62 
PACIFIC SCIENCE, Vol. VI, January, 1952 
3a. In the figure, the mean count of each 
length group is plotted as a deviation from 
the mean count of the sample. 
It will be noted from Figure 3a that, in 
some samples, mean vertebral number tends 
to increase with length, in others it tends to 
decrease, and in still others it shows no ap- 
parent relationship. Using the original un- 
grouped data it was found that either positive 
or negative regressions of vertebral number 
on length could be statistically significant. 
For example, in a sample from Ala Wai Canal 
(5/22/44) the positive regression coefficient 
of 0.0222 over all lengths had a standard 
error of 0.00863 and differed significantly 
from zero (P = 0.02). Also, in a sample from 
Honolulu Harbor (3/3/49) the negative re- 
gression coefficient of —0.0355 over length 
groups 24 to 40 millimeters (Fig. 3a) had a 
standard error of 0.0162 and differed sig- 
nificantly from zero (P = 0.04). 
Before attempting to interpret the above 
results, it should be pointed out that the 
spawning period of the nehu in those locali- 
ties which have been investigated (Ala Wai 
Canal and Kaneohe Bay) extends throughout 
the entire year. Moreover, the period be- 
tween fertilization and hatching of the egg is 
very short— 24 hours or less. 
If, as in many other plastic species, verte- 
bral number is influenced by environmental 
conditions obtaining during the time of em- 
bryonic and early larval development, varia- 
tion in environmental conditions from day 
to day might produce variation in the mean 
vertebral number of successive day-broods. 
In some species, e.g., \i^timg--Clupea pal- 
lasii (Tester, 1938) — there seems to be an 
inverse relationship between mean vertebral 
number and water temperature during the 
period of early development, low mean 
counts being associated with high tempera- 
tures, and vice versa. In others, e.g., the 
Pacific anchovy — EngrauUs mordax (Hubbs, 
1925 )— there seems to be a direct relation- 
ship between mean vertebral number and 
salinity, the low mean counts being associ- 
A 
ALAWAI 
CANAL 
3/12/44 
ALAWAI 
CANAL 
S/22/44 
ALAWAI 
CANAL 
10/25/46 
ALAWAI 
CANAL 
6/I6/47 
HONOLULU 
HAROOR 
12/27/43 
HONOLULU 
HAROOR 
3 / 3/49 
HONOLULU 
HARBOR 
7/16/49 
PEARL 
HARBOR 
7/30/47 
pearl 
HARBOR 
XT/s/p. 
KANEOHE 
GAY 
10/25/47 
KANEOHE 
BAY 
3/10/48 
KANEOHE 
BAY 
KIHEI 
KIHEI 
3/15/46 
KIHEI 
9/13/46 
HILO 
GAY 
6 / 24/46 
KAWAIHE 
BAY 
\IZ9j^9 
KAWAIHE 
DAY 
3/23/49 
2i 27 33 39 45 51 57 63 69 
21 27 33 39 45 51 57 63 69 
k 
I 
LENGTH CLASSES (MM) 
Fig. 3. a, Variation in mean vertebral count with 
length; B, length frequency distributions of the samples. 
ated with less saline, brackish water condi- 
tions, and vice versa. Therefore, in the nehu, 
upward or downward trends in temperature 
or salinity, extending over a period of days or 
weeks, might produce downward or upward 
trends in mean vertebral number within the 
length range of a sample, thus explaining the 
occurrence of both positive and negative re- 
gressions. It is not necessary to postulate an 
inherited tendency for fast-growing, larger 
individuals of one brood to have a higher 
count than slow-growing, smaller individuals 
of the same brood, although the tendency 
may still exist. 
The length frequency distributions of the 
