Novltates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

AMERICAN MUSEUM Novltates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3195, 17 pp., 6 figures, 1 table May 16, 1997

Egg Surface Structure and Larval Cement Glands in Nandid and Badid Fishes with Remarks on Phylogeny and Biogeography*

RALF BRITZ'

ABSTRACT Egg surface structure and larval morphology of procedure. The genus Afronandus is tentatively the Nandidae and Badidae were studied with assigned to this monophyletic group because it SEM, and cement organs of larval Nandus and shares with the other three African-South Amer- Badis were histochemically stained using the PAS ican Nandidae the character of adhesive filaments technique. The study is supplemented with data at the vegetal egg pole. Comparison of egg and on reproductive behavior. The Asian Nandidae larval structure between the Nandidae and Badi- differ from the African-South American Nandi- dae revealed no characters indicating a close re- dae in important features of reproductive behavior lationship of the two families. as well as egg and larval structure. No synapo- Lundberg's (1993) hypothesis, which explains morphy for the family Nandidae could be identi- the distribution of the African-South American fied. The genera Polycentropsis, Polycentrus, and Nandidae by dispersal through seawater, is reject- Monocirrhus, however, form a monophyletic ed on the basis of the ecological preferences of group on the basis of the following synapomor- these Nandidae. The age of origin of African- phies: eggs with a unique surface pattern of nar- South American Nandidae is hypothesized to date row ridges running radially from the micropyle; back at least to the late Cretaceous, that is, before larvae with a multicellular cement gland on top the separation of Africa and South America. of the head; and adults with a unique spawning

* This paper is dedicated to the memory of the influential zoologist Hans M. Peters who died December 13, 1996, at the age of 89. ' Postdoctoral Fellow, Department of Herpetology and Ichthyology, American Museum of Natural History. Present address: Lehrstuhl fur Spezielle Zoologie, Universitat Tubingen, Auf der Morgenstelle 28, 72076 Tubingen, Germany.

