BULLETIN DE L ' INSTITUT ROYAL DES SCIENCES NATURELLES DE BELGIQUE,

BULLETIN VAN HET KONINKLIJK BELGISCH INSTITUUT VOOR NATUURWETENSCHAPPEN,

BIOLOGIE, 66 suppl. : 81-88, 1996 BIOLOGIE, 66 suppl. : 81-88, 1996

Choosing among Poriferan morphological characters within the cladistic paradigm

by Eduardo HAJDU & Rob W.M. VAN SOEST

Abstract Introduction

A protocol is proposed by which certain classes of characters may be selected for use in phylogenetic reconstruction due to their overall better consistency in phylogenies. Alternatively, they could receive additional weight as opposed to classes that show an overall poor consistency and should be down weigh­ted. We recognized 16 classes of characters within previously published data matrices of 17 poriferan phylogenies involving 221 OTUs (Operational Taxonomie Units, = terminal taxa). Only five classes occur in samples of large enough size that can allow discussion of observed trends. The high consistency observed in choanosomal architecture characters is possibly an artefact. Megascleres and microscleres show opposing results and these are discussed with reference to function and adapta­tion. In general, results are deemed preliminary because sample sizes are too small for the majority of recognized classes of characters, and different classes of characters may perform dif­ferently in different taxa, a suspicion which calls for an even larger sample base. Keywords : cladistics, character selection, Porifera, phy­logenetic systematics.

Resumo

Um protocolo é proposto através do qual algumas classes de ca-racteres podem ser selecionadas para reconstruïSo filogenética por sua consistência sensivelmente meihor, ou podem receber peso adicional opostamente a classes que apresentam baixa consistência e deveriam receber pesos menores. Nós reconhe-cemos 16 classes de caracteres dentre as matrizes de dados pu-blicadas para 17 filogenias de poriferos compreendendo 221 OTUs (Unidades Taxonómicas Operacionais, = taxons termi­nals). Apenas cinco classes ocorrem em amostras grandes o su-ficiente para que se possa avan^ar uma discussSo dos padroes observados. A alta consistência observada em caracteres da ar-quitetura coanosomal é provavelmente um artefato. Megascleras e microscleras apresentam resultados opostos, e estes sao discutidos com referenda a fun^ao e adapta^ao. No ge-ral, OS resultados sao vistos como preliminares porque o ta-manho das amostras é muito pequeno para a maioria das classes de caracteres reconhecidas, e porque diferentes classes de carac­teres podem ter desempenhos diferenciados em taxons distin-tos, uma suspeita que implica na necessidade de uma base amos-tral ainda mais ampla.

Palavras chave : cladistica, sele^ao de caracteres, Porifera, sistemética filogenética.

The recognition of homologies is a crucial issue in systematics because it is through homologies that relationships among taxa are inferred (e.g. DARWIN, 1859; PINNA, 1991; FOREY et al, 1992; NELSON, 1994; SMITH, 1994). PINNA (1991) stated that ho­mologies sensu latu may actually be classified into primary (those that generate hypotheses of homo­logy) and secondary (those that legitimate hypo­theses of homology). Primary homologies are those detected on the basis of similarity alone (e.g. axial condensation in Axinellidae and Raspailiidae), while secondary homologies are the outcome of a pattern-detecting analysis (e.g. acanthostyles of Raspailiidae and remaining poecilosclerids). This interpretation supports PATTERSON'S (1982) equa­tion of homology with synapomorphy.

A phylogenetic analysis may be outlined as a two step procedure then, where the initial assembly of a data matrix (= compilation of a set of conjectures of homology) is followed by parsimony analysis of relationships. The initial conjectures of homology will be either supported, refuted or left undecided. This can be translated in, respectively, scores for the same character state emerging as synapomorphy, homoplasy, or both alternatives being possible in the resulting tree profile. In other words, primary conjectures of homology can be classified, after a phylogenetic analysis, into those which appear to be correct, those which may be wrong and those where no assignment seems possible.

