5  Conclusions

1. There are 39 species referable to Pleurothallis in Cuba, although 3 taxa with the atypical pollinium number of 8 should be removed to a separate genus. The endemic portion is ~50 %.
2. In Cuba, open and humid places like gallery forests and spatially diverse (sub‑)montane rainforests are the habitats with the highest α-diversity. Petrologic features play an important role, too, since almost all endemics are associated with one single rock type, of which the most important is serpentine. The highest altitudinal concentration of species can be found in the colline to submontane belt, i.e. 300-1200 m a.s.l., with the peak as low as 600 m a.s.l. Annual precipitation of more than 1200 mm/a is a prerequisite, but an additional high amount of unmeasured humidity and precipitation is provided in all habitats.
3. Horizontal distribution in Cuba correlates with the ecological amplitude, i.e. endemics are the most stenoecious taxa, while wide spread species are generally euryoecious. The latter correlation seems to be often extended into reproduction biology, since autogamy seems to be much more prevalent in wide spread than in an endemic taxa. Yet, endemic taxa are geographically not as restricted as assumed earlier.
4. Morphological, palynological and molecular data shows, that the Cuban taxa are not monophyletic.
5. There is much indication for a mainly post-Pleistocene colonisation and radiation in the Arc.
6. The Antillean flora of Pleurothallis has its origin almost exclusively in Central America. Floristic enrichment of the Greater Antilles can be explained best by late season hurricanes which originate in the S Caribbean Sea and pass frequently over Central America and the W Greater Antilles.
7. Besides mountainous Central America, the species of Pleurothallis have a secondary centre of diversity in the East Cuba – Jamaica – West Hispaniola triangle. This is probably based on: a) geological, hence, pollinator diversity; b) intensive hurricane activity in this area, causing frequent translocation; c) favourable climatic conditions due to the relief and high precipitation.
8. Classic dispersal, with subsequent anagenesis of the founder population is the principal way of speciation in these orchids in the Greater Antilles. Dispersal events can proceed from the mainland (25% of the Antillean species) or from within the archipelago (50%). In any case it results in allopatric distribution. [page 159↓]Besides, sympatric speciation and hybridisation occur, but seem to play a minor role. About 25% of the taxa did not enter speciation in the process of dispersal.
9. Only three lineages have further radiated, i.e. entered cladogenesis, in the West Indies. They represent ~50% of the Antillean taxa.
10. Speciation processes led to geographical and ecological vicariance in most cases. They were apparently triggered by founder events, i.e. the establishment of small extra-zonal populations under new environmental (pollinator?) conditions.

With the treatment of the genus fro the Flora de Cuba, a substantial part of the Cuban orchid flora has been liberated of century-old errors and misconceptions. About 1/6 of the Cuban orchid species were revised orientated to the standard currently employed in modern Floras and revisions. Most of the Antillean endemics have been illustrated for the first time, providing easy visual access for the user. This is especially important in the case of the spatially shaped orchid flowers which are difficult to describe. Descriptions and illustrations have been completed with data on distribution and ecological preferences. With this information one of the most diverse neotropical plant groups on the largest Antillean island has been characterised in detail. The collected data form the base for the phytogeographical and evolutionary studies of the present thesis as well as for future research.

Phylogeographical aspects of an Antillean orchidaceous group have been studied for the first time now, tackling some of the most essential questions of Antillean biogeography, i.e. how the Arc was colonised and which conditions caused the intensive speciation in that region. While morphologically based phylogenetic trees have been used already earlier to discuss modes of migration and speciation in the Caribbean, the present paper combines molecular data with most recently collected information on horizontal, vertical and ecological distribution. Genetic methods had not been employed before to study the genesis of the Caribbean orchid floras. Thus, in the absence of reliable phylogenetic and high resolution phytogeographical data, studies on orchid speciation have focused especially on gene flow and other characteristics at the population level. The present results suggest a rather simple type of speciation in pleurothallid orchids, i.e. dispersal with subsequent exposure to new environmental conditions, combined with a specific genetic evolution like selfing. Founder events, i.e. allopatric speciation, account apparently for most speciation processes in Antillean Pleurothallis species.

Due to the limited sample size and the absence of material from other islands in the molecular study, future research should focus in the first line on the inclusion of more Antillean endemics in the molecular matrix. Moreover, more taxa should be added from the mainland, to sufficiently represent the source area of Caribbean Pleurothallis. Apart [page 160↓]from this fine-tuning of the present method some of the hypotheses established in this study should be studied in the future to provide further details of orchidaceous speciation on islands.

1. Is the geological restriction of Pleurothallis on the other Greater Antilles islands as widespread as on Cuba? Which petrologic types play a major role?
2. Does selfing represent the initial mode of propagation within new orchid populations? How does this process influence genetic variability?
3. Which are the pollinators of Pleurothallis in the Antilles? Do they show a similar pattern of distribution as their respective orchid species, i.e. can geological traits be synonymised with pollinator distribution?

It becomes clear from the questions suggested that genetics will play an important role in future research. In this context it should be emphasised that molecular tools should not be detached from other aspects, like morphology or ecology. It is frustrating to see the vast number of molecular papers compared with the small amount of information that has been drawn from them considering evolution of morphology, ecological adaptation etc. There is still a tremendous treasure of information that awaits recovery simply by linking the new phylogenetic trees with phytogeographical, ecological and many other patterns. Another important point is the importance of this non-molecular information to test the reliability of molecular data, i.e. the feedback of morphology, anatomy, distribution etc. to genetics. To check molecular trees with the help of other information is especially important where large taxa have been extremely undersampled in phylogenetic studies.

As was pointed out earlier in this study, for the Pleurothallidinae we have substantial data now, concerning macro- and micromorphology, anatomy as well as genetics. The subtribe should now be one of the most thoroughly studied orchid groups. Hopefully, these data will be merged into a new classification, providing a system that reflects phylogenetics and evolution much better than any of the former constructions that were based mainly on but one data set.

As a probably more topical issue, the results of this thesis which concern ecology, phytogeography and speciation of Antillean Pleurothallis represent important sources of infomation in conservation matters. Orchids are traditionally used as a political argument in this connection. In the Caribbean, which forms one of the 10 hot spots not only due to its biodiversity, but owing to the tremendous loss of prime forests, too, these biological data represent perhaps the most valuable result of this study.