[page 112↓]

4  Discussion

4.1 General taxonomy

The subtribe Pleurothallidinae, despite the fact that it has been neglected scientifically for a long time, is now probably one of the scientifically best known orchidaceous groups. The morphological base was laid by Luer in his comprehensive Icones Pleurothallidinarum, which, after more than a decade of thorough preparatory investigation, were started with a very detailed taxonomic system of this group (Luer 1986a). The complex genus Pleurothallis itself was reclassified in 1986 (Luer 1986b). In the following years this system was subject only to minor shifts or additions by the author. Luer delivered a detailed database of descriptions with morphological, chorological and taxonomic information on each taxon reviewed. Morphological data were complemented by anatomical characters examined by Pridgeon et al. (s. reference list on p. 4). Luer’s classification is strictly morphological, focussing above all on shape, level of connation, and position of generative organs as well as inflorescence types, and number of pollinia. Traditionally, the latter plays an important role in pleurothallid classification. Being crucial in classifying genera meant not a classification according to the number, but to maintain a consistent number within generic boundaries. However, even this is not a general rule and so we find one genus having 6 and 8 pollinia (Brachionidium Lindl.) and another with two polymorphic pairs of pollinia (Dresslerella Luer). Moreover, according to Luer (2001), Pleurothallis would now comprise both species with 2 and with 8 pollinia, since he transferred the Cuban Octomeria excentrica Luer to Pleurothallis. Palynological data of pleurothallid orchids had been scarce and only sampled randomly (Schill & Pfeiffer 1977; Zavada 1983, 1990) until a broad survey on pleurothallid pollen morphology was prepared by (Stenzel 2000). Additional taxa, especially of the underrepresented genus Pleurothallis itself, were studied at the outset of the present work (Stenzel 2004b). Finally, Pridgeon & al. (2001) published a phylogenetic analysis based on a three-gene data set, which was the base for the most devastating “taxonomic storm” since Kuntze’s time (Kuntze 1891), with more than 500 transfers and new creations (Pridgeon & Chase 2001). Not surprisingly, this radical reclassification of Luer’s system provoked a sharp rebuttal (Luer 2002) of Pridgeon’s and Chase’s interpretations. However, despite Luer’s reproach of nomenclatural inaccuracies, which in fact made 10% of Pridgeon & Chase’s new combinations invalid (Barros 2002, Pridgeon & Chase 2002), he had little to offer against the new system. So, what could have been an interesting discussion of morphological vs. molecular views, ended up in a rather sulking “leave it as it is” (Luer 2002). Other than Luer’s barren reply, who simply turned all taxonomical proposals by Pridgeon & Chase (2001) down, responses to the new system have been cautious (Hammel & al. 2002, Jost [page 113↓]& Endara 2002), principally acknowledging the main results and pointing out the various weaknesses in Pridgeon’s and Chase’s nomenclatural methodology. Thus, these authors doubt that the new system will be adopted in floristic works until the new generic concepts have been formalised and clearly articulated (Hammel & al. 2002). So far only one constructive approach has been published, resulting in a critical revision of the nomenclatural changes (Barros 2002). Ignoring the scattered nomenclatural awkwardnesses, the main weak points in Pridgeon & Chase’s taxonomic methodology shall be discussed briefly.

  1. Morphology vs. DNA – Pridgeon & al. (2001) see the great advantage of molecular data in avoiding the “homoplasy rife” when searching for relationships. However, translating phylogeny into taxonomy, they could not avoid to fall back upon Luer’s system, which is based purely on morphological features. Ironically, they continue the classification that they are just criticising. It should be stressed that more than 3/5 of the taxonomic reclassifications were based indirectly on morphology, since they are made according to Luer’s system! So, if the new classification is built on solid ground it is due to Luer’s intuition and experience, too.
    Unfortunately, morphological issues were discussed almost only if they were in concordance with molecular data, e.g. in the transfer of Myoxanthus subgen. Satyria Luer, which was not represented in the DNA matrix, but shares anatomical characters with subgenus Silenia, which was included in the molecular study. Direct morphological inconsistencies within DNA based clades were usually ignored.
  2. Absence of type material – Of the genera sensu Luer, that were taxonomically altered in any way, only 2/3 were represented by their type taxon. The same applies to his subgeneric taxa of Pleurothallis that were validly1 transferred to other genera. Section Tomentosae Luer of Pleurothallis subgen. Acianthera was reclassified without being sampled at all.
    Likewise, 5 of the 7 resurrected or otherwise changed genera had not been represented by the respective type (Acianthera, Anathallis, Andinia, Phloeophila, Stelis / Pleurothallopsis, Specklinia). In some morphologically homogenous genera (Stelis s. Luer) this may not appear to be necessary. In groups that are polymorphous or, like Phloeophila Hoehne & Schltr. (Pleurothallis subgen. Acianthera sect. Phloeophilae sensu Luer (1986b), which even proved to be [page 114↓]polyphyletic in the same study [!] the inclusion of the type species ought to be a must.
  3. Sample size – The addition of another 30+ taxa from Cuba has altered the general topology of the ITS tree presented in Pridgeon & al. (2001) only slightly. In some instances, the effect was stronger (p. 116), which illustrates the partially provisional status of the topology presented by Pridgeon & al. (2001). Cases of clear conflict between morphology and genetics should have been better excluded from the subsequent nomenclatural changes by Pridgeon & Chase.

Despite all deficiencies, neither Luer’s works nor the results delivered by Pridgeon & al. can be underestimated, since each one provides the data that the other one lacks. Together with the palynological data (Stenzel 2000, 2004b) which are completed in the present work, they form an incomparable pleurothallid database. By combining these 3 sources of phylogenetically relevant information it should be possible to create a system much more natural than those published to date.

As indicated, palynological and molecular data of the material examined in this work yielded further insight into relationships and evolutionary processes. Prior to the present study another 40 taxa mainly from the underrepresented genus Pleurothallis, which had not been examined till then, were added to the palynological survey. Together with the data by Stenzel (2000) and the present work they represent a broad base for a discussion of taxonomic and phylogenetic issues. Concerning molecular data, this study adds another 33 taxa to the 70 species of Pleurothallis studied by Pridgeon & al. (2001). In the following, the impact of both palynological and molecular results on the systems proposed by Luer and Pridgeon & Chase will be discussed. The term “original tree/topology” hereafter refers to the tree from the large ITS matrix analysis by Pridgeon & al. (2001: 2296-2298), “Cuban Analysis” refers to the MP tree of the Cuban taxa alone and “Complete Analysis” describes the analysis of the mixed matrix (Cuban taxa and those from Pridgeon & al.). The latter resulted in several trees, the strict consensus of which is presented in the following figures. To facilitate easy orientation, clades containing Cuban taxa are marked with the same letter as in the Cuban Analysis (Fig. 67, Fig. 68).


[page 115↓]

Fig. 69: Complete Analysis. First portion of one of the 10 MP trees.

(L=4326, RI=0.69, 452 parsimony-informative of 549 variable characters). Tree based on the ITS data set from the present study (in bold type) and the matrix by Pridgeon & al. (2001). Only those genera and species screened by Pridgeon & al. that neighbour Cuban taxa are shown in the tree. Branch lengths in normal type, bootstrap support percentages >50% in bold. Arrows indicate groups absent in the strict consensus tree. Numbers behind species names identify Cuban vouchers in multiple samples (e.g. Stenzel 606). Clade namings are in concordance with those in Fig. 67 and Fig. 68.


[page 116↓]

Fig. 70: Complete Analysis. Second portion of one of the 10 MP trees.

(L=4326, RI=0.69, 452 parsimony-informative of 549 variable characters). Tree based on the ITS data set from the present study (in bold type) and the matrix by Pridgeon & al. (2001). Only those genera and species screened by Pridgeon & al. that neighbour Cuban taxa are shown in the tree. Branch length in normal type, bootstrap support percentages >50% in bold. Numbers behind species names identify Cuban vouchers in multiple samples (e.g. Stenzel 1298). Arrows indicate groups absent in the strict consensus tree. Clade namings are in concordance with those in Fig. 67 and Fig. 68. This tree was computed without the incomplete sequences of P. ghiesbreghtiana. The species was inserted in the tree, based on another analysis (see discussion in the text).


[page 117↓]

Clade A – The three Cuban endemics, Pleurothallis ekmanii, P. excentrica, and P.flabelliformis’ had not been part of the study by Pridgeon & al. (2001). Their inclusion caused interesting changes in the topology. Originally, Brachionidium (6 and 8 pollinia) had fallen in the next higher subclade, along with all genera that have 4 pollinia and Myoxanthus with usually 2, rarely 4 (Stenzel 2000) pollinia. Together with the additional 33 taxa from Cuba, Brachionidium switched now to Clade A, a position already suggested by matK data (Pridgeon & al. 2001). Thus, all taxa with 8 pollinia are grouped in one and those with 4 pollinia in another clade. The second substantial change in the basal groups is the position of a member of Pleurothallis subgen. Kraenzlinella (Kuntze) Luer, P. erina cea Rchb. f., which fell sister to Brachionidium in the original tree. Besides striking differences in gross morphology, this species has only 2 pollinia, whereas the others, with the exception of Myoxanthus have 4 or even more. P. erinacea now ended up in Clade B (see next paragraph), the taxa of which share the feature of 2 pollinia.

Clade B – This group comprises what Pridgeon & Chase (2001) circumscribed as the resurrected Acianthera Scheidw. based on their molecular data. It combines Pleurothallis subgenera Acianthera, Arthrosia and Sarracenella of Luer’s system.
Pollen morphology shows Acianthera as a homogenous entity (Stenzel 2004b). The aciantheroid surface of the pollinia together with a punctate or rarely fossulato-granulate and rugulate sculpture is characteristic for all species examined. However, these characteristics are not very specific and represent rather a plesiomorphic status. Thus, similar conditions can be found among the more “primitive” Pleurothallidinae with four or eight pollinia, e.g. Restrepiella. Among the genera with two pollinia it is unique, however. Molecular data of Cuban taxa reflect the natural classification of the genus as circumscribed by Pridgeon & Chase (2001). Though clade B receives weak to moderate bootstrap percentages in the Cuban Analysis and no support in the Complete Analysis, it is maintained in the strict consensus trees of both analyses. Moreover, Pridgeon, Solano & Chase (2001) found 100% bootstrap support for their set of taxa in plastid sequences which are evolutionary more stable than ITS regions. This should be related to the early split off of the group and subsequent accumulation of multiple mutations in variable sites (ITS) which lead to a low signal/noise ratio. Sect. Tomentosae Luer of subgen Acianthera, which could not be included by Pridgeon, Solano & Chase, was represented in the present study by Pleurothallis bissei, which fell within the Acianthera clade as sister to Luer’s subgenus Antilla in the Cuban analysis. Pleurothallis subgen. Antilla Luer seems to be a natural group within Acianthera. However, it contains more taxa than actually assumed (Luer 2000). According to molecular data, Pleurothallis papulifolia, an endemic from Oriente, belongs to this subclade too. As pointed out before, P. erinacea came out in all trees as sister to the Antilla subclade while it had been placed as sister to Brachionidium in the original tree. This species belongs to subgen Kraenzlinella (Kuntze) [page 118↓]Luer, the affinities with subgen. Acianthera of which had been pointed out by Luer before (1994). Its position within Acianthera received no bootstrap support >50%. However, morphological and palynological data strongly suggest this alliance. Concerning Luer’s subclassification of subgen. Acianthera (1986b), Pridgeon & Chase were loath to adopt his division, owing to the paraphyly of many of his sections. Data of the Cuban taxa yet increase the incompatibility of Luer’s subgeneric system with molecular data.
Morphologically, Acianthera is difficult to describe. Though all species lack an annulus, this is found in other ‘primitive’ genera too and likely represents a plesiomorphism. Species of subclade Antilla Luer, along with Pleurothallis papulifolia and subgen. Kraenzlinella (P. erinacea), all of which form a clade in the Complete Analysis, have variably papillose, verrucate, or pubescent ovaries. Other tendencies within the group are leaf edges, which are decurrent to a variable degree on the ramicaul, which in turn may be surcate to strongly winged. Succulence is found in leaves and even flower parts, which are often papillose or pubescent. None of the characteristics is exclusively confined to this group, however.

Clade C – Pridgeon & Chase combined two of Luer’s subgeneric taxa in the genus Anathallis Barb. Rodr. Both are represented by Cuban plants: subgen. Acuminatia (Lindl.) Luer (1999b) by P. obovata and subgen. Specklinia sect. Muscosae Lindl. by P. sertularioides. Palynological features back the molecular findings of a close alliance between Acuminatia and Specklinia, because of the advanced gemmate sculpture found in the Cuban plants and other species (Stenzel 2004b). Molecular data from P. obovata, which was not included by Pridgeon & al. (2001), reflect this relationship too.
Pridgeon & Chase found subgen. Specklinia polyphyletic and transferred sect. Hymenodanthae Barb. Rodr. and sect. Muscosae en bloc to the resurrected genus Specklinia Lindl. (see further down) and Anathallis resp. However, the monophyly of Luer’s two sections is doubtful. His morphological distinction between sect. Muscosae and sect. Hymenodanthae is extremely blurred and virtually every characteristic of the one can be found in the other too. The mixed concept of the two sections is evident if we compare P. sertularioides (Muscosae) and P. spiculifera Lindl (Hymenodanthae). The two species are not only morphologically similar but share the same highly advanced gemmate sculpture type with Luer’s subgen. Acuminatia. The close relationship becomes clear when the two species are examined in detail. It turns out that they are merely miniature versions of many species in subgen. Acuminatia, among them Pleurothallis obovata! Ironically, this subgenus had been earlier treated as section of subgen. Specklinia (Luer 1986b). On the other hand, there are species treated under sect. Muscosae by Luer and transferred without molecular data by Pridgeon & Chase to Anathallis, which do not show the typical gemmate sculpture but a levelled surface which in turn can be found in species of sect. Hymenodanthae (Stenzel 2004b: P. fuegii Rchb. f.). In summary, the species [page 119↓]treated as Pleurothallis subgen. Specklinia sect. Muscosae by Luer and transferred to Anathallis sensu Pridgeon & Chase are not monophyletic.

