GR-01302; No of Pages 26

Gondwana Research xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Gondwana Research
journal homepage: www.elsevier.com/locate/gr

GR Focus Review

Late Cretaceous plume-induced subduction initiation along the southern margin of the
Caribbean and NW South America: The first documented example with implications for
the onset of plate tectonics
Scott A. Whattam a,⁎, Robert J. Stern b
a
Department of Earth and Environmental Sciences, Korea University, Seoul 136-701, Republic of Korea
b
Geosciences Department, University of Texas at Dallas, Richardson, TX 75083-0688, USA

a r t i c l e i n f o a b s t r a c t

Article history: Plate tectonics is the governing theory that unifies the Earth Sciences and is unique to Earth. The sinking of
Received 8 February 2014 lithosphere in subduction zones drives plate tectonics but exactly how and why subduction begins (subduction
Received in revised form 19 July 2014 initiation, SI) remains enigmatic. Most SI models require exploitation of existing lithospheric weaknesses but
Accepted 19 July 2014
these are now produced by plate tectonics; so how did the first subduction zones form? One possibility is SI
Available online xxxx
along a plume head-cold lithosphere interface, but no examples have been documented. On the basis of three
Handling Editor: T. Gerya key observations, we show here that the Late Cretaceous tectonic evolution of Central America, NW South
America and the Leeward Antilles is consistent with plume-induced SI (PISI) which nucleated along the southern
Keywords: and western margins of the Caribbean Plate (Caribbean Large Igneous Province, CLIP) at ~100 Ma. (1) Trace ele-
Mantle plume ment chemistry of most 100 Ma and younger units interpreted as CLIP that are exposed along the southern mar-
Subduction initiation gin of the Caribbean Plate and NW South America, record subduction additions which increased with time
Volcanic arc beginning at 100 Ma. These ‘plume- and arc-related’ (PAR) units are distinguishable from: (a) global oceanic pla-
Caribbean teau basalts (OPB); (b) 140–110 Ma OPB along the western edge of the CLIP; and (c) post-100 Ma OPB in the
South America
northernmost CLIP. Whereas the older OPB are compositionally identical to global OPB, the younger northerly
CLIP units are similar to oceanic island basalts and record lower degrees of partial melting than PAR units exposed
along the southern and western peripheries of the CLIP. Both the older and the younger northerly CLIP units lack
evidence of subduction modification. (2) There is no known hiatus between CLIP and younger arc units, suggest-
ing continuous tectono-magmatic evolution from plume to arc. (3) Generation of the CLIP and earliest, overlying
and crosscutting arc units overlaps in time, space, chemical and isotopic compositions; both units are consistent
with derivation from Galapagos Plume-like mantle which became increasingly subduction-modified with time.
These observations illustrate that formation of the CLIP and earliest arc volcanism reflects partial melting of
the same hybrid plume-subduction-modified source in a single, rapidly evolving tectonic environment. The
scale of Late Cretaceous PISI in the SW Caribbean realm is consistent with the expected large scale of lithospheric
collapse as seen for other SI examples, extending some 1400 km from southern Costa Rica–Panama to western
Colombia and 1700 km from Ecuador to the Leeward Antilles (Aruba and Curacao). This first documented exam-
ple of PISI may be relevant to the start of plate tectonics which may have started when subduction began around
Precambrian plume heads. Establishment of lava chemostratigraphy designed to establish the composition of
early SI magmatic successions might provide keys for deciphering whether SI occurred along a collapsed trans-
form (like the IBM arc; early MORB-like lavas) or around a mantle plume (like the Caribbean; early plateau-
like lavas). Recognition of this first example of PISI further suggests that interaction of a sufficiently large
plume head with sufficiently dense oceanic lithosphere in Precambrian time may have triggered the modern re-
gime of plate tectonics.
© 2014 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2. Oceanic plateau- and arc-related oceanic units in Central America and NW South America . . . . . . . . . . . . . . . . . . . . . . . . . . . 0

⁎ Corresponding author. Tel.: +82 2 3290 3172; fax: +82 2 3290 3189.
E-mail address: whattam@korea.ac.kr (S.A. Whattam).

http://dx.doi.org/10.1016/j.gr.2014.07.011
1342-937X/© 2014 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
2 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

3. Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0

3.1. Caribbean data compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.2. Compilations of basalt geochemistry from other relevant tectonic environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.3. Age data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4.1. Age of the CLIP and the timing of inception of arc igneous activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4.2. Changes in magma composition at and after arc inception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4.2.1. Temporal variations of tholeiitic and calc-alkaline affinities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4.2.2. Temporal variations of mantle depletion and degrees of partial melting as recorded by Nb/Yb . . . . . . . . . . . . . . . . . . 0
4.2.3. Temporal variations of mantle fugacity as recorded by Ti/V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4.2.4. Temporal variations of subduction additions: Ba/Th, Ba/Nb, and Th/Nb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4.3. Isotopic constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.1. Synopsis of geochemical and isotopic constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.2. Significance of 99–79 Ma arc-related, felsic granitoid suites: Implications for the timing of subduction initiation . . . . . . . . . . . . . . 0
5.3. Comparison of Caribbean PISI sequences and other SI examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.4. Is there a hiatus between Late Cretaceous CLIP and arc igneous activity? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.5. Critical evaluation of existing models for SW Caribbean subduction zone formation . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.5.1. Little evidence for collision-induced SI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.5.2. Necessity of west-dipping subduction to facilitate eastwards emplacement of PAR units upon South America . . . . . . . . . . . 0
5.6. Advantages of a PISI model for understanding Caribbean tectonic evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5.7. Implications of the PISI model for understanding how plate tectonics began . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0

1. Introduction
Subduction zones return old, cold, oceanic lithosphere to the deep
mantle and generate distinctive arc magmas. However, exactly how
and why subduction initiation (SI) (see Table 1 for a list of abbrevia-
tions used and their definitions) occurs remains controversial. It is con-
tentious whether SI requires a priori convergence (e.g., McKenzie, 1977;
Gurnis et al., 2004) or can occur spontaneously (due to collapse of dense
lithosphere above underlying asthenosphere, Kemp and Stevenson,
1996; Stern, 2004; Mart et al., 2005; Nikolaeva et al., 2010). Attempted
subduction of buoyant lithosphere (e.g., with thick crust) can lead to
collision-induced SI (e.g., Cloos, 1993) as in the case of the Miocene
Solomon Arc system where collision of the Ontong Java Plateau shut
down the south-dipping subduction zone and induced a new north-
dipping subduction zone on the south side of the arc (Cooper and
Taylor, 1985). This is an example of induced nucleation of a subduction
zone (INSZ), reflecting continued plate convergence to form a new
subduction zone (Stern, 2004) as reproduced by geodynamic models
(e.g., Hall et al., 2003). It is also possible that new subduction zones
can form without pre-existing convergence as in the case of spon-
taneous nucleation of a subduction zone (SNSZ; Stern, 2004). Recent
results from geodynamic modeling suggest that SNSZ is also possible
(Dymkova and Gerya, 2013). For both INSZ and SNSZ, a pre-existing
lithospheric weakness is required to overcome the great strength of
oceanic lithosphere, for example a transform fault (Casey and Dewey,
1984; Stern, 2004) or fossil spreading center (Casey and Dewey, 1984).
A better understanding of how new subduction zones form today is
also relevant to the question of how the modern regime of plate tecton-
ics began on Earth (see summary in Korenaga, 2013). Because most of
the force driving plate motions today results from the sinking of dense
lithosphere in subduction zones (Lithgow-Berelloni and Richards,
1995), the question of how plate tectonics began largely depends on
understanding how the first subduction zone formed. Addressing this
question requires understanding how large-scale lithospheric weak-
nesses formed in the absence of plate tectonics. Lithospheric weak-
nesses such as transform faults formed during Phanerozoic time have
been generated by plate tectonics, but how did lithospheric weaknesses
form prior to the inception of plate tectonics? Recognition of the SW
Caribbean as the first plausible PISI example further suggests that a sim-
ilar mechanism may have been responsible for making the first subduc-
tion zone in Precambrian time. Interaction of a sufficiently large plume
head with sufficiently dense oceanic lithosphere could have caused
the lithospheric weakness that formed the first subduction zones and
catalyzed the modern episode of plate tectonics.
In spite of it being simple and plausible, PISI is not widely considered
an important way that new subduction zones form. The early sugges-
tion of Niu et al. (2003) that subduction could initiate along the periph-
ery of an oceanic plateau finds support from numerical models (Ueda
et al., 2008; Burov and Cloetingh, 2010). Perhaps the most important
reason that the hypothesis has not gained wider currency is that until
now no convincing examples of PISI have been identified in the rock re-
cord. In this paper, we demonstrate that the Late Cretaceous tectonic
evolution of Central America and NW South America is consistent
with a PISI model. It is widely recognized that subduction in this region
began around the southern and western margins of the Caribbean Large
Igneous Province oceanic plateau (hereafter referred to as CLIP), but the
transformation from plume activity to form the CLIP and subduction to
generate younger arc igneous activity is poorly understood. We compile
published geochemical and geochronological data to show that the gen-
eral outlines of a PISI model explain some otherwise enigmatic features
of the Late Cretaceous tectonic evolution of the Caribbean Plate.
2. Oceanic plateau- and arc-related oceanic units in Central America
and NW South America
Oceanic plateaus (OP) are large mafic igneous provinces comprising
oceanic crust that is much thicker than the 5–7 km formed by normal
seafloor spreading (Condie and Abbott, 1999). OP form when a mantle
plume head – or other large-scale asthenospheric upwelling – ruptures
the lithosphere and erupts voluminous basalt. These outpourings are
often referred to as Large Igneous Provinces (LIP). Oceanic crust of the
Caribbean Plate is often 15–20 km thick (Bowland and Rosencrantz,
1988) and is widely interpreted as an OP that formed above the
Galapagos Plume before drifting into the gap between North and
South America where it was captured to make up the Caribbean Plate
(e.g., Kerr et al., 2000).

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 3

Table 1 and NW South America are compositionally distinct from those of

Abbreviations and definitions. the pre-110 Ma plateau and post-100 Ma plateau units in the central
Abbreviation/ Definition and northerly segments of the plate. Whereas pre-110 Ma and post-
acronym 100 Ma igneous activity in the center and northerly sections of the
AB Aruba Bathoith Caribbean Plate entailed plume-magmatism only, post 100-Ma
AF Amaime Formation (Western Cordillera, western Colombia) magmatism along the southern periphery of the Caribbean Plate and
AQB Antioquia Batholith (Central Cordillera, western Colombia) NW South America was a hybrid of plume and subduction zone contri-
ALF Aruba Lava Formation
butions. Late Cretaceous igneous rocks are common in NW Costa Rica in
AP Azuero Plateau
BB Buga Batholith (Western Cordillera, western Colombia) the west to western Ecuador in the south to the Leeward Antilles in the
BCC Bulk continental crust east. They are also exposed in southern Hispaniola and have been drilled
BF Barrosso Formation (Western Cordillera, western Colombia) in the middle of the Caribbean Sea. Evidence of subduction contribu-
BUC Bólivar Ultramafic Complex (Colombia) tions disappears along northernmost NW South America (i.e., western
CBA Chagres–Bayano Arc (eastern Panama)
CAVA (modern) Central American Volcanic Arc
Colombia) and the Leeward Antilles at 70–65 Ma, coincident with
CAVAB (modern) Central American Volcanic Arc basalt proposed NW South America–forearc collision at 73–70 Ma and em-
CC Central Cordillera (of western Colombia) placement of OP units eastwards upon South America above a west-
CD Curacao Diorites dipping subduction zone (Vallejo et al., 2009).
CDVC Cabo de la Vela Complex (Guajira peninsula, northernmost
Because there are no obvious breaks between the end of plume
Colombia)
CLIP Caribbean Large Igneous Province (an OP) magmatism and the beginning of arc magmatism, we use the term
CLF Curacao Lava Formation ‘plume and arc-related’ (PAR) units to refer to Late Cretaceous igneous
DSDP Deep Sea Drilling Project sequences interpreted by other researchers as having formed during
GAA Golfito Arc (southern Costa Rica) this post-100 Ma hybrid igneous environment. We identify three coter-
GAA Greater Antilles Arc
minous PISI traces defined by PAR sequences exposed in Central
HFSE High-field strength elements (e.g., Nb, Zr, Ti)
IBM Izu-Bonin–Mariana (a convergent margin in the western Pacific) America (Tr1), coastal Colombia and Ecuador (Tr2), and in the Leeward
IBMVA Modern IBM volcanic arc Antilles (Tr3; Fig. 1). Each of these traces is ~ 1000 km long and each
INSZ Induced nucleation of subduction zone preserves geochemical evidence of plume magmatism evolving into
LIP Large Igneous Province (thought to form when mantle plume
subduction-related magmatism with time. Our goal is to elucidate
impinges base of the lithosphere)
MORB Mid-ocean ridge basalt (pure asthenospheric melt)
how the CLIP and the three 1000-km long traces evolved to the modern
NC Nicoya Complex (western Costa Rica) arc that now occupies Central America and Ecuador–Colombia. All
ODP Oceanic Drilling Project workers who have considered this evolution agree that the transition
OIB Ocean island basalt (tholeiitic and alkalic basalts of within-plate to arc magmatism began sometime in the Late Cretaceous but disagree
oceanic volcanoes)
about exactly when. In Panama and SE-most Costa Rica, Buchs et al.
OP Oceanic plateau
OPB Oceanic plateau basalt (plume basalt) (2010) identified depletion of high field strength elements (HFSE)
PAR Plume- and arc-related and concluded that arc igneous activity began by 75 Ma. In western
PG Pujilí Granite (western Ecuador) Ecuador, Vallejo et al. (2009) used a similar argument to conclude that
PISI Plume-induced subduction initiation
arc igneous activity began by 85 Ma. In Aruba and Curacao, Wright
PU Pallatanga Unit (western Ecuador)
RCA Rio Cala Arc (western Ecuador)
and Wyld (2011) interpreted the beginning of arc igneous activity as
SAA Sona–Azuero Arc (southern Panama) the intrusion of the oldest granitoids dated at 89 Ma. These inferences
SDB Serranía de Baudó (complex, Western Cordillera, western for the timing of SI in these regions form the basis of our designations
Colombia) of PISI traces (Tr1, Tr2, Tr3, Figs. 1, 2).
SI Subduction initiation
Part of the uncertainty in interpreting when the switch from plume
SIRO Subduction initiation rule ophiolites (e.g., SSZ ophiolites)
SNSZ Spontaneous nucleation of subduction zone to arc magmatism occurred reflects an assumption that the three traces
SSZ Supra-subduction zone; i.e. formation in a forearc, arc, or evolved independently, which could be falsified if these three traces are
backarc basin, above a subduction zone all related to the same plume-to-arc tectonomagmatic transition. The
THI Tholeiitic index: tholeiitic suites have THI N1; calc-alkaline
uncertainty of the transition is also reflected in reinterpretation of
suites have THI b 1
Tr1 PISI trace 1
Central American PAR units as having arc affinities for some sequences
Tr2 PISI trace 2 previously thought to be CLIP fragments (Wörner et al., 2009; Buchs
Tr3 PISI trace 3 or eastern et al., 2010; Wegner et al., 2011). Stratigraphic relations between
VAB Volcanic arc basalt (subduction-modified basalt) plateau and arc units are generally conformable where these can be ex-
VF Volcanic Formation (western Cordillera, western Colombia)
amined (e.g., western Ecuador, Vallejo et al., 2009; Panama, Buchs et al.,
2010) and not separated by any documented unconformity or sedimen-
tary interval (Fig. 2). We take this to indicate that magma sources
Although most of the CLIP is submerged beneath the Caribbean Sea, evolved from plume- to subduction-related without interruption. In
many Late Cretaceous igneous complexes in Central America (Costa the following sections we use geochemical approaches to examine this
Rica and Panama), NW South America (Colombia and Ecuador) and transition in greater detail.
the southern Caribbean (Aruba and Curacao, Leeward Antilles) (Fig. 1)
are interpreted as CLIP fragments (see references in Table 2). Basal 3. Materials and methods
units and sometimes all of these basaltic sequences are interpreted
as uplifted or accreted CLIP (Fig. 2). Early studies suggest that the CLIP Below we summarize how we compiled geochemical and age data
was mostly constructed between 92 and 88 Ma with minor magmatic for our study. Further details of geochemical and geochronological
pulses at 124–112 Ma and 76–72 Ma (Sinton et al., 1998; Kerr et al., data and tectonic interpretations are provided in the GR Online Supple-
2003 and references therein) but some later studies (e.g., Serrano mentary Document.
et al., 2011) emphasize an extended period of essentially uninterrupted
magmatic activity spanning the entire Cretaceous. 3.1. Caribbean data compilation
We show below that there were two main phases of distinct plateau
construction at 140–110 Ma and after 100 Ma and that post-100 Ma We compiled geochemical, geochronological and isotopic data from
OPB-related lavas along the southern margins of the Caribbean Plate southern Central and NW South American PAR complexes (Figs. 1

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
4 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

70oW 60oW
Plateau- and arc-related units documented
Greater Antilles
in this study (mainly Late Cretaceous, on-
land or drilled seafloor)
(Early Cretaceous ophiolites)
N
Hispaniola
Other plateau- or arc-related ophiolitic or 20oN
seafloor units 14
15
11
CARIBBEAN 152
LAA
PLATE 1001 151 150
146
system
13
Costa 153
LA 8
Rica PISI Tr3 12
10
400 km
1b 1c Panama 7
2b 9 10oN
1d 6b
1a 2a
6a
COCOS PISI Tr1 6bii
PLATE 3
Colombia
NAZCA PLATE 4
5b
6biii
SOUTH
PISI Tr2 5aii AMERICAN
5a 0oN
Galapagos
PLATE
Islands 5aiii
5ai Ecuador

Plateau- and arc-related units

1a Nicoya Complex,
“old plateau”, 140-110 Ma (OP)
“young plateau”, 98-82 Ma (OP)
1b Tortugal Complex, 91-88 Ma (OP)
1c Herradura Complex, 88-82 Ma (OP)
1d Golfito Complex
1di Azuero Plateau ?, 89-85 Ma (OP)
1dii Golfito Arc, 75-66 Ma (PA)
2a Azuero Complex,
2ai Azuero Plateau, 89-85 Ma, (OP)
2aii Sona-Azuero Arc 73-69 Ma (PA, A)
2b Chagres-Bayano Arc, 70-39 Ma (A)
3 Serranía de Baudó Complex, 79-72 Ma (OP)
a
4 Gorgona Island, 91-81 Ma (OP)
b
5a Pallatanga Unit, 90-85 Ma (OP)
5ai Piñon Formation, 90-85 Ma (OP)
c
5aii Totoras Amphibolite, 87-82 Ma (meta OP )
5aiii Pujilí Granite, 87-81 Ma (A)
5b Rio Cala arc, ~85-64 Ma (A)
d PISI Tr3, PISI trace 3
6a Central Cordillera, 97-86 Ma, (OP)
6ai Antioquia Batholith, 95-86 Ma (A)
6b Western Cordillera, 94-75 Ma (OP) (K)
e
6bi Volcanic Formation (101-98) (V)
6bii Bólivar UM Complex, 99-87 Ma (OP)
6biii Buga Batholith, 92-89 Ma (A)
7 Cabo de la Vela Complex ~83-65 Ma (A)
8 Aruba
f
8a Aruba Lava Formation,103-89 Ma (OP)
8b Aruba Batholith, 90-79 Ma (A)
9 Curaçao
g
9a Curaçao Lava Formation, 120-62 Ma (OP)
9b Curaçao Diorite Intrusions, 87-85 Ma (A)
10 Bonaire, Albanian-Cenomanian (A, GAA)
11 Beata Ridge, 82-75 & 56-54 Ma (OP)
12 DSDP Leg 15, 97-87 Ma (OP)
13 ODP Leg 165, Site 1001, 82-80 Ma (OP)
14 Dumisseau Formation, Haiti, 98-82 Ma (OP)
15 Duarte Formation, Dominican Rep. 88-85 Ma (OP)
Plume-induced Subduction Initiation (PISI) traces
PISI Tr1, PISI trace 1
PISI Tr2, PISI trace 2
localities in which felsic, subduction-related,
arc-like intrusive units temporally overlap the
associated ~100-80 Ma oceanic plateau
basement they intrude (see unit ages above
and Fig. 2)

Fig. 1. Distribution of Cretaceous to Eocene Central American, NW South American and southern Caribbean (mostly post-100 Ma) plume- and arc-related (PAR) units. Also shown are
the traces of the circa 100–90 Ma PISI event (dashed lines, see text in box). In the large box, unit names, locations are followed by age ranges (uncertainties rounded to the nearest
Ma) and the current interpretation as oceanic plateau (OP), proto-arc (PA) or arc (A). Thick arrows represent present directions of plate movement. Superscripts (besides unit names
or ages in box): a, age range of Kerr et al. (1996a) and Sinton et al. (1998), but we note that Serrano et al. (2011) record a much wider range in apparent ages of ~87–68 Ma; b, the
Pallatanga Unit is considered to be correlative with other circa 90–85 Ma oceanic plateau units in western Ecuador including the Piñon Formation, and the Pedernale, Guaranda and related
units (see Mamberti et al., 2003; Allibon et al., 2008); c, our interpretation is that the Totoras amphibolite represents the metamorphic sole of the 90–85 Ma western Ecuadorian oceanic
plateau unit, see GR Online Supplementary Document; d, the oldest ages for the Rio Cala Arc will differ depending on whether one considers the La Portada Formation as part of the Rio Cala
Arc or another system (see Kerr et al., 2002a and Vallejo et al., 2009); terminal magmatism at the Rio Cala Arc sensu stricto is likely about 72 Ma (Vallejo et al., 2009); e, the Volcanic For-
mation also includes the Barosso and Amaime formations (Villagómez et al., 2011) of the Western Cordillera (see Fig. 2); f, age range takes into consideration the circa 100 Ma radiometric
age reported in Wright and Wyld (2011); g, controversy exists over the interpreted age range of the CLF (see Wright and Wyld, 2011); age incorporates the circa 115 Ma age reported in
Wright and Wyld (2011). K and V in bold italics in brackets after 6b, Western Cordillera and 6bi, Volcanic Formation indicates that these references are from Kerr et al. (1997) and
Villagómez et al. (2011), respectively. Complete references for all units shown are provided in Table 2. Abbreviations (on map): LA, Leeward Antilles; LAA, Lower Antilles Arc.

and 2; references in Table 2). Data are from (as numbered in Figs. 1
and 2): (1a) the Nicoya, (1b) Tortugal, (1c) Herradura and (1d) Golfito
complexes, Costa Rica; (2a) the Azuero Complex which comprises the
(2ai) Azuero Plateau and the (2aii) Sona–Azuero Arc; and (2b) the
Chagres–Bayano Arc, Panama; (3) the Serranía de Baudó Complex,
westernmost Colombia; (4) Gorgona Island, off of western Colombia;
(5a) the Pallatanga and (5ai) Piñon Units and related oceanic plateau
basement units and (5b) the Rio Cala and Macuchi arcs, western
Ecuador; the (6a) Central Cordillera, (6b, 6bi) Western Cordillera and
(6bii) Bolivar Ultramafic Complex, western Colombia; the (7) Cabo de
la Vela ultramafic–mafic Complex, northernmost Colombia (Guajira
Peninsula); (8a) Aruba Lava Formation and (9a) Curaçao Lava Forma-
tion, Leeward Antilles; (11) the Beata Ridge, (12) DSDP Leg 15 and
(13) ODP Leg 165, Site 1001, Caribbean Sea LIP basalts; and (14) the
Dumisseau Formation, Haiti and (15) Lower Duarte Formation,
Dominican Republic. In addition, we also compiled data for felsic, arc-
related units which intrude or are spatially associated with plateau
units in western Ecuador (5aiii, Pujili Granite), western Colombia (6ai,

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 5

Table 2
Summary of geologic units, tectonic setting, age and references.

Unit/complex Complex/unit Region Interpretation Age References

No. (fr. Fig. 1) (OP, PA, or A) (Ma)

PISI Tr1
1a-old Nicoya Complex w. CR OP 140–130, 110 Hoernle et al. (2004)
1a-young Nicoya Complex w. CR OP 98–82 Sinton et al. (1997); Hauff et al. (2000a);
1b Tortugal Complex w. CR OP 91–88 Alvarado et al. (1997); Hauff et al. (2000a)
1c Herradura Complex w. CR OP 88–82 Sinton et al. (1997), Hauff et al. (2000a)
1d Golfito Complex w. CR OP Assumed to be ~80 Hauff et al. (2000a)
1d Golfito Complex s. CR PA/A Maastrichtian (75–66) DiMarco et al. (1995); Buchs et al. (2010)
2a Azuero Complex s. PAN
2ai Azuero Plateau OP Coniachian–early Santonian Kolarsky et al. (1995); Buchs et al. (2009),
(89–85) Buchs et al. (2010); Wörner et al. (2009);
Wegner et al. (2011)
2aii Azuero Arc PA, A 73–69 Wörner et al. (2009), Buchs et al. (2010),
Wegner et al. (2011)
2b Chagres–Bayano Arc e. PAN A 70–39 Wörner et al. (2009), Wegner et al. (2011),
Montes et al. (2012)
PISI Tr2
3 Serranía de Baudó Complex w. COL OP 79–72 Kerr et al. (1997)
4 Gorgona Island w. COL OP 91–81 Kerr et al. (1996a), Arndt et al. (1997),
Walker et al. (1999); Révillon et al. (2000a)
4 Gorgona Island 101–59 Serrano et al. (2011)
5a Pallatanga Unit w. ECU OP 90–85 Hughes and Bermúdez (1997), Boland et al.
(2000), Kerr et al. (2002a), Hughes and Pilatsig
(2002), Mamberti et al. (2003), Vallejo et al. (2009)
5ai Piñon Formation Turonian, 90–86 Jaillard et al. (1995, 2009), Luzieux et al. (2006)
5aii Tortoras Amphibolite w. ECU meta. OP 87–82 Beaudon et al. (2005), Vallejo et al. (2006)
5aiii Pujilí Granite w. ECU A 87–81 Spikings et al. (2005), Vallejo et al. (2006, 2009)
NA San Juan w. ECU OP 136–110 Lapierre et al. (2000)
NA San Juan w. ECU OP 101–98 Mamberti et al. (2004)
5b Rio Cala Arc w. ECU A 85–64 Hughes and Bermúdez, 1997, Hughes and Pilatsig
(2002); Kerr et al. (2002a), Vallejo et al.(2006, 2009),
Chiaradia (2009), see also Boland et al. (2000)
NA Macuchi Arc w. ECU A 44–33 Kerr et al., 2002a, Vallejo et al. (2009 and refs. therein),
Chiaradia (2009)
6a Central Cordillera w. COL OP 97–86 Kerr et al. (1997, 2002b)
6ai Antioquia Batholith w. COL A 95–86 Villagómez et al. (2011)
6b Western Cordillera w. COL OP Cenomanian–Conachian Barrero (1979), Bourgois et al. (1987), Kerr et al. (1997),
(97–87), 94–75 see also Sinton et al. (1998) and Kerr et al. (2002b)
6bi Volcanic, Amaime, & Barroso w.COL OP 101–98 Villagómez et al. (2011)
Fms. (of WC)
6bii Bólivar UM Complex (WC) w. COL OP 99–87 Kerr et al. (2004), Serrano et al. (2011),
Villagómez et al. (2011)
6biii Buga Batholith (WC) A 92–89 Villagómez et al. (2011)
PISI Tr3
7 Cabo de la Vela MUMC n. COL/LA A 83–65 Weber et al. (2009)
8 Aruba LA
8a Aruba Lava Fm. (ALF) OP Turonian, 94–89 MacDonald (1968), White et al. (1999), see also
Wright and Wyld (2011)
8b Aruba Batholith A 90–82 White et al. (1999), van der Lelij et al. (2010),
Wright and Wyld (2011)
9 Curaçao LA
9a Curaçao Lava Fm. (CLF) OP mid-Albian (~110), Wiedmann (1978), Kerr et al. (1996b), Sinton et al.
120–78 (1998), Wright and Wyld (2011)
9a 94–62 Loewen et al. (2013)
9b Curaçao Diorite Intrusives A 87–85 Wright and Wyld (2011), see also Sinton et al. (1998)
10 Bonaire LA A (of the GAA) NA (as comprises GAA) see Wright and Wyld (2011)
Center of Caribbean Plate
11 CLIP, Leg 15 Carib. Sea OP 97–87 Sinton et al. (1998)
12 Beata Ridge Carib. Sea OP 82–75, 56–54 Sinton et al. (1998), Révillon et al. (2000b),
(see also Donnelly et al., 1973)
13 CLIP, Site 1001 Carib. Sea OP 82–80 Sinton et al. (2000), Kerr et al.(2009)
Northern Caribbean Plate
14 Dumisseau Formation Haiti OP 110–82 Sen et al. (1988), Sinton et al. (1998), Loewen et al. (2013)
15 Lower Duarte Formation Dom. Rep. OP 88–85 Lapierre et al. (1997, 1999)

This table. References for datasets of Caribbean plume- and arc-related units used in this study. The datasets include mostly age and geochemical data.
Footnotes:
Note that all radiometric ages listed here and elsewhere in the text encompass calculated ages plus and minor their uncertainties rounded to the nearest Ma, unless stated otherwise
(e.g., as in Fig. 3). Radiometric ages are determined by 40Ar/39Ar or U–Pb zircon; all 40Ar/39Ar ages listed are plateau ages except those of the older Nicoya Complex (Hoernle et al.,
2004) which are isochron (and agree within error of the uncertainties with the plateau ages). Plateau ages listed for the Curaçao and Dumisseau formations of the study of Loewen
et al. (2013) include only those deemed as reliable plateau ages (by Loewen et al., 2013). Similarly, the age range for Gorgona of the study of Serrano et al. (2011) includes plateau
ages only. The only K–Ar ages listed are those for the Cabo de la Vela Complex. Biochronologic ages are in italics. The radiometric age of 94–89 Ma for the ALF is based on a U–Pb age
of 97.3 ± 8.2 Ma of an ALF gabbro as reported by Wright and Wyld (2011), whereas the Turonian age is based on ammonites intercalated with ALF basalts (MacDonald, 1968). Abbrevi-
ations: A, arc; COL, Colombia; CR, Costa Rica; ECU, Ecuador; GAA, Greater Antilles Arc; LA, Leeward Antilles; meta. OP, metamorphosed OP; MUMC, mafic-ultramafic complex OP, oceanic
plateau; PA, proto-arc; PAN, Panama; OP; WC, Western Cordillera (of western Colombia).

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
6 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

key tectonomagmatic units PISI trace 1 PISI trace 2 PISI trace 3

arc Northwest Southeast North South West East
arc (felsic)
proto-
Caribbean
(felsic subduction-related, Pacific coast coast Pacific Coast Caribbean Coast (Leeward Antilles)
arc + arc arc-like intrusives that
proto- temporally overlap the
arc associated ~90 Ma oceanic western southern southern eastern northernmost Aruba Curacao Bonaire
western Colombia western Ecuador
plateau basement units
plateau they intrude)
Costa Rica Costa Rica Panama Panama coastal inland Colombia

Ma
2a
AC 2b
PALEOGENE

40
EOCENE
NW South America-
50
5a, 5b
forearc collision 6b, 6bi west east
1d SAA CBA
5a, 9
60 PALEOCENE
GC (2aii) (2b) WC (6b) (K) 6a, b
7 Greater
a
Maastrichtian 1a-c GA 3 6ai 5aiii Antilles
L. CRETACEOUS

70
bulk of CLIP formation (senso lato) (1dii) PG (5aiii) RCA CDVC 8 CD Arc
(50% of all post-100 Ma Campanian ? SDBC CC (6a) pelag. (9b)
BB (5b)
radiometric ages) ? ? seds. AB
Santonian
young AP (1di) AP (2ai) AP (2ai) (8b)
10
Coniacian AQB
90 Turonian plat. BUC (6biii) (6ai)
Cenomanian (6bii) ALF ?
100
? PU (5a) PU (5a) PU (5a) (8a)
~ 10 m.y. stratigraphic no sed. horizon
Albian conformable
magmatic gap
E. CRETACEOUS

110 or unconform.? contact? CLF

oldest post-100 Ma (our interpretation,
radiometric ages old VF, BF as arc, see upper (9a)
120 Aptian plat. & AF (6bi) (V) &
(9% of all ages) caption & text)
of WC lower
Barremian NC
130
Hautervian timing of PISI
Valanginian
140

Fig. 2. Simplified stratigraphic relations of Cretaceous to Eocene Central American, NW South American and southern Caribbean plume- and arc-related (PAR) units from this study to
demonstrate relations between underlying units interpreted as plateau and overlying ones interpreted as arc. Units numbered as in Fig. 1. The bulk of CLIP formation appears to have oc-
curred at 95–85 Ma (thick, yellow, horizontal line, see Fig. 3). We infer that the time of plume-induced subduction initiation (PISI) began during this time (~95–85 Ma). The collective data
suggests that the SI event is traceable along southern Central America (PISI trace 1, Costa Rica and Panama); northwestern south America (PISI trace 2, western Ecuador and western
Colombia); and in the southern Caribbean in the Leeward Antilles (PISI trace 3, Aruba and Curacao). References and ages for PAR units are in Table 2 and the majority of stratigraphic re-
lations are ‘constructed’ on the basis of these ages; exceptions include the stratigraphic columns for 1d and 2a which are from Buchs et al. (2010) and the stratigraphic columns for 5a, b
which are from Vallejo et al. (2009). Unit acronyms: AC, Azuero Complex; ALF, Aruba Lava Formation; AF, Anime Formation; AP, Azuero Platteau; AQB, Antioquia Batholith; BB, Buga
Batholith; BF, Barosso Formation; BUC, Bólivar Ultramafic Complex; CBA, Chagres–Bayano Arc; CC, Central Cordillera; CD, Curaçao Diorites; CDVC, Cabo de la Vela Complex; CLF, Curacao
Lava Formation; GA, Golfito Arc; NC, Nicoya Complex; PG, Pujilí Granite; PU, Pallatanga Unit; RCA, Rio Cala Arc; SDBC, Serranía de Baudó Complex; VF, Volcanic Formation; WC, Western
Cordillera. K and V in bold italics in brackets after 6b, Western Cordillera and 6bi, Volcanic Formation indicates that these references are from Kerr et al. (1997) and Villagómez et al. (2011),
respectively. Superscripts, main figure: a, represents the age range for the Nicoya, Tortugal and Herradura complexes, which might represent separate units (see Bandini et al., 2008;
Baumgartner et al., 2008) and not the Nicoya Complex sensu stricto; b (beside 5aiii), the (original) stratigraphic relation between the Pujilí granite and Pallatanga Unit are uncertain,
hence the question mark (see Vallejo et al., 2006).

