AMERICAN MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3058, 11 pp., 6 figures March 24, 1993

Egg Attachment Systems in the Family Cichlidae (: Labroidei), with Some Comments on Their Significance for Phylogenetic Studies

MELANIE L. J. STIASSNY1 AND JASON G. MEZEY2

ABSTRACT Among substrate-spawning two quite Outgroup data from the Pomacentridae are distinct egg attachment systems are described us- equivocal and do not permit conclusive assign- ing scanning electron microscopy. In the phylo- ment ofpolarity for variation within the Cichlidae. genetically primitive Asian and Madagascan cich- However, while we are unable to determine the lids examined, the eggs are nonadhesive and are plesiomorphic attachment system, the attached to the substrate, or to one another, by a presence of a distinct mucus-filament attachment tuft offilaments arising from the pole opposite the system is unique among percomorph taxa exam- micropyle. A different system appears to charac- ined to date and is interpreted here as further sup- terize the Afro-Neotropical clade; in these fishes port for the monophyly of the Afro-Neotropical the eggs are highly adhesive with attachment to Cichlidae. the substrate facilitated by a thick mucus layer and numerous surface filaments.

INTRODUCTION Recent investigations of the reproductive Gerstner, 1992). It is this finding, and our biology of a phylogenetically primitive cich- attempts to interpret such an attachment sys- lid fish from Madagascar, polleni tem within a phylogenetic framework, that Bleeker, 1868, have revealed an unusual egg has stimulated this review of cichlid egg ul- attachment system in this taxon (Stiassny and trastructure (Mezey, 1992). Wickler (1956a,

Associate Curator, Department of Herpetology and Ichthyology, American Museum of Natural History. 2 Undergraduate student intern.

Copyright © American Museum of Natural History 1993 ISSN 0003-0082 / Price $1.60 2 AMERICAN MUSEUM NOVITATES NO. 3058

