THE BIOTA OF LONG-DISTANCE DISPERSAL. II. LOSS OF
DISPERSIBILITY IN PACIFIC COMPOSITAE
SBXRWIH CARLQUIST
Claremont Graduate School and Rancho Santa Ana Botanic Garden, Claremont, California
Reprinted from EVOLUTION, Vol. 20. Xo. 1. April 8, 1966
pp. 30-48
Made in United States of A merica
T H E BIOTA O F LONG-DISTANCE DISPERSAL. I I . LOSS O F
D I S P E R S I B I L I T Y I N P A C I F I C COMPOSITAE
SHERWIN CARLQUIST
Claremont Graduate School and Rancho Santa Ana Botanic Garden, Claremont,
California
Accepted August 14, 1965
dispersibility can be afforded. Moreover,
Polynesian Compositae represent not one,
but a multiplicity of independent systematic
groups within the family, also offering
replications of any evolutionary phenomenon induced by insular conditions.
Interestingly, Pacific Compositae are
overwhelmingly American in affinity, thus
providing alien notes in Polynesian floras,
which are predominantly Indo-Malaysian
in character. The fact that Polynesian
Compositae are mostly American in origin
derives from phytogeographic circumstance.
The western Pacific is poor in Compositae,
whereas the family is abundantly developed
in the western portions of North and South
America. Polynesian C o m p o s i t a e with
American affinities can mostly be traced
to mainland genera and even species with
relative clarity. To estimate the degree to
which dispersibility has been lost, one must
have an approximate idea of the ancestral
fruit form. Pacific species in which dispersal mechanisms are still relatively intact are also important, for they show
earlier stages in loss of dispersibility.
The peculiar morphology of fruits in
certain Pacific Compositae, such as Dendroseris litoralis (Fig. 7G) or Bidenx macroc ar pa (Fig. 2K) first suggested to the
writer a phenomenon which could be designated "loss of dispersibility." In addition, the prominence of loss of flight among
insular insects and birds (viz., Zimmerman,
1948) suggested that among insular plants,
a similar loss of vagility might occur. Ecological considerations have also proved suggestive. The reasons for loss of dispersibility in insular plants prove not unlike the
reasons for ecological shift and loss of
vagility in insular insects given by Darling*
ton (1943) and Wilson (1959).
Comparison of fruits and seeds in the
endemic floras of oceanic islands with those
of related species of mainland floras reveals
that the insular species often have fruits
or seeds notable for diminished dispersibility. This observation, which has been offered elsewhere (Carlquist, 1965, 1966)
needs further explanation, documentation,
and qualification. One way of demonstrating this phenomenon is a survey of the
floras of oceanic islands of sufficient age
and ecological opportunity to have fostered
appreciable evolutionary change. The writer
is currently preparing such a survey for
the Hawaiian flora. Another means of
demonstrating the phenomenon is to examine fruits or seeds of a large family,
well represented on islands, notable for
rapid evolution, and sensitive adjustment
to new environments. This latter method,
suggested as a means of analyzing various
tendencies on islands (Carlquist, 1966), is
particularly suitable for study of loss of
dispersibility. Among plants, the family
Compositae is an outstanding example of
evolutionary plasticity; it is also the largest
family of flowering plants.
Viewed on other grounds, Compositae are
also an excellent group for such a survey.
Fruits of Compositae are often provided
with exceptionally prominent dispersal
mechanisms. Any diminution of these devices can be noticed readily. The excellence
of dispersal mechanisms of mainland Compositae have permitted them to range far
into the Pacific, so that representation of
the family is sufficient for study of autochthonous alteration of fruit morphology
on islands. The presence of Compositae
on numerous islands is advantageous, for
each island or archipelago serves as a replication, so multiple confirmations of loss of
EVOLUTION 2 0 :
30-48. March, 1966
M)
DISPERSAL IN PACIFIC COMPOSITAE
T H E THEORY
Compositae have been able to reach Polynesian islands both because of good dispersal mechanisms and because of ability
to enter pioneer situations, establish, and
adapt rapidly to new conditions. As a
family, Compositae contain a preponderance of species adapted to pioneer or disturbed habitats and a minority of species
suited primarily to stable habitats (e.g.,
forests). "Weedy" species are favored as
Immigrants to oceanic islands and, as island floras mature over periods of geological
time, forest species are evolved from pioneering species. The process of ecological
shift in insular stocks involves several concomitant changes.
Plants of open habitats usually have excellent dispersal mechanisms, whereas those
of stable habitats often do not. This correlation is related, in part, to the fact that
open habitats are far-flung and open for
occupation—and vanish—over short periods of time. Fruits and seeds of plants of
open habitats are usually light. This is
suited not only to good dispersibility but
also to the fact that seeds of such plants
germinate in sunny locations where a seedling can succeed with minimal storage of
food materials within the seed.
On continental areas, stable habitats are
filled with a large assemblage of species.
Such niches are thus pre-empted, and pioneering species succeed in invading stable
habitats with relative infrequency. In insular situations, the stable-habitat species
are lacking—they make poor immigrants
because of poor dispersibility and specialized
ecological requirements. Islands have potential forest regions, however, and pioneering species can exploit these relatively uncontested regions by adaptation to new
ecological requirements.
In the process of such an adaptive shift,
pioneering species tend to acquire features
which mimic those of mainland species
from stable habitats. One such feature is
precinctiveness: specialization to a more restricted set of ecological tolerances—conditions which are fulfilled in geographical
31
areas which are rather small and which
do not change rapidly. Such precinctiveness
lowers the value of an efficient dispersal
mechanism. The majority of sites favorable to an ecologically specialized species
will lie within a short radius of a plant and,
with poor dispersal, many seeds liberated
by such a plant will succeed in establishing.
A dispersal mechanism can dwindle to the
level of efficiency necessary to maintain a
species within regions suited to it ecologically.
In most species of insular Compositae,
there seems little danger that efficient dispersal will carry an excessive number of
fruits to unsuitable sites or out to sea. This
is a possibility, however, with wind-dispersed Compositae such as the groups shown
in Figs. 6 and 7. Abundant pappus could
acquire a negative selective value in such a
case.
Species which have a diminished dispersibility may still be able to disseminate
throughout an archipelago. For example,
Bidens sandvicensis (Fig. 2H) lacks a dispersal mechanism capable of long-distance
transport, but this species has been able to
reach most of the major islands in the
Hawaiian chain—Kauai, Oahu, Maui, and
Hawaii.
Continental forest trees often have large
seeds. Abundant storage in such seeds permits seedlings to establish in shady locations where they must rely upon such
storage until they penetrate upper, betterlighted levels. Some of the insular Compositae studied here appear to mimic this
habit. If increase in seed size without concomitant increase in dispersal mechanism
occurs, dispersibility is proportionately diminished.