Antioquia and 6bii, Buga batholiths), Aruba (8b, Aruba Batholith) and forearc basalts (n = 29) are interpreted to have formed in a similar
Curacao (9b, Curacao diorites). These compilations include data for manner during SI (Whattam and Stern, 2011; Reagan et al., 2013). Ad-
~ 400 samples. Datasets for 11-15 are of igneous rocks removed from ditionally, we also use compositions of mid-oceanic ridge basalts
subduction-affected regions along the three PISI traces and thus provide (MORB) and oceanic island basalts (OIB) (Sun and McDonough, 1989)
useful baseline information about the plume mantle source, unaffected and bulk continental crust (BCC) (Rudnick & Gao, 2003) for further
by subduction inputs, against which data for PAR sequences can be com- comparison.
pared. Further details about samples and methods of data manipulation
are provided in the GR Online Supplementary Document. 3.3. Age data

3.2. Compilations of basalt geochemistry from other relevant tectonic We compiled 100 radiometric dates (5 Ma bins) in addition to bio-
environments stratigraphic age determinations of samples from units interpreted as
CLIP and present these in Fig. 3. We used this database to assign ages
To better illuminate the tectonic environments where Late Creta- to each PAR unit so that geochemical characteristics are securely placed
ceous Caribbean PAR sequences formed, we also compiled geochemical in a temporal framework.
data for lavas from representative examples of three distinct tectono-
magmatic associations: (1) oceanic plateau basalts (OPB); (2) volcanic 4. Results
arc basalts (VAB); and (3) basalts erupted during subduction initia-
tion. The OPB dataset (n = 353) compiles tholeiitic basalts from four Compiled geochemical and geochronological datasets allow us to
OP in the Pacific Ocean: Ontong Java (Mahoney et al., 1993; Tejada demonstrate two main features that – in conjunction with stratigraphic
et al., 2002), Manihiki (Hoernle et al., 2010) and Hikurangi (Hoernle relations and tectonic considerations – are consistent with a PISI model
et al., 2010) plateaus and the Shatsky Rise (Sano et al., 2012); and the (see Section 5) for southern Central American and NW South American
Kerguelen Plateau (Mahoney et al., 1995; Frey et al., 2002; Neal et al., plume- and arc-related complexes: (1) the chemistry of most post-
2002), Indian Ocean. Although many more datasets exist for global 100 Ma units along Central America and NW South America interpreted
OPB the compilation of nearly 400 OPB samples we have amassed is as CLIP show a progressive evolution to higher degrees of partial melt-
more than adequate to serve for comparison with Caribbean OPB. The ing and increasing subduction additions with time beginning at
volcanic arc basalt (VAB) datasets comprise Quaternary lavas from the 100 Ma; and (2) the rapid transformation from plateau to arc environ-
Central American and Izu-Bonin Mariana (IBM) arc (as compiled by ment beginning ~100 Ma.
Jordan et al., 2012). The Central American Arc dataset comprises lavas
(arc basalts, n = 1052) from Quaternary volcanoes in Costa Rica, 4.1. Age of the CLIP and the timing of inception of arc igneous activity
Nicaragua, El Salvador and Guatemala. Izu-Bonin Marianas forearc
basalts (Reagan et al., 2010) formed at the beginning of subduction be- The summary of age information (Fig. 3) reveals two salient fea-
neath this arc. Data compiled for ophiolites that follow the ‘subduction tures: (1) the existence of two phases of plateau construction before
initiation rule’ of Whattam and Stern (2011) includes lower and upper and after 100 Ma; and (2) a main pulse of magmatic activity between
ophiolite unit basalts and basaltic andesites that comprise the Pindos, 95 and 85 Ma. The first phase between 140 and 110 Ma appears to be
Mirdita, Troodos and Semail ophiolites (as compiled by Whattam and volumetrically subordinate as it comprises only 8% of all radiometric
Stern, 2011). The subduction initiation rule ophiolite basalts and IBM dates but it remains uncertain as to the actual volumetric percentage

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 7

Biochronologic ages of units the frequency of ages between 100 and 95 Ma (n = 8), 95–90 Ma
interpreted as mafic OP
a (n = 16) and 90–85 Ma (n = 30) followed by decreasing frequency
140 130 120 110 100 90 80 70 60 50 40
thereafter. We interpret this to indicate that the major pulse of CLIP
magmatic activity occurred between 95 and 85 Ma (46% of all radiomet-
CRETACEOUS TERTIARY
ric age determinations of mafic rocks interpreted as CLIP) and that CLIP
Early Late PALEOGENE magmatic activity subsequently waned.
Our age compilation (Fig. 3) is broadly consistent with earlier studies

PALEOCENE
MAASTRICHTIAN
HAUTERIVIAN
VALANGINIAN

concluding that CLIP plume activity mainly occurred between 92 and

CENOMANIAN

CAMPANIAN
BARREMIAN
BERRIASIAN

EOCENE
SANTONIAN
CONIACIAN
TURONIAN
APTIAN

ALBIAN 88 Ma with minor magmatic pulses at 124–112 Ma and 76–72 Ma

(Sinton et al., 1998; Kerr et al., 2003). A more recent study (Serrano
et al., 2011) concludes that CLIP plume activity continued uninterrupted
until the end of the Cretaceous. Ages younger than ~ 80 Ma however
Tr 1: Costa Rica & Panama are mostly from the middle of the Caribbean Sea and Curaçao in the
(1, 2) Azuero Plateau (PAN,2ai) Leeward Antilles. Most post-80 Ma ages elsewhere are confined to
Gorgona Island off western Colombia (Serrano et al., 2011) and the
Golfito Arc Serranía de Baudó unit (Kerr et al., 1997) of western Colombia (see
(CR, 1dii)
Supplementary Figs. 1 and 2 which split the radiometric ages into re-
Tr 2: western Ecuador & gions and discrete units). In the case of the Gorgona Complex, earlier
western Colombia Piñon Formation (ECU,5a)
(3, 4, 5, 6) studies (Kerr et al., 1996a and studies cited therein) yield ages of 91–
81 Ma; only three samples from the dataset of Serrano et al. (2011)
Western Cordillera (COL,6b)
yield reliable younger circa 75–70 Ma ages which is similar to the age
of terminal magmatism in the Serranía de Baudó Complex (~ 72 Ma,
Tr 3: Aruba & Curaçao
(8, 9) Kerr et al., 1997). Moreover, there has been disagreement as to which
Curaçao Lava Aruba (lower)
Formation (9a) Lava Formation (8a) units represent plateau and which represent arc. For example, Hauff
et al. (2000a,b) considered the 75–66 Ma Golfito Complex of southern
Central segment Costa Rica to be a plateau sequence, whereas Buchs et al. (2010)
of the CLIP in Caribbean Sea
Venezuela Basin reinterpreted this as arc. Thus, if the former interpretation is held, ‘pla-
teau’ magmatism extended until 66 Ma but only until ~80 Ma if the lat-
30 ter interpretation is considered. Similarly, and as we develop in
Radiometric ages of units SI (Buchs et al., 2010, Panama
25 interpreted as mafic OP &
& SW Costa Rica)
Section 4.2, both the Serrania de Baudo and Gorgona units interpreted
felsic arc-related plutonics as plateau record obvious subduction-derived modifications and termi-
Frequency

SI (Vallejo et al., 2009,

20 intrusive into OP
biochronologic ages
15 (from a)
10
mafic units interpreted as
OP, all regions (1, 3-6, 8, 9, 11-15)
felsic, arc-related units intrusive
5 into OP see caption)
140 130 120 110 100
Fig. 3. Distribution and frequency of (a) biostratigraphic and (b) radiometric age determi-
nations of plateau- and arc-related units of this study along traces 1, 2 and 3 (Tr1, Tr2, Tr3)
and the central segment of the Caribbean Sea. Radiometric dates are binned on the basis of
calculated ages without consideration of uncertainties. The gray bar in (a) is for visual aid
to underscore that most biochronologic age determinations correlate with the highest fre-
quency of radiometric age determinations (at ~95–85 Ma, as shown in (b)). The distribu-
tion and frequency of radiometric age determinations subdivided on the basis of region
and individual units are provided in Figs. A1 and A2, respectively, of the GR Online Supple-
mentary Document. References for the age of units plotted are provided in Table 2 apart
from the Venezuelan Basin age data which is from Donnelly et al. (1973). The numbers
in italics in brackets beside the region or unit name corresponds to the unit numbers
assigned in Fig. 1.
of older vs. younger basalts because more older lavas must be buried
under younger lavas. These older ages come from the western and
southeastern limits of the study area. In the west, these ages are from
the Nicoya Complex, Costa Rica (five of eight pre-100 Ma ages)
(Hoernle et al., 2004). In the southeast, these ages are found in the
San Juan oceanic plateau unit, western Ecuador dated as 123 ± 13 Ma
via Sm–Nd isochron (Lapierre et al., 2000). In the northeast, early
plume activity is documented for the 97.3 ± 5.2 Ma lower Aruba Forma-
tion and 112.7 ± 7.3 Ma Curacao Lava Formation (see Wright and Wyld,
2011).
Although it is not possible to evaluate the extent to which age
distributions may be biased, we note that the radiometric and biostrat-
igraphic ages agree reasonably well (Fig. 3). We do not know how
representative the number of ages is of the volumes of CLIP magmas
through time, but Fig. 3 nonetheless demonstrates a steady climb in
western Ecuador)
nation of magmatism at 75–70 Ma is consistent with the timing of ter-
SI (Wright and Wyld, 2011,
Aruba & Curacao) mination of subduction-related magmatism elsewhere along the
western margin of NW South America. For example, the cessation of
magmatism at Gorgona and Serranía de Baudó coincides with the termi-
b nation of arc magmatism to the north (Cabo de la Vela Complex by
~ 74 Ma, Weber et al., 2009) and to the south (Rio Cala Arc, western
90 80 70 60 50 40 Ecuador by ~ 72 Ma, Vallejo et al., 2009) and is coincident with South
Ma America-forearc collision at 73–70 Ma (Vallejo et al., 2009) and the em-
placement of subduction-modified plateau units upon western South
America soon thereafter.
CLIP plume activity mainly occurred between 95 and 85 Ma (Fig. 3)
and may have effectively buried Early Cretaceous oceanic crust. Younger
CLIP ages are essentially confined to the center of the CLIP (e.g., Beata
Ridge, terminal magmatism at ~ 54 Ma (Révillon et al., 2000b)) and
the Leeward Antilles (e.g., Curaçao, terminal magmatism at ~ 62 Ma,
(Loewen et al., 2013)).
Our goal is to understand how the CLIP and the three 1000-km long
PISI traces evolved to the modern arc that now occupies Central
America and Ecuador–Colombia. We assume that subduction of various
Pacific oceanic plates (Farallon, Cocos, Nazca) beneath Tr1 and Tr2 has
been continuous since east-dipping subduction began in the Late Creta-
ceous. The first step to understanding what happened is to reconstruct
when it happened. Plume activity was especially intense at 95–85 Ma
and arc volcanism in Tr1 began by at least 75 Ma (Buchs et al., 2010)
and in Tr2 by 85 Ma (Vallejo et al., 2009). Independent constraints on
the minimum age of subduction are 60 Ma based on the oldest accreted
Galapagos hotspot tracks (Hoernle et al., 2002). Arc igneous activity and
therefore subduction was very brief along Tr3 in the Leeward Antilles
but left distinctive calc-alkaline intrusive suites in Aruba and Curacao
(Aruba Batholith, ~ 92–78 Ma, White et al., 1999; van der Lelij et al.,
2010; Wright and Wyld, 2011; Curacao Diorites, ~ 82–80 Ma, Wright
and Wyld, 2011) which indicates that subduction took place 89–
86 Ma along Tr3 (Wright and Wyld, 2011). These ages are broadly con-
sistent with results from studies of the Villa de Cura belt of northern
Venezuela, where blueschist facies metamorphism is interpreted to

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 8
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-

Table 3
Summary of geologic units, age, and diagnostic chemical features.

Partial melting proxy Subduction proxy

Unit/ Complex/unit Age Region Interpretation THI n Ti/V Nb/Yb Ba/Th Ba/Nb Th/Nb
complex

No. (Ma) (OP, PA, or A) mean TNSS: Fe4.0, mean range n mean range n mean range n mean range n mean range n
(fr. Fig. 1) Fe8.0

PISI Tr1
1a-old Nicoya Complex 140–130, 110 w. CR OP NC 1 1.82 1.65–2.07 5 87.12 73.42–118.73 5 5.97 4.88–8.39 5 0.07 0.07–0.07 5
1a-young Nicoya Complex 98–82 w. CR OP 1.50 49: 3, 31 21.20 16.90–27.63 42 1.72 1.36–2.38 30 201.78 27.27–1414.81 30 11.94 1.97–92.49 30 0.06 0.04–0.08 30
1b Tortugal Complexa 91–88 w. CR OP
depleted basalt w. CR OP 19.26 1 1.34 1 100 1 7.28 1 0.07 1
(BC-17)
enriched basalt w. CR OP 18.57 1 10.3 1 88.69 1 6.43 1 0.07 1
(BC-18)
BC-16 (relative 60.00

S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

‘depletedness’
unknown)
1c Herradura Complex 88–82 w. CR OP NC 22.54 17.83–40.62 8 1.63 1.41–1.96 4 132.41 116.67–153.33 4 8.58 7.07–9.62 4 0.06 0.06–0.07 4
1d(H) Golfito Complex assumed to w. CR OP 1.15 4: 1, 1 22.31 17.74–25.45 4 1.25 1.06–1.63 3 208.55 82.61–355.56 3 21.14 5.29–32.43 3 0.10 0.06–0.14 3
be ~80
1d(B) Golfito Complex Maastrichtian s. CR PA/A 1.21 8: 1, 3 18.02 14.64–21.32 4 0.95 0.68–1.20 4 644.28 400.00–877.27 3 63.22 51.10–85.02 4 0.10 0.08–0.13 3
(75–66)
2a Azuero Complex s. PAN
Azuero Plateau Coniacian-early OP
Santonian 89–85
depleted basalts 0.97A 33: 1, 29 22.20 17.05–38.86 32 1.43 0.70–1.77 32 202.94 39.13–1350.00 32 14.44 2.70–95.86 32 0.07 0.06–0.22 32
enriched basalts NA 40.00 38.92–41.07 4.76 4.31–5.22 2 242.44 51.96–432.93 2 15.89 3.30–28.49 2 0.06 0.06–0.07 2
2ai Azuero Arc 73–69 PA, A 0.77 38: 13, 4 24.69 14.62–33.04 14 1.41 0.65–2.66 14 292.34 623.81–4092.71 14 32.47 9.11–63.60 14 0.13 0.08–0.46 14
2b Chagres-Bayano Arc 70–39 e. PAN A 1.02 73: 20, 10 15.25 5.69–24.47 24 0.68 0.17–4.21 23 806.54 65.22–6657.14 24 287.18 15.31–1696.82 23 0.43 0.09–1.74 23
PISI Tr2
3 Serranía de Baudó 79–72 w. COL OP
Complex
depleted basalts NC 23.18 15.72–23.18 18 1.35 0.92–1.96 6 118.96 83.87–225.00 6 22.54 4.29–75.38 18 0.12 0.07–0.24 6
enriched basalts 0.66 3: 1, 2 47.64 45.08–48.94 3 5.69 1 875.63 1 36.92 6.58–70.41 3 0.08 1
4 Gorgona Island 91–81 w. COL OP
depleted
bas NC 15.72 1 0.66 0.55–0.78 3 339.24 121.82–556.67 2 10.76 9.38–12.10 3 0.06 0.02–0.09 2
pics NC 11.71 10.65–12.96 6 0.22 0.16–0.34 4 225.28 60.00–313.33 3 23.40 7.20–40.87 4 0.11 0.09–0.13 3
koms NC 16.23 11.94–23.68 14 0.40 0.29–0.54 9 2231.75 85.33–7885.00 4 86.82 4.91–358.41 9 0.12 0.05–0.31 4
enriched basalts NC 30.70 17.08–41.24 3 8.07 4.42–11.24 4 74.94 1 3.36 1.68–5.00 4 0.06 1
4(S) Gorgona Island 101–59
depleted basb 69–59 NC 17.42 1 1.82 1 245.46 1 14.48 1 0.06 1
depleted koms assumed to NC 15.54 14.87–16.22 2 0.28 0.28–0.29 2 154.39 106.55–202.23 2 10.42 7.22–13.62 2 0.07 0.07–0.07 2
be Late
Cretaceous
enriched koms 87–84 NC 10.82 1 5.34 1 55.24 1 3.81 1 0.07 1
enriched basalts 71–68 NC 18.56 1 1.93 1 44.92 1 2.83 1 0.06 1
gabbro 101–96 NC 15.58 1.53 1 143.14 1 8.36 1 0.06 1
5a Pallatanga Unit 90–85 w. ECU OP
& othersc
basaltsd NC 21.65 16.51–28.14 34 1.65 0.75–1.92 27 279.20 19.00–2475.38 34 16.99 1.82–136.94 31 0.08 0.04–0.25 32
picrites NC 17.67 16.58–19.09 4 0.92 0.75–1.36 4 232.52 140.75–398.00 4 9.81 6.70–13.23 4 0.04 0.03–0.05 4
ankaramites NC 21.86 18.85–26.10 6 3.58 1.80–6.42 5 256.93 31.52–604.55 6 14.08 1.74–41.30 6 0.07 0.05 6
5b Rio Cala Arc 85–64 w. ECU A 1.16 12: 3, 4 14.66 13.55–15.80 3 1.92 0.29–2.51 8 201.98 103.33–247.60 8 147.91 34.44–222.06 8 0.70 0.33–0.90 8
NA Macuchi Arc 44–33 w. ECU A NC 17.63 14.87–20.07 3 0.61 0.44–0.86 3 288.31 78.05–605.26 3 183.75 32.99–466.22 3 0.49 0.29–0.77 3
6a Central Cordillera 97–86 w. COL OP
basalts 1.49 23: 1, 11 20.39 7.70–36.91 22 2.68 0.54–7.24 4 78.01 60.47–104.65 3 8.03 0.73–21.43 21 0.09 0.08–0.10 3
picrites NC 34.29 15.97–80.72 17 6.66 1.55–19.52 15 224.38 36.84–603.64 4 15.55 2.79–59.52 17 0.15 0.04–0.41 4
 6b Western Cordillera 94–75 w. COL OP NC 22.61 18.44–36.43 26 1.49 0.77–1.89 12 399.98 43.55–1380.00 9 40.70 6.28–172.50 12 0.08 0.07–0.10 9
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-

6bi Volcanic, Amaime, 101–98 w.COL OP 18.74 14.72–22.72 9 1.69 1.24–2.71 9 513.88 48–1975 9 37.14 4.02–158.84 9 0.07 0.05–0.10 9
& Barroso Fms.
(of the Western
Cordillera)
6bii Bólivar UM 99–87 w. COL OP NA 22.76 1 1.68 1 47.45 1 3.26 1 0.07 1
Complex (WC)
basalt
PISI Tr3
7 Cabo de la Vela 83–65 n. COL/LA A
MUMC
lower gabbros 12.51 11.70–13.32 2
(basaltic–andesites)
B
lastest stage dykes 1.69 4: 2, 1 17.27 13.72–20.81 2 0.43 0.26–0.60 2 571.50 461.50–681.50 2 485.60 426.00–545.20 2 0.86 0.80–0.92 2
(basaltic–andesites)
8 Aruba LA
8a Aruba Lava Fm. Turonian, 94–89 OP NC 1.95 1.76–2.15 7 156.43 83.78–332.43 7 11.48 6.74–26.74 7 0.07 0.07–0.08 7
(ALF)
9 Curaçao LA
9a Curaçao Lava Fm. mid-Albian OP
(CLF) (~110), 120–78
basalts NC 18.65 17.29–24.50 14 2.09 1.73–2.60 12 73.04 28.55–111.05 3 9.43 1.94–59.68 14 0.06 0.05–0.07 3

S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

picrites NC 20.76 17.67–26.85 12 2.67 2.13–5.00 8 93.19 19.92–204.42 4 7.16 1.26–13.86 12 0.08 0.06–0.10 4
basalt 78–74 NC 22.62 1 2.19 1 45.38 1 1.97 1 0.04 1
9a (L)R 94–62
basalt 93–92 NC 17.69 1 1.99 1 107.38 8.13 1 0.08
basalt 77–72 NC 19.96 1 1.83 1 88.96 1 5.81 1 0.07 1
basalts 64–62 NC 19.32 18.93–19.71 2 1.56 1.40–1.72 2 65.90 47.87–83.94 6.13 3.33–8.94 2 0.09 0.07–0.11 2
Center of Caribbean Plate
11 CLIP, Leg 15 97–87 Carib. Sea OP
basalts NC 1.41 0.49–1.81 10 44.41 33.33–49.58 10 3.21 2.44–3.74 10 0.07 0.06–0.08 10
enriched basalts NC 9.62 9.22–10.02 2 25.48 25.00–25.96 2 2.06 1.99–2.13 2 0.08 0.08–0.08 2
R
12 Beata Ridge 82–75, 56–54 Carib. Sea OP
depleted dolerites 82–78, 56–54 23.47 23.22–23.73 1.53 1.16–1.89 2 66.03 47.70–84.37 2 5.81 5.21–6.40 2 0.09 0.08–0.11 2
enriched basalts 79–75 70.91 43.06 1 115.33 1 2.85 1 0.02 1
dolerites (relative 79–78 30.31 21.33–39.30 2 2 22.23 12.92–31.54 2 2
‘depletedness’
unknown)
13 CLIP, Site 1001 82–80 Carib. Sea OP NC 22.77 20.62–25.64 14 0.47 0.36–0.57 14 161.24 60–921.43 14 8.41 2.93–52.02 14 0.05 0.04–0.06 14
basalts
Northern Caribbean Plate
R
14 Dumisseau 110–82 Haiti OP
Formation
110–100 NA 21.19 1 1.04 1 60.00 1 6.07 1 0.10 1
96–82 1.07 12: 1, 2 46.44 38.65–53.12 12 5.19 4.54–5.93 12 50.44 33.33–62.96 12 3.69 2.39–4.56 12 0.07 0.07–0.08 12
15R Lower Duarte 88–85 Dom. Rep. OP 6.93 4.69–9.18 2 0.06 0.06–0.07 2
Formation

Tectono- THI n Ti/V Nb/Yb Ba/Th Ba/Nb Th/Nb

magmatic
mean TNSS: Fe4.0, mean range n mean range n mean range n mean range n mean range n
suite
Fe8.0

OPB 1.21 353: 7, 152 27.96 17.40–50.51 351 2.33 0.24–11.45 287 77.75 23.21–621.88 27 5.52 2.0–38.0 27 0.08 0.03–0.13 299
OIB NC 75.00 22.22 87.50 7.29 0.08
N-MORB NC 35.00 0.76 52.50 2.70 0.05
SIRO 0.96 58: 10, 13 29.51 12.20–48.43 34 0.51 0.18–1.38 12 50.05 8.11–188.89 12 5.05 0.55–22.67 14 0.10 0.06–0.15 12
lower
IBMFAB 1.04 18: 1, 9 15.22 12.03–18.45 12 0.52 0.35–0.70 12 89.28 20.77–337.50 12 7.15 2.89–23.28 12 0.09 0.07–0.19 12
SIRO 1.03 66: 16, 12 13.44 8.50–23.71 11 0.39 0.13–0.93 9 230.03 75.00–418.52 8 53.93 17.39–87.72 9 0.26 0.11–0.32 8
upper
CAVA 0.83 583: 287, 50 21.32 9.99–57.80 146 2.48 0.38–12.21 97 600.46 37.04–2085.85 101 191.58 2.33–750.91 110 0.37 0.03–1.44 91
IBMVA 1.19 1436: 836, 24 16.37 8.45–21.96 138 0.37 0.10–3.38 138 478.48 26.79–1110.42 163 168.70 49.00–1023.53 195 0.55 0.12–2.33 137

9
10 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

have accompanied the initiation of subduction at 96.3 ± 0.4 Ma (Smith
et al., 1999).
4.2. Changes in magma composition at and after arc inception
We cannot yet resolve all of the details of how the three subduction
zones (i.e., along Tr1, Tr2, Tr3) formed and whether or not these three
zones formed in a single event and shared a common early history.
Nevertheless, a record of how their mantle sources evolved is available
in the geochemistry of their lavas coupled with geochronological con-
straints. Below, we use geochronologically-constrained chemical com-
positions to probe the mantle source of PAR melts to elucidate how
the following tectonically-sensitive characteristics changed with time:
(1) Tholeiitic vs. calc-alkaline magmatic differentiation as resolved by
SiO2 vs. FeOt/MgO and Na2O + K2O–FeOt/MgO (AFM) relations and
tholeiitic index (THI, Section 4.2.1). Differentiation of subalkaline igne-
ous suites following the distinctive calc-alkaline trend of no iron enrich-
ment as silica increased reveals a clear supra-subducton zone (SSZ)
tectonic environment. These differences largely reflect the much higher
magmatic water contents in arc magmas. (2) Mantle source depletion
and relative degrees of partial melting that generated these melts, as
gauged by Nb/Yb (Section 4.2.2). Arc magmas are generated by melting
mantle that is much more depleted than that which generates MORB
and OIB. (3) Oxygen fugacity of the mantle source as monitored by Ti/
V (Section 4.2.3). Peridotite xenoliths show that the oxygen fugacity of
the uppermost mantle furthest away from subduction zones – for exam-
ple that of mantle plumes – has oxygen fugacities equal to or 1–2 log
units lower than the fayalite–magnetite–quartz (FMQ) oxygen buffer,
whereas arc peridotite is more oxidized, up to 4 log units above FMQ
(Frost and McCammon, 2008; Evans et al., 2012). If the source region
received simultaneous plume and subduction contributions this may
be reflected in variations of Ti/V ratios. Finally, (4) Relative contribu-
tions from the subducted slab as identified by subduction markers of
Ba/Th, Ba/Nb and Th/Nb (Section 4.2.4). Progressive squeezing and
heating of subducted oceanic crust and sediments firstly releases hy-
drous fluids which are followed by sediment melts. Trace elements
such as Ba and Th are sensitive to such inputs. We carry out these geo-
chemical interrogations below using the geochemical proxies outlined
above. Table 3 provides the mean and range of these proxies (THI, Nb/
Yb, Ti/V, and Ba/Th, Ba/Nb, Th/Nb) for unit lavas of this study; Table 3
also provides the means and ranges of these proxies for OPB, VAB, N-
MORB, OIB and subduction initiation basalts (see Section 3.2) for com-
parison with plume- and arc-related units of this study.
Our PISI model (Section 5) predicts that SI occurred after 100 Ma due
to lithospheric weakening accompanying the arrival of the volumetri-
cally dominant post-100 Ma Caribbean mantle plume and that evidence
of this SI event is preserved in PAR sequences along the three ~1000 km
long traces: Tr1 in NW Costa Rica through Panama; Tr2 from western
Ecuador through Colombia; and Tr3 through the westerly Leeward
Antilles. PAR units in these three traces should have chemical composi-
tions that are: (1) similar to each other and distinct from ‘normal’ OPB
elsewhere; and (2) distinct from the pre-100 Ma unit (e.g., 140–
110 Ma Nicoya Complex fragment) and post-100 Ma units in regions
further north where subduction modification of the mantle plume is
unlikely to have occurred (e.g., ODP drill sites in the middle of the
Caribbean Plate, central Hispaniola in the northern segment of the
CLIP). Our model thus further predicts that both the older Nicoya
Complex and CLIP units drilled by ODP to be more like global OPB
than post-100 Ma PAR units exposed along the three PISI traces.
Below, we test the PISI model using the aforementioned key geo-
chemical proxies to highlight progressive post-100 Ma changes in the
melting source of PAR units with time, including tholeiitic vs. calc-
alkaline affinities (Section 4.2.1), partial melting and source fertility
(Section 4.2.2), source oxygen fugacity (Section 4.2.3), and subduction
component additions (Section 4.2.4). Constraints from alteration-
resistant radiogenic isotopes (Nd and Pb) are summarized in Section 4.3.
4.2.1. Temporal variations of tholeiitic and calc-alkaline affinities
Magmatic differentiation trends with or without iron enrichment
provide information about tectonic environment of formation. Igneous
sequences with calc-alkaline affinities (no iron enrichment) are diag-
nostic of convergent margin suites whereas basalts from multiple
tectonic environments (including convergent margins) show tholei-
itic (iron enrichment) trends. Thus, arc units are sometimes tholeiitic
but calc-alkaline suites are invariably subduction-related (Barbarin,
1999). We use SiO2 vs. FeOt/MgO (Miyahsiro, 1974; Arculus, 2003)
and Na2O–FeOt–MgO (AFM, total alkalis–iron–magnesium) (Kuno,
1968; Irvine and Baragar, 1971) relations and the tholeiitic index
(THI) (Zimmer et al., 2010) to discern tholeiitic vs. calc-alkaline affini-
ties through time for plateau- and arc-related units of this study.
SiO2 vs. FeOt/MgO relations demonstrate that most lavas and intru-
sives of basaltic or sub-basaltic composition from units interpreted
as CLIP overlap the tholeiitic–calc-alkaline subdivision of Miyashiro
(1974) and plot as medium-Fe subalkaline series of Arculus (2003)
(Fig. 4a). On the AFM plot (Fig. 4b) all Caribbean units display obvious
Fe-enrichment trends (Fig. 4b) however, a minority of PAR basalts ex-
posed along Tr1, Tr2 and Tr3 plot as calc-alkaline. Moreover, composi-
tions of rocks along these traces evolve to calc-alkaline affinities with
time and increasing differentiation (see also Fig. 12a which demon-
strates the calc-alkaline affinity of Late Cretaceous felsic, arc-like units
from western Ecuador and Colombia and the Leeward Antilles along
Tr2 and Tr3). For example, whereas most of the circa 100–75 Ma
Ecuadorian and Colombian samples plot within the medium-Fe
series field, the 79–72 Ma Serranía de Baudó Complex (westernmost
Colombia) plot almost entirely within the low-Fe (calc-alkaline) field
(Fig. 4a, second plot from the top). It is noteworthy that the oldest
units and those of the central and northernmost Caribbean Plate exhibit
the strongest Fe-enrichment trends. The only units which have basalts/
basaltic intrusives that display nearly vertical trends in SiO2 vs. FeOt/
MgO space and which plot (almost) entirely within the high-Fe series
are ones from the Nicoya Complex in Costa Rica and the Beata Ridge
and Hispaniola, from the center and northernmost segments of the

Notes to Table 3:
This table. Tholeiitic index (THI), Nb/Yb, Ti/V, Ba/Th, Ba/Nb and Th/Nb of volcanic rocks of Caribbean plume- and arc-related units from this study which are plotted in Figs. 4–9. Also listed
are these geochemical proxies of oceanic plateau basalts (OPB), oceanic island basalts (OIB), N-MORB, lower and upper units of subduction initiation rule ophiolites (SIRO lower, SIRO
upper), Izu-Bonin Mariana forearc basalts (IBMFAB), the Central American Volcanic Arc (CAVA) and the Izu-Bonin Mariana Volcanic Arc (IBMVA) volcanic arc. References for the Caribbean
units are provided in Table 2 and references for OPB, SIRO, IBMFAB, CAVA and IBMVA are listed in Section 3.2. Data for OIB and N-MORB are from Sun and McDonough (1989).
Footnotes:
Note that all radiometric ages listed here and elsewhere in the text encompass calculated ages plus and minus their uncertainties rounded to the nearest Ma, unless stated otherwise
(e.g., as in Fig. 3). See Table 2 caption for other age details. Letters H, B, S, and L beside Unit/Complex numbers in column one are to denote that the data is from Hauff et al. (2000a),
Buchs et al. (2010), Serrano et al. (2011) and Loewen et al. (2013). Superscript R, beside Unit/Complex numbers in column one is to denote that all samples of these units have correspond-
ing radiometric ages. Superscripts beside unit names in the Complex/Unit (second) column: a, as all Tortugal alkali basalts and picrites have total oxides b97 wt.% and thus do not pass our
total oxides filter of 97–102 wt.% (see GR Online Supplementary Document), ratios were not calculated for these; b, Serrano et al. (2011) interpret this 69–59 Ma basalt as an enriched
basalt but its flat chondrite-normalized REE pattern suggests that is a depleted basalt; c, ratios presented include those for other basement considered as correlatives of the circa 90–
85 Ma Pallantanga Unit including the Piñon, Pedernales and Guaranda units (see Mamberti et al., 2003); d, one Pallatanga Unit sample (97 Ma13, Mamberti et al., 2003) with anomalously
high Ba (4779 ppm) has been omitted from our calculations as we assume this to be a typo, e.g., the remaining samples exhibit a mean of 57 ppm Ba with a range from 8 to 322 ppm Ba.
Superscripts beside THI mean in THI column: A, as no Azuero depleted basalts have 3–5 wt.% MgO, calculation of THI was done with a single sample with 5.30 wt.% MgO; B, Fe4.0 and Fe8.0
were calculated via upper dykes and lower gabbros, respectively.

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 11

a b
units interpreted as:
oceanic plateau: arc:
depleted basalt, picrite, komatiite basalt
LREE-enriched (OIB-like)
t
8
komat. &
basalt bas. and. andesite dacite & rhyolite
FeO
picrite
7
Costa Rica (1a-d) & Panama
(2ai, 2aii, 2b) thol
6

high-Fe
thol
5

Fe
-e
nri
4
Tr1

ch
med-Fe

m
en
3

t
2 calc-alk
low-Fe calc-alk
1 Symbol color key:
CR PAN

8
western Colombia (3,4, 6c, d)
7 & western Ecuador (5c, d)

6
thol
thol
5
Tr2
4

3
calc-alk
2
FeO t/MgO

Symbol color key:

COL: WC, CC, SDB
1 COL: Gorgona
ECU calc-alk
8
Leeward Antilles (Aruba, 8a,
7 Curaçao, 9a & Cabo de la Vela
Ultramafic Complex, Guajira
Peninsula, Colombia, 7)
6 thol thol
5
Tr3
4

2 calc-alk
calc-alk
1 Symbol color key:
Curaçao Aruba

8
Caribbean Sea (11-13) & Central Hispaniola (Haiti, 14)
7

6 thol
thol
5

2 calc-alk
calc-alk
1 Symbol color key:
Caribbean Sea Haiti

40 45 50 55 60 65 70 75 80
Na2O+K2O MgO
SiO2 wt. %

Fig. 4. (a) SiO2 vs. FeOt/MgO (Miyashiro, 1974) and (b) AFM (Kuno, 1968; Irvine and Baragar, 1971) relations of lavas and intrusives of PAR units exposed along: Tr1 in Costa Rica and
Panama; Tr2 in western Colombia and Ecuador; Tr3 in Aruba and Curacao; and the central and northern segments of the CLIP in the Caribbean Sea and Hispaniola (Haiti, Dominican
Republic). Superimposed on (a) are the low-, medium- and high-Fe series of Arculus (2003). In (a) and (b) (top plots) the dashed, gray shaded and yellow regions represent distributions
of samples from the 140–110 Ma and 98–82 Ma Nicoya Complex units, respectively; in (a) and (b) (second plots from top), the dashed, gray shaded and yellow regions represent distri-
butions of the 101–98 Ma Volcanic and Amaime formations of the Western Cordillera (Colombia) and the 79–72 Ma Serranía de Baudó Complex, respectively; in (a) and (b) (bottom plots)
the dashed, pink shaded and yellow regions represent distributions of 82–80 Ma ODP Leg 165, Site 1001 Caribbean Sea basalts and 82–75 Ma and 56–54 Ma Beata Ridge (central Caribbean
Plate) Caribbean Sea lavas and intrusives, respectively. The white stars in (a) and (b) (top plots) represent the only dated sample (AN86, 94.7 ± 0.9 Ma) available from the younger 98–
82 Ma Nicoya Complex. In (b) (top plot) one sample from the 98–82 Ma Nicoya Complex and two samples from the 70–39 Ma Chagres–Bayano Arc (Panama) with N70 wt. SiO2 fall outside
of the plot with FeOt/MgO N 8. In (b) (bottom plot) one Duarte Formation intrusive with 54 wt.% SiO2 (described as a diorite) plots outside the plot with FeOt/MgO N 8. See Fig. 11 for SiO2
vs. FeOt/MgO relations of circa 95–79 Ma subduction-related, felsic, arc-like intrusives of NW South America and the Greater Antilles. The numbers in italics in brackets beside the region or
unit name corresponds to the unit numbers assigned in Fig. 1.

