1Despite high sensitivity of fluvial environments to both allogenic (eustasy, tectonics, climate) and autogenic (local subsidence mainly related to sediment compaction, substrate and human activity) factors, important economic activities and densely populated cities in the Mediterranean region are concentrated on alluvial and coastal plains. Under the threat of future global warming and increasing sea-level rise (Edwards, 2007; IPCC, 2007; Gehrels, 2010), the role exerted by floodplains as flood retention areas and local drinking water reservoirs becomes of crucial interest. This leads to the urgency of an efficient management plan and predictive evolution models aimed at the preservation of alluvial plains. However, efficient land planning and realistic evolution models must rely on the accurate comprehension of fluvial system dynamics and driving forces, including the development of human societies.
2Historical and palaeoenvironmental (pollen) data have documented intense human land use (agriculture, draining, dredging and land reclamation) during the past 2000-3000 years, which led to drastic changes in vegetation cover, soil erosion and sediment supply in the Mediterranean fluvial catchments (Hooke, 2006). Since the Middle Age, several alluvial plains were also affected by widespread interventions on river courses, including embankments, channelisation, dam constructions and gravel-sand extractions (Hooke, 2006). Furthermore, human settlements and activities were strongly influenced by fluvial landscapes and drinking water availability since prehistoric times. The combination of human land use and late Holocene climatic changes resulted in high-frequency changes in river course dynamics, sedimentation rates and sediment redistribution patterns, recorded within worldwide alluvial successions (Lewin et al., 2005; Hooke, 2006; James and Marcus, 2006; Macklin et al., 2006; De Moor et al., 2008; Lespez et al., 2008; Houben et al., 2009; Lespez et al., 2010). Although sedimentation in fluvial-dominated contexts is intermittent and discontinuous due to ongoing river incision and reworking (Lewin and Macklin, 2003), accurate and reliable estimates of past environmental changes can be extrapolated from alluvial successions by the application of a multi- or trans-disciplinary approach, involving stratigraphy, geomorphology and archaeology (Hudson et al., 2008). In this respect, geoarchaeology is considered a new powerful tool for the reliable reconstruction of human-land interactions during prehistoric and historic times and to improve the chronology of past fluvial changes (Brown, 1997, 2008; Uribelarrea and Benito, 2008; Arnaud-Fassetta et al., 2010). A well-documented long history of human settlements along Mediterranean rivers and coastlines guarantees a high amount of archaeological data to be combined with on-site geomorphological and stratigraphical records. Multi-source databases, derived from several disciplines, have been recently used to reconstruct the palaeogeographic and palaeohydrographic changes affecting the Mediterranean delta plains during the late Holocene (Bruneton et al., 2001; Arnaud-Fassetta et al., 2003; Benvenuti et al., 2006; Bini et al., 2009; Arnaud-Fassetta et al., 2010; Piovan et al., 2010). Since prehistoric times, the Arno coastal plain (northern Tuscany, Italy; fig. 1) has been one of the most populated areas of the Mediterranean basin. This is documented by several archaeological sites, some of them dating back to the late Neolithic (ca. 5000-6000 a cal. BP), which provide opportunities for multi-proxy investigations. Late Holocene human settlements in the Arno river plain were probably influenced by a dense and unstable hydrographic network composed of several channels (Della Rocca et al., 1987; Marchisio et al., 1999; Bruni and Cosci, 2003). These studies, which bring together historical and remote sensing data derived from aerial photographs and satellite images, revealed a dense network of palaeomeanders in the shallow subsurface of the coastal plain. However, timing of the onset of these palaeochannels with respect to the two main fluvial courses that currently flow in northern Tuscany (Arno River and Serchio River) remained hypothetical.
3This study aims to improve our knowledge about the mid-late Holocene landscape evolution of the Arno coastal plain, providing an accurate characterisation of fluvial patterns in proximity of the city of Pisa, also in relation to the historical context. Mid-late Holocene palaeogeographic and palaeohydrographic scenarios are obtained through detailed reconstruction of the depositional architecture of the subsurface alluvial succession, integrated with the available geomorphological and archaeological data.
4The Arno River is one of the main Italian rivers reaching the Tyrrhenian Sea, with a catchment area of about 8230 km2 and a river length of 241 km km(Dapporto et al., 2001; Pranzini, 2001). It flows from the northern Apennines through the Arno plain (Tuscany, Italy) and enters the sea ca. 10 km kmwest of Pisa, forming a cuspate delta system (Federici and Mazzanti, 1995). The Arno Delta mostly developed during the last 2500-3000 years, as documented by the occurrence of archaeological remains dated to the Iron Age within the innermost outcropping beach-ridge (Pranzini, 2001). The Arno coastal plain, which is ca. 450 km2-wide, is bounded by the Pisani Mountains to the NE and the Leghorn-Pisa hills to the south (fig. 1 and fig. 2A), while the northern boundary is not well defined, owing to complex superposition of Arno and Serchio palaeochannels (Bruni and Cosci, 2003; fig. 1). The landward portion of the Arno coastal plain is extremely flat, with the exception of two morphological relieves (known as “Coltano and Castagnolo islands”) cropping out 5-7 km kmsouth of Pisa (Carosi et al., 2011). Fine-grained overbank sedimentation and a dense network of secondary channels, mainly artificial ditches built on former river channels, occur. The present Arno plain is mainly composed of clays, silts and subordinate fine to medium sands, formed by the aggradation of crevasse splays and natural levees. Organic-rich clays are also recorded, in correspondence of reclaimed swamp areas, and palaeochannels plugged by silty-sand sediments locally crop out (Della Rocca et al., 1987; Carratori et al., 1994; Carosi et al., 2011). The seaward part of the coastal plain is marked by the presence of a set of ancient beach-ridges. These crop out up to 6 km kminland from the modern coastline and reach 3.5-4 m in elevation (fig. 1).
