mafic magmas from mount baker in the northern cascade arc, washington
TRANSCRIPT
ORIGINAL PAPER
Mafic magmas from Mount Baker in the northern Cascade arc,Washington: probes into mantle and crustal processes
Nicole E. Moore • Susan M. DeBari
Received: 23 February 2011 / Accepted: 26 August 2011
� Springer-Verlag 2011
Abstract Five mafic lava flows located on the southern
flank of Mount Baker are among the most primitive in the
volcanic field. A comprehensive dataset of whole rock and
mineral chemistry reveals the diversity of these mafic lavas
that come from distinct sources and have been variably
affected by ascent through the crust. Disequilibrium tex-
tures present in all of the lavas indicate that crustal pro-
cesses have affected the magmas. Despite this evidence,
mantle source characteristics have been retained and three
primitive endmember lava types are represented. These
include (1) modified low-K tholeiitic basalt (LKOT-like),
(2) typical calc-alkaline (CA) lavas, and (3) high-Mg
basaltic andesite and andesite (HMBA and HMA). The
Type 1 endmember, the basalt of Park Butte (49.3–50.3
wt% SiO2, Mg# 64–65), has major element chemistry
similar to LKOT found elsewhere in the Cascades. Park
Butte also has the lowest overall abundances of trace ele-
ments (with the exception of the HREE), indicating it is
either derived from the most depleted mantle source or has
undergone the largest degree of partial melting. The Type 2
endmember is represented by the basalts of Lake Shannon
(50.7–52.6 wt% SiO2, Mg# 58–62) and Sulphur Creek
(51.2–54.6 wt% SiO2, Mg# 56–57). These two lavas are
comparable to calc-alkaline rocks found in arcs worldwide
and have similar trace element patterns; however, they
differ from each other in abundances of REE, indicating
variation in degree of partial melting or fractionation. The
Type 3 endmember is represented by the HMBA of Tarn
Plateau (51.8–54.0 wt% SiO2, Mg# 68–70) and the HMA
of Glacier Creek (58.3–58.7 wt% SiO2, Mg# 63–64). The
strongly depleted HREE nature of these Type 3 units and
their decreasing Mg# with increasing SiO2 suggests frac-
tionation from a high-Mg basaltic parent derived from a
source with residual garnet. Another basaltic andesite unit,
Cathedral Crag (52.2–52.6 wt% SiO2, Mg# 55–58), is an
Mg-poor differentiate of the Type 3 endmember. The calc-
alkaline lavas are least enriched in a subduction component
(lowest H2O, Sr/PN, and Ba/Nb), the LKOT-like lavas are
intermediate (moderate Sr/PN and Ba/Nb), and the HMBA
are most enriched (highest H2O, Sr/PN and Ba/Nb). The
generation of the LKOT-like and calc-alkaline lavas can be
successfully modeled by partial melting of a spinel lherz-
olite with variability in composition of slab flux and/or
mantle source depletion. The HMBA lavas can be suc-
cessfully modeled by partial melting of a garnet lherzolite
with slab flux compositionally similar to the other lava
types, or less likely by partial melting of a spinel lherzolite
with a distinctly different, HREE-depleted slab flux.
Keywords Cascade arc � Mount Baker � Basalt �Geochemistry � Petrology � Mantle � Slab
Introduction
Primitive basalts in volcanic arc settings are valuable
recorders of magmatic flux from the mantle to the crust. By
examining the mass balance of input and output from
subduction zones, we can determine both how magma is
generated in these settings, and how this balance affects the
Communicated by T. L. Grove.
Electronic supplementary material The online version of thisarticle (doi:10.1007/s00410-011-0686-4) contains supplementarymaterial, which is available to authorized users.
N. E. Moore (&) � S. M. DeBari
Geology Department, Western Washington University,
MS 9080, Bellingham, WA 98225, USA
e-mail: [email protected]
123
Contrib Mineral Petrol
DOI 10.1007/s00410-011-0686-4
growth and evolution of continental crust (Hirschmann
et al. 2000). Primitive magmas from around the world have
been used to probe the mantle input into magmatic arcs
(e.g., Hildreth and Moorbath 1988; Sisson and Bronto
1998; Class et al. 2000; Grove et al. 2002; Walker et al.
2003).
The processes of magma generation in arcs may vary
depending on the age of the subducting slab (cf. Hirsch-
mann et al. 2000). The Cascade magmatic arc (Fig. 1) is
one of the hottest endmembers of arcs worldwide, with a
subducting plate that is only a few million years old at the
trench. It is a continental arc built on juvenile accreted
terranes in a manner similar to the Southern Volcanic Zone
(SVZ) of the Andes (Hildreth 2007; Hildreth and Moorbath
1988). As such, the lack of old continental basement allows
mafic magmas to be effective probes into magma genera-
tion processes.
Studies of primitive magmas in the Cascade arc have
been concentrated in the middle and southern Cascades
where mafic magmas are abundant (e.g., Bacon et al. 1997;
Borg et al. 1997; 2000; Conrey et al. 1997; Reiners et al.
2000; Elkins-Tanton et al. 2001; Grove et al. 2002; Strong
and Wolff 2003; Leeman et al. 1990; 2005; Smith and
Leeman 2005; Schmidt et al. 2008). The earliest studies by
Leeman et al. (1990) and Bacon et al. (1997) distinguished
three endmember types of primitive magma, interpreted to
represent three different mantle sources for the Cascade
arc. These three magma types consist of hydrous calc-
alkaline basalts (CAB), alkalic ocean-island-type basalts
(OIB), and dry, decompression-related low-potassium
olivine tholeiites (LKOT, aka HAOT of Bacon et al. 1997
and Grove et al. 2002), which share some characteristics
with mid-ocean ridge basalts.
Equivalent studies of primitive magmas are less com-
mon in the northern part of the Cascade arc. In this portion
of the arc, known as the Garibaldi Belt (Fig. 1), the volume
of mafic lavas is dramatically less than in the south. Green
and coworkers analyzed a restricted sample set of primitive
lavas from a transect of the Garibaldi Belt (Green and
Harry 1999; Green and Sinha 2005; Green 2006) and
determined that the slab age, depth of fluid loss, and degree
of partial melting in the mantle wedge all decrease north-
ward into this segment. In their studies, compositions
ranged from calc-alkaline basalts in the southern part of the
Belt (Glacier Peak and Mount Baker) to alkalic basalts in
the north. Green and Sinha (2005) also determined that the
component responsible for the slab signature in the calc-
alkaline basalts is from fluid input, not from partial melting
of sediment. However, a more detailed study of mafic lavas
from Glacier Peak showed that there are multiple mantle
sources for the basalts, including LKOT and calc-alkaline
endmembers (Taylor 2001), with strong isotopic hetero-
geneity (e.g., DeBari et al. 2005 and unpublished
manuscript).
In this study, our aim was to see whether some of the
same variability existed in primitive lavas from Mount
Baker, and whether these lavas could provide insight into
processes of magma generation beneath this part of the arc.
This study presents detailed petrologic and geochemical
analyses of the most mafic lavas erupted from Mount Baker
throughout its history.
Tectonic and geologic setting
The Cascade volcanic arc extends from northern California
to southern British Columbia along the west coast of North
America. The volcanic centers begin in the south at Lassen
Peak and extend northward through Mount Meager in
Canada (Fig. 1). Arc volcanism is generated by the slightly
oblique subduction of the young and hot Juan de Fuca Plate
under the North American Plate at about 45 mm/year
(Riddihough 1984). The dip of the subducting slab beneath
Fig. 1 The Cascade volcanic arc and Mount Baker (MB) study area.
Orange shaded regions and circles represent areas of mafic volcanism
after Borg et al. (1997). Black lines indicate segments of the Cascade
arc as described by Guffanti and Weaver (1988). Triangles represent
major volcanic centers, abbreviated as follows: MM Mount Meager,
MC Mount Cayley, MG Mount Garibaldi, MB Mount Baker, GPGlacier Peak, MR Mount Rainier, MSH Mount St. Helens, MA Mount
Adams, SVF Simcoe Volcanic Field, MH Mount Hood, MJ Mount
Jefferson, TS Three Sisters, NV Newberry Volcano, CLV Crater Lake
Volcano, MMc Mount McLoughlin, MLV Medicine Lake Volcano,
MS Mount Shasta, LVC Lassen Volcanic Center. Rates of subduction
are from McCrory et al. (2004)
Contrib Mineral Petrol
123
Washington and northern Oregon is estimated to be *20�(McCrory et al. 2004). Along-strike variations in magma-
tism and structure are the basis for a subdivision of the arc
into five major segments (Guffanti and Weaver 1988;
Schmidt et al. 2008). The northernmost segment is the
Garibaldi Belt, which is separated from the rest of the
Cascades by a 90-km-wide volcanic gap. This segment is
distinct because a structural change in the Juan de Fuca
plate off the coast of central Washington causes the overall
trend of the arc to change from N–S to NW–SE. A marked
decrease in mafic volcanism occurs northward along the
arc, resulting in few isolated mafic centers in the Garibaldi
Belt (Fig. 1).
Mount Baker (3,286 m) is the northernmost volcano of
the Cascade arc in the United States. It is one of only two
U.S. volcanoes located within the Garibaldi Belt (Fig. 1),
situated between Glacier Peak to the south and Mount
Garibaldi to the north. This primarily andesitic stratovol-
cano is part of a volcanic field that has been active since
about 1.3 Ma, with the most recent magmatic eruption
occurring at 6.5 ka (Hildreth et al. 2003). Mount Baker
is *350 km east of the trench above the subducting Juan
de Fuca plate. The top of the plate lies at depths
of *80–90 km beneath Mount Baker (McCrory et al.
2004) and is dipping at *15� (Hyndman et al. 1990). The
subducting plate age is *18 million years at this location
(Green and Harry 1999). Beneath the volcanic center, the
crust is approximately 40–45 km thick (Mooney and
Weaver 1989), and the basement is composed of accreted
Paleozoic and Mesozoic oceanic rocks. The eruptions in
the volcanic field have largely taken place through the
metamorphosed Nooksack Group and Chilliwack Group,
which are primarily composed of alternating volcanic and
sedimentary rocks that were generated in an oceanic setting
(Tabor 1994; Tabor et al. 2003).
Although Mount Baker is an andesitic volcano, about
1% of the total eruptive products in the volcanic field are
basaltic (Hildreth et al. 2003). Sources of mafic volcanism
at Mount Baker are primarily monogenetic outlier vents
and cinder cones and are generally restricted to the
southern flank of the volcano. This study focuses on five
units identified as the most mafic at Mount Baker, based on
major element data and mapping by Hildreth et al. (2003).
The MgO content reported by Hildreth et al. (2003) for
these five units is 4.0–8.4 wt%, indicative of potentially
primitive magmas (Mg# [ 60, where Mg# = 100*[Mg/
(Mg ? FeT)]). The units include the (1) basalt of Park
Butte, (2) basaltic andesite of Cathedral Crag, (3) basaltic
andesite of Tarn Plateau, (4) basalt of Lake Shannon, and
(5) basalt of Sulphur Creek (Fig. 2; italicized unit names
are from Hildreth et al. 2003). These flows range in age
from Middle Pleistocene to Holocene and K–Ar ages listed
below are from Hildreth et al. (2003).
The basalt of Park Butte (716 ± 45 ka) is a flow rem-
nant from an unknown vent, located 7.5 km southwest of
Grant Peak, the true summit of Mount Baker (Fig. 2). It is
located *400 m east of the Park Butte Lookout. The
remnant is *30 m thick, defined by a thin, discontinuous
red basal layer of flow breccia. The majority of the remnant
is talus, but some basalt columns are found in place
overlain by till. The columns are weathered to a charac-
teristic buff color, but internally reveal a coarse, inter-
granular texture. The lavas contain abundant olivine and
plagioclase (Table 1). In some boulders near the base of the
talus, mingling textures are observed between this and
another mafic, oxidized, and vesicular lava.
Less than 0.5 km to the north and slightly east of the
Park Butte unit lies the glacially eroded basaltic andesite of
Cathedral Crag (Fig. 2; 331 ± 18 ka). The unit crops out
in a large knob *300 m in length and contains both
prismatic vertical and irregular horizontal columns. No
vent has been located and it is unclear whether this unit
contains multiple flows (Hildreth et al. 2003). Plagioclase,
clinopyroxene, and olivine are abundant (Table 1), and
many samples also contain large (4–22 mm) felsic xeno-
liths with surrounding reaction rims.
The basaltic andesite of Tarn Plateau (203 ± 25 ka) is
a mesa-shaped unit that nearly banks against the eastern
portion of the Park Butte lavas (Fig. 2). It rises approxi-
mately 150 m above the floor of the valley that separates it
from Cathedral Crag. The unit crops out in a nearly square
fashion, *0.5 km on a side. Irregular columnar joints are
visible in cross section from across the valley, but the flat
top surface has been glacially scoured and fractured. As
with the Park Butte and Cathedral Crag units, the location
of the source vent is unknown. Samples are rich in clino-
pyroxene, plagioclase, and olivine (Table 1).
The southernmost and farthest outlying mafic unit in this
study is the basalt of Lake Shannon (94 ± 21 ka), located
1 km west of Lake Shannon, and more than 15 km
southeast of the summit of Mount Baker (Fig. 2). The most
prominent feature of this flow unit is a conical knob that
rises about 400 m above Baker Lake Road. The eruption
may have been englacial (not entirely subglacial, Hildreth
et al. 2003), as the majority of the outcrop is composed of
hyaloclastite tuff (*100 m thick). Where they occur, lavas
are vesicular and contain abundant plagioclase with lesser
olivine (Table 1).
The basalt of Sulphur Creek is a compositionally zoned
flow unit (basalt and basaltic andesite) that erupted from a
cinder cone in Schreibers Meadow *8 km south of the main
edifice (Fig. 2). The age of the unit is constrained by 14C ages
from carbon buried by scoria near the vent (8,750 ± 50 to
8,850 ± 50 years BP, Tucker et al. 2007; radiocarbon cali-
brated to 9.8 ka by Scott et al. 2001). Exposures are typically
massive or blocky, but sometimes irregularly jointed.
Contrib Mineral Petrol
123
Although lavas proximal and medial to the cone are basaltic
andesite (up to 55 wt% SiO2 and *4.5 wt. % MgO), the most
distal portions (those that form a lava fan near Baker Lake)
are generally basaltic (51 wt% SiO2 and 5.5 wt% MgO;
Green 1988; Hildreth et al. 2003; Tucker and Scott 2009;
Baggerman and DeBari 2011). The basalt is rich in plagio-
clase and olivine, and the basaltic andesite also contains
abundant clinopyroxene (Table 1).