A dispersal mechanism bears, of course,
a close relationship to the nature of the
agent of dissemination. In many of the
Compositae discussed here (e.g., Bidens),
distribution by seabirds seems likely. A
dispersal device associated with seabirds is
favored in plants adapted to areas habitually occupied by such birds—coastal bluffs,
for example. Many of the fruits carried off
32
SHERVVIN CARLQUIST
by seabirds will reach coastal sites, falling
to the ground when birds preen, for example. Fruits of a Bidens species adapted
to inland sites would probably not be
carried off to coastal sites in excessive
numbers, for visits to inland areas by seabirds would probably be few. However,
Compositae which adapt to inland scrub or
forest are not efficiently dispersed within
inland areas by seabirds, and the connection between dispersal mechanism and dispersal agent is broken. With this connection
broken the dispersal mechanism may no
longer have a strong selective value and
may vanish.
A dispersal mechanism may be refashioned to suit a new type of dissemination. The peltate appendage of fruit-enfolding bracts of Lecocarpus (Fig. 4E) is
unique, and may have originated in the
Galapagos Islands.
Alteration of fruit morphology in various
respects other than the dispersal mechanism itself might have "side-effects" which
would result in degeneration of the dispersal mechanism. Pleiotropic genes might
have such an effect. Changes which lead to
gigantism throughout the plant body might
effectively cancel a dispersal mechanism
which depends on small fruit or seed size.
Such gigantism may have occurred in
Argyroxiphium, Wilkesia, and Fitchia speciosa.
A dispersal mechanism which is not "enforced" by high selective advantage may be
expected eventually to vanish. Tolerances
in this regard may be rather wide, however.
One need not expect perfect correlation between ecology and dispersal mechanism; a
mechanism could persist in some or all individuals of a species for a protracted period after its value has disappeared. Once
disappeared, changes in the fruit which
would be deleterious in mainland relatives
may accumulate. For example, the grotesque sculpturings of some achenes described here (e.g., Dendroseris lit oralis,
Fig. 7G) may be explainable in this fashion.
The conditions described above are not
unique to islands, and counterparts may be
found on mainland areas. The totality of
the patterns described here, however, is
distinctive and forms a portion of what
may be called an ''insular syndrome" of
evolutionary changes.
MATERIALS AND M E T H O D S
Herbarium specimens of virtually all of
the species mentioned below were available.
For the use of these, I wish to thank the
curators of herbaria at the Bernice P.
Bishop Museum, Rancho Santa Ana Botanic Garden, Chicago Natural History
Museum, and San Diego Museum of Natural History. Excellent specimens of Juan
Fernandez Compositae were given to the
writer by the late Dr. Carl Skottsberg. A
list of specimens, together with citation of
authors' names, seems too bulky for presentation here, and is not required in view
of the fact that this paper is not primarily
systematic in outlook. Such a list, however,
has been prepared and is available upon
request.
Nomenclature for Pacific Compositae is
according to the following authors: for
Bidens, Sherff (1937b); for Oparanthus,
Sherff (1937a); for Fitchia,
Carlquist
(1957) and Carlquist and Grant (1963);
for Scalesia, Howell (1941); for Macraea,
Harling (1962); for Lipochaeta,
Sherff
(1935); for Argyroxiphium and Wilkesia,
Carlquist (1959); for Juan Fernandez
Compositae, Skottsberg (1922, 1951, 1958).
Ecological data are obviously important
in a study of this sort. The writer gained
an acquaintance with general vegetational
features during his field work in the Revillagigedos, Hawaiian, and Society islands.
Travel in the Society Islands was aided by
National Science Foundation Grant NSFG23396. Thanks are due Dr. Elmo C. Hardy
and Dr. George W. Gillett for their assistance during my field work in the Hawaiian
Islands. Additional ecological data of a
simple sort were gleaned from labels of
specimens and from the works of Brown
(1935), Sherff (1937b), Stewart (1911),
Harling (1962), Skottsberg (1953, 1954),
DISPERSAL IN PACIFIC COMPOSITAE
Zimmerman (1948), and Papy (19541955).
A uniform scale of drawing has been
employed throughout this paper, except for
Fig. 3, where large size of fruits in Fitchia
necessitated a scale half that of the other
figures.
GENUS
BIDENS
Morphology.—Bidens
(tribe Heliantheae,
subtribe Coreopsidinae) is well represented
by endemic species in the Pacific (Figs. 1
and 2). Bidens pilosa, a widespread tropical weed, has been introduced by man to
many Pacific islands and is included in this
study only as a representative of a mainland species. The Pacific species of Bidens
may be regarded as close to the B. pilosa
complex, and this species (Fig. 1, upper
left) probably represents a reasonable approximation of the stock or stocks of Bidens
which established on Pacific islands aboriginally. For each of the insular species
shown in Figs. 1 and 2, a typical achene is
shown (or for a few, variables in morphology are illustrated). Achene morphology
does vary considerably within some species
of Bidens.
The dispersal mechanism of Bidens is
evidently quite efficient, for the genus has
reached, by natural means, many of the
smaller islands of Polynesia. However, the
genus appears to have "bypassed" many
seemingly suitable islands (Rapa, Mangareva, and the Austral islands other than
Marotiri); this suggests that chance has
operated in dispersal of Bidens. A single
collection of B. australis on Tonga has been
made; although not unlikely, this record
needs confirmation. If B. australis is present on Tonga, one would expect this or
other Bidens species on Samoa or other
islands.
Xone of the endemic Pacific species of
Bidt ns seemingly have dispersal mechanisms as efficient at that of B. pilosa. Five
features may be said to comprise the dispersal mechanism in B. pilosa: (1) awns
long compared to body of the achene; (2)
awns spreading; (3) awns armed with stiff
retrorse barbs (unicellular thick-walled tri-
33
chomes); (4) upward-pointing hairs on
lateral margins of the achene (biseriate
trichomes usually terminated by a single
cell, not pairs of cells); (5) upwardly-pointing hairs on the dorsiventral margins (center of flattened faces of the achene). The
length of awns seems important because it
controls the number of barbs and thus the
likelihood of attachment to animal fur or
bird feathers. Even without barbs, some
degree of adhesion can be effected by long
awns, especially if they are spreading. The
presence of retrorse barbs on the awns in
conjunction with upward-pointing hairs on
the achene body seems important. Oriented
counter to each other, these structures seem
more likely to secure a firmer adhesion
than either alone would achieve. Any depreciation in the five features listed above
may be said to constitute a loss of dispersibility.
Among the South Pacific species of
Bidens, only a few retain the above-mentioned features to a marked degree. Those
best equipped in this regard seem to be the
trio of species from the southeasternmost
islands: B. hendersonensis, B. mathewsii,
and B. saint johniana. All three species have
numerous hairs on the achene body (dorsiventral as well as lateral margins, except
for some populations of B.
hendersonensis).
Similarly well equipped is the unnamed
species from Socorro Island, Mexico (Fig.
I, upper right).
Second in potential dispersibility among
South Pacific species are those with short,
but still barbed, awns and with hairs on
the achenes (lateral margins only): B.
australis, B. lantanoides, B. mooreensis, and
B. cordijolia. Confirmation of efficiency of
dispersal in these species is shown by the
fact that two (B. australis, B. lantanoides),
like B. hendersonensis, occur on more than
one island, whereas all other South Pacific
species are restricted to single islands.
A third category is represented by species
with awns barbed but achene body glabrous: B. glandulijera, B. aoraiensis, and
B. glabrata (p.p.). Bidens orojenensis and
B. glabrata (p.p.) lack barbs on awns but
SHERVVIN CARLQUIST
34
MEXICO
B. pilosa
(AMERICAN MAINLAND)
B.beckiono
^B. polycepholo
^ - ^ ^
B |grdimi
\
B.
cordifol.o
~v
' B. sp.