CLIP, respectively. Some post-100 Ma Nicoya lavas and intrusives (yel- (AN86, 94.7 ± 0.9 Ma, white star) from the younger unit has corre-
low shade, Fig. 4a, top plot) also plot along a vertical trend in FeOt/ sponding major and trace element chemistry (one dated circa 88 Ma
MgO space alongside 140–110 Ma Nicoya lavas (dashed gray shade, plagiogranite sample has corresponding chemistry but total oxides
Fig. 4a, top plot); however, we note that only one dated sample b97 wt.% and thus fails our chemical filter, see Section 1 of

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
12 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

Supplementary Data). We note also that the only LREE-enriched lavas or
intrusives reflecting lower degree partial melting (e.g., Dumisseau For-
mation basalts, Loewen et al., 2013) occur in units in the central and
northern segments of the Caribbean Plate furthest away from the trench
and which received nil subduction additions (see below).
The THI allows iron enrichment or depletion of a given magma series
to be quantified (THI N and b 1 are tholeiitic and calc-alkaline respec-
tively). THI is calculated for a magma series and is defined as Fe4.0/
Fe8.0, where Fe4.0 and Fe8.0 represent the average FeOt of magmas with
3–5 and 7–9 wt.% MgO respectively (Zimmer et al., 2010). The applica-
bility of THI is limited by the scarcity of evolved (~4 ± 1 wt.% MgO) PAR
igneous rocks; we were only able to reconstruct THI for nine PAR suites.
Nonetheless, a plot of THI vs. time (Fig. 5) for units where THI could be
calculated demonstrates similarities to trends seen in SiO2 vs. FeOt/MgO
and AFM space and two important points. First, the oldest post-100 Ma
units on the southern side of the Caribbean Plate and NW South America
(98–82 Ma components of the Nicoya Complex and the 97–96 Ma
Central Cordillera, western Colombia) are clearly tholeiitic, with iden-
tical THI of ~ 1.50. Second, THI decreased with time along the south-
western margin of the CLIP. By 89–85 Ma, THI decreased to b1 as
recorded by the Azuero Plateau and to an OPB-like THI of ~1.2 as record-
ed by the Golfito complex (samples interpreted as plateau by Hauff
et al., 2000a,b and as arc by Buchs et al., 2010 represented by a dark
purple square and light purple circle respectively, on Fig. 5). The Azuero
Plateau records a slightly calc-alkaline THI of 0.97 (Table 3), much lower
than average OPB (THI ~ 1.2). Along with other geochemical criteria
described below, this feature suggests a hybrid plume-subduction
environment of formation, which in the case of the Azuero Plateau,
was achieved by 89–85 Ma.
4.2.2. Temporal variations of mantle depletion and degrees of partial
melting as recorded by Nb/Yb
Nb is a highly incompatible trace element whereas Yb is a moderate-
ly incompatible element and neither is transported by hydrous fluids
released from the subducted slab. Consequently, the Nb/Yb ratio de-
creases as the mantle source region is progressively depleted by melt-
ing, so basalts with low Nb/Yb are derived from more depleted mantle
Costa Rica (1a, 1dii), Panama (2ai, 2aii, 2b), western
Colombia (6a), western Ecuador (5b) & Haiti (14)
oceanic plateau:
units interpreted as:
arc:
depleted basalt basalt
1.6
1.5
SI (Wright and Wyld, 2011,
1.4 SI (Buchs et al.,
Aruba & Curacao) 2010, Panama &
1.3 SI (Vallejo et al., 2009, SW Costa Rica)
western Ecuador)
1.2
THI
than are basalts with higher Nb/Yb (assuming basalts were not generat-
ed in equilibrium with garnet). It is not clear from Nb/Yb alone whether
source depletion reflects a single event (perhaps at the time that the
melt was generated) or the cumulative effects of multiple melting epi-
sodes. Fig. 6 shows that OIB and OPB are derived from a mantle source
that is less depleted (higher Nb/Yb) than is the source region of MORB
(lower Nb/Yb), which is less depleted than the source region of arc ba-
salts like those of the IBMVA (lowest Nb/Yb) (Pearce et al., 2005). Nb/Yb
variations with time can thus be used as one perspective on the transi-
tion from plume to subduction-modified sources. Thus, we plot Nb/Yb
as a proxy of mantle fertility and the relative degree of partial melting
vs. time in Fig. 6 to highlight both differences in the degrees of partial
melting experienced by pre-vs. post-100 Ma units and to demonstrate
that the mantle source was increasingly depleted with time along
both the southern and eastern margins of the Caribbean Plate beginning
at 100 Ma.
The 140–110 Ma Nicoya plateau unit was derived from fertile man-
tle (mean Nb/Yb = 1.8) (Fig. 6a, Table 3) slightly less than mean OPB
(mean Nb/Yb = 2.3). It is important to note that Nb/Yb scatters widely
for 95 Ma and younger basalts erupted away from the three traces, but
most of the scatter is to higher, undepleted values (Fig. 6d). In contrast, a
shift to lower Nb/Yb (more depleted source and/or higher degree melt-
ing) with time is seen in Tr1 (Costa Rica and Panama) (Fig. 6a) and Tr2
(Fig. 6b). Nb/Yb variations are more complicated for Tr2 through west-
ern Colombia and western Ecuador which in part reflects the large var-
iation in partial melting experienced by Gorgona basalts, picrites and
komatiites (Fig. 6b) (e.g., Kerr et al., 1996a). Tr3 PAR sequences in
Aruba and Curaçao also trend toward lower Nb/Yb with time (Fig. 6c)
but only a 20 Ma record is preserved. The central and northern seg-
ments of the CLIP show great scatter and no clear trend with time how-
ever, apart from the more depleted mantle source for 82–80 Ma
Caribbean Sea basalts relative to that of 97–87 Ma Caribbean Sea basalts
(Fig. 6d).
Overall, Nb/Yb decreases with time from an early plume-like to a
later arc-like phase, supporting the idea that the mantle source region
was progressively depleted over this tectonomagmatic transition. A
similar trend is seen in Nb/Y vs. Zr/Y (see GR Online Supplementary
Document and Supplementary Fig. 3). Nb/Yb variations are thus consis-
tent with the PISI model. An important point is that a transition to lower
Nb/Yb and hence greater degrees of partial melting is evident by at least
95 Ma (Fig. 6a, b) which suggests that a significant change in the melt-
ing regime occurred between 100 and 95 Ma which is at least 6 Ma
younger than current interpretations of the transition from plume-
to arc (i.e., timing of SI as indicated by faint pink, blue and purple
vertical bars in Figs. 5, 6 and subsequent figures). A spatial trend in de-
grees of partial melting has also been noted by Loewen et al. (2013) who
OPB
demonstrated a lower degree of partial melting for Dumisseau
IBMVAB
Formation (Haiti) basalts relative to CLIP lavas further to the south

1.1 iron depletion tholeiiitic

(e.g., those of the Curaçao Lava Formation). As we show below, these
1.0
calc-alkaline
trends toward increasing degrees of partial melting with time and ac-
0.9
Symbol color key: CAVAB cording to location (greater degrees of melting in southern and eastern
0.8 CR PAN
COL (CC) ECU PAR units relative to CLIP in the center and northerly segments of the
0.7 Haiti
Caribbean Plate) are mirrored by concomitant trends toward higher
0.6
140 130 120 110 100 90 80 70 60 50 40 subduction contributions with time.
145 135 125 115 105 95 85 75 65 55 45 35

Ma 4.2.3. Temporal variations of mantle fugacity as recorded by Ti/V

Fig. 5. Tholeiiitic index (THI) (Zimmer et al., 2010) vs. time for post-100 Ma plume- and
arc-related units along trace 1 in Costa Rica and Panama and trace 2 in western Ecuador
and western Colombia. The faint pink, blue and purple vertical bars demarcate the timing
of inferred subduction initiation along: trace 3 in Aruba and Curacao (89 Ma, Wright and
Wyld, 2011) on the basis of ages of felsic intrusives of the Aruba Batholith and Curaçao
Diorites which intrude the Aruba and Curaçao lava formations (interpreted as CLIP);
trace 2 in western Ecuador (85 Ma, Vallejo et al., 2009); and trace 1 in SW Costa Rica
and Panama (75 Ma, Buchs et al., 2010), respectively; see text for other details. Note
that the Golfito Complex is interpreted as oceanic by Hauff et al. (2000a,b) (purple square)
and as arc by Buchs et al. (2010). Unit numbers (in italics in brackets) are as assigned as in
Fig. 1.
Ti/V ratios are sensitive to fluctuations in fO2 (Shervais, 1982) and
the transition from plume to SSZ environment should be reflected in
progressively more oxidized mantle. The basis for the Ti/V relation lies
in multiple oxidation states of V in magmas (+ 3, + 4, and + 5), so V
changes from a slightly to a strongly incompatible element as mantle
oxygen fugacity increases (Canil, 1997). Ti is + 4 and is not redox-
sensitive. Low Ti/V (b 20) in basalt indicates an oxidized mantle source,
as is most commonly associated with SSZ environments, whereas high
Ti/V indicates reduced mantle sources akin to those of MORB and OIB
(Shervais, 1982). We plot Ti/V versus time for PAR units (Fig. 7) to

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 13

units interpreted as: examine if and how the oxygen fugacity in the mantle source changed
oceanic plateau: arc:
depleted basalt, picrite, komatiite basalt with time.
enriched &/or alkalic basalt, picrite, komatiite, ankaramite Ti/V decreases markedly from OIB (~ 75) to MORB (~ 35, Shervais,
relative ‘depletedness’ unknown or ambiguous
1982) and OPB (28, n = 351) (Table 3, Fig. 7). Arc basalts typically
PISI trace 1 have Ti/V b20, e.g., the IBM volcanic arc basalts record a mean Ti/V of
30
OIB 16 and range from 8 to 22 (n = 138). Quaternary Central American Vol-
Costa Rica (1a-d) & Panama Symbol color key:
(2ai, 2aii, 2b) CR canic arc basalts record a slightly higher mean (21) and a much larger
10 PAN
range (10–58) than ‘typical’ arc basalts, possibly reflecting the inherited
source fertility & relative CAVAB plume component in the mantle source of Central American Arc lavas
degree of partial melting
(Gazel et al., 2009, 2011) (Table 3).
Nb/Yb

OPB
Nicoya Mean Ti/V of most suites along Tr1 (Costa Rica and Panama) is
1 Complex
melting regime
MORB similar to that of the mean Ti/V = 21 of modern Central American Arc
plume-only: deep, low degree, ~93-85 Ma
basalts (Fig. 7a, Table 3). Although some units exhibit large ranges
plume only: shallow, intermediate IBMVAB
degree, >110 Ma (e.g., Ti/V ranges from 17 to 39 for depleted basalts of the Azuero
hybrid-plume/subduction flux:
shallow, high degree, ~90-70 Ma a Plateau, Table 3), the mean Ti/V generally increases with time until
0.1 about 70 Ma after which Ti/V drastically falls (Fig. 7a, Table 3). Only
PISI trace 2
30
the 70–39 Ma Chagres–Bayano Arc of Panama shows a significantly
OIB lower Ti/V of 15, similar to that of IBMVA basalts (elevated Ti/V of 60
western Colombia (3,4, 6c, d) Symbol color key:
10
& western Ecuador (5c, d) COL: WC, CC, SDB and 40 are for a single Tortugal basalt of unknown ‘depletedness’ and
COL: Gorgona
ECU for the enriched Azuero Plateau basalt suite, n = 2, Table 3). Tr2 Ti/V in-
CAVAB creases with time in a similar manner to Tr1 if one considers only the
Nb/Yb

OPB mean Ti/V of Colombian units but the scatter in Ti/V of 90–80 Ma
140-110 Ma Nicoya Complex
to 33 Ma
Colombia lavas complicates the evolution in general (Fig. 7b). Variabili-
1
MORB ty here is again dominated by large ranges of Ti/V in Gorgona lavas. For
example, depleted Gorgona basalts and komatiites display similar
IBMVAB
means of Ti/V = 16, identical to IBMVA basalts and Gorgona picrites
b exhibit the lowest mean Ti/V of all units (12) (Table 3) similar to Ti/V
0.1
of boninites. Ecuador PAR sequences evolve to low, arc-like Ti/V by
PISI trace 3
30 ~ 80 Ma, when the Rio Cala arc formed (Fig. 7b). Ti/V for Tr3 PAR
OIB
Leeward Antilles (Aruba, 8a, Curaçao, 9a & Cabo de la Vela
sequences shows little change, ranging only from ~ 18–20 for all 5
10 Ultramafic Complex, Guajira Peninsula, Colombia, 7) sequences examined (Fig. 7c). CLIP sequences away from the three
Symbol color key:
Curaçao
PISI traces appear to decrease with time but never reach arc-like values
CAVAB
Aruba & CVC b20 (Fig. 7d).
Nb/Yb

OPB
Unfortunately, there are no V analyses available for the 140–110 Ma
140-110 Ma Nicoya Complex
1 Nicoya OPB but it is important to note that Ti/V is much higher than typ-
MORB ical arc basalts (b20) and quite variable away from the three traces
IBMVAB (Fig. 7d). For example, mean values range from ~70 for enriched basalts
to ~25 for depleted dolerites from the Beata Ridge in the center of the
c Caribbean Plate (Table 3). Similarly, lavas of the Dumisseau Formation,
0.1
Haiti & Dominican Rep. Haiti record Ti/V that range from 33 to 63 (Table 3). As these values
center of CLIP, Caribbean Sea are much higher than those of typical arc basalts this indicates deriva-
30 tion from a mantle source that was more reduced than that supplying
OIB
Caribbean Sea (11-13) & 43.06 Symbol color key: arc magmas.
Central Hispaniola (Haiti, Caribbean Sea
10
14 & Dominican Rep., 15) Hispaniola That Ti/V remained relatively high along Tr1 and Tr2 (generally
higher than 20) but decreased erratically to lower, arc-like values b20
CAVAB
especially after 75 Ma may indicate that the plume mantle source was
Nb/Yb

OPB
not strongly oxidized by subduction until ~ 20 Ma after the main
140-110 Ma Nicoya Complex
1 higher degree of melting with time
plume phase. The PISI model does not readily explain why oxidation
MORB
in middle of plume
took so long but is possibly the result of extended plume contributions
IBMVAB until ~ 75 Ma, after which plume contributions appear to rapidly de-
crease (e.g., Fig. 3). We note that even modern Central American Arc
d lavas have slightly higher Ti/V than expected for arcs. Perhaps the
0.1
140 130 120 110 100 90 80 70 60 50 40 plume continued to supply reduced mantle which buffered the oxygen
Ma fugacity of the evolving mantle source.

Fig. 6. Mean and range of Nb/Yb vs. time for plume- and arc-related units from this study
(squares and circles represent units interpreted as oceanic plateau (OP) and arc, respec-
tively; see top box) from (a) PISI trace 1 in Costa Rica and Panama, (b) PISI trace 2 in
western Ecuador and western Colombia, (c) PISI trace 3 in the Leeward Antilles in the
Cabo de la Vela Complex, Curaçao and Aruba, and (d) the central and northern segments
of the Caribbean Plate in the Caribbean Sea and Central Hispaniola. The small squares in
(b) are Gorgona samples of Serrano et al. (2011). References, Nb/Yb ratios and ages for
all units in (a)–(d) are provided in Tables 2 and 3 for further correlation. References
for the oceanic island basalt (OIB), mid-oceanic ridge basalt (MORB), Central American
Volcanic Arc (CAVA) and the Izu Bonin–Marianas Volcanic Arc (IBMVA) are given in
Section 3.2. The faint purple, blue and pink vertical bars demarcate the timing of inferred
subduction initiation along traces 1, 2 and 3, respectively; see Fig. 5 caption and text for
further details. Unit numbers (in italics in brackets) are as assigned as in Fig. 1.
4.2.4. Temporal variations of subduction additions: Ba/Th, Ba/Nb, and
Th/Nb
Hydrous fluids released from the subducted slab carry fluid-mobile
elements (such as Rb, Ba, Sr, and Pb) up and into the overlying mantle.
Because the upper surface of the slab is progressively heated as it sinks
to greater depth, subducted sediments carried to depths of 100 km or
more can melt, transporting other incompatible trace elements such
as Th into the overlying mantle (Elliott et al., 1997). Fluid and melt are
sometimes referred to as “shallow” and “deep” subduction components,
respectively, and together comprise the total subduction component
that is released from the subducting plate. Fluids and melts released

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
14 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

units interpreted as: from the sinking slab ascend into the overlying mantle, metasomatizing
oceanic plateau: arc:
depleted basalt, picrite, komatiite basalt it and causing it to melt. Ratios of fluid- and melt-mobile Ba and melt
enriched &/or alkalic basalt, picrite, komatiite, ankaramite
relative ‘depletedness’ unknown or ambiguous (only)-mobile Th with trace elements sensitive to mantle composition
such as Nb or Yb determined in basalts can be used to monitor this
PISI trace 1 addition (Pearce et al., 2005). To better illustrate trends in subduction
80
OIB additions of post-100 Ma units described above, we plot Ba/Th (shallow
Costa Rica (1a-d) & Panama BC16 Symbol color key:
(2ai, 2aii, 2b) CR subduction additions), Th/Nb (deep subduction additions) and Ba/Nb
PAN
40 (total subduction additions) (Elliott et al., 1997; Pearce et al., 2005) vs.
MORB
30 mantle source fugacity
OPB
time in Figs. 8–10. As some studies suggest that Ba is an unreliable
petrogenetic indicator due to its potential mobility during hydrother-
CAVAB
Ti/V

20
mal alteration, we discuss why this appears not to be the case in Supple-
IBMVAB
mentary Section 1 of the GR Online Supplementary Document.
10
forearc The variation of these ‘subduction-sensitive’ trace element ratios in
boninite
post-100 Ma PAR units can be compared spatially and temporally to
a those in the center and northern segments of the Caribbean Plate and
5 with the 140–110 Ma Nicoya Complex, the latter two of which received
PISI trace 2
80 no discernible subduction inputs apart perhaps from the 82–80 Ma
OIB
to 81 Symbol color key: Caribbean Sea basalts. For example, the ODP Leg 165 basalts are some-
western Colombia (3,4, 6c, d) COL: WC, CC, SDB
& western Ecuador (5c, d) COL: Gorgona what anomalous recording a mean Ba/Th similar to IBM volcanic arc ba-
40 ECU
MORB salts and which range to values higher than mean Central American Arc
30
OPB lavas (Fig. 8d). Apart from these Leg 165 basalts, the shallow subduction
CAVAB component in central and northern Caribbean Plate igneous rocks is
Ti/V

20

IBMVAB
limited, with mean Ba/Th ~25–115, about the range expected for global
to 33 Ma MORB, OPB, and OIB (Fig. 8d). Apart from the Leg 165 basalts and post-
10 forearc 80 Ma Beata Ridge dolerites which show Ba/Nb that ranges to values in-
boninite
termediate to that of OIB and arc basalts (Fig. 9d) the total subduction
b component of Caribbean Sea and Hispaniola basalts is similarly modest,
5
PISI trace 3 with Ba/Nb ~2–28, again mostly similar to what is expected for global
80
MORB, OPB, and OIB (Fig. 9d). Little evidence of a deep subduction com-
OIB
ponent is observed, with Th/Nb b 0.1, again mostly similar to what is ex-
Leeward Antilles (Aruba, 8a, Curaçao, 9a & Cabo de la Vela
Ultramafic Complex, Guajira Peninsula, Colombia, 7) pected for global MORB, OPB, and OIB (Fig. 10d). Similarly, the Ba/Th,
40
MORB Ba/Nb and Th/Nb of 140–110 Ma Nicoya Complex basalts are very sim-
30
OPB ilar to OPB and OIB (Figs. 8a, 9a, 10a) and record no evidence of subduc-
CAVAB tion modification. The similarity in the older Nicoya Complex basalts
Ti/V

20
with OPB is also apparent in N-MORB normalized plots (Supplementary
IBMVAB
Symbol color key:
Fig. 4). The aforementioned ranges provide a useful baseline for com-
10 Curaçao forearc parison with PAR sequences along Tr1, Tr2, and Tr3.
CDVC boninite
In general, most volcanic rocks of post-100 Ma PAR units in Tr1 and
c Tr2 display evidence of shallow and total subduction additions as
5
Haiti & DR
gauged by elevated Ba/Th and Ba/Nb (Figs. 8, 9); conversely, post-
center of CLIP, Caribbean Sea 100 Ma units in Tr3 in the Leeward Antilles and the northern and central
80 segments of the Caribbean Plate generally do not. Exceptions include
OIB
Caribbean Sea (11-13) & Symbol color key: the Aruba Lava Formation basalts which have mean Ba/Th ~ 156 and
Central Hispaniola (Haiti, Caribbean Sea
Hispaniola
Ba/Nb ~11 (Table 3) similar to similarly-aged units along Tr1 and Tr2
14 & Dominican Rep., 15)
40
MORB and the (probably ~70 Ma) Cabo de la Vela ‘latest stage’ dykes which
30
OPB have Ba/Th and Ba/Nb higher than Central American Arc basalts
CAVAB (Figs. 8d, 9d). Tr1 PAR units have Ba/Th (~ 100–200) intermediate to
Ti/V

20

IBMVAB OIB and IBM volcanic arc basalts until ~ 70 Ma when Ba/Th rose to
modern day Central American Arc values of 200–600; Ba/Th reached
forearc
10
boninite 600–800 by the end of the Cretaceous in the Chagres–Bayano arc and re-
mains high in the modern Central American Arc (Fig. 8a, Table 3). The
d total subduction component (Ba/Nb) began to increase above values ex-
5
140 130 120 110 100 90 80 70 60 50 40 pected for OIB–OPB (~8) by 85 Ma (Ba/Nb of ~10–20) and rose to ~20–
Ma 60 by ~70 Ma (Fig. 9a, Table 3). Ba/Nb for Chagres–Bayano arc lavas is
indistinguishable from that for modern arc lavas from Central America
Fig. 7. Mean and range of Ti/V vs. time for plume- and arc-related units from this study or IBM arc basalts (Fig. 9a). Evidence of the deep subduction component
(squares and circles represent units interpreted as oceanic plateau (OP) and arc, respec- (Th/Nb) rose more slowly, exceeding baseline values for MORB, OIB, and
tively; see top box) from (a) PISI trace 1 in Costa Rica and Panama, (b) PISI trace 2 in west- OPB by ~70 Ma (Fig. 10a).
ern Ecuador and western Colombia, (c) PISI trace 3 in the Leeward Antilles in the Cabo de
la Vela Complex, Curaçao and Aruba, and (d) the central and northern segments of the
The record of subduction inputs preserved in Tr2 PAR sequences
Caribbean Plate in the Caribbean Sea and Central Hispaniola. The small squares in suggests that this mantle source was modified earlier by subduction
(b) are Gorgona samples of Serrano et al. (2011). References, Ti/V and ages for all units inputs and at a faster rate than was the Tr1 source (e.g., compare
in (a)–(d) are provided in Tables 2 and 3 for further correlation. References for the oceanic Figs. 8a and 9a with 8b and 9b). Although Colombia PAR sequences scat-
island basalt (OIB), mid-oceanic ridge basalt (MORB), Central American Volcanic Arc
ter widely, the oldest, circa 100 Ma western Colombia PAR sequence
(CAVA) and the Izu Bonin–Marianas Volcanic Arc (IBMVA) are given in Section 3.2. The
faint purple, blue and pink vertical bars demarcate the timing of inferred subduction initi- (western Cordillera) has mean Ba/Th (shallow subduction inputs)
ation along traces 1, 2 and 3, respectively; see Fig. 5 caption and text for further details. much higher (~ 650) than plume mantle (OIB and OPB, ~ 70–90 and
Unit numbers (in italics in brackets) are as assigned as in Fig. 1. 140–110 Ma Nicoya Complex, ~ 100) and even higher than Central

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 15

units interpreted as:

units interpreted as: oceanic plateau: arc:
oceanic plateau: arc: depleted basalt, picrite, komatiite basalt
depleted basalt, picrite, komatiite basalt enriched &/or alkalic basalt, picrite, komatiite, ankaramite
enriched &/or alkalic basalt, picrite, komatiite, ankaramite relative ‘depletedness’ unknown or ambiguous
relative ‘depletedness’ unknown or ambiguous
PISI trace 1
PISI trace 1
500
8000 Costa Rica (1a-d) & Panama to 1697 CAVAB
(2ai, 2aii, 2b)
Costa Rica (1a-d) & Panama Symbol color key:
(2ai, 2aii, 2b) CR IBMVAB
PAN 100 total subduction
shallow subduction component
1000

Ba/Nb
component Symbol color key:
Ba/Th

CAVAB CR
PAN
OIB
IBMVAB 10 Nicoya Complex
Nicoya OIB OPB
100 Complex
OPB
MORB
MORB
a a
1
20
PISI trace 2
PISI trace 2 500
8000 western Colombia (3,4, 6c, d) CAVAB
& western Ecuador (5c, d) to 33 Ma
western Colombia (3, 4, 6c, d) Symbol color key:
& western Ecuador (5c, d) COL: WC, CC, SDB IBMVAB
COL: Gorgona 100 Symbol color key:
ECU COL: WC, CC, SDB
1000 COL: Gorgona

Ba/Nb
ECU
Ba/Th

CAVAB
to 33 Ma
OIB
IBMVAB 10
OIB OPB
100 140-110 Ma Nicoya Complex 140-110 Ma Nicoya Complex
OPB
BUC MORB
MORB
BUC
b to 0.73 b
PISI trace 3
PISI trace 3 500
8000 Leeward Antilles (Aruba, 8a, to 545.2 CAVAB
Curaçao, 9a & Cabo de la Vela
Leeward Antilles (Aruba, 8a, Symbol color key: Ultramafic Complex, Guajira IBMVAB
Curaçao, 9a & Cabo de la Vela Curaçao Peninsula, Colombia, 7)
100
Ultramafic Complex, Guajira Aruba & CDVC
Symbol color key:
Peninsula, Colombia, 7) Curaçao
Ba/Nb

1000
Aruba & CVC
Ba/Th

CAVAB
OIB
10
IBMVAB
OIB OPB
100 140-110 Ma Nicoya Complex 140-110 Ma Nicoya Complex
OPB MORB
MORB c
c 1
20 Haiti
Haiti & DR center of CLIP, Caribbean Sea
center of CLIP, Caribbean Sea
500
8000 Caribbean Sea (11-13) & Central Hispaniola (Haiti, 14 & CAVAB
Dominican Rep., 15)
Caribean Sea (11-13) & Symbol color key: IBMVAB
Central Hispaniola (Haiti, Caribbean Sea 100
14 & Dominican Rep. 15) Hispaniola Symbol color key:
Caribbean Sea
Hispaniola
Ba/Nb

1000
Ba/Th

CAVAB
OIB
10
IBMVAB
OIB OPB
140-110 Ma Nicoya Complex 140-110 Ma Nicoya Complex
100
OPB MORB
MORB d
d 1
20 140 130 120 110 100 90 80 70 60 50 40
140 130 120 110 100 90 80 70 60 50 40 145 135 125 115 105 95 85 75 65 55 45 35
Ma Ma

Fig. 8. Mean and range of Ba/Th vs. time for plume- and arc-related units from this study
(squares and circles represent units interpreted as oceanic plateau (OP) and arc, respec-
tively; see top box) from (a) PISI trace 1 in Costa Rica and Panama, (b) PISI trace 2 in west-
ern Ecuador and western Colombia, (c) PISI trace 3 in the Leeward Antilles in the Cabo de
la Vela Complex, Curaçao and Aruba, and (d) the central and northern segments of the
Caribbean Plate in the Caribbean Sea and Central Hispaniola. The small squares in
(b) are Gorgona samples of Serrano et al. (2011). References, Ba/Th and ages for all units
in (a)–(d) are provided in Tables 2 and 3 for further correlation. References for the oceanic
island basalt (OIB), mid-oceanic ridge basalt (MORB), Central American Volcanic Arc
(CAVA) and the Izu Bonin–Marianas Volcanic Arc (IBMVA) are given in Section 3.2. The
faint purple, blue and pink vertical bars demarcate the timing of inferred subduction initi-
ation along traces 1, 2 and 3, respectively; see Fig. 5 caption and text for further details.
Unit numbers (in italics in brackets) are as assigned as in Fig. 1.
Fig. 9. Mean and range of Ba/Nb vs. time for plume- and arc-related units from this study
(squares and circles represent units interpreted as oceanic plateau (OP) and arc, respec-
tively; see top box) from (a) PISI trace 1 in Costa Rica and Panama, (b) PISI trace 2 in west-
ern Ecuador and western Colombia, (c) PISI trace 3 in the Leeward Antilles in the Cabo de
la Vela Complex, Curaçao and Aruba, and (d) the central and northern segments of the
Caribbean Plate in the Caribbean Sea and Central Hispaniola. The small squares in
(b) are Gorgona samples of Serrano et al. (2011). References, Ba/Nb and ages for all
units in (a)–(d) are provided in Tables 2 and 3 for further correlation. References for the
oceanic island basalt (OIB), mid-oceanic ridge basalt (MORB), Central American Volcanic
Arc (CAVA) and the Izu Bonin–Marianas Volcanic Arc (IBMVA) are given in Section 3.2.
The faint purple, blue and pink vertical bars demarcate the timing of inferred subduction
initiation along traces 1, 2 and 3, respectively; see Fig. 5 caption and text for further details.
Unit numbers (in italics in brackets) are as assigned as in Fig. 1.

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
16 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

units interpreted as: American Arc lavas (Fig. 8b, Table 3). 90 Ma Colombian units have Ba/Th
oceanic plateau: arc:
depleted basalt, picrite, komatiite basalt that are mostly higher than OIB and OPB and Ecuador units have Ba/Th
enriched &/or alkalic basalt, picrite, komatiite, ankaramite (~ 200) twice that of background OIB–OPB–MORB (b 100) (Fig. 8b,
relative ‘depletedness’ unknown or ambiguous
Table 3). Total subduction inputs monitored by Ba/Nb are also mostly
PISI trace 1 much higher for Tr2 PAR sequences (10–200) than plume mantle (b 8;
0.8
to 1.74
IBMVAB Fig. 9b).
Costa Rica (1a-d) & Panama
(2ai, 2aii, 2b)
Tr3 PAR sequences generally have mean Ba/Th, Ba/Nb and Th/Nb
CAVAB that scatter around baseline values for plume mantle (OIB, OPB, 140–
deep subduction
component
110 Ma Nicoya Complex), although Aruba Lava Formation basalts ex-
Th/Nb

Symbol color key:

hibit Ba/Th and Ba/Nb higher than plume and similar to post-100 Ma
CR Tr1 and Tr2 units (Figs. 8–10).
PAN OIB
0.1 In the GR Online Supplementary Document we also provide N-
OPB MORB- and primitive mantle-normalized plots (Supplementary
Nicoya Complex
MORB Figs. 4, 5) to demonstrate that many lavas of units interpreted as plateau
a exhibit high, arc-like LILE/HFSE ratios consistent with a SSZ formation.
0.03 Additionally as discussed in Section 5, circa 100–80 Ma felsic intrusives
PISI trace 2
0.8 that cut slightly older units interpreted as plateau exhibit unequivocal
to 0.90
IBMVAB subduction additions and HFSE depletions indicative of a supra-
to 33 Ma
western Colombia (3,4, 6c, d)
& western Ecuador (5c, d) subduction zone (SSZ) environment.
CAVAB
Th/Nb

Symbol color key: 4.3. Isotopic constraints

COL: WC, CC, SDB
COL: Gorgona
ECU OIB
0.1 We use isotopic data to evaluate whether or not the mantle source
OPB for PAR sequences can be linked to the mantle plume responsible for
140-110 Ma Nicoya Complex
MORB CLIP formation, which is widely accepted to be the Galapagos Plume
to 0.02 b (Duncan and Hargraves, 1984; Hauff et al., 1997). Sr isotopic com-
PISI trace 3
positions are easily affected by seafloor weathering and low grades of
0.8 metamorphism, so we neglect this to focus on alteration-resistant Nd
IBMVAB
Leeward Antilles (Aruba, 8a, mean, 0.86 and Pb systems. Hf isotopic data for PAR sequences would also be useful
from 0.80-0.92
Curaçao, 9a & Cabo de la Vela but there is little of this available yet. For these reasons, we focus on
Ultramafic Complex, Guajira CAVAB
143
peninsula, Colombia, 7) Nd/144Nd and 207Pb/204Pb that has been corrected for radiogenic
growth after eruption. Plots of initial, age-corrected 206Pb/204Pb versus
Th/Nb

143
Symbol color key: Nd/144Nd for the Caribbean igneous rocks of interest are shown on
Curaçao
Aruba & CDVC OIB Fig. 11, along with fields defined by Quaternary lavas from the
0.1
Galapagos plume (GP in Fig. 11a) and Central American (volcanic) arc
OPB
140-110 Ma Nicoya Complex (CAVA in Fig. 11a). Modern plume and arc lavas occupy similar fields,
MORB
suggesting a similar mantle source today. The modern Galapagos field
c is corrected for 90 Ma of radiogenic growth (dashed fields in Fig. 11)
0.03
Haiti & DR for comparison with ~90 Ma CLIP and PAR sequences. Isotopic data for
center of CLIP, Caribbean Sea lavas from the central Caribbean, away from subduction influences,
0.8
IBMVAB
show similar compositions, although ODP Site 1001 lavas come from a
Caribbean Sea (11-13) & Central Hispaniola (Haiti & Dominican
Republic, (14, 15)
more depleted source (Fig. 11g). Igneous rocks from Tr1, Tr2, and Tr3
CAVAB mostly fall within the age-corrected field for Galapagos plume lavas,
which is consistent with the interpretation that these melts were de-
rived from a broadly similar mantle source.
Th/Nb

Symbol color key:

Caribbean Sea
Hispaniola OIB
0.1
OPB
5. Discussion
140-110 Ma Nicoya Complex
MORB
Below we explore the implications of our compiled data for Late
0.02 d Cretaceous magmatic evolution in the SW Caribbean for testing the
0.03
140 130 120 110 100 90 80 70 60 50 40 PISI model. First we summarize the tectonic significance of the geo-
Ma chemical and isotopic insights presented in Section 4 as well as the sig-
nificance of felsic, subduction-related intrusions in the Leeward Antilles
Fig. 10. Mean and range of Th/Nb vs. time for plume- and arc-related units from this study and NW South America. We then compare the Caribbean PISI sequences
(squares and circles represent units interpreted as oceanic plateau (OP) and arc, respec-
with other SI examples. Because the PISI model predicts a continuous
tively; see top box) from (a) PISI trace 1 in Costa Rica and Panama, (b) PISI trace 2 in west-
ern Ecuador and western Colombia, (c) PISI trace 3 in the Leeward Antilles in the Cabo de evolution from plume- to arc-igneous sequences, we discuss the key
la Vela Complex, Curaçao and Aruba, and (d) the central and northern segments of the question of whether or not there is a hiatus between Late Cretaceous
Caribbean Plate in the Caribbean Sea and Central Hispaniola. The small squares in plume and arc igneous activity. We then discuss the advantages of the
(b) are Gorgona samples of Serrano et al. (2011). References, Th/Nb and ages for all
PISI model for understanding the Late Cretaceous tectonic evolution of
units in (a)–(d) are provided in Tables 2 and 3 for further correlation. References for the
oceanic island basalt (OIB), mid-oceanic ridge basalt (MORB), Central American Volcanic
this region and outline a critical test of competing models for the tecton-
Arc (CAVA) and the Izu Bonin–Marianas Volcanic Arc (IBMVA) are given in Section 3.2. ic evolution of this region: the presence or absence of a hiatus between
The faint purple, blue and pink vertical bars demarcate the timing of inferred subduction early plume and late arc magmatic successions. Finally, we briefly con-
initiation along traces 1, 2 and 3, respectively; see Fig. 5 caption and text for further details. sider how documenting the Caribbean PISI model advances our under-
Unit numbers (in italics in brackets) are as assigned as in Fig. 1.
standing of how, if not when, plate tectonics began on Earth.

