Content uploaded by Hannes F. Paulus
Author content
All content in this area was uploaded by Hannes F. Paulus on May 12, 2020
Content may be subject to copyright.
Looks matter: changes in flower form affect pollination
effectiveness in a sexually deceptive orchid
D. RAKOSY*†,M.CUERVO‡,H.F.PAULUS*&M.AYASSE‡
*Integrative Zoology, Faculty of Life Sciences, University Vienna, Vienna, Austria
†Systematic and Evolutionary Botany, Faculty of Life Sciences, University Vienna, Vienna, Austr ia
‡Institute of Evolutionary Ecology and Conservation Genomics, University Ulm, Ulm, Germany
Keywords:
flower form;
orchids;
pollination effectiveness;
pollinator-mediated selection;
sexual deception.
Abstract
Many species of the sexually deceptive genus Ophrys are characterized by
insect-like flowers. Their form has been traditionally considered to play an
important role in pollinator attraction and manipulation. Yet, the evolution
of the floral form remains insufficiently understood. We hypothesize that
pollinator-mediated selection is essential for driving floral form evolution in
Ophrys, but that form components are being subjected to varying selection
pressures depending on their role in mediating interactions with pollinators.
By using the Eucera-pollinated Ophrys leochroma as a model, our aim has
been to assess whether and in what manner pollination effectiveness is
altered by experimental manipulation of the flower form. Our results show
that floral form plays an essential and, so far, underestimated role in ensur-
ing effective pollination by mechanically guiding pollinators towards the
reproductive structures of the flower. Pollinators are significantly less effec-
tive in interacting with flowers having forms altered to resemble those of
species pollinated by different hymenopteran genera. Further, those compo-
nents used by pollinators as gripping points were found to be more effective
in ensuring pollinia transfer than those with which pollinators do not
directly interact. Thus, mechanically active and inactive components appear
to be under different selection pressures. As a consequence, mechanically
active components of the flower form could reflect adaptations to the inter-
action with particular pollinator groups, whereas mechanically inactive com-
ponents can vary more freely. Disentangling selection patterns between the
functionally different components of flower form may provide valuable
insights into the mechanisms driving the morphological diversification of
sexually deceptive pollination systems.
Introduction
Angiosperm flowers are characterized by a remarkable
diversity of forms, some of the most bizarre being found
in sexually deceptive orchids (van Der Cingel, 2001a, b).
Sexual deception relies primarily on the chemical
mimicry of receptive female insects (e.g. Schiestl, 2005;
Ayasse et al., 2011; Bohman et al., 2016; Johnson &
Schiestl, 2016). However, many species are also charac-
terized by a morphologically complex and often insect-
like third petal, serving as a landing platform for polli-
nators (hereafter referred to as the lip). Although tradi-
tionally considered important, there is little empirical
evidence for the contribution of the insectiform lip to
pollinator attraction and pollination (Gaskett, 2011;
Johnson & Schiestl, 2016).
In principle, the form of the lip might act as a visual
signal to mate-searching males and also ensure the cor-
rect alignment between the body of the pollinator and
the reproductive parts of the plant (Kullenberg, 1961,
1973; Phillips et al., 2014; De Jager & Peakall, 2015;
G€
ogler et al., 2015; Martel et al., 2016). Selection on the
lip form could be thus imposed through both the sen-
sory preferences of the pollinator and the mechanical
Correspondence: D. Rakosy, Integrative Zoology, Faculty of Life Sciences,
University Vienna, Althanstrasse 14, 1090 Vienna, Austria.
e-mail: demetra.rakosy@gmail.com
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL.
1
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
doi: 10.1111/jeb.13153
fit between the pollinator and the lip (De Jager & Pea-
kall, 2015). So far, empirical evidence provides stronger
support for flower form variation being shaped by
selection for an improved mechanical fit between the
lip and the pollinator’s body (Gaskett, 2012; G€
ogler
et al., 2009, 2015; Phillips et al., 2013; De Jager & Pea-
kall, 2015; but see also Ellis & Johnson, 2010). By
selectively facilitating the mechanical fit to a particular
pollinator species, floral form was shown to reinforce
other traits in ensuring reproductive isolation (G€
ogler
et al., 2009, 2015; De Jager & Peakall, 2015). However,
most sexually deceptive orchids rely solely on floral
scent for reproductive isolation and the adaptive signifi-
cance of floral form in these species remains little
understood (e.g. Johnson & Schiestl, 2016).
Considering the available evidence, pollinator-
mediated selection would be expected to mainly act by
stabilizing the floral form, leading to an optimized
mechanical fit between pollinators and flowers (Cress-
well, 1998; Gaskett, 2012). However, most orchids are
characterized by an imperfect match between their lip
and the body of the pollinator (or that of its female)
(Benitez-Vieyra et al., 2009; Gaskett, 2012; De Jager &
Peakall, 2015). This might suggest that selection on
flower form is either weak or absent. Alternatively,
varying selection pressures acting on different compo-
nents of the form of the lip could lead to the observed
incomplete correspondence between flowers and their
pollinators (van der Niet et al., 2010). The lip consti-
tutes a very complex structure, comprising several
shape and size components, only a fraction of which
may be involved in effectively guiding pollinators
towards the reproductive structures of the flowers (Pau-
lus, 2007; Gaskett, 2012). The mechanical fit between
pollinators and flowers may thus be mediated through
only a subset of the components making up the overall
form of the lip (e.g. the calli on the lip of the Australian
Chiloglottis or the lateral lobes on the lip of European
Ophrys species).
We hypothesize that components with which pollina-
tors functionally interact will be under strong pollina-
tor-mediated selection, whereas those with which
pollinators do not effectively interact are more likely to
be shaped by relaxed selection and other stochastic fac-
tors. Such a dichotomy is known to occur between ‘be-
haviourally active’ and ‘behaviourally inactive’
components of floral scent and has been essential for
understanding the evolution and diversification of sex-
ually deceptive orchids (e.g. Schiestl et al., 2000; Ayasse
et al., 2011). We thus propose to also differentiate
between mechanically active and inactive components
of floral form. We define mechanically active compo-
nents as those areas or points of the lip which pollina-
tors use as fixed gripping or contact points during
pollination, whereas mechanically inactive components
relate to areas or points on which the pollinators move
around freely or with which they do not come into
contact during pollination. Selection would thus be
expected to only optimize the mechanical fit between
the mechanically active components of the form of the
lip and the pollinator’s body. As far as we know, no
study has investigated whether mechanically active and
inactive components of the floral form can be subjected
to differential selection pressures. Moreover, we do not
know the way in which these potentially variable selec-
tion pressures shape the evolution of the overall flower
form in sexually deceptive orchids.
The Mediterranean orchid genus Ophrys is among the
most well-known examples for pollination through sex-
ual deception (Gaskett, 2011). Its pollinators, males of
various hymenopteran species, are lured to the flowers
primarily through the mimicry of the sexual phero-
mone of their females (Schiestl et al., 1999, 2000;
Ayasse et al., 2003, 2011; Schiestl, 2005). Na€
ıve males
are usually unable to discriminate between a conspeci-
fic female and the orchid. This leads them to attempt to
copulate with the flowers, a process during which the
effective transfer of pollinia is ensured. The mimicry of
mate-recognition signals ensures that the relationship
between Ophrys species and their pollinators is generally
species specific (Paulus & Gack, 1990; Xu et al., 2011).