Analytical methods
A total of 38 samples were collected from the 5 flow units
described above (see Electronic Supplement 1 for sample
locations). Sample collection for the Sulphur Creek basalt
was limited to the mafic toe of the flow.
Whole rock analyses
Abundances of major element oxides and 6 trace elements
(Ni, Cr, V, Ga, Cu, and Zn) were determined by X-ray
fluorescence (XRF) analysis at the Washington State Uni-
versity (WSU) GeoAnalytical Laboratory. Rock powders
were prepared at Western Washington University (WWU)
by grinding select rock chips in a tungsten carbide SPEX
Shatterbox grinding mill, following the method of Johnson
et al. (1999). Powders were fused into beads using a 2:1
mix of dry dilithium tetraborate flux to dry rock powder, in
a 1,000�C furnace for 10 min. Beads were then re-ground
and re-fused by the above method to ensure homogeneity
of each sample. The samples were analyzed at WSU using
a ThermoARL Advant’XP ? sequential XRF spectrome-
ter. A thorough description of sample preparation as well as
Fig. 2 Geologic map of Mount Baker from Hildreth et al. (2003).
The target mafic lavas for this study are labeled with their respective
ages. Each box is color coded to match the symbol color used for each
unit throughout this study. Early Pleistocene units are in brown,
except for the rhyodacites of the Kulshan caldera in red. Middle
Pleistocene units are in green, Late Pleistocene units in purple.
Holocene eruptive unit is in orange
Contrib Mineral Petrol
123
Ta
ble
1S
um
mar
yo
fp
etro
gra
ph
yan
dm
iner
alch
emis
try
Flo
wM
od
alA
bu
nd
ance
Min
eral
Mo
des
Siz
eR
ang
eD
iseq
uil
ibri
um
Tex
ture
sO
lM
g#
Cp
xM
g#
An
Pla
g
Par
kB
utt
eb
asal
tP
hen
ocr
yst
s:3
5–
40
%O
l:5
0–
65
%0
.3–
3.8
mm
Em
bri
ms,
alt
infr
actu
res
66
–8
27
0–
81
37
–7
8
Cp
x:*
1%
0.2
–1
.3m
mE
mb
rim
s
Pla
g:
35
–4
5%
0.2
–2
.3m
mS
iev
edco
res,
no
rman
do
scil
lato
ryzo
nin
g
Gro
un
dm
ass:
60
–6
5%
Ab
ov
e?
ox
ides
,o
px
and
trap
atit
e
20
–6
00
lmN
on
eo
bse
rved
57
–5
86
3–
75
59
Cat
hed
ral
Cra
g
bas
alti
can
des
ite
Ph
eno
cry
sts:
25
–3
5%
Ol:
5–
10
%0
.2–
2.5
mm
Sev
ere
reso
rpan
dem
bri
ms
70
–7
47
4–
88
39
–8
6
Cp
x:
13
–2
0%
0.3
–2
.4m
mE
mb
rim
s,re
sorb
edco
res,
rev
zon
ing
Pla
g:
70
–8
0%
0.2
–3
.1m
mE
mb
rim
s,si
eved
core
s,n
orm
zon
ing
Gro
un
dm
ass:
65
–7
5%
Ab
ov
e?
ox
ides
,o
px
,
trb
add
eley
ite
15
–6
00
lmF
elsi
cx
eno
lith
sin
han
dsa
mp
leN
A7
66
2–
64
Tar
nP
late
aub
asal
tic
and
esit
e
Ph
eno
cry
sts:
25
–4
0%
Ol:
10
–2
0%
0.2
–3
.2m
mE
mb
rim
s(s
om
eto
core
),al
tin
frac
ture
s7
7–
85
78
–8
76
7–
88
Cp
x:
45
–6
5%
0.2
–3
.0m
mE
mb
rim
s,si
eved
core
s,n
orm
zon
ing
Pla
g:
25
–4
0%
0.2
–4
.8m
mR
eso
rp,
siev
edco
res,
no
rmzo
nin
g
Gro
un
dm
ass:
60
–7
5%
Ab
ov
e?
op
x,
ox
ides
10
–4
00
lmN
on
eo
bse
rved
NA
70
47
–6
2
Lak
eS
han
no
nb
asal
tP
hen
ocr
yst
s:3
0–
35
%O
l:3
0–
35
%0
.1–
2.2
mm
Res
orb
edco
res,
frac
ture
dan
dem
bri
ms
78
–8
1N
A,
rare
55
–6
5
Cp
x:\
2%
0.1
–0
.2m
mN
on
eo
bse
rved
Pla
g:
60
–7
0%
0.1
–3
.2m
mE
mb
siev
edco
res,
no
rman
dre
vzo
nin
g
Gro
un
dm
ass:
65
–7
0%
Ab
ov
e?
ox
ides
,
trb
add
eley
ite
10
–5
00
lmN
on
eo
bse
rved
70
43
–6
95
6–
57
Su
lph
ur
Cre
ekb
asal
tP
hen
ocr
yst
s:1
5–
25
%O
l:2
0–
35
%0
.1–
1.1
mm
Res
orb
edco
res,
sev
erel
yem
bri
ms
70
–8
5N
A,
rare
48
–6
5
Cp
x:\
1%
0.2
–1
.1m
mN
ear-
tota
lab
sorp
tio
nw
her
eo
ccu
r
Pla
g:
65
–8
0%
0.2
–4
.1m
mS
iev
edco
res,
emb
rim
s,re
vzo
nin
g
Gro
un
dm
ass:
75
–8
5%
Ab
ov
e?
op
x,
ox
ides
10
–4
00
lmN
on
eo
bse
rved
64
70
–7
54
4–
59
Ol
Mg
#(M
g/(
Mg
?F
eT)
iseq
uiv
alen
tto
Fo
(fo
rste
rite
)co
nte
nt
fro
mT
able
2
Ol
oli
vin
e,C
px
clin
op
yro
xen
e,O
px
ort
ho
py
rox
ene,
Pla
gp
lag
iocl
ase,
Tr
trac
e,E
mb
emb
ayed
,N
orm
no
rmal
,R
evre
ver
se,
Alt
alte
red
,R
eso
rpre
sorp
tio
n,
NA
no
ne
anal
yze
d
Contrib Mineral Petrol
123
analytical procedure, accuracy, and precision are detailed
by Johnson et al. (1999).
For XRF analysis, the GeoAnalytical Lab at WSU esti-
mates accuracy by comparing the known values of nine
USGS standard samples with compositions determined by
preparation and analysis at WSU for each of these USGS
standard samples. For all major elements except Na, the
difference between the known values and WSU results are
less than the difference expected between two random
samples selected from the same rock unit. Maximum mea-
sured differences (in weight %) between known values and
WSU results for major elements are as follows: \0.60% for
SiO2 and FeO; \0.40% for MgO, B 0.20% for Al2O3 and
Na2O; \0.10% for TiO2, CaO, K2O, and P2O5; and \0.01%
for MnO. For trace elements, values obtained by XRF for Ni,
Cr, and V can be considered only semiquantitative below
30 ppm. The remainder of trace elements analyzed by XRF
are satisfactorily precise and accurate down to values from 1
to 3 ppm.
All 14 naturally occurring rare earth elements (REE)
along with an additional 13 trace elements (Ba, Rb, Y, Nb,
Cs, Hf, Ta, Pb, Th, U, Sr, Sc, Zr) were analyzed in 32
samples by high-precision inductively coupled plasma
source mass spectrometry (ICP-MS) analysis at the WSU
GeoAnalytical Laboratory. Rock powders were prepared at
WWU by grinding select rock chips in an alumina ceramic
SPEX Shatterbox grinding mill. Powders were fused into
beads using a 1:1 mix of dry dilithium tetraborate flux to
dry rock powder, in a 1,000�C furnace for 10 min. Beads
were then re-ground using the above method to ensure
homogeneity of each sample. The resulting powders were
sent to WSU and analyzed using an Agilent 4500 ? ICP-
MS. Analytical procedure as well as accuracy and precision
are detailed by Knaack et al. (1994).
For ICP-MS analyses, the GeoAnalytical Lab at WSU
estimates accuracy by comparing the known values of fif-
teen USGS standard samples with results from two beads
of each of the USGS standard samples that have been
prepared and analyzed at WSU. Maximum measured dif-
ferences (in ppm) between known values and WSU
results for ICP-MS-analyzed trace elements are as fol-
lows: \40 ppm for Ba; \30 ppm for Sr and Zr; B 5 ppm
for La, Ce Pr, Nd, Sm, Dy, Hf, Th, Pb, and Sc; \3 for Rb,
Y, and Nb; \1 ppm for Gd, Er, and U; \0.5 ppm for Eu,
Tm, Yb, Ta, Ho, and Cs; and B 0.1 ppm for Tb and Lu.
Precision is generally better than 5% for the REE and 10%
for the other trace elements.
Isotopic ratios for Pb, Sr, and Nd were measured for 6
samples, 2 each from the following units: the Tarn Plateau,
Park Butte, and Lake Shannon. The most primitive and
most differentiated samples in each flow unit were chosen
to represent the probable isotopic compositional range of
these individual flows. Field samples were reduced to chips
by the method described above and then ground into
powders using an alumina ceramic SPEX Shatterbox
grinding mill. Powders were sent to the University of
Washington Isotope Geochemistry Lab and analyzed using
a Nu Instruments multiple collector inductively coupled
plasma mass spectrometer (MC-ICP-MS).
Samples measured for Pb isotopes were dissolved in
concentrated HF ? 8 N HNO3, dried, and re-dissolved. An
aliquot of 2 samples, NM-LS6 and NM-TP6, were
reserved, dried, and passed twice through ion-exchange
columns (for detailed procedure of entire Pb separation
chemistry, see Nelson 1995). Both aliquots (simply dis-
solved and purified) of samples NM-LS6 and NM-TP6
were analyzed, as well as a second aliquot of sample NM-
LS2 (run as a repeat) to test for reproducibility. Analytical
procedure for Pb analysis is fully described in Harkins et al.
(2008). Pb isotopic compositions are normalized to NIST
981, with isotopic standards of 16.9356, 15.4891, and
36.7006 for 206Pb/204, 207Pb/204Pb, and 208Pb/204Pb,
respectively. Error on Pb analyses is ± 150 ppm (2r)
or ± 0.015%.
Samples measured for Sr and Nd isotopes were dis-
solved in a 10:1 mixture of concentrated HF-HNO3. After
drying, samples were fumed with HClO4 to decompose
fluorides. Final dissolution and sample equilibration were
achieved by adding 6 M HCl. Column separation was
conducted first to partition the light rare earth elements
(LREE) from Sm and Nd and second to partition Sm from
Nd. Analytical procedure for Sr isotope analysis is fully
described in Nelson (1995) and in Gaffney et al. (2007) for
Nd. Sr isotopic compositions are normalized to87Sr/86Sr = 0.710240 for NIST 987, which was the aver-
age value yielded for the standard. Nd isotopic composi-
tions are normalized to the La Jolla value of143Nd/144Nd = 0.511843. Error on Sr and Nd analyses
is ± 30 ppm (2r) or ± 0.3 epsilon units. The 145Nd/144Nd
was monitored as an additional accuracy check. This
yielded a value of 0.348413 for all samples of the standard,
which is within ± 17 ppm (2r) of the accepted value of
0.348417.
Mineral analyses
Compositions of olivine, pyroxenes, plagioclase, Fe–Ti
oxides, and chromium spinel were obtained using a JEOL
Superprobe 733 Electron Probe Microanalyzer (EPMA)
equipped with 4 wavelength-dispersive X-ray spectrome-
ters and an energy-dispersive X-ray spectrometer at the
University of Washington. A standardized set of natural
and synthetic minerals were used for calibration. Acceler-
ating voltage was set to 15 keV, with a beam current of 10
nA for plagioclase and 15 nA for olivine, pyroxene, and
oxides. Beam diameter was \1 lm for all minerals except
Contrib Mineral Petrol
123
plagioclase, which was analyzed at 3 lm. Element peaks
were counted for a minimum of 20 s, and for a maximum
of 40 s, or when 0.4% statistical error was achieved.
Analytical error is \2% for major elements and \12% for
trace elements. Data corrections were performed using the
CITZAF X-ray Correction Program based on the method of
Armstrong (1988).
Summary of petrography and mineral chemistry
Petrographic characteristics and mineral chemistry are
described in detail for each unit in Electronic Supplement
2. These data are summarized below and in Table 1, and
compositional data for all mineral phases are listed in
Tables 2 and 3 (olivine and pyroxene) and Electronic
Supplements 3 and 4 (plagioclase and oxide minerals).
The mafic flows from Mount Baker are dominated by
either Mg-rich clinopyroxene (basaltic andesites of Tarn
Plateau and Cathedral Crag; Mg# 76–88) or olivine (basalts
of Park Butte, Lake Shannon, Sulphur Creek; *Fo66–85)
with variable amounts of plagioclase and Fe–Ti oxides. In
the basaltic andesites of Tarn Plateau and Cathedral Crag,
olivine phenocrysts are also common (Fo70–85), but in
lesser abundance than clinopyroxene and plagioclase. In
contrast, the basalts of Park Butte, Lake Shannon, and
Sulphur Creek contain only rare phenocrysts of clinopy-
roxene (\1%, Mg# \ 75). The Park Butte basalt has
olivine as the most abundant phenocryst phase, followed by
plagioclase, whereas in basalts of Lake Shannon and Sul-
phur Creek, plagioclase is most abundant, followed by
olivine.
