Content uploaded by Wujian Xiong
All content in this area was uploaded by Wujian Xiong on Apr 29, 2022
Content may be subject to copyright.
American Journal of Botany 107(10): 1355–1365, 2020; http://www.wileyonlinelibrary.com/journal/AJB © 2020 Botanical Society of America • 1355
Since Sprengel (1793) described detailed interactions between plants
and insects, research interest on pollination has greatly increased.
Approximately 87.5% of owering plants depend on animals, par-
ticularly insects for pollination (Ollerton et al., 2011). To date, four
insect orders are most frequently reported as pollinators: bees and
wasps (Hymenoptera), butteries and moths (Lepidoptera), true
ies (Diptera), and beetles (Coleoptera). In contrast, other insect
groups, such as cockroaches (Blattodea), true bugs (Hemiptera),
net-winged insects (Neuroptera), and crickets (Orthoptera) have
rarely been recorded and have no clear adaptations to ower feed-
ing (Kevan and Baker, 1983; Wardhaugh, 2015) with the possible
exception of scorpionies (Mecoptera) (Ollerton, 2017). However,
a few studies (Appendix S1) have established that pollination by
cockroaches may be more common than previously thought. So far,
cockroaches are known to act as pollinators of 11 plant species. In
Apocynaceae, cockroaches have been shown to act as pollinators of
the South American asclepiad Ditassa capillaris, but only as one of
a wide range of other pollinators from several other orders (unpub-
lished data of Domingos-Melo et al. [Ollerton et al., 2019]). However,
none of the above studies provided direct proof of fruit set following
single visits only by cockroaches. However, they did establish that
some oral traits, such as nocturnal opening and distinctive scents,
may be associated with a “cockroach pollination syndrome”.
Vincetoxicum is a genus of the subfamily Asclepiadoideae (fam-
ily Apocynaceae sensu Endress and Bruyns, 2000; Endress et al.,
2007). Species of this plant family are quite well studied for their
complex oral morphology and specic pollination mechanisms
(Ollerton, 1998; Ollerton and Liede, 1997, 2003; Ollerton et al., 2019).
Specialized cockroach pollination in the rare and
endangered plant Vincetoxicum hainanense in China
Wujian Xiong1, Je Ollerton2, Sigrid Liede-Schumann3, Wanyi Zhao4, Qiancai Jiang5, Hongmei Sun5, Wenbo Liao4,6, and Wenhui You1,6
Manuscript received 15 April 2020; revision accepted 16 June 2020.
1 Shanghai Key Lab for Urban Ecological Processes and Eco-
Restoration,School of Ecological and Environmental Sciences,East
China Normal University, Shanghai 200241, China
2 Faculty of Arts, Science and Technology,University of Northampton,
Waterside Campus, Northampton NN1 5PH, UK
3 Department of Plant Systematics,University of Bayreuth, Bayreuth
4 State Key Laboratory of Biocontrol and Guangdong Provincial Key
Laboratory of Plant Resources,School of Life Sciences,Sun Yat-sen
University, Guangzhou 510275, China
5 Zhongshan State-owned Forest Resources Protected Center,
Zhongshan 528400, China
6Authors for correspondence (e-mail: email@example.com,
Citation: Xiong, W., J. Ollerton, S. Liede-Schumann, W. Zhao, Q.
Jiang, H. Sun, W. Liao, and W. You. 2020. Specialized cockroach
pollination in the rare and endangered plant Vincetoxicum
hainanense in China. American Journal of Botany 107(10):
PREMISE: Species of Apocynaceae are pollinated by a diverse assemblage of animals. Here
we report the rst record of specialized cockroach pollination in the family, involving an
endangered climbing vine species, Vincetoxicum hainanense in China. Experiments were
designed to provide direct proof of cockroach pollination and compare the eectiveness of
other ower visitors.
METHODS: We investigated the reproductive biology, pollination ecology, pollinaria
removal, pollinia insertion, and fruit set following single visits by the most common insects.
In addition, we reviewed reports of cockroaches as pollinators of other plants and analyzed
the known pollination systems in Vincetoxicum in a phylogenetic context.
RESULTS: The small, pale green owers of V. hainanense opened during the night. The
owers were not autogamous, but were self-compatible. Flower visitors included beetles,
ies, ants and bush crickets, but the most eective pollinator was the cockroach Blattella
bisignata, the only visitor that carried pollen between plants. Less frequent and eective
pollinators are ants and Carabidae. Plants in this genus are predominantly pollinated by
ies, moths and wasps.
CONCLUSIONS: Globally, only 11 plant species are known to be cockroach-pollinated.
Because their range of oral features encompass similarities and dierences, dening a
“cockroach pollination syndrome” is dicult. One commonality is that owers are often
visited by insects other than cockroaches, such as beetles, that vary in their signicance
as pollinators. Cockroach pollination is undoubtedly more widespread than previously
thought and requires further attention.
KEY WORDS Apocynaceae; Asclepiadoideae; breeding system; cockroach pollination;
oral biology; pollination system; pollinaria removal; pollinia insertion; reproductive
biology; Vincetoxicum hainanense.
1356 • American Journal of Botany
Vincetoxicum contains about 150 species that are widely distributed
in Africa, Eurasia, India, and Australia (Endress et al., 2018). Only one
species Vincetoxicum philippicum (syn. Tylophora parviflora) was
previously known to open its owers and produce a scent at night
(Meve et al., 2002), but no other details about the pollinators and pol-
lination ecology of this species are available at yet. During our eld-
work in 2017, we found another species of Vincetoxicum with the
same features, Vincetoxicum (Merrillanthus) hainanense (Chun and
Tsiang, 1941; Liede-Schumann and Meve, 2018) (Apocynaceae), a
twining species from South China, the pollination ecology of which
is so far unstudied. Vincetoxicum hainanense has small (ca. 2.2 cm
in diameter), pale green, dish-shaped owers that open at night and
produce a heavily nauseating odor similar to decayed wood or rot-
In this study, we focused on the pollination and breeding system
of V. hainanense. We examined the eectiveness of cockroaches
and other oral visitors as pollinators of these owers by assess-
ing the quality (amount of pollen deposited per visit) and quantity
(number of visits) components of pollination (Herrera, 1987) to
determine their contributions to reproductive success. To this end,
we determined the number of mature fruits that V. hainanense was
able to set from pollinia inserted by cockroaches and other groups
of visitors to obtain insights in the functional roles of the visitors.
