remote islands, which has been termed ocean siphoning
(Mitrovica and Peltier, 1991). Whether the sea-level curve
at far-field sites peaked abruptly around 6000 years ago
and fell since then (through ocean siphoning) or whether
there has been post-6000 year melt, with a more gradual
peak around 4000 years is difficult to discriminate from
atolls, because it was necessary for the reef rim to accrete
to sea level before evidence would be preserved (Nunn
and Peltier, 2001). The atoll rims have grown from the
surface of the Pleistocene limestone to catch up with sea
level, so the timing of the initiation of a reef flat in
mid Holocene varies geographically from atoll to atoll.
Hydro-isostatic adjustments mean that the elevation of
mid-Holocene reefs can also vary between atolls.
The conglomerate platform on some atolls contains
within it evidence of corals in growth position that formed
a part of a former reef flat (see Conglomerates). Fossil
microatolls are especially useful in this respect, but other
reef flat corals such as Heliopora can also be important in
differentiating units within the conglomerate platform that
mark former reef flat surfaces and indicate higher than present sea level, and overlying storm deposits. Radiocarbon
dates on corals from conglomerate record inputs of corals
detached during storms. For example, ages of 4000–3000
years BP have been obtained on corals from within the conglomerate platform on the Cocos (Keeling) Islands
(Woodroffe et al., 1994). Ages from other atolls seem
broadly comparable, with evidence for variation in Kiribati-Tuvalu (the Gilbert-Ellice chain), from north (where it
may have been as old as 4000 years ago) to south (where
reefs may have reached sea level since 2000 years ago,
McLean and Hosking, 1991). Radiocarbon ages as old as
5500 years BP have been reported from atoll surfaces in
the Tuamotu Archipelago (Pirazzoli and Montaggioni,
1988).
Since the reef rim reached sea level and the reef flat was
formed, there has been further carbonate sediment produced which has continued to infill the lagoon (Purdy
and Gischler, 2005) and has seen the accumulation of sediments to form reef islands around the margin. Several
researchers suggested that island formation on atolls
occurred as a result of this slight fall of sea level (Cloud,
1952; Schofield, 1977a, b; Dickinson, 2004).
David and Sweet (1904) undertook mapping of the
islands and reef flats around Funafuti Atoll, and considered that large Porites corals in growth position indicated
that the sea had been above its present level relative to the
atoll in the past. Similarly Cloud (1952) described outcrops of Heliopora in growth position, above the elevation
that it presently reaches on the reef flat as evidence of
emergence on Onotoa. There has been an ongoing debate
about the extent to which conglomerate of this type is
formed by storms. An expedition to the Caroline and Marshall Islands (CARMARSEL expedition) specifically to
resolve whether the conglomerate was a storm deposit or
as an indicator of higher sea level reached no consensus
(Shepard et al., 1967; Newell and Bloom, 1970). The most
accurate reconstructions of former sea level have been
derived where a fossil sea-level indicator can be related
to its modern equivalent (see Sea-Level Indicators),
and the two most appropriate types of indicator are
microatolls of massive coral and reef flat outcrops of
Heliopora.
Outcrops of conglomerate have been used to infer
higher sea level in the Maldives (Gardiner, 1903; Sewell,
1936), but the evidence is fragmentary. Dated in situ coral
from Addu Atoll was interpreted to infer that the reef flat
had reached modern sea level by around 3000 years BP
(Woodroffe, 1993). Two recent subsequent studies have
proposed detailed sea-level curves for the Maldives during
the Holocene, but differ on whether or not it is possible
to identify evidence to support sea level higher than present
in mid Holocene (Gischler et al., 2008; Kench et al., 2009).
Based on detailed mapping of conglomerate around Cocos,
several in situ microatolls have been radiocarbon dated and
indicate that there has been a gradual fall of sea level from
an elevation 0.5–0.8 m above present over the past 3000
years (Woodroffe et al., 1990b; Woodroffe, 2005).
In the case of islands in the Pacific Ocean there has
been a similar debate. The radiocarbon ages reported by
Schofield (1977a) from Kiribati and Tuvalu appear to be
from corals from the conglomerate that were not in their
growth position. The conglomerate in the northern part
of the Gilbert chain is composed of a lower unit that
contains Heliopora in its growth orientation at a few localized sites (Falkland and Woodroffe, 1997; Woodroffe and
Morrison, 2001), overlain by an upper unit of disoriented
cemented coral clasts. There have been several attempts to
infer sea level either geographically (Grossman et al.,
1998) or at a site (e.g., French Polynesia, Pirazzoli et al.,
1987; Pirazzoli and Montaggioni, 1986, 1988; Funafuti,
Dickinson, 1999), but these have not discriminated the
in situ corals from the more extensive larger conglomerate
outcrops. Large sea-level oscillations or abrupt changes
appear unlikely, and studies that identify large anomalous
fluctuations of sea level have generally been rebutted. For
example, evidence for an abrupt fall of sea level around
1300 AD inferred by Nunn (1998) has been criticized by
Gehrels (2001) both on the basis of how dating evidence
was handled as well as because evidence from a wide geographical area is brought together without regard to spatial
variability.