According to the most recent phylogeny of the genus,
Ophrys has diverged by two fundamental pollinator
shifts from wasps to solitary bees, followed by subse-
quent radiations to various species, mainly within the
solitary bee genera Eucera and Andrena (Breitkopf et al.,
2014). In this context, Ophrys flowers can be classified
into three form types, each attracting different pollina-
tor genera: wasp-type flowers (pollinators: Agrogorythes
and Dasyscolia species), tenthredinifera-type flowers (pol-
linators: Eucera and Andrena species) and fusca-type
flowers (pollinators: Andrena species) (Delforge, 2005;
Paulus, 2007) (Fig. 1). It thus appears that the evolu-
tion of novel floral form types in Ophrys is associated
with shifts between pollinator genera, whereas shifts
within genera have resulted in comparatively minor
form changes. Whether pollinators actually select for
particular lip forms in Ophrys has, however, never been
tested.
Because pseudocopulation is essential in order for
pollinia transfer to take place, a close mechanical fit
between Ophrys flowers and their male pollinators
would be expected. However, the genus is also notori-
ous for high intraspecific variation in the form of the
lip (e.g. Delforge, 2005). This pattern could be the con-
sequence of differential selection shaping a stronger fit
towards mechanically active components of the lip
form, leaving mechanically inactive components to vary
more freely. However, whether in Ophrys adaptive dif-
ferences actually occur between mechanically active
and inactive components of flower form remains
unclear. Knowledge of the differential selection pres-
sures acting on components of flower form may provide
valuable insights into the evolution and diversification
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
2D. RAKOSY ET AL.
of sexually deceptive flowers. In fact, distinguishing
between mechanically active and inactive components
of flower form may help to solve such conceptual con-
flicts as those between the observed high floral varia-
tion and the expected stabilizing pollinator-mediated
selection in specialized pollination systems (e.g. Cress-
well, 1998).
Here, we have used an experimental approach to
elucidate whether and in what manner pollinator-
mediated selection drives the evolution and diversifica-
tion of floral forms in Ophrys. By performing overall
manipulations of flower form and directed manipula-
tions of its mechanically active and inactive compo-
nents, our aim has been to assess 1) the way changes
in the overall floral form, corresponding to the three
floral types, affect the behaviour of pollinators as well
as the mechanical fit between the flowers and their
pollinators and 2) whether changes in the mechanically
active components of the flower form have a stronger
impact on the behaviour of the pollinator and the
mechanical fit than inactive components. Our study
confirms that the flower form in Ophrys is likely to be
under pollinator-mediated selection and that selection
pressures vary between mechanically active and inac-
tive components of flower form.
Materials and methods
Study species and sites
The orchid and its pollinator: Ophrys leochroma and
Eucera kullenbergi
Ophrys leochroma Delforge, 2005 is a well-defined spe-
cies, belonging to the O. tenthredinifera complex (Paulus
& Hirth, 2012). Its distribution range spans the islands
of the east Mediterranean basin (including Crete),
where it can be found flowering from the middle of
March to the beginning of April. It is pollinated
throughout its range exclusively by Eucera kullenbergi
males (Paulus & Hirth, 2012; Hirth & Paulus, 2016).
The long-horn bee E. kullenbergi is widespread in the
east Mediterranean basin, where it occurs in a wide
range of open habitats. Males emerge during the middle
of March, some days prior to their females. During the
following 2–3 weeks, they usually aggregate at and
between nesting sites where they patrol in search of
unmated females (Ayasse et al., 2001).
Morphologically, O. leochroma shows all the features
common for tenthredinifera-type flowers: square to
trapezoid form, distinct lateral lobes and a reflexed api-
cal appendix (Fig. 1). In O. tenthredinifera, a relative of
O. leochroma, histochemical and cytological analyses
have been used to identify the distribution of scent pro-
duction on the lip (Francisco & Ascensao, 2013). It was
thus shown that, as in most Ophrys species, scent is pro-
duced in the epidermal cells on the surface of the lip
and in the tissue within the apical appendix and along
a narrow portion of the margins of the lip (Vogel,
1962; Francisco & Ascensao, 2013). The sexual phero-
mone analogue of O. leochroma was thereby recently
shown to comprise a blend of aldehydes, alcohols, fatty
acids and hydrocarbons (Weber, 2012; Cuervo et al.,
2017).
Experimental sites
All experiments were performed at two neighbouring
sites near Neapoli, Crete (Fig. S1). At the first site,
about 40–50 E. kullenbergi males patrol an area of three
or four square metres, whereas at the second site hun-
dreds of males patrol an area of about 40 square
metres. The orchid does not occur in these areas, so
males can be presumed to be naı
¨ve to its sexual phero-
mone analogue. The number of males and the size of
the swarming areas are ideal for performing pollination
experiments.
Flower form manipulation
Identification of mechanically active and inactive
components of floral form
In order to be able to generalize our results beyond our
model species, we have focused on the four major struc-
tures of the lip, common to most Ophrys species: the stig-
matic cavity, the lateral lobes, the distal-side lobes and
the appendix (Fig. 1). Mechanically active and inactive
components of the lip form were identified with the aid
of pollination experiments. We thereby photographically
recorded the behaviour of E. kullenbergi males during
pseudocopulation with individually presented flowers
picked from four O. leochroma populations (Fig. S1).
From the photographic series, only those frames were
analysed in which males could be seen removing polli-
nia. We thereby assessed the frequency with which
males used the four flower structures as fixed contact or
gripping points during 36 effective pseudocopulations (1
pseudocopulation/flower) (Video S3). In all cases, males
contacted the stigmatic cavity during pseudocopulations
(100%), whereas in 97.2% of the cases, they also relied
on the lateral lobes as gripping points. In contrast, males
moved around more freely on the distal-side lobes and
the appendix, contacting these structures only during
30.5% and 25% of the pseudocopulations. As a conse-
quence, we classified the lateral lobes and the stigmatic
cavity as mechanically active components and the distal-
side lobes and the appendix as mechanically inactive
(Fig. 1 a-d).
Manipulation of mechanically active and inactive
components of the labellum
We collected O. leochroma inflorescences with an equal
number of unpollinated flowers from two populations
in Crete and kept them in water during the duration of
the experiments (Fig. S1). Manipulations were per-
formed by removing certain areas of the lip. According
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 3
to the changes made, the manipulations were classified
into three groups: 1) nonmanipulated flowers (=con-
trol) (C); 2) flowers with overall changes in floral form
resembling a) the form of wasp-pollinated Ophrys spe-
cies (rectangular flowers as found in O. insectifera,
O. regis–ferdinandii, etc.) (WT) and b) the flower form
characteristic of the Andrena-pollinated Ophrys fusca
clade (FT); and 3) flowers with manipulations made to
a) areas of the lip that the males use as attachment
points to the flowers (mechanically active components
of flower form), namely the lateral lobes (WLL) and
the stigmatic lobe (WSL); b) areas with which males do
not closely interact (mechanically inactive compo-
nents), namely the appendix (WA) and the distal sides
(WDS) (Figs 1 and 2).
