ENDEMISM: ORIGINS and IMPLICATIONS a Myers, S De Grave

ENDEMISM: ORIGINS AND IMPLICATIONS A Myers, S de Grave

To cite this version:

A Myers, S de Grave. ENDEMISM: ORIGINS AND IMPLICATIONS. Vie et Milieu / Life & Envi- ronment, Observatoire Océanologique - Laboratoire Arago, 2000, pp.195-203. hal-03186913

HAL Id: hal-03186913 https://hal.sorbonne-universite.fr/hal-03186913 Submitted on 31 Mar 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. VIE ET MILIEU, 2000, 50 (4) : 195-204

ENDEMISM: ORIGINS AND IMPLICATIONS

A.A. MYERS* S. DE GRAVE** * Department of Zoology and Animal Ecology, National University of Ireland Cork, Lee Maltings, Prospect Row, Cork, Ireland ** Oxford University Muséum of Natural History, Parks Road, Oxford, 0X1 3PW, U.K.

ENDEMISM ABSTRACT. - Ail taxa are endémie and occur in nested distributions at a range CLADISTICS of spatial scales. Distinguishing endémies as either neoendemics or palaeoendemics may not be of practical value in analytical biogeography, but distinguishing extinction mediated endémies (cryptoendemics) from endémies which never had a significantly wider range (euendemics) would be useful for interpreting the history of géographie areas. Only cladistic phylogeny provides a tool for distinguishing thèse two types of endémie.

ENDÉMISME RÉSUMÉ. - Tous les taxons sont endémiques et présentent des répartitions em- CLADISTIQUE boîtées selon les échelles spatiales considérées. La distinction de néoendémiques et de paléoendémiques parmi les endémiques peut ne pas avoir de valeur pratique en biogéographie analytique; en revanche, distinguer les endémiques dus à des extinctions (cryptoendémiques) par rapport aux endémiques qui n'ont jamais pré- senté de distribution significativement plus large (euendémiques) pourrait être utile pour interpréter l'histoire des aires géographiques. Seule, la phylogénie cladistique fournit un outil permettant de distinguer ces deux types d'endémiques.

ENDEMISM IN BIOGEOGRAPHY the Guianan lowlands of South America. When considered globally, this is not a disjunct distribution. Within the Guianan lowlands, however, the Endemism has long been a central thème in species are separated into four major basins, the biogeography, but it is often misinterpreted as re- Orinoco basin, the Essequibo-Cuyuni basin, the ferring only to "narrow" distributions, inferring Atlantic basin and the Amazon basin (Rodriguez that only some taxa are endémie. Hengeveld & Campos 1998), a clearly disjunct distribution. (1990) for example notes "it is hard to distinguish Disjunction can corne about either through a between endémie species and "wides"...". Lincoln vicariant event, or through jump dispersai. et al. (1982) define an endémie as "restricted to a Vicariance may resuit from tectonic activity (plate particular geographical région", a définition which movement, or orogeny), from eustasy or from lo- can be seen to apply, quite correctly, to ail organ- cal climate changes (often linked to tectonic and isms. eustatic events). Jump dispersai occurs when Since ail taxa are endémie, it becomes neces- propagules cross an intervening uninhabitable ter- sary to catégorise them on the basis of their range rain to colonise a new area. This is difficult to dis- (extensive or restricted) and of their disjunction. cern as a naturally occurring event due to its rela- Both range and disjunction are continua and dé- tive rareness coupled with the short period of time pend upon the spatial scale (Table I, Fig. 1). The over which scientific records have been kept. It ranges of taxa vary from cosmopolitan through may also be obscured by the frequency of human continental or océan basin scales, to single sites. mediated transport (anthropochore dispersai). They ail show some degree of disjunction, since Endémies then, can be classified hierarchically, populations, species and higher taxa are ail sepa- consisting of a séries of nested sets of areas of oc- rated from one another at some spatial scale. A cupancy by taxa, the largest set of which, under taxon occurring uniquely in a local area might not current knowledge, is the entire world. If the be considered to have a disjunct distribution at a biogeographic patterns resulting from multiple global scale. On the other hand it clearly could be nested distributions can be shown to be statisti- considered to have a disjunct distribution on a lo- cally non-random, they demand a biogeographic cal spatial scale. The decapod crustacean genus interprétation. Some areas can be shown to have Fredius Pretzman, for example, is found only in greater densities of species which are endémie to 196 MYERS A., DE GRAVE S.

