356
neously monitor population dynamics and ecosystem processes. The ecosystems in the Ecotron are
not intact, real ecosystems, but carefully constructed, "artificial" ecosystems. Nevertheless,
these artificial systems are modeled on real communities and embody many essential features of
natural communities. The Ecotron has been used to
study relationships between biotic diversity, climate change, and ecosystem processes (e.g.,
Naeem et al. 1994).
Results from growth chamber studies can provide guidance for field work and mathematical
model development. There have been attempts to
combine laboratory studies with concurrent field investigations (e.g., Rawat and Purohit 1991). This is
an encouraging trend, as growth chamber manipulations, whether micro- or mesocosm, are oflimited
usefulness without the context of field studies (Carpenter 1996; Lawton et al. 1993; Verhoef 1996).
Field Manipulations:
General Considerations
The most beneficial studies for understanding ecosystem response to global climate change are largescale, long-term (up to 10 years or more) field manipulations. Such studies provide context for
laboratory work and aid in designing and testing
dynamical models. Methods vary with the type of
ecosystem under study and the manipulation variable. We limit our discussion to whole-ecosystem
climate (primarily warming) and ultraviolet-B
(UV-B) radiation manipulations, and begin with
some basic considerations.
Global warming will be neither uniform nor constant over the globe, and, because of differences in
soil and vegetation, not all systems will respond
similarly to climate change. Site selection criteria
have therefore included anticipated level of local
climate change and likely sensitivity of dominant
vegetation and biogeochemical processes to climate
change (NSFESP 1991). Field manipulations of climate require careful planning of experimental design and monitoring. The size of the experiment
will depend on the system as well as the method;
for example, while experimental plots may be fairly
small in an Arctic tundra site and still encompass a
representative portion of the entire ecosystem (Debevec and Maclean 1993), this may not be the case
in other systems. It has been suggested that plot size
Karin P. Shen and John Harte
should be large enough to encompass the rooting
system of several individuals of the dominant plant
species (NSFESP 1991). Treatments should be
spaced to avoid the treatment effect influencing the
control and should be as identical as possible to the
controls, ideally differing only in the variable to
be manipulated. This may necessitate using a
"dummy" experimental set-up on the control plots,
which does not involve the climate manipulation
but introduces the same unintended effects as the
treatment apparatus.
Climate manipulation experiments alter many
ecosystem characteristics. Key parameters to monitor are those that integrate several aspects of ecosystem function and are especially sensitive to climate change. Because there is significant diurnal
variation in most soil and vegetation processes, and
because fine root and organic matter profiles vary
with depth, it is desirable to monitor soil temperature and moisture at least several times a day and
at more than one depth. Other key soil parameters
include carbon stocks, nitrogen (total and available
inorganic), at least one measure of soil microbial
activity (such as bottle respiration, field CO 2 fluxes,
or actual active microbial counts), and nitrogen
mineralization (NSFESP 1991). Key vegetation parameters include phenology, above- and belowground biomass, leaf area index, stomatal conductance, water use efficiency, photosynthetic rates,
and community composition (presence or absence).
Replication of climate manipulations over a natural landscape-scale climatic gradient, such as a
mountain slope, can be useful (NSFESP 1991). Topographic gradients provide a means of looking at
ecosystem response to climatic variation over
longer time scales than are accessible in manipulation experiments, and have been used as surrogates for field manipUlations (see, for example,
Burke et al. 1995; Lashof 1989; Stanton et al. 1994;
Tate 1992; Townsend et al. 1995; Vitousek and
Matson 1991). Most gradient studies have focused
on soil carbon because of the important potential
role of climate-induced carbon release from ecosystems as a feedback to climate (e.g., Trumbore et
al. 1996). Adaptations to climate along a topographic gradient are characterized by many time
constants, making the substitution of differences
over a topographic gradient for changes expected
from future global warming problematic. Microbes
and insects may adapt quickly, whereas physical
neously monitor population dynamics and ecosystem processes. The ecosystems in the Ecotron are
not intact, real ecosystems, but carefully constructed, "artificial" ecosystems. Nevertheless,
these artificial systems are modeled on real communities and embody many essential features of
natural communities. The Ecotron has been used to
study relationships between biotic diversity, climate change, and ecosystem processes (e.g.,
Naeem et al. 1994).
