Acanthurus triostegus sandvicensis — RANDALL 
If spawning of the manini occurred only dur- 
ing a definite small segment of the month, the 
dotted curve of Figure 15 would probably be 
bimodal, one of the modes centered on a low 
and the other on a high standard length. Since, 
however, spawning appears to occur through- 
out the month and merely more prominently in 
one part than another, most of the young re- 
sulting from fertilization at a time of low spawn- 
ing intensity will come into shoal areas at a 
more nearly average size during the interval be- 
tween periods of high tide-pool recruitment and 
obscure the postulated bimodal effect. The 
extremes in size of this small group would enter 
the tide-pool zone at the time of peak recruit- 
ment; thus, if the samples were large enough, 
both curves would extend to the same limits. 
The maintenance of populations of manini 
around small islands such as Johnston Island, 
the shores of which are constantly swept by a 
strong current in one direction, would seem to 
demand some means on the part of the develop- 
ing pelagic larvae to remain close to land. The 
presence of small eddy systems around an is- 
land might afford the explanation, provided the 
larvae can stay within such systems. The ap- 
parent differentiation of the manini at Johnston 
Island (Randall, 195 6£: table 2) and also of 
Ctenochaetus strigosus (Randall, 195 5<7) would 
seem to rule out the possibility that the Johns- 
ton Island populations are derived primarily 
from fishes which spawn in the Hawaiian Is- 
lands proper. 
While examining museum specimens of trans- 
forming manini from many localities, a definite 
variation in size with locality was noticed. The 
standard lengths of the available early transform- 
ing museum specimens were measured. These 
lengths (except those of Hawaiian Islands speci- 
mens, already presented graphically) are given 
in Table 7 with localities and with collection 
dates when known. 
A correlation exists between the size of ma- 
nini at transformation and the temperature of 
the water of the locality. The warmest region 
listed in the table is the East Indies where the 
mean monthly sea surface temperature may reach 
85° F. (29.4° C.) (Hydrographic Office, 225). 
Here we find the smallest size at transformation, 
20-21 mm. Small standard lengths, averaging 
247 
less than 23.5 mm., are found in other warm 
areas such as the Palau, Marianas, Gilbert, and 
Marshall islands (all regions where the mean 
monthly sea surface temperature exceeds 81° 
F., or 27.3° C, during the entire year), and 
southern India in January, and Okinawa in 
September. Transforming specimens from re- 
gions where the temperature is 80° F. (26.6° 
C.) or less throughout the year or during the 
month of the date of collection are large, 24 
mm. or greater in standard length. These areas 
include Durban (S. Africa), New South Wales, 
Ningpo (China), Hawaiian Islands, Mangareva, 
the Marquesas Islands ( in August ) , and Clarion 
Island. The large transformation size of the 
specimens from the Phoenix Islands, where the 
sea surface temperature exceeds 82° F. (27.8° 
C. ) all year, is an obvious exception to the above. 
Perhaps the manini population in these islands 
has differentiated in this respect, although the 
Phoenix group is not markedly isolated. 
In order to preclude the possibility that the 
different transformation sizes of the different 
areas are entirely due to genetic factors (re- 
flecting possible races of the manini in all these 
areas) a comparison was made of the size at 
transformation of specimens obtained at Oahu 
in an early cool and a late warm part of the 
season. Thirty-nine first-day transforming ma- 
nini were obtained in collections from April 1 1 
to May 8, 1953. The standard lengths of these 
specimens are plotted as the solid line of Figure 
16. The dotted line enclosing the stippled por- 
tion of the graph represents the lengths of 38 
first-day transforming manini taken from July 
7 to October 4, 1953. The average sea surface 
temperature at Oahu from February to May is 
75.7° F. (24.2° C.) ; the average from July to 
October is 79.1° F. (26.2° C), based on av- 
erage temperatures from 1941 to 1947 presented 
by Leipper and Anderson (1950). The mean 
length of the April-May group is 26.256 mm.; 
the mean length of the July-October group is 
25.447 mm. Although the difference in length 
appears highly significant, a t test was made by 
the group comparison method (Snedecor, 1948: 
80) . The resulting t value of 3.11 gives a prob- 
ability of nearly 0.001 of obtaining a value 
greater than this; thus it is extremely unlikely 
that samples as diverse as the above could be 
