See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273158845

Reflexed flowers in Aeschynomene amorphoides (Fabaceae: Faboideae): A

mechanism promoting pollination specialization?

Article in Botanical Journal of the Linnean Society · April 2015

DOI: 10.1111/boj.12264

CITATIONS READS

15 569

4 authors, including:

Samuel Carleial Cesar Dominguez

Universität Konstanz Instituto de Ecología Universidad Nacional Autónoma de México
18 PUBLICATIONS 251 CITATIONS 141 PUBLICATIONS 2,207 CITATIONS

SEE PROFILE SEE PROFILE

Teresa Terrazas
Universidad Nacional Autónoma de México
362 PUBLICATIONS 5,451 CITATIONS

SEE PROFILE

All content following this page was uploaded by Samuel Carleial on 24 September 2020.

The user has requested enhancement of the downloaded file.

bs_bs_banner

Botanical Journal of the Linnean Society, 2015, 177, 657–666. With 4 figures

Reflexed flowers in Aeschynomene amorphoides

(Fabaceae: Faboideae): a mechanism promoting
pollination specialization?

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

SAMUEL CARLEIAL1,2, ALFONSO DELGADO-SALINAS1*, CÉSAR A. DOMÍNGUEZ3 and
TERESA TERRAZAS1
1
Instituto de Biología, Departamento de Botánica, Universidad Nacional Autónoma de México,
Apartado Postal 70-233, 04510, México, Distrito Federal, Mexico
2
University of Konstanz, Department of Biology, Universitätsstrasse 10, D-78457, Konstanz, Germany
3
Instituto de Ecología, Departamento de Ecología Evolutiva, Universidad Nacional Autónoma de
México, Apartado Postal 70-275, 04510, México, Distrito Federal, Mexico

Received 29 August 2014; revised 16 October 2014; accepted for publication 9 January 2015

This study aims to understand the role of floral traits in determining the reproductive biology of the leguminous
shrub Aeschynomene amorphoides, endemic to western Mexico, which has unusually orientated flowers. We
investigated the floral biology, pollination and breeding system based on a combination of morphological studies
and field experiments, using controlled pollinations in a natural environment. The architecture and reflexed
position of A. amorphoides flowers facilitate precise placement of pollen on the body of the pollinator, but this has
a cost to A. amorphoides in terms of available flower resources. These costs to reproduction success are set against
the attraction of a specialized pollinator, Tetraloniella jaliscoensis, which is capable of manipulating this unique
pollination system in papilionoid (or flag) flowers. © 2015 The Linnean Society of London, Botanical Journal of
the Linnean Society, 2015, 177, 657–666.

ADDITIONAL KEYWORDS: Aeschynomene – Chamela – flower orientation – papilionoid flower – pollina-

tion – western Mexico.

INTRODUCTION nation mechanisms in Faboideae have been classified

into four types, based mainly on the different ways the
Most species and genera in subfamily Faboideae or
pollen and stigma protrude from the keel petals and
Papilionoideae of Fabaceae (= Leguminosae) have
touch a pollinator. These different mechanisms can be
typical pea or papilionoid flowers. These zygomorphic
operated either only once, as in the explosive type, or
flowers comprise a standard (vexillum or banner) petal
several times, as in the valvular type, in which anthers
(adaxially placed), two wing or lateral petals and two
and stigma protrude from the keel, but move back to
(usually partially fused and abaxially placed) keel
their original position after pollinator visitation; piston
petals, which conceal the androecium and gynoecium
type, where pollen is pushed out of the keel tip; or
(Faegri & van der Pijl, 1971). There is considerable
brush type, where pollen is exposed from the keel by a
variation in this subfamily in size, colour, texture, and
style pollen brush as pollinator visits take place
the interlocking and lever mechanisms among the
(Arroyo, 1981; Endress, 1994). Adaptive floral diver-
petals. These modifications of the floral ground plan
gence, however, can be present among closely related
have evolved in parallel with the evolution of a wide
genera or even within a single genus. Faboideae are a
diversity of potential pollinators (Leppik, 1966). Polli-
highly diverse group of plants and are present
throughout the world, but are especially abundant in
*Corresponding author. E-mail: [email protected] seasonally dry tropical forests (SDTFs). In Mexico,

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666 657
658 S. CARLEIAL ET AL.

SDTFs house c. 6000 plant species, of which about half of Aeschynomene spp. are scarce. Griffith & Forseth
are endemic and c. 10% are legumes (Villaseñor & (2002, 2003) studied seed dispersal of A. virginica (L.)
Ortiz, 2014). Britton, Sterns & Poggenb. and its reproductive com-
The genus Aeschynomene L. subgenus Ochopodium petence in open areas. McKellar, Deren & Quesenberry
Vogel comprises a group of species with a high level of (1991) reported that A. americana L. has low rates of
endemism in the Mexican tropical dry forests (Rudd, cross-pollination in natural environments and under
1955). Their breeding systems are mostly unknown. cultivation. Ramírez & Brito (1990) reported that there
One species in this section is Aeschynomene amor- is an increase in seed-set in the self-compatible species
phoides (S.Wats.) Rose ex B.L.Rob., which has showy A. pratensis Small after hand-pollination. Regarding
branches bearing numerous inflorescences with Aeschynomene pollen as a food resource for bees, pollen

