Abstract—A species' habitat preference is intcrpretablc both as a response to present‐day conditions and as a result of evolutionary response to historical conditions. The taxon cycle and taxon pulse have been proposed as hypotheses that allow prediction of patterns of habitat specialization within a lineage. Both are based on common assumptions: (1) habitat specialization is largely irreversible in a lineage, (2) ecological specializations arise in a center of origin, and (3) dispersal events leading to current distributions can be ascertained. Eight taxa of Antillean, Mexican, and Central American Carabidae, for which cladistic hypotheses of relationship have been proposed, are used to test the generality of the taxon cycle and pulse. The patterns of habitat utilization predicted by the taxon cycle and taxon pulse hypotheses are tested by comparing cladistic transformations of habitat preference to randomly generated patterns of data generated under a null hypothesis. Evolutionary changes in habitat are interpreted using Camin‐Sokal coding, which assumes irreversible habitat shifts and a predetermined ecological ground state (assumptions of the taxon cycle and taxon pulse). An observed pattern is considered to demonstrate the taxon cycle or pulse when it results in an explanation of the habitat shifts that is more parsimonious than 95 % of the randomly generated patterns. Of the eight carabid groups, only one exhibits a statistically significant pattern supporting the taxon cycle and pulse. The failure of the taxon cycle and pulse as generally predictive hypotheses may be due to historical changes in climate that permit episodes of range expansion for species previously restricted to small ranges, and habitat shifts and specialization that do not progress in a linear transformation series.Habitat shifts are also analyzed using Farris optimization, resulting in the most parsimonious transformation series of habitats, subject to a predefined ordering of habitats, while allowing reversals. Significance of an observed pattern of habitat shifts under Farris optimization implies habitat constancy relative to cladogenesis, and step‐wise changes in habitat preference. Two of the eight groups exhibit significant patterns of habitat utilization under Farris optimization, indicating that vicariance of areas of like habitat has been the predominant factor generating species diversity in these groups. The other groups do not exhibit habitat constancy, suggesting rapid changes in habitat preference relative to spe