 1994; Morhange et al., 2001) orcomprised between 1 and 2 m (Lambeck and Bard, 2000; Vella andProvansal, 2000).

The sand barrier in front of the Bagnas lagoon corresponds tothe southern termination of the lido closing the Thau lagoon(Fig. 1B), so that we can suppose that the barrier morphodynamicsof the Thau and Bagnas lagoonal systems were similar during theHolocene. The geosystem coupling the Thau lagoon and its coastalback barrier is probably older than 7000 cal yr BP (Court-Piconet al., 2010). During the last millennia, the marine transgressionand the regional NNE-SSW longshore drift induced a retrograda-tional dynamics along the lido of the Thau lagoon generating alandward movement of the shoreline and sandbars on a few hun-dreds of meters (Barusseau et al., 1996; Tessier et al., 2000; Certainet al., 2005; Court-Picon et al., 2010). The landward displacement ofthe shoreline along the Thau lido did not exceed 300 m since theAntiquity (Tessier et al., 2000). This displacement of a few hundredmeters during the last millennia can be considered as small giventhe distance between the current coastline and the B1 drilling site,which is located at ca. 2.5 km from the shoreline (Fig. 1D), so that itshould be recorded in the cored sequence only the major stormevents.

The sedimentological multi-proxy study and the sedimentaryfacies of the B1 core show that terrestrial sediment inputs into thelagoon clearly increased from about 6 m depth, i.e. from ca.600 cal yr AD (Fig. 2). This sedimentological change was probablynot due to a regional hydrographical reorganization of the fluvialnetwork toward the Bagnas lagoon, by considering the elevatedpalaeotopography of the ante-Holocene substratum to the North ofAgde. Indeed, the Bagnas lagoon was probably isolated throughoutthe Holocene from the floodplain of the H�erault River to the West,except perhaps for major flood episodes during which alluvial in-puts may have reached the Bagnas pond. Hence, the above-mentioned sedimentological change in the B1 core could resultrather from local palaeo-morphological modifications such as aclosure of the communications between the lagoon and the sea.This hydrogeomorphic process would have led to a shift toward anisolated lagoonal environment in relation to the final closure of thesandy barrier by coastal hydrodynamics, as evidenced in the Pierre-Blanche lagoon to the North of the study area around 1000 cal yr BP(Sabatier et al., 2010b, 2012). Therefore, in the case of the Bagnaslagoon, the sandy barrier was probably continuous after ca.600 cal yr AD. Before this date, the lagoonal systemwas less isolatedfrom the sea and presented probably a weaker barrier with largepermanent channel, which controlled inflow of marine water intothe system, typical from a protected lagoon environment (Sabatieret al., 2010b, 2012). However, the multiple layers of fine sedimenttypes appearing throughout the whole record show that this backbarrier area was experiencing quiescent sedimentation during themost part of the last 3000 years. This lagoonal system was thusprotected behind the barrier system over that time.

Finally, although there were minor sea-level fluctuations andshoreline changes during the past 3000 years, these changes wereprobably inadequate to alter drastically the depositional environ-ment of the Bagnas lagoon and hence the sensitivity of this site inrecordingmajor paleostorms. It was consequently assumed that theSr record from the Bagnas lagoonal sequence is sufficiently reliableto be used as a proxy characterizing the frequency of major stormevents in the NW Mediterranean, but is possibly less consistent torecord their intensity during the last millennium owing to theprogressive filling and closure of the lagoon.

5.2. Comparison with global and regional proxies

In addition to the storm periods SP1 to SP10, the Bagnas Sr

record was divided into five main phases of storminess activitybased on the number of storm episodes per SP (Fig. 9.1), i.e. threephases of higher storm activity from 1200 cal yr AD to probably theend of the XIXth century AD (phase 1), from 150 to 800 cal yr AD(phase 3), and from at least the onset on the sequence until550 cal yr BC (phase 5), and two phases of low storm activity from800 to 1200 cal yr AD (phase 2) and from 550 cal yr BC to 150 cal yrAD (phase 4).

The storm record from the Bagnas lagoon was firstly comparedto other storminess proxies in Western Europe. The three last highstorm activity periods (HSAPs) evidenced by Sabatier et al. (2012)from the Pierre-Blanche lagoon in the NW Mediterraneanoccurred at 2800e2600, 1950e1400, and 400e50 cal yr BP(Fig. 9.2). The main differences from our record concerns the stormperiods SP3 and SP5, which are not included respectively before thefirst HSAP and after the second HSAP. The three last Holocene stormperiods (HSP) evidenced by Sorrel et al. (2009, 2012) in NorthernEurope occurred at 3300e2400,1900e1050, and 600e250 cal yr BP(Fig. 9.3). Considering the uncertainties of respective age-depthmodels, the main difference from the Bagnas storm record corre-sponds to a shorter first HSP, so that the storm periods SP1 and SP3are not associated with this first HSP. The record of cliff-top stormdeposits (CTSDs) in the British Isles from Hansom et Hall (2009) islimited to the phases 1 to 3 reported in the Bagnas sequence andshows four stormy episodes at 400e550, 700e1050, 1300e1900,and since 1950 cal yr AD (Fig. 9.4). The two oldest episodes can berelated to the phase 3 and the youngest episode to the phase 1.However, the CTSD associated with the phase 3 does not record theSP7 storm period and extends to the first half of SP4.

The comparison was then carried out with climate proxiesdistributed in the North Atlantic andWestern Europe. The phases ofhigher storm activity 1 and 5 match very well the phases of glacieradvances in the Swiss Alps that occurred respectively at1300e1860 cal yr AD and 1000e600 cal yr BC from Holzhauseret al. (2005) (Fig. 9.5). In contrast, the extension of alpine glaciersduring the phase 3 was very brief and restricted to the SP6 stormepisode. Besides, we can point out short pulses of glacier advancesat 800e900 and 1100e1200 cal yr AD arising respectively at theend of the phase 3 and the start of the phase 1. Concerning theHolocene Cold Events (HCEs) evidenced by Wanner et al. (2011)from a global multi-proxy compilation, the first and third HCEsare chronologically very similar to the phases of high storminess 1and 5 (Fig. 9.6). Nonetheless, the second HCE related to the phase 3is limited to the SP6 storm period, excluding the SPs 5 and 7.

The comparative analysis with the climate periods deducedfrom records of d18O and d13C on Foraminifera or speleothems intheWesternMediterranean (Fig. 9.7e9.11) shows that the phases ofhigher storm activity found in the Bagnas lagoon occurred gener-ally during the cold episodes of the Little Ice Age (phase 1), the DarkAges Cold Period (phase 3, and more particularly the very intensestorm periods SP5 and SP6), and the Iron Age Cold Period (phase 5).Furthermore, extreme storm waves were recorded on the FrenchMediterranean coast to the East of the Rhone delta (Fig. 1B) duringthe Little Ice Age (Shah-Hosseini et al., 2013). On the contrary, theMedieval Warm Period (phase 2) and the Roman Warm Period(phase 4) are characterized by a lower storm activity in the Bagnaslagoon. Grauel et al. (2013) added to this chronology the RomanClassical Period between 0 and 200 cal yr AD, which is character-ized by a decrease of temperature after ca. 90 cal yr AD synchron-ised with the entrance in the phase 3. However, in the detail, somediscrepancies appear between the limit of the stormy phases inLanguedoc and the climatic periods evidenced in the WesternMediterranean. For example, the Medieval Warm Period can eitherend or start respectively in the early stage of the phase 1 or in thelate stage of the phase 3, while the Roman Warm Period can

Fig. 9. Comparison of the storm activity in the Bagnas lagoon with palaeoclimaticproxies for the last 2800 years. (1) Storm episode per storm period in the Bagnaslagoon, (2) High storm activity periods (HSAP) in the NW Mediterranean Sea (Sabatier


terminate in the beginning of the phase 3. Besides, the storm periodSP4 in the phase 2 occurred during the Medieval Cold #2 evidencedby Frisia et al. (2005).

The three main phases of higher storminess activity 1, 3, and 5occurred during the periods of colder temperature in centralGreenland from Alley (2000) (Fig. 9.12). The two main coldest ep-isodes from 100 to 800 cal yr AD and from 1200 to 1800 cal yr ADcorrespond respectively to the phase 3 (SP5 to SP7) and the phase 1(SP1 to SP3) in the Bagnas lagoon. More specifically, the veryintense stormy period of SP5 corresponds to the coldest episode inthe central Greenland over the past 3000 years recorded at630e750 cal yr AD. By contrast, the periods of less intense storm-iness activity SP4 and SP8/SP9 match respectively the two mainphases of higher temperatures in central Greenland between 800and 1200 cal yr AD and between ca. 500 cal yr BC and 50 cal yr AD.

