Page 10 - Vacchi_Marriner_2016

M. Vacchi et al. / Earth-Science Reviews 155 (2016) 172–197 181
±0.15 m for rotary corers and vibracorers to ±0.05 m for hand coring percussion corer (±0.15 m of sampling error) and has a surface eleva-
(Hijma et al., 2015). The sample thickness is also incorporated into the tion of +2 m MSL obtained using a high accuracy GPS (±0.05 m of
sampling error term. For older bulk sediment samples, this may be as levelling error) and extends to −15 m MSL. The base of the core
large as 0.25 m (e.g., Engelhart et al., 2015). We also calculated the (−15 m to −11 m) is composed of a dark grey unit dominated by
angle of borehole error as a function of the overburden of the sample, silts and clay (47–98%). A charcoal (±0.05 m of thickness error) was
taken in this study to be 1% (Tornqvist et al., 2008). We also included sampled 12.1 m below the surface (0.12 m of angle error) and
an environmental error that can be as up to 0.5 m in saltmarsh and la- −10.1 below MSL. Radiocarbon dating of the charcoal yielded
goonal sediments not showing a clear depositional environment (see 6500 ± 30 14 C years. The mollusc faunal density is low and dominated
Section 3.1.2). The vertical error of the archaeological index points is by lagoonal taxa (C. glaucum, A. segmentum) and the upper muddy-
not constant and is strongly related to the archaeological interpretations sand assemblage in sheltered areas (Cerithium vulgatum). The ostracods
provided in the original papers. However, preservation of fossil biological are dominated by brackish water (C. torosa) and freshwater (Candona cf.
zones on the archaeological structures may provide vertical uncertainties lactea, Candona cf. compressa, Potamocypris variegata, Darwinula
of up to 0.05 m (Marriner and Morhange, 2015). stevesoni, Illyocypris gibba) species. The lagoonal taxa and low faunal
Because of the lack of specific Mediterranean studies, we decided to densities indicate that the charcoal was deposited in an inner or semi-
not include an error term for potential changes in the paleotidal range enclosed lagoon. Thus we assigned the charcoal sample a reference
(e.g., Hill et al., 2011). Nonetheless, it cannot be excluded that this water at the midpoint between 0 and −1 m MSL (−0.5 m MSL) and
error may have affected the indicative range of the data points during an indicative range of 0 to −1 (±0.5 m). We are aware that charcoal
the Holocene. found in lagoons may have been transported and yield ages older than
To account for sediment compaction (e.g., Edwards, 2006), we the lagoon. Nonetheless, this error is inferior to the error margins asso-
subdivided the saltmarsh and lagoonal index points into basal and inter- ciated with radiocarbon dating. In this instance, multiple samples (char-
calated categories (Horton and Shennan, 2009). Basal samples are those coal and peats) yielded coherent 14 C ages for this unit (Marriner et al.,
recovered from within the sedimentary unit that overlies the incompress- 2012b) supporting the chronology for this lagoonal environment.
ible substrate, but not directly at the intersection between the two The calculation of RSL and age, including the error terms, for this
(e.g., Engelhart and Horton, 2012). Basal samples may have undergone index point was:
some consolidation since deposition. Intercalated samples correspond to
organic horizons in between clastic layers and, therefore, they are RSL ¼ ‐10 :1m− ‐0:5mð Þ
generally most prone to compaction (e.g., Hijma et al., 2015). Where strat- ¼ ‐9:6m
2 2
igraphic information was unavailable for an index point, we conservative- Error ¼ ∑ð0:5m indicative range þ 0:05 m levelling error
2
2
ly interpreted it as intercalated. Index points from archaeological markers, þ 0:2m sampling and thikness errors þ 0:12 m angle error Þ 1=2 ¼ 0:55 m
fixed biological and beachrock samples are virtually compaction free.
14
Age ¼ 6500 30 Cyears ¼ 7326 ka 7471 ka BP 2σð Þ:
3.5. Age of sea-level indicators
In our database, the age of the samples was estimated using radio- 4. Predictions of RSL
14
carbon ( C) dating of organic material from salt and fresh water
marshes, marine and lagoonal shells, beachrock bulk cement as well The RSL model predictions presented in the following sections have
as from the archaeological age of coastal structures. Because the produc- been obtained by solving the Sea-level Equation (SLE, Farrell and Clark,
tion of atmospheric radiocarbon has varied through geological time, 1976). The SLE, which describes the spatiotemporal variations of sea-
radiocarbon ages were calibrated into sidereal years with a 2σ range. level associated with the melting of late Pleistocene ice sheets, has
All samples were calibrated using CALIB 7.0. We employed the IntCal13 been solved numerically by means of an improved version of the open
and Marine13 (Reimer et al., 2013) datasets for terrestrial samples and source code SELEN (Spada and Stocchi, 2007). SELEN assumes a laterally
marine samples, respectively. Where available, information on the nec- homogeneous, spherical, incompressible and self-gravitating Earth with
essary reservoir correction was taken either from the Marine Reservoir Maxwell rheology. It includes the effects of rotational fluctuations on
Database (Reimer and Reimer, 2001) or from published values. All index sea-level (Milne and Mitrovica, 1998) and accounts for horizontal mi-
points are presented as calibrated years before present (ka BP), where gration of shorelines following the method outlined by Peltier (2004).
year 0 is AD 1950 (Stuiver and Polach, 1977). A concern with old radio- In all our computations, we have employed the ICE-5G model of
carbon ages is the correction for isotopic fractionation (Tornqvist et al., Peltier (2004) to predict a nominal RSL curve, based on a three-layer
2015). This became a standard procedure at most laboratories by the approximation of the multi-layered viscosity profile VM2 (Table 2). To
late 1970s (Stuiver and Polach, 1977), but some laboratories have only account for the uncertainties in the viscosity profiles, we performed fur-
applied this correction since the mid-1980s (Hijma et al., 2015). ther runs varying the viscosity profiles in each layer within a reasonable
In the database, the majority of ages were analyzed after 1990 and, range; the minimum and maximum viscosity values are shown in
therefore, most are not subject to this potential error. For the 31 ages Table 2. The thickness of the elastic lithosphere has been kept constant
that are affected, we followed the procedure of Hijma et al. (2015) to (90 km) in all of our calculations.
correct for isotopic fractionation. The age of the archaeological RSL
data-points is given both by the period of construction and/or by the 5. Results
14
C dating of biological indicators fixed on the coastal structures
(Morhange and Marriner, 2015). In our database, the age of the archae- We re-assessed 917 radiocarbon and archeologically dated RSL data-
ological RSL data-points is restricted to the last 5.0 ka BP. points along the western Mediterranean Sea. We reconstructed the RSL
histories of 22 regions using a database composed of 469 index points
3.6. Example of the production of a lagoonal index point from Malta and 177 limiting points (Appendix A, references of the original sources
in Appendix B, Fig. 5A,B). We excluded 271 sea-level datapoints (113
In order to better explain our new methodology for the production index points and 158 limiting points), which were deemed to not be ap-
of index points in the Mediterranean, we here provide an example propriate for the RSL reconstructions (Appendix C). For instance, we
from a marsh coring in Malta. The sedimentary sequence from the discarded the RSL data-points that may have been significantly
coastal plain of Burmarrad is presented in Fig. 4 (Marriner et al., affected by local co-seismic tectonic uplift or subsidence. For example,
2012b). The core BM1 (35.93 N°; 14.41° E) was obtained using a near Punta delle Pietre Nere (North Apulia region, #21), Mastronuzzi