Fig. 1 – Geomorphological map
Fig. 1 – Carte Géomorphologique
Main morphological features and depositional environments of the Arno and Serchio River plains (slightly modified from B. Della Rocca et al., 1987). 1: outer coastal plain; 2: inner coastal plain; 3: alluvial plain; 4: palaeomeander; 5: palaeoshoreline.
Principales caractéristiques géomorphologiques et milieux de sédimentation des plaines de l’Arno et du Serchio (d’après B. Della Rocca et al., 1987, légèrement modifié). 1 : plaine côtière distale ; 2 : plaine côtière proximale ; 3 : plaine alluviale ; 4 : paléoméandre ; 5 : ancienne ligne de rivage.
Fig. 2 – Location map
Fig. 2 – Carte de localisation
A: Location map of the study area reporting the subsurface database available for the Arno coastal plain. The dotted lines trace the inferred Arno palaeovalley boundaries. Section trace of fig. 3 is also reported. B: Planview of the study area around the city of Pisa showing the database used for this study. Reference cores are reported as light gray dots. Traces of cross-sections discussed in text and reported in fig. 6A to C are shown as bold lines. 1: archaeological site; 2: inferred palaeovalley; 3: well log; 4: pre-Holocene deposits; 5: Holocene deposits; 6: outcropping beach-ridges.
A : Localisation de la région d’étude avec mention des données du sous-sol de la plaine de l’Arno. Les traits en pointillés indiquent les limites présumées de la paléovallée de l’Arno. La localisation de la coupe sur la fig. 3 est aussi indiquée. B : Vue en plan de la région d’étude autour de la ville de Pise avec indication de la banque des données utilisée dans cette étude. Les forages de référence sont marqués par des points gris-clair. La localisation des coupes sur la fig. 6A à C sont indiquées en gras. 1 : site archéologique ; 2 : paléovallée présumée ; 3 : forage ; 4 : dépôts pré-holocènes ; 5 : dépôts holocènes ; 6 : cordons littoraux affleurants.
5The Arno coastal plain is located in the southern, inshore portion of an extensional basin, known as Viareggio basin, which developed since the late Tortonian (ca. 8 Ma) due to the opening of the Tyrrhenian Sea and the counter-clockwise migration of the Apenninic foredeep-foreland system (Patacca et al., 1990; Pascucci et al., 2001; Pascucci, 2005). The late Quaternary basin-fill succession, buried beneath the Arno coastal plain, consists of a repeated alternation of shallow-marine, coastal and alluvial deposits formed, under subsiding conditions, in response to interglacial-glacial cycles (Fancelli et al., 1986; Aguzzi et al., 2005, 2007). In proximity of the present Arno river mouth, coastal marine sediments tentatively attributed to the Tyrrhenian transgression (MIS, Marine Isotopic Stage 5e) have been recently recorded at ca. 100-m depth. Upwards, a thick continental succession (ca. 40-m thick) that accumulated during the last glacial period (MIS 4-2) caps the lower transgressive-regressive sequence, and is overlain in turn by a 50-m thick succession of Holocene (MIS 1) deposits (Aguzzi et al., 2007). The upper transgressive-regressive cycle (T-R sequences in the sense of A.F. Embry, 1993, 1995) is composed of marine, estuarine and alluvial deposits and shows remarkable thickness variations. This feature, along with abrupt lateral facies changes from soft Holocene clays to stiff, pre-Holocene deposits, has been interpreted to indicate the presence of an incised-valley developed in response to the last glacial sea-level fall (Aguzzi et al., 2007; Amorosi et al., 2008). The incised-valley system, 5-7-km wide and 40-m thick, is perpendicular to the present coastline, and roughly follows the modern Arno River course. It shows a peculiar T-shaped plan view geometry at proximal locations (fig. 2A), possibly caused by the confluence of two palaeochannels in proximity of the city of Pisa during the last glacial period. Small-scale cyclic alternations of early transgressive (ca. 13000-8000 a cal. BP) estuarine and paludal deposits compose the incised-valley fill at proximal locations (fig. 3), while predominant estuarine clays are recorded at distal locations (Amorosi et al., 2008). On the top of the Arno valley fill, a laterally extensive, 3-15-m thick sedimentary unit (locally known as “pancone”), composed of a homogeneous succession of soft clays and silty clays containing a brackish fauna (lithofacies P1; Rossi et al., in press) is recorded from the Pisani Mountains to the innermost outcropping beach-ridge. “Pancone” spatial distribution reflects the development of a wide lagoon at the turnaround from transgressive to highstand conditions (ca. 8000 a cal. BP; Amorosi et al., 2008). Lagoonal deposits, formed up to about 6000 a cal. BP (Benvenuti et al., 2006), show lateral transition to paludal, backswamp fine-grained sediments. This succession is capped by 10-20-m of highstand fluvio-deltaic deposits, composed predominantly of silt and clay encasing sandy bodies (fluvial channels; Amorosi et al., 2008; fig. 3). Detailed sedimentological analyses performed by M. Benvenuti et al. (2006) at the archaeological site of Pisa S. Rossore, located ca. 1 km kmnorth of the present Arno River course (fig. 1 and fig. 2A), showed a vertical stacking pattern of sand bodies erosively overlying the “pancone” and recording the recent history of Pisa riverine harbour during the Etruscan-Roman period (fig. 4). Above a succession of pre-Roman fine sands and mud, on which an Etruscan palisade was built, four main units (units 1-4; fig. 4) encasing well-preserved Roman ships and associated cargo materials were recorded. These units were interpreted as due to catastrophic, high-frequency (centennial-scale) overbank floods likely related to high-magnitude hydro-climatic events (Benvenuti et al., 2006). In this respect, pollen analyses performed on thin muddy layers interbedded with the sandy units confirmed the occurrence of climatic oscillations and fluctuations in soil drainage during the warm Roman period (Mariotti Lippi et al., 2007). The fluvio-deltaic succession grades seaward into beach-barrier sands and shallow-marine clays (Amorosi et al., 2008).
