Content uploaded by Laura Vary
Author content
All content in this area was uploaded by Laura Vary on Mar 19, 2014
Content may be subject to copyright.
271
Int. J. Plant Sci. 175(3):271–286. 2014.
䉷2014 by The University of Chicago. All rights reserved.
1058-5893/2014/17503-0001$15.00 DOI: 10.1086/674452
INCIDENCE, CORRELATES, AND ORIGINS OF DIOECY IN
THE ISLAND FLORA OF NEW CALEDONIA
Mark A. Schlessman,
1,
* Laura B. Vary,* Je´roˆme Munzinger,
2,
† and Porter P. Lowry II ‡
,
§
*Department of Biology, Box 187, Vassar College, 124 Raymond Avenue, Poughkeepsie, New York 12604-0187, USA; †Institut de Recherche
pour le De´ veloppement (IRD), Unite´ Mixte de Recherche (UMR) botAnique et bioinforMatique de l’Architecture de Plantes (AMAP),
Laboratoire de Botanique et d’E
´cologie Ve´ge´tale Applique´es, Herbarium Noume´a, Noume´a F-98848, Nouvelle-Cale´donie;
‡Missouri Botanical Garden, PO Box 299, St. Louis, Missouri 63166-0299, USA; and §Muse´ um National
d’Histoire Naturelle, De´partement Syste´matique et E
´volution (UMR 7205),
CP 39, 57 rue Cuvier, 75231 Paris Cedex 05, France
Editor: Lynda F. Delph
Premise of research. Because it is an inherently risky sexual system, dioecy is globally rare. Attempts to
explain unusually high incidences of dioecy on certain islands have generated a considerable literature on the
relationships among dioecy, its ecological correlates, establishment after transoceanic dispersal, and postdis-
persal speciation. Nevertheless, few studies of dioecy on islands have included considerations of the origins
and maintenance of dioecy on islands along with determinations of its incidence.
Methodology. We used the literature, herbarium specimens, and fieldwork to determine the incidence of
dioecy in the native angiosperm flora of New Caledonia. We inferred thenumber and characteristics of colonists
needed to account for the extant dioecious flora. We made traditional species-based numerical assessments of
associations between dioecy on New Caledonia and woodiness, plain flowers, fleshy fruit, habitat, and en-
demism, and we constructed a phylogenetic tree for New Caledonia’s native angiosperms to investigate cor-
related evolution of dioecy and those associated traits.
Pivotal results. This study is the first comprehensive survey of sexual systems for the flora of New
Caledonia. One-fifth of New Caledonia’s native angiosperms are dioecious. Dioecy is numerically overrep-
resented among species that are woody, have plain flowers, have fleshy fruit, occur in rainforest, or are endemic.
However, we found strong evidence for correlated evolution only for dioecy and woodiness, plain flowers,
and fleshy fruit. Dioecious groups with more of the widely accepted morphological correlates of dioecy tend
to be more speciose. Approximately 90% of the colonists that gave rise to the extant dioecious flora were
themselves dioecious. Approximately 60% of the colonists have two or more dioecious descendants, and those
descendants comprise more than 90% of the extant dioecious species.
Conclusions. Successful dispersal and establishment of already dioecious colonists and autochthonous
speciation of dioecious lineages are primarily responsible for the high incidence of dioecy on New Caledonia.
There were relatively few postdispersal transitions to dioecy. The associations of dioecy with woodiness, plain
flowers, and fleshy fruit result from correlated evolution that occurred prior to dispersal to New Caledonia,
while the associations of dioecy with rainforest habitat and endemism appear to result from autochthonous
speciation of dioecious lineages. With ∼4% of the world’s dioecious species occurring only there, New Cal-
edonia should be a rich source of new information on the evolutionary ecology of dioecy. Realization of this
potential will require both further study and concerted efforts to preserve the native flora.
Keywords: dioecy, New Caledonia, island floras, Baker’s law.
Online enhancements: supplementary tables, appendixes.
Introduction
Dioecy, in which an individual is either female or male, is
the most extreme form of sexual specialization in angiosperms.
1
Author for correspondence; e-mail: schlessman@vassar.edu.
2
Present address: IRD, UMR AMAP, la Recherche Agronomique
pour le De´veloppement (CIRAD) TA A51/PS2, 34398 Montpellier
Cedex 05, France.
Manuscript received May 2013; revised manuscript received October 2013; elec-
tronically published February 17, 2014.
While dioecy has the unique advantage of completely pre-
venting self-pollination and fertilization, it is also requires
transfer of pollen from one particular type of individual to
another, making it inherently risky. In addition to the risks of
obligate cross-pollination, dioecy imposes several other evo-
lutionary handicaps. Compared to hermaphroditism, dioecy
adversely affects seed dispersal (Heilbuth et al. 2001), colo-
nizing ability (Baker 1955), and the evolutionary success of
clades (Heilbuth 2000). Consequently, only 4% (Richards
1986) to 6% (Renner and Ricklefs 1995) of the world’s an-
giosperms are dioecious.
272 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Along with being globally rare, dioecy is unevenly distrib-
uted with respect to the ecological characteristics and the bio-
geography of angiosperms. Dioecious species tend to be trop-
ical and woody and to have small, plain flowers and fleshy
fruits (Renner and Ricklefs 1995; Sakai and Weller 1999), and
the evolutionary success of dioecious clades compared to non-
dioecious sister clades increases with the number of these four
major ecological correlates present in the dioecious clades (Va-
mosi and Vamosi 2004).
Continental temperate floras generally have lower incidences
of dioecy, whereas woody tropical floras and the floras of
certain islands have the highest (Adam and Williams 1994;
Sakai and Weller 1999; Carpenter et al. 2003; Gross 2005;
Matallana et al. 2005; Vary et al. 2011). With respect to is-
lands, Hawaii received early attention because the high inci-
dence of dioecy there appeared to contradict Baker’s law that
hermaphroditic, genetically self-compatible species should be
the most successful at colonizing islands via long-distance dis-
persal (Baker 1955; Gilmartin 1968). Proposed explanations
for such exceptions to Baker’s law have included the hypoth-
eses that postarrival conditions such as inbreeding in small
colonizing populations favored the autochthonous evolution
of dioecy on islands (Carlquist 1966; Baker 1967) and that
certain ecological correlates of dioecy, such as fleshy bird-dis-
persed fruit, favored transoceanic long-distance dispersal of
dioecious colonists to islands (Bawa 1980). Dioecy on islands
continues to play a significant role in both empirical and the-
oretical studies on the evolution of mating systems and dis-
persal syndromes in plants (Baker and Cox 1984; Thomson
and Brunet 1990; Pannell and Barrett 1998; Sakai and Weller
1999; Barrett 2006; Pannell 2006; Cheptou and Massol 2009;
Cheptou 2012).
Relatively few studies have gone beyond determinations of
the incidence of dioecy to examine both the origins and evo-
lutionary ecology of dioecy in island floras. Those that have
were focused on small, geologically young volcanic islands
with relatively few native species (e.g., Hawaii: Sakai et al.
1995a, 1995b; Juan Fernandez Islands: Bernardello et al.
2001) or on specific components of much larger floras (e.g.,
rainforest trees of New Caledonia: Carpenter et al. 2003; lit-
toral forest of Madagascar: Vary et al. 2011). With the notable
exception of New Zealand (Webb et al. 1999), there are no
comprehensive studies for geologically older islands with larger
floras. Here we examine dioecy in the native flora of New
Caledonia, which includes a geologically ancient fragment of
Gondwana and has an extremely species-rich, highly endemic
indigenous flora.
In an early analysis of dioecy on islands, Baker and Cox
(1984) found very strong positive correlations between the
incidence of dioecy and two factors: proximity to the equator
and maximum island height, with island height as a proxy for
mountainous terrain with moist forests (Baker and Cox 1984).
New Caledonia has an equable tropical climate and consid-
erable topographic diversity, with mountains rising to 1628
m. It supports a predominantly woody flora with origins dating
back to the late Eocene emergence of Grande-Terre and has
extensive areas of dense evergreen rainforest (Jaffre´ et al. 2001;
Lowry et al. 2004). Thus, one might expect a high incidence
of dioecy there, especially among rainforest species and en-
demics. Indeed, in their comprehensive review of dioecy in
angiosperms, Renner and Ricklefs (1995) noted that a number
of entirely or partially dioecious genera were endemic to New
Caledonia. While earlier works report high incidences of
dioecy for particular taxa (Schlessman et al. 1990a, 2001) or
ecological groups (Carpenter et al. 2003), prior to this study
there has been no attempt to determine the incidence of dioecy
or any other sexual system for the entire flora. Here we ask:
Is there a high incidence of dioecy in the native angiosperm
flora of New Caledonia? Does the flora of New Caledonia
provide evidence for correlated evolution of dioecy and wood-
iness, plain flowers, fleshy fruit, rainforest habitat, and en-
demism? Does the success of a dioecious group on New Cal-
edonia increase with the number of ecological correlates it
possesses? Were most of the colonizing species that established
the extant dioecious flora already dioecious, or were they non-
dioecious as predicted by Baker’s law? What is the relative
importance of frequent colonization versus postcolonization
speciation in accounting for the current number and systematic
distribution of dioecious species on New Caledonia? Is there
any evidence for dispersal of dioecy from New Caledonia to
elsewhere? To address these questions, we have determined the
sexual systems of the native species, inferred the ancestry and
origins of the dioecious component of the flora, and analyzed
the entire flora to seek evidence for correlated evolution of
dioecy with endemism, rainforest habitat, woodiness, plain
flowers, and fleshy fruit.
Methods
Study Area
New Caledonia is a tropical archipelago spanning the lati-
tudes 18⬚00
Sto23⬚50
S and situated Ⳳ1200 km E of Australia
and Ⳳ1500 km NNE of New Zealand. The main island of
Grande-Terre (16,595 km
2
) is Gondwanan, having separated
from Australia and New Zealand as early as 85 Mya (Willford
and Brown 1994; Lee et al. 2001). The Loyalty Islands, Isle
of Pines, and several smaller islands add another 2377 km
2
.
New Caledonia is the smallest biodiversity hot spot (Lowry et
al. 2004), with more than 3150 indigenous angiosperms, of
which 77%–79% are endemic (Lowry 1998; Lowry et al.