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

purple–red flowers. The unusual orientation of the grains of A. brevipes Benth. and A. martii Benth. were
flowers of A. amorphoides has never been described in recovered in post-emergence nests of Centris tarsata
taxonomic treatments, and there is no published (Dórea et al., 2010). Von der Ohe & Dustmann (1996)
account of the floral orientation and its effect on the mentioned that A. americana flowers could contribute
species pollination and breeding system. almost 16% of the pollen found in commercialized
In this study, we aim to gain an understanding of the Central American honey.
reproductive biology of A. amorphoides and to examine
floral morphology and pollination mechanisms associ-
ated with the reflexed flowers. In addition, pollination FLORAL BIOLOGY AND POLLINATION
observations and crossing experiments were carried To study and illustrate the flowers, dissections using
out to study the breeding system of A. amorphoides. an Olympus SZH10 stereoscope were made. Flower
description includes petal shape, anther–stigma dis-
tance and flower orientation. The pollen/ovule (P/O)
MATERIAL AND METHODS ratio (Cruden, 1977) was obtained from a single flower
STUDY SITE to give an estimate of pollen and ovule numbers
Field work was conducted at the Chamela Biological produced per flower. To estimate the number and
Station of the National Autonomous University of viability of pollen grains, acetocarmine and lactophe-
Mexico, UNAM (19°29′N; 105°01′W), in October and nol blue were used, respectively (Kearns & Inouye,
November 2008 and 2009. The Chamela region is 1993). Ovule number was counted directly after flower
largely dominated by SDTFs. The climate is tropical dissections (N = 60). Flower production per plant and
dry to sub-humid warm, the Aw type of Köppen- the average number of flowers per inflorescence were
Geiger’s system (Peel, Finlayson & McMahon, 2007). estimated by choosing 20 individuals evenly distrib-
Mean temperature in the area is 24.6 °C, and there is uted inside the forest and in open areas. For each of
a well-defined rainy season (average annual precipita- these individuals, five intermediate reproductive
tion: 748 mm) occurring from June to October branches were taken and their respective number of
(García-Oliva, Camou & Maass, 2002). flowers were counted (N = 20 × 5). Additionally, nectar
qualitative composition was determined by gas chro-
matography at the Chromatography Lab (LCR–LSA–
STUDY SPECIES IQUI) of the Chemistry Institute at Universidad
Of the 80 species of Faboideae occurring in the Nacional Autónoma de México. Flower structures were
Chamela region (Lott, 2002), A. amorphoides (tribe first dehydrated in an ethanol series, dried, coated
Dalbergieae: Aeschynomene subgenus Ochopodium) with gold (Emitech K500) and then examined under a
stands out for its dense, massive blooming. Based Hitachi 2460-N scanning electron microscope. Newly
mainly on its morphology, plants of this species have collected specimens were deposited in the National
been classified in four different genera (Adesmia Herbarium of Mexico (MEXU).
DC., Brya P.Browne, Drepanocarpus G.Mey. and To identify the pollinators and study their behaviour
Hedysarum L.), and there are at least five known on the flowers of A. amorphoides, field observations
synonyms (Rudd, 1975). Aeschynomene amorphoides is were carried out from 12:00 to 18:00 h (for a total of
endemic to the tropical dry forests of the states of 80 h) in small groups of three to five plants. Observa-
Colima, Jalisco, Nayarit and Sinaloa in western tions were made only when weather conditions were
Mexico (Rudd, 1955). It occurs as a shrub up to 3 m tall suitable for insect activity. To record pollination
in open areas and as a tree up to 8 m tall inside these events, photographs were taken using a Nikon D200
forests. It can be recognized by its leaves with 40–80 digital camera with a 105-mm f/2.8 lens and ring flash.
leaflets with a pronounced mucronate apex and inflo- Insect visitors to A. amorphoides flowers were col-
rescences occurring as axillary or terminal panicles. lected and killed with ethyl acetate (Kearns & Inouye,
Investigations on the pollination and reproduction 1993). After identification, the specimens were pre-

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
 POLLINATION IN REFLEXED pAPILIONID FLOWERS 659

served and deposited in the National Insect Collection (N = 20 × 4 × 16). In the hand-pollinated treatments,
of Mexico (CNIN-UNAM). A legitimate pollination by pollen was transferred to stigmas using a small brush
an insect was recorded when a stigma was contacted between 14:30 and 15:30 h, when stigmas were recep-
and pollen was deposited on the same part of the body tive. When fruits were visibly developing, they were
of the insect. Insects that did not fit these criteria were collected and the number of developing ovules was
considered nectar-thieves or herbivores. Moreover, counted. Fruit-set was determined as the number of
visitation frequencies were classified based on the flowers used divided by the number of fruits produced
number of visits made by species per day. Frequent, in the treatment.
occasional and rare visitors were determined according
to the number of visits: > 20 times, five to 20 times, and
DATA ANALYSES