Table I. - Classification of endémies with examples.

Range Characteristics Explanation Example

Global Unlimited biogeographically, Taxon with good powers of dispersai Gnathophyllum (Holendemic) limited only by ecological and colonisation (high compétitive (Fig. 1) and physiological tolérances. advantage) and with sufficient time for establishment

Conjunct broad Broad, more or less continuons Taxon expanding or contracting ils Periclimenella (Euryendemic) or contiguous distribution, limited range(s), or in quasi stasis or régressive (Fig. 2) by biogeographic barriers (aging) taxon beginning to contract its range(s), or in quasi stasis

Conjunct narrow Restricted more or less continuous Progressive taxon recently evolved Macrohectopus (Stenoendemic) or contiguous distribution limited yet to expand its range, or evolved (Fig. 2) by biogeographic barriers in situ (phylogenesis) from a clade constrained historically by biogeographic barrier(s), or relict (old) taxon.

Disjunct Two or more widely disjunct 1. Vicariated taxon (Rhoendemic) distributions. a) Euendemics, isolated by tectonic Globosolembos francanni Reid changes in geography Amphiatlantic (Cape Verde and Florida) b) Cryptendemics, isolated by Iguanidae extintion of intervening populations New World + Madagascar/Fiji or taxa Extinet in Africa 2. Taxon jump-dispersed across a biogeographic barrier.

a) Natural Bubulcus ibis (L.) jump dispersed from Africa to South America in 1920's b) Anthropochore Elminius modestus Darwin Spread in 20* century from New Zealand, ? Australia to Europe and South Africa

Stenoendemic

Euryendemic (suturing)

iflà i Vicariance X 1 Rhoendemics * A Jump dispersai T Euryendemic

Stenoendemic

Fig. 1. - Schematic représentation of the formation of endémie types. ENDEMISM 197

themselves than have others. Thèse have been de- and vicariance. The use of cladistic methodology to scribed as "areas of endemism" and have formed answer biogeographic questions has been further the nucleus of much biogeographic classification. developed in récent years by several authors, nota- Areas of endemism are interpreted as areas of high bly Humphries & Parenti (1986, 1999). endemicity separated from other such areas by a In cladistic biogeogra^hy, a phylogeny is first decided gap. The gap may be defined as an area of determined, after which, known distributions of the low endemicity between the two areas of high taxa in the phylogeny are superimposed upon the endemicity e.g. the Sinaloan gap between the terminal nodes of the cladogram. It is the ances- Cortez and the Mexican marine Provinces. Alterna- tor-descendent branching of a clade which indi- tively it may be defined as the région of co-occur- cates the séquence (although not the timing) of rence of endpoints in species distributions vicariant events and hence the relationship (and (Hastings 2000). relative âge) of each area. Areas of endemism are The first biogeographer to employ endemism as not a primary component of any data set analysed an analytical, rather than a descriptive tool, was in cladistic biogeography and the number of probably Wallace (1876) who used the numbers of endémies in a given area (cf. Wallace above) is ir- endémies as a measure of island âge. He hypothes- relevant to any hypothèses of area relationships. In ised that the more endémie species that occurred on contrast, endémies have been used as the primary an island, the more archaic that island was likely to biogeographic data set in parsimony analysis of be, the assumption being that older islands had ex- endemicity (PAE) devised by Rosen & Smith perienced a longer period of time for the occur- (1988). PAE uses a cladistic method, but bypasses rence of phylogenesis. the requirement for a prior phylogenetic analysis. Croizat (1958), developed a biogeographic Taxa are treated as équivalent to characters and methodology (panbiogeography) for transforming their distributions are the équivalent of character the recorded distributions of individual taxa into states. In cladistic terminology, taxa which occur simplified pathways (tracks) and then combining in one place only are équivalent to autapomorphies, several of thèse individual pathways into a sum- those occurring in ail of the studied sites are équiv- mary (generalised track) which could be used to alent to plesiomorphies and as in phylogenetic hypothesise the historical évolution of the biota cladistics, neither have information content and are concerned. This method has been further refined or accordingly omitted from the analysis. Taxa occur- applied by Page (1987), Craw (1989), Morrone & ring in some but not ail of the areas under study Lopretto (1994) and others. Meanwhile, Hennig (équivalent to synapomorphies) and which also (1966) developed a method for reconstructing lack an extrinsic distribution, are the only ones phylogenies based on evolutionary descent. It was considered to have biogeographic information con- left to Nelson & Platnick (1981) to synthesise thèse tent in this method. The assumption in PAE is that two methods in their séminal book on cladistics two (or more) areas sharing endémie taxa are in 198 MYERS A., DE GRAVE S.