Phenocryst and microphenocryst textures from all of the
units at Mount Baker show that disequilibrium conditions
existed prior to their eruption. Disequilibrium textures are
pervasive in phenocrysts, common in microphenocrysts, but
absent in the groundmass. The most mafic units do not
always contain the most mafic phenocrysts. For example,
Tarn Plateau basaltic andesites contain the most magnesian
olivine phenocrysts and most anorthitic plagioclase pheno-
crysts (Electronic Supplement 3), but Cathedral Crag
basaltic andesite contains the most magnesian clinopyroxene
Table 2 Representative olivine compositions
Sample SiO2 MgO FeO* CaO MnO NiO Total Fo
Park Butte
PB5_Ph_c 37.32 31.98 28.67 0.24 0.47 0.03 98.71 67
PB5_Ma_c 36.72 27.49 34.95 0.13 0.54 0.06 99.90 58
PB8_Ph_c 40.10 42.64 16.80 0.11 0.26 0.13 100.04 82
PB8_Ph_r 37.44 32.35 29.99 0.19 0.48 0.03 100.48 66
PB8_Ma_c 36.35 26.60 36.37 0.09 0.53 0.00 99.94 57
Cathedral Crag
CC1_Mph_c 38.25 35.28 25.65 0.16 0.57 0.05 99.95 71
CC1_Ph_c 38.01 34.87 26.11 0.19 0.52 0.05 99.75 70
CC7_Ph_c 38.45 37.12 23.45 0.13 0.37 0.09 99.61 74
Tarn Plateau
TP3_Ph_c 40.08 44.90 13.98 0.11 0.22 0.12 99.41 85
TP3_Ph_r 39.17 39.26 21.15 0.15 0.30 0.10 100.13 77
TP4_Ph_c 39.15 41.89 17.79 0.15 0.28 0.12 99.39 81
Lake Shannon
LS1_Mph_c 39.47 41.69 17.28 0.21 0.28 0.21 99.14 81
LS4_Ma_c 37.55 34.15 25.35 0.30 0.40 0.04 97.80 71
LS4_Mph_c 39.04 40.18 19.20 0.21 0.29 0.19 99.12 79
LS4_Ph_c 39.11 39.69 19.78 0.21 0.33 0.15 99.25 78
Sulphur Creek
SC1_Mph_c 38.82 37.93 22.12 0.19 0.41 0.07 99.55 75
SC1_Mph_r 37.84 33.93 25.99 0.33 0.55 0.02 98.66 70
SC4_Ph_c 40.72 44.15 13.54 0.16 0.20 0.20 98.97 85
SC4_Ph_r 38.57 36.29 23.63 0.21 0.53 0.10 99.34 73
SC4_Ma_c 37.40 30.96 30.56 0.31 0.66 0.02 99.91 64
Individual mineral grains analyzed in each sample number are designated as Ph for phenocryst, Mph for microphenocryst, and Ma for matrix.
Each is also assigned a letter to indicate location of the analysis, c core or r rim, Fo forsterite content
Contrib Mineral Petrol
123
phenocrysts with less forsteritic olivine (Tables 1, 2). While
mineral chemistry can vary greatly within units, the range in
Mg# of olivine and clinopyroxene, as well as anorthite
content of plagioclase, is fairly consistent throughout all of
the flow units (Table 1, Electronic Supplement 5).
Chromite occurs only as inclusions in olivine in all lava
flows, indicating a restricted period of crystallization at
high temperature. The most forsteritic olivines (Fo [ 80)
contain chromite with Cr# (100*Cr/[Cr ? Al]) that are
distinct for each flow. Tarn Plateau chromites have the
highest Cr# (Cr# *70), whereas Lake Shannon chromites
have the lowest (Cr# *30).
Abundant Fe–Ti oxides are present in the groundmass
of all units in this study. Coexisting mineral pairs of
magnetite and ilmenite in the Park Butte, Cathedral Crag,
and Tarn Plateau lavas were used to calculate pre-eruptive
oxygen fugacity (Electronic Supplement 4). Equilibrium
of coexisting pairs was determined using the Bacon and
Hirschmann (1988) test. Oxygen fugacity for all three of
these units lies near or within 1 log unit above the nickel–
nickel oxide (NNO) buffer and ranges from -11.6 to
-12.9 log units for Park Butte, -12.4 to -14.2 for
Cathedral Crag, and -9.4 to -11.9 for Tarn Plateau. Low
equilibration temperatures were obtained by the coexis-
ting pairs (Electronic Supplement 4), thus the fugacity
values could be considered suspect. However, Shaw
(2011) determined similar values (DQFM ?1.0 to ?1.3)
using sulfur speciation of melt inclusions in olivine from
Table 3 Representative pyroxene compositions
Sample SiO2 Al2O3 TiO2 FeO* MnO MgO CaO Na2O Cr2O3 Total En Fs Wo Mg#
Clinopyroxene
Park Butte
PB5_Ma_c 51.65 1.75 0.82 9.18 0.24 15.56 19.76 0.35 0.06 99.36 45 15 41 75
PB8_Mph_r 51.61 1.45 0.90 11.30 0.30 14.35 18.92 0.47 0.06 99.35 42 18 40 69
PB8_Ma_c 51.15 2.53 0.75 11.01 0.33 15.03 18.08 0.55 0.74 100.16 44 18 38 71
Cathedral Crag
CC1_Mph_c 51.21 3.29 0.64 6.70 0.21 15.89 21.03 0.38 0.01 99.38 46 11 43 81
CC1_Ph_c 51.93 2.72 0.55 7.19 0.19 15.55 20.97 0.33 0.01 99.44 45 12 43 79
CC1_Ph_r 51.56 1.54 0.91 9.67 0.39 15.74 18.69 0.50 0.02 99.01 45 16 39 74
CC7_Ph_c 49.84 4.96 1.20 6.04 0.14 14.97 21.55 0.50 0.45 99.66 44 10 46 82
CC7_Ph_r 52.54 3.24 0.33 4.22 0.11 17.27 21.08 0.35 0.93 100.07 50 7 44 88
CC7_Ma_c 51.89 2.03 0.94 9.17 0.22 16.06 19.33 0.38 0.09 100.11 46 15 40 76
Tarn Plateau
TP3_Mph_c 52.46 2.51 0.35 4.36 0.10 16.78 22.28 0.24 0.58 99.66 48 7 45 87
TP3_Ma_c 50.74 2.00 1.09 11.62 0.29 15.05 18.08 0.33 0.01 99.23 44 19 38 70
TP4_Mph_c 51.33 2.51 0.52 5.90 0.17 16.07 21.20 0.22 0.16 98.09 46 10 44 83
Sulphur Creek
SC1_Ma_c 50.21 2.65 1.13 9.30 0.29 15.29 19.21 0.46 0.06 98.60 45 15 40 75
SC4_Ma_c 50.78 2.66 1.46 9.77 0.27 14.42 19.78 0.58 0.01 99.73 42 16 42 72
Orthopyroxene
Park Butte
PB5_Ma_c 52.05 0.52 0.35 20.48 0.57 20.56 4.18 0.12 0.03 98.86 59 33 9 64
PB8_Ma_r 52.71 0.47 0.49 20.43 0.47 20.60 3.76 0.11 0.00 99.04 59 33 8 64
PB8_Ma_c 50.39 0.63 0.29 23.53 0.55 20.08 2.80 0.12 0.00 98.39 57 37 6 60
Cathedral Crag
CC7_Ma_c 53.85 1.16 0.46 14.91 0.37 26.21 2.28 0.03 0.06 99.31 72 23 5 76
Tarn Plateau
TP3_Ma_c 52.90 1.52 0.52 19.03 0.41 23.12 2.15 0.06 0.02 99.72 65 30 4 68
TP4_Ma_c 52.27 1.13 0.58 18.54 0.60 23.40 2.18 0.09 0.00 98.78 66 29 4 69
Sulphur Creek
SC4_Ma_c 35.78 1.53 0.11 32.80 0.77 27.34 0.97 0.32 0.00 99.63 59 40 2 60
Abbreviations as in Table 2. En enstatite content, Fs ferrosilite content, Wo wollastonite content
Mg# = Mg/(Mg ? Fe*)
Contrib Mineral Petrol
123
tephra associated with the Sulphur Creek lava flow
(equates to Fe2?/FeT of 0.83). Similar values were also
determined for other Mount Baker units (Baggerman and
DeBari 2011).
Whole rock chemistry
Major elements
Representative major element chemical data for the five
mafic lava flow units from Mount Baker are presented in
Table 4. All units are subalkaline and according to the
AFM diagram of Irvine and Baragar (1971), all are calc-
alkaline (not shown). However, the Park Butte lavas plot
directly on the dividing line between the calc-alkaline and
tholeiitic fields. Using the classification of Le Maitre et al.
(1989), the Cathedral Crag, Tarn Plateau, Lake Shannon,
and Sulphur Creek lavas are all categorized as medium-K,
while the Park Butte unit falls just into the low-K field
(Fig. 3).
Harker variation diagrams for major elements are pre-
sented in Fig. 4. Park Butte lavas have the most Si-poor
compositions and are exclusively basaltic (49.3–50.3 wt%
SiO2) with 7.9–8.4 wt% MgO. Basaltic andesite is the sole
rock type found at Cathedral Crag (52.1–52.9 wt% SiO2)
with *4 wt% MgO. Tarn Plateau is basaltic andesite
(53.6–54.0 wt% SiO2), with the exception of one sample
that borderline qualifies as basalt (51.8 wt% SiO2). This
unit is relatively Mg rich for its SiO2 content, with 7.0–7.8
wt% MgO, and we designate the unit as high-Mg basaltic
andesite (HMBA; cf. Grove et al. 2002; 2005; Tatsumi
2006). This high-Mg character is shared by another Mt.
Baker lava, the Glacier Creek andesite (plotted on Fig. 4 as
Mt. Baker HMA for comparison; data from Baggerman and
DeBari 2011). The Lake Shannon and Sulphur Creek units
both include basalt and basaltic andesite (50.7–52.6%, and
51.2–54.6 wt% SiO2, respectively). However, there is a
distinct compositional gap in Sulphur Creek lavas between
basalts with B52 wt% SiO2 and basaltic andesites
with C54.5 wt% SiO2 (cf. Hildreth et al. 2003; Baggerman
and DeBari 2011). Basalts in both of these units are poorer
in MgO than Park Butte low-K lavas and Tarn Plateau
HMBA.
There are no clear trends in major elements between the
different units. Interestingly, Mg# is not correlated with
silica content (Fig. 4), as Tarn Plateau basaltic andesites
have both the highest silica content and the highest Mg#
(68–70) of all units in the study. The most Si-poor unit,
Park Butte, has the second highest Mg# (63–65). The other
basaltic andesite unit, Cathedral Crag, has the lowest Mg#
(55–58), with the exception of one sample (NM-CC7) that
has Mg# 72 due to olivine accumulation. Both MgO
(Fig. 4) and FeO (Table 4) are highest in Park Butte
(7.9–8.4 wt% MgO, 8.9–9.7 wt% FeO).
There is a relatively small range in Al2O3 across four of
the units (15.7–17.9 wt%), but the fifth, Cathedral Crag, is
notably enriched in Al2O3 (20–20.2 wt%, Fig. 4). This is
reflected in the high abundance of plagioclase phenocrysts
in Cathedral Crag lavas (see Table 1) but not in positive Eu
anomalies, as discussed below.
The abundances of TiO2, CaO, and Na2O are distinctly
different between units (Fig. 4). The Sulphur Creek and
Lake Shannon samples have the highest TiO2 ([1.4 wt%)
and Na2O ([4 wt%), and Tarn Plateau and Park Butte have
the lowest (\0.95 wt% TiO2, \3 wt% Na2O). Conversely,
CaO is greatest in the Park Butte and Tarn Plateau samples
(abnormally high for the HMBA of Tarn Plateau) and
lowest in Sulphur Creek and Lake Shannon. The high-Mg
basaltic andesites from Tarn Plateau have the lowest TiO2
(*0.7 wt%).
Although major element compositions vary significantly
between the units, variations within units are limited,
especially for Al2O3, TiO2, and Na2O. Lake Shannon is the
only flow unit that shows a trend in Mg#, MgO and FeO.
Sulphur Creek mafic lavas (B52 wt% SiO2) show no var-
iation, but there is a trend across the compositional gap to
basaltic andesite (Fig. 4). In contrast, K2O content does
generally increase with silica within each individual unit
and across the range of the flow units, forming a sublinear
trend (Fig. 3).
In summary, major elements distinguish three types of
primitive magmas at Mount Baker: low-K basalt (Park
Butte, similar to low-K olivine tholeiites elsewhere in the
Cascades), calc-alkaline basalts (Lake Shannon and Sul-
phur Creek), and high-Mg basaltic andesites and andesites
(Tarn Plateau and Glacier Creek). Major elements also
distinguish between groups of less primitive basaltic
andesites (Cathedral Crag and Sulphur Creek).
Trace elements
Compatible trace element compositions vary widely among
the units, and in some cases within units. In this study,
abundances of both Ni and Cr are relatively low compared
to other primitive Cascade calc-alkaline and LKOT basalts
(Fig. 5). Park Butte LKOT and Tarn Plateau HMBA have
similar Ni content (52–69 ppm), but Park Butte has higher
Cr than Tarn Plateau at a given Ni abundance. Lake
Shannon basalt has the highest Ni content of all the flow
units (62–90 ppm) and is the only lava that exhibits a
predictable decrease in both Ni and Cr as silica decreases.
However, this unit has much lower Cr relative to the Tarn
Plateau HMBA and Park Butte low-K basalt.
