at SciVerse ScienceDirect

Quaternary International 288 (2013) 81e96

Contents lists available

Quaternary International

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

The Grotta dei Fiori (Sardinia, Italy) stratigraphical successions: A key for inferringpalaeoenvironment evolution and updating the biochronology of the Pleistocenemammalian fauna from Sardinia

Rita Teresa Melis a,*, Bassam Ghaleb b, Roberto Boldrini c, Maria Rita Palombo c

aDipartimento Scienze Chimiche e Geologiche, Università degli Studi di Cagliari, via Trentino 51, 09127 Cagliari, ItalybUniversité du Québec à Montréal Centre GEOTOP CP 8888, Succ. Centre-Ville Montréal, Québec, Canada H3C 3P8cDipartimento di Scienze della Terra, Università di Roma “La Sapienza”, CNR, Istituto di Geologia Ambientale e Geoingegneria, Piazzale A. Moro 5, 00185 Roma, Italy

a r t i c l e i n f o

Article history:Available online 26 April 2012

* Corresponding author. Dipartimento Scienze ChiUniversity, via Trentino 51, 09147 Cagliari, Italy.

E-mail address: rtmelis@unica.it (R.T. Melis).

1040-6182/$ e see front matter � 2012 Elsevier Ltd adoi:10.1016/j.quaint.2012.04.032

a b s t r a c t

In Sardinia (western Mediterranean, Italy) studies of Quaternary climatic changes, faunal turnovers, andevolution of environments have mostly been performed autonomously in various scientific fields. TheGrotta dei Fiori Cave, GFC (Carbonia, south-western Sardinia) is one of the numerous limestone caves insouth-eastern Sardinia that contain deposits of large and small mammals with intercalated flowstones. Amultidisciplinary study (palaeontological, sedimentological, soil micromorphological analyses, and U/Thdating) was performed on three stratigraphical successions (SA, SC and SD) for a finer time resolutionand palaeoclimatical setting of the Quaternary fossiliferous deposits cropping out in the GFC karst cave.Relying on U/Th dating, results obtained by the stratigraphical and soil micromorphological analyses,integrated with a previously performed stable isotope analysis, indicate that the studied fossiliferoussuccessions were deposited under warm, alternate wet and dry climatic conditions, albeit at differenttimes. In particular, sediments of SC, older than 500 ka, were possibly deposited during the early MiddlePleistocene, those of SA in a time interval from about 500 to 250 ka, while the SD is older than about350 ka. The presence of Microtus (Tyrrhenicola) henseli in deposits older than 500 ka is of particularinterest, with regard to time and mode of the endemic vole evolution and to the biochronologicalarrangement of the late Early to Middle Pleistocene Sardinian fauna.

� 2012 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

In Sardinia (westernMediterranean, Italy) studies of Quaternaryclimatic changes, palaeoenvironment evolution and mammalianfauna dynamics and turnover have mostly been performed auton-omously in various scientific fields. Some disagreements amongspecialists about the Sardinian Quaternary stratigraphy are some-times due to the scantiness of either precise biostratigraphicalconstraints or reliable numerical dating of sedimentary deposits(see e.g. Andreucci et al., 2010; Catto, 2010; Coltorti et al., 2010;Thiel et al., 2010 as regard to coastal deposits). In addition, remainsof Quaternary mammal fossils have been mainly found in deposits(cave, karst fissure fillings, isolated pockets cropping out in quarriesor natural sections, fluvial-lacustrine, aeolian or beach deposits oflimited thickness and extension) whose age is only roughly known

miche e Geologiche, Cagliari

nd INQUA. All rights reserved.

or poorly defined. This makes correlations difficult and any bio-chronological setting fairly approximate. The scantiness of firmchronological constraints has been a major limiting factor inunravelling the evolution of Sardinianmammalian fauna during thePleistocene/early Holocene, a time characterised by large-scalefluctuations in global climate. The problem of correctly arranginglocal faunal assemblages (LFAs) in a chronological succession(taking into account the time of first and last appearances of taxa,their evolutionary degree as well as the structure of faunalassemblages) is amplified in the case of the insular impoverishedand disharmonic faunas by the rapid evolution of endemic settlersgenerally followed by a period of evolutionary stasis, during whichmorphological and dimensional variation might increase.

On the basis of available data the latest Early Pleistocene to EarlyHolocene Sardinian LFAs, characterized by the presence of theendemic vole Microtus (Tyrrhenicola), have been arranged indiscrete faunal subcomplexes, FSC (Orosei 2 and Dragonara, splittedin Dragonara 1, Dragonara 2 and Dragonara 3 by Palombo and Rozziunpublished data) (Table 1). The Orosei 2 FSC includes few localitiesand has a gap in documentation. Therefore the actual beginning of

Table 1Chronological range chart of mammalian taxa documented in the Pliocene to Early Holocene Sardinian local faunal assemblages (updated and modified from Palombo,2009a,b).

R.T. Melis et al. / Quaternary International 288 (2013) 81e9682

the FSC is affected by some uncertainties, for example the occur-rence of the primitive vole, likely derived from an advancedAllophaiomys species (i.e. Allophaiomys ruffoi Pasa, 19) (Minieriet al., 1995; Marcolini et al., 2006a). This could suggest that itsarrival was earlier than 1.3e1.2 Ma. On the other hand, other taxacharacteristic of this faunal complex (e.g. the first representative ofthe “wild dog” Cynotherium found at Monte Tuttavista togetherwith Microtus (Tyrrhenicola) sondaari (Marcolini et al., 2006a, and,may be, Praemegaceros aff. P. sardus (Van der Made and Palombo,2006) from Sadali) might have entered Sardinia during thepronounced post-Jaramillo lowering of sea level (?MIS 24).

In the Orosei 2 FSC pre-existing taxa such as Rhagapodemusminor (Brandy, 1978), Prolagus figaro López Martinez, 1975, Macacacf. M. majori Azzaroli, 1946, Pannonictis and a leporid are recordedfor the last time. In the following Dragonara 1 FSC, archaic canidsand deer occurred together with advanced small mammals (Talpatyrrhenica Bate, 1945, “Asoriculus” similis (¼ Nesiotites similis)(Hensel, 1855), Microtus (Tyrrhenicola) henseli (Major, 1882), Rhag-amys orthodon (Hensel, 1856) and Prolagus sardus (Wagner, 1829)).The age of the transition from Orosei 2 to Dragonara 1 FSC is highlyproblematic due to the incompleteness of stratigraphical andpaleontological evidence. The only chronological constraint is a ESRdating of about 450� 20% ka B.P. given for the Praemegaceros sardusfound at Santa Lucia (Iglesias, south-western Sardinia) (Van derMade and Palombo, 2006).

The following Dragonara 2 FSC includes small mammals alreadypresent in the Dragonara 1 FSC, as well as advanced representativesof the endemic large mammal lineages (Cynotherium sardous Stu-diati, 1857 and Praemegaceros cazioti (Depéret, 1897)) and newsettlers such as the dwarf elephantMammuthus lamarmorai (Major,1883) and otters (cf. Palombo, 2009a; Palombo et al., 2012). At that

time of Dragonara 3 FSC (Last Glacial-Early Holocene), Mammuthusand otters were not recorded, Prolagus sardus is characterized by anincrease in size, and P. cazioti shows a moderate decrease in sizeand proportionally shortened metapodials as documented at Cor-beddu Cave. The species survived in Sardinia until the Early Holo-cene as shown by the age of about 7500 years BP reported fora skull of P. cazioti from Juntu cave (Benzi et al., 2007), while smallmammals lasted until protohistorical (M. henseli) and historicaltime (Prolagus sardus) (Palombo, 2009a and references therein).

Although in Sardinia most of Quaternary vertebrate remainscome from karst and cave filling sediments caves, studies on caveshave focused primarily on the speleogenesis evolution, mineralogyand morphology of speleothems (De Waele et al., 2001; De Waeleand Forti, 2006; Pagliara et al., 2010). Quite surprisingly, precisedates of speleothems are scarce and paleoenvironmental recon-structions have rarely been investigated. On the other hand, karstfilling sediments are known to offer good information about theQuaternary climatic variations, which could be a useful support topaleoenvironmental reconstructions and biochronological studies,especially when precise dates are not available (Sala, 1980; Ek andQuinif, 1988).

Preliminary multidisciplinary studies (stratigraphical, palae-ontological, and geochemical) already performed on a fossiliferoussedimentary sequence in Grotta dei Fiori cave (south-westernSardinia), allowed a tentative attribution of the stratigraphicsuccession to the early Middle Pleistocene (?MIS 11) (Boldrini et al.,2010 and references therein).

The aim of this paper is to provide a more consistent chrono-logical framework for the deposits of the three main stratigraphicalsections thus far identified in different chambers of the Grotta deiFiori cave, by means of new U/Th dating, stratigraphical,

Fig. 2. Geological sketch of Sulcis area and position of Grotta dei Fiori Cave along theRio Cannas Valley; 1) Quaternary deposits; 2) Oligo-Miocene vulcanic products; 3)Eocene-Lower Oligocene Cixerri Formation; 4) Monte Argentu Formation (?Upper-Middle Ordovician) metasandstones; 5) Lower Cambrian-Lower Ordovician (CabitzaFormation) shales, metasandstones; 6) Lower Cambrian (Gonnesa Formation) “DolomiaRigata”(Grey dolostones), and “Calcare Ceroide” (darker massive limestone); 6)Cambro-Ordovician (Nebida Formation) shales; 7) stratification and its inclination25e35�; 8) main faults (modified from Carmignani et al., 2001).

