Divergence and biogeography of the recently evolved Macaronesian red Festuca (Gramineae) species inferred from coalescence‐based analyses

Abstract

Studying the biogeography and the phylogeography of the endemic Macaronesian red Festuca species (Loliinae, Poaceae) is of prime interest in understanding the speciation and colonization patterns of recently evolved groups in oceanic archipelagos. Coalescence-based analyses of plastid trnLF sequences were employed to estimate evolutionary parameters and to test different species-history scenarios that model the pattern of species divergence. Bayesian IM estimates of species divergence times suggested that ancestral lineages of diploid Macaronesian and Iberian red fescues could have diverged between 1.2 and 1.57 Ma. When empirical data were compared to coalescence-based simulated distributions of discordance and p-distance statistics, two species-history models were chosen in which the first branching lineage derived in Canarian Festuca agustinii. Its sister lineage could have involved a recent polytomy leading to the Madeiran Festuca jubata, the Azorean Festuca francoi + Festuca petraea and the continental Festuca rivularis lineages (Canarian model) or the sequential branching of lineages leading to F. jubata and finally to the sister clades of F.rivularis and F.francoi + F.petraea (Sequential model). Nested clade phylogeographic analysis (NCPA) and a first adapted host-parasite co-evolutionary ParaFit method were used to detect the phylogeographic signal. NCPA inferred long-distance colonizations for the entire diploid red Festuca complex, but allopatric-fragmentation and isolation-by-distance (IBD) patterns were inferred within archipelagos. In addition, the ParaFit method suggested a generalized pattern of a stepping-stone model at all hierarchical levels. Maximum-likelihood-based dispersal-extinction-cladogenesis (DEC) models were superimposed on the Sequential model species tree. The three-independent-colonization (3IC) model was the best supported biogeographic scenario, concurring with previous analysis based on multilocus AFLP data.