198
Fischer et al.
Fairbanks 1992». Indeed, the stable oxygen record
of G. ruber shows a good correspondence to the
measured sea surface temperatures depicted in
Fig. Id. Except in summer 1988, the maximum
values slightly exceeded the measured SSTs in
absolute numbers; SST minimum values were also
slightly higher. Compared to the measured SSTs,
an offset in the order of 1-2 months in the timing
of the peak temperatures deduced from G. ruber
can be observed (e.g. in 1989 and 1990).
L. inflatayields lower temperatures (up to SoC
based onjuveniles, September, 1988) compared to
G. ruber(Fig. Ic). According to Fabry and Deuser
(1992) and Kalberer et al. (1993), this mesopelagic
species precipitates aragonite in a water depth of
approximately 50 m, although it migrates from 50
to 350 m on a die I cycle (Wormuth 1981). Thus, it
should reflect lower temperatures than G. ruber.
As a rule, temperatures of L. injlata and G. ruber
converge in winter-spring and diverge in late summer and fall. Merely, one exception can be observed during summer and fall of 1991 when L.
inflata and G. ruber revealed almost similar temperatures, even if a factor of uncertainty regarding the temperature estimations had been considered (vital effect of G. ruber, calcite-aragonite
offset). Generally, maximum values in temperature
during fall and summer recorded by L. injlatawere
2_3 0 lower compared to the measured values, the
minimum values occurring in winter-spring, however, differed only slightly in absolute numbers
(Fig. lc, d). The offset in the timing of temperature maxima and minima in comparison to the
measured SSTs appears to be smaller compared
the reconstructions based on G. ruber, probably due
to faster sinking rates of pteropods. This finding
is supported by a long-term study performed by
Jasper and Deuser (1993) suggesting rapid growth
and rapid sinking of shells.
The reconstructed temperature patterns show
variable seasonal changes from year to year. Highest seasonal temperature variability was obtained
from both species in 1988, lowest seasonality in
1991 (Fig. Ic). This corraborates SST measurements (Fig. Id). A plot comparing the calculated
temperatures produced by both species to the
measured SST is shown in Fig. 2; the relationship
between both departs from the 1:1-line. We obtained significant correlations between measured
SSTs and calculated temperatures for both species,
with G. ruber yielding about 1.5°C higher values
in the lower temperature range and up to 2°C
higher values in the upper range (applying -0.35%0
as vital effect), compared to L. injlata. The relationship between measured SSTs and calculated
water temperatures for G. ruber becomes stronger
(R2=0.50, N=36), when considering a time shift
between foraminiferal production in the surface
water and collection in the sediment trap of about
one month (R2=0.65). This correction has no significant effect on the correlation coefficient obtained from L. injlata which obviously sank much
faster than G. ruber.
Water temperatures estimated from G. ruber
(white), a species assumed to live in the upper few
tens of meters (Deuser and Ross 1989; Ravelo and
Fairbanks 1992), exceeded the measured SSTs, in
particular in the lower temperature range «22°C,
Fig. 2). A higher vital effect of 0.5%0 (Fairbanks
et al. 1982) would not increase the correlation coefficient significantly, but would lead to a better
correlation between measured and estimated temperatures, in particular in the intermediate temperature range (around 22°C). However, the slope of
the regression line is still too low, a fact that results
in an overestimation in the lower temperature range
and an underestimation in the higher range.
For better comparison of the temperature curves
shown in Fig. 1, we calculated the seasonal means
(Fig. 3), showing highest seasonality in all curves
in 1989 and lowest seasonality in 1991. Increased
stratification during summer and fall can be inferred
from the diverging temperatures calculated from
L. injlata and G. ruber living in different depths
whereas increased mixing and cooling during winter and spring (coastal upwelling and advection of
cold waters) is documented by converging temperatures. The closest correspondence in absolute
numbers was observed in 1991 when summer-fall
temperatures also were the lowest which is indicative for a more continuous upwelling and stronger
mixing of surface and subsurface water masses.
Lowest annual mean temperatures were also
obtained in 1991, considering both the measured
values (21.6°C) as well as the reconstructed SSTs
from G. ruber (21.8°C) and L. injlata (20.8°C, see
Fischer et al.