Parsimony has been established as the most power­ful and logically sound method to sort among pri­mary conjectures of homology, indicating the status of each - correct, wrong or dubious (FARRIS, 1983; SOBER, 1988; PINNA, 1991; STEWART, 1993), and maximizing those primary conjectures of homo­logy that end up as secondary homology. Since it is the objective of systematists to find the greatest possible number of corroborated primary homolo­gies - otherwise phylogenetic reconstruction would

82 E. HAJDU & R.W.M. VAN SOEST

become a chance method with no objectivity - a procedure indicating which, if any, classes of cha­racters consistently yield more primary homologies that are correct would be welcome. This would al­low systematists to focus on those classes of charac­ters where similarities are less misleading, and can thus be more safely translated into true homologies.

Phylogenetic systematics made a step forward rela­tive to numerical taxonomy, in recognizing that primitive similarity, plesiomorphy, can be quite misleading for the assessment of relationships among taxa. The use of outgroups is a way of fo­cusing attention onto characters that are derived and thus relevant for the assessment of relationships within a selected ingroup. Nevertheless, even among derived character states there will often be those that are really homologous, those that are homoplastic, and those that must remain undecided after the analysis. Homoplastic characters may also be im­portant in the establishment of tree topology (DAVIS et ai, 1993), but their occurrence should be as few as possible given the assembled data matrix or the obtained tree runs the risk of being not the one that best reflects the original data. Surely, seeking par­simony minimizes the level of homoplasy present in the final cladogram(s), but even so this level fre­quently remains very high. This may be disappoin­ting in the sense that assembly of a data matrix is quite often an elaborate, time consuming enterprise, and the belief is widespread that low consistency (= high homoplasy) is synonymous with low support [but see GOLOBOFF's (1991) ideas on decisiveness]. If the assembly of data matrices could somehow be focused on those characters that consistently show less homoplasy, perhaps some optimization of the assembly protocol could be obtained, i.e. some cha­racters could either be ignored, or at least have their assessment postponed to some future analysis. It is now becoming accepted that total evidence is the best approach to a phylogenetic analysis (e.g. KLUGE, 1989). However, debate goes on as to whe­ther or not independent classes of characters (defined in terms of process partitions; sensu BULL et ai, 1993) should be better kept isolated (e.g. KLUGE & WOLF, 1993; MIYAMOTO & FITCH, 1995). A corollary derived from the rational behind this latter approach is that the more characters that are analyzed from each class of characters, and the more concurrence between classes of characters, the more sound the phylogenetic results will be. Nevertheless, systematic practice dictates how 'total', total evidence will be for a particular study. Often, the sole guideline for choosing characters is the author's familiarity with this or that technique (which restricts study to one or a few character environments) - anatomy of hard parts, ultrastructure, cytology, secondary metabolites, isozyme electrophoresis, DNA/RNA sequencing (see NOVACEK & WHEELER, 1992).

With the above in mind, we decided to focus on a procedure for the selection of characters that ap­

plies to characters belonging to a single environ­ment. The simple discarding of characters on the basis of personal beliefs on how they perform on phylogenetic analyses is very subjective. We suggest that particular classes of characters within a single character environment might be selected on the ba­sis of their overall better consistency, an idea which parallels BEGLE's (1991) scheme forjudging the re­lative usefulness of gain characters as opposed to losses. Another possible use of such an approach could be the establishment of relative levels of confidence on particular classes of characters within a single environment, thus allowing a priori weigh­ting schemes to be adopted.

Materials and methods

CLASSES OF CHARACTERS Classes were established on the basis of current practice, in parallel with recognized assemblages of characters in the latest glossary of poriferan mor­phological terms (BOURY-ESNAULT e/a/., in prep.). There are classes pertaining to more than one cha­racter environment, but those approaching reaso­nable dimensions belong to the domain of the so called 'classical' approach to sponge taxonomy. Other classes were set aside due to their very low number of occurrences within the sampled data sets (Table 1).