Experimental set-up
Behavioural response of pollinators
Experiments were conducted at the large E. kullenbergi
site during the last 2 weeks of March 2016 and 2017,
between 08:30 and 14:00, provided that the cloud
coverage did not exceed 50%. This corresponds to
the time frame and weather conditions in which
males are active. In 2016, manipulations to the
flower form were performed by removing parts of the
lip, the removed tissue being discarded. All inflores-
cences were replaced after 1 day. In 2017, we
repeated this protocol in order to assess whether the
results obtained are reproducible across years. In addi-
tion, we also performed experiments in which the
removed flower tissue was re-attached below the
flowers (out of reach of approaching males) with an
insect pin in order to supplement the total amount of
scent emitted by manipulated flowers (De Jager &
Peakall, 2015). In this case, each inflorescence was
replaced after 30 min.
In order to ensure that the picked inflorescences
remained attractive towards their pollinators, we per-
formed baiting experiments at the beginning of each
day. Inflorescences were only kept if they elicited pseu-
docopulations. Consequently, inflorescences that
proved to be attractive to the male pollinators were
assigned arbitrarily to the seven manipulation groups
and then tested in a random order. The testing location
was switched between every trial, the distance between
locations measuring 3-4 m. For each inflorescence, the
behaviour of five males was filmed with a Nikon Cool-
pix P330 camera. The behaviour of the males was
assessed from the video recordings and classified as
short landings (<1 s), long landings (>1 s) and pseudo-
copulations (see Videos S1–S3).
In 2016, a total of 560 males and 28 inflorescences
were tested (4 inflorescences/manipulation). In 2017,
320 males and 28 inflorescences could be tested for
experiments without re-attached flower tissue (4 inflo-
rescences/manipulation) and 695 males and 49 inflores-
cences for experiments with re-attached flower tissue
(7 inflorescences/manipulation). The number of males
tested was limited in 2017 by poor weather conditions.
Duration of pseudocopulation
Although it is expected that longer pseudocopulation
times increase the number of massulae deposited on
the stigma, this hypothesis has so far not been tested
(De Jager & Peakall, 2015). In 2017, we therefore per-
formed hand and natural pollination experiments in
order to assess the relationship between the number of
massulae deposited and the duration of the pseudocop-
ulation. For hand pollinations, pollinia were removed
using a toothpick and applied by repeated movement to
the stigma of another plant for 5, 10, 20, 30 and 60 s,
respectively (corresponding to the most frequent pseu-
docopulation times observed in the experiments
(a) (b) (c) (d)
Fig. 1 Illustration of the three flower types in Ophrys: (a) Wasp-pollinated flowers O. insectifera (WT) are characterized by a very slender,
almost rectangular lip form; (b) Eucera-pollinated flowers, here Ophrys leochroma, are characterized by a squared or trapezoid lip form; (c)
Andrena-pollinated fusca-type flowers, here O. iricolor (FT), are characterized by a triangle-shaped or narrow trapezoid lip form; (d) Eucera
kullenbergi pseudocopulating on O. leochroma. Areas used by males as gripping points (marked with green arrows/dots) are similar in all
three flower types. Regions of the lip with which males interact less accurately are marked with red arrows/dots.
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
4D. RAKOSY ET AL.
performed in 2016). For natural pollination experi-
ments, we chose the smaller E. kullenbergi population,
where a higher frequency of pollinia-carrying males
could be achieved. For this purpose, we allowed males
to remove pollinia from presented inflorescences. After
about 60 min, the flowers were offered again and the
duration of pseudocopulation of pollinia-carrying males
was recorded with a stopwatch. We measured the dura-
tion of pseudocopulations from when the males first
contacted the thecae of the pollinia until they left the
flower. The 10 hand-pollinated and the 13 naturally
pollinated flowers were individually marked and stored
in 70% ethanol. The number of deposited massulae
was then counted under a microscope (x20). Low sam-
ple sizes are due to the very fast habituation of males
in small populations.
In order to assess whether changes in floral form can
affect the duration of pseudocopulation, we measured
the duration of pseudocopulation in the control and the
manipulated flowers in both 2016 and 2017. Sample
sizes correspond to the number of pseudocopulations
reported in the behavioural experiments described
above.
Experiments designed to control for possible scent bias
Currently, there is no evidence which would suggest
that in Ophrys individual components of the floral scent
are produced in different regions of the lip. However,
manipulations performed by removing portions of the
lip are likely to bare the risk of altering the total
amount of scent produced by the flowers. Theoretically,
this might lead to a dose-dependent behavioural
response in males, which could bias the frequency and/
or duration of pseudocopulations on experimentally
altered flower forms. In the following, we describe
three scenarios that reflect the ways in which alter-
ations of the scent production during floral manipula-
tions could have biased our results and the steps taken
to control for each type of bias:
1) The reduction in the scent-producing floral tissue could
lead to less-effective long-distance attraction. This is
highly unlikely to have biased our results as we
performed our experiments directly in the swarming
areas of the males, thus circumventing the necessity
for long-distance attraction. Nevertheless, we have
addressed this issue by:
a comparing the average number of males which
approached control and manipulated flowers dur-
ing a 1-min time frame (i.e. overall attractivity). If
alterations of the scent-producing flower tissue
would have affected long-distance attraction, we
would expect the total number of attracted males
to decrease in relation to the amount of
tissue removed, thus following the pattern:
WT < WDSL < FT < WLL < WA < WSL ≤C.
b comparing the overall attractivity between flow-
ers in which the removed flower tissue was
discarded and flowers in which this tissue was
re-attached. If removing floral tissue has biased
long-distance attraction, we would expect to
find an increase in the overall attractivity in
flowers in which the removed flower tissue was
re-attached.
2) The reduction in the scent-producing floral tissue could lead
to less-effective short-distance attraction and reduced pseu-
docopulation attempts. In this case, we would expect
that manipulations in which the greatest amount of
tissue was removed would show less frequent land-
ings and pseudocopulations than manipulations
where less of the total lip surface was affected. Our
results should thus reflect the following pattern:
WT<WDSL<FT<WLL<WA<WSL≤C. The same
results would be expected for the duration of pseudo-
copulations. Under these circumstances, we would
also predict that scent supplementation in manipu-
lated flowers would reduce or remove differences
between the manipulation groups.
3) There is no or only a negligible dose-dependent effect on
pollinator behaviour. In this case, we would expect no
significant difference in the number of approaches,
short and long landings between manipulation
groups. Instead, the number and duration of pseu-
docopulations would be lowest in manipulations
Fig. 2 Illustrations of control (C) and manipulated Ophrys leochroma flowers: fusca type (FT); wasp type (WT); flower without lateral lobes
(WLL); flower without stigmatic lobe (WSL); flower without appendix (WA); flower without distal-side lobes (WDSL).
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 5
resembling other Ophrys floral types. Manipulations
in which mechanically active components were
removed would in turn be less effective in securing
pseudocopulations than those where mechanically
inactive components were removed: (WT and
FT) ≤(WLL and WSL) < (WDSL and WA) ≤C. Pol-
linators would also be expected to show a consistent
response between manipulated flowers with and
without scent supplementation.
Statistics
Behavioural responses to manipulation of the flower
form
The response of each male was treated as an indepen-
dent case during statistical testing. Although we cannot
exclude that some males visited more than one flower,
the large number of males and their ability to learn to
avoid flowers makes this event unlikely (Ayasse et al.,
2000). Differences between the observed and expected
number of responses between each manipulation group
were tested using a chi-square test of goodness of fit.
