Page 19 - 23

ARTICLE IN PRESS
K. Lambeck et al. / Quaternary Science Reviews 23 (2004) 1567–1598
The comparisons for the three sites from western
Sicily (16, 17, 18) that have been assessed as being
tectonically stable are illustrated in Fig. 3d. The
observation from Capo Zafferano (16) at B9.5 ka BP
is, as discussed above, an upper limit estimate and the
observation suggests that the cave was occupied when
the platform was well above the zone of storm-wave
action. The Marettimo (18) observation from the same
depth is from lithophaga within a speleothem. The
speleothem age itself is much older, B24 ka BP, and
indicative of a long hiatus between its growth and the
subsequent marine colonisation. The lithophaga age is
similar to that of the Mesolithic site at Capo Zafferano
and this implies that either the lithophaga age is too old
or that the assumption of tectonic stability for
Marettimo is invalid. The MIS 5.5 shoreline is, however,
well developed here at an elevation that is consistent
with tectonic stability and either the lithophaga ages are
unreliable or the site has undergone periods of almost
compensating uplift and subsistence. The two observa-
tions from San Vito Lo Capo (17) both lie near the
lower limit of the predicted values (see inset, Fig. 3c).
The younger beach rock observations from eastern
(19) and (20) northern Sardinia yield age–depth results
that are consistent with the model predictions but before
about 8000 BP the three observed depths are consis-
tently higher than the predicted values for the epoch of
observation. This suggests one of several possibilities:
inappropriate isostatic–eustatic model parameters,
beach-rock ages that are too old, or that the older
beach rock formed at supra-tidal elevations. The
archaeological observation at B7 ka from Caccia Cape
(21) is an upper limit and is compatible with the beach
rock observation from the north coast. The evidence at
B10 ka BP from both Marettimo (Fig. 3d) and Sardinia
could point to the predictions being too deep, but this
would be inconsistent with the Versilia Plain results for
the same epoch (Fig. 3a). Also, the observed spatial
variability is too large to attribute these discrepancies to 5.3. Subsiding sites
the choice of model parameters.
The two observations from Djerba (31), Tunisia,
(Fig. 3f) indicate that here there has been little sea-level possible subsidence. At Palinuro (12) (Fig. 4c) the
change over the past 6000 years, consistent with the
predictions that Late Holocene levels here occur higher
than in Italy (compare the predictions in Fig. 3f with
those for the other sites). The marine shells used to date
the deposits are not in situ and it is possible that they
represent storm deposits in which case their elevations
may be too high.
5.2. Tectonically uplifting sites
Figs. 4a and b illustrate the comparisons of observed
and predicted sea levels for areas undergoing tectonic
uplift as discussed in Section 4. Both observed and
tectonically corrected observations are illustrated. The
1585
accuracy estimates of the latter, st.c.obs, include the
uncertainty in the tectonic correction according to
st:c:obs ¼ ðs2obs þ u2s2t þ t2s2uÞ; ð3Þ
where sobs is the standard deviation of the observed
position at time t, u and su are the rate and its precision
of tectonic uplift given by (1), and st is the precision of
the age determination at time t. The comparisons for the
Rome plain (6) are illustrated in Fig. 4a along with the
result for Punta della Vipera (5). As in Fig. 3 upper and
lower estimates of the predictions are illustrated. A
number of the core samples are dated wood fragments
found in clay horizons that correspond to upper limits
to sea level. Also, depending on the source of the wood
fragments the ages may predate the time of first
flooding. The other data are from salt marsh deposits
that are indicative of levels above but close to sea level.
These latter observations are mostly in good agreement
with the model values back to about 10,000 years BP
whereas the clay-wood samples lie consistently above
their corresponding predicted values. In view of the
latter corresponding to upper-limiting heights and
lower-limiting ages we adopt the salt marsh results in
preference to the clay-horizon results.
The comparisons for eastern Sicily (15) and southern
Calabria (13, 14, 22, 23, 24) are shown in Fig. 3b. The
spatial variability of the model predictions across this
region is small when compared with observational
uncertainties and the predictions shown correspond to
the average value for these sites. The tectonically
corrected observations for Sybaris (23) lie near the
lower limit of the predicted values and this may indicate
that the assumed uplift rate is too high. At the other
locations the tectonically reduced observations all lie
near the upper limits of the corresponding model
predictions.
Figs. 4c and d summarize the comparisons of
observed and predicted sea levels for sites subject to
subsidence rate discussed above is small and may not be
significant. The terrestrial age for the speleothem at
13.9 ka is too old when compared with the serpulid and
lithophaga ages of B8.6 ka from the same depth and is
indicative of a long hiatus between the last preserved
terrestrial deposits and the subsequent marine growth.
Thus we do not consider this observation further. With
the exception of the samples from this depth, the other
observations from this cave site are consistent with the
model predictions. At Fondi (9) (Fig. 4c), the compar-
isons between observations and the model values are
satisfactory for the four older points. For the younger
data, however, the observed values lie consistently at
much shallower levels than recorded elsewhere at the