Directory UMM :Data Elmu:jurnal:J-a:Journal of Asian Earth Science:Vol18.Issue5.2000:
Journal of Asian Earth Sciences 18 (2000) 603±631
Tertiary evolution of the Eastern Indonesia Collision Complex
T.R. Charlton
Ridge House, 1 St. Omer Ridge, Guildford, Surrey, GU1 2DD, UK
Received 14 January 1999; accepted 6 July 1999
Abstract
Eastern Indonesia is the zone of interaction between three converging megaplates: Eurasia, the Paci®c and Indo-Australia.
The geological basis for interpretations of the Tertiary tectonic evolution of Eastern Indonesia is reviewed, and a series of plate
tectonic reconstructions for this region at 5 million year intervals covering the last 35 million years is presented.
The oldest reconstruction predates the onset of regional collisional deformation. At this time a simple plate con®guration is
interpreted, consisting of the northward-moving Australian continent approaching an approximately E±W oriented, southwardfacing subduction zone extending from the southern margin of the Eurasian continent eastwards into the Paci®c oceanic
domain. Beginning at about 30 Ma the Australian continental margin commenced collision with the subduction zone along its
entire palinspastically-restored northern margin, from Sulawesi in the west to Papua New Guinea in the east. From this time
until ca 24 Ma, the Australian continent indented the former arc trend, with the northward convergence of Australia absorbed
at the palaeo-northern boundary of the Philippine Sea Plate (the present-day Palau-Kyushu Ridge).
At ca 24 Ma the present-day pattern of oblique convergence between the northern margin of Australia and the Philippine Sea
Plate began to develop. At about this time a large portion of the Palaeogene colliding volcanic arc (the future eastern
Philippines) began to detach from the northern continental margin by left-lateral strike slip. From ca 18 Ma oblique southwarddirected subduction commenced at the Maramuni Arc in northern New Guinea. At ca 12 Ma the Sorong Fault Zone strike-slip
system developed, eectively separating the Philippines from the Indonesian tectonic domain. The Sorong Fault Zone became
inactive at ca 6 Ma, since which time the tectonics of eastern Indonesia has been dominated by the anticlockwise rotation of the
Bird's Head structural block by some 30±408.
Contemporaneously with post-18 Ma tectonism, the Banda Arc subduction±collision system developed o the northwestern
margin of the Australian continent. Convergence between Indo-Australia and Eurasia was accommodated initially by northward
subduction of the Indian Ocean, and subsequently, since ca 8 Ma, by the development of a second phase of arc-continent
collision around the former passive continental margin of NW Australia. 7 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction
Eastern Indonesia, situated at the intersection of the
Alpine-Himalayan and Circum-Paci®c orogenic belts,
is the location of the Earth's fundamental convergent
triple junction. Long-term interaction of these three
plates (taking the de®nition of plates at the coarsest
scale) has resulted in the present-day situation where a
complex arrangement of platelets with poorly de®ned
plate boundaries cover an area equivalent to most of
E-mail address: [email protected] (T.R. Charlton).
western Europe or the western United States. It is my
contention, however, that this complexity is only a
relatively recent development, and that back in time,
through the Tertiary, the tectonic situation rapidly
simpli®es, so that by the mid-Palaeogene the fragmented platelets have resolved themselves into a relatively
simple pattern of coherent plates. Starting from this
simple pre-Neogene con®guration, it is the aim of this
paper to generate the complexity of the present-day by
a few fairly simple changes in regional dynamics, with
these changes largely explicable in terms of plate
boundary interactions taking place within the evolving
eastern Indonesia region.
1367-9120/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 7 - 9 1 2 0 ( 9 9 ) 0 0 0 4 9 - 8
604
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Fig. 1. Regional tectonic setting of the eastern Indonesia collision zone. Inset: approximate convergence vectors for the three main plates Eurasia
(EU), Paci®c (PA) and Indo-Australia (AU). M.S.C.Z.: Molucca Sea collision zone. N.B.B.: North Banda Basin. S.B.B.: South Banda Basin.
The region under consideration comprises the eastern half of Indonesia from Sulawesi island in the west
to the Papua New Guinea border in the east (Fig. 1).
Peripheral regions beyond these limits are, quite deliberately, only treated in the most general terms. The
reason for this exclusive focus on eastern Indonesia is
that, as will be discussed in the following section, there
is no clear consensus on how the larger region viewed
at a sub-global scale has evolved through the Tertiary.
2. External constraints on the collision model
The Eastern Indonesia Collision Complex involves
the interaction of three or four major plates: Indo-
Australia, Eurasia and the Paci®c (or Caroline and
Philippine Sea) plates. The gross kinematics of these
plates (Fig. 1) provides an important framework
within which to interpret the evolution of this region.
In particular, the relative movements of the two main
continental fragments, Australia and Southeast Asia,
should provide a good constraint on the evolution of
the collision complex. However, this constraint is not
as strong as could be desired because the two continents form the end-points in a long global plate
motion circuit (Southeast Asia±Eurasia±Africa±Antarctica±Australia), and uncertainties at each stage in
the circuit are compounded in the ®nal result.
It is commonly stated in the literature (e.g. AudleyCharles et al., 1988; Packham, 1990; Rangin et al.,
Fig. 2. Palaeolatitudinal change for a nominal point on the Australian continent (108S, 1248E, currently located in southern West Timor). The individual curves are linear interpolations between measured points (the dots) from published palaeogeographic maps. All apart from the Rangin
et al. (1990) curve are measured directly from the position of Timor on the maps; the Rangin et al. (1990) curve is extrapolated from the
southern tip of the Aru Islands (ÿ38). In some of the maps for older time intervals, extrapolation was necessary from another point on the Australian Plate as the principal reference point had moved o the southern edge of the maps. The two curves from Soeding et al. (1997) are generated from dierent reference parameters: the palaeomagnetically-derived curve [PM] is based on the North America reference frame (Harrison
and Lindh 1982); the hotspot [HS] curve uses the Mueller et al. (1993) hotspot reference frame. The grey shaded area is derived from the rotation
poles of Australia relative to Antarctica as calculated by Royer and Sandwell (1989). The lower limit of the shading assumes that Antarctica has
remained latitudinally ®xed, whilst the upper limit assumes that the ridge system south of Australia has remained ®xed. Regions below the `static
Antarctica' curve would require Antarctica to be moving northward towards SE Asia, whilst the region above would require Antarctica to be
moving away from SE Asia. Together these bounds may provide long-term limits on the likely northward movement of Australia. The lower
graph shows the latitudinal uncertainty in km of the eleven map-derived curves at 5 Ma intervals, as given by the dierence between the highest
and lowest palaeolatitudinal estimate. The dashed line is the same data, but excluding the Dercourt et al. (1991) curve. The palaeolatitudes from
the present study are not shown, but are virtually indistinguishable from the Paleomap curve since 25 Ma.
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Fig. 2.
605
606
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
1990; Hall, 1996) that the position of the Australian
continent through time is reasonably well constrained.
This, however, is not strongly supported by Fig. 2
which shows the palaeolatitude of a point on the Australian continental margin (108S, 1248E, currently
located in the Timor collision zone) through the Tertiary, as interpreted by these authors, together with a
number of other published reconstructions. This shows
that interpreted mean rates of northward movement
for Australia during the last 30 Ma vary by a factor of
two from about 40±80 km/m.y. Only ®fteen million
years ago there was as much as 880 km (88) dierence
in interpreted palaeolatitude between the extremes,
which is close to the present-day north±south spread
of the entire eastern Indonesia collision complex. Even
excluding the Dercourt et al. (1993) track, which is
clearly out of line with the others, only reduces the
uncertainty to ca 700 km at 30 Ma (dashed line in the
lower part of Fig. 2). This allows an uncertainty of up
to ca 9 Ma for the time at which a particular point on
the northern Australian continental margin reached a
particular latitude, given the current `absolute' northward motion of Australia at ca 75 km/m.y. (see
below).
The faster rates in Fig. 2 are associated with models
that utilise a hot spot reference frame, whilst the
models with a slower northward motion use a palaeomagnetically-derived reference framework. This is illustrated in Fig. 2 by the reconstructions generated in the
Soeding et al. (1997) program, which allows calculation of continental palaeopositions in both reference
frames. Thus the latitudinal spread in Fig. 2 is principally a result of the methodology used in determining
the palaeolatitudes (hotspot or palaeomagnetic frameworks). This explains the spread, but does not remove
the problem for the palaeomagnetically-derived reconstructions that the rate of northward movement of
Australia has apparently slowed considerably since 10
Ma relative to latitude. Reconstructions in the hotspot
reference framework generally do not show this apparent slowing.
Present-day plate motion calculations suggest relatively high convergence rates between Australia and
Eurasia. The plate motion calculations of Minster and
Jordan (1978), which are determined in the hotspot
reference frame, indicate that the reference point (108S,
1248E) is moving ca 198 east of north at a rate of
about 79.5 km/m.y. (i.e., a northward velocity of ca
75 km/m.y.). In the NUVEL-1A model of de Mets et
al. (1990, 1994) the relative convergence of Australia
with Eurasia is 79.8 km/m.y. bearing 0168 Ð i.e. a
northward velocity of 76.5 km/m.y. assuming Eurasia
to be stationary. These ®gures are at the top of the
range of long-term northward displacements for Australia shown in Fig. 2, in line with long-term northward displacements in the hotspot reference frame, but
considerably faster than those constructed in the
palaeomagnetic framework.
The palaeoposition of Southeast Asia is also less
than adequately constrained. Taking Java island as a
representative element of this platelet, there is signi®cant disagreement in the published literature as to its
movement history during the Tertiary. Thus Rangin et
al. (1990) and Daly et al. (1991) inferred clockwise and
overall westward displacement of Java through the
Tertiary, whilst Hall (1996) interpreted overall anticlockwise and eastward displacement. Similar disagreement is shown in the palaeomagnetic literature with
regard to the larger region, with many authors proposing large anticlockwise rotations of Southeast Asia
during the Tertiary (e.g. Fuller et al., 1991), whilst
others have concluded little or no overall rotation
(Lumadyo et al., 1993). However, it should be added
that most palaeomagnetic studies ®nd only small
changes in latitude for Southeast Asia through the
Tertiary.
In summary, whilst the globally determined palaeocontinental positions and instantaneous plate motion
calculations provide important external constraints on
the tectonic evolution of eastern Indonesia, they cannot at present provide more than a very generalised
framework. Tying an evolutionary model too closely
to any particular global or very-large-regional tectonic
model is probably not sensible at this stage because
there is such a wide variation between alternatives.
The model developed here assumes fairly high relative velocities between Australia and Southeast Asia in
line with the hot spot reference frame, principally
because this seems to ®t best with inferences on the
timing of collision arising from interpretation of eastern Indonesia geology. A constant northward velocity
is assumed here for Australia since the Eocene because
most calculations of the motion of Australia based on
hot spot traces (e.g. Duncan, 1981; Wellman, 1983) or
relative to Antarctica (Royer and Sandwell, 1989;
Veevers et al., 1991) do not recognise major changes in
the rate of northward displacement through this
period. In terms of available models, the chosen path
of Australia most closely follows the Paleomap reconstructions of Scotese and co-workers (see Scotese,
1999).