METHODS Seventeen published data matrices for monophyle-tic groups of poriferan taxa totaling 221 OTUs were used to assess the performance of sixteen classes of characters in terms of their consistency. Data matrices were analyzed with SWOFFORD's (1993) program PAUP 3.1.1. Shorter trees were searched through the heuristic algorithm following the steps outlined in HAJDU (1995). The performance (j)) of each class of characters was calculated according to the following equation :

p = hc/n . 100%

where he is the number of occurrences of characters assigned to a particular class that obtained the highest consistency (c = 1.0) after PAUP's reweighting procedure, and n is the total number of characters pertaining to the same particular class among the 210 characters sampled in our study.

Table 2 shows the taxa for which data matrices were available, authorship of the compilations, as well as number of OTUs and characters for each of them.

Results

Table 3 shows the five classes of characters that had large enough sample sizes. All of these pertain to classical morphological characters, which include data on the shape of sponges as well as on the ana­tomy of their hard parts (spongin skeletons inclu­ded).

Cladistics and character selection 83

Table 1. Number of occurrences of each class of characters within the 17 sampled poriferan data sets (refer to Table 2).

chemistry

choanosomal architecture characters (CA)*

cytology

colour

consistency

ectosomal architecture characters

features of the aquiferous system

habit

habitat

larval morphology

megascleres : present or absent, categories (Mp/a+c)*

megascleres : shape, dimensions (Ms/cf)*

microscleres : present or absent, categories (mp/a+c)*

microscleres : shape, dimensions (ms/d)*

reproduction

surface

total

3

25

1

1

2

7

1

13

2

1

23

29

41

51

1

9

210

* abbreviations for character classes used in the text for discussion of observed trends.

Discussion

Our approach is guided by systematic practice among specialists of Porifera - characters most of­ten employed are those related to habit (such as shape, color, surface, consistency) and anatomy of hard parts (spicules, architecture). Whether or not habit should be combined in a single data matrix with data derived from the anatomy of hard parts, and whether all of the anatomy of hard parts belong to a single process in the sense of BULL et al. (1993; cf MIYAMOTO & FITCH, 1995) are questions beyond the scope of our study.

A more serious concern to us is the possible inade­quacy of the chosen protocol. Specifically, C [the ensemble consistency index, KLUGE & FARRIS (1969)] and c (character consistency) have been shown to be dependent on the size of the data ma­trix, a serious concern to us due to the fact that the 17 data matrices employed ranged from 4 to 32 OTUs. Larger numbers of taxa were shown to pro­duce lower values of C and c (e.g. ARCHIE, 1989, 1990). The retention index, /•, of FARRIS (1989a, b) which is thought to perform better in regards to

character comparisons (ARCHIE, 1990) will be the same as c, when c = 1.0, and will thus produce no change in our results because only those scores of characters where consistency was maximum (c = 1.0) were selected. Precisely the same reasoning applies to another index proposed by FARRIS (1989a), viz. the rescaled consistency index, re. A different performance relative to autapomorphies is irrelevant here since these were discarded in the present analyses.

Another concern to us is the fact that some classes of characters are poorly represented within the sampled data sets (e.g. cytology, color, habitat, lar­vae - Table 1), and accordingly no general picture may be drawn for them until more data is available. Nevertheless, five classes show a reasonable number of occurrences, viz. 23 to 51 times. These were : choanosomal architecture (CA), presence/absence of megascleres (Mp/a+c), shape/dimensions of mega­scleres (Ms/cf), presence/absence of microscleres (mp/a+c), and shape/dimensions of microscleres (ms/d). If we establish above 50 % of scores of c = 1.0 as a minimum requirement for considering a

84 E. HAJDU & R.W.M. VAN SOEST

Table 2.

Seventeen used data matrices*, their source references and number of OTUs.

