parrot), the Auckland Islands teal (Anas auck-
landica), the Galápagos cormorant (Phalacrocorax
harrisi), the three Hawaiian ibises (Apteribis), and
four Hawaiian ducks (Anas); these 13 species repre-
senting at least eight evolutionary transitions.
Taking all these factors into account, Roff con-
cluded that there is an insular tendency to flight-
lessness, although this could only be formally
shown for the rail group. A separate analysis by
McCall (1998) records that flightlessness has
evolved independently in at least eleven extant bird
families, eight of which currently contain both
volant and non-volant species. They noted that the
volant species in these families tend to be charac-
terized by relatively short wings. As short-winged
species pay a relatively high energetic cost to fly,
their findings support the hypothesis that the ener-
getic costs of flight provides a key factor in select-
ing for flightlessness in these families.
On the topic of flightless rails, Diamond (1991a)
reported a newly discovered species, Gallirallus
rovianae, which is believed to be either flightless or
weak-flying. It was collected in 1977 from New
Georgia, Solomon Islands. The ancestral form is
volant and more boldly patterned than the new
species. Diamond regards G. rovianaeand its rela-
tives on other oceanic islands as cases of convergent
evolution, whereby 11 groups of rails have inde-
pendently evolved weak-flying or flightless forms
and also, in several cases, reduction in boldly pat-
terned plumage, on widely scattered oceanic
islands. These weak- or non-flying rails do not fare
well in contact with people, and as many as half of
those extant at the time of European discovery have
subsequently become extinct. Moreover, subfossil
remains of other extinct species have been found on
most of those Pacific islands that have been
explored by palaeontologists, and it is likely that
there are many more extinct forms awaiting discov-
ery (Steadman 1997a). There is thus clearly a strong
selection pressure for reduced flying ability in rails,
and this appears to be related to the energetic and
weight burden of unnecessary flight muscle on
islands lacking terrestrial mammalian predators.
From the plant world, the genus Fitchia(in the
Asteraceae), endemic to the Polynesian islands
of the south Pacific, exemplifies reduction in
dispersability (Carlquist 1974; Williamson 1981). Its
closest continental relatives are herbs with spiked
fruits transported exozoically (i.e. by external
attachment to an animal), but in Fitchiathe spikes
are relict, the fruits are much larger, the plants are
trees, and their fruits drop passively to the forest
floor. Endemic Pacific island species of Bidens(also
in the Asteraceae) illustrate various steps in the loss
of barbs and hairs and other features favouring dis-
persiveness, these changes being accompanied by
ecological shifts from coast to interior and, in some
cases, wet upland forest habitats (Carlquist 1974;
Ehrendorfer 1979). It might be argued that these
examples suffer from a certain circularity of reason-
ing, with traits being used to infer the reasons for
their selection. To allay such concerns, a demon-
stration is needed of such changes actually occur-
ring. Once more, the Asteraceae provide the
evidence.
Cody and Overton (1996) monitored populations
ofHypochaeris radicataandLactuca muralis, both
wind-dispersed members of the Asteraceae, on 200
near-shore islands in Barkley Sound (Canada). The
islands ranged from a few square metres in area up
to about 1 km^2. Significant shifts in diaspore mor-
phology occurred within 8–10 years of population
establishment, the equivalent in these largely bien-
nial plants of no more than five generations. The
diaspore in these species consists of two parts, a
tiny seed with a covering (the achene), which is sur-
rounded by or connected to a much larger ball of
fluff (the pappus). The clearest findings were for
Lactuca muralis, the species with the largest sample
size. Founding individuals had significantly
smaller seeds (by about 15%) than typical of main-
land populations, illustrative of a form of a non-
random founder effect termed immigrant selection
(Brown and Lomolino 1998, p. 436). Seed sizes then
returned to mainland values after about 8 years,
but pappus volumes decreased below mainland
values within about 6 years. If the pappus is
thought of as a parachute and the seed the payload,
then the ratio between the two indicates the dis-
persability of the diaspore (Diamond 1996).
Figure 7.2 provides a schematic representation of
the trends in dispersability. Cody and Overton
were able to quantify these phenomena becauseNICHE SHIFTS AND SYNDROMES 183