,NUKU HIVA
.UAPOU'
oHIVAOA
^
MARQUESAS is.
B.glandulifera-^
soaErr
Is.
W B. uapensis
B. henryi
B. collina
fsS ^
B. mathewsii
B hendersonensis vor. oenoensis
B. saintjohmano
B. hendersonensis
DISPERSAL IN PACIFIC COMPOSITAE
have, like B. aoraiensis, markedly spreading awns.
Lack of awns (except as ineffectual
vestiges) is shown by the Marquesan species of Bidens other than B. cordijolia. All
the Marquesan Bidens, however, have hairy
achenes (lateral margins only, except for
B. collina). Similarities in achene morphology and other features suggest origin of
Marquesan species from a single introduction. The presence of achene hairs alone
would probably suffice to secure transport
via bird feathers throughout that archipelago.
The two species from Raiatea show the
greatest loss of dispersibility among South
Pacific Bidens species, for they lack any
hairs or appendages (except for vestigial
awns and a few hairs near the achene summit in B. deltoidea).
Within the Hawaiian species of Bidens
(Fig. 2) a wide gamut in achene morphology can be seen. For purposes of illustration, these have been grouped into categories based upon features of morphology
related to potential dispersibility. Equipped
with all features related to dispersibility
are the four species of Fig. 2A. One of
these (B. hillebrandiana)
has spreading
awns, another (B. nematocera)
notably
long awns. Diminution of awn length is
shown by the species of Fig. 2B, whereas
those of Fig. 2C have awns of varying
length but lack achene-body hairs. Nearabsence of awns and restriction of hairs to
the base of the achene is shown in the
species of Fig. 2D; a few hairs occur near
the summit of the achene body in B.
cuneata and B. jorbesii. Lack of awn
vestiges combined with a hairy achene body
characterizes the species of Fig. 2E, which
thus resemble the Marquesan species. In
35
Fig. 2F are species which possess a glabrous
achene body and only the most vestigial
awns. Awn vestiges in B. menziesii and B.
mauiensis are flat and tooth-like, and are
seemingly functionless. A few hairs crown
the achenes of B. cosmoides, which are
otherwise glabrous.
The species of the upper half of Fig. 2
have no peculiarities not also seen in the
South Pacific species. In the lower half of
Fig. 2, however, are features of achene
morphology unique to the Hawaiian Islands. Achenes of B. campylotheca
(Fig.
2G) are glabrous, have vestigial awns, and
are bent into an arc. Achenes of B. degeneri have no awn vestiges (only callous
margins at the achene summit), and maybe straight or curved like those of B. campylotheca. More prominently curved are
the species of Fig. 2H, which often exhibit
achenes twisted into three or more helices.
Although achenes of these species are narrow in diameter, the embryo is relatively
large and occupies nearly the entire length
of the achene. The fact that several species
have curved or helical achenes suggests
that there is a selective value for this form,
but I can offer no suggestions as to the
possible function of this shape. Two of
these species (B. campylotheca, B. sandvicensis) occur on four of the major islands,
respectively. This suggests that this achene
form can be effective for dispersal over
relatively short distances. Achenes of the
species in Fig. 2H differ from each other in
minor respects.
The remaining Hawaiian species (Fig.
2I-L) have large flat achenes which can
sometimes be said to be winged, and which
have awns vestigial, lacking, or highly
modified. Bidens ctenophylla
(Fig. 21)
has awns vestigial or none, broad black
FIG. 1. Comparison of fruits of endemic species of Bidens from southeastern Polynesia, shown geographically. These species, together with those from Hawaii, represent the extent of this genus in the
Pacific. Inset, upper right, an unnamed species native to Socorro Island (Revillagigedos Islands,
Mexico). Inset, upper left, a typical achene of Bidens pilosa; this fruit probably approximates the
achene type ancestral to the Pacific species of Bidens. Bidens pilosa is native to mainland America.
Species omitted from this map include B. ahnei, which may be conspecific with B. polycephala. and
B. hivoanu, which is a synonym of Oparanthus albus.
SHERWIX CARLQUIST
36
\J
B. hillebrandiana
A
B
B. populifolia
B. molokaiensr
B. skottsbergii
B micrantha
B. micronthoides ' _ £ •
4,-,!
m
B. woimeana
B. amplectens
A
B. howanensis.P
A-J
M
B. fecunda
B. nematocera
distans
B. volida
\ f
B. stokesii
M
B. napaliensis
Q
B cuneala
L?
B. forbesii
w
B. asymmetnca
B. conjuncta . i
j
M
tli
coartata
£
B. puichella
Q
B. salicoides
B. perversa
F
/j5j
E osplenioides
ft
B. mauiensis
IsQ
B. cervicata
ft
B. cosmoides
flfk
B. grociloides
ft
B. compylotheco
B. clenophylla
B. sandvicensis
B. waiar.ensis
^1
B. fulvKceno
B. degeneri
I
' •Wl'
u
8. menziesii
B obtusilobo
«
1_>
Iff I
^ mocrocorpa
B. magnidisca
DISPERSAL IN PACIFIC COMPOSITAE
achenes which occasionally bear a tooth
along the lateral margins. Achenes of B.
wiebkei (Fig. 2J) are flat, have awns vestigial or none, and bear a few stiff hairs on
lateral margins of achenes. These hairs
have broad multeriate bases. Some achenes
of B. wiebkei are distorted in shape, and
have wavy lateral margins. Achenes of B.
macrocarpa (Fig. 2K) are the largest
among the Hawaiian species. They have
broad yellowish wings and are glabrous.
Flat awn-like appendages are modified in
conjunction with the formation of wings.
and are thus different from true awns. A
few hairs are present on these awn-like
appendages, but most hairs are not retrorse.
Achenes of B. magnidisca are like those of
B. ctenophylla, but have awns like those of
B. macrocarpa. with yellowish wing-like
webbing between the awns and the body of
the achene; or awns may be lacking, with
narrow yellowish wings at the achene summit. The species of Fig. 2I-L may be flat
enough to suggest transport by wind, but
such transport would be effeclive over only
short distances at best.
Ecology.—The
correlations between
achene morphology of insular Bidens species and ecology of these species, respectively, are remarkably clear. Of the South
Pacific species (Fig. 1), those with the best
dispersal mechanisms are all coastal species.
Bidens sain!johniana
occupies Marotiri,
which can be regarded as nothing more
than a few rocky islets inhabited by seabirds. Oeno and Henderson islands, occupied by B. hendersonensis, are small low
islands on which seabirds are abundant.
Bidens mathewsii grows on exposed sea
cliffs of Pitcairn Island. In other South
Pacific Bidens species more marked dimi-
37
nution of dispersibility has occurred: these
are the species which grow in localities inland from the immediate coast. Those species which have awns still barbed occupy
relatively low elevations in open scrubby
forest or on grassy ridges. The unnamed
Bidens from Socorro Island grows in open
dry forest at about 1000 feet. Bidens australis has been collected at low elevations
on Tahaa recently: it has not been collected recently on the other islands of its
range, a fact which may be due to destruction of low-elevation forests by burning
and farming. This may be true of B. lantanoides also. Bidens aoraiensis and B.
orojenensis are not low-elevation species,
but they grow on exposed rocky outcrops
of ridges high on Tahiti, and could not in
any way be considered forest species. Bidens
glandtdijera occurs on exposed grassy ridges
of montane Bora Bora.
