Specialization is a well-understood concept. A species adapts to a niche and becomes increasingly capable of exploiting that niche, and less capable of exploiting other niches. But there is something unclear about specialization: why not generalize? It should be advantageous to become less particular about diet and habitat; but this means of enhancing fitness is rare.
When we consider the various adaptations in animal forms, it seems odd that this seemingly infinite creativity should reach an end, leaving a lineage to face extinction. Readaptation to a previously abandoned niche is not unknown, but the morphological transformations involved seem awkward, different from the sublime plasticity which allowed the original adaptation. The general theory of irreversibility based on the improbability of history repeating itself or running backwards is not sufficient to explain why species get locked into niches, why they can't leave a niche as easily as they entered it.
The fossil record shows that radiation into a niche and the evolution of specialized morphology may take place rapidly, in geological terms. One would expect that with this kind of adaptivity a lineage could adapt to another niche. The history-doesn't-repeat-itself principle of irreversibility explains why species may not re-evolve the very same form they once had; but this does not explain why species are constrained against adapting somehow with the original degree of flexibility, when the pressure of natural selection demands it.
The problem is solved with the more solid explanation of irreversibility discussed above. The reductive process of evolution in segmented organisms, demonstrated by the hard evidence of evolutionary change, is a one-way street. Parts that are lost are not regained, because the mechanism of their loss is the degradation of genetic information, the corruption of DNA sequences.
The species that would escape its specialization faces the problem of number of anatomical elements; it has lost the parts that are unnecessary for it in its present niche, and the parts that remain are distorted to suit that niche. Adaptation was easy when the species consisted of a mass of symmetrical unspecialized parts; but it is different when this nascent morphology is lost.
There must be selective pressure for generalization or niche-broadening as well as for specialization. But if the mechanism for specialization is an irreversible ratchet, involving the permanent loss of parts, while generalization and retention of parts and potentiality involves stasis in number of parts at best, then the overall trend should be toward specialization.
There is a Procrustean aspect to evolution: whatever doesn't fit the niche, whatever isn't used, tends to get lopped off by natural selection, as it is an unnecessary drain upon resources. Episodic mass extinctions may be more understandable if this factor is recognized, as the longer an environment remains stable, the more vulnerable its occupants will be to changing circumstances. After many millions of years of environmental stability, a small change in climate--or a single geological catastrophe--might be disruptive enough to cause mass extinctions. Since it takes time for this sort of ecological vulnerability to develop, a certain periodicity in mass extinctions is to be expected. But this conception does not explain any supposed regularity in the timing of these episodes; and this is not to say that really massive extinctions are not triggered by extraordinary events.
forward to Stripes