2004; Morat et al. 2012). This flora is both larger and less
well known than those of Hawaii and New Zealand (table 1).
Approximately 40% of the recognized species have not re-
ceived recent (post-1967) taxonomic treatment (Jaffre´etal.
2001), and perhaps 300 or more species are not yet formally
recognized (Morat 1993; Swenson et al. 2007). The current
checklist of the native flora (Morat et al 2012) includes 3053
recognized angiosperm species, some of which are as yet
unpublished.
Many New Caledonian angiosperm groups have both dis-
junct distributions involving other fragments of Gondwana
and generalized (plesiomorphic) traits that led systematists to
characterize them as ancient. Botanists and biogeographers
have debated the relative importance of vicariance (Raven and
Axelrod 1972) and long-distance dispersal (Carlquist 1966) as
explanations for the present distributions of these ancient
Gondwanan taxa. The geologic record shows that after sep-
aration from Australia and New Zealand, Grande-Terre was
submerged during the late Cretaceous and Paleocene and then
SCHLESSMAN ET AL.—DIOECY ON NEW CALEDONIA 273
Table 1
Species Richness, Endemism, and Dioecy in the Native Floras of
New Caledonia, Hawaii, and New Zealand
New
Caledonia Hawaii
New
Zealand
Land area (km
2
) 19,103 16,887 268,680
Angiosperm species 3051 971 2066
% of species endemic 78 89 84
Dioecious species:
Total 640 143 Ⳳ225
Endemic (%) 567 (89) 136 (95) NA
% of world’s dioecious
species (total Ⳳ
14,620):
Total 4.4 1.0 1.5
Endemic 3.9 .9 NA
Sources. New Caledonia: Morat et al. (2012) and this study; Ha-
waii: Sakai et al. (1995a, 1995b); New Zealand: Godley (1979); total
world’s dioecious species: Renner and Ricklefs (1995).
Note. NA pnot available.
resurfaced during the late Eocene (Brothers and Lillie 1988;
Willford and Brown 1994; Hall 2001; Cluzel et al. 2006; Schel-
lart et al. 2006). While some authors have suggested that the
extant Gondwanan component of the flora could be descended
from pre-submergence ancestors that survived on emergent
portions of Grande-Terre (Lee et al. 2001; Ladiges and Cantrill
2007) or on geologically ephemeral nearby islands (Pelletier
2006), the emerging consensus is that the extant flora has
evolved from colonists that arrived via transoceanic dispersal
after the resurfacing of Grande-Terre Ⳳ37 Mya (Aitchison et
al. 1995; McLoughlin 2001; Grandcolas et al. 2008; Cruaud
et al. 2012).
Incidence of Dioecy on New Caledonia
Our attention to information on the sexual systems of New
Caledonian angiosperms has spanned more than a decade, and
we consulted hundreds of publications. We used three types
of literature to determine sexual systems: (a) the Flore de la
Nouvelle-Cale´ donie et De´pendances (Aubre´ville et al. 1967–)
and other floras of the Pacific region, (b) primary taxonomic
literature for families not covered in the floras, and (c) sec-
ondary references (Kubitzki 1990–) to corroborate informa-
tion from aand bor for families not covered there. We also
examined specimens in the herbaria of the Muse´um National
d’Histoire Naturelle in Paris (P; in 2003 and subsequently)
and the Institut de Recherche pour le De´ veloppement in Nou-
me´ a, New Caledonia (NOU; in 2004 and subsequently). In
the herbaria, we focused on taxa for which the literature was
lacking, unclear, or conflicting and also made spot checks of
other taxa to confirm determinations from the literature.
During 2 wk of fieldwork in New Caledonia (Grande-Terre;
January 4–18, 2004), we evaluated the accuracy of our as-
sessments and attempted to resolve questions that had arisen
from our literature and herbarium studies. At each of 10 sites
chosen to provide a representative sampling of substrate and
vegetation types, we examined and collected all accessible in-
digenous species that were flowering (excluding protected spe-
cies and those whose sexual system was impossible to deter-
mine in the field). We focused especially on four problematic
families: Euphorbiaceae, Primulaceae, Rutaceae, and Sapin-
daceae. We deposited voucher specimens at NOU, P, and the
Missouri Botanical Garden in St. Louis (MO). We were able
to assign most collections to a genus in the field. We completed
species identifications at NOU and P, consulting experts when
necessary. In the field, we examined the sex expression of 155
species, i.e., 5% of the indigenous angiosperm flora. For the
few cases in which our field determinations contradicted in-
formation from the literature, newer publications or consul-
tations with colleagues confirmed our fieldwork.
We classified each species as hermaphroditic, all individuals
bearing bisexual flowers (bisexual pperfect, with both pistils
and stamens); monoecious, all individuals bearing both female
(pistillate) and male (staminate) flowers; andromonoecious, all
individuals bearing both bisexual and male flowers; gyno-
monoecious, all individuals bearing both bisexual and female
flowers; dioecious, separate male (staminate flowers, occa-
sionally also a few bisexual or pistillate flowers) and female
(pistillate flowers, occasionally also a few bisexual or staminate
flowers) individuals; androdioecious, some individuals bearing
only male flowers, others with bisexual flowers (or both sta-
minate and pistillate flowers); gynodioecious, some individuals
bearing only female flowers, others with bisexual flowers (or
both male and female flowers); or undetermined. This classi-
fication encompasses the range of variation that we encoun-
tered. It is necessarily morphological (but see below), and it
follows well-accepted definitions of flowering plant sexual sys-
tems (Darwin 1877; Lloyd and Bawa 1984; Richards 1986;
Sakai and Weller 1999).
Hermaphroditism, monoecy, andromonoecy, and gynomon-
oecy are monomorphic sexual systems; i.e., the distribution of
sexual phenotypes is unimodal. All individuals presumably
have the same genotype for sex expression and can potentially
exhibit the same range of sex expression (Lloyd and Bawa
1984; Barrett 2002). Dioecy, androdioecy, and gynodioecy are
forms of sexual dimorphism; i.e., the distribution of sexual
phenotypes is bimodal, and there are presumably two geno-
types for sex expression (Lloyd and Bawa 1984; Barrett 2002).
Sexual dimorphism can involve leakiness (Baker and Cox
1984) or inconstancies (Lloyd and Bawa 1984), i.e., the oc-
casional production of flowers with atypical sex expression.
For example, Tirel and Veillon (2002) reported that some New
Caledonian species of Pittosporum (Pittosporaceae) were
strictly dioecious (all individuals studied bore either staminate
flowers or pistillate flowers but not both), while in other species
some individuals bore pistillate flowers, some bore staminate
flowers, and others bore both types of flowers. Here we treat
the latter group as dioecious (with inconstancies or leakiness)
rather than placing them in a separate category such as para-
dioecious, polygamodioecious, or subdioecious. Finally, sev-
eral relatively well-studied species, e.g., Polyscias pancheri
(Schlessman et al. 1990b), that appeared on first inspection to
be androdioecious (males and hermaphrodites with apparently
bisexual flowers) have been shown to be functionally dioecious
because the pollen in the apparently bisexual flowers of the
female plants is nonfunctional or absent. Here we also classify
such cryptically dioecious species as dioecious.
274 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Correlates of Dioecy on New Caledonia
Scoring traits. We scored species with determined sexual
systems for five ecological traits: endemism (from Morat et al.
2012), habitat (seven vegetation types; Morat et al. 2012),
habit (woody vs. herbaceous; literature and personal obser-
vations), flower type (petals !10 mm and white, green, or
greenish white pplain; petals 110 mm or brightly colored p
showy; following Vamosi and Vamosi 2004; literature, per-
sonal observations, images from the internet when necessary),
and fruit type (fleshy or not, following Vamosi and Vamosi
2004; literature and personal observations). A few species with
determined sexual systems were excluded from analyses of
endemism and habitat because those data were absent from
Morat et al. (2012).
Evidence for associations of dioecy with other traits.We
sought evidence for associations between dioecy and purported
ecological correlates through the traditional method of species-
based contingency table analyses. We interpreted overrepre-
sentation of dioecious species among all species with the pur-
ported correlate (observed 1expected) to suggest that the
purported correlate might favor one or more of the following:
transitions from other sexual systems to dioecy, speciation of
already dioecious lineages, or persistence of dioecious lineages.
For woodiness, plain flowers, fleshy fruit, and endemism,
we used two-by-two contingency tables, e.g., dioecious or not
by woody or not. We also performed a two-by-seven contin-
gency table analysis for association of dioecy with the seven
possible combinations of presence or absence of the three mor-
phological traits: e.g., woody, plain, and fleshy; woody, plain,
and dry; woody, showy, and dry, and so on. Last, we performed
a two-by-seven analysis for the distribution of dioecious spe-
cies among the seven vegetation types recognized by Morat et
al. (2012). We scored species that occurred in more than one
habitat as present in each of them.
Traditionally, the statistical significance of such contingency
table comparisons has been evaluated with tests of indepen-
dence such as the x
2
(Bawa 1980; Renner and Ricklefs 1995;
Sakai et al. 1995a; Carpenter et al. 2003; Gross 2005; Ma-
tallana et al. 2005). However, using such tests of independence
to evaluate hypotheses about correlated evolution is inappro-
priate because phylogenetic history violates the assumption
that a particular combination of traits (e.g., dioecious and
woody) may occur in each species independently of its presence
in any other species (Felsenstein 1985; Harvey and Pagel
1991). To address this concern, we used contingency table
comparisons solely as a heuristic method for identifying as-
sociations that might be biologically important and for gen-
erating hypotheses about correlated evolution. We refrained
from using overrepresentation of dioecy among species with
a purported correlate as evidence for correlated evolution of
dioecy with that trait (see below). Here we report statistical
tests of independence (x
2
tests, JMP 10.0.0, SAS, Cary, NC)
solely for illustrative and comparative purposes.
Tests for correlated evolution of dioecy with other traits.
To assess evidence for correlated evolution of dioecy with other
traits, we employed Pagel’s maximum likelihood method for
two binary characters as implemented in Mesquite 2.75 (Pagel
1994; Maddison and Maddison 2003). This test compares the
likelihoods for two models of transitions between character
states, one in which transitions between the two states of each
character are independent of those in the other and one in
which changes in the hypothesized dependent character (here
sexual system, dioecious or not) depend on the state of the
independent character (e.g., woody or not). Statistical signif-
icance for rejecting the null hypothesis of independence is de-
termined by the magnitude of the difference between the two
likelihoods.