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 17

0.5134
CAVA & GP 0.5132
DMMa GP at 5.1. Synopsis of geochemical and isotopic constraints
DMMb 2.8-0 Ma
Galapagos Islands 0.5128 BSE HIMU
0.5132 (2.8-0 Ma, grey circles) GP at 90 Ma
We recognize three coterminous PISI traces, each ~ 1000 km long,
Nd/ Nd

0.5124 EMI
EMII
0.5120
along the southern margins of the Caribbean Plate and NW South
144

16 18 20 22
0.5130
America where ~100–90 Ma plume-dominated tectonic environments
CAVA (2.6-0 Ma, Galapagos Plume at 90 Ma evolved into convergent margins during Late Cretaceous time. The isoto-
143

0.5128 white circles)

pic data indicate that Caribbean igneous rocks were derived from a man-
a tle source similar to that of the modern Galapagos plume, and the
units interpreted as:
combined geochemical and geochronological data provide insights
oceanic plateau: arc: about when during the Late Cretaceous this mantle source began to be
depleted basalt, picrite, komatiite basalt
enriched &/or alkalic basalt modified by subducted inputs. We know that a subduction zone existed
0.5134 as early as 89 Ma along Tr3 because calc-alkaline plutonic rocks of
western & southern 0.5132
this age are known from Aruba and Curacao (e.g., Wright and Wyld,
GP at 2.8-0 Ma
Costa Rica (1a-d) (from a) 0.5128
0.5132 2011). PAR sequences changed throughout Late Cretaceous time from
Nd/ Nd

0.5124

0.5120 strongly tholeiitic suites expected for OPB toward increasingly calc-
144

16 18 20 22
0.5130
NC (OP, 1a,
alkaline, arc-like compositions (Figs. 4, 5). Nb/Yb ratios show that mantle
140-110 Ma, 98-82 Ma)
AP (OP, 2ai, 89-85 Ma)
sources became increasingly depleted and arc-like with time for Tr1; Tr2
143

0.5128 TC TC (OP, 1b, 91-88 Ma) Nb/Yb variations with time are complex (Fig. 6). Mantle oxidation state as
HD (OP, 1c, 88-82 Ma)
b NC (all) GC (H) (OP, 1d, 75-66 Ma) monitored by Ti/V decreases to lower, arc-like values but only after 80–
0.5134 70 Ma, indicating that the plume mantle source was not strongly oxidized
Panama (2aii, b) 0.5132 by subduction until ~20 Ma after the main plume phase (Fig. 7). Tr1 and
0.5132
0.5128
Tr2 igneous rocks show increasing contributions of shallow and total sub-
Nd/ Nd

0.5124
0.5120
duction components as gauged by elevated Ba/Th and Ba/Nb, respectively
144

16 18 20 22
0.5130 (Figs. 8, 9). Tr1 PAR units have initial (~95 Ma) mean Ba/Th of ~100–200,
intermediate to that of OIB/OPB and IBMVA basalts or similar to IBMVA
143

0.5128 basalts and by ~ 70 Ma, Ba/Th had risen to 200–600, transitional to

SAA (A, 2aii, 73-69 Ma)
CBA (A, 2b, 70-39 Ma) c IBMVA and Central American Arc basalts; Ba/Th reached 600–800 by
0.5134
the end of the Cretaceous in the Chagres–Bayano Arc and remains high
western Colombia (3, 4, 6b) 0.5132 in the modern Central American Arc. The total subduction component
0.5132 Gorgona 0.5128
(Ba/Nb) began to increase above values expected for OIB–OPB earlier, ris-
Nd/ Nd

komatiites 0.5124

0.5120
ing to ~10–20 by 85 Ma and 20–60 by ~70 Ma. Evidence of the deep sub-
144

16 18 20 22
0.5130 duction component (Th/Nb) in Tr1 lavas rises exceeds baseline values for
MORB, OIB, and OPB by ~70 Ma (Fig. 10). The record of subduction inputs
143

0.5128 in Tr2 PAR sequences shows that this mantle source was modified earlier
WC (OP, 6b, 94-75 Ma)
by subduction inputs than was the Tr1 source. Colombia PAR sequences
d Gorgona E-basalts SDB (OP, 3, 79-72 Ma)
scatter widely but have Ba/Th (shallow subduction inputs) that are most-
0.5134
western Ecuador (5a, b) 0.5132 ly and sometimes much higher than plume mantle (OIB, OPB and 140–
0.5132
0.5128 110 Ma Nicoya Complex) by 90 Ma. Ecuador sequences also have high
Nd/ Nd

0.5124
hyaloclastite Ba/Th (~200), about double that of background OIB–OPB–MORB values
0.5120
144

0.5130 picrites
16 18 20 22
(b 100) and higher than IBMVA basalts. Total subduction inputs moni-
ankaramites tored by Ba/Nb are also mostly much higher for Tr2 PAR sequences
(10–200) than plume mantle (b8). The geochemical data show that
143

basalts
0.5128 a
GRU & PDU (OP, 5a, 89-85 Ma) early plume-related OPB igneous activity in the SW Caribbean Plate
b
RCA, LDA & GGA (A, 5b, ~85-64 Ma ) e evolved rapidly into arc igneous activity before the end Cretaceous con-
0.5134
sistent with a general PISI model.
Aruba & Curaçao (8a, b, 9a) 0.5132

0.5128
0.5132
Nd/ 144Nd

0.5124
Fig. 11. Initial, age-corrected 206Pb/204Pb vs. 143Nd/144Nd of (a) lavas of the 2.8–0 Ma
0.5120
16 18 20 22 Galapagos Islands (White et al., 1993) (or Galapagos Plume, GP) and basalts of the Quater-
0.5130
nary Central American Volcanic A (CAVA) (as compiled by Jordan et al., 2012) versus (c–g)
Late Cretaceous to Eocene southern Caribbean plume- and arc-related (PAR) units. Mantle
143

0.5128 CLF (OP, 9a, 120-74 Ma) domains in the inset (BSE, DMMa, DMMb, EMI, EMII, HIMU) are from Zindler and Hart
ALF (OP, 8a, 103-89 Ma)
(1986) and the field for the GP at 90 Ma is from Hauff et al. (2000b). Abbreviations and ref-
f AB (A, 8b, 90-79 Ma)
erences for the Caribbean PAR units: (b), NC, Nicoya Complex (140–110 Ma, Hoernle et al.,
2004; 98–82 Ma, Sinton et al., 1997; Hauff et al., 2000a,b); (c) DSDP Leg 15 (Hauff et al.,
Caribbean Sea (12, 13) 0.5132
2000b; ODP Site 1001, Kerr et al., 2009); (c) CBA, Chagres–Bayano Arc (Wegner et al.,
0.5128
0.5132 Site 1001 2011); SAA, Sona–Azuero Arc (Lissinna, 2005; Wegner et al., 2011); (d) the fields for
Nd/ 144Nd

0.5124
Gorgona komatiites and enriched- (E-) basalts are from Hauff et al. (2000b) based on ref-
0.5120
16 18 20 22 erences therein; SDB, Serranía de Baudó and WC, Western Cordillera (Hauff et al., 2000b);
0.5130
(e) GRU, Guaranda Unit, PDU, Pedernales Unit (Mamberti et al., 2003); RCA, Rio Cala Arc,
LDA, La Derecha Arc, GGA, Guaragua Arc (Allibon et al., 2008); (f) AB, Aruba Batholith and
143

0.5128
DSDP Leg 15 (OP, 12, 97-87 Ma)
ALF, Aruba Lava Formation (White et al., 1999); CLF, Curaçao Lava Formation (Hauff et al.,
ODP Site 1001 (OP, 13, 82-80 Ma) g 2000b). DSDP Leg 15 (Hauff et al., 2000b); ODP Site 1001 (Kerr et al., 2009). Superscripts:
0.5126 a, the GRU and PDU are considered as circa 90–85 Ma correlatives of the Pallatanga Unit
18.0 18.2 18.4 18.6 18.8 19.0 19.2 19.4 19.6 19.8 20.0 and the Piñon Formation and other similarly aged western Ecuador oceanic plateau
206 204
units (see Mamberti et al., 2003); b, whereby the RCA represents arc constructed upon
Pb/ Pb the Pallatanga Unit (interpreted as CLIP, see Figs. 1 and 2 and Mamberti et al., 2003; Vallejo
et al., 2009) the LDA and GGA are interpreted as arc materials constructed upon the Piñon
Formation which is considered to represent 90–85 Ma oceanic plateau analogous to the
Pallatanga Unit and other similar units in western Ecuador (see Mamberti et al., 2003
and Allibon et al., 2008). Units numbered as in Fig. 1.

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
18 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

8
5.2. Significance of 99–79 Ma arc-related, felsic granitoid suites: basalt bas. and. andesite dacite & rhyolite

Implications for the timing of subduction initiation 7 Central & West. cordilleras., W. COL
BUC pegmatites & felsite dykes c
iti
While absent from Tr1, 99–79 Ma arc-related, felsic ‘granitoid’ units 6 6bii, WC), 99-87 Ma
o lei
Antioquia Batholith,(6ai), 95-86 Ma th
intrude ~ 100–80 Ma oceanic plateau basement in Tr2 (western high-Fe

FeO t/MgO
Buga Batholith (6bii), 92-89 Ma
5 e
Ecuador, western Colombia) and Tr3 (Aruba and Curaçao) (Fig. 1 and Leeward Antilles lin
Aruba Batholith (8b), 90-79 Ma a lka
listed with ages in Table 2; see also the GR Online Supplementary 4 lc-
Curacao Diorites (9b), 87-85 Ma ca
Document). In Aruba and Curaçao, CLIP sequences are intruded by 90– med-Fe
3
79 Ma and 87–85 Ma felsic, arc-related plutonic rocks, respectively
(White et al., 1999; van der Lelij et al., 2010; Wright and Wyld, 2011; 2
Aruba Batholith, Curacao Diorites, Table 2). In western Colombia, low-Fe
1
the 95–86 Ma Antioquia and the 93–89 Ma Buga batholiths intrude
~ 100–80 Ma oceanic plateau basement in the Central and Western
a
Cordilleras, respectively (Villagómez et al., 2011). Further north, in the 40 45 50 55 60 65 70 75 80
Western Cordillera of Colombia, ~ 91 Ma pegmatites of the Bólivar
Ultramafic Complex (BUC) (Kerr et al., 2004; Villagómez et al., 2011)
SiO2 wt. %
intrude basement interpreted as CLIP. Although the BUC is also
interpreted as plateau (Kerr et al., 2004), the chemical composition of 100
WPG
BUC andesites is similar to contemporaneous felsic intrusives elsewhere b
in western Colombia and in Ecuador, Aruba and Curacao (see below). In
western Ecuador the 87–81 Ma arc-related Pujilí Granite is entrained in 10
a tectonized zone of the 90–85 Ma Pallatanga Unit interpreted as CLIP Buga Bath

Nb ppm
(Vallejo et al., 2006, 2009). Antiq Bath
Some workers consider these plutonic rocks as tonalite– 1 Bólivar
trondhjemite–granodiorite-like or adakitic units (e.g., Wright and ORG
Wyld, 2011), reflecting the consensus that these are subduction-related. VAG Aruba
Curacao
These plutonic felsic rocks show calc-alkaline affinities (Fig. 12a, b), and .1
Central American Arc, CR & PAN
are most reasonably interpreted as juvenile arc granitoids (Fig. 12c). An SAA (2aii) & CBA (2b), 75-39 Ma
arc origin is further supported by marked HFSE-depletions (Supp. 35-16 Ma
16-6 Ma
Fig. 6) but the only felsic unit that exhibits a convincing origin from a .01
garnet-bearing protolith is the youngest 87–81 Ma Pujilí Granite (west- 1 10 100 1000
ern Ecuador) with chondritic HREE and Y concentrations (Supplementary Y ppm
Fig. 6c, d). Furthermore, all of the aforementioned felsic suites fall within
the range of REE patterns for 73–69 Ma Sona–Azuero Arc (Panama) gran- 10
no
ites (Supplementary Fig. 6i); show similar N-MORB patterns as the youn- rm
al
co ma c
ger Panamanian granites (Supplementary. Fig. 6b–j); and are identical to nt tu
in re
e co
younger 75–39 Ma granites of the Panamanian segment of the early Cen- pr nt nt
im al in
i ar en
c o ti v e c 8.65 ta
tral American Volcanic arc system in Y vs. Nb space, including Bólivar Ul- n t isl la
rc
in
tramafic Complex felsic pegmatites and felsite dykes (Fig. 12b). en a nd
ta
Rb/Zr

1 BUC l a a rc
Chondrite-normalized REE plots and N-MORB normalized incompatible fel. peg.
rc &

element patterns of Bolivar Ultramafic Complex basalts, pegmatites and (n = 1) 82.9 93.5
11.05
felsite dykes are provided in Supplementary Fig. 7. 71.1
86.2 10.19
As some of the felsic arc units pre-date the lower range of the 92.1

associated plateau unit, this demonstrates that these are not always in-
trusive but rather co-magmatic (e.g., as in Curaçao, see Table 2). An un-
ambiguous arc origin is supported by the marked HFSE-depletions and
LILE enrichments and as these felsic suites usually only slightly post-
date the CLIP basement they cut, these felsic plutons provide evidence
that SI quickly followed plume emplacement, consistent with PISI.
5.3. Comparison of Caribbean PISI sequences and other SI examples
The Caribbean PISI example shares significant similarities with
other examples of subduction initiation, but association with a plume
makes it unique. One important similarity is that the oldest units of
both Caribbean PISI and IBM SI sequences show no subduction input;
arc-like affinities appear only after a few million years (Ishizuka et al.,
2011). The MORB-like affinities of lower basaltic units of many
Tethyan-type ophiolites and Izu-Bonin forearc sequences are com-
positionally distinct from overlying arc-like units and this dichotomy
has led some workers to infer two tectonic environments: first at a
mid-ocean ridge and second at a volcanic arc (see Whattam and Stern,
2011 and references therein). For example, IBM forearc basalts
(Reagan et al., 2010) comprising the inner slope of the Izu-Bonin trench
were originally interpreted as trapped, older Philippine Sea MOR crust
(DeBari et al., 1999). A similar scenario has been posited for SW Pacific
CBA PAN Canal
.1
(n = 5) BUC Vijes dykes
(n = 7)
.05
1 10 100
Nb ppm
Fig. 12. (a) SiO2 vs. FeOt/MgO, (b) Y vs. Nb (a) and (c) Nb vs. Rb/Zr plots of circa 95–79 Ma
subduction-related, felsic, arc-like intrusives of NW South America and the Greater
Antilles, 99–87 Ma Bólivar Ultramafic Complex (BUC), Western Cordillera, western
Colombia and younger 75–6 Ma granitoids of the Central American Volcanic Arc in Costa
Rica (CR) and Panama (PAN). The subalkaline discrimination plot in (a) is from Miyashiro
(1974) and the subfields of high-, medium- and low-Fe are from Arculus (2003). Plots
and fields in (b) and (c) are from Pearce et al. (1984) and Brown et al. (1984), respectively.
All data in (b) are of samples with 65–75 wt.% SiO2 apart from two BUC Vijes dykes with
76 wt.% SiO2. References for all 35–16 Ma Costa Rican and Panamanian granitoids are
from Kessler et al. (1977), Drummond et al. (1995 and references listed therein), Wörner
et al. (2009) and Wegner et al. (2011). References for the 95–79 Ma felsic plutonic units
are listed in Table 2. Other abbreviations in (b): CBA, Chagres–Bayano Arc (Panama), CR,
Costa Rica, PAN, Panama, ORG, ocean ridge granite, SAA, Sona–Azuero Arc (Panama) VAG,
volcanic arc granite, WPG, within-plate granite. Numbers beside symbols in (c) represent
ages in Ma. Units numbered as in Fig. 1.
ophiolites but with the lower MORB-like sequences representing
trapped backarc basin crust (see Whattam, 2009). Explanations for the
Late Cretaceous evolution of the Central American Arc are similar,

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx
except in this case the substrate was OPB (e.g., Buchs et al., 2010) as op-
posed to MORB or backarc basin crust. As in the case of Late Cretaceous
Tethyan ophiolites and IBM forearc, there is no break between lower
oceanic tholeiites and upper arc sequences (although more work is
needed to test this for Caribbean PAR sequences). The PISI model for
Late Cretaceous plume- and arc-related sequences provides a simple ex-
planation for SI as well as the magmatic evolution, especially the appar-
ent absence of a hiatus between lower CLIP and upper arc sequences.
The principal difference with SI models for IBM and Tethyan ophiolites
explored by Whattam and Stern (2011) is that the SW Caribbean Arc
is built on plume-like sequences as opposed to MORB-like ones. These
relations suggest that Late Cretaceous arrival of the Caribbean plume
head triggered development of the Late Cretaceous arcs in Central
America, Leeward Antilles, and NW South America. As expected, SI asso-
ciated with near coeval plume-related magmatism imparted a plume
‘flavor’ on the earliest-formed South Caribbean plume- and arc-related
unit lavas and intrusives as documented in Section 4. Apart from these
plume-inherited characteristics, the Central American and NW South
American plume- and arc-related units exhibit the same evolution as
SI units of Tethyan-type ophiolites and intra-oceanic forearc crust,
i.e., a progression to arc-like affinities with time. This difference in earli-
est SI magma chemistries may be key for discriminating between SI
events triggered by plume (OIB- or OPB-like) from those not catalyzed
by plume emplacement (MORB-like).
In Section 4 we demonstrated the overlapping trace element
chemistry and isotopic composition of earlier and later plume- and
arc-related units. We note here that overlapping trace element and par-
ticularly isotopic affinities between earliest circa 90–85 Ma underlying
units interpreted as CLIP and younger, unambiguous arc-like products
have been recognized for complexes in Panama (Buchs et al., 2010),
Colombia (Kerr et al., 2004), Ecuador (Allibon et al., 2008), and the
Leeward Antilles (White et al., 1999) which are consistent with
substantial plume-subduction interactions. For example, on the basis
of trace element chemistry, Buchs et al. (2010) note that proto-arc
lavas and intrusives of the 75–66 Ma Golfito Complex (southern Costa
Rica) are compositionally bracketed by CLIP and arc-like end-
members. Using the circa 85–64 Ma Rio Cala Arc unit which was con-
structed atop the 90–85 Ma Pallatanga Unit (interpreted as CLIP) in
Ecuador as another example, Nd and Pb isotopic compositions
show that Rio Cala Arc lavas are consistent with mixing Pacific MORB-
mantle, subducted pelagic sediments and an oceanic plateau compo-
nent (Allibon et al., 2008). Another example of mixed plume-arc sources
comes from circa 90–79 Ma Aruba Batholith subduction-related
magmas which are isotopically similar to the underlying 103–89 Ma
Aruba Lava Formation unit (interpreted as CLIP) they intrude (White
et al., 1999). All of these observations are consistent with plume-
subduction interactions and apart from the Golfito Complex, provide
evidence for subduction initiation quickly following or accompanying
plume emplacement.
5.4. Is there a hiatus between Late Cretaceous CLIP and arc igneous activity?
A key test of our SW Caribbean PISI model lies in understanding the
transition between older plume-related CLIP units and younger,
subduction-related arc sequences. Our Caribbean PISI model considers
that subduction zones associated with Tr1–Tr3 formed in response to
the Caribbean Plume and that the transition from plume to subduction
happened while the plume was still active. In the cases of forearc-
derived Tethyan ophiolites of the Mediterranean–Persian Gulf region
and the Izu-Bonin forearc – all of which have been interpreted as
forming in response to subduction initiation – there is a continuous
transition upward from early MORB-like to younger arc-like sequences,
without an appreciable break in the record (Reagan et al., 2010;
Whattam and Stern, 2011; Stern et al., 2012). In the case of the Late
Cretaceous evolution of the SW Caribbean, an important break in the
stratigraphic record such as a major unconformity or an intervening
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
19
sedimentary sequence would challenge the PISI model and support
models that interpret older CLIP and younger arc sequences as forming
in independent tectonic environments. In southern Panama, there exists
an apparent ~ 10 Ma time gap between radiometric ages for the Early
Sona–Azuero (89–85 Ma) ‘plateau’ and the Later Sona–Azuero proto-
arc (initiated at 75–73 Ma; Buchs et al., 2010). The significance of this
hiatus is unclear because no known unconformity or sedimentary pack-
age separates the two units, nor are the two units known to be fault-
bounded. However, basalts of the Azuero Plateau record subduction ad-
ditions and N-MORB signatures that overlap those of the younger
Azuero Arc (at least one sample interpreted as Azuero Plateau also ex-
hibits primitive-mantle normalized HFSE anomalies along with other
lavas interpreted as plateau elsewhere; see the GR Online Supplementa-
ry Document). If the hiatus is real, it may support the idea that a signif-
icant change in tectonic environment occurred in the Late Cretaceous,
calling for models entailing early plateau and later arc, at least in south-
ern Panama. If the apparent hiatus is an artifact of inadequate radiomet-
ric ages then the PISI model is supported.
The apparent temporal gap for Panama is illustrated in Fig. 2, which
also shows representative stratigraphic columns for other southern
Caribbean PAR units documented in this study. Further field and geo-
chronologic studies could resolve this question by testing the reality of
the 10 Ma gap, for example by documenting a heretofore unrecognized
interval of Late Cretaceous sedimentary rocks, hardground, or unconfor-
mity separating the youngest CLIP lavas and the oldest proto-arc lavas,
or by obtaining more reliable radiometric ages targeted to locate the
hiatus. We reiterate that more focused stratigraphic and radiometric
studies are needed, but that until the presence of such a break is
documented, we should consider further the PISI model for the Late
Cretaceous tectonic evolution of the SW Caribbean region.
5.5. Critical evaluation of existing models for SW Caribbean subduction
zone formation
5.5.1. Little evidence for collision-induced SI
An important consideration in deciding whether the Caribbean PISI
model has merit is whether or not it answers more questions than
existing models for SI in Tr1, Tr2, and Tr3. It is especially important to
evaluate other models for why these subduction zones formed. Existing
explanations for SI in Tr1, Tr2, and Tr3 rely on mechanisms of induced
nucleation of subduction zone (INSZ; Stern, 2004). INSZ explanations
identify collisions at one subduction zone as causing a new one to
form. We show below that existing models for SI in the region of interest
do not adequately explain how these traces formed.
The reason for Central American (Tr1) SI is controversial. The
Greater Antilles Arc (GAA) was magmatically active beginning in the
Early Cretaceous (perhaps as early as ~ 135 Ma; Pindell et al., 2011).
Some tectonic models (e.g., Duncan and Hargraves, 1984; Pindell and
Barrett, 1990; Hoernle et al., 2002; Mann, 2007) suggest that east-
dipping subduction beneath Central America began after the CLIP
jammed a NE-dipping subduction system beneath the GAA (see Fig. 4
of Hoernle et al., 2002). This interpretation was especially popular
in early tectonic models but recent studies challenge this explana-
tion. We are not convinced by the evidence supporting interpretation
of an Early Cretaceous, NE-dipping subduction zone beneath the GAA.
Lebron and Perfit (1993) argued that the mid-Cretaceous (~ 120–
100 Ma) unconformity in Hispaniola and Puerto Rico could have
reflected uplift accompanying CLIP-GAA collision, which led to a sub-
duction polarity reversal, and that this unconformity also separated vol-
canic suites of distinct compositions (older arc tholeiites and younger
calc-alkaline sequences; Lebron and Perfit, 1993). Jolly et al. (2008)
agreed that Puerto Rico volcanic sequences were interrupted by mid-
Cretaceous unconformities but noted that other Cretaceous arc succes-
sions in surrounding regions were conformable and concluded that
the stratigraphic breaks were not due to collision. This interpretation
is supported by the observation that mid-Cretaceous unconformities
20 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

are common all around the Pacific basin, perhaps due to superplume ac- North
tivity at this time (Vaughan, 1995). It appears that the subduction zone
beneath GAA dipped to the SW throughout its ~90 Ma lifespan, the same 100 Ma a
polarity that the subduction zone beneath the Lesser Antilles has today.
García-Casco et al. (2008) summarized evidence that the western plume-subduction
GAA collided during latest Cretaceous–earliest Tertiary time with a interface; site of plume
thickened sedimentary pile they called “Caribeana” (see also Pindell
prominent density
& compositional
Tr2, Ecuador
et al., 2011). A similar interpretation comes from studies of GAA expo-
contrast between
young plateau &
& Colombia
surrounding older
sures in Hispaniola, with subduction continuing into the Eocene (Jolly oceanic lithosphere subduction vectors at any point
along plume periphery are at right
et al., 2001). The timing of this event appears to be at least 10 Ma too Tr1, Costa Rica angles to the periphery & directed
towards plume center
late to have triggered Central American SI and there is no evidence to & Panama
support a tectonic regime other than that of continuous SW dipping
subduction beneath the GAA. We conclude that this “soft” collision did
not trigger SI beneath Central America.
100-90 Ma ‘pinching out’
b
of plume-derived
GAA arc volcanism continued until North America entered and ‘pinching out’ LIP lithosphere
of plume-derived in Leeward Antilles
jammed the S-dipping subduction zone beneath the GAA. This collision LIP lithosphere ra
pid ea
t
h inge retr
began as early as 70–60 Ma (Lázaro et al., 2009) and continued until
in NW Costa Rica Tr2, Ecuador
~ 40 Ma (Iturralde-Vinent et al., 2008). The long-lived nature of the Tr1, Costa & Colombia
GAA is also reflected in the N70 Ma history of arc volcanism document- Rica & Panama
ed in Puerto Rico, from ~120 to ~45 Ma (Jolly et al., 2001). Terminal col- nascent plume-derived
lision between the Cuban segment of the GAA and North America also LIP lithospere

occurred in the Paleogene (Iturralde-Vinent et al., 2008; van Hinsbergen
et al., 2009), with arc volcanism continuing up to ~ 62 Ma (Rojas-
Agramonet et al., 2011). We conclude that subduction continued
beneath parts of the GAA until ~ 45 Ma, significantly later than the
beginning of subduction beneath Central America in the Late Creta-
ceous. For this reason, GAA-North America collision is also unlikely to
have triggered Central America SI. The observation that collisions on
the other side of the Caribbean plate either did not occur, were too
“soft”, or were too young to have induced Tr1 subduction favors a
non-collisional induced SI model.
5.5.2. Necessity of west-dipping subduction to facilitate eastwards
emplacement of PAR units upon South America
A fundamental assumption in many tectonic models for the Late
Cretaceous evolution of the southern Caribbean and NW South
America realm posit ‘obduction’ of plateau to the east above an east-
dipping subduction zone beneath NW South America (e.g., Kerr et al.,
1999; Hastie and Kerr, 2010; Hastie et al., 2013). However, this is me-
chanically implausible as obduction must occur in a direction opposing
subduction zone dip; if subduction was to the east then obduction
would be to the west which would have emplaced NW South America
upon the plateau (i.e., the plateau would have been subducted, at least
partially, eastwards beneath South America). Eastward obduction of
CLIP units upon South America requires a W-dipping subduction zone
as illustrated by Vallejo et al. (2009; see their Fig. 10). Original east-
dipping subduction requires a subduction flip above which western
Colombian and western Ecuadorian plume- and arc related units were
emplaced, but there is no evidence for such a polarity reversal as
discussed above. Furthermore, a subduction zone polarity flip would
be expected to lead to a magmatic gap during the tectonic reconfigura-
tion and a discernible change in chemistry between initial plume-only
assemblages and post-flip subduction-modified assemblages. However
this is not what we see in western Ecuadorian and Colombian plume-
and arc-related units but instead an uninterrupted interval of post-
100 Ma hybrid subduction-plume magmatism which records – albeit
somewhat erratic – increases in Ba/Th, Ba/Nb and Th/Nb with time
(Figs. 8–10).
West-dipping subduction in the vicinity of NW South America is re-
quired in any tectonic model regardless of the mode of SI as eastward
emplacement requires westward subduction. West-dipping subduction
is also a requirement of our PISI model (Figs. 14, 15) because original
subduction vectors at any point along the periphery of the ~ 100 Ma
plume would necessarily have been directed toward the center of the
plume (Fig. 13a) (Burov and Cloetingh, 2010). Rapid retreat of the slab
hinge and the ‘pinching out’ of PISI-derived lithosphere at the NW and
Rapid slab hinge retreat & ‘pinching out’ of increasingly subduction-modified,
plume-derived lithosphere in NW & NE corners in NW Costa Rica & Leeward Antilles
90-85 Ma c
Tr2, Ecuador
future CR-PAN arc & Colombia
Tr1, Costa Rica future western COL-
western ECU arc
& Panama
establishment of proto-arcs along southern & eastern margins of CLIP in Costa Rica
& Panama and western Ecuador & western Colombia
Fig. 13. Plan cartoon depiction of (a) initial PISI at 100 Ma, (b) rapid slab hinge rollback at
100–90 Ma and (c) establishment of arc construction at 90–85 Ma along the southern and
eastern margins of the CLIP in Tr1 in Costa Rica and Panama and in Tr2 in western Ecuador
and western Colombia. See text for details.
NE corners of the LIP-subduction interface (Fig. 13b) may explain the
termination of the PISI-derived lithosphere in NW Costa Rica and the
Leeward Antilles. Further discussion on the necessity of west-dipping
subduction to facilitate emplacement of SW South American PAR units
is provided in the GR Online Supplementary Document.
5.6. Advantages of a PISI model for understanding Caribbean tectonic
evolution
On the basis of geochemical, geochronological and tectonic con-
straints we infer that the Late Cretaceous tectono-magmatic evolution
of southern Central America and NW South America was characterized
by an episode of spontaneous nucleation of subduction along the pe-
riphery of a plume that was emplaced ~100 Ma (Fig. 14a–e; tectonic re-
constructions are provided in Fig. 14f–j). Our “SW Caribbean PISI
model” builds on geodynamic models presented by Ueda et al. (2008)
and Burov and Cloetingh (2010). The principal way that the SW Carib-
bean PISI model differs from their models is that SI occurred only on
one side of the plume (Fig. 15), not symmetrically all around CLIP.
This is not surprising, because subduction beneath the NE margin
to form the GAA occurred some tens of Ma before SI on the three SW
Caribbean traces began. The Caribbean Plume event may have been
some sort of backarc basin magmatism associated with the GAA, but it
is not necessary for the PISI model that the precise nature of the
plume event be understood, only that sufficiently dense lithosphere
was adjacent to a large enough plume such that plume emplacement

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx 21

NORTHWEST SOUTHEAST WEST EAST WEST EAST

A 140-110 Ma plateau A
0
NA
300°C
100 Ma f 89-85 Ma h
depth (km)
700°C
50 1100°C
1300°C Chortis 20oN 20oN
Block GAA

100 rising
100 Ma future
(~45-40 m.y.)
‘older’ Nicoya
a plume A Complex CLIP
adjoined?
0 140-110 0oN 0oN
Ma plateau IOA
300°C
A SA construction off
depth (km)

700°C GHS
50 1100°C future CR & PAN
1300°C
flattening ~ 1000 km & western South America
100 Ma
o o o
100oW 80oW 60oW 100 W 80 W 60 W
100

b -0.001 m.y.
95 Ma g 70 Ma ?
i
0 slab sinking 20 N
o
20 N
o

300°C asthenosphere
depth (km)

700°C
flooding, & SI
50 1100°C
1300°C
proto-
100 lithosphere weakening 100 Ma CLIP
o
0N 0N
o

c above rising plume -0.02 m.y .