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

fewer than five times per day, respectively. Further-
more, pollen samples from the bodies of captured To test whether fruit-set was affected by daytime and
insects were acetolysed and then analysed under light by pollination treatment, generalized linear mixed-
or scanning electron microscopy to verify whether effects models (GLMMs) with a binomial distribution
they were from A. amorphoides (modified from were used. Concerning the first experiment (stigma
Torres-Colín, Fuentes-Soriano & Delgado-Salinas, receptivity experiment), daytime was included as a
2010). fixed effect, whereas for the second experiment (cross-
ing experiment) treatment was used as the fixed effect.
For both analyses, area and individuals were assigned
BREEDING SYSTEM as random effects. To select the best model, a restricted
To determine the period of stigma receptivity and the maximum-likelihood test (REML) was used. Addition-
time of day when flowers of A. amorphoides were ally, a Bayesian inference approach was used to
more likely to be successfully pollinated, controlled confirm the best-supported model. All statistical
hand-pollinations were performed (stigma receptivity analyses were carried out in R 3.0.2 software (R
experiment). Different groups of flowers were polli- Development Core Team, 2013), with the packages
nated with pollen from at least three different pollen ‘lme4’ used to test GLMMs and ‘arm’ for Bayesian
donors, at increasing times after anthesis. First, ten inference.
groups of ten pre-anthesis flowers in four chosen (by
chance) plants from a population before anthesis were
RESULTS
bagged (N = 10 × 10 × 4). Then, to each group, a polli-
nation time was randomly assigned: immediately after FLORAL BIOLOGY AND POLLINATION
anthesis, 30 min after anthesis and every 30 min until Flowers of A. amorphoides are produced in dense
flower closure. By using random sampling, the likeli- reduced compound racemes or panicle-like inflores-
hood of bias was reduced. Finally, when fruits were cences which deeply contrast in colour with vegetative
visibly starting to develop after 1 week, they were branches (Fig. 1A). This probably makes them visible
collected and the number of developing ovules or seeds to pollinators arriving from above the plant. Inflores-
was counted. cences are axillary or terminal, further enhanced with
To assess the breeding system and the fruit-set by floral units in the axils of leaves and at the base of the
natural pollination in A. amorphoides, controlled pol- main panicles. One or two flowers open on each floral
linations on caged plants were carried out (crossing axis per day. Consequently, a typical branch can carry
experiment). The crossing experiment consisted of four about 18 reduced inflorescence units, and each inflo-
treatments on 16 plants chosen entirely by chance rescence produces about 12 flowers throughout the
located in open areas (N = 9) or inside a forest (N = 7). season. Therefore, a typical plant can produce up to
The four treatments were as follows: (1) control thousands of flowers by the end of a flowering season,
(testing for autogamy), in which no hand-pollinations although only a small portion of these develop into
were done; (2) crossing (between-plant pollinations), in fruits.
which flowers were hand-pollinated with pollen from On average, a flower of A. amorphoides was 4.31
different plants (one to three donors); (3) natural (± 0.39) mm long, and has a short life of c. 5 h, from
pollination, in which flowers were left outside the 13:00 to 18:00 h. It presents the typical papilionoid
cages, thus permitting visits by insects; and (4) selfing flower ground plan (Endress, 1994), with petals modi-
(within-plant pollinations), in which flowers were fied into standard, wings and keel. However, the
hand-pollinated with their own pollen. Due to the standard petal was usually held parallel to the branch
small flower size and difficulty in performing emascu- that supported the inflorescence, and the wings and
lations, no attempts were made to test for agamos- keel were situated above the standard (Fig. 2A). In
permy. In total, 20 flowers for each treatment on each addition, the curved keel remained tightly in contact to
plant were used, resulting in 1280 flowers in total the standard. Consequently, the entire flower served as

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
660 S. CARLEIAL ET AL.

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

Figure 1. Flowers and pollination of Aeschynomene amorphoides. A, branch showing flowers and flower buds on dense
short inflorescences. B, flower in natural position showing the keel with a red maculae. C and D, the pollinator
Tetraloniella jaliscoensis. C, a bee landing on the flower and forcing its way past the keel and standard petals to the
nectar. D, a bee foraging on nectar. Notice the yellowish orange pollen of A. amorphoides on the corbicula of the bee. E,
schematic representing the pollination process of A. amorphoides (from left to right); arrows show flower movement due
to manipulation by a bee.

Figure 2. Flower morphology and position of Aeschynomene amorphoides. A, flower showing the reflexed position, where
the keel (*) is upside down and closely in contact with the standard petal (♦). Wing petals (◊) are parallel with and close
to the keel. B, longitudinal section of a flower with left wing and keel petals and half of the calyx removed to allow
observation of reproductive organs (androecium and gynoecium). Notice the spatial difference between anthers and stigma
almost at the tip of the keel (arrow). Scale bar, 1 mm.

a landing platform for pollinators, and the keel tip to another, in this case from the standard to the keel,
substituted any colour markings (maculae) on the has already been reported in Faboideae, particularly in
standard petal (a feature of many papilionoid flowers). those with beaked keels (i.e. Phaseolus L.; Sousa-Peña,
Furthermore, the keel tip shows a strong purple–red Wong-León & Delgado-Salinas, 1996).
coloration that changes according to the flower phase Concerning the reproductive organs, androecium
(Fig. 1B). The transference of maculae from one petal and gynoecium are hidden within the keel (Fig. 2B).

