258
of < 1 kg N ha - 1 yr - 1 (Kapustka and Dubois 1987;
Grant and Binkley 1987; Silvester 1989; Binkley
1996).
These studies suggest that a controlled, artificial
ecosystem with known boundaries can yield important information about nutrient cycling processes that cannot be determined from the largerscale whole-ecosystem budget approach where
spatial variability is high and many processes cannot be measured at a spatially explicit scale. However, these experiments are expensive, are rarely
replicated (San Dimas was not replicated by species), and take several decades before the first
meaningful analyses can be made. The artificial nature of fill soil may have a large effect on process
rates, making extrapolations to undisturbed ecosystems difficult. The small scale of the mesocosms
also pose some limitations on interpretation, just as
plot-scale studies cannot always be representative
of ecosystem-level dynamics. Although the measured output from these mesocosms quite accurately represent the true outputs, only a narrow portion of an ecosystem is measured, as deep soil,
groundwater, and riparian dynamics are not
measured.
Nitrogen-I5 Studies
at the Ecosystem Scale
Because the watershed budget approach cannot
trace internal movement or storage of ions, various
authors through the years have used isotopes to follow the fate of ecologically relevant elements. The
use of phosphorus-32 e 2 p) or other radioactive
tracers in large-scale experiments in natural terrestrial systems is not common (but see Caldwell et
al. 1985), but 15N has been added to natural ecosystems in a large number of studies either as enriched fertilizer (Mead and Pritchett 1975; Broadbent and Carleton 1978; Nadelhoffer et al. 1995;
Koopmans et al. 1996; Jordan et al. 1997; Nadelhoffer et al. 1998b; Tietema et al. 1998, to name a
few) or at background, throughfall input concentrations (White and Howes 1994; Starns et al. 1991;
Koopmans et al. 1996; Buchmann et al. 1995, 1996;
Seely and Lajtha 1997; Nadelhoffer et al. 1998a).
Unfortunately, less than a handful have used 15N at
background throughfall input levels and have
Kate Lajtha
traced movement over long time periods and in a
number of ecosystem pools, including output pools
such as deep lysimeters.
Buchmann et al. (1995, 1996) used 15N as a
tracer at background levels in the Fichtelgebirge,
Germany. Because there was a 20% decrease in the
recovery of both 15N03 and 15NH4 tracer after 15
months, they estimated that this reflected the
amount of loss from the system via leaching. The
highest retention of 15N was in organic soil horizons, followed by mineral soils, uptake by understory herbs, and uptake by overstory trees. Quite
similar patterns have since been seen in many other
studies with 15N tracer additions, where inputoutput budget studies alone could not have measured the retention rates in these different pools.
Similarly Koopmans et al. (1996), using quite low
as well as fertilizer levels of 15N addition in Nsaturated sites in the Netherlands, budgeted the
added N into vegetation uptake, soil retention, and
leaching losses, and found more uptake of label by
trees in a Douglas fir site than in a Scots pine site.
Seely and Lajtha (1997) added 15N to throughfall
to trace seasonal differences in N retention in a
coastal watershed in the United States and to examine soil texture effects on N retention, and found
patterns similar to those of Buchmann et al. (1995),
but with much higher leaching losses.
Other studies using background levels of 15N for
other purposes, such as estimating nitrification
rates, have not measured many internal pools, but
have measured leaching losses oflabel (e.g., Stams
et al. 1991). Similarly, Hart et al. (1993) compared
the fate of deposited N in a grassland and a plantation forest using background-level 15N tracer, but
because the purpose of the experiment was to analyze internal N processes, the plot sizes used were
too small to make ecosystem-level budgets. However, they were able to identify important retention
mechanisms in the soils examined, and found that
the grassland retained significantly more of the deposited N than the forest. In most of these studies
the recovery of 15N was relatively high initially,
often close to 100%, but declined with time, suggesting that there might be a very real time limitation on following these experiments.