3. Regional structural elements
Eastern Indonesia forms the zone of interaction
between the Eurasian/Southeast Asian and Australian
continents, and the Paci®c and its sub-plates (the
Philippine Sea and Caroline plates: Fig. 1). In the following two sections the principal structural elements of
this region will be outlined. The main bounding elements are described in this section, followed in Sec-
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
tion 4 by the smaller crustal elements within the collision zone.
3.1. Southeast Asia
In the context of eastern Indonesia, the Southeast
Asian Plate (a sub-element of the Eurasian Plate) consists of the Sundaland continental block and the Celebes Sea oceanic fragment (Fig. 1). Eastern Sundaland
comprises Java and Borneo islands and the intervening
shelfal Java Sea. To the south, Sundaland is bounded
607
by the Sunda Arc subduction system. In the southeast
the margin of Sundaland is transitional into the Southwest Sulawesi structural province discussed later. This
transitional region is characterised structurally by
Palaeogene rift grabens inverted by Miocene compression, which has produced a distinctive cross-sectional expression of `Sunda folds' (e.g. Letouzey et al.,
1990). In the eastern Sundaland margin these inversion
structures predominantly trend E±W, but near to Sulawesi the trend changes markedly to nearly N±S
(Fig. 3).
Fig. 3. Principal structural elements of eastern Sundaland and western Sulawesi, and the main structural lineaments of SE Sulawesi. Adapted
from Letouzey et al. (1990), Kavalieris et al. (1992), Bergman et al. (1996) and Parkinson (1996). Bathymetry at 1000 m intervals.
608
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Further north the Makassar Straits between Borneo
and Sulawesi probably developed initially from the
same Palaeogene phase of extension recognised on the
eastern Sundaland margin, although the degree of
extension in the Makassar Straits was considerably
greater. The South Makassar Basin (Fig. 3) is thought
to be underlain by continental crust strongly attenuated by Palaeogene extension (Situmorang, 1982),
whilst the North Makassar Basin may be underlain by
Palaeogene oceanic crust, as is found in the Celebes
Sea to the north (Weissel, 1980; Smith et al., 1990).
The eastern and southern margins of the Makassar
Straits and Celebes Sea respectively comprise fold and
thrust belts (described in more detail later) associated
with the Neogene westward displacement of Sulawesi.
Consequently the Makassar Straits additionally possess
some facets of a foreland basin superimposed on the
older rifted margin structure (Bergman et al., 1996).
island arc system, rather than a pre-collisional disaggregation of the Australian margin as implied by the
allochthonous terrane models.
Immediately to the west of the Australian continent
is Mesozoic oceanic crust of the Indian Ocean. The
western continental margin of Australia was formed
by rifting phases in the Late Jurassic and Early Cretaceous, with extension directions implied by ocean ¯oor
magnetic lineation patterns oriented between about
WNW±ESE and NW±SE. This oceanic crust is subducting northward beneath the southern and eastern
Sunda Arc (from southern Sumatra eastwards), whilst
the Australian continental margin is in collision with
the Banda Arc system, which is the direct eastward
extension of the Sunda Arc. Further west still, the
Indian Ocean is underlain by Cenozoic oceanic crust,
which is being subducted at the SW continental margin
of Sundaland beneath most of Sumatra.
3.2. Australia
3.3. The Paci®c
The northern part of the Australian continental
block comprises Australia, the Northwest and Arafura
shelf seas, and the southern half of New Guinea island
(Irian Jaya province of Indonesia and Papua New Guinea: Fig. 1). The boundary between autochthonous
Australian crust essentially undeformed during the
Tertiary and the collision complex to the north is
marked by the Timor-Tanimbar Trough thrust front,
the extensional front of the Aru Trough, the TareraAiduna strike-slip fault, and the Central RangesPapuan Foldbelt deformation front. Immediately
north of this orogenic front is deformed Australian
crustal material (the parautochthonous zone) and then
crustal elements allochthonous to the Australian margin. Many parts of the parautochthonous zone are
linked directly to the autochthon to the south, and can
be restored by standard palinspastic techniques where
sucient data is available. The precise point of origin
of other `Australian' crustal elements (e.g. the Bird's
Head and Banggai-Sula crustal fragments to be discussed below) is more questionable, and some authors
(Pigram and Panggabean, 1984; Pigram and Davies,
1987; Struckmeyer et al., 1993, etc.) have suggested
that these terranes have had long and complex kinematic histories independent of the main Australian
Plate. Detailed discussion of these so-called allochthonous models is beyond the scope of this paper, but in
summary I would suggest that there is no convincing
geological argument that requires an allochthonous
origin for these terranes, and on the contrary their
geological anities strongly support a relatively local
origin. In this paper I suggest that the present structural isolation of these terranes from autochthonous
Australia is the result of processes acting after initial
collision of a coherent Australian continent with an
The northeastern quadrant of this region comprises
the Paci®c ocean crustal domain. The Paci®c Plate
proper is separated from eastern Indonesia by two
smaller plates of essentially oceanic crustal type: the
Philippine Sea and Caroline plates (Fig. 1). The Caroline Sea comprises oceanic crust formed between approximately 34±29 Ma (magnetic anomalies 12-10:
Weissel and Anderson, 1978). Relative motion between
the Caroline and Paci®c plates is currently small, and
the distinction of a separate Caroline Plate at the present day is questionable. Whether the Caroline Sea
formed a distinct plate at any time after its formation
is not clear.
The southern part of the Philippine Sea also has
only small motion relative to the Paci®c Plate, but
overall has a very dierent trajectory, rotating about a
pole close to the northern apex of the plate relative to
Eurasia, and about a pole NE of present-day Halmahera relative to the Paci®c (e.g. Ranken et al., 1984;
Seno et al., 1993). The western part of the Philippine
Sea Plate formed in two phases of sea¯oor spreading
during the Palaeogene, whilst the eastern third of the
plate formed in two phases of mid-late Tertiary backarc or interarc spreading forward of the Palau-Kyushu
Ridge (e.g. Hilde and Lee, 1984; Mrozowski and
Hayes, 1979). Hall et al. (1995a,b) and Hall (1996)
have suggested that the development of the Philippine
Sea Plate is intimately linked with the evolution of
eastern Indonesia because they interpret the Halmahera±Bacan±Waigeo region as an integral part of the
Philippine Sea Plate since the Palaeogene. This interpretation will be discussed in more detail subsequently.
The Caroline and Philippine Sea plates are separated
in the south by the divergent Ayu Trough (Fig. 1).
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
According to Weissel and Anderson (1978), the most
likely interpretation of the Ayu Trough is that it developed in two phases of sea¯oor spreading: from 12±6
Ma at a relatively high spreading rate (40 km/m.y. full
spreading rate at the southern end of the trough), and
from 6±0 Ma at a slower rate (8 km/m.y.). The opening of this triangular trough accommodated approximately 208 of relative rotation between the Philippine
Sea and Caroline plates.
3.4. The Philippines
At the present day the Philippine archipelago forms
an obliquely convergent buer zone between the Philippine Sea and Eurasian plates. The main structural
elements are N±S oriented subduction zones east and
west of the archipelago, and a left-lateral wrench system (the Philippine Fault) also with a predominantly
N±S orientation. The eastern trench system (the Philippine Trench) and the Philippine Fault are relatively
young features, apparently only having developed in
the last few million years (e.g. Cardwell et al., 1980;
Rangin et al., 1990; Aurelio et al., 1991). However,
geological interpretations indicate that left-lateral
shear may have been active regionally since the Oligocene (Karig et al., 1986), and westward-directed subduction pre-dating that at the Philippine Trench is
recorded in the Wadati-Benio zone associated with
the Sangihe Arc, which links Mindanao island in the
southern Philippines with northern Sulawesi. This seismic zone is shown by Cardwell et al. (1980) to extend
as far north as the central Philippines.
Palaeomagnetism indicates that much of the Philippines has been translated northward through the Tertiary. For instance McCabe et al. (1985) suggested that
before the Neogene the entire Philippines region was
located in equatorial latitudes, and was translated
northward in the late Palaeogene or early Neogene.
The regional tectonic model of Rangin et al. (1990)
interpreted a similar northward translation of the Philippine arc terranes. However, Fuller et al. (1991) have
cautioned that this may be an over-simplication.
Karig et al. (1986) noted the paucity of accretionary
complexes in the Philippines. Most of the archipelago
is composed of non-continental volcanic arc terranes
ranging in age from Cretaceous through to the present
day. In the central Philippines islands of Panay and
Bicol the main arc activity took place between 50±30
Ma, ceasing in early Oligocene times (Rangin et al.,
1990). Ophiolite terranes are locally present (e.g. the
Zambales and Angat Terranes of Luzon: Karig et al.,
1986), but are of lesser importance regionally than
they are in eastern Indonesia. Continental terranes are
found in the west-central part of the Philippines (the
North Palawan Terrane: e.g. McCabe et al., 1985),
and in SW Mindanao in the south of the archipelago
609
(Pubellier et al., 1991). These are usually interpreted as
fragments of the Eurasian continental margin, but it
will be speculatively suggested in the evolutionary
model presented later that the SW Mindanao continental fragment may have originated from the pre-collisional Australian continental margin.
Most of the complexity of the Philippines island arc
history is beyond the scope of this study. Apart from
SW Mindanao, the outline of which is shown in the
older reconstructions presented subsequently, the
remainder of the arc terranes in the eastern Philippines, from Luzon in the north to eastern Mindanao
in the south, are treated here as a loosely de®ned volcanic arc terrane, the East Philippines Terrane.
4. The Eastern Indonesia Collision Complex
4.1. Southwest, west-central and north Sulawesi and the
Sangihe Arc
Southwest Sulawesi has a transitional boundary with
eastern Sundaland, and originated as part of that continent (Fig. 3). However, palaeomagnetic declination
plots by Panjaitan and Mubroto (1993) suggest that
Southwest Sulawesi has at least locally undergone anticlockwise rotations of up to 808 since the Miocene,
with one Pliocene site rotated 708 anticlockwise. This
probably represents local block rotation associated
with the Walanae left-lateral fault system (Fig. 3)
which was active during the Plio-Pleistocene (Berry
and Grady 1987).
The ®rst signi®cant Tertiary structural event recognised in SW Sulawesi is the phase of Palaeogene extension also widely recognised on the eastern Sunda shelf,
marked by extensional tectonics and arc volcanism,
which, at least locally, commenced as early as the
Palaeocene (Polve et al., 1997). Subsequently, during
the Middle Eocene±Middle Miocene, the western half
of SW Sulawesi (present-day coordinates) formed part
of an extensive eastern Sundaland carbonate platform
(Fig. 3; Wilson and Bosence, 1996). In the eastern half
of the peninsula arc volcanism may have continued
contemporaneously through much of this period (van
Leeuwen, 1981; Priadi et al., 1994). From the Middle
Miocene volcanism was widespread across both halves
of SW Sulawesi, although geochemically this younger
volcanism re¯ects an extensional structural environment rather than typical island arc conditions
(Yuwono et al., 1988).