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
 POLLINATION IN REFLEXED pAPILIONID FLOWERS 661

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

Figure 3. Scanning electron micrographs of Aeschynomene amorphoides flower inner structures. A, stigma with lobes
closed in a flower prior to anthesis. B, stigma with lobes opened after flower anthesis. C, longitudinal section of basal
portion of flower, showing the location of nectar reservoir (asterisk). D, nectar reservoir at the base of the gynoecium
(asterisk). Scale bars, 50 μm (A and B); 1 mm (C); 200 μm (D).

Despite their proximity, stigma and anthers remained Different insects visited flowers of A. amorphoides,
spatially separated by at least 0.67 (± 0.84) mm including bees and butterflies (Table 1). Of these, only
throughout flower life. After anthesis, dehisced pollen female bees of Tetraloniella jaliscoensis (Apidae) regu-
accumulated into a mass inside the keel and in this larly promoted pollination (Fig. 1C, D). Twenty-three
way was presented to pollinators. We calculated a P/O to 35 bees were recorded throughout visiting hours, but
ratio of c. 9500, with 98% of the pollen being viable no assessment was made to estimate how many times
(based on lactophenol blue staining). Eighty per cent of any one individual bee visited an inflorescence.
flowers produced two ovules and the rest produced Tetraloniella bees moved rapidly from one flower to the
three ovules. The style is distally narrowed and the next on each inflorescence. Each visit took c. 2 s, and
stigma, c. 30 μm in diameter, supports up to four pollen one to four bees could be actively foraging on a plant at
grains on its surface. Stigma lobes are present, one time. In contrast to the other flower visitors,
opening during anthesis and closing when flowers T. jaliscoensis showed consistent behaviour and visit-
were old or had been pollinated (Fig. 3A, B), but no ing time. It actively transferred pollen between flowers
sensitivity to pressure as a means against self- using a specific small area between its fore- and
pollination was detected (Newcombe, 1924). Further- mid-legs on the anterior ventral portion of the thorax
more, nectar is produced at the base of the gynoecium (prosternum). Acetolysis analysis also showed that
and is stored in a reservoir formed by the lower portion pollen grains found on specific areas of the bodies of
of the staminal tube (Fig. 3C, D). This reservoir can these bees were mostly from A. amorphoides. Moreo-
retain up to 1 μL of nectar, but if it is not depleted it ver, this bee species was frequently observed visiting
sometimes overflows from the nectar reservoir, and from 12:00 to 16:00 h, the exact time during which the
nectar thieves can forage at the base of the flower. flowers were open.

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
662 S. CARLEIAL ET AL.

Table 1. Flower visitors and pollinators of Aeschynomene amorphoides

Order Family Species Category Frequency

Hymenoptera Apidae Tetraloniella jaliscoensis Pollinator Frequent

Vespidae Brachygastra azteca Nectar thief Frequent
Apidae Apis mellifera Nectar/pollen thief Occasional
Trigona fulviventris Nectar/pollen thief Occasional
Formicidae Cyphomyrmex sp. Nectar thief Rare
Pseudomyrmex sp. Nectar thief Rare

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

Lepidoptera Lycaenidae Cyanophrys miserabilis Nectar thief/herbivore Frequent
Eurema daira Nectar thief/herbivore Occasional
Leptotes cassius striata Nectar thief/herbivore Occasional
Coleoptera Meloidae Unidentified Herbivore Rare

Visitors were classified as pollinators, nectar thieves and/or herbivores according to their behaviour on the plant.
Frequency was established according to the number of times species were observed visiting flowers. In Lepidoptera, adults
are nectar thieves, whereas their larvae forage on leaves.

The process of pollination in A. amorphoides is confirmed. All nectar thieves on A. amorphoides con-
depicted in Figure 1E. It begins with the keel-aligned sistently reach the nectar source via the staminal slit
landing of the pollinator. The weight of the bee forces and thus do not come into contact with pollen or stigma
the flower to bend downwards. In this phase, the and do not effect pollination.
flower temporarily presents the usual papilionoid ori-
entation with its standard in a vertical position. Now,
BREEDING SYSTEM
the bee reaches the nectary with its proboscis. Sub-
sequently, pollen collection takes place as the bee Hand pollination in the stigma receptivity experiment
leaves and the flower goes back to its original posi- produced fruits in all daytime treatments, but no
tion. This floral mechanism maximizes the chance of significant differences were found among them (χ2 =
a pollinator making contact with both stigma and 3.55, d.f. = 9, P = 0.679). This suggests that flowers of
pollen, thereby promoting pollination. A. amorphoides were always receptive to pollen
Nectar sugar composition of A. amorphoides con- despite their short life.
sists principally of glucose, mannose, arabinose and In the crossing experiment, we found significant
rhamnose, and thus differs from common sugar- differences among treatments (χ2 = 115.95, d.f. = 3,
species nectar (glucose, fructose and sucrose; Baker & P < 0.001) (Fig. 4). Flowers of the selfing treatment set
Baker, 1983). The presence of mannose in nectar has fewer fruits than those of the crossing treatment
been shown to be toxic to Apis mellifera and other (Z = −3.69, P < 0.001), suggesting that cross pollen
insects (Adler, 2000). However, no sign of intoxication improves fruit-set in A. amorphoides, although the
or disorientation was perceived on either pollinator species is still able to produce fruits by selfing. Both
bees or nectar thieves. In fact, wasps (Brachygastra crossing and selfing treatments, however, did not
azteca) and other nectar thieves preferred to forage differ in fruit-set from natural pollination (Z = −2.47,
specifically on this resource. P = 0.057; Z = −1.26, P = 0.568, respectively). Overall,
Visitation by other flower visitors differs from that of fruit-set was < 30% (crossing, 27.41%; natural pollina-
T. jaliscoensis. The short-tongued bee species A. mel- tion, 19.21%; selfing, 15.36%). The control treatment
lifera and Trigona fulviventris collected the pollen left (no pollinations performed) produced an insignificant
on petals by T. jaliscoensis. Likewise, Brachygastra fruit-set of 1.46%. Concerning seed-set, we found that
azteca was able to reach the nectar reservoir at the crossing produced on average more seeds per fruit than
base of the flowers even before anthesis. Furthermore, selfing (Z = −2.74, P = 0.017).
ants of the genera Cyphomyrmex and Pseudomyrmex
were also attracted to the overflowing nectar, and the
DISCUSSION
butterfly species Cyanophrys miserabilis, Eurema
daira and Leptotes cassius striata were occasionally FLORAL BIOLOGY
seen sucking nectar from the flowers. Larvae of these The pollination system by which flowers of A. amor-
butterfly species appear to use the foliage of A. amor- phoides operate is similar to that described as ‘pump
phoides as a food source, although this needs to be type’ by Delpino (in Endress, 1994) or as ‘piston type’