We addressed the following specic questions: (1) What oral vis-
itors pollinate V. hainanense? (2) What is the dierence in polli-
nation eectiveness among major pollinator groups? (3) How does
the pollination system of V. hainanense compare to other species
within the same clade of Tylophorinae? (4) What are the implica-
tions of our ndings for understanding the ecology and evolution
of cockroach pollination systems?
MATERIALS AND METHODS
Study site and species
e study was carried out between March and August of 2018
and February and July of 2019 in Tianxin Forest Park, Zhongshan
(22°24′39″N, 113°28′21″E), Guangdong Province, southern China.
During the observation periods in the 2 years, there were 10 sunny
days, alternating with 3 cool rainy days. Annual average tempera-
ture at the locality is 20.3°C and annual average precipitation was
1858.2 mm (http://www.zsqx.com/).
Vincetoxicum hainanense is an endangered twining vine spe-
cies narrowly distributed in Hainan and Guangdong provinces; it
reaches 3.9 m long and 2.8 m wide. During our eld investigations,
wild individuals were rarely found due to the destruction and de-
cline in quality of the habitat. In 2015, a single population of about
15 mature individuals was found in Zhongshan, Guangdong. In
2017, we started a Protection Research Program. To date, this popu-
lation is the largest one known in mainland China. e population
grows in sandy soils at the fringe of a forest, close to streams, and
also appears in mixed evergreen and deciduous forest. e forest
habitats of V. hainanense are dominated by trees of Cinnamomum
bodinieri, Cinnamomum burmanni, and Aquilaria sinensis, with an
understory of Alpinia zerumbet, Alocasia macrorrhiza, Bambusa
chungii, Ficus hirta, Commelina communis, and Urena lobata. At
this site, V. hainanense grows in three patches separated by tens to
hundreds of meters, each with 3–5 individuals, and there was no
variation among patches in observed visitation patterns. Specimens
were deposited at the Herbarium of Sun Yat-sen University,
Phenology and oral biology
e lengths of the growing season (from appearance of new leaves
to leaf drop) and owering and fruiting seasons were recorded in
2018 and in 2019. e oral biology was studied in detail using 10
tagged inorescences per patch, on which we recorded the ower
numbers per inorescence, owering period, ower lifespan, odor
production, and stigmatic status. ese owers were observed daily
as they opened, closed, and abscised. However, the owers were too
small to allow extraction of nectar with capillary tubes.
Pollen to ovule ratio and stigma receptivity
e pollen to ovule ratio (P/O) was estimated based on randomly se-
lected individuals following the method of Cruden (1977) and Dafni
(1992). In total, we collected 30 pre-anthesis owers from six individ-
uals and stored them in 70% v/v ethanol in glass bottles. e ovules
in each ovary and pollen grains in each ower were counted using a
light microscope, and the mean and standard error were estimated.
Stigma receptivity at dierent times during owering was estimated
using the benzidine–H2O2 method (Dafni, 1992). Peroxidase activity
on the stigma surface was estimated as described by He et al. (2017).
Observations of ower visitors were carried out for 40 h during
peak owering time, including days that had little to no rain, from
April 2018 and from March 2019 until the end of owering. During
an initial 5-day survey of owers, few insects were seen visiting, just
occasional Apis cerana, ants and unidentied ies, none of which
acted as pollinators. We therefore focused on nocturnal visitors,
which typically arrived at the owers shortly aer anthesis. Plants
were observed for 20–30 sessions in each patch; for each session, we
focused on six inorescences for 30 min from 19:00 to 24:00 in red
light to minimize the light eect on ower visitors. We noted all in-
sects visiting the owers and their behavior. e captured visitors of
each insect species were brought back to the laboratory, then iden-
tied and classied to the lowest taxonomic level possible. We also
examined them for the presence of pollinaria. Voucher specimens
of ower visitors were preserved in 70% v/v ethanol and deposited
at the Herbarium of SYSU, Guangzhou, China.
Pollinaria removal, pollinia insertion, and reproductive success
in major visitor groups
We performed a single-visit pollination treatment in 2019 using
the six major visitor groups identied: (1) SV1, single visit involv-
ing nymphs of Blattella bisignata; (2) SV2, single visit involving
adults of B. bisignata; (3) SV3, single visit involving adults of
Opisthoplatia orientalis; (4) SV4, single visit involving Carabidae
sp.; (5) SV5, single visit involving Diptera sp.; (6) SV6, single visit
involving Tetramorium sp. In these experiments, all SV2 were be-
tween plants, all other groups are within one plant. We covered six
ower buds on each of 300 inorescences with ne mesh nylon
bags; when owers opened, we exposed them for one visit by an
individual of each insect group. Following single-visit exposure,
half of the visited owers were collected and preserved in 70%
October 2020, Volume 107 • Xiong et al.—Cockroach pollination in Vincetoxicum hainanense • 1357
v/v ethanol to quantify pollinaria removal and pollinia insertion.
With a dissecting microscope, the gynostegium was removed from
the corolla, owers with pollinaria removed and pollinia inserted
were counted, and the removed pollinaria and inserted pollinia
were counted for each ower. In total, we collected and examined
32 owers for SV1 and SV2, 10 owers for SV3, 38 owers for SV4,
30 owers for SV5, and 40 owers for SV6. e other half of the
owers were covered to quantify fruit production, we recorded the
number of mature seed pods that resulted from the visits. In total,
the caged number of single visits was 140 for SV1, 90 for SV2, 30
for SV3, 120 for SV4, 90 for SV5, 110 for SV6.
e owers were too small to pollinate by hand, so we used four
treatments to understand the breeding system of V. hainanense: (1)
Autonomous self-pollination: In each patch, 180 ower buds were en-
closed in ne mesh nylon bags to exclude insects. (2) Open-pollination:
Fruit set was assessed aer open-pollination by tagging 453 (2018)
and 470 (2019) unbagged ower buds, then observing them weekly
from the day the owers rst opened until mature fruits formed. (3)
Self-pollination: bçased on the results of pollinaria removal, pollinia
insertion, and reproductive success in major visitor groups, we treated
SV1, SV4, and SV6 as self-pollination. (4) Cross-pollination: We
treated SV2 as cross-pollination. For all four treatments, we recorded
the number of pods initiated, aborted, and matured. Mature fruits
were collected in July 2019, 4 months aer owering, their dry mass
was weighed, and numbers of seeds were counted.
e data set of Liede-Schumann et al. (2016; study 15858 of TreeBase,
http://purl.org/phylo/ treeb ase/phylo ws/study/ TB2:S15858) was
used to assess the phylogenetic position of V. hainanense, add-
ing all relevant partial sequences for this species (ENA numbers
LR700196–LR700200, LR792268). We applied maximum likelihood
(ML) tree inference and nonparametric bootstrapping (BS) analysis
and Bayesian inference (BI) to our data set. e ML tree inference
and BS relied on RAxML v. 8.2.10 (Stamatakis, 2014), which imple-
ments a general-time-reversible substitution model for nucleotides
(Rodriguez et al., 1990), as congured on the CIPRES platform
(Miller et al., 2010). e Bayesian inference analysis was implemented
in MrBayes 3.2 (Huelsenbeck and Ronquist, 2001; Ronquist and
Huelsenbeck, 2003) with partitions unlinked, parameters estimated
independently, and four simultaneous independent runs. Records of
pollinators from the Pollinators of Apocynaceae Database (Ollerton
et al., 2019) were mapped onto the resulting phylogeny.