Fairbanks 1992». Indeed, the stable oxygen record
of G. ruber shows a good correspondence to the
measured sea surface temperatures depicted in
Fig. Id. Except in summer 1988, the maximum
values slightly exceeded the measured SSTs in
absolute numbers; SST minimum values were also
slightly higher. Compared to the measured SSTs,
an offset in the order of 1-2 months in the timing
of the peak temperatures deduced from G. ruber
can be observed (e.g. in 1989 and 1990).
L. inflatayields lower temperatures (up to SoC
based onjuveniles, September, 1988) compared to
G. ruber(Fig. Ic). According to Fabry and Deuser
(1992) and Kalberer et al. (1993), this mesopelagic
species precipitates aragonite in a water depth of
approximately 50 m, although it migrates from 50
to 350 m on a die I cycle (Wormuth 1981). Thus, it
should reflect lower temperatures than G. ruber.
As a rule, temperatures of L. injlata and G. ruber
converge in winter-spring and diverge in late summer and fall. Merely, one exception can be observed during summer and fall of 1991 when L.
inflata and G. ruber revealed almost similar temperatures, even if a factor of uncertainty regarding the temperature estimations had been considered (vital effect of G. ruber, calcite-aragonite
offset). Generally, maximum values in temperature
during fall and summer recorded by L. injlatawere
2_3 0 lower compared to the measured values, the
minimum values occurring in winter-spring, however, differed only slightly in absolute numbers
(Fig. lc, d). The offset in the timing of temperature maxima and minima in comparison to the
measured SSTs appears to be smaller compared
the reconstructions based on G. ruber, probably due
to faster sinking rates of pteropods. This finding
is supported by a long-term study performed by
Jasper and Deuser (1993) suggesting rapid growth
and rapid sinking of shells.
The reconstructed temperature patterns show
variable seasonal changes from year to year. Highest seasonal temperature variability was obtained
from both species in 1988, lowest seasonality in
1991 (Fig. Ic). This corraborates SST measurements (Fig. Id). A plot comparing the calculated
temperatures produced by both species to the
measured SST is shown in Fig. 2; the relationship
between both departs from the 1:1-line. We obtained significant correlations between measured
SSTs and calculated temperatures for both species,
with G. ruber yielding about 1.5°C higher values
in the lower temperature range and up to 2°C
higher values in the upper range (applying -0.35%0
as vital effect), compared to L. injlata. The relationship between measured SSTs and calculated
water temperatures for G. ruber becomes stronger
(R2=0.50, N=36), when considering a time shift
between foraminiferal production in the surface
water and collection in the sediment trap of about
one month (R2=0.65). This correction has no significant effect on the correlation coefficient obtained from L. injlata which obviously sank much
faster than G. ruber.
Water temperatures estimated from G. ruber
(white), a species assumed to live in the upper few
tens of meters (Deuser and Ross 1989; Ravelo and
Fairbanks 1992), exceeded the measured SSTs, in
particular in the lower temperature range «22°C,
Fig. 2). A higher vital effect of 0.5%0 (Fairbanks
et al. 1982) would not increase the correlation coefficient significantly, but would lead to a better
correlation between measured and estimated temperatures, in particular in the intermediate temperature range (around 22°C). However, the slope of
the regression line is still too low, a fact that results
in an overestimation in the lower temperature range
and an underestimation in the higher range.
For better comparison of the temperature curves
shown in Fig. 1, we calculated the seasonal means
(Fig. 3), showing highest seasonality in all curves
in 1989 and lowest seasonality in 1991. Increased
stratification during summer and fall can be inferred
from the diverging temperatures calculated from
L. injlata and G. ruber living in different depths
whereas increased mixing and cooling during winter and spring (coastal upwelling and advection of
cold waters) is documented by converging temperatures. The closest correspondence in absolute
numbers was observed in 1991 when summer-fall
temperatures also were the lowest which is indicative for a more continuous upwelling and stronger
mixing of surface and subsurface water masses.
Lowest annual mean temperatures were also
obtained in 1991, considering both the measured
values (21.6°C) as well as the reconstructed SSTs
from G. ruber (21.8°C) and L. injlata (20.8°C, see