Jökull - 01.01.2015, Blaðsíða 76

Jónsdóttir et al.
Distance (km)
0 1 2 3 4 5
34.6
34.55
34.5
34.45
34.4
34.35
34.65
8
7
6
5
4
3
20
40
60
80
D
e
p
th
(
m
)
20
40
60
80
D
e
p
th
(
m
)
Temperature (°C)
Salinity (psu)
Distance (km)
0 1 2 3 4 5
Temperature (°C)
Salinity (psu)
9
8
7
6
5
34.6
34.4
34.2
34
33.8
Figure 2. Temperature and salinity profiles in Arnarfjörður, taken in June 2005 (left) and October 2006 (right)
(www.hafro.is). – Hita- og seltumælingar frá Arnarfirði, teknar í júní 2005 (t.v.) og október 2006 (t.h.).
Radiocarbon dating
Radiocarbon dating was performed on four marine bi-
valve shells (Table 1) with an accelerator mass spec-
trometer at INSTAAR, Laboratory for AMS Radio-
carbon Preparation and Research in Boulder, Col-
orado. The global average reservoir correction of
about 400 years was applied to the 14C dates for the
core (Reimer et al., 2013) (∆R = 0).
Foraminifera
Samples for the foraminifera analysis were taken at
ca. every 10 cm, in total 66 samples. The sam-
ples were wet sieved with 63, 106 and 1000µm
sieves and then dried. Foraminifera were picked from
the size fraction 106–1000µm. Although it is of-
ten recommended to count and identify at least 300
foraminifera individuals from each sample, statistical
test on foraminiferal counts, counting in total 200 ver-
sus 300 individuals, on a core from a nearby fjord in-
dicated no significant difference in species % or fau-
nal diversity (Ólafsdóttir, 2004). As the Arnarfjörður
faunal assemblages are also of low diversity >200
benthic foraminifera specimens were picked and iden-
tified to species from each sample.
Statistical analysis
Cluster Analyses (CA) was used to subdivide the fau-
nal record into assemblage zones using the software
Past with squared euclidian, paired group (UPGM)
and constrained clustering. A Principal Component
Analysis (PCA) method was used to better under-
stand the association between foraminiferal assem-
blages and hydrographic and environmental factors.
The PCA was performed on the foraminiferal data
set, with the XLSTAT software. The data set con-
tained only species with an abundance of ≥2% in at
least one sample. The BWT reconstruction, in Arn-
arfjörður, was performed with transfer function ap-
plied to a down-core data set using the C2 program
(Juggings, 2010). The transfer function is based on a
calibration set by Sejrup et al. (2004). The statistical
routine, weighted averaging partial least squares (WA-
PLS) was applied on the training set. In an Arctic
fjord environment E. excavatum f. clavata is a fairly
common species due to its tolerance of high sediment
accumulation rates and fluctuating salinity (Hald et
al., 1994). To avoid the issue of too low temperature
estimates, a square root was applied on the training
set to diminish the weighing of E. excavatum on the
76 JÖKULL No. 65, 2015