ARTICLE IN PRESS

Quaternary Science Reviews 24 (2005) 2120–2136

Evidence for sea-level oscillations by the ‘‘characteristic thickness’’ of

marine deposits from raised terraces of Southern Calabria (Italy)
Bernard Dumasa,, Pierre Guérémyb, Jeannine Raffyc
a
Département de Géographie, Université de Paris-Val de Marne, 61 Avenue du Général de Gaulle, 94010 Créteil cedex, France
b
Département de Géographie, Université de Reims Champagne-Ardenne, avenue Taittinger, 51100 Reims cedex, France
c
Institut de Géographie, Université Panthéon-Sorbonne, 191 rue Saint-Jacques, 75005 Paris, France
Received 9 March 2004; accepted 12 December 2004

Abstract

Clastic deposits associated with marine terraces can provide evidence for detecting eustatic oscillations. The method is based on
the characteristic thickness (CT) of beach formations, which are generally deposited on a wave-cut platform, between the breakpoint
and the swash upper limit. On any present-day coast, CT reflects equilibrium related to current shore processes with a constant sea
level. It represents a local threshold (Tcr) as, when sea-level rises, CT increases by a value equal to the rise amplitude. This amount is
reduced in uplifted areas by the magnitude of coeval uplifting. The amplitude of sea level variations (detected by CT) can be
calculated knowing CT
Tcr, uplift rate, and the duration of an oscillation. The method is applied to the clastic deposits of 23 marine
terraces stepped between 52 and 172 m and shaped over the MIS 5–MIS 4 period, along the Calabria coast of the Messina Strait.
Not only the great orbital oscillations MIS 5.5, MIS 5.3, MIS 5.1 and MIS 3 are recorded, but also several rapid and low-amplitude
events are detected into each substage. The most surprising result is the ten small oscillations during MIS 5.5, the duration of which
ranges between 2 ka and less than 1 ka, while sea-level oscillates from 0.5 to 12 m in amplitude. By calculating initial relative sea level
highstands and also lowstands preceding them, detailed eustatic oscillations of the curve are drawn.
r 2005 Elsevier Ltd. All rights reserved.

1. Introduction et al., 2004). Sea-level curves are inferred from the

height–age relationships of raised reefs and are con-
Recently numerous studies dealing with detecting sea strained by more and more accurate radiometric dating
level changes since the Last Interglacial period and from corals, which are known to be the most reliable
estimating their frequency and magnitude have been material. Uplifted coasts such as Barbados and the
published. In summary, two parallel approaches, which Huon Peninsula exhibit numerous emergent coastlines
are now converging, have been explored. Some eustatic and are worldwide references for sea-level changes,
curves were derived from benthic or planktonic for- especially since marine isotope stage 5 (MIS 5). A recent
aminifera isotopic records (Shackleton, 2000; Lea et al., eustatic curve (Cutler et al., 2003) is based on U-series
2002; Waelbroeck et al., 2002; Jouzel et al., 2002). dates from these coral terraces while other U-series
Another approach is based upon geomorphologic dates, from Western Australia (Stirling et al., 1998), a
analysis of coral marine terraces (Esat et al., 1999; tectonic stable area, are commonly used for age
Lambeck and Chappell, 2001; Potter et al., 2001; controls.
Chappell, 2002; Lambeck et al., 2002; Schellmann and Conversely, marine terraces associated with clastic
Radtke, 2002, 2004; Potter and Lambeck, 2003; Potter sediments (in short clastic terraces) are seldom used to
detect sea-level changes (Hearty, 1986; Hillaire-Marcel
Corresponding author. Tel: +33 1 45 85 09 51. et al., 1986, 1996; Hearty and Dai Pra, 1992; Kindler
E-mail address: [email protected] (B. Dumas). et al., 1997; Zazo, 1999; Jedoui et al., 2002, 2003; Dumas

0277-3791/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.quascirev.2004.12.011
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2121

et al., 2002). Because corals are often missing in these Outer edge of the Coastline
deposits, attempts for dating mollusc shells have been wave-cut platform Inner edge of the beach
made. But as U-series dates from molluscs are con- Upper swash limit
Breakpoint
troversial because they are not reliable, other methods Shoreline +5
(AAR, TL, OSL and ESR) have been used to get a Surf zone Swash zone
geochronology from clastic deposits belonging to uplift
Hb
marine terraces. But molluscs or fine sands do not give
0 CT 0m
accurate data. hb
Another approach consists of taking into account the
thickness of marine deposits associated with these
terraces. To this end, the aim of this paper is to first -5
explain what we name ‘‘characteristic thickness’’ (CT) 0 10 20 30 40 50 m
and how CT can be used for detecting sea-level Fig. 1. Definition of the characteristic thickness (CT) of beach
oscillations and calculating their amplitude. The second deposits (dotted area) above a shore platform, from any present-day
objective is (1) to apply the method to the impressive coast. CT only depends on current shore processes. It is equal to the
flight of marine terraces stretching along the Calabria sum of Hb (storm waves height) and hb (water depth) at the breaker
zone.
coast of the Strait of Messina, (2) to test the validity of
the method and (3) to verify the assumption that the
terraces have rather a glacio-eustatic origin than a
coseismic one (Dumas et al., 1999), that is an important Coastline
problem in this unstable area. The attempt will be
carried out on the 52–175 m a.s.l. terraces, which belong Sea level
to MIS 4 and MIS 5 periods. So, Calabria results could CT=Tcr
be compared to other uplifted areas and recent eustatic
curves published by Lambeck et al. (2002), Cutler et al. (A)
(2003), and Potter and Lambeck (2003). Coastline

Sea level
2. The CT of beach deposits and its use to calculate sea-
Slr Slr CT
level changes
Tcr

The approach consists of defining what is the CT of

beach deposits from field observations on present-day (B)
coasts, then outlining how this thickness varies with sea- Fig. 2. Variations of the characteristic thickness of beach deposits
level oscillations, on the one hand, and uplifting on the (CT) controlled by Slr. (A) In tectonically stable areas and under stable
other and how it reveals as the amplitude of sea-level sea level, CT is equal to a threshold we name critical thickness (Tcr).
variations as the amount of uplift. (B) When sea level is rising, the shoreline and the coastline are shifting
upward and landward. New beach deposits (big dotted) accumulate on
Along present-day coasts, mechanical wave-action
former beach. CT increases by a value equal to the amplitude of sea
proceeds from the upper limit of sediment deposition, level change.
due to the swash during high tide and storm waves
(Short, 1999), to the lower limit of wave erosion, which
is controlled by the breaking wave depth at the seaward processes (Fig. 2A). In this case, the CT is equal to a
boundary of the surf zone. On rocky coasts (Trenhaile, value that we name ‘‘critical thickness’’ (Tcr).
2002), breaking waves cut a shore platform below mean
CT ¼ T cr .
sea level. It is within this zone that the sediments, which
compose the beach, are or can be deposited. We call CT When sea level is rising, an accommodation space is
of beach deposits the vertical height between the upper created in the coastline environment: both vertical
limit of the swash zone and the lower surf limit at the (upward) and horizontal (landward) displacements of
breakpoint (Fig. 1). Assuming that the height of the the shoreline occur (Fig. 2B). In agreement with the
upper swash limit is equal to the breaker height (Hb), the sequence stratigraphy model, a sea-level rise involves the
breaker depth (hb) at the breakpoint is deduced from the deposition of a retrogradational parasequence (new
ratio H b =hb ¼ 0:78 (Trenhaile, 2002). Fig. 1 shows that beach sediments), which lies on the top of the former
the CT is equal to H b þ hb . beach and increases CT by a value equal to the sea-level
In tectonically stable areas and under stable sea-level rise amplitude. So, every time the CT exceeds the critical
conditions, the CT of detrital deposits is the result thickness, which is a threshold fitted for each studied
of a dynamic equilibrium controlled by current shore area, it reveals a sea-level rise. In these conditions,
 ARTICLE IN PRESS
2122 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

the CT corresponds to the vertical difference in level I

between a relative sea-level lowstand and the subsequent
D

Amplitude of
relative sea-level highstand. Given the CT and the

sea level fall

l
II fal
el
critical thickness, it is possible to calculate the amplitude ev
a-l

Sea-level rise
Se Se
of sea-level rise (Slr) with the simple formula: a-

CT-Tcr
lev
el
Slr ¼ CT
T cr . (1) ris Slf = CT-Tcr+D
Slr = CT-Tcr e
The more the CT exceeds Tcr, the higher the Time
(A)
amplitude of the sea-level rises.
Comparing the CT value corresponding to the same I
sea-level rise in a raised area (Fig 3B) with that of a Uf
stable area (Fig. 3A), one can see that the CT value is D

sea level fall

reduced. It decreases by the magnitude of uplift (Ur)

Worldwide
II
occurring during sea-level rise. So, in this case,

Sea-level rise
calculating the amplitude of sea-level rise requires

CT-Tcr
adding Ur (Fig. 4B). Formula (1) becomes:
Slf = CT-Tcr+D-Uf
Slr ¼ ðCT
T cr Þ þ Ur; (2) Slr = CT-Tcr
(B) Time
where Ur is an indispensable correction, which implies
to know the uplift rate and the duration of the sea- I
level rise. Ur
It is also possible to calculate the amplitude of a Uf
D
sea-level fall (Slf) from the CT of deposits related to the Slr = CT-Tcr+Ur