ultrastructure beyond that provided by the light microscopy studies of Wickler (1956a, 1956b, 1962). The recent papers ofMooi (1990) and Mooi et al. (1990) provide excellent detailed data on egg ultrastructure in some demersal spawning perciforms. Comparative data of a similar quality are scarce, but Mooi's anal- A yses highlight the potential for additional studies of egg surface structure in other per- Fig. 1. Schematic representation of Wickler's comorph taxa. (1956a, b) (A) "p"-type and (B) "I"-type eggs. PREPARATION AND MATERIALS PREPARATION TECHNIQUES 1956b) provided a baseline light microscopy study of cichlid eggs in a review of egg at- All eggs were obtained immediately post tachment systems in a range of Asian, Neo- fertilization from stocks. After tropical, and African taxa. While Wickler's careful removal from the substrate, the eggs study was essentially ecological in orienta- were fixed in 0.2M glutaraldehyde solution tion, he noted that among the Cichlidae two for a 24 hr period. Ovarian eggs of a gravid quite distinct egg types are recognizable (fig. formalin preserved female Paratilapia pol- 1). In Wickler's "p"-type, as represented in leni were carefully cleaned of excess ovarian the Asian Etroplus, the eggs are at- tissue and refixed in glutaraldehyde solution tached to the substrate by a filament tuft at for 24 hr. Our protocol involved successive one pole, the egg surface is without a mucus washings with 0.2M Na Cacodylate buffer, coat, and is nonadhesive. Wickler's "l"-type immersion in osmium tetroxide, dehydration eggs, common to most African and Neotrop- through a graded series of alcohols, and final ical species, lack the filament tuft and instead immersion in a 50% ethyl alcohol/50% amyl are attached along their longitudinal axes by acetate solution. Treated eggs were then im- an encasing mucus/filament layer. Although mersed in amyl acetate for 24 hr before crit- he presented no phylogenetic argument, ical-point drying and splutter coating with Wickler believed that the "l"-type egg is char- gold for SEM viewing. acteristic of "genuine cichlids," while "p"- type eggs are found in forms only "marginally MATERLAL EXAMINED considered as cichlids" (Wickler, 1956a). Cichlidae There is little published information on egg ultrastructure among labroid fishes (Kauf- Asian: Etroplus maculatus (Bloch, 1795); man and Liem, 1982). The embiotocids are Madagascan: (Bleeker, viviparous (Wourms, 1981) and therefore of 1868); Neotropical: "Cichlasoma (Archocen- limited utility in the present context. Some trus)" nigrofasciatus (Gunther, 1869), Her- data on the egg morphology of various po- otilapia multispinosa (Gunther, 1866), macentrid species are available (e.g., Shaw, "Cichlasoma (Amphilophus)" citrinellum 1955; Swerdloff, 1970; Re, 1980, Thresher, (Gunther, 1864); African: zillui (Ger- 1984, Mooi, 1990), but detailed information vais, 1848), pulcher (Boulen- on labrid egg structure, particularly that of ger, 1901), Lamprologus mustax Poll, 1978. the demersal eggs of the nest building Euro- pean Labrinae, are virtually nonexistent Pomacentridae: Amphiprion sp. (Richards and Leis, 1984; Russell, 1976). Somewhat surprisingly given the vast liter- METHODS ature on their reproductive behavior and bi- Throughout this study we have attempted ology (e.g., Baerends and Baerends Van Roon, to determine character polarity using the out- 1950; Breder and Rosen, 1966; Keenleyside, group method (Maddison et al., 1984). How- 1991; Barlow, 1991), there is little detailed ever, in the absence of a well-supported information on substrate-spawning cichlid egg scheme of labroid interrelationships, selec- 1993 STIASSNY AND MEZEY: EGG ATTACHMENT IN CICHLIDAE 3 tion of appropriate perciform outgroups for pomacentrid eggs for use in this study. Cyn- the present study is highly problematical. thia Gerstner's preliminary studies ofthe eggs Based on similarities in pharyngeal anatomy, of Paratilapia laid the foundation for the Stiassny (1981) suggested a possible relation- present study, and we are grateful to her for ship between the labroids and two perciform all of her help. Thanks also to Carlos Rod- families; the Sparidae and the Gerreidae. Un- riques, William Barnett, Jennie Ono, and fortunately, what little information that is Peling Fong for their patience and skill in the available on egg structure in these families SEM laboratory. Jackie Lazzaro ably cared indicates that their eggs, like those ofthe great for aquarium fish stocks and successfully majority of marine perciforms (Breder and goaded Etroplus maculatus into spawning. Rosen, 1966; Leis and Trnski, 1989; Mooi et Klaus Kallman kindly helped with the Ger- al., 1990) are pelagic and typically without man translation of a number of important any attachment structures or surface orna- papers. For many helpful comments on ear- mentation (Lo Bianco, 1956; Breder and Ro- lier drafts ofthis paper we thank Randy Mooi, sen, 1966; Thresher, 1984), and are therefore Mario de Pinna, and Tony Gill. This work oflittle utility in helping to polarize variation was supported by NSF Research Experiences in attachment structures ofthe demersal eggs for Undergraduates (REU) Site Awards to the of labroids. senior author. Funding from the NSF REU Equally problematical are the precise in- awards supported the summer research of trarelationships ofthe labroid families them- Jason Mezey (DIR 9001058), Cynthia Gerst- selves. Stiassny and Jensen (1987) reviewed ner (DIR 9001058), and Jackie Lazzaro (DIR the morphological data bearing on this ques- 9200351). tion and favored the most parsimonious scheme in which the Cichlidae form the sister RESULTS group to a clade containing the Embiotocidae and a pomacentrid-labrid subclade (Stiassny POMACENTRIDAE and Jensen, 1987: fig. 19). However, previous Pomacentrids lay a plaque ofadhesive de- hypotheses of labroid intrarelationships dif- mersal eggs which are guarded by the male, fer from this scheme and place the Poma- sometimes assisted by the female, usually un- centridae as the basal labroid clade forming til hatching as pelagic larvae some 2 to 7 days the sister group to the Cichlidae and a labrid- after spawning. The eggs, which are typically embiotocid subclade (Stiassny, 1980; Kauf- elliptical, range in size from 0.49 to 4.50 mm man and Liem, 1982). Given the uncertainty along the highest dimension (Thresher, 1984) of labroid intrarelationships (Stiassny and and are attached to the substrate by a dense Jensen, 1987), coupled with the absence of tuft of filaments. Most descriptions of po- pertinent information in other labroids, it macentrid eggs are based on field observa- seems reasonable to treat the demersal tions or light microscopic examination, and spawning Pomacentridae as an operational consequently, details of attachment filament outgroup for polarity assessment ofegg struc- structure and surface ornamentation are lack- ture variation within the family Cichlidae. ing. Information on the egg structure in the de- Mooi (1990) examined ovarian eggs from mersal spawning labrid subgroup Labrinae three pomacentrid species, Paraglyphidodon might bear on this question, but until we are nigroris, Chromis weberi, and Amphiprion able to examine spawned eggs of these fishes clarki. He noted the presence of a complex little more can be added. Our assumption of two-layered sheet that partially or totally en- outgroup relationships is admittedly tenta- cases the eggs. The outer layer is an anasto- tive, but is open to further testing and re- mosing network of filaments while the inner finement as more data become available. layer consists of longer filaments closely ap- posed to, or continuous with, those of the ACKNOWLEDGMENTS outer layer. He suggested that both of these layers peel off the egg surface to form an ad- Our thanks go to Cynthia Gerstner, Ron hesive disc or cup which attaches the egg to Coleman, and Jason Schulterbrandt for kind- the substrate (Mooi, 1990: 465). ly sending us an assortment of cichlid and We have been unable to confirm the pres- 4 AMERICAN MUSEUM NOVITATES NO. 3058