The Marquesan species occupy grassy
ridges and open forest of mid-elevations, as
do B. dcltoides and B. moor een sis of the
Society Islands. Bidens raiateensis, with
the poorest dispersal means of any South
Pacific Bidens, grows in wet upland forest.
Thus, in the South Pacific species, the farthest penetrations into forest sites are accompanied by greatest diminution of dispersal mechanism.
Among the Hawaiian species, similar ecological correlations occur. The species of
Fig. 2A are occupants of seacoast bluffs
(B. hillcbrandii, B. molokaiensis, B. nematocera) or open, low-elevation slopes (B.
waimeana). The species of Fig. 2 B - F occupy open low-elevation sites except for B.
micranthoides (open places in mid-elevation
dry koa forest, Kauai), B. hawaiiensis
(open lava fields of Mauna Kea, Hawaii).
FK; 2. Comparison of fruits of endemic Hawaiian species of Bidens. The species are arranged
into groups according to achene morphology. The groupings shown are not intended to represent natural groups, although in some cases they do represent apparently closely related species. In each of
the groupings shown here, one achene is drawn in its entirety; achenes of the remainder within a
grouping agree in details of achene body, but differ in details of awns, etc., at the achene apex.
Achenes may differ in size, however, within a grouping. Some species of Bidens are variable in achene
morphology. A typical achene is represented for each. For further explanation, see text. Scale
(lower right) applies to all figures.
38
SHERVVIN CARLQUIST
0 . albus
MARQUESAS IS.
NUKU H I V A I
Fcuneato
Fcuneota
subsp. tohoaensis
F
1oh itensis
F
.
nu,ans
TUAMOTUS
^.V
COOK
TAHITI
IS
SOCIETY
IS.
0. rapensis
0. intermedius
0. coricceus
Bidens
pilosa
\ /
.
| 5 mm |
ff rapensis
F mangarevensis
FK;. 3. Comparison of fruits of all known species of Filchia and Oparanlhus, shown geographically.
Note that the scale of magnification is half that of the other figures in this paper. An achene of Bidens
pilosa is shown, bottom center, for comparison of size and morphology.
DISPERSAL IN PACIFIC COMPOSITAE
B. conjuncta (wet but open grassy sites of
higher Oahu and M a u i ) , and B. cosmoides
(rain forest of upper Kauai). The species
in the lower half of Fig. 2 could be said to
occupy relatively wet upland forests mostly.
These species, in general, have large embryos suited to germination in shady forest
sites. A notable exception is B. ctenophylla, Fig. 21 (open aa lava slopes near
Puuwaawaa, Hawaii). This might be a
case of a species which has been derived
from a forest-inhabiting ancestor rather
than a coastal one. Another exception is
B. degeneri (arid regions of lower Oahu,
Molokai, and M a u i ) .
Despite the exceptions as noted, the
progress of Hawaiian species of Bidens
from seacoast to interior and the accompaniment of this adaptive shift by loss of
dispersal mechanism seem clear. One can
hypothesize that Bidens originally established in coastal sites where seabirds nested.
The fact that many coastal areas of the
Hawaiian Islands are not now occupied by
seabirds is probably a result of human occupancy. There seems to be a degree of
flexibility in loss of dipersal mechanism.
Pacific species of Bidens are well suited
for entry into Polynesian forests because
they are perennial and can become large
shrubs up to 10 feet in height. They are
not, however, strongly woody and seem not
to have advanced as far toward arborescence as, for example, Fitchia. There seems
little question that the woodier species of
Bidens do represent specialization—adaptation to forest situations.
GENUS
OPARANTHUS
The four species of this genus (Fig. 3)
are native to the Marquesas and Rapa.
This recently recognized genus (Sherff,
1937a) may be considered a derivative
from an early immigrant stock belonging to
Heliantheae, subtribe Coreopsidinae. The
distribution of Oparanthus suggests relictism of a recent order or poor dispersal
capacity. Anatomical studies (Carlquist,
1957) show relationships to Coreopsidinae,
such as Bidens and Coreopsis, as well as
39
to Fitchia.
Oparanthus has fertile ray
achenes (at right in each species, Fig. 3)
and sterile (male) disk achenes. The disk
achenes—which have no function in reproduction—retain a single awn in O. rapensis
and O. coriaceus. The ray achenes are flattened (curved in 0. rapensis), although not
broadly winged. Awns are modified into
wing-like structures in O. albus, reminiscent
of Bidens macrocarpa or B. magnidisca.
Achene body and awns in Oparanthus are
completely glabrous. Thus, achenes have
a low dispersibility. This correlates well
with ecology and habit of Oparanthus. All
species are shrubs or trees, markedly
woody, and thus have a growth form frequent in Polynesian forests.
Oparanthus
seems to represent a more complete adaptation to the rain forest than does Bidens,
and probably represents a longer history
in Polynesia.
GENUS
FITCHIA
Fitchia (Fig. 3) seems related to Coreopsidinae, but by virtue of homogamous
capitula and other features seems worthy
of recognition within a separate subtribe,
Fitchiinae (Carlquist, 1957). Fitchia, like
Oparanthus, is a truly woody genus, ranging from small shrubs to true trees. We
may suspect that it, too, represents a South
American ancestor whose mainland relatives have now vanished. The achenes of
Fitchia are large. The smallest occur in
the Society Islands species; F. mangarevensis and F. rapensis have fruits of intermediate size, whereas F. speciosa has gigantic
achenes—probably the largest in the family Compositae. All species are provided
with long awns, although in all species except F. speciosa and F. cordata, the awns
are thin and break easily from the achene
body. Achenes in all species are denselycovered with upwardly appressed hairs, as
are the awns. The fact that awns do not
have retrorse barbs does not seem strongly
disadvantageous in dispersal. Fruits with
hairs oriented in a single direction disperse
quite well (e.g., grasses). The ancestors of
Fitchia seem likely to have been brought
K)
SHERWIN CARLQUIST
to Polynesia in feathers of birds, perhaps
seabirds.
One might speculate that the ancestors
of Fitchia had smaller fruits than do the
contemporary species. The relatively large
size of achenes in the Society Island species seems no severe disadvantage, for this
group has become distributed throughout
the high islands of this archipelago. The
southern pair of species, F. rapensis and F.
mangarevensis, are very close and their intermediate fruit size might have originated
in southeastern Polynesia. However, ancestors have evidently succeeded in dispersing
across the distance between Mangareva and
Rapa with this size of achene, for it is difficult to imagine that achenes of this size
could have evolved independently on the
two islands when other features of the two
species are so much alike. The large achene
size of F. speciosa seems likely to have
originated on Rarotonga from ancestors
with achenes like those of F. cordata. Not
only does F. speciosa represent the farthest
penetration of the genus into the Pacific,
it is also by far the most arborescent species in the genus. Fruit size correlates well
with size of plant in Fitchia. The Society
Island species are chiefly shrubs which occur in scrubby forest on ridges. Fitchia
rapensis and F. mangarevensis are small
trees, whereas F. speciosa is a tall tree of
deep forest. The large fruit size of F.
speciosa mimics perfectly the large fruits
characteristic of continental forest trees,
such as oaks or pines. It seems an ultimate
point in adaptation to rain forest.