The ice-rafted debris (IRD) indices from Bond et al. (1997, 2001)show generally higher values during the three phases of higherstorminess activity (Fig. 9.13 and 9.14), which correspond to thethree last Bond cycles 0, 1, and 2. In detail, the storm periods SP1,SP2, and SP3 can be associated respectively to the three main IRDpeaks at ca. 1750e1850, 1500e1600, and 1250e1350 cal yr ADduring the cycle 0. For the cycle 1, the storm periods SP5 and SP6present a noteworthy synchronicity with the two main IRD peaksbetween 400 and 800 cal yr AD. Besides, the storm periods SP8 andSP9 in the phase 4 are chronologically concomitant to the two littlepeaks of the stacked IRD curve at 0e100 and 300e350 cal yr BC. Thestorm period SP10 corresponds to the high values of IRD at the endof the Bond event #2 (ca. 2600e3200 cal yr BP).We can so concludeconcerning the drift ice indices to a very good correspondencebetween the SPs and the phases of higher IRDs.

The mean grain-size of the Holocene Icelandic loess at theHolma site from Jackson et al. (2005) reveals a long period of coarsematerials deposited by more powerful winds from about 200 cal yrBC to 800 cal yr AD, the most part of the highest values of meangrain size corresponding to the phase 3 in the Bagnas sequence(Fig. 9.15). The deposition of finer loess occurred during the phasesof low storm activity 2 and 4 respectively between 800 and1200 cal yr AD and between ca. 200 and 400 cal yr BC. The phase 5is characterized by intermediate grain size values, while four peaksof coarser loess deposition appear during the phase 1. The twolatter peaks match relatively well the storm periods SP1 and SP2,

et al., 2012), (3) Holocene storm periods (HSP) in NW Europe (Sorrel et al., 2012), (4)Cliff-top storm deposits (CTSDs) in British Isles (Hansom and Hall, 2009), (5) Phases ofglacier advances in the Swiss Alps (Holzhauser et al., 2005), (6) Holocene Cold Events(HCE) from a global multi-proxy compilation (Wanner et al., 2011), (7) Climatic periodsbased on d18O and d13C of Globigerinoides ruber (white) in the central MediterraneanSea (Grauel et al., 2013), (8) Climatic periods based on d18O of speleothems from a cavein Northern Italy (Frisia et al., 2005), (9) Climatic periods based on d13C of speleothemsfrom three caves in Northern Iberia (Martin-Chivelet et al., 2011), (10) Climatic periodsbased on d18O of several species of Foraminifera from the Tagus Prodelta (W Portugal)on the Atlantic Iberian margin (Lebreiro et al., 2006), (11) Climatic periods based onpollen influx from a sedimentary sequence in NW Iberia (Desprat et al., 2003), (12)Temperature in central Greenland (GISP2) (Alley, 2000), (13) Hematite-stained grainsfrom the MC52 and VM29-191 coring sites (Bond et al., 1997, 2001), (14) Stacked ice-rafted debris (IRD) index (Bond et al., 1997, 2001), (15) Mean grain size of HoloceneIcelandic loess deposits at Holma site (Jackson et al., 2005), (16) Subpolar NorthAtlantic LO09-14 sea-surface temperatures (SST) (Berner et al., 2008), (17) Sea-surfacetemperature (SST) warm anomalies in the subtropical Atlantic at site ODP658C(DeMenocal et al., 2000), (18) Sea-surface temperature (SST) cold anomalies in thesubtropical Atlantic at site ODP658C (DeMenocal et al., 2000), (19) Annual average ofsea-surface temperatures (SST) from d18O data in the Sargasso Sea (Keigwin, 1996). LIA:Little Ice Age, MWP: Medieval Warm Period, MW: Medieval Warm, MC: Medieval Cold,DACP: Dark Ages Cold Period, RCP: Roman Classical Period, RWP: Roman Warm Period,IACP: Iron Age Cold Period. Dark orange rectangles on the Sr curve from the Bagnaslagoon represent the number of storm episodes per storm period. (For interpretation ofthe references to colour in this figure legend, the reader is referred to the web versionof this article.)


while the two former can be associated to the storm episodes S4and S5 in the storm period SP3.

The SST records release a more contrasted picture of the rela-tionship with the storminess activity deduced from the Bagnaslagoonal sequence. On thewhole, the SSTanomalies in the subpolarNorth Atlantic from Berner et al. (2008) (Fig. 9.16) and the SSTs inthe subtropical Atlantic from DeMenocal et al. (2000) (Fig. 9.17 and9.18) exhibit higher temperatures during the phases of intensestorminess 3 and 5, and high or low temperatures during the phase1. However, the annual average SSTs in the Sargasso Sea fromKeigwin (1996) show an inverse pattern with the most often lowtemperatures during the phases of intense storminess activity(Fig. 9.19).

In first instance, this general review of palaeoclimatologicalproxies compared with the Sr-based storm record from the Bagnaslagoon unveils that the three main phases of higher storminessactivity in the NW Mediterranean during the Late Holocene werecontemporaneous with periods of global cooling in the NorthernHemisphere, as previously suggested by Dezileau et al. (2011),Sabatier et al. (2012), and Sorrel et al. (2012).

Finally, we have tested the statistical relationships between thestorminess activity deduced from the Bagnas sequence and the SSTrecords or proxies evidencing North Atlantic coolings (IRD, tem-peratures in central Greenland, grain size of Icelandic loess). A datatransformation was required in order to harmonize the variousdataset owing to their irregular sampling step, their differenttemporal resolution, and their different degree of stationarity. Allthe dataset were resampled with a 5-yr step and detrended on theconsidered time series (165e2635 cal yr BP) using a least-squareslinear regression and mean adjusted residuals. Additionally, thetime series were smoothed with a 270-yr low-pass FFT filterapplying a Savitsky-Golay least-squares polynomial smoothingwith a minimum norm constraint (Mann, 2004). It appears that theSr-based storm record of the Bagnas sequence is significantlycorrelated at p < 0.05 (n ¼ 495) with the IRD stack (r ¼ 0.30),temperatures in central Greenland (r ¼ �0.38), mean grain size ofIcelandic loess (r ¼ 0.27), and SSTs in the subpolar North Atlantic(r ¼ 0.74) and subtropical Atlantic (r ¼ 0.68).

5.3. Origin of storminess variability

Amongst the possible forcing mechanisms driving the climatevariability were evoked the orbital variations, solar activity, sulphuremissions in the atmosphere during large volcanic eruptions,forcing through land cover change, or greenhouse gas injected inthe atmosphere (Jones and Mann, 2004; Wanner et al., 2008).

The main phases of storminess activity discussed above matchthe 1470 ± 500 yr cycles of ice-rafted debris (IRD) indices in NorthAtlantic evidenced by Bond et al. (1997, 2001). These authorsassociated this cyclicity to a solar forcing mechanism, which couldhave been amplified by the production of the North Atlantic DeepWater (NADW) affected by surface hydrographical changes. Byrevisiting the IRD data and other well-known series in the NorthAtlantic fromwavelet analyses, Debret et al. (2007, 2009) showed acyclicity of 1500 yr since the Mid-Holocene supposedly related toan internal forcing due to the thermohaline circulation (THC). Be-sides, Sorrel et al. (2012) found that there was no impact of solarforcing on storminess activity in Northern Europe for millennialcycles, but did not rule out a possible influence of this externalforcing for shorter timescales.

In the Western Mediterranean area, a ca. 1750 yr oscillation wasevidenced by Fletcher et al. (2013) for the last 6 kyr from thewavelet analysis of a pollen dataset tracking the change of forestvegetation in Southern Spain. This ca. 1750 yr mid-to late-Holoceneoscillation could be closely coupled to North Atlantic surface ocean

circulation dynamics and possibly driven by an internal oscillationin deep-water convection strength (Fletcher et al., 2013). Moreover,this millennial oscillation could reflect shifts between a prevailingstrong and weak state of the zonal flow with impacts respectivelysimilar to the positive and negativemodes of the present-day NorthAtlantic Oscillation (Fletcher et al., 2013). Rodrigo-Gamiz et al.(2014) performed spectral analyses on geochemical ratios from asedimentary sequence in the East Alboran Sea basin (E of Gibraltar)to find a primary periodicity peak at 1515 years possibly related toforcing mechanisms linked to North Atlantic thermohaline circu-lation and atmospheric changes in the North Atlantic area. Theseforcing mechanisms would have modified the NAO patterns andthe Inter-Tropical Convergence Zone (ITCZ) in North Africa, with theconsequent displacement of the African monsoon rain belt(Rodrigo-Gamiz et al., 2014). Martin-Chivelet et al. (2011) used alsospectral analysis from a d13C record of speleothems in NorthernIberia to evidence a 1300 yr cycle possibly related to changes insolar irradiance and North Atlantic circulation patterns.

Concerning the 270 ± 30 yr storminess periodicity found in theNW Mediterranean from the frequency analysis of the Bagnasstrontium record, the spectral coherence between the Sr datasetand high-resolution proxy records representative of external orinternal climate forcing was analysed in a tentative attempt torelate this centennial-scale variability of storminess to a causalfactor (Fig. 10). Analyses of coherence were carried out in previousstudies with the aim of establishing a relationship between local orregional proxies and global climate proxies (Viau et al., 2006;Kravchinsky et al., 2013). The solar forcing was tracked with the10Be-based record of total solar irradiance (DTSI) (Steinhilber et al.,2009) as well as the tree-ring 14C residuals (D14C) from IntCal13(Reimer et al., 2013), which is used as an indicator of the radio-carbon 14C cosmogenic production and indirectly of the solar ac-tivity (Kovaltsov et al., 2012).