Further north in western central Sulawesi, following
the Palaeogene rifting event already outlined in the
south, island arc volcanism was established during
Late Eocene and Oligocene times (Priadi et al., 1994,
Polve et al., 1997). Shoshonitic and high-K igneous activity took place from the Middle Miocene or slightly
610
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
earlier through to the Pliocene or Quaternary (18±3
Ma according to Bergman et al., 1996; as young as 0.6
Ma according to Priadi et al., 1994). However R. Hall
(personal communication, 1999) questions the reliability and volumetric signi®cance of the earlier part
of this age range, considering the high-K igneous activity to be younger than 11 Ma, whilst Polve et al.
(1997) dated the peak of this igneous phase to between
13±10 Ma.
Structurally west-central Sulawesi consists of a westward-directed fold and thrust belt with a thrust front
in the Makassar Straits (Fig. 3; Coeld et al., 1993).
Radiometric dating (Bergman et al., 1996) suggests
that this foldbelt developed between about 13±5 Ma.
An earlier phase of ophiolite obduction on the eastern
margin of this province (Lamasi Complex) is dated at
about 21 Ma.
The north arm of Sulawesi forms a similar volcanoplutonic belt. The western end of the north arm has a
basement of Sundaland continental crust, whilst the
eastern and central segments of the arm are built upon
oceanic crust (Fig. 3; Kavalieris et al., 1992). The oldest Tertiary volcanism is of island arc type, and is
dated as Eocene±Oligocene in age. After a volcanic
hiatus with associated tectonism, a second phase of
volcanism is dated between 22±16 Ma (Early Miocene). A third phase of volcanism commenced at about
9 Ma, and continues through to the present-day at the
eastern end of the north arm. The latter volcanism is
associated with the active Sangihe Arc to the north,
which is related to westward consumption of the
Molucca Sea Plate. According to Hamilton (1979), the
Sangihe Arc originated in the early Middle Miocene,
and was particularly active through to the Late Miocene.
Palaeomagnetic studies suggest that the north arm
of Sulawesi has undergone approximately 20±258 of
clockwise rotation since the Miocene (Surmont et al.,
1994). This rotation has been accommodated by
underthrusting of Celebes Sea oceanic crust beneath
the north arm of Sulawesi at the North Sulawesi
Trench (Hamilton, 1979; Silver et al., 1983a). The
hangingwall of the North Sulawesi Trench has a fold
and thrust belt structural style (Neben et al., 1998, ®g.
5) comparable to the fold and thrust belt on the eastern margin of the Makassar Straits. Oset between the
oshore North Sulawesi foldbelt and the east Makassar Straits foldbelt is taken up on the Palu-Koro strike
slip fault (Fig. 3).
4.2. Eastern Sunda-Banda Arc
The eastern Sunda Arc is the segment of the arc east
of the main Sundaland continent up to its intersection
Fig. 4. Bathymetry/location map of the Banda Sea±Banda Arc region. Oshore bathymetry at 1000 m interval with water depths >4000 m
shaded. H.F.: Hamilton Fault, K.F.: Kioko Fault.
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
with the colliding Australian margin at the latitude of
Sumba island (Fig. 4). East of this the feature continues as the Banda Arc, with direct geological continuity between the two, particularly in the volcanic arc.
Although the eastern Sunda Arc separates oceanic
crust of the Indian Ocean from oceanic crust of the
southern Banda Sea, the arc itself is, at least in part,
built on a basement of Sundaland continental crust.
This basement is exposed in Sumba (Chamalaun et al.,
1981) and in the allochthonous (overthrust non-Australian) sequence of Timor (Earle, 1983; AudleyCharles, 1985). The arc has a history of igneous activity through much of the Tertiary, with arc volcanism recognised in the Palaeocene (Masu Volcanics,
Sumba), Eocene (Metan Volcanics, Timor), Early Oligocene (Kur island, western Kai group: Honthaas et
al., 1997), Early Miocene (Jawila Volcanics, Sumba;
volcanics associated with the Noil Toko Formation in
Timor, e.g. de Waard, 1957; Rosidi et al., 1981),
Middle Miocene (ca 12 Ma in Wetar: Abbott and Chamalaun, 1981), Late Miocene (Manamas Volcanics,
Timor: e.g. Bellon, in Linthout et al., 1997) and
younger. Arc volcanism has not been identi®ed during
the Late Oligocene. East of Wetar Island the eastern
Banda volcanic arc is apparently only as old as Late
Miocene or Pliocene (van Bemmelen, 1949; Bowin et
al., 1980), although Kur Island located in the innermost part of the Kai forearc complex includes arc volcanics of Early Oligocene age (Honthaas et al., 1997).
In the forearc ridge to the south of the present volcanic arc, Sumba, located at the transition from oceanic subduction at the Sunda Arc to continental
collision around the Banda Arc, is entirely composed
of non-Australian crustal elements. In Timor and adjacent islands, non-Australian crustal material originating in the pre-collisional forearc complex is thrust onto
the most distal parts of the Australian continental
margin sequence, which is itself strongly imbricated by
thrusting. Further west in the Tanimbar Islands an
allochthonous sequence analogous to that on Timor
and Sumba is volumetrically very reduced or absent;
the imbricated Australian margin succession is by far
predominant (Charlton et al., 1991b). Similarly the
Kai Islands are composed predominantly of Australian-anity crustal material (Charlton et al., 1991a),
although radiometric dating of igneous and metamorphic rocks on Kur Island in the extreme west of
the Kai group suggests an allochthonous origin for
these (Honthaas et al., 1997).
Fortuin et al. (1994, 1997) have identi®ed a phase of
rifting along the axis of the Banda-eastern Sunda volcanic arc in the Middle-Late Miocene which led to the
isolation of Sumba and the Timor allochthon from the
volcanic arc, and to the development of the Savu
Basin. Rapid subsidence of the Savu Basin commenced
at about the Early/Middle Miocene boundary (ca 16
611
Ma: Fortuin et al., 1997), approximately contemporaneously with the main volcanic arc to the north of the
basin becoming inactive (Barberi et al., 1987). For at
least part of the subsequent period (?Middle-Late Miocene), active volcanism was apparently transferred to
the southern margin of the basin (Fortuin et al., 1994).
This period coincides closely with the time of generation (ca 16±9.5 Ma) and emplacement (ca 8 Ma) of
ultama®c suites in Timor, Seram and possibly intervening parts of the eastern Banda Arc (Linthout et al.,
1997). It will be suggested subsequently that these
events are linked, and are related to Middle-Late Miocene southeastward expansion of the Sunda Arc prior
to Late Miocene collision around the Banda Arc.
4.3. South Banda Basin
The South Banda Basin is a fragment of oceanic
crust situated within the curve of the Banda Arc
(Fig. 4). The age of the crust has been variously interpreted as Mesozoic (Lapouille et al., 1985; Lee and
McCabe, 1986), Palaeogene (Barber, 1979) and Neogene (Hamilton, 1979; Hall, 1996; Honthaas et al.,
1997). One of the principal arguments that has been
used in favour of a Mesozoic age is the parallelism
and apparent lateral near-continuity of magnetic
anomalies in the South Banda Basin with the known
Mesozoic anomalies in the Indian Ocean. However,
the apparent continuity is likely to be no more than
coincidental because oceanic crust that formerly lay
north of the Northwest Shelf in the vicinity of presentday Timor is now represented by the Wadati-Benio
zone dipping north from the Sunda-Banda Arc.
Restoring this to its pre-collisional position on the
Australian passive margin indicates that the presentday South Banda Basin cannot have originated closer
to Timor than present-day Sulawesi or Buru, which
greatly reduces the apparent continuity between the
two anomaly sets.
A recent study by Honthaas et al. (1997) strongly
suggests a late Neogene age for the South Banda
Basin, broadly contemporaneous with the North
Banda Basin (see below). Dredging on the northern
¯ank of the basin has yielded fossil-bearing volcaniclastic sediments and radiometrically dated basalts giving consistent ages of 7±3 Ma for the volcanism,
interpreted as the age of sea¯oor spreading in the
South Banda Basin.
4.4. Banda ridges
The North and South Banda Basins are separated
by a series of ENE±WSW trending topographic highs
usually described collectively as the Banda Ridges.
Two main en echelon ridges can be recognised: the
Sinta Ridge to the NW and the Lucipara Ridge to the
612
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
SE. Dredge samples from the Lucipara Ridge (Silver et
al., 1985, Honthaas et al., 1997) comprise a mixture of
continental metamorphic and sedimentary rocks,
together with basic volcanics. Two metamorphic rocks
yielded radiometric ages of about 11 and 22 Ma (Silver
et al. 1985). Dredge samples from the Sinta Ridge (Villeneuve et al., 1994) consist of continental margin carbonate and clastic sediments including Triassic reef
limestones. Both ridges therefore appear to be composed predominantly of continental crust.
4.5. SE Sulawesi-Buton-Tukangbesi
This region, consisting of the southeast arm of Sulawesi together with its topographic extension oshore
to the southeast, is treated here as a single structural
domain because, contrary to a number of alternative
interpretations (e.g. Smith and Silver, 1991; Davidson,
1991), I will suggest that it has formed a structurally
coherent body throughout the Neogene.
In broad outline, the region consists of a continental
margin terrane of Australian anity overthrust by the
East Sulawesi Ophiolite (to be discussed in more detail
in the following section). In the SE of this region is the
Tukang Besi Platform. Little geological and geophysical data is available from this area, but gravity values
suggest a continental terrane (Ali et al., 1996), and the
absence of signi®cant sedimentary section imaged on
available seismic data (Ali et al., 1996) suggests that
the Buton foldbelt does not extend onto the platform,
as previously interpreted by Davidson (1991). The
platform is probably a relatively undeformed fragment
of Australian-anity continental crust (Hamilton,
1979).
Westward the platform passes into a Neogene
molassic basin in front of the Buton fold and thrust
belt. The deeper parts of the foldbelt comprise sedimentary sequences with clear stratigraphic anities to
the Banda forearc islands of Buru, Seram and Timor
(e.g. Smit Sibinga, 1928; Davidson, 1991; Smith and
Silver, 1991), which are are in turn widely interpreted
to be composed for the most part of Australian continental margin sequences. Minor fragments of a formerly more extensive ophiolite complex are locally
preserved in the structurally higher levels of the Buton
foldbelt (Smith and Silver, 1991). In SE Sulawesi the
ophiolite complex is more extensive, where it overlies
Australian-anity continental metamorphic basement
in the west, and overlies or is imbricated with Australian-anity cover sequences towards the east (Surono,
1996).
In Buton the age of collision between the continent
and the ophiolite complex can be dated stratigraphically as Oligocene, based on the youngest age of the
pre-collisional sequence (Upper Eocene-?Lower Oligocene: Smith and Silver, 1991) and the oldest syn- to
post-orogenic sediments (planktonic foraminiferal
zone N3±N4 or latest Oligocene-earliest Miocene in
the Bulu-1 well: P.T. Robertson Utama Indonesia,
personal communication). Davidson (1991) also mentions ages as old as planktonic foraminiferal zone
N3/N4 for the syn- to post-orogenic Tondo Formation in South Buton, whilst Surono (1996) dated
molassic sediments in SE Sulawesi to the Early Miocene, and interpreted a latest Oligocene age of collision between the continental terrane and the
ophiolite belt. This is signi®cantly earlier than the
Middle Miocene initiation of collision interpreted by
Smith and Silver (1991).