ence of a complex two-layered network in spawned pomacentrid eggs (fig. 2). Rather, as reported by other authors who illustrate spawned pomacentrid eggs (e.g., Abudefduf: Shaw, 1955; Re, 1980. Chromis: Swerdloff, 1970; Thresher, 1984. Amphiprion: Thresh- er, 1984, and Glyphidodontops: Thresher, 1984) we find a single filament tuft arising from around the micropyle that attaches each egg to the substratum (fig. 2A, B). The re- maining egg surface is smooth, nonadherent, and without surface filaments or mucus coat- ing (fig. 2C). A similar, though less well-de- fined, meshlike network encases the ovarian eggs of Paratilapia polleni (fig. 4D). As this encasing network is not present in spawned Paratilapia eggs (fig. 4C), we suggest that the outer layer described in pomacentrids by Mooi (1990) is possibly ovarian in origin and may be shed at or prior to spawning leaving the inner filament layer, originating around the micropyle, as the attachment structure. CICHLIDAE Cichlids are often divided into two poly- phyletic behavioral groups: substrate-spawn- ers and mouthbrooders (e.g., Baerends and Baerends van Roon, 1950; Fryer and Iles, 1972; Noakes and Balon, 1982; Keenleyside, 1991; Barlow, 1991). Substrate-spawning is the plesiomorphic cichlid condition and mouthbrooding, a derived condition, has arisen independently at least three (and prob- ably many more) times within the Afro-Neo- tropical cichlid radiation (Stiassny and Gerst- ner, 1992). Wickler (1956a, 1956b) did not include mouthbrooding cichlid lineages in his initial studies. However, some details of the surface structure of mouthbrooded eggs are given in a later paper (Wickler, 1962), as well as in Kraft and Peters (1963) and Bern and Avtalion (1990). Mouthbrooded eggs are all characterized by an increase in size, a loss of the adhesive mucous coat, and a reduction or loss of surface filaments (Fryer and Iles, 1972). Here we restrict our review to the eggs of substrate-spawning taxa representing each of the major geographical cichlid clades (Stiass- Fig. 2. Egg morphology ofAmphiprion sp. (A) ny, 1991: fig. 1.3). Single egg, micropyle obscured by attachment fil- Asian Cichlidae: Eggs of the Orange aments; (B) filament tuft surrounding micropyle; Chromide, Etroplus maculatus, bear a marked (C) surface ofthe chorion with no filaments. Scale superficial similarity to those of pomacen- bars are in micrometers. trids. They are elliptical and measure 1.5 mm 1993 STIASSNY AND MEZEY: EGG ATTACHMENT IN CICHLIDAE 5

