410 
PACIFIC SCIENCE, Vol. XX, October 1966 
PERCIFORM ORIGINS 
The perciform fishes are generally believed to 
have arisen among the Beryciformes. Their suc- 
cess, as compared to that of the Beryciformes or 
indeed of any other order, cannot be attributed 
to any one major advance, but seems to be the 
result of an assimilation and integration of a 
number of minor improvements over beryci- 
form features. Among the characters separating 
the Perciformes from the Beryciformes, some of 
the better documented are: 
(1) Pelvic fins with no more than 5 soft 
rays. However, Channa ( Ophiocephalus ) has 
6 segmented rays, and the flatfishes, presumably 
derived from the Perciformes, have up to 13 
(Norman, 1934). 
(2) Pelvic bones extending between and 
directly attached to the cleithra. The several 
exceptions to this seem to fall into two cate- 
gories (Regan, 1909) : those groups some or 
all of which seem never to have attained such 
an attachment — the Mugiliformes, Anabantidae, 
Channidae, and Nannatherina (Regan, 1940) ; 
and those groups in which some or all of the 
species seem to have secondarily lost such 
an attachment — Stromateidae, Tetragonuridae, 
Gempylidae, and Trichiuridae. 
(3) Orbitosphenoid, antorbital, and nodules 
between the pelvic fin rays and the pelvic girdle 
absent as separate entities. There is no known 
percoid that retains any of these bones in the 
adult. Such bones, all present in the beryciform 
family Holocentridae, have been lost, however, 
in numerous fishes besides the Perciformes. 
(4) Branched caudal rays 15 or fewer. Re- 
duction in the number of branched caudal rays 
in the percoids is rather commonplace. A few 
round-tailed forms are known to have more 
than 15 (Gosline, I960). 
(5) In the percoids, as contrasted with most 
beryciform fishes, there are basically five cir- 
cumorbital bones behind the lacrimal, and a 
subocular shelf, if present, tends to be restricted 
to the second (but see Katayama, 1959: Figs. 
3-5). In the Beryciformes, except Holocentri- 
dae, there appear to be only four circumorbitals 
(Patterson, 1964), and the subocular shelf tends 
to spread over more than one of them. In this 
character, as in the generally high degree of 
ossification, it is the berycoids that seem to be 
unusual as compared with other acanthopteran 
fishes (see below). 
That an integration of the above characters 
did not occur all at once is shown by the groups 
of modern fishes which seem to have stopped 
short part way along the path of beryciform- 
perciform evolution: e. g., the Mugiliformes, 
anabantoid-channoid group, and apparently 
Nannatherina (see above paragraphs). 
It has generally been postulated, implicitly or 
expressly, that the Perciformes has had a single 
origin among the Beryciformes. In 1964, how- 
ever, Patterson suggested four separate origins 
for perciform families among the Beryciformes. 
Specifically these are: 
BERYCIFORM ANCESTORS PERCIFORM DERIVATIVES 
Polymixiidae » Scorpidae, Monodactylidae, 
and Kyphosidae 
Sphenocephalidae ■ — » Serranidae 
Aipichthyidae » Menidae and Carangidae 
Pharmacichthyidae » Acanthuroidei 
Pycnosteroididae — ? Chaetodontidae 
Dinopterygidae — ? ? -» Centrarchidae 
Thus, according to Patterson’s view the basal 
percoid families would have at least three and 
possibly five independent derivations from the 
Beryciformes, and the Acanthuroidei would 
have evolved from a sixth. Such a viewpoint 
deserves discussion in considerable detail. 
Thanks to Patterson’s (1964) excellent re- 
descriptions and figures of the Cretaceous bery- 
coids Berycopsis, Homonotichthys, Hoplopteryx, 
and Caproberyx, it is possible to make a detailed 
comparison between these forms and the better 
known of modern berycoids (Starks, 1904). 
They fit together nicely. Thus Berycopsis and 
Homonotichthys belong in the same family with 
the modern Polymixia (Starks, 1904) ; Hoplop- 
teryx with the modern Trachichthys; and Capro- 
beryx with the modern holocentrids. These 
Cretaceous and modern forms together make 
up a sort of central core of the known Beryci- 
formes, forming a congruent, easily recogniz- 
able, and clearly definable group of fishes. 
Thus, when Patterson suggests Berycopsis, a 
Cretaceous polymixiid that he has described 
in detail, as an ancestral type for the percoid 
families Scorpidae, Monodactylidae, and Kypho- 
sidae, what he is saying is clear. Unfortunately, 
