288 
PACIFIC SCIENCE, VoL XXII, July 1968 
TABLE 2 
Comparison of Dioecism and Woodiness in the Flora of Three Areas 1 
AREA 
TOTAL 
SPECIES 
WOODY 
SPECIES 
PER CENT 
WOODY OF 
TOTAL 
WOODY AND 
DIOECIOUS 
PER CENT DIOECIOUS 
OF WOODY SPECIES 
Hawaii 
730 
325 
44% 
32 
9% 
So. Calif. 
2,335 
613 
26% 
22 
3% 
Ecuador 
658 
282 
42% 
19 
7% 
1 Data compiled from Hillebrand (1888) for Hawaii, Munz (1935) for southern California, and Diels (1938) for Ecuador. 
sons between woodiness and dioecism in the 
three areas, Hawaii, California, and Ecuador. 
This comparison exposes an even greater differ- 
ence between the floras of Hawaii and Cali- 
fornia. Based on the total number of woody 
species that could be tabulated from Hillebrand’s 
Flora and the number from Munz’s Manual, 
the percentage of those showing dioecism is 3 
for California and 9 for Hawaii. The two-by- 
two contingency table indicates a signicant dif- 
ference (P < .005). The percentage of woody 
species showing the dioecious habit in the 
Hawaiian flora is so much greater than that in 
the Californian flora that one can be assured 
that some factor is responsible for the differ- 
ence. It could not be merely random chance. 
Our question — Is there a signicantly higher 
percentage of dioecious species in the Hawaiian 
flora than in a continental flora? — can be an- 
swered in the affirmative on the basis of the 
samples from Hawaii and from southern Cali- 
fornia. Relative to the comparison of the per- 
centage of dioecism in woody species in Hawaii 
and in Ecuador, the difference, while less, is 
still signicant (P = .005). It is of interest to 
compare the present results with estimates of 
percentage dioecism from some other areas. 
Baker (1966, personal communication) has 
calculated that the approximate percentage of 
dioecious species in west tropical Africa was 
3.2 per cent and for all of California, 2.4 per 
cent. Parsons (1958) found approximately 3-9 
per cent dioecism for the Angiosperms of Aus- 
tralia. Milliner (1966, personal communication) 
estimated that approximately 20 per cent of the 
indigenous flora of New Zealand was dioecious. 
EVOLUTION OF DIOECISM IN HAWAIIAN PLANTS 
The problem remains of explaining why 
there is a higher percentage of an outcrossing 
type of breeding system (dioecism) in Hawaii, 
an area that apparently can be attained only by 
long distance dispersal. The possibility of an 
alternative explanation for the origin of the 
Hawaiian flora will be discussed briefly later. 
Let us assume that at least a majority of the 
indigenous species now present either have 
arrived themselves by long distance dispersal or 
have evolved from ancestors that had arrived 
in this way. This is the more likely explanation. 
Fosberg (1948) suggests that if on an average 
one seed plant arrived and became successfully 
established every 20,000 to 30,000 years the 
present flora would be accounted for. His cal- 
culations are based on the number of indigenous 
species now present and the estimate of ". . . 
5-10 million years of above- water history for 
the entire Hawaiian chain.” This estimate is in 
line with the ones given today by geophysicists 
(Woollard, 1965) who suggest that the young- 
est island may be about 2 million years old and 
the older islands about 5 million years old. 
To explain this apparent contradiction be- 
tween Baker’s Law and high dioecism in 
Hawaii the author believes that the dioecious 
species present have evolved this habit in situ 
and are descended from species or strains that 
were hermaphroditic when they arrived and be- 
came established. That there has been time for 
this sort of change to dioecism to occur is evi- 
dent. Approximately 80 per cent of the Hawai- 
ian flora consists of endemics (St. John, 1946). 
These endemic species had time to evolve. 
Dioecism can occur quite rapidly. Jones (1932, 
1934) experimentally produced a dioecious 
maize in four generations. Warmke and Blakes- 
lee (1940) established a dioecious tetraploid 
race in a species of Melandrium ( = Lychnis , 
Cariophyllaceae) . H. Lewis (1966) has pointed 
out that differences in chromosome numbers 
between members of the same family in differ- 