Amplitude of
sea level fall
subsequent Slr. As the CT of marine deposits represents II
the height of the accommodation space available for
Sea-level rise

sedimentation between two positions of sea level, it is

CT-Tcr

interpreted as the height of space available for deposi-

Slf = Slr+D-(Uf+Ur)
tion generated by a sea-level fall preceding a sea-level Ur Ur
rise, this space being more or less filled in over the
Time
next transgression (Figs. 3A and 4A). The amplitude of (C)
sea-level fall is equal to Fig. 4. Calculating the amplitude of sea-level oscillations between two
terraces associated with two successive highstands (I and II) separated
Slf ¼ Slr þ D, (3)
by a lowstand: (A) in tectonically stable areas; (B) when uplifting
occurs during sea-level fall; (C) when uplifting occurs over the whole
oscillation.
Coastline

sea level
Slr Slr where Slf is the amplitude of sea-level fall, Slr the
CT
Tcr amplitude of sea-level rise, and D the difference in level
between two successive sea-level highstands.
(A) We can substitute for Slr from Eq. (1) into Eq. (3),
which becomes
Coastline
Slf ¼ ðCT
T cr Þ þ D. (4)
sea level CT
CT This formula only applies if sea-level changes occur in
Slr
a stable area (Fig. 4A). When uplifting occurs during a
Ur Ur sea-level fall, it raises the shoreline I, and this tectonic
(B) forcing generates a relative sea-level fall (Fig. 4B), which
is not a worldwide regression of the sea, but is equal to
Fig. 3. Variations of the characteristic thickness of beach deposits the uplift magnitude (Uf) occurring during sea-level fall.
(CT) controlled by tectonics. When sea level is rising (same amplitude It must be subtracted to obtain the real amplitude of the
on A and B) during uplifting (B), CT is reduced by a value equal to the
uplift amount (Ur) and the landward displacement of the shoreline and
eustatic regression
the coastline is also reduced, compared to A. Thick dashed line
indicates the position of the base of beach deposits in stable area (A). Slf ¼ ðCT
T cr Þ þ D
Uf: (5)
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2123

If uplifting lasts over the whole regression-transgres- 3. Evaluating the amplitudes of sea-level fluctuations
sion cycle (Fig. 4C), Eq. (5) becomes strictly applying the CT method to terraces of Southern Calabria

Slf ¼ ðCT
T cr Þ þ Ur þ D
Uf
Ur; (6) In Southern Calabria, along the coast of the Strait of
Messina, previous studies (Dumas et al., 1999, 2000,
i.e. the simplified formula: Slf ¼ ðCT2T cr Þ þ D2Uf. 2002), which have intended to establish detailed
Comparing the amplitude of a sea-level fall with that geomorphologic maps of the terraces and their corre-
of the following sea-level rise, one obtains the vertical sponding coastline (Fig. 6), allow one to draw a
difference or eustatic gap (e) between two successive synthetic profile with 23 terraces from 52 to 172 m a.s.l.
marine highstands (Fig. 7A). The elevation is that of the inner edge of each
terrace, which is measured at the top of the highest
e ¼ Slf
Slr; (7) observed marine deposit, located at the foot of their
where (e) is positive if sea-level fall amplitude is greater corresponding sea cliff. Measurements have been carried
than sea-level rise amplitude. It is the case in Fig. 4C. (e) out repeatedly with seven high-precision altimeters,
is negative if sea-level fall amplitude is smaller than sea- often double-checked with a theodolite or with photo-
level rise amplitude. grammetry based upon GPS data.
The difference in elevation (D) between two successive The topographic surface of every terrace corresponds
shorelines is equal to the algebraic sum to the top of clastic formations, unequally thick, more
often than not, beginning by coarse deposits (blocks,
D ¼ e þ U, (8) cobbles and pebbles) shifting upwards to finer forma-
tions (gravels and sands). Marine sediment shows planar
where U ¼ ðUf þ UrÞ.
stratification (Fig. 8) typical of foreshore layers, which
It shows the respective places of glacio-eustatism and
are up dipping landward close to the coastline (Fig. 9)
tectonics to explain the stepped marine terraces in
while seaward some of them are shifting into clinos-
uplifted areas. If an uplift magnitude is added to a
tratification, probably corresponding to shore face
positive eustatic gap between two subsequent marine
deposits. These detrital formations were deposited in a
highstands, it increases the vertical height, which
coastal environment under shallow-water depth; they lie
separates the shorelines I and II that correspond to
very often on a wave-cut platform, which looks like the
these highstands (Fig. 4B and C). As a consequence,
transgressive surface in the sequence stratigraphy.
marine terraces are clearly stepped and more easily
Sometimes, in narrow bays or in valleys, some terraces
observed in the field. This fact explains that the uplifted
occupy a lateral position compared with the setting of
coasts have recorded more eustatic cycles than stable
the former terraces. They appear as inset terraces, built
areas, during the Quaternary period. If (e) is negative
up into incised valleys, flooded by the sea, like rias. The
and greater than U (Fig. 5), the terrace shaped during a
149.4-m terrace, in the Campanella valley, South of the
first highstand is destroyed by the second one, which
Commenda plateau, gives one of the best examples.
shapes a new terrace. In this case only this younger
Measurements of CT have been carried out from
terrace can be observed in the field. So, according to
every marine deposit. The CT corresponds to the
unequal relative amplitudes of sea-level falls and
difference in level between the top of beach clastic
subsequent sea-level rises, it is possible that all eustatic
deposits, which are thought to be the upper swash limit,
cycles could not be recorded in coastal landforms.
and the base of these deposits lying upon a wave-cut
platform or an unconformity surface, as far as it can be
observed seawards. The result is a minimal value
II
because the retreating sea cliff of the next terrace may
Slf-Slr = e < 0 have eroded the outer edge of the marine deposits. As
D the thickness of deposits varies from one point to
Sea-level rise

e > Uf+Ur
another, because the geological framework of the area
CT-Tcr

I
presents alternating horsts (crystalline and Tertiary
Ur
+ rocks) cut by abrasion platforms and Plio-Quaternary
Uf
sedimentary basins where depositional terraces prevail,
sea-level
fall

Ur we retained the highest value of the CT we have

Ur measured in the field, for each terrace. Fig. 7 shows that
Time four of them exceed 30 m while a fifth is close to 20 m.
The high values correspond to five terraces, associated
Fig. 5. Destruction of a raised marine terrace when the amplitude of
the next sea-level highstand is greater than the former sea-level fall. In with thick deposits, which are build-up terraces,
this case, the marine oscillation corresponding to the deleted terrace composed of retrogradational parasequences, the coast-
cannot be detected in the field. line of each of them being, respectively, raised at 171.9,
 ARTICLE IN PRESS
2124 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

9 8 7 6 5 4
3
2

163.7
Calabro
11 149.9
Salice 10 1
170.4
140.8 154.5

138.0
146.5
149.4 170
Prior to MT VI
l l a
i 153.7
h
g
n 155.2 176.5
A 165.0
129.3
i
d 159.5
o
a n MT VI
P i Villa
150.2 S. Giuseppe
129.5

134.6
141.9

ga 129.3
Lun 123.8 Belfatto
alle
te V 114.8
ren
Tor
86.4 128.3 138.1 MT VI o
71.1 141.7 c
132.7 i
l
l
s co a
nce G
Fra
an i
n teS d
e
Torr
106.3 a
r 1
96.5 115.7 a
i m
Valle t
t u 2
70.3 a i
78.3 90.0 r F
64.5 T 3
85.9 90.9 MT V
60.8 4
81.9 95.9
N
86.1
73.6 MT IV 5

65.9 90.0
61.7 0 100 500 m 6
MT III
52.4
7
MT II
8

Fig. 6. Example of detailed mapping of the stepped series of marine terraces shaped during MIS 5 and MIS 4 in Southern Calabria. Five main
terraces (MT) include a lot of small terraces. (1) MT VI ¼ MIS 5.5; (2) MT V ¼ MIS 5.3; (3) MT IV ¼ MIS 5.1; (4) MT III ¼ MIS 4; (5) MT
II ¼ MIS 3; (6) coastline with its elevation (m); (7) inferred coastline; and (8) linking coastline segments.

119.5, 88.9, 73.6 and 52.4 m. These terraces are clearly inset and present much lesser thick deposits, the
separated in elevation that permits to distinguish them associated CT of which ranging from 5.4 to 14.7 m. A
without any doubt, and to carry out reliable links set of terraces, which is composed of a thick depositional
between the different sectors all along the studied terrace and a lot of inset terraces is called main terrace
coastal area (Fig. 10). In each of them, terraces are (MT); it is a geomorphologic frame of reference in the
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2125

VI 200
171.9
V 180
IA
Tyrrhenian IV 160
ABR

Sea
140
CAL

III 119.5
Messina

SICILY
Reggio II 120
Calabria
88.9
100
73.6
I 80
52.4 marine deposits
60
Present-day
beach CT 40
8.5 20
(A)
s.l.
0
500 1000 1500 2000 m

altitude (m) 52.4 61.7 65.9 73.6 78.3 82.7 85.9 88.9 96.0 105 115.1 119.5 123.7 128.5 133.5 136.8 141.8 146.0 149.4 156.6 161.8 167.2 171.9

CT (m) 34.4 5.6 5.8 19.6 7.3 11.8 10.5 34.9 10.1 >10.1 14.7 42.1 9.7 9.4 7.0 5.7 >7.2 10.3 11.6 7.6 5.4 7.7 120

CT-Tcr (m) 29.4 0.6 0.8 14.6 2.3 6.8 5.5 29.9 5.1 >5.1 9.7 37.1 4.7 4.4 2.0 0.7 >2.2 5.3 6.6 2.6 0.4 2.7 115
(B)

Fig. 7. Synthetic cross-section of 23 stepped marine terraces along the Calabria coast of the Strait of Messina (A) and the characteristic thickness of
deposits associated with each marine terrace (B).

terraces. Fig. 7 shows five main terraces above a sixth,

which is the Holocene plain.