INTRODUCTION Nandidae, or leaffishes, occur in fresh wa- mented by data obtained from ovarian eggs ters of Southeast Asia, West Africa, and of the poorly known West African species South America (Berra, 1981; Nelson, 1994), Afronandus sheljuhzkoi. After evaluating the a disjunct distribution that has been noted by phylogenetic significance of these new data, a number of ichthyologists and zoogeogra- the hypothesis recently proposed by Lund- phers (see Lundberg, 1993 for a review). In berg (1993) to explain the current distribu- the past, seven genera, Nandus, Afronandus, tion of the family Nandidae and its sub- Polycentropsis, Polycentrus, Monocirrhus, groups was examined. Badis, and Pristolepis, have been arranged in various combinations with different numbers ACKNOWLEDGMENTS of families (Gunther, 1861; Boulenger, 1904; Jordan, 1923; Weber and De Beaufort, 1936; I am deeply indebted to R. Rucks (Kiel) Berg, 1958; Greenwood et al., 1966; Liem, and R. Donoso-Buchner (Dorsten) for pro- 1970; Lauder and Liem, 1983; Nelson, viding eggs and larvae of Polycentropsis, 1994). They are usually classified as mem- Monocirrhus, and Badis and for sharing with bers of the Percoidei (Johnson, 1984; Nelson, me their profound knowledge of nandid be- 1994), although sometimes a closer relation- havior. I am grateful to the individuals at the ship to the Anabantoidei (labyrinth fishes) or Lehrstuhl fur Spezielle Zoologie, Universitat the Channidae (snakeheads) has been postu- Tiubingen, especially W. Maier for the op- lated (Gosline, 1968, 1971; Nelson, 1969; portunity to maintain specimens of N. nan- Rosen and Patterson, 1990). Barlow et al. dus and B. badis at his institute, H. Schop- (1968) compared breeding behavior, egg and pmann for his expert skills with SEM, and larval morphology, and osteology of Badis M. Hohloch and M. Meinert for a variety of and Polycentrus and found striking differ- technical advice and assistance. G. Teugels, ences between the two genera. As a conse- Musee Royale de L'Afrique Centrale quence, they removed Badis from the Nan- (MRAC; Tervuren), kindly lent a specimen monotypic family, of Afronandus shelhjuhzkoi for SEM of its didae and erected a new eggs. At the American Museum of Natural Badidae. Subsequently, Pristolepis also was P. to the History (New York), the help of Fong- considered to be related only remotely Melville with SEM was greatly appreciated. Nandidae (Liem, 1970; Liem and Green- The present study was supported by a grant wood, 1981). Liem (1970) restricted the fam- of the Landesgraduiertenforderung Baden- ily Nandidae to five genera, Nandus, Afron- Wurttemberg at the Lehrstuhl fur Spezielle andus, Polycentropsis, Polycentrus, and Zoologie and subsequently by a Kalbfleisch Monocirrhus, and provided an osteological Postdoctoral Fellowship at the Department of definition of the group. Nevertheless, he ad- Herpetology and Ichthyology of the Ameri- mitted that "no single osteological feature can Museum of Natural History. distinguishes the family from other per- The manuscript was read and greatly im- coids" (op. cit. p. 82). The monophyly of the proved by the comments of M. Toledo-Piza, Nandidae is still questionable and its phylo- G. Nelson, and J. Atz and by the reviews of genetic inter- and intrarelationships remain S. 0. Kullander and K. F Liem. unresolved (Lundberg, 1993). Aquarium breeding of three geographical- MATERIALS AND METHODS ly separated Nandidae-the Southeast Asian Nandus nandus, West African Polycentropsis Freshly spawned, fertilized eggs and lar- abbreviata, and South American Monocir- vae of N. nandus, P. abbreviata, M. poly- rhus polyacanthus-as well as the Southeast acanthus, and B. badis were prepared for Asian Badis badis provided the opportunity SEM using the procedure described in Britz to investigate whether egg and larval char- et al. (1995). They were observed and pho- acters reported for Polycentrus by Barlow et tographed with a Cambridge Stereoscan 250 al. (1968) are actually representative of the Mk2. Nandidae. These results have been supple- Ovarian eggs of A. sheljuhzkoi were ob- 1997 BRITZ: NANDID AND BADID FISHES 3 tained from a preserved female specimen tained about 70 ripe eggs with a diameter of (MRAC 73-05-P-4669-4674). After remov- 1.1-1.3 mm. The size of spawned, fertilized ing maternal tissue by use of fine forceps, eggs may be slightly larger due to formation eggs were critical-point dried in a Balzers of the perivitelline space (Laale, 1980). The CPD 030, coated with 150 A gold-palladium, vegetal egg pole bears a tuft of filaments that and observed and photographed using a originates from a circular area on the zona Zeiss DSM 950. radiata (fig. IE, F) and supposedly serves to To confirm the position and shape of ce- attach the egg to the substrate. The micropyle ment glands or individual cement cells, lar- is situated on the opposite pole (fig. IG) and vae of N. nandus (2.5 days postspawning) has a diameter of almost 4 ,um. The zona and B. badis (-3 days postspawning) were radiata near the micropyle does not show any studied histochemically. The larvae were striking surface structure (fig. 1H) apart from stained in toto for mucopolysaccharides in the numerous canal openings typical of many accordance with the PAS (= periodic acid teleost eggs (Stehr and Hawkes, 1983; Riehl, Schiff reaction) technique of Peters and 1991). Berns (1982) and then photographed with a POLYCENTROPSIS ABBREVIATA: The male of Zeiss Tessovar. this other West African nandid usually builds a nestlike structure of air bubbles under RESULTS leaves that are floating at the water surface; EGG STRUCTURE IN THE NANDIDAE eggs are attached to the underside of these leaves (Rucks, 1992). The male guards the NANDUS NANDUS: Two Southeast Asian spe- clutch, which may consist of 300-350 eggs cies of the genus Nandus, for which infor- (Scheel, 1964b; Rucks, 1992). Eggs measure mation is available, spawn several thousand 1.3-1.4 mm in diameter. SEM reveals that small, translucent eggs that adhere to plants their vegetal pole attaches to the leaf with the and other substrates and show no parental aid of a stalk of radially arranged fiber bun- care (for N. nandus: Parameshwaran et al., dles (fig. 2A). Each bundle originates from 1971; personal obs.; for N. nebulosus: Rucks, the zona radiata of the egg (fig. 2B) and ends 1973, 1996). Respective data for the newly at a distance of -0.5 cm (fig. 2A). Bundles described third species in the genus, Nandus are about 10 ,um wide and consist of many oxyrhynchus (Ng et al., 1996), are lacking. smaller individual fibers that had a width of Spawned eggs of N. nandus measure 0.7-0.8 0.2-0.3 ,um each (fig. 2B). The zona radiata mm in diameter. SEM demonstrated that the shows a distinct honeycomblike surface pat- animal pole of the egg adheres to the sub- tern where the fibers originate (fig. 2B). The strate (fig. IA); to observe the micropyle, the animal pole bears a considerable number of egg had to be removed from the substrate radial ridges which run radially from the oval (fig. iB). A circular area around the micro- micropylar pit situated in a craterlike eleva- pyle bears a dense carpet of short filaments tion (fig. 2C). The micropyle also has a (fig. IC, D) that appear to be primarily re- slightly oval shape, 4 ,um long and 2 ,um sponsible for the attachment of the egg to the wide (fig. 2D). substrate. The micropyle is situated in the MONOCIRRHUS POLYACANTHus: The South middle of this area and has a diameter of -2 American nandid M. polyacanthus attaches ,um (fig. IC, D). The remaining egg surface eggs to the underside of leaves of aquatic possesses fine, irregular wrinkles (fig. IC) plants. The male guards the eggs, of which that are covered by a thin layer of unknown there may be more than 200 (Richter, 1973). substance, which also may have adhesive Eggs are slightly larger than those of Poly- properties inasmuch as smaller bits of plants centropsis, measuring 1.7-1.8 ,um in diame- or mud stick to it. ter. Their vegetal pole sits on a peduncle of AFRONANDUS SHELJUHZKOI: The breeding fibers -0.5 mm long that spread out distally behavior of this rare West African nandid is and adhere to the substrate (fig. 2E). The pe- unknown, although some cursory remarks by duncle originates from an area at the vegetal Scheel (1964a) point to the existence of male pole smaller than that area in Polycentropsis, parental care. Ovaries of one female con- but the zona radiata shows a similar honey- 4 AMERICAN MUSEUM NOVITATES NO. 3195

Fig. 1. A-D. Nandus nandus. SEM photograph of freshly spawned egg. A, Egg attached to Riccia moss, scale bar 200 ,um; B, view of animal pole, scale bar 200 ,um; C, micropylar region at higher magnification, scale bar 20 jLm; D, close-up of the micropyle, scale bar 4 ,um. E-H. Afronandus sheljuhzkoi. SEM photograph of ovarian egg. E, view of vegetal pole with stalk of fibers, scale bar 200 jim; F, region of zona radiata from which fibers originate, scale bar 20 jim; G, animal pole with micropyle, scale bar 50 jim; H, close-up of the micropyle, scale bar 2 jim. 1997 BRITZ: NANDID AND BADID FISHES 5

