68 
PACIFIC SCIENCE, VoL XVII, January 1963 
Johnsen indiscriminately attributed both kinds 
of variation to direct modification by the en- 
vironment. 
A further similarity between the results of 
Johnsen s investigation and the present study 
lies in the course of divergence of the fin index 
(calculated from his data). There is a geo- 
graphic cline in the index which is minimal in 
the Baltic Sea and increases to the west and 
then independently south and north in the North 
Sea, but is maximal to the north. 
In Gillie bt by s mirabilis , a genetic cline is 
also suggested by comparisons of population 
means of combined fin ray counts. Variation 
between year classes and a comparison of Salton 
Sea fish with their parental stock indicate that 
slower development is correlated with an in- 
crease in the total number of fin elements. If 
the developmental rates within the populations 
are altered by climatic temperatures, and this 
is a reasonable assumption, then the combined 
counts should decrease from north to- south. But 
the combined counts are about the same for all 
populations, with a few exceptions, and the ex- 
ceptions are obvious deviants. 
One could argue for a cline of combined 
counts for the seven populations in California. 
But the counts increase, not decrease, steadily 
from San Francisco south to San Diego. Such 
differences from north to south could be at- 
tributed to increasingly saline water to- the south, 
or to later spawning seasons to the north. Yet 
the northernmost habitat, the Alviso salt ponds, 
is highly saline, and spawning commences at 
about the same time throughout the range of 
mirabilis . 
There seems to be homeostasis of the com- 
bined counts in spite of changes in latitude, 
since over the rest of the range of mirabilis the 
means are about the same from population to 
population. This suggests genetic compensation, 
or relatively complete acclimation of develop- 
mental rates. The most obvious deviant, the 
population from just north of San Felipe, is 
also atypical in body form. 
The variance of the countable characters in 
each population is even more stable than the 
mean of the characters. It does not appear pos- 
sible to estimate the respective roles of heredity 
and environment in producing this constancy. 
Even though the means of the counts were 
found to shift within limits from year class to 
year class, the variances showed no statistical 
differences. The environmental changes respon- j 
si hie for the changes in mean counts, most likely 
temperature, apparently do not influence the 
variances. This would seem to indicate an in- 
trinsic homeostasis where variability is con- 
cerned. 
Photoperiod might determine variability. Day j 
length is constant from year to year on the 
same day at a given location. Since the change 
in photoperiod over the spawning season in- 
creases faster at higher latitudes, one can demon- || 
strate a positive, although perhaps false, correla- 
tion between degree of change in day length and I 
variance. Interestingly, estimates of the variances ) 
of the different fins in populations of a different 
but closely related species, Gillichthys seta ( Bar- 
low, 1961*), fall directly in the midst of the 
corresponding data (Fig. 10) for mirabilis . \ 
Somewhat further aside, in two different popula- 
tioes of seta the measurements of four body j 
parts, not counts, having different means had 
the same variances (Barlow, 1961*). 
If the variance of each character is largely 
genetically determined, then the data indicate 
two different types of dines in mirabilis. Based j 
on the variances, there would be a simple linear 
relationship, north to south, irrespective of 
whether the populations are situated on the 
Pacific Coast or on the east or the west coasts of 
the Gulf of California. The V-like arrangement 
deduced from the distribution of the means of 
the counts, the fin indices, and measurements of 
body parts have already been described and jj 
will be summarized below. 
The possible relationship between the dif- jj 
ferential of the photoperiod and the variance j 
cannot be considered as evidence for or against |j 
modification or genetic determination. Even ■ 
though the environment originally may have 
induced the variance in question, the phenotype 
easily could have become genetically reinforced. 
The kind and pattern of the differences found . 
between the populations of mirabilis seem too *1 
complex to be accounted for on the assumption 
that the dissimilarities result chiefly from direct 
environmental modification. The most obvious j 
part of the explanation is that the divergences i 
