全 文 ：Recieved: 2012–04–11　　　　Accepted: 2012–04–26
Supported by National Basic Research Program of China (973 Program) (2007CB411600), the Knowledge Innovation Project of The National Key
Technology Rand D Program (2008BAC39B02), the Chinese Forest Biodiversity Monitoring Network and the Foreign Exchange
Program National Founder (31011120470)
Author: ZHU Peng, PhD student, major in molecular ecology. E-mail: rocerzhu2008@126.com
* Corresponding author. E-mail: wzf@scib.ac.cn
珍稀濒危物种格木传粉方式和交配系统的初步研究
朱鹏1,2,3, 王峥峰1,2*, 叶万辉1,2, 曹洪麟1,2, Saravanan THAVAMANIKUMAR4
(1. 中国科学院华南植物园， 中国科学院退化生态系统植被恢复与管理重点实验室， 广州 510650; 2. 中国科学院华南植物园， 数字化植物
园广东重点实验室， 广州 510650; 3. 中国科学院大学， 北京 100049; 4. Department of Forest and Ecosystem Science, University of Melbourne,
Creswick, VIC 3363, Australia)
摘要： 为研究格木(Erythrophleum fordii Oliv.)的传粉方式和交配系统，在鼎湖山国家级自然保护区对珍稀濒危树种格木进行野
外观察 , 花的微器官(花粉和柱头)用扫描电镜观察，并进行控制授粉实验和遗传分析。结果表明，格木具有颜色亮丽和散发香
味的总状花序，格木花粉包被厚重的粘性物质，柱头没有易于捕获花粉的结构，风媒传粉没有坐果，推断格木应该是通过虫媒
传播花粉，或者主要是通过虫媒传播。遗传分析和在不同繁殖环境的两棵目标树的坐果情况表明，格木偶尔可以自交产生后
代，但异交占据着绝对的优势，鼎湖山格木种群的异交率达到 90.6%。格木传粉方式和交配系统的研究有利于对格木进行保护
和保育。
关键词： 格木； 传粉方式； 交配系统； 遗传分析
doi: 10.3969/j.issn.1005–3395.2013.01.005
Preliminary Studies on Pollination and Mating System of Rare and
Endangered Plant Erythrophleum fordii Oliv.
ZHU Peng1,2,3, WANG Zheng-feng1,2*, YE Wan-hui1,2, CAO Hong-lin1,2, Saravanan THAVAMANIKUMAR4
(1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of
Sciences, Guangzhou 510650, China; 2. Guangdong Key Laboratory of Digital Botanical Garden, South China Botanical Garden, Chinese Academy
of Sciences, Guangzhou 510650, China; 3. University of Chinese Academy of Sciences, Beijing 100049, China; 4. Department of Forest and Ecosystem
Science, University of Melbourne, Creswick, VIC 3363, Australia)
Abstract: Although a few efforts have been taken to conserve rare and endangered species Erythropleum fordii
Olive., information about its pollination and mating system still remains unknown. The pollination and mating
system of E. fordii through performing the field observation, SEM observation of pollen and stigma, controlled
pollination experiment and genetic analyses in Dinghushan Nature Reserve, China were studied. Erythrophleum
fordii has brightly colorful inflorescences and sends forth delicate odor, its pollen is coated by strong mucous
substance and its stigmas have not specific structure for pollen capturing, and meanwhile the inflorescences
pollinated by wind did not fruit at all, therefore it was inferred that E. fordii might be entomophilous. According
to genetic analyses and fruiting of two object trees under contrasting reproductive situations, E. fordii can be
selfing occasionally but out-crossing is dominant with out-crossing rate in the Dinghushan E. fordii community for
90.6%. This preliminary elucidation of pollination and mating system would benefit the conservation of E. fordii.
