476 
PACIFIC SCIENCE, Vol. IX, October, 1955 
are elsewhere nearly constant is excessive. 
If one considers B to be the pure Marshall- 
ese form, C to be the Marquesan, and A and D 
a tendency toward the Hawaiian species, then 
the table may be recalculated as is done in 
Table 9. 
Table 9 demonstrates the unity in pectoral 
markings of the two samples from the north- 
ern Line Islands (Christmas) and the southern 
Phoenix Islands (Hull). These two islands are, 
incidentally, some 800 air miles from one 
another. Also, if types B and D really are a 
tendency towards A. sandvicensis , some indica- 
tion of intergradation with the Hawaiian spe- 
cies seems present in both samples. 
Summarizing the data on pectoral markings 
in the Phoenix-Line populations, it may be 
said that these contain to some degree all the 
marking types to be found in the more con- 
stant races to the west (Marshallese), north 
(Hawaiian), and southeast (Marquesan). 
In regard to caudal peduncle markings, a 
similar concentration of variability in the 
Phoenix and Line Island samples could prob- 
ably be demonstrated. However, an analysis 
of the caudal markings suffers from the two 
facts that the Marshallese and Marquesan 
races are not particularly constant in this 
feature and that the Hawaiian marking is to 
some extent intermediate between the Mar- 
quesan and Marshallese pattern. Suffice it, 
then, to say here that the same bilateral asym- 
metry in the coloration on the caudal pedun- 
cle takes place as occurs in the pectoral 
marking, that there does not seem to be any 
correlation between the shape of the marks 
on the caudal peduncle and those at the 
pectoral bases, and finally that there are again 
all gradations between the various types of 
caudal markings. 
The other feature used by Schultz and 
Woods {op. cit.) in separating the forms of the 
Acanthurus triostegus complex is the dorsal, 
anal, and pectoral fin counts. The Hawaiian 
species was found to have higher average 
sample values for each of these fins than 
samples from elsewhere. Schultz and Woods 
go on to note that the somewhat lower water 
temperatures of the Hawaiian area may be 
responsible for this. 
Dorsal, anal, and pectoral counts of certain 
samples from the Hawaiian, Line, and Phoenix 
Islands, along with the average annual water 
temperatures (as calculated from Charts II 
and III at the back of Sverdrup, Johnston, 
and Fleming, 1946) are shown in Table 10. 
Several points about this table need discus- 
sion. The first regards variation within areas. 
Thus, two samples from the single island of 
Oahu have an average difference of nearly 0.2 
of a dorsal ray. Indeed, one suspects that 
different populations from the same area 
might differ by perhaps 0.3 dorsal ray, 0.2 anal 
ray, and 0.1 pectoral ray, though available 
data is insufficient to prove this. At least 
nothing less than such amounts should be 
considered geographically significant. Sec- 
ond, the various island groups investigat- 
ed seem to have rather different average 
counts, as summarized in Table 10. The 
Phoenix (Hull Island) sample, well to the 
south of the Line Islands and still farther away 
from Hawaii seems to be intermediate be- 
tween the Line and Hawaiian Island samples. 
Another Phoenix Island lot counted by 
Schultz and Woods (1948: table I) indicates 
the same thing. Certainly, no genetic inter- 
mixing between the Hawaiian and Line Island 
samples is indicated. If one attempts to ex- 
plain the change in average count by tem- 
perature effect, the Line Island samples create 
the same stumbling block as for introgression, 
for temperatures in the Phoenix Islands seem 
to be higher (and should therefore give lower, 
not higher, average counts than for the Line 
Island lots). 
Summarizing for the Acanthurus sandvicen- 
sis-triostegus complex, the following points 
may be made. There is no sign of intergrada- 
tion between A. sandvicensis and A. triostegus 
at Johnston; the pure Hawaiian form is rep- 
resented there. In the Line and Phoenix Island 
samples there is some indication of the A. 
sandvicensis pectoral marking in some sped- 