Southeast Sulawesi and Buton are cut by a number
of large faults striking NW±SE (the Matano, Lasolo
(or Lawanopo), Kolono, Kolaka, Kioko and Hamilton faults: Figs. 3 and 4). The Matano Fault is a
left-lateral strike-slip fault based on geology (Ahmad,
1977) and seismology (McCarey and Sutardjo,
1982), and it has been generally assumed that the
other faults have similar strike-slip displacements.
However, most of the other faults do not show the
geological characteristics of large strike-slip faults:
there is for instance no clear braiding of fault
strands, or linked oblique transpressional/transtensional folding and faulting. Instead the faults have
been mapped regionally as nearly straight lineaments
with gentle folding subparallel to the fault trend (e.g.
folding of the Tokala and Meluhu formations near
the SE end of the main Lasolo Fault lineament on
the Lasusua±Kendari map sheet: Rusmana et al.,
1993b). These faults have more of the characteristics
of large normal faults. Facies patterns in the Late
Triassic suggest that these major block faults have
existed since at least that time (Charlton, submitted),
and have probably been reactivated several times
since. The relatively young strike-slip oset on the
Matano Fault may be such a reactivation with a
dierent sense of displacement.
The SE arm of Sulawesi is separated from the SW
arm by the Gulf of Bone. Seismic sections across the
gulf (e.g. Guntoro, 1996) suggest an extensional origin for the embayment, and Hamilton (1979) interpreted a Middle Miocene age for the commencement
of extension. The end of extension is marked on the
seismic lines by an unconformity separating normal
faulted and gently folded section below from essentially undeformed strata above. At the head of the
gulf in the Malili geological map sheet (Simandjuntak
et al., 1991a) this unconformity can be traced
onshore, where it corresponds to the base of the
Bonebone and Tomata formations. These formations
are dated as latest Miocene-earliest Pliocene (planktonic foraminiferal zones N17±N18) based on the age
ranges of foraminifera listed by Simandjuntak et al.
(1991a).
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
4.6. Central Sulawesi and East Sulawesi-Banggai-Sula
As with the SE Sulawesi-Buton-Tukangbesi structural domain, East Sulawesi and the Banggai and Sula
islands are here interpreted as having formed an essentially coherent structural block throughout the Neogene. This similarity extends to the general geological
characteristics, with the East Sulawesi Ophiolite thrust
eastwards onto a continental margin terrane of Australian anity. The East Sulawesi Ophiolite is also thrust
westward onto Sundaland-anity basement in the
Poso region of central Sulawesi. As with SE Sulawesi
and Buton, the age of collision between the continental
terrane and the East Sulwesi Ophiolite has been widely
dated to the Middle Miocene (e.g. Garrard et al.,
1988), but again I will suggest that there is evidence
for an earlier onset of collision in the Oligocene. In the
East Arm of Sulawesi there is also a second phase of
fold and thrust belt development in the Pliocene
(Davies, 1990).
The geological evolution of central Sulawesi has
been studied in detail by Parkinson (1991, 1996, etc.).
The Pompangeo metamorphic complex was interpreted
as Sundaland continental margin basement, perhaps
formed by the accretion of a Gondwanan terrane in
the mid Cretaceous (ca 110 Ma) when initial mediumhigh pressure metamorphism developed. During the
Oligocene a second phase of metamorphism developed
under high pressure conditions. Subsequently the East
Sulawesi Ophiolite was obducted onto the Pompangeo
basement complex by westward-directed thrusting.
This obduction may have occurred contemporaneously
with obduction of the Lamasi Complex in southwestern Sulawesi at about 21 Ma (cf. Bergman et al.,
1996).
Separating the base of the East Sulawesi Ophiolite
from the underlying Pompangeo Complex is a tectonic
melange complex, the Peleru Melange. The upper contact of the Peleru Melange Complex with the base of
the ophiolite is overprinted by a metamorphic sole
with an inverted thermal gradient, formed by the
obduction of a hot ultrabasic body onto cooler crustal
material. Parkinson interpeted the Oligocene phase of
deformation in central Sulawesi as the result of obduction of young and hot backarc oceanic crust onto the
Sundaland margin prior to collision with the Banggai
Platform in the Middle Miocene. However, several elements of the sub-ophiolite tectonic complex upon
which the inverted metamorphic gradient is overprinted (speci®cally, the Nanaka, Tetambahu and
probably the Matano Broken Formations of the Peleru
Melange Complex) show stratigraphic anities with
the ``Australian'' sequence of the Banggai Platform. As
the metamorphic sole separating the Peleru Melange
Complex from the structurally overlying East Sulawesi
Ophiolite has been dated radiometrically at about 31
613
Ma (Parkinson, 1996), the Banggai Platform must
have arrived by (or at) that time. This Oligocene age
of suturing between western (Sundaland-anity) and
eastern (Australian-anity) Sulawesi is consistent with
the stratigraphically-derived Oligocene age for the
ophiolite-continent collision in Buton and SE Sulawesi.
The origin of the East Sulawesi Ophiolite and indeed
ophiolites in general is not well understood. Geochemistry suggests an origin in a supra-subduction zone setting, possibly related to the backarc Celebes Sea
(Monnier et al., 1995; also Bergman et al., 1996),
which suggests linkage with the Asian/Paci®c plate
margins. On the other hand, palaeomagnetic studies of
Cretaceous or Palaeogene lavas from the east arm of
Sulawesi indicate a relatively high southerly latitude
(ca 208S), possibly not far north of the Australian continent at that time (Mubroto et al., 1994). Whatever
the precise origin of the East Sulawesi Ophiolite, there
is a common association regionally between ophiolite
complexes and subduction forearcs. In the reconstructions presented later the East Sulawesi Ophiolite is
treated as part of an oceanic forearc complex paired
with the Palaeogene volcanic arc terranes of western
Sulawesi.
The East Sulawesi Ophiolite is separated from the
Banggai-Sula continental fragment by the Tomori
Basin (Davies, 1990; Handiwiria, 1990; Abimanyu,
1990). The basinal succession comprises a lower
sequence of shelf carbonates and clastics ranging in
age from Upper Eocene±Upper Miocene, succeeded by
thick molassic sequences of Pliocene±Recent age.
There is no direct evidence for major collision-related
structural development before the Pliocene, and presumably this region lay some distance in front of the
ophiolite-continent collision front during the OligoMiocene period. The present fold and thrust belt structure on the western ¯ank of the Tomori Basin only
developed during the Pliocene (between 5.2±2.8 Ma
according to Davies, 1990).
The Banggai-Sula continental fragment (Pigram et
al., 1985; Garrard et al., 1988) has long been recognised as stratigraphically related to the continental
part of New Guinea island, and hence to the Australian continent (e.g. KlompeÂ, 1954; Visser and Hermes,
1962). These latter authors suggested stratigraphic
similarity with the Bird's Head structural block of western New Guinea, whilst Pigram et al. (1985) argued
for a more distant origin, adjacent to central Papua
New Guinea. Elsewhere (Charlton, 1996) I have
reviewed the evidence which leads me to favour a connection with the Bird's Head block.
The east arm of Sulawesi is separated from the
north arm by the Gulf of Tomini (Fig. 5). An extensional origin for this gulf, as with the Gulf of Bone,
separating the SE and SW arms of Sulawesi can be
inferred, e.g. from regional seismic lines and gravity
614
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Fig. 5. Bathymetry/location map of northeastern Indonesia. Oshore bathymetry at 1000 m interval with water depths >4000 m shaded.
modelling (Silver et al., 1983a). A possible indication
of a Middle Miocene or earlier onset of extension is
given by the Middle Miocene (and younger) Bongka
Formation (Poso map sheet: Simandjuntak et al.,
1991b), which is a deepwater turbiditic sequence
unconformably overlying the central Sulawesi collision
complex. Again latest Miocene±earliest Pliocene formations that unconformably overlie older strata (e.g.
the Lonsio and Kintom formations of the Luwuk map
sheet: Rusmana et al., 1993a) may mark the cessation
of extension in the gulf.
4.7. Seram and Buru
Seram Island has been described as a mirror image
across the Banda Sea of Timor in the south (AudleyCharles et al., 1979). It consists of a northward-directed fold and thrust belt forming the forearc complex
of the northern Banda Arc. However, unlike Timor
but like Tanimbar and Kai, there is no major `Asian'
allochthon within the Seram collision complex. [N.B.
this interpretation is in contrast to the original interpretation of Audley-Charles et al. (1979) who
inferred an allochthonous origin for a major part of
the Seram succession based on similarities with the
Timor allochthon as then recognised, particularly with
the Maubisse Formation of Timor. An Australian and
therefore parautochthonous origin for the Maubisse
Formation is now widely accepted (e.g. Audley-Charles
and Harris, 1990), and the necessity for an extensive
allochthon in Seram is negated]. As with Tanimbar
and the eastern Banda Arc, the volcanic arc in the hin-
terland of Seram (Ambon and adjacent islands) is
essentially Pliocene and younger in age.
Buru Island is also usually considered to be one of
the islands in the Banda forearc chain. However,
although there is close stratigraphic similarity between
Seram and Buru, the two islands show very dierent
structural styles. Whilst Seram consists of an imbricate
stack of thrust sheets, Buru has a relatively simple
anticlinorial structure with the principal fold axis following the long axis of the island. It is likely that the
Buru anticlinorium marks a westward dying out of
Banda forearc deformation, and thus the island forms
a pin-point termination for this convergent system.
Seram and Buru islands are separated by a small triangular marine embayment (Fig. 5) which I interpret
as a triangular pull-apart structure (sphenochasm).
This structure osets the Pliocene volcanic island of
Ambelau south of Buru from the Ambon group south
of Seram, which suggests that it opened during Late
Pliocene±Recent times. This sphenochasm is interpreted in the later reconstructions to have accommodated 458 of late-stage clockwise rotation between
Buru and western Seram.
4.8. Bird's Head-Misool and the Sorong Fault Zone
The Bird's Head-Misool block south of the Sorong Fault and west of the Ransiki Fault (Fig. 5)
forms an essentially coherent and little deformed
structural domain of Australian continental anity.
The region has strong stratigraphic links both to
cratonic and parautochthonous southern New Guinea to the east (Dow and Sukamto, 1986) and to
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
the Banda Arc region to the southwest. There is
thus no strong reason to suspect on stratigraphic
grounds that the Bird's Head structural block is
allochthonous, as has been proposed by Pigram and
co-workers (see above). Neither is there any strong
stratigraphic or structural evidence to support the
contention (Pigram and Panggabean, 1984; Pigram
and Davies, 1987) that Misool together with the
Onin and Kumawa peninsulas of the Bird's Head
formed a separate terrane independent of the main
Bird's Head block prior to the Oligocene. On the
contrary, seismic data (e.g. Perkins and Livsey,
1993, Fig. 2) strongly suggests simple tectonostratigraphic continuity from the Misool-Onin-Kumawa
Ridge into the main part of the Bird's Head block
before the inversion that formed the ridge in the
Pliocene, the inversion being related to the collisional development of the northern Banda Arc.