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
 POLLINATION IN REFLEXED pAPILIONID FLOWERS 663

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

Figure 4. Variation in Aeschynomene amorphoides pollination success in different pollination treatments. Pollination
success is represented as the percentage of flowers that developed into immature fruits by the end of the experiment.
Relative importance of fruits with one, two or three seeds is highlighted in levels of grey inside each bar. Letters on top
of standard error bars represent the statistical significance among treatments. Treatments with letters in common showed
no statistical differences in pollination success.

with secondary pollen presentation dependent on bee (Michener, 2000). Tetraloniella jaliscoensis was
activity on the flower (Leppik, 1966; Westerkamp, recently described by LaBerge & Warwick (2001) from
1997). Nevertheless, A. amorphoides differs from most bees collected in the Chamela Biological Station and
other papilionoid flowers with respect to its flower hence there is not much information concerning its
orientation. Instead of presenting a horizontal posi- biology or foraging preferences. Bees of this species
tion, as in most Faboideae (e.g. Aeschynomene, Arachis have been collected from June to November in Texas
L., Crotalaria L., Glycine Willd., Phaseolus) or resupi- (USA), Chihuahua, Jalisco and Michoacán (Mexico).
nate presentation as present in, for example, Cen- Thus, they are known to be active during the flowering
trosema (DC.) Benth. and Clitoria L. (Faegri & van der season of A. amorphoides, from the end of August to
Pijl, 1971; Proctor, Yeo & Lack, 2003), the standard November.
petal in A. amorphoides has an obliquely horizontal The behaviour of short-tongued Tetraloniella bees on
position, with the wing and keel in close juxtaposition flowers of A. amorphoides enables them to take nectar
above. This presents the entire flower as a small from the nectar reservoir, by crawling down the keel–
landing platform for pollinators, at the same time standard passageway and, in so doing, they have
acting as an effective barrier to insect thieves, which pollen deposited on them as they crawl back out of the
cannot manage to get the nectar and pollen directly. petals. In addition, pollination by Tetraloniella bees is
The only species of Faboideae discussed in the assured, as pollen presented secondarily on the keel tip
literature with similar floral characteristics to occupies exactly the same position as the stigma
A. amorphoides is Machaerium isadelphum (E.Mey.) outside the keel orifice. Observations indicate that this
Standl. Coincidently, recent analyses of DNA sequence interaction is very precise, and only Tetraloniella bees
data established species of Aeschynomene subgenus effectively pollinate flowers of A. amorphoides. Indeed,
Ochopodium (including A. amorphoides) as sister to no other observed flower visitor demonstrated such a
species of Machaerium Pers., placing species of selective behaviour. Additionally, pollinator morphol-
Aeschynomene subgenus Aeschynomene in a distinct ogy precisely fits flower size, allowing pollen deposition
phylogenetic lineage (Lavin et al., 2001; Ribeiro et al., on the bee on the ventral anterior surface of the thorax,
2007). where it is transferred from one flower to another. A
sternotribic pollination system (where pollen is depos-
ited on the anterior ventral portion of the thorax)
POLLINATION in Faboideae is apparently rare; to the best of our
Bees of the genus Tetraloniella (tribe Eucerini, knowledge this is the first time such a precise pollina-
Apidae: Apinae) are generally oligolectic and solitary tion system has been described in a species of this

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
664 S. CARLEIAL ET AL.

subfamily. However, sternotribic anterior–thorax mechanism in A. amorphoides that inhibits some self-
pollination also has been reported for some other pollen tube growth by a late-acting self-incompatibility
plants, such as Amomum subulatum Roxb. (Deka, system or early-acting inbreeding depression (Hao
Sudharshan & Saju, 2011) and species of Pedicularis L. et al., 2012). The first mechanism has been reported in
(Wang et al., 2009). other members of Fabaceae (Gibbs & Sassaki, 1998)
For each ovule there are almost 10 000 pollen and probably explains the fact that crossing yielded
grains produced in one single flower, but the gynoe- more fruits than selfing.
cium contains only two or three ovules and the stig- Fruit-set in A. amorphoides following the crossing
matic surface is large enough to accommodate four treatment was comparable to natural pollination and
pollen grains. Stigma lobes in A. amorphoides pre- higher than for selfing, but is still considered to be