Fig. 2. A-D. Polycentropsis abbreviata. SEM photograph of freshly spawned egg. A, Egg attached to underside of Nymphaea leaf, scale bar 400 pm; B, region of zona radiata from which adhesive fibers originate, scale bar 10 ,um; C, animal pole with micropyle, scale bar 40 ,um; D, close-up of oval micropyle, scale bar 20 pLm. E-H. Monocirrhus polyacanthus. SEM photograph of freshly spawned egg. E, view of vegetal pole of egg that has been removed from leaf, scale bar 400 ,um; F, region of zona radiata from which adhesive fibers originate (compare 2B), scale bar 10 ,um; G, animal pole with micropyle, scale bar 40 ,um; H, close-up of oval micropyle, scale bar 20 j,m; 6 AMERICAN MUSEUM NOVITATES NO. 3195 combed ultrastructure (cf. fig. 2B, F). As in cells confined to the ventral part of the yolk Polycentropsis, narrow ridges extend radially sac (fig. 4C, D). from the oval micropyle, which measures 3 POLYCENTROPSIS ABBREVIATA: Eggs of this ,um in length and 2 ,um in width (fig. 2G, African nandid hatch after 120 hours at 27°C H). (Rucks, 1992, personal commun.). Larvae adhere to leaves with the aid of a cement EGG STRUCTURE IN THE BADIDAE gland on their head. With SEM, this gland BADIS BADIS: The eggs of B. badis adhere can be recognized as a humplike structure on to the substrate and are guarded and fanned the dorsal area of the head of newly hatched by the male (Barlow et al., 1968; Richter, larvae (fig. 4E, F). Between the epidermal 1981; personal obs.). They have a diameter cells, the gland has numerous openings (one of -0.8 mm. Observations made with SEM marked with an arrow) through which the confirm the results of Barlow et al. (1968). glutinous substance is released (fig. 4G). Un- A baglike sheath completely surrounds the fortunately no PAS preparation could be car- egg (fig. 3A, B). The brim of this sheath ad- ried out. hered to the underlying substrate (fig. 3B, C). MONOCIRRHUS POLYACANTHUS: Eggs of the The sheath consists of a very dense network South American leaffish M. polyacanthus of extremely thin fibers, and its adhesiveness hatch after -72 hours (Richter, 1973; Rucks, is affected by the individual fibers that an- personal commun.) and, as in Polycentropsis, chored that area to the substrate (fig. 3C). larvae use their cement glands to adhere to Figure 3A shows a clutch of eggs of which leaves, where they are guarded by the male the sheaths of two (marked with arrows) (Richter, 1973). The cement gland is situated were torn to reveal the egg inside. The egg on top of the head (fig. 4H) and strongly re- surface is irregularly wrinkled (fig. 3D). The sembles the gland of Polycentropsis. circular micropyle, 2.5-3 ,um wide, lies in a No data on Afronandus were available. micropylar pit 15-20 ,um in diameter (fig. 3D). Eggs of Badis that had been imported CEMENT GLANDS OF LARVAL BADIDAE from another location exhibited a quite different surface structure surrounding the mi- BADIS BADIS: Eggs hatch after 2 days, and cropyle (fig. 3E). The wrinkles were confined larvae adhere to the substrate, where they are to a small, distinct area close to the micro- guarded and fanned by the male (Barlow, pyle, and the remaining surface appeared to 1964; Barlow et al., 1968; personal obs.). be smooth (fig. 3E). Adhesion is accomplished with the aid of a cement gland situated at the anteroventral tip of CEMENT GLANDS OF LARVAL NANDIDAE the larval yolk sac. This cement gland was described by Barlow et al. (1968) on the ba- NANDUS NANDUS: Eggs of the Southeast sis of light microscopy, but SEM provided Asian N. nandus hatch within 30-36 hours further insight into its structure. Figure 5A after spawning (Parameshwaran et al., 1971; depicts a ventral view of a larval B. badis. personal obs.). Larvae can be seen attached The adhesive gland can be recognized as an to different kinds of substrate. This is also area on the anterior part of the yolk sac true for N. nebulosus (Rucks, 1996), but where the epidermis possesses a different there is no information for N. oxyrhynchus. sculpturing (fig. SA, B). A roll-like elevation Figure 4A shows an SEM photograph of a marks the anterior border of the gland. Here, larval N. nandus at 2.5 days after hatching. many openings between the epidermal cells Cement cells can be recognized as small pap- can be recognized where the glutinous sub- illalike protuberances that bulge from the stance is released (fig. 5B). Posterior to the epidermis on the ventral side of the yolk sac roll, the epidermis is smoother, but a consid- (fig. 4A, B). A glutinous substance is re- erable number of openings are nevertheless leased through small apertures between the present. epidermal cells (fig. 4B). PAS staining con- The differential staining of PAS prepara- firmed the SEM observation that the cement tion reveals that the gland consists of a con- gland actually consists of scattered single siderable number of individual cells (fig. 1997 BRITZ: NANDID AND BADID FISHES 7

Fig. 3. Badis badis. SEM photograph of freshly spawned eggs. A, Egg clutch with adhesive sheathes of two eggs (marked by arrows) opened to reveal egg inside, scale bar 1 mm; B, egg with covering sheath, scale bar 200 urm; C, periphery of sheath with individual fibers anchored to substrate, scale bar 10 urm; D, micropylar region with well-developed wrinkles, scale bar 10 ,um; E, micropylar region of egg of Badis from a different locality, wrinkles confined to area around micropyle, scale bar 10 pum.

5C, D). Cell density is highest at the anterior whereas in the posterior part they appear to roll-like elevation of the gland and decreas- be single layered with each cell having one es toward its posterior end (fig. 5C). At the opening. The delineation of the three-di- roll-like elevation the cement cells are ar- mensional arrangement of the adhesive ranged in multiple layers and possibly have cells, however, requires further histological collective openings through the epidermis, investigation. 8 AMERICAN MUSEUM NOVITATES NO. 3195

C . ~~~~~~~~~~~~~~~~~O

Da Fig. 4. Cement gland of larval Nandidae. A-D. Nandus nandus 2.5 days postspawning. A, Ventral view of larva, arrows mark bulges of individual cement cells, scale bar 100 ,um; B, closeup of epidermis of yolk sac, cement cells indicated as bulges of the epidermis (marked by arrows) with an aperture at the tip of each protuberance, scale bar 20 ,um; C, ventral and D, lateral view of PAS-stained larva, individual cement cells marked by arrows, scale bars 100 pum. E-G. Polycentropsis abbreviata 5 days postspawning. E, head of larva with humplike cement gland (marked by arrow) above eyes, scale bar 200 [Lm; F, G, cement gland at higher magnification, showing irregular openings (marked by arrows) 1997 BRITZ: NANDID AND BADID FISHES 9

Fig. 5. Cement gland of larval Badis badis -3 days postspawning. A, ventral view of larval yolk sac, arrow points to adhesive thread released from openings between epidermal cells, scale bar 100 ixm; B, anterior part of cement gland at higher magnification, with numerous openings through which adhesive substance is released, scale bar 20 ,um; C, ventral and D, lateral view of PAS-stained larva, scale bars 100 jxm.