3.1. CT and T cr values

To determine if a marine terrace has been shaped by a

sea-level rise or not, it is necessary to know if its CT
exceeds the critical thickness. This last value has been
measured on the beaches of the Strait of Messina,
assuming minimal differences between present-day
coastal dynamics and those of the Late Pleistocene
period. On this microtidal coast (tide range: 0.33 m) the
height of the upper swash limit has been measured on
ten points; it ranges from 1.6 to 2.8 m (mean
Fig. 8. Beach stacked deposits associated with a small depositional
value ¼ 2.2 m). In these conditions, applying the for-
terrace near the famous site of Ravagnese (Southern suburbs of Reggio
Calabria, see Fig. 7). Located 100 m far from the 127.8 m-coastline, mula H b =hb ¼ 0:78 (see Section 2), the breaker depth
the cross-section is slightly oblique to the coastline. The sedimentary (hb) along the beaches of the Strait of Messina ranges
body lies with unconformity on a wave-cut platform truncating between 3.6 and 2 m and the present-day CT of beach
Messinian sandstone. Out of the photograph on the left and seaward, deposits, which is the critical thickness, between 6.4 and
it begins with high sphericity clasts of different sizes. Landward and
3.6 m, i.e. a 571.4 m mean value. The subtraction of the
upward, it is mainly composed of well-stratified, medium to fine-
grained sands with scarce pebbles (3–4 m thickness). Above, underlined critical value (T cr ¼ 5 m) from every CT (Fig. 7B)
by disc-shaped pebbles, beach gravels indicate final coarsening-upward enhances the contrast between the top terraces and the
trend. This is related to decreasing water depth at the end of sea-level other small terraces. Because the CT exceeds 5 m, the
rising and perhaps is the beginning of progradational sedimentation corresponding marine deposits cannot be only due to
linked to a maximum stillstand. These deposits reflect one of the MIS
wave breaking and swash, in equilibrium with a stable
5.5 sea-level oscillations.
sea level; so there must be a time period over which
sediments were deposited as sea-level rose. It is
concluded that all terraces are surely due to rises in
Calabria coastal landforms along the Strait of Messina sea level. In this raised coastal environment, it is
(Dumas et al., 2000). At the head of a main terrace, the impossible that relative sea-level rises and thick sedi-
first terrace is called top terrace while the others ments could be related with subsiding phenomena. On
characterized by thinner deposits are named small the contrary, as sea-level rose, a wave-cut platform was
 ARTICLE IN PRESS
2126 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

W E

156.6 m

149.4 m

5m

10 m 5 0

156.6 m

149.4 m

5m

10 m 5 0

Fig. 9. MIS 5.5 small terraces and their deposits at Commenda. Top photograph shows topographic features: four marine terraces (underlined in
white) with, for two of them, the point where the coastline elevation was measured. Thanks to a 90 m-long cross-section (a quarry dug into the
Campanella valley-slope bounding southward the Commenda plateau, see Fig. 11) these two depositional terraces could not be mistaken with a man-
made terrace (dark layers, with irregular base) carved into the 156.6-m terrace. Other man-made terraces cut into higher terraces are seen in the
background and the surface of the 149.4-m terrace has been also flattened for cultivation. At the bottom, an interpretative sketch shows four
sedimentary bodies. At the base, foreset beds (MIS 5.5) are truncated by a transgressive surface (white dashed line) on which lies with unconformity a
transgressive tract (MIS 5.5 maximum highstand). It is also truncated by the upward fining-up beach deposits (conglomerate, then sands) associated
with the 156.6-m terrace. Note that bedding layers are updipping as they are approaching the coastline. Beach deposits associated with the 149.4 m-
terrace are inset into the preceding terrace deposits, but they also truncate the two former marine sequences. Vertical hatched area represents man-
reworked material.

shaped in the breaker zone and was covered by beach 3.2. Chronology and uplift rate
deposits; this setting was progressively shifting landward
and upward until the maximum sea-level highstand was The chronology of the important eustatic events is
reached. So, near the coastline, beach deposits were based upon the data, which were obtained from marine
stacking and moreover they were shifting seawards into deposits belonging to terraces VI, V, IV, and II (Fig. 11).
shallow-water marine formations. The amino-acid geochronology constrains formations
Indeed, in four cases, the CT exceeds the 5 m- VI, V, IV to substage 5 (Hearty, 1986; Hearty et al.,
threshold of less than 1 m. It means that if taking the 1986; Dumas et al., 1988). Two of them (VI and V)
6.4 m-maximum value for the threshold the four terraces contain the Strombus bubonius fauna, which has
raised at 61.7, 65.9, 136.8, and 161.8 m could not have disappeared after MIS 5, while Glycymeris and Ceras-
been considered in relation with eustatic oscillations. toderma provide aminozone E; they are older than
But the measurement of CT in the field is a minimal 100 ka (TL and OSL dates, Balescu et al., 1997). They
value and in a raised coastal area, any CT ought to be can be attributed to substage MIS 5.5 or MIS 5.3. But as
increased of the uplift magnitude occurring during sea- these formations are inset formations (see Fig. 2 in
level rise (see above Eq. (2) and Fig. 3C). In contrast, Dumas et al., 2002) marine deposits VI correspond to
only with (CT-5) or with (CT-6.4) there is no doubt MIS 5.5 and formation V to MIS 5.3. The formation IV,
about the glacio-eustatic origin of the other 19 terraces, inset in formation V, gives a younger corrected TL or
and the five thick formations associated with top OSL age (90 and 77 ka) and belongs to aminozone C, so
terraces revealing five important eustatic events. it must be attributed to MIS 5.1. Consequently, the
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2127

Fig. 10. The set of main terraces (roman numerals), along the Calabria coast of the Messina Strait (Northern part). (1) VI ¼ MIS 5.5; (2) V ¼ MIS
5.3; (3) IV ¼ MIS 5.1; (4) III ¼ MIS 4; (5) II ¼ MIS 3; (6) I ¼ Holocene plain; (7) coastline with its elevation (m); (8) inferred coastline; and (9)
linking coastline segments.
 ARTICLE IN PRESS
2128 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

periodicity over the whole interval between each

V Main terrace numbering Str.b. important sea-level peak because it is unlikely that the
Bovetto VI
IR-OSL Str.b. terraces are equally time-spaced. Moreover, 11 terraces
Corrected TL Ravagnese 1 116 ka were shaped between 128 and 105 ka and the same
Str.b. Strombus bubonius number between 105 and 58 ka. So we have to
101 ka
investigate another tentative approach to estimate the
Str.b.
IV V Ravagnese 2 most conclusive age of these terraces.
aminozone C As every eustatic oscillation, which generates a
116 ka
Trombaca 102 ka terrace, is composed of a sea-level fall followed by a
90 ka
sea-level rise (Fig. 4A), the duration of an oscillation is
77 ka equal to the sum of the duration of both events. It can
Acciarello be evaluated from the rate of sedimentation during sea-
aminozone E
level rise through CT
Tcr ( ¼ CT-5) and from the rate
64 ka of generating an available space occurring during sea-
peak 3.3 peak 5.1 peak 5.3 peak 5.5
60 70 80 90 100 110 120 130 level fall through (CT-5+D). In other words, the
duration of an oscillation can be considered propor-
3 4 5.1 5.2 5.3 5.4 5.5 6
tional to 2(CT-5)+D. Strictly speaking, in an uplifted
area as is Southern Calabria, it would be necessary to
Fig. 11. Geochronological data of Calabria marine formations with take into account the amount of uplift occurring during
their error bars.
the eustatic oscillation (Fig. 4C). But as in an oscillation,
Ur must be added to (CT-5) and Uf subtracted from
coastline of each top terrace is correlated to the MIS 5.5, (CT-5+D), the difference (Ur–Uf) is equal to 0 if sea-
MIS 5.3, MIS 5.1 high sea levels, which are known to level rise and sea-level fall have the same duration and it
occur at 12871 ka (Stirling et al., 1998; Muhs, 2002), could be neglected in other cases.
105.170.6 ka (Gallup et al., 2002; Cutler et al., 2003) In a series of eustatic oscillations lasting between t1
and 82.970.4 ka (Edwards et al., 1997), respectively. and t0 , the ratio between the duration of each oscillation
The formation II is constrained by a corrected TL date and the duration t1 2t0 is equal to the ratio between
(6478 ka) and the terrace II can be assigned to the first 2(CT-5)+D of every oscillation and the sum of 2(CT-
peak sea level of MIS 3, which is thought to be 58–60 ka 5)+D relative to all the oscillations from t0 to t1
old (Chappell, 2002). The main terrace III is older than X
t1
64 ka and younger than MIS 5.1. So it can correspond to tn ðt1
t0 Þ
1 ¼ ½2ðCT-5Þ þ D ½2ðCT-5Þ þ D
1 ,
the reef complex IV (71.671.6 ka) at the Huon t0
Peninsula (Chappell et al., 1996a). (10)
With these data, we can calculate the uplift rate U,
applying the formula where tn is the duration of the oscillation n, t1–t0 the
P
t1
U ¼ ½ðH
2:2Þ
Slt
1 , (9) total duration of the oscillations ½2ðCT-5Þ þ D
1 the
t0
where H ¼ 175.5 m, the elevation of the MIS 5.5 sum of accumulating sediments and available accom-
top coastline at Nocella where was found the highest modation space, over the period t1–t0.
S. bubonius outcrop (Dumas et al., 1987), 2.2 the mean This postulates that the rate of sedimentation is
height (m) of the sub-aerial beach, Sl ¼ +5 m, i.e. the constant with time.
relative sea level of the upper MIS 5.5 highstand So it is possible to calculate the duration of the
deduced from field observations in Sardinia by Kindler oscillation n applying the formula
et al. (1997) and in Tunisia by Jedoui et al. (2002, 2003); X
t1
and t ¼ 128 ka. tn ¼ ½2ðCT-5Þ þ D ðt1
t0 Þ ½2ðCT-5Þ þ D
1 .
As a result and assuming constant uplifting since the t0
Last Interglacial period, Calabria coastal area along the (11)
Strait of Messina is uplifted at the mean rate of 1.3 m/ka.
The formula is applied to the time interval 128–58 ka
with the sea-level peak at 105 ka used as age-control.
3.3. Calculating the age of the marine terraces and the
During each oscillation, the duration of Slr and Slf
amplitude of sea-level changes
are obtained applying the following formulae:
To carry out the application of the method, we have Duration of Slr ¼ ðCT-5Þ ðt1
t0 Þ
to estimate the age of the marine terraces, which were X
t1
shaped during each substage between the age controls: ½2ðCT-5Þ þ D
1 , ð12Þ
128, 105, 83 and 58 ka. We cannot calculate a mean t0
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2129