The observed frequency of responses in control flowers
was considered as the expected frequency when com-
paring manipulated to control flowers. For comparisons
between manipulation groups, the expected frequencies
were calculated from the observed frequencies of the
group to which the comparison was assessed. Because
of differences in the number of males in each treatment
group, expected frequencies were individually adjusted
to correspond to the number of males in each group.
Changes in the duration of pseudocopulation due to
manipulation of the flower form
We tested whether the duration of pseudocopulation
can predict the number of massulae deposited on the
stigma by means of a linear regression analysis. Further,
pairwise differences in the duration of the pseudocopu-
lation between groups were compared by means of
independent-sample t-tests.
Experiments designed to control for possible scent bias
The overall attractiveness of control and manipulated
flowers was compared using a Kruskal–Wallis test. The
same procedure was used in 2017 to compare the over-
all attractiveness of flowers with and without re-
attached flower tissue.
All statistics were performed using SPSS 20.0.0,
whereas graphical outputs were created in SPSS and
Photoshop CS5 (version 12.0.4 x32).
Results
Our results showed that the experimental alteration of
the floral form had a significant effect on changes in
the frequency and duration of the responses elicited in
the male pollinators. This effect was unlikely to have
been biased by alterations of chemical cues induced by
the manipulation of the lip because:
1) No difference in the effectiveness of long-distance attrac-
tion was found. Instead, the overall attractivity of
control and manipulated flowers was statistically
the same, similar results being found also between
manipulated flowers (2016 without re-attached
flower tissue A : X
2
= 9.344, d.f. = 6, P= 0.155; 2017
without re-attached flower tissue: X
2
= 10.761, d.f. = 6,
P= 0.096; 2017 with re-attached flower tissue: X
2
=
3.422, d.f. = 6, P= 0.754). Scent supplementation did
not improve overall attractivity as no significant differ-
ences were found between experiments where the
removed flower tissue was discarded and those in which
the flower tissue was re-attached (2017: X
2
= 0.106,
d.f. = 6, P=0.744).
2) Short-distance attraction and pseudocopulation rates did
not change in relation to the amount of tissue removed:
We thus found that in manipulated flowers short
landings increased, whereas the number of long
landings remained relatively unchanged as pseudo-
copulations became less frequent. However, differ-
ences found in the frequency of landings and
pseudocopulations did not decrease proportionally
to the increase in the amount of tissue removed
within each manipulation group. Neither did the
duration of pseudocopulations (Table 1a, b). In addi-
tion, no difference in the repertoire of pollinator
behaviour was detected between flowers for which
we explicitly controlled for any potential effect on
floral scent, and those we did not (Tables 2a, b and
3a, b).
3) Our results are consistent with the third hypothesis,
suggesting that floral manipulations had no or only
a negligible dose-dependent effect on floral scent
production. Changes in the behaviour of the polli-
nators can be thus attributed to the alteration of the
floral form.
Frequency of landings and pseudocopulations
The response pattern of E. kullenbergi males to alter-
ations of the flower form in O. leochroma was found to
remain consistent between years and experimental set-
ups (Tables 1–3, Fig. 3a–c). Manipulated flowers thus
usually elicited significantly more landings and less
pseudocopulations than the control (Tables 1–3,
Fig. 3a–c). Although the frequency of landings was
relatively similar between manipulated flowers, we
found that the frequency of pseudocopulations varies
according to the type of manipulation (Tables 1–3,
Fig. 3a–c). Manipulations to the overall flower form
thereby elicited the lowest number of pseudocopula-
tions, followed by manipulations of mechanically
active components of the flower form. Manipulation
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
6D. RAKOSY ET AL.
Table 1 Experiments without re-attached flower tissue (2016): a) Comparisons between the frequency of behavioural responses in male Eucera kullenbergi towards the control and
manipulated Ophrys leochroma flowers: statistical values for chi-square test of goodness of fit (lower left half); b) pairwise comparisons of the amount of time males spent
pseudocopulating with the control and manipulated orchid flowers: statistical values for independent-sample t-test (upper right half).
b) Duration of pseudocopulation
C FT WT WLL WSL WA WDSL
Ct=6.04 t=6.77 t=4.7 t=3.07 t=2.11 t=-2.00
d.f. =52.3, P=0.0000001** d.f. =48.43, P=0.0000001** d.f. =55.36, P=0.000018** d.f. =53.03, P=0.002* d.f. =78.24, P=0.019* d.f. =65.22, P=0.025*
FT X
2
=66.49, d.f. =2,
P=0.000001** t=-1.39 t=2.52 t=2.71 t=4.57 t=3.7
SL
r
=16.2 LL
r
=17 K
r
=33.2 d.f. =12, P=0.05 d.f. =34, P=0.009* d.f. =16.79, P=0.008* d.f. =38.24, P=0.000049** d.f. =26.41, P=0.0005**
WT X
2
=159.44, d.f. =2,
P=0.000001**
X
2
=24.23, d.f. =2,
P=0.000005** t=2.40 t=3.54 t=5.55 t=4.45
SL
r
=40 LL
r
=0K
r
=-40 SL
r
=15 LL
r
=-8 K
r
=7 d.f. =28, P=0.012* d.f. =14.72, P=0.002* d.f. =33.69, P=0.000003** d.f. =23.89, P=0.00017**
WLL X
2
=19.68, d.f. =2,
P=0.000053**
X
2
=10.19, d.f. =2,
P=0.006*
X
2
=51.73, d.f. =2,
P=0.000001** t=1.13 t=2.84 t=2.28
SL
r
=12 LL
r
=5K
r
=17 SL
r
=-8 LL
r
=4K
r
=12 SL
r
=28 LL
r
=5K
r
=23 d.f. =18.17, P=0.135 d.f. =42.25, P=0.004* d.f. =28.14, P=0.02*
WSL X
2
=51.93, d.f. =2,
P=0.000001**
X
2
=1.21, d.f. =2,
P=0.55
X
2
=46.36, d.f. =2,
P=0.000001**
X
2
=10.02, d.f. =2,
P=0.007* t=1.07 t=0.93
SL
r
=18 LL
r
=14 K
r
=32 SL
r
=5LL
r
=3K
r
=2SL
r
=24 LL
r
=13 K
r
=11 SL
r
=6LL
r
=8K
r
=14 d.f. =46, P=0.145 d.f. =37, P=0.09
WA X
2
=28.57, d.f. =2,
P=0.00001**
X
2
=11.51, d.f. =2,
P=0.03*
X
2
=42.93, d.f. =2,
P=0.000001**
X
2
=0.96, d.f. =2,
P=0.62
X
2
=12.61, d.f. =2,
P=0.02* t=0.07
SL
r
=18 LL
r
=1K
r
=19 SL
r
=5.7 LL
r
=7.7 K
r
=13.3 SL
r
=25 LL
r
=1K
r
=24 SL
r
=3LL
r
=4K
r
=1SL
r
=3LL
r
=12 K
r
=15 d.f. =55, P=0.25
WDSL X
2
=27.33, d.f. =2,
P=0.000001**
X
2
=7.16, d.f. =2,
P=0.028*
X
2
=112.72, d.f. =2,
P=0.000001**
X
2
=0.69, d.f. =2,
P=0.71
X
2
=6.02, d.f. =2,
P=0.0049
X
2
=1.14, d.f. =2,
P=0.57
SL
r
=15 LL
r
=5K
r
=20 SL
r
=6LL
r
=4K
r
=10 SL
r
=25 LL
r
=5K
r
=20 SL
r
=3LL
r
=0K
r
=3SL
r
=2LL
r
=8K
r
=10 SL
r
=0LL
r
=4K
r
=4
a) Behavioural responses
*P<0.05 and **P<0.001. C, control flowers; FT, fusca-type flowers; WT, wasp-type flowers; WLL, flowers without lateral lobes; WSL, flowers without stigmatic lobe; WA, flowers
without appendix; WDSL, flowers without distal-side lobes; SL, short landings; LL, long landings; K, pseudocopulations.
