192 
PACIFIC SCIENCE, Vol. XVIII, April 1964 
lar to L. oxycephalus , and he mentioned no dif- 
ferences other than the somewhat lower somite 
count of acuticeps . D’Ancona (1928) rejected 
affinity of oxycephalus to acuticeps because of 
the ostensible lack of pigment in oxycephalus . 
Bertin (1936) doubted that they are related, 
but he gave no reasons. On present knowledge, 
I consider L. oxycephalus too incompletely 
known to be identified with certainty. 
C. Leptocephalus magnaghii 
D’Ancona (1928:109) rejected Regan’s view 
that L. acuticeps resembles L. oxycephalus , and 
suggested instead that acuticeps should be com- 
pared with his own newly described leptoce- 
phalus from the Red Sea, L. magnaghii ( op. cit. : 
44; pi 3, figs. 4, 5). He stated that the somite 
counts of acuticeps and magnaghii show no 
essential difference ( acuticeps , total 207, pre- 
anal 174; magnaghii, total 205-219, preanal 
157-161, excluding a metamorphosing larva 
with shortened gut). He suggested that acuti- 
ceps might even be considered a younger stage 
of magnaghii, but he decided to recognize both 
forms because of a difference in lateral pig- 
mentation that he did not think could be ex- 
plained by either individual or growth-stage 
variation; kec, the presence of a midlateral longi- 
tudinal row of melanophores along each side in 
magnaghii and the absence of such rows in 
acuticeps. 
L . acuticeps and L. magnaghii differ more 
widely than D’Ancona realized. L. magnaghii 
does resemble acuticeps in total somite count, 
in its long straight gut, and in the low number 
of caudal rays, but it differs in virtually all other 
significant features of morphology and pigmen- 
tation that can be compared. D’Ancona was 
unaware that acuticeps has the unusual pattern 
of three internal blotches, for Regan did not 
mention or figure these spots and Bertin’s com- 
ments on them had not yet been published. The 
Scripps fish collection includes extensive series 
of magnaghii- like larvae from the eastern trop- 
ical Pacific. These leptocephali are identifiable 
as heterocongrid eels on the basis of compared 
somite and vertebral counts, congrid characters 
of metamorphosing specimens, and remnants of 
the larval color pattern that are retained by 
juveniles of Taenio conger sp. in the Scripps col- 
lection. The close agreement of the Scripps 
larvae with D’Ancona’s detailed description and 
excellent illustrations of magnaghii (including 
a metamorphosing specimen) suggests that this 
larval name was based on heterocongrid larvae, 
possibly of two species. 
The relatively greater importance of charac- 
ters other than the somite count, in the critical 
comparison of L. magnaghii and L. acuticeps, 
exemplifies the paradox that data on somite 
counts can be both essential and misleading. The 
eels comprise such a large and complex group 
that totally unrelated forms may independently 
have the same, or broadly overlapping, ranges of 
variation in vertebral counts. For example, at 
least ten families 3 are already known to contain 
species with vertebral counts that fall within 
the range of about 145-155. Thus, an uniden- 
tified leptocephalus that has a somite count 
within this range might belong to any one of at 
least ten families. The successful identification 
of eel larvae requires the use of many additional 
characters. The visceral anatomy supplies more 
informative clues to the family affinities of a 
leptocephalus than does the somite count. 
D. Congrid Larvae 
Characters of congrid larvae . So far as I can 
find, only D’Ancona and Bertin have tried to 
assign Leptocephalus acuticeps to a family in the 
eel classification. Each author considered it to 
be a congrid, but neither stated his reasons. A 
detailed comparison of acuticeps with congrid 
larvae should reveal whether it properly belongs 
with them, but two formidable difficulties ham- 
per this comparison: the wide disagreement 
among taxonomists as to the composition of the 
family Congridae, and the resultant uncertainty 
over the criteria for defining congrid vs. non •• 
congrid larvae. The present paper is hardly the 
place for an attempt to settle the natural bound- 
aries of this family, yet some sort of limits 
must be indicated in order to permit useful 
larval comparisons. 
The type genus, Conger, is the base line for 
comparison of larval stages as well as of adults 
3 Heterenchelidae, Muraenidae, Synaphobranchidae, 
Ilyophidae, Nessorhamphidae, Echelidae, Ophichthi- 
dae, Congridae, Muraenesocidae, Serrivomeridae. 