The Bird's Head block was aected by an important
phase of deformation during the Oligocene, most notably giving rise to the NW±SE trending Central Bird's
Head (Vogelkop) Monocline (Visser and Hermes,
1962). Resultant uplift and erosion of the Kemum
basement block to the north of the monocline led to
the shedding of an extensive clastic sequence (the
Tertiary evolution of the Eastern Indonesia Collision Complex
T.R. Charlton
Ridge House, 1 St. Omer Ridge, Guildford, Surrey, GU1 2DD, UK
Received 14 January 1999; accepted 6 July 1999
Abstract
Eastern Indonesia is the zone of interaction between three converging megaplates: Eurasia, the Paci®c and Indo-Australia.
The geological basis for interpretations of the Tertiary tectonic evolution of Eastern Indonesia is reviewed, and a series of plate
tectonic reconstructions for this region at 5 million year intervals covering the last 35 million years is presented.
The oldest reconstruction predates the onset of regional collisional deformation. At this time a simple plate con®guration is
interpreted, consisting of the northward-moving Australian continent approaching an approximately E±W oriented, southwardfacing subduction zone extending from the southern margin of the Eurasian continent eastwards into the Paci®c oceanic
domain. Beginning at about 30 Ma the Australian continental margin commenced collision with the subduction zone along its
entire palinspastically-restored northern margin, from Sulawesi in the west to Papua New Guinea in the east. From this time
until ca 24 Ma, the Australian continent indented the former arc trend, with the northward convergence of Australia absorbed
at the palaeo-northern boundary of the Philippine Sea Plate (the present-day Palau-Kyushu Ridge).
At ca 24 Ma the present-day pattern of oblique convergence between the northern margin of Australia and the Philippine Sea
Plate began to develop. At about this time a large portion of the Palaeogene colliding volcanic arc (the future eastern
Philippines) began to detach from the northern continental margin by left-lateral strike slip. From ca 18 Ma oblique southwarddirected subduction commenced at the Maramuni Arc in northern New Guinea. At ca 12 Ma the Sorong Fault Zone strike-slip
system developed, eectively separating the Philippines from the Indonesian tectonic domain. The Sorong Fault Zone became
inactive at ca 6 Ma, since which time the tectonics of eastern Indonesia has been dominated by the anticlockwise rotation of the
Bird's Head structural block by some 30±408.
Contemporaneously with post-18 Ma tectonism, the Banda Arc subduction±collision system developed o the northwestern
margin of the Australian continent. Convergence between Indo-Australia and Eurasia was accommodated initially by northward
subduction of the Indian Ocean, and subsequently, since ca 8 Ma, by the development of a second phase of arc-continent
collision around the former passive continental margin of NW Australia. 7 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction
Eastern Indonesia, situated at the intersection of the
Alpine-Himalayan and Circum-Paci®c orogenic belts,
is the location of the Earth's fundamental convergent
triple junction. Long-term interaction of these three
plates (taking the de®nition of plates at the coarsest
scale) has resulted in the present-day situation where a
complex arrangement of platelets with poorly de®ned
plate boundaries cover an area equivalent to most of
E-mail address: [email protected] (T.R. Charlton).
western Europe or the western United States. It is my
contention, however, that this complexity is only a
relatively recent development, and that back in time,
through the Tertiary, the tectonic situation rapidly
simpli®es, so that by the mid-Palaeogene the fragmented platelets have resolved themselves into a relatively
simple pattern of coherent plates. Starting from this
simple pre-Neogene con®guration, it is the aim of this
paper to generate the complexity of the present-day by
a few fairly simple changes in regional dynamics, with
these changes largely explicable in terms of plate
boundary interactions taking place within the evolving
eastern Indonesia region.
1367-9120/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 7 - 9 1 2 0 ( 9 9 ) 0 0 0 4 9 - 8
604
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Fig. 1. Regional tectonic setting of the eastern Indonesia collision zone. Inset: approximate convergence vectors for the three main plates Eurasia
(EU), Paci®c (PA) and Indo-Australia (AU). M.S.C.Z.: Molucca Sea collision zone. N.B.B.: North Banda Basin. S.B.B.: South Banda Basin.
The region under consideration comprises the eastern half of Indonesia from Sulawesi island in the west
to the Papua New Guinea border in the east (Fig. 1).
Peripheral regions beyond these limits are, quite deliberately, only treated in the most general terms. The
reason for this exclusive focus on eastern Indonesia is
that, as will be discussed in the following section, there
is no clear consensus on how the larger region viewed
at a sub-global scale has evolved through the Tertiary.
2. External constraints on the collision model
The Eastern Indonesia Collision Complex involves
the interaction of three or four major plates: Indo-
Australia, Eurasia and the Paci®c (or Caroline and
Philippine Sea) plates. The gross kinematics of these
plates (Fig. 1) provides an important framework
within which to interpret the evolution of this region.
In particular, the relative movements of the two main
continental fragments, Australia and Southeast Asia,
should provide a good constraint on the evolution of
the collision complex. However, this constraint is not
as strong as could be desired because the two continents form the end-points in a long global plate
motion circuit (Southeast Asia±Eurasia±Africa±Antarctica±Australia), and uncertainties at each stage in
the circuit are compounded in the ®nal result.
It is commonly stated in the literature (e.g. AudleyCharles et al., 1988; Packham, 1990; Rangin et al.,
Fig. 2. Palaeolatitudinal change for a nominal point on the Australian continent (108S, 1248E, currently located in southern West Timor). The individual curves are linear interpolations between measured points (the dots) from published palaeogeographic maps. All apart from the Rangin
et al. (1990) curve are measured directly from the position of Timor on the maps; the Rangin et al. (1990) curve is extrapolated from the
southern tip of the Aru Islands (ÿ38). In some of the maps for older time intervals, extrapolation was necessary from another point on the Australian Plate as the principal reference point had moved o the southern edge of the maps. The two curves from Soeding et al. (1997) are generated from dierent reference parameters: the palaeomagnetically-derived curve [PM] is based on the North America reference frame (Harrison
and Lindh 1982); the hotspot [HS] curve uses the Mueller et al. (1993) hotspot reference frame. The grey shaded area is derived from the rotation
poles of Australia relative to Antarctica as calculated by Royer and Sandwell (1989). The lower limit of the shading assumes that Antarctica has
remained latitudinally ®xed, whilst the upper limit assumes that the ridge system south of Australia has remained ®xed. Regions below the `static
Antarctica' curve would require Antarctica to be moving northward towards SE Asia, whilst the region above would require Antarctica to be
moving away from SE Asia. Together these bounds may provide long-term limits on the likely northward movement of Australia. The lower
graph shows the latitudinal uncertainty in km of the eleven map-derived curves at 5 Ma intervals, as given by the dierence between the highest
and lowest palaeolatitudinal estimate. The dashed line is the same data, but excluding the Dercourt et al. (1991) curve. The palaeolatitudes from
the present study are not shown, but are virtually indistinguishable from the Paleomap curve since 25 Ma.
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Fig. 2.
605
606
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
1990; Hall, 1996) that the position of the Australian
continent through time is reasonably well constrained.
This, however, is not strongly supported by Fig. 2
which shows the palaeolatitude of a point on the Australian continental margin (108S, 1248E, currently
located in the Timor collision zone) through the Tertiary, as interpreted by these authors, together with a
number of other published reconstructions. This shows
that interpreted mean rates of northward movement
for Australia during the last 30 Ma vary by a factor of
two from about 40±80 km/m.y. Only ®fteen million
years ago there was as much as 880 km (88) dierence
in interpreted palaeolatitude between the extremes,
which is close to the present-day north±south spread
of the entire eastern Indonesia collision complex. Even
excluding the Dercourt et al. (1993) track, which is
clearly out of line with the others, only reduces the
uncertainty to ca 700 km at 30 Ma (dashed line in the
lower part of Fig. 2). This allows an uncertainty of up
to ca 9 Ma for the time at which a particular point on
the northern Australian continental margin reached a
particular latitude, given the current `absolute' northward motion of Australia at ca 75 km/m.y. (see
below).
The faster rates in Fig. 2 are associated with models
that utilise a hot spot reference frame, whilst the
models with a slower northward motion use a palaeomagnetically-derived reference framework. This is illustrated in Fig. 2 by the reconstructions generated in the
Soeding et al. (1997) program, which allows calculation of continental palaeopositions in both reference
frames. Thus the latitudinal spread in Fig. 2 is principally a result of the methodology used in determining
the palaeolatitudes (hotspot or palaeomagnetic frameworks). This explains the spread, but does not remove
the problem for the palaeomagnetically-derived reconstructions that the rate of northward movement of
Australia has apparently slowed considerably since 10
Ma relative to latitude. Reconstructions in the hotspot
reference framework generally do not show this apparent slowing.
Present-day plate motion calculations suggest relatively high convergence rates between Australia and
Eurasia. The plate motion calculations of Minster and
Jordan (1978), which are determined in the hotspot
reference frame, indicate that the reference point (108S,
1248E) is moving ca 198 east of north at a rate of
about 79.5 km/m.y. (i.e., a northward velocity of ca
75 km/m.y.). In the NUVEL-1A model of de Mets et
al. (1990, 1994) the relative convergence of Australia
with Eurasia is 79.8 km/m.y. bearing 0168 Ð i.e. a
northward velocity of 76.5 km/m.y. assuming Eurasia
to be stationary. These ®gures are at the top of the
range of long-term northward displacements for Australia shown in Fig. 2, in line with long-term northward displacements in the hotspot reference frame, but
considerably faster than those constructed in the
palaeomagnetic framework.
The palaeoposition of Southeast Asia is also less
than adequately constrained. Taking Java island as a
representative element of this platelet, there is signi®cant disagreement in the published literature as to its
movement history during the Tertiary. Thus Rangin et
al. (1990) and Daly et al. (1991) inferred clockwise and
overall westward displacement of Java through the
Tertiary, whilst Hall (1996) interpreted overall anticlockwise and eastward displacement. Similar disagreement is shown in the palaeomagnetic literature with
regard to the larger region, with many authors proposing large anticlockwise rotations of Southeast Asia
during the Tertiary (e.g. Fuller et al., 1991), whilst
others have concluded little or no overall rotation
(Lumadyo et al., 1993). However, it should be added
that most palaeomagnetic studies ®nd only small
changes in latitude for Southeast Asia through the
Tertiary.
In summary, whilst the globally determined palaeocontinental positions and instantaneous plate motion
calculations provide important external constraints on
the tectonic evolution of eastern Indonesia, they cannot at present provide more than a very generalised
framework. Tying an evolutionary model too closely
to any particular global or very-large-regional tectonic
model is probably not sensible at this stage because
there is such a wide variation between alternatives.
The model developed here assumes fairly high relative velocities between Australia and Southeast Asia in
line with the hot spot reference frame, principally
because this seems to ®t best with inferences on the
timing of collision arising from interpretation of eastern Indonesia geology. A constant northward velocity
is assumed here for Australia since the Eocene because
most calculations of the motion of Australia based on
hot spot traces (e.g. Duncan, 1981; Wellman, 1983) or
relative to Antarctica (Royer and Sandwell, 1989;
Veevers et al., 1991) do not recognise major changes in
the rate of northward displacement through this
period. In terms of available models, the chosen path
of Australia most closely follows the Paleomap reconstructions of Scotese and co-workers (see Scotese,
1999).
