70
Hale and Pflaumann
perature. These qualities, combined with the fact
that almost all modem species spend at least a portion of their lives in the uppermost water layers
within the photic zone (Be 1967; Be and Tolderlund
1971; Fairbanks et a1.1982; Oberhanslietal. 1992;
Kemle-von Miicke and Oberhiinsli this volume),
make this group of organisms a practicable tool for
studying past ocean surface conditions. Although
individual species may tolerate a range of temperatures, some of which can be rather wide, the combined occurrence of30 to 40 species or forms in
most sediment samples (except at high latitudes)
provides a wealth of environmental information.
However, evaluating abundance data of
foraminiferal assemblages involves more than a
straightforward compilation of temperature ranges
of the species present. Because the abundance
distribution of an individual species varies within its
absolute temperature range (Mix 1987), accurate
counts of relative abundances in an assemblage
allow us to infer past conditions that would be impossible to estimate with simple presence/absence
data. It is also important to mention, as Kemle-von
Miicke and Oberhansli (this volume) point out, that
many foraminiferal species react sensitively to other
hydrographical factors such as oxygenation, nutrient levels, upwelling conditions, and thermocline
depths. In spite of this complexity, it is assumed that
all species contribute some degree of information
relating to the temperature of the water in which
they live and reproduce.
Temperature is a critical parameter for models
of ancient ocean conditions primarily because different water masses are often distinguished by their
mean temperatures or temperature ranges. In addition, estimates of paleotemperature may prove to
be useful in reconstructing past salinity conditions
(see Wolff et al. this volume), which are also important to ocean modeling efforts. Since the early
work ofUrey (1947), Epstein et al. (1953), and
Emiliani (1955), the relationship between oxygen
isotope measurements of biogenic carbonate and
temperature of shell formation has been well
known. However, the application of isotope measurements as a tool to reconstruct past temperatures
has been hampered by an inability to specify the
isotopic composition of the water in which the carbonate was formed, a critical element in the equation to determine temperature from shell measurements. Turning the problem around, if we had a
reliable, independent method for determining
paleotemperature, it would then be possible to calculate the oxygen isotopic composition ofthe ambient water (OISO) from the isotopic signal preserved in fossil calcite. Recent efforts to calculate
paleosalinity (e.g. Rostek et al. 1993; Wang et al.
1995; Wolffetal. this volume) depend on the OISOw
value because of its observable (albeit regionally
variable) relationship with salinity in the modem
ocean.
Attempts to develop an independent method of
paleotemperature determination from preserved
biogenic material have taken several approaches,
the most promising of which are transfer functions,
modem analog techniques, and an alkenone technique employing unsaturation ratios of long-chain
ketones (U~7 index) that are linearly related to water temperature (Schneider et al. 1995; Sikes and
Keigwin 1994; Rosell-Mele et al. 1995). Both transfer and modem analog techniques rely on comparison of fossil assemblages with large modem coretop
data sets, called calibration or reference data sets.
The reference samples are, in tum, associated with
modem sea-surface temperatures (SST's) through
carefully maintained meteorological data sets (e.g.
Levitus 1982). Both of these methods are also
based on several common assumptions, the most
significant of which are: constant ecology of the
species through the time in question, a systematic
relationship between faunal content at the sediment
surface and SST, and cross-hemisphere similarity
of ecological responses by the organisms being used
for the analysis (Pflaum ann et a!. 1996; for a more
detailed discussion see Imbrie and Kipp 1971, and
Prell 1985). The transfer technique, based on
grouping species by Q-mode factor analysis, has
been widely employed and is well known (e.g. Kipp
1976; Molfino et al. 1982; McIntyre et a!. 1989),
the classic application being that of the CLIMAP
group to obtain a global database of SST values
from the last glacial maximum (LGM) (Climate:
Long-Range Investigations, Mapping, and Prediction (CLIMAP) 1981). However, this technique
suffers from a drawback in the form of data gaps
or unreliable data in cases where the ancient faunal
composition differs significantly from modem fau-
Hale and Pflaumann
perature. These qualities, combined with the fact
that almost all modem species spend at least a portion of their lives in the uppermost water layers
within the photic zone (Be 1967; Be and Tolderlund
1971; Fairbanks et a1.1982; Oberhanslietal. 1992;
Kemle-von Miicke and Oberhiinsli this volume),
make this group of organisms a practicable tool for
studying past ocean surface conditions. Although
individual species may tolerate a range of temperatures, some of which can be rather wide, the combined occurrence of30 to 40 species or forms in
most sediment samples (except at high latitudes)
provides a wealth of environmental information.
