Abstract:
In this note I want to point out a potentially concerning bias in evolutionary hypothesis testing. In
evolutionary research, deductive hypothesis testing is the most common way of achieving information
on causal relationships (QUINN & DUNHAM 1983; LOEHLE 1987). Two general classes of
hypotheses can be identified in research on trait evolution, although these may not always be explicitly
stated. Hypotheses of the first class are descriptive in that they specifically relate to how traits have
evolved. These hypotheses typically focus on evolutionary trajectories, but may also include rates of
evolutionary change or the nature of the evolutionary process itself (PAGEL 1997). Hypotheses of this
class are often tested using phylogenetic methods (e.g., SWOFFORD & MADDISON 1987;
MADDISON 1994; SCHLUTER 1995; SCHLUTER et al. 1997; PAGEL 1999), and generally make
no assumptions regarding the causes for evolutionary events. Hypotheses belonging to the second class,
on the other hand, are explanatory in that they focus on explaining why a particular evolutionary
process has occurred. These hypotheses therefore rest on an assumption that a specific course of
evolution has taken place. They are generally analysed using a fitness maximization framework, and in
contrast to descriptive hypotheses are not frequently tested using phylogenetic approaches (WADE &
KALISZ 1990).