The 270-yr storminess periodicity is highly coherent with theDTSI, while the coherence with the D14C record is less pronounced(Fig. 10). More specifically, the storm period SP10 arose during atransitional phase of increasing solar influx following a deep solarirradiance low around 2650 cal yr BP (Van Geel et al., 1996; Vieiraet al., 2011), the so-called Homeric Low (Van Vliet-Lano€e et al.,2014). The higher radiocarbon 14C cosmogenic production in theatmosphere during the storm period SP9 indicates a lower solaractivity, but less marked than during the Homeric Low. This lowersolar radiative influx is visible on the DTSI signal only for the onsetof the SP9, the most part of this SP having a positive DTSI.

The storm periods SP8 to SP6 occurred during phases of highsolar activity from both 14C and 10Be records. This high solar activitycontinued during the storm period SP5 from the residual 14C curve,while the DTSI record exhibits the lowest solar irradiance from thelast 2500 years in the first half of this SP at ca. 650 cal yr AD. Thestorm period SP4 shows a mixed situation with low D14C levelsindicating a high solar energy input, but with a decrease of the totalsolar irradiance at the same time. The storm period SP3 is charac-terized by a large decrease of solar activity during the Wolf Mini-mum from 1280 to 1340 AD (Stuiver et al., 1997; Vonmoos et al.,2006). However, the two storm episodes S4 and S5 of SP3 aremore particularly associated with respective increasing or highlevels of solar irradiance (Fig. 10).

The storm period SP2 occurred during a general phase of lowsolar activity encompassing the Sp€orer and Maunder minimarespectively from 1415 to 1535 AD and from 1645 to 1715 AD (Eddy,1976; Stuiver et al., 1997; Usoskin and Kovaltsov, 2004; Vonmooset al., 2006), the later corresponding to a period of very low sun-spot number from telescopic observation since the year of 1620 AD(Hoyt and Schatten, 1998; Solanki et al., 2004). However, in thedetail, SP2 occurred more precisely during a phase of increasing

Fig. 10. Spectral coherence between the strontium concentration in the Bagnas B1 sequence and global climate proxy records. (a) Strontium content in the Bagnas B1 sequence, (b)difference of total solar irradiance from the value of the PMOD composite during the solar cycle minimum of the year 1986 AD (1365.7 W m�2) as given in Fr€ohlich (2009)(Steinhilber et al., 2009), (c) tree-ring 14C residuals from IntCal13 (Reimer et al., 2013), (d) global volcanic stratospheric sulphate injection based on 54 ice core records from theArctic and Antarctica (Gao et al., 2008), (e) NAO index based on a set of 48 annually resolved proxy records distributed around the Atlantic Ocean (Ortega et al., 2015). Analysis of thespectral coherence with the following parameters: Blackman-Tuckey cross-spectrum method, linear trend removed, Bartlett or Welch windows, 30% of the series.


solar activity between the Sp€orer andMaunder minima. Finally, thestorm period SP1 also took place during a phase of high solar ac-tivity, although this one was slightly lower during the DaltonMinimum around AD 1800 (Hoyt and Schatten, 1998).

Miller et al. (2012) suggested that the phase of low solar activityinitiated in the late thirteenth century AD and associated with thefirst cooling period of the Little Ice Age was triggered by volcanismand sustained by sea-ice/ocean feedbacks. Miller et al. (2012)showed that explosive volcanism produces abrupt summer cool-ing with low summer insolation, and that cold summers could bemaintained by sea-ice/ocean feedbacks long after volcanic aerosols

are removed from the atmosphere. These aerosols scatter somesolar radiation back into space and absorb both solar and terrestrialradiation, implying a cooling of the ground surface and a heating ofthe stratosphere (Robock, 2000). Moreover, the impacts of tropicalexplosive volcanism or extremely large Northern Hemisphere mid-latitude volcanic eruption on climate and the carbon cycle resultfrom a global and regional atmospheric surface temperature cool-ing and a multi-decadal decrease in atmospheric pCO2 due to car-bon uptake by the land and ocean (Fr€olicher et al., 2011;Segschneider et al., 2013). As an assumption, this volcanic distur-bance on the carbon cycle could explain the lowest coherence


between the Bagnas Sr content and the D14C record compared tothe DTSI record.

A series of moderate to large sulphate injections occurred dur-ing the thirteenth century AD, so that the cumulative volcanicsulphate flux in this century was 2e10 times larger than this in anyother century within the last millennium (Gao et al., 2008). Thisperiod of high sulphate stratospheric injections could be respon-sible of the high DTSI decrease during the storm period SP3.Amongst these sulphate injections, the super-eruption of theSamalas volcano (Indonesia) in 1257 AD yielded the largest volcanicsulphur release to the stratosphere over the past 7000 years(Oppenheimer, 2003; Lavigne et al., 2013). However, although therewas perturbation of solar radiative influx by sulphur emissions inthe atmosphere during large volcanic eruptions in the secondmillennium AD, the occurrence of storm periods in the NW Medi-terranean according to a cycle of 270 years does not seem to havebeen impacted by this volcanic forcing, as confirmed by the lowspectral coherence for this periodicity (Fig. 10).

The analysis of spectral coherence suggests that an externalforcing as the solar activity could have primarily driven the 270 yrstorminess pattern in the NW Mediterranean for the last 2700years. Multi-centennial solar-driven cycles with global or regionalsignificance have been reported around 200e210 and 340e350 yr(Andrews et al., 2003; Wanner et al., 2008; Duan et al., 2014; Soonet al., 2014), the first corresponding to the well-known Suess (or deVries) cycle associated to solar modulation (Stuiver and Braziunas,1993; Wagner et al., 2001). In the Central Mediterranean, Tarricoet al. (2009) evidenced cycles of 200 and 350 years from a d18Orecord on Foraminifera. These cycles were supposedly related tovariations of temperature in Europe (Tarrico et al., 2009). InNorthern Iberia, Martin-Chivelet et al. (2011) recognised a climaticcycle of 230 years based on d18O of speleothems, possibly related tochanges in solar irradiance and North Atlantic circulation patterns.However, a multi-centennial cyclicity with a periodicity in therange of 240e300 yr studied here was rarely observed up to now.The periodicity the closest to 270 ± 30 yr was brought out from aLate Holocene varve-thickness record from a lacustrine sequence incentral Iceland, which revealed cycles of 55, 130, and 290 years forthe last 3 kyr (Olafsdottir et al., 2013).

5.4. Atmospheric modes

The cyclogenesis in the Western Mediterranean is triggered bythemajor North Atlantic synoptic systems (Trigo et al., 2002). In theNorth Atlantic realm, atmospheric modes such as the Artic Oscil-lation (AO) and the North Atlantic Oscillation (NAO) can influencethe climate variability (Jones and Mann, 2004; Wanner et al., 2008;Goudeau et al., 2015). The NAO is considered as one of the mostprominent and recurrent modes of atmospheric circulation vari-ability in the North Atlantic and Europe (Hurrell et al., 2003). In theliterature, some works established relationships between thestorminess activity, cyclogenesis and NAO modes. At short time-scales, Andrade et al. (2008) evidenced that the periods of higherstorminess activity in the Azores area between 1865 AD and the latetwentieth century are generally characterized by lower values ofthe NAO index, while Pinto et al. (2009) found that extreme cy-clones in the North Atlantic occur more frequently during strongpositive NAO phases in the second half of the twentieth century. Forclimate changes at the millennial timescale, various schemesshowed a transition from weakened to strengthened cyclogenesis/storminess and NAO respectively from the MCA to the LIA (Dawsonet al., 2003, 2007; Van Vliet-Lano€e et al., 2014). Besides, Sorrel et al.(2012) and Trouet et al. (2012) noted that the MCA was character-ized by a phase of low storm activity contemporaneous to apersistent positive NAO.

However, our Sr-based storminess record does not exhibit aspectral coherence with the NAO index reconstructed by Ortegaet al. (2015) for the past millennium (Fig. 10). Furthermore, themillennial-scale synchronism of the NW Mediterranean and theNorthern European storm activity, already observed by Sabatieret al. (2012) and Sorrel et al. (2012), seems to be in disagreementwith the storm track seesaw between southern and northernEurope for the present-day NAO (Hurrell, 1995; Hurrell et al., 2003).The N/S in-phase pattern for storm activity in Europe has led someauthors to suggest that the NAO seesaw was probably not a majormechanism driving the long-term storminess trend during theHolocene (Dezileau et al., 2011; Sabatier et al., 2012). The absence ofspectral coherence between the NAO index from Ortega et al.(2015) and the Sr concentration in the Bagnas lagoon seems toconfirm this view, at the least for the last millennium. Besides, thesynchronous storminess activity in Northern Europe and WesternMediterranean concurs with the N/S in-phase climatic humiditypattern in Spain and Norway evidenced by Dermody et al. (2012),but it does not match the humidity pattern from Fletcher et al.(2013) showing a predominant opposition on millennial time-scales between southern and northern Europe during the mid tolate Holocene.

