2. Methods of Estimating Aboveground Net Primary Productivity
35
ductivity and senescence. In contrast, ecosystems
made up of species with different phenologies and
life cycles, such as early-and late-season species,
C 3 and C 4 , or annuals and perennials, have a high
overlap of productivity, senescence, and decomposition. For example, the senescence of the earlyseason C 3 species in the Pampas masked the productivity of the late-season C4-dominated species
(Sala et al. 1981). The overlap error becomes less
important in ecosystems with lower diversity of
species or functional groups. One approach to reduce the overlap error has been to measure productivity by species and then sum them to calculate
total productivity.
Estimates of Aboveground Biomass
The discussion about estimates of primary productivity in fast turnover ecosystems assumed that a
time series of biomass estimates was available. This
section discusses the different ways of obtaining
biomass estimates and focuses on how to take into
account spatial variability; the temporal variability
issue and sampling frequency are discussed in the
following section. This section focuses on the issues related exclusively to biomass estimates and
does not deal with statistical sampling issues. Sampling is a large section of statistics devoted to the
theory and solution of applied problems and its
scope goes beyond that of this chapter (Scheaffer
et al. 1979).
In grasslands and shrublands, harvests of aboveground biomass have been the most frequently used
technique. The number of samples is mostly determined by the variability of the ecosystem and the
budget available or the accepted error. The size and
shape of the sampling quadrats try to minimize the
variance of the estimate (Greig-Smith 1983). The
optimal size of the sampling quadrats is related to
the pattern of vegetation. The variance of the estimates is maximum when the size of vegetation
patches coincides with the quadrat size and decreases as they become larger or smaller. Another
consideration is that as the quadrat size becomes
smaller, the length of the quadrat boundary per unit
area increases. Consequently, the edge effect increases because the person clipping biomass includes individuals or parts of individuals that
should not be included and vice versa. Different
shapes also affect the magnitude of the edge effect.
Circular plots have the lower boundary/area ratio,
while rectangular strips located perpendicular to
boundaries between patches show lower variance
than rectangles and circles. Rectangular strips encompass efficiently the natural variability of vegetation, which otherwise would require huge circles
or rectangles. The sampling scheme chosen, which
could be random, stratified random, or systematic,
depends on the patterns of variability of biomass
and on the available budget (see Scheaffer et al.
[1979] for a description of sampling techniques'
costs and benefits).
After the harvest in the field, biomass is transferred to a freezer as soon as possible to avoid mass
losses as a result of plant respiration. Later, material
is separated into the different compartments, which
can be specific compartments or functional compartments, such as green biomass, standing dead,
and litter, depending on the objectives of the study.
Finally, the different samples are oven-dried at
70°C for two days to obtain dry weights.
Harvest techniques are expensive and, consequently, many double sampling techniques have
been developed to reduce costs. Double sampling
techniques consist of establishing the correlation
between an expensive variable, such as biomass,
and a less-expensive variable, and then using the
inexpensive variable in the correlation for future
estimates of the expensive variable. Double sampling techniques work as long as the treatment or
factor under study does not affect the relationship
between the expensive and inexpensive variables.
For example, if the purpose of the experiment is to
assess the effect of fertilization on productivity and
the fertilization alters the relationship between
greenness and productivity, then double sampling
using greenness is not a good technique.
Capacitance is another variable used for double
sampling (Currie et al. 1973; Neal et al. 1976; Neal
and Neal 1973; Vickery et al. 1980). Changes in
canopy capacitance correlate well with aboveground biomass and small easy-to-use devices have
been developed that provide instant readings. Capacitance is mostly correlated with leaf area, and
meters need to be calibrated often when water content of the canopy changes. The correlation with
dry weight changes with species composition. This
is a simple technique mostly used in extension
when rapid estimates are needed.
35
ductivity and senescence. In contrast, ecosystems
made up of species with different phenologies and
life cycles, such as early-and late-season species,
C 3 and C 4 , or annuals and perennials, have a high
overlap of productivity, senescence, and decomposition. For example, the senescence of the earlyseason C 3 species in the Pampas masked the productivity of the late-season C4-dominated species
(Sala et al. 1981). The overlap error becomes less
important in ecosystems with lower diversity of
species or functional groups. One approach to reduce the overlap error has been to measure productivity by species and then sum them to calculate
total productivity.
Estimates of Aboveground Biomass
The discussion about estimates of primary productivity in fast turnover ecosystems assumed that a
time series of biomass estimates was available. This
section discusses the different ways of obtaining
biomass estimates and focuses on how to take into
account spatial variability; the temporal variability
issue and sampling frequency are discussed in the
following section. This section focuses on the issues related exclusively to biomass estimates and
does not deal with statistical sampling issues. Sampling is a large section of statistics devoted to the
theory and solution of applied problems and its
scope goes beyond that of this chapter (Scheaffer
et al. 1979).
In grasslands and shrublands, harvests of aboveground biomass have been the most frequently used
technique. The number of samples is mostly determined by the variability of the ecosystem and the
budget available or the accepted error. The size and
shape of the sampling quadrats try to minimize the
variance of the estimate (Greig-Smith 1983). The
optimal size of the sampling quadrats is related to
the pattern of vegetation. The variance of the estimates is maximum when the size of vegetation
patches coincides with the quadrat size and decreases as they become larger or smaller. Another
consideration is that as the quadrat size becomes
smaller, the length of the quadrat boundary per unit
area increases. Consequently, the edge effect increases because the person clipping biomass includes individuals or parts of individuals that
should not be included and vice versa. Different
shapes also affect the magnitude of the edge effect.
Circular plots have the lower boundary/area ratio,
while rectangular strips located perpendicular to
boundaries between patches show lower variance
than rectangles and circles. Rectangular strips encompass efficiently the natural variability of vegetation, which otherwise would require huge circles
or rectangles. The sampling scheme chosen, which
could be random, stratified random, or systematic,
depends on the patterns of variability of biomass
and on the available budget (see Scheaffer et al.
[1979] for a description of sampling techniques'
costs and benefits).
After the harvest in the field, biomass is transferred to a freezer as soon as possible to avoid mass
losses as a result of plant respiration. Later, material
is separated into the different compartments, which
can be specific compartments or functional compartments, such as green biomass, standing dead,
and litter, depending on the objectives of the study.
Finally, the different samples are oven-dried at
70°C for two days to obtain dry weights.
Harvest techniques are expensive and, consequently, many double sampling techniques have
been developed to reduce costs. Double sampling
techniques consist of establishing the correlation
between an expensive variable, such as biomass,
and a less-expensive variable, and then using the
inexpensive variable in the correlation for future
estimates of the expensive variable. Double sampling techniques work as long as the treatment or
factor under study does not affect the relationship
between the expensive and inexpensive variables.
For example, if the purpose of the experiment is to
assess the effect of fertilization on productivity and
the fertilization alters the relationship between
greenness and productivity, then double sampling
using greenness is not a good technique.
Capacitance is another variable used for double
sampling (Currie et al. 1973; Neal et al. 1976; Neal
and Neal 1973; Vickery et al. 1980). Changes in
canopy capacitance correlate well with aboveground biomass and small easy-to-use devices have
been developed that provide instant readings. Capacitance is mostly correlated with leaf area, and
meters need to be calibrated often when water content of the canopy changes. The correlation with
dry weight changes with species composition. This
is a simple technique mostly used in extension
when rapid estimates are needed.