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

sumably could function as a selective barrier to pollen relatively low. An increase in fruit-set has also been
when closing the stigma. As only a few pollen grains reported for Aeschynomene pratensis after hand polli-
are necessary for successful fertilization in any given nations (Ramírez & Brito, 1990). Similarly, McKellar
flower, any extra pollen deposited on a stigma is et al. (1991) reported that A. americana presented low
wasted or collected by bees. A similar pattern is found rates (20–30%) of cross pollination success in natural
in perennial peanut species (Arachis), in which environments and under cultivation. Results from the
stigmas support up to three pollen grains, contrasting crossing experiment show that A. amorphoides is self-
with annual species with wider stigmas that have no compatible, although with a low pollination success
lobes (Lu, Mayer & Pickersgill, 1990). This pollen ratio of c. 17%. This result is similar to the one reported
surplus is, however, also important for the plant, by Arroyo (1981) for Arachis and Stylosanthes Sw.,
possibly serving to attract pollinators. both members of tribe Dalbergieae.
Field observations suggest that T. jaliscoensis has a That some flowers produced fruits even in the
close relationship with A. amorphoides. Pollen analy- absence of pollinators in the control treatment indi-
sis also corroborates that there is specificity in this cates that A. amorphoides may produce a few flowers
interaction because, at least in the Chamela region, where anthers and stigma are in direct contact and
these bees rely on flowers of A. amorphoides as a spontaneous self-pollination takes place, although
major food source. To what extent this fidelity is such flowers were not detected in our study. Never-
maintained throughout the geographical distribution theless, these few fruits could also be the product of
of A. amorphoides is, however, not known. Further- agamospermy, and because we did not test this, this
more, it is noteworthy that this apparently strict reproductive process cannot be ruled out.
pollinator selection occurs in a region where bee Fruits with different seed numbers could be due to
diversity is extremely high. In the Chamela region, ovule reduction (80% of ovaries presented two ovules
228 species in 87 genera and seven families of bees instead of three), pollen limitation (few pollen grains
have been reported (Ayala, 2004). carried by bees in the pollen deposition area) or seed
abortion by other causes (Knight et al., 2005). A larger
number of one-seeded fruits from selfing and natural
BREEDING SYSTEM pollination could be due to pollen limitation or inher-
The main finding of this study is that A. amorphoides itable seed abortion. Lack of resources can be dis-
requires pollination by Tetraloniella bees to set fruit. counted because the studied plants and different
The adaptive explanation for the morphological treatment branches showed an increase in fruit-set
changes shown by flowers of A. amorphoides is that and no apparent sign of systematic fruit abortion.
they have allowed a more precise pollination process, Field experiments also showed a low seed/ovule ratio
directing the specific placement of pollen on the body in all treatments. Bawa & Buckley (1989) reported
of the bees, while reducing the probability of a stigma that in Fabaceae seed-set is uniformly greater in
receiving foreign pollen and, consequently, increasing self-compatible species than in self-incompatible ones.
reproductive isolation of this species. By using an In relation to Faboideae, they found self-compatible
experimental approach, we demonstrate that T. jalis- species, such as Desmodium incanum DC., with
coensis bees are the single pollen vector of A. amor- seed-set close to 80%. This contrasted with self-
phoides and we quantify the differences between self- incompatible species, namely Gliricidia sepium (Jacq.)
and cross-pollinated flowers. Kunth ex Walp., Dalbergia retusa Hemsl. and Ptero-
Fruit-set of A. amorphoides in the crossing carpus rohrii Vahl, with seed-set of < 50%. Bullock
experiments demonstrated a high degree of self- (1995), however, considered only SDTF species and
compatibility. Observations also show that pollinators estimated seed/ovule ratios for self-compatible species
tend to move from flower to flower on the same plant, even lower than 30%. According to him, this low
thus promoting geitonogamy. Nevertheless, results of fecundity can be explained by pollinator limitation.
the selfing treatment indicate that there is possibly a Aeschynomene amorphoides seems to follow this same

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
 POLLINATION IN REFLEXED pAPILIONID FLOWERS 665

pattern, as results from the crossing experiment gave assistance in scanning electron microscopy and to R.
low seed-sets: natural pollination, 20%; selfing, 17%; Ramírez-Delgadillo, M. Abarca, and E. Ramírez García
crossing, 28%. for providing field images. We thank J. César Montero
Bullock (1985) established a self-incompatibility and D. Martínez Almaguer for their assistance with
index (ISI: self-pollination/cross-pollination success) figure designs. Finally, we are grateful for the support
for plants in the Chamela region. He estimated that given by K. Renton and the personnel of the Chamela
ISI values between 0.40 and 0.99 are characteristic of Biological Station throughout this project.
self-compatible species, such as in Caesalpinia calad-
enia Standl., C. pulcherrima (L.) Sw., Lonchocarpus
REFERENCES
guatemalensis Benth. and L. lanceolatus Benth.