Other trace element compositions differ greatly between
the units in this study, and trends within units are variably
Contrib Mineral Petrol
123
Ta
ble
4R
epre
sen
tati
ve
wh
ole
rock
maj
or
and
trac
eel
emen
tco
mp
osi
tio
ns
Unit
Par
kB
utt
eC
athed
ral
Cra
gT
arn
Pla
teau
Lak
eS
han
non
Sulp
hur
Cre
ek
Sam
ple
NM
-PB
2N
M-P
B5
NM
-PB
6N
M-P
B8
NM
-CC
1N
M-C
C5
NM
-CC
6N
M-T
P2
NM
-TP
3N
M-T
P4
NM
-TP
6N
M-L
S1
NM
-LS
2N
M-L
S6
NM
-SC
1N
M-S
C2
NM
-SC
3N
M-S
C6
Majo
rel
emen
ts(w
t%)
SiO
249.4
549.9
150.0
549.5
952.4
652.1
652.4
853.9
753.6
053.9
053.9
051.3
152.4
850.7
051.2
451.5
551.4
951.4
4
TiO
21.0
64
1.0
63
1.0
41
1.0
13
1.1
34
1.0
53
1.1
35
0.9
19
0.9
05
0.9
22
0.8
95
1.4
63
1.4
39
1.4
20
1.6
20
1.6
35
1.6
34
1.6
37
Al 2
O3
17.3
016.9
317.1
017.3
220.1
020.0
320.0
916.1
715.8
416.1
515.7
417.7
617.8
717.9
117.3
417.5
517.5
217.4
7
FeO
*9.6
69.1
59.3
98.9
06.4
26.3
26.4
06.9
27.0
26.9
56.9
48.5
08.0
88.3
38.5
08.8
08.7
48.7
7
MnO
0.1
68
0.1
66
0.1
65
0.1
61
0.1
13
0.1
12
0.1
13
0.1
29
0.1
34
0.1
31
0.1
32
0.1
57
0.1
50
0.1
55
0.1
64
0.1
66
0.1
65
0.1
65
MgO
8.1
48.1
68.3
88.1
83.9
74.1
73.8
17.3
57.7
97.0
37.8
86.3
65.2
86.4
25.3
95.4
85.3
75.4
0
CaO
9.1
79.1
99.2
19.2
99.3
29.5
19.3
39.4
79.7
69.4
19.6
68.8
48.0
28.8
18.2
48.4
08.2
98.3
3
Na 2
O3.2
13.1
63.2
33.2
03.8
33.7
03.8
72.8
92.8
02.9
42.8
34.0
84.2
43.9
14.3
64.4
14.3
74.3
9
K2O
0.4
50.4
50.4
40.4
20.9
80.9
40.9
80.9
50.8
60.9
50.9
10.6
40.8
80.5
50.8
60.8
50.8
50.8
6
P2O
50.1
50
0.1
60
0.1
55
0.1
49
0.3
32
0.3
15
0.3
34
0.1
84
0.1
78
0.1
84
0.1
76
0.2
81
0.3
06
0.2
73
0.4
29
0.4
27
0.4
29
0.4
31
Tota
l98.7
698.3
399.1
798.2
498.6
698.3
198.5
498.9
598.8
998.5
799.0
699.4
198.7
498.4
898.1
599.2
798.8
598.8
9
Mg#
63.8
65.1
65.2
65.8
56.4
58.1
55.5
69.0
69.9
68.0
70.4
61.1
57.8
61.7
757.1
56.6
56.3
56.3
Tra
ceel
emen
tsby
XR
F(p
pm
)
Ni
67
65
69
65
20
25
19
56
61
52
63
85
62
90
38
38
37
37
Cr
253
235
257
243
23
28
22
198
233
176
234
164
115
170
93
93
91
92
V174
192
185
182
201
191
205
180
181
181
176
178
166
173
202
203
202
201
Ga
17
18
18
17
20
19
19
16
17
17
17
18
18
17
17
18
18
18
Cu
24
26
25
26
44
45
31
40
43
36
40
60
58
69
38
34
40
37
Zn
82
79
82
77
69
68
70
65
67
65
65
76
77
75
85
85
86
84
Tra
ceel
emen
tsby
ICP
-MS
(ppm
)
La
8.0
47.8
18.0
87.6
214.8
116.1
117.0
612.8
112.6
712.4
012.3
411.4
313.2
611.5
016.2
115.8
216.3
716.0
6
Ce
18.4
918.4
418.2
117.6
733.0
536.4
538.2
329.2
428.6
128.3
427.4
326.8
630.4
226.5
438.8
537.8
038.9
838.3
6
Pr
2.6
32.6
02.6
62.4
94.5
34.9
75.2
44.0
03.9
63.8
63.8
23.7
24.1
23.7
25.3
55.2
05.3
95.3
2
Nd
11.8
811.6
611.9
711.1
719.1
421.1
321.9
416.8
416.7
716.2
416.1
516.3
517.7
316.3
023.1
422.9
423.0
823.0
9
Sm
3.1
63.0
83.0
62.9
34.1
14.4
84.7
53.7
63.7
43.6
23.6
24.2
74.3
94.1
25.6
75.6
15.6
25.5
7
Eu
1.1
71.1
51.1
71.1
11.3
31.4
61.5
21.2
31.2
11.2
21.1
71.5
51.5
31.5
01.8
61.8
81.8
71.8
9
Gd
3.5
33.5
33.5
73.3
13.5
54.0
04.2
23.6
03.6
53.4
63.4
44.7
34.7
84.6
25.9
15.8
45.9
55.8
4
Tb
0.6
30.6
20.6
30.5
80.5
40.6
00.6
30.5
80.5
70.5
60.5
50.8
20.8
10.8
00.9
80.9
70.9
90.9
7
Dy
4.0
13.9
84.0
33.6
43.1
13.4
53.6
43.4
73.4
73.3
43.3
35.1
65.0
05.0
86.0
65.9
86.0
35.9
7
Ho
0.8
50.8
50.8
60.7
90.6
10.6
80.7
10.7
10.7
00.6
60.6
81.0
81.0
51.0
71.2
41.2
21.2
71.2
2
Er
2.3
62.3
42.3
22.1
71.6
21.8
01.8
51.8
91.8
91.8
11.7
92.9
22.8
92.9
13.3
73.3
33.4
23.3
2
Tm
0.3
40.3
50.3
40.3
20.2
30.2
50.2
60.2
70.2
70.2
60.2
60.4
40.4
30.4
30.4
90.4
80.4
90.4
9
Yb
2.1
12.1
32.1
62.0
41.3
81.5
31.6
21.6
81.6
51.6
41.5
92.6
82.6
22.6
53.0
32.9
92.9
93.0
0
Lu
0.3
30.3
40.3
30.3
20.2
10.2
40.2
50.2
60.2
60.2
50.2
50.4
20.4
20.4
20.4
70.4
70.4
80.4
8
Ba
207
202
207
203
403
405
431
350
322
333
315
212
283
212
284
279
287
280
Th
0.6
30.6
40.6
00.6
61.6
31.8
41.9
02.1
62.0
82.1
22.0
51.0
81.5
11.1
51.3
91.2
81.4
51.3
6
Contrib Mineral Petrol
123
developed (Fig. 6). Park Butte LKOT has the lowest
abundances of both large-ion lithophile elements (LILE,
Ba, Rb, and Pb) and high-field-strength elements (HFSE,
Zr, Th, and Ce) compared to the other four flow units
(Table 4). Abundances of Ba, Sr, Th, Yb, Pb, and Ce in the
Park Butte basalt are comparable to values published for
LKOT elsewhere in the Cascade arc (Bacon et al. 1997).
Values of Rb, Ba, Pb, and Th generally increase with silica
between units but do not show much variation within units
(except for Lake Shannon and the shift across the com-
positional gap of Sulphur Creek). Zr, Nb, and Y values
vary widely between units, but are constant for all samples
within a given flow unit. Sr values in the Tarn Plateau
HMBA and Cathedral Crag lavas vary widely within those
units and are much higher than in the other units
(750–900 ppm and 1,000–1,200 ppm, respectively). Nb
and Ta display incompatible behavior in Lake Shannon
samples, but values within the other units are relatively
constant (Table 4). Sulphur Creek lavas have anomalously
high Zr, Nb, Y, and Yb (Figs. 5, 6). These lavas show a
slight decrease in Y and Yb with silica (Fig. 7). Overall,
most trace elements have relatively clustered values within
each flow unit (with the exception of Sr in Cathedral Crag
samples, see Fig. 5).
The NMORB-normalized trace element diagram
(Fig. 6) reveals patterns very typical of calc-alkaline arc
lavas. All the samples are enriched in LILE and exhibit
distinctive negative Nb and Ta anomalies that are repre-
sentative of arc magmas whose generation has been
influenced by a subduction component. The lavas all have
pronounced enrichment in Ba, U, Pb, and Sr. Park Butte
LKOT is the least enriched in these trace elements and
shifts to a flatter, more MORB-like pattern on the right side
of the diagram, with moderate enrichment compared to the
other units. This is similar to patterns for LKOT from other
Cascade volcanoes (Bacon et al. 1997). The Tarn Plateau
HMBA and Cathedral Crag basaltic andesites have the
lowest abundances of HFSE on the right of the diagram,
but have intermediate Nb and Ta relative to the other flow
units. This type of pattern is similar to high magnesium
andesites (HMA) from Mount Shasta (Grove et al. 2002).
The Sulphur Creek and Lake Shannon calc-alkaline basalts
display a third type of trend, with intermediate HFSE and
the highest Nb and Ta; however, the relative enrichment of
the Sulphur Creek samples is slightly higher than those
from Lake Shannon.
On the chondrite-normalized REE diagram (Fig. 7),
Park Butte LKOT samples have the least enriched LREE
signature and flattest overall patterns, with (LREE)N
*20–35 and (La/Sm)N B 1.7. Tarn Plateau HMBA has
moderate LREE enrichment and a relatively steep REE
pattern, with (LREE)N values from 40 to 55 and (La/Sm)N
at *2.2. Cathedral Crag, the most Mg-poor lava, has highTa
ble
4co
nti
nu
ed
Unit
Par
kB
utt
eC
athed
ral
Cra
gT
arn
Pla
teau
Lak
eS
han
non
Sulp
hur
Cre
ek
Sam
ple
NM
-PB
2N
M-P
B5
NM
-PB
6N
M-P
B8
NM
-CC
1N
M-C
C5
NM
-CC
6N
M-T
P2
NM
-TP
3N
M-T
P4
NM
-TP
6N
M-L
S1
NM
-LS
2N
M-L
S6
NM
-SC
1N
M-S
C2
NM
-SC
3N
M-S
C6
Nb
2.7
42.7
82.6
42.6
44.2
34.6
14.8
73.5
83.3
93.5
03.3
75.1
06.3
34.9
97.2
17.0
97.3
17.2
2
Y21.3
821.2
021.2
619.9
815.6
517.1
618.0
718.0
518.1
617.4
017.3
527.2
026.2
826.7
731.2
230.9
931.4
731.0
6
Hf
2.0
32.0
71.9
61.9
22.2
82.5
12.6
12.9
82.8
32.8
82.8
03.3
13.5
93.2
04.3
74.2
64.3
64.3
0
Ta
0.1
70.1
80.1
80.1
70.2
50.2
70.2
70.2
40.2
30.2
30.2
20.3
60.4
60.3
60.4
80.4
60.4
90.4
8
U0.2
40.2
50.2
40.2
70.5
50.6
30.6
50.7
80.7
20.7
00.7
30.4
80.6
70.5
00.6
10.5
80.6
30.6
0
Pb
1.6
41.6
81.7
91.7
92.9
43.3
63.7
43.5
13.3
13.1
63.1
62.7
03.6
42.8
33.4
23.2
63.4
93.4
0
Rb
4.8
5.4
5.6
5.7
9.9
12.1
11.6
12.4
10.1
11.6
11.7
8.1
12.1
7.7
9.8
8.3
9.2
9.0
Cs
0.0
50.0
60.0
50.0
70.0
90.1
50.1
20.2
00.2
20.1
70.1
70.1
80.2
70.1
90.2
60.2
00.2
30.2
5
Sr
521
491
499
499
1,0
51
1,2
06
1,2
27
854
853
858
833
477
506
490
559
563
560
562
Sc
28.3
30.1
30.0
28.3
17.7
20.7
21.2
25.8
26.7
26.4
28.2
26.6
21.4
25.7
25.4
25.9
26.2
26.1
Zr
76
77
73
72
87
97
100
107
103
105
102
145
157
141
195
192
198
195
All
anal
yse
sper
form
edat
the
WS
UG
eoA
nal
yti
cal
Lab
.F
eis
report
edas
all
FeO
*.
Mg#
=M
g/(
Mg
?F
e2?
),F
e2?
isca
lcula
ted
as0.8
5F
eO*.
For
det
ails
on
accu
racy
,se
e‘‘
Anal
yti
cal
met
hods’
’se
ctio
n
Contrib Mineral Petrol
123
LREE enrichment and the steepest REE pattern, with
(LREE)N values from 45 to 70 and (La/Sm)N at *2.3. All
of the samples from both Park Butte and Lake Shannon
have minor positive Eu anomalies, and one Sulphur Creek
sample appears to have a very slight negative Eu anomaly.
Notably, the Cathedral Crag samples with high modal
0
0.5
1
1.5
2
2.5
3
45 47 49 51 53 55 57 59 61 63
K2O
(wt.
%)
SiO2 (wt.%)
Park Butte (LKOT-like)
Cathedral Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Creek (CA)
Mt. Baker HMA
High-K(Calc-Alkaline)
Medium-K(Calc-Alkaline)
Low-K(Tholeiitic)
Shoshonitic
Fig. 3 K2O versus SiO2
classification diagram. Fields
from Le Maitre et al. (1989).
Abbreviations for lava types are
indicated in parentheses next to
their respective unit names:
basaltic andesite (BA), high-Mg
basaltic andesite (HMBA), high-
Mg andesite (HMA), low-K
olivine tholeiite (LKOT-like),
and calc-alkaline (CA). Mt.
Baker HMA from the Glacier
Creek unit of Baggerman and
DeBari (2011). Black and browndashed fields encompass fields
of other Cascade primitive calc-
alkaline basalt (CAB) and
LKOT (data compiled from
GEOROC database 2008)
15
16
17
18
19
20
21
48 50 52 54 56 58 60
Al 2
O3 (w
t.%
)
Uncertainty
0.5
1.0
1.5
2.0
48 50 52 54 56 58 60
TiO
2(w
t.%
)
SiO2 (wt.%)
Uncertainty
52
56
60
64
68
72
48 50 52 54 56 58 60 M
g #
6
7
8
9
10
11
48 50 52 54 56 58 60
CaO
(w
t.%
)
Uncertainty
2
4
6
8
10
48 50 52 54 56 58 60
Mg
O (w
t.%
)
Park Butte (LKOT-like)
Cath. Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Cr. (CA)
Felsic SC
Mt. Baker HMA
Uncertainty
2
3
4
5
48 50 52 54 56 58 60
Na 2
O (
wt.
%)
SiO2 (wt.%)
Uncertainty
Fig. 4 Select major element
variation diagrams. Symbols,
data sources, and abbreviations
as in Fig. 3. Cathedral Crag
sample circled in black is
accumulative in phenocrysts
(see ‘‘Whole rock chemistry’’).
Open purple triangles are more
felsic Sulphur Creek samples
from Hildreth et al. (2003).
Maximum uncertainty in
analyses shown at bottom left.Mg# = Mg/(Mg ? Fe2?), Fe2?
is calculated as 0.85 FeO*
Contrib Mineral Petrol
123
plagioclase, high Al2O3, and high Sr do not have significant
positive Eu anomalies. This suggests that high modal pla-
gioclase is not accumulative, but is a result of high alumina
content in the melt.
Isotopic composition
Isotopic compositions for Sr, Nd, and Pb were analyzed for
the three most primitive flow units: Park Butte, Tarn Pla-
teau, and Lake Shannon (Table 5). The most mafic and
most felsic sample in the Park Butte and Lake Shannon
units were selected to represent the probable isotopic
compositional range of each (note that the range of silica
content is *1 and 2%, respectively, in these units). The
range in silica content for Tarn Plateau is less than the error
associated with XRF analysis, and thus two samples with
identical silica values were analyzed. The samples fall
within or just outside the mantle array on the eNd versus87Sr/86Sr (Fig. 8a) and plot within the field for primitive
lavas from other parts of the Cascade arc.
Measured 87Sr/86Sr values in the analyzed units have a
range of 0.70309–0.70335 (Table 5; Fig. 8a). This range is
wider within the Lake Shannon and Park Butte lavas than
between the units. In the Lake Shannon samples, 87Sr/86Sr
300
500
700
900
1100
1300
48 50 52 54 56 58 60
Sr
(pp
m)
0
4
8
12
16
20
24
48 50 52 54 56 58 60
Rb
(pp
m)
SiO2 (wt.%)
0
50
100
150
200
250
48 50 52 54 56 58 60
Zr
(pp
m)
100
200
300
400
500
48 50 52 54 56 58 60
Ba
(pp
m)
10
20
30
40
48 50 52 54 56 58 60
Y (p
pm
)
0
2
4
6
8
10
48 50 52 54 56 58 60
Nb
(pp
m)
SiO2 (wt.%)
0
50
100
150
200
250
300
350
45 50 55 60
Ni (
pp
m)
Park Butte (LKOT-like)
Cath. Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Cr. (CA)
Felsic SC
Mt. Baker HMA
Cascades CABCascades LKOT
0
100
200
300
400
500
600
45 50 55 60
Cr
(pp
m)
Cascades LKOT
Cascades CAB
Fig. 5 Select minor and trace
element variation diagrams.