Key words: Erythrophleum fordii; Pollination; Mating system; Genetic analysis
热带亚热带植物学报　2013， 21(1)： 38~44
Journal of Tropical and Subtropical Botany
第1期 39
Erythrophleum fordii Oliv. is a precious
hardwood tree species belonging to the informal
group Dimorphandra of the tribe Caesalpinieae
(Fabaceae: Caesalpinioideae) and naturally occurring
in tropical and subtropical areas of South China
and Vietnam[1–2]. In addition to E. fordii, 15 other
congener species can also be found in tropical areas of
Africa, tropical and subtropical areas of Eastern Asia
and North Australia[2]. The E. fordii trees produce
quite hard, heavy, durable and erosion-resistant wood,
popularly called as ‘iron wood’, which is generally
used for furniture, bridge, shipbuilding, etc.[3]. In
addition, E. fordii is also a species with medicinal
properties[4] used by native Chinese as invigoration
and blood circulation promoting agents[5]. Because
of its superior wood and medicinal uses, it has been
historically over-exploited leading to the serious
population reduction[6]. To protect it from further
population reduction, it has been listed in both China
and Vietnam plant red data book[7–8] and also placed
into the ‘endangered’ category by IUCN[1].
Erythrophleum fordii normally occurs as
a canopy tree species in the south subtropical
evergreen broad-leaved monsoon forest, thus it
plays an important role in maintaining some forest
communities in the south subtropical areas. Few
studies have been conducted on the conservation
biology of E. fordii, such as endangered mechanism[6],
seed germination[9], in situ and ex situ conservation[10],
population dynamics[11–12], phenotypic variation of
fruit and seed related to geographic distribution[13],
and so on. However, we still know little about the
pollination and mating system of E. fordii, which are
limited factors for endangered species’ conservation.
Therefore, we expected to learn about the pollination
and mating system of E. fordii to promote our
conservation for this endangered precious species and
in turn for its communities.
1 Material and Methods
1.1 Species and study site
Erythrophleum fordii has small single flowers
with ten stamens and one pistil, which form racemes
and in turn panicle[14]. Single small flower does not
have colorful corolla but only the calyx-like green
corolla[14]. The flowering of E. fordii is prosperous
yearly after maturing but fruiting is different among
years[6].
Our study site is located in the first national
nature reserve of China, Dinghushan Nature Reserve
(23°10′ N / 112°34′ E). This reserve is designated
into the MAB (Man and the Biosphere) programme
in 1979. The main vegetation is lower subtropical
evergreen broadleaf forest[11]. In contrast to the
disturbed surrounding forests, the Dinghushan Nature
Reserve comprises rare primary forests of at least 400
years of age (www.unesco.org). The annual average
rainfall is about 1956 mm with distinct dry and wet
season, of which the dry season occurs between
September and February with an annual average
temperature of 20.9℃[15].
In the core area of the reserve, there is a small
E. fordii dominated community (Circa 2 hm2), the
unique E. fordii community. Additionally, one mature
single E. fordii individual was found near the office
area (referred to MST later) in the reserve with
strong human disturbance, 1.5 km away from the E.
fordii community. Apart from the mature E. fordii
individuals in the community and the MST, no other
mature E. fordii individuals could be found in the
reserve. Comparing mature E. fordii individuals in the
community which could exchange their pollens easily,
MST has a quite different reproductive environment
with receiving pollens from other E. fordii individuals
hardly because of its isolated status.
1.2 Flower visitor observation
Since the tree height of a mature E. fordii
is normally between 20 m and 30 m, we set up
bamboo shelves with a platform on the top of shelves
surrounding the trunk of trees to facilitate flower
visitor observation, flower specimen collection and
controlled pollination experiment later. One mature
tree in the community (referred to PT later) and
the MST were selected as the object trees. During
朱鹏等：珍稀濒危物种格木传粉方式和交配系统的初步研究
40 第21卷热带亚热带植物学报
florescence (May) flower visitors were observed on the
top of the object trees in the morning (9∶00 – 11∶30 am)
and afternoon (2∶00 – 5∶30 pm) in 10 consecutive
days. We could not capture visitor specimens to conduct
further research in the laboratory because the insect
visitors observed normally were far from our platform.
1.3 Stigma and pollen observation
Fresh single flowers and anther specimens were
collected and fixed in 4% glutaraldehyde in 0.1 mol L-1
phosphate buffer at pH 7.4 and then brought back
for stigma, anther and pollen examination using light
microscopy (LM) and scanning electron microscopy
(SEM) in the laboratory. When doing the observation
with SEM, the fixed samples were rinsed three times
in the same buffer and fixed again with 1% osmium
tetroxide. Samples were then mounted on metal stubs
using double-sided adhesive tape and sputter-coated
with platinum using JFC-1600 auto fine coater (JEOL
Ltd., Tokyo, Japan). The stigmas were observed with
a ZEISS-510Meta LSM (German) and a JSM-6360LV
SEM (JEOL Ltd., Tokyo, Japan), anthers with a LM
and a JSM-6360LV SEM and pollen grains with a
JSM-6360LV SEM.