Data matrices : Numbers of ingroup OTUs : characters

Acarnus (Poecilosclerida) 21 : 19

VAN SOEST, HOOPER & HIEMSTRA, 1991) Ceratopsion/Thrinacophora (Poecilosclerida) 16 : 10

HOOPER & LEVI, 1994) Clathria procera spp. group (Poecilosclerida) 9 : 7

HOOPER «& LEVI, 1994)

Didiscus (Halichondrida) 7 : 1 1 HIEMSTRA & VAN SOEST, 1991; emended**)

Fistulose iophonids (Poecilosclerida) 19 : 14 VAN SOEST, ZEA & KIELMAN, 1994)

Halichondrida 4 : 4 VAN SOEST, DiAZ & POMPONI, 1990)

Halichondriidae (Halichondrida) 12 : 10 VAN SOEST, DIAZ & POMPONI, 1990)

Keratosa (Dendroceratida, Dictyoceratida, Verongida) 7 : 8 VAN SOEST, 1991)

A^cale, the 'curved-assemblage' (Poecilosclerida) 32 : 27 HAJDU, 1995)

Mycalidae + Hamacantha (Poecilosclerida) 1 0 : 1 1 HAJDU & DESQUEYROUX-FAÜNDEZ, 1994)

Pachastrellidae-I (Astrophorida) 1 1 : 1 1 MALDONADO, 1994)

Pachastrellidae-IV (Astrophorida) 1 0 : 1 0 MALDONADO, 1994)

Poecilosclerid families (Poecilosclerida) 9 : 8 DESQUEYROUX-FAÜNDEZ & VAN SOEST, submitted)

Ptilocaulis/Reniochalina (Halichondrida) 14 : 9 HOOPER & LEVI, 1994)

Rhabderemia (Poecilosclerida) 23 : 20 VAN SOEST & HOOPER, 1993)

Tethyidae (Hadromerida) 8 : 1 8 SARA & BURLANDO, 1994)

Tetractinellid-hadromerid-hemiasterellid relations 9 : 13 Astrophorida, Desmophorida, Hadromerida, Spirophorida) VAN SOEST, 1991)

total 221 : 210

* Other phylogenies are available, but they either do not comprise monophyletic assemblages or data ma­trices were not provided. ** The New Zealand Oamaru fossil species, D. hindei, has been added to the data matrix prior to the analysis. Six out of the 16 used characters pertain to the discorhabds and can be scored for the fossil species on the basis of the drawing provided by HINDE & HOLMES (1891).

Cladistics and character selection 85

Table 3. Relative performance in terms of percent occur­rences of c = 1.0, when c is tabulated for every single character pertaining to each of five classes of characters, which have a reasonable number of oc­currences among the 17 sampled poriferan data matrices (refer to Table 2). Abbreviations accord­ing to Table 1.

CA 64% (16/25)

Mp/a+c 74% (17/23)

Ms/d 52 % (15/29)

mp/a+c 39% (16/41)

ms/d 59% (30/51)

class of characters trustworthy, we could say that Ms/d (52 %, barely on the other side) and mp/a+c (39 %) are less trustworthy classes of characters than CA (64 %), Mp/a+c (74 %) and ms/d (59 %). We would like to see these results corroborated by analyses of additional data matrices before atta­ching any predictive value to them, for instance as differential a priori weighting of particular classes of characters. On the other hand results obtained here, although preliminary, may be useful as a ba­seline to decide between discordant phylogenetic hypotheses in cases where characters in conflict belong to classes suggested to be distinctly trustworthy.

Care must be taken with respect to the apparently good performance of choanosomal architecture characters. More than any other class of characters, choanosomal architecture is usually described with very broad descriptive terms. FROMONT & BERGQUIST (1990) argued that habitus characters such as possession of fistules need not be homolo­gous in different higher taxa. Similarly, HAJDU et al. (1994) stressed that skeletal characters are likely to be constrained by the availability of the building blocks, e.g. the size and form of the spicules, and the amount of spongin. The predictability of a given skeletal structure from the nature of the spicules is such that modeling is possible (KAANDORP, 1991). Our concern here is that widespread poriferan systematic procedures may be misleading our proposed method in respect to consistency of choanosomal architecture characters. On the other hand, since results obtained here are the most parsimonious given the data matrices analyzed, and high consistency equals congruence with other characters, we must accept, at least for the time being, choanosomal architecture characters as a relatively trustworthy class of characters (64 % of c = 1.0).