3. Regional structural elements
Eastern Indonesia forms the zone of interaction
between the Eurasian/Southeast Asian and Australian
continents, and the Paci®c and its sub-plates (the
Philippine Sea and Caroline plates: Fig. 1). In the following two sections the principal structural elements of
this region will be outlined. The main bounding elements are described in this section, followed in Sec-
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
tion 4 by the smaller crustal elements within the collision zone.
3.1. Southeast Asia
In the context of eastern Indonesia, the Southeast
Asian Plate (a sub-element of the Eurasian Plate) consists of the Sundaland continental block and the Celebes Sea oceanic fragment (Fig. 1). Eastern Sundaland
comprises Java and Borneo islands and the intervening
shelfal Java Sea. To the south, Sundaland is bounded
607
by the Sunda Arc subduction system. In the southeast
the margin of Sundaland is transitional into the Southwest Sulawesi structural province discussed later. This
transitional region is characterised structurally by
Palaeogene rift grabens inverted by Miocene compression, which has produced a distinctive cross-sectional expression of `Sunda folds' (e.g. Letouzey et al.,
1990). In the eastern Sundaland margin these inversion
structures predominantly trend E±W, but near to Sulawesi the trend changes markedly to nearly N±S
(Fig. 3).
Fig. 3. Principal structural elements of eastern Sundaland and western Sulawesi, and the main structural lineaments of SE Sulawesi. Adapted
from Letouzey et al. (1990), Kavalieris et al. (1992), Bergman et al. (1996) and Parkinson (1996). Bathymetry at 1000 m intervals.
608
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Further north the Makassar Straits between Borneo
and Sulawesi probably developed initially from the
same Palaeogene phase of extension recognised on the
eastern Sundaland margin, although the degree of
extension in the Makassar Straits was considerably
greater. The South Makassar Basin (Fig. 3) is thought
to be underlain by continental crust strongly attenuated by Palaeogene extension (Situmorang, 1982),
whilst the North Makassar Basin may be underlain by
Palaeogene oceanic crust, as is found in the Celebes
Sea to the north (Weissel, 1980; Smith et al., 1990).
The eastern and southern margins of the Makassar
Straits and Celebes Sea respectively comprise fold and
thrust belts (described in more detail later) associated
with the Neogene westward displacement of Sulawesi.
Consequently the Makassar Straits additionally possess
some facets of a foreland basin superimposed on the
older rifted margin structure (Bergman et al., 1996).
island arc system, rather than a pre-collisional disaggregation of the Australian margin as implied by the
allochthonous terrane models.
Immediately to the west of the Australian continent
is Mesozoic oceanic crust of the Indian Ocean. The
western continental margin of Australia was formed
by rifting phases in the Late Jurassic and Early Cretaceous, with extension directions implied by ocean ¯oor
magnetic lineation patterns oriented between about
WNW±ESE and NW±SE. This oceanic crust is subducting northward beneath the southern and eastern
Sunda Arc (from southern Sumatra eastwards), whilst
the Australian continental margin is in collision with
the Banda Arc system, which is the direct eastward
extension of the Sunda Arc. Further west still, the
Indian Ocean is underlain by Cenozoic oceanic crust,
which is being subducted at the SW continental margin
of Sundaland beneath most of Sumatra.
3.2. Australia
3.3. The Paci®c
The northern part of the Australian continental
block comprises Australia, the Northwest and Arafura
shelf seas, and the southern half of New Guinea island
(Irian Jaya province of Indonesia and Papua New Guinea: Fig. 1). The boundary between autochthonous
Australian crust essentially undeformed during the
Tertiary and the collision complex to the north is
marked by the Timor-Tanimbar Trough thrust front,
the extensional front of the Aru Trough, the TareraAiduna strike-slip fault, and the Central RangesPapuan Foldbelt deformation front. Immediately
north of this orogenic front is deformed Australian
crustal material (the parautochthonous zone) and then
crustal elements allochthonous to the Australian margin. Many parts of the parautochthonous zone are
linked directly to the autochthon to the south, and can
be restored by standard palinspastic techniques where
sucient data is available. The precise point of origin
of other `Australian' crustal elements (e.g. the Bird's
Head and Banggai-Sula crustal fragments to be discussed below) is more questionable, and some authors
(Pigram and Panggabean, 1984; Pigram and Davies,
1987; Struckmeyer et al., 1993, etc.) have suggested
that these terranes have had long and complex kinematic histories independent of the main Australian
Plate. Detailed discussion of these so-called allochthonous models is beyond the scope of this paper, but in
summary I would suggest that there is no convincing
geological argument that requires an allochthonous
origin for these terranes, and on the contrary their
geological anities strongly support a relatively local
origin. In this paper I suggest that the present structural isolation of these terranes from autochthonous
Australia is the result of processes acting after initial
collision of a coherent Australian continent with an
The northeastern quadrant of this region comprises
the Paci®c ocean crustal domain. The Paci®c Plate
proper is separated from eastern Indonesia by two
smaller plates of essentially oceanic crustal type: the
Philippine Sea and Caroline plates (Fig. 1). The Caroline Sea comprises oceanic crust formed between approximately 34±29 Ma (magnetic anomalies 12-10:
Weissel and Anderson, 1978). Relative motion between
the Caroline and Paci®c plates is currently small, and
the distinction of a separate Caroline Plate at the present day is questionable. Whether the Caroline Sea
formed a distinct plate at any time after its formation
is not clear.
The southern part of the Philippine Sea also has
only small motion relative to the Paci®c Plate, but
overall has a very dierent trajectory, rotating about a
pole close to the northern apex of the plate relative to
Eurasia, and about a pole NE of present-day Halmahera relative to the Paci®c (e.g. Ranken et al., 1984;
Seno et al., 1993). The western part of the Philippine
Sea Plate formed in two phases of sea¯oor spreading
during the Palaeogene, whilst the eastern third of the
plate formed in two phases of mid-late Tertiary backarc or interarc spreading forward of the Palau-Kyushu
Ridge (e.g. Hilde and Lee, 1984; Mrozowski and
Hayes, 1979). Hall et al. (1995a,b) and Hall (1996)
have suggested that the development of the Philippine
Sea Plate is intimately linked with the evolution of
eastern Indonesia because they interpret the Halmahera±Bacan±Waigeo region as an integral part of the
Philippine Sea Plate since the Palaeogene. This interpretation will be discussed in more detail subsequently.
The Caroline and Philippine Sea plates are separated
in the south by the divergent Ayu Trough (Fig. 1).
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
According to Weissel and Anderson (1978), the most
likely interpretation of the Ayu Trough is that it developed in two phases of sea¯oor spreading: from 12±6
Ma at a relatively high spreading rate (40 km/m.y. full
spreading rate at the southern end of the trough), and
from 6±0 Ma at a slower rate (8 km/m.y.). The opening of this triangular trough accommodated approximately 208 of relative rotation between the Philippine
Sea and Caroline plates.
3.4. The Philippines
At the present day the Philippine archipelago forms
an obliquely convergent buer zone between the Philippine Sea and Eurasian plates. The main structural
elements are N±S oriented subduction zones east and
west of the archipelago, and a left-lateral wrench system (the Philippine Fault) also with a predominantly
N±S orientation. The eastern trench system (the Philippine Trench) and the Philippine Fault are relatively
young features, apparently only having developed in
the last few million years (e.g. Cardwell et al., 1980;
Rangin et al., 1990; Aurelio et al., 1991). However,
geological interpretations indicate that left-lateral
shear may have been active regionally since the Oligocene (Karig et al., 1986), and westward-directed subduction pre-dating that at the Philippine Trench is
recorded in the Wadati-Benio zone associated with
the Sangihe Arc, which links Mindanao island in the
southern Philippines with northern Sulawesi. This seismic zone is shown by Cardwell et al. (1980) to extend
as far north as the central Philippines.
Palaeomagnetism indicates that much of the Philippines has been translated northward through the Tertiary. For instance McCabe et al. (1985) suggested that
before the Neogene the entire Philippines region was
located in equatorial latitudes, and was translated
northward in the late Palaeogene or early Neogene.
The regional tectonic model of Rangin et al. (1990)
interpreted a similar northward translation of the Philippine arc terranes. However, Fuller et al. (1991) have
cautioned that this may be an over-simplication.
Karig et al. (1986) noted the paucity of accretionary
complexes in the Philippines. Most of the archipelago
is composed of non-continental volcanic arc terranes
ranging in age from Cretaceous through to the present
day. In the central Philippines islands of Panay and
Bicol the main arc activity took place between 50±30
Ma, ceasing in early Oligocene times (Rangin et al.,
1990). Ophiolite terranes are locally present (e.g. the
Zambales and Angat Terranes of Luzon: Karig et al.,
1986), but are of lesser importance regionally than
they are in eastern Indonesia. Continental terranes are
found in the west-central part of the Philippines (the
North Palawan Terrane: e.g. McCabe et al., 1985),
and in SW Mindanao in the south of the archipelago
609
(Pubellier et al., 1991). These are usually interpreted as
fragments of the Eurasian continental margin, but it
will be speculatively suggested in the evolutionary
model presented later that the SW Mindanao continental fragment may have originated from the pre-collisional Australian continental margin.
Most of the complexity of the Philippines island arc
history is beyond the scope of this study. Apart from
SW Mindanao, the outline of which is shown in the
older reconstructions presented subsequently, the
remainder of the arc terranes in the eastern Philippines, from Luzon in the north to eastern Mindanao
in the south, are treated here as a loosely de®ned volcanic arc terrane, the East Philippines Terrane.
4. The Eastern Indonesia Collision Complex
4.1. Southwest, west-central and north Sulawesi and the
Sangihe Arc
Southwest Sulawesi has a transitional boundary with
eastern Sundaland, and originated as part of that continent (Fig. 3). However, palaeomagnetic declination
plots by Panjaitan and Mubroto (1993) suggest that
Southwest Sulawesi has at least locally undergone anticlockwise rotations of up to 808 since the Miocene,
with one Pliocene site rotated 708 anticlockwise. This
probably represents local block rotation associated
with the Walanae left-lateral fault system (Fig. 3)
which was active during the Plio-Pleistocene (Berry
and Grady 1987).
The ®rst signi®cant Tertiary structural event recognised in SW Sulawesi is the phase of Palaeogene extension also widely recognised on the eastern Sunda shelf,
marked by extensional tectonics and arc volcanism,
which, at least locally, commenced as early as the
Palaeocene (Polve et al., 1997). Subsequently, during
the Middle Eocene±Middle Miocene, the western half
of SW Sulawesi (present-day coordinates) formed part
of an extensive eastern Sundaland carbonate platform
(Fig. 3; Wilson and Bosence, 1996). In the eastern half
of the peninsula arc volcanism may have continued
contemporaneously through much of this period (van
Leeuwen, 1981; Priadi et al., 1994). From the Middle
Miocene volcanism was widespread across both halves
of SW Sulawesi, although geochemically this younger
volcanism re¯ects an extensional structural environment rather than typical island arc conditions
(Yuwono et al., 1988).
Further north in western central Sulawesi, following
the Palaeogene rifting event already outlined in the
south, island arc volcanism was established during
Late Eocene and Oligocene times (Priadi et al., 1994,
Polve et al., 1997). Shoshonitic and high-K igneous activity took place from the Middle Miocene or slightly
610
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
earlier through to the Pliocene or Quaternary (18±3
Ma according to Bergman et al., 1996; as young as 0.6
Ma according to Priadi et al., 1994). However R. Hall
(personal communication, 1999) questions the reliability and volumetric signi®cance of the earlier part
of this age range, considering the high-K igneous activity to be younger than 11 Ma, whilst Polve et al.