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 7
Table 2 Experiments without re-attached floral tissue (2017): a) Comparisons between the frequency of behavioural responses in male Eucera kullenbergi towards the control and
manipulated Ophrys leochroma flowers: statistical values for chi-square test of goodness of fit (lower left half); b) pairwise comparisons of the amount of time males spent
pseudocopulating with the control and manipulated orchid flowers: statistical values for Mann–Whitney U-test (upper right half).
b) Duration of pseudocopulation
C FT WT WLL WSL WA WDSL
Ct=2.27 t=10.58 t=3.1 t=2.14 t=0.46 t=1.02
d.f. =40, P=0.015* d.f. =19.2 P=0.000001** d.f. =43, P=0.015* d.f. =39, P=0.019* d.f. =57, P=0.32 d.f. =53, P=0.155*
FT X
2
=19.64, d.f. =2,
P=0.000054** t=3.13 t=0.31 t=0.08 t=1.84 t=1.96
SL
r
=5LL
r
=10 K
r
=15 d.f. =9, P=0.006* d.f. =10.45, P=0.38 d.f. =7, P=0.47 d.f. =25, P=0.039* d.f. =21, P=0.032*
WT X
2
=20.46, d.f. =2,
P=0.000036
X
2
=0.52, d.f. =2,
P=0.77 t=2.37 t=2.91 t=8.28 t=6.38
SL
r
=6LL
r
=9K
r
=15 SL
r
=–1LL
r
=2K
r
=1 d.f. =8.61, P=0.022* d.f. =8, P=0.01* d.f. =23.46,
P=0.000001**
d.f. =22,
P=0.000002**
WLL X
2
=15.74, d.f. =2,
P=0.00038**
X
2
=2.41, d.f. =2,
P=0.3
X
2
=1.35, d.f. =2,
P=0.51 t=0.21 t=2.48 t=2.06
SL
r
=4LL
r
=10 K
r
=14 SL
r
=2LL
r
=1K
r
=3SL
r
=2LL
r
=0K
r
=2 d.f. =7.75, P=0.42 d.f. =28, P=0.01* d.f. =9.42,
P=0.034*
WSL X
2
=31.73, d.f. =2,
P=0.000001**
X
2
=3.02, d.f. =2,
P=0.22
X
2
=5.96, d.f. =2,
P=0.048*
X
2
=4.57, d.f. =2,
P=0.10 t=1.76 t=1.91
SL
r
=3LL
r
=18 K
r
=21 SL
r
=3LL
r
=4K
r
=2SL
r
=4LL
r
=6K
r
=3SL
r
=1LL
r
=6K
r
=4 d.f. =24, P=0.046* d.f. =20, P=0.036*
WA X
2
=5.11, d.f. =2,
P=0.078
X
2
=10.61, d.f. =2,
P=0.049*
X
2
=10.41, d.f. =2,
P=0.006*
X
2
=8.51, d.f. =2,
P=0.014*
X
2
=67.65, d.f. =2,
P=0.000001** t=0.49
SL
r
=4LL
r
=3K
r
=7 SLr =-2 LL
r
=-8 K
r
=10 SL
r
=3LL
r
=7K
r
=10 SL
r
=0LL
r
=8K
r
=9SL
r
=1LL
r
=-19 K
r
=17 d.f. =38, P=0.315
WDSL X
2
=2.848, d.f. =2,
P=0.24
X
2
=10.13, d.f. =2,
P=0.006*
X
2
=10.15, d.f. =2,
P=0.006*
X
2
=6.83, d.f. =2,
P=0.033*
X
2
=19.08, d.f. =2,
P=0.00007**
X
2
=0.95, d.f. =2,
P=0.62
SL
r
=2LL
r
=4K
r
=6SL
r
=3LL
r
=7K
r
=-10 SL
r
=4LL
r
=5K
r
=10 SL
r
=2LL
r
=6K
r
=9SL
r
=1LL
r
=14 K
r
=15 SL
r
=2LL
r
=2K
r
=0
a) Behavioural responses
*P<0.05 and **P<0.001. C, control flowers; FT, fusca-type flowers; WT, wasp-type flowers; WLL, flowers without lateral lobes; WSL, flowers without stigmatic lobe; WA, flowers
without appendix; WDSL, flowers without distal-side lobes; SL, short landings; LL, long landings; K, pseudocopulations.
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
8D. RAKOSY ET AL.
Table 3 Experiments with re-attached flower tissue (2017): a) Comparisons between the frequency of behavioural responses in male Eucera kullenbergi towards the control and
manipulated Ophrys leochroma flowers: statistical values for chi-square test of goodness of fit (lower left half); b) pairwise comparisons of the amount of time males spent
pseudocopulating with the control and manipulated orchid flowers: statistical values for Mann–Whitney U-test (upper right half).
b) Duration of pseudocopulation
C FT WT WLL WSL WA WDSL
Ct=5.26 t=12.1 t=1.10 t=4.20 t=0.181 t=1.79
d.f. =45, P=0.000004** d.f. =42.8, P=0.000001** d.f. =48, P=0.14 d.f. =48, P=0.000114** d.f. =52, P=0.43 d.f. =62, P=0.039*
FT X
2
=53.43, d.f. =2,
P=0.000001** t=1.07 t=2.79 t=1.29 t=4.87 t=3.01
SL
r
=10 LL
r
=20 K
r
=-30 d.f. =10, P=0.155 d.f. =21, P=0.006* d.f. =21, P=0.11 d.f. =35, P=0.00002** d.f. =35, P=0.003*
WT X
2
=105.08, d.f. =2,
P=0.000001**
X
2
=10.11, d.f. =2,
P=0.006 t=4.37 t=3.01 t=9.96 t=6.42
SL
r
=19 LL
r
=14 K
r
=33 SL
r
=10 LL
r
=5K
r
=5 d.f. =12.69, P=0.005* d.f. =13.54, P=0.005* d.f. =30.41, P=0.000001 d.f. =29.41, P=0.000001**
WLL X
2
=34.81, d.f. =2,
P=0.000001**
X
2
=2.37, d.f. =2,
P=0.31
X
2
=24.25, d.f. =2,
P=0.000005** t=1.98 t=1.09 t=0.25
SL
r
=7LL
r
=18 K
r
=25 SL
r
=3LL
r
=1K
r
=4SL
r
=13 LL
r
=4K
r
=9 d.f. =24, P=0.03* d.f. =38, P=0.14 d.f. =38, P=0.4
WSL X
2
=47.01, d.f. =2,
P=0.000001**
X
2
=0.69, d.f. =2,
P=0.71
X
2
=19.41, d.f. =2,
P=0.00006**
X
2
=0.36, d.f. =2,
P=0.84 t=3.92 t=2.07
SL
r
=9LL
r
=20 K
r
=29 SL
r
=2LL
r
=0K
r
=2SL
r
=12 LL
r
=4K
r
=8SL
r
=1LL
r
=1K
r
=2 d.f. =38, P=0.00018** d.f. =38, P=0.023*
WA X
2
=10.23, d.f. =2,
P=0.006*
X
2
=41.71, d.f. =2,
P=0.000001**
X
2
=106.44, d.f. =2,
P=0.00001**
X
2
=13.82, d.f. =2,
P=0.001**
X
2
=16.51, d.f. =2,
P=0.0003** t=1.72
SL
r
=6LL
r
=4K
r
=10 SLr =3LL
r
=15 K
r
=18 SL
r
=12 LL
r
=10 K
r
=22 SL
r
=1LL
r
=14 K
r
=15 SL
r
=2LL
r
=15 K
r
=17 d.f. =52, P=0.046*
WDSL X
2
=10.77, d.f. =2,
P=0.005*
X
2
=31.13, d.f. =2,
P=0.000001
X
2
=89.28, d.f. =2,
P=0.000001**
X
2
=10.51, d.f. =2,
P=0.005*
X
2
=17.91, d.f. =2,
P=0.00013**
X
2
=5.31, d.f. =2,
P=0.07
SL
r
=2LL
r
=13 K
r
=15 SL
r
=9LL
r
=7K
r
=16 SL
r
=19 LL
r
=2K
r
=21 SL
r
=5LL
r
=6K
r
=11 SL
r
=7LL
r
=-7 K
r
=14 SL
r
=52 LL
r
=9K
r
=3
a) Behavioural responses
*P<0.05 and **P<0.001. C, nonmanipulated flowers; FT, fusca-type flowers; WT, wasp-type flowers; WLL, flowers without lateral lobes; WSL, flowers without stigmatic lobe; WA,
flowers without appendix; WDSL, flowers without distal-side lobes; SL, short landings; LL, long landings; K, pseudocopulations.