Duration of Slr ¼ ðCT-5 þ DÞ ðt1
t0 Þ 4.2. Age of sea level highstands

X
t1
½2ðCT-5Þ þ D
1 . ð13Þ The age of each sea-level highstand (equal to that of
t0 each terrace) follows directly from the duration of
oscillations (Table 1). We can only obtain an approx-
These durations are used to calculate Ur and Uf, with
imate but not an absolute age. Yet, some results appear
formulae (2) and (5), which allow getting amplitudes of
in good agreement with coral U-dates, particularly over
every sea-level rise and sea-level fall. The relative sea-
the 58–105 ka period. In Table 1, the top terrace
level highstand corresponding to each terrace is calcu-
88.9 m a.s.l., which is associated with a thick clastic
lated with Eq. (9), the relative sea-level lowstand, which
precedes each of them, resulting from the subtraction of formation (29.9 m) and which has been attributed to
the sea-level rise amplitude. The successive sea-level MIS 5.1 (see above Section 3.2), is 83.6 ka old, that is
highstands and sea-level lowstands permit to draw the very close to the reliable sea-level peak at 82.970.4 ka
(Edwards et al., 1997) or 84.370.7 ka in Barbados
relative eustatic curve.
according to Potter et al. (2004). The age of the terrace
78.3 m high could be 77.1 ka, i.e. a date that is
considered as the end of substage 5.1 in Barbados
4. Results (76.270.4 ka in Cutler et al., 2003) or as the ‘‘younger
MIS 5a’’ (76.771 ka in Potter et al., 2004). In this last
4.1. Duration of sea-level oscillations recent paper, Potter et al. indicate that in the Huon
Peninsula ‘‘Dating of reef IV gives a mean age of 71 ka
Assuming that the rate of uplift and that of and appears to represent a sea-level oscillation during
sedimentation were constant with time in Southern the transition from MIS 5 to 4’’. This information could
Calabria, along the Strait of Messina, the CT of marine validate the age (71.6 ka), which is calculated for the first
deposits allows calculating the duration of each eustatic terrace belonging to MIS 4, that is measured at
oscillation, i.e. the interval between two successive 73.6 m a.s.l. For MIS 5.3 it does not exist many reliable
highstands or two successive marine terraces. Table 1 U-dates. With an age estimate of 94.6 ka, the terrace
shows 22 oscillations with unequal durations, occurring raised at 96 m could be correlated to the Huon Peninsula
between 128 and 58 ka. Substage 5.5 includes 11 ‘‘sub-reef Vamain, which is thought to represent an
oscillations; among them, three have a very short earlier sea-level rise prior to the peak of MIS 5a, an
duration (less than 1 ka), six others lasted between 1 interpretation that is supported by evidence of relatively
and 2 ka and only one has a duration higher than 2 ka. high sea level at Florida by 93 ka BP’’ (Potter et al.,
From Lambeck et al. (2002), Termination II lasted 10 ka. 2004). An age of 101.1 ka calculated for the terrace
After the important oscillation (9.5 ka) just prior to the 115.1 m high coincides with the U-dating of 101 ka
105 ka sea-level highstand, the three oscillations of MIS obtained in Barbados (Potter et al., 2004) for the
5.3 lasted more than 3 ka while MIS 5.1 is characterized classical MIS 5.3 but it could be, in our opinion, a good
by three shorter oscillations (1.5–2.9 ka) after 83 ka. As candidate for another oscillation of MIS 5.3. Within the
regards the oscillations of MIS 4, they were the shortest time interval between 128 and 105 ka, it does not exist
ones, ranging between 1.5 ka and less than 1 ka. reliable age-controls, which could be compared with our

Table 1
Duration of eustatic oscillations from 128 to 105 ka and from 105 to 58 ka

Coastline elevation (m) 119.5 123.7 128.5 133.5 136.8 141.8 146.0 149.4 156.6 161.8 167.2 171.9
CT-5m 37.1 4.7 4.4 2 0.7 4 2.2 5.3 6.6 2.6 0.4 2.7 115
D (m) 4.2 4.8 5 3.3 5 4.2 3.4 7.2 5.2 5.4 4.7
Oscillation duration (ka) 9.5 1.7 1.7 0.9 0.8 1.0 1.7 2.5 1.3 0.8 1.2
Sea-level rise duration (ka) 4.5 0.6 0.5 0.2 0.09 0.3 0.6 0.8 0.3 0.05 0.3 10
Sea-level fall duration (ka) 5 1.2 1.1 0.6 0.7 0.8 1.1 1.7 1 0.7 0.9
Terrace age estimate ( ka) 105 114.5 116.2 117.9 118.8 119.6 120.6 122.3 124.8 126 126.8 128
Coastline elevation (m) 52.4 61.7 65.9 73.6 78.3 82.7 86.5 88.9 96 105 115.1 119.5
CT-5m 29.4 0.6 0.8 14.6 2.3 6.8 5.5 29.9 5.1 45.1 9.7 37.1
D (m) 9.3 4.2 7.7 4.7 4.4 3.8 2.4 7.1 9 10.1 4.4
Oscillation duration (ka) 11.2 0.9 1.5 5.6 1.5 2.9 2.2 11 3.1 3.3 3.9
Sea-level rise duration (ka) 4.8 0.1 0.1 2.4 0.4 1.1 0.9 4.9 0.8 0.8 1.6
Sea-level fall duration (ka) 6.4 0.8 1.4 3.2 1.1 1.8 1.3 6.1 2.3 2.5 2.3
Terrace age estimate (ka) 58 69.2 70.0 71.6 77.1 78.6 81.5 83.7 94.6 97.8 101.1 105
Results are obtained applying Eq. (11), (12) and (13).
 ARTICLE IN PRESS
2130 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

171.9

119.5
37.1

18.2
5.8
42.9
results, in spite of detection of one (Bloom et al., 1974;

128

105
115

13
128

5
Plaziat et al., 1998) or two oscillations (Esat et al., 1999;
Schellmann and Radtke, 2004).

11.1
6.2

29.3
4.7

1.7
1.2

4.4

3.0
4.3. Amplitude of sea-level oscillations

126.8
167.2

101.1
115.1

17.5
2.7

1.4

9.7
0.4
3.1

2.0
11.8
Table 2 and Fig. 12 indicate the amplitude of the sea-

4.9

12.0
5.4

3.4
0.9

10.1

29.5
3.2
level variations. The largest amplitudes of sea-level rises
(more than 35 m) correspond to the orbital MIS 5.5,

0.06
161.8

97.8

23.3
0.4

2.9

4 5.1
0.5

1.0
6.2
MIS 5.3, MIS 5.1, MIS 3 peaks: 128 m for Termination

126

105
II against 140 m after Potter and Lambeck (2003), 43 m

11.1
6.6

34.4
5.2

1.2

3.0
against 42 m for MIS 5.3 (Lambeck et al., 2002), 36 m

10

9
against about 46 m for MIS 5.1 (Cutler et al., 2003) and

124.8
156.6

94.6

28.3
2.6

0.4

6.6
3.4

5.1

1.0
6.2
36 m against 26 m for MIS.3 (Chappell, 2002). A 35-m

96
drop of sea level is recorded during MIS 5.2 against 44 m

11.6

18.2

29.2

57.4
7.2

2.2

7.1

7.8
after Cutler et al. (2003). So, as coral-reef terraces,
Calabria clastic terraces have perfectly recorded these

