298
The Biology of Sea Turtles, Vol. II
Models are our primary method to evaluate alternative hypotheses for the causes
of population decline and future trends caused by changes in vital rates. Extrapolation
of measured trends can be useful, but a model that projects population growth as a
function of ASM and vital rates may more accurately predict changes in that trend.
For example, Spotila et al. (2000) extrapolated the decline of leatherback turtles
nesting at Playa Grande, Costa Rica, but also estimated the response of the nesting
population to increased adult recruitment through nest protection efforts. Models for
Kemp’s ridleys (Heppell et al., 2002a) project a decrease in the population growth
rate starting around 2010 because of a decrease in the nest survival rate. These
analyses were deterministic projections that can only generally predict changes in
population size, but serve as a useful baseline to compare with actual population
trends that we may witness in the future.
In Australia, more detailed data allowed development of more elaborate stochastic difference equation models for green and loggerhead sea turtles (Chaloupka,
2002; Chaloupka and Limpus, 2002). Chaloupka (2001) outlined many advantages
of developing fully stochastic models, if the data are available. One basis for his
analyses is that variation is probably greater in processes that influence fecundity
and survival in the egg and early juvenile stages than it is for survival later in the
life history. The qualitative effect of this variability on previous conclusions from
deterministic models is that reproduction and early survival can have large effects
on the variability in sea turtle abundance, although the average growth rate over
long time periods may be similar in both models. This has been shown in comparisons of deterministic and stochastic models for fish populations (Quinlan and
Crowder, 1999) and in viability analysis (Wisdom et al., 2000). Of course, one could
use a variety of model structures from simple deterministic models to stochastic,
individual-based, spatially explicit models (Letcher et al., 1998; Walters et al., 2002).
Our approach has been to use the most appropriate modeling form as constrained
by our questions and the available data. However, more complex models can be used
as tools for the heuristic evaluation of population dynamics under a suite of assumptions about vital rates, variability, and density-dependent effects. The utility of such
exercises to management will soon be apparent for new assessments of Pacific
leatherback populations (M. Chaloupka, personal communication).
Population models are essential for conservation and management because sea
turtles are such late-maturing species. A biologist who studies loggerhead, green,
or hawksbill turtles might be lucky to see two generations in his or her lifetime.
Long life and late ASM, coupled with variable abundance in space and time, decrease
our ability to observe population changes that may lead to recovery or extinction.
We are only beginning to understand how sea turtle population dynamics operate,
through a combination of long-term surveys, mark–recapture studies, and population
models. Much of our work is highly speculative, but has already contributed to the
management and conservation of sea turtles worldwide.
1123 book.book Page 298 Tuesday, November 12, 2002 7:43 AM
The Biology of Sea Turtles, Vol. II
Models are our primary method to evaluate alternative hypotheses for the causes
of population decline and future trends caused by changes in vital rates. Extrapolation
of measured trends can be useful, but a model that projects population growth as a
function of ASM and vital rates may more accurately predict changes in that trend.
For example, Spotila et al. (2000) extrapolated the decline of leatherback turtles
nesting at Playa Grande, Costa Rica, but also estimated the response of the nesting
population to increased adult recruitment through nest protection efforts. Models for
Kemp’s ridleys (Heppell et al., 2002a) project a decrease in the population growth
rate starting around 2010 because of a decrease in the nest survival rate. These
analyses were deterministic projections that can only generally predict changes in
population size, but serve as a useful baseline to compare with actual population
trends that we may witness in the future.
In Australia, more detailed data allowed development of more elaborate stochastic difference equation models for green and loggerhead sea turtles (Chaloupka,
2002; Chaloupka and Limpus, 2002). Chaloupka (2001) outlined many advantages
of developing fully stochastic models, if the data are available. One basis for his
analyses is that variation is probably greater in processes that influence fecundity
and survival in the egg and early juvenile stages than it is for survival later in the
life history. The qualitative effect of this variability on previous conclusions from
deterministic models is that reproduction and early survival can have large effects
on the variability in sea turtle abundance, although the average growth rate over
long time periods may be similar in both models. This has been shown in comparisons of deterministic and stochastic models for fish populations (Quinlan and
Crowder, 1999) and in viability analysis (Wisdom et al., 2000). Of course, one could
use a variety of model structures from simple deterministic models to stochastic,
individual-based, spatially explicit models (Letcher et al., 1998; Walters et al., 2002).
Our approach has been to use the most appropriate modeling form as constrained
by our questions and the available data. However, more complex models can be used
as tools for the heuristic evaluation of population dynamics under a suite of assumptions about vital rates, variability, and density-dependent effects. The utility of such
exercises to management will soon be apparent for new assessments of Pacific
leatherback populations (M. Chaloupka, personal communication).
Population models are essential for conservation and management because sea
turtles are such late-maturing species. A biologist who studies loggerhead, green,
or hawksbill turtles might be lucky to see two generations in his or her lifetime.
Long life and late ASM, coupled with variable abundance in space and time, decrease
our ability to observe population changes that may lead to recovery or extinction.
We are only beginning to understand how sea turtle population dynamics operate,
through a combination of long-term surveys, mark–recapture studies, and population
models. Much of our work is highly speculative, but has already contributed to the
management and conservation of sea turtles worldwide.
1123 book.book Page 298 Tuesday, November 12, 2002 7:43 AM