Fitchia is a more primitive genus than
Bidens and Oparanthus and may have, during a longer occupancy in Polynesia, perfected this adaptation. Even if ancestors
of contemporary Fitchias had fruits as
large as those of the Society Island species,
subsequent increase in fruit size has undoubtedly occurred within Polynesia, and
this increase is correlated with degree of
entry into forest habitats—a farther penetration than that of Bidens. Although no
direct evidence can be adduced, one suspects that ancestors of Fitchia did arrive in
Polynesia by means of fruits smaller than
those of contemporary species, and perhaps
rather like those of Bidens. The absence
of Fitchia from the Marquesas, Hawaiian,
and Austral islands (other than R a p a ) , all
of which offer suitable ecological sites (F.
speciosa has escaped from cultivation on
Oahu) suggests its present dispersibility is
effective only over rather short ranges. A
feature not related to fruit size but probably deleterious to dispersibility is the nodding habit of the heads. Heads shatter at
maturity, but instead of exposing fruits
above foliage, tend to drop fruits beneath
the parent plant.
GENUS
SCALESIA
Endemic to the Galapagos Islands, Scalesia belongs to Heliantheae subtribe Verbesininae, and is probably closely related
to several verbesininoid genera, of which
Verbesina (Fig. 4A) can be considered an
example. Of the 18 species of Scalesia.
only a few (5. microcephala, Fig. 4B, 5.
pednnculata, S. darwinii, S. snodgrassii, S.
hopkinsii, S. cordata) have one or two
awns, usually rather short. Scalesia cordata
has the longest awns (1-2 mm.) but does
not seem markedly more dispersible than
the other species. T h e remainder of the
species have only small ''callous" pappus
rudiments (S. aspera, Fig. 4C) or none at
all. The achenes are not hairy. Hairs on
achenes of Compositae serve not only for
adhesion to dispersal agents, but for release
from the head as well. At maturity, achenebody hairs flex outward when dry, forcing
the fruits out of the head. Thus, achenes
of Scalesia have no device for release from
the heads. Moreover, involucres in Scalesia
remain closed at maturity, whereas those
of many other Verbesininae open out, facilitating release of achenes. There seems
little question that Scalesia is in the process
of losing means of dispersal. Its dispersal
throughout the archipelago might be due
to seed-eating birds. Some of the species
appear to have reached their present stations after loss of awns, because awnless
species occur on many of the islands. None
DISPERSAL IX PACIFIC
COMPOSITAE
41
5 mm
FIG. 4. Comparison of achenes of species of Heliantheae, subtribe Melampodiinae (E) and subtribe
Verbesininae (all others). All achenes are from disk flowers unless otherwise stated. Mainland species are included for purposes of comparison with the insular species. A, Vigitiera linearis (Mexican
mainland). B, Scalesia microcephala (Albemarle Island, Galapagos Islands). C, Scalesia aspera (Indefatigable Island, Galapagos Islands). D, Macraea laricifolia (Charles Island, Galapagos Islands). E,
Lecocarpus pinnatijidus
(Charles Island, Galapagos Islands); ray achene and the bract with peltate wing which permanently enfolds the achene are shown in longitudinal section. F, Zexmenia
brevifotia (Mexican mainland). G, Lipochaeta rockii (Molokai, Hawaiian Islands). H, Lipochaeta
remyi (Oahu, Hawaiian Islands). I. Lipochaeta lavarum (Maui, Hawaiian Islands).
of the species of Scalesia appear to retain
the same efficiency or means of dispersal
responsible for their original introduction
to the Galapagos. Adherence within bird
feathers seems a likely means for that introduction.
Adaptation to new ecological situations
does not appear to have played a dominant
role in loss of dispersibility. While species
retaining awns are in some cases upland
species, in others they are not: 5. cordata
is most abundant in the ''Transition Zone"
which is a dry scrubby region, and S. snodgrassS, which also retains awns, occurs in
open habitats near sea level. Scalesia pedunculata, which is an upland tree species,
has awns in some populations, none in
others. Presence and absence of awns may
be observed within a single head in some
species, such as S. microcephala (Harling.
1962). The Galapagos Islands do not, of
course, offer the broad ecological range of
the mountainous Polynesian islands. Loss
of dispersibility in Scalesia might be due to
a change in dispersal mechanism (eaten by
birds instead of carried within feathers) or
it might be the by-product of other evolutionary changes (selective value of dispersibility under island conditions insufficient to maintain continued production of
awns, for example).
GENUS
MACRAEA
Macraea, a monotypic genus from the
Galapagos (Fig. 4D) was formerly united
with Lipochaeta, but is worthy of recognition as a separate genus (Harling, 1962).
Harling claims that Macraea is close to
American species of Wedelia and Aspilia,
whereas Lipochaeta has been generally conceded to have close affinities to Zexmenia
(Sherff, 1935). All of these genera belong
to subtribe Verbesininae of Heliantheae,
and perhaps are not distant from one another. Macraea laricifolia occurs at lower
elevations of four of the larger Galapagos
Islands. Achenes are winged and crowned
by a circle of short scales. Outer achenes
in each head are broader and warty. The
dispersal mechanism seems to be one which
is suited to relatively short distances. A
few achenes in Macraea, however, have one
or two awns like those of Verbesina (Fig.
4A), and Macraea may be in the process of
losing these awns.
GENUS
LECOCARPUS
A genus which seems an excellent example of alteration to a new mode of dispersal
is the Galapagos endemic Lecocarpus, a
monotypic genus which belongs to Heliantheae, subtribe Melampodiinae. In this
genus, ray achenes are fertile; each is enclosed within a highly modified bract which
bears a peltate wing. The wing is thin and
leaf like. The two genera closest to Lecocarpus are Melampodium, in which fruitenclosing bracts are like an additional seed
coat and are not winged, and Acanthospermum, in which enclosing bracts bear
SHERWIN CARLQUIST
FIG. 5. Comparison of achenes of Heliantheae, subtribe Madiinae. All achenes shown are disk
achenes. A, Madia nutans (Californian mainland). B, Layia platyglossa (Californian mainland).
C, Wilkesia gymnoxiphium (Kauai, Hawaiian Islands). D, Argyroxiphium virescens (Maui, Hawaiian
Islands). E, Argyroxiphium sandwichense (Maui. Hawaiian Islands). F, Dubaulia latifolia (Kauai,
Hawaiian Islands). G, Dubaulia knudsenU (Kauai, Hawaiian Islands).
hooks (Hoffmann, 1890-1893). The morphology and weight of bract-enclosed fruits
of Lecocarpus suggest wind-dispersal across
relatively short distances. Lecocarpus occurs on two of the Galapagos Islands.
Either the winged bract has been evolved
since arrival of the ancestral stock, or
Lecocarpus has now vanished from the
American mainland. More study needs to
be devoted to this genus.
