In the spirit of reducing the complexity of the evaluations, a
series of statistical indicators was chosen in order to quantitatively
characterize the performance. The most suitable indicators for
model benchmarking and evaluation are residual indicators
(Table 5) which calculate the difference between observed and
modelled data pairs (Thunis et al., 2011; Bennett et al., 2013). Statistics
were calculated separately for all species but values are given
in a domain-wide basis. Scatter plots were used as diagnosis diagrams,
evaluating in every case if data pairs distributed along a
unity-slope line through the origin.