1.4 Controlled pollination experiment
Controlled pollination experiment was designed
with five treatments for each object tree: (1) Net
bagging (NB): four panicles without any single flower
blooming were bagged using four nylon net bags
(1.5 mm × 1.5 mm) to protect from animal pollination
but allow wind pollination and tagged; (2) Paper
bagging (PB): four panicles without any single flower
blooming were bagged using four closed paper bags
to test self-pollination and tagged; (3) Human-assisted
pollination with pollens from the same tree (HPS);
(4) Human-assisted pollination with pollens from
the other object tree (HPO); (5) We also tagged four
panicles without any treatment as control. However,
during doing HPS and HPO, we found that the
followers were strongly impaired after androecium
wipe-off because small single flowers (approximately
width × height = 2 mm × 10 mm and length of stamen =
4 mm) are too squashing together at the raceme, so
we doubted the impairment actually would affect the
ability of fruiting, thus finally we did not complete
HPS and HPO.
1.5 Examination of mating system using genetic
analyses
The exhaustive survey for E. fordii individuals
in the E. fordii community by setting up 10 m × 10 m
small survey boxes over the whole community.
Overall, 528 E. fordii individuals were found in
the whole community, but only 498 individuals
were sampled for further genetic analyses. We did
not sample those small individuals of H < 0.3 m
and less than 10 leaves to avoid anthropogenic
mortality leading to disturbing population’s natural
regeneration. Tree height (H) and diameter at breast
height (DBH) were measured for individuals of H ≥
2 m and only H was measured for individuals of H <
2 m. According to Huang[6] and our field observation,
based on DBH and/or H all sampled individuals were
classified into four cohorts: (i) adult (DBH ≥ 15 cm,
N = 78), (ii) juvenile (DBH < 15 cm and H ≥ 2 m,
N = 29), (iii) sapling (2 m > H ≥ 0.5 m, N = 124) and
(iv) seedling (H < 0.5 m, N = 267). And under MST
we found three small individuals (0.3 – 0.5 m at tree
height, Table 1), which plus MST were sampled for
further genetic analyses either.
Total DNA was extracted from tissue preserved in
dry condition with silica gel using the cetyltrimethyl
ammonium bromide (CTAB) method[16]. Nine
microsatellites isolated previously[17] were used to
genotype all the samples. PCR amplified products
were separated by 6% denaturing polyacrylamide
gel electrophoresis (PAGE) and visualized with
silver staining. Two genetic analyses were carried
out to study the mating system including genotype
comparison and mating rate calculation based on
parentage analysis.
Genotype comparison　　Due to MST is an
isolated adult in office area, the small individuals
found under it were unlikely to be grown up from
seeds delivered from other E. fordii trees given its
第1期 41
seed properties of gravity dispersal and not attracting
birds, but rather produced by MST. Therefore we did
direct genotype comparison between MST and the
three small individuals under it to check whether the
small individuals had come from the out-crossing of
MST with pollens from other trees or from selfing of
MST. If small individuals are from selfing, we would
not expect to find alleles in them that are not identical
to those of MST and we could also expect to find the
inclination of the heterozygous loci in MST becoming
homozygous loci in small individuals.
Parentage analysis　　Parentage analysis can
statistically assign individuals with their most likely
parents and it has been used as an effective method
to investigate the mating system. We performed the
parentage analysis by using Cervus[18–20] with using all
adults as candidate parents (n = 78) for saplings (n =
124) and seedlings (n = 267). The type of parentage
analysis used here was the model of parent pair
(sex unknown) considering self fertilization. Once
individuals were assigned with their most likely
parents, we then calculated both mating rates of
out-crossing and selfing of E. fordii in the E. fordii
community. In the assignments, only the individuals
assigned with one pair of parents with the mother and
the father being the same tree were viewed as selfing,
all other cases including one parent, one pair of
parents with the mother and the father being different
trees and multiple pairs of parents were viewed as
out-crossing.