122.3
149.4

83.7
88.9

10.6
6.6

29.9

21.2
7.6

6.3
36.2
important sea-level events and another significant sub-

1
orbital oscillation of medium-amplitude (about 18 m) at
the beginning of MIS 4 (71 ka). These results confirm the

7.3

17.9

6.2

27.4
3.4

1.4

2.4

1.7
validity of the method taking into account the CT of
marine deposits. As they hold with the obvious
120.6
146.0

81.5
86.5

20.8
5.3

11.8
0.8
6.1

5.5

1.2
6.7
oscillations, there is no reason that they would not be
validated when CT is smaller. Between the great orbital

5.4

17.1

8.4

29.1
4.2

1.0

3.8

2.2
variations of sea level, minor oscillations corresponding
to low-amplitude events have been recorded: they range
119.6
141.8

78.6
82.7
4 2.2

0.3

14.6
2.5

6.8

1.4

20.9
8.2
from less than 1 m to about 8 m for sea-level rises and
from 4 to 12 m for sea-level falls. Consequently, CT
4.8

19.3
0.9

4.4

5.3

26.2
1.4
5

surely permits identifying smaller but significant values

Minimal amplitude of sea-level rise and sea-level fall from 128 to 105 ka and from 105 to 58 ka

and detecting metre-scale sea-level rises, three of them

118.8
136.8

77.1
78.3

18.5

23.4
0.7

0.1
0.8

2.3

0.5
2.8
being smaller than 1 m. The limit of detection of the
amplitude of sea-level rises depends on the uplift

15.2

38.6
3.3

4.5
0.8

4.7

4.1
magnitude Ur and the duration of the rise in sea level.

23

When CT is equal to Tcr, no sediment thickness permits

117.9
133.5

71.6
73.6

20.7

14.6

20.9
0.3
2.3

3.0
17.7
to detect a sea-level rise. In this case, cf. Eq. (2), sea-level
2

rise becomes equal to Ur that means it cannot be

8.0

6.7
1.5

28.7

7.7

1.8

27.7
discriminated from the effect of uplifting. So, the limit of
5

detection increases as uplift rate increases. It also

128.5

23.6

70.0
65.9

26.7
4.4

0.7
5.0

0.8

0.2
1.0

depends on the duration of sea-level rise and increases

116

with it. In Calabria, with an uplift rate of 1.3 m/ka, the

limit of detection is about 0.4 m when the sea-level rose
4.8

8.0

31.6
1.5

4.2

3.8

30.4
1.0

Results are obtained applying Eqs. (5), (6),and (9).

during a third of 1000 years, and it rises to 1.16 m with

duration three times longer (1 ka). In each small
114.5
123.7
4.7

26.2

69.2
61.7
0.7
5.4

0.6

29.7
0.1
0.7

oscillation, the drop of sea level is always superior or

equal to sea-level rise. It means that when sea-level rose,
34.8

30.5

60.2
4.2

6.5

9.3

8.2

61

it could reach the prior sea level highstand but never

overpasses it.
119.5
37.1

18.2

52.4
29.4
5.8
42.9

35.6

24.6
6.2
105

58

On the eustatic curve (Fig. 13), the three substages are

well-delimited by very low relative sea level:
123 m for
Uplift during sea-level rise (m)

Uplift during sea-level rise (m)

Uplift during sea-level fall (m)

Uplift during sea-level fall (m)

the penultimate glaciation event, about
60 m for both

Sea-level rise amplitude (m)

Sea-level rise amplitude (m)

Sea-level fall amplitude (m)

Sea-level fall amplitude (m)

Relative high sea level (m)

Relative high sea level (m)

Relative low sea level (m)

Relative low sea level (m)

Terrace age estimate (ka)

Terrace age estimate (ka)

MIS 5.4 and MIS 5.1–MIS 4 transition, then
57 m for

Coastline elevation (m)

Coastline elevation (m)

MIS 5.2. The classical MIS 5.3, MIS 5.1 and MIS 4 sea
level maxima are at about
18,
22,
21 m, respec-
tively. All these values are similar to those deduced from
Table 2

CT-5m

CT-5m

the above-mentioned references (eustatic curves and

D (m)

D (m)

papers), but for MIS 3 our value (
25 m) is in

 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2131

m
130
120
110
100
90
80
MIS 4 MIS 5.1 MIS 5.3 MIS 5.5
70
sea-level rises

60
50
40
30
Amplitude

20
10
0
58 69 70 72 77 79 82 84 95 98 101 105 115 116 118 119 120 121 122 125 126 127 128
sea-level falls

-10
-20
-30
-40
Age (ka) estimate of each terrace

Fig. 12. Histogram of successive amplitudes of sea-level rises and sea-level falls. The values follow directly from Table 2.

20

MIS 4 MIS 5.1 MIS 5.3 MIS 5.5 +5

+
0 +
++
-10 + +
++ +
-21 -21 -18 -18 +
-20 -25 -23
+ + + + ++++ + +
+ +
+ ++ -28 +
++ +
-26 -18
+
+++ + + + + + ++
Relative sea-level (m)

+
-40 +
-39

-60 + +
+
-60 -57 -61

-80

-100

-120
-123 +

-140
55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140

Age (ka)

Fig. 13. The relative sea-level curve from 140 to 58 ka deduced from the stepped marine terraces in Southern Calabria.

disagreement with Cutler et al. (2003) who indicate amplitude oscillations, sea level progressively fell until

74 m at 60.670.3 ka. The most striking feature is the about –10 m. A greater drop (12 m) was the beginning
ten small oscillations over MIS 5.5. Between 128 and of a second period, lasting about 6 kyr (125–119 ka) and
114 ka, they are inscribed in a ‘‘descending’’ curve, including four oscillations. The first has the greatest
which puts relative sea level highstands lower and lower amplitude of all the small MIS 5.5 oscillations, sea-level
from +5 to
26 m, into three main steps. After the MIS rising from
18 to
10 m. Through the next three
5.5 sea level peak and until 125 ka, through three low- others, both rises and falls of the sea decreased (Fig. 12),
 ARTICLE IN PRESS
2132 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

and relative sea level varied between
10 and
20 m. 2002; Potter and Lambeck, 2003) probably for lack of
Then in the third part of the curve, it dropped and data on the position of relative lowstands. Another
oscillated between –20 and –30 m through three oscilla- difference is the duration of the MIS 5.5 sea level
tions, during which both fall and rise amplitudes maximum. In contrast with the writings of Chen et al.
increased, before the MIS 5.2 minimal lowstand (1991), Stirling et al. (1998), and Lambeck et al. (2002)
(
61 m). With MIS 5.3, sea-level rose up to
18 m; the sea did not remain some millennium above 0 m
the two first highstands of this substage appear similar during what they called a ‘‘plateau’’ ending at 120, 116
as are the lowstand between them and the following one. or 118 ka, respectively.
Dropping of the sea intervened with the third and fourth The recent results from Barbados (Schellmann and
oscillations. Ten metres separate the first and the fourth Radtke, 2002, 2004) and the Huon Peninsula put in
sea level highstands. During MIS 5.1, the sea remained evidence three coral terrace build-ups during MIS 5.3
below
20 m and its variations were weaker as the against one previously. The same number of transgres-
fourth and last highstand was ‘‘high enough’’ (Cutler sive events is indicated from the Huon Peninsula raised
et al., 2003), i.e. only 2 m lower than the first:
23 terraces between 107 and 90 ka (Fig. 1 in Lambeck and
against
21 m. It is noticed that, after MIS 5.1–MIS 4 Chappell, 2001; Fig. 2 in Lambeck et al., 2002). In
transition, sea-level rose up again to near
21 m. Calabria, four terraces were shaped during the same
period. It is now admitted that two terraces can be
attributed to MIS 5.1 in Barbados (Potter et al., 2004).
5. Discussion Two more oscillations are recorded in Calabria.
That the number of marine terraces is higher in
5.1. Comparing the sea level curve from Southern Calabria than in Barbados could be explained by the
Calabria with other eustatic records for MIS 5 and MIS 4 uplift rate which is more than three- or four-fold higher
in one case (1.3 m/ka) than in the other (from 0.29 to
The detailed relative sea-level curve from Calabria 0.45 m/kyr). If the ‘‘Calabria curve’’ provides values of
data (Fig. 13) does not look like the eustatic curves relative sea level not too far from truth, it is to be noted
published by Lambeck et al. (2002), Cutler et al. (2003), that, in weakly uplifted coast, many small oscillations
and Potter and Lambeck (2003). Such a high frequency could be recorded into the same reef tract without
eustatic variability over the MIS 5–MIS 4 period has possibility of distinguishing them from morphological
never been reported from anywhere in the world. The features. For example, as MIS 5.1 sea level highstands
greatest discrepancy is observed for the Last Interglacial have nearly the same initial relative elevation, their
period when eleven sea-level oscillations are detected associated deposits can merge into a unique tract. So it
against only one to three from other coastal countries. is difficult to detect more than one oscillation. Leaving
In tectonically stable areas, no more than two terraces accuracy and reliability of dating out of account, this
and consequently, no more than two oscillations have possibility could explain the dispersion in ages of
been noticed during this time interval. It is the case in samples belonging to the apparently same stratigraphic
some papers related to coral reefs in Oahu Island, position. The same reason could explain that generally
Hawaiian Archipelago (Sherman et al., 1993), in the only one or two MIS 5.5 highstands are detected
Bahamas (Hearty and Kindler, 1995; Neumann and through marine terraces. In weakly uplifted areas each
Hearty, 1996; Hearty and Neumann, 2001) or along the of these landforms represents the record of the oscilla-
Egyptian coast of the Red Sea (Plaziat et al., 1998). tions included in the main steps of the MIS 5.5 eustatic
Other authors have drawn the same conclusion from curve as deduced from Calabria terraces (see above in
two superposed clastic beaches at Majorca Island Section 4.3). Moreover, as sea level during the 5.3
(Hillaire-Marcel et al., 1996), in Sardinia (Kindler et maximal peak returned approximately at the same level
al., 1997) and in Tunisia (Jedoui et al., 2002, 2003). In all as it was at about 118 ka, it could have overlapped, or
cases, the relative highstands are found only some swept out the sediments deposited during the preceding
metres a.s.l. In moderate and high raised coastal areas, it third part of MIS 5.5. Unfortunately, because of the
is worth noting that the counting of MIS 5.5 coral lack of dating, it is impossible to correlate the more
terraces has recently increased (T4, T5a, T5b on important oscillation of MIS 5.5 (12-m fall and 8-m
Southern Barbados, according to Schellmann and rise) with other events. Is it the Plaziat et al.’s second
Radtke, 2002, 2004; coral tracts VIIa, VIIb, VIIc in MIS 5.5 highstand? Is the 149.4 m terrace equivalent
the Huon Peninsula following Esat et al., 1999) while in to the reef VIIb? And do the second and third
former papers they were only one or two, respectively. Barbados terraces correspond to the 149.4 and the
But it remains far below the 11 Calabria terraces. 128.5-m terraces?
Moreover, all the MIS 5.5 oscillations detected in The argumentation related to uplift rate cannot be
Barbados and Huon Peninsula are not drawn on applied to the Huon Peninsula (uplift rate up to
recently published eustatic curves (Lambeck et al., 3.2 m/ka) with its only two or three MIS 5.5 coral
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2133