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 9
of mechanically inactive components elicited the high-
est number of pseudocopulations among the manipu-
lation types (Tables 1–3, Fig. 3a–c). Differences within
manipulation groups were less pronounced. Wasp-type
flowers (WT) generally elicited significantly less pseu-
docopulations than fusca-type flowers (FT). Flowers
with removed lateral lobes (WLL) usually elicited
pseudocopulation as frequently as flowers without the
stigmatic lobe (WSL), whereas flowers without distal-
side lobes (WDSL) were as successful as flowers with-
out the appendix (WA). The only major difference
was found in the number of long landings, the fre-
quency of males landing without pseudocopulating on
the lips increasing in 2017. However, this increase was
observed in experiments with and without re-attached
floral tissue.
Duration of pseudocopulations
The simple linear regression analysis we performed
revealed that the number of massulae deposited on the
stigma can be predicted based on the duration of the
pseudocopulation. We thereby found a significant
regression equation for both the hand-pollinated flow-
ers (R
2
=0.376, F
(1,18)
=10.854, P=0.004) and flowers
pollinated by E. kullenbergi males (R
2
=0.604,
F
(1,13)
=19.864, P=0.001) (Fig. 4a, b). For hand polli-
nation, the number of massulae deposited was
Fig. 3 Male Eucera kullenbergi behavioural response to control and manipulated Ophrys leochroma flowers. I –control flowers (C); II –
overall manipulation of flower form: fusca-type (FT) and wasp-type (WT) flowers; III –manipulation of mechanically active components of
floral form: flowers without lateral lobes (WLL) and without the stigmatic lobe (WSL); IV –manipulation of mechanically inactive
components of flower form: flowers without appendix (WA) and without the distal-side lobes (WDSL). The responses of the males were
classified as short landings, long landings and pseudocopulations. Significance values are for overall comparisons between groups. Each
letter designates a manipulation group, whereas different cases indicate significant differences between the number of responses in each
behavioural group. (a) Experiments conducted in 2016 without scent supplementation (i.e. without re-attaching the removed flower parts
to the manipulated flowers). (b) Experiments conducted in 2017 without scent supplementation (i.e. without re-attaching the removed
flower parts to the manipulated flowers). (c) Experiments conducted in 2017 with scent supplementation (i.e. by re-attaching the removed
flower parts to the manipulated flowers).
Fig. 4 Simple linear regression analysis showing that the number of massulae deposited during one pseudocopulation event can be
predicted from the duration of the pseudocopulation: the number of massulae increases when pseudocopulations last longer. (a) Hand
pollination experiments; (b) natural pollination experiments.
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
10 D. RAKOSY ET AL.
predicted to increase by 0.364 with each second of
pseudocopulation, whereas for flowers pollinated by
E. kullenbergi males, the number of massulae was pre-
dicted to increase by 0.26 massulae with each second of
the pseudocopulation.
The time that males spent pseudocopulating with the
flowers varied significantly depending on the type of
manipulation (Kruskal–Wallis, 2016: X
2
=37.32,
d.f. =6, P=0.000002; 2017 with re-attached flower
tissue: X
2
=41.57, d.f. =6, P=0.000001; 2017 with
re-attached flower tissue: X
2
=25.1, d.f. =6,
P=0.000328). These results were found to be mostly
consistent between years and experiments with and
without re-attached flower tissue (Fig. 5a–c). Manipu-
lated flowers thereby elicited significantly shorter pseu-
docopulations than the control (P<0.05; Tables 1–3,
Fig. 5a–c). The strongest decrease in pseudocopulation
time in comparison with control flowers was observed
in fusca- and wasp-type flowers (Fig. 5a–c). The dura-
tion of pseudocopulation in groups in which mechani-
cally active components were manipulated usually
showed significantly shorter pseudocopulation times
than in groups in which mechanically inactive compo-
nents were manipulated. Pairwise differences within
these two groups were, however, not significant
(Tables 1–3, Fig. 5a–c). Thus, pollinators spent similar
amounts of time pseudocopulating with flowers with
removed lateral lobes and flowers in which the stig-
matic lobe was removed. The same was true for flowers
in which the appendix had been removed and those
without distal-side lobes. The only difference was found
between the duration of pseudocopulation in flowers
without lateral lobes (WLL), which showed an increase
in the duration of pseudocopulations in experiments in
which the removed flower tissue was re-attached.
Discussion
The striking visual and morphological resemblance
between many sexually deceptive orchids and their
insect models has traditionally been assumed to have
been shaped by pollinator-mediated selection (Agren
et al., 1984; Paulus, 2007; Gaskett, 2012). Several
recent experimental studies conducted across cases of
pollinator sharing among sexually deceptive orchids
have supported this assumption by demonstrating the
plausibility of pollinator-mediated selection (De Jager &
Peakall, 2015; G€
ogler et al., 2015). Our study extends
this work in a different dimension by experimentally
manipulating floral morphology within a species. We
could thus show that in O. leochroma, the form of the
lip is essential for ensuring the contact between the
pollinators and the reproductive structures of the flow-
ers. However, the effectiveness of pollen transfer varied
significantly depending on which components of floral
form were manipulated. We could thus provide the first
empirical evidence showing that mechanically active
components of flower form are under stronger pollina-
tor selection than mechanically inactive components.
This finding sheds new light into the way pollinator-
mediated selection may have driven floral form evolu-
tion in Ophrys.
