168 
Psyche 
[Vol. 91 
no disadvantage; it is a valuable hedge against off-season repro- 
duction. In such benign climates the photoperiodic threshold for 
diapause inhibition may be under intermittent selection at both ends 
of the curve. Non-diapause overwintering may be fairly frequent in 
some California butterflies, especially P. rapae. At the latitude of 
Sacramento it is definitely rare in P. zelicaon, since midwinter 
larvae were not observed in the first 12 yr of this study. In coastal 
southern California, however, this species flies more or less all 
winter (Emmel and Emmel, 1973; Orsak, 1977; Shapiro, unpub- 
lished) and larvae may often be found during the shortest days of 
the year. Although diapause intensity is reduced and thresholds are 
moderately shifted relative to further north, even the San Diego 
populations retain the ability to diapause (Sims, 1980). How often 
they use it under field conditions may depend on the timing of 
autumn reproduction on a year-by-year basis. The hypothesis that 
both thresholds of a Type III curve are adaptive and under selection 
is testable in principle by examining latitudinal shifts in critical 
photoperiods. 
We have had under study a culture of P. rapae from Xochimilco, 
D.F., Mexico, the southernmost (ca. 19° N) population of this 
species in the Americas; it also displays a Type III curve with the 
short-day inhibition threshold shifted upward (Shapiro, in prepara- 
tion). 
In climates where the onset of cold is more rapid than the change 
in photoperiod, that is, where strong air-mass contrasts exist and 
thermal lethality can occur with great suddenness, it is difficult to 
envision the low end of a Type III curve as adaptive. Cold nights 
should assure conservatism in the critical photoperiod, so that 
nearly all individuals will be determined as diapausers before the 
inhibition threshold can be reached. Larvae of P. rapae can be 
found into December and rarely into January at Philadelphia and 
New York City as well when early winter conditions are mild, but 
they never seem to survive. On the other hand, the occasional very 
early onset of lethal cold in such areas should select for more 
conservative diapause induction than the average conditions seem 
to warrant. 
The natural selection of diapause characteristics, already compli- 
cated by its nature as a recurrent (cyclical) process operating every 
4th and 5th generation, is further complicated by the ability of 
stochastic variation to override it occasionally in benign climates. 