Review of cockroach-pollinated owers and their features
To assess the ower characteristics associated with cockroach pol-
lination, we collected information from previous studies on spe-
cies pollinated by cockroaches. With the results of this study added
to the assessment, we compared those oral features qualitatively.
We used one-way ANOVA followed by a Tukey test for multi-
ple comparisons to test the dierences of mature fruit set in dif-
ferent pollination treatments, including single-visit pollination,
autonomous self-pollination, and open-pollination. Analyses were
done using SPSS version 20.0 (IBM, Armonk, NY, USA). In the text,
data are presented as mean values and their standard errors.
Phenology and oral biology
In 2018, new leaves appeared in the middle of January, owering
began in the middle of March, peaked toward the end of April,
and nished in the middle of May. Fruiting began in the middle
of April, and the rst fruits began to dehisce and release seeds
in mid-July. Later-developing pods reached maturity and released
seeds up to early August, the leaves dropped from the middle of
September. In 2019, the phenology was similar, but 2–3 weeks
Vincetoxicum hainanense has axillary cymes of pale green ow-
ers. Each inorescence has 22.0 ± 7.1 (mean ± SE) owers, but only
1–5 of the owers in an inorescence are in ower at any one time.
e lifespan of an intact inorescence is 16 to 20 days, with each
ower lasting 5–7 days. During anthesis, one petal opens rst, and
then the other four petals open sequentially, all within 5–8 min,
usually at around 19:00. On the rst day, the corolla is pale green. By
the third day, the petals have become pale yellow. Aer the ower
has been open 5 or 6 days, all petals turn deep yellow, and owers
oen remain on the plant for 1 or 2 days aer closing.
Due to the small size of the owers (see below), we were unable
to test for nectar. But the behavior of ower visitors suggests that
nectar is present. e owers have a heavy, nauseatingly sweet scent,
reminiscent of rotting fruit. Although the odor was present during
the day, it became noticeably stronger at sunset and remained strong
during the night. On hot humid days, the smell was oen apparent
some distance from the plant.
e owers have ve-lobed, yellow corollas measuring approx-
imately 22.1 ± 1.5 (mean ± SE) mm in diameter, and have ve ca-
lyx glands (Fig. 1A). Follicles are fusiform, 12.3–16.5 cm long, and
produce about 100 ovate hairy seeds inside. e two free ovaries
of owers are surrounded by ve united stamens. On the gynoste-
gium, ve stigma slits are arranged. Anther appendages are ovate
and membranous, overlying the style-head. Each pollinarium has
two oblong pollinia, measuring about 224 × 110 µm. Pollen grains
are irregularly shaped and measure ca. 14 µm in diameter.
P/O ratio and stigma receptivity
e mean number of pollen grains per pollinium was 491.3 ± 20.5
(mean ± SE, N = 30) and the average number of ovules per ovary was
180.2 ± 5.8 (mean ± SE, N = 30). us, because there are 10 pollinia
and two ovaries per ower, the P/O ratio was 13.6 ± 0.5 (mean ± SE).
Stigmas were recorded as receptive during the rst 3 days of ower-
ing in 100% of observations, with the greatest receptivity on day 3;
none of the owers was receptive on or beyond day 4.
During our periods of observation, at least seven species of insects and
arachnids, representing seven families, were observed and collected.
Two species of cockroaches, Blattella bisignata and Opisthoplatia
orientalis, were the most frequent visitors (Table 1). With an average
1358 • American Journal of Botany
of 11.1 and 4.7 visits per hour, nymphs of B. bisignata and adults of
O. orientalis visited multiple inorescences and owers within one
plant, while adults of B. bisignata visited multiple owers between
separate plants in rapid succession. And while hunting for nectar
they stayed on a ower for 1.9–2.6 min and picked up pollinaria and
deposited pollinia. Diptera also had a higher number of visits, they
visited multiple inorescences between separate plants, but no pol-
linaria were found attached to their body. Ants (Hymenoptera) and
Coleoptera visited inorescences during evenings to collect nectar,
but they did not move between separate plants. Pollinaria were found
attached to their legs or mouth. With the lowest number of visits per
hour, Orthoptera and Araneida also came to visit inorescences at
night, but they rarely touched the reproductive parts as they stayed
on the petals and rarely moved. A number of larger insects had pol-
linaria attached on their legs or mouth (Table 1). Figure 2 shows the
visitor groups observed in all patches in Zhongshan.
Pollinaria removal and pollinia insertion by major groups of
Our data show that nymphs of B. bisignata were the most eective visi-
tors (Table 2), with 93.8% of the single visited owers examined having
pollinaria removed, and 25.0% with pollinia
inserted. e average number of pollinaria
removed and pollinia inserted per ower
is 4.0 and 0.4, respectively. Adults of B.
bisignata and Carabidae also had a higher
pollinaria removal and pollinia insertion
rates, whilst adults of O. orientalis and
Tetramorium sp. had lower removal and in-
sertion rates. Among 30 owers examined
that had single visits by dipterans, no pol-
linaria were removed, and no pollinia were
found inserted. Comparing the data from
Table 2, we can see that the rate of pollinaria
removal is exactly correlated with the rate of
pollinia insertion for the six major groups of
visitors (r = 0.92, p < 0.01). In all cases, de-
posited pollinia were no longer attached to
the translator arm. In three owers (15.8%),
there was a double insertion of two pollinia
in dierent anther slits of one ower.
Eect of major visitor groups on plant
Mature fruit production in V. hainanense
depended on dierent groups of major vis-
itors (Fig. 3) with signicant dierences in
fruit set observed (F5, 12 = 12.60, P < 0.001).