GENUS
LIPOCHAETA
Lipochaeta (Fig. 4 G - I ) contains 24 species endemic to the Hawaiian Islands. A
species alleged to be native to New Caledonia proves to be Wedelia uniflora (Harling, 1962). Lipochaeta characteristically
occupies lowland sites. Compared with
relatives such as Zexmenia (Fig. 4 F ) , species of Lipochaeta have relatively poor dispersal mechanisms. Although some, such
as L. remyi (Fig. 4G), retain both aristae
and wings on achenes, most have short
wings and no aristae (Fig. 4 H ) or no wings
and only a few short and ineffectual aristae
(Fig. 41). Unlike those of Zexmenia, Lipochaeta achenes lack hairs.
ARGYROXIPHIUM,
WILKESIA,
DUBAUTIA
AND
The genera Argyroxiphium,
Wilkesia,
and Dubautia (including Railliardia) must
be considered endemic Hawaiian representatives of Heliantheae, subtribe Madiinae.
the "tarweeds," which are otherwise restricted to western North America, with
one species in Chile. Some mainland tarweeds (Fig. 5A, B) have achenes with
pappus scales or plumose bristles, devices
which seem effective for wind dispersal or
adhesion within bird feathers or animal fur.
Other mainland tarweeds have fertile ray
achenes which have little pappus but which
are enfolded within sticky bracts, and thus
have the dispersal mechanism transferred
from the achene to the bract. This seems
true, to some extent, also of Argyroxiphium,
but few ray flowers are present in that
genus, and fertile disk achenes have no
such adherent bracts. All achenes of Wilkesia (Fig. 5C) are disk achenes, without adherent bracts. T h e scales which
crown the relatively gigantic achenes of
these genera seem relatively inefficient, but
might become caught in birds' feathers.
Hairs are few or absent on these achenes.
When mature, heads in these genera shatter,
and most fruits fall near the parent plant.
Wilkesia is restricted to Kauai, but Argyroxiphium occurs on both Maui and Hawaii. Both Wilkesia and
Argyroxiphium
seem gigantic versions of mainland tarweeds. With such relatively large fruits, an
exceptionally good dispersal mechanism
DISPERSAL IN PACIFIC COMPOSITAE
43
FIG. 6. Comparison of achenes of species of Cynareae (A-C) and Senecioneae ( D - G ) . Continental
species are shown for comparison with insular ones; all achenes are disk achenes. Where pappus
bristles or setae are shown to the right and above the achene ( A - E ) . a portion of the setae are caducous. In B, C, and E. only a small number of setae remain attached to the achene at maturity. In D.
some setae may break from the fruit, but the majority are persistent. Setae are shown as broken
lines in B and C because they tend to crumble easily and are rarely intact on mature achenes. A, Centaurea tweediei (Argentina). B, Centaurodendron dracaenoides (Masatierra, Juan Fernandez Islands).
C, Yunquea tenzii (Masatierra, Juan Fernandez Islands). D, Senecio macrantha (Australia). E, Robinsonia gayana (Masatierra, Juan Fernandez Islands). F, Rhetinodendron berterii (Masatierra, Juan
Fernandez Islands). G, Symphyochaela macrocephala (Masatierra, Juan Fernandez Islands).
would be required if long-distance dispersibility were to be retained. Thus, dispersibility of Argyroxiphium and Wilkcsia appears to have fallen well behind that of
ancestral forms.
Dubautia (Fig. 5F-G) has retained
better dispersal mechanisms, owing to possession of the plumose-bristle type of pappus. This genus may be related closely to
the preceding two, and a hybird between
Argyroxiphium and Dubautia has been reported. Some species, such as D. knudsenii
(Fig. 5G), do, however, have pappus rather
short in proportion to achene; these bristles (or scales) do not spread well at maturity. Moreover, D. knudsenii has relatively few achene-body hairs, and bracts
of the head do not reflex well to release the
achenes; the heads are nodding and tend
to drop achenes to the ground. Appropriately, D knudsenii is a rain forest tree,
whereas species with longer bristles and
capitals which open better, such as I). latifolia (Fig. 5F), are characteristic of exposed situations. There does appear to be
dispersal diminution in relation to ecology
in Dubautia, and the genus may be in a
state of transition in this respect. Better
dispersibility in Dubautia may account for
the fact that the genus occurs the length of
the chain of major Hawaiian Islands, and
has more species (27) and has occupied
more diverse ecological sites than Argyroxiphium or Wilkesia.
JUAN FERNANDEZ CYNAKKAK
The single entry of the tribe Cynareae
into the Pacific is a pair of closely related
genera, Centaurodendron (two species)
and Yunquea (one species), endemic to the
Juan Fernandez Islands. These genera are
probably closely related to Centaur ea sect.
Plectocephalus, which have fruits as shown
in Fig. 6A. In such Centaureas, pappus
bristles are numerous, strong, and although
somewhat caducous, are often well retained
on the achene summit. Pappus bristles
spread outward well when dry, forcing mature achenes out of the head. Bracts of the
head reflex well at maturity, liberating
achenes. In Centaurodendron (Fig. 6B),
on the contrary, bracts of the involucre do
44
SHERVVIN CARLQUIST
not reflex appreciably, so that fruits are
well retained within the head. Furthermore, achenes bear relatively short pappus
bristles which "come off at the slightest
touch, and there can be little doubt that
they are caducous and useless for the
transport of the achene" (Skottsberg,
1938). Moreover, the bristles break easily
into tiny fragments. Mature fruits are
even larger than those of Centaurca, and
would require an even more efficient pappus to equal dispersibility of Centaurea.
On all counts, Centaurodendron
exhibits
greatly depreciated dispersibility. Fruits
are released only with gradual weathering
of heads. Achenes of Yunquea (Fig. 6C)
are similar to those of Centaurodendron in
all respects except that they have an oddly
convolute surface and are irregular in
shape—features which would certainly
hinder dispersal. Such distortion of the
achene may represent mutations which, although they would be unfavorable in a
species with an efficient dissemination
mechanism, are not deleterious in Yunquea,
where dispersibility is already minimal.
Centaurodendron
and Yunquea are rosette trees of rain forest. This adaptation is
an unusual one for Cynareae, and the adaptation of these genera appears to be "complete."
J U A N FERNANDEZ SENECIOXEAK
Also native to the Juan Fernandez Islands are three closely-related endemic
genera of Senecioneae: Robinsonk
(Fig.
6 E ) , with five species;
Rhetinodendron
(Fig. 6 F ) , with one species; and Symphyochaeta (Fig. 6 G ) , with one species. Typical mainland Senecioneae (Fig. 6D) have
many fine pappus setae which are longer
than the achene body, persistent to somewhat caducous, and which flex outward at
maturity, both liberating achenes from the
head and providing a parachute-like structure for wind-flotation. Release of achenes
is aided by reflexing of involucral bracts.
In Robinsonia (Fig. 6 E ) , achenes are relatively small, but the pappus bristles are
notably short and clearly caducous. Ma-
ture female heads of Robinsonias are cupules which do not open well and which
contain nearly bare fruits which are dispersed mainly by falling from the head.