SA-IOA collision &
0
ophiolite emplacement
300°C
depth (km)

o o o o o o
700°C 100 W 80 W 60 W 100 W 80 W 60 W
50 1100°C
1300°C key for a-e
dry lithos. peridotite, partially 60 Ma j
100 plume head emplacement 100 Ma dry asthenos. molten
? 20 N
o

hydrated mantle 140-110 Ma thickened

d & lithosphere rupture -0.05 m.y.
key for f-j oceanic crust (plateau)
140-110 Ma remnant
oceanic plateau
0 oceanic plateau
partially molten
300°C active sub. zone
depth (km)

700°C solidified
remnant sub. zone
50 1100°C arc proposed transform 0oN
inception of SE-
1300°C active subduction vector dipping subduction
remnant
100 subduction 95 Ma emplaced
MORB-like asthenos.
+ enriched, hydrated
& arc construction
along NW South America

150
e initiation ophiolite-arc
fragment
plume component
plate motion vector 100oW 80oW 60oW

0 100 200 300 400 500 600 700 800 900

width (km)

Fig. 14. PISI model (a-e) and 100-60 Ma tectonic reconstructions (f-j). PISI model is based on and modified from Ueda et al. (2008). A (west) and A’ (east) for (a–e) is shown in (f). In
(a), the (b) plume rises to the lithosphere base and flattens. (c) A central plume wedge rises and weakens the surrounding lithosphere. (d) The partially molten plume head ruptures
the lithosphere and begins to spread. (e) Subduction initiates in response to plume emplacement. (f) The Farallon Plate comprised a segment of circa 140–100 Ma thickened oceanic
crust represented at least in part, by the 140–110 Ma segment of the Nicoya Complex, western Costa Rica (see text); this older thickened oceanic crust was likely the substrate upon
which the volumetrically dominant, post-100 Ma CLIP was constructed. As depicted in (e), emplacement of the mantle plume possibly above the Galapagos hotspot (GHS)
(f) instigates (g) lithosphere sinking, subsequent asthenospheric flooding and ultimately SI. (h) SI results in arc construction along Costa Rica and Panama, western South America in
Ecuador and Colombia and the Leeward Antilles (Aruba and Curacao) (PISI traces 1–3, Fig. 1). (i) Collision of South America with the arc system off of western South America
at ~ 70 Ma results in emplacement of intra-oceanic arc (IOA) ophiolitic fragments (dark yellow) in Colombia and Ecuador (see text for details). (j) E/SE dipping subduction nucleates
to the west of South America. M.y. = millions of years.

caused a lithospheric weakness (Fig. 15b) that allowed flanking dense lith-
osphere to flex downward (Fig. 15c). It was not coincidental that SI imme-
diately followed plume head emplacement but rather that SI was catalyzed
by the plume head. As terminal magmatism of the 140–110 Ma plateau
occurred ~10 Ma prior to the start of post-100 Ma CLIP magmatism, we
suggest that pre-existing compositional and density contrasts between
the 140–110 Ma plateau and its surrounding ‘normal’ oceanic lithosphere
(Niu et al., 2003) created a favorable site for subduction to subsequently
nucleate upon younger plateau extrusion beginning at 100 Ma.
Because it is the dense lithosphere adjacent to the plume that sinks
to form a subduction zone, the PISI model predicts that subduction
zones will dip beneath and toward the center of the plume as explained
above (Burov and Cloetingh, 2010; and as shown in Fig. 13a). For Tr1,
the Central American subduction zone has dipped continuously east-
ward since it formed in the Late Cretaceous. The subduction zone be-
neath Tr3 also resulted in a subduction zone that dipped northward,
beneath the CLIP plume. This would have quickly brought northern
South America into this subduction zone, jamming it and shutting it
down by ~73 Ma as observed. The original dip of the subduction zone
beneath Tr2 is not easily constrained, but the presence of former E-
dipping subduction zones along the Pacific margin of South America
may have influenced its development.
The recognition of increasing subduction additions and degrees of
partial melting and overlapping isotopic composition between post-
100 Ma plume and arc in the oldest sequences interpreted as CLIP are
key arguments for our PISI model. As the rising plume-head arrived at
the base of the lithosphere, the latter was weakened via heating,
pervasive melt infiltration (Ueda et al., 2008) and diking. Lithosphere
above the plume became more buoyant as flood basalts thickened the
crust and heating thinned the mantle root (Laske et al., 2007), while
oceanic lithosphere beyond the plume-affected (rejuvenated) region
remained cold and dense. The lithospheric transition between rejuve-
nated and unaffected lithosphere localized differential vertical motions
between buoyant and dense lithosphere (Fig. 15b). This lithospheric
transition would have been arcuate in map view, centered on the
plume axis. Dense lithosphere near this transition was able to flex
downward around the plume margins (Fig. 15c). Sinking of old litho-
sphere may have been enhanced as plume-related lavas buried parts
of the old seafloor (Fig. 14g). Downward-flexing lithosphere ultimately
sank deep enough that motion changed from vertical to edge-on and
this marked when the new subduction zones formed, with attendant
development of arc magmatic systems around the nascent LIP margins
(Fig. 13c). The magmatic progression of this plume-arc evolution is pre-
served in the rock record from Tr1–Tr3 that we have studied.
Our interpretation of the Late Cretaceous tectono-magmatic evolu-
tion of this region is consistent with numerical geodynamic experi-
ments (Ueda et al., 2008; Burov and Cloetingh, 2010) and explains
systematic temporal changes in magma compositions outlined above,
at the same time reconciling many of the inconsistencies and difficulties
of current tectonic models for the region. PISI is also consistent with the
geochemistry and isotopic composition of many units which are inter-
mediate between “plume” and “arc” affinities and similar in some re-
spects to nascent SI units. Our PISI model is also consistent with the
lack of evidence for any hiatus separating the ‘CLIP’ basement fragments

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
22 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx

SW NE

110 Ma 140-110 Ma plateau Greater Antilles Arc a

Pacific normal (thin) oceanic crust Atlantic

waning “early” 140-110 Ma

mantle plume magmatism

100 Ma zone of plume-weakened

lithosphere
BUOYANT
‘old’ plateau b
DENSE
old thick
lithosphere mantle plume(s)
upwelling
plume-modified
asthenosphere

95-85 Ma Caribbean Large Igneous Province OP

proto-arc
c
thin plume-modified lithosphere
spreading
thick OP crust
incipient sinking plu
me
su flo asthenosphere
b
ad duc w counter-flow
d it
i on ti on
retards plume
s penetration

70 Ma Central American Volcanic Arc system

hybrid plume-arc crust d
Farallon
P
la
te

Fig. 15. Cartoon depicting construction of (a) the early 140–110 Ma oceanic plateau followed by the subsequent evolution of PISI between (b) 100 Ma, (c) 95–85 Ma and (d) 70 Ma.

in Panama and western Ecuador from the overlying arc units
(e.g., Vallejo et al., 2009; Buchs et al., 2010; see also Fig. 2).
We recognize that a circum-CLIP subduction zone, or its fossil re-
mainder, must have undergone considerable shape transformation/
passive deformation to arrive at the present configuration shown in
Fig. 1: a semi-circular, U-shape, has transformed, likely via a V-shape,
into a Y-shape (Tr1 = left arm, Tr3 = right arm, and Tr2 = vertical).
The details of how and when this complex oroclinal bending was ac-
complished are not clear, but interactions with NW South America
must be the key. Understanding this reconfiguration will require more
work and is beyond the scope of our study.
5.7. Implications of the PISI model for understanding how plate tectonics
began
Confirmation of the PISI model provides important new insights into
Earth's tectonic evolution. As plate tectonics must have been begun
without the sorts of plate tectonic-induced lithospheric weakness that
are today called for in most SI models, articulation of the Caribbean
PISI example helps us understand how the first subduction zones re-
quired to start global plate tectonics began. Because most of the force
driving plate motions today results from the sinking of dense litho-
sphere in subduction zones (Lithgow-Bertelloni and Richards, 1995),
the question of how plate tectonics began must focus on understanding
how the first subduction zone formed. The key to understanding this is
knowing what caused lithospheric weakness of sufficient extent
(hundreds to 1000s of kilometers long) along which oceanic lithosphere
could collapse. One possibility is large meteorite impact (Hansen, 2007),
but impacts sufficiently large to weaken a large enough region of the
lithosphere may only have occurred in Hadean times, when lithosphere
was probably not dense enough and strong enough to subduct. Bercovici
and Ricard (2014) recently suggested that lithospheric damage –
perhaps resulting from delamination – promoted shear localization
and weak zones. They argued that this lithospheric fabric combined
with transient mantle flow to form the first subduction zone and plate
tectonics, but it is not clear that lithospheric weak zones would be
weak enough and long enough to allow lithosphere to sag and sink be-
neath underlying asthenosphere. Another possibility is plume-induced
subduction initiation (PISI) whereby a mantle plume head (or other
large-scale asthenospheric upwelling) weakened the lithosphere when
it impinged on its base. Lithosphere weakening results from thermal ef-
fects and dike injection (Burov et al., 2007). PISI may have been more
important in Precambrian times than it is today, because plume heads
were likely hotter and larger and lithosphere was thinner and weaker.
The Caribbean PISI example suggests that the modern plate tectonics
regime became effective when early subduction zones formed as a lith-
ospheric response to mantle upwelling similar to what we have de-
scribed for the Late Cretaceous tectonic evolution of the SW Caribbean.
Once early subduction zones were established and the horizontal trans-
lation of lithospheric plates was initiated, transform faults and other
kinds of lithospheric weaknesses could be generated allowing for more
subduction zones to form. We should consider seriously that interaction
of a sufficiently large plume head with sufficiently dense oceanic litho-
sphere may have triggered the modern episode of plate tectonics.
6. Conclusions
On the basis of geochemical, isotopic, geochronological, stratigraphic
and tectonic considerations we infer that the principal catalyst for
forming subduction zones around the southern margin of the Caribbean

Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx
Plate and NW South America in the Late Cretaceous was the arrival of a
mantle plume head circa 100 Ma which fed the near simultaneous erup-
tion of the CLIP. Three fundamental observations support this conclu-
sion: (1) There exists two geochemically and temporally distinct
plateaus that were generated at 140–110 Ma and after 100 Ma. Distinct
compositional differences exist between post-100 Ma units exposed
along Tr1 and Tr2 (southern periphery of the Caribbean Plate and NW
South America) and (i) the older plateau; and (ii) units in the center
and northerly sections of the CLIP. Lavas and intrusives of the younger
units exposed in Tr1 and Tr2 record evidence of subduction additions
which increased with time beginning ~ 100 Ma; conversely, the older
140–110 Ma plateau and the post-100 Ma units in the northern and cen-
tral regions of the CLIP record no evidence of subduction modification.
(2) There are no obvious or documented breaks between post-100 Ma
plume-related igneous sequences (CLIP) and overlying arc units.
(3) Generation of the CLIP and earliest, overlying arc units overlap in
time, space and trace element chemical and isotopic compositions.
Compositions of the CLIP and arc units are both consistent with deriva-
tion from mantle associated with the Galapagos Plume which became
increasingly subduction-modified with time beginning at 100 Ma.
This is the first convincing example of plume-induced subduction
initiation, and the validation of this model opens doors for many ave-
nues of further research, ranging from the need for more detailed stud-
ies of circum-Caribbean igneous suites to investigating why some
mantle plumes evolve into subduction zones and others do not. We em-
phasize that many aspects of this model require further testing. For ex-
ample, did the pre-existing subduction regime further south in South
America and along the NE margin of the Caribbean plate contribute to
PISI in the SW Caribbean? Perhaps this explains why Caribbean plume
emplacement triggered SI around the Caribbean but not along other
oceanic plateaus (e.g., the Ontong Java Plateau). Another important
question is whether our interpretation of the Late Cretaceous tectonic
development of this complex region advances understanding of how
other subduction zones originated. If the PISI model is valid, it is unlikely
that it only applies for the Caribbean plume-arc system and there should
be other episodes of plume-induced subduction initiation. PISI may be
useful for understanding how the modern Cascadia subduction zone
formed which has been interpreted as forming in response to the arrival
of the Yellowstone mantle plume in the Eocene (Wells et al., 1984).
Another PISI candidate can be found in the Neoproterozoic of South
China, where there is vibrant controversy as to whether 825–720 Ma ig-
neous rocks formed in a mantle plume/rift setting or collision/arc
setting (Wang et al., 2009). Finally, confirmation of the PISI model pro-
vides important new insights into Earth's tectonic evolution. PISI may
have been more important in Precambrian times than it is today, be-
cause plume heads were likely hotter and larger and lithosphere was
thinner and weaker. The Caribbean PISI example suggests that the
modern-day plate tectonics regime may have begun when early sub-
duction zones formed in response to arrival of a mantle plume head at
the base of old, dense oceanic lithosphere. Establishment of early sub-
duction zones and horizontal translation of lithospheric plates generat-
ed lithospheric weaknesses; subsequently, the proliferation of
subduction zones heralded the onset of the modern plate tectonics
regime.
Acknowledgments
We wish to thank T. Gerya for his positive comments on an earlier
version of this manuscript and his invitation to submit this manuscript
to Gondwana Research. K. Ueda and an anonymous reviewer are grate-
fully acknowledged for their meticulous, detailed and constructive re-
views which greatly improved this paper. SAW acknowledges the
financial support from the Smithsonian Tropical Research Institute in
Panama City, Panama during his tenure there in 2010–2011. This is a
UTD Geosciences contribution #1261.
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
23
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.gr.2014.07.011.
References
Allibon, J., Monjoie, P., Lapierre, H., Jaillard, E., Bussy, F., Bosch, D., Senebier, F., 2008. The
contribution of the young Cretaceous Caribbean Oceanic Plateau to the genesis of
late Cretaceous arc magmatism in the Cordillera Occidental of Ecuador. Journal of
South American Earth Sciences 26, 355–368.
Alvarado, G.E., Denyer, P., Sinton, C.W., 1997. The 89 Ma Tortugal komatiitic suite, Costa
Rica: implications for a common geological origin of the Caribbean and Eastern Pacific
region from a mantle plume. Geology 25, 439–442.
Arculus, R.J., 2003. Use and abuse of the terms calcalkaline and calcalkalic. Journal of
Petrology 44, 929–935.
Arndt, N.T., Kerr, A.C., Tarney, J., 1997. Dynamic melting in plume heads: the forma-
tion of Gorgona komatiites and basalts. Earth and Planetary Science Letters 146,
289–301.
Bandini, A.N., Flores, K., Baumgartner, P.O., Jackett, S.-J., Denyer, P., 2008. Late Cretaceous
and Paleogene radiolaria from the Nicoya Peninsula, Costa Rica: a tectonostratigraphic
application. Stratigraphy 5, 3–21.
Barbarin, B., 1999. A review of the relationships between granitoid types, their origins and
their geodynamic environments. Lithos 46, 605–626.
Barrero, D., 1979. Geology of the Central Western Cordillera, west of Buga and Roldanillo,
Colombia. Publicaciones Geologicas Especials del INGEOMINAS 4, 1–75.
Baumgartner, P.O., Flores, K., Bandini, A., Girault, F., Cruz, D., 2008. Upper Triassic to
Cretaceous radiolaria from Nicaragua and northern Costa Rica—The Mesquito com-
posite oceanic terrane. Ofioliti 33, 1–19.
Bercovici, D., Ricard, Y., 2014. Plate tectonics, damage, and inheritance. Nature 508,
513–516.
Bourgois, J., Toussaint, J.F., Gonzales, H., Azema, J., Calle, B., Desmet, A., Murcia, L.S.,
Acevedo, A.P., Parra, E., Tournon, J., 1987. Geological history of the Cretaceous
ophiolitic complexes of NW South America (Colombian Andes). Tectonophysics
143, 307–327.
Beaudon, E., Martelat, J.E., Amortegui, A., Lapierre, H., Jaillard, E., 2005. Metabasites de la
cordillere occidentale d'Equateur, temoins du soubassement oceanique des Andes
'Equateur. Comptes Rendus Geoscience 337, 625–634. http://dx.doi.org/10.1016/j.
crte.2005.01.002.
Boland, M.L., Pilatasig, L.F., Ibadango, C.E., McCourt, W.J., Aspden, J.A., Hughes, R.A., Beate,
B., 2000. Geology of the Cordillera Occidental of Ecuador between 0° and 1° N.
Proyecto de Desarrollo Minero y Control Ambiental, programa de Informacion
Cartografica y Geológica, CODIGEM-BGS, Quito, Informe. 10.
Bowland, C.L., Rosencrantz, E., 1988. Upper crustal structure of the western Colombian
Basin, Caribbean Sea. Geological Society of America Bulletin 100, 534–546.
Brown, G.C., Thorpe, R.S., Webb, P.C., 1984. The geochemical characteritics of granitoids in
contrasting arcs and comments on magma sources. Journal of the Geological Society
of London 141, 413–426.
Buchs, D.M., Arculus, R.J., Baumgartner, P.O., Baumgartner-Mora, C., Ulianov, A., 2010. Late
Cretaceous arc development on the SW margin of the Caribbean Plate: insights from
the Golfito, Costa Rica, and Azuero, Panama, complexes. Geochemistry, Geophysics,
Geosystems 11. http://dx.doi.org/10.1029/2009GC002901.
Buchs, D.M., Baumgartner, P.O., Baumgartner-Mora, C., Bandini, A.N., Jackett, S.-J.,
Diserens, M.-O., Stucki, J., 2009. Late Cretaceous to Miocene seamount accretion and
mélange formation in the Osa and Burica peninsulas (southern Costa Rica): Episodic
growth of a convergent margin. In: James, K.H., Lorente, M.A., Pindell, J.L. (Eds.), The
Origin and Evolution of the Caribbean Plate. Geological Society (London) Special Pub-
lication. 328, pp. 411–456.
Burov, E., Cloetingh, S., 2010. Plume-like upper mantle instabilities drive subduction
initiation. Geophysical Research Letters 37, L03309. http://dx.doi.org/10.1029/
2009GL041535.
Burov, E., Guillou-Frottier, L., d'Acremonte, E., Le Pourhiet, L., Cloetingh, S., 2007. Plume
head-lithosphere interactions near intra-continental plate boundaries. Tectonophysics
434, 15–38.
Canil, D., 1997. Vanadium partitioning and the oxidation state of Archaean komatiite
magmas. Nature 389, 842–845.
Casey, J.F., Dewey, J.F., 1984. Initiation of subduction zones along transform and accreting
boundaries, triple junction evolution, and forearc spreading centers—implications for
ophiolite geology and obduction. In: Gass, I., Lippard, S.J., Shelton, A.W. (Eds.),
Ophiolites and oceanic lithosphere. Geological Society of London Special Publication.
13, pp. 269–290.
Chiaradia, M., 2009. Adakite-like magmas from fractional crystallization and melting-
assimilation of mafic lower crust (Eocene Macuchi arc, Western Cordillera, Ecuador).
Chemical Geology 265, 468–487.
Cloos, M., 1993. Lithospheric buoyancy and collisional orogenesis: subduction of oceanic
plateaus, continental margins, island arcs, spreading ridges and seamounts. Geologi-
cal Society of America Bulletin 105, 715–737.
Cooper, P., Taylor, B., 1985. Polarity reversal in the Solomon Islands. Nature 314, 428–430.
Condie, K.C., Abbott, D.H., 1999. Oceanic plateaus and hotspot islands: identification and
role in continental growth. Lithos 46, 1–4.
DeBari, S.M., Taylor, B., Spencer, K., Fujioka, K., 1999. A trapped Philippine Sea plate origin
for MORB from the inner slope of the Izu-Bonin Trench. Earth and Planetary Science
Letters 174, 183–197. http://dx.doi.org/10.1016/S0012-821X(99)00252-6./S0012-
821X(99)00252-6.
24 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx
DiMarco, G., Baumgartner, P.O., Channell, J.E.T., 1995. Late Cretaceous‐early Tertiary pa-
leomagnetic data and a revised tectonostratigraphic subdivision of Costa Rica and
western Panama. In: Mann, P. (Ed.), Geologic and Tectonic Development of the
Caribbean Plate Boundary in Southern Central America. Special Paper of the Geolog-
ical Society of America. 295, pp. 1–27.
Donnelly, T.W., Melson, K., Kay, R., Rogers, J.J.W., 1973. Basalts and dolerites of Late
Cretaceous age from the central Caribbean. In: Edgar, N.T., Saunders, J.B. (Eds.), Initial
Reports of the Deep Seal Drilling Project. vol. 15. U.S. Government Printing Office,
Washington, D.C., pp. 989–1012.
Drummond, M.S., Bordelon, M., de Boer, J.Z., Defant, M.J., Bellon, H., Feigenson, M.D., 1995.
Igneous petrogenesis and tectonic setting of plutonic and volcanic rocks of the
Cordillera de Talamanca, Costa Rica–Panama, Central American arc. American Journal
of Science 295, 875–919.
Duncan, R.A., Hargraves, R.B., 1984. Plate tectonic evolution of the Caribbean region in the
mantle reference frame. Geological Society of America Memoir 162, 81–93.
Dymkova, D., Gerya, T., 2013. Porous fluid flow enables oceanic subduction initiation on
Earth. Geophysical Research Letters 40, 5671–5676.
Elliott, T., Plank, T., Zindler, A., White, W., Bourdon, B., 1997. Element transport from
slab to volcanic front at the Mariana arc. Journal of Geophysical Research 102,
14,991–15,019.
Evans, K.A., Elburg, M.A., Kamenetsky, V.S., 2012. Oxidation state of subarc mantle. Geol-
ogy 40, 783–786.
Frey, F.A., Weis, D., Yu. Borisova, A., Xu, G., 2002. Involvement of continental crust in the
formation of the Cretaceous Kerguelen Plateau: new perspectives from ODP Leg 120
sites. Journal of Petrology 43, 1207–1239.
Frost, D.J., McCammon, C.A., 2008. The redox state of the Earth's mantle. Annual Review of
Earth and Planetary Sciences 36, 389–420.
García-Casco, A., Iturralde-Vinent, M.A., Pindell, J., 2008. Latest Cretaceous collision/
accretion between the Caribbean Plate and Caribeana: origin of metamorphic ter-
ranes in the Greater Antilles. International Geology Review 50, 781–809.
Gazel, E., Hoerle, K., Carr, M.J., Herzberg, C., Saginor, I., van den Bogaard, P., Hauff, F.,
Feigenson, M., Swisher III, C., 2011. Plume-subduction interaction in southern Central
America: mantle upwelling and slab melting. Lithos 121, 117–134.
Gazel, E., Carr, M.J., Hoernle, K., Feigenson, M.D., Szymanski, D., Hauff, F., van den Bogaard,
P., 2009. Galapagos-OIB signature in southern Central America: mantle refertilization
by arc–hot spot interaction. Geochemistry, Geophysics, Geosystems 10, Q02S11.
http://dx.doi.org/10.1029/2008GC002246.
Gurnis, M., Hall, C., Lavier, L., 2004. Evolving force balance during incipient subduction.
Geochemistry, Geophysics, Geosystems 5, Q07001. http://dx.doi.org/10.1029/
2003GC000681.
Hall, C., Gurnis, M., Sdrolias, M., Lavier, L.L., Müller, R.D., 2003. Catastrophic initiation of
subduction following forced convergence across fracture zones. Earth and Planetary
Science Letters 212, 15–30.
Hansen, V.L., 2007. Suduction origin on early Earth: a hypothesis. Earth and Planetary Sci-
ence Letters 35, 1059–1062.
Hastie, A.R., Kerr, A.C., 2010. Mantle plume or slab window? Physical and geochemical
constraints on the origin of the Caribbean oceanic plateau. Earth-Science Reviews
98, 283–293.
Hastie, A.R., Mitchell, S.F., Treloar, P.J., Kerr, A.C., Neill, I., Barfo, D.N., 2013. Geochem-
ical components in a Cretaceous island arc: the Th/La–(Ce/Ce*)Nd diagram and
implications for subduction initiation in the inter-American region. Lithos
162–163, 57–69.
Hauff, F., Hoernle, K., van der Bogard, P., 2000a. Age and geochemistry of basaltic com-
plexes in western Costa Rica: contributions to the geotectonic evolution of Central
America. Geochemistry, Geophysics, Geosystems 1 (Paper number 1999GC000020).
Hauff, F., Hoernle, K., Tilton, G., Graham, D.W., Kerr, A.C., 2000b. Large scale recycling
of oceanic lithosphere over short time scales: geochemical constraints from the
Caribbean Large Igneous Province. Earth and Planetary Science Letters 174,
247–263.
Hauff, F., Hoernle, K., Schmincke, H.-U., Werner, A., 1997. A Mid Cretaceous origin for the
Galapagos hotspot: volcanological, petrological and geochemical evidence from Costa
Rican oceanic crustal segments. Geologische Rundschau 86, 141–155.
Hoernle, K., Hauff, F., van den Bogaard, P., 2004. 70 m.y. history (139–69 Ma) for the
Caribbean large igneous province. Geology 32, 697–700.
Hoernle, K., Hauff, F., van den Bogaard, P., Werner, R., Mortimer, N., Geldmacher, J.,
Garbe−Schönberg, D., Davy, B., 2010. Age and geochemistry of volcanic rocks
from the Hikurangi and Manihiki oceanic plateaus. Geochimica et Cosmochimica
Acta 74, 7196–7219.
Hoernle, K., van den Bogaard, P., Werner, R., Lissinna, B., Hauff, F., Alvarado, G., Garbe-
Schonberg, D., 2002. Missing history (16–71 Ma) of the Galapagos hotspot: implica-
tions for the tectonic and biological evolution of the Americas. Geology 30, 795–798.
http://dx.doi.org/10.1130/0091-7613.
Hughes, R., Bermúdez, R., 1997. Geology of the Cordillera Occidental of Ecuador between
0°00′ and 1° 00′S: Quito, Ecuador, Proyecto de Desarollo Minero y Control Ambiental.
Programa de Información Cartográficay Geológica Report 4 (CODIGEM–British Geo-
logical Survey), (75 pp.).
Hughes, R.A., Pilatsig, L.F., 2002. Cretaceous and Tertiary terrane accretion in the Cordillera
Occidental of the Andes of Ecuador. Tectonophysics 345, 29–48. http://dx.doi.org/10.
1016/S0040-1951(01)00205-0.
Ingle, S., Mahoney, J.J., Sato, H., Coffin, M.F., Kimura, J.-I., Hirano, N., Nakanishi, N., 2007.
Depleted mantle wedge and sediment fingerprint in unusual basalts from the Manihiki
Plateau, central Pacific Ocean. Geology 35, 595–598.
Irvine, T.N., Baragar, W.R.A., 1971. A guide to the chemical classification of common vol-
canic rocks. Canadian Journal of Earth Sciences 8, 523–548.
Ishizuka, O., Tani, K., Reagan, M.K., Kanayama, K., Umino, S., Harigane, Y., Sakamoto, I.,
Miyajima, Y., Yuasa, M., Dunkley, D.J., 2011. The timescales of subduction initiation
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
and subsequent evolution of an oceanic island arc. Earth and Planetary Science Let-
ters 306, 229–240.
Iturralde-Vinent, M.A., Otero, C.D., García-Casco, A., Van Hinsbergen, D.J.J., 2008. Paleo-
gene foredeep basin deposits of north-central Cuba: a record of arc-continent colli-
sion between the Caribbean and North American Plates. International Geology
Review 50, 863–884.
Jaillard, E., Lapierre, H., Ordoñez, M., Álava, J.T., Amórtegui, A., Vanmelle, A., 2009. Accreted
oceanic terranes in Ecuador: southern margin of the Caribbean Plate? Geological
Society, London, Special Publications 328, 469–485.
Jaillard, E., Ordoñez, M., Benitez, S., Berrones, G., Jimenez, N., Montenegro, G.,
Zambrano, I., 1995. Basin development in an accretionary, oceanic-floored fore-
arc setting: Southern coastal Ecuador during Late Cretaceous–late Eocene time.
In: Tankard, A.J., Suárez Soruco, R., Welsink, H.J. (Eds.), Petroleum Basins of
South America: American Association of Petroleum Geologists (AAPG) Memoir.
62, pp. 615–631.
Jolly, W.T., Lidiak, E.G., Dickin, A.P., Wu, T.-W., 2001. Secular geochemistry of Central
Puerto Rican Island Arc lavas: constraints on Mesozoic tectonism in the Greater
Antilles. Journal of Petrology 42, 2197–2214.
Jolly, W.T., Lidiak, E.G., Dickin, A.P., 2008. The case for persistent southwest-dipping
Cretaceous convergence in the northeast Antilles: geochemistry, melting
models, and tectonic implications. Geological Society of America Bulletin 120,
1036.
Jordan, E.K., Liu, W., Stern, R.J., Carr, M.J., Feigenson, M.D., Gill, J.B., Lenert, K., 2012. Geo-
chemical Data Library for Quaternary Lavas of Lavas from the Magmatic Front of
the Central America and Izu-Bonin Mariana Arc: Explanation to Accompany Excel
Workbook CentAm & IBM Geochem Database version 1.02: MARGINS Subduction
Factory Synthesis and Integration Project April 4, 2012.
Kemp, D.V., Stevenson, D.J., 1996. A tensile flexural model for the initiation of subduction.
Geophysical Journal 125, 73–94.
Kessler, S.E., Sutter, J.F., Issigonis, M.J., Jones, L.M., Walker, R.L., 1977. Evolution of copper
porphyry mineralization in an oceanic island arc: Panama. Economic Geology 72,
1142–1153.
Kerr, A.C., Iturralde-Vinent, M., Saunders, A.D., Babbs, T.L., Tarney, J., 1999. A new plate
tectonic model of the Caribbean: implications from a reconnaissance of Cuban
Mesozoic volcanic rocks. Geological Society of America Bulletin 111, 1581–1599.
http://dx.doi.org/10.1130/0016-7606(1999)111b1581:ANPTMON2.3.CO;2.
Kerr, A.C., Aspden, J.A., Tarney, J., Pilatsig, L.F., 2002a. The nature and provenance of accret-
ed oceanic terranes in western Ecuador: geochemical and tectonic constraints.
Journal of the Geological Society of London 159, 577–594. http://dx.doi.org/10.
1144/0016-764901-151.
Kerr, A.C., Marriner, G.F., Arndt, N.T., Tarney, J., Nivia, A., Saunders, A.D., Duncan, R.A.,
1996a. The petrogenesis of Gorgona komatiites, picrites and basalts: new field, petro-
graphic and geochemical constraints. Lithos 37, 245–260.
Kerr, A.C., Marriner, G.F., Tarney, J., Nivia, A., Saunders, A.D., Thirlwall, M.J., Sinton, C.
W., 1997. Cretaceous basaltic terranes in western Colombia: elemental, chrono-
logical and Sr–Nd isotopic constraints on petrogenesis. Journal of Petrology 38,
677–702.
Kerr, A.C., Pearson, D.G., Nowell, G.M., 2009. Magma source evolution beneath the
Caribbean oceanic plateau: new insights from elemental and Sr–Nd–Pb–Hf isotopic
studies of ODP Leg 165 Site 1001 basalts. In: James, K.H., Lorente, M.A., Pindell, J.L.
(Eds.), The Origin and Evolution of the Caribbean Plate. Geological Society of
London Special Publications. 328, pp. 809–827.
Kerr, A.C., Tarney, J., Kempton, P.D., Pringle, M., Nivia, I., 2004. Mafic pegmatites intruding
oceanic plateau gabbros and ultramafic cumulates from Bolívar, Colombia: evidence
for a ‘wet’ mantle plume? Journal of Petrology 45, 1877–1906.
Kerr, A.C., Tarney, J., Marriner, G.F., Klaver, G.T., Saunders, A.D., Thirlwall, M.F., 1996b. The
geochemistry and petrogenesis of the Late-Cretaceous picrites and basalts of Curacao,
Netherlands Antilles: a remnant of an oceanic plateau. Contributions to Mineralogy
and Petrology 124, 29–43.
Kerr, A.C., Tarney, J., Kempton, P.D., Spadea, P., Nivia, A., Marriner, G.F., Duncan, R.A.,
2002b. Pervasive mantle plume head heterogeneity: Evidence from the late Creta-
ceous Caribbean–Colombian Oceanic Plateau. Journal of Geophysical Research 107.
http://dx.doi.org/10.1019/2001JB000790.
Kerr, A.C., White, R.V., Thompson, P.M., Tarnez, J., Saunders, A.D., 2003. In: Bartolini, C.,
Buffler, R.T., Blickwede, J. (Eds.), The Circum-Gulf of Mexico and the Caribbean: hy-
drocarbon habitats, basin formation, and plate tectonics: American Association of
Petroleum Geologists (AAPG) Memoir. 79, pp. 126–188.
Kerr, A.C., White, R.V., Saunders, A.D., 2000. LIP reading: recognizing oceanic plateaux in
the geological record. Journal of Petrology 41, 1041–1056.
Kolarsky, R.A., Mann, P., Monechi, S., Meyerhoff, H.D., Pessagno Jr., E.A., 1995. Stratigraphic
development of southwestern Panama as determined from integration of marine
seismic data and onshore geology. In: Mann, P. (Ed.), Geologic and Tectonic Develop-
ment of the Caribbean Plate Boundary in Southern Central America. Special Paper of
the Geological Society of America. 295, pp. 159–200.
Korenaga, J., 2013. Initiation and evolution of plate tectonics on Earth: theory and obser-
vations. Annual Review of Earth and Planetary Sciences 41, 117–151.
Kuno, H., 1968. Differentiation of basalt magmas. In: Hess, H.H., Poldervaart, A.A. (Eds.),
Basalts: The Poldervaart Treatise on Rocks of Basaltic Composition, 2. Interscience,
New York, pp. 623–688.
Lapierre, H., Bosch, D., Dupuis, V., Polve, M., Maury, R., Hernandez, J., Monie, P.,
Yeghicheyan, D., Jaillard, E., Tardy, M., Mercier de Lepinay, B., Mamberti, M.,
Desmet, A., Keller, F., Senebier, F., 2000. Multiple plume events in the genesis of the
peri-Caribbean Cretaceous oceanic plateau province. Journal of Geophysical Research
105, 8403–8421. http://dx.doi.org/10.1029/1998JB900091.
Lapierre, H., Dupuis, V., de Lépinay, B.M., Bosch, D., Monié, P., Tardy, M., Maury, R.C.,
Hernandez, J., Polvé, Yeghicheyan, D., Cotten, J., 1999. Late Jurassic oceanic crust
 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx
and Upper Cretaceous Caribbean Plateau picritic basalts exposed in the Duarte Igne-
ous Complex, Hispaniola. Journal of Geology 107, 193–207.
Lapierre, H., Dupuis, V., de Lépinay, B.M., Tardy, M., Ruíz, J., Maury, R.C., Hernandez, J.,
Loubet, M., 1997. Is the Lower Duarte Igneous Complex (Hispaniola) a remnant of
the Caribbean plume-generated oceanic plateau? Journal of Geology 105, 111–120.
Laske, G., Phipps Morgan, J., Orcutt, J.A., 2007. The Hawaiian SWELL pilot experiment—
Evidence for lithosphere rejuvenation from ocean bottom surface wave data. In:
Foulger, G.R., Jurdy, D.M. (Eds.), Plates, plumes, and planetary processes. Geological
Society of America Special Paper. 430, pp. 209–233.
Lázaro, C., García-Casco, A., Rojas Agramonte, Y., Kröner, A., Neubauer, F., Iturralde-Vinent,
M., 2009. Fifty-five- million-year history of oceanic subduction and exhumation at
the northern edge of the Caribbean plate (Sierra del Convento mélange, Cuba).
Journal of Metamorphic Geology 27, 19–40.
Lebron, M.C., Perfit, M.R., 1993. Stratigraphic and petrochemical data support subduction
polarity reversal of the Caribbean Island Arc. Journal of Geology 101, 389–396.
Lissinna, B., 2005. A profile through the Central American Landbridge in western Panama:
115 Ma interplay between the Galápagos Hotspot and the Central American Subduc-
tion Zone. (Ph.D. thesis) Christian-Albrechts University, Kiel, Germany, (102 pp.).
Lithgow-Bertelloni, C., Richards, M.A., 1995. Cenozoic plate driving forces. Geophysical
Research Letters 22, 1317–1320.
Loewen, M.W., Duncan, R.A., Kent, A.J.R., Krawl, K., 2013. Prolonged plume volcanism in
the Caribbean Large Igneous Province: new insights from Curacao and Haiti. Geo-
chemistry, Geophysics, Geosystems 14. http://dx.doi.org/10.1002/ggge.20273.
Luzieux, L.D.A., Heller, F., Spikings, F., Vallejo, C.F., Winkler, W., 2006. Origin and Creta-
ceous tectonic history of the coastal Ecuadorian forearc between 1°N and 3°S: paleo-
magnetic, radiometric and fossil evidence. Earth and Planetary Science Letters 249,
400–414. http://dx.doi.org/10.1016/j.epsl.2006.07.008.
MacDonald, W.D., 1968. Communication. Status of geological research in the Caribbean.
14. University of Puerto Rico, p. 40.
Mahoney, J.J., Jones, W.B., Frey, F.A., Salters, V.J.M., Pyle, D.G., Davies, H.L., 1995. Geochem-
ical characteristics of lavas from Broken Ridge, the Naturaliste Plateau and southern-
most Kerguelen Plateau: early volcanism of the Kerguelen hotspot. Chemical Geology
120, 315–345.
Mahoney, J.J., Storey, M., Duncan, R.A., Spencer, K.J., Pringle, M., 1993. Geochemistry and
geochronology of the Ontong Java Plateau. In: Pringle, M., Sager, W., Sliter, W.,
Stein, S. (Eds.), The Mesozoic Pacific: Geology, Tectonics, and Volcanism: Geophysical
Monograph 77. American Geophysical Union, pp. 233–261.
Mamberti, M., Lapierre, H., Bosch, D., Jaillard, E., Ethiene, R., Hernandez, J., Polvé, M., 2003.
Accreted fragments of the Late Cretaceous Caribbean–Colombian Plateau in Ecuador.
Lithos 66, 173–199.
Mamberti, M., Lapierre, H., Bosch, D., Jaillard, E., Hernandez, J., Polvé, M., 2004. The Early
Cretaceous San Juan Plutonic suite, Ecuador: a magma chamber in an oceanic pla-
teau? Canadian Journal of Earth Sciences 41, 1237–1258.
Mann, P., 2007. Overview of the tectonic history of northern Central America. In: Mann, P.
(Ed.), Geologic and tectonic development of the Caribbean plate boundary in north-
ern Central America. Geological Society of America Special Paper. 428, pp. 1–19.
Mart, Y., Aharonov, E., Mulugeta, G., Ryan, W., Tentler, T., Goren, L., 2005. Analogue model-
ling of the initiation of subduction. Geophysical Journal International 160, 1081–1091.
McKenzie, D.P., 1977. The initiation of trenches: a finite amplitude instability. In: Talwani,
M., Pitman III, W.C. (Eds.), Island arcs, deep sea trenches and back-arc basins.
American Geophysical Union, Maurice Ewing Series. 1, pp. 57–61.
Miyashiro, A., 1974. Volcanic rock series in island arcs and active continental margins.
American Journal of Science 274, 321–355.
Montes, C., Bayona, G., Cardona, A., Buchs, D.M., Silva, C.A., Morón, S., Hoyos, N., Ramírez,
D.A., Jaramillo, C.A., Valencia, V., 2012. Arc–continent collision and orocline forma-
tion: closing of the Central American seaway. Journal of Geophysical Research 117.
http://dx.doi.org/10.1029/2011JB008959.
Neal, C.R., Mahoney, J.J., Chazey, W.I., 2002. Mantle sources and the highly variable role of
continental lithosphere in basalt petrogeneisis of the Kerguelen Plateau and the
Broken Ridge LIP: results from ODP Leg 193. Journal of Petrology 43, 1177–1205.
Nikolaeva, K., Gerya, T.V., Marques, F.O., 2010. Subduction initiation at passive margins:
numerical modeling. Journal of Geophysical Research 115, B03406. http://dx.doi.
org/10.1029/2009JB006549.
Niu, Y., O'Hara, M.J., Pearce, J., 2003. Initiation of subduction zones as a consequence of lat-
eral compositional buoyancy contrast within the lithosphere: a petrological perspec-
tive. Journal of Petrology 44, 851–866. http://dx.doi.org/10.1093/petrology/44.5.851.
Pearce, J.A., Harris, N.B.W., Tindle, A.G., 1984. Trace element discrimination diagrams for
the tectonic interpretation of granitic rocks. Journal of Petrology 25, 956–983.
Pearce, J.A., Stern, R.J., Bloomer, S.H., Fryer, P., 2005. Geochemical mapping of the Mariana
arc-basin system: implications for the nature and distribution of subduction compo-
nents. Geochemistry, Geophysics, Geosystems 6, Q07006. http://dx.doi.org/10.1029/
2004GC000895.
Pindell, J.L., Barrett, S.F., 1990. Geological evolution of the Caribbean region: a plate tec-
tonic perspective. In: Dengo, G., Case, J.E. (Eds.), The Caribbean Region, the Geology
of North America volume H. Geological Society of America, pp. 405–432.
Pindell, J., Maresch, W.V., Martens, U., Stanek, K., 2011. The Greater Antillean Arc: Early
Cretaceous origin and proposed relationship to Central American subduction mé-
langes: implications for models of Caribbean evolution. International Geology Review
54, 131–143.
Reagan, M.K., Ishiuka, O., Stern, R.J., Kelley, K.A., Ohara, Y., Blichert-Toft, J., Bloomer, S.H.,
Cash, J., Fryer, P., Hanan, B.B., Hickey-Vargas, R., Ishii, T., Kimura, J.I., Peate, D.W.,
Rowe, M.C., Woods, M., 2010. Fore-arc basalts and subduction initiation in the Izu-
Bonin–Mariana system. Geochemistry, Geophysics, Geosystems 11, Q03X12. http://
dx.doi.org/10.1029/2009GC002871.
Reagan, M.K., McClelland, W.C., Guillaume, G., Goff, K.R., Peate, D.W., Ohara, Y., Stern, R.J.,
2013. The geology of the southern Mariana fore-arc crust: implications for the scale of
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
25
Eocene volcanism in the western Pacific. Earth and Planetary Science Letters 380,
41–51.
Révillon, S., Arndt, N.T., Chauvel, C., Hallot, E., 2000a. Geochemical study of ultramafic vol-
canic and plutonic rocks from Gorgona Island, Colombia: the plumbing system of an
oceanic plateau. Journal of Petrology 41, 1127–1153.
Révillon, S., Hallot, E., Arndt, N.T., Chauvel, C., Duncan, R.A., 2000b. A complex history for
the Caribbean Plateau: petrology, geochemistry, and geomorphology of the Beata
Ridge, South Hispaniola. Journal of Geology 108, 641–666.
Rojas-Agramonet, Y., Kröner, A., García-Casco, A., Somin, M., Iturralde-Vinent, J., Wingate,
D., Liu, Y., 2011. Timing and evolution of Cretaceous Island Arc magmatism in Central
Cuba: implications for the history of arc systems in the Northwestern Caribbean.
Journal of Geology 119, 619–640.
Rudnick, R.L., Gao, S., 2003. Composition of the continental crust. In: Rudnick, R.L.,
Holland, H.D., Turekian, K.K. (Eds.), Treatise on Geochemistry. 3, pp. 1–64.
Sano, T., Shimizu, K., Ishikawa, A., Senda, R., Chang, Q., Kimura, J.-I., Widdowson, M., Sager,
W.W., 2012. Variety and origin of magmas on Shatsky Rise, northwest Pacific Ocean.
Geochemistry, Geophysics, Geosystems 13, Q08010. http://dx.doi.org/10.1029/
2012GC004235.
Sen, G., Hickey-Vargas, R., Waggoner, D.G., Maurrasse, F., 1988. Geochemistry of basalts
from the Dumisseau Formation, southern Haiti: implications for the origin of the
Caribbean Sea crust. Earth and Planetary Science Letters 87, 423–437.
Serrano, L., Ferrari, L., López Martínez, M., Maria Petrone, C., Jaramillo, C., 2011. An integra-
tive geologic, geochronologic and geochemical study of Gorgona Island, Colombia: im-
plications for the formation of the Caribbean Large Igneous Province. Earth and
Planetary Science Letters 309, 324–336.
Shervais, J.W., 1982. Ti-V plots and the petrogenesis of modern and ophiolitic lavas. Earth
and Planetary Science Letters 59, 110–118.
Sinton, C.W., Duncan, R.A., Denyer, P., 1997. Nicoya Peninsula, Costa Rica: a single suite of
Caribbean oceanic plateau magmas. Journal of Geophysical Research 102, 507–515.
http://dx.doi.org/10.1029/97JB00681.
Sinton, C.W., Duncan, R.A., Storey, M., Lewis, J., Estrada, J.J., 1998. An oceanic flood basalt
province within the Caribbean plate. Earth and Planetary Science Letters 155, 221–235.
Sinton, C.W., Sigurdsson, H., Duncan, R.A., 2000. Geochronology and petrology of the igne-
ous basement at the lower Nicaraguan Rise, Site 1001. Proceedings of the Ocean
Drilling Program, Scientific Results, Leg 165. Ocean Drilling Program, Texas A&M
University, College Station, Texas, pp. 233–236.
Smith, C.A., Sisson, V.B., Avé Lallemant, H.G., Copeland, P., 1999. Two contrasting
pressure–temperature–time paths in the Villa de Cura blueschist belt, Venezuela:
possible evidence for Late Cretaceous initiation of subduction in the Caribbean. Geo-
logical Society of America Bulletin 111, 831–848.
Spikings, R.A., Winkler, W., Hughes, R.A., Handler, R., 2005. Thermochronology of alloch-
thonous terranes in Ecuador: unraveling the accretionary and post-accretionary his-
tory of the Northern Andes. Tectonophysics 399, 195–220. http://dx.doi.org/10.1016/
j.tecto.2004.12.023.
Stern, R.J., 2004. Subduction initiation: spontaneous and induced. Earth and Planetary
Science Letters 226, 275–292.
Stern, R.J., Reagan, M., Ishizuka, O., Ohara, Y., Whattam, S., 2012. To understand subduc-
tion initiation, study forearc crust: to understand forearc crust, study ophiolites. Lith-
osphere http://dx.doi.org/10.1130/L183.1.
Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic systematic of oceanic basalts:
implications for mantle composition and processes. In: Saunders, A.D., Norry, M.J.
(Eds.), Magmatism in the Ocean Basins. Geological Society of London Special Publica-
tion. 42, pp. 313–345.
Tejada, M.L.G., Mahoney, J.J., Neal, C.R., Duncan, R.A., Petterson, M.G., 2002. Basement geo-
chemistry and geochronology of Central Malaita, Solomon Islands, with implications
for the origin and evolution of the Ontong Java Plateau. Journal of Petrology 43,
449–484.
Ueda, K., Gerya, T., Sobolev, S.V., 2008. Subduction initiation by thermal–chemical
plumes: numerical studies. Physics of the Earth and Planetary Interiors 171, 296–312.
Vallejo, C., Spikings, R.A., Luzieux, L., Winkler, W., Chew, D., Page, L., 2006. The early inter-
action between the Caribbean Plateau and the NW South American plate. Terra Nova
18, 264–269.
Vallejo, C., Winkler, W., Spikings, R.A., Luziuex, L., Heller, F., Bussy, F., 2009. Mode and
timing terrane accretion in the forearc of the Andes in Ecuador. In: Kay, S.M.,
Ramos, V.A., Dickinson, W.R. (Eds.), Backbone of the Americas: shallow subduction,
plateau uplift, and ridge and terrane collision. Geological Society of America Memoir.
204, pp. 197–216. http://dx.doi.org/10.1130/2009.1204.
van der Lelij, R., Spikings, R.A., Kerr, A.C., Kounov, A., Cosca, M., Chew, D., Villagomez, D.,
2010. Thermochronology and tectonics of the Leeward Antilles: evolution of the
southern Caribbean Plate boundary. Tectonics 29, TC6003. http://dx.doi.org/10.
1029/2009TC002654.
van Hinsbergen, D.J.J., Iturralde-Vinent, M.A., van Geffen, P.W.G., García-Casco, A., van
Benthem, S., 2009. Structure of the accretionary prism, and the evolution of the Paleo-
gene northern Caribbean subduction zone in the region of Camagüey, Cuba. Journal of
Structural Geology 31, 1130–1144.
Vaughan, A.P.M., 1995. Circum-Pacific mid-Cretaceous deformation and uplift: a
superplume-related event? Geology 23, 491–494.
Villagómez, D., Spikings, R., Magna, T., Kammer, A., Winkler, W., Beltrán, A., 2011. Geo-
chronology, geochemistry and tectonic evolution of the Western and Central cordil-
leras of Colombia. Lithos 125, 875–896.
Walker, R.J., Storey, M.J., Kerr, A.C., Tarney, J., Arndt, N.T., 1999. Implications of 187Os iso-
topic heterogeneities in a mantle plume: evidence from Gorgona Island and Curacao.
Geochimica et Cosmochimica Acta 63, 13–728.
Wang, X.-C., Li, X.-H., Li, W.-X., Li, Z.-X., 2009. Variable involvements of mantle plumes in
the genesis of mid-Neoproterozoic basaltic rocks in South China: a review. Gondwana
Research 15, 381–395.
26 S.A. Whattam, R.J. Stern / Gondwana Research xxx (2014) xxx–xxx
Weber, M.B.I., Cardona, A., Paniagua, F., Cordani, U., Sepúlveda, L., Wilson, R., 2009.
The Cabo de la Vela Mafic–Ultramafic Complex, Northeastern Colombian Caribbean
region: a record of multistage evolution of a late Cretaceous intra-oceanic arc. In:
James, K.H., Lorente, M.A., Pindell, J.L. (Eds.), The origin and evolution of the
Caribbean Plate. Geological Society of London Special Publications. 328, pp. 547–566.
Wegner, W., Wörner, G., Harmon, R.S., Jicha, B.R., 2011. Magmatic history and evolution of
the Central American land bridge in Panama since Cretaceous time. Geological Socie-
ty of America Bulletin 123, 703–734.
Wells, R.E., Engebretson, D.C., Snavely Jr., P.D., Coe, R.S., 1984. Oregon plate motions and
the volcano-tectonic evolution of western Oregon and Washington. Tectonics 3,
275–284.
Whattam, S.A., 2009. Arc-continent collisional orogenesis in the SW Pacific and the nature,
source and correlation of emplaced ophiolitic nappe components. Lithos 113, 88–114.
Whattam, S.A., Stern, R.J., 2011. The ‘subduction initiation rule’: a key for linking
ophiolites, intra-oceanic forearcs, and subduction initiation. Contributions to Mineral-
ogy and Petrology 62, 1031–1045. http://dx.doi.org/10.1007/s00410-011-0638-z.
White, R.V., Tarney, J., Kerr, A.C., Saunders, A.D., Kempton, P.D., Pringle, M.S., Klaver, G.T.,
1999. Modification of an oceanic plateau, Aruba, Dutch Caribbean: implications for
the generation of continental crust. Lithos 46, 43–68. http://dx.doi.org/10.1016/
S0024-4937(98)00061-9.
White, W.M., McBirney, A.R., Duncan, R.A., 1993. Petrology and geochemistry of the
Galapagos Islands: portrait of a pathological mantle plume. Journal of Geophysical
Research 98 (B11), 19533–19563.
Wiedmann, J., 1978. Ammonites from the CLF, Curaçao, Caribbean. Geologie & Mijnbouw
57, 361–364.
Wörner, G., Harmon, R.S., Wegner, W., 2009. In: Kay, S.M., Ramos, V.A., Dickinson, W.R.
(Eds.), Backbone of the Americas: shallow subduction, plateau uplift, and ridge and
terrane collision. Geological Society of America Memoir. 204, pp. 183–196. http://
dx.doi.org/10.1130/2009.1204.
Wright, J.E., Wyld, S.J., 2011. Late Cretaceous subduction initiation on the eastern margin
of Caribbean–Colombian oceanic plateau: one great arc of the Caribbean (?).
Geosphere 7, 468–493. http://dx.doi.org/10.1130/GES00577.1.
Zimmer, M., Plank, T., Hauri, E.H., Yogodzinski, G.M., Stelling, P., Larsen, J., Singer, B., Jicha,
B., Mandeville, C., Nye, C.J., 2010. The role of water in generating the calc-alkaline
trend: new volatile data for Aleutian magmas and a new tholeiitic index. Journal of
Petrology 51, 2411–2444.
Zindler, A., Hart, S., 1986. Chemical geodynamics. Annual Review of Earth and Planetary
Sciences 14, 493–571.
Please cite this article as: Whattam, S.A., Stern, R.J., Late Cretaceous plume-induced subduction initiation along the southern margin of the Carib-
bean and NW South America: The first doc..., Gondwana Research (2014), http://dx.doi.org/10.1016/j.gr.2014.07.011
Scott A. Whattam holds a BSc from Carleton University,
Canada and a PhD from the University of Hong Kong, SAR
China. He is an Associate Professor of Petrology at Korea
University (Seoul, Republic of Korea). He held post-doctoral
positions at Seoul National University (Republic of Korea),
Rutgers University (US) and the Smithsonian Tropical Re-
search Institute (Panama City, Panama). His research inter-
ests include understanding the relation between ophiolites,
forearc crust and subduction initiation; and cosmochemistry
and early solar system studies. His land- and sea-based re-
search has been conducted on ophiolite and arc-related ter-
ranes in the SW Pacific, North Pacific, East Asia and Central
America. His NASA-funded, laboratory-based experimental
petrology studies have focused on constraining the thermal
and temporal evolution of earliest solar nebula components such as ferromagnesian sili-
cate chondrules and Ca–Al-rich refractory inclusions. He has served on the Editorial Board
of International Geology Review since July 2013.
Robert J. Stern is a Professor of Geosciences at the University
of Texas at Dallas, where he has been a member of the faculty
since 1982. He received his undergraduate training at the
University of California at Davis, and his PhD at the Scripps
Institute of Oceanography at UC San Diego. He carried out
post-doctoral studies at the Department of Terrestrial Mag-
netism, Carnegie Institution of Washington, and he has been
a visiting scholar at Stanford University, Caltech and ETH. His
research focuses on understanding how subduction zones
operate, how convergent margin magmatic systems form
and evolve, how the continental crust is generated and
destroyed, how new subduction zones form, and when Plate
Tectonics began. This research agenda requires studies on
land and at sea and many scientific collaborations. Informa-
tion about his research can be found on Google Scholar. He is the Editor-in-Chief of Inter-
national Geology Review.
 GR Online Supplementary Document for:

