Constraints on Adakite Existence

1
Constraints on Adakite Existence
Colin Macpherson University of Durham
2
Hi! You have covered a fairly diverse range of
topics so far in this class that employ a wide
range of techniques to understand magmatism and
volcanism at convergent plate margins. I hope
that you will be up for some more geochemistry. I
realise that not everyone is conversant in trace
element ratios and isotopes but all arguments
surrounding adakites rely on their geochemistry.
I have tried to incorporate sufficient
explanatory text or figures to help you
understand these. The two papers you really need
to read are 1. Defant Drummond (1990) from
Nature, and 2. Macpherson et al (2006) from
EPSL. Several others are mentioned in this file
for the really keen!
3
Normal Arc Magmatism

• subducted lithosphere releases hydrous fluids
  and, possibly, silicate melts,
• these infiltrate the overlying mantle wedge
  lowering the solidus of the mantle peridotite
  there,
• partial melting of peridotite produces basaltic
  magma.

4

• many processes operate within arc lithosphere to
  produce diverse compositions from the primary
  basaltic flux.
• in general, though, most arc lavas lie along the
  basalt-andesite-dacite-rhyolite differentiation
  trend as a result of differentiation at crustal
  pressures.

5
So, What is an Adakite?
Defant and Drummond (1990) suggest that
distinctive geochemical trends in some dacites
and andesites cannot be produced by low pressure
differentiation. This group of rocks, which they
termed adakites, have

• relatively high alumina content,
• intermediate silica content
• low concentrations of heavy rare earth elements,
  and
• elevated Sr/Y ratios

6
Why Does This Geochemical Signature Matter?

• relatively high alumina content,
• intermediate silica content
• low concentrations of heavy rare earth elements,
  and
• elevated Sr/Y ratios