Adults of B. bisignata were the most eec-
tive ower pollinators; nymphs of B. bisig-
nata, Tetramorium sp. (ants) and beetles
(Carabidae) were the next most eective.
No fruit set was found in owers visited by
adults of O. orientalis or dipterans (Fig. 3).
When all insects were excluded, none of
the owers set fruit.
Breeding system and fruit set
e open-pollination and bag-pollination treatments in 2018 and
2019 (Fig. 3) showed that for the 540 bagged owers, no visibly
enlarged ovaries were initiated, and none bore green follicles, in-
dicating that V. hainanense possesses no potential for automatic
selng. Of the 453 (2018) and 470 (2019) unbagged open-polli-
nated owers, 26 (6.0%) and 27 (5.8%) produced mature fruits.
Usually only one fruit was produced per ower, but among 110
fruits examined, and 15 owers (13.6%) produced double fruits.
e number of lled and unlled seeds per capsule in open polli-
nation was 86.3 ± 13.4 (mean ± SE) and 5.3 ± 2.5 (mean ± SE). e
dry mass of lled seeds per capsule was 2.2 ± 0.4 g (mean ± SE).
Fruit set diered signicantly between self-pollination (SV1,
SV4, and SV6) and cross-pollination (SV2) (Fig. 3) (F3, 8 = 16.21,
P < 0.001). Fruit set in self-pollination averaged 1.4~3.6% per
ower, while in cross-pollination it was 7.7%. e self-pollina-
tion produced 74.2 ± 6.2 (mean ± SE) lled seeds and 5.0 ± 2.3
(mean ± SE) unlled seeds per capsule, and dry mass of lled
seeds was 2.0 ± 0.2 g (mean ± SE). e cross-pollination averaged
92.2 ± 12.4 (mean ± SE) lled seeds and 4.2 ± 1.3 (mean ± SE)
unlled seeds, and averaged 2.3 ± 0.3 g (mean ± SE) dry mass
FIGURE 1. (A) Flower of Vincetoxicum hainanense. (B) Top of a gynostegium. (C) Side of gynostegium.
Sc, stigmatic chamber. S, slit. (D) pollinarium with a pair of pollinia joined by a corpuscle.
October 2020, Volume 107 • Xiong et al.—Cockroach pollination in Vincetoxicum hainanense • 1359
per capsule. Not all the initiated pods reached maturity, the pro-
portion of aborted pods aer open pollination was 15.2%, 24.1%
aer self- and 8.3% aer cross-pollination.
Vincetoxicum hainanense is retrieved together with a sam-
ple of V. cissoides from Papua New Guinea and V. philippicum
from the Philippines in a fully supported clade, representing
the rst-branching clade in non-African Vincetoxicum (Fig.
4). Vincetoxicum cissoides is a widespread and variable species
in the Philippines, Malaysia, Indonesia, Brunei, and Papua New
Guinea, characterized by large, inated and possibly oating
fruits (Schneidt, 1999); however, nothing about its oral biology
is known. Vincetoxicum philippicum, a Philippine endemic, is
the only species known in Vincetoxicum with strictly nocturnal
(i.e., they close during the day), heavily scented owers (Meve
et al., 2002). Conrmed or potential pollinators of Vincetoxicum
include ies, wasps, butteries, and possibly even birds (Fig. 4),
in pollination systems that vary from the highly specialized to
the extremely generalized, with at least one obligately autoga-
mous species (Yamashiro and Maki, 2006; Yamashiro et al., 2008;
Nakahama et al., 2013).
Cockroach-pollinated owers and their features
We found eight studies related to cockroach pollination of owers
from nine plant families and 11 species (Table 3). ese plant species
with cockroach pollination share some oral features, but there is no
consistent motif of a cockroach pollination syndrome: their owers
are pale yellow, green, or white; they can be strongly scented or un-
scented; open at night and/or during the day; produce nectar, pollen,
or liquid secretions as a reward; and be monoecious or dioecious.
One of the most consistent features of their pollination ecology is
that they all have pollinators other than cockroaches (Table 3).
In our study of the pollination biology of Vincetoxicum hainan-
ense, we demonstrated that the species is pollinated mainly by cock-
roaches with a smaller, but still signicant, contribution by ants and
Carabidae. Due to the combined eect of ecient pollen transfer
and higher outcrossing rate, cockroaches are the most important
pollinators compared to other insect taxa. Inside Vincetoxicum,
knowledge of oral biology and pollination is hitherto almost
restricted to the two temperate clades, the Far Eastern Clade and
the Vincetoxicum s.s. clade (Ollerton et al., 2019). In both, y pol-
lination is predominant, while moths have been reported only
from the Far Eastern Clade, and wasps seem more common in the
Vincetoxicum s.s. clade. Pentatropis nivalis, a member of the sister
genus of Vincetoxicum, has been reported as having a fairly gen-
eralist pollination system. For V. cernuum, the only African mem-
ber of Vincetoxicum studied, bird and beetle pollination have also
been reported (Woodell, 1979). From the rather large-owered V.
biglandulosum from New Caledonia, buttery pollination has been
reported (Kato and Kawakita, 2004). Unfortunately, no pollinator
observations are available for nocturnal Philippine V. philippicum,
V. cissoides, or any of the Australian species.
Despite the small size of each pollinium, we determined the
number of pollen grains per pollinium was 491.3, and the number
of ovules per ovary about 180.2. us the P/O ratio of V. hainanense
is 13.6 on average. is low P/O ratio is in the range of other 10
species of Asclepiadoideae (3.8–18.4) reported by Cruden (2000).
P/O ratio is known to be lower in Asclepiadoideae for their pollen
grains are packaged in pollinia (Wyatt et al., 2000). A low P/O ratio
TABLE 1. Nocturnal oral visitors to Vincetoxicum hainanense owers with number of their visits, mean time staying, and pollen attachment on their bodies.