5.5. Societal relevance

Concerning the humaneenvironment interactions, the newmulti-centennial storminess pattern revealed by the Bagnas recordcould give a remarkable opportunity to study the impact of majorstorm events on human settlements in the Western Mediterraneanduring the last millennia. For example, the development of theRoman civilization across the Western Mediterranean during aperiod of quiet storm activity, the so-called Roman Warm Period,was succeeded by the twilight of theWestern Roman Empire in theLate Antiquity for which is recorded the highest storm activity overthe past three millennia. The analysis of the economical and soci-etal impacts of major storm periods on the ancient societies, andnotably the impacts on maritime trade, port infrastructures, andcoastal natural resources, will require a crossover study of sedi-mentary, historical, and archaeological archives.

In terms of future storminess activity, the prediction of theoccurrence of storm event is a major concern for populationinhabiting the coastal areas prone to cyclonic hazard. The stormloss models developed from numerical simulations based on amulti-scenario ensemble of the changes in the frequency and in-tensity of European windstorms under future climate conditionsshow an increase of insured loss potentials at the end of the 21stcentury in Europe (Pinto et al., 2007, 2012; Della-Marta and Pinto,2009). However, the knowledge of the past storminess variability isan important factor for predicting the future occurrence of majorstorm events. For example, by assuming a continuation of the 270-yr storminess periodicity in the future, the climaxof the next periodof high storm activity could occur around 2100 AD in the WesternMediterranean with a possible intensification of the tempest fre-quency during the next decades, which would in turn increase thevulnerability of populations settled along the Western Mediterra-nean coastal areas. Nevertheless, this is in contradiction with theclimate models simulated under enhanced Greenhouse Gas forcingconditions that predict a reduction in the total number of cyclonesand associated windstorms crossing the Mediterranean regionduring the 21st century (Nissen et al., 2014). If correct, the pre-dictions from these model simulations could indicate that theglobal warming and high atmospheric CO2 increase experiencedduring the last century and possibly due to anthropogenic forcingcould be interfering with the climate mechanisms at the origin ofthe multi-centennial storminess variability evidenced in the


Western Mediterranean from the Bagnas lagoonal sequence.

6. Conclusion

The reconstruction of paleostorm events may supply reliableproxies to record the past climate variability. In coastal wetlands,the lagoons contain sedimentary sequence in which may be inter-bedded tempestites issued from storm event. In the NW Mediter-ranean, the Bagnas pond is filled with a lagoonal sequence, whichallowed the reconstruction of the storminess activity over the last3000 years. A principal component analysis (PCA) of thegeochemistry and sedimentology of the sediments in the water-shed of the lagoon and from the coastal barriers was undertaken inorder to discriminate spatially the source of detrital inputs into thelagoon. The PCA reveals that the sedimentary supply from thecoastal barriers are characterized by high strontium concentrationsand very low magnetic susceptibility. A multi-proxy analysis of theSr content, the magnetic susceptibility, and the grain size of theBagnas lagoonal sequence evidences the presence of twenty-sevenstorm episodes during the Late Holocene. Spectral analyses of theSr concentrations show a multi-centennial storminess variabilitydefining ten major storm periods with a periodicity of 270 years.

A comparison of the Bagnas Sr record with paleoclimate proxiesin North Atlantic, Western Europe, and Western Mediterraneanshows that the phases of high storminess activity occurred duringcold periods. Moreover, the comparison with other storm proxiesexhibits an in-phase storminess pattern between Northern Europeand the Western Mediterranean. The highest storminess activityoccurred from 400 to 820 cal yr AD during the Dark Ages ColdPeriod, with an increase in the frequency of storm episodes. TheBagnas sequence records also periods of high storminess activityduring the end of the Iron Age Cold Period and during the Little IceAge. By contrast, the Roman Warm Period and the Medieval WarmPeriod were characterized by phases of low storm activity.

The results of spectral analyses show that the 270-yr storminesscycle is coherent with the total solar irradiance record and seem toindicate that the Late Holocene solar-driven multi-centennialvariability of the cyclogenesis in the Western Mediterranean wassteered by an external climate forcing. Besides, the absence ofspectral coherence between our storm record and the phases of theNorth Atlantic Oscillation for the last millennium suggests that thismode of atmospheric circulation variability was not a key drivingfactor in the triggering and timing of the multi-centennial stormi-ness pattern evidenced in the Western Mediterranean. Neverthe-less, further studies will be required to study potential relationshipsbetween the major storm periods and other modes of climaticvariability such as the Atlantic Multidecadal Oscillation or theAtlantic Meridional Overturning Circulation.

Acknowledgements

This work is a part of the DYLITAG program funded by the LABEXARCHIMEDE from a grant of the French funding agency “AgenceNationale de la Recherche” (program “Investissement d’Avenir”ANR-11-LABX-0032-01). The drill cores were performed by theArcheoEnvironnement technical platform (CNRS UMR 5140) withthe help of Philippe Blanchemanche and Gael Piques. The grain sizeanalyses were performed at the Geosciences Laboratory (CNRS andUniversity of Montpellier). The AMS 14C ages were undertaken bythe Centre de Datation par le RadioCarbone (CDRC, CNRS UMR 5138and University of Lyon 1, ARTEMIS program). The ADENA associa-tion authorized access to the protected nature reserve of the Bagnaspond. We thank Dr. Henning A. Bauch (editor) and the threeanonymous reviewers for their helpful remarks and suggestions.

References

Alley, R.B., 2000. The Younger Dryas cold interval as viewed from central Greenland.Quat. Sci. Rev. 19, 213e226.

Andrade, C., Trigo, R.M., Freitas, M.C., Gallego, M.C., Borges, P., Ramos, A.M., 2008.Comparing historic records of storm frequency and the North Atlantic Oscilla-tion (NAO) chronology for the Azores region. Holocene 18 (5), 745e754.

Andrews, J.T., Hardadottir, J., Stoner, J.S., Mann, M.E., Kristjansdottir, G.B., Koc, N.,2003. Decadal to millennial-scale periodicities in North Iceland shelf sedimentsover the last 12000 cal yr: long-term North Atlantic oceanographic variabilityand solar forcing. Earth Planet. Sci. Lett. 210, 453e465.

Bache, F., Olivet, J.-L., Gorini, C., Aslanian, D., Labails, C., Rabineau, M., 2010. Evo-lution of rifted continental margins: the case of the gulf of lions (WesternMediterranean Basin). Earth Planet. Sci. Lett. 292, 345e356.

Barruol, G., Granet, M., 2002. A Tertiary asthenospheric flow beneath the southernFrench Massif Central indicated by upper mantle seismic anisotropy and relatedto the west Mediterranean extension. Earth Planet. Sci. Lett. 202, 31e47.

Bartholy, J., Pongra

cz, R., Pattantyu

s-A

braha

m, M., 2009. Analyzing the genesis,intensity, and tracks of western Mediterranean cyclones. Theor. Appl. Climatol.96, 133e144.

Barusseau, J.-P., Akouango, E., Ba, M., Descamps, C., Golf, A., 1996. Evidence for shortterm retreat of the barrier shorelines. Quat. Sci. Rev. 15, 763e771.

Berger, A., Melice, J.-L., Hinnov, L., 1991. A strategy for frequency spectra of Qua-ternary climate records. Clim. Dyn. 5, 227e240.

Berner, K.S., Koç, N., Divine, D., Godtliebsen, F., Moros, M., 2008. A decadal-scaleHolocene sea surface temperature record from the subpolar North Atlanticconstructed using diatoms and statistics and its relation to other climate pa-rameters. Paleoceanography 23, PA2210.

Blanchemanche, P., 2009. Crues historiques et vendanges en Languedocm�editerran�een oriental : la source, le signal et l'interpr�etation. Arch�eologie DuMidi M�edi�eval 27, 225e235.

Bond, G., Showers, W., Cheseby, M., Lotti, R., Almasi, P., DeMenocal, P., Priore, P.,Cullen, H., Hajdas, I., Bonani, G., 1997. A pervasive millennial-scale cycle in NorthAtlantic Holocene and glacial climates. Science 278, 1257e1266.

Bond, G., Kromer, B., Beer, J., Muscheler, R., Evans, M.N., Showers, W., Hoffmann, S.,Lotti-Bond, R., Hajdas, I., Bonani, G., 2001. Persistent solar influence on NorthAtlantic climate during the Holocene. Science 294, 2130e2136.

Bony, G., Marriner, N., Morhange, C., Kaniewski, D., Perinçek, D., 2012. A high-en-ergy deposit in the Byzantine harbour of Yenikapı, Istanbul (Turkey). Quat. Int.266, 117e130.