With Mesquite 2.75, we manually constructed a phyloge-
netic tree for 3040 of the 3053 species in Morat et al. (2012;
3047 with determined sexual systems minus 7 lacking habitat
data). For each species, we included data on sexual system
(dioecious or not), endemism (endemic or not), habitat (occurs
in rainforest or does not), habit (woody or herbaceous), flower
type (plain or showy), and fruit type (fleshy or dry).
We followed Soltis et al. (2011) for phylogenetic relation-
ships among families, and we consulted numerous other
publications to resolve relationships within families and genera
as fully as possible (see literature cited and table S1; tables S1–
S3 available online). Because we needed to draw on a number
of sources to construct the topology of our tree, it was im-
possible to assign meaningful relative branch lengths; there-
fore, we set all branch lengths to 1.
As with most phylogenetic tests for correlated evolution, the
Pagel (1994) method requires a completely resolved tree. How-
ever, some New Caledonian genera and most New Caledonian
species have not been included in molecular phylogenetic stud-
ies, so initially our tree included numerous polytomies, espe-
cially at the infrageneric level. We made every effort to min-
imize the number of polytomies while retaining all of the
meaningful information we had about relationships among
taxa. We collapsed phylogenetically uninformative polytomies
and clades (all species with identical states for all characters)
into single terminal taxa. When infrageneric polytomies could
not be fully collapsed, we reduced the number of terminal taxa
by collapsing species with the same sets of character states into
single operational taxonomic units. A Nexus file for our tree,
a key to taxon abbreviations, and a list of additional sources
we consulted are provided in appendixes A–C, available
online.
Our final tree still had many polytomies, and it was nec-
essary to generate randomly resolved trees for analysis. We
generated a set of 100 different randomly resolved trees and
ran tests for correlated evolution on each one. For each of
those 500 tests (5 possible correlates #100 trees), we based
estimates of statistical significance (Pvalues) on the recom-
mended minimum of 1000 simulation replicates (Maddison
and Maddison 2003).
Morphological correlates and species richness. We used
linear regression (JMP 10.0.0; SAS) to examine the relationship
between the species richness of a dioecious group (genus or
clade with two or more genera) and the number of major
morphological correlates of dioecy (woodiness, plain flowers,
fleshy fruit) possessed by that group. We defined species rich-
ness as the number of extant dioecious species on New Cal-
edonia. If we determined that two or more ancestral colonists
had given rise to a dioecious group, we apportioned species
among ancestors when the literature provided a basis for doing
so (e.g., Ficus in tables 3, S1), or we simply divided the species
equally among the ancestors (e.g., Canarium in tables 3, S1).
SCHLESSMAN ET AL.—DIOECY ON NEW CALEDONIA 275
Table 2
Frequencies (%) of Sexual Systems among Species of Angiosperms in the
Native Floras of Three Pacific Island Groups and the World
Sexual system New Caledonia Hawaii New Zealand World
Dioecy 21.0 14.7 12–13
G
Ⳳ19
W
4
Y
,6
RR
Gynodioecy .1 3.8 2
G
Ⳳ7
W
7
R
Androdioecy .1 0 NA NA
Total dimorphic 21.2 18.5 14–16
G
Ⳳ23
W
10
L
Hermaphroditism 61.2 62.4 NA 72
R
Monoecy 12.8 7.6 9
G
5
R
Andromonoecy 3.7 4.5 NA 1.7
Y
Gynomonoecy 1.0 3.9 NA 3
R
Total monomorphic 78.8 78.4 84–86
G
Ⳳ77
W
90
L
Sources. New Caledonia: this study; Hawaii: Sakai et al. (1995b); New Zealand: G pGodley (1979),
WpWebb et al. (1999); world: L pLloyd and Bawa (1984), R pRichards (1986), RR pRenner and
Ricklefs (1995), Y pYampolsky and Yampolsky (1922).
Note. In sexually dimorphic systems, the distributions of sexual phenotypes within populations are bi-
modal, presumably reflecting the expression of two distinct sexual genotypes, while in monomorphic systems,
the distribution of sexual phenotypes is unimodal, presumably reflecting the expression of a single sexual
genotype (Lloyd and Bawa 1984; Barrett 2002).
We predicted that the evolutionary success of a group on New
Caledonia would increase with the number of major correlates
possessed by the group.
Origins of Dioecy on New Caledonia
The sources used to determine sexual systems and construct
the phylogenetic tree also provided information to estimate the
number of ancestral colonist species (i.e., species that arrived
on New Caledonia following long-distance dispersal) required
to account for origins of the extant dioecious flora, to infer
the sexual systems and ecological traits of those ancestral col-
onists, and to assess the extent of autochthonous speciation
that occurred after colonists became established. In many
cases, traits were invariant within genera or even families, and
no further analysis was necessary to assign character states to
hypothesized ancestors. When traits varied, we compared the
distributions of character states within and among clades con-
taining the extant dioecious taxa (in-groups) and their estab-
lished or likely sister clades (out-groups). We considered the
character state common to an in-group and its out-group as
ancestral for the in-group and assigned that state to the an-
cestral colonist (or colonists) for the in-group. Our ability to
determine the exact numbers of colonizing ancestors and au-
tochthonous speciation events was frequently hampered by
insufficient phylogenetic data. In several cases it was clear that
some autochthonous speciation within a genus had occurred,
but we could not infer the exact number of colonizing ances-
tors involved. Thus, the numbers of hypothesized ancestors
we report are often minimum estimates. Synopses of our rea-
soning on colonizing ancestors and their attributes, including
literature citations, are given in table S1. During our exami-
nations of published phylogenetic and biogeographic studies,
we also sought evidence for the dispersal of dioecy from New
Caledonia to elsewhere.
Results
Incidence of Dioecy in New Caledonian Angiosperms
Using the published literature and our herbarium and field
observations, we were able to assign sexual systems to 99.8%
of the indigenous angiosperm flora (3047 of the 3053 species
listed in Morat et al. 2012; table S2). We found that a re-
markable 21% of the indigenous species were dioecious (640
of 3047; tables 1–3, S1, S2). Dioecious species occurred in
one-third (52 of 158) of the angiosperm families and in one-
sixth of the genera (108 of 690). Instances of leakiness or
inconstancy, i.e., occasional production of unisexual flowers
and bisexual flowers or both kinds of unisexual flowers on the
same individual (see Methods), were reported for seven genera
representing six families: Casuarina and Gymnostoma (Cas-
uarinaceae), Myrtastrum (Myrtaceae), Canarium (Bursera-
ceae), Euroshinus (Anacardiaceae), Myrsine (Primulaceae),
and Pittosporum (Pittosporaceae; tables 3, S1). Rare occur-
rences of andro- and gynodioecy brought the incidence of sex-
ual dimorphism (dioecy ⫹androdioecy ⫹gynodioecy) to
21.2% (table 2).
Ecological Traits Associated with Dioecy
on New Caledonia
Major correlates of dioecy. The four widely accepted eco-
logical correlates of dioecy are tropical distribution, woody
habit, plain flowers, and fleshy fruit (Vamosi et al. 2003). By
definition, all indigenous New Caledonian angiosperms have
tropical distributions. In addition, we found that 76% of the
indigenous angiosperm species are woody, 56% have plain
flowers, 46% have fleshy fruit, and 25% have all three traits
(table S2). Even though the overall frequency of woodiness
was very high, dioecy was still more common than expected
among woody species (fig. 1, W). Similarly, the observed num-
bers of dioecious species with plain flowers and fleshy fruit
are both much greater than expected (fig. 1, P, F). To make
276
Table 3
Origins and Ecology of the Indigenous Dioecious Flora of New Caledonia
Order, family,
and genus
Species (endemic) Ancestors Ecology of extant species
Total Dioecious No. SS Habitat Woody? Plain flowers? Fleshy fruit?
Amborellales:
Amborellaceae:
a
Amborella
a
1(1) 1 1 u F ⫹⫹ ⫹
Chloranthales:
Chloranthaceae:
Ascarina 2(2) 2 2 d F ⫹⫹ ⫹
Piperales:
Piperaceae:
Piper 4(3) 4 1 d F – ⫹⫹
Laurales:
Lauraceae:
Litsea
b
15 (15) 15 1 d M ⫹⫹ ⫹
Adenodaphne
a
4(4) 4 M ⫹⫹ ⫹
Monimiaceae:
Hedycarya
b
9(9) 9 1 d F ⫹⫹ ⫹
Kibaropsis
a
1(1) 1 F ⫹⫹ ⫹
Alismatales:
Cymodoceaceae:
Cymodocea 2(0) 2 2 d R – ⫹–
Halodule 2(0) 2 2 d R – ⫹–
Syringodium 1(0) 1 1 d R – ⫹–
Thalassodendron 1(0) 1 1 d R ⫹⫹ –
Hydrocharitaceae:
Enhalus 1(0) 1 1 d R – ⫹⫹
Halophila 3(0) 2 2 d R – ⫹⫹
Thalassia 1(0) 1 1 d R – ⫹⫹
Vallisneria 1(0) 1 1 d R – ⫹⫹
Dioscoreales:
Dioscoreaceae:
Dioscorea 1(0) 1 1 d F – ⫹⫹
Pandanales:
Pandanaceae:
Frecinetia 24 (19) 24 1 d F ⫹⫹ ⫹
Pandanus 21 (20) 21 1 d F ⫹⫹ ⫹
Liliales:
Smilacaceae:
Smilax 6(6) 6 1 d F ⫹⫹ ⫹
Asparagales:
Asparagaceae:
Lomandra 1(1) 1 1 M ⫹⫹ –
Asteliaceae:
Astelia 1(1) 1 1 d F – ⫹⫹
Poales:
Poaceae:
Spinifex 1(0) 1 1 d G – ⫹–
Ranunculales:
Menispermaceae:
Hypserpa 3(2) 3 1 d F ⫹⫹ ⫹
Pachygone 2(2) 2 1 d F ⫹⫹ ⫹
Stephania 1(0) 1 1 d F ⫹⫹ ⫹
Tinospora 1(1) 1 1 d F ⫹⫹ ⫹
Ranunculaceae:
Clematis 3 (1) 3 1 d L,M,N ⫹––
Fabales:
Polygalaceae:
Balgoya
a
1(1) 1 1 H F ⫹⫹ ⫹
Rosales:
Moraceae:
Ficus 35 (27) 4 (0) 2 d L ⫹⫹ ⫹
Maclura 1 (0) 1 1 d L,G,N ⫹⫹ ⫹
277
Table 3
(Continued )
Order, family,
and genus
Species (endemic) Ancestors Ecology of extant species
Total Dioecious No. SS Habitat Woody? Plain flowers? Fleshy fruit?
