694 
Fishery Bulletin 96(4), 1 998 
Table 2 
Heuristic search results for an analysis of 106 characters 
for 53 ingroup taxa. 
Addition 
sequence 
Branch- 
swapping 
algorithm 
Tree 
length 
(steps) 
Number 
of trees 
found 
Cumulative 
total of 
unique 
trees 
as is 
tbr 
403 
112 
112 
as is 
spr 
403 
80 
112 
as is 
nni 7 
405 
16 
128 
closest 
tbr 
403 
16 
128 
closest 
spr 
403 
16 
128 
closest 
nni 
403 
16 
128 
simple 
tbr 
403 
112 
128 
simple 
spr 
403 
16 
128 
simple 
nni 7 
409 
48 
128 
random 2 
tbr 
403 
128 
128 
random 2 
spr 
403 
128 
128 
random 2 
nni 
403 
128 
128 
7 Trees found by this method were not of minimum length and not 
added to the total number of trees observed. 
2 One hundred replicates for each random addition sequence. 
quences using the “nearest neighbour interchange” 
branch-swapping were the only combinations that 
did not find trees with 403 steps (Table 2). The re- 
sults with 100 “random” addition replicates were not 
different from those observed with the three other 
methods (Table 2). This nonrigorous test suggests 
that heuristic search combinations were effective in 
finding all of the most parsimonious cladograms. 
A quantitative estimate of data decisiveness (Golo- 
boff, 1991) is not possible for an analysis of 53 taxa 
which requires calculation of mean length for all 
possible cladograms. However, a generalized state- 
ment of data decisiveness suggests that data for this 
analysis are decisive. “Data are strongly decisive if 
one or more cladograms explaining them is very much 
shorter than others, and only weakly decisive if all 
possible cladograms are not very different for each 
other in length” (Forey et al., 1992). In total, only 
128 unique cladograms trees were observed (Table 
2), which represents a minute fraction of possible 
trees, and only a slightly larger fraction of those ac- 
tually examined during the search. Assuming that 
all trees of 403 steps were found, there must be many 
trees that have more than 403 steps. Although the 
frequency distribution of trees with number of steps 
cannot be determined, it is assumed that the num- 
ber of trees with more than 403 steps must increase 
dramatically given that the maximum number of 
steps is 1384. 
The majority-rule consensus of these trees (Fig. 1) 
illustrates clades observed in 50% or more of the 128 
results. Clades found in less than 100% of the trees 
are indicated in parentheses as the percent of trees 
observed at this node. Only 5 of 47 resolved nodes 
were observed in less than 100% of the trees and only 
2 were observed in less than 75% of the trees. Ex- 
amination of character distribution in the tree re- 
veals the homoplastic nature of many characters used 
in the analysis. This is reflected by the low consis- 
tency index (ci=0.33) observed in the 128 most par- 
simonious cladograms (Rohlf, 1982). 
Low consistency index is expected in studies of 
interrelationships for large numbers of taxa 
(Sanderson and Donoghue, 1989). The expected con- 
sistency index for a study of 53 taxa would be 0.14 
with an equation of linear regression derived from 
60 previous cladistic studies (Sanderson and 
Donoghue, 1989). This consistency index suggests 
that there is less homoplasy describing the interre- 
lationships of the Pleuronectidae than in other stud- 
ies of this size. The retention index (ri=0.79) indi- 
cates that homoplasy is occurring at terminal nodes 
and not internal nodes, which, in turn, suggests that 
relationships at the level of subfamily, tribe, and gen- 
era are not based purely on homoplastic morpholo- 
gies (Forey et al., 1992) and that the strength of this 
analysis can be used to determine relationships at 
this level. The retention index also suggests that this 
analysis is not effective at determining relationships 
near terminal ends, such as interrelationships among 
species within a genus. Consequently, the character 
analysis is restricted to the level of subfamily and 
tribe, and intragenera analysis will be explored only 
for relationships well corroborated by uniquely de- 
rived character states. 
Monophyly of the Pleuronectidae 
Ten synapomorphies define the Pleuronectidae. Dis- 
tributions of these states were also surveyed in the 
literature for taxa within the bothoid lineage 
(Hensley and Ahlstrom, 1984) and basal lineages 
within Pleuronectiformes (Chapleau, 1993). 
1 Ocular-side frontal articulated with mesethmoid. 
Outgroup taxa with ocular-side prefrontal sepa- 
rating frontal from mesethmoid. Distribution of 
this structure within other pleuronectiform taxa 
reveals that the frontal on the ocular side is not 
articulated with the mesethmoid in Psettodidae, 
Citharidae, Paralichthyidae, Taeniopsettinae, 
and some genera of Bothinae ( Arnoglossus , 
Psettina, Asterorhombus, Japonolaeops, Laeops, 
Komoharaia, Neolaeops, and C hascanopsetta) and 
is observed to be in contact with the mesethmoid 
only in the Bothinae genera ( Parabothus , 
