Molecular phylogenetic studies have suggested major clades in the subtribe Aeridinae of Orchidaceae. However, systematic morphological charaters is still lacking. We carried out this study on the pollinium morphologies of subtribe Aeridinae with emphases on the aperture types and the pollinium sculpture to find out their systematic signification in light of a molecular phylogeny. The results indicate that most species have cleft type and few with porate type or none. The shapes of most pollinium are flat spheroid but not constant at genera levels. The longth scope of the pollinium is 204-2606 um and the width scope is 178-1524 um. The long and wide pollinium are easily found in more primitive groups. The short and narrow pollinium are commonly found in more evolutionary groups. Shapes and sizes of individual units of pollinium vary from uniformity to difference with few exceptions: isometrical polygons in Phalaenopsis of Phalaenopsis clade, Vanda clade and Aerides clade and irregular shape and different sizes in rest samples. Individual units of most pollinium are tetrads and individual pollen grains are easy to distinguish within the tetrad. The degree of discriminability of individual pollen grains is constant at clades level and changes following a tendency except Aerides rosa: unrecognizable in Chiloschista clade - easily distinguished in Phalaenopsis clades, Vanda, Aerides and Gastrocohilus clades - partly distinguished in Trichoglottis and Cleisostoma clade. Exine sculpture characters are consistent in each clades except Cleisostoma clade and Phalaenopsis clade. It has a somewhat clear changing tendency in consist with the molecular phylogeny with few exceptions: from negative reticulum of Phalaenopsis clade - reticulum of Chiloschista clade - levelled psilate surface of Vanda and Aerides clades – bulgy with psilate surface in Gastrocohilus and Trichoglottis clades - two types in Cleisostoma clade. Conclusionarily, these pollinium characters are not systematic but useful characters in determining major systematic groups in subtribe Aeridinae.

Molecular phylogeny of pollinium samples—5 DNA sequences including ITS, atpI-H, matK, psbA-trnH and trnL-F were downloaded from Genebank and an alternative species from the same genus was used when DNA sequences of the pollinium samples were unavailable (Appendix 2). Sequence alignments were made with Clustal X (Thompson et al. 1997) and refined manually with BioEdit (Hall 1999). Phylogenetic analyses were performed under maximum parsimony (MP). Some unavailable sequences in combined data analyses were treated as missing.

Parsimony analyses were conducted in PAUP* 4.0b10 (Swofford 2002), and all characters were equally weighted and unordered. The test settings included 1000 replications of random addition sequence and heuristic search with TBR branch swapping. To estimate the clade support, heuristic searches were completed for 1000 bootstrap replicates with 10000 random sequence additions (holding one tree at each step) (Harbaugh and Baldwin 2007).

The datasets were also analyzed under Bayesian inference using MrBayes v.3.2.3 (Ronouist and Huelsenbeck 2003). The best-fit model were selected with Modeltest 3.06 (Posada and Crandall 1998). From a random starting tree, a Bayesian analysis was run for 1000000 generations with four simultaneous chains and sampling trees every 100 generations. The first 300 samples for each run were discarded as burn-in to ensure that the chains reached stationarity. Majority-rule consensus trees were constructed on those trees sampled after generation 30000.