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 83

sedimentological, micromorphological and palaeontological anal-ysis. This could also provide an account of the Sardinian environ-ments during the Middle Pleistocene.

2. Geological background

The Grotta dei Fiori Cave (GFC) is located in the Sulcis region(south-west Sardinia, Fig. 1) characterized by Palaeozoic sedimentsdeposited during the Caledonian cycle from the Lower Cambrian tothe Lower Ordovician, and the Variscan cycle from Upper Ordovi-cian to Upper Carboniferous (Carmignani et al., 2001) (Fig. 2).

The Cambro-Ordovician sedimentary succession is composed ofthe Nebida Formation (Lower Cambrian), consisting of siliciclasticsedimentary rocks with carbonate intercalations; the GonnesaFormation (Lower Cambrian) characterised by “Dolomia Rigata”(grey dolostones), and “Calcare Ceroide” (darker massive limestone)(Pillola, 1989; Bechstadt and Boni, 1996); the Cabitza Formationconsisting of shales and metasandstones (Lower Cambrian-LowerOrdovician), and the Monte Argentu Formation (?Upper-MiddleOrdovician) represented by metasandstone and metasiltites (Leoneet al., 1991, 2002).

The Tertiary sediments outcropping in the area are clearlyseparated from the Palaeozoic rocks by EeW faults of the Oligo-Miocene age of regional importance. The observed lithologies arepresented by continental sediments of the Cixerri Formation(Eocene-Lower Oligocene) and alkaline-calcalkaline volcanicproducts of Oligo-Miocene volcanic activity.

The Quaternary in the region consist of eolian-alluvial-slopesediments and karst deposits. These latter deposits are charac-terised by speleothem concretions, collapsed breccia and terra rossawith fossil remains of terrestrial fauna.

The actual structural settlement of the area was established bydeformative tectonic events: the Sardinian Phase (Lower-MiddleOrdovician) with a tectonic style with an EeW direction and theVariscan Phase that produced a straight stress with a NeS directioncausing intense cleavage of the shales. The Variscan uplifts werefollowed by several pulses of extensional tectonics causingrepeated openings of fractures during the Mesozoic as well ascirculation of hydrothermal fluids (Boni et al., 1992, 2001). Thisevent was followed by deep karstification in the Cambriancarbonate rocks (Forti and Perna, 1982). The last tectonic phase,connected to the Alpine Orogenesis (OligoceneeMiocene),produced striking EeW faults and associated NNEeSSW faults,

Fig. 1. Location of Grotta dei Fiori Cave in South-West Sardinia (Italy).

related to the rifting of the SardinianeCorsican microplate(Carmignani et al., 2001).

3. The Grotta Dei Fiori Cave (GFC)

The GFC is situated on the western side of the Rio Cannas Valleyin the Paleozoic relief delimited by Cixerri graben to the north andCarbonia-Giba-Narcao graben to the south-east, near the town ofCarbonia and about 10 km west of the Golfo di Palmas coast. Thegeological structure and underground water circulation affectedthe morphology of the GFC, which has a maximum depth of about45m and extends for about 800m2mainly in the Dolomia Rigata” ofthe Gonnesa Formation (Lower Cambrian). GFC consists of gallerieselongated along the principal fault directions and chambersoccurring at fault intersections (NNSeSSW and EeW), ceilingchannels, and scallops clearly showing the flow of waters (cf. Bartaand Tarnai, 1998; Klimchouk, 2007, 2009).

The karst system of GFC, sub-horizontal and dendritic, devel-oped at different levels (Fig. 3). It is now accessed by three artificialentrances (EA, EB, EC), made by miners for onyx extraction at analtitude of 135 m, and by a natural opening created by the collapseof the roof in the Main Hall (MH) (Villani, 2001). Karst evolution ofGFC has been highly complex and no firm evidence exists to detectthe genesis or age of the cave. However, according to the studiescarried out on many caves of the Sulcis-Iglesiente region (Forti andPerna, 1982; Civita et al., 1983; De Vivo et al., 1987; Bini et al., 1988;Cortecci et al., 1989; Ludwig et al., 1989; De Waele and Forti, 2006;Pagliara et al., 2010), it is possible to hypothesize that the first karstphenomena of GFC occurred during the Upper Cambrian/LowerOrdovician when a large part of the carbonate formations wereexposed on the land surface (Forti and Perna, 1982; Carmignaniet al., 2001).

Fig. 4. Schematic outline of the first lower molar (M1) of a vole showing thenomenclature used in text.

Fig. 3. Cave map of the Grotta dei Fiori system showing the location of the sections.Section A (SA), Section C (SC), and Section D (SD); entrance A (EA), entrance B (EB),entrance C (EC). (Modified from: Gruppo Ricerche Speleologiche “E. A. Martel”Carbonia, 1977).

Fig. 5. Schematic outline of the main morphotypes of the first lower molar (M1) ofMicrotus (Tyrrhenicola) henseli (modified from Mezzabotta et al., 1995).

R.T. Melis et al. / Quaternary International 288 (2013) 81e9684

During the Upper Triassic and OligoceneeMiocene the GFC mayhave been affected by another phase of intense karstification, suchas that described by Pagliara et al. (2010) in the Iglesiente caves (seeDe Vivo et al., 1987; Cortecci et al., 1989; Ludwig et al., 1989), andtriggered the deposit of some speleothems (De Waele and Forti,2006; Pagliara et al., 2010).

The cavewas then filled by red clay sediments and subsequentlycompletely emptied by increasing water circulation mainly duringthe Early Pleistocene tectonic activity. Strips of red clay sedimentsare still present on the ceiling of the chambers. During the late Earlyand Middle Pleistocene the cave was partially refilled by rockfall,spelothems and alluvial deposits, which were affected by erosionduring the Late Pleistocene.

4. The main sedimentary successions: stratigraphy, UeThdating; Palaeontology

Three sedimentary successions (SA, SC, SD) were studied, two(SA and SC) are located in the Main Hall (MH) (Fig. 3) and one (CD)in the Chamber of Flowers (FC) of the GFC (Fig. 3).

4.1. Methods

4.1.1. Sedimentological and micromorphological analysisSamples from each layer were taken for sedimentological and

micromorphological analysis. Granulometric analysis of the sand(<2 mm), silt and clay fractions used the pipette method (Violante,2000). The percentage of calcium carbonate was determined witha Dietric-Fruhling Calcimeter following Violante (2000). Mineral-ogical analysis was carried out on bulk powder (randomly oriented)

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 85

and clay fractions (parallel oriented specimens separated bypipette) by means of a Spectro X-Lab XRF (Philips PW 1710) withCuka radiation. Qualitative clay-mineral identification of <2 mmfraction was obtained by search match ADP software (Laviano,1987).

Undisturbed samples were collected for micromorphologicalstudy. Thin sections were analysed using a polarising microscope.Micromorphological features were described according to Bullocket al. (1985).

4.1.2. Palaeontological analysisAmong the mammalian species found in the fossiliferous layers

of GFC sedimentary successions, this study was focused on therichest and best-preserved specimens of the arvicolid Microtus(Tyrrhenicola), because the evolutionary morphological anddimensional trends of this endemic lineage are, at least at themoment, the most studied and best-known among PleistoceneSardinian taxa.

Morphological and morphometrical analyses on the first lowermolar (M1)ofMicrotus (Tyrrhenicola)were focusedoncharacteristics(relative length of the anteroconid complex (ACC)), morphology of

Fig. 6. Sketch of main measurements of the first lower molar (M1) of Microtus (Tyr-rhenicola). V1 ¼width of mesial part; V2 ¼ total width; V3 ¼ length of posterior part ofBRA2; V4 ¼ length of posterior part of BSA2; V5 ¼ maximum length between AC andT4T5 axis; V6 ¼ total length Elongation of TTC; V9 ¼ distance BRA3 from T4T5 axis;V10 ¼ distance BRA4 from T4T5 axis; V11 ¼ distance LRA4 from T4T5 axis;V12¼maximum length betweenAC and T4T5 axis; V13¼ distance LSA4 fromT4T5 axis;V14 V13 ¼ distance LSA4 from T4T5 axis; V14 ¼ distance LSA5 from T4T5 axis;V15 ¼ distance BSA4 from T4; V17 ¼ distance BRA3 from T4; V18 ¼ distance BRA4 fromT4; V19¼ distance LRA4 from T4; V20¼ distance LRA5 from T4; V21¼Maximun lengthbetween T4 and T5; V22 ¼ distance LSA5 from T4; V23 ¼ distance LSA6 from T4;V24¼width of mesial part of T4T5 axis; V25 ¼maximum length between AC and T2T3axis; V26 ¼ width of mesial part of T2T3 axis; V27 ¼ maximun length between T2 andT3; V31 ¼ minimum distance LRA4-BRA3; V32 ¼ minimum distance BRA3-LRA3;V33 ¼ minimum distance LRA3-BRA2; V34 ¼ minimum distance BRA2-LRA2;V35 ¼ angle between T6-T7 axis and tangent to PL (posterior Lobe); V36 ¼ anglebetweenT4-T5 axis and tangent to PL; V37¼ angle betweenT2-T3 axis and tangent to PL.

the anterior cap (AC), degree of pinching of the neck, development ofthe sixth triangle (T6), presence of the seventh (T7) and ninth (T9)triangles) regarded as the most useful for discriminating betweenarchaic and advanced morphotypes (Fig. 4).

M1 from the GFC were attributed to the main morphotypesrecognized by Mezzabotta et al. (1995, 1996) according to theirmorphological features (Fig. 5). The frequency of each morphotypein each layer was calculated.