Budillon, F., Esposito, E., Iorio, M., Pelosi, N., Porfido, S., Violante, C., 2005. Thegeological record of storm events over the last 1000 years in the Salerno Bay(Southern Tyrrhenian Sea): new proxy evidences. Adv. Geosci. 2, 123e130.

Camuffo, D., Secco, C., Brimblecombe, P., Martin-Vide, J., 2000. Sea storms in theAdriatic Sea and the Western Mediterranean during the last millennium. Clim.Change 46, 209e223.

Certain, R., Tessier, B., Barusseau, J.-P., Courp, T., Pauc, H., 2005. Sedimentary balanceand sand stock availability along a littoral system. The case of the western Gulfof Lions littoral prism (France) investigated by very high resolution seismic.Mar. Petrol. Geol. 22, 889e900.

Collins, E.S., Scott, D.B., Gayes, P.T., 1999. Hurricane records on the South Carolinacoast: can they be detected in the sediment record? Quat. Int. 56, 15e26.

Costa, P.J.M., Andrade, C., Dawson, A.G., Mahaney, W.C., Freitas, M.C., Paris, R.,Taborda, R., 2012. Microtextural characteristics of quartz grains transported anddeposited by tsunamis and storms. Sediment. Geol. 275e276, 55e69.

Court-Picon, M., Vella, C., Chabal, L., Bruneton, H., 2010. Pal�eo-environnementslittoraux depuis 8000 ans sur la bordure occidentale du Golfe du Lion: le lido del'Etang de Thau (carottage SETIF, S�ete, H�erault). Quaternaire 21 (1), 43e60.

Das, O., Wang, Y., Donoghue, J., Xu, X., Coor, J., Elsner, J., Xu, Y., 2013. Reconstructionof paleostorms and paleoenvironment using geochemical proxies archived inthe sediments of two coastal lakes in northwest Florida. Quat. Sci. Rev. 68,142e153.

Dawson, A.G., Eliot, L., Mayewski, P., Lockett, P., Noone, S., Hickey, K., Holt, T.,Wadhams, P., Foster, I., 2003. Late-Holocene North Atlantic climate 'seesaws',storminess changes and Greenland ice sheet (GISP2) palaeoclimates. Holocene13 (3), 381e392.

Dawson, A.G., Hickey, K., Mayewski, P.A., Nesje, A., 2007. Greenland (GISP2) ice coreand historical indicators of complex North Atlantic climate changes during thefourteenth century. Holocene 17 (4), 427e434.

Debret, M., Bout-Roumazeilles, V., Grousset, F., Desmet, M., McManus, J.F.,Massei, N., Sebag, D., Petit, J.-R., Copard, Y., Trentesaux, A., 2007. The origin ofthe 1500-year climate cycles in Holocene North-Atlantic records. Clim. Past 3,569e575.

Debret, M., Sebag, D., Crosta, X., Massei, N., Petit, J.-R., Chapron, E., Bout-Roumazeilles, V., 2009. Evidence from wavelet analysis for a mid-Holocenetransition in global climate forcing. Quat. Sci. Rev. 28, 2675e2688.

Della-Marta, P.M., Pinto, J.G., 2009. Statistical uncertainty of changes in winterstorms over the North Atlantic and Europe in an ensemble of transient climatesimulations. Geophys. Res. Lett. 36, L14703.

DeMenocal, P., Ortiz, J., Guilderson, T., Sarnthein, M., 2000. Coherent high- and low-latitude climate variability during the Holocene warm period. Science 288,2198e2202.

Dermody, B.J., De Boer, H.J., Bierkens, M.F.P., Weber, S.L., Wassen, M.J., Dekker, S.C.,2012. A seesaw in Mediterranean precipitation during the roman period linked

http://refhub.elsevier.com/S0277-3791(15)30133-5/sref1























































































































to millennial-scale changes in the North Atlantic. Clim. Past 8, 637e651.Desprat, S., Sanchez-Goni, M.F., Loutre, M.-F., 2003. Revealing climatic variability of

the last three millennia in northwestern Iberia using pollen influx data. EarthPlanet. Sci. Lett. 213, 63e78.

D�ezes, P., Schmid, S.M., Ziegler, P.A., 2004. Evolution of the European Cenozoic RiftSystem: interaction of the Alpine and Pyrenean orogens with their forelandlithosphere. Tetonophysics 389, 1e33.

Dezileau, L., Castaings, J., 2014. Extreme storms during the last 500 years fromlagoonal sedimentary archives in Languedoc (SE France). M�editerran�ee 122,131e137.

Dezileau, L., Sabatier, P., Blanchemanche, P., Joly, B., Swingedouw, D., Cassou, C.,Castaings, J., Martinez, P., Von Grafenstein, U., 2011. Intense storm activityduring the little ice age on the French Mediterranean coast. Palaeogeogr.Palaeoclimatol. Palaeoecol. 299, 289e297.

Donnelly, J.P., Webb, T., 2004. Back-barrier sedimentary records of intense hurricanelandfalls in the Northeastern United States. In: Murnane, R.J., Liu, K.-B. (Eds.),Hurricanes and Typhoons: Past, Present, and Future. Columbia University Press,New York, pp. 58e95.

Donnelly, J.P., Woodruff, J.D., 2007. Intense hurricane activity over the past 5,000years controlled by El Nino and the West African monsoon. Nature 447,465e468.

Donnelly, J.P., Butler, J., Roll, S., Wengren, M., Webb, T., 2004. A backbarrier over-wash record of intense storms from Brigantine, New Jersey. Mar. Geol. 210,107e121.

Duan, F., Wang, Y., Shen, C.-C., Wang, Y., Cheng, H., Wu, C.-C., Hu, H.-M., Kong, X.,Liu, D., Zhao, K., 2014. Evidence for solar cycles in a late Holocene speleothemrecord from Dongge Cave, China. Sci. Rep. 4 (5159), 1e7.

Dubar, M., 2006. Approche climatique de la p�eriode romaine dans l'est du Var :recherche et analyse des composantes p�eriodiques sur un concr�etionnementcentennal (Ier-IIe si�ecle apr. J.-C.) de l'aqueduc de Fr�ejus. Arch�eoSciences 30,163e171.

Eddy, J.A., 1976. The maunder minimum. Science 192, 1189e1202.Emanuel, K., 2005. Genesis and maintenance of “Mediterranean hurricanes”. Adv.

Geosci. 2, 217e220.Fita, L., Romero, R., Luque, A., Emanuel, K., Ramis, C., 2007. Analysis of the envi-

ronments of seven Mediterranean tropical-like storms using an axisymmetric,nonhydrostatic, cloud resolving model. Nat. Hazards Earth Syst. Sci. 7, 41e56.

Flaounas, E., Raveh-Rubin, S., Wernli, H., Drobinski, P., Bastin, S., 2015. Thedynamical structure of intense Mediterranean cyclones. Clim. Dyn. 44,2411e2427.

Fletcher, W.J., Debret, M., Sanchez-Goni, M.F., 2013. Mid-Holocene emergence of alow- frequency millennial oscillation in western Mediterranean climate: im-plications for past dynamics of the North Atlantic atmospheric westerlies.Holocene 23 (2), 153e166.

Frisia, S., Borsato, A., Sp€otl, C., Villa, I.M., Cucchi, F., 2005. Climate variability in theSE Alps of Italy over the past 17 000 years reconstructed from a stalagmiterecord. Boreas 34, 445e455.

Fr€ohlich, C., 2009. Evidence of a long-term trend in total solar irradiance. Astron.Astrophys. 501, L27eL30.

Fr€olicher, T.L., Joos, F., Raible, C.C., 2011. Sensitivity of atmospheric CO2 and climateto explosive volcanic eruptions. Biogeosciences 8, 2317e2339.

Gao, C., Robock, A., Ammann, C., 2008. Volcanic forcing of climate over the past1500 years: an improved ice core-based index for climate models. J. Geophys.Res. 113, D23111.

Gaertner, M.A., Dominguez, M., Gil, V., Sanchez, E., 2009. Risk of tropical cyclonesover the Mediterranean Sea in a climate change scenario. In: Elsner, J.B.,Jagger, T.H. (Eds.), Hurricanes and Climate Change. Springer US, New York,pp. 235e250.

Goff, J., McFadgen, B.G., Chagu�e-Goff, C., 2004. Sedimentary differences between the2002 Easter storm and the 15th-century Okoropunga tsunami, southeasternNorth Island, New Zealand. Mar. Geol. 204, 235e250.

Goudeau, M.-L.S., Reichart, G.-J., Wit, J.C., De Nooijer, L.J., Grauel, A.-L.,Bernasconi, S.M., De Lange, G.J., 2015. Seasonality variations in the CentralMediterranean during climate change events in the Late Holocene. Palaeogeogr.Palaeoclimatol. Palaeoecol. 418, 304e318.