2 Results
2.1 Flower visitors
The beetles (Cerambycidae) , butterf l ies
(Papilionidae), bees and wasps were observed to be
the main visitors of E. fordii (Fig. 1). We did not find
any birds visit the inflorescences.
2.2 Stigma and pollen observation
Stigma has an obvious ornament-absent stigmatic
cavity (Fig. 2: A). Pollen grains released as isopolar,
tricolporate monads with a size of a mean P (polar
length) × mean E (equatorial width) = 19.4 μm ×
36.9 μm and the surface of pollen grains were psilate-
finely perforate or perforate (Fig. 2: D,E). Strong
sticky coats were observed on the surface of pollen
grains making them tightly stick to each other (Fig. 2:
C – E).
2.3 Controlled pollination
In the controlled pollination experiment, the
treatments of both PB and NB of two object trees did
not bear any fruits. For the Control, MST did not bear
fruits either while PT did. For the whole fruiting of
two object trees, MST was found only having sparse
fruits while PT having a mass of fruits.
2.4 Examination of mating system using genetic
analyses
Genotype comparison between the three small
individuals under MST and the MST (Table 1)
showed that four loci (Gm 1048, Gm 2024, Gm 4040
and Gm 4058) which were heterozygous in MST had
two, three and four loci becoming homozygous in O-1,
O-2 and O-3 respectively with all alleles identical
to the alleles in MST. All the other loci which were
homozygous in MST were still homozygous in all the
three small individuals with identical alleles between
tree small individuals and MST.
朱鹏等：珍稀濒危物种格木传粉方式和交配系统的初步研究
Table 1 Genotype comparison between mature single tree found in office area (MST) and three small individuals under MST
Individual name DBH or Height
Microsatellite locus
Gm 1048 Gm 1052 Gm 2024 Gm 2062 Gm 2065 Gm 4040 Gm 4058 Gm 5b3 Gm 65
MST 42.8 cm 2/3 2/2 2/4 1/1 5/5 6/9 3/5 4/4 2/2
O-1 0.33 m 2/2 2/2 4/4 1/1 5/5 6/9 3/5 4/4 2/2
O-2 0.42 m 2/2 2/2 2/2 1/1 5/5 9/9 3/5 4/4 2/2
O-3 0.36 m 2/2 2/2 2/2 1/1 5/5 6/6 3/3 4/4 2/2
MST: Mature single E. fordii isolated in office area; O-1, O-2, O-3: No. 1, No. 2, and No. 3 offspring under MST, respeatively.
42 第21卷热带亚热带植物学报
Fig. 2 Stigma and pollens under SEM. A. Style and stigma; B. Close-up stigma of Erythrophleum africanum showing one pollen grain within stigmatic
cavity[24]; C. Dehisced anther (insert in C for the dehisced anther examined under LM); D – E. Pollens.
Fig. 1 Flower visitors during florescence. Beetle (A,C), butterflies (B), bee (D) and wasp (E).
In parentage assignment (Table 2), 385 individuals
out of 391 seedlings and saplings were assigned with
parents from adults in the E. fordii community, of
which 36 individuals were assigned with one pair of
parents having mother and father being the same tree,
i.e. selfing rate of 0.094, and 349 individuals were
assigned with parents having mother and father being
different trees, i.e. out-crossing rate of 0.906.
3 Discussion
For non-hydrophytic plants, anemophily and
第1期 43
entomophily are two main pollination types of the
cross-pollination. Anemophilous plants are usually
characterized by the reduction of the normal means
of attracting and rewarding animal visitors like dull-
colored corolla and faint scent, while entomophilous
plants often develop the inverse means[21–22]. For E.
fordii, the inflorescences with mass of single small
flowers having orange anthers tightly distributing on
them produce a very bright color and the flowers also
send forth a delicate odor, which would attract animal
visiting. Meanwhile, in our field survey, beetles,
butterflies, bees and wasps were observed visiting
on the flowers during florescence. In addition, the
pollen grains of E. fordii were found to be clothed
by strong mucous substance, which would increase
the dispersing difficulty for pollen by wind, although
they were found to belong to pollens of relative small
size according to Harder’s[23] review and have psilate-
finely perforate or perforate ornamentation surface.