Measurements were taken following Van der Meulen (1973),Marcolini et al. (2006a,b) and further modifications by Boldrini(2010) (Fig. 6). Principal Component Analysis (PCA) was carriedout using as input all measurements as well as selected variables(13 linear measurements).

To calculate the SDQ index (Schmelzband-Differenzierungs-Quozient) the thickness of the enamel of M1 was measuredaccording to Heinrich (1978). The SDQ index was calculated as themean of the ratio of the thicknesses of posterior and anteriorenamel bands in each triangle. Morphological traits were analysedand measurements were taken on microphotographs obtained bymeans of a scanning electronic microscope (FEI Quanta, jsm-5600lv) in low vacuum mode.

4.1.3. UeTh datingIn order to investigate the chronology of the GFC, six samples

were taken from the flowstones of: sections SA, SC, SD, section cutby mining activity and flowstone above debris slope. These weredated by the uraniumethorium method.

For U/Th dating, pieces of flowstone ranging from 2 g to 7 g werecut using a diamond saw for each UeTh analysis. The flowstonepieces were dissolved using nitric acid in a Teflon beaker intowhicha weighted amounts of mixed spike 233Ue236Ue229Th had beenplaced and evaporated. Samples were dissolved using concentratedHNO3 and around 10 mg of iron carrier was added. U and Th werecoprecipitated with Fe(OH)3. The precipitate was washed withwater, dried and then dissolved in 6NHCl. TheUeTh separationwasperformed on a 2ml volume AG1X8 anionic resin. Thorium fractionwas recovered by elution with 6 N HCl and the U and Fe fraction by

Fig. 7. Sketch section of Main Hall showing relationships between sediment and datedspeleothems. Sections of the chambers, have been drawn not to scale from surveyswhile deposits and speleothems, have been simplified and exaggerated for visibility. Toimprove clarity of the UeTh ages are usually shownwithout errors, which may be readon Table 1. Legend: 1) carbonate rock; 2) coarse sediments; 3) clayey sediments; 4)flowstone; 5) debris cone; 6) rockfall; 7) fossil remains.

Fig. 8. The Main Hall; a) view of debris cone in the western end of the Upper Chamber; b) section of flowstones in the Upper Chamber cut by mining activity; c) flowstoneoverlaying the debris cone.

R.T. Melis et al. / Quaternary International 288 (2013) 81e9686

H2O. The U fraction was purified on a 0.2 ml volume U-Teva(Elchrom industry) resin. The Fe was eluted with 3 N HNO3 and theU fractionwith 0.002 N HNO3. Thorium purificationwas carried outon a 2 mL AG1X8 resin in 7 N HNO3 and eluted with 6 N HCl. Afterdrying, a final purification was performed on a 0.2 ml AG1X8 resinin 7 N HNO3 and Th was eluted with 6 N HCl. UeTh measurementswere performed using a VG sector TIMS fitted with an electrostatic

Table 2Uranium-series data for six flowstones from Grotta dei Fiori Cave. Samples are given aflowstone above the debris cone in the Upper Chamber of Main Hall; SAA2 flowstone of seflowstone of section D in the Chamber of Flowers.

Sample 238U ppp � 232Th ppb � (234U/23

SCC15 74.1119 0.2907 3.3071 0.0168 1.0872CDF1 33.2359 0.4576 3.5469 0.0141 1.0909SAA2 2009.0968 7.5342 0.7196 0.0034 1.0147FL 1 85.0821 0.2487 4.0598 0.0020 1.1815FL 2 163.1009 0.5071 7.3820 0.0501 1.0330SDD17 812.8249 3.2722 0.6253 0.0057 1.3253

filter and a Daly ion counter. The U and Th fractions were depositedon a single zone-refined rhenium filament between two layers ofcolloidal graphite. U and Th isotopes were measured in peakjumping mode on the Daly counter. The overall analytical repro-ducibility was estimated from replicate measurement of a coralfrom Timor Island that dates from the last interglacial age anda uraninite standard. Precision is usually better than �0.5%.

prefix: SCC15 flowstone of the section C in the Lower Chamber of Main Hall; CDFction A; FL1 and FL2 flowstones in the section cut by mining in the Main hall; SDD17

8U) � (230Th/238U) � 230Th-age �0.0135 1.1246 0.0125 >500 ka0.0041 1.0023 0.0072 248.2610 83450.0040 1.0491 0.0059 >500 ka0.0069 0.8479 0.0136 130.8410 44230.0038 1.0835 0.0144 >500 ka0.0069 1.3611 0.0164 327.0062 4514

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 87

4.2. The Main Hall (MH)

The Main Hall contains a number of sequences, two of whichhave been selected for study because of their completeness andinterest in investigating the relationships between sediments,speleothems, and the accumulation history of the fossiliferousdeposits. MH is one of the largest halls lying in the upper sub-horizontal layer of the GFC karst system. MH consists of twosuperimposed chambers (upper and lower), connected by a verticalshaft, likely to have originated after a collapse of the ceiling of thelower chamber (Fig. 7). In the lower chamber, remains of reddishfossiliferous deposits (SC), covered by a flowstone, are still presenton the walls. The deposits probably filled the whole chamber andwere eroded by water circulation after the ceiling collapse.

In the inner area of the upper chamber, extensive speleothems,several meters deep and covering a large part of the limestone floor,were affected by mining during the 1960s, when activity exposeda section of about 2 m (Fig. 8b). U/Th dating of the lowermost anduppermost flowstones (Table 2) indicates that the sedimentationoccurred from before 500 ka until about 130 ka. The absence ofdetrital deposits across the flowstone sequence suggests that atleast this part of the chamber had been closed during this time (cf.Moriarty et al., 2000). Below the natural entrance, remains of

Fig. 9. The SA section. a) View of SA section; b) thin section of A3 layer showing the section oaragonite level, B: a level of mosaic calcite crystals (XPL).

a debris cone, about 10 m thick, are present at the western end ofthe chamber (Fig. 8a). The deposit, although chaotic in the lowerpart, shows an éboulis ordonnée stratification characterised byalternate thin coarse (gravel) and fine (sand) layers, probablydeposited during a cold period (cf. Tricart and Cailleux, 1967; Pena,1998). The cone is overlaid by a flowstone CDF (dated to248.3 � 8.3 ka U/Th), extending from the ceiling to the floor(Fig. 8c).

The thin section of the flowstone shows alternate layers ofcalcite and aragonite that indicate changes in the depositionalenvironment caused by climatic oscillations (cf. Railsback et al.,1994; Frisia et al., 2002). Periods of minimum rain and warmclimate would have promoted the deposit of aragonite, whilstperiods with abundant rain and low temperatures would haveallowed the deposit of calcite (cf. Frisia, 2003; Baker et al., 2008).Similar aragonite-calcite couplets were observed by Pagliara et al.(2010) in some flowstones in caves north of the GFC (Iglesienteregion), and believed to be related to either changes in the climateconditions or to a climate charactherized by seasonal contrasts,which occurred in the area during the Quaternary. FollowingMoriarty et al. (2000) and Quinif (2006), the éboulis ordonnéesediments probably deposited under dry-cold climatic conditions,when a sparse vegetation covered the slope (MIS 8), while the

f a molar of Microtus (Tyrrhenicola); c) thin section of A2 flowstone, A: deposition of an

Fig. 10. Stratigraphic sketch of SA section. 1) Mud cracks; 2) flowstone; 3) angular andsub-angular clasts of carbonate and metamorphic rocks; 4) rockfall deposit with blocksand clast of limestone, dolomite, metamorphic rocks and flowstone; 5) fossil remains.

R.T. Melis et al. / Quaternary International 288 (2013) 81e9688

flowstone deposited during a wet period, when the vegetation wasdense (MIS 7). Actually, taking into account the U/Th dating, theflowstone was probably deposited during the transition fromMIS 8and MIS 7, when the climate was characterized by alternating coldandwarm phases (Tzedakis, 2005; Roucoux et al., 2006, 2008; Spötlet al., 2006).

4.2.1. The SA stratigraphical successionThe fossiliferous SA section, about 4 m thick, is located near the

artificial entrance A (Fig. 9a). The main results of the granulometricanalysis, CaCO3 and clay minerals are reported in Table 2.

4.2.1.1. Stratigraphy and micromorphology. The SA stratigraphicalsuccession (Fig. 10) includes, from the bottom to the top: A1)partially exposed, no fossiliferous red clayey sediments (5YR5/6Munsell, 1998) showing mud cracks filled by calcite; A2) a flow-stone about 20 cm thick; A3) a fossiliferous rockfall deposit, about2 m thick; A4) a relic of a detrial cone, showing an éboulis ordonéestratification characterised by alternate coarse and fine thin layers;A5) a flowstone about 20 cm thick. The A4 and A5 deposits are thesame cropping out in the upper chamber and described above.

The well-sorted A1 deposit probably accumulated in a pondedarea of the cave that received episodic slow moving water. Thepresence of abundant mud-cracks filled by calcite indicates thata dry period followed the red clayed deposit sedimentation.

The A1 layers are overlaid by a thick flowstone dated U/Th tomore than 500 ka. The flowstone’s thin section shows a successionof thin layers of columnar calcite and acicular aragonite, indicatingperiodical climate oscillations (cf. Railsback et al., 1994; Frisia et al.,2002) (Fig. 9c).

