2. Methods of Estimating Aboveground Net Primary Productivity
41
from the estimate of wood production for the study
period. Consequently, tree mortality results in a
large underestimation error when the method of differences in stand biomass is used to calculate wood
production.
Optimal Methodology
to Estimate ANPP
The optimal methodology to assess ANPP depends
on the objective of the study. Analysis of the errors
associated with estimates of productivity provides
the tools to choose the best method given certain
objectives and budget. Methods that try to minimize ELUs necessarily increase ELOs. The sources
of ELUs are the missing peaks and the simultaneous nature of productivity and decomposition. The
way to reduce these errors is to increase the sampling frequency to reduce the number of missing
peaks and troughs. The increase in the frequency
also decreases the possibility of overlap between
productivity and decomposition. Estimating productivity by species and then adding them up to
assess total NPP also reduces ELUs, as discussed
above. However, the increase in sampling frequency and the estimates by species all increase
ELOs since they reduce the true value of NPP and
increase the variance of the estimates of B 1 - BO.
Similarly, reductions in the sampling frequency and
in the number of components analyzed reduce
ELOs but increase ELUs.
If the purpose of the study is to obtain an estimate of annual production, and the ecosystem has
a clear seasonality, the best method will be a single
harvest at peak biomass. If the ecosystem has even
productivity distributed throughout the year or frequent peaks and troughs, a more frequent sampling
scheme will be desirable. One possibility of resolving the tradeoff between reducing ELUs and ELOs
is to use a detailed method and estimate the overestimation error using the algorithm developed by
Sala et al. (1988).
It is important to distinguish between methods
that are conceptually more correct and methods that
provide an answer that is closer to the true value.
Methods that study the independent pattern of individual species or those that take into account numerous functional compartments are closer to the
concept of primary production. However, the results that they yield may be further from the real
value than those resulting from simpler methods.
Here it is crucial to take into account the uncertainty associated with each one of the terms used
in the calculation. O'Neill (1973) pointed out that
uncertainty of the results increases as the uncertainty of the individual components increases. In
the production case, higher variability in the biomass estimates means larger overestimation error.
An extreme example is to try to estimate annual
production from gas exchange estimates from individualleaves or small patches that then are extrapolated to longer time frames and larger areas.
Photosynthesis is a fast variable that varies in a
matter of minutes and consequently the errors associated with extrapolating from the leaf to the
hectare and from minutes or hours to months and
years are so large that they make the results irrelevant. Although the annual estimate of production
through the gas exchange method may contain
large errors, it may still be very close to the concept
of productivity.
Summary
There is not a best method to measure aboveground
primary productivity. Decisions about methods are
even more complicated, because methods that reduce one kind of error increase other kinds of errors. It is somehow counterintuitive that the more
complicated and expensive methods, which take
into account most of the flows involved in primary
productivity, may yield results with the largest errors. Although these methods are closest to the concept of primary productivity, they may yield results
that are the farthest from the real value of productivity. Uncertainty in the variables used to estimate
primary productivity results in greater uncertainty
in the final estimates of productivity. Because of
the way productivity is calculated, uncertainty in
the input variables always results in overestimation
of productivity. This chapter described of the kinds
of errors associated with each method and how they
relate to the ecosystem characteristics, as well as
the costs and benefits of the different methodological alternatives. The best method will depend on
the characteristics of the ecosystem, such as turn-
41
from the estimate of wood production for the study
period. Consequently, tree mortality results in a
large underestimation error when the method of differences in stand biomass is used to calculate wood
production.
Optimal Methodology
to Estimate ANPP
The optimal methodology to assess ANPP depends
on the objective of the study. Analysis of the errors
associated with estimates of productivity provides
the tools to choose the best method given certain
objectives and budget. Methods that try to minimize ELUs necessarily increase ELOs. The sources
of ELUs are the missing peaks and the simultaneous nature of productivity and decomposition. The
way to reduce these errors is to increase the sampling frequency to reduce the number of missing
peaks and troughs. The increase in the frequency
also decreases the possibility of overlap between
productivity and decomposition. Estimating productivity by species and then adding them up to
assess total NPP also reduces ELUs, as discussed
above. However, the increase in sampling frequency and the estimates by species all increase
ELOs since they reduce the true value of NPP and
increase the variance of the estimates of B 1 - BO.
Similarly, reductions in the sampling frequency and
in the number of components analyzed reduce
ELOs but increase ELUs.
If the purpose of the study is to obtain an estimate of annual production, and the ecosystem has
a clear seasonality, the best method will be a single
harvest at peak biomass. If the ecosystem has even
productivity distributed throughout the year or frequent peaks and troughs, a more frequent sampling
scheme will be desirable. One possibility of resolving the tradeoff between reducing ELUs and ELOs
is to use a detailed method and estimate the overestimation error using the algorithm developed by
Sala et al. (1988).
It is important to distinguish between methods
that are conceptually more correct and methods that
provide an answer that is closer to the true value.
Methods that study the independent pattern of individual species or those that take into account numerous functional compartments are closer to the
concept of primary production. However, the results that they yield may be further from the real
value than those resulting from simpler methods.
Here it is crucial to take into account the uncertainty associated with each one of the terms used
in the calculation. O'Neill (1973) pointed out that
uncertainty of the results increases as the uncertainty of the individual components increases. In
the production case, higher variability in the biomass estimates means larger overestimation error.
An extreme example is to try to estimate annual
production from gas exchange estimates from individualleaves or small patches that then are extrapolated to longer time frames and larger areas.
Photosynthesis is a fast variable that varies in a
matter of minutes and consequently the errors associated with extrapolating from the leaf to the
hectare and from minutes or hours to months and
years are so large that they make the results irrelevant. Although the annual estimate of production
through the gas exchange method may contain
large errors, it may still be very close to the concept
of productivity.
Summary
There is not a best method to measure aboveground
primary productivity. Decisions about methods are
even more complicated, because methods that reduce one kind of error increase other kinds of errors. It is somehow counterintuitive that the more
complicated and expensive methods, which take
into account most of the flows involved in primary
productivity, may yield results with the largest errors. Although these methods are closest to the concept of primary productivity, they may yield results
that are the farthest from the real value of productivity. Uncertainty in the variables used to estimate
primary productivity results in greater uncertainty
in the final estimates of productivity. Because of
the way productivity is calculated, uncertainty in
the input variables always results in overestimation
of productivity. This chapter described of the kinds
of errors associated with each method and how they
relate to the ecosystem characteristics, as well as
the costs and benefits of the different methodological alternatives. The best method will depend on
the characteristics of the ecosystem, such as turn-