Symbols, data sources, and
abbreviations as in Figs. 3, 5.
More felsic Sulphur Creek
samples are from Baggerman
and DeBari (2011). Black and
brown dashed lines in Ni and Cr
diagrams encompass fields of
other Cascade primitive calc-
alkaline basalts (CAB) and low-
K olivine tholeiites (LKOT; data
compiled from GEOROC
database 2008)
Contrib Mineral Petrol
123
0.1
1
10
100
Cs Rb Ba Th U Nb Ta K La Ce Pb Sr P Nd Zr Hf Sm Eu Ti Tb Y Yb Lu
Sam
ple
/NM
OR
B
Park Butte (LKOT-like)
Cathedral Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Creek (CA)
Diff. SC (Mixed)
Mt. Baker HMA
Mt. Shasta LKOT
Mt. Shasta PMA
Fig. 6 NMORB-normalized
trace element diagram for
representative samples.
Normalization values from Sun
and McDonough (1989). Data
sources and abbreviations as in
Fig. 3. Diff. SC (mixed) is a
Sulphur Creek basaltic andesite
sample calculated by mixing of
mafic and felsic endmember
magmas (Baggerman and
DeBari 2011). Mt. Shasta
LKOT and PMA (Primitive
magnesian andesites) from
Grove et al. (2002). The high-
Mg lavas and derivatives (Tarn
Plateau HMBA, Cathedral Crag
BA, and Mt. Baker HMA) are
most depleted in HREE and plot
nearly atop one another
10
100
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb Lu
Sam
ple
/Ch
on
dri
te
Park Butte (LKOT-like)
Cathedral Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Creek (CA)
Diff. SC (Mixed)
Mt. Baker HMA
NMORB
Mt. Shasta LKOT
Mt. Shasta PMA
4
8
12
16
20
48 50 52 54 56 58 60
La
(pp
m)
SiO2 (wt.%)
PBCCTPLSSCMB HMA
0
1
2
3
4
48 50 52 54 56 58 60
Yb
(pp
m)
SiO2 (wt.%)
Fig. 7 Chondrite-normalized
rare earth element (REE)
diagram. Normalization values
and NMORB from Sun and
McDonough (1989). Data
sources and abbreviations as in
Figs. 3 and 6. Select REE (La
and Yb) versus SiO2 are plotted
to show changes in pattern
across the range of each flow
unit
Contrib Mineral Petrol
123
is higher in the more felsic samples, but in the Park Butte
samples, this ratio is higher in the more mafic samples.
Tarn Plateau samples have constant 87Sr/86Sr. The range of87Sr/86Sr does not extend to the low values seen in the
Columbia segment of the Cascade arc (Schmidt et al.
2008), nor does it extend to the high values seen in the
Central segment of the arc. The values are typical of the
range seen in the southern Cascades of California (Schmidt
et al. 2008).
Measured eNd values in the analyzed units are between
?6.8 and ?8.0 (Table 5; Fig. 8a). In contrast to 87Sr/86Sr,
the range is wider between units than within units, and
within flow unit variation is within analytical error.
All three units plot just above the NHRL (Northern
Hemisphere Reference Line) on the 207Pb/204Pb versus206Pb/204Pb diagram (Fig. 8b), but plot just below the
NHRL in the 208Pb/204Pb versus 206Pb/204Pb diagram (not
shown). Pb isotopic ratios range from 18.75 to 18.86 for206Pb/204Pb, 15.54–15.56 for 207Pb/204Pb, and 38.20–38.29
for 208Pb/204Pb (Table 5). All Pb isotope ratios increase
with silica in the Lake Shannon lavas, and the same is true
for 206Pb/204Pb and 208Pb/204Pb in the Park Butte samples;
however, 207Pb/204Pb decreases as silica increases in Park
Butte samples.
Summary of chemical characteristics
Overall, major and trace element chemistry reveals that the
flow units can be categorized as three distinct primitive
endmember magma types: calc-alkaline basalts, LKOT-
like basalts, and high-Mg basaltic andesite (with probable
lower-Mg derivatives).
The calc-alkaline lavas are represented by the basalts of
Lake Shannon and Sulphur Creek. They exhibit charac-
teristic enrichment in large-ion lithophile elements (LILE:
Ba, Rb, K, Pb) and relative depletions in Nb and Ta
(Fig. 6). However, they display the highest overall abun-
dances of Nb, Ta, Zr, and middle REE among the units.
The LKOT-like endmember is represented by the basalt
of Park Butte, which has major element chemistry similar
to LKOT found elsewhere in the Cascades (Bacon et al.
1997, Conrey et al. 1997; Leeman et al. 1990). Park Butte
also has the lowest abundances of Nb, Ta, Zr, and middle
REE among the units.
The third magma type, high-Mg basaltic andesite
(HMBA), is represented by the basaltic andesite of Tarn
Plateau. It has high MgO for a given silica content, and
Mg# higher than the basalts of this study (Fig. 4). This flow
unit is strongly depleted in heavy rare earth elements
(HREE; Fig. 7), but has intermediate Nb, Ta, Zr, and
middle REE compared to the other lavas (Fig. 6). Other
high-Mg andesites with similar trace element characteris-
tics are present at Mount Baker (the Glacier Creek unit, or
Mount Baker HMA, see Baggerman and DeBari 2011;
Figs. 5, 6, 7). The basaltic andesite of Cathedral Crag is not
a high-Mg lava, but it shares the same depletion in heavy
REE, with even higher (La/Yb)N.
Isotopic compositions do not show any systematic
trends for these lava types. Within flow unit variability in87Sr/86Sr is as great as the variability between units,
Table 5 Sr, Nd, and Pb isotopic compositions
Sample 87Sr/86Sr eSr143Nd/144Nd eNd
206Pb/204Pb 207Pb/204Pb 208Pb/204Pb SiO2
Park Butte
NM-PB3 0.703347 -16.4 0.512997 7.0 18.748 15.552 38.215 49.25
NM-PB6 0.703157 -19.1 0.512988 6.8 18.806 15.547 38.238 50.05
Tarn Plateau
NM-TP4 0.703090 -20.0 0.513011 7.3 18.778 15.547 38.224 53.90
NM-TP6 0.703094 -20.0 0.513027 7.6 18.771 15.543 38.212 53.90
NM-TP6 (2nd solution) – – – – 18.771 15.542 38.205 53.90
Lake Shannon
NM-LS2 0.703291 -17.2 0.513026 7.6 18.861 15.554 38.283 52.48
NM-LS2 (repeat) – – – – 18.861 15.555 38.286 52.48
NM-LS6 0.703193 -18.6 0.513049 8.0 18.837 15.548 38.249 50.70
NM-LS6 (2nd solution) – – – – 18.833 15.548 38.244 50.70
Analyses performed at the University of Washington Isotope Geochemistry Laboratory by Dr. Bruce Nelson. Error on Nd and Sr analyses
is ± 30 ppm (2r) or ± 0.3 epsilon units; error on Pb analyses is ± 150 ppm (2r) or ± 0.015%. NM-LS2 (repeat) is a second analyses of the
same solution
NM-TP6 (2nd solution) and NM-LS6 (2nd solution) are second aliquots of the dissolved sample that went through the entire Pb separation
chemistry
Contrib Mineral Petrol
123
whereas the compositional range of Nd isotopes within
each unit is negligible (well within analytical error), and
the spread between the units is very small as well.
H2O contents
Sisson and Grove (1993) presented a method for estimating
H2O content of magmas. By calculating the exchange of Ca
and Na in plagioclase versus liquid, the resulting Kd rela-
tionship (defined as [Ca/Na]plag/[Ca/Na]liquid) can be cor-
related with H2O content. Since high water content leads to
more calcic plagioclase, as H2O of the magma increases, so
does KdCa-Na of plagioclase–liquid.
Figure 9 shows Ca/Na relationships and resulting esti-
mates of H2O contents using the whole rock compositions
of CaO and Na2O as a proxy for liquid and associated
plagioclase grains. The plagioclase grains selected for use
with this hygrometer include phenocrysts and micro-
phenocrysts with the highest An content to represent the
earliest-formed grains. Only the unresorbed portions of the
cores were analyzed. Estimated H2O content of the units
are as follows: *2% for Lake Shannon and Sulphur
Creek, *3 wt% H2O for Park Butte, *4 wt% H2O for
Tarn Plateau, and *4.5 wt% H2O for Cathedral Crag.
These estimated H2O contents are used in the following
sections for interpreting eruption temperatures, contribu-
tions from the subducting slab, and for modeling fluid
compositions. The results of the hygrometer are supported
by the following arguments. First, the unit that gives the
highest H2O content estimate, the Cathedral Crag basaltic
andesite, also has the highest Al2O3 with no associated Eu
anomaly. Gaetani et al. (1993) demonstrated that plagio-
clase crystallization is suppressed in H2O-rich magmas,
causing elevated Al2O3 content. Second, H2O content
estimates calculated for the Sulphur Creek basalt (2%) are
consistent with the 2.3% H2O measurement obtained by
Shaw (2011) using FTIR of olivine-hosted melt inclusions
from tephra associated with the Sulphur Creek lava.
Finally, measured H2O contents (by FTIR) for LKOT-like
lavas at Glacier Peak are also not very different from calc-
alkaline lavas (2.0 wt% as compared to 2.2 wt%; Shaw
2011).
Magmatic temperatures and equilibration pressures
Magma equilibration temperatures can be determined from
major element compositions using the olivine–liquid ther-
mometer of Suguwara (2000) and H2O liquidus depres-
sions from Medard and Grove (2008). Using the estimated
H2O contents from Fig. 9, the LKOT-like lavas of Park
Butte record the highest liquidus temperatures of
1,110–1,120�C. The calc-alkaline lavas record magmatic
temperatures of 1,070–1,100�, with the more evolved
Sulphur Creek lavas recording temperatures on the cooler
end of this range. The high-Mg basaltic andesites of Tarn
Plateau record similar temperatures (1,065–1,085�), a
15.48
15.52
15.56
15.60
15.64
15.68
15.72
19.0018.50
207P
b /2
04P
b
206Pb/ 204Pb
(b)
1
2
3
4
5
6
7
8
9
0.7028 0.7032 0.7036 0.7040 0.7044
εNd
87Sr/ 86Sr
Park Butte (LKOT-like)
Tarn Plateau (HMBA)
Lake Shannon (CA)
(a)
Fig. 8 Select isotopic compositions. Abbreviations as in Fig. 3. a eNd
versus 87Sr/86Sr and b 207Pb/204Pb versus 206Pb/204Pb. Black dashedfields indicate the range of compositions for Cascade primitive lavas
as compiled in Schmidt et al. 2008. Error bars are symbol sized or
smaller where not visible on the diagrams. NHRL from Hart (1984),
Bulk Cascadia subducted sediment from Plank and Langmuir (1998),
and Juan de Fuca MORB from White et al. (1987). ‘‘M’’ indicates the
most mafic sample analyzed from the Park Butte and Lake Shannon
flow units
0
5
10
15
20
0 1 2 3 4 5
Ca/
Na
Pla
gio
clas
e
Ca/Na Liquid (WR)
Park Butte (LKOT-like)Cath. Crag (BA)Tarn Plateau (HMBA)Lake Shannon (CA)Sulphur Creek (CA)
Fig. 9 Ca–Na exchange in plagioclase and liquid. Ca/Na (molar)
liquid is from whole rock compositions. Estimates of H2O content (in
wt%) are based on Kd after the method of Sisson and Grove (1993).
Note that two points exist for the Park Butte unit that plot nearly atop
one another. See text for details
Contrib Mineral Petrol
123
result of a much larger liquidus depression because of
higher H2O contents. The more differentiated Cathedral
Crag lavas record significantly lower temperatures
of *985�. Equilibration pressures using the Si-activity
barometers of Albarede (1992) and Putirka (2008) are
variable, but yield the lowest pressures for Tarn Plateau
and Cathedral Crag lavas (5–8 kbar), intermediate pres-
sures for the calc-alkaline lavas of Lake Shannon and
Sulphur Creek (6–10 kbar), and the highest pressures for
the LKOT-like lavas of Park Butte (9.5–10.5 kbar).
Discussion
The focus of this discussion is the characterization of
subducted slab and mantle inputs for the production of
mafic lavas (Mg# 55–70) beneath the northern Cascade arc
at Mount Baker. We begin the discussion by first estab-
lishing that these whole rock compositions are indeed
representative of mantle-derived liquid compositions. This
requires the assessment of mineral–whole rock equilibria
and effects of crustal processes such as assimilation and
magma mixing.
Assessment of equilibrium and crustal effects
Because the primitive lavas in this study are moderately to
strongly phenocryst rich, we used olivine–liquid distribu-
tion coefficients to ascertain whether olivine compositions
are in equilibrium with whole rock compositions (i.e.,
whether whole rocks represent equilibrium liquids). Oliv-
ine core compositions from Tarn Plateau and Lake Shan-
non lavas, as well as select samples from Park Butte,
represent near-equilibrium conditions (Electronic Supple-
ment 6). In contrast, olivine compositions from Cathedral
Crag, Sulphur Creek, and most (but not all) from Park
Butte, are more Fe rich than would be expected from their
whole rock Fe/Mg compositions. Hence, whole rock
compositions are too mafic to be in equilibrium with their
enclosed olivine, suggesting that these three units may have
experienced some (few %) accumulation of olivine.
Olivine accumulation will produce variation in the
whole rock composition of lavas, specifically in MgO,
SiO2, and Ni, and has been documented for other flows at
Mount Baker (Baggerman and DeBari 2011). Within flow
unit variations in MgO and SiO2 exist within flow units,
some of which cannot be ascribed to the standard error in
XRF analysis (Fig. 4). Thus, SiO2, MgO, and Ni contents
of the Park Butte, Cathedral Crag, and Sulphur Creek lavas
may not be reflective of liquids, and magmatic tempera-
tures described above may be maxima. However, olivine in
the Tarn Plateau high-Mg basaltic andesite does record
near-equilibrium conditions (Electronic Supplement 6),
thus their high-Mg nature is a liquid characteristic not
reflective of olivine accumulation.
Although some of the units in this study may have been
affected by olivine accumulation, the most incompatible
element abundances, those which clearly define mantle
source compositions, will be unaffected by this process.
