LEIS: TETRAODONTIFORMES 



463 



TYLER 



WINTERBOTTOM 



I TRIACANTHODIDAE 



2 TRIACANTHIDAE 



3 BALISTIDAE 



4 MONACANTHIDAE 



5 OSTRACIIDAE 



6 TETRAODONTIDAE 



7 DIODONTIDAE 



8 MOLIDAE 



Fig. 253. The two published adult-based phylogenies of tetraodon- 

 tiform fishes which were tested by the ELH-based phylogeny. These 

 phylogenies were modified by omitting the two families which could 

 not be included in the ELH-based phylogeny. After Tyler (1980) and 

 Winterbottom (1974a). Numbers refer to the families listed at bottom. 

 Rosen's unpublished phylogeny is not shown. 



ostraciid relationships (Figs. 252, 253), indicating that further 

 study is required. Winterbottom ( 1 974a) and Tyler ( 1 980) place 

 the ostraciids with the balistoids, a view reinforced by a recent 

 reassessment of their data (Winterbottom and Tyler, 1983). The 

 present phylogeny and Rosen's (pers. comm.), however, place 

 the ostraciids with the tetraodontoids. A relationship between 

 ostraciids and tetraodontoids was suggested by Sakamoto and 

 Suzuki (1978) based on general similarity of larvae. 



The three adult-based phylogenies regard the tetraodontids 

 as the sister group of the diodontids (Fig. 253). This differs 

 significantly from the ELH-based phylogeny (Fig. 252) which 

 regards the tetraodontids as the sister group of all other tetra- 

 odontoids (including ostraciids). The trichotomy between these 

 "other tetraodontoids" in Fig. 252 cannot be resolved at present. 

 Further study is indicated. 



The balistoids (Monacanthidae-Balistidae) branch off in a 

 convincing manner, but not without problems. The phylogeny 

 as depicted in Fig. 252 requires that demersal eggs (1) be in- 

 dependently derived in balistoids and tetraodontids. Although 

 this is quite possible, it brings into question the validity of using 

 demersal eggs as a derived character to define the Balistoidea. 

 Morph C monacanthids lack the preopercular spine cluster (14) 



which characterises all other balistoids. I conclude that this is 

 a secondary loss and that the delayed development of the dorsal 

 fin spine in Morph C is independently derived (thus not indi- 

 cating a relationship with triacanthoids). 



All phylogenies agree on the close relationship of monacan- 

 thids and balistids. Indeed, in the present study (Fig. 252), they 

 were separated by only two ELH characters, (17b) loss of a fin 

 spine, and (7) parental care of eggs, about which there is little 

 information and which is variable in tetraodontids. Although 

 the present phylogeny is nominally consistent with Matsuura's 

 (1979) phylogeny, Winterbottom (1974a) considered monacan- 

 thids and balistids to be subfamilies, and the ELH-based phy- 

 logeny presented here has done little to clarify this conflict. 



There is some indication from ELH characters of divergences 

 within families, but the amount that can be said is severely 

 limited by the small number of taxa for which ELH characters 

 are known. The diodontids seem very conservative but some 

 species of Chilomycterus have a specialized ontogenetic stage 

 between larvae and juveniles (" Lyosphaera"): this supports re- 

 moval of these species to a separate genus (study in progress). 

 Within the ostraciids, the two subfamilies are separated by de- 

 gree of chorion ornamentation, and to a lesser degree by de- 

 velopment at hatching. The specialized "Molacanthus" stage 

 separates Afola and Masturus from Ranzania in the Molidae. 

 Balistids seem very conservative in development. Tetraodontids 

 vary greatly in development at hatching, parental care of eggs, 

 and perhaps in a number of other characters. Too few taxa are 

 known within the triacanthodids and triacanthids for any state- 

 ments to be made here. Monacanthids have the most variation 

 in ELH characters within the order, some of which has already 

 been referred to (Aboussouan and Leis, this volume). There 

 seems to be a great deal of potential in the use of ELH characters 

 for phylogenetic studies in the Monacanthidae, but first, devel- 

 opmental series for more species and genera must be established. 

 I have attempted to use ELH characters independently as a 

 test of phylogenies based on adult characters. Where the two 

 types of phylogenies support each other, confidence in the phy- 

 logeny is increased. Where differences appear, further study, or 

 re-interpretation of existing data is called for to resolve the 

 differences. 



In conclusion, the present classification should be viewed with 

 caution because there are relatively few taxa for which eady life 

 history information is available. Monophyly of the tetraodon- 

 tiform fishes could not be established using ELH characters. 

 The present ELH-based phylogeny and those of Winterbottom 

 (1974) and Rosen (pers. comm.) agree in the creation of a sep- 

 arate suborder for triacanthoid fishes; Tyler (1980) disagrees 

 with this placement. Tyler (1980) and Winterbottom (1974a) 

 agree in placing the Ostraciidae in the Balistoidea, in contrast 

 to inclusion of the Ostraciidae within the Tetraodontoidea as 

 proposed here and by Rosen (pers. comm.). My placement of 

 the Tetraodontidae is in conflict with previous phylogenies based 

 on adult characters. In other areas, the ELH-based phylogeny 

 is in agreement with the three adult-based phylogenies. The 

 different placements of the Tetraodontidae and in particular the 

 Ostraciidae in the present classification warrant further inves- 

 tigation of tetraodontiform interrelationships. 



Department of Ichthyology, The Australian Museum, P.O. 

 Box A285, Sydney, N.S.W., 2000, Australia. 