Late Cretaceous Plume-induced Subduction Initiation along the

southern margin of the Caribbean and NW South America: The
first documented example with implications for the onset of
Plate Tectonics

Scott A. Whattama*, Robert J. Sternb

a
Department of Earth and Environmental Sciences, Korea University, Seoul 136-701, Korea
(whattam@korea.ac.kr)
b
Geosciences Department, University of Texas at Dallas, Richardson, TX 75083-0688, USA

This Supplementary Document (7595 words) comprises three written Sections: Section 1, Nature
of samples and geochemical data manipulation; Section 2, Distribution and frequency of
radiometric ages at the level of region and discrete unit; Section 3: Additional details on the
chemical compositions of Caribbean plume- and arc-related units and the onset of arc igneous
activity; Section 4, Additional tectonic considerations; eight figures (Supp. Figs. 1-8); and
references.

Words: 7631, excluding references, captions

1. Nature of samples and data manipulation

1.1 Nature of samples and geochemical data manipulation

Our data compilation focuses on volcanic rocks. Samples are usually basalts but also

include picrites, komatiites and ankaramites from western Colombian, Ecuadorian and Curacaos

basement units interpreted as CLIP which occur in addition to basalts. We also plot samples

described as (basaltic) dykes, diabases, dolerites and basaltic clasts. Apart from Figs. 4 and 5

where we also plot intrusive rocks to elucidate subalkaline affinities and Fig. 12 and

Supplementary Figs. 6, 7 where we plot 99-87 Ma felsic, arc-like intrusive units (see main text)

we do not consider plutonic rocks such as gabbros or pyroxenites (e.g., from Bolivar) as these

often reflect cumulate processes that strongly affect trace element concentrations; we also

exclude tuffs, breccias and metamorphic rocks as these are especially vulnerable to chemical

changes due to alteration.

In datasets where both XRF and ICP data are presented, we use the latter. Oxides are

recalculated on an anhydrous (volatile-free) basis and normalized to 100%. Only samples which

yield oxide totals of 97-102 wt.% (excluding LOI) are plotted. In some cases, this resulted in

omitting entire chemical suites such as the Tortugal (Costa Rica) alkaline suite where all alkali

basalts yield oxide totals much lower than 97 wt. %. An exception to the 97-102 wt. % filter is

the dataset of subduction initiation rule ophiolites (SIRO), for which only samples with LOI < 7

wt.% are plotted (see Whattam and Stern, 2011). Another exception to the 97-102 total oxide wt. %

filter is the use of Leeward Antilles Curacao diorite intrusive data as both samples yield total

oxides between 96-97 wt.%. In cases where compositions are reported as ‘dry’, i.e., pre-

normalized on an anhydrous basis (e.g., as in Révillon et al., 2000; Mamberti et al., 2003; Kerr et

al., 1996a, b; 1997), samples are not normalized but only samples with LOI < 7 wt.% are plotted.

2
Rare earth elements (REE) and other trace elements are expressed in ppm and trace elements are

plotted only for samples with accompanying major element data; an exception is our use of the

databases of DSDP Leg 15 CLIP samples (described as basalts) as only trace element

abundances were reported by Sinton et al. (1998).

1.2 Subsolidus elemental mobility

The compiled data are for rocks that are variably altered. We cannot directly assess the

extent to which the composition of these samples have been affected by alteration, but we accept

that data reported for samples that pass the filters mentioned in Supplementary Section 1.1 above

mostly preserve primary igneous compositions. Alteration has certainly added water and has

variably mobilized alkali metals (Na, K, Rb, Cs) and to a lesser extent alkaline earths (Mg, Ca,

Sr, Ba) and complexed some elements (especially U). For this reason, we largely base our

interpretations on fluid-immobile elements, especially the rare earth elements (REE), high field

strength elements (HFSE, Ti, Nb, Zr), the transition metal V and the subduction-mobile elements

Ba and Th. We discuss in more detail below the use of Th, Nb and Ba and why Ba should be

considered as subduction- and not alteration-mobile in the current study.

1.3 Alteration- vs. subduction-derived element mobility

Some studies of Late Cretaceous plume- and arc-related units of Caribbean region

suggest that Ba is an unreliable petrogenetic indicator due to its possible mobility during

hydrothermal alteration. For this reason, we discuss below why this appears not to be the case in

this study.

The large ion lithophile elements (LILE) of Ba, Sr, K, Rb, and Cs and the high field

3
strength elements (HFSE) of La, Th and U are enriched in arc volcanics. These elevated

abundances may reflect addition of fluids or melts from subducted sediments to the arc melting

source (Plank and Langmuir, 1993). Moderate alteration can affect abundances of alkali metals

and alkaline earths but not high-field strength elements (HFSE) Th, Nb, Zr (and Ti) including the

rare earth elements (REE, e.g., La), which are generally not mobilized during greenschist

metamorphism and seafloor weathering and thus generally provide excellent petrogenetic

markers. The significance of the depletion of Nb and Ta in arc magmas is controversial and may

either reflect ultra-depleted peridotite source or partition into the refractrory phases such as

rutile . Thus, Th/Nb is a powerful tool for identifying coupled arc ‘signatures’ which record both

source depletion and slab-derived sedimentary contributions to the source.

In contrast to the HFSE, LILE can potentially be mobilized during secondary processes,

including greenschist facies metamorphism, hydrothermal activity, and seafloor weathering that

many or perhaps most Caribbean plume- and arc-related units have been subjected to. As a result,

many studies of Caribbean plume- and arc- units have dismissed Ba as unreliable for preserving

primary magmatic information citing its lack of correlation with Zr (e.g., Kerr et al., 1997; 2009;

Révillon et al., 2000; Hastie et al., 2013). These studies conclude that the lack of correlation

indicates that alteration remobilized Ba. We challenge this conclusion, for two main reasons.

Firstly, the lack of correlation between Zr and Ba is not necessarily evidence of secondary Ba

remobilization instead of subduction zone addition. Moreover, other studies of Caribbean PAR

units suggest that whereas Ba has been affected by alteration in the plateau units, elevated Ba

abundances along with other LILE are taken as evidence of subduction-influence in younger

units interpreted as arc (e.g., see Buchs et al., 2010 and references therein with respect to the

Azuero Arc). We see no reason to accept that elevated Ba/Nb and Ba/Th should be considered as

4
due to alteration in slightly older units but as reliable indicators of subduction processes in

younger units, especially because a lack of correlation between Ba and Zr is apparent in both

units. As shown in studies of earliest SI lavas (Reagan et al., 2010; Whattam and Stern, 2011)

and in N-MORB normalized plots of earliest Caribbean arc units (Supplementary Fig. 4) HFSE

depletion does not instantly appear upon subduction initiation but becomes gradually more

pronounced, as do other subduction markers (e.g., Ba/Th, Ba/Nb, Th/Nb) (Figs. 8-10). This is

due to the coupled processes of source depletion and slab-derived subduction component

additions over the course of forming a new subduction zone. Second, there is no correlation

between Ba and an indicator of alteration such as loss on ignition (LOI).

Several other lines of evidence indicate that Ba abundances in most Caribbean volcanic

rocks is a robust marker of subduction fluid additions and not alteration-derived mobility: 1,

increases in Ba/Th and Ba/Nb ratios, particularly for Tr1 units are not erratic, but rather increase

steadily with time (Figs. 8, 9); 2, the progression to higher Ba/Th and Ba/Nb with time is

mirrored by that of Th/Nb (Fig. 10); 3, the oldest (pre-100 Ma) Nicoya Complex basalts and

hence ones which should have been to subjected to more alteration than younger basalts, show

negligible increases in Ba/Th and Ba/Nb between 140-110 Ma (Figs. 8a, 9a)); 4, Ba/Nb and

Ba/Th ratios are similar between volcanic and intrusive rocks; 5, Ba/Th and Ba/Nb ratios are

lowest where and when we expect them to be lowest, e.g., in the oldest segment of the middle of

the Caribbean Plate and in the northern segment of the CLIP (Figs. 8d, 9d). For example, Ba/Nb

and Ba/Th ratios range from 2.44-3.74 and 33.3-49.6 in 97-87 Ma Caribbean Sea CLIP tholeiitic

basalts (Sinton et al., 1998) versus 2.9-52 and 60-191 in 82-80 Ma Caribbean Sea tholeiitic

basalts whereas mean Ba/Nb and Ba/Th ratios of Dumisseau Formation (Haiti) basalts are

similar regardless of age (respectively 6.1 and 60 in 110-100 Ma and a 3.7 and 50.4 in 96-82 Ma

5
Dumisseau Formation (Haiti) basalts (Loewen et al., 2013) (Table 3); and 6, Ba/Nb and Ba/Th

ratios increase progressively from MOR-like, subduction initation rule ophiolites (SIRO) lower

unit basalts and Izu Bonin forearc basalts (Ba/Nb and Ba/Th of 5.1, 7.2 and 53, 89 respectively;

MORB Ba/Nb, Ba/Th = 3, 53) to VAB-like SIRO upper unit basalts (Ba/Nb, Ba/Th = 54, 230) to

VAB (IBM and Central American VAB have Ba/Nb, Ba/Th of 142, 208 and 192, 600) (Table 3)

as expected from the subduction initation rule (Whattam and Stern, 2011). The aforementioned

observations are summarized in Table 3.

1.4 Geochronological data manipulation

40
Ages (U/Pb, Ar/39Ar or biostratigraphic) cited in the main text and here represent the

range of reported magmatic ages (including uncertainties) rounded to the nearest Ma. An

exception is Fig. 3 where calculated radiometric ages are plotted without consideration of

uncertainties.

2. Distribution and frequency of radiometric ages at the level of region and discrete unit

Fig. 3 shows the broad (on the scale of the Caribbean Plate) distribution and frequency

of biochronologic (Fig. 3a) and radiometric (Fig. 3b) magmatic ages for Caribbean plume- and

arc-related units. Suppementary Figs. 1 and 2 demonstrate the distribution and frequency of

radiometric ages at the levels of region and discrete unit, respectively.

3. Additional details on the chemical compositions of Caribbean plume- and arc-related

units and the onset of arc igneous activity

6
3.1 Mantle fertility and relative degrees of partial melting: Nb/Y vs. Zr/Y

The Zr/Y-Nb/Y plot (Fitton et al., 1997) (Supplementary Fig. 3) is commonly used in

various papers to assess degrees of melting undergone by Caribbean plateau- and arc-related

units but is rarely used for comparison with other global OPB and we have never seen it used for

comparison with younger Caribbean units interpreted as arc constructed upon the plateau. With

respect to these two important points, the Zr/Y-Nb/Y plot demonstrates: (1) a markedly more

HSFE-depleted source for basalts interpreted as CLIP than almost all other OPB, indicating that

the CLIP source experienced more melt depletion than other OPB; and (2) the difficulty in

discriminating early plateau from later arc. For example, the 75-66 Ma Golfito Complex was

originally interpreted as plateau by Hauff et al. (2000) but reinterpreted as arc by Buchs et al.

(2010). Such confusion is understandable as samples of the former dataset are more ‘plateau-like’

with Zr/Y compositions more enriched than primitive mantle whereby samples of the latter

dataset are more ‘arc-like’ with compositions more depleted than primitive mantle

(Supplementary Fig. 3e). The 89-85 Ma Azuero Plateau and the 73-69 Ma Azuero Arc show an

identical trend with samples from each overlapping and extending from compositions more

enriched to less enriched than primitive mantle (Supplementary Fig. 3f). A perusal of

Supplementary Fig. 3 demonstrates that this trend with time is in fact indicative of all units

interpreted as plateau except for those from the northern segment of the CLIP in central

Hispaniola (Supplementary Fig. 3l) although the trend is somewhat muted at either ends of the

extent of plume-induced subduction initation (PISI) hybrid crust in NW Costa Rica

(Supplementary Fig. 3d) and the Leeward Antilles (Supplementary Fig. 3j). We suggest that this

ubiquitous feature reflects formation in a hybrid subduction-plume environment established after

100 Ma. Specifically, we interpret this high degree of melting (indicated by low Zr/Y and Nb/Y)

7
to be the result of the interaction of hydrous fluids (from the subducted slab) interacting with

hotter-than-normal mantle (due to the plume).

Supplementary Fig. 3 also demonstrates that depleted basalts and related intrusive rocks

of southern Central and NW South American plume- and arc-related units were generated from a

(i) similar, HFSE-depleted mantle source that was (ii) either gradational from OPB to arc or arc-

like. In Zr/Y vs. Nb/Y space the vast majority (>97 %) of OPB have compositions more enriched

than primitive mantle (Zr/Y = 2.56, Nb/Y = 0.16); in fact, the only OPB that have compositions

more depleted than primitive mantle are ones interpreted as having formed from a ‘subduction-

contaminated’ source (a subset of Manihiki Plateau basalts, Ingle et al., 2007; U1349 basalts,

Shatsky Rise, Geldmacher et al., 2012) (Supplementary Fig. 3a). Similar to these ‘subduction-

contaminated’ OPB, many basalts and related rocks interpreted as CLIP exhibit compositions

that are more depleted than primitive mantle and suggest formation from a source akin to IBM

forearc basalts and subduction initiation rule ophiolite basalts (Supplementary Figs. 3b, c). OPB

generally plot within or just below the Icelandic (i.e., plume-OIB) array from OIB to MORB

(Supplementary Fig. 3a); IBM VAB and subduction initiation rule ophiolite units on the other

hand, plot well below the OIB field and range to sources more depleted than MORB

(Supplementary Figs. 3b, c).

The only volcanic rocks interpreted as CLIP that exhibit OIB-like compositions are the

relatively uncommon enriched basalts (comprising various units) and enriched picrites of the

Central Cordillera (Supplementary Fig. 3h); and basalts from the northern CLIP (Dumisseau and

Lower Duarte formations in Haiti and the Dominican Republic, Supplementary Fig. 3l) which

were too far north of the trench to be affected by subduction modification beginning at or soon

after 100 Ma. Caribbean basalts and related rocks interpreted as CLIP generally show a hybrid

8
plateau-arc affinity (Supplementary Figs. 3d-i) although the depleted Gorgona basalts and

komatiites show evidence of extreme degrees of melt extraction similar to many IBMVA basalts

(Supplementary Fig. 3g). The only exception is basalts of the northern segment of the CLIP in

Central Hispaniola which exhibit plateau/OIB-only affinities (Y analyses of the 140-110 Ma

Nicoya unit are not available for comparison). Similarly, Quaternary Central American VAB

overlap the OPB-IBM arc field, suggesting derivation from a source with both plateau and arc-

like affinities Supplementary Fig. 3b) indicative of the inherited plume component in the Central

American Arc source (Gazel et al., 2009; 2011).

3.2 N-MORB normalized patterns

We plot N-MORB normalized abundances of the mean composition of each volcanic

rock type of each post-100 Ma unit of this study vs. the mean composition of 140-110 Ma

Nicoya Complex basalts (Supplementary Fig. 4) to demonstrate three fundamental features

consistent with our PISI model: (1) the similarity between 140-110 Ma Nicoya Complex basalts

with global OPB; (2) the dissimilarity of 140-110 Ma Nicoya Complex basalts and OPB with

volcanic rocks of most other post-100 Ma units; and (3) the rapid development of these

dissimilarities beginning after plume emplacement at 100 Ma. Although the morphology of N-

MORB normalized patterns of most post-100 Ma CLIP basalts are broadly similar to those of the

140-110 Ma Nicoya Complex basalts and OPB in general (Supplementary Fig. 4), further

scrutiny reveals fundamental differences between pre- and post-100 Ma CLIP basalts.

Differences between most post-100 Ma units and pre-100 Ma basalts of the Nicoya

Complex (and hence ‘normal’ OPB) are manifest in N-MORB normalized plots as demonstrable

variations in markers of shallow subduction additions (Ba/Th), deep subduction additions (Th/Nb)

9
and total subduction additions (Ba/Th) (Pearce et al., 2005; see also main text); and N-MORB

normalized concentrations of incompatible elements including the HFSE ( Nb, Zr, Ti)

(Supplementary Fig. 4)). Most post-100 Ma units record discernible subductions additions as

well as lower N-MORB normalized HFSE abundances than the 140-110 Ma Nicoya Complex

basalts; the only post-100 Ma volcanic rocks which either do not show these features and/or are

similar to pre-100 Ma basalts are the volumetrically subordinate enriched/alkalic basalts and

CLIP basalts in the most northerly sections of the Caribbean Plate in central Hispaniola (Haiti

and Dominican Republic) (Supplementary Fig. 3r).

The similarity between the mean N-MORB normalized pattern of the 140-110 Ma

Nicoya Complex basalts and ‘global’ OPB (n = 353) is striking (Supplementary Fig. 4a).

Specifically, this plot demonstrates identical ratios of Ba/Nb, Th/Nb and Ba/Th for global OPB

and 140-110 Ma Nicoya Complex basalts (5.52, 0.08, and 77.75 vs. 5.97, 0.07 and 87.12) (Table

3) and similar, relatively high N-MORB normalized concentrations of Nb, Zr, and Ti (2.62, 1.50

and 1.61 vs. 3.05, 1.45 and 1.40). These features are indicative of derivation from similar,

relatively HFSE-enriched sources which underwent relatively low degrees of partial melting and

received nil subduction additions.

Discernible differences between the older 140-110 Ma and post-100 Ma basalts and

related volcanic rocks is evident even within the same complex as younger 98-82 Ma Nicoya

Complex exhibit marked increases in subduction markers and relative degrees of depletion of

most incompatible elements vs. the pre-100 Ma Nicoya basalts . The most obvious changes are

the much higher ratios of Ba/Th and Ba/Nb and lower N-MORB normalized concentrations of

the HFSE of Nb, Ti and Zr and the more depleted nature in general of all incompatible elements

of the younger Nicoya basalts vs. the older ones (Supplementary Fig. 3b). For example, whereby

10
the older Nicoya basalts record Ba/Nb and Ba/Th of 5.97 and 87.12 similar to global OPB

(Ba/Nb and Ba/Th of 5.52 and 77.75) the post 100-Ma Nicoya basalts record ratios of Ba/Nb

(11.94) and Ba/Th (201.78) that are double those of the pre-100 Ma Nicoya basalts (Table 3).

Similarly, whereby the 140-110 Ma Nicoya basalts record relatively high N-MORB normalized

concentrations of Nb (2.62), Zr (1.50) and Ti (1.61), the 98-82 Ma Nicoya basalts conversely

record much lower N-MORB normalized abundances of Nb (1.71), Zr (0.87) and Ti (0.97).

Depleted basalts of the slightly younger 88-82 Ma Herradura and 91-88 Ma Tortugal Complexes

to the immediate north and east of the Nicoya Complex (Fig. 1) record slightly higher Ba/Th and

Ba/Nb (means of 7.28, 100 and 8.52, 132.41 respectively) (Supplementary Fig. 3c, Table 3)

which demonstrates an evolution to higher shallow subduction additions with time.

Depleted basalts of the oldest post-100 Ma units along the southern margin of the CLIP

in Costa Rica and Panama (Nicoya, Herradura, Tortugal, Golfito, Azuero Plateau) record low

means of Th/Nb (0.6-0.10) similar to both the 140-110 Ma Nicoya basalts (mean Th/Nb of 0.07)

and global OPB (mean Th/Nb of 0.08) (Table 3). However, depleted basalts of the 89-85 Ma

Azuero Plateau (mean Th/Nb of 0.07) record a range to relatively high Th/Nb (0.06-0.22) similar

to but more modest than the 73-69 Ma Sona-Azuero Arc basalts (with a low mean Th/Nb of 0.10

but which ranges from 0.08-0.46) (Table 3). These features illustrate that in general, a shallow

subduction component only is recorded in the oldest post-100 Ma units along Costa Rica and

Panama.

To demonstrate that some units previously or currently interpreted as plateau are instead

more consistent as hybrid subduction-plateau units and that hybrid subduction-plateau units

began forming at or just after 100 Ma, we consider the: (1) 75-66 Ma Golfito Complex,

southernmost Costa Rica (Hauff et al., 2000; Buchs et al., 2010); (2) the 89-85 Ma Azuero

11
Plateau vs. the 73-69 Ma Sona-Azuero Arc, southern Panama (Wörner et al., 2009; Buchs et al.,

2010l Wegner et al., 2011); (3) the 79-72 Ma Serrania de Baudo unit, westernmost Colombia

(Kerr et al., 1997); and (4) the 101-98 Ma Volcanic, Barroso and Amaime formations

(Villagomez et al., 2011) which comprises the Western Cordillera of western Colombia.

As mentioned previously, the Golfito Complex was originally interpreted as plateau by

Hauff et al. (2000) but subsequently reinterpreted as arc by Buchs et al. (2010). As we have

shown in previous plots (e.g., Zr/Y vs Nb/Y, Supplementary Fig. 3), bimodal interpretations as

plateau and arc are understandable based on the particular dataset that one considers. For

example, whereby the Golfito basalts of the dataset of Hauff et al. (2000) exhibit Nb/Y higher

than primitive mantle consistent with many global OPB, the basalts of the dataset of Buchs et al.

(2010) conversely display Nb/Y less than primitive mantle, consistent with the composition of

many arc basalts (Supplementary Fig. 3). As these basalts comprise the same unit however, it is

reasonable to interpret that this unit collectively represents the transition from and overlap of

plateau and subduction modifications and not two discrete tectonic environments of formation.