The stigma of E. fordii does not have any specific
structures promoting pollen capturing while only
has an ornament-absent stigmatic cavity, similar to
some other Erythrophleum species[24], which would
make the stigma have a low ability to capture the
pollens delivered only by wind. Furthermore, in our
controlled pollination experiment, net bagging of both
object trees that was designed to test wind pollination
did not bear fruits at all. Accordingly, E. fordii might
deliver its pollen by insects rather than by wind, i.e.
entomophilous plant. However, more evidences from
direct observation of pollination behaviors of insects
are needed for further confirming the inference here.
Petit & Hampe[25] summarized that many
trees can self but not one is prominently selfing[26].
According to the direct genotype comparison between
the MST and the three small individuals found under
it which were likely produced by MST, we found
that the four loci which were heterozygous in MST
had two, three and four loci becoming homozygous
in O-1, O-2 and O-3, respectively, with all alleles
identical to the alleles in MST, and meanwhile all
the other loci which were homozygous in MST were
still homozygous in three small individuals. Give
the isolated status of the MST, the long distance
between MST and E. fordii community which might
be the only heterogenous pollen source for MST
and the selfing inclination exhibited from genotype
comparison, the three small individuals might have
been produced by the selfing of MST, especially the
O-3 which had all loci being homozygous. Further,
through parentage analysis between seedlings/
saplings and adults in the E. fordii community, 36
seedlings/saplings were assigned with one pair of
parents having mother and father being the same
tree, which also gave the evidence that E. fordii
can be selfing. However, the selfing rate was quite
lower than out-crossing rate (90.6%), showing that
E. fordii is dominantly out-crossing. Meanwhile, in
our field survey, we found that the inflorescences
without controlled pollination treatment of two object
tree (PT and MST) had quite different amounts of
fruits with PT having a mass of fruits but MST only
having exiguous fruits. PT was growing in the E.
fordii community while MST is isolated in the office
Table 2 Results of parentage assignment and from which the rates of selfing and out-crossing calculated
Types of parent assignment for offspring Number Rate
SP (Selfing)　 36 0.094†
DP (Out-crossing) 349 0.906†
One parent 21 ---
One pair of parents 302 ---
Multiple pairs of parents 26 ---
No assigned parents 6 ---
Total of offspring 391# ---
SP: Offspring having mother and father being the same tree; DP: Offspring having mother and father being different trees; †: Calculated from
excluding the offspring with no assigned parents; #: Sum of saplings (n = 124) and seedlings (n = 267).
朱鹏等：珍稀濒危物种格木传粉方式和交配系统的初步研究
44 第21卷热带亚热带植物学报
area, so PT could easily obtain heterogenous pollens
from other trees accompanying with it but MST
inversely. The contrasting reproductive situations
and the contrasting productions of PT and MST also
suggested that E. fordii should be a dominantly out-
crossing species and might have some mechanism to
constrain selfing although E. fordii can be selfing.
Acknowledgements　　We thank Prof. HUANG Zhong-liang
for his help in setting up the bamboo shelves and WU Lin-fang
for his help to collect the samples in the field. We also thank
HU Xiao-ying for her help on the microscopy and XU Gou-
liang and Prof. WU Hong-ji for their helps to identify insects.
References
[1]　 Nghia N H. Erythrophleum fordii [OL]// IUCN. IUCN Red List
of Threatened Species. Version 2011.2.1998 [2011–12–26]. http://
www.iucnredlist.org
[2]　 Chen D Z, Zhang D X, Kai L. Erythrophleum R. Br [M]// Flora
Peipublicae Popularis Sinicae, Tomus 10. Beijing: Science Press,
2010: 1–49.(in Chinese)
[3]　 Fang X F, Fang B Z. Wood physical and mechanical properties
of Erythrophleum fordii in southern Fujian [J]. J Fujian For Sci
Techn, 2007, 34 (2): 146–147.(in Chinese)
[4]　 Chen J S, Zhen S. Chinese Virose Plant [M]. Beijing: Science
Press, 1987: 1–321.(in Chinese)
[5]　 Du D, Qu J, Wang J M, et al. Cytotoxic cassaine diterpenoid-
diterpenoid amide dimers and diterpenoid amides from the leaves
of Erythrophleum fordii [J]. Phytochemistry, 2010, 71(14/15):
1749–1755.