tracts. In this case as uplifting is very high, erosive steeper and when sea-level rises have had more or less
processes could have obliterated some geomorphic short duration, wave erosion did not have enough time
proofs. It can be given in mind that the major coral to shape large landforms so it is still less easy to
terraces include some benches (Y. Ota, pers. informa- distinguish erosional benches. This explains that some
tion), which have not been numbered because they are authors (Westaway, 1993; Miyauchi et al., 1994) who
regarded as coseismic events. described the Calabria terraces, particularly those which
appear very large upon the horsts, have not reported the
5.2. Why so many eustatic events are recorded in series of small terraces with metre-scale sea-cliffs, which
Southern Calabria? are not important features in the landscape. As they can
escape notice, small terraces are included into a large
5.2.1. Geomorphologic survey and geological setting terrace that, consequently, decreases the identified
Compared to other coastal regions, Southern Calab- number of marine terraces and as a consequence the
ria appears as one of the best favourable areas to number of oscillations. In our opinion, a main terrace
observe a great number of eustatic events. This is composed of several small glacio-eustatic terraces
conclusion rests first upon geomorphologic and geolo- with the head terrace corresponding to an important
gical data. Whereas main terraces bounded landward by worldwide eustatic transgression associated with thick
several-metre high sea-cliffs are easily located from field deposits. These results depend on repeatedly accurate
observations or on aerial photographs, it is far more field survey.
difficult detecting small terraces from topographic
features when the coastlines are separated by less than 5.2.2. Connections between general uplifting, uplift rate
5 m, the outer edge of terraces being only 1 m above the and oscillation amplitudes
next coastline. Moreover, in a densely cultivated region In Calabria, the recording of eustatic events is getting
such as Southern Calabria, man-made terraces look like easy by the geological setting but it also depends on
marine terraces and small scarps generated by terrace interference between the uplift rate and the oscillation
cultivation have similar heights as small sea-cliffs amplitudes. In nearly all the cases (Figs. 12 and 13),
(Fig. 9); they spread confusion into the flight of marine successive relative highstands appear lower and lower
terraces especially when natural landforms are shaded or from 128 to 58 ka. It is as though the setting of terraces
destroyed. So the detection of these small terraces needs was the result of uplifting, generating a ‘‘forced
accurate field survey, making use of road-cut cross- regression’’ of the sea (Coe et al., 2003, p. 68), or was
sections and quarries opened into slope valleys. In such shaped during sea level stillstands breaking a very large
particular field conditions, one can observe the inset worldwide dropping of the sea without oscillations. As
landforms and the cut-in-fill formations. regional uplifting raised the coastal area, each shoreline
It is not a chance if so many terraces have been escaped from destruction by the successive relative
detected, revealing numerous sea-level oscillations in stillstands. These assumptions do not allow for the
this part of Southern Calabria. In summary, the proof that each terrace is separated from the following
structural framework is composed of horsts (granite one by a sea-level fall, the amplitude of which is very
covered by Tertiary layers) separated by Plio-Pleistocene often greater than the uplift magnitude occurring during
sedimentary basins, in which the valleys are downcutted. the corresponding sea-level fall. Moreover except the
During Termination II, thanks to an important relief greatest rising of the sea at the head of each substage,
with high uplifted hinterland, a large quantity of sands, sea-level rise amplitudes are always smaller than sea-
gravels, pebbles were supplied to the transgressive sea; level falls. It is an important feature, which gives one
as a result, thick loose materials were deposited into the but not the only explanation for the stepped terraces.
flooded valleys or against fault scarps bounding the In fact the vertical difference between the elevation of
horsts. It is mainly into these materials that depositional two successive terraces is never equal to the only
terraces were generated from cut-and-fill processes. uplift magnitude; this result demonstrates the effect
Before the formation of each terrace, downcutting was of eustatism (Table 3), without which the Calabria
both controlled by uplifting and sea-level fall and it was staircase of marine terraces would not be shaped. The
easy in soft material. These hollow landforms have 149.4-m terrace is one of the best examples. As at
naturally nested the following sea-level rises. These least equal downcutting amplitude is necessary prior
favourable and sensitive areas have recorded a great nesting the deposit thickness (CT ¼ 11.6 m), it becomes
number of glacio-eustatic events while depositional obvious that uplifting which is nearly ten-fold smaller
terraces very often shift laterally to abrasion platforms (Uf ¼ about 2 m) cannot be the major cause for
shaped into the hard rocks of the horsts. These shore explaining the valley cutting, contrary to the sea-level
platforms result from wave action but the associated fall amplitude (nearly 12 m). Even when the eustatic
veneer deposits keep from using their CT as the effect appears so weak that it can be doubtful, for
‘‘signature’’ of a rise in sea level. Where slopes were example, between terraces 167.2 and 161.8 m, high
 ARTICLE IN PRESS
2134 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

Table 3
Difference in elevation (D) between two successive terraces explained by the algebraic sum of eustatic gap (e) and uplift amount (U) during a whole
oscillation (Eq. 8)

Terrace estimate age (ka) 105 114.5 116.2 117.9 118.8 119.6 120.6 122.3 124.8 126 126.8 128
Terrace elevation (m) 119.5 123.7 128.5 133.5 136.8 141.8 146 149.4 156.6 161.8 167.2 171.9
Eustatic gap (e) (m)
8.1 2.6 2.8 2.2 4 4 2.9 1.2 4.0 3.6 4.4 3.1
Uplift (U) amount (m) 12.3 2.2 2.2 1.1 1 1.3 2.2 3.2 1.6 1.0 1.6
D (m) 4.2 4.8 5 3.3 5 4.2 3.4 7.2 5.2 5.4 4.7
Terrace estimate age (ka) 58 69.2 70 71.6 77.1 78.6 81.5 83.7 94.6 97.8 101.1 105
Terrace elevation (m) 52.4 61.7 65.9 73.6 78.3 82.7 86.5 88.9 96 105 115.1 119.5
Eustatic gap (e) (m)
5.1 3.0 5.7
2.5 2.5 0.1
0.4
7.1 4.9 5.8
0.6
Uplift (U) amount (m) 14.4 1.2 2.0 7.2 1.9 3.7 2.8 14.2 4.1 4.3 5.0
D (m) 9.3 4.2 7.7 4.7 4.4 3.8 2.4 7.1 9 10.1 4.4