Rhetinodendron
(Fig. 6F) has female
heads with narrow involucres usually containing only three fruits each. The achenes
are hairy, but these hairs do not spread
outward appreciably upon drying, and
achenes tend to remain within the involucres. The short pappus bristles, although persistent, do not reflex but are
permanently upright. Involucres open only
a little at maturity.
In Symphyochaeta
(Fig. 6G), female
achenes are notably large. The achenes are
not hairy, and ridges on the achene body
are knobby. The pappus bristles are short,
and united into a tube, thus definitively
cancelling any possibility of wind dispersal.
Involucral bracts do not reflex at maturity.
All three Juan Fernandez genera of Senecioneae show advanced stages of dispersal
loss. As in Centaurodendron and Yunquea,
this alteration is accompanied by entry into
the rain forest habitat, an ecological preference of relatively few Senecioneae.
CICHORIEAE
The achene of Stephanomcria
juncea
(Fig. 7A) serves as typical of mainland
relatives of insular Pacific Cichorieae. The
achene is crowned by numerous pappus
setae; these reflex at maturity, as do involucral bracts. The setae in this species
are persistent. Also belonging to the subtribe Stephanomeriinae is
Munzothamniis
blairii (Fig. 7B), an endemic of southern
California's offshore island of San Clemente. This species does not seem to show
appreciable diminution of dispersibility.
However, pappus setae are more caducous
than in many Cichorieae.
Munzothaninus
is a shrub of exposed slopes.
The remainder of the genera of Cichorieae shown in Fig. 7 belong to subtribe
Dendroserinae, a subtribe conceded to be
close to Stephanomeriinae (Stebbins, 1953).
Thamnoseris is a genus of shrubs, probably
monotypic, native to the Desventuradas
DISPERSAL IX PACIFIC COMPOSITAE
I
5 mm
45
|
FIG. 7. Comparison of achenes of species of Cichorieae. A mainland species (A) is included for
comparison with insular species. A, Lygodesmia juncea (southwestern United States). B, Munzotkamnus blairii (San Clemente Island, California). C, Thamnoseris lacerata (San Ambrosio Island, Desventuradas Islands). D, Hesperoseris gigantea (Masafuera, Juan Fernandez Islands). E, Rea micrantha
(Masatierra, Juan Fernandez Islands). F, Phoenicoseris pinnata (Masatierra, Juan Fernandez Islands).
G, Dendroseris litoralis (Santa Clara, Juan Fernandez Islands).
Islands (Chile), where it grows on barren
lava slopes (Skottsberg, 1963). Involucres
of Thamnoseris are composed of bracts
with large swollen bases; involucres apparently do open when mature and dry to
release fruits (Skottsberg, 1947). Pappus
setae are very short and ''fall very easily"
from the achene body (Skottsberg, 1937).
These pappus features suggest loss of dispersibility.
T h e Juan Fernandez Cichorieae (Fig.
7 D - G ) form an unusually rich assemblage
which Skottsberg (1951) concluded were
best regarded as four genera, although
Stebbins (1953) has demurred. Of the four
genera, Hesperoseris (Fig. 7D) is monotypic, and possesses achenes with relatively
good potential dispersal ability, although
pappus setae are relatively caducous. Rea,
a genus of three species, has less dispersibility. In R. micrantha (Fig. 4 E ) , pappus
setae are very short and caducous. Involucral bracts are not caducous, although
they are in R. pruinata.
Phoenicoseris pinnata has achenes which
are oddly angled, sometimes triquetrous,
but which have strikingly convolute surfaces (Fig. 7 F ) . This, in part, is related
to embryo volume, but the considerations
cited above for Yunquea may also apply
here. Pappus setae in P. pinnata are too
short to offer appreciable lift to achenes released into the air. Setae do not reflex well
in mature dry achenes and are quite caducous. Moreover, mature involucres open
poorly when dry, and seeds are tardily and
irregularly released.
All four species of Dendroseris
have
achenes with fantastically sculptured flatfish achenes (Fig. 7G). The shape and
size of these is related to the embryos with
broad cotyledons (Skottsberg, 1922). In
their large size, fruits of Dendroseris represent an evolutionary culmination in Cichorieae comparable to that of Fitchia speciosa in Heliantheae. Also contributing to
minimized dispersibility in Dendroseris is
the tuft of short, caducous pappus setae,
far too short and too few either to release
achenes from the involucre or to aid their
aerial flotation. Achenes are, in fact, retained at maturity within the well-closed
involucres composed of massive involucral
bracts. The reasons for highly contorted
achene shapes may be the same as those
given above for Yunquea.
The aboriginal vegetation of the Juan
Fernandez Islands may be said to have
been scrubby at lower elevations, rain forest above and in canyons. Most of the
Juan Fernandez Cichorieae inhabit rain
forest areas. Palm-like habits in Phoenicoseris and Dendroseris suggest the high
degree of adaptation to rain forest. Not
in rain forest are Rea pruinata and Dendroseris litoralis; the former—which is pro-
SHERWIN CARLQUIST
46
vided with relatively good dispersal means
—grows in coastal valleys, the latter on
grassy offshore islets of the Juan Fernandez. The habit and fruit morphology of
Dendroseris suggests adaptation to rain
forests, so D. lit oralis may be a secondaryentry into drier lowlands situations.
ASTEREAE
Endemic genera of Astereae in the Pacific
include Remya and Tetramolopium
(Hawaiian Islands), and
Darwiniothamnus
(Galapagos Islands). In addition, endemic
species of Erigeron occur on the Juan
Fernandez Islands. Achenes of Remya
bear a few vestigial bristles, and are doubtless an instance of loss of dispersibility.
Remya is difficult to discuss in this regard,
however, because its affinities cannot be
expressed in terms of particular genera yet.
Remya is a rain forest shrub. Many species of Tetramolopium seem to have a typical astereoid pappus, but at least one species (T. remyi) has few setae. Tetramolopium characteristically forms small perennial shrubs in open habitats. Darwiniothamnus and the Juan Fernandez species
of Erigeron appear to have normal pappus.
Darwiniothamnus can be said to grow on
pioneer habitats.
SUMMARY
AND
CONCLUSIONS
Loss of dispersibility in Pacific Compositae seems strongly dependent on (a)
ecological opportunity, (b) relative amount
of time available on islands for evolution,
and (c) cessation of the relationship between dispersal mechanism and dispersal
agent possessed by ancestors.
The contrast between the Galapagos Islands and the Juan Fernandez Islands is
interesting. The Juan Fernandez Islands
can be said to consist mostly of dry forest
and rain forest (aboriginally), so that immigrant Compositae have adapted to situations other than those characteristic of
Compositae, such as dry lowland open
habitats. Loss of dispersibility has characterized the various groups of Juan Fernandez Compositae quite clearly. Extreme
cases of loss of dispersibility, such as these,
feature increase in fruit size without concomitant increase in appendages which serve
in dissemination, diminution or malformation or loss of those appendages, and alteration in mechanism of release of fruits.
Thus, in addition to loss of dispersibility
of the fruits themselves, the involucres may
become incapable of opening at maturity,
and pappus or achene-body hairs which
facilitate ejection of achenes from the mature head may be altered or lost.