Clade D – P. nummularia had been treated in section Phloeophilae (Hoehne & Schltr.) Luer of subgen. Acianthera by Luer (1986b). In the present study, it fell sister to the morphologically similar P. peperomioides. The two form a clade with the morphologically distinct genera Ophidion Luer and Luerella Braas, which led Pridgeon & Chase to raise sect. Phleophilae to generic status. In the absence of palynological data the nomenclatural changes were drawn only on the base of molecular affinities. However, the grouping with Luerella is only supported by the trnL bootstrap test, and the inclusion of Ophidion is not backed by any of the bootstrap trees with values >50% (ITS, matK, trnL). Moreover, another member of sect. Phloeophilae, P. raduliglossa, ended up within clade B (Acianthera) in Pridgeon’s study, indicating polyphyly in Luer’s concept of Phloeophilae. Unfortunately, Pridgeon & al. do not indicate the source of their material. Thus, it was not possible to check for an erroneous determination. The genus Phloeophila sensu Pridgeon & Chase (2001) remains therefore a rather questionable taxon.

Clade E – One of the most radical taxonomic changes was the tremendous expansion of Stelis Sw. by Pridgeon & Chase. Several of Luer’s subgenera of Pleurothallis (1986b) formed a clade with the genus Stelis s.str. itself in the study by Pridgeon & al. Two of these subgenera have representatives in Cuba, Pleurothallis domingensis (subgen. Crocodeilanthe (Rchb. f. & Warsz.) Luer) and Pleurothallis racemiflora (Sw.) Lindl. (subgen. Dracontia Luer). Palynological examinations revealed a great uniformity within the broadened generic limits drawn by Pridgeon & Chase. The prevalent tendency is a reduction of the sporoderm towards octomerioid conditions (Stenzel 2004b), a pattern which can be found in the two Cuban species too. P. domingensis and P. racemiflora are situated in rather isolated positions. The latter forms a clade with other members of subgen. Dracontia (P. cobanensis, P. powellii, P. tuerckheimii), the inner structure of which is not supported by the bootstrap test, however. P. domingensis is found with other members of subgen. Crocodeilanthe as sister to Stelis. The inner structure of this clade is not robust either, however the group itself is moderately supported by the bootstrap test (81%).

Clade F – This clade contains the type of genus Pleurothallis, P. ruscifolia (Jacq.) R. Br. The genus as circumscribed sensu Luer 1986b) comprises a great range of discordant palynological patterns suggesting the grouping of many unrelated taxa (Stenzel 2000). Those taxa assembled in Pleurothallis s.str. (Pridgeon & Chase 2001), on the contrary, show an uniform pollen morphology with a levelled surface and psilate to punctate sculptures (Stenzel 2004b). Palynological data was available for most of the subgenera proposed by Luer and sequenced by Pridgeon, Solano & Chase. The concordant pollen [page 120↓]morphology of those subsections of sect. Pleurothallis (sensu Luer) which could not be included by Pridgeon & al. backed the concept of a narrowed definition of Pleurothallis. Yet, subgenera Pleurobotryum (Barb. Rodr.) Luer and Kraenzlinella, which despite absent or insufficient molecular data, had been included in Pleurothallis by Pridgeon & Chase (2001) show a closer relationship with Acianthera instead. Representatives of both taxa have the typical combination of triangular pollinia without caudicles and with a punctate sculpture.
The two Cuban species, which had been classified in subgen. Pleurothallis sect. Pleurothallis by Luer (1986b), P. ruscifolia and P. pruinosa, were grouped with the two accessions of P. ruscifolia separated by P. pruinosa. Similar discordant topologies can be found in the next clade and will be discussed later (p. 124).
The third species that fell in the Pleurothallis s.str. clade, P. ghiesbreghtiana 2, could be sequenced only partially (ITS1). The inclusion of the incomplete sequence had led to a number of distortions in the gene trees as well as substantial drops in bootstrap values, due to the loss of phylogenetic information of ITS2 and apparent insufficiencies in the MP algorithm (Pridgeon, pers. commun.). P. ghiesbreghtiana was analysed separately in the clade then and added to the tree afterwards. Yet, its position remains questionable for it doesn’t match several autapomorphies of Pleurothallis s.str., e.g. apical anthers and a lepanthoid pollen surface and sculpture. Instead, palynological data strongly suggest an affiliation with the neighbouring clade E, i.e. Stelis s.l. This idea was recently confirmed by the work of R. Solano (pers. commun.). Moreover, a treatment of P. ghiesbreghtiana in Anathallis (Pridgeon & Chase 2001) is not only irrelevant as indicated by molecular data but is refuted by pollen morphology, too. Species of Anathallis show a distinct gemmate pattern, not found in any of the accessions of P. ghiesbreghtiana.

Clade G – This group comprises no less than 14 Cuban taxa, most of which, if treated at all, have been accommodated in the subgenus Specklinia in various sections (Luer 1986b). The addition of the Cuban species referable to this subgenus more than doubled the sample size that was employed by Pridgeon & al. (2001). However, compared with the number of species attributable it is still hopelessly underrepresented. This may account for the low bootstrap values received. For the respective sections of subgen. Specklinia alone, Luer enumerated more than 70 species, a list that still lacks many Antillean taxa and is probably completely outdated concerning its volume.
Still, the inclusion of the Cuban taxa had a striking effect on the original topology in some cases. Morphologically puzzling combinations in the original topology, as [page 121↓] Acostaea~Pleurothallis setosa, were rearranged in a more natural pattern now. The tie of Acostaea to the P.-grobyi-subclade had been suggested by matK sequences (Pridgeon & al. 2001). The subdivision of sect. Hymenodanthae, based on the congested inflorescence, which was acknowledged even by Luer as a mere “key character”, is unlikely to reflect true relationships and a condensed rachis has apparently evolved several times in different lineages (brighamii, condylata, corniculata 3, fulgens, trichyphis). It should be mentioned, however, that all examined taxa with condensed inflorescences except P. trichyphis show the same punctate to granulate sculpture, so that the ladderized appearance of this sculpture type along the distal spine of Clade G (tribuloides through brighamii) might still be a product of undersampling as shown by the inclusion of the Cuban taxa. Concerning the unexpected “unrelatedness” of different accessions of the same species as seen in P. brighamii, see the discussion on p. 125.
With the addition of pollen morphology, there are three groups of concordant elements.
1) In the morphological homogenous aristata-clade (subgen. Specklinia sect. Muscariae Luer), relationships are neatly mirrored by pollen morphology. Two groups may be distinguished, one comprising aristata, helenae and setosa (not from Cuba) with obovate (flattened turbinate) pollinia and an almost aciantheroid pattern, the other unifying the endemic taxa llamachoi, longilabris, and obliquipetala, with clavate pollinia and psilate tetrads that show a tendency to fuse their edges. P. mucronata shows intermediate features which is reflected by a low bootstrap value. 2) The P.-grobyi-subclade (including Acostaea) comprises typical representatives of sect. Hymenodanthae subsect. Longicaulae (Barb. Rodr.) Luer with elongate racemes and a reticulate to gemmate octomerioid pollen morphology. Even the shape of the pollinia is very stable (lentiform with abruptly narrowed caudicles). Consistency of pollen characteristics was found in other species of this group too (Stenzel 2004b). However, some taxa placed by Luer in this subsection are clearly related to Clade C (Anathallis s. Pridgeon & Chase 2001) with which they share the advanced gemmate sculpture (Stenzel 2004b: P. spiculifera Lindl.). 3) The P.-endotrachys-subclade comprises morphologically inconsistent members, which share, however, a homogenous pollen morphology with levelled tetrads and a punctate sculpture. This is found nowhere else in the clade, except for the relatively closely positioned genus Scaphosepalum, which is habitually similar to P. endotrachys (Chase 1985) with which it shares distichous inflorescences and distinct conduplicate flower bracts.
[page 122↓]The Cuban endemics P. gemina, grisebachiana, shaferi, trichyphis, and wrightii which form the most distant subclade, comprise (pollen-) morphologically discordant features. This may be the result of undersampling, which is indicated by long branches, especially among the deeper splits.

As a result, the addition of the Cuban data had the following effects on the original tree: the overall resolution in the strict consensus tree rose substantially and nearly all of the collapsed branches in the original consensus tree are resolved now. Puzzling combinations in the original ITS tree were rearranged in a more natural order, which had often been indicated before by data from more stable DNA regions like matK and trnL-F (Pridgeon & al. 2001), e.g. Brachionidium-Octomeria, Acostaea-Pleurothallis grobyi-alliance. In the following, some taxonomic suggestions drawn from the broadened insight in pleurothallid phylogeny shall be listed:


[page 123↓]

4.2  Relationships of the Cuban taxa – molecular evidence

Comparing the molecular results of this study, it becomes evident that the Cuban Flora of Pleurothallis does not represent a monophyletic group. Instead, it is nourished by different evolutionary lineages.
A great number of the Cuban species of Pleurothallis is not closely related with each other and many occupy even rather isolated positions in the system (P. bissei, P. domingensis, P. nummularia, P. obovata, P. pruinosa, P. racemiflora, P. ruscifolia, P. sertularioides). Other groups appear clustered in the molecular tree. However, morphological traits and long branch lengths indicate missing samples, i.e. they also represent independent lineages (P. odontotepala, P. rubroviridis, P. testaefolia, P. wilsonii). Finally, there are tight clusters of species, cases in which molecular, morphological and palynological data indicates a close relationship (subgen. Antilla, i.e. lower branch of clade B; subgen. Specklinia sect. Muscariae and sect. Hymenodanthae subsect. Longicaulae).

Thus, the Cuban spectrum of Pleurothallis, comprises groups of closely related taxa as well as distantly related ones. This pattern is found in many of the Cuban genera, both orchidaceous and non-orchidaceous (Alain 1958, Borhidi 1996, Samek 1973) and coincides with the general notion of island biogeography as a product of immigration and radiation (MacArthur & Wilson 1963, 1967). A close examination of the groups reveals two unusual features: a) differences in rates of morphological (phenotype) and molecular (genotype) evolution, and b) differences in tree topologies inferred from morphological and molecular data, which, however, should not be confused with the overall discrepancies between morphology and genetics. These features will be discussed in the following

Molecular vs. morphological evolution: pace and branch lengths

The first inconsistency concerns the unusual distribution of branch lengths. Rate heterogeneity among taxa is a widespread phenomenon (reviewed in Wendel & Doyle 1998) and is attributed to 1) variation in generation times, assuming a clocklike mutation rate, 2) non-hierarchical molecular evolution, e.g. recombination between paralogous or xenologous DNA, and 3) insufficient taxon sampling and/or extinction.
Since taxon sampling density is one of the major problems in this study, due to the limited number of Antillean taxa included, it is irrelevant to discuss substitution rate heterogeneity among deeper splits, since they are most probably brought about by just this methodical deficiency. However, even among the finest splits (Fig. 69, Fig. 70), which, on the base of additional gross and pollen morphological data, are assumed to be sister taxa, one can observe that branch lengths (genetic distance) often does not coincide with morphological differentiation (species concept based on morphological distance). On the one hand, we [page 124↓]find pairs of species which exhibit a low number of mutations (e.g. P. gemina~wrightii: 0 mutations; P. longilabris~obliquipetala: 5). On the other hand, there are instances, where substitution rates within specific boundaries strongly exceed even those between species (e.g. P. sertularioides: 4, P. tribuloides: 9; P. trichophora: 13; P. ruscifolia: 17). Since the Cuban pleurothallid species are usually well defined by morphology, these observations touch the underlying question if substitution rates are concordant with changes in morphological characters, i.e. if molecular evolution reflects morphological evolution and vice versa. In the species of Pleurothallis studied it is apparent that some taxa have maintained the original sequence while differentiating into morphologically distinct species, whereas others have undergone considerable molecular evolution while retaining a characteristic morphology. This is partially due to the fact that, when using ITS, we deal only with a small fragment of the pleurothallid nuclear genome, which, owing to the putatively non-coding character, is little exposed to environmental selection involved in speciation processes (reviewed in Baldwin & al. 1995). Differences between the molecular (ITS) and morphological pace of evolution should be therefore a priori no surprise. Moreover, even within the ITS region mutations do not occur randomly, partially because of the chemical behaviour of nucleotides (p. 104), partially because certain regions are more conserved than others (Fig. 64).

