Progress in Organismal Design

Abstract

The question of whether or not large-scale historical trends in the design of organisms represent progress (i.e., improvement of design) has attracted attention since the beginning of evolutionary thought but has proved extremely resistant to objective analysis. Some of this difficulty reflects the range of time scales and phylogenetic contexts for which the question might be posed, but a more fundamental problem involves the definition of improvement itself. When improvement of design is discussed within the context of current evolutionary thought, it is frequently portrayed as the expected outcome of the sustained operation of natural selection on variation within populations. Such an interpretation of large-scale historical trends in morphology frequently involves some degree of orthoselection and/or adaptive replacement. An alternative interpretation is that many such trends reflect little more than the Markovian aspect of the evolutionary process. Trends do have a finite, and not always small, probability of occurrence in systems whose underlying causal structure behaves in a pseudorandom fashion. According to this second interpretation, the appropriate level for causal analysis may be well below that at which the trend is manifested. This is equivalent to suggesting that the causal basis of the trend may be extremely heterogeneous. Both selectionist and Markovian models of morphologic change can be tested, but it is always a particular model, defined by its own assumptions and boundary conditions, that is corroborated or refuted, not the whole class of selectionist or Markovian models. Certain large-scale morphologic trends documented in the fossil record (e.g., increases in brain size within mammals) indeed appear to represent the intermittent or sustained operation of directional selection. Other trends (e.g., changes in morphologic complexity or in amount of genetic information within most phyla or classes) show patterns that are best interpreted as a simple random walk or a branching process (depending on the phylogenetic structure of the problem). However, a large number of trends (e.g., body size within certain higher taxa) can be explained by “diffusion” models or models of branching processes in which directionality is imposed by the position of the initial state of the system relative to the total range of accessible states. Because of the frequency of environmental change, the multiplicity of factors underlying fitness, the possibility of frequency-dependent and epistatic interactions among features, and the consequent possibility of nontransitive fitness relations between phenotypes, selection acting within populations frequently, though not inevitably, fails to produce unidirectional trends. The extent to which unidirectional trends dominate, or fail to dominate, the fossil record is therefore not a measure of the adequacy of neo-Darwinian mechanisms as causes of large-scale patterns in evolution.