In the Galapagos Compositae, Scalesia
may be said to show decrease in dispersibility, but not in a spectacular fashion. No
true rain forest may be said to exist on
the Galapagos Islands, and most of the
islands can be said to consist of pioneer
habitats. The Galapagos Islands may be
geologically newer than the Juan Fernandez,
although data on this point are lacking.
For these reasons, Galapagos Compositae
would be expected to show relatively little
change from ancestral types compared to
Juan Fernandez Compositae.
The Pacific species of Bidens also confirm sensitively the guiding role of ecology
in diminution of dispersal devices. In
Bidens, only the species in pioneering sites
near the immediate coastline retain a fully
developed dispersal apparatus, and the degree of modification is roughly proportionate to adaptation to stable rain forest.
Such adaptation brings with it specialization with respect to particular conditions
and localities of small geographical extent,
where high dispersibility may be disadvantageous or of neutral selective value, and
a dispersal mechanism may fail to be maintained in the history of a species. The
Pacific Bidens species as a whole appear
rather recent, and Oparanthus and Fitchia,
which have become truly woody and show
clear habital adaptation to rain forest, may
have antedated Bidens in Polynesia.
Increase in fruit size as a way of adaptation to shady forest conditions by virtue
of a larger food storage on which seedlings
can draw seems an operative factor in loss
of dispersibility. Fitchia speciosa, the most
DISPERSAL IN PACIFIC COMPOSITAE
truly arboreal of Pacific Compositae, has
gigantic fruits which are probably the largest in Compositae.
Diminution of dispersibility may be expected to follow rapidly if the mode of dispersal characteristic of a species is suddenly changed. This event seems likely to
happen following immigration to islands—a
rather drastic occurrence. Either a species
changes to a new mode of dispersal (e.g.,
Lecocarpus) is maintained by a slower pace
of dissemination, or dispersal is achieved
merely by direct fall of fruits. In the
latter case "unfavorable" mutations altering fruit morphology can be tolerated. This
may account for grotesque fruits of Yunquea, Phoenicoseris, Dendroseris, and some
species of Bidens.
Detailed comparisons of morphology of
achenes with ecological requirements seem
to provide multiple confirmations of the
validity of the theory outlined in the beginning of this paper. Some degree of flexibility in correlations between achene morphology and ecology of a species may be
expected. A dispersal mechanism may be
in a state of transition, or a poorly functioning dispersal apparatus may be retained because it does not have a strongly
negative selective value.
Loss or alteration of dispersal mechanism is an evolutionary curriculum by no
means restricted entirely to oceanic islands.
Some of the principles outlined above may
apply to some mainland situations which
are insular in character (e.g., mountaintops), or other areas which offer ecologically specialized conditions. The totality
of patterns presented in this paper may be
said to be characteristic of oceanic islands,
however.
LITERATURE CITED
BROWN, F. B. H. 1935. Flora of southeastern
Polynesia. III. Dicotyledons. Bull. Bishop
Mus., 130: 1-386.
CARLQUIST, S. 1957. The genus Fitchia (Compositae). Univ. California Publ. Bot., 2 9 : 1144.
. 1959. Studies on Madinae: anatomy,
cytology and evolutionary relationships. Aliso,
4 : 171-236.
. 1965. Island life: a natural history of
-47
islands of the world. Natural History Press,
New York, 451 pp.
. 1966. The biota of long distance dispersal.
I. Principles of dispersal and evolution. Quart.
Rev. Biol, (in press).
CARLQUIST, S., AND M. L. GRANT.
1963.
Studies
in Fitchia (Compositae): novelties from the
Society Islands; anatomical studies. Pac. Sci.,
17: 282-298.
DARLINGTON, P. J.,
JR.
1943.
Carabidae
of
mountains and islands: data on the evolution
of isolated faunas, and on atrophy of wings.
Ecol. Monogr., 13: 37-61.
HARLINC, G. 1962. On some Compositae endemic to the Galapagos Islands. Acta Horti
Bergiani, 2 0 : 63-120.
HOFFMANN, O. 1890-1893. Compositae. In, A.
Engler and K. Prantl (eds.). Die natiirlichen
Pflanzenfamilien, 4 ( 5 ) : 87-402. Verlag William Engelmann, Leipzig.
HOWELL, J. T. 1941. The genus Scalesia. Proc.
California Acad. Sci., 4th ser., 2 2 : 221-271.
PAI-Y, H. R. 1954-1955. Tahiti et les Ues voi
sines. La vegetation des Ues de la Societe ct
de Makatea. Trav. Lab. Forest. Toulouse,
5 ( 2 : 1 ) : 1-386.
SHERFF, E. E. 1935. Revision of Telramolopium,
Lipochaeta, Dubaulia and Railliardia. Bull.
Bishop Mus., 1 3 5 : 1-136.
. 1937a. Some Compositae of southeastern
Polynesia (Bidens, Coreopsis, Cosmos, and
Oparanthus). Occ. Papers Bishop Mus., 12
(19): 1-19.
. 1937b. The genus Bidens. Part I. Field
Mus. Nat. Hist., Bot. Ser., 16: 1-346.
SKOTTSBERG, C.
1922.
The phanerogams of the
Juan Fernandez Islands. Nat. Hist. Juan Fernandez and Easter Islands, 2 : 95-240.
. 1937. Die Flora der Desvcnturadas-Inseln.
Goteborgs Kungl. Vetensk.-O. Vitterh.-Samh.
Handl., ser. B, 5 ( 6 ) : 1-87.
. 1938. On Mr. C. Bock's collection of
plants from Masatierra (Juan Fernandez), with
remarks on the flowers of Centaurodendron.
Acti Horti Gotoburgensis, 12: 361-373.
. 1947. Eine kleine Pflanzensammlung von
San Ambrosio (Islas Desventuradas, Chile).
Acti Horti Gotoburgensis, 17: 49-57.
. 1951. A supplement to the pteridophytes
and phanerogams of Juan Fernandez and
Easter I. Nat. Hist. Juan Fernandez and
Easter Islands, 2 : 763-792.
. 1953. The vegetation of the Juan Fernandez Islands. Nat. Hist. Juan Fernandez and
Easter Islands, 2 : 793-960.
. 1954. A geographical sketch of the Juan
Fernandez Islands. Nat. Hist. Juan Fernandez
and Easter Islands, 1: 89-192.
. 1958. Uber Yunquea tenzii Skottsb. Ber.
Deutsch. Bot. Ges., 7 1 : 45-52.
. 1963. Zur Naturgeschichte der Insel San
48
SHERWIN CARLQL 1ST
Ambrosio (Islas Desventuradas, Chile). 2.
Bliitenpflanzen. Arkiv for Bot., ser. 2, 4 : 465488.
STEBBINS, G. L., J R .
1953.
A new classification
of tribe Cichorieae, family Compositae. Madrono, 12: 33-64.
STEWART, A. 1911. A botanical survey of the
Galapagos Islands. Proc. California Acad. Sci.,
4th ser., 1: 7-288.
WrxsoN, E. O. 1959. Adaptive shift and dispersal in a tropical ant fauna. Evolution, 1 3 :
122-144.
ZIMMERMAN, E. C. 1948. Insects of Hawaii. I.
Introduction. Univ. Hawaii Press, Honolulu.