Developing predictive models - Conservation and Sustainable Use - page 160

of the diagram? If not, then this needs careful thought to check that the model
is logically constructed. It might be worth considering a different shape for some
of the boxes—e.g. rectangles for quantities, diamonds for decision points.
● The model is sometimes simply a representation of how some quantity moves
around a system (like energy in food web diagrams, or individuals in population
models—see conceptual model example 1 below). Other times it represents
causal links —events A, B and C lead to a decision X, which has effect Y
(see conceptual model example 2 below). It is important not to mix up the
two kinds of diagram.
● It's useful to run through the diagram in both directions, forwards and back-
wards, and check that the logic holds whichever way you go.
● Ask yourself whether each step of the model translates into an equation of the
form y f ( x ), i.e. the outcome box is a function of the input box(es). If yes,

then the conceptual model is usable as a basis for a mathematical model.

The conceptual model is a summary of your current understanding of the system,
informed either by the literature, or by the data collection and analysis we've discussed
in Chapters 2-4. At this conceptualising stage another important step is to think
about the uncertainty in our understanding of the system, and whether there are any
feasible alternative model structures . The outcome of a model is fundamentally
determined by the structure of its causal links, so if only one conceptual model is used,
this has already to a large extent determined what your answers are. This comes back
to the issue of confronting models with data discussed in Section 4.4.1. In population
models, for example, there are usually several potentially valid forms for density
dependence and dispersal functions. Similarly, in economic models, we make
assumptions about people's attitudes to risk and the relationship between money and
utility (see Section 5.4.2), which may or may not be valid.
We could also consider using two different types of model and comparing their
output, for example, comparing a simple lumped model with an age-structured
model. This will tell us whether the added complexity of the structured model leads
to a better fit to the data or a worse one (see Section 5.4). Another fundamental
question is whether we are including the right variables in the model? Is there a
missing variable ? For example, habitat type may determine animal abundance as
well as hunting pressure (Box 4.4).
The most important component of developing a conceptual model is to get a
really good understanding of the system through reading and/or fieldwork.
Reading as many papers as possible on models developed for similar systems to
yours will give you ideas about the type of model which is useful for addressing
your research questions.

5.3.1.1 Conceptual model example 1—Red deer
Figure 5.1 shows the flow of individuals through different age classes. We
have modelled three age classes using the symbol N — year 1 (juveniles), year 2
(non-breeding sub-adults) and adults (all other ages). The symbol S represents