Paranisitra longipes Chopard, 1925

Several hypotheses were previously proposed about the evolution of acoustic devices and signals, using the phylogeny previously reconstructed with the morphological char- acters only (Robillard and Desutter-Grandcolas, 2004a,b,c).

In molecular and combined analyses, the mute species Swezwilderia bryani and Paranisitra longipes are never sister clades; despite the unstable position of Swezwilderia, the hypothesis of multiple and independent losses of acoustic communication (Alexander, 1962; Otte, 1992) is corroborated here in Eneopterinae. These losses however appear as direct evolutionary events, involving no intermediate state between functional acoustic devices and their loss. These results thus partly invalidate previous hypotheses about the loss of acoustic communication in crickets, generally considered as a sequential, regressive process (Alexander, 1962; Otte, 1992).

The analysis of the patterns of evolution of male fore-wing venation revealed that the diversiWcation of acoustic devices originates from two processes, a continuous and regular modification process, responsible for slight venation changes; and an irregular, more intense punctuated process, allowing the emergence of divergent venations in Lebinthini, Nisitrus, Salmanites and in the clade (Ponca–Ligypterus). Among these clades, Lebinthini, Nisitrus and (Ponca–Ligypterus) are clearly supported by molecular and/or combined analyses. As mentioned above, the monophyly of Salmanites is not clearly recovered by combined analysis, though it is unambiguous on morphological basis and in the separate analyses of 12S, 16S and cytb. All the clades characterised by massive changes of venation and supporting the punctuated evolution hypothesis are well supported by our results. Previous conclusions on the evo- lution of acoustic devices in Eneopterinae are thus corroborated.

Concerning acoustic signals, a hypothesis of adaptive radiation for high-frequency calling in Lebinthini has been derived from the analysis of spectral characteristic of call- ing songs (Robillard and Desutter-Grandcolas, 2004b). This hypothesis is supported by our results: Lebinthini species are always grouped together, and even if Swezwilderia may jump into this clade, it would not invalidate the radia- tion hypothesis, because it is mute.

Further analyses of spectral components of the calling song also revealed a case of convergence involving three diVerent ways to produce high frequencies (Robillard and Desutter-Grandcolas, submit.), which occur independently in Eneoptera guyanensis, Cardiodactylus and (Agnotecous– Lebinthus). The topologies obtained here support the inde- pendent origins of these acoustic novelties: despite its unstable position in the tree, Eneoptera is never the sister group of Lebinthini, and Cardiodactylus and (Agnotecous– Lebinthus) are most often monophyletic and sister clades, as in the morphological phylogeny. [1]

Références