That very much depends on context, and that is one of the keys to understanding how evolution works.
If the context is one where the major threats to an organism comes from other things like itself in the same population, then competitive systems dominate, and the individuals are forced to adapt to some set of minima on the “complexity landscape”.
If the context is one where threats to individuals in a population come more from external factors that may be mitigated to some degree by cooperative effort, then cooperation can establish, and with the addition of strategies to detect and remove cheats, stabilise. In cooperative contexts, complexity is free to increase, as there is true freedom to explore.
Contexts are always changing.
The context in one valley may be quite different from the valley next door.
If sea levels rise every thousand generations or so, cutting off access between valleys, then fall and allow mixing for a thousand generations or so (roughly what the ice age cycle does in many parts of the planet), then there is plenty of room for variations on themes to develop new skill sets and explore new niches.
The space of possible selection pressures is hugely dimensional.
Often a few of those clearly dominate over all others, and sometimes the mix is more even.
The numeric complexity of the resulting permutations on systems is mind numbingly huge.
We seem to be the result of some 4 billion years of that process (as, it seems, is every other living organism – from bacteria to blue whales).
Ted Howard NZ's Blog 