The distinctly opposed reliability of megascleres versus microscleres is remarkable. While Mp/a+c performed better than Ms/d, the opposite trend was observed in microscleres. Megascleres presumably play a more relevant structural role in most sponges due to their prominent occurrence in skeletons (DENDY, 1921) - e.g. coring and/or echinating fibres, supporting ectosomal tangential skeletons. The functional role for microscleres is obscure (cf RIDLEY & DENDY, 1887; DENDY, I.e.), and even where a particular function has been demonstrated (e.g. VACELET & BOURY-ESNAULT, 1995) it seems unlikely that a similar role could not be performed by a megasclere, or another kind of microsclere. In the case of the Asbestopluma reported by VACELET & BOURY-ESNAULT (1995), Spines in the megas­cleres, if they would have been present, would most probably perform the same function as the small anisochelae, viz. trapping of zooplankton. It seems to us then that the evolution of megascleres and mi­croscleres need not be guided by the same factors. WHEELER (1986) quoted the following sentence by DARWIN (1859) : "The less any part of the organi­zation is concerned with special habits, the more important it becomes for classification." This is the essence of what is known as Darwin Principle (MAYR, 1979), viz. that characters of low adaptive value are important echoes of underlying genetic similarities.

Extrapolating this idea to the characters under consideration here is not entirely straightforward as will be shown below. The trend observed in megas­cleres of higher c levels in presence/absence + cate­gories as opposed to shape/dimensions could sug­gest that changes in shape/dimensions are less deci­sive for sponge functions and thus more easily achieved independently (i.e. oxeas and styles may have the same function). On the other hand, com­plete loss of categories seems more problematic in terms of sponge function and thus less likely to happen independently (i.e. megascleres are back­bones essential for sponge sustenance, their loss may mean collapse of architecture). These are pos­sible interpretations for adaptively constrained morphological characters. Microscleres, according to RIDLEY & DENDY (1 887), DENDY (1921), HAJDU et al. (1994) and others, are most likely non-adaptive characters. As argued above, it is unlikely, in most cases, that microscleres have an unequivocal func­tion in poriferan architecture. If an anisochela is believed to act as a hook, there is no clear reason why such a function could not be performed by a sigma, or by an acanthostyle. Accordingly, mp/a+c has a generalized low level of c. Why do we observe a high level of c for ms/d ?

MACBETH (1980) in discussing the concept of irre­versibility postulated that evolution never retraces its steps in a big way, but it often reverses one or another change. There is no clear border between irreversibility of the 'big' phenomena and the more

86 E. HAJDU & R.W.M. VAN SOEST

likely reversibility of the 'smaller' ones. LAURENT (1983) compared the former with "the unreeling of a movie film", and the later with the simpler idea of loss and reappearance of characters (even of com­plex ones). In our view, the improbability of a big reversion in microscleres can be equated with the relatively high stability of shapes {ms/d : 59 % of c = 1.0). This is so even though some shapes are believed to pertain to a single transformation series. In support of this idea HAJDU et al. (1994) pointed out that only three species are known that may have two distinct chelae morphotypes. This suggests that although chelae may have evolved in the direction of palmate through arcuate to anchorate, develop­ment of a derived morphotype does not imply going through the more primitive ones as in an ontogenetic series. This argument should apply to other forms of microscleres too. In other words, once something new is evolved, information (genetic control) on how to build the former stage is lost. On the other hand, mp/a+c has a low c be­cause structures can be lost and regained relatively easily depending on supressor genes (e.g. LAURENT, 1983), and their own supressors - given that they are non-adaptive structures.

A final concern relates to the fact that different cha­racters may perform distinctly well in terms of c within different taxa, i.e. c could be taxon-specific. A much larger data base is needed in order to pro­perly verify these suspicions. An alternative interpretation of the results may be ventilated though, in which, except for the 74 % c -1.0 obtained by presence/absence + categories of megascleres, results may all suggest high levels of homoplasy, and consequently low confidence levels in most classes of characters. Since the reasonably large classes considered here all pertain to the do­main of the anatomy of hard parts, one could be tempted to say : "Let us do chemistry then!". KELLY-BORGES (pers. comm.) commented that she obtained similar consistencies for molecular data. The message may be then that homoplasy is high in different character environments, and that no, or perhaps only a very few classes of characters are more trustworthy than the others. If such a scenario is true, hard labor of the 'blind' systematist is the only tool to assemble a data matrix, and parsimony the only tool to extract phylogeny out of it.