Fig. 3 – Facies architecture of the Arno palaeovalley fill
Fig. 3 – Architecture des faciès de la paléovallée de l’Arno
Facies architecture of the Arno palaeovalley fill at proximal locations, showing the small-scale cyclic alternations of Lateglacial-early Holocene estuarine and paludal deposits. Above the palaeovalley fill an homogeneous lagoonal succession (“pancone”) is overlain by fine-grained paludal and floodplain sediments encasing sandy channel bodies. Reference cores for which detailed facies analysis is available are box-bordered. TS: Transgressive Surface. 14C dates are reported as calibrated ages BP (partially modified from A. Amorosi et al., 2008). Section trace is reported in fig. 2A. 1: clay/silt; 2: sand; 3: fossil; 4: organic-rich layer; 5: indurated horizon; 6: palaeosol; 7: estuarine deposits; 8: lagoonal deposits; 9: paludal deposits; 10: floodplain deposits; 11: fluvial channel sands; 12: Last Glacial Maximum substrate.
Architecture des faciès de la paléovallée de l’Arno dans la zone proximale, montrant une succession stratigraphique cyclique à grande échelle (alternance de dépôts lagunaires et palustres). Recouvrant le remblaiement de la paléovallée, la succession lagunaire (“pancone”) est surmontée par des dépôts fins de marais ou de plaine d’inondation incluant des unités sableuses de paléochenal. Les forages de référence sont encadrés. TS : surface de transgression. Les dates obtenues par le radiocarbone sont des valeurs calibrées (d’après A. Amorosi et al., 2008, légèrement modifié). Le tracé de la coupe est indiqué in fig. 2A. 1 : argiles/limons ; 2 : sables : 3 : fossile ; 4 : unité riche en matière organique ; 5 : niveau induré ; 6 : paléosol ; 7 : dépôts d’estuaire ; 8 : dépôts lagunaires ; 9 : dépôts palustres ; 10 : dépôts de plaine d’inondation ; 11 : unité sableuse de chenal ; 12 : substrat du Dernier Maximum Glaciaire.
Fig. 4 – Stratigraphic model of the archaeological site of Pisa S. Rossore
Fig. 4 – Stratigraphie du site archéologique de Pise S. Rossore
A: Cross section of the site from exposures and core data (from M. Benvenuti et al., 2006). B: Stratigraphic correlation among representative cores in the site area (from M. Benvenuti et al., 2006).
A : Coupe stratigraphique du site reconstituée d’après les données des forages (d’après M. Benvenuti et al., 2006). B : Corrélation stratigraphique entre plusieurs forages (d’après M. Benvenuti et al., 2006).
6The study area falls within the Arno coastal plain (fig. 1 and fig. 2A), where a large georeferenced subsurface database consisting of about 2600 well logs was provided by Municipality of Pisa and Geological Survey of Pisa Province in collaboration with the University of Pisa. The database includes a variety of stratigraphic logs ranging in depth between about 15 m and 200 m, derived from continuous cores (ca. 350 cores), water wells and cone penetration tests. As this dataset exhibits a high variability in quality and spatial distribution, our attention was focused around the city of Pisa (fig. 2B), characterised by an elevated concentration of subsurface data (cores, penetration tests). Moreover, important information regarding past landscape features of the Pisa area rely upon several archaeological findings, dating from the Iron Age to late Mediaeval times (Benvenuti et al., 2006), and abundant documentary sources (ancient maps and historical documents) dating back to the Roman period (Bruni and Cosci, 2003, Martini et al., 2010). Stratigraphic analysis of the floodplain surrounding the city of Pisa was based on about 400 subsurface data (ca. 200 cores and 200 penetration tests) with a maximum spacing of about 1.5 km (fig.f 2B). Despite the inhomogeneous data distribution and low resolution of several stratigraphic logs, the differentiation between sandy and fine-grained deposits could be established robustly. The main lithofacies composing the mid-late Holocene fluvio-deltaic succession were identified by means of sedimentological analyses. Fifteen continuous cores were described in detail and used as reference data in terms of texture, colour, sedimentary structures and accessory materials, such as plant remains, wood fragments, calcareous nodules and macrofossils (see fig. 2B for location). Following A. Amorosi and N. Marchi (1999), reference cores were also used to calibrate cone penetration tests (CPT), improving lithofacies identification from Qt-cone resistance profiles and FR-friction ratio values (Robertson, 1986). To reconstruct the alluvial architecture and the main phases of hydrographic evolution of the Arno coastal plain, stratigraphic correlations were performed along seventeen, evenly distributed cross-sections oriented perpendicular to the present Arno River flow direction (fig. 2B). Where possible, the reliability of past landscape and fluvial dynamics reconstructions was tested by comparing our results with geomorphological, archaeological and historical data. The chronological framework of the studied succession was supported by archaeological and radiocarbon datings of artifacts (palisade, pier fragment and ship) found at the Pisa S. Rossore site (see fig. 1 and fig. 2A for location, and fig. 4; Benvenuti et al., 2006) and four radiocarbon ages (tab. 1) performed on mollusk shells and woods, three of which published in M. Benvenuti et al. (2006). All ages (in years) are calibrated (cal.) BP.