Fig. 3. - The distribution of the euryendemic shrimp genus Periclimenella Bruce (closed circles) and of the stenoendemic amphipod genus Macrohectopus Stebbing (closed square). some way more "related" to each another, as a re- America, but not in Africa. Few biogeographers suit of a lack of historical biogeographic barriers would posit a closer ancestral relationship between between them, than either is to any other area that Madagascar and South America than between Mad- does not share with them those same taxa. agascar and Africa. Iguanids almost certainly did Humphries (2000), it should be noted, strongly occur in Africa but became extinct there (presum- criticises this method, asserting that, by bypassing ably through compétition with more advanced aga- taxic homology, it has corrupted the meaning of mid lizards). Thus, in cladistic terms, iguanids do area homology. In PAE, the taxa are derived char- occur in Africa but in modified form (they have acter states (synapomorphies) of the areas. Because been lost rather than having never been présent). It PAE makes the inference that areas uniquely shar- could be argued that the same assumptions about ing taxa are more related one to another than to ar- "loss" are made in phylogenetic cladistics with re- eas lacking thèse same taxa, it dépends upon an as- gard to missing character states. However, data for sumption that extinction has not significantly falsifying a given hypothesis are more readily modified the distribution patterns. called upon in phylogenetic analysis. In a phylo- If a biogeographic analysis based on endémies is genetic analysis, characters may be polarised by to be meaningful, it is necessary to distinguish outgroup analysis, thus facilitating the identifica- between on the one hand, absent having never been tion of lost (rather than never evolved) states. PAE présent in the area (équivalent to symplesiomorphic is not amenable to outgroup analysis. Clearly the in phylogenetic cladistics) and on the other hand, outgroup would have to be an area outside the area lost from the area i.e. extinct (équivalent to of study which by the criteria set in PAE could ex- synapomorphic). Any analysis of endémies needs ist only in the state "absent". In addition, in phylo- to be based entirely upon "synapomorphies", which genetic cladistic analyses a number of unrelated include both "présent now" and "présent once but character types may be analysed, allowing multiple now extinct". Iguanid lizards, for example, occur corroboration, whereas PAE uses only one charac- among other places, in Madagascar and South ter type, the taxon. ENDEMISM 199

Fig. 4. - The distribution and phylogeny of the shrimp genus Discias Rathbun (for character matrix see appendix 1).