The Cathedral Crag sample that has anomalous major
element chemistry and petrography (i.e., much higher
modal olivine than the rest of the unit) has a REE pattern
similar to the other samples in that unit (Fig. 7), indicating
that the REE budget is controlled by the melt rather than
mineral accumulation.
Xenolith textures described earlier suggest that assimi-
lation of exotic material plays an insignificant role in
generating the geochemical signature of the flow units. As
previously mentioned, xenoliths have been noted in only
one unit of this study, the basaltic andesite of Cathedral
Crag. The two types of xenoliths in this lava contain thin
reaction rims; however, the boundaries are very sharp,
indicating minimal chemical exchange between the xeno-
liths and the magma. Additionally, isotopic ratios do not
reflect assimilation of crustal material. The analyzed flows
have relatively low Sr and Pb isotopic ratios, relatively
high Nd isotopic ratios, and are comparable to other Cas-
cade primitive lavas (Fig. 8).
Magma mixing and mingling seem to be likely causes of
pervasive disequilibrium textures found in all of the units
(e.g., Green 1988; Baggerman and DeBari 2011), but has
not dramatically altered trace, REE, and some major ele-
ment abundances. Baggerman and DeBari (2011) showed
that mixing between Sulphur Creek basalt and Mount
Baker dacite can effectively produce the basaltic andesites
on the Si-rich side of the Sulphur Creek compositional gap.
In this scenario, 70% basalt mixed with 30% dacite yields a
close match with most major, trace, and REE compositions
of the basaltic andesite. However, even after this mixing,
the trace element chemistry of the Sulphur Creek basaltic
andesite samples is comparable to that of the Sulphur
Creek basalt (Figs. 6, 7). This suggests that magma mixing
had a minimal impact on the trace element signature of the
lavas, and thus mantle source characteristics are still
discernable.
In summary, despite pervasive textural evidence for
disequilibrium, these lavas appear to retain the trace ele-
ment signature of their mantle-derived parents. This is
demonstrated by: (1) the lack of variation in many of the
most highly incompatible elements, even in the units that
have experienced olivine accumulation; (2) insignificant
addition of crustal material and lack of an obvious crustal
radiogenic source; and (3) minimal difference in trace
element chemistry between mixed lavas and their mafic
endmembers, even with significant addition by a felsic
endmember. Together, these observations suggest that
Contrib Mineral Petrol
123
distinct mantle source characteristics are still interpretable
from the mafic units of this study.
Characterization of mantle and slab sources beneath
the Mount Baker volcanic field
Endmember magma types
Three endmember primitive magma types were originally
described for the middle and southern Cascades: (1) a
typical hydrous, arc-like calc-alkaline basalt; (2) a depleted
and anhydrous low-K olivine tholeiite (LKOT), alterna-
tively called high alumina olivine tholeiite (HAOT); and
(3) an enriched, intraplate-type ocean island basalt (OIB;
e.g., Bacon et al. 1997; Borg et al. 1997; 2000; Conrey
et al. 1997; Grove et al. 2002; Leeman et al. 1990; 2005;
Reiners et al. 2000; Smith and Leeman 2005). More
recently, primary high-Mg andesites have been described at
Mount Shasta (Grove et al. 2002). Major and trace element
criteria are specific to each endmember group.
In this study, we present evidence for three endmember
magma types that bear resemblance to the types described
above. These include a calc-alkaline endmember, a LKOT-
like endmember (with modified trace element characteris-
tics), and a high-Mg basaltic andesite endmember. None of
the lavas of this study resemble OIB.
Calc-alkaline lavas in the Cascade arc are generally
medium-K (Fig. 3) with higher silica content than LKOT
magmas. The Lake Shannon and Sulphur Creek lavas from
Mount Baker are typical examples and are characterized by
enrichment of LREE relative to HREE, and enrichment of
LILE (Ba, Sr, Pb, Rb) relative to HFSE (Nb, Ta, Zr, Hf).
These lavas have nearly identical incompatible trace ele-
ment patterns (Figs. 6, 7), but abundances are higher in the
Sulphur Creek samples for a given Si content (Fig. 5).
These units were probably derived from similar sources,
with variable amounts of partial melting and/or fractional
crystallization.
The LKOT magmas in the Cascade arc are characterized
by [17 wt% Al2O3, generally \0.1 wt% K2O, and *2.5
wt% Na2O (Bacon et al. 1997). The Park Butte basalt fits
these criteria, containing high Al, but low K and Na
compared to other Mount Baker units (Figs. 3, 4). The
incompatible trace element abundances that are particularly
reflective of mantle source (i.e., Nb, Ta, Zr, and Hf) are
similar in Cascade LKOT, MORB, and the Park Butte
basalts. Cascade LKOT have distinct negative Nb and Ta
anomalies, as well as depleted Zr and Hf (Bacon et al.
1997, see Mount Shasta LKOT in Fig. 6), suggesting that a
depleted mantle source is responsible for their chemical
signature. However, Cascade LKOT show more enriched
LREE and LILE (e.g., Ba, Sr, Rb, Pb) than MORB, more
typical of back-arc basin basalts. The Park Butte basalt is
even more enriched than typical Cascade LKOT in these
elements (Figs. 6, 7), suggesting that magma generation
processes for these ‘‘LKOT-like’’ magmas are distinct in
this northern part of the arc.
High-Mg lavas occur throughout the Cascade arc
(Bacon et al. 1997), but have been studied in detail in the
southern Cascade arc at Mount Shasta (Grove et al. 2002;
2005) and in the northern Cascade arc at Glacier Peak
(Taylor 2001). In these studies, high-Mg lavas are thought
to be derived by fluid-induced melting of depleted mantle
with significant subduction component. The high-Mg lavas,
including Mount Baker’s Tarn Plateau basaltic andesite
(HMBA, this study) and associated Glacier Creek andesite
(HMA, Baggerman and DeBari 2011), have high Mg#
([60), high MgO wt% ([7%) and higher Ni and Cr than
other lavas with similar Si contents. In addition, Sr/Y ratios
are higher in the high-Mg lavas from Shasta and Mount
Baker ([40) than in typical arc andesites/basaltic andesites
(\30, see Electronic Supplement 7). Their La/Yb ratios are
also high ([La/Yb]N [5), reflective of low heavy REE
(Fig. 7) with higher Dy/Yb than the other units.
One distinct feature of northern Cascade arc high-Mg
lavas (Mt. Baker and Glacier Peak) compared to Mt. Shasta
is that basaltic andesites have the highest Mg# and MgO
contents (Mg# 68–70; 52–54 wt% SiO2) rather than
andesites (Mg #63–64; *58% SiO2). The HMAs are
clearly differentiates of the HMBAs (Figs. 4, 5) and cannot
represent slab melts as has been suggested for HMAs in
some arcs (e.g., Kelemen et al. 2003). Mount Baker’s
Cathedral Crag basaltic andesites share the characteristic
trace element trends of the Tarn Plateau HMBA end-
member, but not the high-Mg character. They are most
likely low-Mg differentiates of that endmember.
Slab components responsible for magma generation
The extent to which each mafic flow unit at Mount Baker
has been affected by a subduction component can be
assessed by utilizing LILE/HFSE ratios (e.g., Ba/Nb),
proxies such as primitive-mantle-normalized Sr/P ratios
([Sr/P]N; Borg et al. 1997), and estimated H2O contents.
The HFSE and P are fluid-immobile and are left behind
when a hydrous phase is transferred from the slab (e.g.,
Brenan et al. 1994), whereas LILE (Rb, Sr, Ba, K) are fluid
mobile and are enriched in slab fluids. Figure 10a shows a
clear correlation between Ba/Nb and (Sr/P)N. The calc-
alkaline Lake Shannon and Sulphur Creek basalts appear to
have been least modified by a subduction component, while
Tarn Plateau high-Mg basaltic andesites and Cathedral
Crag derivatives appear to have been most modified by slab
flux. Interestingly, the lava with the most LKOT-like sig-
nature, Park Butte basalt, falls between the other units.
Calculated H2O contents also generally increase with (Sr/
Contrib Mineral Petrol
123
P)N (Fig. 10b) and coincide with the endmember lava
types. These results are unusual, since LKOT lavas would
be expected to have the weakest slab flux signature. It is
evident that variable influence of the subduction compo-
nent is partially responsible for the distinct character of
each endmember magma type.
Modeling mantle and slab contributions
In this section, we model melting to match conditions in
the mantle wedge (mineralogy and % melting) that gen-
erate geochemical characteristics of the various endmem-
ber magmas at Mount Baker. In the first model, the source
for Park Butte LKOT and the calc-alkaline lavas (Lake
Shannon and Sulphur Creek) is assumed to be spinel
lherzolite (5% cpx). However, due to the strongly depleted
HREE character of the Tarn Plateau HMBA and Cathedral
Crag lavas, garnet must be present in their source. Garnet
can contribute to the depleted HREE signature of these
lavas by several possible mechanisms: melting of a mantle
source containing garnet (e.g., model of Borg et al. 1997;
supported by experiments of Gaetani and Grove 1998),
fluids/melt contributed from a subducting slab containing
garnet (e.g., Kelemen et al. 2003), or sub-Moho garnet
fractionation (e.g., Macpherson 2008). Since there is no
textural evidence that garnet fractionation has occurred in
any of the lavas of this study, we prefer either of the first
two options. However, given that we cannot distinguish
whether the garnet source is from the slab or the mantle, we
model both possibilities. Matching the partial melts that
result from these models against lava compositions tests
the validity of these hypotheses.
Equilibrium, or batch melting, is assumed in this model
because it is consistent with a fluid flux–induced melting
process (Grove et al. 2002). In this type of melting, the
partial melt continuously reacts and re-equilibrates with the
residual crystalline solid, and melt is segregated in a single
event. Modal melting (where minerals contribute elements
to partial melts in the same proportions as they appear in
the solid) is also assumed in this case to reduce the number
of unknown variables. We use the batch modal melting
equation defined by Shaw (1970) for determining trace
element concentrations at various melt fractions (F).
Trace element concentrations of depleted MORB mantle
(DMM) used in the model are from Workman and Hart
(2005) and partition coefficients were compiled from sev-
eral sources (Electronic Supplement 8). Partial melting was
modeled at various melt fractions to determine the best fit
to the most primitive REE pattern (middle to heavy REE)
from each magma type (sample NM-PB8 for the LKOT
type, samples NM-LS6 and NM-SC1 for the calc-alkaline
type, and sample NM-TP6 for the high-Mg basaltic
andesite). Best fit was determined by dividing the modeled
HREE concentrations by the measured HREE values from
each sample, with the best fit approaching modelHREE/
measuredHREE = 1 (Figs. 11, 12). Modeled compositions
and parameters are reported in Table 6.
The focus of this melt modeling is to most closely
approximate the middle and heavy REE compositional
patterns of the lavas, since these are dominantly mantle-
derived, whereas LREE are largely slab-derived. Thus, the
results presented in this step are expected to show poor
matches with LREE. This poor match will be resolved in
the next step where subduction fluids are added.
Melting of a spinel lherzolite mantle source with the
assemblage Ol ? Cpx ? Opx ? Spl in the proportions
76:5:15:4 can effectively model the average HREE pattern
of the LKOT-like lava (Park Butte) and the calc-alkaline
lavas (Lake Shannon and Sulphur Creek; Fig. 11). Melting
percentages required to produce these HREE abundances
are 4.2% for Sulphur Creek, 6% for Lake Shannon, and
10% for Park Butte. Variation in melt fraction is the most
logical explanation for the similarly shaped HREE pattern
at different HREE abundances seen in these three units;
however, a variably depleted mantle source cannot be ruled
0
20
40
60
80
100
120
Ba/
Nb
Park Butte (LKOT-like)
Cathedral Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Creek (CA)
Mt. Baker HMA
(a)
0
1
2
3
4
5
6
0 1 2 3 4 5 6
0 1 2 3 4 5 6
H2O
wt.
%
(Sr/P)N
(b)
Fig. 10 Trace element ratios and estimated H2O wt% versus
primitive-mantle-normalized Sr/P [(Sr/P)N]. Abbreviations and data
sources as in Fig. 3. a Ba/Nb versus (Sr/P)N. b Estimated H2O wt%
versus (Sr/P)N. Normalization values from Sun and McDonough
(1989). Estimated H2O wt% from Fig. 9 and for Sulphur Creek from
Shaw (2011)
Contrib Mineral Petrol
123
out. Using enriched MORB (E-DMM of Workman and
Hart 2005) mantle instead of DMM increases calculated
melt fractions by only a small amount (\0.05%), but does
significantly change required slab flux (see Discussion
below).
Partial melting of a garnet lherzolite source with the
assemblage Ol ? Cpx ? Opx ? Grt in the proportions
73:5:20:2 at 7% melt fraction can effectively model the
average HREE pattern of the Tarn Plateau high-Mg
basaltic andesite (Fig. 12). Melting of a spinel lherzolite
source with the assemblage Ol ? Cpx ? Opx ? Spl in the
proportions 76:5:15:4 at 16% melt fraction produces a less
robust match for the majority of the HREE (Fig. 12).
Garnet, rather than spinel, is a necessary component in this
model to successfully reproduce the majority of the
observed HREE (the spinel melt only closely approximates
Yb and Lu, see Fig. 12b, c). Cathedral Crag lavas are not
included in this model since we assume that they are
derivatives of Tarn Plateau-like magmas after crustal
differentiation.
In the following discussion, we model the REE com-
position of the subduction component, following the
assumption that the trace element composition of the
magma is derived from only two components: a mantle
1
10
100
1000
10000
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb Lu
Sam
ple
/Ch
on
dri
te
PB Lava
LS Lava
SC Lava
PB Melt (F=0.10)
LS Melt (F=0.06)
SC Melt (F=0.042)
PB Fluid
LS Fluid
SC Fluid
Mt. Shasta Fluid 85-15
Mt. Shasta Fluid 95-13
DMM
(a)
0.00.51.01.5
Dy
Ho Er
Yb Lu
CL/C
Ob
(c)
0.0
0.5
1.0
1.5
Dy
Ho Er
Yb Lu
CL/C
Ob
(d)
0.0
0.5
1.0
1.5
Dy
Ho Er
Yb Lu
CL/C
Ob
(b)
Fig. 11 Mantle partial melt and slab fluid models for Park Butte
(PB), Lake Shannon (LS), and Sulphur Creek (SC) lavas. See text for
discussion. a Chondrite-normalized REE abundances of representa-
tive lavas, modeled melt, and modeled slab fluids. Normalization
values from Sun and McDonough (1989), depleted MORB mantle
(DMM) from Workman and Hart (2005). Melts are from a depleted
spinel lherzolite source with the proportions Ol ? Cpx ? Opx ?