DISCUSSION lationships of its five genera were available com- With the exclusion of the genus Badis by (Lundberg, 1993). Liem, in a personal Barlow et al. (1968) and the genus Pristo- munication that was cited by Cracraft (1974), lepis by Liem (1970) and Liem and Green- claimed that the group Polycentrus + Mon- wood (1981), the family Nandidae sensu ocirrhus is the sister-group of Polycentrop- Liem (1970) comprises only the Southeast sis, but he provided no supporting evidence Asian genus Nandus, the two West African for this hypothesis. genera Afronandus and Polycentropsis, and The present SEM study of egg and larval the two South American genera Polycentrus morphology, which includes four nandid spe- and Monocirrhus. Even in this restricted cies, demonstrates that the egg and larval state, however, no synapomorphies were pro- characters reported for Polycentrus (Barlow posed to support the monophyly of the Nan- et al., 1968) are not uniformly expressed didae and no hypotheses about the interre- within the Nandidae (see table 1). Nandus

between epidermal cells, scale bar in F 40 ,um, in G 4 jim. H, Monocirrhus polyacanthus 9 days postspawning. Head region of larva with humplike cement gland (marked by arrow) above eyes, scale bar 400 ,um. "0 LZ :3 0) - C.) 0 00 z 0 .0 0) 0) ;~ 3 lp 0 z cX c _ J z l 00 a -CAD .- IC I aC. 0 z - C," 0) D :4. c 0 . - ) I" 12 <~D

1 ie Po .o I r D 3 0) "0 3 I n

CO)~~~~~~~~~C n tDt00D >to E t

-o ". 0 CO 0)-u 0 0) 0. ".&. . CO > "3 -.%.".Z. ."..CO

. co CZ 0) 0 Z:: *5-a E C.) . &- . E 0) 0) to C. _ .~ I CZ It > 0) (A o~ -a 0t .0 0 cl l l C.- ~< C-- C, PC5-

PC 00~~~~ Z XZ COCO o < Z 0 00tD0°3

"0 "0 0 0 o- 5- CZ 0) .0 0 CtO 11 <0) 0 cI 0) 0001.1 0 E c o t tu U "t - 00 T* = c . l.4