In the A3 rockfall deposit, fragments of the A2 flowstone anddolomite are included in a yellow red to reddish-yellow sandy loam(5YR5/6 to 5YR6/6 Munsell, 1998). In some parts of the depositremains of small mammals are present in some reworked brecciablocks, well-cemented by calcite (30% of CaCO3). Micromorpho-logical analyses of the breccia blocks revealed the presence ofangular lithorelicts of limestone and metasandstone, roundedlithoreticts of weathered volcanic rocks and sub-rounded feldsparand quartz minerals. The round-edged volcanic lithorelicts wereprobably deposited on the slope of the Cannas valley by the windduring an arid period, and might belong to the Oligo-Miocenevolcanic rocks outcropping in the Carbonia-Giba-Narcao graben(SWof Carbonia, Gonnesa basin). The presence in the breccia blocksof black nodules characterised by sharp boundaries, which havedifferent dimensions and random distribution, suggests that thenodules derived from soils developed along the slope at a period ofintense seasonal changes (cf. Federoff et al., 2010). The identifiedsmall mammal remains, Prolagus and Microtus (Tyrrhenicola), aretoo poorly-preserved to be studied in detail. In thin sections theyappear to be impregnated by microsparite and included in a sandymatrix, cemented by calcite (Fig. 9b). All in all, A3 deposit repre-sents the collapse of the outer wall of the MH. Taking intoconsideration the dating of the A2 (>500 ka) and A5 flowstone(248.3 � 8.3 ka U/Th), the collapse possibly happened during theearly Middle Pleistocene, before MIS 8.

4.2.2. The SC stratigraphical successionThe SC section, about 5 m thick, is located in the lower chamber

(Fig.11). In Table 3 themain results of granulometry, CaCO3 and clayminerals are reported.

4.2.2.1. Stratigraphy and micromorphology. Two units have beendepicted (Melis et al., 2002; Boldrini et al., 2010). The lower unit(UA), is characterised by fine sediments, while across the upper unit(UB) coarse deposits prevail. The transition from UA to UB is

marked by three thin flowstones, interbedded by two red clayeylayers, each about 10 cm thick (C7). The stratigraphical successionends with a flowstone dating before 500 ka U/Th.

UA consists of 5 clayey layers (C1, C3, C4, C5 and C6) from red(2,5 YR 4/8) to yellowish red (5 YR 5/8).

The clayey layer C1, partially exposed for about 70 cm, resultsfrom a sedimentation by slow moving water, shows on the toppolygonal mud cracks filled by calcite concretions (already namedC2 by Melis et al., 2002; Boldrini et al., 2010), as the layer A1 of SA,which indicates a dry phase.

The layers C3 (30 cm), C4 (60 cm), C5 (40 cm) and C6 (25 cm) arecharacterised by a low percentage of CaCO3 (5e7%) and a clayeytexture. The characteristics of C6 show that sedimentation likelyoccurred due to decantation of stagnant water coming mainly fromdripping water (70% of clay and mottles). Fossil remains werepresent in C3 and C5, while in C4 they are very scarce and in C6 arenot present. The lack of fossils in the C6 layer is probably due to thescarce water flow. Results obtained by XRD analysis show that theclay fraction contains kaolinite and illite in all layers, while a smallamount of chlorite is present in the layers C1 and C3. The origin ofclay mineral in caves is normally difficult to demonstrate. Cavesediments of detrital origin usually contain more than one claymineral (Foos et al., 2000) that derived from surface soils. Boeroand Franchini (1992) found that in xeric sites, illite and kaoliniteare themain claymineral phases in ‘terra rossa’ soil. In thin sectionsthe coarse fraction (silt to fine sand size) of all 5 layers is mainlycomposed of sub-rounded quartz, feldspars, angular lithorelicts of

Fig. 11. Stratigraphic sketch of the SC section. 1) Clayey sediments; 2) mud cracks; 3) sandy-clay-loam sediments; 4) angular and sub-angular clasts of carbonate and metamorphicrocks; 5) flowstone; 6) fossil remains (modified from Melis et al., 2002).

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 89

limestone and metasandstone, with strongly-weathered femicminerals. Some rounded fragments of weathered volcanic rocksandwell-rounded Fe/Mn nodules are present in all the layers. Theseelements, of detrital origin, were transported into the cave byepisodic low water flows. Red, sub-rounded pedorelicts consistingof red-yellow silt-clay coatings can be found (Fig. 12e). The pedor-elicts, originating from the erosion of outside rubefied soil whichoccurred during a period characterized by poor vegetation cover,show that the soil developed under warm and seasonal climateconditions.

This is confirmed by the presence of Fe/Mn nodules (Fig. 12c)that generally form in soil subjected to seasonal changes(Schwertmann, 1985; Schwertmann and Taylor, 1989; Elles andRabenhortst, 1994). The micro-graded bedding of clay and silt(Fig. 12d) in the C6 layer confirms that the sedimentary accumu-lation took place mostly due to decantation or slow-movingfloodwater, while the formation of fissures probably occurred

during a dry period after the sedimentation. The three thin flow-stones, that mark the passage with the upper Unit B, show alter-nating layers of calcite and aragonite confirming the depositoccurred under seasonal conditions (cf. Railsback et al., 1994).

The UB unit consists of 7 red clayey and coarse layers (C8, C9,C10, C11, C12, C13 and C14) (Fig. 12a). The C8 (15 cm thick), C10(35 cm) and C12 (20 cm) layers are characterized by red (2.5 YR 4/8)to yellow-red colours (5 YR 5/8), a clayey texture, numerousnodules of Fe/Mn and scarce fossil remains. However, layers C9, C11,C13 and C14 are composed of a coarse chaotic assortment ofangular and sub-angular limestone and metamorphic rock frag-ments in a clayey and sandy-clay loammatrix. The characteristics ofthis deposit and the presence of allochthonous rock fragments,correspond to sediments inside the caves caused by stronger waterflows, probably during storm events.

In all these layers, there are many remains of small mammalspresent, while few bones of a carnivore (Cynotherium sp., Palombo

Table 3Grain size distribution, CaCO3 and clay minerals of SA, SC and SD section in theGrotta dei Fiori Cave.

Sections Samplesnumber

Grain size < 2 mm CaCO3

(%)Clay minerals

Sand (%) Silt (%) Clay (%)

SA A4 72.6 9.6 17.9 46.2A3 54.5 35.6 55.8 45.5 Kaolinite-IlliteA1 26.5 17.7 55.8 10.1 Kaolinite-Illite

SC C14 49.2 15.0 35.8 17.0 Kaolinite-Illite-ChloriteC13 51.2 17.4 31.5 14.5 Kaolinite-IlliteC12 41.5 15.2 43.4 6.3 Kaolinite-Illite-ChloriteC11 35.3 19.4 45.4 18.0C10 36.0 18.1 45.9 4.0 Kaolinite-Illite-ChloriteC9 20.2 22.2 57.7 9.3 Kaolinite-IlliteC8 36.2 17.6 46.3 7.0 Kaolinite-IlliteC7 38.7 17.2 44.1 7.0C6 9.8 22.9 67.4 6.5 Kaolinite-Illite-ChloriteC5 20.2 22.2 57.7 7.0 Kaolinite-IlliteC4 31.5 19.4 49.1 7.0 Kaolinite-IlliteC3 36.7 14.3 49.1 5.3 Kaolinite-IlliteC2 49.2 10.3 40.6 28.3 Kaolinite-IlliteC1 32.2 15.3 52.6 15.3 Kaolinite-Illite-Chlorite

SD D16 62.5 26.8 10.7 18.0 Kaolinite-Illite-ChloriteD15 68.7 18.5 12.8 15.3 Kaolinite-IlliteD14 60.8 29.5 9.7 14.0 Kaolinite-Illite-ChloriteD13 62.4 20.1 17.5 16.2 Kaolinite-IlliteD12 60.1 20.9 19.0 9.0 Kaolinite-IlliteD11 70.0 14.6 15.4 7.2 Kaolinite-IlliteD10 61.0 29.5 9.7 16.3 Kaolinite-IlliteD9 72.0 15.6 12.4 8.4D8 40.5 35.2 24.3 7.3 Kaolinite-IlliteD7 40.2 45.0 14.8 6.9D5 61.2 22.0 16.8 7.4 Kaolinite-IlliteD4 22.5 55.5 22.0 6.5 Kaolinite-IlliteD3 32.5 48.5 19.0 7.0 Kaolinite-IlliteD2 20.5 58.5 21.0 7.1 Kaolinite-IlliteD1 40.3 45.5 14.2 7.0 Kaolinite-Illite

R.T. Melis et al. / Quaternary International 288 (2013) 81e9690

and Sotnikova unpublished data) and well-preserved bird remains(Pavia and Bedetti, 2003) were found in the C14 layer. In all layers,the clay minerals are kaolinite, Illite and chlorite that probablyderived from erosion of surface soils developed under seasonalclimate conditions (cf. Boero and Franchini, 1992).

At a microscopic scale, the sediment of the matrix in the coarselayers (C9, C11, C13 and C14) is heterogeneous sandy loam andconsists of limestone, metamorphic and volcanic rock fragmentswith different degrees of weathering. Mineral material is mainlycomposed of sub-angular quartz (from coarse to fine silt), feldsparsand few sub-rounded weathered amphiboles. The same mineralcomposition is present in the clayey layers (C8, C10 and C12).Angular and sub-rounded pedorelicts are present in all layers, andare characterised by yellow-red silt-clay coatings and rounded darkFeeMn nodules (Fig. 12a). Dark mottles in C13 and C14 indicatestagnant water.