Mean no. visits/h
on ower (min)
Total no. insects
Body parts with
Blattella bisignata (nymphs) 11.1 ± 3.4 2.6 ± 0.2 152 ＋legs, mouth
B. bisignata (adults) 7.0 ± 2.2 1.9 ± 0.2 206 ＋legs, mouth
Opisthoplatia orientalis (adults) 4.7 ± 1.9 2.1 ± 0.3 126 －
Unidentiﬁed sp. 3.7 ± 1.2 2.5 ± 0.2 46 ＋mouth
Unidentiﬁed sp. 4.3 ± 2.0 6.9 ± 0.4 128 －
Tetramorium sp. 4.2 ± 2.0 4.1 ± 0.4 38 ＋Legs
Hexacentrus sp. 1.4 ± 0.5 10.4 ± 0.4 36 －
Unidentiﬁed sp. 1.2 ± 0.6 4.5 ± 0.3 31 －
1360 • American Journal of Botany
FIGURE 2. Principal ower visitors of Vincetoxicum hainanense. (A) Cecidomyiidae sp. (Diptera). (B) Tetramorium sp. (Hymenoptera). (C) Unidenied
sp. (Araneida). (D) Hexacentrus sp. (Orthoptera). (E) Carabidae sp. (Coleoptera). (F) Blattella bisignata (Blattodea), nymph on a ower. (G) Blattella bisig-
nata (Blattodea), adult on a ower. (H) Opisthoplatia orientalis (Blattodea), adult on inorescence. Scale bar = 5 mm.
October 2020, Volume 107 • Xiong et al.—Cockroach pollination in Vincetoxicum hainanense • 1361
reects a trade-o between pollen investment and reproductive
return, with minimal energetic investment, but large reproductive
return if pollination is successful (Cruden, 1977).
Successful pollination in plants is dependent on the timings of
ower opening, stigma receptivity and anther dehiscence (Renata
et al., 2006). In V. hainanense, stigmas were receptive for the rst 3
days, with maximum receptivity on the third day of anthesis.
e results of the V. hainanense pollination experiments showed
that many groups of insects visited the owers at night and trans-
ported pollinaria, indicating that they were potential pollinators.
However, their functional roles in pollination vary. e nymphs
of Blattella bisignata, ants, and Carabidae were regularly spotted
crossing owers and inorescences on the same plant. e adults
of B. bisignata were regularly found ying between separate plants.
e adults of O. orientalis were also spotted on the same plants,
but they have a much larger body and were barely able to touch the
reproductive organs of owers. us, we regard nymphs of B. bisig-
nata, ants, and Carabidae to be more likely responsible for self and
geitonogamous pollination, and adults of B. bisignata to be more
likely responsible for cross-pollination. In contrast, Diptera also vis-
ited owers frequently, but they made no contribution to fruit set.
Worker ants are ightless with a limited movement range, and tradi-
tionally have been regarded as poor pollinators (Beattie et al., 1984;
Proctor et al., 1996). However, recent studies found remarkable
cases of ant pollination, which result in viable seed set (Svensson,
1985; Gómez et al., 1996; Beattie, 2006) even in another genus of
small-owered asclepiad, Ditassa (Domingos-Melo et al., 2017).
Intriguingly, in one of these species (D. capillaris), cockroaches also
form part of the pollinating fauna (unpublished data of Domingos-
Melo et al. [Ollerton et al., 2019]). Carabidae probably are predators
(Lövei and Sunderland, 1996) and generally not regarded as pol-
linators (Knop et al., 2017), but they could be looking for prey on
the owers and pollinating as they move around. In V. hainanense,
both ants and Carabidae visited the inorescences throughout the
evening and resulted in fruit set, while Carabidae had a higher re-
moval and insertion rate. Based on these observations, we consider
Carabidae more eective than ants, at least in this population.
e average fruit set under open-pollinated conditions was
5.9%, a little higher than the 0.33–5.0% reported in Asclepias
(Wyatt and Broyles, 1990). e proportion of owers that ma-
tured double fruits was 13.6%, consistent with the 5.0–24.4%
recorded in Asclepias (Wyatt and Broyles, 1990). Our present
data from experimental bag-pollination, self-pollination, and
cross-pollination of V. hainanense, however, indicate automatic
self-pollination (autogamy) does not occur, and geitonogamy re-
sulted in signicant lower fruit set than cross-pollination. In an
attempt to explain the relatively low fruit set in asclepiads, two
hypotheses have been developed: environmental resource limita-
tion and insucient pollination. In this experiment, we observed
only one follicle developed, with one carpel aborted. A similar
phenomenon was found in some species of Asclepias (Willson
and Price, 1980). Wyatt (1980) concluded that early competition
among ovaries within the umbel is the main factor that leads to
low fruit set. Another possible cause of low level in fruit set may
be geitonogamy, which will result in
fruit set only when the plant individual
is self-compatible. In a self-incompati-
ble species, geitonogamy may result in
reduced fruit set through pollen dis-
counting and the outcross-pollen in-
terference (Holsinger et al., 1984; Lloyd
and Webb, 1986; Schoen and Lloyd,
1992; Klinkhamer and de Jong, 1993;
Finer and Morgan, 2003). From the re-
sults of self and cross-pollination, and
pollinia removal and insertion, we con-
clude that low fruit set in V. hainanense
can be partially explained by compatible
pollen limitation. We reach this conclu-
sion because there is a low (an average of
1.5–1.9%), but denite, level of self-com-
patibility (Appendix S3), and among
those pollinators, only adults of B. bisig-
nata could carry outcross pollen and
pollinate between plants, thus making a
TABLE 2. Pollinaria removal and pollinia insertion in Vincetoxicum hainanense owers after single visits by the major insect groups.
No. owers with
pollinaria removed (%)
No. owers with
pollinia inserted (%)
SV1 (Nymphs of Blattella bisignata) 32 30 (93.8) 8 (25.0) 4.0 0.4
SV2 (Adults of B. bisignata) 32 26 (81.3) 6 (18.8) 3.0 0.3
SV3 (Adults of Opisthoplatia orientalis) 10 2 (20.0) 0 (0.0) 0.5 0.0
SV4 (Carabidae) 38 24 (63.2) 3 (7.9) 1.8 0.1
SV5 (Diptera) 30 0 (0.0) 0 (0.0) 0.0 0.0
SV6 (Tetramorium sp.) 40 4 (10.0) 2 (5.0) 0.4 0.1
FIGURE 3. Mean fruit set (±SD) by Vincetoxicum hainanense after dierent pollination treatments
in 2018 and 2019. Open = open pollination, Bag = autonomous self-pollination, SV1 = single visit
involving the nymphs of Blattella bisignata, SV2 = single visit involving the adults of B. bisignata, SV3
= single visit involving the adults of Opisthoplatia orientalis, SV4 = single visit involving the Carabidae
sp., SV5 = single visit involving the Diptera, SV6 = single visit involving the Tetramorium sp., Values
are number of owers. Seed sets with the same letters did not dier signicantly at the 5% level. Test
outputs are provided in Appendix S2.
1362 • American Journal of Botany
signicantly higher contribution to fruit set. Evidence of simul-
taneous limitation by both resources and pollen in V. hainanense
will require further exploration.