aE .: - 0) OON00 0 C

E A o nX E C.) ~~ ~ ~ ~ ~ ~ ct tD3t 3t E .t CZ t.0D t taC~ D c 3 o Dt ~ "I O a3_tI 1997 BRITZ: NANDID AND BADID FISHES I1I differs from the African and South American and has not been reported for any other per- Nandidae in several of these characters, an comorph. This unique character is considered observation that correlates with some strik- a synapomorphy of these three genera, thus ing differences between the two groups in the lending support to Liem's belief that Poly- osteological features studied by Liem (1970). centropsis is more closely related to Poly- centrus and Monocirrhus than to other Nan- PHYLOGENETIC RELATIONSHIPS AMONG didae. It is unlikely that the lack of ridges in NANDID GENERA the eggs of Afronandus is an artifact due to the study of unfertilized ovarian eggs be- POSITION OF ADHESIVE FILAMENTS: There is cause several investigations have demon- a striking similarity between African and strated that specific surface structures are al- South American Nandidae in the location of ready present in the ripe ovarian eggs of adhesive filaments at the vegetal egg pole bony fishes (Stehr and Hawkes, 1983; Mooi, (figs. lE, 2A, E); in contrast, adhesive 1990; Mooi et al., 1990; Britz et al., 1995). threads are developed at the animal pole in CEMENT GLANDS: In Polycentropsis (fig. Nandus (fig. lB, C). 4E, F), Polycentrus (Barlow et al., 1968), In an attempt to polarize this character, and Monocirrhus (fig. 4H) the cement gland searching among putatively related taxa is a multicellular organ on top of the head. proved to be of little help. Although badid In Nandus, it is represented by scattered in- eggs also adhere to the substrate, their ad- dividual cement cells that are confined to the hesive filaments do not attach to the egg's ventral and lateral parts of the yolk sac (fig. surface at all. Anabantoidei primitively pos- 4A-D). sess nonadhesive eggs that either float on the Unfortunately, only limited data from oth- water surface, are guarded in nests of foam, er percomorphs are available for comparison. or are orally incubated (Forselius, 1957; In a remarkable pioneering study of the mor- Breder and Rosen, 1966; Vierke, 1975, 1978, phology and evolution of cement glands in 1991a; Britz, 1995; Britz et al., 1995). Eggs cichlids, Peters and Berns (1982) investigat- of channids similarly float at the water sur- ed these organs with both SEM and histo- face and are not adhesive (Willey, 1910; chemical PAS staining techniques. In all the Armbrust, 1963, 1967; Yapchiongco, 1963; cichlids that have been examined, this struc- Vierke, 1991b). ture consists of three pairs of glands, one sit- SEM data from other percomorph families uated between the eyes and the other two on are scarce, and wide-ranging comparisons top of the head (Jones, 1937; llg, 1952; Ben- are therefore impossible at the present. How- nemann and Pietzsch-Rohrschneider, 1978; ever, adhesive filaments on the vegetal egg Peters and Berns, 1982; Hamlett, 1990). In pole have been reported for pseudoplesiopine addition to their larger number, cichlid ce- pseudochromoids (Mooi, 1990) and for some ment glands differ in structure from those de- cichlids (Wickler 1956a, 1956b; Stiassny and scribed for the three Afro-American Nandi- Mezey, 1993). Neither group is closely re- dae. Cichlid cement glands have only one lated to the Nandidae. wide, central, apical opening through which Based on this evidence, the presence of the adhesive substance is released, whereas adhesive filaments at the vegetal pole of the the cement gland of Polycentropsis and Mon- egg in Afronandus, Polycentropsis, Polycen- ocirrhus (and Polycentrus ?) has numerous trus, and Monocirrhus is interpreted here as openings. Although data about larval cement a synapomorphy of the African-South Amer- glands in other percomorphs are rare, some ican Nandidae. information can be gathered from Ilg (1952). RADIAL RIDGES RUNNING FROM THE MICRO- She found multicellular glands with a central PYLE: The characteristic pattern of ridges on opening not only in cichlids but also in the the micropylar region of the eggs of Poly- Ambassidae, in which only one pair of centropsis (fig. 2C, D), Polycentrus (see Bar- glands develops. low et al., 1968: 442 and pl. Id), and Mon- Scattered individual cement cells have ocirrhus (fig. 2G, H) is not present in Nandus been described in the Anabantoidei Pseudos- (fig. iB, C, D) or Afronandus (fig. 1G, H) phromenus cupanus (Jones, 1940; Padman- 12 AMERICAN MUSEUM NOVITATES NO. 3195 abhan, 1955), Betta splendens, Trichogaster mer, 1982). Among percomorph groups pos- trichopterus, and Macropodus opercularis sibly related to Nandidae, parental care oc- (Ilg, 1952), and Ctenopoma damasi and C. curs in the majority of Anabantoidei (For- muriei (Morike, 1977). These cells are scat- selius, 1957; Breder and Rosen, 1966; Vier- tered in the epidermis of the anterior part of ke, 1975, 1978, 1991a; Cambray, 1990, in the body and do not form multicellular press; Britz, 1995), in the channids (Willey, glands as in cichlids or ambassids. A supple- 1910; Armbrust, 1963, 1967; Yapchiongco, mentary study by the author has confirmed 1963; Ng and Lim, 1990; Vierke, 1991b), the presence of such scattered individual ce- and in the Badidae (Barlow et al., 1968). ment cells for the Anabantoidei Anabas tes- However, these taxa's reproductive styles tudineus, Ctenopoma weeksii, Belontia sig- show numerous differences and are hardly nata, Colisa labiosa, Trichogaster trichop- comparable to those of the genera Polycen- terus, Macropodus concolor, Pseudosphro- tropsis, Polycentrus, and Monocirrhus. The menus dayi, Parosphromenus paludicola, Anabantoidei (Forselius, 1957; Vierke, 1975; Trichopsis vittatus, and Betta imbellis, in Cambray, 1990, in press; Britz, 1995), the which the cells are scattered in group-specific Channidae (Yapchiongco, 1963; Vierke, patterns over the yolk sac, head, and anterior 1991b), and the Badidae (Barlow et al., trunk region. These scattered cement cells in 1968) release eggs and sperm during a well- Anabantoidei resemble the condition in Nan- developed spawning embrace (fig. 6A, B, C), dus, but because of the lack of comparative a behavior that seems to be present also in data in other percomorphs, little can be said Nandus (Parameshwaran et al., 1971; Rucks, about their phylogenetic significance. Their 1973, 1996). This spawning embrace is en- occurrence in centrarchid larvae (Ilg, 1952), tirely lacking in Polycentropsis, Polycentrus, however, may indicate that this is a wide- and Monocirrhus. In these three Nandidae, spread character. the female assumes an upside-down position The only other percomorph group with lar- and attaches eggs to the substrate while the val cement glands that has been studied to a male hovers close by in a normal position significant extent is the Badidae (Barlow et (fig. 6D, E). To fertilize the eggs, the male al., 1968; present study). Though also mul- remains in a normal position or bends side- ticellular, the badid cement gland is situated ways, releasing his sperm and washing them at the tip of the yolk sac (fig. 5A-D). Thus, to the eggs with the help of fin movements the type of cement gland of larval Polycen- (for Polycentrus: Rucks, 1960; Barlow, tropsis, Polycentrus, and Monocirrhus ap- 1967; Zukal, 1971; for Monocirrhus: Richter, pears to be unique among percomorphs and 1973; for Polycentropsis: Rucks, 1992). This can be considered another synapomorphy of special spawning procedure is tentatively in- this assemblage. Unfortunately, nothing is terpreted as another synapomorphy of the known about cement glands in Afronandus. genera Polycentropsis, Polycentrus, and It can be summarized that the investigation Monocirrhus. of breeding behavior, egg structure, and larval morphology has not yielded a single PHYLOGENETIC RELATIONSHIPS BETWEEN THE character common to all Nandidae that could NANDIDAE AND BADIDAE be interpreted as a synapomorphy of this family. However, the genera Polycentropsis, Barlow et al. (1968) justified the exclusion Polycentrus, and Monocirrhus showed great of Badis from Nandidae mainly on the basis similarities in three character complexes that of differences between Badis and Polycen- were interpreted as synapomorphies support- trus. It was shown above that reproductive ing their monophyly. Afronandus is tenta- ethology and egg and larval morphology tively assigned to this group until more in- vary among the five nandid genera. However, formation about its breeding behavior and Badis differs significantly from the leaffishes larval structure becomes available. in all these characters. BREEDING BEHAVIOR: Parental care exists Eggs of Badis are completely surrounded in a variety of percomorphs; usually the male by a sheath of fibers without actually being guards the eggs and often the offspring (Blu- attached to them (fig. 3A, B), a situation not 1997 BRITZ: NANDID AND BADID FISHES 13