It is interesting, however, that one pair of species, despite the proneness of ITS to mutations, shares the same sequence. P. gemina and P. wrightii are abundant in Cuba’s Oriente today, nevertheless, P. gemina had never been collected prior to the 1980ies. This and the common ITS sequence indicate a recent origin for the new species. Pairs of species exhibiting little genetic distance have been reported from other orchidaceous groups, too (Cox & al. 1997, Borba & al. 2002). ITS, despite its popularity in studies aimed at the species level, fails to reflect putatively recent phenotypic differentiation. This has been observed also in other orchidaceous (Borba & al. 2002, Van den Berg & al. 2000) and angiospermous taxa (Panero & al. 1999: in Macronesian Asteraceae). If we assume a clock-like rate in ITS evolution among lower taxonomic ranks at least, these results contradict the assumption (Soto Arenas 1996) that orchidaceous speciation processes are long-time events, although it may be the case in certain orchids (Ackerman & Ward 1999). In the case of the Cuban endemics P. gemina and P. wrightii, we possibly observe an example of what was coined “evolutionary explosion” by Gentry (1982), i.e. the genesis of new species within very short periods of time (Gentry & Dodson 1987). Gross morphology is quite similar with some synapomorphies shared by both taxa: similar creeping habit, verrucate leaves etc. Respective autapomorphies comprise single, two-flowered racemes with whitish flowers (P. gemina) and several single-flowered racemes with purple flowers (P. wrightii). Hybrids with a mixture of the paternal features were found on one occasion (Stenzel 2001). The speciation process was triggered perhaps by a [page 125↓]spontaneous mutation in one population, that met favourable environmental conditions and set up a reproductive barrier at once. The rest of the pool remained untouched. Except for leaf shape and overall plant size, herbarium specimens of P. wrightii do not show any morphological variation during the last 150 years. Thus, it was a process of splitting off rather than of parallel stepwise divergence which should be a requirement in sympatric speciation where back-crossing is an incidence easily imaginable.
Cox & al. (1997) present a much more complicate explanation for the divergence between morphological and molecular evolution rates. Certain special molecular evolution modes, e.g. reticulate inheritance, may explain inconsistencies between gene and species trees. However, in this particular case it ignores the simple fact, that ITS is not a coding region for morphological and physiological traits that are related with floral and ecological adaptations. Imbalances between the evolution of morphological, physiological and genetic characters seems to be a common trait in orchids (Cox & al. 1997; Borba & Semir 2001; Borba & al. 2000, 2001a, 2001b, 2001c, 2002). On condition that ITS represents a kind of a molecular clock among finer splits, the only but essential conclusion of different branch lengths among pairs of species is that speciation occurs at different rates of time. This hypothesis coincides with findings of different patterns and paces in orchidaceous evolution in this study and elsewhere (Gentry & Dodson 1987, Ackerman & Ward 1999, Tremblay & Ackerman 2001).

Molecular vs. morphological evolution: topology

Before going into detail, it has to be stressed, that most of the peculiar topologies that are subsequently discussed, did not receive bootstrap support above 50%. However, they are found in the strict consensus of the Complete Analysis without exception.

The second unusual pattern observed in the molecular tree should not be confused with the overall differences between relationships derived from morphological and molecular data. Rather it refers to those instances, where at low taxonomic ranks doubtless similar morphology is contradicted by differences in sequences and consecutively by topology. This can be observed in morphologically similar taxon pairs, however, it is most striking when comparing different samples from one and the same morphological well defined species. As one would expect, many of these pairs of species or samples reflect the close relationship by falling sister to each other (e.g. P. ghiesbreghtiana, P. trichophora; P. nummularia~peperomioides; P. testaefolia~melanochthoda; P. racemiflora~powellii; clade G: from P. mucronata onwards). However, in other cases they were, unexpectedly, separated by morphologically very different taxa. As an example, the Cuban endemic P. rubroviridis was separated from the morphological very similar P. s icaria by a number of (Cuban) taxa of comparably much less morphological affinities. Interestingly, the P.-s icaria-rubroviridis subclade consists almost exclusively of species that were [page 126↓]morphologically classified by Luer as section Sicariae in subgen. Acianthera which underlines the morphological affinities.
P. grisebachiana, a Cuban endemic, is morphologically very similar to the continental Pleurothallis grobyi Batem. ex. Lindl., sharing even such fragile features as lip coloration. Both species fell in a “disjunct” position comparable to that of P. rubroviridis and P. sicaria. Whilst considering species pairs may be often a matter of taste, classifying species should be less subjective. Yet, even within specific limits the described pattern can be observed. Perhaps the most striking instance is represented by P. bri ghamii. The Cuban sample fell within a group of other Cuban species. The accession of P. brighamii sequenced by Pridgeon & al. (2001], on the other hand, was separated from the Cuban plant by 7 nodes, which span even distinct genera, like Scaphosepalum. Theoretically, there are three explanations: unverified plant material, inadequate topology and reticulate evolution. As to the source of material, it was not possible to check for wrong determinations in the material used by Pridgeon & al. However, when those samples from this study that are represented by Cuban vouchers, too (P. brighamii, P. tribuloides, P. ruscifolia, P. sertularioides) where added to the Cuban matrix (ITS15C), all pairs came out as sisters! Considering the reliability of the topology, we have to face the limitations brought about by incomplete sampling. However, it is unlikely that sample pairs at such a low taxonomic rank which are so profoundly separated should become sisters by adding more taxa. On the contrary, species samples that had been pairs in the Cuban Analysis (33 taxa + Pridgeon’s accessions of taxa which are present in Cuba, too) became separated in the complete MP study (>200 taxa). Thus, the two accessions of P. brighamii fell sister to each other in the Cuban Analysis, and became separated only in the Complete one. Moreover, this “disjunct” topology was observed several times in clearly unrelated clades reducing the probability that we face an artefact due to an inappropriate molecular or data-processing methodology.
To explain these inconsistencies, special features in the evolution of ITS, resulting in differences between the gene tree and the species tree (Pamilo & Nei 1988; review in Soltis & Soltis 1998), might be the key. Distortions in the gene tree may have been brought about by “inadvertent analysis of paralogous ITS” copies, i.e. by additional ITS copies that originated within the branch of the species tree. This hypothesis was used to explain unusual molecular topologies among morphologically similar species in the subfamily Cypripedioideae (Orchidaceae) (Cox & al. 1997)]. In the Cuban Analysis, however, a second peculiar pattern was observed. It is striking that the “disjunct” positions of closely related taxa or infraspecific samples always involve other Cuban samples as a separating block (Fig. 71). In these instances the gene tree reflected more a phytogeographical pattern than the morphologically defined species tree, since it groups the Cubans on one side and the continentals on the other side.


[page 127↓]

Fig. 71: Generalised example of a “disjunct” topology in the Complete Analysis involving either two closely related taxa or two accessions from one and the same taxon. Note the grouping of the Cuban samples. Positions of the Cuban and the non-Cuban sequence are marked with l.

Thus, this distortion by phytogeographic patterns may be explained much easier by xenologous than by paralogous ITS sequences, i.e. reticulate evolution. It may be the result of hybridisation processes in sympatric, i.e. Antillean, populations of different taxa, which is supported by the well known proneness of orchids to hybridisation even at higher taxonomic ranks. What seems to be quite logical and compelling, becomes more complicated when we try to explain the creation of hybridogenous genetic information. Unfortunately, little is known how concerted evolution in the intergeneric spacers takes place (Soltis & Soltis 1998). Yet, a possible scenario would be both the homogenisation of hybridogenous ITS sequences towards either of the parental copies, as observed in Gossypium allopolyploids (Wendel & al. 1995) in combination with the positive selection of either of the ancestral phenotypes. A similar explanation has been offered by (Cox & al. 1997) concerning the polyphyly of two taxonomically synonymised species of Paphiopedilum (Orchidaceae). Although the return to one of the parental phenotypes might appear unlikely, it must be emphasised here that floral features might be under high pro-parental selective pressure due to the presence of the traditional pollinator sets.
To add even more to the complexity in the case of P. brighamii, however, both orthologous and xenologous information must be assembled in the same ITS copy, since voucher pairs like the two accession of P. brighamii came out as sisters in the Cuban Analysis, while the Complete Analysis placed them in separate positions. Theoretically this is quite possible, since manipulation of ITS subregions or loci must be an inherent feature in concerted evolution. Indeed, Buckler-IV & al. (1997) found evidence of recombinant ITS sequences that combine portion of two distinct copies in Tripsacum (Poaceae), Bubbia (Winteraceae) and Nicotinia (Solanaceae).
The role of hybridisation in the evolution of island floras has been demonstrated in other angiospermous families, too. Similar to the topology of P. brighamii, Francisco-Ortega & al. (1996) found subspecies and different accessions of Argyranthemum species [page 128↓](Asteraceae) in a paraphyletic topology. The authors interpreted this phenomenon in the Macronesian flora by either introgression from other taxa as is suggested here, too.

(Pamilo & Nei 1988) recommend the employment of various markers with a different history, preferably both plastid and nuclear loci, to reduce the risk of irrelevant molecular information due to specific traits of ITS evolution. Results from Pridgeon & al. (2001: 2301), however, have shown that common plastid markers (matK, trnL) failed to resolve relationships among species of subgen. Specklinia (clade G). To test the hypothesis of local hybridisation as a cause for incongruence between gene and species trees, it will be essential to add other chloroplast markers which are evolutionary less stable. Similarly, cytological examinations of chromosomal features could help in identifying hybrids and additional PCRs using denaturating detergents (DMSO) should be conducted to check for functional vs. non-functional PCR products (Buckler-IV & al. 1997).

The observed pattern is rarely seen in genetic studies on orchids, since most of these projects deal either with phylogenetic issues at a larger scale or with population genetics in restricted areas (Borba & al. 2000, 2001a). The simultaneous study of sample sets defined by area (floras) and by systematics (phylogeny) including both distantly and closely related species or even multiple samples from different but well known localities is a requirement for the findings in the present study. Some studies are aimed at the phylogeny of distantly related taxa, thus, spanning a great array of species. Logically, sequencing of closely related taxa is usually avoided to guarantee an evenly spread sample set of the group under study. Multiple sampling of species is rare in molecular studies on Orchidaceae aimed at the phylogeny of large-scale taxa. Most studies include only one or two species with double accessions. Van den Berg & al. 2002 double sampled several taxa. In this study the two accessions of one species (Cymbidium ensifolium (L.) Sw.) were separated by an other (C. kanran Mak.), however, with bootstrap support < 50% and a collapsed branch in the strict consensus tree. Unfortunately, this interesting result was not discussed by the authors. Cox & al. (1997) reports a case where a taxonomically synonymised, hence morphologically very similar pair of species was separated by several other taxa in the ITS tree. This is exactly what was found here in P. brighamii and P. grisebachiana-P. grobyi. Again, the authors fail to further discuss these findings.

4.3  Genesis of the Antillean Pleurothallis flora

Reliable phylogenetic data along with distribution patterns are available for one of the major groups of Antillean orchids for the first time. This allows a detailed discussion of the genesis of Antillean orchid flora.

[page 129↓]Colonisation of the Antillean arc by pleurothallid orchids

In order to assess the history of pleurothallid evolution in the Greater Antilles it is important to consider the age of the orchid family in general and that of Pleurothallidinae in particular. This issue has been discussed controversially in literature and personal communication (Arditti 1992). While some authors favour a relatively early origin of the family (Stebbins 1950: early Cretaceous), and the estimates differ considerably (Arditti (1992) citing from pers. comm.: Cretaceous to early Miocene), there is a widespread consensus now in that “most extant groups are probably very young” (Arditti 1992). Garay (1960), although supporting a Cretaceous origin, suggested a probable secondary “expansion” during the post-Pleistocene time. To my mind, a relatively late origin and/or radiation is quite comprehensible and is backed by the following traits: 1. almost complete absence of fossil records (Doyle 1973; Schmid & Schmid 1977), although this may be in part due to the herbaceous growth and other physiological features (Wolter & Schill 1985); 2. high diversity at the species level with an overwhelming specialists-generalists ratio, which should mark intensive adaptive radiation; 3. relative genetic stability in contrast to the enormous morphological plasticity (Neyland & Urbatsch 1995) which characterises active speciation. Considering the presumable young age of the family it is quite safe to assume that advanced taxa like Pleurothallidinae were absent from the Antillean arc until the present island constellation was reached (Iturralde-Vinent & MacPhee 1999: late Eocene to middle Miocene, 35-14ma).
However, another factor may render the age of the family of secondary importance. Pleistocene climate fluctuations are known to have severely altered floral and faunal belts, not only in temperate zones but in (sub)tropical regions, too (Leyden 1994; cf. Curtis & al. 2001). Glacation led to temperature depressions in the Circum-Caribbean, too, and there is even geomorphologic evidence for quaternary glaciation above 2200-2300 m in Hispaniola (Schubert & Medina 1982). As a result, mountainous vegetation belts were lowered, which was detected in several neotropical locations (Van der Hammen (1973): Colombian Andes; cited in Leyden (1984): Costa Rica and Panama, by 500-1000 m). A second effect is represented by a severe aridization in the Circum-Caribbean lowlands (Leyden 1984: N Guatemala; Grimm & al. 1993: Florida; Street-Perrott & al. 1993: Jamaica) with a parallel suppression of humid forests. Leyden (1984) and Leyden & al. (1994) found that mesic tropical forests were generally absent in the northern lowlands of Guatemala from 36-10000 years BP. Similarly, Köhler (pers. commun.) found evidence for striking anatomical adaptations to severe drought stress in Cuban Buxus even in those taxa that live under humid conditions at present. Climatic oscillation affected not only terrestrial habitats, but marine ones, too. Several studies (cf. Curtis & al. 2001) indicate that both air and sea surface temperatures decreased by 5°-8°C during the last glaciation compared with present conditions, hence, leading to a drastic drop in oceanic [page 130↓]evaporation.
A switch back to humid conditions between 10.000-8.500 BP is inferred from several Circum-Caribbean localities (Curtis & al. 2001).
Of course, without exact data that are gathered directly from the Antillean arc (only Street-Parrot 1993), generalising conclusions have to be drawn with extreme care. Specific local features, like the relief, may have altered the palaeoclimatic process regionally, as is seen in differing data for the onset of humid conditions in the Circum-Caribbean (cf. Curtis & al. 2001).
Nevertheless, the whole climatic turn-over must have had a severe impact on the pleurothallid flora. The specific impact on the distribution of microphytic orchids, like pleurothallids, must be due to changes in both biotic and abiotic environmental factors. First, despite general adaptations of orchids to drought, i.e. succulence, CAM, velamen, etc. (Arditti 1992), which is a temporal but prevalent climatic feature in all epiphytic habitats (Freiberg 1992, Gentry & Dodson 1987), many Pleurothallidinae show a higher affinity to even more humid habitats as is reflected in their ecological centre of diversity, the wet (sub)montane rain- and cloud forests. Data from this study reveal a demand of >1200-1400 mm/a precipitation. Yet, the predominantly occurrence in habitats with a specific microclimate (p. 89) even in areas with an increased annual rainfall, suggests that a high level of humidity is required. Together with absent pseudobulbs, the reduction and disintegration of the sporoderm (Stenzel 2000) may be interpreted along this line too. Consequently, these affinities should make them susceptible to drought stress, although, as a consequence of absent pseudobulbs, some members of Pleurothallidinae have shifted succulence to the leaves and/or flower tissues. Cuban species show these traits, above all in the P.-ekmanii-group and to a lesser extent in several taxa in Acianthera (clade B). However, other groups, e.g. Pleurothallis subgen. Specklinia (clade G), Platystele and Lepanthopsis, show little or no sign of succulence. Conclusively, certain groups of Pleurothallidinae must have suffered from Pleistocene aridity to a greater extent than macrophytic orchids. Second, aridity must have had an indirect double impact on pleurothallid habitats. While a general drop in atmospheric humidity led to the retreat of rainforests and seasonal forests, a subsequent dry-up of rivulets, affected the principal habitat of Pleurothallis in Cuba (p. 89). Another habitat favourable for Pleurothallis are the mountainous condensation belts, however, only were closed forests exist. In the high-mountainous open elfin-thickets and shrublands of the Turquino group almost no species of Pleurothallis occur. Thus, a humid yet generally more open vegetation even in cloudy ranges, would not have been a Pleistocene refuge for Pleurothallis. Third, it should not be neglected, that faunal fluctuations followed floral ones and, in the case of insect dependent orchids, floral distribution followed faunal in turn which must have even reinforced the impact of Pleistocene climate oscillations.