Spl = 76:5:15:4 at F = 0.10 for Park Butte, F = 0.06 for Lake
Shannon, and F = 0.042 for Sulphur Creek. Mt. Shasta calculated
fluids from Grove et al. (2002). Plots b–d are ratios of modeled melt
HREE abundances to measured HREE abundances to demonstrate
best-fit melt fraction for each unit. Fit is perfect where the ratio of
(CL)/(COb) or modeled abundance to observed abundance equals 1
Contrib Mineral Petrol
123
partial melt and a hydrous subduction component. This
subduction component may either be a supercritical fluid or
a small degree, H2O-rich partial melt. Following the
method of Grove et al. (2002), a mass balance equation can
be used to calculate the concentration of each trace element
contributed by the subduction component and is defined as
Cf ¼ Clava � XmCmð Þ= Xfð Þ
where Cf is the concentration of the trace element con-
tributed by the subduction component, Clava is the con-
centration of the trace element in the lava (from Table 4),
Cm is the concentration of the trace element contributed by
the mantle partial melt (from melt modeling above), Xm is
the weight fraction of the lava derived from the mantle
melt, and Xf is the weight fraction of the lava that was
contributed by a subduction component. Results are
presented in Table 6. The H2O contents estimated by
Ca–Na exchange in plagioclase–liquid (Fig. 9) are used as
an approximation of Xf (2% for Lake Shannon and Sulphur
Creek, 3% for Park Butte, 4% for Tarn Plateau).
The REE abundances of the subduction component for
each of the units are illustrated in Figs. 11 and 12. They
display a broad range of compositions, but all are enriched
in LREE and depleted in HREE. The calc-alkaline lavas
are distinct in that their fluid components have much higher
overall REE abundances than both the LKOT and HMBA.
Adding the appropriate proportions (2–4%) of each of these
fluids to the corresponding melts of the mantle wedge
successfully produces the range of primitive compositions
seen at Mount Baker. Similar results are obtained when the
subduction flux model of Portnyagin et al. (2007) is used
(not shown).
1
10
100
1000
10000
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb Lu
Sam
ple
/Ch
on
dri
te
TP Lava
TP Garnet Source Melt (F=0.07)
TP Garnet Source Fluid
TP Spinel Source Melt (F=0.16)
TP Spinel Source Fluid
Mt. Shasta Fluid 85-15
Mt. Shasta Fluid 95-13
DMM
(a)
0.0
0.5
1.0
1.5
Dy
Ho Er
Yb
Lu
CL/
CO
b
Garnet Melt Fit(b)
0.0
0.5
1.0
1.5
Dy
Ho Er
Yb
Lu
CL/
CO
b
Spinel Melt Fit(c)
Fig. 12 Mantle partial melt and slab fluid models for Tarn Plateau
(TP) lavas. See text for discussion. a Chondrite-normalized REE
abundances of representative lava, modeled melt, and modeled slab
fluids. Normalization values from Sun and McDonough (1989),
depleted MORB mantle (DMM) from Workman and Hart (2005).
Melts are from a depleted garnet lherzolite source with the
proportions Ol ? Cpx ? Opx ? Grt = 73:5:20:2 at F = 0.07, and
a depleted spinel lherzolite source with the proportions Ol ? Cpx ?
Opx ? Spl = 76:5:15:4 at F = 0.16. Mt. Shasta calculated fluids
from Grove et al. (2002). Plots b, c are ratios of modeled melt HREE
abundances to measured HREE abundances to demonstrate best-fit
melt fraction for each unit. Fit is perfect where the ratio of (CL)/(COb)
or modeled abundance to observed abundance equals 1
Contrib Mineral Petrol
123
An important caveat to the model results presented in
Figs. 11, 12 is that we have assumed that both LKOT-like
lavas and calc-alkaline lavas come from a similarly
depleted source (DMM). If we instead assume that calc-
alkaline lavas come from a more enriched mantle source
(E-DMM of Workman and Hart 2005) than the LKOT-like
lavas, then the calculated subduction components are less
enriched and more comparable to that calculated for the
LKOT-like lavas (Electronic Supplement 9). Given the
data we have, we cannot distinguish between these models.
Another important point is that the HMBA lavas require
some residual garnet for their generation, either in the
mantle wedge as a garnet lherzolite source (producing an
HREE-depleted mantle melt) or in the slab (producing an
HREE-depleted subduction component). The models pre-
sented cannot distinguish between them. However, if one
magma type at Mt. Baker requires a strongly HREE-
depleted subduction flux, why isn’t this flux evident in all
the lavas? If we assume a garnet lherzolite source, the
subduction component is similar in HREE character to that
of the other lavas (cf. Figures 11, 12) as is the degree of
melting (7% as compared to 16%). It seems more plausible
to envision a mantle wedge comprising both spinel lherz-
olite (shallow) and garnet lherzolite (deep) producing
similar degree partial melts with variable HREE than a
subducting slab producing such diverse fluid components.
Summary model of mafic magma evolution at Mount
Baker
The schematic illustration in Fig. 13 summarizes one
possible scenario for the evolution of mafic magmas at
Mount Baker. As the Juan de Fuca plate subducts, varying
degrees and compositions of a subduction component
infiltrate the mantle wedge beneath the volcano. This slab
flux assists in melting two distinct mantle regions: spinel
lherzolite and garnet lherzolite. This melting occurs at
various fractions, and the resulting melts stall at different
levels in the crust as they ascend. Differentiation occurs in
the crust at a range of pressures and to differing degrees,
producing magmas with diverse chemistry, resulting in the
three endmember lava types observed at Mount Baker:
LKOT-like, calc-alkaline, and high-Mg basaltic andesite.
LKOT-like lavas are the result of the highest degrees of
partial melting from a spinel lherzolite source. However,
they are atypical of LKOTs from elsewhere in the arc due
to their greater slab signature (both H2O content and trace
element characteristics). This may be due to the lack of an
active back-arc spreading system at this latitude. Their
distinct compositional characteristics compared to calc-
alkaline lavas can be attributed either to a more depleted
mantle source with a compositionally similar slab flux
(Electronic Supplement 9) or to a similar mantle source
Table 6 Modeled mantle partial melt and slab fluid REE compositions
DMM Spinel source Garnet source Fluid composition
4.2% 6% 10% 16% 7% SC LS PB TP Sp TP Grt
La 0.192 3.77 2.79 1.77 1.14 2.43 650.6 437.9 196.8 281.1 250.3
Ce 0.550 10.24 7.69 4.95 3.23 6.68 1,505.1 950.2 429.0 608.3 525.5
Pr 0.107 1.86 1.42 0.93 0.62 1.24 187.3 116.4 52.8 80.6 65.7
Nd 0.581 9.69 7.48 4.97 3.31 6.23 737.4 448.5 211.6 324.5 254.4
Sm 0.239 3.48 2.78 1.91 1.30 2.33 130.1 69.9 35.9 59.1 34.5
Eu 0.096 1.27 1.03 0.73 0.51 0.85 36.6 24.1 13.1 17.2 9.0
Gd 0.358 4.28 3.55 2.57 1.82 2.71 103.5 57.2 27.1 42.2 20.9
Tb 0.070 0.80 0.67 0.49 0.35 0.50 12.8 7.0 3.5 5.4 1.9
Dy 0.505 5.34 4.53 3.38 2.45 3.07 60.9 32.2 12.1 24.3 9.4
Ho 0.115 1.13 0.97 0.74 0.54 0.63 10.7 6.3 2.2 4.1 2.0
Er 0.348 3.09 2.69 2.09 1.57 1.74 26.5 13.5 4.6 7.2 3.1
Yb 0.365 2.82 2.51 2.01 1.54 1.51 20.6 9.7 3.0 2.8 3.6
Lu 0.058 0.46 0.41 0.32 0.25 0.25 2.4 0.9 0.4 0.3 0.3
Xf – – – – – 2.0% 2.0% 3.0% 4.0% 4.0%
Xm – – – – – 98.0% 98.0% 97.0% 96.0% 96.0%
Depleted MORB mantle (DMM) starting composition from Workman and Hart (2005)
PB Park Butte basalt, TP Tarn Plateau basaltic andesite, LS Lake Shannon basalt, SC Sulphur Creek basalt. TP Sp and TP Grt are calculated
fluids based on spinel (16%) and garnet melts, respectively. Modeled compositions of partial melts (4.2, 6, 10, and 16%) from spinel lherzolite
(‘‘spinel source’’) are from a mantle source with Ol ? Cpx ? Opx ? Spl in the proportions 76:5:15:4. Modeled composition of partial melt (7%)
from garnet lherzolite (‘‘garnet source’’) are from a mantle source with Ol ? Cpx ? Opx ? Grt in the proportions 73:5:20:2. Xf is the weight
fraction of lava contributed by fluid in the fluid composition model and is based on estimated H2O content of the Mt. Baker mafic magmas (see
Fig. 9). Xm is the weight fraction of lava contributed by the mantle melt in the fluid composition model
Contrib Mineral Petrol
123
fluxed by a less enriched subduction component (Fig. 11).
This latter model is comparable to recent studies in the
central Oregon Cascades (Rowe et al. 2009), where melt
modeling suggested that a fertile mantle source (more
enriched than depleted MORB) can produce LKOT, calc-
alkaline, and OIB endmember primitive magmas, and
called upon variability in a subduction component to pro-
duce the different endmembers.
Flow age correlation with mantle and slab indicators
There is no correlation between age of the units and
petrogenetic indicators such as Sr/Y and (Sr/P)N (Fig. 14).
The LKOT-like Park Butte is the oldest unit and has
intermediate values of both mantle- and slab-influenced
trace element ratios. The Tarn Plateau and Cathedral Crag
basaltic andesites are intermediate in age among the units,
but have the highest values of these ratios (Fig. 14). The
Lake Shannon and Sulphur Creek lavas are youngest and
have the lowest values. This lack of correlation suggests
that different mantle and slab sources are being tapped
over the life of the Mount Baker volcanic field. Initially, a
depleted spinel lherzolite mantle with a moderate sub-
duction component was responsible for the generation of
mafic magma (LKOT-like Park Butte). Over time, an
additional mantle source with residual garnet and a larger
contribution from a subduction component became the
primary input for magma petrogenesis (Tarn Plateau
HMBA and Cathedral Crag basaltic andesite). This source
was tapped as recently as 14,000 years ago with the
eruption of the high-Mg andesites of Glacier Creek
(Baggerman and DeBari 2011). Spinel lherzolite contin-
ued to contribute to magma generation in the Lake
Shannon and Sulphur Creek units in the last
100,000 years, but with the lowest degree of slab input
over the life of mafic volcanism (Fig. 14). This suggests
that both mantle sources are being tapped simultaneously
at different depths in the mantle wedge beneath Mount
Baker, with variable amounts and compositions of slab
input.
Fig. 13 Schematic illustration of mafic magma evolution at Mt.
Baker. Mafic magmas are generated at various melt fractions when
slab fluid flux contributes to melting of mantle with different residual
mineralogy: a depleted spinel lherzolite and a depleted garnet
lherzolite. The resulting melts stall at different levels in the crust as
they ascend, where differentiation occurs at a range of pressures and
to varying degrees. This produces magmas with diverse chemistry,
resulting in the three endmember lava types observed at Mount Baker.
Depth estimates from McCrory et al. (2004), Robinson and Wood
(1998), and Thompson (1975)
0
20
40
60
80
Sr/
Y
Park Butte (LKOT-like)
Cathedral Crag (BA)
Tarn Plateau (HMBA)
Lake Shannon (CA)
Sulphur Creek (CA)
Mt. Baker HMA
0
1
2
3
4
5
6
0 200 400 600 800
0 200 400 600 800
(Sr/
P) N
Age (ka)
Fig. 14 Sr/Y and (Sr/P)N versus flow unit age. Symbols and
abbreviations as in Fig. 3. Mt. Baker HMA from Baggerman and
DeBari (2011). Ages from Hildreth et al. (2003)
Contrib Mineral Petrol
123
Summary and conclusions
The chemical and petrologic characteristics of five mafic
lava units from the Mount Baker volcanic field in the
northern Cascade arc reveal a diversity of near-primitive
compositions that require distinct mantle sources and
varying subducting slab influence to explain their petro-
genesis. Three distinct endmember magma types are rep-
resented that cannot be related by fractional crystallization
or other crustal processes. These include LKOT-like, calc-
alkaline, and high-Mg basaltic andesite.
The LKOT-like basalt of Park Butte is the result of partial
melting of a spinel lherzolite mantle that was moderately
fluxed by a subduction component. The calc-alkaline basalts
of Lake Shannon and Sulphur Creek may have been derived
from the same mantle source (DMM) or a more enriched one
(E-DMM), but at lower degrees of partial melting (discerned
from higher trace element abundances). They also have a
smaller contribution from the subducting slab (as evidenced
by lower H2O contents and lower values of slab fluid indi-
cators such as Ba/Nb and [Sr/P]N).
High-Mg basaltic andesite (HMBA) of Tarn Plateau was
likely derived by partial melting of a garnet lherzolite
mantle source that had the highest flux of subduction
component (highest H2O contents and [Sr/P]N). However,
we cannot rule out a spinel lherzolite mantle source with a
subduction component that is more HREE-depleted than
that required for the other lava types. Other units on Mount
Baker with similar trace element patterns must be derived
from this same source (e.g., high-Mg andesite of Glacier
Creek from Baggerman and DeBari (2011) and basaltic
andesite of Cathedral Crag). High-Mg andesites at Mt.
Baker have lower MgO and Mg# than high-Mg basaltic
andesite and are fractionated versions of the latter.
The age of the various endmembers lacks correlation
with petrogenetic indicators. This suggests that both mantle
sources have been tapped simultaneously throughout the
history of mafic volcanism in this region and that the
mantle has been variably fluxed by a subduction compo-
nent over time as well.
Acknowledgments Funding for this project was provided by the
Geological Society of America Parke D. Snavely, Jr. Cascadia
Research Award, the USGS Cascades Volcano Observatory Kleinman
Grant for Volcano Research, and Western Washington University.
The manuscript benefited from reviews by Pete Stelling, Elizabeth
Schermer, and two anonymous reviewers.
References
Albarede F (1992) How deep do common basaltic magmas form and
differentiate? J Geophys Res 97:10997–11009
Armstrong JT (1988) Bence-Albee after 20 years: review of the
accuracy of a-factor correction procedures for oxide and silicate
minerals. In: Newbury DE (ed) Microbeam analysis. San
Francisco Press, pp 469–476
Bacon CR, Hirschmann MM (1988) Mg/Mn partitioning as a test for
equilibrium between coexisting Fe-Ti oxides. Am Mineral
73(1–2):57–61
Bacon CR, Bruggman PE, Christiansen RL, Clynne MA, Donelly-
Nolan JM, Hildreth W (1997) Primitive magmas at five Cascade
volcanic fields: melts from hot, heterogeneous sub-arc mantle.
