Jökull - 01.12.2006, Qupperneq 74

David W. McGarvie et al.
and more-evolved rhyolites progressing to younger
and less-evolved rhyolites seen on Figures 6 and 7
over a c. 300 ka period, and supported by data in Mc-
Garvie (1985) and Ívarsson (1992).
In the absence of detailed information on the pro-
duction of rhyolitic magmas through time (which
does not yet exist for Torfajökull, except during the
Holocene), the general trends shown on Figures 6
and 7 (which span four glacial maxima, see Figure 5)
can be interpreted as suggesting that variations in ice
thickness may have had no effect on the composi-
tion of erupted melts. This contrasts with how basaltic
magma systems respond to ice loading and unload-
ing, as they show excellent evidence of higher produc-
tion rates (and subtly-different compositions) during
deglaciation (e.g. Jull and Mackenzie, 1996; Slater
et al., 1998; Maclennan et al., 2002). The reason(s)
for this difference require further investigation, but it
is suggested that these may be linked to differences
in magma generation, with basalt dominated by man-
tle processes and rhyolite dominated by crustal pro-
cesses. In emphasising this difference, it is notewor-
thy that immediately following the last deglaciation,
whilst basaltic magma systems across Iceland were
erupting at rates >30 times higher than at present, no
corresponding pulse in Holocene rhyolitic eruptions
occurred at Torfajökull, Iceland’s largest active rhyo-
lite central volcano.
CONCLUSIONS
• Five new Ar-Ar ages of Pleistocene rhyolites
from the Torfajökull central volcano range from
384 ka to 67 ka, and span four glacial peri-
ods (Fuhne, OI stage 10), Drenthe (OI stage 8),
Warthe (OI stage 6), and Weichselian (OI stages
4 and 2).
• For three of the ages there is good correlation
between the climatic conditions indicated by
the oxygen isotope record (cold, during glacial
periods), and the field evidence indicating erup-
tion of rhyolite into thick ice sheets. This repre-
sents the first step towards developing a rhyolite
tuya proxy for past ice sheet thicknesses.
• Two Ar-Ar ages of 67±9 ka and 72±7 ka
that were obtained on widely-separated tuyas of
the postulated c. 16 km3 ring fracture eruption
(McGarvie, 1984) strengthen the argument that
a large eruptive event took place during the We-
ichselian, with the eruption probably occurring
during (cold) OI stage 4.
• Whole-rock geochemical data confirm a trend
of decreasing peralkalinity with time, with the
most recent (Holocene) rhyolites being dom-
inated by subalkaline compositions. A near-
linear trend of trace element concentrations
with time requires further investigation, as it
may simply be an artefact of the small data
set. If the near-linear trend is real, then it
raises the question of how a magma system
can evacuate batches of magma of successively
less-evolved composition over a period span-
ning c. 300 ka, in a fashion that mimics the
enrichment-depletion trace element signatures
of individual compositionally-zoned eruptions.
• Further work should involve improvements in
the Ar-Ar method to reduce uncertainties on the
ages of younger (<400 ka) rhyolites, which will
enable better information on ice sheet thick-
nesses for constraining paleoclimate models.
ACKNOWLEDGEMENTS
John Watson (OU) is thanked for doing the XRF anal-
ysis. Special thanks go to Mark Davies (OU) for moti-
vation and good humour during fieldwork at Torfajök-
ull. Thanks also go to Smári, Nina, Eydís, and Daði at
Landmannalaugar for good cheer and a solid roof dur-
ing poor weather. Discussions with Ray Macdonald,
John Smellie, Jennie Gilbert, Harry Pinkerton, Gret-
ar Ívarsson and Magnús Tumi Guðmundsson helped
to shape various parts of this paper. Ben Edwards and
an anonymous reviewer are thanked for thorough and
constructive comments that improved the final paper.
Finally, we wish to thank Bryndís Brandsdóttir for her
editorial acumen.
72 JÖKULL No. 56