[page 131↓]

Although Holocene palynological data for Cuba is not available (Borhidi 1996) in order to draw at least a rough picture of how the vegetation recovered during the last 10.000 years, it is very likely that small epiphytic orchids, among them Pleurothallidinae, have re-[?]colonised the arc only after the last glaciation, when rising temperatures led to favourable levels of air humidity and temperature.Antillean Pleurothallis should range therefore among the youngest lineages within this mainly continental genus. This is considered a general trait of insular biotas with respect to their continental sister (Carlquist 1995).
This assumption puts aside all discussion concerning the existence of historical dry land extension, plate tectonics, and land bridges (Borhidi 1996, Iturralde-Vinent & MacPhee 1999, Hedge 2001). Closely connected with these palaeogeographic phenomena, is the, mainly zoogeographic driven, dispute, whether dispersal or vicarianceaccounts for the creation of the Caribbean biota (reviewed by Page & Lydeard 1994). However, the herein assumed post-glaciation evolution of Pleurothallis in the Caribbean is embedded in rather stable geological and climatic constellations, which yield no background for an application of the vicariance model. The two models have been developed and applied mainly in the zoogeographic field, where speciation dates are assumed to have paralleled the geological evolution of the Antillean arc (Hedges & al. 1992; Woods & Sergile (2001) with various papers; see refs. in Page & Lydeard 1994: 21-22), rather than the recent Holocenic era. In contrast, the debate was touched upon only marginally by botanists (cf. Prance’s discussion of Haffer 1981). Although the dispersal-vicariance dispute seems to be irrelevant in this study, it should be mentioned here (cf. p.155).

Characteristics of orchidaceous dispersal. – The biogeographical assumption of a recent colonisation and radiation on the Antillean arc is admittedly unusual in a region where evolutionary processes have been traditionally discussed in the light of millions of years. Two major features connected with this issue, dispersal and speciation rates, will be briefly discussed now to provide further evidence for the hypothesis.
When comparing the studies on West Indian phytogeography it is frustrating how little attention is paid to the specific means of dispersal and chorologically relevant ecological traits to elucidate processes that led to the geographic patterns found (Dietrich 1989a; Judd 2001). This is mostly due to the fact that the present distribution in the West Indies is seen as a result of primarily Tertiary geological processes (revised in Iturralde-Vinent & MacPhee 1999; Borhidi 1996; Woods & Sergile 2001), which leaves little to discuss concerning the ecological influence on biogeography, i.e. migration in birds for avichory, air currents for anemochorous plants etc. When discussing the particular case of the genesis of Antillean Orchidaceae, dispersal observations made on the island of Krakatau are of crucial importance. The study of the re-colonisation of the entirely destroyed volcanic archipelago, 40-80 km off the Sumatra shore, showed the high vag ility of [page 132↓]orchidaceous seeds. The first orchids were found after 13 years (Van der Leeuwen 1936). After another 40 years the number of species had risen to 25. Considering the low rate of growth and initial failure of successful colonisation, seeds must have reached Krakatau several years before 1896 even (Gandawijaja & Arditti 1983). Close & al. (1978) examined the probability of wind dispersal between Australia and New Zealand, separated by 2000 km, stating that seeds might reach the islands with the aid of prevailing W winds within 1-3 days. Their review of literature on orchidaceous Circum-Tasmanian phytogeography, revealed strong affinities between SE Australia and New Zealand, including a report of presumably recent migration in Cryptostylis. These findings, the Krakatau experience, in vitro studies of the floating capability of orchidaceous seeds by the same authors, as well as empirical studies of neophytic plant migration (Dod 1986a; Stern 1988) underline the strong potential for long distance dispersal in orchids within rather short periods of time.

Speciation rate. – While dispersal in orchids has been accepted as a probably frequent and fast phenomenon, speciation is traditionally considered a slow process even in this family (Soto Arenas 1996). Dietrich (1989a) cites subtribe Angraecinae, and the genera Bulbophyllum and Polystachya as examples for the West African / Madagascar – Caribbean disjunction in concordance with Wegener’s theory on continental drift. On the other side, she doubts, Orchidaceae will be of any use in elucidating the Caribbean paleoclimatic and paleogeographic history, a statement which is not further discussed, unfortunately. Trejo-Torres & Ackerman (2001) found floristic similarities among Caribbean islands with a common geological history and related these affinities to existing or absent historical landbridges, among other causes. Dod (1984a) used the particular orchid flora of SW Hispaniola to show a separation of Massif de la Hotte from the rest of this mountain range until the Pliocene.
In contrast, the likelihood of explosive evolution has been discussed only sporadically in Orchidaceae (Gentry, 1982; Gentry & Dodson 1987). If any it has received mostly negative or at best ironic reception (Holm-Nielsen & al. 1989). Only recent studies in other plant groups, combining molecular and palaeogeographic data, seem to have cleared this issue of its speculative character. Richardson & al. (2001a) show the explosive speciation of Phylica (Rhamnaceae) in the Cape region during the glacial epoch which they attribute to processes connected to aridization (ecological areal fragmentation, pollinator shift etc.). Likewise, Richardson (2001b) found evidence for explosive radiation in Inga (Fabaceae) since the bridging of the Panaman Isthmus (3.5 ma), stating that the intensely active speciation may be attributed to this event, later phases of Andean orogeny and Quaternary climatic oscillation. Long before speciation events have been dated with the help of molecular clocks, Gentry (1989) had proposed that close to ½ of the neotropical flora might have originated by explosive evolution. Interestingly these studies connect saltation with rather young epochs, usually not older than the Quaternary. Even younger, [page 133↓]i.e. post-glacial speciation, is widely ignored (Bateman & DiMichele 2003), however.

Phytogeographic evidence for migration routes. – Assuming an origin of Pleurothallidinae outside the Antilles and a colonisation of the islands at the earliest when most of the geographic conditions of both continents and the archipelago were fixed, the Greater Antilles may have been colonised directly via two routes. One starts in Central America and leaves the continent at the Yucatan shore passing Cuba and Jamaica, towards Hispaniola and Puerto Rico. The other would lead from the Guyanas via the Lesser towards the Greater Antilles.
As was shown, Antillean taxa of Pleurothallis are mainly island endemics (80%) or of neotropical distribution (9%). Based on present distribution, only 14% of the taxa are geographically informative with respect to the colonisation form continental areas. Present distribution patterns with 11% of the taxa being exclusively of Greater Antilles – Central America (–South America) distribution, strongly support the route via Central America. The MP analysis (Fig. 52) groups (East) Cuba and Jamaica together with Central America. Hispaniola and Puerto Rico have closer affinities to the Lesser Antilles and South America. The assumed floristic division of western and eastern Greater Antilles (Judd 2001) and respective affinities with Central America and the Lesser Antilles presents a tempting pattern, simply by the geographic neighbourhood of the respective source areas. A closer look at the species in concern reveals, however, that there are no taxa confined to the eastern Greater and the Lesser Antilles and/or South America. In fact, all but one taxa (P. discoidea) that occur in the Lesser Antilles (P. aristata, P. discoidea, P. imraei, P. pruinosa, P. ruscifolia, P. wilsonii) are Pan-Caribbean elements. They may or may not be absent on some of the islands, but are present in Central America and South America. Thus, the distribution of P. discoidea, if not a collection artefact, is the only example in Pleurothallis where colonisation may have taken place via the Lesser Antilles. All other taxa could have reached the Greater Antilles just as well via Central America. The topology of the tree is influenced by species absence too, which groups Puerto Rico, the Lesser Antilles and South America, although not a single taxon is confined to this area. Hispaniola, in spite of its high species diversity, is included in this group probably due to the scarcity of shared taxa with the Western Antilles and Central America.
The disjunction Cuba – South America has been often cited and apparently generally overestimated in Cuban literature on biogeography. Pleurothallis provides an illustrative example of how the growth of phytogeographic and taxonomic knowledge has improved our notion of Caribbean biogeography. Bearing the temporary character of floristic inventories in mind, it is astonishing that the general phytogeographical proportions of historical sources (Acuña Galé 1939, León & Schweinfurth 1946) are largely congruent with modern ones, despite the taxonomical and systematical problems of these works.


[page 134↓]

Tab. 12: Pleurothallis of Cuba: phytogeographical data of different sources in comparison with the present notion, drawn from this study.

 

Acuña Galé 1939

León & Schweinfurth 1946

present study

total

39

38

39

island endemics

21

23

19

Greater Antilles endemics

6

5

8

+Central America endemics

1

1

4

+South America endemics

0

0

3

+Lesser Antilles endemics

3

3

0

+South America endemics

0

1

0

+South America endemics

2

3

0

neotropical elements

1

1

5

Differences are mainly to be found among finely resoluted biogeographical relationships, especially those that include South America. Comparing these sources it is clear that many distributions originally thought to be disjunct (Greater Antilles-South America) or of restricted areal (island endemics, Greater Antilles-Lesser Antilles, Greater Antilles-Central America), have turned out to be collection artefacts or were based on wrong classifications (Stenzel & Llamacho 2002).

Studies on orchidaceous phytogeography. – The close floristic affinities of the orchid floras between Cuba, Jamaica and Central America have been stated already by Fawcett & Rendle (1910). On the contrary, Dietrich (1989a) and Trejo-Torres & Ackerman (2001) presented different views. Both studies used the whole orchidaceous spectrum of the Antillean. Dietrich found Cuba in closest floristic neighbourhood with the Bahamas and Puerto Rico [!], which is obviously due to methodical incompatibilities. In Dietrich’s work the level of relationship is expressed by the percentage of the Cuban taxa in foreign floras. Thus, areas with indistinct floras, composed of wide spread species, showed the strongest affinities4. Trejo-Torres & Ackerman (2001) found that “the Guyanas form a sister group to the Greater Antilles” which is evidently a result of the exclusion of (sub)mountainous Central America from the study. According to their data, the Hispaniolian flora is more similar to the Cuban than to the Jamaican one. This, again, may be a methodical artefact, brought about either by different sample sizes, e.g. floras with different numbers of species, which may yield groupings by species number. It may be also due to the employment of a Lundberg outgroup (all species coded absent) which, as the present study shows, influences both tree number and topology. On the other hand, Pleurothallis may well show closer affinities with Jamaica, while other orchidaceous genera do not. There are numerous examples for both ties: Homalopetalum, a monotypic genus endemic to the Sierra Maestra massif and the Blue Mountains, and, on the other hand, Domingoa, a monotypic genus found in Cuba, Hispaniola, and a few other islands to the East but not in Jamaica. A third methodical difference makes data obtained in the [page 135↓]present study difficult to compare with those from Trejo-Torres & Ackerman. Pleurothallid affinities with Jamaica are brought about largely by a certain floristic similarity between Cuba, Jamaica and Central America, rather than by species exclusively shared by the two islands. Neither Dietrich nor Trejo-Torres & Ackerman included Central America except for some randomly picked isolated regions, e.g. Yucatan (Trejo-Torres & Ackerman), El Salvador or Nicaragua (Dietrich), which show but a portion of the Central American taxa. Finally, it should be stressed that the floristic exchange between the western Greater Antilles and Central America is still an active process, as can be seen in the most recent discovery of the Central American orchid Catasetum integerrimum Hook. in Guanahacabibes, Cuba’s westernmost tip (Díaz & Cabrera 1985).