Sparattosyce
a
2(2) 2 1 d F ⫹⫹ ⫹
Streblus 2(1) 2 1 d F ⫹⫹ ⫹
Trophis 1 (0) 1 1 d F,L,G,N ⫹⫹ ⫹
Urticaceae:
Dendrocnide 3(0) 3 1 d F – ⫹–
Nothocnide 1 (0) 1 1 d F,G,N – ⫹–
Pipturus 2(0) 2 1 d N – ⫹–
Fagales:
Myricaceae:
Canacomyrica
a
1(1) 1 1 d F ⫹⫹ ⫹
Casuarinaceae:
Casuarina
c
3(2) 3 1 d M,G ⫹⫹ –
Gymnostoma
c
8 (8) 8 1 d F,M,R ⫹⫹ –
Cucurbitales:
Cucurbitaceae:
Zehneria 3(1) 3 1 d N – ⫹⫹
Celastrales:
Celastraceae:
Celastrus 1(0) 1 1 d ,M,N ⫹⫹ –
Elaeodendron 6(5) 3(2) 1 d F ⫹⫹ ⫹
Menepetalum
a
4(4) 4 1 H F ⫹⫹ –
Salaciopsis
a
6(6) 6 1 u F ⫹⫹ –
Oxalidales:
Cunoniaceae:
Hooglandia
a
1(1) 1 1 H F ⫹⫹ ⫹
Pancheria
a
27 (27) 27 1 d M ⫹⫹ –
Malphigiales:
Euphorbiaceae:
Acalypha 4(3) 1(1) 1 m? F ⫹⫹ –
Baloghia 13 (12) 9 (8) 1 u F ⫹⫹ –
Bocquillonia
a
14 (14) 14 1 d F ⫹⫹ –
Claoxylon 1(1) 1 1 d F ⫹⫹ –
Cleidion 12 (12) 10 1 u F ⫹⫹ –
Excoecaria 1(0) 1 1 d F ⫹⫹ –
Fontainea 1(0) 1 1 d F ⫹⫹ ⫹
Macaranga 5(5) 5 1 d F ⫹⫹ –
Mallotus 1(0) 1 1 d F ⫹⫹ –
Picrodendraceae:
Austrobuxus 15 (15) 15 1 u F ⫹⫹ ⫹
Phyllanthaceae:
Antidesma 1(1) 1 1 d F ⫹⫹ ⫹
Bischofia 1(0) 1 1 d F ⫹⫹ ⫹
Malpighiaceae:
Rhyssopterus 4(3) 4 1 d M ⫹⫹ –
Balanopaceae:
Balanops 7(7) 7 1 d F ⫹⫹ ⫹
Putranjivaceae:
Drypetes 1(0) 1 1 d F ⫹⫹ ⫹
Salicaceae:
Xylosma
b
20 (20) 20 1 d F ⫹⫹ ⫹
Lasiochlamys
a
11 (11) 11
Calophyllaceae:
Mammea 2(1) 2 1 d F,G ⫹–⫹
Clusiaceae:
Garcinia 13 (13) 13 (13) 1 d F ⫹–⫹
Myrtales:
Myrtaceae:
Carpolepis 3(3) 1 1 H F,M ⫹––
Myrtastrum
a,c
1(1) 1 1 H F ⫹⫹ ⫹
Table 3
(Continued )
Order, family,
and genus
Species (endemic) Ancestors Ecology of extant species
Total Dioecious No. SS Habitat Woody? Plain flowers? Fleshy fruit?
Sapindales:
Burseraceae:
Canarium
c
4(4) 4 2 d F ⫹⫹ ⫹
Anacardiaceae:
Euroschinus
a,c
7(7) 7 1 d F ⫹⫹ ⫹
Pleiogynium 1(0) 1 1 d L ⫹⫹ ⫹
Semecarpus 6(6) 6 1 d F ⫹⫹ ⫹
Sapindaceae:
Allophylus 2(0) 2 1 d F ⫹⫹ –
Dodonaea 2(0) 1 1 d F ⫹⫹ –
Rutaceae:
Comptonella
a
8(8) 8 1 u F,M ⫹⫹ ⫹
?Picrella
a,b
3(3) 3 L ⫹⫹ ⫹
Crossosperma
a
2(2) 2 1 H F ⫹⫹ ⫹
Medicosma 15 (15) 11 1 H M ⫹⫹ –
Melicope 6(6) 6 1 d F ⫹⫹ ⫹
Sarcomelicope 9(8) 9 1 d F ⫹⫹ ⫹
Zanthoxylum 8(8) 8 1 d F ⫹⫹ ⫹
Simaroubaceae:
Soulamea 11 (11) 11 1 d F ⫹⫹ –
Meliaceae:
Aglaia 1(0) 1 1 d F,L ⫹⫹ ⫹
Anthocarapa 1(0) 1 1 d F ⫹⫹ ⫹
Dysoxylum 9(8) 9 1 d F ⫹⫹ –
Xylocarpus 2(0) 2 2 d G ⫹⫹ –
Malvales:
Thymelaeaceae:
Lethedon 15 (15) 15 1 d F ⫹⫹ ⫹
?Solmsia
a,b
2(2) 2
Santalales:
Balanophoraceae:
Hachettea
a
1(1) 1 1 d F – ⫹–
Caryophyllales:
Nepenthaceae:
Nepenthes 1(1) 1 1 d F ⫹––
Phytolaccaceae:
Monococcus 1(0) 1 1 d F,L,N ⫹⫹ –
Nyctaginaceae:
Pisonia 4(2) 3(1) 1 u F,L ⫹⫹ –
Ericales:
Ebenaceae:
Diospyros 32 (29) 32 4 d F ⫹⫹ ⫹
Primulaceae:
Myrsine
c
39 (37) 39 1 d F ⫹⫹ ⫹
Lamiales:
Boraginaceae:
Cordia 2(0) 2 2 u G ⫹–⫹
Gentianales:
Rubiaceae:
Antirhea 2(2) 2 1 d F ⫹⫹ ⫹
Guettarda 11 (10) 11 F ⫹⫹ ⫹
Timonius 1(0) 1 G ⫹⫹ ⫹
Tinadendron 2(1) 1 FL ⫹⫹ ⫹
Atractocarpus 12 (12) 12 1 d F ⫹–⫹
Randia 9(9) 9 ⫹–⫹
Gardenia 7(7) 1 ⫹–⫹
Cyclophyllum 16 (16) 13 1 d F ⫹–⫹
Gea 6(6) 6 1 d FM ⫹⫹ ⫹
SCHLESSMAN ET AL.—DIOECY ON NEW CALEDONIA 279
Table 3
(Continued )
Order, family,
and genus
Species (endemic) Ancestors Ecology of extant species
Total Dioecious No. SS Habitat Woody? Plain flowers? Fleshy fruit?
Aquifoliales:
Aquifoliaceae:
Ilex 2(0) 2 1 d F ⫹⫹ ⫹
Asterales:
Phellinaceae:
a
Phelline
a
14 (14) 14 1 H F ⫹⫹ ⫹
Apiales:
Araliaceae:
Meryta 16 (16) 16 1 d F ⫹⫹ ⫹
Polyscias 26 (25) 18 (18) 1 Am F ⫹⫹ ⫹
Pittosporaceae:
Pittosporum
c
45 (45) 40 1 u F ⫹––
Totals 640 (566) 109
Note. Orders follow APG III (2009); families and genera follow FLORICAL (Morat et al. 2012, which also follows APG III). Underlined
taxa signify that all taxa included are likely in the same clade. Numbers of endemic dioecious species given in parentheses only if not clear from
data for the entire genus. Ancestors: S psexual system, d pdioecious, H phermaphroditic, Am pandromonoecious, m pmonoecious,
upunknown; the total number of colonists is the minimum estimate. Habitats according to Morat et al. (2012); whenever possible, only the
predominant habitat is listed: F prainforest, L psclerophyll forest, M pmaquis, G phalophytic, N pdisturbed, R pwetland. For other
ecological traits, a plus sign indicates yes, minus no. Inferred ecological traits of ancestors are the same as the ecological traits for the extant
taxa in all cases except Crossosperma, for which we infer that the ancestor had dry rather than fleshy fruit (see table S1).
a
Endemic to New Caledonia.
b
Autochthonous derivation of the second genus from an ancestor resembling the first genus.
c
“Leakiness” or “inconstancy” reported.
the comparison another way, 98% of the dioecious species are
woody, 84% have plain flowers, and 69% have fleshy fruit.
Traditional species-based tests for independence of trait com-
binations (presented here for illustrative purposes only, not
hypothesis testing) indicated very strong associations between
dioecy and each of the three traits (x
2
p218 for woodiness,
234 for plain flowers, 161 for fleshy fruit; df p1, Palways
!0.0001, np3049). Similarly, in all tests for correlated evo-
lution of dioecy with woodiness, plain flowers, and fleshy fruit,
differences between the likelihoods for independent and de-
pendent models were always large and very highly significant
(Palways 0; table 4).
The distribution of character state combinations for wood-
iness, plain flowers, and fleshy fruit among dioecious species
differed significantly from expectation based on the frequencies
of the combinations in the entire flora (fig. 1, right-hand group
of bars; x
2
p568, df p7, P!0.0001). Species with all three
traits are very strongly overrepresented, while those that were
woody with plain flowers and dry fruit are only moderately
overrepresented and all other trait combinations were
underrepresented.