Tab. 1 – Published radiocarbon dates
Tab. 1 – Dates radiocarbone publiées
Sample
|
Sample depth (in m b.s.l.)
|
Dating material
|
Conventional
C14 ages
|
Calibrated
C14 ages
|
Lithofacies associations
|
References
|
S1-17.10
|
13.5
|
Mollusc shells
|
7370±60
|
6000-
5750 BC
|
Lagoonal deposits (pancone)
|
Amorosi et al. (2009)
|
Beta-171194
|
unknown
|
Cerastoderma shell
|
5800±80
|
4430-
4045 BC
|
Lagoonal deposits (pancone)
|
Benvenuti et al. (2006)
|
Beta-171195
|
unknown
|
Clastic wood
|
2000±40
|
80 BC-
AD 80
|
Channel deposits
|
Benvenuti et al. (2006)
|
Beta-171196
|
unknown
|
Clastic wood
|
1980±50
|
80 BC-
AD 120
|
Channel deposits
|
Benvenuti et al. (2006)
|
7In the area surrounding the city of Pisa (fig. 2B), the fluvio-deltaic succession resting on “pancone” is composed of predominantly fine-grained, cohesive deposits locally interbedded with sands; the overall thickness ranges between 10 m and 20 m. Three main lithofacies associations are identified according to texture, organic matter, fossils content and diagnostic penetration test features (Qt profile and FR values), which are strictly connected to the grain size and the consolidation of deposits (Robertson, 1986). Their description and interpretation in terms of depositional environments is reported below.
8Across the study area, lithofacies Sw is recorded with a highly variable thickness, ranging between 1 m and 7 m. This lithofacies association is composed of a homogeneous succession of dark soft clays and silty clays containing scattered shells and fragments of small terrestrial and freshwater gastropods, mainly belonging to the genus Bithynia. Abundant plant remains and large wood fragments are also recorded. These vegetal remains and the widespread occurrence of decomposed organic materials, occasionally forming peaty layers, document elevated organic matter content and groundwater level. Cm- to dm-thick fine sandy layers rarely occur (fig. 5A). Low Qt values (lower than 10 kg/cm2) and relatively regular Qt profiles characterise this lithofacies. FR values commonly range between 3.5% and 6%, which are typical of soft organic-rich clays (Robertson, 1986). Sharp peaks of Qt occur in correspondence of peats or thin sandy layers and wood trunks, interrupting the homogenous clayey succession. Lithology, texture, high organic matter content and lack of marine and brackish fossils indicate a fully terrestrial, wet and low-energy depositional setting rich in woody vegetation, such as a freshwater swampland formed in a delta plain and occasionally reached by river floods (sandy layers). Cone penetration parametres comparable to those observed within this lithofacies association have been recorded in the Holocene succession of the Po Plain and regarded as diagnostic of a paludal environment (Amorosi and Marchi, 1999).
9The swamp lithofacies association is commonly overlain by lithofacies Of, which is mainly composed of dry, massive, fine-grained sediments (clay, silty and sandy silts) with low organic matter content. Carbonate nodules and yellow-brown mottles due to iron and manganese oxides provide evidences of subaerial exposure (fig. 5 B and C). Few plant remains occur locally, while no shells or mollusc bioclasts are found. Occasionally, less than 1-m thick coarse-grained layers composed of yellow-brown, very fine-fine sands are recorded. Relatively low to medium Qt values (ca. 10-30 kg/cm2) and irregular Qt profiles, due to the occurrence of calcareous nodules and thin sandy layers, characterise this lithofacies. FR commonly ranges between 5% and 9%, values typical of floodplain silts and clays (Robertson, 1986). On the basis of sedimentological features, lithofacies Of is interpreted as the product of fluvial overbank deposition outside the river channel, within a dry, alluvial depositional setting subject to subaerial exposure. Thin sandy layers possibly were formed during small-scale floods derived from nearby active channels, recording levee or crevasse deposition. As the low degree of resolution of several stratigraphic logs precludes identification of the individual depositional facies that compose the alluvial overbank succession, lithofacies Of includes floodplain, levee and crevasse sediments.
10This lithofacies association, which shows a variable thickness of 1-8 m, consists of gray or brown, fine to coarse sands occasionally rich in organic remains, as wood or plant fragments, and reworked marine mollusc bioclasts. Sparse pebbles and mud-clasts are also locally encountered. High-quality data commonly evidence a distinctive internal fining-upward trend and a sharp erosional lower boundary, separating lithofacies Ch from the underlying lagoonal (fig. 5D), swamp or overbank deposits. No sedimentary structure was found in the cores. High Qt values (commonly >30 kg/cm2) and low FR values, ranging between 0.5-4%, invariably characterise lithofacies Ch. According to diagnostic sedimentological and cone penetration features, this lithofacies association is interpreted mostly as channel bodies cutting the highstand deltaic-alluvial plain succession at various stratigraphic levels. Channels bodies, recorded at the lowest stratigraphic levels, show exclusive stratigraphic relationships with swamp deposits (lithofacies Sw) and less frequently “pancone” (fig. 6 A to C).
Fig. 5 – Lithofacies associations of the fluvio-deltaic succession
Fig. 5 – Associations de lithofaciès dans la succession fluvio-deltaïque
Representative photographs of continuous cores used as reference data for depicting the main lithofacies associations of the Arno fluvio-deltaic succession. The white arrows point towards the top of the core. A: Organic-rich swamp deposits (lithofacies Sw) overlying a homogeneous succession of lagoonal clays (“pancone”) - core Kindu. B: Alluvial overbank clays and silts (lithofacies Of) - core v-S4. C: Alluvial overbank deposits (lithofacies Of) overlying organic-rich swamp clays (lithofacies Sw) - core v-S5. D: Channelised sand containing bioclasts (lithofacies Ch) and its erosional lower boundary onto lagoonal clays (“pancone”) - core BP.