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

(Fabaceae), which often colonize disturbed areas and Adler LS. 2000. The ecological significance of toxic nectar.
have high reproductive capacities. Similarly, A. amor- Oikos 91: 409–420.
phoides presented ISI values of 0.58. Nevertheless, Arroyo MTK. 1981. Breeding systems and pollination biology
self-compatibility is not the single cause by which in Leguminosae. In: Polhill RM, Raven PH, eds. Advances in
A. amorphoides colonizes and succeeds in disturbed or legume systematics: part 1. Kew: Royal Botanic Gardens,
open areas at the edge of the tropical dry forest. Bee 723–769.
pollination is still required for fruit-set, and without Ayala R. 2004. Fauna de abejas silvestres (Hymenoptera:
T. jaliscoensis bees plants from A. amorphoides could Apoidea). In: García-Aldrete AN, Ayala R, eds. Artrópodos
not reproduce efficiently. The maintenance of T. jalis- de Chamela. Mexico City: Universidad Nacional Autónoma
coensis populations is essential to retain successful de México, 193–219.
Baker HG, Baker I. 1983. Floral nectar sugar constituents
reproduction of A. amorphoides, at least in the
in relation to pollinator type. In: Jones CE, Little RJ, eds.
Chamela region.
Handbook of experimental pollination ecology. New York:
In this study, our aim was to understand how floral
Van Nostrand-Reinhold, 117: 141.
mechanisms in A. amorphoides with reflexed flowers
Bawa KS, Buckley DP. 1989. Seed : ovule ratios, selective
and associated bee behaviour function, and how the seed abortion, and mating systems in Leguminosae. In:
close relationship between the two organisms facili- Stirton CH, Zarucchi JL, eds. Advances in legume biology.
tates successful plant reproduction in the local envi- Monographs of the Missouri Botanical Garden 29: 243–262.
ronment. However, this apparent specialization may Bullock SH. 1985. Breeding systems in the flora of a tropical
not be a good indication of dependency because deciduous forest in Mexico. Biotropica 17: 287–301.
even among specialists a spatio-temporal variation Bullock SH. 1995. Plant reproduction in Neotropical dry
might exist. Interaction between A. amorphoides and forests. In: Bullock SH, Mooney HA, Medina E, eds. Sea-
T. jaliscoensis may not be the same across the entire sonally dry tropical forests. Cambridge: Cambridge Univer-
geographical range where the two species coexist. sity Press, 277–303.
McVaugh (1987: 260–261), in his floristic treatment Cruden RW. 1977. Pollen–ovule ratios: a conservative indi-
of western Mexico, commented that with: ‘. . . abun- cator of breeding systems in flowering plants. Evolution 31:
dant material [of A. amorphoides] now at hand, the 32–46.
fruits of this species usually consist of a single article Deka TN, Sudharshan MR, Saju KA. 2011. New record of
[one-seeded fruit]; on some plants 1–articulate and bumble bee, Bombus breviceps Smith as a pollinator of large
2–articulate fruits are present in about equal cardamom. Current Science 100: 926–928.
numbers; 3–articulate fruits are rare in most speci- Dórea MC, Aguiar CML, Figueroa LER, Lima LCLE,
mens, or wanting.’ This statement confirms that the Santos FAR. 2010. Pollen residues in nests of Centris
reproductive success of A. amorphoides in large parts tarsata Smith (Hymenoptera, Apidae, Centridini) in a tropi-
cal semiarid area in NE Brazil. Apidologie 41: 557–567.
of its distributional range is similar to the results
Endress PK. 1994. Diversity and evolutionary biology of
obtained in our crossing experiments.
tropical flowers. Cambridge: Cambridge University Press.
Faegri K, van der Pijl L. 1971. The principles of pollination
ACKNOWLEDGEMENTS ecology. Oxford: Pergamon Press.
García-Oliva F, Camou A, Maass JM. 2002. El clima de la
The Consejo Nacional de Ciencia y Tecnología región central de la costa del Pacífico Mexicano. In: Noguera
(CONACyT) is thanked for financial support to S.C. We FA, Vega-Rivera JH, García-Aldrete AN, Quesada-Avendaño
thank M. F. Fay, G. Lewis and M. Lavin for their M, eds. Historia natural de Chamela. Mexico City: Univer-
helpful comments on the manuscript, and also J. E. sidad Nacional Autónoma de México, Instituto de Biología,
Fornoni Agnelli and J. Núñez Farfán for their inspir- 3–10.
ing ideas. Insect identification was carried out by the Gibbs P, Sassaki R. 1998. Reproductive biology of Dalbergia
bee specialist R. Ayala and by A. Ibarra Vázquez of the miscolobium Benth. (Leguminosae-Papilionoideae) in SE
Colección Nacional de Insectos (IBUNAM). We are Brazil: the effects of pistillate sorting on fruit-set. Annals of
deeply grateful to B. Mendoza Garfias for technical Botany 81: 735–740.

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666
 666 S. CARLEIAL ET AL.