Vicariance biogeography then, based on phylo- palaeoendemic (Engler 1882). Paleoendemics should, genetic cladistics, provides the most robust in theory, provide a signal of ancient biogeography biogeographic technique, with the fewest assump- provided that it is possible to distinguish real ab- tions. However, it requires well founded sences from extinctions. In practice, distinguishing phylogenies which at this stage are missing for the endémies as either new or old may not be particularly vast majority of taxa. PAE may be flawed by its informative in a biogeographic analysis. The limited lack of an underlying phylogeny, but it can, if the distribution of neoendemics is assumed to be due to signal is greater than the background noise, pro- the fact that they have not yet had time to expand vide, in the absence of a sound phylogeny, a their ranges. This may not hold true for jump "rough and ready" technique for highlighting dispersers, which on entering a "new" area in which vicariant events, particularly when more than one they have had no récent evolutionary history find va- level of the taxonomic hierarchy is used (see for cant niches or poorly compétitive résidents (see Sax example Myers 1991). The use of more than one & Brown 2000). There is also, a spatial component taxonomic group may increase the confidence lev- (independent of the temporal component) of range els further, because it is parsimonious to assume expansion, i.e. the présence of biogeographic barri- that extinction events would not have affected dis- ers, which cannot be ignored. Organisms trapped on tantly related taxa in an identical way. Différences an island (either a true island or a continental island) in dispersai potential between taxa, however, need may be archaic (by définition palaeoendemics), but if careful considération, because such différences can they have had no opportunity during their evolution- strongly influence a hypothesis (see for example ary history to expand their range, they will give the Myers 1993) same biogeographic signal as neoendemics. If, they have undergone in situ phylogenesis then the resuit will be a flock of taxa of variable âges. The âge of a taxon is not critical; of importance is whether it once THE SPECTRE OF EXTINCTION had a much wider range since modified by extinctions. Most species will have undergone some popu- lation extinctions in their evolutionary history, but it is necessary in most types of biogeographic analysis A taxon having a limited geographical range at- to be able to discern those species which have modi- tributable to recency of origin (because theoretically fied their distributions significantly through prior ex- there has been insufficient time for it to expand its tinctions. Clearly the biogeographic interprétation of range) has been termed a neoendemic. A taxon with a endemism dépends upon the spatio-temporal scale of limited geographical range even though it has a con- the analysis and the level in the taxonomic hierarchy. sidérable evolutionary âge, i.e. a relict having under- In the early Tertiary, adapid lemurs are known as fos- gone extensive extinctions, has been termed a 200 MYERS A., DE GRAVE S.

Fig. 5. - The distribution and phylogeny of the Bemlos clade (Amphipoda: Aorinae) (phylogeny from Myers 1988).

sils from America (Notharctinae) and Europe certain that the areas under study are not (Adapinae) (Hill 1953). At some stage Adapinae paraphyletic. If endémie taxa in the areas under reached Madagascar where they have since undergone study were at one time distributed extrinsically and explosive phylogenesis, producing a high diversity of have only since become restricted to the areas un- in situ endémie taxa. The Malgache lemurs can there- der study (by extinctions in the extrinsic area), then fore be defined either as relict paleoendemics (the the areas analysed are in fact paraphyletic. Any as- adapid clade) or as in situ neoendemics (the species sumptions about biogeographic relationship would, of Lemuridae, Indridae and Daubentoniidae). Either in such a scénario, be fundamentally flawed. Even définition is correct for a given context. if the taxa concerned have never had an extrinsic It would be particularly valuable to be able to distribution, they may once have been more widely distinguish between distributions mediated by ex- distributed within the area (thus confounding the tensive extinction, here referred to as crypto- signal of area relationships) or may even have been endemics and those whose distributions have not cosmopolitan within the area (symplesiomorphic) been significantly altered by prior extinctions, here so confounding any inferences of relationship. referred to as euendemics (see Table I). Indeed, It is clearly not possible to détermine the real without this knowledge, we may not be able to in- distribution of any taxon through its history, even terpret areas of endemicity effectively. In gênerai, with the existence of a fossil record. Phylogenetic disjunct species are likely to be euendemics. If they analyses, however, provide insights that enable had been isolated for considérable periods of time parsimonious décisions to be made between the they would almost certainly have changed from claims of competing hypothèses. their ancestors through anagenetic processes (sélection or genetic drift). Endémie sister taxa Frequently some members of a taxon are and higher taxa represent in situ évolution stenoendemics whilst others are euryendemics. In (phylogenesis) from a common ancestor and may such cases, a clade in which an ancestral taxon is widespread and the derived taxa have a restricted be euendemics or cryptoendemics. The higher in distribution will suggest a différent history from a the taxonomic hierarchy the older the ancestral géographie relationship. clade in which one or more of the derived taxa are widespread. The genus Discias Rathbun (Caridea, Disciadidae), for example, is widespread across ail world océans (Fig. 4). Two species, D. atlanticus UNDERSTANDING ENDEMIC PATTERNS Gurney and D. exul Kemp are euryendemics, the former in ail océans, the latter in the Indo-west Pa- cific. Two species D. serratirostris Lebour and D. If endemism is to be used as a primary database serrifer Rathbun are rhoendemics, the former in the for any biogeographic analysis, it is necessary to be western Atlantic, the other in the Eastern Pacific ENDEMISM 201