al., 1968; Lauder and Liem, 1983). However, a similar spawning embrace occurs also in channids (fig. 6C; Yapchiongco, 1963; Ng and Lim, 1990; Vierke, 1991b) and seems to be present in Nandus (Parameshwaran et al., 1971; Rucks, 1973, 1996). Accordingly, Britz (1995) concluded that such a spawning embrace should be considered a plesio- morphic character at the level of Badidae and B Anabantoidei and does not indicate a sister- group relationship between both taxa. Unfor- tunately, the data in this study do not further clarify the phylogenetic relationships of Bad- idae. No synapomorphies with Nandidae could be identified. Future re- C study of the productive behavior and egg and larval structure of Pristolepis should yield new data and lead to a better understanding of the phylogenetic relationships of Badis. Moreover, the difference in egg surface structures of the Badis collected in two different localities in the present study might prove to be evidence for the existence of additional species of Badis and demonstrates the need for a thor- ough revision of this genus.

BIOGEOGRAPHIC IMPLICATIONS Recently Lundberg (1993) reviewed Fig. 6. Spawning posture in A, Badis badis, Afri- from Britz (1995) after photograph in Richter can-South American freshwater fish clades (1981); B, Anabas testudineus, from Britz (1995); and discussed different models to explain C, Channa bleheri, from Britz (1995) after pho- their recent geographic distribution. As a pre- tograph in Vierke (1991b), sexes not given; D, requisite for building biogeographic hypoth- Polycentrus schomburgkii, after photograph in eses, the phylogeny of the taxa involved has Zukal (1971); E, Monocirrhus polyacanthus, after to be satisfactorily resolved (Nelson and photograph in Richter (1973). Male stippled, fe- Platnick, 1981). Lundberg (1993) pointed out male white. that there are no well-supported hypotheses of relationships among the different nandid genera, apart from Liem's belief (cited in present in any nandid. The micropylar region Cracraft, 1974) that Polycentropsis is the sis- has neither ridges as in Polycentropsis, Po- ter group of Polycentrus + Monocirrhus, lycentrus, and Monocirrhus nor the circular which has found supportive evidence in this area of carpetlike fibers described for N. nan- study. As shown above, a monophyletic dus. group of the three genera Polycentropsis + The cement gland of Badis is multicellular Polycentrus + Monocirrhus can be hypoth- but situated at the tip of the yolk sac. Badis esized on the basis of several synapomor- shows an unusual spawning embrace (fig. phies, and therefore at least one African- 6A; Barlow, 1962; Barlow et al., 1968) that South American relationship has been estab- is not present in Polycentropsis, Polycentrus, lished. and Monocirrhus (fig. 6D, E) but shared with Lundberg (1993) stressed that an origin of Anabantoidei (fig. 6B). This was one of the Nandidae in the early Cretaceous before sep- reasons why Badidae and Anabantoidei were aration of India from Africa or in the later hypothesized to be sister groups (Barlow et Cretaceous before separation of Africa and 14 AMERICAN MUSEUM NOVITATES NO. 3195

South America would "imply unacceptably For the reasons given, it is hypothesized ancient ages of origin for this family and its here that the distribution of the African- basic subgroups" (p. 187). For this reason, South American Nandidae is best explained Lundberg (1993: 187-188) considered by a simple drift vicariance model and their "postdrift dispersals of nandids through ma- presence on the African-South American rine habitats ... more likely." The data pre- landmass before the final separation of the sented here, however, are incompatible with two continents. This explanation suggests a this hypothesis. much older age of origin for the African- Observations by Scheel (1964b) and South American Nandidae than that assumed Rucks (1992) clearly show that Polycentrop- by Lundberg (1993). According to the recent sis is a species that occurs in very soft, acidic overview of the fossil record of teleosts (Pat- waters. Moreover, Polycentropsis, Monocir- the earliest rhus, and terson, 1993a, 1993b), Percomor- Polycentrus breed in captivity only pha or Perciformes incertae sedis are found when they are maintained under such con- in the late Cretaceous. At the ditions (Scheel, 1964b; moment, the Richter, 1973; Rucks, taxon with the lowest to 1992). The most parsimonious conclusion is monophyletic rank that the which the Nandidae can be assigned without most recent common ancestor of this doubt is the monophyletic group already exhibited simi- Percomorpha, although Nandi- lar ecological preferences. Thus, the ecolog- dae have typically been classified as Percoid- ical environment in which the recent Afri- ei since Regan (1913). Johnson (1984), how- can-South American Nandidae occur renders ever, pointed out that Regan's Percoidei is unlikely any dispersal through marine habi- not supported by synapomorphies. The oc- tats after the separation of the African and currence of African-South American Nan- South American landmasses. didae in the late Cretaceous thus appears not The genera Polycentropsis, Polycentrus, so unlikely as claimed by Lundberg (1993). and Monocirrhus show similarities in breed- In addition to osteological differences cit- ing behavior, egg structure, and larval mor- ed by Liem (1970), this study has presented phology that are assumed to have already differences in reproductive biology and egg been present in the last common ancestor of and larval structure between African-South these three genera. It seems unlikely that an American and Asian Nandidae. Therefore, ancestor with these features would have been inasmuch as the monophyly of Nandidae able to survive and reproduce under saltwa- sensu Liem (1970) has not been satisfactorily ter conditions-an additional argument demonstrated, it seems premature to specu- against the postdrift dispersal hypothesis of late about models that try to explain nandid Lundberg (1993). distribution on three continents.