Photographies de carottes montrant les principaux lithofaciès dans la succession fluvio-deltaïque de l’Arno. Les flèches blanches indiquent le sommet de la carotte. A : Dépôt organique de marais (lithofaciès Sw) au-dessus d’une succession homogène d’argiles lagunaires (« pancone ») - forage Kindu. B : Argiles et limons d’inondation (lithofaciès Of) - forage v-S4. C : Argiles organiques de marais (lithofaciès Sw) surmontant des dépôts de débordement (lithofaciès Of) - forage v-S5. D : Sables de chenal incluant des bioclastes (lithofaciès Ch) et sa limite inférieure avec des argiles lagunaires (« pancone ») - forage BP.
Fig. 6 – Depositional architecture of the fluvio-deltaic succession
Fig. 6 – Architecture des dépôts de la succession fluvio-deltaïque
Cross-sections perpendicular to the present Arno River course, depicting lithofacies distribution patterns, stratigraphic architecture of the Arno fluvio-deltaic succession and vertical stacking patterns of fluvio-deltaic units at proximal locations (A), around the city of Pisa (B) and at distal locations (C). Channel bodies cut lagoonal, swamp or alluvial overbank deposits. Units refer to fluvio-deltaic units (see tab. 2 and explanation in text). Reference cores are box-bordered and section traces are reported in fig. 2B. 1: clay and silt; 2: soft clay; 3: sand; 4: peat; 5: organic matter; 6: fossil; 7: evidence of subaerial exposure; 8: lagoonal deposits; 9: swamp deposits; 10: alluvial overbank deposits; 11: fluvial-channel deposits; 12: Last Glacial Maximum substrate; 13: anthropogenic deposits.
Schéma de distribution des faciès et architecture stratigraphique de la succession fluvio-deltaïque de l’Arno sur une coupe transversale au fleuve située dans la zone proximale (A), autour de Pise (B) et et dans la zone distale (C). Les unités de chenal montrent un contact érosif avec les dépôts lagunaires, palustres ou fluviatiles sous-jacents. Le terme « unité » indique des unités fluvio-deltaïques (voir tab. 2 et explication dans le texte). Les forages de référence sont encadrés et les tracés des coupes sont indiqués in fig 2B. 1 : argiles et limons ; 2 : argiles molles ; 3 : sables : 4 : tourbe ; 5 : matière organique ; 6 : fossile ; 7 : traces d'exposition à des processus d’érosion aériens ; 8 : dépôts lagunaires ; 9 : dépôts de marais ; 10 : dépôts de plaine d’inondation ; 11 : dépôts de chenal ; 12 : substrat du Dernier Maximum Glaciaire ; 13 : dépôts anthropiques.
11Identification of lithofacies associations and stratigraphic correlations along seventeen cross-sections (fig. 2B) allowed the reconstruction of the main palaeoenvironmental and palaeogeographic elements developed during the mid-late Holocene in the Arno coastal plain around the city of Pisa.
12Across the study area, laterally extensive lagoonal deposits (“pancone”) are recorded 7-18 m below sea level (fig. 3 and fig. 6A to C), documenting the development of a wide lagoon during the maximum marine ingression, dated to ca. 8000 a cal. BP (Amorosi et al., 2008; Rossi et al., in press). Above “pancone”, a 10-20-m thick highstand fluvio-deltaic wedge is invariably recorded (fig. 3 and fig. 6A to C). This succession documents the gradual infilling of the lagoonal basin and the establishment of a river-dominated environment, resulting from coastal progradation and an increase in fluvial activity since around 6000 a cal. BP (Amorosi et al., 2008). The fluvio-deltaic wedge shows a sedimentary architecture typical of a high-accommodation setting (Catuneanu, 2006), characterised by isolated to locally amalgamated channel fills (fig. 6A to C). In the Arno coastal plain, a remarkable amount of fluvial accommodation was guaranteed during the mid-late Holocene period by highstand sea-level conditions (Lambeck et al., 2004) and a subsiding context. The former reflects the eustatic sea-level trend (Fairbanks, 1989), while the latter is connected both to the extensional tectonic regime of the northern Tuscany and the compaction of organic-rich soft deposits buried beneath the Arno plain (Pascucci, 2005; Rossi et al., in press). The lenticular channel sand bodies (lithofacies Ch) of variable thickness (1-8 m) occur within predominantly fine-grained deposits mainly pertaining to swamp or alluvial overbank lithofacies. Swamp deposits (lithofacies Sw; fig. 6A to C) in the lower part (6-8 m) of the fluvio-deltaic wedge indicate sedimentation in interdistributary (between channel) areas and suggest the development of an extensive delta plain during the earliest phases of progradation. Thin fine sandy layers, occasionally recorded within high-quality stratigraphic data, are interpreted to represent levee or crevasse deposits connected to adjacent distributary channels. Unfortunately, the low resolution of several stratigraphic data precludes a clear recognition of these deposits across the entire study area. Moreover, levee or crevasse sediments are lithologically similar to the coarser portion of swamp deposits in a low-gradient and low-energy depositional setting as the Arno delta plain. Correlative distributary channels locally cut 4-5 m into the underlying lagoonal succession (fig. 6A to C). The transition from a deltaic to an alluvial environment is evidenced by the upward occurrence of channel bodies (lithofacies Ch) encased in alluvial overbank facies that are interpreted to include floodplain, levee and crevasse deposits (lithofacies Of; see Results paragraph). The superposition of an extensive, 5-7 m thick alluvial plain succession onto the deltaic succession is invariably recorded across the Pisa area.