Species richness of dioecious groups, and hence the amount
of postcolonization speciation that probably occurred in those
groups, increased with the number of major morphological
correlates of dioecy that they possessed (fig. 2). However, the
differences among means were not statistically significant be-
cause the genus Pittosporum, with only one correlate (wood-
iness) and 40 dioecious species, was a clear outlier (linear re-
gression; Fp2.10, df p1, 97, Pp0.151). When we excluded
Pittosporum (for illustrative purposes only), the mean species
richness for groups with only one correlate decreased sub-
stantially, and differences among means became statistically
significant (linear regression; Fp7.09, df p1, 96, Pp
0.009).
Habitat and endemism. The distribution of dioecious spe-
cies among habitats shows that dioecy is noticeably more com-
mon than expected in rainforest, slightly overrepresented in
maquis, minimally overrepresented in sclerophyll, and under-
represented in all other vegetation types (fig. 3; table S3). Our
traditional test of independence suggested that dioecious spe-
cies are indeed significantly overrepresented in rainforest
(x
2
p17, df p1, P!0.0001, np3041), but the results of
our phylogenetic tests for correlated evolution were decidedly
mixed. Because most of the unresolved polytomies in our phy-
logenetic tree included both forest taxa and nonforest taxa,
rejection of the null hypothesis of independent character state
transitions depended strongly on how those polytomies were
resolved. Just 44 of the 100 randomly resolved trees gave sta-
tistically significant differences between log likelihoods for the
independent and dependent models (table 4).
Even though the overall proportion of endemics on New
Caledonia is very high at 78% (Morat et al. 2012), dioecy was
still overrepresented among endemic taxa (fig. 1). While our
traditional test of independence indicated statistical signifi-
cance (x
2
p54, df p1, P!0.0001, np3049), our phy-
logenetic tests did not support correlated evolution of dioecy
and endemism. Only nine of the 100 randomly resolved trees
produced statistically significant Pvalues (table 4).
Eight of the nine randomly resolved trees that supported
correlated evolution of dioecy and endemism also supported
correlated evolution of dioecy and occurrence in the rainforest
habitat. Thus, while all 100 randomly resolved trees provided
280 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Fig. 1 Numbers of dioecious angiosperms native to New Caledonia exhibiting the three morphological traits considered major ecological
correlates of dioecy. Single traits (three groups on left) and combinations of traits (eight groups on right) arranged in decreasing order of expected
frequencies in the entire native angiosperm flora. W pwoody, h pherbaceous, P pplain flowers, sh pshowy flowers, F pfleshy fruit,
dr pdry fruit.
Table 4
Tests for Correlated Evolution of Dioecy with Other Traits
Difference Pvalue
Trait Mean Range Range No. !.05
Plain flowers 23.65 21.66–24.89 0 100
Woody 21.67 21.19–22.65 0 100
Fleshy fruit 15.51 13.29–17.03 0 100
Rainforest 4.08 2.13–7.15 .002–.730 44
Endemic 4.25 3.44–6.27 .015–.980 9
Note. Maximum likelihood tests (Pagel 1994) implemented with
Mesquite 2.75 (Maddison and Maddison 2003). Differences in log
likelihoods (independent model ⫺dependent model) and their asso-
ciated Pvalues were determined for 100 different random resolutions
of the polytomies in our phylogenetic tree.
support for correlated evolution of dioecy and the three mor-
phological correlates (woodiness, plain flowers, fleshy fruit),
only eight supported correlated evolution of all five of the
purported correlates that we evaluated.
Origins of Dioecious Lineages on New Caledonia
Characteristics of ancestral colonists. We estimated that
a minimum total of 109 ancestral colonist species was required
to account for the extant indigenous dioecious flora (table 3).
We were able to infer the sexual systems for 104 of thecolonists
and concluded that 95 of them (91%) were dioecious, while
only 9 had some other sexual system (7 hermaphroditism, 1
monoecy, 1 andromonoecy). Even if all of the ancestors with
unknown sexual systems were found to be nondioecious, the
dioecious proportion would remain very high at 87% (95/
109). Eight of the inferred ancestors (7%) may have exhibited
leaky or inconstant dioecy (tables 3, S1). Except for Crossos-
perma (ancestor with dry fruit to New Caledonian descendants
with fleshy fruit; table S1), we found no evidence that the
morphological traits of ancestors differed from those of their
extant descendants; thus, 99 (91%) had plain flowers and 90
(83%) were woody, but only 67 (61%) had fleshy fruit.
Postcolonization speciation in dioecious groups. While
the quality of phylogenetic information varied considerably
among groups, we were able to infer that 46 (42%) of our
hypothesized ancestral colonists have just one extant descen-
dant (indeed, the hypothesized ancestor and the extant species
may be the same), while as many as 63 (58%) may have two
or more descendants (table 3). Since the later 58% of colonists
account for 93% (595/640) of the extant dioecious species, it
appears that autochthonous speciation (i.e., postcolonization
speciation on New Caledonia) has contributed substantially to
the current high incidence of dioecy. Groups that appear to
have undergone the most extensive autochthonous speciation
are Pittosporum (Pittosporaceae, 45 spp.); Myrsine (formerly
Rapanea; Primulaceae, 39 spp.); Xylosma ⫹Lasiochlamys
SCHLESSMAN ET AL.—DIOECY ON NEW CALEDONIA 281
Fig. 2 Species richness of dioecious groups (single genera or groups
of related genera) in the indigenous angiosperm flora of New Cale-
donia with different numbers of the three major morphological cor-
relates of dioecy (woodiness, plain flowers, fleshy fruits). The trend of
increasing numbers of species with increasing numbers of correlates
is not statistically significant (filled bars) unless the outlier group Pit-
tosporum (1 correlate, 40 species) is excluded (filled bars; linear
regressions).
Fig. 3 Observed and expected numbers of dioecious angiosperms
native to New Caledonia that are endemic or that occur in each of
the seven habitats (vegetation types) recognized by Morat et al. (2012).
Expected numbers are based on the overall frequency of endemics or
the frequencies of species occurring in each habitat for the entire flora.
(Salicaceae, 31 spp.); Pancheria (Cunoniaceae, 27 spp.); Frey-
cinetia (Pandanaceae, 24 spp.); Pandanus (Pandanaceae, 21
spp.); Litsea ⫹Adenodaphne (Lauraceae, 19 spp.), each per-
haps derived from a single colonizing ancestor; and Diospyros
(Ebenaceae, 32 spp.), for which there were probably four dif-
ferent colonists (tables 3, S1).
New Caledonia as a Source of Dioecy Elsewhere
We found evidence for dispersal of dioecious taxa from New
Caledonia to elsewhere for the genera Sarcomelicope (Ruta-
ceae), Soulamea (Simaroubaceae), Polyscias (Araliaceae), and
Pittosporum (Pittosporaceae). All species of Sarcomelicope ex-
cept one are endemic to New Caledonia. The exception, Sar-
comelicope simplicifolia, occurs also on Australia, Lord Howe
Island, and Norfolk Island. The most recent molecular phy-
logenetic results are consistent with dispersal from New Cal-
edonia elsewhere, but further work is necessary to confirm this
(Bayly et al. 2013). Phylogeographic analysis suggests that the
ancestor of Soulamea terminaloides, a dioecious species en-
demic to the Seychelles, may have dispersed from New Cale-
donia (Clayton et al. 2009). For Polyscias, there may have
been three separate dispersals from New Caledonia: Polyscias
cissodendron to Vanuatu and Lord Howe Island, Polyscias
jokesii to Fiji, and Polyscias schmidii to Vanuatu (Plunkett et
al. 2001; Plunkett and Lowry 2010). For Pittosporum, Gem-
mill et al. (2002) suggested possible dispersal from New Cal-
edonia to Figi/Tonga and then to Hawaii. More recent analyses
(C. E. C. Gemmill, personal communication) appear to confirm
that suggestion (though independent dispersals to Figi/Tonga
and Hawaii cannot be ruled out). More recent work also in-
dicates dispersal from New Caledonia to New Zealand (C. E.
C. Gemmill, personal communication).
Discussion
Incidence of Dioecy on New Caledonia
At 21%, the incidence of dioecy on New Caledonia is more
than three times that for angiosperms as a whole and among
the highest known for island floras. The incidence of sexual
dimorphism (dioecy ⫹androdioecy ⫹gynodioecy) is 21.2%,
which is similar to Hawaii and New Zealand (table 2), where
gynodioecy is more common. The combination of great species
richness and a high incidence of dioecy gives New Caledonia
perhaps the highest concentration of dioecious angiosperms
per unit area in the world (table 1). New Caledonia has 2.5
times as many dioecious species as New Zealand, which is 14
times larger, and more than 4 times as many as Hawaii, which
is only 12% smaller. We estimated that 4.4% of the world’s
dioecious species occur on New Caledonia and 3.9% of the
world’s dioecious species are found only there.
This study is the first to assess the diversity of sexual systems
in the entire native flora of New Caledonia. We believe that
our multipronged approach of combining literature search,
herbarium work, and direct observation in the field has yielded
the most accurate assessment possible at this time. While the
numbers of species with various sexual systems will surely
change as more detailed studies of floral biology are completed
and new taxa are discovered and described, we expect the
relative proportions to remain largely the same.
Correlates of Dioecy on New Caledonia
In our study, explicitly phylogenetic tests strongly supported
correlated evolution of dioecy with woodiness, plain flowers,
and fleshy fruit, thus confirming that each of those morpho-
logical traits as an ecological correlate of dioecy. Our work
also supports the finding that dioecious lineages are most spe-
ciose when they posses all three of these morphological traits
282 INTERNATIONAL JOURNAL OF PLANT SCIENCES
and also occur in the tropics (Vamosi and Vamosi 2004). Our
analyses indicate that ∼90% of the colonists that established
the extant dioecious lineages on New Caledonia were them-
selves dioecious and that each of those dioecious colonists
possessed at least two of the morphological correlates. From
this, we conclude that with respect to the dioecious lineages
that successfully colonized New Caledonia and left extant de-
scendants, essentially all of the events of correlated evolution
of dioecy and the three morphological correlates occurred else-
where, prior to dispersal to New Caledonia. Thus, dispersal
to New Caledonia and successful establishment there has been
an important ecological sieve (Pannell 2006) through which
dioecious species with the most correlates are the most likely
to be successful colonists. Moreover, low floristic diversity
soon after the Eocene reemergence of Grande-Terre may have
limited competition of dioecious colonists with nondioecious
ones, mitigating the disadvantages of dioecy relative to her-
maphroditism and facilitating establishment of the dioecious
colonists (Heilbuth et al. 2001). Therefore, we discuss the sig-
nificance of fleshy fruits, plain flowers, woodiness, and tropical
habitat for dioecy on New Caledonia in terms of dispersal,
establishment, and subsequent speciation of already dioecious
lineages.
