REVISION OF ANISOCHROMINAE 
through the scale at the midposterior edge of the hypural plate. 
‘Scales in transverse series’ were counted anterodorsally from the 
anal-fin origin to the dorsal-fin base and are presented in the form “x 
+ 1 +y =z’ where ‘x’ is the number of scales between the anterior 
lateral line and the anal-fin origin, ‘1’ is the anterior lateral-line 
scale, ‘y’ is the number of scales above the anterior lateral line to the 
dorsal-fin base, and ‘z’ is the total number of scales in the series. 
Scale-row number for the position of the first ctenoid scale was 
determined by aligning the scale along an anterodorsal scale row to 
the anterior lateral line, then recording the number (counting from 
the origin of the lateral line) of the tubed scale in that row. Gill-raker 
counts were of the outer rakers on the first arch; the angle raker is 
included in the lower-limb (second) count. Counts of pseudobranch 
filaments included all rudiments. Because counts of gill rakers and 
pseudobranch filaments could not be made without damaging the 
branchiostegal membranes, these were only made on a few speci- 
mens. 
Nomenclature of cephalic laterosensory pores is clarified in Fig. 
1, and follows Winterbottom (1986), except for the nomenclature of 
those of the posttemporal area, which follows Gill et al. (2000). 
Vertebral counts are given in the form precaudal + caudal = 
total. Caudal vertebrae are defined as those with a haemal spine, 
and include the terminal urostylar complex (which was counted as 
a single vertebra). The pattern of insertion of supraneural 
(predorsal) bones and anterior dorsal-fin pterygiophores within 
interneural spaces is given as an ‘anterior dorsal-fin pterygiophore 
formula’ modified from the ‘predorsal formula’ of Ahlstrom et al. 
(1976). Each supraneural is represented by an ‘S,’ neural spines 
are represented by slashes, and pterygiophores are represented by 
‘2’ (indicating a pterygiophore that bears a supernumerary ray 
and a serially associated ray), or ‘1’ (indicating a pterygiophore 
that bears only a serially associated ray). A superscript ‘v’ indi- 
cates where a supraneural bone was present only as a vestige. 
Parentheses enclose elements that may be absent. An ‘anterior 
anal-fin pterygiophore formula’ is also presented, where the 
slashes represent haemal spines. Epineural counts are of the inter- 
muscular bones that have been traditionally identified as epipleural 
ribs by perciform workers; this terminology follows Johnson & 
Patterson (1993: 557, ‘the bones conventionally called epipleurals 
in those fishes [atherinomorphs and perciforms and their rela- 
tives] are homologous with the epineurals, not the epipleurals, of 
non-acanthomorph fishes’ ). Posterior epineurals were often poorly 
ossified, and difficult to count accurately on radiographs; counts 
should therefore be considered approximate. Osteological features 
were determined from radiographs and from cleared-and-stained 
specimens, which were prepared following the methods of Taylor 
& Van Dyke (1985). Examination of ligaments was enhanced by 
transfer of cleared-and-stained specimens from glycerol to 70% 
ethanol. 
We use the terms ‘female’ and ‘male’ in preference to ‘ocellated 
phase’ and ‘terminal phase’ of Springer et al. (1977). We appreciate, 
however, the deficiencies of this terminology. As noted by Springer 
et al.(p. 7) for A. straussi: ‘specimens with ocellated stage colour 
pattern are smaller and generally females, and specimens with 
terminal stage colour pattern are larger and are males, but there is 
considerable overlap. The overlap is not surprising and is compar- 
able to that found in other transforming hermaphrodites.’ 
Count and morphometric value ranges are given first for all 
specimens, followed, where variation was noted, by values for the 
holotype in parentheses; where bilateral counts were recorded from 
the holotype, both counts are given and separated from each other by 
a Slash, the first count being the left count. Frequency distributions 
for selected meristic characters are summarised in Tables 1-11. 
193 
SYSTEMATICS 
ANISOCHROMINAE SMITH, 1954 
Anisochromidae Smith, 1954: 298. 
Anisochromis Smith, 1954 
Anisochromis Smith 1954: 298 (type species, Anisochromis kenyae 
Smith, 1954, by original designation and monotypy). 
MONOPHYLY 
The following eight autapomorphies diagnose Anisochromis as 
monophyletic: 
1. Ectopterygoid and mesopterygoid well separated from palatine. 
In pseudochromines and pseudoplesiopines the palatine is loosely 
connected to the ectopterygoid via a short ligament and to the 
mesopterygoid via a short, narrow ribbon of cartilage. In 
congrogadines the palatine is closely applied, and often tightly 
bound, to the mesopterygoid; the ectopterygoid lacks an anterior 
process, and is disassociated from the dorsal part of the 
palatoquadrate (e.g., Godkin & Winterbottom, 1985: fig. 6; Gill 
et al., 2000: fig. 3). In anisochromines the palatine is broadly 
separated from the ectopterygoid and mesopterygoid: the 
ectopterygoid is truncated anteriorly (though closely applied to 
the mesopterygoid), and lacks a ligamentous connection to the 
palatine; the mesopterygoid is connected to the palatine via a 
relatively long, broad, strap-like ligament (which lacks cartilage) 
(Fig. 2). 
2. Preopercle well separated from skull. In congrogadines, pseudo- 
chromines and pseudoplesiopines the dorsal tip of preopercle is 
in close proximity to the skull, and the preopercular laterosensory 
canal communicates with the pterotic canal via a short membra- 
nous canal; the dorsal tip of the preopercle reaches to near the 
dorsal margin of the hyomandibula (e.g., Gill et al., 2000: figs 1, 
3). In anisochromines the preopercle is truncated dorsally, not 
reaching the dorsal margin of the hyomandibula (reaching to or 
slightly above the opercular process of the hyomandibula), and 
well-separated from the skull (Fig. 2); the preopercular 
laterosensory canal communicates with the pterotic canal via a 
relatively long membranous canal. 
3. Dorsal insertion of posterior mandibulohyoid ligament. Springer 
et al. (1977) proposed that Anisochromis is autapomorphic in 
having a modified orientation of a cord-like ligament extending 
from the anterior ceratohyal to the dentary, from insertion near 
the symphysis (pseudochromines and pseudoplesiopines) to in- 
sertion on the coronoid process. However, we present a different 
interpretation of this character. There are actually two more-or- 
less cord-like ligaments connecting the dentary and anterior 
ceratohyal in anisochromines, pseudochromines and 
pseudoplesiopines. [We use the general term ‘mandibulohyoid’ 
for these ligaments following Greenwood (1995), though we do 
not intend to imply homology with the mandibulohyoid liga- 
ments of lower teleosts.] The first of these extends from the 
medial or dorsal surface of the anterior ceratohyal, in the vicinity 
of a dorsal notch in the bone (possibly homologous with the 
beryciform foramen of McAllister, 1968), to the dentary sym- 
physis. The second mandibulohyoid ligament extends from the 
lateral surface of the anterior ceratohyal at a point just anterior to 
the anterior/posterior ceratohyal suture to either the posterior end 
of the ventral process (pseudochromines and pseudoplesiopines) 
