The rule based ecosystem response model uses rules specified by experts to estimate the response of various species to the lake salinity.
The rules specified by the biologist types implied quite a bit of complexity, things like complex queries (any adjacent, all cells, any cell etc), and a large degree of parallelism, (what is a beastie doing in next door cell?)
This complexity conflicted with the requirement that the user must be able to fiddle easily with the rules.
By effectively creating a new programming language, in which these rules could easily be formulated, I hoped to separate the complexity of the programming from the complexity of the biology. So fiddling with the biology or the program won't require major changes in the other.
The full detail on the rule based model may be found in the associate program manual. (This exists in a GNU ``info'' hypertext online form, on the world wide web and in paper manual form.)