The flowstone on the top of SC (C15) dates from prior to 500 kaU/Th, Thin section reveals that the flowstone is composed ofalternating aragonite and calcite layers. Calcite is predominant andof greater thickness. The calcite crystals appear organized ina columnar fabric. These crystals show few structural deformationsand grow in perfect optical continuity with the lower ones, despitethe presence of clay impurities, which did not stop the growth ofeach individual crystal.

All this data confirms that sediments of UB, like those of UA,came from the erosion of soil developed outside the cave underwarm and seasonal climate conditions. Taking into account the U/Th dating of the flowstone closing the SC succession, the hypothesisthat these deposits formed during MIS 15 cannot be ruled out.

4.2.2.2. The endemic vole Microtus (Tyrrhenicola). In the SC section,several fossiliferous layers yielded a fairly rich, unreworked sample

of small mammals, although not all layers are equally rich. LayersC3, C5, C10, C13 and C14 contain the richest bone concentrations. Ineach fossiliferous layer the following species were found: T. tyr-rhenica, “Nesiotites” similis,M. (Tyrrhenicola) henseli, R. orthodon andProlagus sardus. Together with as yet unstudied chiropters, the voleand the ochotonid are the most abundant in each layer, while themole is poorly represented by only a few humeri, sometimesincomplete. Scanty remains of a large canid close to Cynotherium(Palombo and Sotnikova unpublished data) and some birds (Ciconianigra, Gypaetus barbatus, Aquila sp., Columba livia, Pyrrhocoraxpyrrhocorax, Pyrrhocorax graculus, Corvus corone, Corvus corax,Emberiza cirlus/schoeniclus) were found in the uppermost fossilif-erous layer (Melis et al., 2002; Pavia and Bedetti, 2003).

As mentioned above, only Microtus (Tyrrhenicola) has beenstudied in detail (Boldrini, 2008, 2010; Boldrini et al., 2010). Resultsobtained by morphological and statistical bivariate and multivar-iate analyses performed on the samples of each fossiliferous layer,as well as the value of the SDQ index of each sample, on the onehand indicate that the Microtus (Tyrrhenicola) populations main-tained a substantial uniformity during the time of the SC deposits,but on the other stress the great variability of this endemic vole,both in dimensions and morphology. Nearly all M1 share somefeatures: T1 and T2 are both weakly confluent, and the buccal re-entrant angles (BRA) are more elongated than the lingual re-entrant ones (LRA), while the anteroconid complex (ACC) showsthe highest morphological variability. It ranges in shape fromarchaic types, with T6 just barely outlined, to the most advancedones, with well-outlined T7 and T9, sometimes irrespectively oftooth size. Moreover, the increase in dimensions, the augmentationof the ACC complexity, and a reduction in the wideness of the neckdo not strictly correlate. This is the case, for instance of some M1found in layer C11, characterized by small dimensions andadvanced ACC, and some others found in the successive C13 layer,which, conversely, are definitively larger in size but have a fairlyprimitive ACC (Fig. 13). A few of the smallest specimens fall in thedimensional field of the archaic M. (Tyrrhenicola) sondaari fromMonte Tuttavista (western Sardinia), but show features moreadvanced than those of these Monte Tuttavista specimens. Thefrequency of differentmorphotypes in each layer (Fig.14), as well asthe average value of the SDQ index (Fig. 15), stress the lack of anyevolution along the SC, being as the most archaic morphotype 1 isonly present in upper layers C11 and C13 of UB, while the speci-mens of the lowermost layer C1 and the uppermost layer C14 showthe highest and lowest SQD indexes respectively.

All in all, despite the small dimensions of some species, the voleof the SC succession can be confidently ascribed toM. (Tyrrhenicola)henseli because of its quite well-developed anteroconid, T6 and T7slightly to fully outlined, T9 frequently present, the neck betweenLRA4 and BRA3 not large and T4-T5 sometimes confluent.

Regarding the relative development of ACC compared to otherSardinian populations, the SC sample falls among the less advancedpopulations of this species, close to the Cava Alabastro (Minieriet al., 1995), Siniscola E (Mezzabotta et al., 1995) and Capo FigariII ones (Fig. 13). For the latter, an age of about 350 ka has beensuggested (cf. Van Der Made, 1999; Palombo, 2009a).

4.3. Chamber of Flowers (FC)

The Chamber of Flowers, which developed along the NWeSEfractures, is located on a GFC layer below that of the Main Hall.The FC is connected to other layers by steeply sloping passages.

Today it is possible to enter the chamber through an artificialentrance (EC) made by onyx miners. In the FC numerous speleo-thems and concretions, shaped like flowers, cover the walls. Largerockfall blocks and speleothem fragments cover a large part of the

Fig. 12. Particular of the SC section and microphotographs of sediments. a) View of Unit B of SC; b) pedorelict (P) in calcified matrix of C14 layer (XPL); c) redblack rounded Fe/Mnnodules (N) in the C10 layer (XPL); d) red micro-graded bedding of clay and silt in the C6 layer (PPL); e) Sub-rounded pedorelict characterised by red-yellow silt-clay coatings in theC7 layer (XPL). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 91

chamber floor, while a small portion of the clay sandy fossiliferousdeposits, recovered by a thin flowstone, once filling the FC, are stillpreserved on the innermost walls.

4.3.1. The SD stratigraphical successionThe section (SD) here described is located on the eastern part of

the FC. The stratigraphical data are synthesized in (Fig. 16) andTable 3.

4.3.1.1. Stratigraphy and micromorphology. The section SD, about8 m thick (Fig. 16), is composed of two units: the lower Unit A ischaracterised by fine sediments, whereas in the upper Unit B coarsedeposits prevail.

Unit A consists of 6 red clayey and sandy layers (D1, D2, D3, D4and D5). A few badly-preserved fossil remains are only present inthe D1 and D5 layers: D1 (15 cm thick) shows a red clayey texture,well-cemented; D5 (about 80 cm thick) is yellowish red (5 YR 5/8)with angular gravel in a sandy loam matrix. D2 is a clayey layer,50 cm thick, with intercalate thin (6 cm) well-cemented calciclayers, while D3 consists of 80 cm of red clayey layers alternatingwith sandy-clay loam layers. Mud cracks are present in the clayeyD4 layer, testifying to dry conditions.

Thin sections reveal that the sediments of this UA consist mainlyof well-sorted quartz sand, with limestone and metamorphic lith-orelicts. Some rounded volcanic rock fragments and feldspars are

common. A few fossil remains were observed in the sandy zone ofD1 and D5. Red pedorelicts of silt-clay coatings and dark, roundednodules occurred in all layers, suggesting that the sediments derivefrom the erosion of soils present outside the cave that developedunder a climate with distinct seasonal changes. Black mottles(Fig. 17a), are frequent in the D2 and D4 layers. Their diffusiveboundaries suggest local wet conditions. A thin flowstone (D6) of5 cm thickness separates the UA from the upper UB.

The lower unit UB is composed of coarse layers, most of themcontaining well-preserved, unreworked fossil remains (D11, D12,D13, D14 and D16), and clayey (D7 and D8) to sandy-clay loam(D9, D10 and D15) non-fossiliferous layers. Limestone and meta-morphic rock fragments in the coarse sediments indicate theirallochthonous origin. Moreover, the irregular distribution of theclasts suggests water transport probably during storm events. Thinsection observations show a considerable degree of lithologicalvariability within the layer. Limestone, metasandstone andvolcanic lithorelicts are present within the sandy silt fractions. Thepresence of sub-rounded red pedorelicts (Fig. 17b) and roundednodules of Fe/Mn in almost all the layers indicate downwardtransport of fine material coming from outside soils that devel-oped under a climate characterized by wet and dry seasons. Aflowstone, about 10 cm thick, U/Th dated at 327 � 24.51 ka,characterised, in thin section, by columnar calcite, closes the SCsedimentary succession.

Fig. 13. Total length (V6) of the first lower molar of plotted against the ratio of lengths of anteroconid complex and molar (A/L ¼ (V6�V3)*100/V6). Above: Microtus (Tyrrhenicola)henseli from Grotta dei Fiori. Below: comparison among the samples from Grotta dei Fiori (Sections C and D) and samples of Microtus (Tyrrhenicola) henseli from varios Sardinianlocalities ranging in age from the early Middle Pleistocene to the Holocene (data from Mezzabotta et al., 1995; Minieri et al., 1995; Turmes, 2003; Marcolini et al., 2005; Marcoliniet al., 2006a,b; Boldrini, 2010). MTV ¼ Monte Tuttavista.

R.T. Melis et al. / Quaternary International 288 (2013) 81e9692

4.3.1.2. The endemic vole Microtus (Tyrrhenicola). In SD section,scanty and badly-preserved remains of small mammals have beenfound in layers D1, D5, D11 and D13. Conversely, a fairly rich fossilrecord, including M. (Tyrrhenicola) henseli together with R. ortho-don, Prolagus sardus and very scarce remains of T. tyrrhenica and“Nesiotites” similis, has been found in layers D12, D14 and D16. Inthe same fossiliferous layers as yet unstudied chiropters werepresent, but no large mammals and birds.

A short account of the main results obtained by the morpho-logical and biometrical analysis of more than 40 M1 of Microtus(Tyrrhenicola) are presented here. In all layers, morphological andbiometrical traits of the M1 confidently identify them as M. (Tyr-rhenicola) henseli. It is worth noting that despite the average size(V6) of these M1 being smaller than the average size of M1 found

in the fossiliferous layers of SC succession (Fig. 13), in the samplefrom SD section, the frequency of the advanced morphotype ishigher and no specimens belong to the most archaic morphotype 1(Fig. 14). Specimens with complex (but relatively short) ACC, quitelarge and asymmetrical AC, a narrow neck, well-outlined T6 andT7, and from fairly to well-outlined T9 are common. Moreover, itseems that in the uppermost layers, the frequency of advancedmorphotypes progressively increases. In layer D12 only morpho-types 2 and 3 are present, while in layers D14 and D16 morpho-type 4 is present with increasing frequency (Fig. 14). As observedin the SC succession, the SDQ index of M1 in SD succession doesnot show linear variations from the lower to the uppermost layer,and the values fall in the range of variability already defined forthe SC specimens.