Other large scale studies. – It would go definitely beyond the scope of this discussion to compare the numerous studies that have dealt with floristic relationships of the Antilles that have been published within the scope of taxonomic treatments. However, some of the most comprehensive works and those in ecologically similar groups should be mentioned. Alain (1958), and especially Borhidi’s (1996: 259) notion of the origin of the Antillean flora is strongly influenced by the geological history of the archipelago. Both stress the importance of Central America as a floristic source area for the Greater Antilles. Migration via the Lesser Antilles, in contrast, is granted less importance. So far, these general findings are congruent with pleurothallid data. Inconsistencies in other affinities are largely due to the specific ecological preferences of Pleurothallis and allies. Such incompatibility exists towards Florida/Bahamas, owing to the fact, that there are almost no pleurothallids N of Cuba. Other studies with a broader sample size (Trejo-Torres & Ackerman 2001) in contrast, do show that those relationships exist in orchids, too. Samek (1973), on the basis of the Antillean phanerogamous flora, detected closest affinities of Cuba with Hispaniola, Jamaica, and Puerto Rico in descending order, which is not backed by pleurothallid data. In return, Samek found less distinct, but equal similarities of Cuba with Central and South America. The phytogeographical disjunct tie between Cuba and South America has been stated in various works, and was, at least in pleurothallid data, an artefact due to the poorly reviewed orchid flora and a wrong notion of distribution patterns (s. Tab. 12 and accompanying text). Howard (1974), in contrast, found closer relationships between the Greater Antilles and Central America as well as the Lesser Antilles and South America. The latter, however, was qualified in a subsequent paper (cited in Borhidi 1996 as Howard 1982 [?]). Based on generic similarities he summarised the Greater Antilles – Central America – South America distribution as the ‘Western Continental Range’ (ibid. p.29) which he saw as an ‘extension’ [!] of Central American genera across the Greater Antilles. Similar to conditions found in Pleurothallidinae, he considered the Virgin Islands east of Puerto Rico as a phytogeographic terminus. Other relationships that he discussed are only poorly or not at all represented by Pleurothallis. [page 136↓]Concerning the taxonomic level to be chosen, Howard favoured a comparison between genera, although he stated, that similar results will be obtained at the species level (ibid. p.18). In the case of Pleurothallis this would be fatal and much of the chorological information would be lost treating the genus in Luer’s (1986b) circumscription. However, even if taking the system proposed by Pridgeon & Chase (2001), many of relationships between the islands and continental areas would be drowned in genera like Stelis s.l. or Acianthera. In my view it is not advisable to restrict the study a priori to any particular taxonomic unit, since all taxonomic entities are man made (cf. Borhidi 1996: 285) and their taxonomic level does not necessarily reflect the evolutionary history. Phytogeographical studies should be done at all taxonomic levels instead, a goal that is finally envisaged by phylogeography, as will be shown next.
Since distribution patterns are heavily influenced by dispersal units, associated means of dispersal and other traits, it may be expected that other anemochorous plant groups should reflect to a certain degree a distribution paradigm similar to that found in Pleurothallis. Judd (2001: Sect. Lyonia of subgen. Lyonia, Ericaceae ) did not find any extant floristic affinities between the Greater Antilles and the Central American continent, but claimed such for phylogenetic relationships. This is interesting, since it shows that present-day distribution does not necessarily reflect historical migration routes. Like in Pleurothallis, there are only weak affinities with the Lesser Antilles. On the other hand, he found floristically well delimited island floras, i.e. with a high rate of endemism within island boundaries. As in Pleurothallis this applies above all to Hispaniola and E Cuba. The floristic affinities between Lyonia and Pleurothallis may be the result of analogous chorological patterns and a predominantly mountainous distribution. Quite in contrast, data of Pteridophytes presented by Borhidi (1996) show much lower island endemism (~11%) than Pleurothallis does (~50%). Furthermore, the Cuban pteridophytic flora has a predominantly Caribbean and Neotropical relationship. Unfortunately, Borhidi does not show the level of congruence with Central America. He lists a mere 6 taxa (~2%) that are of Central American – Cuban distribution. Interestingly, 4 of them are rainforest elements, an ecological proportion reflected by pleurothallid data. The specific geographical relationships should be mainly the result of the comparably great age of ferns, as stated by Borhidi. However, his second suggestion, fern vagility, is fairly qualified by conditions found in Pleurothallis and other Orchidaceae. Yet, it may hold in connection with the great age and an euryoecious characteristic of these plants.

Molecular evidence for migration routes. – Plotting horizontal distribution over phylogeny, it is possible to follow migration routes and study processes or features linked to speciation events. The main problem in phylogeography is, that for a reliable reconstruction of these routes and conditions, all taxa should be sampled, since each taxon absent from the matrix may represent a valuable link in dispersal and speciation [page 137↓]processes (Baldwin & al. 1998). Due to the limited number of continental and island taxa, only tentative conclusions concerning the closest continental relatives of Antillean taxa can be drawn in the present study (Fig. 69, Fig. 70).
Many of the Antillean endemics that occur on Cuba appear rather isolated in the phylogenetic tree, e.g. P. nummularia, P. racemiflora, P. bissei etc. This gives the impression of an Antillean flora largely dependent on the continental source area with a considerable frequency of colonisation events without further cladogenesis. Even where taxa came out as sisters in the molecular analysis (P. ruscifoliaP. pruinosa, P. trichyphisP. brighamii), this does not automatically imply sisterhood, which is due to the limited sample size. Thus, even in tight clusters of several Cuban species (P. corniculata through P. wrightii in clade G), independent migration to the Antillean arc has to be assumed, where taxa show a distinct morphology, not found elsewhere in the Antillean set of species.
Consequently, three g roups of species can be distinguished according to their evolutionary history (Tab. 13). Most of the Antillean endemics that are present in Cuba have their closest relatives inside the Greater Antilles (~17 spp = 44% of the Cuban species). Morphological data of the species that have evolved while reaching the island arc (~10 spp. = 26%), suggest that most have apparently not radiated into new taxa, however, this can be said with certainty only after the molecular study of the rest of the Antillean pleurothallid flora. The same applies to the 12 taxa (30%) that occur outside the archipelago, too.

Tab. 13: Origin of the Cuban taxa of Pleurothallis (endemics in bold).

Invasive species with no subsequent cladogenesis

Invasive species that evolved into new taxa during dispersal, but no further dichotomy of the lineage

Species that originated within the Greater Antilles

P. aristata

P. brighamii

P. corniculata

P. gelida

P. ghiesbreghtiana

P. obovata

P. pruinosa

P. ruscifolia

P. sertularioides

P. testaefolia

P. tribuloides

P. wilsonii

P. bissei ?

P. domingensis

P. helenae

P. nummularia

P. odontotepala

P. oricola

P. papulifolia

P. racemiflora

P. rubroviridis

P. trichyphis

P. appendiculata

P. caymanensis

P. denticulata

P. ekmanii ?

P. excentrica ?

P. flabelliformis ’?

P. gemina

P. grisebachiana

P. llamachoi

P. longilabris

P. mucronata

P. murex

P. obliquipetala

P. prostrata

P. shaferi

P. trichophora

P. wrightii

The first row lists taxa that are widespread and do not show substantial anagenetic evolution. The two rows to the right show species that either originated while migrating to the Greater Antilles (centre) or evolved from plants already present on the islands (right).


[page 138↓]

In the following, some clusters of Cuban Pleurothallis shall be shortly discussed, concerning their origin and migration to the island arc.

  1. P.-ekmanii-group. This association comprises three morphologically “primitive” species with 8 pollinia (clade A). Backed by (pollen) morphology too, it is safe to claim that this group has no closer relatives among sequenced Cuban taxa (this study), other Antillean or continental taxa sequenced by Pridgeon & al. (2001). The clade fell sister to Octomeria, a genus with 8 pollinia too. The latter has a neotropical distribution, however, with a centre of diversity in S Brazil (Luer 1986a). In the Greater Antilles it is represented by just one species, O. ventii Dietrich (formerly treated as Octomeria tridentata Lindl.). Morphological and palynological traits clearly refute a possible relationship of this species with ekmanii & al., hence, O. ventii must have reached Cuba in an independent dispersal event. Consequently, two explanations exist for the present distribution of ekmanii and its allies. The high genetic distance to the sister group Octomeria (69 steps in Fig. 69) may indicate either loss of diversity by extinction, as observed in other plants of hot spot archipelagos (Baldwin & al. 1998: 412) or undersampling in Octomeria. In connection with the first, the distribution of P. ekmanii and its allies may therefore represent a relict of a formerly wide area of the common ancestor with Octomeria. Forest refuge theories (Prance1973) based vegetation fragmentation during the Pleistocene, although later qualified (Prance 1982), may serve to explain this pattern. Yet, the existence of such a basal pleurothallid clade in Cuba contradicts the assumption of a late colonisation of the Antilles by pleurothallid orchids. On the other hand, the reduced morphology with fused rhizome, ramicauls, sheaths and inflorescence along with succulence developed in virtually all organs of P. ‘flabelliformis’ and P. excentrica may actually indicate adaptations to Pleistocene drought. The three species would therefore inhabit a palaeoendemic area. The post-Pleistocene colonisation of the archipelago, as proposed before may therefore represent a re-colonisation or secondary expansion in reality. Yet, it may appear difficult to explain why such a primitive group survived just in the northern Caribbean, while all other basal Pleurothallidinae have a primarily South American centre of diversity. Although the following suggestion provides no explanation and has often been exaggerated, it should be reminded that the disjunct biogeographical range between the archipelago and northern South America has been repeatedly reported from other groups of vascular plants (Bonnetia, Burmannia etc. according to Borhidi 1996).
    Another explanation aims at a genetic background: hybridisation, i.e. the origin of this group from orchid species present on the same island. The comparatively high step number leading to the ekmanii-branch could then be interpreted freely by the [page 139↓]introgression of xenologous DNA. To search for the presumable parental taxa is considerably speculative, since we do not know how phenological traits are passed on to hybrid generations. Moreover, Rieseberg (1995) stated that most true hybrids do not show intermediate characters. In turn, Borba & al. (2000) found that intermediate features do not necessarily indicate hybridisation in Pleurothallis fabiobarrosii.
    A relict status of P. ekmanii, P. excentrica and P.flabelliformis’ seems to be the most compelling way to explain their history, despite the unusual distribution pattern far from the core area of Pleurothallidinae. After all, (presently extrazonal) relict areas are one of the commonly observed and discussed biogeographical issues.
  2. Pleurothallis subgen. Antilla. – Perhaps the most clear phylogeographic picture can be found in one subclade of clade B, which hosts a number of Cuban endemics (P. papulifolia through P. trichophora), accommodated in the Antillean Pleurothallis subgen. Antilla Luer. This subgenus comprises ± 10 taxa in the Greater Antilles sensu Luer (2000), however, the present molecular and palynological analyses indicate the phylogenetic association with additional taxa, e.g. P. papulifolia, P. murex and perhaps P. caymanensis too.
    The clade fell sister to the Central – South American P. erinacea. The latter belongs to Pleurothallis subgen. Kraenzlinella, a group which shows some morphological affinities to Antilla (verrucate, papillose or scaly ovaries and similarly ornamented capsules; the tendency to enlarged, conduplicate, and oblique flower bracts; unguiculate, basally biauriculate lip, often with two lateral lobules above the claw). The species of this subgenus are of South- and Central American distribution. The Mexican P. hintonii L.O Williams from seasonally dry forests (Luer 1994) shows succulent leaves, which in turn, can be found in the epilithic Cuban endemic P. papulifolia. It is apparently an adaptation to temporal drought (ibid.). Taking into consideration the geographical neighbourhood and morphological and molecular affinities, it is quite reasonable to assume a descent of Antilla via P. papulifolia from either a common ancestor with subgen. Kraenzlinella or from an extant member of that subgenus. In any way Antilla is clearly of Central American origin.
  3. Pleurothallis subgen. Specklinia sect. Muscariae. – This large section was described by Luer to accommodate a number of species with rather similar habit and flower morphology. In the Antilles are at least 6 species, 4 of which are endemic to Oriente. P. helenae is a Greater Antillean endemic and P. aristata is of neotropical distribution. Both occur in Cuba. Although it is tempting to trace back the origin of the Cuban endemics to island populations of P. aristata, molecular data suggest rather an early split off from a common ancestor, which was most likely of continental distribution. This constellation is equal to that of many other Pleurothallis, which, although [page 140↓]occurring throughout the Caribbean, have apparently not brought about new species. P. helenae which, due to certain morphological similarities with P. aristata, was treated by some authors (Nir 2000) as a synonym of P. aristata, turned out to be closer related to other continental species. Relatively high numbers of mutations (Fig. 70) in both P. helenae and P. setosa, must be the result of the limited sample size.