and four D. pascuensis Fransen, D. vernbergi relicts of a once more widely distributed clade. Boothe & Heard, D. musicus Holthuis, and D. Again, a phylogenetic analysis can help to illumi- brownae Kensley are stenoendemics. A phylogen- nate the spatial parameters of the ancestral clade. etic analysis (Fig. 4) reveals that D. atlanticus, the The genus Parisia Holthuis consists of six spe- widespread taxon, is ancestral whilst D. pascuensis cies, four in Madagascar and two in Australia. The for example - a species known only from Easter Is- two Australian species share a number of character land - is a more recently derived taxon. With thèse states and appear to be more closely related to each data, it is parsimonious to hypothesise that D. other than either is to the Malagasy species (Wil- pascuensis is a stenoendemic whose ancestor ar- liams 1964). No cladistic analysis has yet been per- rived at some time in Easter Island and has not un- formed on this genus but if the above relationships dergone significant extinction events elsewhere on are confirmed, there is an orderly relationship be- a global scale. A similar scénario could be applied tween phylogeny and distribution. Under thèse cir- to the other stenoendemics, D. serratirostris, D. cumstances there is no reason to hypothesise other serrifer and D. vernbergi. The euryendemic D. than that each clade evolved in situ, after a exul, has spread widely in the Indo-west Pacific vicariant event had first divided the ancestral but has been unable to cross the océan gap to the taxon. An alternative to this pattern would be real- east Pacific, suggesting that the range expansion ised if a phylogenetic analysis revealed that the occurred after the development of that gap in the closest sister of a taxon was not its nearest géo- late Mesozoic (see Myers 1991). It is also reason- graphie neighbour. The barnacle genus Chthamalus able to hypothesise that the extant distribution of Ranzani may be an example of this. The two spe- D. atlanticus represents a distribution derived at a cies Chthamalus montagui Southward and C. time in the past when the world was topographi- stellatus (Poli), sympatric along north-east Atlantic cally différent and colonisation of ail the major shores are apparently not sister taxa (Dando & océans was still a possibility. A competing hypoth- Southward 1980). C. stellatus is thought to be more esis could be that the ancestral D. atlanticus main- closely related to two species from the west Atlan- tained a limited distribution over most of its evolu- tic, viz. C. angustitergum Pilsbry and C. bisinuatus tionary history and only recently spread rapidly to Pilsbry, whilst C. montagui is thought to be more attain its présent range. Such a jump-dispersal scé- closely related to Pacific species such as C. nario is, however, less parsimonious since it re- challengeri Hoek and C. dalli Pilsbry. Unfortu- quires two assumptions: nately no cladistic analysis has been carried out on 1. that the taxon had a restricted range through Chthamalus to support the relationships suggested most of its evolutionary history and by morphological and electrophoretic studies 2. that it underwent more récent jump-dispersal within the genus. Three possible explanations, into ail the major océans. other than a falsified phylogeny, exist for such a The vicariant hypothesis requires only one as- phylogenetic relationship: sumption, that it became widespread during a pe- 1) They may have been euryendemics which riod in history when barriers to its dispersai were subsequently underwent extensive extinction to not significant. leave a geographically unordered pattern of relicts in widely separated localities. An alternative pattern is illustrated by a phylog- 2) Tectonic activity may have altered the earth's eny of the Bemlos clade of aorid amphipods (Fig. geography to such an extent that the extant distribu- 5). Two gênera, Bemlos Shoemaker and tions can only be understood by plotting them onto Globosolembos Myers are euryendemics, occur- a map of ancient topography ring throughout the tropical océans of the world 3) Jump dispersai may have occurred. and extending into the warm temperate of Austra- lia. Ancestral taxa Meridiolembos Myers and Tethylembos Myers are stenoendemics restricted to the Mediterranean and New Zealand respectively. CONCLUDING COMMENTS The most parsimonious hypothesis is that Tethylembos and Meridiolembos are antitropical relicts of an ancient flock of taxa most of which Higher taxa above the genus level generally have were driven to extinction within the tropics by little biogeographic information content. In the compétition with derived taxa now represented by main, families are cosmopolitan, although this may the widespread tropical gênera Bemlos and hold less for plants. Gênera tend to be more nar- Globosolembos. The sister taxon of Bemlos rowly endémie and species are frequently (Protolembos Myers) and of Globosolembos stenoendemic. Endemism originates in three différ- (Plesiolembos Myers) are stenoendemics. They ent ways: may be either ancestral or derived with respect to 1) by extinction of populations in part of their their euryendemic sister taxon. range. If a taxon is composed of several narrow 2) by range restriction through biogeographic rhoendemics, thèse may be either vicariants, or barriers (after anagenesis or phylogenesis) 202 MYERS A., DE GRAVE S.

