460 



ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 



Zeiformes as a potential sister group to the tetraodontiform 

 fishes. However, at least 7.eus and Capros have a long notochord 

 tip; Capros has early-forming spinule-like scales; and Capros 

 and Zeus (but not Antigoma) larvae are generally similar to 

 some balistoid larvae in body shape and pigmentation. Thus, 

 based on scanty ELH information, there are some suggestions 

 of support for Rosen's proposal of a zeiform-telraodontiform 

 relationship. 



Present knowledge of ELH characters does not help much in 

 determining the inter-ordinal relationships of the tetraodonti- 

 form fishes. This is partially because there are no unique, derived 

 ELH characters shared by all tetraodontiform fishes (see below). 

 In addition the ELH characters which are shared between te- 

 traodontiforms and either acanthuroids or zeiforms (above) are 

 shared with other groups as well, thus lessening the value of 

 these characters in determining relationships: e.g., similar body 

 shape, pigment, reduced number of vertebrae, early-forming 

 spinule-like scales, and elongate notochord tip are found in var- 

 ious combinations in priacanthids, pomacanthids, callionymids 

 andlophiiform fishes (Leis and Rennis, 1983, and relevant chap- 

 ters in this volume). 



Therefore, one must rely on ideas of inter-ordinal relation- 

 ships based on adults. For the purposes of this analysis, the 

 Acanthuroidei and the Zeiformes are considered as alternative 

 sister groups for the Tetraodontiformes. 



Order Tetraodontiformes 



There are relatively few ELH characters which apply to the 

 Tetraodontiformes as a whole, and fewer still which could be 

 considered derived. The only characters which might be con- 

 sidered derived are the late formation of the caudal fin and the 

 various early-forming scale specializations, and both are found 

 in a few other percoid and non-percoid groups. The dermal sac 

 and some other derived characters are probably derived within 

 the order and are of no use in characterizing the order as a 

 whole. Other tetraodontiform characters which are wide-spread 

 among other fishes are: small mouth, gas bladder present, rel- 

 atively few myomeres and fin rays, large head, no bones of the 

 head with spines, oviparity, basically spherical eggs with un- 

 segmented yolk, and transformation to an unspecialized pelagic 

 juvenile at a small size. Therefore, I could find no uniquely 

 derived ELH characters shared by all members of the order. 



Some features of the adults can be considered paedomorphic: 

 large head, lack of certain structures that simply never form 

 (Fraser-Brunner, 1950), delayed ossification of some bones. 



Intra-ordinal relationships 



As noted above, the Acanthuroidei (consisting of the families 

 Acanthuridae, Zanclidae and Siganidae) and the Zeiformes (in- 

 cluding Caproidae after Rosen, pers. comm.) will be considered 

 as alternative sister groups to the Tetraodontiformes. Therefore, 

 characteristics shared with the early life history stages of the 

 Acanthuroidei, and particularly the Acanthuridae (or alterna- 

 tively with the Zeiformes) will be considered primitive. Char- 

 acteristics of acanthuroid larvae are summarized in Leis and 

 Rennis (1983) and Leis and Richards (this volume). Character- 

 istics of zeiform larvae are summarized in Russell (1976) and 

 Tighe and Keene (this volume). 



Two tetraodontiform families cannot be included for lack of 

 information (Aracanidae and Triodonlidae) and these are not 

 considered further. I don't know how seriously these omissions 

 might bias the results. It is assumed the egg characteristics of 



the triacanthodids (which are unknown) are the same as those 

 of the triacanthids. 



Perhaps surprisingly, the acanthuroid and zeiform character 

 states differ for only three of the characters used in the following 

 analysis. For these three, the difference lies in my inability to 

 assign polarity to the character if the zeiforms were chosen as 

 the outgroup. Thus, it makes no difference to the shape of the 

 resulting phylogeny (but does weaken two of the branch points) 

 if the Zeiformes rather than the Acanthuroidei is chosen as 

 outgroup. 



A discussion of the characters used follows (Table 118): (1) 

 Egg type— Acanthurids (and zeiforms) have pelagic eggs, al- 

 though siganids have demersal eggs. The demersal eggs of te- 

 traodontiform fishes and siganids have no adaptations for being 

 demersal other than stickiness or a mucous mass, and seem 

 relatively unspecialized for being demersal. A pelagic egg is 

 considered primitive. (2) Egg size— Acanthuroid eggs are small 

 (< 1 mm), so eggs larger than 1.4 mm are considered derived. 

 However, zeiform eggs are medium to large (0.95-2.0 mm), so 

 if zeiform fishes are accepted as the outgroup, polarity of this 

 character cannot be determined. (3) Oil droplets in eggs— Acan- 

 thuroid eggs (and zeiform eggs) have one or more oil droplets 

 in the yolk. Lack of oil droplets in eggs is considered derived. 

 (4) Egg shape— An egg that is not spherical is considered derived 

 because acanthuroid eggs (and zeiform eggs) are spherical. (5) 

 Chorion sculpture — Sculpturing on the chorion is considered 

 derived because acanthuroid eggs (and zeiform eggs) are un- 

 sculptured. (6) Incubation period— Acanthuroid eggs hatch in 

 about two days or less, and an incubation time longer than this 

 is considered derived. Because incubation period is tempera- 

 ture-dependent, it is possible that some of the differences noted 

 here are artifacts of the different temperatures at which the eggs 

 were reared. However, insofar as it has been possible to com- 

 pare different taxa reared at similar temperatures, the differences 

 in incubation period noted here seem valid. Incubation times 

 of zeiform eggs are poorly known, but may be up to 13 days for 

 Zeus. Therefore, if zeiform fishes are accepted as the outgroup, 

 polarity of this character cannot be determined. (7) Parental 

 care of eggs— There is no parental care of eggs by fishes with 

 pelagic eggs including zeiforms and acanthurids. Siganids lay 

 demersal eggs but no parental care has been reported. Therefore, 

 lack of parental care is considered primitive. (8) Body shape— 

 Acanthuroid (and zeiform) larvae tend to be cylindrical to some- 

 what compressed at hatching and to be compressed by the time 

 flexion is complete, although they may pass through an early 

 preflexion stage which is more rotund. This developmental pat- 

 tern is considered primitive. Some tetraodontiform larvae are 

 extremely rotund throughout development, but this is largely 

 due to a greatly inflated dermal sac (see character 10). (9) Head 

 and gut development — All balistoid fishes but ostraciids hatch 

 with a cylindrical to compressed body. All of these but mona- 

 canthid Morph AB become deeper-bodied and wider in head 

 and gut by the middle of the preflexion stage and then become 

 compressed by flexion. Morph AB monacanthids never become 

 broad in head and gut. Due to the widespread occurrence of the 

 wide body development mode in the suborder, it is considered 

 primitive. (10) Vesicular dermal sac— Some tetraodontiform 

 larvae have the head and trunk enclosed in a vesicular dermal 

 sac, a condition not found in acanthuroids or zeiforms (a very 

 weakly-developed dermal sac without vesicles similar to the one 

 of yolk-sac balistids is found in acanthurids). This sac and its 

 subdermal space seem to be the source of many of the dermal 