Fig. 14. Istogramm showing the frequency of the different morphotypes of the firstlower molars of Microtus (Tyrrhenicola) henseli recognised in each fossiliferous level ofthe SC stratigraphical section.

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 93

5. Remarks

The question of how the different depositional environmentswithin the cave system might have reacted to regional climaticvariability is not yet clear.

Studies on the reconstruction of past environments throughcave deposits (Brain, 1995; Moriarty et al., 2000; Cuenca-Bescóset al., 2005, 2009; Pickering et al., 2007) point out that the prin-cipal guides for alternating accumulation and erosional episodesare assumed to be external factors, predominantly changingclimatic conditions during glacial and interglacial cycles (Quinif andMaire, 1998; Quinif, 2006). Moreover, several studies (Bailey andWoodward, 1997; Goldberg, 2000; Woodward and Bailey, 2000)have shown that the striking spatial variations within individualstratigraphic units in the amount and morphology of coarse debris,its thickness, its internal assemblage, or the carbonate content ofthe fine fraction, often suggest that the climatic control on sedi-mentation styles is strongly obscured by local factors and thereforenot precisely detectable.

Evidence for rapid shifts in depositional conditions has beententatively related to abrupt climate changes (Woodward, 1997a,1997b). Thus, while study of the coarse fraction can perhapsreveal some information about climate, better indicators of envi-ronment change can be obtained from the finer fraction usingmicroscopic techniques (Goldberg and Sherwood, 2006).

Fig. 15. Boxplots showing the range of variation of SDQ index of the first lower molars of M

The study of the GFC fill sediments revealed the depositionalhistory of the cave and its relationships with the external envi-ronment. The results obtained indicate that the study of cavedeposits could be a useful support in inferring the paleoenvir-onmental conditions outside the cave, and on the other hand openup new prospects regarding time and modes of evolution of the M.(Tyrrhenicola) endemic lineage.

5.1. Stratgrafical context and paleoenvironmental inferences

The two sections, SC and SD, even though they are located intwo different karst layers, evolved similarly. Their origins are con-nected to the deposit of sediments carried by low energy waterflow coming from the erosion of soil outside the cave. The lowenergy water flow could perhaps justify the smaller percentage offossil remains present in the lower units (Units A of SC and SD).

The absence of fossils in the clayey layers is instead connected toa situation of stagnant water coming fromwater dripping, as madeevident by the occurrence of mottles. The presence of mud cracksfilled with calcite on the top of some clayey layers of the SA, SC andSD, show an interruption of the depositing and dry conditions.

On the other hand, the characteristics of the pedorelicts, Fe/Mn,and nodules of illite and kalinite, give evidence to a climate withseasonal contrast. Moreover, changes in the vegetation coverageaffected the slope processes. Dense vegetation could have reduced,or perhaps stopped, the sediment supply inside the cave.

The Upper units (UB) of the successions SC and SD, which arerich in fossil remains, show a change in the external conditions,probably in an arid time. These conditions caused a reduction in thevegetation and an increase in the erosion of the slope, as shown bychaotic coarser deposits, which were transported by water witha higher energy.

Given that the hypothesis that deposits of SC succession formedduringMIS 15 cannot be ruled out, those of the SD successionmighthave been deposited during MIS 9 or MIS 11. This hypothesis issupported by the values of d18O obtained by analyzing samples ofsmall mammals from GFC sections, which indicate that at the timeof deposition of the fossiliferous layers, the temperature waswarmer than at present (cf. Boldrini, 2010; Boldrini et al., 2010).

5.2. New evidence on M. (Tyrrhenicola) henseli evolutionarypatterns

M. (Tyrrhenicola) henseli is an endemic fossil vole, widespread inSardinia and Corsica during the Middle Pleistocene to Holocene

icrotus (Tyrrhenicola) henseli in each fossiliferous level of the SC stratigraphical section.

Fig. 16. Stratigraphic sketch of the SD section. 1) Clayey sediments; 2) sandy loam sediments; 3) sandy-clay-loam sediments; 4) angular and sub-angular clasts of carbonate andmetamorphic rocks; 5) flowstone; 6) fossil remains.

R.T. Melis et al. / Quaternary International 288 (2013) 81e9694

(late Bronze Age, Nuraghe Is Paras de Isili, Delussu 2000; SantuAntine de Torralba, Manconi, unpublished data). This taxon showsgreat variability, and a number of different morphotypes have beendescribed by authors even within a single population (cf.

Fig. 17. A selection of the main microstructural features observed in thin section from SD s

Mezzabotta et al., 1995, 1996; Minieri et al., 1995). Moreover,chronological information on each sample is scanty, and most ofthe data suggests a sort of “mosaic” evolutionary pattern, althougha number of differences showed up in the different populations,

ection. a) Black mottles in D4 layer (XPL); b) rounded pedorelict in the D8 layer (XPL).

R.T. Melis et al. / Quaternary International 288 (2013) 81e96 95

suggesting they could be placed in a relative evolutionary order(e.g. Mezzabotta et al., 1995; Minieri et al., 1995; Marcolini et al.,2006a; Palombo, 2006, 2009a,b).

Most of the populations recovered in Sardinia seem to bealready evolved, none of them presenting archaic features, at leastin regard to the first lower molar (M1), thus far considered themost diagnostic skeletal element in inferring the evolutionarydegree in arvicolids. The presence of a small M. (Tyrrhenicola),showing a first lower molar with more archaic features than M.(Tyrrhenicola) henseli, has been claimed as present at Capo Figari(Olbia, north-western Sardinia) (Brandy, 1978), and reported byGinesu and Cordy (1997) and Sondaar (2000) in the fissure X g3 atthe Monte Tuttavista karst network (Orosei, Eastern Sardinia). Onthe basis of the latter material Turmes (2003) described, in anunpublished PhD thesis, a new species (Tyrrhenicola orosei), whichwas then formally described as M. (Tyrrhenicola) sondaari byMarcolini et al. (2006a) on the basis of new, rich material comingfrom the same fissure.

The colonization time and phylogenetic relationships of theSardinian endemic vole have been debated. Van der Meulen (1973)for instance, suggested a derivation from Allophaiomys pliocaenicusand an Early Pleistocene migration, while Mezzabotta et al. (1995)and Minieri et al. (1995) suggested that the ancestor was a quiteadvanced representative of the Allophaiomys ruffoi e Allophaiomysburgondiae lineage, who entered Sardinia during the Early Pleis-tocene to Middle Pleistocene transition. The recent identification ofthe primitive endemic vole M. (Tyrrhenicola) sondaari, which likelyderived from an advanced A. ruffoi, could suggest that its arrivalmay be earlier than 1.3e1.2 Ma.

Whether M. henseli replaced M. sondaari before or during theMiddle Pleistocene is still an unanswered question, but the datafrom SC succession of the GFC indicated thatM. henseliwas alreadypresent in the Early Middle Plestocene, before 500 ka. On the otherhand, morphometrical and morphological data of M1 from SC andSD succession confirm the great intra-population variability of thistaxon, and stress the fluctuation of the average size throughouttime and the lack of correspondence between increase in size andappearances of advanced features, such as a more elongated ACC.This fact suggests that trying to arrange M. henseli populations inan evolutionary order, it is preferable to consider the relativefrequency of morphotypes than biometrical parameters.

Acknowledgments

We are indebted with G. Cuenca-Bescos and an anonymousreviewer for their helpful remarks. We thank Mauro Villani,Guglielmo Caddeo and Daniela Sini for helping us in the sectionsampling. The English version of the manuscript has been revisedby Dr. David Sommers. This work was supported by the project ofUniversity of Cagliari (ex 60% R.T. Melis) and the project MIUR PRIN2008RTCZJH 002 (Unit 2, Resp., M.R. Palombo).

References

Andreucci, S., Clemmensen, L.B., Murray, A.S., Pascucci, V., 2010. Middle to latePleistocene coastal deposits of Alghero, northwest Sardinia (Italy): chronologyand evolution. Quaternary International 222, 3e16.

Bailey, G.N., Woodward, J.C., 1997. The Klithi deposits: sedimentology, stratigraphyand chronology. In: Bailey, G.N. (Ed.), Klithi: Palaeolithic Settlement andQuaternary Landscapes in Northwest Greece. Excavation and Intra-site Analysisat Klithi, Cambridge, vol. 1, pp. 61e94.

Baker, A., Smith, C.L., Jex, C., Fairchild, I.J., Genty, D., Fuller, L., 2008. Annuallylaminated speleothems: a review. International Journal of Speleology 37,193e206.

Barta, K., Tarnai, T., 1998. Geomorphological investigation of cave explorationopportunities in the Mecsek Mountains, S-Hungary. Geografia Fisica e DinamicaQuaternaria 21, 5e8.

Bechstadt, T., Boni, M., 1996. Sedimentological, stratigraphical and ore deposits fieldguide of the autochtonous Cambro-Ordovician of Southwestern Sardinia.Memorie descrittive della Carta Geologica d’Italia 48, 1e390.

