The three main lineages of non-marine ostracoda (here represented by Darwinula stevensoni, Limnocythere inopinata and Eucypris virens) diverged between 350 and 450 My ago and developed fundamentally different evolutionary strategies. Analysis of distribution of reproductive modes in Europe has shown that Cyprididae are the only supra-generic category that consistently shows geographic parthenogenesis; other groups either have reproductive systems not geographically segregated (Candonidae, Cytheroidea) or are ancient asexuals (Darwinuloidea). The classical explanation of geographic parthenogenesis, a faster recolonisation of parthenogens after retreating glaciations with subsequent insufficient competitive superiority of bisexuals cannot fully be invalidated, but metapopulation models, competition experiments and analyses of lower Holocene ostracod assemblages provide support for alternative hypotheses. Eucypris virens and L. inopinata both consist of clusters of specialized clones, with evidence for at least seasonal segregation, whereas D. stevensoni has a general-purpose genotype. As some of these specialized clones are also morphologically recognizable (especially in Limnocythere), this potentially constitutes a refined method of palaeoenvironmental reconstruction. The different strategies are also exemplified by extensive analyses of genetic and morphological variability. Eucypris virens has the highest morphological and genetic variability (>200 identified clones); differences in DNA sequences in this species reach interspecific levels for other taxa. This is probably due to multiple origins of clonal lineages; a polyploid hybrid of a male with a parthenogenetic female was found. That such interactions are possible, is further supported by regular coexistence of parthenogenetic and bisexual females, by observed copulation (video) of a male with a parthenogenetic female and by the fact that in both genetic and morphometric trees, parthenogenetic and bisexual populations can cluster together and have similar variabilities. DNA and morphometric trees are congruent. Darwinula stevensoni has much lower genetic variability: only few clones were identified with allozyme analyses (and these showed clear segregation between lacustrine and riverine habitats). Whereas mitochondrial DNA (CO1) has normal evolutionary rates, the investigated nuclear DNA strand (ITS1) was identical in all investigated populations (ranging from Finland to South Africa, estimated to be 6-8 My apart). This striking absence of genetic variability, which coincides with morphological uniformity, can be due to overall low mutation rate and/or efficient DNA repair systems; it contradicts that ancient asexual lineages are necessarily doomed to extinction because of accumulation of deleterious mutations. If the presence of (an) efficient DNA repair system(s) can be demonstrated in this group, than this could have useful future applications in cancer research. The above results were published in primary journals, but an integrated view was also presented in a book (Martens, K. (ed.) (1998). Sex and parthenogenesis, the evolutionary ecology of reproductive modes in non-marine postcards, Backhuys Publ., Leiden).