3) by jump dispersai followed by anagenesis or Humphries CJ 2000. Form, space and time, which cornes phylogenesis. first? J Biogeography 27: 11-15 Humphries CJ, Parenti LR 1986. Cladistic biogeogra- Whilst no known technique can distinguish ab- phy. Oxford monographs on biogeography No. 2, solutely between thèse différent routes to Clarendon Press, Oxford. endemicity, phylogenetic cladistics provides the Humphries CJ, Parenti LR 1999. Cladistic biogeogra- best method on which to erect a parsimonious hy- phy, interpreting patterns of plant and animal distri- pothesis. bution, 2nd édition. Oxford Biogeography séries No With an awareness of the importance of preserv- 12, Oxford University Press, Oxford. ing the Earth's biodiversity cornes the need not Lincoln RJ, Boxshall GA, Clark PF 1982. A Dictionary only to détermine which areas contain large num- of Ecology, Evolution and Systematics, Cambridge bers of species unique to them, but also how such University Press, Cambridge areas of endemicity originate. Morrone JJ, Lopretto EC 1994. Distributional patterns of Freshwater Decapoda (Crustacea: Malacostraca) in An appréciation of the complexity of origins of southern South America: a panbiogeographic ap- endémies and of the alternative routes to narrow proach. J Biogeography 21: 97-109. endemicity may improve our understanding of how Myers AA 1988. A cladistic and biogeographic analysis areas of endemicity have corne about. of the Aorinae subfamily nov. Crustaceana suppl. 13: 167-192. Myers A A 1991. How did Hawaii accumulate its biota? REFERENCES A test from the Amphipoda. Res letters Evol Biogeo- graphy 1: 24-29 Myers A A 1993. Dispersai and endemicity in gammari- dean Amphipoda. J Nat Hist 27: 901-908. Craw RC 1989. Quantitative panbiogeography: introduc- tion to methods. N Z J Zoology 16: 485-494. Nelson G, Platnick N 1981. Systematics and Biogeogra- Croizat L 1958. Panbiogeography. Vols 1, 2a, 2b, pu- phy; cladistics and vicariance. Columbia University Press, New York. blished by the author, Caracas, Venezuela. Dando PR, Southward AJ 1980. A new species of Chtha- Page RDM 1987. Graphs and generalised tracks: quanti- malus (Crustacea: Cirripedia) characterised by en- fying Croizat's Panbiogeography. Systematic Zoology zyme electrophoresis and shell morphology, with a 36: 1-17. revision of other species of Chthamalus from the Rosen BR, Smith AB 1988. Tectonics from fossils? Ana- western shores of the Atlantic océan. / Mar Biol Ass lysis of reef coral and sea-urchin distributions from UK 60: 787-831. Late Cretaceous to récent, using a new method. Engler A 1882. Versuch einer Entwicklungsgeschichte Gondwana and Tethys (ed. By MG Audley-Charles & der extratropischen Florengebiete (2 vols), Engle- A Hallam): 275-306. Geological Soc Sp Publ: 37. mann, Leipzig. Rodriguez G, Campo MR 1998. A cladistic revision of Hastings PA 2000. Biogeography of the Tropical Eastern the genus Fredius (Crustacea: Decapoda: Pseudothel- Pacific: distribution and phylogeny of chaenopsid fis- phusidae) and its significance to the biogeography of hes. Zool J Lin Soc 128: 319-335. the Guianan lowlands of South America. J Nat Hist Hengeveld R 1990. Dynamic biogeography. Cambridge 32: 763-775. Studies in Ecology, Cambridge University Press, 249 Sax DF, Brown JH 2000. The paradox of invasion. Glo- P- bal Ecol Biogeogr 9: 363-371. Hennig W 1966. Phylogenetic Systematics. Illinois Na- Wallace AR 1876. The geographical distribution of ani- tural History Survey, Urbana, Illinois. mais. 2 vols, Macmillan, London Hill WO 1953. Primates. Comparative anatomy and Williams WD 1964. Subterranean freshwater prawns taxonomy 1. Strepsirhini. Edinburgh University (Crustacea: Decapoda: Atyidae) in Australia, Austr J Press, Edinburgh,. Mar Freshwater Res 15: 93-106. ENDEMISM 203

