Evolutionary walks through a land plant morphospace

A model for mimicking land plant evolution is here expanded and re-evaluate d. The model consists of (1) a morphospace containing on the order of 10(9) phenotypic variants, (2) 15 different fitness landscapes, each defined on the basis of performing one or more of four tasks (i,e, maximizing light in terception, mechanical stability and reproduction, and minimizing total sur face area), and (3) an algorithm driving a search through fitness landscape s for more fit variants. The model is used to predict the effects of the nu mber of simultaneously performed tasks ('complexity'), abrupt changes in en vironmental conditions (mimicked by random replacement of one fitness lands cape with another), and developmental barriers (mimicked by barring searche s from entering specific subdomains in the morphospace) on number and acces sibility of variants occupying fitness maxima, The model predicts that (1) the number and accessibility of fitness peaks will increase (while the diff erence between the relative fitness of peaks and valleys will decrease) in proportion to functional complexity, (2) abrupt shifts in landscapes will i ncrease rather than decrease phenotypic diversity, and (3) obstructed searc hes have an equal or higher probability of reaching fitness peaks than unfe ttered searches. These results follow axiomatically from the way hypothetic al variants are spatially ordered in the morphospace, the manner in which r elative fitness is defined, and the protocol used to locate fitness peaks. A critique of the model's predictions and desiderata for future research ar e provided.