20 



D.J. SIEBERT 



remains as to whether the apomorphic condition of neoteny in 

 Sundasalanx and salangids is a synapomorphy or a homoplasy. 

 Derived, non-paedomorphic, features unique to both Sundasalanx 

 and salangids can corroborate a Sundasalanx-salangid relationship, 

 thereby establishing neoteny as synapomorphic for Sundasalanx 

 and salangids. 



The recent re-examination of lower eutelostean relationships by 

 Johnson & Patterson (1996) provides a context for the search for 

 non-paedomorphic features that might confirm a Sundasalanx- 

 salangid relationship. The relevant groups and characters (numbered 

 sequentially) are: euteleosts - 1) supraneurals develop caudally, 2) 

 uroneural 1 with a membranous outgrowth (stegural), 3) caudal 

 median cartilages present; salmoniforms - 4) epicentrals with carti- 

 lage rods distally, 5) epicentral and epineural intramuscular bones 

 lacking proximal forks; salmonoids + osmeroids - 6) derm- and 

 supraethmoid separate, 7) postemporal fossa open, 8) a single 

 supramaxilla, 9) upper pharyngeal tooth plate 4 absent, 10) anterior 

 epineurals not fused to neural arches; 11) epipleurals absent, 12) 

 distal parts of posterior neural and haemal arches forming a keel, 13) 

 uroneural 2 anterodorsal to uroneural 1 , 14) scales without radii, 15) 

 nuptial tubercles present, 16) diadromy; osmeroids - 17) orbit- 

 osphenoid absent, 18) basisphenoid absent, 19) articular reduced, 

 20) gill rakers toothless, 21) preural centrum 1 and ural centrum 1 

 fused, 22) cleithrum with a narrow columnar process toward cora- 

 coid, 23) no postcleithrum; osmerids - 24) short hyomandibular 

 crest, 25) opercular with anterodorsal notch, 26) levator process 

 present on epibranchial 4, 27) uroneural 1 fused to preural centrum 

 1, 28) caudal scutes absent, 29) extrascapular fragmented into 

 several ossicles, 30) posterior dorsal pterygiophores fused, 31) 

 adipose cartilage present, 32) egg with adhesive membrane; 

 osmerines - 33) otic bulla (=saccular recess) somewhat inflated, 34) 

 keel formed by posterior neural and haemal spines absent; salangins 

 - 35 ethmoid endoskeleton long and unossified, 36) 1st pectoral 

 radial unmodified, 37) 4th pectoral radial multifid distally, 38) males 

 with modified anal fin endoskeleton; and Icefishes (=salangids) - 

 39) anterior margin of metapterygoid above quadrate, 40) antorbital 

 bone absent, 41) 1 supraneural present, 42) 4th pectoral radial 

 articulates with glenoid, 43) dermal plate absent from basibranchials. 



At least 19 of these characters are 'absence characters', or 'revers- 

 als' to a more primitive condition, leaving just 24 as 'presence 

 characters' .As with Roberts' list, resemblance between Sundasalanx 

 and salangids due to the first class characters requires confirmation 

 from congruence with the second class of characters. Character 32, 

 egg with adhesive membrane, was not checked in this study. Of the 

 remaining 24 characters from the second class of characters 

 Sundasalanx can be shown to have only four, fusion of preural 

 centrum 1 with ural centrum 1 (21), levator process on epibranchial 

 4 present (26), uroneural 1 fused to preural centrum 1 (27), and 

 posterior dorsal pterygiophores fused (30). Sundasalanx simply 

 lacks the rest, either because they are primitively absent or because 

 Sundasalanx is so underdeveloped they never appear in its ontogeny. 

 All of the four that are present in Sundasalanx are also present 

 among clupeocephalan, or even elopocephalan, fishes. The evidence 

 for a close relationship among Sundasalanx and icefishes should be 

 regarded as non-existent. 