-7_l Fig. 3. Egg morphology of Etroplus maculatus. (A) Single egg; (B) filament tuft at pole opposite micropyle; (C) surface of chonon studded with short filaments; (D) micropylar pole. Scale bars are in micrometers. along the longest dimension (fig. 3). Each egg study revealed a similar filament tuft attach- is attached to the substrate by a thick filament ment system in the Green Chromide, Etro- tuft but, unlike the situation in the Poma- plus suratensis (Bloch, 1790). centridae, the etropline attachment tuft arises Madagascan Cichlidae: Eggs of the Mad- from the pole opposite the micropyle (fig. 3A, agascan cichlid, Paratilapia polleni, are B, D). The remaining egg surface is nonad- slightly elliptical and measure 1.2 mm along herent, but is studded with short filaments the longest dimension (fig. 4). A tuft of elon- (fig. 3C). These surface filaments increase gate filaments arises from the pole opposite slightly in length around the micropyle (fig. the micropyle (fig. 4A, B), very much as in 3D). No trace of an encasing mucus layer is the manner of the etropline cichlids. How- present. Wickler's (1956a) light microscopy ever, this filament tuft does not attach the 6 AMERICAN MUSEUM NOVITATES NO. 3058

I

~\No i*7'

Fig. 4. Egg morphology ofParatilapia polleni. (A) Single egg; (B) filament tufts oftwo eggs entwined; (C) surface of chorion studded with extremely short filament stubs; (D) surface ofchorion of an ovarian egg; (E) micropylar pole. Scale bars are in micrometers.

Paratilapia eggs to the substrate. Instead, the SA). The eggs are extremely "sticky" and ad- tufts ofindividual eggs entangle with one an- here, most often along their longitudinal axes, other resulting in a moveable egg mass to the substrate by a distinctive mucus/fila- (Stiassny and Gerstner, 1992; fig. 4B). The ment layer (fig. 5B). At points of contact be- remaining egg surface is nonadherent and is tween eggs, egg to egg attachment is also fa- studded with very short filaments which are cilitated by the glutinous mucus coating and little more than filament stubs (fig. 4C). The fine filament threads (Kraft and Peters, 1963). micropyle is ringed by somewhat more elon- No attachment tuft is present and the entire gate filaments (fig. 4E). No trace of an en- surface of the egg is covered with relatively casing mucus layer is present. elongate filaments (e.g., fig. 5C) which are Afro-Neotropical Cichlidae: In this large encased in a dense mucus layer (fig. 5B). Ac- clade the eggs are typically somewhat ellip- cording to Kraft and Peters (1963), glutinous tical and vary in size usually between 1.0 and ovarian secretions adhere to the egg surface 1.5 mm along the longest dimension (see also just prior to spawning. These highly adhesive Wickler, 1956a; Noakes and Balon, 1982; fig. eggs will attach wherever they contact the 1993 STIASSNY AND MEZEY: EGG ATTACHMENT IN CICHLIDAE 7