(1997) dated the peak of this igneous phase to between
13±10 Ma.
Structurally west-central Sulawesi consists of a westward-directed fold and thrust belt with a thrust front
in the Makassar Straits (Fig. 3; Coeld et al., 1993).
Radiometric dating (Bergman et al., 1996) suggests
that this foldbelt developed between about 13±5 Ma.
An earlier phase of ophiolite obduction on the eastern
margin of this province (Lamasi Complex) is dated at
about 21 Ma.
The north arm of Sulawesi forms a similar volcanoplutonic belt. The western end of the north arm has a
basement of Sundaland continental crust, whilst the
eastern and central segments of the arm are built upon
oceanic crust (Fig. 3; Kavalieris et al., 1992). The oldest Tertiary volcanism is of island arc type, and is
dated as Eocene±Oligocene in age. After a volcanic
hiatus with associated tectonism, a second phase of
volcanism is dated between 22±16 Ma (Early Miocene). A third phase of volcanism commenced at about
9 Ma, and continues through to the present-day at the
eastern end of the north arm. The latter volcanism is
associated with the active Sangihe Arc to the north,
which is related to westward consumption of the
Molucca Sea Plate. According to Hamilton (1979), the
Sangihe Arc originated in the early Middle Miocene,
and was particularly active through to the Late Miocene.
Palaeomagnetic studies suggest that the north arm
of Sulawesi has undergone approximately 20±258 of
clockwise rotation since the Miocene (Surmont et al.,
1994). This rotation has been accommodated by
underthrusting of Celebes Sea oceanic crust beneath
the north arm of Sulawesi at the North Sulawesi
Trench (Hamilton, 1979; Silver et al., 1983a). The
hangingwall of the North Sulawesi Trench has a fold
and thrust belt structural style (Neben et al., 1998, ®g.
5) comparable to the fold and thrust belt on the eastern margin of the Makassar Straits. Oset between the
oshore North Sulawesi foldbelt and the east Makassar Straits foldbelt is taken up on the Palu-Koro strike
slip fault (Fig. 3).
4.2. Eastern Sunda-Banda Arc
The eastern Sunda Arc is the segment of the arc east
of the main Sundaland continent up to its intersection
Fig. 4. Bathymetry/location map of the Banda Sea±Banda Arc region. Oshore bathymetry at 1000 m interval with water depths >4000 m
shaded. H.F.: Hamilton Fault, K.F.: Kioko Fault.
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
with the colliding Australian margin at the latitude of
Sumba island (Fig. 4). East of this the feature continues as the Banda Arc, with direct geological continuity between the two, particularly in the volcanic arc.
Although the eastern Sunda Arc separates oceanic
crust of the Indian Ocean from oceanic crust of the
southern Banda Sea, the arc itself is, at least in part,
built on a basement of Sundaland continental crust.
This basement is exposed in Sumba (Chamalaun et al.,
1981) and in the allochthonous (overthrust non-Australian) sequence of Timor (Earle, 1983; AudleyCharles, 1985). The arc has a history of igneous activity through much of the Tertiary, with arc volcanism recognised in the Palaeocene (Masu Volcanics,
Sumba), Eocene (Metan Volcanics, Timor), Early Oligocene (Kur island, western Kai group: Honthaas et
al., 1997), Early Miocene (Jawila Volcanics, Sumba;
volcanics associated with the Noil Toko Formation in
Timor, e.g. de Waard, 1957; Rosidi et al., 1981),
Middle Miocene (ca 12 Ma in Wetar: Abbott and Chamalaun, 1981), Late Miocene (Manamas Volcanics,
Timor: e.g. Bellon, in Linthout et al., 1997) and
younger. Arc volcanism has not been identi®ed during
the Late Oligocene. East of Wetar Island the eastern
Banda volcanic arc is apparently only as old as Late
Miocene or Pliocene (van Bemmelen, 1949; Bowin et
al., 1980), although Kur Island located in the innermost part of the Kai forearc complex includes arc volcanics of Early Oligocene age (Honthaas et al., 1997).
In the forearc ridge to the south of the present volcanic arc, Sumba, located at the transition from oceanic subduction at the Sunda Arc to continental
collision around the Banda Arc, is entirely composed
of non-Australian crustal elements. In Timor and adjacent islands, non-Australian crustal material originating in the pre-collisional forearc complex is thrust onto
the most distal parts of the Australian continental
margin sequence, which is itself strongly imbricated by
thrusting. Further west in the Tanimbar Islands an
allochthonous sequence analogous to that on Timor
and Sumba is volumetrically very reduced or absent;
the imbricated Australian margin succession is by far
predominant (Charlton et al., 1991b). Similarly the
Kai Islands are composed predominantly of Australian-anity crustal material (Charlton et al., 1991a),
although radiometric dating of igneous and metamorphic rocks on Kur Island in the extreme west of
the Kai group suggests an allochthonous origin for
these (Honthaas et al., 1997).
Fortuin et al. (1994, 1997) have identi®ed a phase of
rifting along the axis of the Banda-eastern Sunda volcanic arc in the Middle-Late Miocene which led to the
isolation of Sumba and the Timor allochthon from the
volcanic arc, and to the development of the Savu
Basin. Rapid subsidence of the Savu Basin commenced
at about the Early/Middle Miocene boundary (ca 16
611
Ma: Fortuin et al., 1997), approximately contemporaneously with the main volcanic arc to the north of the
basin becoming inactive (Barberi et al., 1987). For at
least part of the subsequent period (?Middle-Late Miocene), active volcanism was apparently transferred to
the southern margin of the basin (Fortuin et al., 1994).
This period coincides closely with the time of generation (ca 16±9.5 Ma) and emplacement (ca 8 Ma) of
ultama®c suites in Timor, Seram and possibly intervening parts of the eastern Banda Arc (Linthout et al.,
1997). It will be suggested subsequently that these
events are linked, and are related to Middle-Late Miocene southeastward expansion of the Sunda Arc prior
to Late Miocene collision around the Banda Arc.
4.3. South Banda Basin
The South Banda Basin is a fragment of oceanic
crust situated within the curve of the Banda Arc
(Fig. 4). The age of the crust has been variously interpreted as Mesozoic (Lapouille et al., 1985; Lee and
McCabe, 1986), Palaeogene (Barber, 1979) and Neogene (Hamilton, 1979; Hall, 1996; Honthaas et al.,
1997). One of the principal arguments that has been
used in favour of a Mesozoic age is the parallelism
and apparent lateral near-continuity of magnetic
anomalies in the South Banda Basin with the known
Mesozoic anomalies in the Indian Ocean. However,
the apparent continuity is likely to be no more than
coincidental because oceanic crust that formerly lay
north of the Northwest Shelf in the vicinity of presentday Timor is now represented by the Wadati-Benio
zone dipping north from the Sunda-Banda Arc.
Restoring this to its pre-collisional position on the
Australian passive margin indicates that the presentday South Banda Basin cannot have originated closer
to Timor than present-day Sulawesi or Buru, which
greatly reduces the apparent continuity between the
two anomaly sets.
A recent study by Honthaas et al. (1997) strongly
suggests a late Neogene age for the South Banda
Basin, broadly contemporaneous with the North
Banda Basin (see below). Dredging on the northern
¯ank of the basin has yielded fossil-bearing volcaniclastic sediments and radiometrically dated basalts giving consistent ages of 7±3 Ma for the volcanism,
interpreted as the age of sea¯oor spreading in the
South Banda Basin.
4.4. Banda ridges
The North and South Banda Basins are separated
by a series of ENE±WSW trending topographic highs
usually described collectively as the Banda Ridges.
Two main en echelon ridges can be recognised: the
Sinta Ridge to the NW and the Lucipara Ridge to the
612
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
SE. Dredge samples from the Lucipara Ridge (Silver et
al., 1985, Honthaas et al., 1997) comprise a mixture of
continental metamorphic and sedimentary rocks,
together with basic volcanics. Two metamorphic rocks
yielded radiometric ages of about 11 and 22 Ma (Silver
et al. 1985). Dredge samples from the Sinta Ridge (Villeneuve et al., 1994) consist of continental margin carbonate and clastic sediments including Triassic reef
limestones. Both ridges therefore appear to be composed predominantly of continental crust.
4.5. SE Sulawesi-Buton-Tukangbesi
This region, consisting of the southeast arm of Sulawesi together with its topographic extension oshore
to the southeast, is treated here as a single structural
domain because, contrary to a number of alternative
interpretations (e.g. Smith and Silver, 1991; Davidson,
1991), I will suggest that it has formed a structurally
coherent body throughout the Neogene.
In broad outline, the region consists of a continental
margin terrane of Australian anity overthrust by the
East Sulawesi Ophiolite (to be discussed in more detail
in the following section). In the SE of this region is the
Tukang Besi Platform. Little geological and geophysical data is available from this area, but gravity values
suggest a continental terrane (Ali et al., 1996), and the
absence of signi®cant sedimentary section imaged on
available seismic data (Ali et al., 1996) suggests that
the Buton foldbelt does not extend onto the platform,
as previously interpreted by Davidson (1991). The
platform is probably a relatively undeformed fragment
of Australian-anity continental crust (Hamilton,
1979).
Westward the platform passes into a Neogene
molassic basin in front of the Buton fold and thrust
belt. The deeper parts of the foldbelt comprise sedimentary sequences with clear stratigraphic anities to
the Banda forearc islands of Buru, Seram and Timor
(e.g. Smit Sibinga, 1928; Davidson, 1991; Smith and
Silver, 1991), which are are in turn widely interpreted
to be composed for the most part of Australian continental margin sequences. Minor fragments of a formerly more extensive ophiolite complex are locally
preserved in the structurally higher levels of the Buton
foldbelt (Smith and Silver, 1991). In SE Sulawesi the
ophiolite complex is more extensive, where it overlies
Australian-anity continental metamorphic basement
in the west, and overlies or is imbricated with Australian-anity cover sequences towards the east (Surono,
1996).
In Buton the age of collision between the continent
and the ophiolite complex can be dated stratigraphically as Oligocene, based on the youngest age of the
pre-collisional sequence (Upper Eocene-?Lower Oligocene: Smith and Silver, 1991) and the oldest syn- to
post-orogenic sediments (planktonic foraminiferal
zone N3±N4 or latest Oligocene-earliest Miocene in
the Bulu-1 well: P.T. Robertson Utama Indonesia,
personal communication). Davidson (1991) also mentions ages as old as planktonic foraminiferal zone
N3/N4 for the syn- to post-orogenic Tondo Formation in South Buton, whilst Surono (1996) dated
molassic sediments in SE Sulawesi to the Early Miocene, and interpreted a latest Oligocene age of collision between the continental terrane and the
ophiolite belt. This is signi®cantly earlier than the
Middle Miocene initiation of collision interpreted by
Smith and Silver (1991).