Similarly, the 89-85 Ma Azuero Plateau underlies the 73-69 Ma Sona-Azuero Arc; the

interpretation of Buchs et al. (2010) is that these two units represent two different tectonic

environments of formation with the latter being constructed upon the former.

N-MORB normalized signatures of all post-100 Ma units listed above (Golfito,

Azuero/Sona-Azuero, Serrania de Baudo and the Volcanic, Barroso and Amaime formations)

exhibit discernible differences with 140-110 Ma Nicoya Complex and average OPB similar to

that of the 98-82 Ma Nicoya basalts; and the oldest post-100 Ma depleted basalts of the Western

Cordillera of western Colombia record exceptionally high Ba/Nb relative to units along the

southern margin of the CLIP and related units along western Colombia and western Ecuador

12
(Fig.9b, Table 3). For example, the mean N-MORB normalized patterns of basalts from both

datasets of the Golfito Complex (Supplementary Fig. 4d) (Hauff et al., 2000; Buchs et al., 2010)

are similar to each other but distinct from normal OPB and 140-110 Ma Nicoya Complex basalts.

The mean N-MORB normalized signature of basalts of each (Golfito) dataset exhibit lower

concentrations of most incompatible elements including the HFSE (Supplementary. Fig. 4d) and

demonstrably higher Ba/Nb (21.14, Hauff et al., 2000; 63.22, Buchs et al., 2010) and Ba/Th

(208.55, Hauff et al., 2000; 644.28, Buchs et al., 2010) than 140-110 Ma Nicoya Complex

basalts (5.97, 87.12) and global OPB (5.52, 77.75) but low Th/Nb (Table 3) as discussed above.

This trend is similar for the Sona-Azuero region where N-MORB normalized patterns of both the

unit interpreted as plateau and the unit interpreted as arc constructed upon the CLIP are similar

but gradational from relatively older to younger basalts. The mean N-MORB normalized patterns

of the 89-85 Ma Azuero Plateau and 73-69 Ma Sona-Azuero (non-enriched) basalts both show

depletions in most incompatible element concentrations and enrichments in subduction markers

to relative to 140-110 Ma Nicoya Complex basalts and global OPB (Supplementary Figs. 4e, f).

For example, Ba/Nb and Ba/Th of the Azuero Plateau basalts are 14.44 and 202.94 which are

similar to but relatively lower than the Ba/Nb and Ba/Th of the younger Sona-Azuero basalts

(34.47, 292.34), but still much higher than that of global OPB (Ba/Nb and Ba/Th of 5.52, 77.75)

and 140-110 Ma Nicoya Complex basalts (Ba/Nb and Ba/Th of 5.52, 87.12) (Table 3).

The Volcanic, Barosso and Amaime formations of the Western Cordillera of

westernmost Colombia represent the oldest post-100 Ma CLIP unit (101-98 Ma, Villagomez et

al., 2011, or ‘sub-unit’ as this encompasses the 94-75 Ma Western Cordillera of Kerr et al., 1997).

Supplementary Fig. 4h demonstrates the similarity in the N-MORB normalized pattern of the

mean composition of basalts of the 101-98 Ma Volcanic, Barosso and Amaime formations (and

13
94-75 Ma Western Cordillera basalts) to the Golfito, Azuero and Serrania de Baudo units

described above in terms of relative depletions in most incompatible elements and elevations in

Ba/Nb, Ba/Th relative to pre-100 Ma Nicoya Complex basalts and global OPB. However, the

Western Cordilleran basalts generally record much higher Ba/Nb than other units. Mean Ba/Nb,

Ba/Th and Th/Nb of the 101-98 Ma Volcanic, Amaime and Barosso formation basalts are 37.14,

513.88 and 0.07 (with a range of 0.05-0.10) whereby mean Ba/Nb, Ba/Th and Th/Nb of the 94-

75 Ma Western Cordillera basalts of Kerr et al. (1997) are 40.70, 399.98 and 0.08 (with a range

from 0.08-0.10) (Table 3). This appears to suggest that the transformation from plume to

subduction was more rapid along the (initially) eastern margin of the CLIP in western Ecuador

and western Colombia than along the southern margin of the CLIP in Costa Rica and Panama as

discussed in the main text.

The 97-86 Ma Central Cordillera basalts are somewhat anomalous as they record nil

evidence of shallow subduction additions (mean Ba/Th of 78.01) and low Ba/Nb (mean of 8.03)

(Supplementary Fig. 4g, Table 3). In contrast, the Central Cordillera picrites exhibit Ba/Nb

(15.55) and Ba/Th (224.38) similar to 90-85 Ma western Ecuadorian basalts (mean Ba/Nb and

Ba/Th of 16.99 and 279.20), picrites (mean Ba/Nb and Ba/Th of 9.81, 232.52) and ankaramites

(mean Ba/Nb and Ba/Th of 14.08, 256.93) (Supplementary Figs. 4g, k, l) and units in Costa Rica

and Panama. However, Th/Nb indicative of deep subduction additions is relatively higher in the

Central Cordillera basalts (mean of 0.09) than most other western Colombian and Ecuadorian

units and extremely high in the Central Cordillera picrites (mean Th/Nb of 0.15 and a wide range

from 0.04 to 0.41, Table 3).

Evidence of deep subduction additions (e.g., elevated Th/Nb) is generally lacking in

most of the oldest post-100 Ma units. However, as discussed above, some older units (e.g., the

14
89-85 Ma Azuero Plateau, Panama and 97-86 Ma Central Cordillera, western Colombia) range to

relatively high Th/Nb, two or three times that of mean OPB (with Th/Nb of 0.08); e.g., the

depleted basalts of the Azuero Plateau which range to 0.22 or as in the case of the Central

Cordilleran picrites, range to much higher Th/Nb (0.41) (Table 3). Similarly, 90-85 Ma western

Ecuadorian plateau basalts exhibit a similar mean Th/Nb (0.8) and range of Th/Nb (0.04-0.25) to

that of depleted Azuero basalts (0.07, 0.06-0.22) which may suggest the addition of minor deep

subduction contributions.

Similar to Nb/Yb and Ti/V which show great fluctuations in the 91-81 Ma depleted

Gorgona basalts, komatiites and picrites, the N-MORB normalized abundances of the depleted

Gorgona suite exhibits anomalously high incompatible element depletions indicative of very

high degrees of partial melting relative to most other southern Caribbean and NW South

American units (Supplementary Fig. 4j). Total subduction additions as exemplified by Ba/Nb

(mean and range of 10.76, 9.38-12.10) of the depleted Gorgona basalts is about twice that of

global OPB and 140-110 Ma Nicoya Complex basalts (means of 5.52 and 5.97) similar to other

similarly aged post-100 Ma units interpreted as plateau along the southern margin of the CLIP

(e.g., 98-82 Ma Nicoya basalts with mean Ba/Nb of 11.94) and the eastern margin of the CLIP

(e.g., 90-85 Ma Pallatanga Unit of western Ecuador with mean Ba/Nb of 16.99). Mean Ba/Th

(shallow subduction) additions however, are relatively higher in the depleted Gorgona basalts

(mean Ba/Th~330) than units in Costa Rica and Panama and intermediate to those of western

Ecuadorian plateau basalts (mean Ba/Th ~280) and Western Cordilleran basalts (which range

from ~400-500) (Table 3). Mean Ba/Nb increases from the basalts (~11) to the picrites (~23) to

the komatiites (~87) whereby Ba/Nb increases from the picrites to the basalts to the komatiites

15
(means of ~225, 339 and 2200); mean Th/Nb does not vary a lot but increases from basalts (0.06)

to picrites (0.11) to komatiites (0.12) which range to very high Th/Nb (0.050.31).

N-MORB normalized patterns of the CLIP units interpreted as plateau in Aruba and

Curacao and CLIP units in the central Caribbean Plate (Caribbean Sea drilled sites and Beata

Ridge) and northern segment of the Caribbean Plate in central Hispaniola (Haiti and Dominican

Republic) are shown in Supplementary Figs. 4m-r. Volcanic rocks of the oldest post-100 Ma

units in Aruba and Curacao (103-89 Ma Aruba Lava Formation (ALF) basalts, Curacao Lava

Formation (CLF) basalts and picrites presumed to be ~90-88 Ma, Kerr et al. 1996b) both exhibit

low N-MORB normalized incompatible element abundances relative to 140-110 Ma Nicoya

Complex basalts and OPB and but only the ALF exhibits evidence of shallow subduction

additions (Fig. 7c) not present in the older Nicoya basalts. The morphology of the N-MORB

signature of the mean composition of the ALF basalts for instance, is essentially identical to that

of the mean of 98-82 Ma Nicoya Complex basalts (Supplementary Fig. 4m). As well, the ALF

basalts exhibit mean ‘total’ subduction additions (Ba/Nb of 11.48) similar to the younger Nicoya

basalts (mean Ba/Nb of 11.74) but lower shallow subduction additions (mean Ba/Th = 156.43)

than the latter (mean Ba/Th of 201.78), both of which are nonetheless approximately twice that

of the means of 140-110 Ma Nicoya basalts and OPB (Table 3). The CLF basalts and picrites

assumed to be ~90-88 Ma (Kerr et al., 1996b) and one CLF basalt dated as 92.0±1.0 Ma

(Loewen et al., 2013), on the other hand exhibit mean Ba/Th which range from 73.04-107.38 and

mean Ba/Nb that range from 7.16-9.43 (Table 3). These values are more similar to the 140-110

Ma Nicoya Complex basalts (mean Ba/Th of 78.12 and mean Ba/Nb of 5.97) and global OPB

(mean Ba/Th of 77.75 and mean Ba/Nb of 5.52).

16
 Similarly, apart from the 82-80 Ma ODP Leg 165 Ma Caribbean Sea basalts (Kerr et al.,

2009) which exhibit relatively high Ba/Th (161.24) similar to that of the ALF basalts (mean

Ba/Th of 156.43) possibly indicative of shallow subduction additions, there is no evidence of

subduction additions in the center or the northern segments of the Caribbean Plate (Figs. 8-10,

main text). For example, all ~94-82 Ma CLIP basalts from Central Hispaniola have N-MORB

normalized Ba/Th ratios <1 (Supplementary Fig. 4r). )

The most important point that the N-MORB normalized plots underscore is that all post-

100 Ma units are related to plateau extrusion and subduction modification that occurred soon

thereafter, but vary in relative degrees of subduction modification and depth of subduction

additions based on their relative timing of formation and proximity to the 100 Ma plume.

Magmatism between 140-110 Ma as recorded by the older segment of the Nicoya Complex and

after 100 Ma in the northernmost units of the CLIP record plateau construction only; post-100

Ma magmatism along the southern and eastern margins on the other hand entailed the

construction of hybrid subduction-plateau lithosphere which received mostly shallow subduction

additions.

Some of the oldest post-100 Ma units exposed along NW South America, e.g., the 101-

98 Ma Volcanic, Barosso and Amaime Formations of the western Cordillera (Villagmoez et al.,

2011) and the 94-75 Ma of Western Cordillera of Kerr et al. (1997) record anomalously high

(initial) shallow subduction additions (mean Ba/Th of 400-500) intermediate to that of IBM

volcanic arc basalts (~500) and Central American Arc basalts (~600) (Table 3). As these values

are higher than in units of the southern margin of the CLIP in Costa Rica and Panama (e.g., mean

Ba/Th ~200 for the 98-82 Ma Nicoya and 89-85 Azuero Plateau basalts), this may suggest that

17
the transformation from plateau to subduction was more rapid along the eastern margin of the

CLIP relative to the southern margin as alluded to above.

N-MORB normalized and Th/Nb plots also suggest a faster transformation to deep

subduction additions for the NW South American PAR units relative to those of the southern

margin of the CLIP when we consider the 73-69 Ma Sona-Azuero Arc (Panama) and ~85-72 Ma

Rio Cala Arc (western Ecuador) (Supplementary Figs. 4f, l). Whereby the N-MORB normalized

plot of the Sona-Azuero Arc is very similar to that of the underlying 89-85 Ma Azuero Plateau

and both share similar Th/Nb (means of 0.07 and 0.13 respectively) the Rio Cala exhibits

extreme Th/Nb enrichment (mean of 0.70) which is higher than mean IBM arc basalts (0.55) and

much higher than the underlying Ecuadorian oceanic plateau units (mean Th/Nb of underlying

depleted plateau basalts is 0.08). Interestingly, the Rio Cala Arc conversely records low shallow,

subduction additions (mean Ba/Th of 202) relative to the underlying plateau units (mean Ba/Th

of ~280) whereby the Azuero Arc basalts record relatively much higher Ba/Th (~300) relative to

the underlying Azuero Plateau (~200) (Table 3). These data again suggest that the transformation

to deep subduction inputs was relatively more rapid along the initially eastern margin of the

CLIP than along its southern margin.

A more rapid transformation to subduction along the initially eastern margin of the CLIP

relative to its southern margin could be the result of a number of factors: (1) the site of initial

subduction nucleation along the initially eastern margin was closer to the periphery of the plume

head than along the southern margin; (2) there existed gross differences in the morphology of the

plume such that it was thinner along the initially eastern margin of the CLIP than along the

southern margin of the CLIP allowing for more relative subduction modification in an otherwise

equal interval of time; or (3) there existed gross differences in the chemistry of the plume. As

18
PISI along relatively more felsic plumes leads to much faster subduction inception than PISI

along more mafic plumes (Ueda et al., 2008), the plume may have been relatively more felsic

along its eastern margin relative to the southern margin. Whether this is related to the existence

of circa 93-79 Ma felsic, subduction-related intrusives along the initially eastern margin of the

CLIP in western Ecuador, western Colombia, Aruba and Curacao which are not present along the

southern margin of the CLIP in Costa Rica and Panama is uncertain.

3.3 Primitive mantle-normalized plots

We note also here that that some volcanic rocks interpreted as CLIP also display

characteristics suggestive of an arc as opposed to a within-plate environment of formation on

primitive mantle-normalized plots (Supplementary Fig. 5). The primitive-mantle normalized

plots in Supplementary Fig. 5a again illustrate the conspicuous resemblance of the 140-110 Ma

Nicoya Complex basalts with mean global OPB (n > 350 samples with reported major element

analyses but varying completeness of trace elements); however, the primitive mantle normalized

plots do not as effectively discriminate the older and younger Nicoya basalts as do the N-MORB

normalized plots (Supplementary Fig. 4a), but nonetheless illustrate the greater overall depletion

in incompatible elements and the relative greater enrichment of Ba relative to Nb and Th

(Supplementary Fig. 5a).

Other primitive-mantle normalized plots of selected post-100 Ma Caribbean ‘plateau’

volcanic rocks are shown in Supplementary. Figs. 5c-e, which further demonstrate the existence

of ‘arc-like’ signatures. For example, some lavas from the 89-85 Ma Azuero Plateau (Panama),

79-72 Ma Serrania de Baudo unit (western Colombia), 91-81 Ma Gorgona (western Colombia)

and 90-85 Ma Pallatanga (western Ecuador) units display ‘relatively prominent’ negative-Nb

19
anomalies (e.g., PAN-05-003, Azuero Plateau, and SDB10 and PO 1476, Serrania de Baudo and

Pallantanga respectively, Supplementary Figs. 5c-e); and sometimes negative Zr- and Ti-

anomalies (e.g, Azuero Plateau sample PAN-05-003, Supplementary Fig. 5c). A negative Nb-

anomaly with respect to both Th and Ce is commonly used to infer a subduction signature

(Pearce, 1983) but the mean PM-normalized signatures of both subduction initiation rule

ophiolite lower unit (SIRO LU) basalts and IBM forearc basalts (IBMFAB) which formed at the

earliest stage of subduction initation, exhibit negative Nb with respect to Ce only (i.e., primitive

mantle-normalized Th that is less than Nb) (Supplementary Fig. 5b). We note also that primitive

mantle signatures of the mean composition of basalts of both the Golfito Arc (of Buchs et al.,

2010) and the Sona-Azuero Arc (Supplementary Fig. 5c) exhibit negative Nb with respect to Ce

only and not Th, which again reiterates that N-MORB normalized plots (Supplementary Fig. 4)

are superior to primitive mantle normalized plots for detecting subtle subduction modification

signatures.

The primitive mantle normalized signatures of the aforementioned southern CLIP basalts

are unexpected from OPB but consistent with SI magmas. Some chemical characteristics of

subduction-related magmas such as negative Nb-anomalies can be produced when mantle-

derived mafic magmas lacking these signatures assimilate continental crust. There are at least

three reasons why continental crust contamination was unimportant: (1) similar volcanic

sequences are developed in regions where there is no continental crust, such as Costa Rica and

Panama; (2) there is no co-variation of subduction-related signatures with indices of fractionation,

such as decreasing MgO content; and (3) isotopic compositions show no evidence of crustal

contamination (see main text).

20
3.4 Chondrite-normalized REE and primitive mantle-normalized incompatible element plots of

felsic, arc -related intrusives

Chondrite-normalized REE and primitive mantle-normalized plots of 95-79 Ma felsic,

arc-related units are provided in Supplementary Fig. 6 to supplement the text in Section 5.2.

4. Additional tectonic considerations

Here we provide a discussion of the significance and consequences of additional

tectonic considerations for our PISI model.

4.1 Tectonic significance of the Bolivar Ultramafic Complex

The 99-87 Ma Bolivar Ultramafic Complex (BUC) (Kerr et al., 2004; Serrano et al.,

2011; Villagómez et al., 2011) occurs along the SE margin of the Western Cordillera of western

Colombia (6bii in Fig. 1). As it represents one of the oldest post-100 Ma units and temporally

overlaps the oldest post-100 Ma unit (the 101-98 Ma Volcanic, Barosso and Amaime formations

comprising the Western Cordillera, Villagómez et al., 2011) its petrogenetic interpretation is

important for tectonic models of the region.

The BUC comprises a lower section of ultramafic cumulates successively overlain by

layered and isotropic gabbroic rocks which grade into and are intruded by mafic pegmatites

(Kerr et al., 2004); intermediate and felsic pegmatites as well as felsite dykes are also locally

present. Kerr et al. (2004) report negative primitive mantle normalized Nb-anomalies for felsic

and intermediate BUC pegmatites, but conclude that BUC chemistry is not consistent with a

subduction origin and that the BUC instead represents an integral part of the plume-derived CLIP.

Kerr et al (2004) suggest that the negative-Nb anomaly exhibited by the intermediate and felsic

21
pegmatites to be the result of plagioclase accumulation and base this interpretation in part on the

fact that these pegmatites have positive Eu-anomalies. However, BUC basalts have similar but

less fractionated chondrite-normalized REE and N-MORB normalized patterns as 101-75 Ma

Western Cordillera basalts (Supplementary Figs. 7a, b); the Western Cordillera basalts exhibit

among the highest Ba/Th ratios (means of 400-500, which is about 5-6 times that of global OPB

with mean Ba/Th of 78 and 140-110 Ma Nicoya basalts with mean Ba/Th of 87, Figs. 9g, 10b)

indicating substantial shallow subduction contributions. Furthermore, the BUC basalts, mafic

pegmatites and (Vijes) felsite dykes also exhibit, subtle, negative N-MORB normalized Nb-

anomalies (with respect to Th) (Supplementary Figs. 7c-h) similar to the Golfito and Sona-

Azuero Arcs of Costa Rica and Panama but display either no primitive mantle normalized Eu-

anomaly or negative-Eu anomalies (e.g., Fig. 7 of Kerr et al., 2004). Plagioclase accumulation

thus cannot explain the negative-Nb anomalies. As shown in Fig. 11, the Bolivar felsic

pegmatites and felsite dykes are compositionally similar to 99-79 Ma felsic, arc-related

intrusives and ~75-69 Panamanian arc granitoids, all of which are consistent as primitive island

arc granitoids.

We note also here that Mamberti et al. (2004) provide a rather unorthodox interpretation

of negative Nb- and Zr-anomalies in clinopyroxenes and host gabbros of the western Ecuadorian
40
San Juan oceanic plateau unit in which they obtained a Ar/39Ar plateau age of 99.2±1.3 Ma.

Mamberti et al. (2004) suggest the negative HFSE-anomalies to be the result of the exclusion of

these elements from the structure of cumulus clinopyroxene and thus do not indicate a

subduction zone setting. Similar to what we describe for circa 101-75 Ma Western Cordillera

(Colombia) basalts above, 90-85 Ma western Ecuadorian oceanic plateau basalts, picrites and

ankaramites also exhibit amongst the highest Ba/Th (means of 279, 232, 257, Table 3) of all

22
post-100 Ma units. We suggest instead that in both cases (Bolivar and San Juan) the negative

HFSE-anomalies are ‘real’ and indicative of subduction modification that began at or just after

100 Ma.

4.2 Evidence of Late Cretaceous arc construction in the Leeward Antilles

Of the multitude of papers dealing with the Late Cretaceous to Eocene tectonic evolution

of the southern Caribbean and NW South American region, we have never seen cited the paper

by Weber et al. (2009) which describes the formation of the forearc-derived Cabo de la Vela

Ultramafic Complex (CVC) , Guajira Peninsula, northernmost Colombia (7, Fig. 1) . Yet, this

study provides crucial evidence for Late Cretaceous arc construction in the region between the

Leeward Antilles and western Colombia at this time (as the complex is located at the

northeasterly-most extent of Colombia, some 200-300 kms west of Aruba, Fig. 1, its location

could arguably be considered as comprising the southernmost Leeward Antilles). Although

radiometric ages for CVC arc basalts are K-Ar total fusion ages and yield high uncertainties

(three reported ages are 73.5±4.2, 69.5±4.9 and 77.3±5.4 Ma), the combined chemistry

(Supplementary Fig. 8) and geochronological constraints demonstrate forearc construction and

emplacement in the region in the Late Cretaceous. We suggest that the apparent termination of

arc-related magmatism circa 70 Ma (Weber et al., 2009) is related to South America-forearc

collision at 73-70 Ma above an west-dipping subduction system which resulted in emplacement

of arc and ‘plateau’ units soon thereafter (Vallejo et al., 2009), e.g., the 79-72 Ma Serrania de

Baudo unit, western Colombia and the coupled 90-85 Ma oceanic plateau and 85-72 Ma Rio

Cala Arc, western Ecuador.

23
4.3 Significance of a pre-plume subduction regime

Evidence of subduction along NW South America dates back to Jurassic times (e.g.,

Aspden et al., 1987; Jaillard et al., 1990). Although a pre-existing trench could represent a

lithospheric weakness upon which (west-dipping) subduction (e.g., Vallejo et al., 2009)

nucleated off of NW South America following plume extrusion after 100 Ma, to our knowledge

there is no evidence of subduction to the south of Central America before 90 Ma and it is

difficult to envision how a roughly N-S aligned lithospheric weakness in the east could propagate

to a region 100s or 1000s of km to the SW at the foci of Central American volcanic arc SI which

would have occurred along a roughly E-W trending lineament. Nevertheless, the apparent faster

rate of subduction inception along NW South America versus along Central America (see

Section 5, main text) may have been facilitated by the existence of a fossil trench off NW South

America.

4.4 Metamorphic sole beneath the Pallatanga Unit in western Ecuador

Vallejo et al. (2006) report an age of 84.72.2 Ma for the Totoras amphibolite (Jaillard

et al., 2004; Beaudon et al., 2005) which is juxtaposed against unmetamorphosed mafic rocks of

the Pallatanga Unit (PU, interpreted as CLIP) in western Ecuador. According to Jaillard et al.

(2004) and Beaudon et al. (2005) these amphibolites show OPB geochemical affinities and

Beaudon et al. (2005) suggest that the sequence formed by metamorphism of the CLIP.

Metamorphic soles are a common feature of many forearc-derived ophiolites and we suggest that

this amphibolite sequence represents the metamorphosed sole of the PU associated with incipient

west-dipping subduction and formation of the 90-85 PU. It has been recognized for some time

that the age of metamorphic sole formation is essentially the same age or only slightly younger

24
than the precursor supra-subduction zone (SSZ) ophiolite which formed above the subducted and

metamorphosed slab. Examples which report near synchronous ophiolite and sole formation are

found in western Pacific ophiolites, e.g., the Palawan ophiolite of the Philippines (Encanacion et

al., 2001), the Papuan Ultramafic Belt ophiolite of Papua-New Guinea (Lus et al., 2004) and

Tethyan-type ophiolites in general (Hacker et al., 1996; Shervais, 2001, and references therein).

These observations are relevant to western Ecuador in the Late Cretaceous as the PU and

the subsequent Rio Cala Arc constructed atop the PU were emplaced upon NW South America

as a result of South America-CLIP collision (Vallejo et al., 2009), i.e., in the same fashion as

Tethyan-type ophiolites, many of which are characterized by a metamorphic sole (see Whattam

and Stern, 2011 and references therein). As west-dipping SI must have preceded metamorphic

sole formation (see also Section 5.5.2, main text), this lends further credence to our model which

infers SI prior to 85 Ma as the result of plume extrusion.

4.5 Further discussion on necessity of west-dipping subduction to facilitate eastwards

emplacement of PAR units upon South America

Late Cretaceous Tethyan-type ophiolites of the Eastern Mediterranean-Persian Gulf

region represent oceanic lithosphere generated by seafloor spreading in what will ultimately

become the forearc during SI (Whattam and Stern, 2011 and references therein). Such

ophiolites are emplaced above the same subduction zone which generated the ophiolite. As

summarized by Whattam (2009) generation and emplacement of most ophiolites thus mark the

start and demise of a subduction zone. In the case of Late Cretaceous Tethyan ophiolites,

emplacement results when buoyant continental lithosphere jams the subduction zone and the

ophiolitic forearc slides over the partially subducted continental lithosphere. This process is akin

25
to placing a spatula (subducting continental crust) beneath a fried egg (forearc ophiolite) cooked

(generated) on a frying pan (astheospheric mantle); if one continues to slide the spatula beneath

the egg, the latter will eventually slide up the spatula handle (inboard continental crust). An

identical process resulted in the emplacement of Caribbbean plateau- and arc-related units upon

western Ecuador at 73-70 Ma as the result of South America-forearc collision (Vallejo et al.,

2009). We suggest that termination of arc-related magmatism to the north as recorded by the 79-

72 Ma Serrania de Baudo unit, western Colombia and the ~85-70 Ma Cabo de la Vela forearc-

derived complex, northernmost Colombia (Supplementary Section 4, Supplementary Fig. 8) was

also the result of NW South America-forearc collision which resulted in the eastwards

emplacement of these units upon the eastern margin of NW South America soon thereafter.

‘Emplacement’ of Cordilleran-type ophiolites (see reviews of Shervais, 2001 and

Beccaluva et al., 2004) on the other hand, are the result of gradual, in-situ, accretionary uplift

(Shervais, 2001) due to continual or renewed subduction of buoyant crust and sediments beneath

the original subduction zone that generated the ophiolite. Thus, oceanic complexes in the forearc

along the Pacific coast of Costa Rica and Panama can be considered ‘Cordilleran-like’ and those

along western Ecuador and Colombia as ‘Tethyan-like’.

4.6 Tectonic evolution at 140-100 Ma?

It is noteworthy that inferred inception of the Greater Antilles Arc (GAA, or the ‘Great

Arc of the Caribbean’, see Burke, 1988 and earlier papers) at ~ 135 Ma (see Pindell et al., 2011)

coincides with the earliest recognized phase of Caribbean plume-related magmatism at 139 Ma

as preserved in the Nicoya Complex (Hoernle et al., 2004). According to Pindell et al. (2011), an

'Inter-American transform' connected a NE-dipping system beneath North America with a NE-

26
dipping system beneath South America at 141 Ma. At 130 Ma, both NE-dipping systems had

rolled back to the SE and SW-dipping subduction (i.e., earliest GAA subduction) had nucleated

along the former transform so that there existed two subduction zones of opposing polarity, i.e.,

SW-dipping at the proto-GAA and NE-dipping beneath both North and South America (see

Figure 3 of Pindell et al., 2011). If the oldest Nicoya rocks indeed represent earliest plume

magmatism is it possible that both NE- and SW-dipping SI also occurred at 140-135 Ma on

either sides of the 'initial plume' in response to plume emplacement?

Supplemental Figure captions

Supplementary Figure 1. Distribution and frequency of radiometric age determinations of mostly

post-100 Ma plateau- and arc-related units of (a) Costa Rica, (b) western Ecuador and western

Colombia, (c) Aruba and Curacao, (d) the central Caribbean and (e) Central Hispaniola (Haiti

and Dominican Republic) vs. the distribution and frequency of all ages (grey shade). Individual

plume- and arc-related units compiled in the histogram are included in the boxes in each plot.

Units numbered as in Fig. 1 of main text.

Supplementary Figure 2. Distribution and frequency of radiometric age determinations of mostly

post-100 Ma plateau- and arc-related units according to region and unit in (a) Costa Rica and

Panama, (b) western Ecuador and western Colombia, (c) western Colombia only, (d) Aruba and

Curacao and (e) the central Caribbean Plate in the Caribbean Sea and the northern Caribbean

Plate in Central Hispaniola (Haiti and Dominican Republic). The bold L, M, Lx, V and W in

brackets beside San Juan (in b), Volcanic Formation (in c) and in (d) are to indicate that these age

data are from Lapierre et al. (2000), Mamberti et al. (2003), Luzieux et al. (2006), Villagómez et

27
al. (2011) and Wright and Wyld (2011), respectively. References for other units plotted are

provided in Table 2. For biostratigraphic age determinations of the Nicoya Complex, sensu lato,

see Bandini et al. (2008). Units numbered as in Fig. 1.

Supplementary Figure 3. Nb/Y-Zr/Y systematics of (a) oceanic plateau basalts (OPB) comprising

data from the Ontong Java, Manihiki, Hikurangi and Kerguelen (plateaus and the Shatsky Rise;

(b) basalts of the Quaternary Central American (Costa Rica, Nicaragua, Guatemala and El

Salvador) and Izu-Bonin Marianas Volcanic arcs (CAVAB, IBMVAB); and (c) lower and upper

units (LU, UU of subduction initiation rule, SIR, ophiolites and IBM forearc basalts (FAB);

versus (d-l) basalts and related volcanic rocks of post-100 Ma Caribbean plume- and arc-related

(PAR) units, i.e. lower units interpreted as CLIP plateau (squares) and upper units interpreted as

arc sequences (circles) built upon the CLIP plateau. Italicized numbers in brackets beside the

region name correspond to unit numbers assigned in Fig. 1 of the main text. Samples in this and

other geochemical plots are lavas and dykes of basaltic composition except for the SIRO upper

unit samples in (c) which are basaltic-andesites; the depleted Gorgona samples in (g) which also

include komatiites and picrites; the western Ecuador plateau samples in (i) which also comprise

picrites and ankaramites; and the Central Cordillera (CC, western Colombia) and Curacao Lava

Formation (CLF) plateau samples in (h) and (j) which also include picrites. Plot and fields are

from Fitton et al. (1997) and references therein. References for Caribbean and SW American

datasets are provided in Table 2 of the main text; references for OPB, CAVAB, IBMVAB, SIR

ophiolites and IBMFAB are in Section 3.2. See Fig. 1 of for locations. Superscript: a (beside

PUPF in l), in addition to data of the Pallatanga Unit and the Pinon Formation (PUPF), the data

also comprises that of the related Guaranda, Pedernales and other oceanic units in Western

28
Ecuador interpreted as circa 95-90 Ma CLIP (see Mamberti et al., 2003; Allibon et al., 2008).

The italicized V and K in brackets beside VABF (WC) and WC in (h) is to indicate that these are

data of Villagómez et al. (2011) and Kerr et al. (1997) respectively. The italicized R beside unit

abbreviations in (j-l) is to indicate that all samples of these datasets have accompanying

radiometric age data. Unit abbreviations can be ascertained from Fig. 1. Units numbered as in

Fig. 1.

Supplementary Figure 4. N-MORB normalized plots of plateau- and arc-related units from along

trace 1 of plume-induced subduction initiation, (PISI Tr 1) in (a-f) Costa Rica and Panama; PISI

Tr2 in (g-l) western Colombia and western Ecuador; and PISI Tr3, in (m-o) Aruba and Curacao;

and (p-r) the central and northern segments of the CLIP in the Caribbean Sea and Central

Hispaniola (Haiti and Dominican Republic). N-MORB abundances are from Sun and

McDonough (1989). The italicized H and B in (d), V and K in (h), and K and L in (n, o) is to

indicate that these sample sets are from Hauff et al. (2000) and Buchs et al. (2010); Villagómez

et al. (2011) and Kerr et al. (1997); Kerr et al. (1996b); and Loewen et al. (2013), respectively.

The abbreviation ‘ptb’ in (n) stands for ‘presumed to be’. Superscript: a, beside Pallantanga Unit

and Pinon Formation in (k) is to indicate that in addition to data of the Pallantanga Unit and the

Pinon Formation (PUPF), the data also comprises data of the related Guaranda, Pedernales and

other oceanic units in Western Ecuador interpreted as circa 95-90 Ma CLIP (see Mamberti et al.,

2003; Allibon et al., 2008). Unit abbreviations can be ascertained from Fig. 1. Units numbered as

in Fig. 1.

29
Supplementary Figure 5. Primitive mantle (PM)-normalized plots of the mean of (a) oceanic

plateau basalts (OPB), 140-110 Ma Nicoya Complex basalts and 98-82 Ma Nicoya Complex

basalts, (b) Quaternary Central American Volcanic Arc basalts (CAVAB), Izu-Bonin Marianas

Volcanic Arc basalts (IBMVAB), IBM forearc basalts (IBMFAB) and subduction initation rule

ophiolite (SIRO) lower unit basalts; and selected basalts and komatiiites of post-100 Ma plume-

and arc-related (PAR) units in (c) southern Costa Rica and Panama, (d) western Colombia and (e)

western Ecuador. Italicized numbers in brackets beside the region name correspond to unit

numbers assigned in Fig. 1 of main text. All plotted samples are lavas or dykes of basaltic

composition apart from the Gorgona komatiites in (d). In (c-e) selected samples interpreted as

plateau but which display negative-Nb anomalies are plotted alongside the mean composition of

associated plateau and arc lavas and intrusives of basaltic composition. PM values are from Sun

and McDonough (1989). Note the sample size (n) for the units in (a) and (b) are approximate as

these represent the number of samples with major element analyses but not all samples have a

full suite of trace element analyses. References for all Caribbean and NW South American

datasets are in Table 2 of the main text, while references for OPB, CAVAB, IBMVAB,

IBMFAB, and SIRO LU are in Section 3.2 of main text. Unit abbreviations can be ascertained

from Fig. 1. Units numbered as in Fig. 1.

Supplementary Figure 6. Chondrite-normalized REE plots (a, c, d, e, g, i) and N-MORB-

normalized plots (b, d, f, h, j) of 95-79 Ma subduction-related, felsic, arc-like intrusives of NW

South America and the Greater Antilles versus modern-day adakites and Archean tonalite-

trondhejemite-granodiorite (TTG) suites in (a-h) and younger granitoids of the Central American

Volcanic Arc in (i) and (j). Chondrite REE and N-MORB abundances are from Nakamura (1974)

30
and Sun and McDonough (1989), respectively. All data is of samples with 65-75 wt. % SiO2

except the lone Puliji Granite suite sample with 57 wt. %. SiO2. The fields for adakites formed

via partial melting of delaminated (delam.) lower crust, oceanic (ocean.) crust and thickened

lower crust are from references as compiled and listed in Fig. 8 of Whattam et al. (2012). The

mean Archaen TTG chondrite-normalized pattern in (a) is from Moyen and Martin (2012).

References for age and geochemical data of NW South American and Greater Antillean

intrusives are listed in Table 2 of main text and references for the Panamanian arc sequences are

listed in the figure caption of Fig. 11 in the main text. Abbreviations: COL, Colombia; ECU,

Ecuador; LA, Lesser Antilles. Unit abbreviations can be ascertained from Fig. 1Units numbered

as in Fig. 1.

