characterized the environments of perennibranchiate salamanders in just such terms: many perennibranchiates are cavernicolous,2 while others inhabit permanent lakes and streams.

In summary, the permanent fish-free pond and the ephemeral mushroom are environments of a very different order. A salamander will not crowd itself out of a good pond when life is so difficult outside it—especially when the pond may persist as a favorable environment for several thousand years. A mushroom or a leaf, on the other hand, appears infrequently (relative to a tiny insect's size and mobility) and offers much in the very short run. Since it won't last long anyway (and since the next superabundant patch will not appear in the same place), it might as well be exploited to build as much biomass as quickly as possible, so that some fortunate bearer of parental genes may survive to find the next mushroom. In this metaphorical simplification, the r strategy of progenetic insects contrasts sensibly with the K strategy of neotenic salamanders.

The Ecological Determinants of Progenesis

As a further test of my hypothesis, I have tried to categorize the environments for all clear cases of progenesis known to me from the literature. I find a remarkable correlation between progenesis and r- selective habitats (except in cases where selection works primarily for small size itself). I do not know of any instance in which juvenilized morphology per se is clearly the major determinant of progenesis. Rapid maturation or small body size are the primary objects of selection; juvenile morphology is often an incidental (though not inadaptive) by-product. This conclusion is contrary to the conventional view that morphology is of prime importance in the evolution of heterochrony.

Unstable Environments

During the heyday of stability-diversity theory a few years back, much was written about the correlation of low organic diversity with unstable, unpredictable habitats that impose frequent, density- independent, mass mortality upon the few species able to survive. In these environments, fine-tuned morphological specializations would provide no benefit—for there is nothing stable enough to tune them to, and selection must operate primarily for sheer survival against frequent disruption. High r may yield enough offspring (and produce