“One defining feature of cacti is having clusters of spines. Numerous plants have spines of course, but in cacti, spines occur in clusters in the axil of leaves, even though the leaves are usually microscopic. Most cactus morphologists have concluded that cactus spines are either modified leaves or modified bud scales (the difference is inconsequential because bud scales themselves are modified leaves). The leaf-nature of spines is certainly understandable from the point of view of location: spine primordia look just like leaf primordia and are produced at a location where we would expect leaf primordia – at the base of the axillary bud’s shoot apical meristem.
“Evolution appears to have been more complex than would be expected: mature cactus spines do not contain any of the cells or tissues characteristic of leaves, and conversely leaves lack all features characteristic of spines. The two organs have little in common other than developing from leaf primordia. Spines consist of just a core of fibers surrounded by sclereid-like epidermis cells. They have no stomata, no guard cells, no mesophyll parenchyma, no xylem, no phloem. When mature, all cells in a spine are dead, and even when the spine is still growing it has living cells only at its base. Cactus leaves on the other hand … have parenchymatous epidermis cells, guard cells, spongy mesophyll, chlorenchyma, xylem and phloem. So the evolutionary conversion of cactus leaves into spines did not involve a mere reduction of the lamina and then further reduction of midrib and petiole, it instead involved the suppression of all leaf-cell type genes and activation of genes that control formation of fibers, the deposition and lignification of secondary walls, and then programmed cell death. These fiber morphogenesis genes are not activated in any cactus leaf (none at all has fibers), but they are activated of course in the development of wood. It would appear that after an axillary bud apical meristem initiates spine primordia, most leaf genes remain suppressed and instead wood fiber genes are activated. This does not involve all wood genes because vessels are never produced in the spines, just wood fibers. This would be a type of homeotic evolution.”
Mauseth, J. D. 1982. Development and ultrastructure of extrafloral nectaries in Ancistrocactus scheeri (Cactaceae). Botanical Gazette 143: 273 – 277.