6
Weferetal.
there are a number of other potential correction
factors as well, including preferential growth during certain seasons and at certain depths, and dissolution effects (Berger and Gardner 1975; Wefer
and Berger 1991; Wu and Berger 1991). To express
this in shorthand, we write:
OXe = f(T, Ice, Sel, Vit, Dis)
(3),
where T, Ice, Sel, Vit, and Dis, stand for the effects of temperature, ice mass and composition,
selective growth in time and space, vital factors,
and postdepositional dissolution, respectively.
These complications should be kept in mind,
especially considering the fact that oxygen isotopes
are the most frequently used proxies. In practice,
because most proxies are determined on biogenic
calcite or represent the response of a biological
association to the target parameter, natural variability has to be taken into account. It is therefore good
practice to estimate the target parameter using
more than one proxy. Only such a multiproxy approach will lead to a sense of the confidence that
can be placed in the reconstruction. The quality of
the correlation within the calibration set is insufficient evidence for the quality of the transform back
in time because the mixture of parameters influencing a proxy changes through time. Nevertheless, confidence limits derived from calibration sets
are commonly offered as valid error estimates.
Such limits are optimistic.
Types of Proxies
Proxies can be classified according to the type of
sediment property they describe (physical, chemical, isotopic, biological remains) or according to the
target they stand in for (Appendix I). The first
approach equates proxies (correctly) with sediment
properties (e.g. Mg/Ca ratio). The word "proxy"
does not appear in this context. The second approach emphasizes ocean properties, which is what
is of interest to paleoceanographers. Thus, we
speak of proxies for water properties (e.g. temperature, sea-level, nutrient contents, thermocline-thickness, or productivity). In this case, proxies are classified by the targets they represent.
Many sediment properties are directly related
to abundances of particles within the sediment.
Such properties can be expressed as fluxes if a
detailed time scale is known. The name "flux proxies" is sometimes applied after formal conversion
of abundance data (percentages) to flux. It should
be remembered that flux is not being proxied by
such conversions. Instead, flux is being assumed,
rather than proxied. When flux is pro xi ed, it is done
by taking the influx of one component as constant,
and converting the variation of abundance of that
component into rates of accumulation for the entire sediment. The element Al has been used as a
flux proxy in this fashion (Arrhenius 1952), as has
230Th (Anderson et al. 1983; Schwarz et al. 1996).
Confidence in this type of proxying is greatly increased if it can be shown that the assumption of
constant flux for the proxy variable is valid over
some significant portion of the core studied.
In some circumstances an important sediment
property (rather than a property relating to the
ocean) is being proxied by another property, for
example when carbonate is estimated from physical properties, or organic matter from the presence
of radioactive uranium. The proper usage is to put
the intended target in front of the word proxy, to
avoid confusion (hence, carbonate proxy, and C",g
proxy).
In what follows, we shall discuss sediment properties according to the target parameters which
they proxy. We begin with temperature, the single
most important parameter for describing the conditions of past oceans and crucial for climate modelling. We also touch on salinity, bottom water oxygenation and flow, productivity, partial pressure of
CO 2 , and wind.
Temperature Estimates from Microfossil
Assemblages
Most workers would agree that the temperature
distribution at the surface ofthe ocean (SST) is the
single most useful target for reconstruction in
paleoceanography. It provides the most important
clues for modelling of climate and ocean circulation, and for explaining biogeography and productivity patterns. A number of proxies are available
Weferetal.
there are a number of other potential correction
factors as well, including preferential growth during certain seasons and at certain depths, and dissolution effects (Berger and Gardner 1975; Wefer
and Berger 1991; Wu and Berger 1991). To express
this in shorthand, we write:
OXe = f(T, Ice, Sel, Vit, Dis)
(3),
where T, Ice, Sel, Vit, and Dis, stand for the effects of temperature, ice mass and composition,
selective growth in time and space, vital factors,
and postdepositional dissolution, respectively.
These complications should be kept in mind,
especially considering the fact that oxygen isotopes
are the most frequently used proxies. In practice,
because most proxies are determined on biogenic
calcite or represent the response of a biological
association to the target parameter, natural variability has to be taken into account. It is therefore good
practice to estimate the target parameter using
more than one proxy. Only such a multiproxy approach will lead to a sense of the confidence that
can be placed in the reconstruction. The quality of
the correlation within the calibration set is insufficient evidence for the quality of the transform back
in time because the mixture of parameters influencing a proxy changes through time. Nevertheless, confidence limits derived from calibration sets
are commonly offered as valid error estimates.
Such limits are optimistic.
Types of Proxies
Proxies can be classified according to the type of
sediment property they describe (physical, chemical, isotopic, biological remains) or according to the
target they stand in for (Appendix I). The first
approach equates proxies (correctly) with sediment
properties (e.g. Mg/Ca ratio). The word "proxy"
does not appear in this context. The second approach emphasizes ocean properties, which is what
is of interest to paleoceanographers. Thus, we
speak of proxies for water properties (e.g. temperature, sea-level, nutrient contents, thermocline-thickness, or productivity). In this case, proxies are classified by the targets they represent.
Many sediment properties are directly related
to abundances of particles within the sediment.
Such properties can be expressed as fluxes if a
detailed time scale is known. The name "flux proxies" is sometimes applied after formal conversion
of abundance data (percentages) to flux. It should
be remembered that flux is not being proxied by
such conversions. Instead, flux is being assumed,
rather than proxied. When flux is pro xi ed, it is done
by taking the influx of one component as constant,
and converting the variation of abundance of that
component into rates of accumulation for the entire sediment. The element Al has been used as a
flux proxy in this fashion (Arrhenius 1952), as has
230Th (Anderson et al. 1983; Schwarz et al. 1996).
Confidence in this type of proxying is greatly increased if it can be shown that the assumption of
constant flux for the proxy variable is valid over
some significant portion of the core studied.
In some circumstances an important sediment
property (rather than a property relating to the
ocean) is being proxied by another property, for
example when carbonate is estimated from physical properties, or organic matter from the presence
of radioactive uranium. The proper usage is to put
the intended target in front of the word proxy, to
avoid confusion (hence, carbonate proxy, and C",g
proxy).
In what follows, we shall discuss sediment properties according to the target parameters which
they proxy. We begin with temperature, the single
most important parameter for describing the conditions of past oceans and crucial for climate modelling. We also touch on salinity, bottom water oxygenation and flow, productivity, partial pressure of
CO 2 , and wind.
Temperature Estimates from Microfossil
Assemblages
Most workers would agree that the temperature
distribution at the surface ofthe ocean (SST) is the
single most useful target for reconstruction in
paleoceanography. It provides the most important
clues for modelling of climate and ocean circulation, and for explaining biogeography and productivity patterns. A number of proxies are available