Based on our review, all of the cockroach-pollinated plants have
owers that open at night, and most are pale in color. Where scent
has been reported, most of these owers have a strong unpleasant
or sweet odor, also nocturnally emitted. Cockroaches are an ancient
insect group, which evolved at least 300 million years ago in the
Carboniferous period and are now found globally in a wide range of
habitats; most are nocturnal, actively emerging at night to feed omniv-
orously (Bell et al., 2007). Being nocturnal, cockroaches mainly use the
sense of smell to nd food, though some of them are attracted to pale
colors (Brenner et al., 1988). We suggest that cockroach-pollinated
plants possess certain adaptive features for pollination by cockroaches,
especially the nocturnal anthesis and scent emission and pale oral
colors. However, these features overlap to some extent with nocturnal
beetle pollination, though beetle-pollinated owers in Apocynaceae
vary enormously in their phenotypes (Ollerton et al., 2017, 2019).
Recent discoveries from Myanmar amber suggest that attraction and
reward of cockroaches by gymnosperms and angiosperms may be an
ancient pollination system that dates to at least the Cretaceous period
(Hinkelman, 2020; Hinkelman and Vršanská, 2020). Further fossil ma-
terial and extant cases of cockroach pollination need to be found and
described to draw more general conclusions about the origins of this
pollination system and similarities of oral phenotype in dierent
is is the rst record of the pollinators and reproductive biol-
ogy for V. hainanense. Additionally, this study is the rst record
of a cockroach-pollination system in Apocynaceae. We evaluated
the owering phenology, oral morphology, P/O ratio, stigma
FIGURE 4. Phylogenetic tree of Tylophorinae based on trnT–L, trnL–F, and trnH–psbA spacers, trnL and trnG introns, ITS, and ETS derived from Bayesian
analysis using MrBayes. The position of V. hainanense is highlighted, and the species of its clade are spelled out. For other clades, only species for which
pollinator information is available are named. Clades without pollinator information have been summarized by triangles. Clades fully supported by
maximum likelihood analysis and Bayesian inference are indicated by bold lines. Clade designations correspond to Liede-Schumann et al. (2016).
Insect icons in black represent conrmed pollinators; those in gray are ower visitors but not conrmed as pollinators. Insect icons are licensed under
<div>Icons made by <a href = "https://www.ati con.com/autho rs/freepik" title = "Freepik">Freepik</a> from <a href = https://www.ati con.com/
TABLE 3. Flowers reported to be visited and often pollinated by cockroaches and oral features.
Plant family and
species Plant sex
colour Primary pollinator(s) Source
monoecious d, n ﬂ. T not reported pink or yellow cockroaches Schapheer et al.,
Uvaria elmeri monoecious n others, p strong odor like
creamy white or
cockroaches, ﬂies Nagamitsu and
Uvariopsis dioica monoecious d, n others not reported pale yellow ants, cockroaches,
Mertens et al.,
hermaphroditic d, n others, p strong
cockroaches, ants This study
monoecious d, n ﬂ. T not reported yellow cockroaches Schapheer et al.,
monoecious d, n nc, ﬂ. T no detectable
pale pink ants, cockroaches,
monoecious d, n nc, ﬂ. T no detectable
yellow ants, cockroaches Kawakita and
Puya venusta monoecious d, n nc not reported purple cockroaches Schapheer et al.,
Clusia blattophila dioecious n others buttery pale yellow cockroaches, Diptera Vlasáková et al.,
monoecious d, n nc, p sweet
white social wasps, crickets,
Oenothera acaulis monoecious n p not reported white cockroaches Schapheer et al.,
Abbreviations: d: daytime (06:00-18:00); n: nighttime (18:00-06:00); d, n: both day and night. nc: nectar; p: pollen; ﬂ. T: ﬂoral tissue; others: secretions, exudates, etc.
October 2020, Volume 107 • Xiong et al.—Cockroach pollination in Vincetoxicum hainanense • 1363
1364 • American Journal of Botany
receptivity, oral visitors, fruit set, and breeding system and placed
our observations in a phylogenetic context. Above all, we provided
direct proof that mature fruits set predominantly in owers visited
only by cockroaches. As the principal pollinator of V. hainanense,
cockroaches are quite eective in pollination, and adult cock-
roaches of Blattella bisignata are quite essential for the reproduc-
tion of the species. Fruit production is naturally low and likely the
result of compatible pollen limitation. However, many other ques-
tions remain to be addressed, in particular, the phylogeography and
genetic diversity of the species are interesting areas requiring fur-
ther study, as is variation in the pollinators in dierent populations.
We thank L. Feng, J.-W. Pan, F. Ye, G.-B. Tan, and Q.-C. Zou for eld
assistance; Zhongshan State-owned Forest Resources Protected
Center for their co-operation; Dr. Fenglong Jia, Professor of Sun
Yat-sen University, for identication of pollinators and for valuable
information about other insects; and two anonymous reviewers for
valuable comments on the manuscript. is work was supported
by the Financial Fund Project of Zhongshan on the Endangered
Mechanism and Articial Breeding of the Rare and Endangered Plant
Vincetoxicum hainanense (Project No. ZSCG2016-06-044-C0809),
the Fourth National Survey on Chinese Material Medical Resources
Program (Project No. 2017-152-003 and 2018-523-001) from the
State Administration of Traditional Chinese Medicine of China.
W.-J.X. designed the experiments, collected and analyzed the eld
data, and draed the manuscript. W.-Y.Z., Q.-C.J. and H.-M.S.
conducted the eld-pollination experiments. W.-H.Y. and W.-B.L.
gave much advice on the details of all experiments. J.O. and S.L.S.
provided the data on pollinators and phylogeny of Tylophorinae,
advised on interpretation of the results, and revised the manuscript.
All authors read and approved the nal manuscript.
Please see Appendix S4 for raw data of oral visitors, pollinaria
removal and pollinia insertion, and pollination treatments. All se-
quences generated for this study are deposited in ENA (numbers
LR700196–LR700200, LR792268). ey have been added to the
alignment published under study number 15858 of TreeBase, http://
purl.org/phylo/ treeb ase/phylo ws/study/ TB2:S15858.
Additional Supporting Information may be found online in the
supporting information tab for this article.
APPENDIX S1. Studies related to cockroach (Blattodea)
APPENDIX S2. One-way ANOVA to test the dierences of mature
fruit set by Vincetoxicum hainanense in dierent pollination treat-
ments (refer to Fig. 3).
