280
tions of control factors and response variables are
likely to be sampled. Second, some substantial statistical power is gained; if all the assumptions of
regression analysis are met, for instance, each sample (or composite) adds one degree of freedom.
Third, the spatial dimension of landscapes or
regions can be explicitly addressed through autocorrelation analysis. Fourth, patterns and processes
that are not apparent to the investigator a priori may
emerge from the autocorrelation analysis (e.g.,
Robertson et al. 1997). Finally, many relationships
among the variables can be explored through multivariate analysis. These techniques are used much
less commonly than the discrete approaches described above. The most likely explanations are the
limitation imposed by the number of samples that
need to be taken to conduct such analyses, the fact
that many of the most sophisticated multivariate
procedures do not produce predictive equations for
application, and finally, a lesser familiarity with
these geostatistical and multivariate techniques
among most biogeochemists than the more traditional, agronomically based statistics such as analysis of variance.
Spatially Explicit Analyses
Field Analyses
Many of the key processes responsible for environmental degradation involve the movement of elements or compounds within and among components of landscapes and regions. Acid deposition,
eutrophication, and pollutant transport are examples of important environmental problems that involve transport and concentration of substances,
particularly in wetlands or aquatic areas. Identifying source landscape units and quantifying the rate
of transport is often a major goal.
There are three basic approaches to quantifying
source-sink relationships in landscapes or regions:
simple measurements of pools and inference, use
of natural abundance of isotopes to characterize
sources and sinks, and use of a tracer to track
movement.
In the first approach, investigators utilize an understanding of the hydrologic cycle to estimate nutrient movement through landscapes. In the classic
watershed approach (Likens et al. 1977), detailed
Ingrid C. Burke
estimates are made of ecosystem-level biogeochemical budgets (inputs and internal cycling), and
are combined with estimates of stream/river nutrient concentrations and streamflow to estimate total
watershed hydrologic export of elements. This classic approach has unquestionably yielded our most
important knowledge on the basic movement of
elements and particulate matter through terrestrial
systems to aquatic. In addition, it has contributed
crucial information on how humans alter the
landscape- to regional-scale fluxes of nutrients
through direct disturbance (Likens et al. 1977; Bormann et al. 1974; Vitousek and Reiners 1975; Gorham et al. 1979), through altering atmospheric input (Likens et al. 1996), and landscape- (Peterjohn
and Correl 1984; Hanson et al. 1994) to regionaland continental-scale impacts of humans on river
export of nutrients to oceans (Peierls et al. 1991;
Cole et al. 1993; Howarth et al. 1996). Multiple
element analyses provide particularly powerful
tools for exploring changes through time, since
some of the most significant biogeochemical alterations may be reflected more in such relationships
as cation/anion ratios than in single element trends
(Likens et al. 1996).
A second approach is the use of stable isotopes
to estimate sources, sinks, and net fluxes among
compartments of the landscape or region. Elements
in different pools exist with different "isotopic signatures" (0 isotope), which are characteristically
determined by (1) the signature of the source pools
and (2) the fractionation, or preferential processing
of lighter isotopes, for a particular process. The basic approach is to identify the isotopic signatures of
potential sources for the sink pool of interest, as
well as the signature of this sink pool, and then to
use mixing models to calculate the fractional contribution of each source (e.g., Shearer and Kohl
1993). The approach for ecosystem ecologists has
been reviewed in detail (e.g., Peterson and Fry
1987; Lajtha and Michener 1994; Shearer and Kohl
1993), as it is used at scales from organismal to
global, but has particular utility for scales such as
landscape to regional, in which large-scale processes are very difficult to quantify. Geochemists
have long used natural abundance patterns of isotopes to distinguish among important sources such
as weathering, atmosphere, and pollution (e.g.,
Wadleigh et al. 1985; Abert 1995; Finley et al.
1995). One of the first applications by ecosystem
tions of control factors and response variables are
likely to be sampled. Second, some substantial statistical power is gained; if all the assumptions of
regression analysis are met, for instance, each sample (or composite) adds one degree of freedom.
Third, the spatial dimension of landscapes or
regions can be explicitly addressed through autocorrelation analysis. Fourth, patterns and processes
that are not apparent to the investigator a priori may
emerge from the autocorrelation analysis (e.g.,
Robertson et al. 1997). Finally, many relationships
among the variables can be explored through multivariate analysis. These techniques are used much
less commonly than the discrete approaches described above. The most likely explanations are the
limitation imposed by the number of samples that
need to be taken to conduct such analyses, the fact
that many of the most sophisticated multivariate
procedures do not produce predictive equations for
application, and finally, a lesser familiarity with
these geostatistical and multivariate techniques
among most biogeochemists than the more traditional, agronomically based statistics such as analysis of variance.
Spatially Explicit Analyses
Field Analyses
Many of the key processes responsible for environmental degradation involve the movement of elements or compounds within and among components of landscapes and regions. Acid deposition,
eutrophication, and pollutant transport are examples of important environmental problems that involve transport and concentration of substances,
particularly in wetlands or aquatic areas. Identifying source landscape units and quantifying the rate
of transport is often a major goal.
There are three basic approaches to quantifying
source-sink relationships in landscapes or regions:
simple measurements of pools and inference, use
of natural abundance of isotopes to characterize
sources and sinks, and use of a tracer to track
movement.
In the first approach, investigators utilize an understanding of the hydrologic cycle to estimate nutrient movement through landscapes. In the classic
watershed approach (Likens et al. 1977), detailed
Ingrid C. Burke
estimates are made of ecosystem-level biogeochemical budgets (inputs and internal cycling), and
are combined with estimates of stream/river nutrient concentrations and streamflow to estimate total
watershed hydrologic export of elements. This classic approach has unquestionably yielded our most
important knowledge on the basic movement of
elements and particulate matter through terrestrial
systems to aquatic. In addition, it has contributed
crucial information on how humans alter the
landscape- to regional-scale fluxes of nutrients
through direct disturbance (Likens et al. 1977; Bormann et al. 1974; Vitousek and Reiners 1975; Gorham et al. 1979), through altering atmospheric input (Likens et al. 1996), and landscape- (Peterjohn
and Correl 1984; Hanson et al. 1994) to regionaland continental-scale impacts of humans on river
export of nutrients to oceans (Peierls et al. 1991;
Cole et al. 1993; Howarth et al. 1996). Multiple
element analyses provide particularly powerful
tools for exploring changes through time, since
some of the most significant biogeochemical alterations may be reflected more in such relationships
as cation/anion ratios than in single element trends
(Likens et al. 1996).
A second approach is the use of stable isotopes
to estimate sources, sinks, and net fluxes among
compartments of the landscape or region. Elements
in different pools exist with different "isotopic signatures" (0 isotope), which are characteristically
determined by (1) the signature of the source pools
and (2) the fractionation, or preferential processing
of lighter isotopes, for a particular process. The basic approach is to identify the isotopic signatures of
potential sources for the sink pool of interest, as
well as the signature of this sink pool, and then to
use mixing models to calculate the fractional contribution of each source (e.g., Shearer and Kohl
1993). The approach for ecosystem ecologists has
been reviewed in detail (e.g., Peterson and Fry
1987; Lajtha and Michener 1994; Shearer and Kohl
1993), as it is used at scales from organismal to
global, but has particular utility for scales such as
landscape to regional, in which large-scale processes are very difficult to quantify. Geochemists
have long used natural abundance patterns of isotopes to distinguish among important sources such
as weathering, atmosphere, and pollution (e.g.,
Wadleigh et al. 1985; Abert 1995; Finley et al.
1995). One of the first applications by ecosystem