Fleshy fruit. Bawa (1980), Givnish (1980), and others (see
Sakai and Weller 1999) have argued that since dioecy is as-
sociated with fleshy fruits and endozoochorous dispersal of
seeds by birds, dioecy may be especially common on islands
with native floras derived from transoceanic dispersal. Infer-
ences about the colonists that gave rise to the extant flora of
Hawaii appear to support this, as the incidence of fleshy fruit
was only 35% among all hypothesized colonists but rose to
52% for dioecious colonists that gave rise to extant dioecious
groups and to 65% for all dioecious colonists (Sakai et al.
1995b). For New Caledonia we have so far considered only
those colonists that gave rise to the extant dioecious flora. Of
those, 62% overall (68/109) had fleshy fruits, as did 67% (64/
95) of the colonists that were dioecious.
Of course, the absence of fleshy fruits does not preclude
dioecy or prevent transoceanic dispersal. In our analyses, 80%
(32/40) of colonists with dry fruits were also dioecious, and
a number of them would have been well adapted for long-
distance dispersal of diaspores. Examples include Macaranga
and Mallotus (Euphorbiaceae), with dry, capsular fruits con-
taining fleshy seeds (Esser 2003); Austrobuxus (Picrodendra-
cee), with arillate seeds (McPherson and Tirel 1987); and Pit-
tosporum (Pittosporaceae), with bright orange or red seeds
embedded in sticky resin (Cayzer et al. 2000).
Plain flowers. Small, radially symmetrical, plainly colored
flowers visited by generalist pollinators are thought to favor
dioecy because they facilitate (1) gender specialization without
substantial morphological differentiation between male and
female flowers, reducing the ability of pollinators to discrim-
inate between them (Charlesworth 1993), and (2) pollen-pack-
aging strategies that allow increases in male reproductive in-
vestment (e.g., larger male inflorescences) without saturation
of the relationship between investment and actual reproductive
success (the male gain curve; Charlesworth and Charlesworth
1987). These factors should facilitate the establishment of
males in hermaphroditic populations (Thomson and Brunet
1990) and the establishment of females in hermaphroditic pop-
ulations when pollen is the only floral reward (Charlesworth
1993; Vamosi et al. 2003). Plain flowers should also reduce
the likelihood of extinction caused by infrequent pollinator
visits to females of populations with extreme sexual dimor-
phism in floral display (Vamosi and Otto 2002). Scarcity of
specialized pollinators has been suggested as a factor favoring
plain flowers and dioecy on both Hawaii (Bawa 1980; Sakai
et al. 1995a) and New Zealand (Godley 1979; Lloyd 1985).
The entomofauna of New Caledonia is rather poorly known,
with as much as 80% of the species not yet described (J. Cha-
zeau and H. Jourdan, personal communication). The bee fauna
(Apoidea) is now known to include at least 45 species (Don-
ovan et al. 2013), which is comparable to New Zealand and
Hawaii (40 and 54 species, respectively; Pauly and Munzinger
2003). Excluding three recent introductions (Apis mellifera
and two megachilids) there are 32 species of long-tongued bees,
but none of them is common. In contrast there are only 524
described species of Lepidoptera on New Caledonia (Chazeau
1993), compared to 1203 for Hawaii (Bishop Museum 2002).
There are just 17 species of nocturnal moths (Sphingidae) and
no long-tongued flies (Nemestrinidae). The New Caledonian
avifauna includes just six species of nectar feeders (all honey-
eaters; Meliphagidae), and only three of those are found in
the rainforests of Grande-Terre (Hannecart and Le´tocart 1980,
1983).
Broad, community-level studies of plant-pollinator interac-
tions on New Caledonia are almost nonexistent. In their study
of 123 rainforest tree species, Carpenter et al. (2003) reported
that Ⳳ61% had white or pale-colored flowers and that 81%
were pollinated by insects, 13% by wind, and 6% by birds.
In a study of 95 indigenous species distributed among rain-
forest, sclerophyllous forest, maquis, savanna, and mangrove
vegetation, Kato and Kawakita (2004) reported Ⳳ80% insect,
3% wind, 12% bird, and 3% bat pollination. While the most
frequent primary insect pollinators were bees (46.3% of the
plant species), the overwhelming majority of visits were by
introduced honeybees (Apis mellifera). Other insect visitors
were moths (20% of plant species), beetles (8.4%), and flies
(3.2%). The available data suggest that on New Caledonia, a
relative paucity of some groups of specialized pollinators may
have contributed to the high incidence of dioecy by favoring
the establishment of colonists with plain flowers over those
with showy ones. This is consistent with our estimate that 91%
of the ancestors of New Caledonia’s dioecious flora already
had plain flowers.
Woody habit. Woodiness is thought to favor dioecy pri-
marily through indirect means. The large inflorescences pro-
duced by many woody plants may result in substantial gei-
tonogamy, which could cause pollen-stigma interference and
inbreeding (de Jong et al. 1993). If inflorescences are suffi-
ciently large, mechanisms such as herkogamy and dichogamy
may be ineffective in reducing geitonogamy, and selection
would then favor dioecy (Harder and Barrett 1996; Barrett
2002; Vamosi et al. 2003). Since woody plants are perennial
and may have numerous opportunities for successful mating,
the riskiness of dioecy could be mitigated (Pannell and Barrett
1998).
New Caledonia’s equable, tropical climate has clearly been
favorable for woody plants. Prior to the arrival of humans,
extensive evergreen forests likely covered more than 70% of
SCHLESSMAN ET AL.—DIOECY ON NEW CALEDONIA 283
the land area (∼13,000 km
2
; Jaffre´ et al. 2001; Lowry et al.
2004). Though rainforests are now reduced to only ∼4000
km
2
(Lowry et al. 2004), ∼60% of New Caledonia’s angio-
sperms are still found there (Morat et al. 2012). As we noted
above, the incidence of dioecy among woody rainforest species
is extremely high, both on New Caledonia as we have shown
here and elsewhere.
Tropical climate. Vamosi and Vamosi (2004) argued that
tropical climate per se favors dioecy independently of its as-
sociations with morphological correlates such as woodiness.
They suggested that the absence of a severely unfavorable sea-
son permits less synchrony of flowering and fruiting among
species, decreasing interspecific competition for pollinators and
dispersers. Since dioecious species cannot self-pollinate and
have fewer fruit-producing individuals than sexually mono-
morphic species, reduction of such interspecific competition
should be especially advantageous for dioecious species (Heil-
buth et al. 2001). There are few comprehensive phenological
studies of New Caledonian rainforest plants, but Carpenter et
al. (2003) found that as a group, 123 species of rainforest trees
on ultramafic soils flowered and bore fruit continuously
throughout the year. Understanding the reproductive phenol-
ogies of dioecious species may help explain the association of
dioecy with woodiness and rainforest habitats.
Habitat and endemism. The extremely high incidence of
dioecy among New Caledonia’s rainforest species, 27%, is at
the upper end of the range previously reported in studies re-
stricted to tropical forest vegetation (Ⳳ18%–26%; Adam and
Williams 1994; Sakai and Weller 1999; Carpenter et al. 2003;
Gross 2005; Matallana et al. 2005; Vary et al. 2011). Marked
overrepresentation of dioecy among endemics and rainforest
species seems to suggest that the high incidence of dioecy on
New Caledonia is due in part to a climate that supports an
extensive rainforest. Indeed, all but two of the more speciose
dioecious groups mentioned above occur solely or primarily
in rainforest, the exceptions being Litsea ⫹Adenodaphne
(Lauraceae) and Pancheria (Cunoniaceae), which occur pri-
marily in maquis. The lack of support for correlated evolution
of dioecy and occurrence in rainforest habitat or endemism
suggests that while the New Caledonian environment may
have favored postcolonization speciation of already dioecious
lineages, it has not favored postcolonization transitions from
other sexual systems to dioecy.
Origins of Dioecy on New Caledonia
Dioecious colonists versus nondioecious colonists.Wees-
timated that a minimum of 87% of the colonizing species that
gave rise to the extant dioecious flora of New Caledonia were
themselves dioecious. Sakai et al. (1995b) inferred a similarly
high proportion of dioecy among colonists that produced the
dioecious flora of Hawaii (86%; treating polygamo- and sub-
dioecy as dioecy). For New Zealand, Webb et al. (1999) con-
cluded that dioecy arose autochthonously (i.e., on New Zea-
land after colonization) in 22% (15/69) of the entirely or
partially dioecious genera, implying that ancestors of the re-
maining 78% were probably dioecious. Clearly, our results,
as well as those for Hawaii and New Zealand, do not support
the prediction, derived from Baker’s law (Baker 1955), that
dioecy on these islands should be largely derived from her-
maphroditism or another cosexual system after dispersal to
the island. Rather, the available data for all three island groups
favor the alternative view that dioecy itself usually originated
elsewhere and then reached the islands because it was asso-
ciated with other traits (e.g., fleshy fruit) that facilitated long-
distance dispersal (Bawa 1980). Our results for New Caledonia
provide new empirical support for the recent theoretical work
of Cheptou and Massol (2009), who, contrary to Baker’s law,
predicted evolutionarily stable associations between outcross-
ing and high dispersibility and also between selfing and low
dispersibility.
As we noted above, limited competition soon after the Eo-
cene reemergence of Grande-Terre may have mitigated the dis-
advantages of dioecy relative to hermaphroditism and other
monomorphic sexual systems, thus facilitating establishment
of dioecious lineages. Nevertheless, 79% of the extant native
flora is sexually monomorphic, and it is likely that a strong
majority of those species are derived from sexually mono-
morphic ancestors. While it is beyond the scope of this study,
it would be interesting to infer sexual systems for ancestors of
the sexually monomorphic flora and to determine the relative
ages of monomorphic and dimorphic lineages.
Similarly, leakiness or inconstancy may also have facilitated
establishment after long-distance dispersal (Humeau et al.
1999). Though we found that only 7% of the colonizing an-
cestors that gave rise to the extant dioecious flora may have
exhibited leaky dioecy, leakiness does occur in two of the most
speciose dioecious groups (Myrsine, 39 species, and Pittos-
porum, 40 species).