Grauel, A.-L., Goudeau, M.-L.S., De Lange, G.J., Bernasconi, S.M., 2013. A high-reso-lution climate reconstruction based on d18O and d13C of Globigerinoides ruber(white). Holocene 23 (10), 1440e1446.

Hansom, J.D., Hall, A.M., 2009. Magnitude and frequency of extra-tropical NorthAtlantic cyclones: a chronology from cliff-top storm deposits. Quat. Int. 195,42e52.

Harremo€es, P., Topsoe, F., 2001. Maximum entropy fundamentals. Entropy 3,191e226.

Hawkes, A.D., Horton, B.P., 2012. Sedimentary record of storm deposits from Hur-ricane Ike, Galveston and San Luis Islands, Texas. Geomorphology 171e172,180e189.

Hippensteel, S.P., Martin, R.E., 1999. Foraminifera as an indicator of overwash de-posits, Barrier Island sediment supply, and Barrier Island evolution: Folly Island,South Carolina. Palaeogeogr. Palaeoclimatol. Palaeoecol. 149, 115e125.

Holzhauser, H., Magny, M., Zumbühl, H.J., 2005. Glacier and lake-level variations inwest-central Europe over the last 3500 years. Holocene 15 (6), 789e801.

Horton, B.P., Rossi, V., Hawkes, A.D., 2009. The sedimentary record of the 2005hurricane season from the Mississippi and Alabama coastlines. Quat. Int. 195,15e30.

Hoyt, D.V., Schatten, K.H., 1998. Group sunspot numbers: a new solar activity

reconstruction. Sol. Phys. 179, 189e219.Hughen, K.A., Baillie, M.G.L., Bard, E., Beck, J.W., Bertrand, C.J.H., Blackwell, P.G.,

Buck, C.E., Burr, G.S., Cutler, K.B., Damon, P.E., Edwards, R.L., Fairbanks, R.G.,Friedrich, M., Guilderson, T.P., Kromer, B., McCormac, G., Manning, S.,Ramsey, C.B., Reimer, P.J., Reimer, R.W., Remmele, S., Southon, J.R., Stuiver, M.,Talamo, S., Taylor, F.W., Van der Plicht, J., Weyhenmeyer, C.E., 2004. Marine04marine radiocarbon age calibration, 0e26 cal kyr BP. Radiocarbon 46 (3),1059e1086.

Hurrell, J.W., 1995. Decadal trends in the North Atlantic Oscillation: regional tem-peratures and precipition. Science 269, 676e679.

Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M., 2003. An overview of the NorthAtlantic oscillation. In: Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M. (Eds.),The North Atlantic Oscillation: Climatic Significance and Environmental Impact,Geophysical Monograph, vol. 134. American Geophysical Union, WashingtonD.C, pp. 1e35.

Jackson, M.G., Oskarsson, N., Tronnes, R.G., McManus, J.F., Oppo, D.W., Gr€onvold, K.,Hart, S.R., Sachs, J.P., 2005. Holocene loess deposition in Iceland: evidence formillennial- scale atmosphere-ocean coupling in the North Atlantic. Geology 33(6), 509e512.

Jaynes, E.T., 1957. Information theory and statistical mechanics. Phys. Rev. 106 (4),620e630.

Jelgersma, S., Stive, M.J.F., Van der Valk, L., 1995. Holocene storm surge signatures inthe coastal dunes of the western Netherlands. Mar. Geol. 125, 95e110.

Jones, P.D., Mann, M.E., 2004. Climate over past millennia. Rev. Geophys. 42,RG2002.

Keigwin, L.D., 1996. The little ice age and medieval warm period in the sargasso sea.Science 274, 1504e1508.

Knutson, T.R., McBride, J.L., Chan, J., Emanuel, K., Holland, G., Landsea, C., Held, I.,Kossin, J.P., Srivastava, A.K., Sugi, M., 2010. Tropical cyclones and climate change.Nat. Geosci. 3, 157e163.

Kortekaas, S., Dawson, A.G., 2007. Distinguishing tsunami and storm deposits: anexample from Martinhal, SW Portugal. Sediment. Geol. 200, 208e221.

Kouroutzoglou, J., Flocas, H.A., Hatzaki, M., Keay, K., Simmonds, I., 2014. A high-resolution climatological study on the comparison between surface explosiveand ordinary cyclones in the Mediterranean. Reg. Environ. Change 14,1833e1846.

Kouroutzoglou, J., Flocas, H.A., Hatzaki, M., Keay, K., Simmonds, I., Mavroudis, A.,2015. On the dynamics of a case study of explosive cyclogenesis in the Medi-terranean. Meteorol. Atmos. Phys. 127, 49e73.

Kovaltsov, G.A., Mishev, A., Usoskin, I.G., 2012. A new model of cosmogenic pro-duction of radiocarbon 14C in the atmosphere. Earth Planet. Sci. Lett. 337e338,114e120.

Kravchinsky, V.A., Langereis, C.G., Walker, S.D., Dlusskiy, K.G., White, D., 2013. Dis-covery of Holocene millennial climate cycles in the Asian continental interior:has the sun been governing the continental climate? Glob. Planet. Change 110,386e396.

Labaune, C., Tesson, M., Gensous, B., 2008. Variability of the transgressive stackingpattern under environmental changes control: example from the Post-Glacialdeposits of the Gulf of Lions inner-shelf, Mediterranean, France. Cont. ShelfRes. 28, 1138e1152.

Labaune, C., Tesson, M., Gensous, B., Parize, O., Imbert, P., Delhaye-Prat, V., 2010.Detailed architecture of a compound incised valley system and correlation withforced regressive wedges: example of late quaternary Tet and Agly rivers,western Gulf of Lions, Mediterranean Sea, France. Sediment. Geol. 223,360e179.

Laborel, J., Morhange, C., Lafont, R., Le Campion, J., Laborel-Deguen, F., Sartoretto, S.,1994. Biological evidence of sea-level rise during the last 4500 years on therocky coasts of continental southwestern France and Corsica. Mar. Geol. 120,203e223.

Lagmay, A.M.F., Agaton, R.P., Bahala, M.A., Briones, J.B.L., Cabacaba, K.M., Caro, C.V.,Dasallas, L., Gonzalo, L.A., Ladiero, C., Lapidez, J.P., Mungcal, M.T.F., Puno, J.V.,Ramos, M.M.A., Santiago, J., Suarez, J.K., Tablazon, J., 2015. Devastating stormsurges of typhoon Haiyan. Int. J. Disaster Risk Reduct. 11, 1e12.

Lambeck, K., Bard, E., 2000. Sea-level change along the French Mediterranean coastfor the past 30 000 years. Earth Planet. Sci. Lett. 175, 203e222.

Lambert, W.J., Aharon, P., Rodriguez, A.B., 2008. Catastrophic hurricane historyrevealed by organic geochemical proxies in coastal lake sediments: a case studyof Lake Shelby, Alabama (USA). J. Paleolimnol. 39, 117e131.

Lane, P., Donnelly, J.P., Woodruff, J.D., Hawkes, A.D., 2011. A decadally-resolvedpaleohurricane record archived in the late Holocene sediments of a Floridasinkhole. Mar. Geol. 287, 14e30.

Larue, J.-P., 2008. Effects of tectonics and lithology on long profiles of 16 rivers ofthe southern Central Massif border between the Aude and the Orb (France).Geomorphology 93, 343e367.

Lavigne, F., Degeai, J.-P., Komorowski, J.-C., Guillet, S., Robert, V., Lahitte, P.,Oppenheimer, C., Stoffel, M., Vidal, C.M., Surono, D., Pratomo, I., Wassmer, P.,Hajdas, I., Sri Hadmoko, D., De Belizal, E., 2013. Source of the great AD 1257mystery eruption unveiled, Samalas volcano, Rinjani Volcanic Complex,Indonesia. Proc. Natl. Acad. Sci. U. S. A. 110 (42), 16742e16747.

Lebreiro, S.M., Franc�es, G., Abrantes, F.F.G., Diz, P., Bartels-Jonsdottir, H.B.,Stroynowski, Z.N., Gil, I.M., Pena, L.D., Rodrigues, T., Jones, P.D., Nombela, M.A.,Alejo, I., Briffa, K.R., Harris, I., Grimalt, J.O., 2006. Climate change and coastalhydrographic response along the Atlantic Iberian margin (Tagus Prodelta andMuros Ria) during the last two millennia. Holocene 16 (7), 1003e1015.

Liu, K.-B., Fearn, M.L., 2000. Reconstruction of prehistoric landfall frequencies of
























































































































































































































































catastrophic hurricanes in northwestern Florida from Lake sediment records.Quat. Res. 54, 238e245.

Liu, K.-B., Lu, H., Shen, C., 2008. A 1200-year proxy record of hurricanes and firesfrom the Gulf of Mexico coast: testing the hypothesis of hurricaneefire in-teractions. Quat. Res. 69, 29e41.

Loulergue, L., Schilt, A., Spahni, R., Masson-Delmotte, V., Blunier, T., Lemieux, B.,Barnola, J.-M., Raynaud, D., Stocker, T.F., Chappellaz, J., 2008. Orbital andmillennial-scale features of atmospheric CH4 over the past 800,000 years. Na-ture 453, 383e386.

