Historical connectivity and environmental filtering jointly determine the freshwater fish assemblages on Taiwan and Hainan Islands of China

Biogeography of Greater Antillean freshwater fishes, with a review of competing hypotheses

In biogeography, vicariance and long-distance dispersal are often characterised as competing scenarios. However, they are related concepts, both relying on collective geological, ecological, and phylogenetic evidence. This is illustrated by freshwater fishes, which may immigrate to islands either when freshwater connections are temporarily present and later severed (vicariance), or by unusual means when ocean gaps are crossed (long-distance dispersal). Marine barriers have a strong filtering effect on freshwater fishes, limiting immigrants to those most capable of oceanic dispersal. The roles of vicariance and dispersal are debated for freshwater fishes of the Greater Antilles. We review three active hypotheses [Cretaceous vicariance, Greater Antilles-Aves Ridge (GAARlandia), long-distance dispersal] and propose long-distance dispersal to be an appropriate model due to limited support for freshwater fish use of landspans. Greater Antillean freshwater fishes have six potential source bioregions (defined from faunal similarity): Northern Gulf of México, Western Gulf of México, Maya Terrane, Chortís Block, Eastern Panamá, and Northern South America. Faunas of the Greater Antilles are composed of taxa immigrating from many of these bioregions, but there is strong compositional disharmony between island and mainland fish faunas (>90% of Antillean species are cyprinodontiforms, compared to <10% in Northern Gulf of México and Northern South America, and ≤50% elsewhere), consistent with a hypothesis of long-distance dispersal. Ancestral-area reconstruction analysis indicates there were 16 or 17 immigration events over the last 51 million years, 14 or 15 of these by cyprinodontiforms. Published divergence estimates and evidence available for each immigration event suggests they occurred at different times and by different pathways, possibly with rafts of vegetation discharged from rivers or washed to sea during storms. If so, ocean currents likely provide critical pathways for immigration when flowing from one landmass to another. On the other hand, currents create dispersal barriers when flowing perpendicularly between landmasses. In addition to high salinity tolerance, cyprinodontiforms collectively display a variety of adaptations that could enhance their ability to live with rafts (small body size, viviparity, low metabolism, amphibiousness, diapause, self-fertilisation). These adaptations likely also helped immigrants establish island populations after arrival and to persist long term thereafter. Cichlids may have used a pseudo bridge (Nicaragua Rise) to reach the Greater Antilles. Gars (Lepisosteidae) may have crossed the Straits of Florida to Cuba, a relatively short crossing that is not a barrier to gene flow for several cyprinodontiform immigrants. Indeed, widespread distributions of Quaternary migrants (Cyprinodon, Gambusia, Kryptolebias), within the Greater Antilles and among neighbouring bioregions, imply that long-distance dispersal is not necessarily inhibitory for well-adapted species, even though it appears to be virtually impossible for all other freshwater fishes.