sea-level fall amplitude (nearly 5 m) is more than four (Ota et al., 2000; Ota and Yamaguchi, 2004), Taiwan
times larger than uplift magnitude (0.9 m). (Yamaguchi and Ota, 2004), New Zealand (Berryman,
Nevertheless uplifting increases the possibility for 1993), the Huon Peninsula (Chappell et al., 1996b)
distinguishing coastlines. When eustatic gap (e) is appear as wave-cut platforms or terraces with veneer
positive (sea-level fall amplitude greater than sea-level beach deposits or thin coral beds. They correspond to
rise), uplift magnitude increases the vertical height the sudden out-of-water emersion of a wave-cut plat-
between two successive coastlines. Without uplifting form elaborated under stable sea-level conditions. So,
small terraces could not be detected in the field, for the deposits are thin (1–3 m). This setting is inconsistent
example, the 82.7-m terrace from the 86.5 m one. And with constructional terraces the CT of which exceeds
when (e) is negative (two cases only) uplift magnitude that of beaches in equilibrium with stable sea level. In
being greater than the eustatic gap puts the former addition, disturbances in sedimentation (slumps, sliding,
terrace higher than the next highstand and preserves it seismites) ‘‘sedimentary structures and deposits includ-
from destruction. ing those observed during and after earthquakes’’ or
sedimentological indicators of paleoseismicity (Ricci
5.3. Rejecting the coseismic origin of terraces Lucchi, 1995) have never been observed in marine
deposits along the Strait of Messina.
All the arguments inferring a glacio-eustatic origin for
the Calabria terraces lead to rule out the assumption of
a coseismic origin. As successive sea-level rises have 6. Conclusion
generated as many as marine terraces, earthquakes
cannot be involved in their formation. This conclusion Marine terraces without reliable dating material are
seems to be a paradox because the Reggio Calabria area not as famous as are the raised coral-reef terraces for
is a grievously famous seismic area that has been shaken recording the glacio-eustatic variations. However, the
by a 7.5 magnitude earthquake in 1908. But the characteristic thickness (CT) of clastic deposits asso-
comparison between field observations before and after ciated with terraces appears as a good tool to detect such
this earthquake (Loperfido, 1909 in Ghisetti, 1992) and oscillations, knowing that it exists on any beach a local
vertical changes predicted by Boschi et al.’s model fault threshold (Tcr) produced by current shore processes,
(1989, in Valensise and Pantosti, 1992) exhibit multi- which oughts to be deduced. In raised areas, every time
centimetre dislevellements only: a subsidence area with a CT (plus uplift magnitude) exceeds Tcr, it can be argued
maximum amplitude of 0.58 m and an uplift zone of that a rise in sea level has occurred. When the vertical
0.07 m. These magnitudes of deformations are incon- gap between two successive coastlines exceeds the
sistent with the 4 or 5 m mean vertical topographic possibility of being only due to the uplift magnitude, a
interval between two successive coastlines in the study sea-level fall is implied. This method is applied to the
area. In the coastal regions, which have been affected by flight of 23 marine terraces, which overlook the Strait of
earthquakes (Alaska, Plafker, 1965; Costa Rica, Plafker Messina, between 172 and 52 m. It is supported by the
and Ward, 1992; New Zealand, Berryman, 1993), the measurement in the field of sediment thickness that
recorded topographic deformations are unequal within a cannot be only stacked by current shore processes
10 km distance while it is not the case within North and without numerous intervening sea-level oscillations. It
South limits of the set of raised coastlines in Calabria. puts in evidence the glacio-eustatic origin of these
The historic or Holocene coseismic terraces in Japan terraces shaped from 128 to 58 ka, eleven of them
 ARTICLE IN PRESS
B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136 2135

during MIS 5.5, four during both MIS 5.3 and MIS 5.1 Berryman, K., 1993. Age, height and deformation of Holocene
and three during MIS 4. This surprising result, without marine terraces at Mahia Peninsula, Hikurangi subduction margin,
other references in the world to our knowledge, is New Zealand. Tectonics 12 (6), 1347–1364.
Bloom, A.L., Broecker, W.S., Chappell, J.M., Matthews, R.K.,
explained by two favourable fact connections. First, the Mesolella, K.J., 1974. Quaternary sea-level fluctuations on a
inherited geological and geomorphologic evolution tectonic coast. New 230Th/234U dates from the Huon Peninsula,
leads to the deposition of thick clastic tracts in or at New Guinea. Quaternary Research 4, 185–205.
the mouth of valleys and against high fault scarps, Chappell, J., 2002. Sea level changes forced ice breakouts in the last
during Termination II. Most of the study terraces are Glacial cycle: new results from coral terraces. Quaternary Science
Reviews, 1229–1240.
inset into these loose materials. Second the 1.3-m/ka
Chappell, J., Omura, A., Esat, T., McCulloch, M., Pandolfi, J., Ota,
mean uplift rate is high enough to protect each coastline Y., Pillans, B., 1996a. Reconciliation of late Quaternary sea levels
due to a sea-level rise from destruction by the next high derived from coral reef terraces at Huon Peninsula with deep
sea level. If it had been higher, continental erosion sea-oxygen isotope records. Earth and Planetary Science Letters 4,
would have been more active and maybe would have 227–236.
dissected the older terraces. The 1.3 m/ka uplift rate is Chappell, J., Ota, ., Berryman, K., 1996b. Late Quaternary coseismic
uplift history of Huon Peninsula, Papua New Guinea. Quaternary
also responsible for an accurate limit of detection: 0.4 m Science Reviews 15, 7–22.
when the time interval between two terraces is a third of Chen, J.H., Curran, H.A., White, B., Wasserburg, G.J., 1991. Precise
1 ka. So, Southern Calabria has recorded very rapid and chronology of the last interglacial period: 234U–230Th data from
short oscillations. fossil coral reefs in the Bahamas. Geological Society of America
This new approach confirms assumptions or results Bulletin 103, 82–97.
Coe, A.L., Bosence, D.W.J., Church, K.D., Flint, S.S., Howell, J.A.,
deduced from geomorphologic studies carried out for a
Wilson, R.C.L., 2003. The Sedimentary Record of Sea-Level
long time (Dumas et al., 1993). In addition to Change. The open University, Cambridge University press 288p.
calculating approximate dating of the terraces, that it Cutler, K.B., Edwards, R.L., Taylor, F.W., Cheng, H., Adkins, J.,
is better than drawing correlation with some peaks on Gallup, C.D., Cutler, P.M., Burr, G.S., Bloom, A.S., 2003. Rapid sea-
isotopic curves (Dumas et al., 1999, 2000), it permits level fall and deep-ocean temperature change since the last interglacial
establishing a detailed relative sea-level curve. On the period. Earth and Planetary Science Letters 206, 253–271.
Dumas, B., Guérémy, P., Lhénaff, R., Raffy, J., 1987. Découverte à
one hand, the method is validated because for four great 157 m d’altitude, près de Nocella (Italie du Sud), du plus haut
transgressions (MIS 5.5, MIS.5.3, MIS 5.1 and MIS 3) it rivage tyrrhénien connu à ce jour. Comptes Rendus de l’Académie
provides amplitudes of sea-level rises, which are des Sciences de Paris 305 (II), 139–142.
consistent with those published on the subject. On the Dumas, B., Guérémy, P., Hearty, P.J., Lhénaff, R., Raffy, J., 1988.
other hand, the curve appears more detailed than curves Morphometric analysis and amino-acid geochronology of uplifted
shorelines in a tectonic region near Reggio Calabria, South Italy.
deduced from coral-reef terraces or from models. It
Palaeogeography, Palaeoclimatology, Palaeoecology 68, 273–289.
looks like some high-resolution isotopic curves. Dumas, B., Guérémy, P., Lhénaff, R., Raffy, J., 1993. Rapid uplift,
Finally, field observations (inset terraces and cut-in- stepped marine terraces and raised shorelines on the Calabrian
fill formations) and the detected sea-level oscillations coast of Messina Strait, Italy. Earth Surface Processes and
lead to rule out a possible coseismic origin for the Landforms 18, 241–256.
numerous Calabria terraces in spite of the high Dumas, B., Guérémy, P., Raffy, J., 1999. Essai de corrélations entre
des étagements de lignes de rivage soulevées en Calabre méridionale
seismicity of the Reggio Calabria area. (Italie) et des courbes isotopiques à haute résolution entre 130 et
40 ka. Quaternaire 2/3, 107–119.
Dumas, B., Guérémy, P., Lhénaff, R., Raffy, J., 2000. Périodicités de
temps long et de temps court, depuis 400 000 ans, dans l’étagement
des terrasses marines en Calabre méridionale (Italie). Géomorpho-
Acknowledgements
logie: relief, processus, environnement 1, 25–44.
Dumas, B., Guérémy, P., Raffy, J., 2002. Variations rapides du niveau
We are grateful to Stephen Kershaw for his critical de la mer depuis le stade 5e en Calabre méridionale (Italie) et dans
and constructive review; his comments help to strongly la Péninsule de Huon (Nouvelle-Guinée). Quaternaire 13 (3/4),
improve the manuscript. We also thank Martine 1–13.
Marandola, a cartographer of Paris-Val de Marne Edwards, R.L., Cheng, H., Murrel, M.T., Golstein, S.J., 1997.
Protactinium-231 dating of carbonates by thermal ionization mass
University, for drawing the figures. spectrometry: implications for quaternary climate change. Science
276, 782–786.
Esat, T.M., Mc Culloch, M.T., Chappell, J., Pillans, B., Omura, A.,
1999. Rapid fluctuations in sea level recorded at Huon Peninsula
References during the penultimate deglaciation. Science 283, 197–201.
Ghisetti, F., 1992. Fault parameters in the Messina Strait (southern
Balescu, S., Dumas, B., Guérémy, P., Lamothe, M., Lhénaff, R., Italy) and relations with the seismogenic source. Tectonophysics
Raffy, J., 1997. Thermoluminescence dating tests of Pleistocene 210, 117–133.
sediments from uplifted marine shorelines along the southwest Gallup, C.D., Cheng, H., Taylor, F.W., Edwards, R.L., 2002. Direct
coastline of the Calabrian Peninsula (Southern Italy). Palaeogeo- determination of the timing of sea-level change during Termination
graphy, Palaeoclimatology, Palaeoecology 130, 25–41. II. Science 295, 310–313.
 ARTICLE IN PRESS
2136 B. Dumas et al. / Quaternary Science Reviews 24 (2005) 2120–2136

