Geoscience Reference
In-Depth Information
since the isotopic properties of the water are modified by
brine formation associated with the freezing of seawater.
The δ
18
O of foraminifera also depends on salinity and tem-
perature and cannot be interpreted directly by itself in terms
of sea ice formation. Nevertheless, recent studies of isotopic
properties of planktic foraminifera from the Arctic Ocean
suggest that their light δ
18
O values, far from isotopic equi-
librium values expected under the cold Arctic conditions,
are thought to relate to production rates of isotopically light
brines resulting from sea ice formation [e.g.,
Bauch
et al.
,
1997;
Hillaire-Marcel
et al.
, 2004].
Zonneveld and Versteegh
[2008]). Nevertheless, the bioge-
ography of dinocyst taxa and assemblages clearly illustrates
close relationships with sea ice cover extent. The develop-
ment of a large dinocyst database from the analyses of sur-
face sediment samples (0-1 cm in box cores or multicores)
indeed permitted the statistical evaluation of relationships
between dinocysts and the sea ice cover (Figures 1 and 2).
It also allowed the application of transfer functions for
the reconstruction of past sea ice [
de Vernal
and
Hillaire-
Marcel
, 2000;
de Vernal
et al.
, 2001, 2005a, 2005b].
3.2. Accuracy and Limit of the Approach Based on Transfer
Functions Using Dinocysts
3. QUANTIFICATION OF PAST SeA ICe COVeR IN
THe ARCTIC AND SUB-ARCTIC FROM DINOCYST
ASSeMBlAGeS
There are various techniques of transfer functions allow-
ing quantitative reconstructions of past climate or ocean
parameters [e.g.,
Guiot and
de Vernal
, 2007]. Calibration
techniques such as the
Imbrie and Kipp
[1971] method,
the weighted averaging partial least squares regression or
the artificial neural network approach were tested, but they
were not used for reconstructions because the results depend
strongly on the spatial extent of the reference database and
because it seems advisable to avoid assumptions concerning
the type of relationship (i.e., equations) between the dino-
cyst distribution and hydrographical parameters leading to
hazardous extrapolations. We choose to apply the modern
analogue technique (MAT) that is simply based on the com-
parison of past assemblages to modern ones in the reference
database using a dissimilarity index. A detailed description of
the MAT approach is given by
Guiot and
de Vernal
[2007].
Details on the procedures used for sea ice reconstructions
based on dinocyst assemblages are presented by
de Vernal
et
al.
[2001, 2005a, 2005b]. The results of validation exercises
and reconstructions that are presented here are based on a
database of 1189 reference sites that includes 64 dinocyst
taxa. Of the 1189 sites, 584 are characterized by the occur-
rence of sea ice during the 1954 to 2000 A.D. interval, which
is used as the “modern” reference.
The sea ice parameter compiled in the database is the
mean annual sea ice occurrence with an areal concentration
higher than 50%. The mean has been calculated on 1° by
1° (longitude-latitude) grid for the interval spanning 1954-
2000 A.D. based on data provided by the National Snow and
Ice Data Center in Boulder, Colorado. The reconstructions
are expressed as the duration of sea ice in months per year,
a parameter that is linearly correlated to the annual sea ice
concentration [cf., e.g.,
de Vernal
et al.
, 2005a].
One difficulty in any approach based on the analyses of
surface sediment samples compared to hydrographical ob-
servations lies in the fact that the two sets of data represent
different time intervals. On one side, the upper centimeter
3.1. Dinocysts as Proxy of Sea Ice Cover
As mentioned above, at the scale of the Arctic and sub-
arctic seas, most quantitative estimates of the past sea ice
cover were derived from dinocyst assemblages. The rela-
tionship between sea ice cover and dinocyst assemblages is,
however, not a simple function since none of the few dino-
flagellates living in sea ice produces fossilizable cysts. Only
two cyst-forming species are known to dwell in pack ice
environment:
Polarella
glacialis
and
Peridiniella
catenata
[
Matthiessen
et al.
, 2005]. However, their cysts are not re-
covered in sediment after usual laboratory procedures for
the analyses of dinocyst assemblages. Thus, sediments from
areas characterized by multiyear perennial pack ice are usu-
ally barren in dinocysts [e.g.,
Rochon
et al.
, 1999;
de Vernal
et al.
, 2005a]. Nevertheless, there are a few dinocyst taxa
that are known to occur in sediments from areas marked by
seasonal sea ice. In Arctic and subarctic seas,
Islandinium
species are often abundant, and some morphotypes of the
cyst of
Polykrikos
sp. occur exclusively in seasonally ice-
covered marine environments [e.g.,
de Vernal
et al.
, 2001,
2005a]. Many other taxa appear tolerant to sea ice cover and
may occur in high proportions in sea ice environments. This
is notably the case of the ubiquitous taxa
Operculodinium
centrocarpum
and
Brigantedinium
spp. Other taxa such as
Pentapharsodinium dalei
,
Spiniferites elongatus-frigidus
,
and
Impagidinium pallidum
have affinities for subarctic en-
vironments and often characterize areas with some winter sea
ice.
The correspondence between dinocyst assemblages and
sea surface conditions is not “straightforward” since the cyst
stage only represents a fragmentary picture of the original
dinoflagellate population and because biases related to lat-
eral transport of particles or selective preservation cannot be
discarded (for a review, see
de Vernal
and Marret
[2007] and
Search WWH ::
Custom Search