10. Herbivory in Terrestrial Ecosystems
increase in relative growth rate is not high enough
to result in equal productivity of grazed and ungrazed systems, the response is called partial compensation or undercompensation. When the increase is enough to equal the productivity of
ungrazed systems, the response is called full compensation. Finally, when the increase is such that
productivity under grazing is larger than in ungrazed situations, the response is called overcompensation (Belsky 1986; Oesterheld and McNaughton 1991). If there were neither compensatory
growth nor damage, the effect of grazing on productivity would be equivalent to consumption, so
measuring either one would return the value of the
other.
Overcompensation and full compensation may
be detected by simple comparisons of productivity
of grazed and ungrazed treatments. In contrast,
when productivity under grazing is lower than controls, simple determinations of productivity will not
suffice to discriminate between partial compensation and damage. In that case, a determination of
initial biomass of grazed and control plots is required to calculate relative growth rates. Lack of
these data and loose usage of terms frequently results in two different errors: some researchers refer
to "compensatory responses" only in cases in which
overcompensation was observed, ignoring that
there may be compensation even when grazing reduced productivity; whereas others refer to partial
compensation or undercompensation in all cases in
which productivity is reduced by grazing, ignoring
that some of those cases may represent damage.
Approaches
One of the approaches to estimating the effect of
herbivory on production is based on the comparison
of productivity of ungrazed systems with that of
grazed systems from which herbivores have been
temporarily removed. This herbivore exclusion
may be done in different ways: fencing (for large
herbivores), poisoning (for insects), or individual
removals (rodents). We here discuss in more detail
the methods based on fencing because they are the
most frequently used, but most of our considerations are valid for the other methods. As we described above, fencing is also used to estimate consumption. Thus, both consumption and production
may be estimated in the same study by a combi155
nation of cage movements (McNaughton et al.
1996): as we showed before, the difference ofbiomass between caged and uncaged plots estimates
consumption, and, as we will show now, the rate of
accumulation of biomass inside cages estimates
production.
Compensatory growth affects the choice among
different approaches to study the effect of herbivory
on primary productivity (McNaughton et al. 1996).
The moveable-cage method uses moveable exclosures that are moved with a frequency timed to reflect the intensity of herbivory. Compared with the
season-Iong--<:age method, in which herbivores are
excluded for the entire season, it has the advantage
that what is sampled inside moveable cages better
resembles the real grazed system. The season-Iongcage method seems to be suitable to assess the longterm effect of grazing on productivity, rather than
the current effect of the herbivores. For example,
comparing 30-year-old exclosures with continuously grazed areas by placing season-long cages in
the grazed area answers the question, "What is the
effect of long-term grazing on primary productivity
in a year without grazing?" Short-term compensatory regrowth will be missed by this approach and
the grazed systems being evaluated will soon resemble the structure of ungrazed systems, particularly in regions with high productivity (McNaughton et al. 1996). For both methods, however,
production must be simultaneously estimated in
comparable, ungrazed controls (see Chapter 2).
The decision of how frequently to move exclosures and sample biomass in ungrazed controls is
crucial for the moveable-cage method (McNaughton et al. 1996). Since productivity estimates are
affected by the frequency of biomass sampling
(Sala et al. 1988; Biondini et al. 1991; see Chapter
2), care should be taken in order to avoid introducing differences of frequency between grazed and
ungrazed treatments. Once this is taken care of, increasing frequency will maximize both the resemblance between the cages and the grazed system
and the likelihood of detecting regrowth. However,
the statistical errors common to any determination
of productivity will also increase (Sala et al. 1988;
Biondini et al. 1991; McNaughton et al. 1996; see
Chapter 2). There is not a simple rule to decide
about this tradeoff, but certainly sampling effort at
each biomass sampling and harvest frequency will
have to be correlated. Maximizing the detection of
increase in relative growth rate is not high enough
to result in equal productivity of grazed and ungrazed systems, the response is called partial compensation or undercompensation. When the increase is enough to equal the productivity of
ungrazed systems, the response is called full compensation. Finally, when the increase is such that
productivity under grazing is larger than in ungrazed situations, the response is called overcompensation (Belsky 1986; Oesterheld and McNaughton 1991). If there were neither compensatory
growth nor damage, the effect of grazing on productivity would be equivalent to consumption, so
measuring either one would return the value of the
other.
Overcompensation and full compensation may
be detected by simple comparisons of productivity
of grazed and ungrazed treatments. In contrast,
when productivity under grazing is lower than controls, simple determinations of productivity will not
suffice to discriminate between partial compensation and damage. In that case, a determination of
initial biomass of grazed and control plots is required to calculate relative growth rates. Lack of
these data and loose usage of terms frequently results in two different errors: some researchers refer
to "compensatory responses" only in cases in which
overcompensation was observed, ignoring that
there may be compensation even when grazing reduced productivity; whereas others refer to partial
compensation or undercompensation in all cases in
which productivity is reduced by grazing, ignoring
that some of those cases may represent damage.
Approaches
One of the approaches to estimating the effect of
herbivory on production is based on the comparison
of productivity of ungrazed systems with that of
grazed systems from which herbivores have been
temporarily removed. This herbivore exclusion
may be done in different ways: fencing (for large
herbivores), poisoning (for insects), or individual
removals (rodents). We here discuss in more detail
the methods based on fencing because they are the
most frequently used, but most of our considerations are valid for the other methods. As we described above, fencing is also used to estimate consumption. Thus, both consumption and production
may be estimated in the same study by a combi155
nation of cage movements (McNaughton et al.
1996): as we showed before, the difference ofbiomass between caged and uncaged plots estimates
consumption, and, as we will show now, the rate of
accumulation of biomass inside cages estimates
production.
Compensatory growth affects the choice among
different approaches to study the effect of herbivory
on primary productivity (McNaughton et al. 1996).
The moveable-cage method uses moveable exclosures that are moved with a frequency timed to reflect the intensity of herbivory. Compared with the
season-Iong--<:age method, in which herbivores are
excluded for the entire season, it has the advantage
that what is sampled inside moveable cages better
resembles the real grazed system. The season-Iongcage method seems to be suitable to assess the longterm effect of grazing on productivity, rather than
the current effect of the herbivores. For example,
comparing 30-year-old exclosures with continuously grazed areas by placing season-long cages in
the grazed area answers the question, "What is the
effect of long-term grazing on primary productivity
in a year without grazing?" Short-term compensatory regrowth will be missed by this approach and
the grazed systems being evaluated will soon resemble the structure of ungrazed systems, particularly in regions with high productivity (McNaughton et al. 1996). For both methods, however,
production must be simultaneously estimated in
comparable, ungrazed controls (see Chapter 2).
The decision of how frequently to move exclosures and sample biomass in ungrazed controls is
crucial for the moveable-cage method (McNaughton et al. 1996). Since productivity estimates are
affected by the frequency of biomass sampling
(Sala et al. 1988; Biondini et al. 1991; see Chapter
2), care should be taken in order to avoid introducing differences of frequency between grazed and
ungrazed treatments. Once this is taken care of, increasing frequency will maximize both the resemblance between the cages and the grazed system
and the likelihood of detecting regrowth. However,
the statistical errors common to any determination
of productivity will also increase (Sala et al. 1988;
Biondini et al. 1991; McNaughton et al. 1996; see
Chapter 2). There is not a simple rule to decide
about this tradeoff, but certainly sampling effort at
each biomass sampling and harvest frequency will
have to be correlated. Maximizing the detection of