Supplementary Figure 7. Chondrite-normalized REE plots (a, c, d, e, g) and N-MORB-

normalized plots (b, d, f, h, ) of 99-87 Ma Bolivar Ultramafic Complex (BUC) basalts (a, b),

mafic pegmatites (c, d), felsic pegmatites (e, f) and (Vijes) felsite dykes (g, h). Chondrite

normalized REE and N-MORB normalized abundances are from Nakamura (1974) and Sun and

McDonough (1989), respectively. Data for the BUC is from Kerr et al. (2004). In (g) and (h),

only sample Pan 92-14 (Kerr et al., 2004) which has a full suite of trace element analyzes is

plotted. In (a) and (b) the mean composition of 101-98 Ma Volcanic, Barosso and Amaime

formation basalts (Villagómez et al., 2011) of the Western Cordillera, western Colombia and 94-

75 Ma Western Cordillera basalts (Kerr et al., 1997) are plotted against the BUC basalt.

References for other units plotted in (g) and (h) (grey shade) are listed in Table 2 of main text.

Unit abbreviations can be ascertained from Fig. 1. Units are as numbered in Fig. 1.

31
Supplementary Figure 8. Chondrite normalized REE and N-MORB normalized abundances

lower basaltic-andesite gabbros and upper, latest-stage basaltic andesite dykes (a, b) and spinel

Mg# vs. spinel Cr# of spinels of peridotites (c) of the Cabo de la Vela Ultramafic Complex

(Weber et al., 2009), Guajira peninsula, northernmost Colombia (unit 7, Fig. 1). Chondrite

normalized REE and N-MORB normalized abundances are from Nakamura (1974) and Sun and

McDonough (1989), respectively. Fields for spinels from abyssal and forearc spinel peridotites in

(c) are from Dick and Bullen (1984) Parkinson and Pearce (1998), respectively.

References

Allibon, J., Monjoie, P., Lapierre H., Jaillard, E., Bussy, F., Bosch, D., Senebier, F., 2008. The

contribution of the young Cretaceous Caribbean Oceanic Plateau to the genesis of late

Cretaceous arc magmatism in the Cordillera Occidental of Ecuador. Journal of South

American Earth Sciences 26, 355-368.

Bandini, A. N., Flores, K., Baumgartner, P.O., Jackett, S.‐J., Denyer, P., 2008. Late Cretaceous

and Paleogene radiolaria from the Nicoya Peninsula, Costa Rica: A tectonostratigraphic

application. Stratigraphy 5, 3-21.

Beccaluva, L., Coltori, M., Giunta, G., Siena, F., 2004. Tethyan vs. Cordilleran ophiolites: a

reappraisal of distinctive tectonomagmatic features of supra-subduction complexes in

relation to the subduction mode. Tectonophysics 393, 163–174

Beaudon, E., Martelat, J.E., Amortegui, A., Lapierre, H., Jaillard, E., 2005. Metabasites de la

cordillere occidentale d’Equateur, temoins du soubassement oceanique des Andes ’Equateur:

Comptes Rendus Geoscience 337, 625–634, doi: 10.1016/j.crte.2005.01.002.

32
Buchs, D.M., Arculus, R.J. Baumgartner, P.O., Baumgartner-Mora, C., Ulianov, A., 2010. Late

Cretaceous arc development on the SW margin of the Caribbean Plate: Insights from the

Golfito, Costa Rica, and Azuero, Panama, complexes. Geochemistry, Geophysics,

Geosystems 11, doi:10.1029/2009GC002901.

Dick, H.J.B., Bullen, T., 1984. Chromian spinel as a petrogenetic indicator in abyssal and alpine-

type peridotites and spatially associated lavas. Contributions to Mineralogy and Petrology

86, 54-76.

Encanacion, J., Fernandez, D., Mattinson, J., 2001. Subduction initiation by extrusion tectonics?

Evidence from the Palawan Ophiolite, Philippines. Eos Transactions AGU 82(47), Fall

Meeting Supplement, Abstract T42G-03.

Fitton, J.G., Saunders, A.D., Norry, M.J., Hardarson B.S., and Taylor, R.N., 1997. Thermal and

chemical structure of the Iceland plume. Earth and Planetary Science Letters 153, 197–208.

Gazel, E., Hoerle, K., Carr, M.J., Herzberg, C., Saginor, I., van den Bogaard, P., Hauff, F.,

Feigenson, M., Swisher III, C., 2011. Plume-subduction interaction in southern Central

America: Mantle upwelling and slab melting. Lithos 121, 117-134.

Gazel, E., Carr, M.J., Hoernle, K., Feigenson, M.D., Szymanski, D., Hauff, F., van den Bogaard,

P., 2009. Galapagos-OIB signature in southern Central America: mantle refertilization by

arc–hot spot interaction. Geochemistry, Geophysics, Geosystems 10 (32), Q02S11.

doi:10.1029/2008GC002246.

Geldmacher J., Heydolph K., Murphy D., Romanova I., Mahoney J., Hoernle K., 2012.

Combined Sr, Nd, Pb and Hf isotopic constraints on the origin of Shatsky Rise (NW

Pacific). AGU Fall Meeting, DI51A-2339.

33
Hacker, B.R., Mosenfelder, J.L., Gnos E., 1996. Rapid emplacement of the Oman ophiolite:

Thermal and geochronologic constraints. Tectonics 15, 1230–1247.

Hastie, A.R., Mitchell, S.F., Treloar, P.J., Kerr, A.C., Neill, I., Barfo, D.N., 2013. Geochemical

components in a Cretaceous island arc: The Th/La–(Ce/Ce*)Nd diagram and implications

for subduction initiation in the inter-American region. Lithos 162-163, 57-69.

Hauff, F., Hoernle, K., van der Bogard, P., 2000. Age and geochemistry of basaltic complexes in

western Costa Rica: Contributions to the geotectonic evolution of Central America.

Geochemistry, Geophysics, Geosystems 1, Paper number 1999GC000020.

Hoernle, K., Hauff, F., van den Bogaard, P., 2004. 70 m.y. history (139-69 Ma) for the

Caribbean large igneous province. Geology 32, 697-700.

Ingle, S., Mahoney, J.J., Sato, H., Coffin, M.F., Kimura, J-I, Hirano, N., Nakanishi, 2007.

Depleted mantle wedge and sediment fingerprint in unusual basalts from the Manihiki

Plateau, central Pacific Ocean. Geology 35, 595-598. Doi: 10.1130/G23741A.1

Jaillard, E., Ordoñez, M., Suárez, J., Toro, J., Iza, D., Lugo, W., 2004. Stratigraphy of the Late

Cretaceous-Paleogene deposits of the Cordillera Occidental of central Ecuador: geodynamic

implications. Journal of South American Earth Sciences 17, 49–58.

Kerr, A.C., Marriner, G.F., Arndt, N.T., Tarney, J., Nivia, A., Saunders, A.D., Duncan, R.A.,

1996a. The petrogenesis of Gorgona komatiites, picrites and basalts: new field, petrographic

and geochemical constraints. Lithos 37, 245-260.

Kerr, A.C., Tarney, J., Marriner, G.F., Klaver, G.T., Saunders, A.D., Thirlwall, M.F., 1996b.

The geochemistry and petrogenesis of the Late-Cretaceous picrites and basalts of Curacao,

Netherlands Antilles: a remnant of an oceanic plateau: Contributions to Mineralogy and

Petrology 124, 29-43.

34
Kerr, A.C., Marriner, G.F., Tarney, J., Nivia, A., Saunders, A.D., Thirlwall, M.J., Sinton, C.W.,

1997. Cretaceous basaltic terranes in western Colombia: Elemental, chronological and Sr-

Nd isotopic constraints on petrogenesis. Journal of Petrology 38, 677-702.

Kerr, A.C., Pearson, D.G. , Nowell, G.M., 2009. Magma source evolution beneath the Caribbean

oceanic plateau: new insights from elemental and Sr-Nd-Pb-Hf isotopic studies of ODP Leg

165 Site 1001 basalts. In: James, K.H., Lorente, M.A., Pindell, J.L., (Eds.), The Origin and

Evolution of the Caribbean Plate. Geological Society of London Special Publication 328, pp.

809-827.

Kerr, A.C., Tarney, J., Kempton, P.D., Pringle, M., Nivia, I., 2004. Mafic Pegmatites Intruding

Oceanic Plateau Gabbros and Ultramafic Cumulates from Bolívar, Colombia: Evidence for

a ‘Wet’ Mantle Plume? Journal of Petrology 45, 1877-1906.

Lapierre, H., Bosch, D., Dupuis, V., Polve, M., Maury, R., Hernandez, J., Monie, P.,

Yeghicheyan, D., Jaillard, E., Tardy, M., Mercier de Lepinay, B., Mamberti, M., Desmet, A.,

Keller, F., and Senebier, F., 2000, Multiple plume events in the genesis of the peri-

Caribbean Cretaceous oceanic plateau province: Journal of Geophysical Research 105,

8403–8421, doi: 10.1029/1998JB900091.

Loewen, M.W., Duncan, R.A., Kent, A.J.R., Krawl, K., 2013. Prolonged plume volcanism in the

Caribbean Large Igneous Province: New insights from Curacao and Haiti. Geochemistry,

Geophysics, Geosystems 14, doi: 10.1002/ggge.20273.

Lus, W.Y., McDougall, I., Davies, H.L., 2004. Age of the metamorphic sole of the Papuan

Ultramafic Belt ophiolite, Papua New Guinea. Tectonophysics 392, 85–101.

Luzieux, L.D.A., Heller, F., Spikings, F., Vallejo, C.F., Winkler, W., 2006, Origin and

Cretaceous tectonic history of the coastal Ecuadorian forearc between 1°N and 3°S:

35
 Paleomagnetic, radiometric and fossil evidence: Earth and Planetary Science Letters 249,

400-414, doi: 10.1016/j.epsl.2006.07.008.

Mamberti, M., Lapierre, H., Bosch, D., Jaillard, E., Ethiene, R., Hernandez, J., Polvé, M., 2003.

Accreted fragments of the Late Cretaceous Caribbean-Colombian Plateau in Ecuador.

Lithos 66, 173-199.

Mamberti, M., Lapierre, H., Bosch, D., Jaillard, E., Hernandez, J., Polvé, M., 2004. The Early

Cretaceous San Juan Plutonic suite, Ecuador: a magma chamber in an oceanic plateau?

Canadian Journal of Earth Sciences 41, 1237-1258.

Moyen, J - F & Martin, H., 2012. 40 years of TTG research. Lithos 148, 312-336.

Nakamura, N., 1974. Determination of REE, Ba, Fe, Mg, Na and K in carbonaceous and

ordinary meteorites. Geochimica et Cosmochimica Acta 38, 757-775.

Parkinson, I.J., Pearce, J.A., 1998. Peridotites from the Izu–Bonin–Mariana Forearc (ODP

Leg125): Evidence for mantle melting and melt-mantle interaction in a suprasubduction

zone setting. Journal of Petrology 39, 1577-1618.

Pearce, J.A., 1983. Role of the subcontinental lithosphere in magma genesis at active continental

margins. In: Hawkesworth, C.J., Norry, M.J. (Eds.), Continental Basalts and Mantle

Xenoliths, Shiva Publishing, Cheshire, pp. 230-49.

Pearce, J.A., Stern, R.J., Bloomer, S.H., Fryer, P., 2005. Geochemical mapping of the Mariana

arc-basin system: Implications for the nature and distribution of subduction components.

Geochemistry, Geophysics, Geosystems 6, Q07006, doi:10.1029/2004GC000895.

Pindell, J., Maresch, W.V., Martens, U., Stanek, K., 2011. The Greater Antillean Arc: Early

Cretaceous origin and proposed relationship to Central American subduction mélanges:

36
 implications for models of Caribbean evolution: International Geology Review, doi:

10.1080/00206814.2010.510008.

Reagan, M.K., Ishiuka, O., Stern, R.J., Kelley, K.A., Ohara, Y., Blichert-Toft, J., Bloomer, S.H.,

Cash, J., Fryer, P., Hanan, B.B., Hickey-Vargas, R., Ishii, T., Kimura, J.I., Peate, D.W.,

Rowe, M.C., Woods, M., 2010. Fore-arc basalts and subduction initiation in the Izu-Bonin-

Mariana system. Geochemistry, Geophysics, Geosyststems 11, Q03X12,

doi:10.1029/2009GC002871.

Révillon, S., Arndt, N.T., Chauvel, C., Hallot, E., 2000. Geochemical study of ultramafic

volcanic and plutonic rocks from Gorgona Island, Colombia: The plumbing system of an

oceanic plateau. Journal of Petrology 41, 1127-1153.

Serrano, L., Ferrari, L., López Martínez, M., Maria Petrone, C., Jaramillo, C., 2011. An

integrative geologic, geochronologic and geochemical study of Gorgona Island, Colombia:

Implications for the formation of the Caribbean Large Igneous Province. Earth and

Planetary Science Letters 309, 324-336.

Shervais, J.W., 2001. Birth, death, and resurrection: The life cycle of suprasubduction zone

ophiolites. Geochemistry, Geophysics, Geosystems 2, doi:10.1029/2000GC000080.

Sinton, C.W., Duncan, R.A., Denyer, P., 1997. Nicoya Peninsula, Costa Rica: A single suite of

Caribbean oceanic plateau magmas. Journal of Geophysical Research 102, 507-515,

doi:10.1029/97JB00681.

Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic systematic of oceanic basalts:

implications for mantle composition and processes. In: Saunders, A.D., Norry, M.J. (Eds.),

Magmatism in the Ocean Basins. Geological Society of London Special Publication 42, pp

313-345.

37
Vallejo, C., Spikings, R.A., Luzieux, L., Winkler, W., Chew, D., Page, L., 2006. The early

interaction between the Caribbean Plateau and the NW South American plate. Terra Nova

18, 264–269.

Vallejo, C., Winkler, W., Spikings, R.A., Luziuex, L., Heller, F., Bussy, F., 2009. In: Kay, S.M.,

Ramos, V.A., Dickinson, W.R. (Eds.), Backbone of the Americas: Shallow Subduction,

Plateau Uplift, and Ridge and Terrane Collision. Geological Society of America Memoir

n204, pp. 197-216, doi:10.1130/2009.1204.

Villagómez, D., Spikings, R., Magna, T., Kammer, A., Winkler, W., Beltrán, A., 2011.

Geochronology, geochemistry and tectonic evolution of the Western and Central cordilleras

of Colombia. Lithos 125, 875-896.

Weber, M.B.I., Cardona, A., Paniagua, F., Cordani, U., Sepúlveda, L., Wilson, R., 2009. The

Cabo de la Vela Mafic–Ultramafic Complex, Northeastern Colombian Caribbean region: a

record of Multistage evolution of a late Cretaceous intra-oceanic arc. In: James, K. H.,

Lorente, M.A., Pindell, J.L. (Eds.), The Origin and Evolution of the Caribbean Plate.

Geological Society, London, Special Publications 328, pp. 547–566.

Wegner W., Wörner, G. Harmon, R.S., Jicha, B.R., 2011. Magmatic history and evolution of the

Central American land bridge in Panama since Cretaceous time. Geological Society of

America Bulletin 123, 703-734.

Whattam, S.A., 2009. Arc-continent collisional orogenesis in the SW Pacific and the nature,

source and correlation of emplaced ophiolitic nappe components. Lithos 113, 88-114.

Whattam, S.A., Montes, C., McFadden, R.R., Cardona, A., Ramirez, D. and Valencia, V., 2012,

Age and origin of earliest adakitic-like magmatism in Panama: implications for the tectonic

evolution of the Panamanian magmatic arc system: Lithos, v. 142-143, p. 226-244.

38
Whattam, S.A., Stern, R.J., 2011. The ‘subduction initiation rule’: a key for linking ophiolites,

intra-oceanic forearcs, and subduction initiation: Contributions to Mineralogy and Petrology

62, 1031-1045, doi:10.1007/s00410-011-0638-z.

Wörner, G., Harmon, R.S. , Wegner, W., 2009. In: Kay, S.M., Ramos, V.A., Dickinson, W.R.,

(Eds.), Backbone of the Americas: Shallow Subduction, Plateau Uplift, and Ridge and

Terrane Collision. Geological Society of America Memoir 204, pp. 183-196,

doi:10.1130/2009.1204.

Wright, J.E., Wyld, S. J., 2011. Late Cretaceous subduction initiation on the eastern margin of

Caribbean-Colombian oceanic plateau: One Great Arc of the Caribbean (?): Geosphere 7,

468-493, doi: 10.1130/GES00577.1.

39
 30
Costa Rica (1) OP Units:
25 Costa Rica: Nicoya Complex
(1a), Tortugal Complex (1b)
20 & Herradura (1c)

15 all regions
(1, 3-6, 8, 9, 11-15 )

10

5
a
30
western Ecuador & OP Unit
25 western Colombia (3, 4, 5, 6) western Colombia: Serrania de
Baudo (3), Gorgona Island, (4),
Central Cordillera (6a), Western
20 Cordillera (6b), Bolivar Ultramafic
Complex (6bii)
western Ecuador: San Juan Unit
15 (5a), Pinon Fm. (5ai)

10

5
b
30
Aruba & Curacao (8, 9) OP Unit
25 Aruba Lava Formation (8a),
Curacao Lava Formation (9a)
Frequency

20

15

10

5
c
30
Central segments of the OP Units
25 CLIP, Caribbean Sea Beata Ridge (11), DSDP Leg 15
(12), ODP Leg 165, Site 1001 (13)
(11, 12, 13)
20

15

10

5
d
30
Northern segment OP Units
25 of the CLIP, Haiti (14) & Dumisseau Fm., Haiti (14),
Lower Duarte Fm., Dominican
Domincan Republic (15) Republic (15)
20

15

10

5
e
140 130 120 110 100 90 80 70 60 50 40
Ma

Supplementary Fig. 1 Whattam and Stern 2014

 14

12 Costa Rica & Panama OP Unit

(1, 2) all (Costa Rica)
Nicoya Complex (1a)
10

Frequency
Tortugal Complex (1b)
Herradura (1c)
8

4 Tortugal
& Herradura
2
a
16
14 western Ecuador & OP Unit
western Colombia (3, 4, 5, 6) both regions
west. Colombia (3, 4, 6a, 6b, 6bi, 6bii)
12 west. Ecuador (5a, 5ai)
Frequency

Felsic arc-related units intrusive into OP

10
west. COL, AQB (6ai), BB (6biii)
west. ECU, Pujili Granite (5aiii)
8

4 San Juan (V, 2006) & Pinon (Lx)

San Juan (L)
San Juan (M)
2
b
14

12 western Colombia OP Unit

(3, 4, 6a, 6b) all of western Colombia
Central Cordillera (6a)
10
Frequency

western Cordillera (6b, 6bi)

Bolivar Ultramafic Complex (6bii)
8 Serrano de Baudo (3)
Gorgona Island (4)
6

4
Volcanic
Formation (WC)(V)
2
c
14

12 Aruba & Curacao (8, 9) OP Unit

both regions
Aruba Lava Formation (8a)
10
Frequency

Curacao Lava Formation (9a)

Felsic arc-related units intrusive into OP
8
Aruba Batholith (8b)
Curacao Diorites (9b)
6

4
(W) (W)

2
d
14

12 Central & northern segments OP Unit

of the CLIP in Caribbean all regions
Beata Ridge (11)
10 Sea & Haiti (11, 12, 13, 14)
Frequency

DSDP Leg 15 (12)

ODP Leg 165, Site 1001 (13)
8
Dumisseau Fm., Haiti (14)
Lower Duarte Fm., Dom. Rep. (15)
6

2
e
0
140 130 120 110 100 90 80 70 60 50 40
Ma

Supplementary Fig. 2 Whattam and Stern 2014

 10
a Oceanic Plateaus b Intra-oceanic Arcs c SIR ophiolites & IBMFAB
OPB (n = 307)
‘subduction OIB
1 contaminated’
Nb/Y

n=1 AP (from f)

PM
.1 rray ‘continental crust
nd a
Icela
contaminated’ SIRO LU (older, n = 11)
ORB SIRO UU (younger, n = 4)
N-M CAVAB (n = 106) IBMFAB (n = 13)
U1349A-SR >97 % OPB IBMVAB (n = 193)
10
d northwestern Costa Rica (1a-c) e southwestern Costa Rica (1d) f OP unit: Azuero Plateau
AP (2ai, bas), 89-85 Ma)
AP (2ai, enr-bas)
AP enriched Arc units: Sona-Azuero
1
Nb/Y

(from f) & Chagres-Bayano arcs

SAA (2aii, bas), 73-69 Ma
CBA (2b, bas), 70-39 Ma)
OP units: Nicoya, Tortugal
& Herradura complexes
.1 NC (1a, bas), 98-82 Ma OP unit: Golfito Complex
TC (1b, LREE-dep bas), 91-88 Ma GC (H) (1d, bas), 75-66 Ma
TC (1b, LREE-enr bas) Arc unit: Golfito Complex
HD (1c, bas), 88-82 Ma GC (B) (1d, bas) Panama (2a, b)
IBMFAB (from c) IBMFAB (from c)

10
g OP Units: Gorgona Island
Serrania de Baudo
h OP Units: Cent. Cord. & Volcanic, Amaime & Barros Fms., West. Cord.
CC (6a, bas), 97-86 Ma
I OP uni ts: Pal lat an ga Un it & Pi no n Fm.
a
PU PF (5a & 5a i, ba s, pi c, an k), 90 -85 Ma
GI (4, dep-bas, pic, kom), CC (6a, pic) Ar c uni t: Ri o Cal a Ar c
91-81 Ma VABF (V) (WC, 6bi, bas), 101-98 Ma RCA (5b , ba s) , ~8 5- 64 Ma
1
Nb/Y

GI (4, enr-bas) WC (K) (6b, bas), 94-75 Ma)

SDB (3, bas), 79-72 Ma ankaramites
basalts & picrites
SDB (3, enr-bas)

.1
n=3
western Colombia (3, 4) western Colombia (6c, d) western Ecuador (5a, c, d)

10
j Aruba & Curacao (8, 9) k OP units:
DSDP Leg 15 (12, bas), 97-87 Ma l northern segment of CLIP, Haiti (14)
DSDP Leg 15 (12, enr-bas)
ODP Site 1001 (13, bas), 82-80 Ma)
& Dominican Reb. (15)
R
BR (11, dol) 82-78 Ma
1
Nb/Y

R
BR (11, enr-bas), 76.87±2.14 Ma n=1
OP units: Aruba & Curacao Lava Fms. R
BR (11, dol), 55.20±1.4 Ma
ALF (8a, bas), 103-89 Ma
CLF (9a, bas), ptb ~90-88 Ma (K) OP units: Dumisseau Fm., Haiti
CLF R (9a, bas) 75.8±2.0 Ma & Lower Duarte Fm., Dom. Rep.
.1 CLF (9a, pic), ptb ~90-88 Ma (K) R
DF R(14, bas), 93.6±0.8 Ma
R
CLF (9a, bas), 92.0±1.0 Ma
CLF R (9a, bas), 74.2±2.4 Ma
center of CLIP, DF R (14, bas), 93-84 Ma
DF (14, bas), 82.8±0.7 Ma
R
CLF (9a, bas), 64-62 Ma Caribbean Sea (11, 12) LDF (15, diab) 88-85 Ma

.01
1 5 1 5 1 5 10
Zr/Y Zr/Y Zr/Y

Supplementary Fig. 3 Whattam and Stern 2014

 PISI tr 1, SFPH, Costa Rica & Panama (1, 2)
100
OP unit: Nicoya Complex, CR
NC (1a, bas), 140-110 Ma (n = 10)

Rock/N-MORB
10 mean OPB (n = 353)

a
100
OP unit: Nicoya Complex, CR
NC (1a, bas), 98-82 Ma (n = 41)
Rock/N-MORB

10
140-110 Ma Nicoya Complex
mean (from a)

b
100
OP units: Tortugal & Herradura complexes, CR
TC (1b, LREE-dep bas), 91-88 Ma (n = 1)
TC (1b, LREE-enr bas) (n = 1)
Rock/N-MORB

HC (1c, bas), 88-81 Ma (n = 7)

10

c
100
OP unit: Golfito Complex, CR
GC (H) (1d, bas), 75-66 Ma (n = 3)
Arc unit: Golfito Complex, CR
Rock/N-MORB

GC (B) (1d, bas) (n = 4)

10

d
100
OP unit: Azuero Complex, PAN
AP (2ai, bas), 89-85 Ma (n = 33)
AP (2ai, enr-bas) (n = 2)
Rock/N-MORB

10

e
100
Arc units: Sona-Azuero & Chagres-Bayano arcs, PAN
SAA (2aii, bas), 73-69 Ma (n = 14)
CBA (2b, bas), 70-39 Ma (n = 24)
Rock/N-MOR

10

AP (2ai)
1 89-85 Ma
(from e)

f
0.1
Cs Ba Nb La Pb Sr Nd Sm Ti Yb
Rb Th Ce Pr Zr Eu Dy Lu

Supplementary Fig. 4a-f Whattam and Stern 2014

 PISI tr 2a, EFPH, SA, western Colombia &
western Ecuador (3, 4, 5, 6)
100
OP unit: Central Cordillera, w. COL
CC (6a, bas), 97-86 Ma (n = 22)
CC (6a, pic) (n = 17)

Rock/N-MORB
10

g
100
OP units: Volcanic, Amaime & Barros Fms.,
2
Western Cordillera , w. COL
VABF (V) (WC) (6bi, bas), 101-98 Ma (n = 9)
Rock/N-MORB

WC (K) (6b, bas), 94-75 Ma (n = 26)

10

h
100
OP unit: Serrania de Baudo, w. COL
SDB (4, bas), 79-72 Ma (n = 18)
SDB (4, enr-bas) (n = 3)
Rock/N-MORB

10

i
100
OP unit: Gorgona Island, off of w. COL
GI (4, bas), 91-81 Ma (n = 2)
GI (4, dep-pic) (n = 9)
Rock/N-MORB

GI (4, dep-kom) (n =18)

10 GI (4, enr-bas) (n = 3)

j
100
a
OP units: Pallatanga Unit & Pinon Fms.
PUPF (5a & 5ai, bas), 90-85 Ma (n = 38)
PUPF (5a, pic) (n = 4)
Rock/N-MORB

PUPF (5a, ank) (n = 6)

10

k
100
Arc units: Rio Cala & Macuchi arc, w. ECU
RCA (5b, bas), 85-64 Ma (n = 8)
MA ( bas), 44-33 Ma (n = 3)
Rock/N-MORB

10

PU (5a
& 5ai)
1 90-85 Ma
(from k)

l
0.1
Cs Ba Nb La Pb Sr Nd Sm Ti Yb
Rb Th Ce Pr Zr Eu Dy Lu

Supplementary Fig. 4g-l Whattam and Stern 2014

 PISI tr 2b, EFPH, NA, LA (Aruba & Curacao,
8, 9); and Central & northern segments
of CLIP (Carib. Sea, 11, 12, 13 & Haiti, 14)
100
OP unit: Aruba Lava Formation
ALF (8a, bas), 103-89 Ma (n = 7)

Rock/N-MORB
10
98-82 Ma Nicoya Complex
mean (from b)

m
100
OP unit: Curacao Lava Formation
CLF (9a, bas), ptb ~90-88 Ma (K) (n = 14)
R
CLF (L)(9a, bas),75.8±2.0 Ma (n = 1)
Rock/N-MORB

CLF (9a, pic), ptb ~ 90-88 Ma (K) (n = 12)

10

n
100
OP unit: Curacao Lava Formation
R
CLF R (9a, bas), 92.0±1.0 Ma (n = 1)
CLF R (9a, bas) 74.2±2.4 Ma(n = 1)
Rock/N-MORB

CLF (9a, bas) 64-62 Ma (n =2)

10

o
100
OP units: Caribbean Sea
DSDP Leg 15 (12, bas), 97-87 Ma (n = 10)
DSDP Leg 15 (12, enr-bas) (n = 2)
Rock/N-MORB

ODP Leg 165, Site 1001 (13, bas), 82-80 Ma (n = 9)

10

p
100
OP unit: Beata Ridge, Caribbean Sea
R
BR (11, dol), 81.10±0.51 Ma (n = 1)
R
BR (11, dol) 79-78 Ma (n = 2)
Rock/N-MORB

R
BR (11, enr-bas) 76.87±2.14 Ma (n = 1) (inset)
10 R
BR (11, dol) 55.20±1.4 Ma (n = 1)

q
100
OP units: Dumisseau Fm., Haiti &
Lower Duarte Fm., Dominican Rep.
R
DF (14, bas) 93.6±0.8 Ma (n = 1)
R
DF (14, bas) 93-84 Ma (n = 6)
Rock/N-MOR

R
10 DF (14, bas) 82.8±0.7 Ma (n = 1)
R
LDF (15, diab) 88-85 Ma (n = 2)

r
0.1
Cs Ba Nb La Pb Sr Nd Sm Ti Yb
Rb Th Ce Pr Zr Eu Dy Lu

Supplementary Fig. 4m-r Whattam and Stern 2014

 500

Rock/Primitive Mantle Rock/Primitive Mantle Rock/Primitive Mantle Rock/Primitive Mantle Rock/Primitive Mantle
OP unit: Nicoya Complex, CR
NW Costa Rica NC (1a, bas), 140-110 Ma (n = 10)

100
(1a) NC (1a, bas), 98-82 Ma (n = 41)

mean OPB (n = 353)

O PB (n = 353)

10

a
500

CAVAB (n = 190)
b
100 IBMVAB (n = 517)
SIRO LU (n = 37)
IBMFAB (n = 13)

10

Arcs & forearcs

500
OP & Arc units: Azuero Plateau, Sona-Azuero & Chagres-
Bayano arcs, PAN; Golfito Arc, CR
PAN-05-003 AP (2ai, bas) AP mean, 89-85 Ma (n = 33)
c
100 GA (B) (1d, bas) mean, 75-66 Ma, (n =4)
SAA (2aii, bas) mean, 73-69 Ma, (n = 14)
CBA (2b, bas) mean, 70-39 Ma, (n = 24)

10

Panama & Costa Rica (2a-c)

50
western Colombia (3, 4)
d

OP units: Gorgona Island & Serrania de Baudo , w. COL

GI (4, kom): Gor92-20, Gor92-32, Gor94-1, 91-81 Ma
SDB (3, dep bas) mean, 79-72 Ma (n = 18) SDB 10 SDB 13

500
OP & Arc units: Pallatanga Unit & Rio Cala Arc, w. ECU
PU (5a, bas) mean, 90-85 Ma (n = 19) PO 1476 (PU, 5a , bas)
RCA (5b, bas) mean, ~85-64 Ma (n = 8)
100

10

western Ecuador (5c, d) e

1
Cs Ba U K Ce Pr P Zr Eu Dy Yb
Rb Th Nb La Pb Sr Nd Sm Ti Y Lu

Supplementary Fig.5 Whattam and Stern 2014

 arc-related felsic suites
ECU,COL, LA (5a, 6a,b, 8, 9) 95-79 Ma
200 200
100
100 Felsic arc units: Antioquia
& Buga batholiths, w. COL
Rock/Chondrite

(intrusive into Central & Western

Rock/N-MORB
cordilleras, respectively)
AQB (6ai), 95-86 Ma
BB (6biii). 92-89 Ma
10

10
Adakites formed via 1
partial melting of:
delam. lower crust
subduct. ocean. crust
thick lower crust .1
1
a b
200 200
100
100 Felsic arc unit: Pujili Granite, w. ECU
PG (5aiii), 87-81 Ma
Rock/Chondrite

Rock/N-MORB
10

10
1

.1
1
c d
200 200
100
100 Felsic arc unit: Aruba Batholith
AB (8b), 90-79 Ma
Rock/Chondrite

Rock/N-MORB

10

10
1

.1
1
e f
200 200
100
100 Felsic arc unit: Curacao Diorites
CD (9b), 87-85 Ma
Rock/Chondrite

Rock/N-MORB

10

10
1

.1
1
g h
200 200
100
100 Sona-Azuero & Chagres-Bayano arcs, PAN
SAA (2aii), 73-69 Ma
Rock/Chondrite

Rock/N-MORB

CBA (2b), 70-39 Ma felsic arc units from

b, f & h
10

10
1
felsic arc
units from
a, e & g
.1
1
I j
La Ce Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Rb Th K Ce Sr Nd Sm Ti Y Lu
Ba Nb La Eu Dy Yb

Supplementary Fig. 6 Whattam and Stern 2014

 Bolivar UM Complex basalts, pegmatites & felsite dykes COL (6b) 99-87 Ma
100 100
BUC basalt (52 wt.% SiO2) a
development of incipient HFSE depletions (Nb, Zr)
Rock/Chondrite

Rock/N-MORB
10

10
1

OP units: Volcanic, Amaime & Barros Fms.,

2
Western Cordillera , w. COL
VABF (WC) (6bi, bas), 101-98 Ma (n = 9) .1
WC (6b, bas), 94-75 Ma (n = 26) b
100 100
c development of shallow subduction addition
fractionation plus depletion of mantle source with time (Ba/Nb) with slight deep subduction addition (Th/Nb)
Rock/Chondrite

Rock/N-MORB
10
increase in HFSE depletions (Nb, Zr, Ti) with time

10
1
101-75 Ma Western Cordillera 101-75 Ma
basalts & BUC basalt (from a) Western Cordillera
basalts & BUC
mafic pegmatites .1 basalt (from b)
(50-52 wt.% SiO2) d
100 100
e similar to mafic pegmatites with negative HFSE (Nb, Zr, Ti)
anomalies plus progressive source depletion
Rock/Chondrite

Rock/N-MORB

10
felsic pegmatites
(65-75 wt.% SiO2)
10

.1
progressive source depletion f
100 100
Rock/Primitive Mantle

50

93-79 Ma felsic-arc-related granitoids (w. Colombia,

Aruba & Curacao) & 75-39 Ma Panamanian arc
granitoids (n = 20) from Fig. 15i
g 10
Rock/Chondrite

Rock/N-MORB

10 1
Rb Th Nb La Pr P Zr Eu Dy Yb
Ba Ce Sr Nd Sm Ti Lu

a
10 felsite dykes
65-75 wt.% SiO2) 1
93-79 Ma felsic-arc-related granitoids
(w. Colombia, Aruba & Curacao) & 75-39 Ma
Panamanian arc granitoids (n = 20) from Fig. 15i
.1
similar to pegmatites with negative HFSE (Nb, Ti) anomalies h
1 .02
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Rb Th Nb Ce Pr P Nd Sm Ti Y Lu
Ba La Sr Zr Eu Dy Yb

Supplementary Fig. 7 Whattam and Stern 2014

 200 0.70

100 CVC, northernmost a CVC peridotite c

COL (7) ~83-65 Ma spinels
Rock/Chondrite

fractionation plus slab-derived metasomatic

enrichment of mantle source with time Mariana forearc
0.60
peridotites
10

0.50

lower basaltic-andesite gabbros

1 upper, latest-stage bas-andesite dykes
spinel Cr #

0.5
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
0.40
500 increase in degree of partial
increases in subduction additions
with time, e.g., Ba/Zr, Ba/Ti
b melting of increasingly slab-
100 metasomatized, refractory
mantle source
Rock/N-MORB

0.30
mean Ba/Zr mean Ba/Ti
10 = 10.82 = 0.082

Abssyal
1 peridotites
0.20

.1
mean Ba/Zr mean Ba/Ti
= 1.91 = 0.005

.01 0.10
Rb Th Nb La Pb Sr Nd Sm Ti Y Lu 1.00 0.80 0.60 0.40
Cs Ba U K Ce Pr P Zr Eu Dy Yb spinel Mg#

Supplementary Fig. 8 Whattam and Stern 2014