Conclusions

The method outlined here may help in determining which characters are more trustworthy in terms of phylogenetic signal (congruence with other charac­ters), based on their performance on previously cladistically analyzed data sets. These characters could be the ones selected in the future for new analyses, while less trustworthy characters would have their assessment postponed or simply discar­ded. Additionally, less trustworthy characters, if

used in an analysis would need additional care in the scoring of primary conjectures of homology (similar scores for a character state in a data ma­trix). An alternative use of our protocol is to objec­tively establish confidence levels on particular classes of characters, an important building block in an a priori weighting scheme. The data base of po-riferan characters is still too meager, as a reflection of incipient use of cladistic procedures in day-to­day practice among specialists in Porifera (only 17 phylogenies are available which contain published data matrices). This precludes any firmer conclu­sions to be drawn for most of the 16 recognized classes of characters. Nevertheless, five classes have sample sizes approaching the reasonable (22-50). Among these, shape/dimensions of megascleres and presence/absence + categories of microscleres per­form poorly (52 and 39 % of c = 1.0, respectively). Choanosomal architecture characters, pre­sence/absence + categories of megascleres and shape/dimensions of microscleres perform better (64, 74 and 59 % of c = 1.0, respectively). A much broader data base is needed in order to check whe­ther different classes of characters would perform differentially in distinct taxa.

Acknowledgments

This work benefited from the comments of two anonymous reviewers. J.N.A. HOOPER is thanked for providing us with his unpublished data matrices for Ceratopsion / Thrinacophora, the Clathria procera spp. group, and Ptilocaulis / Reniochalina. M. MALDONADO discussed with us his ideas on spicule function and adaptation. M. KELLY-BORGES commented that she obtains similar results with molecular data. They are both deeply acknowled­ged. F.R. SCHRAM and R. SLUYS provided construc­tive comments yielding a much improved version of the manuscript. EH benefited from financial support by Conselho Nacional de Desenvolvimento Cientifico e Tecnológico (CNPq - Brazil, process 201577/90.9).

References

ARCHIE, J.W., 1989. Homoplasy excess ratios : new indices for measuring levels of homoplasy in phylogenetic systematics and a critique of the consistency index. Systematic Zoology, 38 : 253-269.

ARCHIE, J.W., 1990. Homoplasy excess statistics and re­tention indices : a reply to Farris. Systematic Zoology, 39 : 169-174.

BEGLE, D.P., 1991. Relationships of the osmeroid fishes and the use of reductive characters in phylogenetic analysis. Systematic Zoology, 40 : 33-53.

BouRY-EsNAULT, N. & RuTZLER, K. (Eds), in prep. Thesaurus of sponge morphology. Smithsonian Institution Press, Washington DC.

Cladistics and character selection 87

B U L L , J J , H U E L S E N B E C K , C W , C U N N I N G H A M , C W ,

SwoFFORD, D L & W A D D E L L , P J , 1993 Partitioning and combining data in phylogenetic analysis Systematic Biology, 42 384-397

D A R W I N , C , 18^9 On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life J Murray, London, i-ix + 1-506

D A V I S , J 1 , FROHLICH, M W & S O R E N G , R J , 1993 Cladistic

characters and cladogram stability Systematic Botany, 18 188-192

D E N D Y , A , 1921 The tetraxonid sponge spicule a study in evolution Acta zoologica, 2 95-1S2

DESQUEYROUX-FAUNDEZ, R & VAN SOEST, R W M , (in press)

A review of lophon Myxilla and Tedania from the Southeast Pacific Revue suisse de zoölogie

F A R R I S , J S , 1983 The logical basis of phylogenetic ana­lysis In P L A T N I C K , N AND F U N K , V (Eds), Advances in cladistics. Vol 2 Columbia University Press, New York, pp 7-36

F A R R I S , J S 1989a The retention index and the rescaled consistency index Cladistics, 5 417-419