Autochthonous speciation versus frequent colonization.
We found that autochthonous speciation played a significant
role in the evolution of New Caledonia’s extant dioecious flora,
with more than nine-tenths of the extant dioecious species
derived from just 58% of the ancestral colonists. As this con-
clusion is based on our inferences of the minimal number of
ancestors needed to account for the extant dioecious flora, it
is subject to revision if new phylogenetic studies provide evi-
dence for multiple colonizations by species in the same genus.
However, we note that extensive autochthonous speciation ap-
pears to have occurred in the evolution of the dioecious native
flora of Hawaii. We calculated that 85% of the dioecious flora
(134 of 158 dioecious, polygamodioecious, or subdioecious
species; Sakai et al. 1995b) was derived via autochthonous
speciation from just 46% (17/37) of the inferred colonists. We
could not obtain a comparable estimate for any other island
group.
Dioecy Out of New Caledonia?
In all but one of the cases we discovered, it appears that
dispersal of dioecy from New Caledonia to elsewhere has left
just one (Sarcomelicope,Soulamea,Polyscias)oratmosttwo
(Pittosporum to Fiji/Tonga, Pittosporum to New Zealand) ex-
tant descendants. For Sarcomelicope and Polyscias, within-
species disjunctions indicate rather recent dispersals with very
little morphological change. At present, the only possible case
of dispersal of dioecy from New Caledonia leading to a sig-
nificant radiation of dioecious species elsewhere is that of Pit-
tosporum dispersing directly from New Caledonia to Hawaii.
However, it seems more likely that the ancestor of the Ha-
284 INTERNATIONAL JOURNAL OF PLANT SCIENCES
waiian clade of Pittosporum arrived from Fiji/Tonga rather
than directly from New Caledonia (Gemmill et al. 2002).
Future Prospects
With 3.9% of the world’s dioecious angiosperms occurring
only there (table 1), New Caledonia presents an especially rich
and potentially important source of new information for elu-
cidating the maintenance of dioecy and the conditions under
which dioecious lineages have diversified. While this study fo-
cused specifically on the origins of the extant dioecious species,
there are many questions remaining. Expanding our analyses
of numbers and traits of ancestral colonists and postcoloni-
zation speciation to the entire native flora would provide a
more comprehensive assessment of Baker’s law and perhaps
reveal losses of dioecy after dispersal to New Caledonia. It
would also be interesting to compare the ages of dioecious and
nondioecious lineages to test the prediction that most dioecious
lineages were established relatively soon after the Eocene emer-
gence of Grande-Terre, when the flora was relatively sparse
and competition somewhat limited.
As we noted above, knowledge of New Caledonia’s flora is
far from complete. New taxa are described regularly, and for
many species there is little information beyond the original
taxonomic descriptions. Yet even as new species arerecognized
and described, conservationists have recognized New Cale-
donia as a priority hot spot in which anthropogenic fire, in-
vasive species, logging, mining, and urbanization continue to
threaten natural vegetation (Bramwell 2011). The once exten-
sive humid evergreen forest, in which the large majority of
endemic dioecious species occur, has been reduced by at least
70%, from more than 13,000 to ∼4,000 km
2
(Lowry et al.
2004). Truly protected natural areas comprise only 3.4% of
the land area (Morat et al. 2012).
The same attributes of dioecy that make it rare among an-
giosperms, including the necessity of pollen transfer between
two particular types of individuals, the inability of many in-
dividuals to produce seeds, and the need for at least two in-
dividuals to establish a new population, should also make
dioecious species especially vulnerable to anthropogenic ex-
tinction. Recently, Vamosi and Vamosi (2005) showed that
exclusively dioecious families had higher proportions of threat-
ened species than their nondioecious sister groups and that
woodiness also contributes to the high likelihood that dioe-
cious species are listed as threatened. The very high incidence
of dioecy in New Caledonia’s remaining rainforest, along with
the high incidence of dioecy overall, makes the conservation
of New Caledonia’s remaining natural vegetation an even
greater priority.
Acknowledgments
Vassar College provided financial support to L. B. Vary and
M. A. Schlessman. National Geographic Society grant 7555-
03 to M. A. Schlessman funded fieldwork. The environmental
services of the North and South Provinces of New Caledonia
provided collection permits. T. Le Borgne, A. Mouly, and the
staff of IRD Noume´ a assisted in the field. F. Tronchet helped
with identifications, and the staff of the MNHN herbarium
provided access to specimens.
Literature Cited
Adam P, G Williams 1994 Dioecy, self-compatibility and vegetative
reproduction in Australian subtropical rainforest trees and shrubs.
Cunninghamia 7:89–100.
Aitchison JC, GL Clarke, S Meffre, D Cluzel 1995 Eocene arc-con-
tinent collision in New Caledonia and implications for regional
southwest Pacific tectonic evolution. Geology 23:161–164.
Angiosperm Phylogeny Group (APG) III 2009 An update of the An-
giosperm Phylogeny Group classification for the orders and families
of flowering plants: APG III. Bot J Linn Soc 161:105–121.
Aubre´ville A, J-F Leroy, P Morat, HS MacKee, eds 1967– Flore de
la Nouvelle-Cae´donie et De´pendances. Muse´ um National d’Histoire
Naturelle, Paris. 25 vol.
Baker HG 1955 Self-compatibility and establishment after “long-dis-
tance” dispersal. Evolution 9:347–348.
——— 1967 Support for Baker’s law—as a rule. Evolution 21:853–
856.
Baker HG, PA Cox 1984 Further thoughts on dioecism and islands.
Ann Mo Bot Gard 71:244–253.
Barrett SCH 2002 The evolution of plant sexual diversity. Nat Rev
Genet 3:274–284.
——— 2006 The reproductive biology and genetics of island plants.
Philos Trans R Soc B 351:725–733.
Bawa KS 1980 Evolution of dioecy in flowering plants. Annu Rev
Ecol Syst 11:15–40.
Bayly M, DG Holmes, PI Forster, DJ Cantrill,PY Ladiges 2013 Major
clades of Australasian Rutoideae (Rutaceae) based on rbcL and atpB
sequences. PLoS ONE 8:e72493. doi:10.1371/journal.pone
.0072493.
Bernardello G, GJ Anderson, TF Stuessy, DJ Crawford 2001 A survey
of floral traits, breeding systems, floral visitors, and pollination sys-
tems of the angiosperms of the Juan Ferna´ndez Islands (Chile). Bot
Rev 67:255–308.
Bishop Museum 2002 Hawaiian arthropod checklist. http://www2
.bishopmuseum.org/HBS/checklist/query.asp?grppArthropod. Ac-
cessed May 29, 2013.
Bramwell D 2011 Biogeography and conservation of the flora of New
Caledonia. Pages 226–238 in D Bramwell, J Caujape´-Castells, eds.
The biology of island floras. Cambridge University Press,
Cambridge.
Brothers RN, AR Lillie 1988 Regional geology of New Caledonia.
Pages 325–374 in AEM Nairn, FG Stehli, S Uyeda, eds. The ocean
basins and margins. Vol 7B. Plenum, New York.
Carlquist S 1966 The biota of long-distance dispersal. IV. Genetic
systems in the floras of oceanic islands. Evolution 20:433–455.
Carpenter RJ, J Read, T Jaffre´ 2003 Reproductive traits of tropical
rain-forest trees in New Caledonia. J Trop Ecol 19:351–365.
Cayzer LW, MD Crisp, RH Telford 2000 Revision of Pittosporum
(Pittosporaceae) in Australia. Aust Syst Bot 13:845–902.
Charlesworth D 1993 Why are unisexual flowers associated with
wind pollination and unspecialized pollinators? Am Nat 141:481–
490.
Charlesworth D, B Charlesworth 1987 The effect of investment in
attractive structures on allocation to male and female functions in
plants. Evolution 41:948–968.
Chazeau J 1993 Research on New Caledonian terrestrial fauna:
achievements and prospects. Biodivers Lett 1:123–129.
SCHLESSMAN ET AL.—DIOECY ON NEW CALEDONIA 285
Cheptou PO 2012 Clarifying Baker’s law. Ann Bot 109:633–641.
Cheptou PO, F Massol 2009 Pollination fluctuations drive evolution-
ary syndromes linking dispersal and mating systems. Am Nat 174:
46–55.
Clayton JW, P Soltis, DE Soltis 2009 Recent long-distance dispersal
overshadows ancient biogeographical patterns in a pantropical fam-
ily (Simaroubaceae, Sapindales). Syst Bot 58:395–410.
Cluzel D, S Meffre, P Maurizot, AJ Crawford 2006 Earliest Eocene
(53 Ma) convergence in the southwest Pacific: evidence from pre-
obduction dikes in the ophiolite of New Caledonia. Terra Nova 18:
395–402.
Cruaud A, R Jabbour-Zahab, G Genson, S Ungricht, JY Ras-
plus 2012 Testing the emergence of New Caledonia: fig wasp mu-
tualism as a case study and a review of evidence. PLoS ONE 7:
e30941.
Darwin C 1877 The different forms of flowers on plants of the same
species. J Murray, London.
de Jong TJ, NM Waser, PGL Klinkhamer 1993 Geitonogamy: the
neglected side of selfing. Trends Ecol Evol 8:321–325.
Donovan B, J Munzinger, A Pauly, G McPherson 2013 Flower-visiting
records of the native bees of New Caledonia. Ann Mo Bot Gard
99:19–43.
Esser JH 2003 Fruit characters in Malesian Euphorbiaceae. Telopea
10:169–177.
Felsenstein J 1985 Phylogenies and the comparative method. Am Nat
125:1–15.
Gemmill CEC, GJ Allan, WL Wagner, EA Zimmer 2002 Evolution
of insular Pacific Pittosporum (Pittosporaceae): origin of the Ha-
waiian radiation. Mol Phylogenet Evol 22:31–42.
Gilmartin AJ 1968 Baker’s law and dioecism in the Hawaiian flora:
an apparent contradiction. Pac Sci 22:285–292.
Givnish TJ 1980 Ecological constraints on the evolution of breeding
systems in seed plants: dioecy and dispersal in gymnosperms. Evo-
lution 34:959–972.