APPENDIX

Characters and character states analysed 1. Rostral tip acute (0) or rounded (1). 2. Latéral margin of rostrum entire (0) or serrate (1). 3. Dorsal ridge on rostrum présent (0) or absent (1). 4. Rostrum triangular shaped (0) or broad (1). 5. Supra-orbital spine présent (0) or absent (1). 6. Branchiostegal spine présent (0) or absent (1). 7. Mid-dorsal spine on posterior margin of second abdominal segment présent (0) or absent (1). 8. Distolateral tooth of antennal scale présent (0) or absent (1). 9. Three (0), four (1) or more than four (2) pairs of posterior telson spines. 10. Mandibular incisor toothed (0) or entire (1). 11. Mandibular palp with none (0), one (1) or more (2) than one segment. 12. Dactyli of third to fifth pereiopods serrate (0) or entire (1). 13. Exopods on fourth and fifth pereiopods présent (0) or absent (1). 14. Latéral margin of uropodal exopod serrate (0) or entire (1).

Data matrix

D. atlanticus Gurney, 1939 1 0 0 1 1 1 0 0 0 0 ? 1 0 1 D. brownae Kensley, 1983 0 1 1 1 1 1 0 1 0 0 2 1 0 1 D. exul Kemp, 1920 0 1 0 1 1 1 0 1 1 0 2 1 0 1 D. musicus Holthuis, 1981 0 0 0 1 1 1 0 1 0 0 0 1 0 1 D. pascuensis Fransen, 1987 0 1 0 1 1 1 0 1 0 0 1 1 0 1 D. serratirostris Lebour, 1949 0 1 1 1 1 1 0 1 0 0 1 1 0 1 D. serrifer Rathbun, 1902 0 1 1 1 1 1 0 1 1 0 2 1 0 0 D. vernbergi Boothe & Heard, 1987 0 1 0 1 1 1 1 1 2 0 2 1 0 0 T. transkei Kensley, 1983 0 0 1 0 0 0 1 0 2 1 2 0 1 1