Hajdu, E. (1995). Macroevolutionary Patterns Within the Demosponge Order Poecilosclerida. Phylogeny of the marine cosmopolitan genus Mycale, and an integrated approach to biogeography of the seas. Ph.D.Thesis (University of Amsterdam). 1-173.
Macroevolutionary Patterns Within the Demosponge Order Poecilosclerida. Phylogeny of the marine cosmopolitan genus <i>Mycale</i>, and an integrated approach to biogeography of the seas.
Five out of seven chapters in this thesis report on poriferan
systematics. This is so because a phylogenetic framework is either weakly supported or entirely lacking for most groups of sponges. The establishment of such a framework is the necessary first step in revisional studies. The upper bound of my revision was set at the Order Poecilosclerida. The first chapter of results, Chapter 2, reports on the sta tus of this order. Its monophyly an d sister-group relationship to the order Haplosclerida are supported. Characters
thought relevant at this level of universality were discussed, and underlying synapomorphies proposed for the supra ordinal assemblage, as well as for each of the orders. Chelae were added to Van Soest's (1991) list of synapomorphies for both orders, viz., sigmas, toxas and larvae with a bare posterior pole. Crucial to this understanding of both orders was the postulation of the polyphyly of the poecilosclerid family Desmacididae. Isodictya and Cercidochela were transferred from this family to the order Haplosclerida, along with Coelocarieria, a genus formerly assigned to Coelosphaeridae/Cornulidae. The splitting of the Desmacididae goes on to Chap ter 3, until only Desmacidon and Desmapsamma remain in the family. A preliminary revision of over 300 nominal poecilosclerid genera has been under taken here, searching for assemblages supported by synapornorphies, and also to infer the best assignment for several desmacidid genera, artificially kept together because of lack of characters presumed diagnostic for the remaining poecilosclerid families. The likely synapomorphic value of chelae morphotypes (palmate, arcuate, anchorate, unguiferate, birotulate, denticulated = placochelae-related) is stressed, in contrast to the former poecilosclerid system where skeletal architecture played a pivotal role in the classification. This allowed an unequivocal transfer of desmacidid genera to the Cladorhizidae, Coelosphaeridae (new concept), Crambeidae, Guitarridae (rescued assemblage), Iophonidae (rescued assemblage) and Myxillidae (new concept) or its vicinity (= Myxillidae s.l .). Also, poecilosclerid families were
recognised to belong to three major clusters, for which new suborders were erected, viz . Microcionina, Myxillina and Mycalina. Synapomorphies supporting these suborders are respectively: marked
megasclere specialisation + loss of sigmas, tridentate chelae + loss of toxas and mycalostyles + sigmancistra-derivatives. Microcionina + Myxillina were postulated as more closely related to each other than either is to Mycalina, on the basis of two shared novelties, viz., non -verticillate acanthostyles + special ectosomal megascleres. Other poecilosclerid characters are discussed and likely parsimonious in terpretations are forwarded. Chapter 4 focuses on relationships within the Mycalina. The starting point for this chapter was the recognition that Hypsispongia is actually an artefact due to contamination by extraneous microscleres. This prompted me to revise hamacanthid
relationships, resolved with the latter's sister-group relationship to Mycalidae. Desmacellidae is postulated as the sister to Hamacanthidae + Mycalidae. The sister to all three families is the clade comprising Cladorhizidae + Guitarridae. Relationships among these families are still very weakly supported. In the next step (Chapter 5) relationships within the Mycalidae are assessed. A revision of South American Mycale (Mycale) - eight species , of which three were new - revealed several characters never discussed in a phylogenetic perspective before. A phylogenetic terminology is applied which implies homology of categories across species borders, to be recognised on the basis of comparative analyses of taxa. Input taxa for the cladistic analysis performed were all Mycale subgenera, to which Esperiopsis, Hamacantha and Desmacellidae were added, the
latter as an outgroup. Both subgenera Grapelia and Esperiopsis were split into two components each to account for their plasticity (multistate taxa sensu Swofford & Begle, 1993). Results support the
phylogenetic framework established in Chapter 4. Hamacantha is recognised as sister to Mycalidae, Esperiopsis indicated as paraphyletic and the monophyly of Mycale is supported. Subgenera Grapelia + Mycale are those more strongly supported by synapomorphies. The concise aspect of Grapelia and its tropical/southern-temperate distribution were decisive for its selection as first taxon for an analysis of phylogenetic relationships among species (Chapter 6). Nine species were recognised, of which four were new. Mycale (Grapelia) australis was split into three Operational Taxonomic Units (OTUs) as in Chapter 5, to account for its plasticity. The Optimal Character Compatibility Index (OCCI - Rodrigo, 1992) indicated a single most supported tree. The monophyly of the M. (G.) australis OTUs is corroborated and the species has been treated as a single OTU in the subsequent analysis. The resulting cladogram is highly unbalanced: «««(ancorina + vansoesti ) + (menylloides + trichophora)) +carteri) + unguifera) + vaceleti) + australis) + burtoni). Mycale (Mycale) was the natural next target. In Chapter 7 an assemblage of 23 species (four of which new) was found to share some synapomorphies with M. (Grapelia), thus implying the paraphyly of subgenus Mycale. These synapomorphies pertain to microfeatures of the anisochelae1, viz ., a shaft which is curved in profile view, a foot with alae extended downwards forming a basal pore, and head with a reduced height (relative to total anisochela height). Relationships of all 32 species [curved assemblage: 23 Mycale (Mycale) + 9 M. (Grapelia)] are analysed
