Extant and fossil Myrtaceae pollen morphology and their significance to systematics, phylogenetics, evolution, and biogeography of the family

Fossils are needed to calibrate the molecular clock used for dating phylogenies. This has led to a re-examination of the identification and phylogenetic placement of fossils used for calibration. Until recently, fossil pollen has been neglected for calibration due to a belief that pollen lacks sufficiently useful characters. Consequently, taxonomy of extant pollen has also been neglected, and this applies particularly to the pollen of Myrtaceae, which was last fully revised in the 1950s. This study surveys pollen morphology across Myrtaceae, assessing the phylogenetic signal in pollen characters and assessing suitability of fossil pollen for calibration. Fossils identified as suitable were used to calibrate a new dated molecular phylogeny, which was then applied to investigating the evolution and biogeography of Myrtaceae. A study of pollen from over 200 living taxa using Scanning Electron Microscopy (SEM) and 500 taxa using Light Microscopy (LM) found nine distinct pollen types within Myrtaceae that potentially showed a relationship to molecular tribal classification of the family. Optimising pollen characters onto a molecular phylogram, constructed from two chloroplast (matK and ndhF) and one nuclear (ITS) loci from 111 taxa, indicated the potential use of colpal morphology in diagnosing Myrtaceae pollen groups. However, exine pattern, apocolpial island presence and pollen width are homoplasious and relatively uninformative. A review of all formally described Myrtaceidites fossil species identified nine distinct morphotypes, including six that could be used to calibrate molecular dating. One new morphospecies, Myrtaceidites leptospermoides, was described for fossil pollen with syncolpate colpi and a granulate exine pattern. The fit of 26 pollen fossils onto the new molecular phylogeny was measured using parsimony optimisation of characters from extant Myrtaceae pollen. Eight Myrtaceidites fossils were identified as appropriate for calibration based on their placements on the tree. These fossils were used to calibrate a Bayesian phylogenetic analysis and this led to older estimates than have been previously found for the crown ages of tribes such as Eucalypteae and Myrteae, showing the potential of pollen for calibration. A Bayesian phylogenetic analysis using a partitioned relaxed clock and 12 fossil calibrations was used to test biogeographic hypotheses by optimizing the modern geographic location of extant taxa using parsimony reconstruction. Of the 22 tested disjunct sister-groups, up to five could possibly be explained by vicariance, four likely resulted from overland dispersal via new land connections, and 13 were too young for vicariance by continental drift and therefore inferred to be the result of long distance dispersal and establishment (LDDE) events. Holocene fossil pollen from Bega Swamp, New South Wales, Australia, were compared with pollen of twenty-five extant Myrtaceae species from the surrounding Bega Swamp area using visual judgement and a Lucid key constructed for the purpose. It was found that Eucalyptus pauciflora, together with other Eucalyptus species, has occurred in the Bega Swamp area for over 12,500 years. However, interpreting past vegetation composition using Eucalyptus is problematic because extant taxonomic (e.g. subgenus Symphyomyrtus) or ecological (e.g. alpine or wet forest) groups do not form pollen with distinct pollen types and therefore fossils cannot be confidently assigned to any extant Eucalyptus group