F A R R I S , J S , 1989b The retention index and homoplasy excess Systematic Zoology, 38 406-407

F O R E Y , P L , H U M P H R I E S , C J , K I T C H I N G , I L , S C O T L A N D ,

R W , SIEBERT, D J & W I L L I A M S , D M , 1992 Cladistics, a

practical course in systematics The Systematics Association Publication No 10 Oxford Science Publications Clarendon Press, Oxford, pp i+xii + 1-191

FROMONT, J P & BERGQUIST, P R , 1990 Structural Characters and Their Use in Sponge Taxonomy When Is a Sigma Not a Sigma'' In RUTZLER, K (Ed ), New Perspec-tives in Sponge Biology Smithsonian Institution Press, Washington DC & London, pp 273-278

GOLOBOFF, P A , 1991 Homoplasy and the choice among cladograms Cladistics, 7 215-232

H A J D U , 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 PhD Thesis Universiteit van Amsterdam Centrale Drukkerij, Universiteit van Amsterdam, Amsterdam, 174 pp

HAJDU, E & DESQUEYROUX-FAUNDEZ, R , 1994 A synopsis of

South American Mycale (Mycale) (Poec i lo sc l e r ida , Demospongiae), with the description of three new species and a cladistic analysis of Mycalidae Revue suisse de Zoölogie, 101 563-600

H A J D U , E, VAN S O E S T , R W M & H O O P E R , J N A , 1994

Proposal for a phylogenetic classification of poecilosclerid sponges In VAN S O E S T , R W M , VAN K E M P E N , Th M G &

B R A E K M A N , J C (Eds), Sponges in Time and Space Proceedings of the IVth International Ponfera Congress A A Balkema, Rotterdam, pp 123-139

HIEMSTRA, F & VAN SOEST, R W M , 1991 Didiscus verdensis

spec nov (Ponfera Halichondrida) from the Cape Verde Islands, with a revision and phylogenetic classification of the genus Didiscus Zoologische Mededelingen Leiden,6'> 39-52

HINDE, G J & HOLMES, W M , 1891 Sponge remains in the Ter­tiary of New Zealand Linnean Journal of Zoology 24 177-362

HOOPER, J N A & LEVI, C , 1994 Biogeography of Indo-west

Pacific sponges Microcionidae, Raspailiidae, Axinellidae In VAN SOEST, R W M , VAN KEMPEN, Th M G & BRAEKMAN,

J C (Eds), Sponges in Time and Space Proceedings of the IVth International Ponfera Congress A A Balkema, Rotterdam, pp 191-212

KAANDORP, J A , 1991 Modelling growth forms of the sponge Haliclona oculata (Ponfera, Demospongiae) using fractal techniques Marine Biology, 110 203-215

KLUGE, A G , 1989 A concern for evidence and a phylogenetic hypothesis of relat ionships among Epicrates (Boidae, Serpentes) Systematic Zoology 38 7-25

K L U G E , A G & FARRIS , J S , 1969 Quantitative phyietics and the evolution of anurans Systematic Zoology 18 1-32

KLUGE, A G & W O L F , A J , 1993 Cladistics what's in a word Cladistics, 9 183-199

L A U R E N T , R F , 1983 Irreversibility a comment on MACBETH'S interpretations Systematic Zoology, 32 75

M A C B E T H , N , 1980 Reflections on irreversibility Systematic Zoology, 29 402-404

M A L D O N A D O , M , 1994 The taxonomie significance of the short-shafted mesotnaene reviewed by parsimony analysis validation of Pachastrella ovisternata VON LENDENFELD (Demo­spongiae Astrophonda) Bijdragen tot de dierkunde 63 129-148

M A Y R , E , 1979 Principles of systematic zoology McGraw-Hill, New York, 428 pp

M I Y A M O T O , M M & F I T C H , W M , 1995 Testing species

phylogenies and phylogenetic methods with congruence Systematic Biology, 44 64-76

NELSON, G , 1994 Homology and systematics In HALL, B K (Ed ), Homology the hierarchical basis of comparative biology Academic Press, San Diego, pp 101-149