Sundasalanx is highly paedomorphic and as obvious from com- 

 parison to the above lists establishing its relationship presents 

 certain difficulties. It is not, however, wholly paedomorphic and the 

 non-paedomorphic features of Sundasalanx suggest a relationship 

 not with salangids, nor any other euteleostean, but with clupeiforms, 

 and further to dussumieriine clupeids. 



Prootic bullae are found only among clupeomorph fishes and the 

 recessus lateralis is found only among clupeiforms (Greenwood et 



al, 1966; Grande, 1985). Sundasalanx is thus a clupeiform. Estab- 

 lishing the relationships of Sundasalanx within Clupeiformes is less 

 easy. Many external and internal features used to elucidate relation- 

 ships within the order (Grande, 1985) are absent; all scute and scale 

 characters widely used in the identification of clupeiforms are 

 absent from Sundasalanx, which is completely scaleless and lacks 

 ribs and supraneurals. However, Sundasalanx exhibits a derived, 

 highly consolidated, caudal skeleton. Derived features are: 1) fusion 

 of preural centrum 1 (PU1), ural centrum 1 (Ul), ural centrum 2 

 (U2), and the first uroneural into a single element; 2) reduction of the 

 number of epurals to 1 ; 3) reduction of the number of hypurals to 5; 

 4) fusion of the parhypural and hypural 1 ; 5) an expanded hypural 4; 

 and 6) absence of the extensions of the middle caudal fin rays 

 characteristic of clupeiforms. Among clupeiforms only the 

 spratelloidin dussumieriine genus Jenkinsia, a marine, Caribbean 

 endemic, approaches the degree of caudal skeleton consolidation 

 found in Sundasalanx (Fig. 7B). The pertinent modifications are 

 (Grande, 1985): 1 ) fusion of PU 1 , Ul, and U2 into a single element; 

 2) reduction of the number of epurals to one; and 3) expansion of 

 hypural 4 to a size equal to that of hypural 1 . Fusion of PU 1 with U 1 

 is also known in some pellonulines, some engraulids andClupeonella, 

 but these lack the other derived caudal skeleton features of 

 spratelloidins (Grande, 1985). Sundasalanx is thus a spratelloidin, 

 and possibly the sister-group of Jenkinsia. 



DISTINCTIVENESS OF SUNDASALANX 

 PRAECOX 



The discovery of materials seemingly intermediate in eye size 

 between^, praecox andS. microps led Roberts (1984) to suggest that 

 distinction between them at the species level needed further consid- 

 eration, with the implication that there might be only one widespread 

 species of Sundasalanx. Sundasalanx praecox and S. microps were 

 re-examined for this study. I conclude that Roberts' (1981) original 

 assessment of the specific status of S. praecox is correct. It is 

 different from other species of Sundasalanx in so many details that 

 it stands out as the most distinctive of all the described species. No 

 other known species of Sundasalanx possesses any palatal or 

 basibranchial teeth. It has far fewer vertebrae, many more upper jaw 

 teeth, larger pharyngeal tooth plates with many more teeth, larger 

 and more numerous gill rakers, a relatively longer posterior parapelvic 

 bone than any species known from Kalimantan, and lacks a midventral 

 line of melanophores associated with the ventral fin-fold. 



DESCRIPTIONS OF NEW SPECIES 



Sundasalanx malleti Siebert and Crimmen, sp. nov. 



(Fig. 11) 



Holotype. MZB 6096, 26.4 mm SL, Indonesia, Kalimantan 

 Tengah, Barito River basin, Sungai Barito at Muara Laung, dip nets 

 and seines, 20-22 Feb 1991, D. Siebert, A Tjakrawidjaja, O. 

 Crimmen, and A. Effendi. 



Paratypes. MZB 6097 (20), collection data as for holotype. 

 BMNH 1996.7.18.147-311 (164), collection data as for holotype. 

 USNM 320689 (5), collection data as for holotype. 



Referred material. BMNH 1996.7.18.315-324 (10), C&S, In- 

 donesia, Kalimantan Tengah, Barito River drainage, sand bars of 