REFERENCES Armbrust, W. leaf fish, Polycentrus schomburgkii, 1963. Zucht des Schlangenkopfes Ophice- with an account of recurring pseudo- phalus obscurus. Aquarien. Terr. Z. 16: female behavior. Am. Midl. Nat. 78: 298-301. 215-234. 1967. Nachzucht beim Schlangenkopf (Ophi- Barlow, G. K. F and W. cephalus africanus). Ibid. 20: 367-368. W., Liem, Wickler Barlow, G. A. 1968. Badidae, a new fish family-behavior- 1962. Ethology of the Asian teleost Badis al, osteological, and developmental ev- badis. IV. Sexual behavior. Copeia idence. J. Zool. (London) 156: 415- 1962: 346-360. 447. 1964. Ethology of the Asian teleost Badis Bennemann, R., and I. Pietzsch-Rohrschneider badis. V. Dynamics of fanning and oth- 1978. The morphology of the cement gland er parental activities, with comments apparatus of larval Pterophyllum sca- on the behavior of the larvae and post- lare Cuv. & Val. (Cichlidae, Teleostei). larvae. Z. Tierpsychol. 21: 99-123. Cell Tissue Res. 193: 491-501. Barlow, G. W. Berg, L. S. 1967. Social behavior of a South American 1958. System der rezenten und fossilen Fis- 1997 BRITZ: NANDID AND BADID FISHES 15

chartigen und Fische. Berlin: VEB Ver- Greenwood, P. H., D. E. Rosen, S. H. Weitzman, lag. and G. S. Myers Berra, T M. 1966. Phyletic studies of teleostean fishes 1981. An atlas of distribution of the fresh- with a provisional classification of liv- water families of the world. Lincoln: ing forms. Bull. Am. Mus. Nat. Hist. Univ. Nebraska Press. 131: 339-456. Blumer, L. S. Gunther, A. 1982. A bibliography and categorisation of 1861. Catalogue of the acanthopterygian fish- bony fishes exhibiting parental care. es in the British Museum. London: Zool. J. Linn. Soc. 76: 1-22. Taylor and Francis. Boulenger, G. A. Hamlett, W. C. 1904. Teleostei. In S. F Harmer and A. E. 1990. Subcellular structure and function of Shipley (eds.), Fishes, Ascidians, etc., cement secreting glands in Pterophyl- pp. 541-727. London: Macmillan. lum scalare, a transient larval special- Breder, C. M., and D. E. Rosen ization. J. Submicrosc. Cytol. Pathol. 1966. Modes of 22: 27-37. reproduction in fishes. New L. York: The Natural History Press. llg, Britz, R. 1952. Uber larvale Haftorgane bei Teleos- teern. Zool. Jahrb. Abt. Anat. 72: 577- 1995. Zur phylogenetischen Systematik der 600. Anabantoidei (Teleostei, Percomorpha) Jones, S. unter besonderer Berucksichtigung der 1937. On the origin and development of the Stellung des Genus Luciocephalus. cement glands in Etroplus maculatus Morphologische und ethologische Un- (Bloch). Proc. Indian Acad. Sci., B 6: tersuchungen. Ph.D. diss., Tubingen 79-90. Univ., 125 pp. 1940. Notes on the breeding habits and early Britz, R., M. Kokoscha, and R. Riehl development of Macropodus cupanus 1995. The anabantoid genera Ctenops, Luci- (Cuv. & Val.), with special reference to ocephalus, Parasphaerichthys, and cement glands of the early larvae. Rec. Sphaerichthys (Teleostei: Perciformes) Indian Mus. 42: 269-276. as a monophyletic group: evidence Johnson, G. D. from egg surface structure and repro- 1984. Percoidei: development and relation- ductive behavior. Japan. J. Ichthyol. 42: ships. In H. G. Moser et al. (eds.), On- 71-79. togeny and systematics of fishes, pp. Cambray, J. A. 464-498. Lawrence, KS: Allen Press. 1990. Early ontogeny and notes on breeding Jordan, D. S. behavior habitat preference and conser- 1923. A classification of fishes, including vation of the Cape kurper, Sandelia ca- families and genera as far as known. pensis. Ann. Cape Prov. Mus. Nat. Stanford Univ. Publ. Biol. Sci. 3: 77- Hist. 18: 159-182. 243. In press. The spawning behavior and conser- Laale, H. W. vation of the endangered eastern Cape 1980. The perivitelline space and egg enve- Rocky (Sandelia bainsii; Anabantidae), lopes of bony fishes: a review. Copeia South Africa. Environ. Biol. Fishes. 1980: 210-226. Cracraft, J. Lauder, G. V., and K. F Liem 1974. Continental drift and vertebrate 1983. The evolution and interrelationships of distri- the bution. Annu. Rev. Ecol. Syst. 5: 215- actinopterygian fishes. Bull. Mus. Comp. Zool. 150: 95-197. 261. Liem, K. F Forselius, S. 1970. 1957. Comparative functional anatomy of the Studies of anabantid fishes I-III. Zool. Nandidae. Fieldiana Zool. 56: 1-166. Bijdr. Uppsala. 32: 93-597. Liem, K. F, and P H. Greenwood Gosline, W A. 1981. A functional approach to the phylogeny 1968. The suborders of perciform fishes. of the pharyngognath teleosts. Am. Proc. U.S. Natl. Mus. 124: 1-78. Zool. 21: 83-101. 1971. Functional morphology and classifica- Lundberg, J. G. tion of teleostean fishes. Honolulu: 1993. African-South American freshwater Univ. Press of Hawaii. fish clades and continental drift: prob- 16 AMERICAN MUSEUM NOVITATES NO. 3195