are exactly the characteristics that would be
expected of a magma extracted from hydrous
basaltic crust by partial melting.
The animation over the next few pages explains
why Sr and the heavy rare earth elements (to
which Y is closely related) would be affected in
this way.
7
In subducted crust (mostly greenschist facies
rock) amphibole hosts Y ( HREE) while
plagioclase hosts Sr.
8
Phase changes occur as the slab subducts. The
transformation of greenschist to blueschist has
relatively little effect on Y Sr.
9
As P increases plagioclase destabilises, so Sr is
homeless. Garnet, in which Y is very
compatible, appears.
10
If the (now eclogite facies) slab melts then Y
(HREE) will be retained in amphibole and garnet
but Sr will go into melt.
11
Thereforeresidue Y-rich Sr-poormelt
Y-poor Sr-rich i.e. melt has high Sr/Y and low Y
12
What Support for Slab Melting?
Not Adakites
Adakites
Of slab
Defant and Drummond noted that all the rocks they
called adakites were associated with subducted
slab that were young. Therefore, they claimed
that these slab were more likely to melt because
they retained a lot of heat from their (recent)
formation. This diagram shows their evidence base
of a dozen adakitic suites.
13
So What?
Why all the fuss?Why does it matter if slab
melts reach the surface (or arc crust)?
Adakites show many similarities to tonalite,
trondhjemite, and granodiorite (TTG) suites
which are a defining feature of Archean terranes.
The two figures A compare data for ordinary arc
lavas (CA) with adakites. The figures B show TTG
suites.
Archean
Phanerozoic
So, adakites are a potential analogue for Archean
TTG that would help understand Archean tectonics
and crust generation (next slide).
CA
Adakite
14
Martin (1999)
Basalt slab melts
Basalt slab dehydrates
Hydrated peridotite melts
Hot Slab follows geotherm 1 Archean
Cool Slab follows geotherm 3 Phanerozoic
15
OK Lets Recap 1
Adakites defined in 1990 based on their
geochemical similarity to expected slab melts and
association with subduction of slabs
crustal components The model is nice it is
simple, quite intuitive and makes profound
predictions about how the early Earth operated.
16
Some things to discuss - 1
Are you happy that Defant Drummond (1990)
exhausted all alternative explanations for the
geochemical signature of adakites? Are Defant
Drummonds adakites open or closed systems (and
does this matter)? How would you recognise an
adakite in the field?
17
Post-1990
Through the 1990s more rock suites with adakitic
geochemistry were found in modern and ancient
convergent margins.
Not all of these suites were associated with
young slabs.
Therefore, two possible alternatives were
recognised 1. The slab melting model is
wrong. 2. Slabs can reach fusion point through
other ways than just being young. Many of the
workers studying the rocks chose to follow option
2 leading to many slab-melting mechanisms being
inferred. Some of these are illustrated on the
next few slides.
18
Flat Slab
Normal PTt path ( path 3 in slide 13) for slab
Long time at low P therefore different PTt path
( path 1 in slide 14) and more chance to heat
slab above solidus
Adakitic rocks have been found above some areas
where the slab has a shallow dip. e.g. Gutscher
et al (2000, Geology 28, 535-538)
19
Subduction Initiation
Mantle wedge not yet cooled by ongoing subduction
High temperature adjacent to slab, therefore slab
melting
Adakitic rocks were found in the Philippines
where the slab was old but had not been
subducting long. Numerical modelling (Peacock et
al., 1994, EPSL 121, 227-244) suggests that under
these conditions the mantle will be hotter than
mantle adjacent to more mature slabs so may
provide sufficient heat to make the slab
melt. e.g. Sajona et al (1993, Geology 21,
1007-1010)
20
Slab Tears
Some adakitic rocks occur close to tears or
gaps in the slab. This is assumed to expose the
slab interior to relatively high temperatures so
that it can melt. e.g. Yogodzinski et al (2001,
Nature 409, 500-504).
Mantle wedge cooled by ongoing subduction
producing normal arc lavas
High temperature adjacent to slab, therefore slab
melting
21
OK Lets Recap 2
Adakites defined in 1990 based on their
geochemical similarity to expected slab melts and
association with subduction of slabs
2001 use only geochemistry to define slab
melting. By early 2000s three classes of
exception to the rule (that I have listed and
others that I havent) have been introduced. The
model is not so nice now it has lost its
simplicity and can no longer make specific
predictions about how the early Earth operated.
22
A Case Study
Many adakite locations based on relatively few
rocks. Subduction initiation study in Surigao,
Philippines (slide 19) based on three rocks.
Examine Surigao example in more detail with
comprehensive dataset collected in 1999
(Macpherson et al., 2006, EPSL 243, 581-593)
Thorough sampling of Surigao peninsula and wide
array of geochemical and petrological techniques
used.
23
Westwards subduction of Philippine Sea Plate
began in north about 10Ma and has propagated south