13Following this perspective, from landward to distal locations the fluvio-deltaic succession shows low facies variability. However, few and more localised channel bodies are recorded along the seaward-most cross-sections (fig. 6C), suggesting a less articulate fluvial pattern west of Pisa, in the downstream portion of the Arno Plain. On the contrary, in the subsurface around the city of Pisa several palaeochannels are observed (fig.f 6 A to C). These form a vertically stacked pattern of depositional units accumulated from the late Neolithic (ca. 6000 a cal. BP) to the present.
14Clear discrimination of distinct channel generations is locally hampered by channel reoccupation. However, the use of upper boundaries of channel bodies as stratigraphic markers allows identification of five fluvio-deltaic Units (Units I-V; tab. 2) beneath the city of Pisa (fig. 6A to C and fig. 7). Indeed, these depositional surfaces are clearly discernible from core data by means of the sharp change in grain size from a continuous sandy succession (lithofacies Ch) to the overlying fine-grained deposits belonging to lithofacies Sw or Of (fig. 7). Along all the studied cross-sections, channel bodies are observed to cluster at specific depth intervals from landward to seaward locations (tab. 2). Lack of substantial differential subsidence rates across the small Pisa area (less than 30 km2) supports the stratigraphic significance of channel upper boundaries. Indeed, no fault active zones have been recognised so far (Pascucci, 2005) and the homogeneous depositional succession (“pancone”), invariably underlying the fluvio-deltaic wedge (fig. 6A to C) in the surroundings of Pisa, allow us to consider negligible the influence of differential compaction rates.
15Channel bodies show transition to either “pancone”, swamp (lithofacies Sw) or alluvial overbank deposits (lithofacies Of), as a function of their stratigraphic position (tab. 2). Units I and II, which grade laterally into swamp and, less frequently, lagoonal (interdistributary) deposits, developed into a delta plain environment. On the contrary, a fluvial depositional setting can be envisaged for Units IV and V, which are laterally correlative with alluvial overbank sediments. The transition in the study area from a deltaic environment to an alluvial plain is recorded by Unit III, which is associated with both swamp (mostly at distal locations) and alluvial overbank (mostly at proximal locations) deposits (tab. 2 and fig. 6A to C).
16A preliminary chronological calibration of Units I-V is obtained comparing our subsurface data (fig. 6 and fig. 7) with the sedimentary succession exposed at the archaeological site of Pisa San Rossore, located about 1 km NW of Pisa. At that site, the cutbank exposure and shallow cores (up to 20-m depth) revealed a 2-5 m thick sand body, bearing Roman archeological remains, below about five metres of modern floodplain deposits (fig. 4; Benvenuti et al., 2006). The upper boundary of this channel body, which rests on pre-Roman sands, is recorded 3 meters below sea level, and falls in the stratigraphic range of Unit III (tab. 2). Thus, the integration of stratigraphic, archaeological and radiocarbon data (tab. 1) data suggests a pre-Roman age for Units I and II (ca. 6000-2500 a cal. BP), a Roman age (ca. 2500-1600 a cal. BP) for Unit III and a post-Roman age (< 1600 a cal. BP) for Units IV and V. The absence of additional chronological constrains, especially radiocarbon ages, precludes at present a more accurate absolute chronology for Units I-V.
Tab. 2 – Fluvio-deltaic Units (I-V) recognised within the Arno plain
Tab. 2 – Unités fluvio-deltaïques reconnues dans la plaine de l’Arno
Fluvio-deltaic Unit
|
Depth interval of the upper boundary (relative to s.l.)
|
Adjacent lithofacies
|
Estimated age (cal. a BP)
|
Unit I
|
–12.5 m to -7 m
|
Lagoonal and swamp deposits (lithofacies Sw)
|
ca. 6000-2500 (pre-Roman period)
|
Unit II
|
–6 m to -5 m
|
Swamp deposits (lithofacies Sw)
|
|
Unit III
|
–4 m to -3 m
|
Swamp (lithofacies Sw) and alluvial overbank (lithofacies Of) deposits
|
ca. 2500-1600 (Roman period)
|
Unit IV
|
–2 m to 0 m
|
Alluvial overbank deposits (lithofacies Of)
|
ca. 1600 to present (post-Roman period)
|
Unit V
|
+1 m to +5 m
|
Alluvial overbank deposits (lithofacies Of)
|
|
The depth interval where the upper boundaries of channel bodies cluster, their stratigraphic relationships with adjacent deposits and an estimated age are reported for each unit.
Pour chaque unité est mentionné l’intervalle stratigraphique de concentration des toits des unités de paléochenal, les relations stratigraphiques avec les dépôts adjacents et l’âge présumé.
Fig. 7 – Vertical stacking pattern of channel units within reference cores
Fig. 7 – Succession des unités de chenal dans les forages de référence
Sedimentological features and vertical distribution of lithofacies within the mid-late Holocene succession of two reference cores. Channel units are distinguished on the basis of the depth interval of their upper boundary (see tab. 2 and text). 1: clay and silt; 2: sand; 3: soft clay: 4: organic matter; 5: fossil; 6: evidence of subaerial exposure; 7: anthropogenic deposits.
Caractéristiques sédimentaires et répartition verticale des lithofaciès dans la succession à l’Holocène moyen-supérieur. Les unités du paléochenal sont distinguées sur la base de la profondeur de leur toit (voir tab. 2 et texte). 1 : argiles et limons ; 2 : sables ; 3 : argiles molles ; 4 : matière organique ; 4 : fossile ; 6 : traces d'exposition à des processus d’érosion aériens ; 7 : dépôts anthropiques.
17The identification of five fluvio-deltaic units (Units I-V) beneath the city of Pisa reveals a complex palaeogeographic and palaeohydrographic evolution during the mid-late Holocene, characterised by high channel instability. Based on the aforementioned chronology, three main phases of fluvial activity have been identified (pre-Roman, Roman and post-Roman; tab. 2).