lems with a paradigm. In P Goldblatt Patterson, C. (ed.), Biological relations between Af- 1993a. An overview of the early fossil record rica and South America: 156-199. New of acanthomorphs. Bull. Mar. Sci. 52: Haven, CT: Yale Univ. Press. 29-59. M6rike, D. 1993b. Osteichthyes: Teleostei. In M. J. Ben- 1977. Vergleichende Untersuchungen zur ton (ed.), The fossil record 2, pp. 621- Ethologie zweier Labyrinthfischarten, 656. London: Chapman and Hall. Ctenopoma muriei (Boulenger 1906) Peters, H. M., and S. Berns und Ctenopoma damasi (Poll 1939) 1982. Die Maulbrutpflege der Cichliden. Un- (Anabantoidei, Pisces). Ph.D. diss., Tu- tersuchungen eines Verhaltensmusters. bingen Univ. 178 pp. Z. zool. Syst. Evolutionsforsch. 20: 18- Mooi, R. D. 52. 1990. Egg surface morphology of pseudoch- Regan, C. T. romoids (Perciformes: Percoidei), with 1913. The classification of the percoid fishes. comments on its phylogenetic implica- Ann. Mag. Nat. Hist. 12: 111-145. tions. Copeia 1990: 455-475. Richter, H.-J. Mooi, R. D., R. Winterbottom, and M. Burridge 1973. DDR-Erstnachzucht von Blattfischen. 1990. Egg surface morphology, and evolution Aquarien Terr. 20: 220-225. in the Congrogadinae (Pisces: Percifor- 1981. Das "Fisch-Chamaleon": Der Blau- mes: Pseudochromidae). Can. J. Zool. barsch. Aquarien Mag. 446-450. 68: 923-934. Riehl, R. G. 1991. Die Struktur der Oocyten und Eihullen Nelson, J. oviparer Knochenfische-eine Uber- 1969. Gill arches and the phylogeny of fishes, sicht. Acta biol. Benrodis. 3: 27-65. with notes on the classification of ver- Rosen, D. E., and C. Patterson tebrates. Bull. Am. Mus. Nat. Hist. 1990. On Muller's and Cuvier's concepts of 141: 475-552. pharyngognath and labyrinth fishes and Nelson, G. J., and N. Platnick the classification of percomorph fishes, 1981. Systematics and biogeography: cladis- with an atlas of percomorph dorsal gill tics and vicariance. New York: Colum- arches. Am. Mus. Novitates 2983: 57 bia Univ. Press. pp- Nelson, J. S. Rucks, R. 1994. Fishes of the world. (3rd ed.) New 1960. Polycentrus schomburgki(i). Aquarien York: Wiley. Terr. Z. 13: 105-107. Nieuwenhuizen, A. van den 1973. Zum ersten Mal (?!) gezuchtet: Nandus 1959. Badis badis III. Aquarien Terr. Z. 12: nandus-Der Nander. Aquarien Terr. Z. 98-101. 26: 158-160. Ng, P K. L., and K. K. P Lim 1992. Polycentropsis abbreviata. Beobach- 1990. Snakeheads (Pisces: Channidae): natu- tungen und Erkenntnisse. Aquarien ral history, biology and economic im- Terr. Z. 45: 420-423. portance. In L. M. Chou and P. K. L. 1996. Nandus nebulosus-ein Nanderbarsch Ng. (eds.) Essays in zoology, pp. 127- wird vorgestellt. Aquarien Terr. Z. 49: 152. Singapore: Dept. Zool., Natl. 554-556. Univ. Scheel, J. J. Ng, H. H., C. Vidthayanon, and P. K. L. Ng 1 964a. Afronandus sheljuzhkoi Meinken, a rare 1996. Nandus oxyrhynchus, a new species of African nandid. The Aquarium (Norris- leaf fish (Teleostei: Nandidae) from the town) 33(7): 6-9. Mekong basin. Raffles Bull. Zool. 44: 1964b. Polycentropsis abbreviata Boulenger 11-19. 1901. The Aquarium (Norristown) Padmanabhan, K. G. 33(8): 8-10. 1955. Breeding habits and early embryology Stehr, C. M., and J. W. Hawkes of Macropodus cupanus (Cuv. & Val.) 1983. The development of the hexagonally Bull. C.R.I. Trivandrum. 4: 1-45. structured egg envelope of the C-O sole Parameshwaran, S., S. Radhakrishnan, and C. Sel- (Pleurichthys coenosus). J. Morphol. varaj 178: 267-284. 1971. Some observations on biology and life- Stiassny, M. L. J., and J. G. Mezey history of Nandus nandus (Hamilton). 1993. Egg attachment systems in the family Proc. Indian Acad. Sci. 73B: 132-147. Cichlidae (Perciformes: Labroidei), 1997 BRITZ: NANDID AND BADID FISHES 17

with some comments on their signifi- Wickler, W. cance for phylogenetic studies. Am. 1956a. Der Haftapparat einiger Cichlideneier. Mus. Novitates 3058: 11 pp. Z. Zellforsch. 45: 304-327. Vierke, J. 1956b. Unterschiede zwischen den Cichliden- 1975. Beitrage zur Ethologie und Phylogenie Gattungen, speziell Geophagus und der Familie Belontiidae. Z. Tierpsy- Biotodoma, im Haftapparat der Eier. chol. 38: 163-199. Naturwiss. 14: 333-334. 1978. Labyrinthfische und verwandte Arten. Willey, A. Wuppertal-Elberfeld: Engelbert Pfriem 1910. Observations on the nests, eggs and larvae of Ophiocephalus striatus. Spolia Verlag. Zeylan. 6: 108-123. 1991 a. Brutpflegestrategien bei Belontiiden Yapchiongco, J. V. (Pisces, Anabantoidei). Bonn. zool. 1963. Spawning habits of the murrell or Beitr. 42: 299-324. snakehead fish, Ophicephalus striatus 199 lb. Der'- Regenbogen-Channa. Haltung und Bloch. Copeia 1963: 700-702. Zucht von Channa bleheri. Das Aquar- Zukal, R. ium 25(266): 15-19. 1971. Bei Nacht schleicht sich der Rauber aus

Weber, M., and L. F DeBeaufort der Hohle ... Der Vielstachler Poly- 1936. The fishes of the Indo-Australian Ar- centrus schomburgki(i). Aquarien Mag. chipelago. Vol. 7. Leiden: E. J. Brill. 5: 32-35.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last five years are available at World Wide Web site http://nimidi.amnh.org. Or address mail orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y 10024. TEL: (212) 769-5545. FAX: (212) 769-5009. E-MAIL: [email protected]

) This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).