cladistically. A composite of subgenera Aegogropila, Carmia and Paresperella is used as an outgroup. Additionally, five other species of M. (Mycale) are selected to check the monophyly of the target group (curved assemblage) . Multiple most parsimonious trees were obtained as a result of the cladistic analysis. The OCCI proved ineffective to select among rival solutions. I pursued an alternative approach which consisted in subsequently weighting characters by their maximum Cl
(Consistency Index) and quantitatively summarising results through Nelson consensus (= majority rule consensus with other compatible groupings). Results supported the monophyly of M. (Grapelia),
although relationships were slightly altered relative to Chapter 6. Monophyly of the whole curved assemblage is also supported, and monophyly of the remaining M. (Mycale) is possible. Several species groups are recognised. I used the cladogram of the curved assemblage as the raw material for an integrated biogeographic approach, where a current cladistic biogeographic protocol (BPA - Brooks Parsimony Analysis) was complemented by No Assumption analysis, tree reconciliation, PAE (Parsimony Analysis of Endemicity) analysis and panbiogeography. Area cladograms are obtained for the curved
assemblage under the BPA and No Assumption protocols. These preliminary analyses were followed by the addition of other published poriferan phylogenies, viz., Acarnus, Ceratopsion / Thrinacophora, Clathria procera spp. group, Didiscus, Ptilocaulis / Reniochalina and
Rhabderemia, selected on the basis of their possession of considerable distribution in the southern seas. The No Assumption approach is defended as theoretically more consistent, given the expectation that areas may have had multiple patterns of relationships. MLT (multilocality taxa) is introduced as a replacement to the current a prioristic notion of 'widespread' species. General area cladograms are generated through both protocols (BPA + No Assumption) , and individual phylogenies mapped onto them . Statistics derived from these reconciliations (total leaves / TL, leaves added /A, duplications/Dup. losses/L) are found to be size dependent and accordingly, rescaled indices (respectively LNI, ANI , ONI, LoNI) were introduced where the former statistics are divided by the number of OTUs + HTUs (hypothetical taxonomic units - nodes). Real world data
confirmed the preliminary findings on paper examples of Van Soest (submitted) that No Assumption analysis performs better in tree reconciliation studies. Of special interest is the finding that ONI and LoNI were virtually uniform among the seven studied clades, thus implying that duplications and losses may be related to factors extrinsic to the taxa. Duplications have recently been equated with non-vicariant speciation and losses with extinctions. My results support the notion that both phenomena may be guided stochastically . In the case of extinctions, findings corroborate the idea of contingent evolution (e.g. Gould, 1991; Raup, 1993). Panbiogeographic track analysis (fig . 7.14) polarized with the cladogram of the curved assemblage was performed and found to be retroactively illuminating on phylogenetics . An interpretation of ambiguous tracks is suggested where these are postulated to link distinct localities of a single area of endemism. Homoplasy in the paradigm of biogeography is discussed and two classes recognised, those to be minimised (ad hoc hypothesis about the history of taxa or the practice of systematists) and those to be preserved (presumed plurality of patterns - 'geoplasy'). The former should not generate pattern, thus conflicting signal derived from this source is assumed to be weak. The latter may generate conflicting patterns, with conflict assumed to be strong. It is defended that subsequent weighting of taxa (OTUs + HTUs) as characters for their areas of occurrence is an objective way to sort among these classes of conflict. It is postulated that stronger noise will be preserved, and a case is made for recognition of conflicting patterns in a single area cladogram. Finally, I discuss the biogeographic pattern obtained under the No Assumption protocol, both from the general area cladogram (fig. 7.10) and from that of the curved assemblage (fig. 7.4.). Recognition of Gondwanan descendants amid extant tropical marine species is possible. PAE results are meagre due to the low number of MLTs in my data sample. PAE analysis may be a theoretically consistent protocol for the recognition of
areas of ende mism, but it is not for the inference of relationships among them.