Southeast Sulawesi and Buton are cut by a number
of large faults striking NW±SE (the Matano, Lasolo
(or Lawanopo), Kolono, Kolaka, Kioko and Hamilton faults: Figs. 3 and 4). The Matano Fault is a
left-lateral strike-slip fault based on geology (Ahmad,
1977) and seismology (McCarey and Sutardjo,
1982), and it has been generally assumed that the
other faults have similar strike-slip displacements.
However, most of the other faults do not show the
geological characteristics of large strike-slip faults:
there is for instance no clear braiding of fault
strands, or linked oblique transpressional/transtensional folding and faulting. Instead the faults have
been mapped regionally as nearly straight lineaments
with gentle folding subparallel to the fault trend (e.g.
folding of the Tokala and Meluhu formations near
the SE end of the main Lasolo Fault lineament on
the Lasusua±Kendari map sheet: Rusmana et al.,
1993b). These faults have more of the characteristics
of large normal faults. Facies patterns in the Late
Triassic suggest that these major block faults have
existed since at least that time (Charlton, submitted),
and have probably been reactivated several times
since. The relatively young strike-slip oset on the
Matano Fault may be such a reactivation with a
dierent sense of displacement.
The SE arm of Sulawesi is separated from the SW
arm by the Gulf of Bone. Seismic sections across the
gulf (e.g. Guntoro, 1996) suggest an extensional origin for the embayment, and Hamilton (1979) interpreted a Middle Miocene age for the commencement
of extension. The end of extension is marked on the
seismic lines by an unconformity separating normal
faulted and gently folded section below from essentially undeformed strata above. At the head of the
gulf in the Malili geological map sheet (Simandjuntak
et al., 1991a) this unconformity can be traced
onshore, where it corresponds to the base of the
Bonebone and Tomata formations. These formations
are dated as latest Miocene-earliest Pliocene (planktonic foraminiferal zones N17±N18) based on the age
ranges of foraminifera listed by Simandjuntak et al.
(1991a).
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
4.6. Central Sulawesi and East Sulawesi-Banggai-Sula
As with the SE Sulawesi-Buton-Tukangbesi structural domain, East Sulawesi and the Banggai and Sula
islands are here interpreted as having formed an essentially coherent structural block throughout the Neogene. This similarity extends to the general geological
characteristics, with the East Sulawesi Ophiolite thrust
eastwards onto a continental margin terrane of Australian anity. The East Sulawesi Ophiolite is also thrust
westward onto Sundaland-anity basement in the
Poso region of central Sulawesi. As with SE Sulawesi
and Buton, the age of collision between the continental
terrane and the East Sulwesi Ophiolite has been widely
dated to the Middle Miocene (e.g. Garrard et al.,
1988), but again I will suggest that there is evidence
for an earlier onset of collision in the Oligocene. In the
East Arm of Sulawesi there is also a second phase of
fold and thrust belt development in the Pliocene
(Davies, 1990).
The geological evolution of central Sulawesi has
been studied in detail by Parkinson (1991, 1996, etc.).
The Pompangeo metamorphic complex was interpreted
as Sundaland continental margin basement, perhaps
formed by the accretion of a Gondwanan terrane in
the mid Cretaceous (ca 110 Ma) when initial mediumhigh pressure metamorphism developed. During the
Oligocene a second phase of metamorphism developed
under high pressure conditions. Subsequently the East
Sulawesi Ophiolite was obducted onto the Pompangeo
basement complex by westward-directed thrusting.
This obduction may have occurred contemporaneously
with obduction of the Lamasi Complex in southwestern Sulawesi at about 21 Ma (cf. Bergman et al.,
1996).
Separating the base of the East Sulawesi Ophiolite
from the underlying Pompangeo Complex is a tectonic
melange complex, the Peleru Melange. The upper contact of the Peleru Melange Complex with the base of
the ophiolite is overprinted by a metamorphic sole
with an inverted thermal gradient, formed by the
obduction of a hot ultrabasic body onto cooler crustal
material. Parkinson interpeted the Oligocene phase of
deformation in central Sulawesi as the result of obduction of young and hot backarc oceanic crust onto the
Sundaland margin prior to collision with the Banggai
Platform in the Middle Miocene. However, several elements of the sub-ophiolite tectonic complex upon
which the inverted metamorphic gradient is overprinted (speci®cally, the Nanaka, Tetambahu and
probably the Matano Broken Formations of the Peleru
Melange Complex) show stratigraphic anities with
the ``Australian'' sequence of the Banggai Platform. As
the metamorphic sole separating the Peleru Melange
Complex from the structurally overlying East Sulawesi
Ophiolite has been dated radiometrically at about 31
613
Ma (Parkinson, 1996), the Banggai Platform must
have arrived by (or at) that time. This Oligocene age
of suturing between western (Sundaland-anity) and
eastern (Australian-anity) Sulawesi is consistent with
the stratigraphically-derived Oligocene age for the
ophiolite-continent collision in Buton and SE Sulawesi.
The origin of the East Sulawesi Ophiolite and indeed
ophiolites in general is not well understood. Geochemistry suggests an origin in a supra-subduction zone setting, possibly related to the backarc Celebes Sea
(Monnier et al., 1995; also Bergman et al., 1996),
which suggests linkage with the Asian/Paci®c plate
margins. On the other hand, palaeomagnetic studies of
Cretaceous or Palaeogene lavas from the east arm of
Sulawesi indicate a relatively high southerly latitude
(ca 208S), possibly not far north of the Australian continent at that time (Mubroto et al., 1994). Whatever
the precise origin of the East Sulawesi Ophiolite, there
is a common association regionally between ophiolite
complexes and subduction forearcs. In the reconstructions presented later the East Sulawesi Ophiolite is
treated as part of an oceanic forearc complex paired
with the Palaeogene volcanic arc terranes of western
Sulawesi.
The East Sulawesi Ophiolite is separated from the
Banggai-Sula continental fragment by the Tomori
Basin (Davies, 1990; Handiwiria, 1990; Abimanyu,
1990). The basinal succession comprises a lower
sequence of shelf carbonates and clastics ranging in
age from Upper Eocene±Upper Miocene, succeeded by
thick molassic sequences of Pliocene±Recent age.
There is no direct evidence for major collision-related
structural development before the Pliocene, and presumably this region lay some distance in front of the
ophiolite-continent collision front during the OligoMiocene period. The present fold and thrust belt structure on the western ¯ank of the Tomori Basin only
developed during the Pliocene (between 5.2±2.8 Ma
according to Davies, 1990).
The Banggai-Sula continental fragment (Pigram et
al., 1985; Garrard et al., 1988) has long been recognised as stratigraphically related to the continental
part of New Guinea island, and hence to the Australian continent (e.g. KlompeÂ, 1954; Visser and Hermes,
1962). These latter authors suggested stratigraphic
similarity with the Bird's Head structural block of western New Guinea, whilst Pigram et al. (1985) argued
for a more distant origin, adjacent to central Papua
New Guinea. Elsewhere (Charlton, 1996) I have
reviewed the evidence which leads me to favour a connection with the Bird's Head block.
The east arm of Sulawesi is separated from the
north arm by the Gulf of Tomini (Fig. 5). An extensional origin for this gulf, as with the Gulf of Bone,
separating the SE and SW arms of Sulawesi can be
inferred, e.g. from regional seismic lines and gravity
614
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
Fig. 5. Bathymetry/location map of northeastern Indonesia. Oshore bathymetry at 1000 m interval with water depths >4000 m shaded.
modelling (Silver et al., 1983a). A possible indication
of a Middle Miocene or earlier onset of extension is
given by the Middle Miocene (and younger) Bongka
Formation (Poso map sheet: Simandjuntak et al.,
1991b), which is a deepwater turbiditic sequence
unconformably overlying the central Sulawesi collision
complex. Again latest Miocene±earliest Pliocene formations that unconformably overlie older strata (e.g.
the Lonsio and Kintom formations of the Luwuk map
sheet: Rusmana et al., 1993a) may mark the cessation
of extension in the gulf.
4.7. Seram and Buru
Seram Island has been described as a mirror image
across the Banda Sea of Timor in the south (AudleyCharles et al., 1979). It consists of a northward-directed fold and thrust belt forming the forearc complex
of the northern Banda Arc. However, unlike Timor
but like Tanimbar and Kai, there is no major `Asian'
allochthon within the Seram collision complex. [N.B.
this interpretation is in contrast to the original interpretation of Audley-Charles et al. (1979) who
inferred an allochthonous origin for a major part of
the Seram succession based on similarities with the
Timor allochthon as then recognised, particularly with
the Maubisse Formation of Timor. An Australian and
therefore parautochthonous origin for the Maubisse
Formation is now widely accepted (e.g. Audley-Charles
and Harris, 1990), and the necessity for an extensive
allochthon in Seram is negated]. As with Tanimbar
and the eastern Banda Arc, the volcanic arc in the hin-
terland of Seram (Ambon and adjacent islands) is
essentially Pliocene and younger in age.
Buru Island is also usually considered to be one of
the islands in the Banda forearc chain. However,
although there is close stratigraphic similarity between
Seram and Buru, the two islands show very dierent
structural styles. Whilst Seram consists of an imbricate
stack of thrust sheets, Buru has a relatively simple
anticlinorial structure with the principal fold axis following the long axis of the island. It is likely that the
Buru anticlinorium marks a westward dying out of
Banda forearc deformation, and thus the island forms
a pin-point termination for this convergent system.
Seram and Buru islands are separated by a small triangular marine embayment (Fig. 5) which I interpret
as a triangular pull-apart structure (sphenochasm).
This structure osets the Pliocene volcanic island of
Ambelau south of Buru from the Ambon group south
of Seram, which suggests that it opened during Late
Pliocene±Recent times. This sphenochasm is interpreted in the later reconstructions to have accommodated 458 of late-stage clockwise rotation between
Buru and western Seram.
4.8. Bird's Head-Misool and the Sorong Fault Zone
The Bird's Head-Misool block south of the Sorong Fault and west of the Ransiki Fault (Fig. 5)
forms an essentially coherent and little deformed
structural domain of Australian continental anity.
The region has strong stratigraphic links both to
cratonic and parautochthonous southern New Guinea to the east (Dow and Sukamto, 1986) and to
T.R. Charlton / Journal of Asian Earth Sciences 18 (2000) 603±631
the Banda Arc region to the southwest. There is
thus no strong reason to suspect on stratigraphic
grounds that the Bird's Head structural block is
allochthonous, as has been proposed by Pigram and
co-workers (see above). Neither is there any strong
stratigraphic or structural evidence to support the
contention (Pigram and Panggabean, 1984; Pigram
and Davies, 1987) that Misool together with the
Onin and Kumawa peninsulas of the Bird's Head
formed a separate terrane independent of the main
Bird's Head block prior to the Oligocene. On the
contrary, seismic data (e.g. Perkins and Livsey,
1993, Fig. 2) strongly suggests simple tectonostratigraphic continuity from the Misool-Onin-Kumawa
Ridge into the main part of the Bird's Head block
before the inversion that formed the ridge in the
Pliocene, the inversion being related to the collisional development of the northern Banda Arc.
The Bird's Head block was aected by an important
phase of deformation during the Oligocene, most notably giving rise to the NW±SE trending Central Bird's
Head (Vogelkop) Monocline (Visser and Hermes,
1962). Resultant uplift and erosion of the Kemum
basement block to the north of the monocline led to
the shedding of an extensive clastic sequence (the