18Shortly after 6000 cal. BP, the hydrological network of the Arno coastal plain started to evolve in the Pisa area, as documented by the abrupt superposition of channel bodies belonging to Units I and II onto the lagoonal succession (“pancone”). These units are characterised by elongate sand bodies that commonly contain mollusc bioclasts derived from the underlying lagoonal deposits. Despite the overall aggradational trend shown by the fluvio-deltaic succession, which is typical of interglacial, high-accommodation conditions, channels attributed to Units I and II are observed to erode locally ”pancone” down to 5 m. This notable phase of incision, which likely occurred during the pre-Roman period (6000-2500 a cal. BP), has been already documented at the archaeological site of Pisa San Rossore (fig. 4). At this location M. Benvenuti et al. (2006) suggested a possible control on channel erosion by small-scale sea-level lowering, favoured by proximity (about 5 km) of the city of Pisa to the coeval coastline. However, several sea-level markers along the North Tyrrhenian coast evidence that a phase of decelerated sea-level rise, without significant sea-level falls, occurred since about 7000 a cal. BP (Lambeck et al., 2004). Thus, other controlling factors, possibly related to abrupt changes in sediment supply or water discharge, and the erodibility of substrate could have favoured this deep incision. Specifically, high-energy floods could be invoked to explain erosion and dismantling of the former lagoonal basin, as suggested by C. Baeteman (2008) and L. Lespez et al. (2010) for the Belgium coastal plain and the Dives estuary (NW France), respectively. In these instances, a phase of estuarine channel erosion has been related to increased fluvial system energy dated around 3000 a cal. BP and triggered by a period of enhanced storms. Although relatively wet and cool climatic conditions have been recorded by pollen data from the pre-Roman unit erosively overlying “pancone” at the archaeological site of Pisa S. Rossore (Mariotti Lippi et al., 2007), no clear evidence of pre-Roman strong hydrodynamic activity linked to climatic oscillations has been observed so far for the Arno plain.
19If the role of climate variability is difficult to specify, the stratigraphic architecture of Pisa underground unequivocally points out that the pre-Roman fluvial network (Unit I and partially Unit II) developed on a peculiar, highly cohesive substrate (“pancone”; fig. 6 A to C). The underlying lagoonal succession is, in fact, composed of homogeneous deposits whose high content in clay and silt makes them very resistant to lateral erosion (Schumm, 1968). Thus, it is likely that the formation of relatively narrow and deep channels onto “pancone” during the initial stages of fluvial evolution was enhanced by the local stratigraphic context.
20About the fluvial network, the location of Units I and II suggests the presence of an active river channel along the northern edge of the study area, close to the present Morto River course, and an articulate fluvial pattern around Pisa (fig. 8A). This fluvial pattern seems to persist during the Roman period as recorded by Unit III, which shows evidence of reoccupation of previous channels. Unit III is the only chronologically well-constrained unit beneath the city of Pisa. For this reason, stratigraphic data of Unit III were synthesised to produce a palaeogeographic map reporting the main features of the drainage network. To test the validity of our results, the map was subsequently compared with published palaeohydrographic reconstructions, obtained combining historical and geomorphological information (Bruni and Cosci, 2003).
21The Roman period map shows three active palaeochannels that laterally border predominantly wet fine-grained deposits (lithofacies Sw), formed into delta plain interdistributary areas (fig. 8A). The presence of a wet palaeoenvironment is supported by pollen data from the archaeological site of Pisa San Rossore (Mariotti-Lippi et al., 2007). An east-west flowing river branch, interpreted as the palaeo-Morto River, is recognised about 2 km kmnorth of Pisa. The activity of this palaeochannel is probably related to an old branch of Serchio River, known as Oseri, which flowed from the Pisani Mountains toward the Arno plain, reaching the sea close to the present Serchio River mouth (Bruni and Cosci, 2003). A more articulate fluvial pattern occurred around the city of Pisa, where the palaeo-Arno merged with a NE-SW flowing river branch (fig. 8A). The orientation of this palaeochannel suggests its possible attribution to the palaeo-Serchio known as Auser. Thus, according to historical (Strabone chronicles) and geomorphological (buried palaeochannels) data, the palaeo-Serchio flowed from the northeast bordering the Pisani Mountains and merged with the palaeo-Arno at Pisa. Based on aerial photo interpretation and image analysis S. Bruni and M. Cosci (2003) proposed a similar palaeohydrographic scenario for the Arno plain during the Etruscan-Roman age (fig. 8B).
22Around the archaeological site of Pisa San Rossore an abandoned meander no more connected with any active fluvial channel is observed between the palaeo-Morto River and the palaeo-Arno (fig. 8A). Taking into account the subsurface spatial distribution of pre-Roman and Roman fluvial units, the reconstructed orientation of fluvial patterns and the palaeohydrographic reconstruction of S. Bruni and M. Cosci (2003), this palaeomeander, one time site of an important Etruscan and Roman riverine harbour, was likely connected to an old branch of the Auser course (fig. 8A and B). However, the hypothesis of a palaeo-Arno provenance, although less probable, cannot be totally rejected. In this respect, further additional data including well logs and the analysis of the petrographic composition of sands composing Unit III could provide conclusive information to discriminate fluvial provenance and clarify river patterns.