Hearty, P.J., 1986. An inventory of last interglacial (s.l.) age deposits Ota, Y., Sasaki, K., Omura, A., Nozawa, K., 2000. Holocene sea level
from the Mediterranean basin: a study of isoleucine epimerization history and tectonic history related to the formation of coral
and U-series dating. Zeitschrift für Geomorphologie, Supplement- terraces at Kikai Island, Northern Ryukyu Islands. The Quatern-
Band 62, 51–59. ary Research 39 (1), 81–95 (in Japanese with abstract in English).
Hearty, P.J., Dai Pra, G., 1992. The age and stratigraphy of Plafker, G., 1965. Tectonic deformation associated with the 1964
quaternary coastal deposits along the Gulf of Taranto (South Alaska earthquake. Science 148, 1675–1687.
Italy). Journal of Coastal Research 8 (4), 882–905. Plafker, G., Ward, S.N., 1992. Back-arc thrust faulting and tectonic
Hearty, P.J., Kindler, P., 1995. Sea-level highstand chronology from uplift along the Caribbean sea coast during the April 22, 1991
stable platforms (Bermuda and Bahamas). Journal of Coastal Costa Rica earthquake. Tectonics 11 (4), 709–718.
Research 11, 675–689. Plaziat, J.C., Reyss, J.L., Choukri, A., Orsag-Sperber, F., Baltzer, F.,
Hearty, P.J., Neumann, A.C., 2001. Rapid sea-level and climate change et Purser, B.H., 1998. Mise en evidence, sur la côte récifale
at the close of the Last Interglaciation (MIS 5e): evidence from the d’Egypte, d’une régression interrompant brièvement le plus haut
Bahama Islands. Quaternary Science Reviews 20, 1881–1895. niveau du Dernier Interglaciaire (5e): un nouvel indice de
Hearty, P.J., Bonfiglio, L., Violanti, D., Szabo, B.J., 1986. Age of late variations eustatiques à haute fréquence au Pléistocène. Bulletin
Quaternary marine deposits of Southern Italy determined by de la Société Géologique de France 169 (1), 115–125.
aminostratigraphy, faunal correlations and uranium-series dating. Potter, E.K., Lambeck, K., 2003. Reconciliation of sea-level observa-
Rivista Italaliana Paleontologia e Stratigrafia 92, 149–164. tions in the Western North Atlantic during the last glacial cycle.
Hillaire-Marcel, C.L., Carro, O., Causse, C., Goy, J.L., Zazo, C., 1986. Earth and Planetary Science Letters 217, 171–181.
Th/U dating of Strombus bubonius bearing marine terraces in Potter, E.K., Esat, T.M., Lambeck, K., McCulloch, M.T., Radtke, U.,
southern Spain. Geology 14, 613–616. Schellmann, G., 2001. Timing and magnitude of oxygen isotope
Hillaire-Marcel, C.L., Gariepy, C., Ghaleb, B., Goy, J.L., Zazo, C., stage 5a and stage 5c sea level oscillations at Barbados:
Cuerda, J., 1996. U-series measurements in Tyrrhenian deposits implications for the melting history of the Laurentide ice sheet
from Mallorca: further evidence for two last-interglacial high sea during OIS 5 and 6. EOS, Transactions AGU 82 (47), 738.
levels in the Balearic islands. Quaternary Science Reviews 15, 53–62. Potter, E.K., Esat, T.M., Schellmann, G., Radtke, U., Lambeck, K.,
Jedoui, Y., Davaud, E., Ben Ismail, H., Reyss, J.L., 2002. Analyse McCulloch, M.T., 2004. Suborbital-period sea-level oscillations
sédimentologique des dépôts marins pléistocènes du Sud-Est during marine isotope substages 5a and 5c. Earth and Planetary
tunisien: mise en évidence de deux périodes de haut niveau marin Science Letters 225, 191–204.
pendant le sous-stade isotopique marin 5e (Eémien, tyrrhénien). Ricci Lucchi, F., 1995. Sedimentological indicators of paleoseismicity.
Bulletin de la Société Géologique de France 173, 63–72. In: Serva, L., Burton Slemmons, D. (Eds.), Perspectives in
Jedoui, Y., Reyss, J.-L., Kallel, N., Montacer, M., Ben Ismaı̈l, H., Paleoseismology. Association of Engineering Geologists. Sp. Publ.
Davaud, E., 2003. U-series evidence for two high Last Interglacial N1 6, Sudbury (MA), USA, pp. 7–18.
sea levels in southeastern Tunisia. Quaternary Science Reviews 22, Shackleton, N.J., 2000. The 100,000-year ice-age cycle identified and
343–351. found to lag temperature, carbon dioxide and orbital eccentricity.
Jouzel, J., Hoffman, G., Parrenin, F., Waelbroeck, C., 2002. Atmo- Science 289, 1897–1902.
spheric oxygen 18 and sea level changes. Quaternary Science Schellmann, G., Radtke, U., 2002. The coral reef terraces of Barbados:
Reviews 21, 307–314. a guide. International Conference on ‘‘Quaternary Sea-Level
Kindler, P., Davaud, E., Strasser, A., 1997. Tyrrhenian coastal Change’’, Barbados, 26 oct–2 nov 2002, 118p.
deposits from Sardinia (Italy): a petrographic record of high sea Schellmann, G., Radtke, U., 2004. A revised morpho- and chronostrati-
levels and shifting climate belts during the last interglacial graphy of the late and middle pleistocene coral reef terraces on
(isotopique substage 5e). Palaeogeography, Palaeoclimatology, Southern Barbados (West Indies). Earth Science Reviews 64, 157–187.
Palaeoecology 133, 1–25. Sherman, C.E., Glenn, C.R., Jones, A.T., Burnett, W.C., Schwarcz,
Lambeck, K., Chappell, J., 2001. Sea level change through the Last H.P., 1993. New evidence for two highstands of the sea during the
Glacial Cycle. Science 292, 679–686. last interglacial oxygen isotope substage 5e. Geology 21, 1079–1082.
Lambeck, K., Esat, T.M., Potter, E.K., 2002. Links between climate Short, A.D. (Ed.), 1999. Handbook of Beach and Shoreface
and sea levels for the past three million years. Nature 419, 199–206. Morphodynamics. Wiley 379p.
Lea, D.W., Martin, P.A., Pak, D., Spero, H.J., 2002. Reconstructing a Stirling, C.H., Esat, T.M., Lambeck, K., McCulloch, M.T., 1998.
350 ky history of sea-level using planktonic Mg/Ca and oxygen Timing and duration of the Last Interglacial: evidence for a
isotope records of Cocos Ridge core. Quaternary Science Reviews restricted interval of widespread coral reef growth. Earth and
21, 283–293. Planetary Science Letters 160, 745–762.
Loperfido, A., 1909. Livellazione geometrica di precizione eseguita Trenhaile, A.S., 2002. Modelling the development of marine terraces
dall’Istituto Geografico Militare sulla costa orientale della Sicilia e on tectonically mobile rock coasts. Marine Geology 185, 341–361.
sulla costa occidentale della Calabria per incarico del Ministero Valensise, G., Pantosti, D., 1992. A 125 kyr-long geological record of
dell’Agricoltura, Industriale e Commercio. Relazione Commissione seismic source repeatability: the Messina Straits (Southern Italy)
Regia, Roma 0, 131–169. and the 1908 earthquake (Ms 7 1/2). Terra Nova 4, 472–483.
Miyauchi, T., Dai Pra, G., Sylos Labini, S., 1994. Geochronology of Waelbroeck, C., Labeyrie, L., Michel, E., Duplessy, J.-C., McManus,
Pleistocene marine terraces and regional tectonics in the Tyr- J.F., Lambeck, K., Balbon, E., Labracherie, M., 2002. Sea-level
rhenian coast of South Calabria, Italy. Il Quaternario 7 (1), 17–34. and deep water temperature changes derived from benthic
Muhs, D.R., 2002. Evidence for the timing and duration of the last foraminifera isotopic records. Quaternary Science Reviews 21,
interglacial period from high-precision uranium-series ages of corals 295–305.
on tectonically stable coastlines. Quaternary Research 58, 36–40. Westaway, R., 1993. Quaternary uplift of Southern Italy. Journal of
Neumann, A.C., Hearty, P.J., 1996. Rapid sea-level changes at the Geophysical Research 98 (B 12), 21741–21772.
close of the Last Interglacial (substage 5e) recorded in Bahamas Yamaguchi, M., Ota, Y., 2004. Tectonic interpretations of Holocene
Island geology. Geology 24, 775–778. marine terraces, east coast of Coastal Range, Taiwan. Quaternary
Ota, Y., Yamaguchi, M., 2004. Holocene coastal uplift in the western International 115–116, 71–81.
Pacific Rim in the context of late Quaternary uplift. Quaternary Zazo, C., 1999. Interglacial sea levels. Quaternary International 55,
International 120, 105–117. 101–113.