MIS~*MA.-N~~ It v 2- J Fig. 5. Egg morphology of Tilapia zillui. (A) Single egg; (B) mucus layer and mucus-encrusted filaments on surface of egg; (C) surface of chorion covered with elongated filaments; (D) micropylar pole. Scale bars are in micrometers. substrate or each other. When the eggs are mis pulcher are extremely sticky and covered removed from the substrate their point of with a thicker layer ofmucus and have longer attachment is usually evident as a smooth surface filaments than those of other species. area from which most of the mucus and fil- Clearly, our sampling ofNeotropical and Af- aments are absent from the eggs and remain rican lineages is limited; however, Wickler's adherent to the substratum (fig. 5A). (1956a) study incorporated three additional Among the Neotropical and African taxa African species {Hemichromis fasciatus Pe- we examined, variation in egg structure is ters, 1858, Hemichromis bimaculatus Gill, minimal and manifest only in the amount of 1862, and Nanochromis nudiceps (Boulenger, mucus adherent to the eggs and in the length 1899)} and nine additional Neotropical spe- ofthe surface filaments. For example, the eggs cies {Aequidens curviceps [= Laeticara curv- ofthe African chromidotilapine Pelvicachro- iceps (Ahl, 1924)], Aequidens portalegrensis 8 AMERICAN MUSEUM NOVITATES NO. 3058

[= Cichlasoma portalegrensis (Hensel, 1870)], polarity of variation in egg attachment sys- Apistogramma ramirezi [= Papiliochromis tems within the Cichlidae, outgroup data from ramirezi (Meyers and Harry, 1948)], Cich- the Pomacentridae are equivocal. The ap- lasoma severum [= Heros severus Heckel, parent similarity of the filament tuft attach- 1840], Cichlasoma festivum [= Mesonauta ment system in pomacentrids and in the basal festivus (Heckel, 1840)], Pterophyllum scalare cichlid clades is probably spurious. In po- (Lichtenstein, 1832), Symphysodon discus macentrids the attachment tuft arises from Heckel, 1840, Geophagus brasiliensis (Quoy around the micropyle, while in etroplines and and Gaimaird, 1824), and Geophagus cupido Paratilapia the attachment tuft originates [= Biotodoma cupido (Heckel, 1840)}. With- from around the pole opposite the micropyle. in that sample, Wickler (1956a) also noted Mooi (1990) noted similar variation in fila- similar variation in the thickness of the mu- ment location with respect to the micropyle cus covering and in filament dimensions and in the various demersal spawning "pseu- their distribution over the egg surface. For dochromoid" taxa he investigated. For ex- example, Symphysodon and Pterophyllum ample, in most pseudochromid subfamilies, apparently share a specialized configuration filaments originate in the region opposite the in which the mucus and surface filaments are micropyle, while in opistognathids, gram- arranged in a broad "belt" around the girth matids, and plesiopids, the filaments arise of the egg. around the micropyle, as they do in apogo- Interestingly, he also found that the eggs of nids and pomacentrids (Mooi, 1990). Eggs of one African chromidotilapine (Nanochromis two other demersal spawning percomorph nudiceps) and one Neotropical geophagine families, the Gobiidae and Blenniidae, ap- (Biotodoma cupido) differed from the rest of parently also bear an attachment tuft or disc the group. In these two species Wickler located around the micropyle (Lo Bianco, (1956a, 1956b) recorded the presence of a 1956; Wickler, 1956a, 1956b; Russell, 1976; filament tuft by which the eggs are hung from Thresher, 1984). Wickler (1956a) suggested the roof of caves. We have been unable to that the position of the micropyle in sub- obtain spawned eggs of either of these two strate-attached eggs is closely matched by taxa, however Linke (1989) illustrated the eggs spawning behavior. In fishes where the at- ofan undescribed species ofNanochromis. In tachment tuft encircles the micropyle, the mi- this species the eggs are suspended from the cropyle is functionally closed to sperm entry roof of the spawning tube by an adhesive after attachment and fertilization and egg ex- filament up to 1.5 mm in length. According trusion must be closely synchronized. In to Linke (1989: 105) the adhesive filament is cichlids, fertilization can occur after egg at- composed of several individual fibers which tachment has taken place. enclose the egg in "a sort of net resembling While outgroup data from the Pomacen- an inverted parachute." tridae do not permit conclusive assignment ofpolarity for variation within the Cichlidae, the presence of a polar tuft of filaments at- DISCUSSION taching the eggs to the substrate is not un- The monophyly of the family Cichlidae is common among demersal spawning perco- well supported by data from a wide range of morphs. However, as the homology of these anatomical systems (Stiassny, 1981, Zihler, systems in different families remains obscure 1982; Gaemers, 1984) and Stiassny (1991) (see also Mooi, 1990), we are unable to de- provided a recent discussion and summary termine with any certainty the plesiomorphic ofintrafamilial relationships. While her con- cichlid attachment system. For the time be- clusions are tentative, at least at the more ing we are able to do little more than spec- inclusive levels, a hypothesis of cichlid in- ulate that the ancestral cichlid egg attachment trarelationship is now available, and varia- system was probably similar to that of the tion in egg attachment systems can be ex- etropline cichlids (the "p"-type egg of Wick- plored within that phylogenetic framework ler, 1956a). The configuration in the Mada- (Stiassny, 1991: fig. 6). gascan genus, Paratilapia, in which the polar With respect to the determination of the attachment tuft of each egg entwines to form 1 993 STIASSNY AND MEZEY: EGG ATTACHMENT IN CICHLIDAE 9