NOVACEK, M J & WHEELER, Q D , 1992 Introduction Extinct taxa Accounting for 99 999 % of the Earth's Biota In N O V A C E K , M J & W H E E L E R , Q D (Eds), Extinction and Phylogeny Columbia University Press, New York, pp 1-16

PATTERSON, C , 1982 Morphological characters and homolo­gy In JOYSEY, K A & FRIDAY, A E (Eds), Problems of phylo­geny reconstruction Academic Press, London, pp 21-74

P I N N A , M C C DE, 1991 Concepts and tests of homology in the cladistic paradigm Cladistics, 7 367-394

RIDLEY, S O & D E N D Y , A , 1887 Report on the Monaxonida collected by H M S Challenger dunng the years 1873-1876 Report on the scientific Results of the Exploring Voyage of HMS Challenger 1873-76 Zoology 2 0 ( 5 9 ) i-lxviii + 1-275 pp

SARA, M & BURLANDO, B , 1994 Phylogenetic reconstruction

and evolutionary hypotheses in the family Tethyidae (Demospongiae) In VAN S O E S T , R W M , VAN K E M P E N ,

Th M G & BRAEKMAN, J C (Eds), Sponges in Time and Space Proceedings of the IVth International Ponfera Congress A A Balkema, Rotterdam, pp 1 I I - I I 6

S M I T H , A B , 1994 Systematics and the fossil record documenting evolutionary patterns Blakwell Scientific Publications, Oxford, i-viii + 1-223 pp

8 8 E HAJDU & R W M VAN SOEST

S O B E R , E , 1988 Reconstructing the past Parsimony, evolution and inference The Massachusetts Institute of Tech­nology Press, Cambridge - Massachusetts, i-xiii + i-26'S

S O E S T , R W M V A N , 1991 Demosponge higher taxa classification re-exammed In REITNER, J & KEUPP, H (Eds), Fossil and Recent Sponges Springer-Verlag Berl in, Heidelberg & New York, pp S4-71

S O E S T , R W M V A N , D I A Z , M C & P O M P O N I , S A , 1990

Phylogenetic classification of the halichondnds (Porifera, Demospongiae) Beauforlia, 40 (2) 1 ^-62

S O E S T , R W M VAN & H O O P E R , J N A , 1993 Taxonomy,

phylogeny and biogeography of the marine sponge genus Rhabderemia TOPSENJ, 1890 (Demospongiae, Poecilosclerida) Scientia Manna, 'S? (4) 317-351

S O E S T , R W M VAN. HOOPER, J N A & HIEMSTRA, F , 1991

Taxonomy, phylogeny and biogeography of the marine sponge genus Acarnus (Porifera, Poecilosclerida) Beaufortia, 42 (3) 49-88

S O E S T , R W M V A N , Z E A , S & K I E L M A N , M , 1994 New

species of Zyzzya Cornulella Damina and Achehderma (Porifera Poecilosclerida), with a review of fistular genera of lophonidae Bijdragen tot de dierkunde, M 163-192

S T E W A R T , C B , 1993 The powers and pitfalls of parsimony Nature, 361 603-607

SwOFFORD, D L , 1993 PAUP version 3 1 1 Computer program, Smithsonian Institution, Washington Distributed by the Illinois Natural History Survey, Champaign Illinois

V A C E L E T , J & BouRY-EsNAULT, N , 1995 Carnivorous sponges Nature, 373 333-335

W H E E L E R , Q D , 1986 Character weighting and cladistic analysis Systematic Zoology, 35 102-109

Eduardo HAJDU

Institute for Systematics and Population Biology (Zoologisch Museum)

University of Amsterdam PO Box 94766, 1090-GT Amsterdam

The Netherlands

Present address Departamento de Zoologia - Institute de Biociências

Universidade de SSo Paulo - Cx Postal 11461 05422-970 Sao Paulo - Brazil

Rob W M VAN SOEST

Institute for Systematics and Population Biology (Zoologisch Museum)

University of Amsterdam PO Box 94766, 1090-GT Amsterdam

The Netherlands