The role of flower form in mediating the interactions
between orchids of the genus Ophrys and their
pollinators
The sexual pheromone analogue produced by sexually
deceptive orchids is essential for ensuring species-speci-
fic pollinator attraction and for eliciting copulatory
behaviour in male pollinators (Ayasse et al., 2011;
Fig. 5 Mean duration of pseudocopulations on control and manipulated Ophrys leochroma flowers. I –control flowers (C); II –overall
manipulation of flower form: fusca-type (FT) and wasp-type (WT) flowers; III –manipulation of mechanically active components of floral
form: flowers without lateral lobes (WLL) and without the stigmatic lobe (WSL); IV –manipulation of mechanically inactive components
of flower form: flowers without appendix (WA) and without the distal-side lobes (WDSL). Each letter designates a manipulation group,
whereas different cases indicate significant differences between the duration of the pseudocopulations in each behavioural group. (a)
Experiments conducted in 2016 without scent supplementation (i.e. without re-attaching the removed flower parts to the manipulated
flowers). (b) Experiments conducted in 2017 without scent supplementation (i.e. without re-attaching the removed flower parts to the
manipulated flowers). (c) Experiments conducted in 2017 with scent supplementation (i.e. by re-attaching the removed flower parts to the
manipulated flowers).
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 11
Peakall & Whitehead, 2014). However, the effectiveness
of pollen transfer depends on whether males are able to
contact the reproductive parts of flowers during pseu-
docopulation (e.g. Nilsson, 1992). Flower form could
affect the frequency of pollen transfer by mediating pol-
linator attraction and/or by ensuring the mechanical fit
between the body of the pollinator and the flower
(Cresswell, 1998; Benitez-Vieyra et al., 2009; Gaskett,
2012; De Jager & Peakall, 2015; G€
ogler et al., 2015). In
turn, the amount of pollen deposited is likely to depend
on the accuracy of the mechanical fit and the time
males spend attempting to copulate with the flowers
(Cresswell, 1998; De Jager & Peakall, 2015).
Despite the primary role of the sexual pheromone
analogue for mediating the interaction between sexu-
ally deceptive orchids of the genus Ophrys and their
pollinators, our study suggests that floral form plays a
so far underestimated role in ensuring pollinia removal
and deposition. We found that in O. leochroma the
manipulation of the form of the lip did not reduce the
overall attractivity of the flowers towards their pollina-
tors. Instead, most approaching males landed on the
O. leochroma flowers but only started pseudocopulating
when they found proper gripping points on the lip. In
manipulated flowers, this happened less frequently and
pseudocopulations were significantly shorter than in
control flowers. Consequently, the flower form seems
to mediate efficient pollen transfer in Ophrys predomi-
nantly by ensuring the mechanical fit between the
flowers and their pollinators rather than by providing a
visual signal to approaching males. This result is consis-
tent with other recent experimental studies that investi-
gated the function of floral form in a diverse spectrum
of sexually deceptive systems (G€
ogler et al., 2015; Phil-
lips et al., 2013; Phillips et al., 2014; De Jager & Peakall,
2015). Further studies are, however, needed to clarify
the contribution of floral form to visual attraction in
sexually deceptive systems, a contribution that may
strongly vary between species (Benitez-Vieyra et al.,
2009; De Jager & Peakall, 2015).
All studies investigating the function of floral form in
sexually deceptive orchids face one major challenge:
altering floral form without affecting the emission of the
sexual pheromone analogue (De Jager & Peakall, 2015).
For instance in Ophrys where the lip is the main scent-
producing organ, alterations of the form of the lip could
also result in changes in the amount of scent produced
(Francisco & Ascensao, 2013). However, in O. leochroma,
we did not find any conclusive evidence for scent bias
altering the response of males towards the manipulated
flowers. A possible explanation might lie in the fact that
Ophrys flowers produce very high amounts of the sexual
pheromone analogue, often higher than the amount of
sex pheromone in females (Ayasse et al., 2003; Schiestl,
2004). Therefore, even manipulated flowers with
strongly reduced lips can apparently still emit enough
scent to be highly attractive to their male pollinators.
We thus conclude that the differences in the beha-
vioural responses of males towards control and manipu-
lated O. leochroma flowers were mostly due to
alterations of the flower form. This suggests that the
manipulated flowers are less adaptive than the control,
the altered floral form leading to less efficient pollen
transfer and reduced reproductive success.
Flower form and reproductive success: evidence for
pollinator-mediated selection
In sexually deceptive orchids, it is expected that pseu-
docopulations are associated with the removal of one
or both pollinia, whereas successful massulae deposition
requires males to spend a certain period of time pseu-
docopulating (Paulus, 2007). The frequency of pseudo-
copulations could thus be considered a good proxy of
male reproductive success, whereas the duration of
pseudocopulations could act as a proxy of female repro-
ductive success (De Jager & Peakall, 2015). Field obser-
vations have shown that if males align with the
reproductive structures of Ophrys flowers, about 90% of
pseudocopulations are associated with pollinia removals
(H. Paulus, D. Rakosy pers. obs.). By performing both
hand and natural pollination experiments, we could
now show that the number of pollen grains deposited
on the stigma during a single pollination event
increases significantly with the time males spent pseu-
docopulating. The number of pollen grains deposited on
control flowers of O. leochroma was thereby found to be
sufficient to fertilize all available ovules (Paulus, 1988;
Nazarov & Gerlach, 1997). Although more data are
needed, we consider that alterations in the lip form can
not only reduce the frequency of successful pollinia
removals, but will also affect the number of ovules
being fertilized.
Extreme variations in the lip form may therefore
greatly reduce reproductive success (Ayasse et al., 2000;
Schiestl & Ayasse, 2001; Paulus, 2007). One would thus
expect that pollinator-mediated stabilizing selection has
shaped the evolution of species-specific flower form
optima, leading to an increased frequency of accurate
pollinia removal and higher pollen loads on the stigma
(e.g. Cresswell, 1998; Gaskett, 2012; Joly et al., 2016).
The evolution of floral form types in Ophrys
The three major floral types associated with particular
pollinator genera may represent such form optima, each
floral type being adapted to maximize the mechanical fit
between the flowers and their specific pollinators. Our
results show that changes in floral form types produced
the strongest reduction in terms of both the frequency
and the duration of pseudocopulations. These overall
changes in floral form were aimed at resembling the
forms of two relatives of the study group, namely those
pollinated by wasp species and Andrena bees, respectively
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
12 D. RAKOSY ET AL.
(Breitkopf et al., 2014). These species rely on highly
different sexual pheromone analogues for pollinator
attraction, but they also present strikingly different mor-
phologies (Ayasse et al., 2003; St€
okl et al., 2005). Our
experiments show that E. kullenbergi males copulated less
frequently (and if they did, they held on to the flowers
for only a short time) with flowers having forms corre-
sponding to species pollinated by either wasps or Andrena
bees, even though the flowers emitted the specific scent
of their females. The reduced frequency of pseudocopu-
lations and their short duration indicate that, on flowers
manipulated to resemble other floral types, form compo-
nents were missing that the pollinators would normally
use as gripping and/or contact points, thus reducing the
mechanical fit.
