• Cistaceae;
  • ITS;
  • molecular dating;
  • phylogeography;
  • rbcL;
  • seed dispersal;
  • Strait of Gibraltar;
  • trnK-matK;
  • trnS-trnG


Aim  There are few biogeographical and evolutionary studies that address plant colonization and lineage origins in the Mediterranean. Cistus serves as an excellent model with which to study diaspore dispersal and distribution patterns of plants exhibiting no special long-distance dispersal mechanisms. Here we analyse the pattern of genetic variation and divergence times to infer whether the African–European disjunction of C. ladanifer L. is the result of long-distance dispersal or of vicariance events.

Location  Principally the Western Mediterranean region, with a focus on the Strait of Gibraltar.

Methods  We used DNA sequence phylogenetic approaches, based on plastid (rbcL/trnK-matK) and nuclear (ITS) sequence data sets, and the penalized likelihood method, to date the diversification of the 21 species of Cistus. Phylogenetic relationships and phylogeographical patterns in 47 populations of C. ladanifer were also analysed using two plastid DNA regions (trnS-trnG, trnK-matK). These sequence data were analysed using maximum parsimony, Bayesian inference and statistical parsimony.

Results  Dating estimates indicated divergence dates of the C. ladanifer lineage in the Pleistocene. Eight nucleotide-substitution haplotypes distributed on the European (four haplotypes) and African (five haplotypes) sides of the Strait of Gibraltar were revealed from C. ladanifer sequences. Both the haplotype network and the phylogenetic analyses depicted two main Cistus lineages distributed in both Europe and North Africa. An Iberian haplotype forms part of the North African lineage, and another haplotype distributed on both continents is related to the European lineage. Haplotype relationships with respect to outgroup sequences supported the hypothesis that the centre of genetic diversity is in northern Africa.

Main conclusions  Based on lineage divergence-time estimates and disassociation between geographical and lineage haplotype distributions, we inferred at least two intercontinental colonization events of C. ladanifer post-dating the opening of the Strait of Gibraltar (c. 5 Ma). This result supports a hypothesis of long-distance dispersal rather than a hypothesis of vicariance. We argue that, despite limited dispersal abilities, preference for disturbed habitats was integral to historical colonization after the advent of the Mediterranean climate (c. 3.2 Ma), when Cistus species diverged and became established as a dominant element in the Mediterranean scrub.