How many continental lineages have turned into radia tion in the whole Greater Antillean arc is difficult to ascertain, due to the limited sample size of the present molecular screening (1/2 of the Antillean taxa) and the probability of hybridisation, as shown in P. brighamii. A comparison of the Antillean taxa based on morphology (incl. palynology) shows that two of the discussed clusters comprise additional species on the other islands. Only the aristata-group (sect. Muscariae) seems to have radiated exclusively in E Cuba. Furthermore there are a few groups that consist of just two taxa: P. bisseiP. hirsutula and P. migueliicompressicaulis. The rest of the Antillean endemics is morphologically very distinct, indicating that the closest relatives have to be back on the Central American mainland. Thus by comparing these three groups we receive the following portions. Even a conservative estimate would consider at least 50% of the Antillean flora as autochthonous, distributed mainly among three radiative subclades: Pleurothallis subgen. Antilla s.l. (15 taxa), subgen. Specklinia sect. Muscariae (4 taxa) and sect. Hymenodanthae subsect. Longicaulae (16 taxa). The dominance of autochthonous species is greater than in Cuba (44%), due to the frequency of successful migration from the continent which was observed to decrease with distance (MacArthur & Wilson 1963; Thornton 1996; see p. 145). Consequently, the rest of the Antillean species, i.e. up to 36, are the result of successful colonisation events of pleurothallid species. However, this number may increase, i.e. the level of infra-Antillean speciation may drop, if some of the radiative clades should turn out to be polyphyletic. Discounting two taxa that emerged from restricted cladogenesis (endemic species pairs) and 15 species (~20%) that have not morphologically changed during migration, ~20 taxa (~26%) are left that have undergone anagenetic speciation in the course of migration.
From the little that we know about the genesis of orchidaceous island floras it can be assumed that the level of pleurothallid radiation in the Greater Antilles is striking, compared with other oceanic archipelagos. Bateman & DiMichele (2003: 17) report only one autochthonous event among the 12 orchid species present in Macronesia. The other 11 taxa are the result of anagenetic allopatric speciation, i.e. they represent unrelated lineages originating from neighbouring continental areas. The fact that some lineages enter cladogenesis while others do not, can be observed in other Antillean orchids too. Among the ~10 genera of Pleurothallidinae, only a few have radiated, e.g. Lepanthes and Pleurothallis. Bateman & DiMichele stated a general imbalance in the evolutionary potential within clades, with only some groups entering adaptive radiation and saltation [page 141↓]evolution. According to the authors, absence of the appropriate set of pollinating insects as well as different mycorrhizal fungi for particular ontogenetic stages may account for the scarce post-immigration radiation on these islands. Similarly, an accumulation of pre-adaptations (genetic heritage) in successfully radiating groups may have been trigger for subsequent (Bateman & DiMichele 2003). On the other hand, migrated species that did not enter cladogenesis could represent stasigenetic lineages, which are genetically in a conservative phase that prevents active speciation. Which one of the features eventually triggers or prevents radiation cannot be judged here. However, the fact that these three Cuban lineages represent clades that have intensively radiated on the continent too (Luer 1986b), reflects the evolutionary potential inherent to these groups. Another, rather simple, explanation for the unequal rate of speciation in different clades would be the time of colonisation. Available time surely plays a role in evolution. However, two of the lineages that have actively radiated on the islands were presumably among the last colonisers of the arc. Members of these groups show almost no adaptations to drought. They are small, fragile plants, with a reduced sporoderm and delicate flowers. In contrast, it is just the more drought resistant taxa of subgen. Acianthera, with the exception of Luer’s subgen. Antilla, that have ‘failed’ to radiate. These plants show a primitive sporoderm that provides full protection for the pollen masses and tend to have succulent organs including the flowers, features that indicate adaptation to drought stress. Hence, they should have been among the first Pleurothallids that were able to colonise the island arc after the Pleistocene drought.

Until now we have tacitly assumed that migration in the Caribbean is unidirectional. This is strongly suggested by molecular and phytogeographical data. On the other hand, considering the high vagility of orchids, an opposite migration cannot be ruled out completely. Theoretically, two patterns could exist: 1) the re-migration of species that had successfully colonised the islands without anagenetic evolution, i.e. continental mother–, island–, and secondary continental populations belong to the same species. This represents a likely process, since plants that re-migrate are faced with ecological features they are already adapted to and because species that have colonised the islands should have a high colonisation potential. 2) the migration of a new species that either originated from the islands or evolved in the process of dispersal towards the continent. This pattern is theoretically less likely, because it involves active adaptation to the new environment. Nevertheless, if we compare the number of island endemics (68%) to wide spread species in the Greater Antilles, it becomes clear that this process is by far more frequent than simple dispersal. Gathering from the Cuban species sampled, migration towards the island arc must be the predominant route since there is no occasion where continental plants came out in terminal position with island vouchers as basal branches.

[page 142↓]Dispersal and speciation within the island arc

The high level of both total species diversity and endemism in the Caribbean is well known and is reflected in the classification of this area as one of the 10 prime hotspots of biodiversity (Davis & al. 1997). Yet, rates tend to differ sharply, depending on the taxon as well as the area in concern.

Tab. 14: Numbers of species and regional endemics in the Greater Antilles.

 

Phanerogamous plants

Orchidaceae

Pleurothallis

Greater Antilles

Cuba

Hispaniola

Jamaica

Puerto Rico

13000 / ~58%1

6400 / ~50%2

5300 / ~333-39%2

3250 / ~202-27%4

3000 / ~13 %2

71010 / 44%10

3107 / 29%5

2155-3006/ 40%10

206 / 30 %4

150 / 11 %8

72 / ~80 %

39 / ~50 %

40 / ~60 %

23 / ~26 %

11 / 0 %

References: Davis & al. (1997)1, Borhidi (1996: 284)2, Zanoni (1989)3, Adams (1972)4, Dietrich (1989a)5, Dod (1984b)6, Dietrich (1989b)7, Ackerman (1995)8, J. Ackerman (pers. commun.)10 and unpublished data of the author.

Taxon. – In almost all taxa we find the same unequal distribution of diversity and endemism among the next lower ranks. Endemism of Cuban Orchidaceae, although ranging among the most diverse families, is far outnumbered by other angiospermous families like Myrtaceae (88%), Rubiaceae (68%), or Euphorbiaceae (67%), let alone certain less diverse but highly endemic families like Arecaceae (90%) or Ericaceae (92%, all data from Capote & al. (1989). Similarly, there are diverse orchid genera with “only” low endemism, e.g. Epidendrum (27%), or with high endemism, as in Lepanthes (98%) on the Antillean level.
Diversity patterns of Pleurothallis come closer to phanerogamous than to orchidaceous conditions (Tab. 14) in some cases. On the greater islands, Pleurothallis shows a much higher level of endemism than the average orchid data, although some of the information may be positively or negatively exaggerated. The high figure in Hispaniola, e.g., has to be dealt with caution. Dod contributed almost 20 new epithets to Pleurothallis and 13 to Lepanthopsis, another pleurothallid genus. However, of the latter 1/3 are synonymous with older names according to Luer (1991). The number of Pleurothallis given for Hispaniola may be overestimated, Likewise, experience from the taxonomic and phytogeographical revision of Pleurothallis in Cuba has shown, that there is much left to be done in this family, hence overall orchid data for Cuba (Dietrich 1989b) may not exactly reflect real conditions either.

Area. – Although the absolute numbers may differ substantially among the three categories, the general proportions in the archipelago follow a similar pattern (Tab. 14). Diversity and endemism rate reflect the size of the respective island, a correlation that was considered by MacArthur & Wilson (1963) in the equilibrium theory of insular zoogeography. Hispaniola bears about the same number of taxa and endemics as Cuba, whereas the number drops abruptly in Jamaica and even more in Puerto Rico. Island size, [page 143↓]however, is not directly proportional to the rates of diversity. Jamaica has an area comparable to East Cuba, which accommodates the majority (36 of 39) of the Cuban taxa. However, Jamaica has a much lower diversity (23 spp.). Puerto Rico, finally, though not substantially smaller than Jamaica, hosts less than half the number of taxa. Plotting the horizontal distribution of Pleurothallis (Fig. 48) over geography (Fig. 72), the resulting chorological pattern is very uneven. The highest concentration both in total and endemics is found in a triangle that comprises E-Cuba, Jamaica and Hispaniola, specifically the mountainous W and middle Hispaniola. Species diversity abruptly drops towards the edges, i.e. W and middle Cuba, E Hispaniola, Puerto Rico and the Lesser Antilles.

Fig. 72: Pleurothallis: α-diversity in several locations of the Antilles: W-Cuba, middle Cuba, Jamaica, E-Cuba, Hispaniola, Puerto Rico, Lesser Antilles (islands not shown in the map sketch).

At least 50% of the Antillean species of Pleurothallis have reached the island arc from the Central American continent, hence, we have to deal with factors that influence dispersal at first to find out more about the Caribbean diversity.

Means of dispersal.Pleurothallis, like most of the orchids, are wind dispersed, although other means have been repeatedly weighed (Garay 1964, Gandawijaja & Arditti 1983, Thornton 1996). Thus it would be only normal to touch upon meteorological issues when discussing orchidaceous phytogeography. Quite the contrary, these features have been, if mentioned at all (Trejo-Torres & Ackerman 2001: 779), only marginally considered (Borhidi 1996, Garay 1964). Only Cox & al. (1997:187) referred directly to major global air currents and their putative role in the realisation of trans-oceanic distribution of orchids. As mentioned before (p. 130), orchidaceous seeds possess strong long distance dispersal capabilities (Gandawijaja & Arditti 1983). The West Indies lay under direct influence of the trade winds with a main direction from the oceanic NE. These air currents are essentially counterproductive in the colonisation of the Caribbean. Coming from the Atlantic Ocean, they virtually prevent any floristic enrichment of the Antillean arc from both (sub)tropical continental joints. Only two areas benefit from this type of wind, Cuba from the Bahamas, [page 144↓]which, however, does not host any Pleurothallis, and Jamaica from E-Cuba and Hispaniola. Apart from these connections, trade winds can explain neither the floristic enrichment of the archipelago in general, nor the level of diversity in the E-Cuba-Jamaica-Hispaniola triangle, which accommodates 98% of the Greater Antillean Pleurothallis.
These biogeographic traits are best explained with another climatic phenomenon of this area, cyclones. Hurri canes have been repeatedly cited in both zoo- and phytogeographic literature as putative means of dispersal, however, never with reference to particular features, like wind speed, tracks or structural traits. Hurricane-force winds can extend outward to about 40 km from the storm centre (eye) of a small hurricane and to more than 250 km for a large one (all data in this paragraph from the National Hurricane Center – http://hurricanes.noaa.gov/prepare/structure.htm). The area over which tropical storm-force winds occur is even greater, ranging as far out as almost 500 km from the eye of a large hurricane. A hurricane's forward speed averages around 25-35 (-100) km/h. Adding circular wind of the storm itself, the resulting speed ranges between 80-250 km/h. Flora (October 1963), one of the most devastating cyclones that hit Cuba, had wind speeds of 210 km/h. Thus, distances between East Cuba and Jamaica (140 km) or Haiti (77 km) are easily bridged within less than one hour. One crucial issue in discussing the translocation by cyclones is that of wind speed which is necessary for the transport for a specific item. Spectacular reports concerning the translocation of “complete huge trees... loaded by thousands of epiphytes” (Borhidi 1996: 50) are most probably a product of human fantasy. Such big items depend on a considerable wind speed. However, if the needed speed exceeds 75 mph then another problem appears. These strong winds “normally do not extend more than 25-50 nmi from the eye and any airborne material that close could very well wind up in the eyewall instead of being transported significant distances around the eye” (Dr. J. L. Beven, Tropical Prediction Center/National Hurricane Center, pers. comm.). Therefore, it is much more likely that seeds are the mean form of dispersal. For the transport of these vagile organisms, even weak storms would suffice, which increases the chance of successful transport.

Now that the potential for bridging the islands by air currents was shown to exist, the question remains if hurricane track directions are in concordance with floristic exchange routes as indicated by present distribution. In the Greater Antilles there are two main types of cyclones. One part forms in the southern Caribbean and passes Central America before hitting the islands (Fig. 73, Fig. 74). Those storms, rarely reach the eastern Greater Antilles. Instead, they pass over Cuba or Jamaica at best, before proceeding to the continental shelf of the southern USA or the Bahamas. Their route excellently explains the floristic connection Central America–Cuba–Jamaica. The only pleurothallid orchids N of the Tropic of Cancer, Pleurothallis gelida and Lepanthopsis melanantha, have reached Florida most probably via this route. The second type develops W of tropical Africa (Cape [page 145↓]Verde type) over the Atlantic Ocean (Fig. 73) These hurricanes frequently cross the Lesser Antilles and the eastern Greater Antilles before turning north. They should be mainly responsible for floristic exchange between the Lesser Antilles and among the Greater Antilles. P. imraei, which occurs on the continent, Hispaniola and Guadeloupe may have found its way with the help of a Cape Verde hurricane from South America rather than from Central America. In P. discoidea (Jamaica, Trinidad, South America) this is evidently the case, since the plant has not been known from Central America so far.

Fig. 73: Average zones of origin and tracks for hurricanes in September. Picture from the National Hurricane Center (http://www.nhc.noaa.gov/HAW/basics/zones_origin.htm).

Fig. 74: Average zones of origin and tracks for hurricanes in October. Picture from the National Hurricane Center (http://www.nhc.noaa.gov/HAW/basics/zones_origin.htm).

In this connection it should be stressed, that transport is possible in both directions (J. L. Beven, pers. comm.), a feature of hurricanes, that is essential in considering migration between the islands. Due to the circular construction of the storm, a hurricane moving from E to W along the Greater Antilles (Fig. 73), can transport items en route as well as in the opposite direction, e.g. from E Cuba to Haiti, e.g.. This represents an important means of transportation from W to the E in the otherwise trade winds influenced Caribbean region. Only cyclones that originate in the southern Caribbean and head NE across the Greater Antilles provide a similar transport (Fig. 74).


[page 146↓]

As Fig. 75 indicates, tracks of individual storms can differ substantially from average routes. Moreover, hurricanes should not be considered mere points on the map with linear tracks. They represent huge climatic complexes that affect vast areas, ranging 100-700 km in diameter.

Fig. 75: Tracks of the main hurricanes in the last century 1900-1980 (Celeiro & Vásquez 1989).