Reef islands on atolls
A generalized cross-section of an atoll reef island suggests
a typical cross-island morphology comprising a distinct
oceanward ridge and a lesser lagoonward ridge, with a
pronounced swale in the middle. Figure 6 demonstrates
considerable variation in reef island morphology, as well
as human modification. Waves represent the principal processes that build islands. Open ocean swell is filtered at the
reef crest, but a component of the wave energy crosses the
reef flat and reaches the island shore. As a consequence,
60
ATOLLS
(Mitrovica and Peltier, 1991). Whether the sea-level curve
at far-field sites peaked abruptly around 6000 years ago
and fell since then (through ocean siphoning) or whether
there has been post-6000 year melt, with a more gradual
peak around 4000 years is difficult to discriminate from
atolls, because it was necessary for the reef rim to accrete
to sea level before evidence would be preserved (Nunn
and Peltier, 2001). The atoll rims have grown from the
surface of the Pleistocene limestone to catch up with sea
level, so the timing of the initiation of a reef flat in
mid Holocene varies geographically from atoll to atoll.
Hydro-isostatic adjustments mean that the elevation of
mid-Holocene reefs can also vary between atolls.
The conglomerate platform on some atolls contains
within it evidence of corals in growth position that formed
a part of a former reef flat (see Conglomerates). Fossil
microatolls are especially useful in this respect, but other
reef flat corals such as Heliopora can also be important in
differentiating units within the conglomerate platform that
mark former reef flat surfaces and indicate higher than present sea level, and overlying storm deposits. Radiocarbon
dates on corals from conglomerate record inputs of corals
detached during storms. For example, ages of 4000–3000
years BP have been obtained on corals from within the conglomerate platform on the Cocos (Keeling) Islands
(Woodroffe et al., 1994). Ages from other atolls seem
broadly comparable, with evidence for variation in Kiribati-Tuvalu (the Gilbert-Ellice chain), from north (where it
may have been as old as 4000 years ago) to south (where
reefs may have reached sea level since 2000 years ago,
McLean and Hosking, 1991). Radiocarbon ages as old as
5500 years BP have been reported from atoll surfaces in
the Tuamotu Archipelago (Pirazzoli and Montaggioni,
1988).
Since the reef rim reached sea level and the reef flat was
formed, there has been further carbonate sediment produced which has continued to infill the lagoon (Purdy
and Gischler, 2005) and has seen the accumulation of sediments to form reef islands around the margin. Several
researchers suggested that island formation on atolls
occurred as a result of this slight fall of sea level (Cloud,
1952; Schofield, 1977a, b; Dickinson, 2004).
David and Sweet (1904) undertook mapping of the
islands and reef flats around Funafuti Atoll, and considered that large Porites corals in growth position indicated
that the sea had been above its present level relative to the
atoll in the past. Similarly Cloud (1952) described outcrops of Heliopora in growth position, above the elevation
that it presently reaches on the reef flat as evidence of
emergence on Onotoa. There has been an ongoing debate
about the extent to which conglomerate of this type is
formed by storms. An expedition to the Caroline and Marshall Islands (CARMARSEL expedition) specifically to
resolve whether the conglomerate was a storm deposit or
as an indicator of higher sea level reached no consensus
(Shepard et al., 1967; Newell and Bloom, 1970). The most
accurate reconstructions of former sea level have been
derived where a fossil sea-level indicator can be related
to its modern equivalent (see Sea-Level Indicators),
and the two most appropriate types of indicator are
microatolls of massive coral and reef flat outcrops of
Heliopora.
Outcrops of conglomerate have been used to infer
higher sea level in the Maldives (Gardiner, 1903; Sewell,
1936), but the evidence is fragmentary. Dated in situ coral
from Addu Atoll was interpreted to infer that the reef flat
had reached modern sea level by around 3000 years BP
(Woodroffe, 1993). Two recent subsequent studies have
proposed detailed sea-level curves for the Maldives during
the Holocene, but differ on whether or not it is possible
to identify evidence to support sea level higher than present
in mid Holocene (Gischler et al., 2008; Kench et al., 2009).
Based on detailed mapping of conglomerate around Cocos,
several in situ microatolls have been radiocarbon dated and
indicate that there has been a gradual fall of sea level from
an elevation 0.5–0.8 m above present over the past 3000
years (Woodroffe et al., 1990b; Woodroffe, 2005).
In the case of islands in the Pacific Ocean there has
been a similar debate. The radiocarbon ages reported by
Schofield (1977a) from Kiribati and Tuvalu appear to be
from corals from the conglomerate that were not in their
growth position. The conglomerate in the northern part
of the Gilbert chain is composed of a lower unit that
contains Heliopora in its growth orientation at a few localized sites (Falkland and Woodroffe, 1997; Woodroffe and
Morrison, 2001), overlain by an upper unit of disoriented
cemented coral clasts. There have been several attempts to
infer sea level either geographically (Grossman et al.,
1998) or at a site (e.g., French Polynesia, Pirazzoli et al.,
1987; Pirazzoli and Montaggioni, 1986, 1988; Funafuti,
Dickinson, 1999), but these have not discriminated the
in situ corals from the more extensive larger conglomerate
outcrops. Large sea-level oscillations or abrupt changes
appear unlikely, and studies that identify large anomalous
fluctuations of sea level have generally been rebutted. For
example, evidence for an abrupt fall of sea level around
1300 AD inferred by Nunn (1998) has been criticized by
Gehrels (2001) both on the basis of how dating evidence
was handled as well as because evidence from a wide geographical area is brought together without regard to spatial
variability.
Reef islands on atolls
A generalized cross-section of an atoll reef island suggests
a typical cross-island morphology comprising a distinct
oceanward ridge and a lesser lagoonward ridge, with a
pronounced swale in the middle. Figure 6 demonstrates
considerable variation in reef island morphology, as well
as human modification. Waves represent the principal processes that build islands. Open ocean swell is filtered at the
reef crest, but a component of the wave energy crosses the
reef flat and reaches the island shore. As a consequence,
60
ATOLLS