AFRO-NEUTR(UPIUAL. - - - . . - I--, -r- r-% 0--. r-% I P% A I CICHLIDAE-% I -% I I I I rl% A r- -7

,~Z. /cP j

Fig. 6. Strict consensus tree ofcichlid intrarelationships. Modified after Stiassny (1991). The asterisks indicate lineages in which the occurrence of mouthbrooding in some members has been documented. Cross-bar indicates hypothesized level of origin of the derived "l"-type egg. For a list of the genera included in each named lineage, and characters supporting the various clades, see Stiassny (1991). an interconnected and mobile egg mass nus, probably representing a modification of (Stiassny and Gerstner, 1992: fig. 4B), is a the tuft-to-substrate attachment system pres- unique system within the family. It is of in- ent in etroplines. terest to note that a number of other perci- Further information on the attachment form taxa have similar moveable egg masses. system of eggs in the remaining basal cichlid For example, Mooi (1990) described the egg clades, in particular the ptychochromines and balls of pseudochromids, plesiopids and in the aberrant African genus, Heterochro- opistognathids as being formed by the entan- mis, would be of great interest (fig. 6). Al- glement offilaments ofadjacent eggs. In opis- though we have been unable to obtain tognathids the egg ball is formed of filament spawned eggs from these basal cichlid taxa, threads entangled to form braided ropelike de Rham (in litt.) reported that the eggs of strands (Mooi, 1990), a condition seemingly Ptychochromis oligacanthus (Bleeker, 1868) very similar to that observed in Paratilapia. are unlike those of Paratilapia and are at- However, as noted previously, the filaments tached individually to the substrate in a man- in these taxa originate from around the mi- ner similar to that observed in the etropline cropyle rather than from the opposite pole as genus . We anticipate that Pare- in Paratilapia and other basal cichlids. De- troplus, like its sister genus Etroplus (Stiass- spite the superficial similarities between the ny, 1991) will attach its nonadhesive eggs attachment system ofParatilapia and that of individually to the substrate via single fila- some of the "pseudochromoids" described mentous attachment tufts. by Mooi (1990) we interpret the system in The cichlid egg attachment system in which Paratilapia as an autapomorphy of that ge- an adhesive mucus layer encases a filament- 10 AMERICAN MUSEUM NOVITATES NO. 3058 covered egg (the "l"-type egg of Wickler, and African subclades (Stiassny, 1991) in 1 956a) is unique among percomorph taxa ex- which the other clade members, and taxa bas- amined to date and as such is reasonably in- al to them, exhibit the derived mucus/fila- terpreted here as an innovation of the Afro- ment adhesion system. Neotropical Cichlidae (fig. 6). The apparent Prior to this study, Stiassny (1991) iden- similarity between the eggs ofthe Neotropical tified two derived anatomical characters sup- geophagine Biotodoma and the African porting the monophyly of the Afro-Neotrop- chromidotilapine Nanochromis, with those ical Cichlidae (excluding the Zairian genus, of the etroplines (Wickler, 1956a, 1956b) re- Heterochromis). The recognition ofa derived quires further investigation. However, until egg morphology at this level of the cichlid spawned eggs from these taxa become avail- tree adds further support for the monophyly able for study, it is not possible to determine of this extensive cichlid radiation (fig. 6). whether the similarities noted are more than The results of this preliminary study high- superficial. Linke's (1989) description of the light the potential utility ofegg ultrastructure attachment tuft in Nanochromis as "a sort of variation in studies of cichlid phylogenetic net resembling an inverted parachute" cer- intrarelationships. More data from addition- tainly suggests nonhomology with the etro- al taxa are necessary before the precise dis- pline tuft system. It is the case, however, that tribution, extent, and homology of variation whatever the morphological similarities, par- in egg structure within the family can be de- simony requires an assumption ofhomoplasy termined, but our preliminary observations for the egg attachment systems ofBiotodoma suggest that this system is potentially rich in and Nanochromis as both of these taxa are phylogenetically informative variation. embedded within monophyletic Neotropical

