No rattlesnakes in the rainforests: reply to Gosling and Bush.

Gosling & Bush (2005) criticize our interpretation (Wüster et al. 2005) of a phylogeographic analysis of the Neotropical rattlesnake (Crotalus durissus) as supporting a hypothesis of Pleistocene fragmentation of the Amazonian rainforests. We thank these two authors for the opportunity to further discuss issues concerning the ecological and biogeograph-ical history of the Amazon Basin. Gosling and Bush quite rightly comment on the diversity of the ecosystems that are often grouped together as 'Amazonian rainforest', and also note the dynamic changes of community composition throughout the Pleistocene. They go on to comment that these two factors make any extrapolation from present-day distribution patterns difficult, and question the use of C. durissus to determine the possible past extent of rainforest fragmentation. We reject these doubts, and maintain that the presence of C. durissus on both sides of the Amazon Basin does represent evidence of profound changes in the distribution of rainforests in the Pleistocene and cannot be explained by relatively minor changes in rainforest community composition. Despite the dynamic nature of community composition in Neotropical forests, a typical rainforest physiognomy can be defined and can be traced back through the vegeta-tional history of South America to at least the Eocene (Burnham & Johnson 2004). Moreover, the many different identifiable Amazonian forest types all have one factor in common: C. durissus does not occur in them. The few available records of C. durissus in forest zones (e.g. Beebe 1946) are all of occasional specimens from ecotonal zones close to established populations in open formations. Kartabo (6 ° 23 ′ N, 58 ° 41 ′ W), the site of Beebe's records, is situated at the edge of the documented distribution of C. durissus in Guyana (Campbell & Lamar 2004). Numerous intensive herpetofaunal surveys in many parts of the Amazon and adjacent moist forests, including many different forest types, have consistently failed to locate C. durissus in rainforest areas (e. Altogether, this suggests that C. durissus is a species that is unable to persist in a wide range of moist forest types. We reject the suggestion that C. durissus may have occupied rainforest formations in the past as entirely unsupported. Apart from the absence of forest-dwelling populations at present, we also note that no other species of Crotalus occupies tropical rainforests (Campbell & Lamar 2004). Indeed, the closest relatives of C. durissus , such as Crotalus molossus and Crotalus basiliscus , occupy highly seasonal or …