When we are talking about reproductive isolation, what emerging in our minds are that RI between species appears to be an innate property,as proposed by the classic Biological definition of RI that whether two species are a same species or not, determines by the results of their crossed offsprings. However in empirical tests of reproductive isolation on plant species, RI itself was revealed by evaluating interactions (across space and time) of different RI stages,both pre-zygotic and post-zygotic isolation contributed to the total isolation. Usually plants are not locomotive, but use pollen vectors for reproductive interaction between conspecific and heterospecific individuals. The mobility of the pollen vectors are also in some degree finite----even if the vectors are bees (even bees could not guarentee a flight coverage of area with radius of too many kilometers),these would make a large enough plant community has heterogenerous RI barrier with other species across different sites within.
For the same species, some populations of this species may distribute in some places where RI was low between this species and other local species, these palces could be RI hotspots, beacuse selection would favor those traits reducing or avoiding reproductive interference, and selection is urgently on the way. In such a situation, biotic pre-zygotic isolation stages may be the first to be impacted, and the effects would accumulating until affecting the post-zygotic stages; while in some other places, reproductive interferences between this species and other species could be not-so-fierce, reproductive isolation index could be high,selection strength would thus be lower,these palces are henceforth RI coldspots.
I think a geographic mosaic pattern could help to better understand the evolutionary histories and ecological interactions of reproductive isolation.