Mann, M.E., 2004. On smoothing potentially non-stationary climate time series.Geophys. Res. Lett. 31, L07214.

Mann, M.E., Woodruff, J.D., Donnelly, J.P., Zhang, Z., 2009. Atlantic hurricanes andclimate over the past 1,500 years. Nature 460, 880e883.

Martin-Chivelet, J., Munoz-Garcia, M.B., Edwards, R.L., Turrero, M.J., Ortega, A.I.,2011. Land surface temperature changes in Northern Iberia since 4000 yr BP,based on d13C of speleothems. Glob. Planet. Change 77, 1e12.

Mendelsohn, R., Emanuel, K., Chonabayashi, S., Bakkensen, L., 2012. The impact ofclimate change on global tropical cyclone damage. Nat. Clim. Change 2,205e209.

Miller, G.H., Geirsdottir, A., Zhong, Y., Larsen, D.J., Otto-Bliesner, B.L., Holland, M.M.,Bailey, D.A., Refsnider, K.A., Lehman, S.J., Southon, J.R., Anderson, C.,Bj€ornsson, H., Thordarson, T., 2012. Abrupt onset of the little ice age triggeredby volcanism and sustained by sea-ice/ocean feedbacks. Geophys. Res. Lett. 39,L02708.

Morhange, C., Laborel, J., Hesnard, A., 2001. Changes of relative sea level during thepast 5000 years in the ancient harbour of Marseilles, Southern France. Palae-ogeogr. Palaeoclimatol. Palaeoecol. 166, 319e329.

Morton, R.A., Gelfenbaum, G., Jaffe, B.E., 2007. Physical criteria for distinguishingsandy tsunami and storm deposits using modern examples. Sediment. Geol.200, 184e207.

Muscheler, R., 2012. The enigmatic 1,500-year cycle. Nat. Geosci. 5, 850e851.Nissen, K.M., Leckebusch, G.C., Pinto, J.G., Ulbrich, U., 2014. Mediterranean cyclones

and windstorms in a changing climate. Reg. Environ. Change 14, 1873e1890.Nott, J., 2004. Palaeotempestology: the study of prehistoric tropical cyclonesea

review and implications for hazard assessment. Environ. Int. 30, 433e447.Nott, J., 2011. A 6000 year tropical cyclone record from Western Australia. Quat. Sci.

Rev. 30, 713e722.Olafsdottir, K.B., Geirsdottir, A., Miller, G.H., Larsen, D.J., 2013. Evolution of NAO and

AMO strength and cyclicity derived from a 3-ka varve-thickness record fromIceland. Quat. Sci. Rev. 69, 142e154.

Olsen, J., Anderson, N.J., Knudsen, M.F., 2012. Variability of the North AtlanticOscillation over the past 5,200 years. Nat. Geosci. 5, 808e812.

Oppenheimer, C., 2003. Ice core and palaeoclimatic evidence for the timing andnature of the great mid-13th century volcanic eruption. Int. J. Climatol. 23,417e426.

Ortega, P., Lehner, F., Swingedouw, D., Masson-Delmotte, V., Raible, C.C., Casado, M.,Yiou, P., 2015. A model-tested North Atlantic Oscillation reconstruction for thepast millennium. Nature 523, 71e74.

Otvos, E.G., 2011. Hurricane signatures and landformsetoward improved in-terpretations and global storm climate chronology. Sediment. Geol. 239, 10e22.

Page, M.J., Trustrum, N.A., Orpin, A.R., Carter, L., Gomez, B., Cochran, U.A.,Mildenhall, D.C., Rogers, K.M., Brackley, H.L., Palmer, A.S., Northcote, L., 2010.Storm frequency and magnitude in response to Holocene climate variability,Lake Tutira, North-Eastern New Zealand. Mar. Geol. 270, 30e44.

Paillard, D., Labeyrie, L., Yiou, P., 1996. Macintosh program performs time-seriesanalysis. EOS Trans. AGU 77 (39), 379e379.

Pardo-Iguzquiza, E., Rodriguez-Tovar, F.J., 2006. Maximum entropy spectral analysisof climatic time series revisited: assessing the statistical significance of esti-mated spectral peaks. J. Geophys. Res. 111, D10102.

Parris, A.S., Bierman, P.R., Noren, A.J., Prins, M.A., Lini, A., 2010. Holocene paleo-storms identified by particle size signatures in lake sediments from thenortheastern United States. J. Paleolimnol. 43, 29e49.

Parsons, M.L., 1998. Salt marsh sedimentary record of the landfall of HurricaneAndrew on the Louisiana Coast: diatoms and other paleoindicators. J. Coast. Res.14 (3), 939e950.

Pelinovsky, E., Kharif, C., Riabov, I., Francius, M., 2002. Modelling of tsunamipropagation in the vicinity of the French coast of the Mediterranean. Nat.Hazards 25, 135e159.

Percival, D.B., Walden, A.T., 1993. Spectral Analysis for Physical Applications: Mul-titaper and Conventional Univariate Techniques. Cambridge University Press,New York, p. 583.

Pilarczyk, J.E., Dura, T., Horton, B.P., Engelhart, S.E., Kemp, A.C., Sawai, Y., 2014.Microfossils from coastal environments as indicators of paleo-earthquakes,tsunamis and storms. Palaeogeogr. Palaeoclimatol. Palaeoecol. 413, 144e157.

Pinto, J.G., Fr€ohlich, E.L., Leckebusch, G.C., Ulbrich, U., 2007. Changing Europeanstorm loss potentials under modified climate conditions according to ensemblesimulations of the ECHAM5/MPI-OM1 GCM. Nat. Hazards Earth Syst. Sci. 7,165e175.

Pinto, J.G., Zacharias, S., Fink, A.H., Leckebusch, G.C., Ulbrich, U., 2009. Factorscontributing to the development of extreme North Atlantic cyclones and theirrelationship with the NAO. Clim. Dyn. 32, 711e737.

Pinto, J.G., Karremann, M.K., Born, K., Della-Marta, P.M., Klawa, M., 2012. Loss po-tentials associated with European windstorms under future climate conditions.Clim. Res. 54, 1e20.

Rabineau, M., Bern�e, S., Olivet, J.-L., Aslanian, D., Guillocheau, F., Joseph, P., 2006.

Paleo sea levels reconsidered from direct observation of paleoshoreline positionduring Glacial Maxima (for the last 500,000 yr). Earth Planet. Sci. Lett. 252,119e137.

Raji, O., Dezileau, L., Von Grafenstein, U., Niazi, S., Snoussi, M., Martinez, P., 2015.Extreme sea events during the last millennium in the northeast of Morocco.Nat. Hazards Earth Syst. Sci. 15, 203e211.

Raynal, O., Bouchette, F., Certain, R., S�eranne, M., Dezileau, L., Sabatier, P., Lofi, J., BuiXuan Hy, A., Briqueu, L., Pezard, P., Tessier, B., 2009. Control of alongshore-oriented sand spits on the dynamics of a wave-dominated coastal system(Holocene deposits, northern Gulf of Lions, France). Mar. Geol. 264, 242e257.

Reimer, P.J., McCormac, F.G., 2002. Marine radiocarbon reservoir corrections for theMediterranean and Aegean seas. Radiocarbon 44 (1), 159e166.

Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Ramsey, C.B., Buck, C.E.,Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P.,Haflidason, H., Hajdas, I., Hatte

, C., Heaton, T.J., Hoffmann, D.L., Hogg, A.G.,Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W., Niu, M., Reimer, R.W.,Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney, C.S.M., Van derPlicht, J., 2013. Intcal13 and Marine13 radiocarbon age calibration curves 0-50,000 years cal BP. Radiocarbon 55 (4), 1869e1887.

Robock, A., 2000. Volcanic eruptions and climate. Rev. Geophys. 38 (2), 191e219.Rodrigo-Gamiz, M., Martinez-Ruiz, F., Rodriguez-Tovar, F.J., Jimenez-Espejo, F.J.,

Pardo-Iguzquiza, E., 2014. Millennial- to centennial-scale climate periodicitiesand forcing mechanisms in the westernmost Mediterranean for the past 20,000yr. Quat. Res. 81, 78e93.

Sabatier, P., Dezileau, L., Condomines, M., Briqueu, L., Colin, C., Bouchette, F., LeDuff, M., Blanchemanche, P., 2008. Reconstruction of paleostorm events in acoastal lagoon (H�erault, South of France). Mar. Geol. 251, 224e232.

Sabatier, P., Dezileau, L., Briqueu, L., Colin, C., Siani, G., 2010a. Clay minerals andgeochemistry record from northwest Mediterranean coastal lagoon sequence:implications for paleostorm reconstruction. Sediment. Geol. 228, 205e217.