REFERENCES Baerends, G. P., and J. M. Baerends van Roon and evolutionary significance. Breviora 1950. An introduction to the study of the 472: 1-19. ethology of cichlid fishes. Behaviour Keenleyside, M. H. A. Suppl. 1: 1-242. 1991. Parental care. In M. H. A. Keenleyside Barlow, G. W. (ed.), Cichlid fishes: behaviour, ecology 1991. Mating systems among cichlid fishes. In and evolution, pp. 191-224. London: M. H. A. Keenleyside (ed.), Cichlid fish- Chapman and Hall. es: behaviour, ecology, and evolution, Kraft, A. v., and H. M. Peters pp. 173-190. London: Chapman and 1963. Vergleichende Studien uiber die Ooge- Hall. nese in der Gattung Tilapia (Cichlidae, Bern, O., and R. R. Avtalion Teleostei). Z. Zellforsch. 61: 434-485. 1990. Some morphological aspects of fertil- Leis, J. M., and T. Trnski ization in . J. Fish. Biol. 36: 375- 1989. The larvae of Indo-Pacific shorefishes. 381. Honolulu: University of Hawaii Press, Breder, C. M., Jr., and D. E. Rosen pp. 371. 1966. Modes of reproduction in fishes. Nep- Linke, H. tune City: T.F.H. Publications, 941 pp. 1989. The silver-bellied Nanochromis. Trop. Fryer, G., and T. D. Iles Fish Hobbyist 38(1): 102-111. 1972. The cichlid fishes of the Great Lakes of Lo Bianco, S. Africa. Their biology and evolution. Ed- 1956. Uova, larve e stadi Giovanili de Te- inburgh: Oliver and Boyd, 641 pp. leostei. Fauna e Flora del Golfo di Na- Gaemers, P. A. M. poli, Monografia della Stazione Zoolo- 1984. Taxonomic position of the Cichlidae as gia di Napoli 38 (1, 2). 1064 pp. demonstrated by the morphology of Maddison, W. P., M. J. Donoghue, and D. R. their otoliths. Netherlands J. Zool. 34: Maddison 566-595. 1984. Outgroup analysis and parsimony. Syst. Kaufman, L., and Liem, K. F. Zool. 33: 83-103. 1982. Fishes of the suborder Labroidei (Pi- Mezey, J. G. sces: Perciformes): phylogeny, ecology 1992. A comparative scanning electron mi- 1993 STIASSNY AND MEZEY: EGG ATTACHMENT IN CICHLIDAE I1I