Other studies have used tracers in both reference
and fertilized plots; such complete cross-system
studies allow researchers to evaluate retention
mechanisms under different fertilizer and climate
of < 1 kg N ha - 1 yr - 1 (Kapustka and Dubois 1987;
Grant and Binkley 1987; Silvester 1989; Binkley
1996).
These studies suggest that a controlled, artificial
ecosystem with known boundaries can yield important information about nutrient cycling processes that cannot be determined from the largerscale whole-ecosystem budget approach where
spatial variability is high and many processes cannot be measured at a spatially explicit scale. However, these experiments are expensive, are rarely
replicated (San Dimas was not replicated by species), and take several decades before the first
meaningful analyses can be made. The artificial nature of fill soil may have a large effect on process
rates, making extrapolations to undisturbed ecosystems difficult. The small scale of the mesocosms
also pose some limitations on interpretation, just as
plot-scale studies cannot always be representative
of ecosystem-level dynamics. Although the measured output from these mesocosms quite accurately represent the true outputs, only a narrow portion of an ecosystem is measured, as deep soil,
groundwater, and riparian dynamics are not
measured.
Nitrogen-I5 Studies
at the Ecosystem Scale
Because the watershed budget approach cannot
trace internal movement or storage of ions, various
authors through the years have used isotopes to follow the fate of ecologically relevant elements. The
use of phosphorus-32 e 2 p) or other radioactive
tracers in large-scale experiments in natural terrestrial systems is not common (but see Caldwell et
al. 1985), but 15N has been added to natural ecosystems in a large number of studies either as enriched fertilizer (Mead and Pritchett 1975; Broadbent and Carleton 1978; Nadelhoffer et al. 1995;
Koopmans et al. 1996; Jordan et al. 1997; Nadelhoffer et al. 1998b; Tietema et al. 1998, to name a
few) or at background, throughfall input concentrations (White and Howes 1994; Starns et al. 1991;
Koopmans et al. 1996; Buchmann et al. 1995, 1996;
Seely and Lajtha 1997; Nadelhoffer et al. 1998a).
Unfortunately, less than a handful have used 15N at
background throughfall input levels and have
Kate Lajtha
traced movement over long time periods and in a
number of ecosystem pools, including output pools
such as deep lysimeters.
Buchmann et al. (1995, 1996) used 15N as a
tracer at background levels in the Fichtelgebirge,
Germany. Because there was a 20% decrease in the
recovery of both 15N03 and 15NH4 tracer after 15
months, they estimated that this reflected the
amount of loss from the system via leaching. The
highest retention of 15N was in organic soil horizons, followed by mineral soils, uptake by understory herbs, and uptake by overstory trees. Quite
similar patterns have since been seen in many other
studies with 15N tracer additions, where inputoutput budget studies alone could not have measured the retention rates in these different pools.
Similarly Koopmans et al. (1996), using quite low
as well as fertilizer levels of 15N addition in Nsaturated sites in the Netherlands, budgeted the
added N into vegetation uptake, soil retention, and
leaching losses, and found more uptake of label by
trees in a Douglas fir site than in a Scots pine site.
Seely and Lajtha (1997) added 15N to throughfall
to trace seasonal differences in N retention in a
coastal watershed in the United States and to examine soil texture effects on N retention, and found
patterns similar to those of Buchmann et al. (1995),
but with much higher leaching losses.
Other studies using background levels of 15N for
other purposes, such as estimating nitrification
rates, have not measured many internal pools, but
have measured leaching losses oflabel (e.g., Stams
et al. 1991). Similarly, Hart et al. (1993) compared
the fate of deposited N in a grassland and a plantation forest using background-level 15N tracer, but
because the purpose of the experiment was to analyze internal N processes, the plot sizes used were
too small to make ecosystem-level budgets. However, they were able to identify important retention
mechanisms in the soils examined, and found that
the grassland retained significantly more of the deposited N than the forest. In most of these studies
the recovery of 15N was relatively high initially,
often close to 100%, but declined with time, suggesting that there might be a very real time limitation on following these experiments.
Other studies have used tracers in both reference
and fertilized plots; such complete cross-system
studies allow researchers to evaluate retention
mechanisms under different fertilizer and climate