APPENDIX S3. Results of self-pollination on fruit set in
Vincetoxicum hainanense based on three isolated plants.
APPENDIX S4. Complete data sets for (A) nocturnal oral visitors
to Vincetoxicum hainanense owers. (B) pollinaria removal and
pollinia insertion, and (C) fruit set aer dierent pollination treat-
ments in 2018 and 2019.
Beattie, A. J. 2006. e evolution of ant pollination systems. Botanische
Jahrbücher 127: 43–55.
Beattie, A. J., C. Turnbull, R. B. Knox, and E. G. Williams. 1984. Ant inhibition
of pollen function: a possible reason why ant pollination is rare. American
Journal of Botany 71: 421–426.
Bell, W. J., L. M. Roth, and C. A. Nalepa. 2007. Cockroaches: ecology, behavior,
and natural history. Johns Hopkins University Press, Baltimore, MD.
Brenner, R. J., R. S. Patterson, and P. G. Koehler. 1988. Ecology, behavior, and dis-
tribution of Blattella asahinai (Orthoptera: Blattellidae) in central Florida.
Annals of the Entomological Society of America 81: 432–436.
Chun, W. Y., and Y. Tsiang. 1941. Merrillanthus hainanensis. Sunyatsenia 6: 107–
108, pl. 17–18.
Cruden, R. W. 1977. Pollen–ovule ratios: a conservative indicator of breeding
systems in owering plants. Evolution 4: 32–46.
Cruden, R. W. 2000. Pollen grains: why so many? Plant Systematics and Evolution
Dafni, A. 1992. Pollination ecology: a practical approach. Oxford University
Press, Oxford, UK.
Domingos, M. A., T. L. Nadia, and I. C. Machado. 2017. Complex owers and
rare pollinators: Does ant pollination in Ditassa show a stable system in
Asclepiadoideae (Apocynaceae)? Arthropod-Plant Interactions 11: 339–349.
Endress, M. E., and P. V. Bruyns. 2000. A revised classication of the Apocynaceae
s.l. Botanical Review 66: 1–56.
Endress, M. E., S. Liede-Schumann, and U. Meve. 2007. Advances in Apocynaceae:
the enlightenment, an introduction. Annals of the Missouri Botanical Garden
Endress, M. E., U. Meve, D. J. Middleton, and S. Liede-Schumann. 2018.
Apocynaceae. In J. Kadereit and V. Bittrich [eds.], Flowering plants. Eudicots.
e families and genera of vascular plants, vol.15, 207–411. Springer, Cham,
Finer, M. S., and M. T. Morgan. 2003. Eects of natural rates of geitonogamy on
fruit set in Asclepias speciosa (Apocynaceae): evidence favoring the plant’s
dilemma. American Journal of Botany 90: 1746–1750.
Gómez, J. M., R. Zamora, J. A. Hódar, and D. García. 1996. Experimental study of
pollination by ants in Mediterranean high mountain arid habitats. Oecologia
He, G. R., F. R. Hu, J. Ming, C. Liu, and S. X. Yuan. 2017. Pollen viability and stigma
receptivity in Lilium during anthesis. Euphytica 213: 231.
Herrera, C. M. 1987. Components of pollinator ‘quality’: comparative analysis of
a diverse insect assemblage. Oikos 50: 79–90.
Hinkelman, J. 2020. Earliest behavioral mimicry and possible food begging in a
Mesozoic alienopterid pollinator. Biologia 75: 83–92.
Hinkelman, J., and L. A. Vršanská. 2020. Myanmar amber cockroach with protrud-
ing feces contains pollen and a rich microcenosis. Science of Nature 107: 13.
Holsinger, K. E., M. W. Feldman, and F. B. Christiansen. 1984. e evolution of
self fertilization in plants: a population genetic model. American Naturalist
Huelsenbeck, J. P., and F. Ronquist. 2001. MRBAYES: Bayesian inference of phy-
logeny. Bioinformatics 17: 754–755.
Kato, M., and A. Kawakita. 2004. Plant–pollinator interactions in New Caledonia in-
uenced by introduced honey bees. American Journal of Botany 91: 1814–1827.
Kawakita, A., and M. Kato. 2002. Floral biology and unique pollination sys-
tem of root holoparasites, Balanophora kuroiwai and B. tobiracola
(Balanophoraceae). American Journal of Botany 89: 1164–1170.
October 2020, Volume 107 • Xiong et al.—Cockroach pollination in Vincetoxicum hainanense • 1365
Kevan, P., and H. Baker. 1983. Insects as ower visitors and pollinators. Annual
Review of Entomology 28: 407–453.
Klinkhamer, G. L., and T. J. Jong. 1993. Attractiveness to pollinators: a plant’s
dilemma. Oikos 66: 180–184.
Knop, E., L. Zoller, R. Ryser, C. Gerpe, M. Hörler, and C. Fontaine. 2017. Articial
light at night as a new threat to pollination. Nature 548: 206–209.
Liede-Schumann, S., and U. Meve. 2018. Vincetoxicum (Apocynaceae–
Asclepiadoideae) expanded to include Tylophora and allies. Phytotaxa 369:
Liede-Schumann, S., R. Khanum, A. S. Mumtaz, I. Gherghel, and A. Pahlevani.
2016. Going west – A subtropical lineage (Vincetoxicum, Apocynaceae:
Asclepiadoideae) expanding into Europe. Molecular Phylogenetics and
Evolution 94: 436–446.
Lloyd, D. G., and C. J. Webb. 1986. e avoidance of interference between the pre-
sentation of pollen and stigmas in angiosperms. I. Dichogamy. New Zealand
Journal of Botany 24: 135–162.
Lövei, G. L., and K. D. Sunderland. 1996. Ecology and behavior of ground beetles
(Coleoptera: Carabidae). Annual Review of Entomology 41: 231–256.
Mertens, J. E. J., R. Tropek, F. F. Dzekashu, V. Maicher, E. B. Fokam, and Š. Janeček.
2017. Communities of ower visitors of Uvariopsis dioica (Annonaceae) in
lowland forests of Mt. Cameroon, with notes on its potential pollinators.
African Journal of Ecology 56: 146–152.
Meve, U., R. Omlor, and S. Liede. 2002. A new combination in Tylophora
(Apocynaceae, Asclepiadoideae) from the Philippines. Systematics and
Geography of Plants 72: 27–32.
