Phylogenetic trouble unleashed
The first part of my thesis deals with a comprehensive phylogeny of the Bromelioideae subfamily. The family Bromeliaceae is subdivided into eight subfamilies, one of them is the Bromelioideae. Phylogenetic relationships among the Bromelioideae are still poorly understood and many of the extant genera are suspected to be not monophyletic. Especially Aechmea, the largest and most polymorphic genus constitutes many questions and the genus was used as a depot for taxonomically problematic species. The phylogenetic study presented here is the most comprehensive one so far, covering about half of the known species (434 of 965, Table 1) of Bromelioideae. The phylogeny was generated using plastid (atpB-rbcL, matK, rps16, ycf1_1, ycf1_6) and nuclear (AGT1_exon, ETS, G3PDH, PHYC, RPB2) genetic markers. The markers were analysed individually as well as combined using maximum likelihood and Bayesian analysis. The comparison of plastid vs. nuclear data revealed significant differences which were discussed in detail and hypothesised to indicate hybridisation in certain lineages. Nevertheless, the combination of both datasets increased the overall resolution of the phylogeny and was used to discuss the results in the light of previous studies. The entire phylogeny was divided into 32 groups for discussion. These groups represent potential genera or starting points for further studies in order to reorganise the polyphyletic genera of Bromelioideae into monophyletic lineages. Many extant genera of the eu-Bromelioideae were found to be not monophyletic. Monophyly was observed for the genera Acanthostachys, Billbergia, Cryptanthus, Disteganthus, Hoplocrypanthus, Lapanthus, Orthocryptanthus, Orthophytum, Rokautskyia, Ronnbergia, Sincoraea, Wittmackia and the monotypic ones (Deinacanthon, Eduandrea, Fascicularia, Hohenbergiopsis, Pseudananas). The genus concept proposed by Smith and Downs (1979) is therefore rejected, as well as the taxonomic utility of petal appendages, which were mainly used to delimit genera. In summary, this study and recent studies highlighted other morphological characters (e.g. pollen morphology, stigma type) as much more informative. However, no single character should be used to delimit genera and combinations of relevant characters are required. Even the petal appendages can pose a taxonomical important character at certain taxonomic level.
The combination of biogeography and phylogeny revealed that species of some groups which co- occur in a biome or region are also phylogenetically closely related. These groups were not recognised before because the misinterpretation of homoplastic characters led to wrong taxonomical conclusion. For example, the recent re-organisation of the Cryptanthoid group and the re-establishment of Wittmackia with the former Hohenbergia subgen. Wittmackiopsis species highlighted, among other characters, the importance of biogeography. Another case is the subgenus Neoregelia subgen. Hylaeaicum which is geographically and phylogenetically separated from the Nidularioid group and therefore has to be excluded.
5
The large phylogeny presented here gives evidence for multiple invasions of the Brazilian biomes (Amazon Forest, Atlantic Forest, Cerrado, Caatinga) as well as of Central America and the Greater Antilles. It is important to note that the phylogeny is lacking resolution in the deeper nodes. Confident assumptions are therefore hindered and the historical biogeography of Bromelioideae remains cryptic. Anyway, the Atlantic Forest is nowadays the diversity hotspot of the core Bromelioideae and critically endangered. Extensive conservation efforts are required to protect the diverse flora, including the bromeliads.
The genetic markers used so far in bromeliad phylogenies provided only limited variation resulting in often unresolved complexes. The search for additional suitable genetic markers in bromelioid phylogenies yielded the nuclear marker AGT1. The amplified fragment consists of one well conserved exon region as well as a highly variable intron. The intron was too variable for aligning it across the entire bromelioid set. On the other hand, the intron provides relevant information for inferring phylogenies of closely related species groups (e.g. in Ananas, Cryptanthoid group). Furthermore, AGT1 is proposed as a genetic barcode in Bromelioideae because it poses much more information then the commonly used ones (e.g. matK).
Does size matter?
The second part of this thesis deals with the genome size evolution within the family Bromeliaceae. Samples from seven subfamilies were screened with the emphasis on the subfamily Bromelioideae. The data were combined with data from literature and the observed patterns were discussed in relation to known phenomena (e.g. correlations to environment and life form). In the second sub-chapter I have chosen the species Tillandsia usneoides to study the intraspecific genome size variation in combination with morphology and biogeography. Genome size and base composition were measured using the flow cytometry technique.
Bromeliaceae comprises mostly diploid species with predominantly 50 small chromosomes (2n), small genome sizes (0.59-4.11 pg) and normal GC content (36.46-42.21 %) compared to other families. Polyploidy was observed so far in the subfamilies Bromelioideae, Tillandsioideae and Pitcairnioideae. Triploids, tetraploids and potential hexaploids were identified. The genera show significant differences in holoploid genome size and base composition throughout the entire family. GC content is weakly positively correlated with genome size. Significant intraspecific genome size variation has been observed, including polyploidization, but no endopolyploidy and no variation in dioecious species. Within the subfamily Bromelioideae, the observed genome size between the early diverging lineages and the core Bromelioideae supports this division. The differences are due to a higher proportion of polyploids in the early diverging lineages and a significant higher
6
GC content in the core Bromelioideae. Both groups differ in their life strategies and occupy principally different habitats with corresponding morphological adaptations. Hence, the early diverging lineages are predominantly terrestrial and xeromorphic. In contrast, the prevailing epiphytic core Bromelioideae are characterised by a tank habit and mostly adapted to more humid environments. Across the family and the subfamily Bromelioideae in particular, significant genome size differences between the different life forms have been observed, but no correlation to biomes within Brazil.
Tillandsia usneoides is the most widely distributed species of the family Bromeliaceae. It ranges from the southeastern United States to Argentina and Chile. Tillandsia usneoides grows epiphytic and is dispersed by seeds as well as by fragments of the plant. Within the species striking morphological differences can be observed as far as size characters are concerned. Morphotypes have shown to be stable in cultivation while growing under the same conditions. In order to investigate possible reasons for the variation the relative genome size of 75 specimens covering the whole distribution range was measured and combined with morphological, distribution and climatic data. Significant variation in the relative genome size corresponded to the morphological differences and reflected the north-south distribution gradient. Genome size and morphotypes showed a positive correlation, as well as with the mean temperature of the driest and coldest quarter and the minimal temperature of the coldest month.