Results from growth chamber studies can provide guidance for field work and mathematical
model development. There have been attempts to
combine laboratory studies with concurrent field investigations (e.g., Rawat and Purohit 1991). This is
an encouraging trend, as growth chamber manipulations, whether micro- or mesocosm, are oflimited
usefulness without the context of field studies (Carpenter 1996; Lawton et al. 1993; Verhoef 1996).
Field Manipulations:
General Considerations
The most beneficial studies for understanding ecosystem response to global climate change are largescale, long-term (up to 10 years or more) field manipulations. Such studies provide context for
laboratory work and aid in designing and testing
dynamical models. Methods vary with the type of
ecosystem under study and the manipulation variable. We limit our discussion to whole-ecosystem
climate (primarily warming) and ultraviolet-B
(UV-B) radiation manipulations, and begin with
some basic considerations.
Global warming will be neither uniform nor constant over the globe, and, because of differences in
soil and vegetation, not all systems will respond
similarly to climate change. Site selection criteria
have therefore included anticipated level of local
climate change and likely sensitivity of dominant
vegetation and biogeochemical processes to climate
change (NSFESP 1991). Field manipulations of climate require careful planning of experimental design and monitoring. The size of the experiment
will depend on the system as well as the method;
for example, while experimental plots may be fairly
small in an Arctic tundra site and still encompass a
representative portion of the entire ecosystem (Debevec and Maclean 1993), this may not be the case
in other systems. It has been suggested that plot size
Karin P. Shen and John Harte
should be large enough to encompass the rooting
system of several individuals of the dominant plant
species (NSFESP 1991). Treatments should be
spaced to avoid the treatment effect influencing the
control and should be as identical as possible to the
controls, ideally differing only in the variable to
be manipulated. This may necessitate using a
"dummy" experimental set-up on the control plots,
which does not involve the climate manipulation
but introduces the same unintended effects as the
treatment apparatus.
Climate manipulation experiments alter many
ecosystem characteristics. Key parameters to monitor are those that integrate several aspects of ecosystem function and are especially sensitive to climate change. Because there is significant diurnal
variation in most soil and vegetation processes, and
because fine root and organic matter profiles vary
with depth, it is desirable to monitor soil temperature and moisture at least several times a day and
at more than one depth. Other key soil parameters
include carbon stocks, nitrogen (total and available
inorganic), at least one measure of soil microbial
activity (such as bottle respiration, field CO 2 fluxes,
or actual active microbial counts), and nitrogen
mineralization (NSFESP 1991). Key vegetation parameters include phenology, above- and belowground biomass, leaf area index, stomatal conductance, water use efficiency, photosynthetic rates,
and community composition (presence or absence).
Replication of climate manipulations over a natural landscape-scale climatic gradient, such as a
mountain slope, can be useful (NSFESP 1991). Topographic gradients provide a means of looking at
ecosystem response to climatic variation over
longer time scales than are accessible in manipulation experiments, and have been used as surrogates for field manipUlations (see, for example,
Burke et al. 1995; Lashof 1989; Stanton et al. 1994;
Tate 1992; Townsend et al. 1995; Vitousek and
Matson 1991). Most gradient studies have focused
on soil carbon because of the important potential
role of climate-induced carbon release from ecosystems as a feedback to climate (e.g., Trumbore et
al. 1996). Adaptations to climate along a topographic gradient are characterized by many time
constants, making the substitution of differences
over a topographic gradient for changes expected
from future global warming problematic. Microbes
and insects may adapt quickly, whereas physical