The Philippine Tectonic Context
This trench propagating south
Mindanao
24
Philippine Sea Plate is 55Ma where it is
subducting so is too old to melt under normal
geotherm
The Philippine Tectonic Context
Mindanao
25
Adakitic rocks collected from Surigao peninsula
on Mindanao island.
The Philippine Tectonic Context
Mindanao
26
Surigao Peninsula Geology
Trench isover there
M
Large strike-slip fault acts as western
graben-bounding fault
Reactivated back-thrusts act as eastern
graben-bounding fault
Down-thrown area hosting volcanic peak (star M)
and shallow lake (surrounded by very recent
flat-bedded sediments). Suggests crustal
thinning.
27
Northeasterly view from top of volcano M
(Maniayao) with eastern graben bounding fault
indicated
28
Surigao is nice, friendly place for the most
part!
29
In east, normal arc andesites, dacites and
rhyolites (ADRs). In west, lots of
adakites. Perfect opportunity to test 1
Relationship between adakites and ADRs 2
Relationship within the suite of adakites 3
Origin of an adakite suite
30
1 Relationship between adakites and ADRs
All trace element ratios very similar except that
Y is depleted in adakites. Also resemble typical
arc magmas. Suggests similar sources and
processes of formation in source for adakites and
ADRs. ADRs formed by normal arc processes
(slides 3 and 14). Y depletion of adakites is
main difference to ADRs.
31
2 Relationship within the suite of adakites
Significant contrast in behavior of Y between
adakites and ADRs. In adakites Y shows strong
negative correlation with SiO2. Consistent with
Y depletion during differentiation, possibly by
fractional crystallisation of amphibole or
garnet. How can we determine which is involved?
Adakites (west) open circlesADRs (east) black
circles
32
Fractionation of middle (e.g. Dy) from heavy
(e.g. Yb) REE is different for removal of
amphibole as opposed to garnet from a melt.
Davidson et al. (2007)
33
Fractionation of middle (e.g. Dy) from heavy
(e.g. Yb) REE is different for removal of
amphibole as opposed to garnet from a melt.
Davidson et al. (2007)
34
Fractionation of middle (e.g. Dy) from heavy
(e.g. Yb) REE is different for removal of
amphibole as opposed to garnet from a melt.
Davidson et al. (2007)
35
2 Relationship within the suite of adakites
Adakite suite shows positive correlation of Dy/Yb
with SiO2. This is consistent with fractional
crystallisation of garnet from mafic melt. ADRs
are more consistent with fractional
crystallisation of low-pressure crystal
assemblage (amphibole plagioclase).
36
3 Origin of an adakite suite
If Surigao adakites are slab melts then should
have isotope composition of Philippine Sea
Plate ADRs and adakites indistinguishable in
isotope ratios. Both different to PSP. Suggests
similar sources for adakites and ADRs in
metasomatised mantle wedge peridotite.
37
OK Lets Recap 3
The geochemistry of Surigao adakites is very
similar to many other adakitic suites. Adakitic
rocks are largely indistinguishable from
near-contemporaneous ADRs. Trace element ratios
and isotope ratios point to similar sources in
mantle wedge metasomatised by the same type of
slab fluids found in most subduction zones. The
simplest explanation for Surigao adakites is that
they are the produced by differentiation of
hydrous basaltic magma at sufficient pressure to
include garnet in the fractionating
assemblage. In Surigao the crust is estimated to
be 25km thick while garnet require pressures
equivalent to 33km or more. Differentiation must
be sub-Moho.
38
Implications
On slide 17, I said the presence of adakites in
subduction zones with old slabs could be
interpreted in one of two ways 1. The slab
melting model is wrong. 2. Slabs can reach fusion
point through other ways than just being young.
In 1993 the Surigao was interpreted as an
expectional case of slab melting. The more
thorough study outlined in slides 22 to 37
suggests that Surigao adakites can be produced
without slab melting. Is Surigao an oddity, or
are there other examples where high pressure
differentiation of basaltic arc magma can be
invoked?
39
Ecuador
Suites of different ages that appear to be
derived from similar parents but differentiate at
different depth. Chiarardia et al (2004,
Mineralium Deposita 39, 204-222)
40
Chile
Longaivi volcano. Early granet-dominated phase
(mafic enclaves) followed by lower pressure
differentiation. Rodriguiz et al (2007, Journal
of Petrology 48, 2033-2061)
41
Other non-Slab-Melt Models
Differentiation Castillo et al. (1999, CMP)
Camiguin Island, Philippines. Garrison Davidson
(2003, Geology) Nothern Volcanic Zone,
Andes Prouteau Scaillet (2003, J. Pet) Pinatubo
1991 dacite Very low-degree partial melting of
hydrous peridotite Eiler et al. (2007, G-cubed)
42
Some things to discuss - 2
Does the slab melt? What conditions would be
required for those melts to reach the
surface? How can the deep differentiation model
be reconciled with other classic adakite
occurrences? What are the implications of the
deep differentiation model for the adakite - TTG
analogue?