Benzi, V., Abbazzi, L., Bartolomei, P., 2007. Radiocarbon and U-series dating of theendemic deer Praemegaceros cazioti (Depéret) from “Grotta Juntu”, Sardinia.Journal of Archaeological Science 34, 790e794.

Bini, A., Cremaschi, M., Forti, P., Perna, G., 1988. Paleokarst fills in Iglesiente (Sar-dinia, Italy); sedimentary processes and age. Annales de la Société géologiquede Belgique 111, 149e161.

Boero, V., Franchini, A.M., 1992. Crystallization of hematite and goethite in thepresence of soil minerals. European Journal of Mineralogy 4, 539e546.

Boldrini, R., 2008. Microtus (Tyrrhenicola) henseli (Major, 1905) di Grotta dei Fiori(Sardegna sudoccidentale): caratteri evolutivi e interpretazione biocronologica.Geologica Romana 41, 55e64.

Boldrini, R., 2010. Ecomorfologia e geochimica isotopica, quale supporto per analisipaleo ambientali? Il caso dei micro mammiferi quaternari della Sardegna. PhDThesis.

Boldrini, R., Palombo, M.R., Iacumin, P., Melis, R.T., 2010. The Middle Pleistocenefossiliferous sequence of Grotta dei Fiori (Sardinia, Italy): multidisciplinaryanalysis. Bollettino della Società Paleontologica Italiana 49 (2), 123e134.

Boni, M., Iannace, A., Köppel, V., Hansmann, W., Früh-Green, G., 1992. Late to post-Hercynian hydrothermal activity and mineralization in SW Sardinia. EconomicGeology 87, 2113e2137.

Boni, M., Iannace, A., Villa, I.M., Fedele, L., Bodnar, R., 2001. Multiple fluid-flowevents and mineralizations in SW Sardinia: an European perspective. In:Cidu, R. (Ed.), Water-rock Interaction 2001. Balkema, Villasimius, Lisse, TheNetherlands, pp. 673e676.

Brain, C.K., 1995. The Influence of climatic changes on the completeness of the earlyhominin record in Southern African Caves, with particular reference toSwartkrans. In: Vrba, E.S. (Ed.), Palaeoclimate and Evolution, with Emphasis onHuman Origins. Yale, New Haven, pp. 385e424.

Brandy, L.D., 1978. Données nouvelles sur l’évolution du rongeur endémique fossilecorso-sarde Rhagamys F. Major (1905) (Mammalia, Rodentia). Bulletin de laSociété Géologique de France 7 (20), 831e835.

Bullock, P., Fedoroff, N., Jongerius, A., Stoops, G., Tursina, T., 1985. Handbook for SoilThin Section Description. Waine Research Publications, Wolverhampton, UK.

Carmignani, L., Oggiano, G., Barca, S., Conti, P., Salvadori, I., Eltrudis, A., Funedda, A.,Pasci, S., 2001. Geologia della Sardegna. Note illustrative della Carta Geologicadella Sardegna a scala 1:200.000. In: Memorie Descrittive della Carta Geologicad’Italia, vol. 60. Servizio Geologico, Ist. Poligr. e Zecca dello Stato, Roma. 283.

Catto, N., 2010. Comment on Commentary Geomorphology, stratigraphy and faciesanalysis of some Late Pleistocene and Holocene key deposits along the coast ofSardinia (Italy), and Geochronology for some key sites along the coast of Sardinia(Italy): a note from the Editor-in-Chief. Quaternary International 222, 46e47.

Civita, M., Cocozza, T., Forti, P., Perna, G., Turi, B., 1983. Idrogeologia del bacinominerario dell’Iglesiente (Sardegna sud-occidentale). Memorie dell’IstitutoItaliano di Speleologia 2, 1e139.

Coltorti, M., Melis, E., Patta, D., 2010. Geomorphology, stratigraphy and faciesanalysis of some Late Pleistocene and Holocene key deposits along the coast ofSardinia. Quaternary International 222, 19e35.

Cortecci, G., Fontes, J.C., Maiorani, A., Perna, G., Pintus, E., Turi, B., 1989. Oxygen,sulfur, and strontium isotope and fluid inclusion studies of barite deposits fromthe Iglesiente-Sulcis mining district, southwestern Sardinia, Italy. MineraliumDeposita 24, 34e42.

Cuenca-Bescós, G., Rofes, J., García-Pimienta, J.C., 2005. Early Europeans and envi-ronmental change across the Early-Middle Pleistocene transition: smallmammalian evidence from Trinchera Dolina cave, Atapuerca, Spain. In:Head, M.J., Gibbard, P.L. (Eds.), Early-middle Pleistocene Transitions: the Land-ocean Evidence. Geological Society of London, Special Publication, vol. 247,pp. 277e286.

Cuenca-Bescós, G., Straus, L.G., González Morales, M.R., García Pimienta, J.C., 2009.The reconstruction of past environments through small mammals: from theMousterian to the Bronze Age in el Mirón cave (Cantabria, Spain). Journal ofArchaeological Science 36 (4), 947e955.

De Vivo, B., Maiorani, A., Perna, G., Turi, B., 1987. Fluid inclusion and stable isotopestudies of calcite, quartz and barite from karst caves in the Masua Mine,southwestern Sardinia, Italy. Chemie der Erde 46, 259e273.

De Waele, J., Forti, P., 2006. A new hypogean karst form: the oxidation vent.Zeitschrift fur Geomorphologie N.F. Supplement-band 147, 107e127.

De Waele, J., Forti, P., Perna, G., 2001. Hyperkarst phenomena in the Iglesientemining district (SW-Sardinia). In: Cidu, A. (Ed.), Water-rock Interaction 2001.A.A. Balkema, Lisse, pp. 619e622.

Ek, C., Quinif, Y., 1988. Les sédiments détritiques des grottes: aperçu synthétique.Annales de la Société Géologique de Belgique 111, 1e7.

Elles, M.P., Rabenhortst, M.C., 1994. Micromorphological interpretation of redoxprocesses in soils derived from Triassic red bed parent material. In: Ringrose-Voase, A.J., Humphreys, G.S. (Eds.), Soil Micromorphology: Studies in Manage-ment and Genesis, International Society of Soil Science. Elvesier, Amterdam,pp. 171e178.

Federoff, N., Courty, M.A., Zhengtang, G., 2010. Paleosoils and relict soils. In:Stoops, G., Marcellino, V., Mees, F. (Eds.), Interpretation of MicromorphologicalFeatures of Soils and Regoliths. Elsevier, Amsterdam, pp. 623e662.

Foos, A.M., Sasowsky, I.D., LaRock, E.J., Kambesis, P.N., 2000. Detrital origin ofa sedimentary fill, Lechuguilla Cave, Guadalupe Mountains, New Mexico. Claysand Clay Minerals 48 (6), 693e698.

R.T. Melis et al. / Quaternary International 288 (2013) 81e9696

Forti, P., Perna, G., 1982. Le cavità naturali dell’Iglesiente. Memorie dell’IstitutoItaliano di Speleologia II (1), 229.

Frisia, S., 2003. Le tessiture negli speleo temi. Studi Trent. Scienze Naturelle, ActaGeologica 80, 85e94.

Frisia, S., Borsato, A., Fairchild, I.J., McDermott, F., 2002. Aragonite-calcite relation-ships in speleothems (Grotte de Clamouse, France): environment, fabrics andcarbonate geochemistry. Journal of Sedimentary Research 72, 687e699.

Ginesu, S., Cordy, J.M., 1997. Il Monte Tuttavista (Orosei-Galtelli). Edizione Poddighe,Sassari, 1e47.

Goldberg, P., 2000. Micromorphology and site formation at Die Kelders Cave 1,South Africa. Journal of Human Evolution 38, 43e90.

Goldberg, P., Sherwood, S., 2006. Deciphering human prehistory through the geo-archeological study of cave sediments. Evolutionary Anthropology 15, 20e36.

Gruppo Ricerche Speleologiche “E.A. Martel” Carbonia, 1977. Elenco catastale dellegrotte del comune di Carbonia, p. 19.

Heinrich, W.D., 1978. Zur biometrischen Erfassung eines Evolutionstrends beiArvicola (Rodentia, Mammalia) aus dem Pleistozan Thutinges. Saugertiekund-liche Information 2, 3e21.

Klimchouk, A.B., 2007. Hypogene Speleogenesis: Hydrogeological and Morphoge-netic Perspective. National Cave and Karst Research Institute, Carlsbad, USA.pp. 106.

Klimchouk, A.B., 2009. Morphogenesis of hypogenic caves. Geomorphology 106,100e117.

Laviano, R., 1987. Analisi mineralogica quantitativa di argille mediante dif-frattometria di raggi X. Collana di Studi ambientali. Procedure di analisi diminerali argillosi, Ente Nazionale Energie Alternative, 215e234.

Leone, F., Hamman, W., Laske, R., Serpagli, E., Villas, E., 1991. Lithostratigraphic unitsand biostratigraphy of the post-sardic Ordovician sequence in south-westSardinia. Bolletino Societa Paleontogica Italia 30, 201e235.

Leone, F., Ferretti, A., Hammann, W., Loi, A., Pillola, G.L., Serpagli, E., 2002. A generalview of the post-sardic Ordovician sequence from SW Sardinia. Rendu SocietaPaleontogica Italia 1, 51e68.

Ludwig, K.R., Vollmer, R., Turi, B., Simmons, K.R., Perna, G., 1989. Isotopic constraintson the genesis of base-metal ores in southern and central Sardinia. EuropeanJournal of Mineralogy 1, 657e666.

