The establishment of reproductive isolation is considered to be an ultimate result of ecological speciation, but empirical evidence for the latter is limited. We hypothesized that in the absence of inter-specific gene flow and with subtle environmental differences, local adaptation will not require trade-offs in performance across environments and therefore will not result in reproductive isolation. We tested this hypothesis on four iris species, Iris atrofusca, I. atropurpurea, I. petrana and I. mariae, which have non-overlapping geographical distributions, grow naturally in different environmental conditions with respect to amount of rainfall and soil type, and possess suites of diagnostic traits that previously were assumed to result from local selection. We analyzed their genetic (AFLP) and quantitative trait divergence (by means of a common garden experiment), conducted a habitat suitability analysis, examined experimentally the effects of soil type and water availability on plant performance, and tested for postzygotic reproductive isolation using a crossing experiment. Our results supported the hypothesis that eco-geographical isolation does not necessarily lead to local adaptation or to postzygotic reproductive isolation when environmental differences are subtle. Thus, we obtained some evidence that the desert I. atrofusca was reproductively isolated from coastal I. atropurpurea, but not from desert I. mariae, although genetically the first two are more similar to each other than I. atrofusca is to I. mariae.