Dissertation - HQ
Dissertation - HQ
Dissertation - HQ
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
28 Behaviour in models<br />
Using mean population performance in models is not appropriate<br />
when only a small portion of the performance distribution may constitute<br />
the survivors. Therefore, variance around the mean has to be<br />
derived from observations 63 or estimated from published accounts and<br />
incorporated into the model to provide a realistic range of individual<br />
results. Such a probabilistic approach can be accomplished through<br />
individual based models where traits of individual particles can be assigned<br />
following a probability density function. In addition, maximum<br />
values should also be considered because successful recruits may be<br />
the very few “best” individuals of each cohort (Figure 1.1).<br />
1.2.2 Ontogeny of behaviour<br />
Just like morphology, behaviour develops during the pelagic larval stage<br />
from essentially planktonic at its start to nektonic at its end, and the<br />
passive portion of it is likely to be short. In addition to ontogenetic<br />
changes in behavioural ability (e.g. swimming speed), there are frequently<br />
ontogenetic changes in the use of those abilities (e.g. age-related<br />
changes in depth or in swimming direction). The methods for modelling<br />
behaviour need to be adjusted according to the state of knowledge of<br />
physical-biological interactions that result in larval growth. Indeed, most<br />
studies indicate that size (or stage of development) is a better predictor<br />
of behavioural ability than is age 64 .<br />
Ontogeny of behaviour<br />
is best described by size<br />
• When growth is explicitly included in the model, behaviour can<br />
be formulated as a function of size. In addition, as mentioned<br />
above, this relation should not be deterministic and should not<br />
consider only the mean value for the population; associated variation<br />
should be included. In this case, as larvae are subjected<br />
to differential growth, in a model with spatially heterogeneous<br />
resources for example, they will display differential performance<br />
for a given behaviour.<br />
• When larval growth is not resolved in the model or when not<br />
enough information is available to predict a continuous relationship<br />
between size and behavioural performance, either age<br />
or developmental milestones can be used to model behaviours,<br />
possibly in a simplified, step-wise manner. Age and ontogenetic<br />
stages can be expressed by a dimensionless metric such as an<br />
developmental age 65 or ontogenetic index 64 .<br />
1.2.3 Taxonomic resolution of behaviour<br />
Ideally, the behaviour of larvae of the species to be modelled should be<br />
incorporated into the model. Nevertheless, it is important to know the<br />
degree to which the behaviour of a particular species can be generalised<br />
to other taxa, because it is unlikely that we will ever have even partial<br />
information on the behaviour of all fish species. At present, the amount