Can Mineral 35:397–423
Baggerman TD, DeBari SM (2011) The generation of a diverse suite
of Late Pleistocene and Holocene basalt through dacite lavas
from the northern Cascade arc at Mount Baker, Washington.
Contrib Mineral Petrol 161:75–99
Baker MB, Grove TL, Price R (1994) Primitive basalts and andesites
from the Mount Shasta region, N. California: products of varying
melt fraction and water content. Contrib Mineral Petrol
118:111–129
Borg LE, Clynne MA, Bullen TD (1997) The variable role of slab-
derived fluids in the generation of a suite of primitive calc-
alkaline lavas from the southernmost Cascades, California. Can
Mineral 35:425–452
Borg LE, Brandon AD, Clynne MA, Walker RJ (2000) Re-Os isotopic
systematics of primitive lavas from the Lassen region of the
Cascade arc, California. Earth Planet Sci Lett 177:301–317
Brenan JM, Shaw HF, Phinney DL, Ryerson FJ (1994) Rutile-
aqueous fluid partitioning of Nb, Ta, Hf, Zr, U and Th;
implications for high field strength element depletions in island-
arc basalts. Earth Planet Sci Lett 128:327–339
Class C, Miller DM, Goldstein SL, Langumuir CH (2000) Distin-
guishing melt and fluid subduction components in Umnak
Volcanics, Aleutian Arc. Geochem Geophys Geosyst 1 (paper
number 1999GC000010)
Conrey RM, Sherrod DR, Hooper P, Swanson D (1997) Diverse
primitive magmas in the Cascade arc, northern Oregon and
southern Washington. Can Mineral 35:367–396
DeBari SM, Taylor DD, Sisson TW (2005) Basalts and high-Mg
andesites from the Northern Cascade arc (Glacier Peak, Wash-
ington): Insights into mantle and crustal processes. 15th Annual
Goldschmidt conference abstracts. Geochimica Cosmochimica
Acta 69(10):A268
Defant MJ, Drummond MS (1990) Derivation of some modern arc
magmas by melting of young subducted lithosphere. Nature
347:662–665
Elkins-Tanton LT, Grove TL, Donnelly-Nolan J (2001) Hot, shallow
mantle melting under the Cascades volcanic arc. Geology
29:631–634
Gaetani GA, Grove TL, Bryan WB (1993) The influence of water on
the petrogenesis of subduction-related igneous rocks. Nature
365:332–334
Gaetani GA, Grove TL (1998) The influence of water on melting of
mantle peridotite: Contributions to Mineralogy and Petrology,
vol 131. pp 323–346
Gaffney AM, Blichert-Toft J, Nelson BK, Bizzarro M, Rosing M,
Albarede F (2007) Constraints on source-forming processes of
West Greenland kimberlites inferred from Hf-Nd isotope
systematics. Geochim Cosmochim Acta 71:2820–2836
GEOROC Database (2008) Geochemistry of rocks of the oceans and
continents, July 6, 2008. http://georoc.mpch-mainz.gwdg.de/
georoc/start.asp
Green NL (1988) Basalt-basaltic andesite mixing at Mount Baker
volcano, Washington, I. Estimation of mixing conditions.
J Volcanol Geotherm Res 34:251–265
Green NL (2006) Influence of slab thermal structure on basalt source
regions and melting conditions: REE and HFSE constraints from
Contrib Mineral Petrol
123
the Garibaldi volcanic belt. northern Cascadia subduction
system. Lithos 87:23–49
Green NL, Harry DL (1999) On the relationship between subducted
slab age and arc basalt petrogenesis, Cascadia subduction
system, North America. Earth Planet Sci Lett 171:367–381
Green NL, Sinha AK (2005) Consequences of varied slab age and
thermal structure on enrichment processes in the sub-arc mantle
of the northern Cascadia subduction system. J Volcanol Geo-
therm Res 140:107–132
Grove TL, Parman SW, Bowring SA, Price RC, Baker MB (2002)
The role of an H2O-rich fluid component in the generation of
primitive basaltic andesites from the Mount Shasta region, N
California. Contrib Mineral Petrol 142:375–396
Grove TL, Baker MB, Price RC, Parman SW, Elkins-Tanton LT,
Chatterjee N, Muntener O (2005) Magnesian andesite and dacite
lavas from Mt. Shasta, northern California: products of fractional
crystallization of H2O-rich mantle melts. Contrib Mineral Petrol
148:542–565
Guffanti M, Weaver CS (1988) Distribution of Late Cenozoic
Volcanic Vents in the Cascade Range: Volcanic arc segmenta-
tion and regional tectonic considerations. J Geophys Res
93:6513–6529
Halliday AL, Lee D, Tommasini S, Davies GR, Paslick CR, Fitton JD,
James DE (1995) Incompatible trace elements in OIB and
MORB and source enrichment in the sub-oceanic mantle. Earth
Planet Sci Lett 133:379–395
Harkins SA, Appold MS, Nelson BK, Brewer AM, Groves IM (2008)
Lead isotope constraints on the origin of nonsulfide zinc and
sulfide zinc-lead deposits in the Flinders Ranges, South Austra-
lia. Econ Geol 103:353–364
Hart SR (1984) A large scale isotope anomaly in the Southern
Hemisphere mantle. Nature 309:753–757
Hart SR, Dunn T (1993) Experimental cpx/melt partitioning of 24
trace elements. Contrib Mineral Petrol 113:1–8
Hildreth W (2007) Quaternary magmatism in the Cascades—geologic
perspectives. USGS Professional Paper, no. 1744, p 125
Hildreth W, Moorbath S (1988) Crustal contributions to arc magma-
tism in the Andes of Central Chile. Contrib Mineral Petrol
98:455–489
Hildreth W, Fierstein J, Lansphere M (2003) Eruptive history and
geochronology of the Mount Baker volcanic field, Washington.
GSA Bull 115:729–764
Hirschmann M, Wiens D, Peacock S (2000) Subduction factory
science plan, June 15, 2007, p 42. http://www.ldeo.columbia.
edu/margins/SF_Sci_Plan_revised.pdf
Hyndman RD, Yorath CJ, Clowes RM, Davis EE (1990) The northern
Cascadia subduction zone at Vancouver Island: Seismic structure
and tectonic history. Can J Earth Sci 27:313–329
Irvine TN, Baragar WRA (1971) A guide to the chemical classifi-
cation of the common volcanic rocks. Can J Earth Sci 8:523–548
Johnson DM, Hooper PR, Conrey RM (1999) XRF analysis of rocks
and minerals for major and trace elements on a single low
dilution Li-tetraborate fused bead. Adv X-Ray Anal 41:843–867
Kelemen PB, Yogodszinski GM, Scholl DW (2003) Along-Strike
Variation in the Aleutian Island Arc: genesis of high Mg#
andesite and implications for continental crust. In: Eiler J (ed)
Inside the subduction factory, geophysical monograph, vol. 138.
American Geophysical Union, Washington, DC, pp 223–246
Knaack C, Cornelius SB, Hooper PR (1994) Trace element analyses
of rocks and minerals by ICP-MS, March 2, 2009.
http://www.sees.wsu.edu/Geolab/note/icpms.html
Le Maitre RW, Bateman P, Dudek A, Lameyre Le Bas MJ, Sabine
PA, Schmid R, Sorensen H, Streckeisen A, Woolley AR,
Zanettin B (1989) A classification of igneous rocks and glossary
of terms. Blackwell, Oxford
Leeman WP, Smith DR, Hildreth W, Palacz Z, Rogers N (1990)
Compositional diversity of late Cenozoic basalts in a transect
across the southern Washington Cascades: implications for
subduction zone magmatism. J Geophys Res 95:19561–19582
Leeman WP, Lewis JF, Evarts RC, Conrey RM, Streck MJ (2005)
Petrologic constraints on the thermal structure of the Cascades
arc. J Volcanol Geotherm Res 140:67–105
Lepage L (2003) ILMAT: an Excel worksheet for ilmenite-magnetite
geothermometry and geobarometry. Comput Geosci 29:673–678
Macpherson CG (2008) Lithosphere erosion and crustal growth in
subduction zones: Insights from initiation of the nascent East
Philippine Arc. Geology 36:311–314
McCrory PA, Blair JL, Oppenheimer DH, Walter SR (2004) Depth to
the Juan de Fuca slab beneath the Cascadia subduction margin—
a 3-D model for sorting earthquakes. U.S. Geological Survey
Data Series 91, CD-Rom Version 1.2, July 13, 2009, p 13.
http://pubs.usgs.gov/ds/91
McKenzie D, O’Nions RK (1991) Partial melt distributions from
inversion of rare earth element concentrations. J Petrol
32:1021–1091
Medard E, Grove TL (2008) The effect of H2O on the olivine liquidus
of basaltic melts: experiments and thermodynamic models.
Contrib Mineral Petrol 155:417–432
Mooney WD, Weaver CS (1989) Regional crustal structure and
tectonics of the Pacific coastal states; California, Oregon, and
Washington. Geol Soc Am Memoir 172:129–161
Nelson BK (1995) Fluid flow in subduction zones: evidence from
neodymium and strontium isotope variations in metabasalts of
the Franciscan Complex, California. Contrib Mineral Petrol
119:247–262
Plank T, Langmuir CH (1998) The geochemical composition of
subducting sediment and its consequences for the crust and the
mantle. Chem Geol 145:325–394
Portnyagin M, Hoernle K, Plechov P, Mironov N, Khubunya S (2007)
Constraints on mantle melting and composition and nature of
slab components in volcanic arcs from volatiles (H2O, S, Cl, F)
and trace elements from melt inclusions in the Kamchatka arc.
Earth Planet Sci Lett 255:53–69
Putirka K (2008) Thermometers and barometers for volcanic systems.
In: Putirka K, Tepley F (eds) Minerals, inclusions and volcanic
processes, vol 69. Reviews in Mineralogy and Geochemistry,
Mineralogical Soc. Am., pp. 61–120
Reiners PW, Hammond PE, McKenna JM, Duncan RA (2000) Young
basalts of the central Washington Cascades, flux melting of the
mantle, and trace element signatures of primary arc magmas.
Contrib Mineral Petrol 138:249–264
Riddihough RP (1984) Recent movements of the Juan de Fuca Plate
system. J Geophys Res 89:6980–6994
Robinson JAC, Wood BJ (1998) The depth of the spinel to garnet
transition at the peridotite solidus. Earth Planet Sci Lett
164:277–284
Roeder PL, Emslie RF (1970) Olivine-liquid equilibrium. Contrib
Mineral Petrol 29:275–289
Rollinson H (1993) Using geochemical data: evaluation, presentation,
interpretation. Longman, London
Rowe MC, Kent AJR, Nielsen RL (2009) Subduction influence on
oxygen fugacity and trace and volatile elements in basalts across
the Cascade volcanic arc. J Petrol 50:61–91
Schmidt ME, Grunder AL, Rowe MC (2008) Segmentation of the
Cascade Arc as indicated by Sr and Nd isotopic variation among
diverse primitive basalts. Earth Planet Sci Lett 266:166–181
Scott KM, Macias JL, Naranjo JA, Rodriguez S, McGeehin JP (2001)
Catastrophic debris flows transformed from landslides in volca-
nic terrains: Mobility, behavior, and mitigation strategies. U.S.
Geological Survey Professional Paper 1630, p 59
Contrib Mineral Petrol
123
Shaw DM (1970) Trace element fractionation during anatexis.
Geochim Cosmochim Acta 34:237–243
Shaw SD (2011) H2O contents in olivine-hosted melt inclusions from
primitive magmas in the northern Cascade arc [M.S. Thesis].
Western Washington University, p 85
Sisson TW, Bronto S (1998) Evidence for pressure-release melting
beneath magmatic arcs from basalt at Galunggung, Indonesia.
Nature 391:883–886
Sisson TW, Grove TL (1993) Experimental investigations of the role
of H2O in calc-alkaline differentiation and subduction zone
magmatism. Contrib Mineral Petrol 113:143–166
Smith DR, Leeman WP (2005) Chromian spinel-olivine phase
chemistry and the origin of primitive basalts of the southern
Washington Cascades. J Volcanol Geotherm Res 140:49–66
Strong M, Wolff J (2003) Compositional variations within scoria
cones. Geology 31:143–146
Suguwara T (2000) Empirical relationships between temperature,
pressure, and MgO content in olivine and pyroxene saturated
liquid. J Geophys Res 105:8457–8472
Sun SS, McDonough WF (1989) Chemical and isotopic systematics
of oceanic basalts: implications for mantle composition and
processes. Geol Soc Special Pub 42:313–345
Tabor RW (1994) Late Mesozoic and possible early Tertiary
accretion in western Washington State: the Helena-Haystack
melange and the Darrington-Devils Mountain Fault Zone. GSA
Bull 106:217–232
Tabor RW, Haugerud RA, Hildreth W, Brown EH (2003) Geologic
map of the Mount Baker 30 x 60 minute quadrangle, Washing-
ton: U. S. Geological Survey Map I-2660, scale 1:100,000
Tatsumi Y (2006) High-Mg andesites in the Setouchi Volcanic Belt,
Southwestern Japan: analogy to Archean magmatism and
continental crust formation? Annu Rev Earth Planet Sci
34:467–499
Taylor DD (2001) Petrology and geochemistry of primitive lavas near
Glacier Peak, North Cascades, Washington [M.S. Thesis].
Western Washington University, p 86
Thompson RN (1975) Primary basalts and magma genesis II: Snake
River plain, Idaho, USA. Contrib Mineral Petrol 52:213–232
Tucker DS, Scott KM (2009) Structures and facies associated with the
flow of subaerial basaltic lava into a deep freshwater lake: the
Sulphur Creek lava flow, North Cascades, Washington. J Volc-
anol Geotherm Res. doi:10.1016/j.jvolgeores.2008.11.028
Tucker DS, Scott KM, Foit FF, Mierendorf RR (2007) Age,
distribution, and composition of Holocene tephras from Mount
Baker, Cascade arc, Washington, USA. Geol Soc Am Abs
Program 39:66
Van Westrenen W, Blundy J, Wood B (1999) Crystal-chemical
controls on trace element partitioning between garnet and
anhydrous silicate melt. Am Mineral 84:838–847
Walker JA, Roggensack K, Patino LC, Cameron BI, Matıas O (2003)
The water and trace element contents of melt inclusions across
an active subduction zone. Contrib Mineral Petrol 146:62–77
White WM, Hofmann AW, Puchelt H (1987) Isotope geochemistry of
Pacific mid ocean ridge basalt. J Geophys Res 92:4881–4893
Workman RK, Hart SR (2005) Major and trace element composition
of the depleted MORB mantle (DMM). Earth Planet Sci Lett
231:53–72
Contrib Mineral Petrol
123