Consequently, even if there are main tracks and directions, virtually every spot in the northern Caribbean can be affected by a hurricane, even more if we deal with such light items as orchidaceous seeds. Flotation time tests of orchid seeds showed an average of 0,72 km/h sink speed (Gandawijaja & Arditti 1983). Thus, these items do not depend on strong winds to be transported over significant distances. Below the line, the most important phytogeographical function of hurricanes is therefore that they break the ubiquitous trade winds from the NE, that would otherwise effectively counteract colonisation from the Central and South American continents. The role of hurricanes in long distance dispersal has been discussed in other anemochorous plant groups, too (Pteridophyta, L. Regalado, pers. commun.).

It should not be neglected that orchidaceous seeds represent fragile items simply by the small biomass that does not provide physical and physiological buffering. During long distance dispersal, seeds are likely to undergo physical stress like UV radiation, low temperatures and desiccation. Especially the latter was used when arguing against the likelihood of long distance dispersal (Garay 1964). Gandawijaja & Arditti (1983) and Arditti (1992: 611), however, state that the actual space of transport time may be rather short, [page 147↓]which seems to be a fact in the case of dispersal by hurricanes. Moreover, they report (unquoted) tests of freezing and desiccation that allegedly showed the hardiness of orchidaceous seeds.

Distance from the continental source area. – This is a logical conclusion drawn from the probability of successful colonisation which should be negatively correlated with distance (MacArthur & Wilson 1963, Gandawijaja & Arditti 1983, Thornton 1996). This pattern is not only caused by the thinning of dispersal units with a growing distance from the source area. It does, moreover, reflect the growing probability of extinction, since remoter islands benefit less from the rescue effect (Thornton 1996). In the case of the Greater Antilles it should be therefore expected to find a decline in α-diversity from West to East. This pattern should be polarised even more since hurricanes from Central America are less likely to reach the eastern islands (Fig. 73, Fig. 74) diminishing the frequency of direct colonisation. To assess the impact of distance from the mainland, we can only consider those 15 taxa that have colonised the arc without further speciation, since the frequency of successful colonisation in ana- or even cladogenetically active taxa cannot be determined with the present molecular and morphological data. Fig. 49 shows a shallow decline in species from Cuba and Jamaica (both each 12 taxa) over Hispaniola (11) to Puerto Rico (8). It is less contestable when considering only those taxa absent in the Lesser Antilles to exclude colonisation from the other direction. Here we find the islands in the same order with 7, 6, 5, and 4 taxa respectively.

Thus, the low diversity in species in Puerto Rico is partially caused by the great distance from the continent considering the present routes and means of dispersal, which accounts for the impoverished Pleurothallis flora of the Lesser Antilles too! This distribution, that was described as the “Western Continental” by Howard (1974) is found in numerous other anemochorous plant groups too (Howard 1974, Judd 2001). On the other islands of the Greater Antilles, anagenesis and especially cladogenesis accounts for the main part of the flora. The former is clearly dependable on migration frequencies from the continent. The close position of Cuba to the mainland is reflected by the fact that at least 56% of its taxa are directly related to continental ancestors. In contrast, Hispaniola, with a similar diversity, shows a greater portion of autochthonous species which is partially due to its greater distance from Central America. To elucidate the processes that have caused the Greater Antilles patterns of distribution in Pleurothallis, the subsequent discussion must therefore deal with features and processes that influence speciation on the Greater Antilles islands.

[page 148↓]Ecologicalpreferences

Water supply. – As was shown earlier (p. 129), Cuban Pleurothallis require a minimum of 1200-1400 mm with an additional water supply due to a favourable microclimate providing constant high humidity (p. 89) in most habitats. This strongly suggests that non-liquid water (humidity, fog and clouds) plays a major role in the pleurothallid water balance. Apart from the mentioned microclimatic exception, the Cuban climate is marked by two major sources of precipitation, macroclimatically from the NE (trade winds) and mesoclimatically from the mountainous condensation belts. The increased diversity in mountainous belts is easily explained by a positive water balance due to the reduced evaporation (lower temperatures) and additional water supply by clouds and fog. Furthermore, canopies of lowland forests, owing to a restricted water supply (rain) and little retention capacity, can suffer a severe drought in epiphytic habitats during extended periods of the day (Freiberg 1992). Thus, although Cuban species of Pleurothallis occur at virtually all altitudinal belts from sea level to the summits of the Turquino group, the greatest diversity can be found between 300 and 1300 m where at least 10 of the 39 Cuban taxa can be found in all 100 m belts (Fig. 54). The altitudinal concentration of orchids (pleurothallids) at middle to lower mountainous elevations with a substantial decline in both high montane rainforests and at lower altitudes, was reported for Lepanthes (Hespenheide & Dod, 1990) and orchids in general on the island of Puerto Rico (Woodbury 1974). Cuban Pleurothallis, in contrast, can be found in considerable numbers even at lower elevations (200-700 m). This is most probably related to the high species diversity in the submontane Nipe-Baracoa range. Descending vegetation belts in serpentine areas were reviewed by Borhidi (1996: 138) as a general phenomenon. This vertical anomaly is facilitated in part by the high precipitation rates (Borhidi 1996: >3000 mm/a) in this region and the subsequent existence of numerous creeks and rivulets that, almost exclusively, serve as habitats for a great number of Pleurothallis species even at lower altitudes (Fig. 57). The sudden drop in species diversity in high mountain rainforests, on the other hand, is unlikely a result of the high rainfall as suggested by (Woodbury 1974). As indicated, Pleurothallis occurs in a great variety under very wet conditions in Nipe-Baracoa. The observed imbalance is caused rather by an unequal orography in Oriente. Limestone and serpentine mountains do not reach beyond 1200 m. All species that are found above this limit are associated with the geologically more uniform Sierra Maestra range (Fig. 54). Thus, the decline above 1200 m reflects actually the general poorness in species diversity in the Sierra Maestra (17 taxa), compared with that of the diverse lower altitudes of NE Oriente (35 taxa). The imbalance in α-diversity between serpentine and magmatic rock, has been observed in other plants groups, too (Borhidi 1996; E. Köhler pers. commun.: Buxaceae) and might be common. The relative [page 149↓]poorness in Pleurothallis found in the mountainous regions of W and middle Cuba, may be again associated with water supply. Here, instead of lacking precipitation, a petrologic feature may cause the decline in species diversity: habitats in these regions are almost exclusively found on limestone. This karstic rock is known to have little retention capacity for water. Water surplus is often directly drained by subterranean rivers and therefore immediately lost for the (epiphytic) vegetation, which strongly depends on favourable meso- and microclimatic situations. Localities were rivers prevail on the surface belong to classic collections sites for Pleurothallis: Taco Taco in Pinar del Río, Salto de Vegas Grandes in the Escambray mountains.

Geological restriction. – Apart from the secondary influence of the rock type on the water balance, petrologic features turned out to be closely associated with pleurothallid species diversity (Fig. 60). Especially striking is the level of ecological endemism ‘on’ serpentine. With 25% of the Cuban taxa, Pleurothallis surpasses the phanerogamous average by far (Reeves & al. 1996: 14%). The presence of the genera Pleurothallis and Lepanthes on serpentine in NE Cuba has been dramatically missed in favour of the putative dominance in the Turquino region (López Almirall 1994), which was seen as the centre of origin. López Almirall & al. (ibid. p.464) even denied the existence of Lepanthes in Moa, a region where at least 10 species were collected in the course of this study. The restriction of Pleurothallis to specific rock types was an unexpected result of the present study since epiphytism and edaphic issues are not considered necessarily interrelated issues. Moreover, in Cuban Pleurothallis, the relevance of rock types is not restricted to the (facultative) epilithic taxa, as may be assumed. By way of contrast, most of the lithophytic taxa belong to the euryoecious group of widespread taxa (P. corniculata, P. gelida, P. obovata, P. sertularioides, P. tribuloides), whereas petrologic restriction is most prevalent among endemic epiphytes from northern Oriente. Only two taxa are epilithic, P. bissei and P. papulifolia. Petrologic restriction of epiphytic orchids has been described only occasionally. Dietrich (pers. comm.) reported that Oncidium undulatum (Sw.) Salisbury is confined to limestone and Hespenheide & Dod (1990) reported a Lepanthes (‘kárstica’) being restricted to karstic zones. Trejo-Torres & Ackerman (2001) found evidence for a general floristic relationship of the geologically similar islands of the ‘calcareous group’ based on the entire orchidaceous flora, however. Another, though only empirically inferred, trait which is inherent to the ecological distribution of Pleurothallis in Cuba is the fact that the group of species which most frequently inhabits limestone niches is composed primarily of the widespread taxa mentioned above. This calciphilous character of the Pan-Caribbean species should be no surprise on the largely limestone dominated Caribbean Arc.


[page 150↓]

Fig. 76: Restriction of the Cuban taxa to certain types of rock. Included are distinct inter-island species pairs and indicators (‘spp.’) where multiple sister taxa on other islands would be placed according to morphological and palynological data. Since the latter comprise several taxa, geological classification is generalised.

Letters after bold-typed endemics refer to the island (C – Cuba, J – Jamaica, H – Hispaniola).

Which processes precisely influence geological patterns of pleurothallid distribution remains a speculative matter. Theoretically, the limitation of epiphytic orchids to geologically defined areas could be the result of direct physiological adaptation to the chemistry of the respective rock. This mechanism is seen as an ubiquitous and irreversible phenomenon of the terrestrial vegetation on serpentine Borhidi 1996: 133). In epiphytes, however, this may not hold for several reasons. First, it is unknown how high the concentration of heavy metal ions on the bark is, i.e. to what degree, epiphytic plants [page 151↓]get in contact with them. Several Buxus and Leucocroton species which are restricted to ultramafic rock are known to hyperaccumulate Ni in their leaves (Reeves & al. 1996) and some of them do serve as phorophytes for species of Pleurothallis, e.g. P. llamachoi on hyperaccumulating species of Buxus gonoclada compl. and P. ekmanii on Leucocroton sp. It is likely that the bark, which serves often as a physiological waste disposal, reflects the geological traits just as the leaves do, although leaves should be a preferred place of storage for their ephemeral character and due to the fact, that many (sub)tropical trees do not build up a thick bark as do their temperate relatives (Vareschi 1980: 62). Nevertheless, even if we assume that the epiphytic plants come in contact with considerable amounts of serpentine bound heavy metals, a physiological, taxon specific tolerance is rather unlikely, since closest relatives of serpentine endemics are often restricted to non-ultrabasic rock (cf. further down). A post-Pleistocene origin of the Antillean endemics, as assumed in this study, would imply the physiological back- and forth adaptation to the chemistry of different rocks within 10000 years. Secondary ties by locally restricted fungi and/or pollinators are more likely, owing to the fact that both are probably in closer contact with the substrate. Due to the great age of these organisms, a physiological tie to the specific type of rock with a subsequent geologically defined area of distribution is likely. Unfortunately, little is known about both fungi-host and pollinator-plant specifity. Studies on the specifity of endophytic fungi (Currah 1997, Bayman & al. 1997) showed mixed infections in the same plants and a great heterogeneity in number and type of endophyte among species. The latter even presented evidence for shortcomings in widespread identifying methods that may have caused failure in extracting the whole set of fungi present in the root and/or other organs. Therefore, the fungal diversity may be even greater. Logically, co-evolution with a particular fungus of restricted ecological amplitude would narrow down the bottle neck of seed germination even more (Bayman & al. 1997). A widespread specialisation on yet another co-organism beside the pollinator is less likely and in fact, that is what the studies essentially show.

A pollinator dependent geographic restriction, on the other hand, would require a geologically confined biogeography of the insect. Interestingly, conservative estimates on the biogeography of Cuban insects, suggest a closer relationship between E Cuba and Hispaniola (Genaro & Tejuca 2001). Yet, in contrast to the vast number of species, profound studies on pollinator-plant interactions in pleurothallid orchids are by far under-represented (reviewed in Van der Cingel 2001; Chase 1985; Christensen 1992; Dod 1986c; Duque 1993; Mesler & al. 1980; Borba & Semir 2001; Blanco & Barboza 2001) and information on the distribution of pollinators are virtually absent. Borba & Semir found indications for a geologically defined biogeography in Brazilian species of Pleurothallis, which they interpreted implicitly by pollinator biogeography. In Cuba, pollinators of these presumably myophilous plants (Van der Pijl & Dodson 1966; Christensen 1994; Borba & [page 152↓]Semir 2001) should be basically substrate-bound, since most growing sites were found in 0-2.0(-3.0) m height above the ground. The ecology of the insects should therefore receive direct influence from the edaphic chemistry. Similar to the high level of endemism in Borhidi’s Eu-Moanicum (Borhidi 1996, López Almirall & al. 1984) regional ecological restriction could be a zoogeographical phenomenon too. Moreover, dispersal in orchids could be very similar to that of pollinating insects, since Johnson (1969) found evidence for a correlation between meteorological pathways and insect migration “in many cases” (cited in Close & al. 1978). A switch from one to another edaphic type should be therefore associated with the loss of the traditional pollinator set or parts of it, which in turn should effectively prevent expansion when no adaptation to the new environment occurs. This is backed by the geographical structure of pleurothallid endemism in Cuba, which is different from many other plant families. In contrast to the predominantly local, not regional [!] phanerogamous endemism (>500 of ~1000 endemic taxa according to Borhidi (1996) in Nipe-Baracoa, which has been assumed for orchids, too (Dietrich 1989a), most of the serpentine endemics of Pleurothallis are found throughout the area from the Sierra de Nipe to the Jauco, i.e. along the whole spine of serpentine in Oriente. The notion that pleurothallid distribution in Cuba might often be confined to individual ridges and hills turned out to be rather a collection artefact.

