14 



D.J. SIEBERT 



(L. Parenti, pers. comm.). Four new species of Sundasalanx are 

 described below, three from the Barito River, Central Borneo, 

 Indonesia and one from the Kapuas river, West Borneo, Indonesia. 

 The new species of Sundasalanx appear to be larger than S. praecox 

 and S. microps but are very small fishes nevertheless, with observed 

 standard lengths not exceeding 30 mm. 



Roberts (1981) considered Sundasalanx to be so unlike other 

 known fishes he erected a new family for it. While recognising its 

 unusual nature Roberts nevertheless felt it is allied with Salangidae 

 (icefishes), hence the name Sundasalanx, and its familial derivative, 

 Sundasalangidae. The following list of features was presented as 

 evidence in support of this conclusion: 1) jaw suspensorium consist- 

 ing of a single cartilaginous element; 2) 4 th hypobranchial element 

 present; 3) pedunculate pectoral fins; 4)scaleless body; 5) absence of 

 symplectic; 6) absence of circumorbital bones; 7) myotomal muscles 

 not meeting at the ventral midline of the body; and 8) distal two-thirds 

 of maxilla curved beneath the head so that its ventral edge is directed 

 medially. Others have regarded Sundasalanx simply as a salangid, 

 arguing that if its relationship is with salangids then recognition of a 

 Sundasalangidae would render Salangidae paraphyletic (Fink, 1984; 

 Begle, 1991), if co-ordinate ranking is maintained in classification. 

 Johnson & Patterson ( 1 996) recently have proposed a classification of 

 the Salmoniformes in which salangids are nested within the Osmeridae. 

 In their classification salangids are given lower rank. They plus 

 Mallotus make up the tribe Salangini. 



All of the features enumerated above as suggestive of a relation- 

 ship with salangids, except possibly the curve of the posterior 

 portion of the maxilla to beneath the head, are simply aspects of a 

 physiognomy that is paedomorphic to an extreme. They are in fact 

 features suggestive of a larval stage of development and are a 

 consequence of a truncated ontogeny. As such they might not each 

 constitute independent evidence of relationship as each might be the 

 result of the same process or event that altered the development of 

 Sundasalanx. Furthermore, they amount to statements of absence, 

 which renders them ineffectual as evidence of relationship in the 

 absence of corroboration from other characters. Different evidence, 

 and a new radical hypothesis of relationship for Sundasalanx are 

 presented below. 



MATERIALS AND METHODS 



Comparative morphometry is presented with reference either to 

 Standard length (SL) or Head length (HL), each measured as 

 recommended in Hubbs & Lagler (1947). Head width (HW) was 

 taken as the width of the widest part of the head. Eye diameter was 

 measured from camera lucida tracings of heads. Counterstained 

 (C&S) materials were prepared following Dingerkus & Uhler ( 1 977). 

 Since ossification in Sundasalanx is slight some materials were 

 cleared and then stained with alizarin only. Some very lightly 

 ossified elements were only apparent in material prepared this way. 

 Whether very lightly ossified elements were obscured by the blue 

 counterstain or dissolved by the acidic alcian blue solution during 

 the counterstaining process was not determined. Institutional abbre- 

 viations follow Leviton et al. (1985). 



Anatomical notes are based on observations of cleared and stained 

 specimens: 17 Sundasalanx malleti sp. nov., from the Barito River; 

 10 Sundasalanx me sops sp. nov., from the Barito River, 3 Sundasalanx 

 platyrhynchus sp. nov., from the Kapuas River; 2 paratypes of 

 Sundasalanx microps Roberts, from the Kapuas River, 1 paratype of 

 S. praecox Roberts, from peninsular Thailand; and 1 paratype of S. 

 megalops sp. nov., from the Barito River. 



Material Examined. Materials of the new species are listed 

 below in the descriptions of new species. 



Sundasalanx microps, paratypes, CAS 44220, 5 of 9 ale, 2 of 7 

 C&S, Indonesia, Kalimantan Barat, Kapuas R. basin, Kapuas R. 

 mainstream at Kampong Nibung, ca 100 km northeast of Selimbau, 

 5-6 Jul 1976, T.Roberts. 



Sundasalanx praecox, paratypes, BMNH 1981.5.19:80-84, 4 ale. 

 ex., 1 C&S, Thailand, Isthmus of Kra, Khlong Falamee, a tributary 

 of Tale Sap, ca. 2 km W of Pak Payoon, 20 Jun 1970, T.Roberts. 



NOTES ON THE ANATOMY OF SUNDASALANX 

 ROBERTS, 1981 



Small size and light ossification make Sundasalanx difficult subjects 

 to study. Observations and interpretations that augment, or differ 

 from, those of Roberts (1981, 1984) and Kottelat (1991) are pre- 

 sented below. Although the relationships of Sundasalanx are 

 discussed later, comparisons are made here to clupeomorph, 

 clupeiform, or clupeid anatomy as an aid in interpretingSwn dasalanx. 

 This is based on the conclusion that Sundasalanx is a clupeid, not a 

 salangoid. 



Sundasalanx does not look like other juvenile or adult clupeids. 

 Rather, their physiognomy is very much like that of a late stage pre- 

 metamorphosis larva. Although there is a large literature concerning 

 the identification of larval clupeomorphs (McGowan & Berry, 1984) 

 surprisingly little has been written about their internal anatomy 

 (O'Connell, 1981 ). An exception is the gas bladder-inner ear-lateral 

 line system, of which detailed anatomical descriptions are available 

 for a number of clupeiforms (Allen et al., 1976; Blaxter & Hunter, 

 1982; O'Connell, 1981; Shardo, 1996). 



Sundasalanx is distinctive. The peculiar structures Roberts (1981) 

 termed parapelvic cartilages (or bones; Fig. 1) are unknown else- 

 where among teleosts. Sundasalanx also possesses a highly derived 

 caudal skeleton (described below; Fig. 7A). A number of other 

 features listed by Roberts, mostly paedomorphic and not unique to 

 Sundasalanx, contribute to the generic diagnosis (Roberts, 1981). 

 The absence of hypobranchials 1-3 but with presence of hypo- 

 branchial 4, a cartilage bar uniting the shoulder girdle across the 

 ventral midline and a rayless pectoral fin supported by a single 

 cartilaginous plate rather than by separate radials are particularly 

 striking among the list. 



Colour Pattern 



Previous descriptions of Sundasalanx have been based on single 



Fig. 1 Pelvic girdle of Sundasalanx malleti (anterior to left); A = 

 parapelvic bones, B = basipterygium, C = I s ' pelvic fin-ray, D = radial, 

 scale bar = 0.5 mm. 