Sabatier, P., Dezileau, L., Blanchemanche, P., Siani, G., Condomines, M., Bentaleb, I.,Piqu�es, G., 2010b. Holocene variations of radiocarbon reservoir ages in a Med-iterranean lagoonal system. Radiocarbon 52 (1), 1e12.

Sabatier, P., Dezileau, L., Colin, C., Briqueu, L., Bouchette, F., Martinez, P., Siani, G.,Raynal, O., Von Grafenstein, U., 2012. 7000 years of paleostorm activity in theNWMediterranean Sea in response to Holocene climate events. Quat. Res. 2012,1e11.

Segschneider, J., Beitsch, A., Timmreck, C., Brovkin, V., Ilyina, T., Jungclaus, J.,Lorenz, S.J., Six, K.D., Zanchettin, D., 2013. Impact of an extremely largemagnitude volcanic eruption on the global climate and carbon cycle estimatedfrom ensemble earth system model simulations. Biogeosciences 10, 669e687.

Shah-Hosseini, M., Morhange, C., De Marco, A., Wante, J., Anthony, E.J.,Mastronuzzi, G., Pignatelli, C., Piscitelli, A., 2013. Coastal boulders in Martigues,French Mediterranean: evidence for extreme storm waves during the little iceage. Z. für Geomorphol. 57 (4), 181e199.

Siani, G., Paterne, M., Arnold, M., Bard, E., M�etivier, B., Tisnerat, N., Bassinot, F., 2000.Radiocarbon reservoir ages in the Mediterranean Sea and Black Sea. Radio-carbon 42 (2), 271e280.

Solanki, S.K., Usoskin, I.G., Kromer, B., Schüssler, M., Beer, J., 2004. Unusual activityof the Sun during recent decades compared to the previous 11,000 years. Nature431, 1084e1087.

Soloviev, S.L., Solovieva, O.N., Go, C.N., Kim, K.S., Shchetnikov, N.A., 2000. Tsunamisin the Mediterranean Sea 2000 B.C.e2000 A.D. In: Advances in Natural andTechnological Hazards Research, vol. 13. Kluwer Academic Publichers, Dor-drecht, p. 239.

Soon, W., Velasco Herrera, V.M., Selvaraj, K., Traversi, R., Usoskin, I., Arthur Chen, C.-T., Lou, J.-Y., Kao, S.-J., Carter, R.M., Pipin, V., Severi, M., Becagli, S., 2014. A reviewof Holocene solar-linked climatic variation on centennial to millennial time-scales: physical processes, interpretative frameworks and a new multiple cross-wavelet transform algorithm. Earth-Sci. Rev. 134, 1e15.

Sorrel, P., Tessier, B., Demory, F., Delsinne, N., Mouaz�e, D., 2009. Evidence formillennial-scale climatic events in the sedimentary infilling of a macrotidalestuarine system, the Seine estuary (NW France). Quat. Sci. Rev. 28, 499e516.

Sorrel, P., Debret, M., Billeaud, I., Jaccard, S.L., McManus, J.F., Tessier, B., 2012.Persistent non-solar forcing of Holocene storm dynamics in coastal sedimen-tary archives. Nat. Geosci. 5, 892e896.

Steinhilber, F., Beer, J., Fr€ohlich, C., 2009. Total solar irradiance during the Holocene.Geophys. Res. Lett. 36, L19704.

Stuiver, M., Braziunas, T.F., 1993. Sun, ocean, climate and atmospheric 14CO2: anevaluation of causal and spectral relationships. Holocene 3 (4), 289e305.

Stuiver, M., Braziunas, T.F., Grootes, P.M., Zielinski, G.A., 1997. Is there evidence forsolar forcing of climate in the GISP2 oxygen isotope record? Quat. Res. 48,259e266.

Taricco, C., Ghil, M., Alessio, S., Vivaldo, G., 2009. Two millennia of climate vari-ability in the Central Mediterranean. Clim. Past 5, 171e181.

Tessier, B., Certain, R., Barusseau, J.-P., Henriet, J.-P., 2000. �Evolution historique duprisme littoral du lido de l'�etang de Thau (S�ete, Sud-Est de la France). Mise en�evidence par sismique r�eflexion tr�es haute r�esolution. Comptes Rendusl'Acad�emie Sci. Paris 331, 709e716.

Tesson, M., Labaune, C., Gensous, B., 2005. Small rivers contribution to the Qua-ternary evolution of a Mediterranean littoral system: the western gulf of Lion,France. Mar. Geol. 222e223, 313e334.

Thomson, D.J., 1982. Spectrum estimation and harmonic analysis. Proc. IEEE 70 (9),1055e1096.

Thomson, D.J., 1990. Time series analysis of Holocene climate data. Philos. Trans. R.























































































































































































































































Soc. Lond. Ser. A, Math. Phys. Sci. 330, 601e616.Tinti, S., Armigliato, A., Pagnoni, G., Zaniboni, F., 2005. Scenarios of giant tsunamis of

tectonic origin in the Mediterranean. ISET J. Earthq. Technol. 42 (4), 171e188.Toomey, M.R., Donnelly, J.P., Woodruff, J.D., 2013. Reconstructing mid-late Holocene

cyclone variability in the Central Pacific using sedimentary records from Tahaa,French Polynesia. Quat. Sci. Rev. 77, 181e189.

Trigo, I.F., Bigg, G.R., Davies, T.D., 2002. Climatology of cyclogenesis mechanisms inthe Mediterranean. Mon. Weather Rev. 130, 549e569.

Trouet, V., Esper, J., Graham, N.E., Baker, A., Scourse, J.D., Frank, D.C., 2009. Persistentpositive North Atlantic oscillation mode dominated the medieval climateanomaly. Science 324, 78e80.

Trouet, V., Scourse, J.D., Raible, C.C., 2012. North Atlantic storminess and Atlanticmeridional overturning circulation during the last Millennium: reconcilingcontradictory proxy records of NAO variability. Glob. Planet. Change 84e85,48e55.

Usoskin, I.G., Kovaltsov, G.A., 2004. Long-term solar activity: direct and indirectstudy. Sol. Phys. 224, 37e47.

Van Geel, B., Buurman, J., Waterbolk, H.T., 1996. Archaeological and palaeologicalindications of an abrupt climate change in The Netherlands, and evidence forclimatological teleconnections around 2650 BP. J. Quat. Sci. 11 (6), 451e460.

Van Vliet-Lano€e, B., Penaud, A., H�enaff, A., Delacourt, C., Fernane, A., Goslin, J.,Hall�egou€et, B., Le Cornec, E., 2014. Middle- to late-Holocene storminess inBrittany (NW France): part II e the chronology of events and climate forcing.Holocene 24 (4), 434e453.

Vella, C., Provansal, M., 2000. Relative sea-level rise and neotectonic events duringthe last 6500 yr on the southern eastern Rhone delta, France. Mar. Geol. 170,

27e39.Viau, A.E., Gajewski, K., Sawada, M.C., Fines, P., 2006. Millennial-scale temperature

variations in North America during the Holocene. J. Geophys. Res. 111, D09102.Vieira, L.E., Solanki, S.K., Krivova, N.A., Usoskin, I., 2011. Evolution of the solar

irradiance during the Holocene. Astron. Astrophys. 531, A6.Vonmoos, M., Beer, J., Muscheler, R., 2006. Large variations in Holocene solar ac-

tivity: constraints from 10Be in the Greenland ice core project ice core.J. Geophys. Res. 111, A10105.

Wagner, G., Beer, J., Masarik, J., Muscheler, R., Kubik, P.W., Mende, W., Laj, C.,Raisbeck, G.M., Yiou, F., 2001. Presence of the solar de Vries cycle (~205 years)during the last ice age. Geophys. Res. Lett. 28 (2), 303e306.

Wanner, H., Beer, J., Bütikofer, J., Crowley, T.J., Cubasch, U., Flückiger, J., Goosse, H.,Grosjean, M., Joos, F., Kaplan, J.O., Küttel, M., Müller, S.A., Prentice, I.C.,Solomina, O., Stocker, T.F., Tarasov, P., Wagner, M., Widmann, M., 2008. Mid- toLate Holocene climate change: an overview. Quat. Sci. Rev. 27, 1791e1828.

Wanner, H., Solomina, O., Grosjean, M., Ritz, S.P., Jetel, M., 2011. Structure and originof Holocene cold events. Quat. Sci. Rev. 30, 3109e3123.

Woodruff, J.D., Donnelly, J.P., Okusu, A., 2009. Exploring typhoon variability over themid-to-late Holocene: evidence of extreme coastal flooding from Kamikoshiki,Japan. Quat. Sci. Rev. 28, 1774e1785.

Yiou, P., Baert, E., Loutre, M.-F., 1996. Spectral analysis of climate data. Surv. Geo-phys. 17, 619e663.

Zoppi, U., Albani, A., Ammerman, A.J., Hua, Q., Lawson, E.M., Serandrei Barbero, R.,2001. Preliminary estimate of the reservoir age in the lagoon of Venice.Radiocarbon 43 (2A), 489e494.












































































major storm periods and climate forcing in the western

Documents