2. Methods of Estimating Aboveground Net Primary Productivity
37
volves taking measurements of several "easy-tomeasure" parameters, which, when used with regression equations on a small sample of intensively
measured individuals, give information about many
of the "hard-to-measure" parameters, including
woody biomass and net growth increments. These
accessible measurements include DBH, tree height,
and basal circumference, which are translated with
the use of regression equations to measurements of
woody biomass and tree production. In cases in
which even the intensive sampling and harvest of
even a few individuals cannot be done, some generalizations can be made and they are useful when
working in native forests or protected areas where
little background information is available. For example, Whittaker and Marks (1975) suggested as
an estimate of tree biomass:
TB = 0.5 * BABH * TH
(2.4)
where TB is tree biomass, BABH is basal area at
breast height, and TH is tree height. Similarly, tree
production is estimated as:
TNPP = 0.5 * AWl * TH
(2.5)
where TNPP is tree net primary production and
AWl is annual wood increment (at breast height).
Additionally, some general relationships of DBH
and biomass have been developed for specific
regions, such as tropical forests (Brown and Iverson
1992), and in site-specific studies in particular forest ecosystems (e.g., Bormann and Gordon 1984;
Raich et al. 1997; Singh et a1. 1994).
Wood production per unit area can be estimated
in two ways (Binkley et al. 1997). One is to sum
the increments of individual trees in a unit area and
extrapolate to a hectare basis. Individual increments
can be estimated from repeated estimates as described above or using tree cores. The second way
of estimating wood production is to make repeated
estimates of total stand biomass through time. As
will be discussed later, these two different ways of
estimating wood production have interesting implications for the estimates of ANPP error.
Branch production is another component of forest ANPP. In some cases, branch plus trunk production can be estimated using regression equations
with tree height and diameter (Newbould 1970).
When branch production varies among treatments,
an independent assessment has been preferred. Researchers have measured the diameter of branches
at the base and correlated it with direct measurements of branch dry weight.
Once the woody and branch increments and the
litterfall have been accurately assessed, a further
complication that can occur in deciduous forests
(where some part of the year has substantial light
intercepted at the soil surface) is the productivity
of understory biomass. In some cases, this vegetation can represent a substantial component of NPP
and cannot be ignored or assumed to be a constant
fraction of woody production. Peak biomass harvests or sequential harvests over time (such as those
described in the previous section) may be used and
combined with overs tory measurements for the total production estimate.
Shrublands and steppes share characteristics of
grasslands and forests and usually require methods
that are a hybrid of those used in fast and slow
turnover ecosystems. Generally, the two components are assessed separately and combined for a
total measure of NPP. Shrubs present some of the
same challenges as trees, and various methods have
been developed to assess shrub productivity. For
example, Fernandez et al. (1991) developed a
method to measure shrub production in which they
harvested annual growth of shrubs (leaves and
small twigs) in a small quadrat (10 X 25 cm) located on top of several shrub individuals. They also
measured height and two diameters of several individuals. Based on allometric studies of these
shrub species and on data about their densities, they
extrapolated the quadrat results to a shrub individual and finally expressed their results on an areal
basis.
Errors Associated with Estimates
of ANPP
The section on estimating ANPP in fast turnover
ecosystems highlighted a series of methods to estimate ANPP. The more complicated methods tried
to take into account different sources of errors ranging from the missing peaks in biomass to the simultaneous nature of productivity, senescence, and
decomposition. All these errors tend to underestimate ANPP. Sala et al. (1988) named this kind of
error errors leading to underestimation (ELUs).
37
volves taking measurements of several "easy-tomeasure" parameters, which, when used with regression equations on a small sample of intensively
measured individuals, give information about many
of the "hard-to-measure" parameters, including
woody biomass and net growth increments. These
accessible measurements include DBH, tree height,
and basal circumference, which are translated with
the use of regression equations to measurements of
woody biomass and tree production. In cases in
which even the intensive sampling and harvest of
even a few individuals cannot be done, some generalizations can be made and they are useful when
working in native forests or protected areas where
little background information is available. For example, Whittaker and Marks (1975) suggested as
an estimate of tree biomass:
TB = 0.5 * BABH * TH
(2.4)
where TB is tree biomass, BABH is basal area at
breast height, and TH is tree height. Similarly, tree
production is estimated as:
TNPP = 0.5 * AWl * TH
(2.5)
where TNPP is tree net primary production and
AWl is annual wood increment (at breast height).
Additionally, some general relationships of DBH
and biomass have been developed for specific
regions, such as tropical forests (Brown and Iverson
1992), and in site-specific studies in particular forest ecosystems (e.g., Bormann and Gordon 1984;
Raich et al. 1997; Singh et a1. 1994).
Wood production per unit area can be estimated
in two ways (Binkley et al. 1997). One is to sum
the increments of individual trees in a unit area and
extrapolate to a hectare basis. Individual increments
can be estimated from repeated estimates as described above or using tree cores. The second way
of estimating wood production is to make repeated
estimates of total stand biomass through time. As
will be discussed later, these two different ways of
estimating wood production have interesting implications for the estimates of ANPP error.
Branch production is another component of forest ANPP. In some cases, branch plus trunk production can be estimated using regression equations
with tree height and diameter (Newbould 1970).
When branch production varies among treatments,
an independent assessment has been preferred. Researchers have measured the diameter of branches
at the base and correlated it with direct measurements of branch dry weight.
Once the woody and branch increments and the
litterfall have been accurately assessed, a further
complication that can occur in deciduous forests
(where some part of the year has substantial light
intercepted at the soil surface) is the productivity
of understory biomass. In some cases, this vegetation can represent a substantial component of NPP
and cannot be ignored or assumed to be a constant
fraction of woody production. Peak biomass harvests or sequential harvests over time (such as those
described in the previous section) may be used and
combined with overs tory measurements for the total production estimate.
Shrublands and steppes share characteristics of
grasslands and forests and usually require methods
that are a hybrid of those used in fast and slow
turnover ecosystems. Generally, the two components are assessed separately and combined for a
total measure of NPP. Shrubs present some of the
same challenges as trees, and various methods have
been developed to assess shrub productivity. For
example, Fernandez et al. (1991) developed a
method to measure shrub production in which they
harvested annual growth of shrubs (leaves and
small twigs) in a small quadrat (10 X 25 cm) located on top of several shrub individuals. They also
measured height and two diameters of several individuals. Based on allometric studies of these
shrub species and on data about their densities, they
extrapolated the quadrat results to a shrub individual and finally expressed their results on an areal
basis.
Errors Associated with Estimates
of ANPP
The section on estimating ANPP in fast turnover
ecosystems highlighted a series of methods to estimate ANPP. The more complicated methods tried
to take into account different sources of errors ranging from the missing peaks in biomass to the simultaneous nature of productivity, senescence, and
decomposition. All these errors tend to underestimate ANPP. Sala et al. (1988) named this kind of
error errors leading to underestimation (ELUs).