[6]　 Huang Z L, Guo G Z, Zhang Z P. A study about endangered
mechanism of Erythrophleum fordii [J]. Acta Ecol Sin, 1997, 17
(6): 671–676.(in Chinese)
[7]　 Fu L K, Jin J M. China Plant Red Data Book: Rare and
Endangered Plants [M]. Beijing: Science Press, 1992: 1–378.(in
Chinese)
[8]　 Anon. Red Data Book of Vietnam, Science and Technology [M].
Hanoi: Publishing House, 1996: 1–360.
[9]　 Yi G L, Luo J H, Lin G R, et al. Effects of different treatment on
germination of Erythrophleum fordii seeds [J]. J Fujian For Sci
Techn, 2004, 31(3): 68–70.(in Chinese)
[10]　 Nghia N H. Status of forest genetic resources conservation
and management in Vietnam [M]// Luoma-aho T, Hong L T,
Ramanatha V, et al. Forest Genetic Resources Conservation and
Management. Serdang, Malaysia: IPGRI-APO, 2004: 290–301.
[11]　 Shi J H, Huang Z L, Yi W M, et al. Dynamics of Erythrophleum
fordii community and conservation strategies [J]. J NW For
Univ, 2005, 20(3): 65–69.(in Chinese)
[12]　 Wei S G, Li L, Liu H G, et al. Analyses of the dynamic state of
Erythrophleum fordii population [J]. Ecol Environ, 2008, 17(1):
285–289.(in Chinese)
[13]　 Zhao Z G, Guo J J, Sha E, et al. Geographic distribution and
phenotypic variation of fruit and seed of Erythrophleum fordii in
China [J]. Chin Bull Bot, 2009, 44(3): 338–344.(in Chinese)
[14]　 General Station of Tree Seed Work and State-owned Tree Farm
of State Forestry Administration, China. Seeds of Woody Plants
in China [M]. Beijing: Chinese Forestry Press, 2001: 285–286.(in
Chinese)
[15]　 Kong G H. Dinghushan Biosphere Reserve: Ecological Research,
History and Perspective [M]. Beijing: Science Press, 1993: 3–5.
(in Chinese)
[16]　 Doyle J J, Doyle J L. A rapid DNA isolation procedure for small
quantities of fresh leaf tissue [J]. Phytochem Bull, 1987, 19(1):
11–15.
[17]　 Zhu P, Ye W H, Wang Z F, et al. Isolation and characterization
of ten polymorphic microsatellite in the endangered tree
Erythrophleum fordii Oliv. [J]. Conserv Genet, 2009, 10(4):
1017–1019.
[18]　 Marshall T C, Slate J, Kruuk L E B, et al. Statistical confidence
for likelihood-based paternity inference in natural populations
[J]. Mol Ecol, 1998, 7(5): 639–655.
[19]　 Slate J, Marshall T, Pemberton J. A retrospective assessment of
the accuracy of the paternity inference program CERVUS [J].
Mol Ecol, 2000, 9(6): 801–808.
[20]　 Kalinowski S T, Taper M L, Marshall T C. Revising how the
computer program CERVUS accommodates genotyping error
increases success in paternity assignment [J]. Mol Ecol, 2007,
16(5): 1099–1106.
[21]　 Weberling F. Morphology of Flowers and Inflorescences (English
Translation) [M]. Cambridge: Great Britain at the University
Press, 1989: 319–322.
[22]　 Edlund A F, Swanson R, Preuss D. Pollen and stigma structure
and function: The role of diversity in pollination [J]. Plant Cell,
2004, 16(1): S84–S97.
[23]　 Harder L D. Pollen-size comparisons among animal-pollinated
angiosperms with different pollination characteristics [J]. Biol J
Linn Soc, 1998, 64(4): 513–525.
[24]　 Banks H, Himanen I, Lewis G P. Evolution of pollen, stigmas
and ovule numbers at the caesalpinioid-mimosoid interface
(Fabaceae) [J]. Biol J Linn Soc, 2010, 162(4): 594–615.
[25]　 Petit R J, Hampe A. Some evolutionary consequences of being a
tree [J]. Ann Rev Ecol Evol Systs, 2006, 37(1): 187–214.
[26]　 Hamrick J L, Godt M J W. Effects of life history traits on genetic
diversity in plant species [J]. Philos T Roy Soc Lond Ser B: Biol
Sci, 1996, 351(1/2/3): 1291–1298.