Miller, M. A., W. Pfeier, and T. Schwartz. 2010. Creating the CIPRES Science
Gateway for inference of large phylogenetic trees. In 2010 Proceedings of
the Gateway Computing Environments Workshop (GCE), New Orleans, LA,
USA, 1–8. https://doi.org/10.1109/GCE.2010.5676129
Nagamitsu, T., and T. Inoue. 1997. Cockroach pollination and breeding system
of Uvaria elmeri (Annonaceae) in a lowland mixed-dipterocarp forest in
Sarawak. American Journal of Botany 84: 208–213.
Nakahama, N., A. Ushimaru, and Y. Isagi. 2013. Reproductive traits and ower
visitor fauna of Vincetoxicum pycnostelma Kitag. In Nishitani area of
Takarazuka City, Hyogo Prefecture. Bulletin of Kansai Organization for
Nature Conservation 35: 115–123.
Ollerton, J. 1998. Sunbird surprise for syndromes. Nature 394: 726–727.
Ollerton, J. 2017. Pollinator diversity: distribution, ecological function, and con-
servation. Annual Review of Ecology, Evolution, and Systematics 48: 353–376.
Ollerton, J., S. Dötterl, K. Ghorpadé, A. Heiduk, S. Liede-Schumann, S.
Masinde, U. Meve, et al. 2017. Diversity of Diptera families that pollinate
Ceropegia (Apocynaceae) trap owers: an update in light of new data and
phylogenetic analyses. Flora 234: 233–244.
Ollerton, J., and S. Liede. 1997. Pollination systems in the Asclepiadaceae: a survey and
preliminary analysis. Biological Journal of the Linnean Society 62: 593–610.
Ollerton, J., and S. Liede. 2003. Corona structure in cynanchum: link morphol-
ogy to function. Ecotropica 9: 107–112.
Ollerton, J., S. Liede-Schumann, M. E. Endress, U. Meve, A. R. Rech, A.
Shuttleworth, H. A. Keller, M. Fishbein, et al. 2019. e diversity and evolu-
tion of pollination systems in large plant clades: Apocynaceae as a case study.
Annals of Botany 123: 311–325.
Ollerton, J., R. Winfree, and S. Tarrant. 2011. How many owering plants are
pollinated by animals? Oikos 120: 321–326.
Proctor, M., P. Yeo, and A. Lack. 1996. e natural history of pollination. Harper
Collins, Hampshire, UK.
Renata, R., M. Rosa, B. Eric, V. Maria, and H. Huges. 2006. Preliminary stud-
ies toward genetic improvement of annatto (Bixa orellana L.). Scientia
Horticulturae 109: 165–172.
Rodriguez, F., J. L. Oliver, A. Marin, and J. R. Medina. 1990. e general stochastic
model of nucleotide substitution. Journal of Theoretical Biology 142: 485–501.
Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic infer-
ence under mixed models. Bioinformatics 19: 1572–1574.
Schapheer, C., M. M. Lopez-Uribe, A. Vera, and C. A. Villagra. 2017. Distribution,
habitat use and plant associations of Moluchia brevipennis (Saussure, 1864)
(Blattodea: Ectobiidae): an endemic cockroach from Chilean Mediterranean
Matorral biome. Revista Brasileira de Entomologia 61: 114–122.
Schneidt, J. 1999. Taxonomy and biodiversity of the genus Tylophora
(Apocynaceae: Asclepiadoideae) in Malesia. Ph.D. dissertation, University of
Aberdeen, Aberdeen, UK.
Schoen, D. J., and D. G. Lloyd. 1992. Self- and cross-fertilization in plants. III.
Methods for studying modes and functional aspects of self-fertilization.
International Journal of Plant Sciences 153: 381–393.
Sprengel, C. K. 1793. Das Entdeckte Geheimnis der Nature Im Bau und in der
Befruchtung der Blumen. Mayer and Müller, Berlin, Germany.
Stamatakis, A. 2014. RaxML version 8: a tool for phylogenetic analysis and
post-analysis of large phylogenies. Bioinformatics 30: 1312–1313.
Suetsugu, K. 2019. Social wasps, crickets and cockroaches contribute to pollina-
tion of the holoparasitic plant Mitrastemon yamamotoi (Mitrastemonaceae)
in southern Japan. Plant Biology 21: 176–182.
Svensson, L. 1985. An estimate of pollen carryover by ants in a natural popu-
lation of Scleranthus perennis L. (Caryophyllaceae). Oecologia 66: 373–377.
Vlasáková, B., B. Kalinová, M. H. G. Gustafsson, and H. Teichert. 2008.
Cockroaches as pollinators of Clusia a. Sellowiana (Clusiaceae) on
Inselbergs in French Guiana. Annals of Botany 102: 295–304.
Vlasáková, B., J. Pinc, F. Jůna, and Z. K. Varadínová. 2019. Pollination eciency
of cockroaches and other oral visitors of Clusia blattophila. Plant Biology
Wardhaugh, C. W. 2015. How many species of arthropods visit owers?
Arthropod-Plant Interactions 9: 547–565.
Willson, M. F., and P. W. Price. 1980. Resource limitation of fruit and seed produc-
tion in some Asclepias species. Canadian Journal of Botany 58: 2229–2233.
Woodell, S. R. 1979. e role of unspecic pollinators in the reproductive suc-
cess of Aldabran plants. Philosophical Transactions of the Royal Society of
London, B, Biological Sciences 286: 99–108.
Wyatt, R. 1980. e reproductive biology of Asclepias tuberosa: I. Flower number,
arrangement, and fruit-set. New Phytologist 85: 119–131.
Wyatt, R., and S. B. Broyles. 1990. Reproductive biology of milkweeds (Asclepias):
recent advances. In S. Kawano [ed.], Biological approaches and evolutionary
trends in plants, 255–272. Academic Press, Toronto, ON, Canada.
Wyatt, R., S. B. Broyles, and S. R. Lipow. 2000. Pollen–ovule ratios in milkweeds
(Asclepiadaceae): an exception that probes the rule. Systematic Botany 25:
Yamashiro, T., and M. Maki. 2006. A comparative study of the reproductive char-
acter and genetic diversity of an autogamous Tylophora matsumurae and
its progenitor Tylophora tanakae (Apocynaceae-Asclepiadoideae). Plant
Systematics and Evolution 256: 55–67.
Yamashiro, T., A. Yamashiro, J. Yokoyama, and M. Maki. 2008. Morphological
aspects and phylogenetic analyses of pollination systems in the Tylophora-
Vincetoxicum complex (Apocynaceae-Asclepiadoideae) in Japan. Botanical
Journal of the Linnean Society 93: 325–341.