Godley EJ 1979 Flower biology in New Zealand. N Z J Bot 17:441–
466.
Grandcolas P, J Murienne, T Robillard, L Desutter-Grandcolas, H
Jourdan, E Guilbert, L Deharveng 2008 New Caledonia: a very old
Darwinian island? Philos Trans R Soc B 363:3309–3317.
Gross CL 2005 A comparison of the sexual systems in the trees from
the Australian tropics with other tropical biomes: more monoecy
but why? Am J Bot 92:907–919.
Hall R 2001 Cenozoic reconstructions of SE Asia and the SW Pacific:
changing patterns of land and sea. Pages 35–56 in I Metcalfe, JMB
Smith, M Morwood, ID Davidson, eds. Faunal and floral migrations
and evolution in SE Asia–Australia. Lisse, Netherlands.
Hannecart F, Y Le´ tocart 1980 Oiseaux de Nlle Cale´donie et des Loy-
aute´s. I. Clark & Matheson, Auckland.
——— 1983 Oiseaux de Nlle Cale´donie et des Loyaute´s. II. Clark &
Matheson, Auckland.
Harder LD, SCH Barrett 1996 Pollen dispersal and mating patterns
in animal-pollinated plants. Pages 140–190 in DG Lloyd, SCH
Barrett, eds. Floral biology: studies on floral evolution in animal-
pollinated plants. Chapman & Hall, New York.
Harvey PH, MD Pagel 1991 The comparative method in evolutionary
biology. Oxford University Press, Oxford.
Heilbuth JC 2000 Lower species richness in dioecious clades. Am Nat
156:221–241.
Heilbuth JC, KL Ilves, SP Otto 2001 The consequences of dioecy for
seed dispersal: modeling the seed-shadow handicap. Evolution 55:
880–888.
Humeau L, T Pailler, JD Thompson 1999 Cryptic dioecy and leaky
dioecy in endemic species of Dombeya (Sterculiaceae) on La Re´-
union. Am J Bot 86:1437–1447.
Jaffre´ T, P Morat, J-M Veillon, F Rigault, G Dagostini 2001
Composition et caracte´risation de la flore indige`ne de Nouvelle-
Cale´donie. Institut de Recherche pour le De´veloppement, Noume´a.
Kato M, A Kawakita 2004 Plant-pollinator interactions in New Cal-
edonia influenced by introduced honey bees. Am J Bot 91:1814–
1827.
Kubitzki K, ed 1990– The families and genera of vascular plants.
Springer, Berlin. 11 vol.
Ladiges PY, D Cantrill 2007 New Caledonia–Australia connections:
biogeographic patterns and geology. Aust Syst Bot 20:383–389.
Lee DE, WG Lee, N Mortimer 2001 Where and why have all the
flowers gone? depletion and turnover in the New Zealand Cenozoic
angiosperm flora in relation to palaeogeography and climate. Aust
J Bot 49:341–356.
Lloyd DG 1985 Progress in understanding the natural history of New
Zealand plants. N Z J Bot 23:707–722.
Lloyd DG, KS Bawa 1984 Modification of the gender of seed plants
in varying conditions. Evol Biol 17:255–388.
Lowry PP 1998 Diversity, endemism, and extinction in the flora of
New Caledonia: a review. Acad Sin Monogr 16:181–206.
Lowry PP, J Munzinger, P Bouchet, H Ge´ raux, A Bauer, O Langrand,
RA Mittermeier 2004 New Caledonia. Pages 193–197 in R Mit-
termeier, GP Robles, M Hoffman, J Pilgrim, T Brooks, CG Mitter-
meier, JL Lamoreux, GAB da Fonseca, eds. Hotspots revisited:
earth’s biologically richest and most threatened ecoregions.
CEMEX, Mexico City.
Maddison WP, DR Maddison 2003 Mesquite: a modular system for
evolutionary analysis. Version 2.75. http://mesquiteproject.org.
Matallana G, T Wendt, DSD Araujo, FR Scarano 2005 High abun-
dance of dioecious plants in a tropical coastal vegetation. Am J Bot
92:1513–1519.
McLoughlin S 2001 The breakup history of Gondwana and its impact
on pre-Cenozoic floristic provincialism. Aust J Bot 49:271–300.
McPherson G, C Tirel 1987 Euphorbiace´es I. Pages 1–225 in P Morat,
HS MacKee, eds. Flore de la Nouvelle-Cale´donie et De´pendances.
Vol 14. Muse´um National d’Histoire Naturelle, Paris.
Morat P 1993 Our knowledge of the flora of New Caledonia: en-
demism and diversity in relation to vegetation types and substrates.
Biodivers Lett 1:72–81.
Morat P, T Jaffre´, F Tronchet, J Munzinger, Y Pillon, JM Veillon, M
Calopin 2012 The taxonomic database FLORICAL and character-
istics of the indigenous flora of New Caledonia. Adansonia 34:177–
219.
Pagel M 1994 Detecting correlated evolution on phylogenies: a gen-
eral method for the comparative analysis of discrete characters. Proc
R Soc B 255:37–45.
Pannell JR 2006 Effects of colonization and metapopulation dynam-
ics on the evolution of plant sexual systems. Pages 223–238 in LD
Harder, SCH Barrett, eds. Ecology and evolution of flowers. Oxford
University Press, Oxford.
Pannell JR, SCH Barrett 1998 Baker’s law revisited: reproductive as-
surance in a metapopulation. Evolution 52:657–668.
Pauly A, J Munzinger 2003 Contribution a` la connaissance des Hy-
menoptera Apoidea de Nouvelle-Cale´donie et de leurs relations avec
la flore butine´e. Ann Soc Entomol Fr 39:153–166.
Pelletier B 2006 Geology of the New Caledonian region and its im-
plications for the study of the New Caledonian biodiversity. Pages
19–32 in CE Payri, B Forges, eds. Compendium of marine species
of New Caledonia. IRD, Noume´a.
Plunkett GM, PP Lowry II 2010 Paraphyly and polyphyly in Polyscias
sensu lato: molecular evidence and the case for recircumscribing the
“pinnate genera” of Araliaceae. Plant Divers Evol 128:23–54.
Plunkett GM, PP Lowry II, MK Burke 2001 The phylogenetic status
of Polyscias (Araliaceae) based on nuclear ITS sequence data. Ann
Mo Bot Gard 88:213–230.
Raven PH, DI Axelrod 1972 Plate tectonics and Australasian paleo-
biogeography. Science 176:1379–1386.
286 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Renner SS, RE Ricklefs 1995 Dioecy and its correlates in flowering
plants. Am J Bot 82:596–606.
Richards AJ 1986 Plant breeding systems. Allen Unwin, London.
Sakai AK, WL Wagner, DM Ferguson, DR Herbst 1995aBio-
geographical and ecological correlates of dioecy in the Hawaiian
flora. Ecology 76:2530–2543.
——— 1995bOrigins of dioecy in the Hawaiian flora. Ecology 76:
2517–2529.
Sakai AK, SG Weller 1999 Gender and sexual dimorphism in flow-
ering plants: a review of terminology, biogeographic patterns, eco-
logical correlates, and phylogenetic approaches. Pages 1–31 in MA
Geber, TE Dawson, LF Delph, eds. Gender and sexual dimorphism
in flowering plants. Springer, New York.
Schellart WP, GS Lister, VG Toy 2006 A late Cretaceous and Cenozoic
reconstruction of the southwest Pacific region: tectonics controlled
by subduction and slab rollback processes. Earth-Sci Rev 76:191–
233.
Schlessman MA, DG Lloyd, PP Lowry 1990aEvolution of sexual
systems in New Caledonian Araliaceae. Mem NY Bot Gard 55:105–
117.
Schlessman MA, PP Lowry, DG Lloyd 1990bFunctional dioecism in
the New Caledonian endemic Polyscias pancheri (Araliaceae). Bio-
tropica 22:133–139.
Schlessman MA, GM Plunkett, PP Lowry, DG Lloyd 2001 Sexual
systems of New Caledonian Araliaceae: a preliminary phylogenetic
reappraisal. Edinb J Bot 58:221–228.
Soltis DE, SA Smith, N Cellinese, KJ Wurdack, DC Tank, SF Brock-
ington, NF Refulio-Rodriguez, et al 2011 Angiosperm phylogeny:
17 genes, 640 taxa. Am J Bot 98:704–730.
Swenson U, J Munzinger, I Bartish 2007 Molecular phylogeny of
Planchonella (Sapotaceae) and eight new species from New Cale-
donia. Taxon 56:329–354.
Thomson JD, J Brunet 1990 Hypotheses for the evolution of dioecy
in seed plants. Trends Ecol Evol 5:11–16.
Tirel C, J-M Veillon 2002 Pittosporaceae. Pages 1–178 in P Morat,
ed. Flore de la Nouvelle-Cale´donie. Vol 24. Muse´um National
d’Histoire Naturelle, Paris.
Vamosi JC, SP Otto 2002 When looks can kill: the evolution of sex-
ually dimorphic floral display and extinction of dioecious plants.
Proc R Soc B 269:1187–1194.
Vamosi JC, SP Otto, SCH Barrett 2003 Phylogenetic analysis of the
ecological correlates of dioecy in angiosperms. J Evol Biol 16:1006–
1018.
Vamosi JC, SM Vamosi 2004 The role of diversification in causing
the correlates of dioecy. Evolution 58:723–731.
——— 2005 Present day risk of extinction may exacerbate the lower
species richness of dioecious clades. Divers Distrib 11:25–32.
Vary LB, DL Gillen, M Randrianjanahary, PP Lowry II, AK Sakai, SG
Weller 2011 Dioecy, monoecy, and their ecological correlates in the
littoral forest of Madagascar. Biotropica 43:582–590.
Webb CJ, DG Lloyd, LF Delph 1999 Gender dimorphism in indig-
enous New Zealand seed plants. N Z J Bot 37:119–130.
Willford GE, PJ Brown 1994 Maps of late Mesozoic-Cenozoic Gond-
wana breakup: some palaeogeographic implications. Pages 5–13 in
RS Hill, ed. History of the Australian vegetation: Cretaceous to
Recent. Cambridge University Press, Cambridge.
Yampolsky C, H Yampolsky 1922 Distribution of sex forms in the
phanerogamic flora. Bibl Genet 3:1–62.