8. Whaling Models for Cetacean Conservation
115
such extraordinary rates of increase of 7% or more are likely to be inaccurate
estimates. Nevertheless, these many observations of population resilience imply
that other baleen whales can recover after having been overexploited, although
many years may be required given the low reproductive rates of these large
species.
Testing the CLA with Simulations
The IWC Scientific Committee conducted extensive computer simulation trials to
evaluate the CLA and to ensure that it met the IWC’s three objectives. Models
were constructed that included considerable detail about whale life histories including age and sex structure, variable vital rates, and a variety of assumptions
about depensation and density dependence. In addition, thousands of simulations
were conducted to evaluate the consequences of biases and errors in survey data
and prior catch data. To simulate changes in future environments, simulations
were conducted in which the carrying capacity declined or underwent cyclic
oscillations.
The Cooke procedure was selected because it was most robust to the various
modeling scenarios. The simplicity of the Cooke procedure was a benefit because
estimation of fewer parameters was required, and therefore the power associated
with input data was higher. The many conservative cushions incorporated into the
CLA outlined above would seem to be sufficient to prevent harvests of whales in
depleted stocks.
Of the algorithms evaluated by the Scientific Committee, the Cooke procedure
best met the objectives provided by the IWC. However, the IWC has not provided
guidelines on precisely what constitutes acceptable performance of the CLA, and
the simulations demonstrate that there are some circumstances under which it
performs poorly (Taylor et al. 1994). Using the population size at the end of 100
years to evaluate the CLA, model performance appears poor when current stock
size is overestimated (Taylor et al. 1994). Deriving reliable estimates of whale
populations over extensive areas of ocean is a daunting task and one that can be
highly sensitive to observer bias and sampling protocol. If estimates were made
that indicated a number higher that the true population size, quotas set by the CLA
will be excessive and drive the population to low levels. The slow rate at which
the CLA learns is partly responsible for this weakness in model performance
(Young 1993).
Simulations also showed that the CLA performs poorly when historical catches
are underreported (Taylor et al 1994). The CLA uses historical catch to backcalculate preexploitation stock size, and if only a fraction of the actual harvest is
reported, the model will target a harvested population that is far too low. For some
stocks, this can be a serious problem. Right Whales have been protected since the
1930s, yet recent reports document Soviet exploitation of Right Whales in the
Okhotsk Sea, near Tristan da Cunha in the south Atlantic, near Kurile Island in the
Pacific, and from factory ships in the Antarctic (Best 1988; Yablokov 1994).
115
such extraordinary rates of increase of 7% or more are likely to be inaccurate
estimates. Nevertheless, these many observations of population resilience imply
that other baleen whales can recover after having been overexploited, although
many years may be required given the low reproductive rates of these large
species.
Testing the CLA with Simulations
The IWC Scientific Committee conducted extensive computer simulation trials to
evaluate the CLA and to ensure that it met the IWC’s three objectives. Models
were constructed that included considerable detail about whale life histories including age and sex structure, variable vital rates, and a variety of assumptions
about depensation and density dependence. In addition, thousands of simulations
were conducted to evaluate the consequences of biases and errors in survey data
and prior catch data. To simulate changes in future environments, simulations
were conducted in which the carrying capacity declined or underwent cyclic
oscillations.
The Cooke procedure was selected because it was most robust to the various
modeling scenarios. The simplicity of the Cooke procedure was a benefit because
estimation of fewer parameters was required, and therefore the power associated
with input data was higher. The many conservative cushions incorporated into the
CLA outlined above would seem to be sufficient to prevent harvests of whales in
depleted stocks.
Of the algorithms evaluated by the Scientific Committee, the Cooke procedure
best met the objectives provided by the IWC. However, the IWC has not provided
guidelines on precisely what constitutes acceptable performance of the CLA, and
the simulations demonstrate that there are some circumstances under which it
performs poorly (Taylor et al. 1994). Using the population size at the end of 100
years to evaluate the CLA, model performance appears poor when current stock
size is overestimated (Taylor et al. 1994). Deriving reliable estimates of whale
populations over extensive areas of ocean is a daunting task and one that can be
highly sensitive to observer bias and sampling protocol. If estimates were made
that indicated a number higher that the true population size, quotas set by the CLA
will be excessive and drive the population to low levels. The slow rate at which
the CLA learns is partly responsible for this weakness in model performance
(Young 1993).
Simulations also showed that the CLA performs poorly when historical catches
are underreported (Taylor et al 1994). The CLA uses historical catch to backcalculate preexploitation stock size, and if only a fraction of the actual harvest is
reported, the model will target a harvested population that is far too low. For some
stocks, this can be a serious problem. Right Whales have been protected since the
1930s, yet recent reports document Soviet exploitation of Right Whales in the
Okhotsk Sea, near Tristan da Cunha in the south Atlantic, near Kurile Island in the
Pacific, and from factory ships in the Antarctic (Best 1988; Yablokov 1994).