Griffith AB, Forseth IN. 2002. Primary and secondary seed von der Ohe K, Dustmann JH. 1996. Identifizierung einer
dispersal of a rare, tidal wetland annual, Aeschynomene bisher nicht zugeordneten Pollenform in mittelamerikanis-
virginica. Wetlands 22: 696–704. chen Honigen (sog, Cuba-Form): Aeschynomene americana L
Griffith AB, Forseth IN. 2003. Establishment and reproduc- (Leguminosae, Papilionoideae). Apidologie 27: 157–163.
tion of Aeschynomene virginica (L.) Britton, Sterns & Peel MC, Finlayson BL, McMahon TA. 2007. Updated
Poggenb. (Fabaceae) a rare, annual, wetland species in world map of the Köppen-Geiger climate classification.
relation to vegetation removal and water level. Plant Ecology Hydrology and Earth System Sciences 4: 439–473.
167: 117–125. Proctor M, Yeo P, Lack A. 2003. The natural history of
Hao YQ, Zhao XF, She DY, Xu B, Zhang DY, Liao WJ. pollination. Portland: Timber Press.
2012. The role of late-acting self-incompatibility and early- R Development Core Team. 2013. R: a language and envi-
acting inbreeding depression in governing female fertility in ronment for statistical computing. Vienna: R Foundation

Downloaded from https://academic.oup.com/botlinnean/article/177/4/657/2416513 by guest on 24 September 2020

monkshood, Aconitum kusnezoffii. PLoS ONE 7: e47034. for Statistical Computing. Available at: http://www.R-
Kearns CA, Inouye DW. 1993. Techniques for pollination project.org
biologists. Boulder: University Press of Colorado. Ramírez N, Brito Y. 1990. Reproductive biology of a tropical
Knight TM, Steets JA, Vamosi JC, Mazer SJ, Burd M, palm swamp community in the Venezuelan llanos. American
Campbell DR, Dudash MR, Johnston MO, Mitchell RJ, Journal of Botany 77: 1260–1271.
Ashman TL. 2005. Pollen limitation of plant reproduction: Ribeiro RA, Lavin M, Lemos-Filho JP, Filho CV, Santos
pattern and process. Annual Review of Ecology, Evolution, FR, Lovato MB. 2007. The genus Machaerium (Legumi-
and Systematics 36: 467–497. nosae) is more closely related to Aeschynomene sect. Ocho-
LaBerge WE, Warwick C. 2001. Revision of the bees of the podium than to Dalbergia: inferences from combined
genus Tetraloniella in the New World (Hymenoptera: sequence data. Systematic Botany 32: 762–771.
Apidae). Illinois Natural History Survey Bulletin 36: 67–162. Rudd VE. 1955. The American species of Aeschynomene.
Lavin M, Pennington RT, Klitgaard BB, Sprent JI, de Contributions from the United States National Herbarium
Lima HC, Gasson PE. 2001. The dalbergioid legumes 32: 1–172.
(Fabaceae): delimitation of a pantropical monophyletic Rudd VE. 1975. Studies in Aeschynomene III: series Sco-
clade. American Journal of Botany 88: 503–533. pariae in Mexico and Central America. Phytologia 31: 431–
Leppik EE. 1966. Floral evolution and pollination in Legu- 434.
minosae. Annales Botanici Fennici 3: 299–308. Sousa-Peña M, Wong-León A, Delgado-Salinas A. 1996.
Lott EJ. 2002. Lista anotada de las plantas vasculares de Pollination dynamics and evolution in the Phaseolus coc-
Chamela-Cuixmala. In: Noguera FA, Vega-Rivera JH, cineus L. complex. In: Pickersgill B, Lock JM, eds. Advances
García-Aldrete AN, Quesada-Avendaño M, eds. Historia in legume systematics. Kew: Royal Botanic Gardens, 75–
natural de Chamela. Mexico City: Universidad Nacional 81.
Autónoma de México, Instituto de Biología, 99–136. Torres-Colín L, Fuentes-Soriano S, Delgado-Salinas A.
Lu J, Mayer A, Pickersgill B. 1990. Stigma morphology and 2010. Estudio palinológico del género Macroptilium (Benth.)
pollination in Arachis L. (Leguminosae). Annals of Botany Urb. (Leguminosae) en México y Centroamérica. Acta
66: 73–82. Botanica Mexicana 91: 51–69.
McKellar MA, Deren CW, Quesenberry KH. 1991. Out- Villaseñor JL, Ortiz E. 2014. Biodiversidad de las plantas
crossing in Aeschynomene. Crop Science 31: 476–478. con flores (División Magnoliophyta) en México. Revista
McVaugh R. 1987. Flora Novo-Galiciana: a descriptive Mexicana de Biodiversidad 85: S134–S142.
account of the vascular plants of western Mexico. Ann Arbor: Wang H, Yu WB, Chen JQ, Blackmore S. 2009. Pollen
The University of Michigan Press, 580–598. morphology in relation to floral types and pollination syn-
Michener CD. 2000. The bees of the world. Baltimore: Johns dromes in Pedicularis (Orobanchaceae). Plant Systematics
Hopkins University Press. and Evolution 277: 153–162.
Newcombe FC. 1924. Significance of the behavior of sensi- Westerkamp C. 1997. Keel blossoms: bee flowers with adap-
tive stigmas II. American Journal of Botany 11: 85–93. tations against bees. Flora 192: 125–132.

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 177, 657–666

View publication stats