Speciation

Until now it was shown, that species distribution depends mostly on factors related to dispersal, (palaeo)climatic traits and ecological preferences. However, these features explain only in part why speciation has been so intense just in the triangle of Cuba’s Oriente, Jamaica and western Hispaniola. To elucidate the underlying patterns, speciation processes would have to be analysed in detail. This is where the content of data gathered during this study limits further discussion. In the absence of data covering such important steps in the orchidaceous life circle as pollination, it is almost impossible to draw serious conclusions. Thus the following paragraph on speciation is aimed mainly on the development of some hypotheses on speciation, which are based on the present distribution and ecological preferences of the Cuban taxa.

Allopatric speciation. – Phylogeographic data of the Antillean species of Pleurothallis indicate that speciation has occurred both in allopatric and sympatric processes. The respective portion is difficult to ascertain owing to the limited molecular sample size. Allopatric events have led at least to ¼ of the Antillean Pleurothallis flora, i.e. those species which migrated from the continent. Among autochthonous taxa, allopatric speciation seems to be the rule too. First, there are many examples for groups of species with mutually exclusive distribution, e.g. in subgen. Antilla. Second, the ITS based phylogenetic tree shows often greater genetic distances between sympatric species of [page 153↓]Cuba than among those with no overlapping distribution (e.g. P. mucronata-llamachoi-longilabris, P. shaferi-grisebachiana) which is probably due to the omission of the closest relatives that occur on other islands. Third, there is much evidence for frequent geological vicariance in the three radiative lineages (Fig. 76). Although many cases of vicariance can be inferred only indirectly by unusual branch lengths, e.g. subgen. Antilla, some species pairs can be inferred directly from the tree. It turned out that vicariance occurs in three qualities.

  1. Geographic and geological vicariance, i.e. species occur on both different islands and types of rock. This is apparently the most frequent case and occurs in the grisebachiana-group of clade G (subgen. Specklinia sect. Hymenodanthae) and the papulifolia-group of clade B (subgen. Antilla). Two examples: P. shaferi- P. simpliciflora (Hispaniola, limestone) and P. bissei (Cuba, serpentine) – P. hirsutula (Jamaica, limestone). In subgen. Antilla, the three Cuban endemics are restricted to limestone, serpentine and magmatic rock, while their sister taxa occur in Jamaica and most frequently in Hispaniola.
  2. Petrologic vicariance within islands. This can be found in the aristata-group. All but one endemics are restricted to serpentine and are of sympatric distribution (Fig. 76). Genesis of this group may have been the result of several allopatric speciation events combined with a change of the geological background: P. llamachoi (serpentine) àP. obliquipetala (volcanic rock) àP. longilabris (serpentine). This type of vicariance is rather rare. It represents actually only a physiographic subtype of #1. Geological combined with island vicariance is much more frequent, which may be due to the extended areals of serpentine, volcanic rock and limestone in Cuba, Jamaica and Hispaniola respectively.
  3. Geographic vicariance on the same petrologic ground. This can only be inferred indirectly from the molecular tree, however it is backed by morphological and palynological data. Among the Cuban taxa sampled no pair of species falls in this category. However, Hispaniola with its extended and numerous limestone areas may have species that show this type of vicariance, unless they form sympatric pairs.

Sympatric speciation, in turn, seems to be a rather rare event. It may have been the case in the aristata-group (Fig. 76), assuming that P. longilabris originated from P. llamachoi before it gave rise to P. obliquipetala. However, P. obliquipetala may be descended equally from P. llamachoi, as was just pointed out. Sympatry can be observed most clearly in P. gemina and P. wrightii, two endemics of the Nipe-Baracoa-range which share the same ITS sequence. Moreover, sympatric evolution may occur in Haiti more frequently, where a lot of taxa from the three radiative lineages have been described [page 154↓]uniformly from limestone habitats.

Thus, allopatric speciation accounts for the majority of infra-Antillean speciation in Pleurothallis, resulting frequently in inter-island and geological vicariance, i.e. ecological shift. Hence, pleurothallid evolution in the West Indies is based mainly on founder events, a common pattern of speciation in archipelagos (Crawford & al. 1987). Baldwin & al. (1998: 426ff) reviewed studies on the evolution on oceanic islands, finding examples for both inter- and infra-island radiation, depending on the taxon and the island group. Similarly, speciation events may or may not parallel environmental shifts (Francisco-Ortega & al. 1996). Unfortunately, little reference is made to chorological patterns and geographical features in the review (ibid. p.429). Judd (2001) found most radiation within Antillean island boundaries in Lyonia. Although the seeds are anemochorous this is most probably due to the lower vagility, compared with pleurothallid seeds.

Speciation triggering founder effects comprise the following traits.

  1. Change of the genetic structure due to increased inbreeding and subsequent increase in homozygosity (Mayr 1954). In this context it should be pointed out that observations in the field suggest, that founder events in Cuban Pleurothallis are most probably launched by selfing. Species pairs are separated by 30-700 km. Considering the presumably low radius of pollinators, small dipters above all, selfing must be an inevitable result of this situation. In orchids, the impact on the genetic structure, like increased homozygosity, should be therefore even stronger.
    Autogamy, suggested by degenerated pollinia, complete fruit set with or without anthesis, is widespread among the widespread Cuban taxa (Tab. 5), indicating that selfing is more common among Pleurothallis than assumed (Catling 1990). It has not been observed yet in Cuban endemics. Cross-pollination is ensured by a predominatly subsequent anthesis at first sight. Yet, a closer look reveals that several traits have been developed to allow parallely open flowers. Thus, P. wrightii has one-flowered inflorescences, which are produced in pairs on one stem. The recently from P. wrightii evolved P. gemina, by contrast, produces single two-flowered inflorescences, however, with a simultaneous anthesis! The same applies to P. shaferi. Plants of the endemic subgen. Antilla have many-flowered inflorescences with parallel anthesis. Even in species with subsequently multi-flowered inflorescences several secondary stems show synchronised flowering, thus ensuring multiple open flowers on one and the same plant. Borba & al (2001b) showed that in Brazilian Pleurothallis selfing results in dramatically low fruit set with decreased seed viability. However, intact [page 155↓]seeds remained in almost all tests. Selfing, though payed with a substantial drop in viable seeds, may be therefore a viable step in a colonisation event.
  2. Random change of allele frequency due to genetic drift (Wright 1931).
  3. Change of the environment resulting in new directions of selection (Wright 1931). This is one of the essentials of founder events, the role of which can be observed in Antillean Pleurothallis by the frequency of ecological vicariance (pollinator set distribution defined by geological traits).
  4. Templeton (1980) added the feature of genetic variability in the founder population to his model of genetic transilience. Genetic variability was found to be a common trait in pleurothallid (Borba & al. 2001a; Tremblay & Ackerman 2001) and other orchids (Ackerman & Ward 1999), based on isozyme data. Templeton (1980: 1013) pointed out that isozyme data, which is widely used to ascertain the genetic variability in populations and species may not reflect the genetic variability that is essential for adaptations of pollination and major traits at the life cycle level. Similarly, under certain conditions selfing is not necessarily accompanied by loss of adapability (cf. Takebayashi & Morrell 2001: 1144). As founder events, as was hypothesized, may be based on single plants among pleurothallid orchids, genetic variability as inferred from isozyme allele variation, may be in fact of secondary importance for the adaptive potential in the new environment.

The prerequisite for a founder event, reproductive isolation, is most likely not induced by distance, although this parameter, as was shown, clearly influences dispersal, and therefore gene flow, too Ackerman & Ward 1999). Although only indirectly inferred, most geologically restricted taxa in Cuba are apparently confined to their areal by pollinator distribution, which is illustrated by the fact that most of them occur exactly within the complete geologically defined area. These species are therefore obviously limited to an area that is much smaller than the potential extent based on dispersal traits (seed production, air currents etc.). This, in turn, is backed by the multiple speciation events that have occurred in Pleurothallis in the Caribbean, which per se represent extra-zonal migration. Moreover, the potential of pleurothallid dispersal can be seen in widespread taxa with obligate or facultative autogamy (Tab. 5), which are independent of pollinator distribution5. Based on this ‘artificial’ geographical restriction, it would be natural to assume that there is at least a moderate gene flow within populations of endemic taxa, which in turn would account for the scarcity of sympatric speciation and low probability of [page 156↓]speciation due to genetic drift. However, several studies have shown, that even restricted gene flow does not inevitably result in genetic and morphological differentiation, leading to speciation (Ackerman & Ward 1999; Borba & al. 2001a; Tremblay & Ackerman 2001). As Ackerman points out, founder populations must be ‘sufficiently geographically isolated’ and, as indicated in this study, must be confronted with a new environment, i.e. potential pollinator sets. Interestingly there are indications, that the insect fauna of E Cuba is most similar to that of Hispaniola (Genaro & Tejuca 2001). As we have seen in Pleurothallis, biotic similarity may indicate congruent dispersal patterns, hence, there may be concordant patterns of distribution and speciation in Diptera. Unfortunately, little or none information is available concerning ecological patterns of distribution in flies, which in turn could coincide with pleurothallid phytogeography. A correlation between the ecogeography of the two organisms would elegantly demonstrate the evolutionary potential of co-evolution.

As a result, the pleurothallid floristic richness within the triangle E Cuba – Jamaica – Hispaniola can be traced back to the following features. First, geological diversity has apparently brought about a great variety of pollinators suitable for these orchids. The floristic richness of Cuba’s serpentine flora has been related traditionally to the age of the region (Borhidi 1996: 129). López Almirall et al 1984: 447), in contrast, implicitly indicate that unfavourable conditions (physiological aridity, ion toxicity etc.) and consequently high selection pressure have led to accelerated speciation. This is essentially the pattern in Pleurothallis! This plant groups shows that intense speciation in this region is mainly due to the spatial proximity of limestone and serpentine and, to a lesser degree, volcanic rock. In fact, the pleurothallid endemism rate of W Hispaniola (Hespenheide & Dod 1990: Lepanthes; data from herbarium material of Pleurothallis), which is composed almost exclusively of limestone, may reflect that of serpentine areas in Cuba’s N Oriente. Second, the triangle comprises the most extensive mountainous area in the Greater Antilles, providing optimal climatic conditions for microphytic epiphytes. Third, changes in the environment trigger speciation. In the Greater Antilles, this is enhanced by the mentioned hurricane activity, which ‘increases’ the instability of environmental characteristics directly by the mechanic impact on the habitat (Walker & al. 1980; Rodríguez-Robles & al. 1990) and indirectly by frequent translocation of plants and seeds.

Considering the vicariance-dispersaldiscussion mentioned before (s. p.130), the pleurothallid way of speciation provides a good example for the irrelevance of this dispute. In my mind, dispersal and vicariance processes depend simply on the point of view, whether an separating barrier is effective prior (dispersal) or after establishment of the ancestral areal (vicariance). The irrelevance of this classification becomes evident in the case of the South America – Africa disjunction (Wolfe 1981). Following the slow drift of the [page 157↓]two continents it is impossible to determine, when exactly dispersal stopped and vicariance set in, even more when completely ignoring specific dispersal traits. Both schools have the problem of static assumptions at a certain point of their train of thought: ‘dispersalists’ proceed along the idea of a stable geological constellation (Page & Lydeard 1994) and vicariance biogeography is based on the inalterability of distribution, i.e. ancestral and present distribution is the same (Bremer 1992). Consequently the former was most popular before plate tectonics was established, whereas the latter gained much benefit from Wegener’s theory (Page & Lydeard 1994). However, both processes exist, and dispersal should be more prevalent in vagile taxa within a stable, not uniform!, environment, while vicariance in sedentary taxa within an inconstant area. Finally vicariance will not do without dispersal (Hedges & al. 1994), and dispersal results in vicariant taxa as well. Whether a barrier, oceanic gap, or mountains etc., effectively hinder gene flow right from the moment the ancestral areal has colonised or later due to secondary changes in the environment, can often assessed only after phenotype or genotype differentiation has started. Moreover, these processes are reversible, when gene flow is resumed due to changes in environmental factors. Given the present-day distribution is congruent with the historical, Caribbean data of pleurothallid phylogeography suggest a strict dispersal pattern, mainly due to orchidaceous vagility and the stable geographical environment. In contrast, if it was that part of the pleurothallid flora resulted from forest fragmentation due to climatic oscillations during the Holocene (Curtis & al. 2001), then the speciation events would be clearly referable to the vicariance model. This, however, is fairly unlikely, given the great vagility of orchidaceous seeds.


Footnotes and Endnotes

1 Some taxa were tentatively included in Pleurothallis s.str., “pending DNA sequencing“ (Pridgeon & Chase 2001).

2 All recent taxonomic publications in this connection (Luer 1986b, 2000; Pridgeon & Chase 2001; Pridgeon & Chase 2002) cite Pleurothallis ghiesbreghtiana A. Rich. & Galeotti under the invalid name Pleurothallis racemiflora Lindl. ex Lodd.

3  Pleurothallis corniculata, though classified among those plants with a congested inflorescence (subsect. Apodae-Caespitosae) by Luer, must have evolved from a racemose ancestor by reducing the flower number rather than the rachis length. There is one collection with a racemose inflorescence (duplicate of HAJB 80113 in herb. Greuter) that shows the presumably ancestral state with racemose inflorescences.

4 According to this point of view, Grand Cayman would have had the strongest floristic affinities with Cuba, since it shares 100% of its Pleurothallis flora – P. caymanensis – with Cuba.

5 Of course, there are more reasons that have added to the successful migration in these plants, e.g. high fruit set.



© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 3.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
05.04.2004