Edaphic “islands” have long been thought to have reduced gene flow among them, but how much is gene flow restricted, and are there asymmetries between seed and pollen dispersal? Using DNA sequence data in exemplar taxa from the gypsophile flora of the Chihuahuan Desert, we demonstrate that seed movement is often very limited, even among relatively close exposures of gypsum. Likewise, there is some evidence for persistently low population sizes from plastid data. However, pollen flow seems to be more variable among taxa, with some evidence that pollination syndrome affects nuclear gene flow.