However, evaluating abundance data of
foraminiferal assemblages involves more than a
straightforward compilation of temperature ranges
of the species present. Because the abundance
distribution of an individual species varies within its
absolute temperature range (Mix 1987), accurate
counts of relative abundances in an assemblage
allow us to infer past conditions that would be impossible to estimate with simple presence/absence
data. It is also important to mention, as Kemle-von
Miicke and Oberhansli (this volume) point out, that
many foraminiferal species react sensitively to other
hydrographical factors such as oxygenation, nutrient levels, upwelling conditions, and thermocline
depths. In spite of this complexity, it is assumed that
all species contribute some degree of information
relating to the temperature of the water in which
they live and reproduce.
Temperature is a critical parameter for models
of ancient ocean conditions primarily because different water masses are often distinguished by their
mean temperatures or temperature ranges. In addition, estimates of paleotemperature may prove to
be useful in reconstructing past salinity conditions
(see Wolff et al. this volume), which are also important to ocean modeling efforts. Since the early
work ofUrey (1947), Epstein et al. (1953), and
Emiliani (1955), the relationship between oxygen
isotope measurements of biogenic carbonate and
temperature of shell formation has been well
known. However, the application of isotope measurements as a tool to reconstruct past temperatures
has been hampered by an inability to specify the
isotopic composition of the water in which the carbonate was formed, a critical element in the equation to determine temperature from shell measurements. Turning the problem around, if we had a
reliable, independent method for determining
paleotemperature, it would then be possible to calculate the oxygen isotopic composition ofthe ambient water (OISO) from the isotopic signal preserved in fossil calcite. Recent efforts to calculate
paleosalinity (e.g. Rostek et al. 1993; Wang et al.
1995; Wolffetal. this volume) depend on the OISOw
value because of its observable (albeit regionally
variable) relationship with salinity in the modem
ocean.
Attempts to develop an independent method of
paleotemperature determination from preserved
biogenic material have taken several approaches,
the most promising of which are transfer functions,
modem analog techniques, and an alkenone technique employing unsaturation ratios of long-chain
ketones (U~7 index) that are linearly related to water temperature (Schneider et al. 1995; Sikes and
Keigwin 1994; Rosell-Mele et al. 1995). Both transfer and modem analog techniques rely on comparison of fossil assemblages with large modem coretop
data sets, called calibration or reference data sets.
The reference samples are, in tum, associated with
modem sea-surface temperatures (SST's) through
carefully maintained meteorological data sets (e.g.
Levitus 1982). Both of these methods are also
based on several common assumptions, the most
significant of which are: constant ecology of the
species through the time in question, a systematic
relationship between faunal content at the sediment
surface and SST, and cross-hemisphere similarity
of ecological responses by the organisms being used
for the analysis (Pflaum ann et a!. 1996; for a more
detailed discussion see Imbrie and Kipp 1971, and
Prell 1985). The transfer technique, based on
grouping species by Q-mode factor analysis, has
been widely employed and is well known (e.g. Kipp
1976; Molfino et al. 1982; McIntyre et a!. 1989),
the classic application being that of the CLIMAP
group to obtain a global database of SST values
from the last glacial maximum (LGM) (Climate:
Long-Range Investigations, Mapping, and Prediction (CLIMAP) 1981). However, this technique
suffers from a drawback in the form of data gaps
or unreliable data in cases where the ancient faunal
composition differs significantly from modem fau-