croscope study of cichlid eggs and its 1991. Phylogenetic intrarelationships of the implications on cichlid phylogeny. Pro- family Cichlidae: an overview. In M. H. ceedings Natl. Conf. Undergrad. Res. A. Keenleyside (ed.), Cichlid fishes: be- VI(1992): 1-6. haviour, ecology and evolution, pp. 1- Mooi, R. D. 35. London: Chapman and Hall. 1990. Egg surface morphology ofpseudochro- Stiassny, M. L. J., and C. L. Gerstner moids (Perciformes: Percoidei) with 1992. The parental care behaviour of Parati- comments on its phylogenetic implica- lapia polleni (Perciformes, Labroidei), a tions. Copeia (2): 455-475. phylogenetically primitive cichlid from Mooi, R. D., R. Winterbottom, and M. Burridge Madagascar, with a discussion of the 1990. Egg surface morphology, development, evolution ofmaternal care in the family and evolution in the Congrogadinae (Pi- Cichlidae. Environ. Biol. Fishes 34:219- sces: Perciformes: Pseudochromidae). 233. Can. J. Zool. 68: 923-934. Stiassny, M. L. J., and J. S. Jensen Noakes, D. L. G., and E. K. Balon 1987. Labroid intrarelationships revisited: 1982. Life histories oftilapia; an evolutionary morphological complexity, "key inno- perspective. In R. S. V. Pullin and R. vation", and the study of comparative H. Lowe-McConnell (eds.), The biology diversity. Bull. Mus. Comp. Zool. 151: and culture of tilapias, ICLARM Con- 269-319. ference Proceedings 7, Manila. Swerdloff, S. N. Re, P. 1970. Behavioral observations on Eniwetok 1980. The eggs and newly hatched larvae of damselfishes (Pomacentridae: Chromis) Abudefduf luridus (Cuvier, 1830) (Pi- with special reference to the spawning sces: Pomacentridae) from the Azores. of Chromis caeruleus. Copeia (2): 371- Publ. Mus. Fac. Cienc. Lisboa 7(8): 109- 374. 115. Thresher, R. E. Richards, W. J., and J. M. Leis 1984. Reproduction in reef fishes. Neptune 1984. Labroidei: development and relation- City: T.F.H. Publs., 399 pp. ships. In Moser et al. (eds.), Ontogeny Wickler, W. and systematics of fishes, pp. 542-547. 1956a. Der Haftapparat einiger Cichliden-eier. Spec. Publ. No. 1, American Society of Z. Zellforsch. 45: 304-327. Ichthyologists and Herpetologists. 1956b. Unterschiede zwischen den Cichliden- Russell, F. S. Gattungen, speziell Geophagus und Bio- 1976. The eggs and planktonic stages of Brit- todoma, im Haftapparat der Eier. Na- ish marine fishes. London: Academic turwiss. 43: 333-334. Press, 524 pp. 1962. Ei-Attrapen und Maulbruten bei afri- Shaw, E. S. kanischen Cichliden. Z. Tierpsychol. 19: 1955. The embryology of the Sargent Major, 129-164. Abudefdufsaxatalis. Copeia (2): 85-89. Wourms, J. P. Stiassny, M. L. J. 1981. Viviparity: the maternal-fetal relation- 1980. The anatomy and relationships of two ship in fishes. Am. Zool. 21: 473-515. genera of African cichlid fishes. Ph.D. Zihler, F. thesis, University of London. 1982. Gross morphology and configuration of 1981. The phyletic status of the family Cich- digestive tracts of Cichlidae (Teleostei, lidae (Pisces: Cichlidae): a comparative Perciformes): phylogenetic and func- anatomical investigation. Netherlands tional significance. Netherlands J. Zool. J. Zool. 31: 275-314. 32: 544-571. Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024.

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.