Marcolini, F., Arca, M., Kotsakis, T., Tuveri, C., 2005. The endemic vole Tyrrhenicola(Arvicolidae, Rodentia) from Monte Tuttavista (Sardinia, Italy): new perspec-tives for phylogeny and biochronology. Bolletì de la Societat d’Historia Naturalde les Balears 12, 185e192.

Marcolini, F., Tuveri, C., Arca, M., Kotsakis, T., 2006a. Microtus(Tyrrhenicola) sondaarin. sp. (Arvicolidae, Rodentia) from Monte Tuttavista (Sardinia, Italy). HellenicJournal of Geosciences 41, 73e82.

Marcolini, F., Tuveri, C., Arca, M., Canu, C., Demontis, G., Kotsakis, T., 2006b. Analisidello smalto dentario di Microtus(Tyrrhenicola) (Arvicolidae, Rodentia, Mam-malia) dal Monte Tuttavista (Sardegna, Italia): nuove prospettive bio-cronologiche. Sardinia, Corsica et Baleares antiquae 2, 11e29.

Melis, R.T., Palombo, M.R., Bedetti, C., Fenza, P., Pavia, M., Mureddu, M., Villani, M.,2002. Evoluzione speleogenetica e paleoambientale della grotta dei Fiori SA/CA218 (Carbonia-Sardegna sud-occidentale). Anthéo Bollettino Gruppo SpeleoArcheo Giovanni Spano di Cagliari 6, 71e88.

Mezzabotta, C., Masini, F., 1996. Influence of the climatic environmental variationson the vole Microtus (Tyrrhenicola) henseli at the late Pleistocene-Holocenetransition. Il Quaternario 9 (2), 721e724.

Mezzabotta, C., Masini, F., Torre, D., 1995. Microtus (Tyrrhenicola) henseli, endemicfossil vole from Pleistocene and Holocene localities of Sardinia and Corsica:evolutionary patterns and biochronological meaning. Bollettino Società Pale-ontologica Italiana 34, 81e104.

Minieri, M.R., Palombo, M.R., Scarano, M., 1995. Microtus(Tyrrhenicola) henseli(Major, 1882) del Pleistocene superiore di Cava Alabastro (Is Oreris; Iglesias;Sardegna sudoccidentale). Geologica Romana 31, 51e60.

Moriarty, K.C., McCulloch, M.T., Wells, R.T., McDowell, M.C., 2000. Mid-Pleistocenecave fills, megafaunal remains and climate change at Naracoorte, SouthAustralia: towards a predictive model using UeTh dating of speleothems.Palaeogeography, Palaeoclimatology, Palaeoecology 159, 113e143.

Munsell, Colour, 1998. Munsell Soil Colour Charts. Koll Morgen Instruments,Maryland.

Pagliara, A., DeWaele, J., Forti, P., Galli, E., Rossi, A., 2010. Speleothems hypogenicSanta Barbara Cave system (South-WestSardinia, Italy). Acta Carsologica, Post-ojna 39 (3), 551e564.

Palombo, M.R., 2006. Biochronology of the Plio-Pleistocene terrestrial mammals ofSardinia: the state of the art. Hellenic Journal of Geosciences 41, 47e66.

Palombo, M.R., 2009a. Biochronology and turnover patterns of endemic Plio-Pleistocene mammals from Sardinia (Western Mediterranean). IntegrativeZoology 4, 367e386.

Palombo, M.R., 2009b. Body size structure of the Pleistocene Mammaliancommunities on Islands: what support for the island rule? Integrative Zoology4, 341e356.

Palombo, M.R., Ferretti, M.P., Pillola, G.L., Chiappini, L., 2012. A reappraisal of thedwarfed mammoth Mammuthus lamarmorai (Major, 1883) from Gonnesa(south-western Sardinia, Italy). Quaternary International 255, 158e170.

Pavia, M., Bedetti, C., 2003. The late Pleistocene fossil avian remains from Grotta deiFiori, Carbonia (SW Sardinia, Italy). Bollettino della Società PaleontologicaItaliana 42, 163e169.

Pena, J.L., 1998. Los estudios sobre procesos de clima frío en Espaena: balance yperspectivas. In: Gómez-Ortiz, A., Salvador, F., Schulte, L., García-Navarro, A.(Eds.), Procesos biofísicosactuales en medios fríos. Universidad deBarcelona,Barcelona, pp. 43e54.

Pickering, R., Hancox, P.J., Lee-Thorp, J.A., Grün, R., Mortimer, G.E., McCulloch, M.,Berger, L.R., 2007. Stratigraphy, UeTh chronology, and paleoenvironments atGladysvale Cave: insights into the climatic control of South African hominin-bearing cave deposits. Journal of Human Evolution 35, 602e619.

Pillola, G.L., 1989. Trilobites du Cambrien inférieur du SW de la Sardaigne, Italie.Paleontographica Italica 78, 1e174.

Quinif, Y., 2006. Complex stratigraphic sequences in Belgian caves correlation withclimatic changes during the Middle, the Upper Pleistocene and the Holocene.Geologica Belgica 9 (3e4), 231e244.

Quinif, Y., Maire, R., 1998. Pleistocene deposits in Pierre Saint-Martin Cave. FrenchPyrenees Quaternary Research 49, 37e50.

Railsback, L.B., Brook, G.A., Chen, J., Kalin, R., Fleisher, C.J.,1994. Environmental controlson the petrology of a Late Holocene speleothem fromBotswanawith annual layersof aragonite and calcite. Journal of Sedimentary Research 64, 147e155.

Roucoux, K.H., Tzedakis, P.C., de Abreu, L., Shackleton, N.J., 2006. Climate andvegetation changes 180,000 to 345,000 years ago recorded in a deep-sea coreoff Portugal. Earth and Planetary Science Letters 249, 307e325.

Roucoux, K.H., Tzedakis, P.C., Frogley, M.R., Lawson, I.T., Preece, R.C., 2008. Vege-tation history of the marine isotope stage 7 interglacial complex at Ioannina,NW Greece. Quaternary Science Reviews 27, 1378e1395.

Sala, B., 1980. Interpretazione Crono-Bio-Stratigrafica dei depositi Pleistocenici dellaGrotta del Broion (Vicenza). Geografia Fisica Dinamica, Quaternaria 3, 66e71.

Schwertmann, U., 1985. The effect of pedogenic environments on iron oxidesminerals. Advances in Soil Science 1, 172e200.

Schwertmann, U., Taylor, R.M., 1989. Iron oxides. In: Dixon, J.B., Weed, J.B. (Eds.),Minerals in Soil Environments, second ed. Soil Science Society of America,Madison, Wisconsin, pp. 379e438.

Sondaar, P.Y., 2000. Early human exploration and exploitation on islands. Tropics 10,203e230.

Spötl, C., Mangini, A., Richards, D.A., 2006. Chronology and paleoenvironment ofmarine isotope stage 3 from two high-elevation speleothems, Austrian Alps.Quaternary Science Reviews 25, 1127e1136.

Thiel, C., Coltorti, M., Tsukamoto, S., Frechen, M., 2010. Geochronology for somekey sites along the coast of Sardinia (Italy). Quaternary International 222,36e47.

Tricart, J., Cailleux, A., 1967. Le modelé des régions périglaciaires. SEDES, Paris, 512 pp.Turmes, M., 2003. Les micrommamifères (Rongeurs, Insectivores, Lagomprphes)

quaternaires du karst du Monte Tuttavista (Sardaigne, Italie): étude morpho-logique et biométrique-microévolution en milieu insulaire. Ph.D. Thesis, Uni-versité de Liège.

Tzedakis, P.C., 2005. Towards an understanding of the response of southern Euro-pean vegetation to orbital and suborbital climate variability. Quaternary ScienceReviews 25, 1585e1599.

Van Der Made, J., 1999. Biogeography and stratigraphy of the Mio-Pleistocenemammals of Sardinia and the description of some fossils. In: Reumer, J.W.F.,de Vos, J. (Eds.), Elephants Have a Snorkel!, Deinse A, pp. 337e360.

Van der Made, J., Palombo, M.R., 2006. Large deer from the Pleistocene of Sardinia.Hellenic Journal of Geosciences 41, 163e176.

Van der Meulen, A.J., 1973. Middle Pleistocene smaller mammals from the MontePeglia (Orvieto, Italy) with special reference to the phylogeny of Microtus(Arvicolidae, Rodentia). Quaternaria 17, 1e144.

Villani, M., 2001. La valle di Cannas e la Grotta dei Fiori (Carbonia Ca). Speleologia44, 89e90.

Violante, P., 2000. Metodi di analisi chimica del suolo. Franco Angeli, Roma, Italy,536 pp.

Woodward, J.C., 1997a. Late Pleistocene rockshelter sedimentation at Klithi. In:Bailey, G.N. (Ed.), Klithi: Palaeolithic Settlement and Quaternary Landscapes inNorthwest Greece. Klithi in Its Local and Regional Setting, vol. 2. MacDonaldInstitute for Archaeological Research, Cambridge, pp. 361e376.

Woodward, J.C., 1997b. Late Pleistocene rockshelter sedimentation at Megalakkos.In: Bailey, G.N. (Ed.), Klithi: Palaeolithic Settlement and Quaternary Landscapesin Northwest Greece. Klithi in Its Local and Regional Setting, vol. 2. MacDonaldInstitute for Archaeological Research, Cambridge, pp. 377e393.

Woodward, J.C., Bailey, G.N., 2000. Sediment sources and terminal Pleistocenegeomorphological processes recorded in rockshelter sequences in NorthwestGreece. In: Foster, I.D.L. (Ed.), Tracers in Geomorphology. Wiley Chichester,pp. 521e551.