Steps by which some higher taxa have evolved (e.g., birds, mammals) have been well documented in the fossil record. Intermediate forms give evidence of both mosaic evolution of different characters and of changes in the form and function of specific characteristics.
Major changes in characteristics evolve not by large jumps (saltations), but generally evolve gradually, through intermediate stages. The evolution of some characters does include effects of mutations with moderately large effects. Complex structures such as eyes evolve by rather small, individually advantageous steps. They may acquire functional integration with other features so that they become indispensable.
Homologous characters may be based on similar or the same networks of regulatory gene interactions. Novel features have arisen, in at least some cases, by the recruitment of integrated genetic and developmental pathways in new contexts or combinations.
Some fundamental characteristics of developmental processes and organismal integration may enhance evolvability, the capacity of a genome to produce variants that are potentially adaptive.
The fossil record provides examples of both gradual change and the pattern called punctuated equilibria a rapid shift from one static phenotype to another. The hypothesis that such shifts require speciation is not widely accepted because responses to selection do not depend on speciation.
The long-term average rate of evolution of most characters is very low because long periods of little change (stasis) are averaged with short periods of rapid evolution, or because the character mean fluctuates without long-term directional change. The highest rates of character evolution in the fossil record are comparable to rates observed in current populations and can readily be explained by known processes such as mutation, genetic drift, and natural selection.
Stasis and low rates of character evolution can be explained by genetic constraints, stabilizing selection (owing largely to habitat tracking), or gene flow among divergently selected populations that may prevent or reverse the evolution of divergent phenotypes. In some cases, speciation appears to be correlated with higher rates of phenotypic and molecular evolution.
Long-term trends may result from individual selection, species selection, or constraints that bias the direction of evolution between character states. Driven trends, whereby the entire frequency distribution of a character among species in a clade shifts in a consistent direction over time, are distinguished from passive trends, in which variation among species (and therefore the mean of the clade) expands from an ancestral state that is located near a boundary (such as a minimal body size).
Probably no feature exhibits a trend common to all clades in the tree of life. Features such as genome size and structural complexity display passive trends, in that the maximum has increased since very early in evolutionary history, but such changes have been inconsistent among lineages. There is no clear evidence of trends in measures of adaptedness, such as the longevity of species or higher taxa, in geological time.
Certain aspects of evolution are predictable, especially in the short term, and may be manifested by convergent evolution. However, long evolutionary histories are probably contingent: that is, particular evolutionary events would have differed, or would not have occurred, if any of a great many previous events had been different. Unique events such as the emergence of human intelligence may have been highly contingent and improbable.
If “progress” implies movement toward a goal, then there can be no progress in evolution. If “progress” implies betterment or improvement, improvement can be seen only relative to a species’ environment or way of life.