Developing predictive models - Conservation and Sustainable Use - page 157

It is important to recognise that a model is a highly simplified representation of
reality. Making a model that is so detailed that it captures reality near-perfectly is
a pointless exercise. On the other hand, there is a lot of justifiable criticism and
suspicion about models in the conservation world, which comes from modellers
abstracting so far from reality that their results are not practically useful. So there
is a trade-off between simplicity and realism . Too simple, and the model doesn't
capture the key elements of the system and so it can't predict effectively. Too com-
plex, and the model becomes a black box, in which you have no understanding of
the reasons behind results—which negates the purpose of the model in the first
place, and is very dangerous when you come to make predictions.
The first step is to create a conceptual model of the system. This in itself will be
useful in clarifying your assumptions about how the system works. But it can't be
used to quantify the effects of changes in the system or in parameter values, or of
uncertainty, on the outcome. This means it is very limited as a predictive tool—it
is more like an expression of the understanding that the statistical models in
Chapter 4 would give you. Hence it is preferable to go on to produce a mathema-
tical model .
Analytical models are of limited use in most of the situations that we are cover-
ing in this topic. They need to be very simple in order that they can be solved, so
bio-economic systems are usually too complex for meaningful analytical models to
be built. For example, the simple models used to obtain reference points in Box 4.1
are analytical models. These models also require mathematical skills to solve them.
Simulation models have the weakness of only telling you about the solution to
the model under the particular set of parameter values that you have used. We
will focus on simulation models in this chapter—but bearing in mind that the
ability of analytical models to produce a mathematical solution to a problem is a
big strength.
Equilibrium models are static—they don't take time into account. They are useful
inasmuch as it is important to know whether exploitation is tending towards a
sustainable equilibrium or to population extirpation. However, bio-economic
systems are dynamic, meaning that they need to be modelled over time. There are
inherent time-lags through, for example, the effects of hunting on population
dynamics and the effect of price and cost changes on hunter behaviour. There are also
constant shocks to the system moving it away from equilibrium, such as changes in
the weather affecting animal populations, as well as deterministic trends in impor-
tant variables that are constantly changing the equilibrium itself (such as habitat loss,
human population growth or technology improvements). Equilibrium models can
be solved analytically, but we will focus on dynamic simulation models. These are
more flexible, realistic and also easier to build and run because they don't require
advanced mathematical abilities—just the ability to correctly specify an equation.
The logistic model in Chapter 1 was a lumped model, in which the population
size was given the symbol N , and all individuals were implicitly assumed to be the
same. However, populations are in reality made up of different components,
whether they are populations of animals, plants, hunters or consumers. The extent