23The activity of this palaeochannel seems to be strongly influenced by local hydro-climatic conditions. Indeed, M. Benvenuti et al. (2006) interpreted the high-magnitude flood deposits distinguished within the channel-fill succession (units 1-4; fig. 4) as due to high-frequency hydro-climatic events. Pollen data from the finest-grained fraction of Roman Units 1-4 seem to confirm this interpretation (Mariotti-Lippi et al., 2007). Irregular climatic conditions and maybe increased precipitations have also been invoked by D. Camuffo and S. Enzi (1995), H. Bruneton et al. (2001) and G. Arnaud-Fassetta (2002) as the main controlling factors of the high fluvial activity recorded in the Tiber Delta and Rhône River catchments during the same interval of time. However, the supposed climatic oscillation was of small-amplitude and not recorded across the entire Mediterranean basin. Thus, we are inclined to consider climate as not the only factor contributing to river channel instability. Probably, during the Roman period the significant intensification of human land use (deforestation and agriculture), documented all over Europe (Macklin et al., 2006) as well as in the Arno plain by pollen and archaeological/historical data (Mariotti-Lippi et al., 2007), rendered the fluvial systems more sensitive to climatic changes and hydrological events than before. Indeed, coeval phases of intense alluvial aggradation have been recorded in other European plains and related to an increase in soil erosion favoured by human land use changes (Lespez, 2003; Lespez et al., 2008).
24On the contrary, the spatial distribution of channel bodies in Units IV and V clearly reflects post-Roman human direct interventions (channelisation, dams, dredging and embankment), which led to a less dense fluvial network in the Pisa area (fig. 6A to C). Along the cross-sections, fluvial Units IV and V are less dispersed than Units I-III. In particular, Unit V occurs mainly in association with the present Arno River course and no evidence of palaeo-Serchio branches is observed. This is a consequence of the embankment during the late Middle Ages of both palaeo-Serchio (Auser), which was forced to flow northward to reduce flooding risks, and palaeo-Morto River (Bruni and Cosci, 2003; Martini et al., 2010).
Fig. 8 – Palaeogeography of the Arno plain during the Roman period
Fig. 8 – Paléogéographie de la plaine de l’Arno au cours de l’époque romaine
A: Palaeoenvironmental reconstruction of the Pisa area during the Roman period. Estimated flow directions of palaeochannels are evidenced by white arrows. The white star indicates the location of the archaeological site of Pisa San Rossore. B: Palaeohydrographic reconstruction of the Arno coastal plain during the Etruscan-Roman period obtained through integrated historical and geomorphological data (from S. Bruni and M. Cosci, 2003). 1: delta plain; 2: alluvial plain; 3: pre-Roman network; 4: Roman palaeochannel; 5: present drainage network.
A : Reconstitution des milieux de sédimentation dans la région de Pise à l’époque romaine. Les flèches blanches indiquent les directions présumées des paléochenaux. L’étoile blanche indique la localisation du site archéologique de Pise S. Rossore. B : Reconstitution paléogéographique de la plaine côtière de l’Arno à l’époque étrusco-romaine obtenue par intégration de données hystoriques et géomorphologiques (d’après S. Bruni et M. Cosci, 2003). 1 : plaine deltaïque ; 2 : plaine alluviale ; 3 : réseau hydrographique pré-romain ; 4 : paléochenal romain ; 5 : réseau de drainage actuel.
25Facies analysis and correlation of subsurface data (cores and penetration tests) document a stratigraphic architecture of the mid-late Holocene fluvio-deltaic succession, buried beneath the Arno coastal plain, typical of an interglacial, high-accommodation setting. Isolated to locally amalgamated lenticular channel bodies (1-8-m thick) encased within predominantly fine-grained deposits compose the 10-20-m thick fluvio-deltaic wedge, which overlies an homogeneous succession of lagoonal deposits dated ca. 8000-6000 a cal. BP. Stratigraphy of channel bodies beneath the city of Pisa reveals a vertical stacking pattern of five deltaic-fluvial units (Units I-V). Integration with sedimentological, archaeological and radiocarbon data from a neighbouring archaeological site allowed us to obtain a preliminary chronological framework of the Units and identify three main phases of fluvial activity.
26This study provides evidence for the reliability and validity of a stratigraphic-based approach, supported by archaeological data, to reconstruct the palaeogeographic changes affecting coastal and alluvial plains during the mid-late Holocene. Specifically, the high-resolution reconstruction of the depositional architecture characterising fluvio-deltaic successions can provide new insights into the drainage system dynamics. However, further stratigraphic and palaeoenvironmental data (mainly pollen spectra and provenance analyses), as well as a more accurate chronology, are required to better define the chronology of the Arno plain fluvial events and to contribute to our knowledge and understanding of the factors controlling river behaviour.
27The major results of this study can be summarised as follows: (i) The oldest Units (I and II), formed during pre-Roman period (ca. 6000-2500 a cal. BP), record the first phases of deltaic progradation, which led to the infilling of the wide lagoonal basin previously developed in the area surrounding Pisa. (ii) Subaqueous erosion and dismantling of the former lagoonal basin (with subsequent deposition of Units I-II) were likely favoured by low erodibility of the highly cohesive, homogeneous lagoonal clays. (iii) At time of deposition of Units I-II the interdistributary areas were mainly represented by swamps (upper delta-plain setting). A gradual transition from a deltaic to alluvial environments occurred during the Roman times. (iv) Spatial distribution of Unit III suggests a fluvial network similar to the one depicted by pre-Roman channels (Units I-II), with evidence of channel reoccupation. An active river channel flowed along the northern edge of the study area, close to the present Morto River course, and an articulate drainage network existed around the city of Pisa. In particular, our data documents the confluence at Pisa of the palaeo-Arno with the palaeo-Serchio River. (v) Less than 1 km kmnorthward of Pisa, an abandoned meander, site of an important Etruscan and Roman riverine harbour, is also recorded by Unit III. The subsurface spatial distribution of pre-Roman and Roman fluvial units and the reconstructed fluvial patterns suggest for this palaeomeander a palaeo-Serchio provenance.
This work was founded by the Municipality of Pisa in the form of a grant to Giovanni Sarti. The authors thank M. Benvenuti and two anonymous reviewers for insightful critiques that improved the manuscript.