Testing the hypothesis that biological modularity is shaped by adaptation: Xylem in the Bursera simaruba clade of tropical trees.

The study of modularity allows recognition of suites of character covariation that potentially diagnose units of evolutionary change. One prominent perspective predicts that natural selection should forge developmental units that maximize mutual functional independence. We examined the module-function relation using secondary xylem (wood) in a clade of tropical trees as a study system. Traditionally, the three main cell types in wood (vessels, fibers, and parenchyma) have respectively been associated with three functions (conduction, mechanical support, and storage). We collected samples from nine species of the simaruba clade of Bursera at fifteen sites and measured thirteen anatomical variables that have traditionally been regarded as reflecting the distinct functions of these cell types. If there are indeed (semi) independently evolving modules associated with functions, and cell types really are associated with these functions, then we should observe greater association between traits within cell types than between traits from different cell types. To map these associations, we calculated correlation coefficients among anatomical variables and identified modules using cluster and factor analysis. Our results were only partially congruent with expectations, with associations between characters of different cell types common. These results suggest causes of covariation, some involving selected function as predicted, but also highlighting the tradeoffs and shared developmental pathways limiting the evolutionary independence of some cell types in the secondary xylem. The evolution of diversity across the simaruba clade appears to have required only limited independence between parts.