76
Hale and Ptlaumann
variations in the warm-season curves. Cores in the
central and eastern equatorial Atlantic do not show
this pattern (McIntyre et al. 1989; Mix and Morey
1996). For the cold-season temperatures, this excursion is even more pronounced in both our cores,
as are the amplitudes of temperature variation in
general. The average difference between cold- and
warm-season values in the tropical cores is about
2°, with the greater differences consistently occurring during the cooler intervals.
Temperature variations in the Rio Grande Rise
cores are characterized by higher amplitudes
throughout most oftheir lengths. The warm temperatures in GeoB 1312 range approximately from
19° to 24°, and those in GeoB 1309 from 19°to23°
C. Intervals with the highest temperatures occasionally correspond to interglacial stages (e.g.
stages 5, 7 and II), but the relationship is so vague
that it may be coincidental. Where warmer temperatures do occur during interglacials, it is commonly more pronounced in the early part of the
stage. Correlation coefficients obtained in comparisons of SIMMAX estimated temperature with the
8 18 0 curves for these cores were only around 0.25.
The absence of a clear pattern of temperature
variation through time may be real, or it may be
significantly influenced by low sedimentation rates
in this region (-I cm/ky) combined with bioturbation
effects, which tend to mix the sediments and blend
foraminiferal assemblages from significantly different environments through time. This blurring effect
is evident in the oxygen isotope curves, which lack
the defmition and detail of those in the tropical cores,
even though the sampling resolution is twice as
dense (every 5 cm in the Rio Grande Rise compared to every 10 cm in the tropical cores). The
Stage boundaries are much more difficult to pinpoint in this area, although the general shapes of
the isotope curves are easily recognized. This may
also help to explain why the temperature curves
presented for these two cores, which are geographically very close, have quite different shapes in their
detail. Between the warm and cold season curves
there is an average difference of about 4°C,
with the greater differences appearing during the
warmer intervals.
The uppermost sample in each ofthe four cores
was taken at 3 cm below the sediment/water interface. The cold and warm temperatures calculated for these core-top samples agree very well
with the modem temperature range data presented
in Fig. 2. Unfortunately, we have no reliable record
or other tool to test the accuracy of the calculated
temperatures into the past. However, we can apply other statistical or "parastatistical" techniques
to analyze the mass of data that has been accumulated in order to investigate more thoroughly
how this method of temperature calculation functions, and which elements of the data set have
greater influences on the results.
We can begin by looking at individual species
abundance patterns throughout the samples in all
cores. Intuitively, we would assume that certain
species would be more plentiful in samples with
warm temperature estimations and that others
would appear with greater numbers in samples
where cooler values were calculated, depending on
the modem-day temperature preferences of the
various species. By looking at scatter plots of relative species abundance to estimated temperature
- using all the samples from the four cores as a
single database - we can qualitatively assess which
species contribute significantly to cooler or warmer
estimates. These plots are presented in Fig. 4 and
illustrate the presence of 21 species in all ofthe
samples according to the estimated temperatures
(distance-weighted, warm season) for each
sample. Species not illustrated in Fig. 4 (a-c)
(Sphaeroidinella dehiscens, Globigerinella
digitata, Neogloboquadrina pachyderma (sinistral form), and Globigerina quinqueloba) were
only present in abundances near or below the limits of precision for the counting procedure (averaging less than 1%).
Three of the 21 species graphed did not show
a visually significant trend of preference for higher
or lower temperatures in our opinion. These are the
last three species in the figure, Globigerinoides
conglobatus, Globigerina rubescens, and
Orbulina universa. The remaining 18 species plotted show a discernible visual pattern of preference
for either cooler (plots on the right side) or warmer
(plots on the left side) temperatures. Note that the
scales of the y-axes, abundance values, are not
constant among the various species. Two obvious
groups of points are recognizable in each plot: one
Hale and Ptlaumann
variations in the warm-season curves. Cores in the
central and eastern equatorial Atlantic do not show
this pattern (McIntyre et al. 1989; Mix and Morey
1996). For the cold-season temperatures, this excursion is even more pronounced in both our cores,
as are the amplitudes of temperature variation in
general. The average difference between cold- and
warm-season values in the tropical cores is about
2°, with the greater differences consistently occurring during the cooler intervals.
Temperature variations in the Rio Grande Rise
cores are characterized by higher amplitudes
throughout most oftheir lengths. The warm temperatures in GeoB 1312 range approximately from
19° to 24°, and those in GeoB 1309 from 19°to23°
C. Intervals with the highest temperatures occasionally correspond to interglacial stages (e.g.
stages 5, 7 and II), but the relationship is so vague
that it may be coincidental. Where warmer temperatures do occur during interglacials, it is commonly more pronounced in the early part of the
stage. Correlation coefficients obtained in comparisons of SIMMAX estimated temperature with the
8 18 0 curves for these cores were only around 0.25.
The absence of a clear pattern of temperature
variation through time may be real, or it may be
significantly influenced by low sedimentation rates
in this region (-I cm/ky) combined with bioturbation
effects, which tend to mix the sediments and blend
foraminiferal assemblages from significantly different environments through time. This blurring effect
is evident in the oxygen isotope curves, which lack
the defmition and detail of those in the tropical cores,
even though the sampling resolution is twice as
dense (every 5 cm in the Rio Grande Rise compared to every 10 cm in the tropical cores). The
Stage boundaries are much more difficult to pinpoint in this area, although the general shapes of
the isotope curves are easily recognized. This may
also help to explain why the temperature curves
presented for these two cores, which are geographically very close, have quite different shapes in their
detail. Between the warm and cold season curves
there is an average difference of about 4°C,
with the greater differences appearing during the
warmer intervals.
The uppermost sample in each ofthe four cores
was taken at 3 cm below the sediment/water interface. The cold and warm temperatures calculated for these core-top samples agree very well
with the modem temperature range data presented
in Fig. 2. Unfortunately, we have no reliable record
or other tool to test the accuracy of the calculated
temperatures into the past. However, we can apply other statistical or "parastatistical" techniques
to analyze the mass of data that has been accumulated in order to investigate more thoroughly
how this method of temperature calculation functions, and which elements of the data set have
greater influences on the results.
We can begin by looking at individual species
abundance patterns throughout the samples in all
cores. Intuitively, we would assume that certain
species would be more plentiful in samples with
warm temperature estimations and that others
would appear with greater numbers in samples
where cooler values were calculated, depending on
the modem-day temperature preferences of the
various species. By looking at scatter plots of relative species abundance to estimated temperature
- using all the samples from the four cores as a
single database - we can qualitatively assess which
species contribute significantly to cooler or warmer
estimates. These plots are presented in Fig. 4 and
illustrate the presence of 21 species in all ofthe
samples according to the estimated temperatures
(distance-weighted, warm season) for each
sample. Species not illustrated in Fig. 4 (a-c)
(Sphaeroidinella dehiscens, Globigerinella
digitata, Neogloboquadrina pachyderma (sinistral form), and Globigerina quinqueloba) were
only present in abundances near or below the limits of precision for the counting procedure (averaging less than 1%).
Three of the 21 species graphed did not show
a visually significant trend of preference for higher
or lower temperatures in our opinion. These are the
last three species in the figure, Globigerinoides
conglobatus, Globigerina rubescens, and
Orbulina universa. The remaining 18 species plotted show a discernible visual pattern of preference
for either cooler (plots on the right side) or warmer
(plots on the left side) temperatures. Note that the
scales of the y-axes, abundance values, are not
constant among the various species. Two obvious
groups of points are recognizable in each plot: one