If particular pollinator groups differ in the efficiency
with which they interact with floral types, then selection
would be expected to lead to an increased accuracy in
the mechanical fit and a more effective pollinia transfer
by the specialized pollinator (e.g. Cresswell, 1998). If
morphological and behavioural differences between the
pollinators are large enough, such adaptations could in
time lead to morphological differentiation even between
closely related species (Sedeek et al., 2014). A recent
study has found that in Ophrys species sharing a common
pollinator, morphological differences can maintain
reproductive isolation (G€
ogler et al., 2015). It appears,
however, unlikely that if the floral scent barrier between
Ophrys species, which rely on the sexual pheromone ana-
logue for reproductive isolation, would break down,
morphological differences alone would be able to filter
out unspecific visitors (Cortis et al., 2009; Johnson &
Schiestl, 2016). Instead, our results suggest that flower
form acts to maximize reproductive success in a system
which is severely pollinator limited (Tremblay et al.,
2005). In most species, floral form is thus likely to com-
plement the function of floral scent, by increasing the
probability that the specific pollinator can ensure effi-
cient pollen transfer. The functional interaction between
floral scent and other floral traits, such as form and col-
our, in ensuring successful pollen transfer will, however,
require some additional attention.
Pollinator-mediated selection and the evolution of
mechanically active and inactive components of
floral form
Strong pollinator-mediated selection is not only
expected to lead to the evolution of floral forms
adapted to specific pollinator groups, but to also reduce
the amount of intraspecific variation (e.g. Cresswell
et al., 1998). However, the genus Ophrys is notorious
for high intraspecific variation in its floral form (Del-
forge, 2005); this has been traditionally interpreted as
either the outcome of hybridization or relaxed selec-
tion (Soliva & Widmer, 2003; Tremblay et al., 2005;
Devey et al., 2008). Alternatively, pollinators may be
able to select for particular lip forms, but the strength
of selection may vary between the various components
of the lip. This would leave mechanically inactive
components to be more variable than mechanically
active ones. Although the concept of classifying flower
form components into mechanically active and inactive
has been previously discussed, we here provide the
first evidence for the potential of different selection
pressures on these two components of the lip form
(Gaskett, 2012).
In O. leochroma, we have classified the stigmatic and
the lateral lobes as mechanically active and the appen-
dix and the distal-side lobes as mechanically inactive.
According to our expectations, differences in the
response frequency and duration of pseudocopulations
were relatively clear cut between the two trait compo-
nents. We thus found that changes in the mechanically
active components of floral form reduced the frequency
and duration of pseudocopulations more strongly than
changes in mechanically inactive components. The
presence of mechanically active components seems thus
essential for ensuring effective pollen transfer. It can
therefore be concluded that mechanically active traits
are likely to be under stronger selection pressure by
pollinators than mechanically inactive traits.
Weaker selection on particular lip form components
probably results in higher variation. Another possibility
is that mechanically inactive components are shaped by
other selective mechanisms such as selection for visual
display. High variation in particular traits might also be
an adaptation to reduce avoidance learning in males, a
mechanism that was first shown in relation to variation
in the sexual pheromone analogue (Ayasse et al., 2000,
2011; St€
okl et al., 2005). Variation may also be a conse-
quence of varying selection pressures across years due
to changing preferences in the males. However, our
results were highly consistent over a 2-year period and
no evidence was found for a change in the preference
of males for particular flower manipulations. Thus, the
evolution of the complex and highly variable floral
form in Ophrys is likely to be driven by differential
selection pressures acting on the two form components.
The mechanically active and inactive components
identified in O. leochroma can be expected to be also
found in most other Ophrys species. We thus consider
that the evolution of variously shaped lateral lobes (e.g.
short in O. tenthredinifera vs. extremely long in
O. scolopax) and stigmatic lobes (e.g. O. tenthredinifera vs.
O. sphegodes) has been driven by pollinator-mediated
selection for improved mechanical fit. The mechanically
inactive components are in turn much more conserved.
While occurring in most Ophrys species, both the lateral
lobes and the appendix show little differentiation
among species, but high variation within species. Thus,
the position of the lateral lobes can range from spread
to completely folded beneath the lip, whereas the
appendix can be triangular to trapezoid, contacting the
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 13
lip or bent to various degrees. This suggests that in most
Ophrys species, variation in these two structures is likely
to be strongly influenced by stochastic factors. How-
ever, further experiments, preferably conducted with
several pollinator species, are necessary to confirm the
adaptive function of the various components of floral
form throughout Ophrys.
Outlook
Mechanically active and inactive components of floral
form can be found not only in the genus Ophrys, but also
in other sexually deceptive orchids and other flowers
that mechanically guide their pollinators towards their
reproductive parts (e.g. flowers with spurs, trap flowers)
(Whittall & Hodges, 2007; Phillips et al., 2014). Failure to
distinguish between these traits in selection studies will
limit our further understanding of floral evolution. With
the rise of three-dimensional (3D) geometric morpho-
metrics and 3D printing, new pathways are opening up
allowing the finer manipulation of both components of
floral form and the integration of this variation with that
in other traits, such as colour and scent. This would
enable the better quantification of the amount of selec-
tion pressure on various traits and trait components and
provide a better understanding of the relative role of pol-
linator-mediated selection and stochastic factors on the
evolution of flower phenotypes.
Acknowledgments
We thank Y. Staedler for his help in planning the
experiments, E. Rakosy-Tican for proofreading the
manuscript, T. Jones for improving our English and two
anonymous reviewers for their substantial contribution
to improving previous versions of the manuscript. The
research has been supported by funds provided by the
Vienna Orchid Society and the Institute of Evolutionary
Ecology and Conservation Genomics, University Ulm.
Collection permits were granted by the Ministry of
Reconstruction of Production, Environment and Energy,
Directorate General for the Protection and Development
of Forests and the Rural Environment in Athens
(Greece) (permission number 121894/490).
Conflict of interest
The authors do not have any conflict of interest to
declare.
References
Agren, L., Kullenberg, B. & Sensenbauch, T. 1984. Congru-
ences in pilosity between three species of Ophrys (Orchi-
daceae) in their hymenopteran pollinators. Nova Acta Reg.
Soc. Sci. Upsaliensis 5:15–25.
Ayasse, M., Schiestl, F.P., Paulus, H.F., L€
ofstedt, C., Hansson,
B., Ibarra, F. et al. 2000. Evolution of reproductive strategies
in sexually deceptive orchid Ophrys sphegodes: how does
flower-specific variation of odor signals influence reproduc-
tive success? Evolution 54: 1995–2006.
Ayasse, M., Paxton, R.J. & Teng€
o, J. 2001. Mating behaviour
and chemical communication in the order Hymenoptera.
Annu. Rev. Entomol. 46:31–78.
Ayasse, M., Schiestl, F.P., Paulus, H.F., Ibarra, F. & Francke,
W. 2003. Pollinator attraction in a sexually deceptive orchid
by means of unconventional chemicals. Proc. R. Soc. Lond. B
Biol. Sci. 270: 517–522.
Ayasse, M., St€
okl, J. & Francke, W. 2011. Chemical ecology
and pollinator-driven speciation in sexually deceptive orch-
ids. Phytochemistry 72: 1667–1677.
Benitez-Vieyra, S., Medina, A.M. & Cocucci, A.A. 2009. Vari-
able selection patterns on the labellum shape of Geoblasta
pennicillata, a sexually deceptive orchid. J. Evol. Biol. 22:
2354–2362.
Bohman, B., Flematti, G.R., Barrow, R.A., Pichersky, E. & Pea-
kall, R. 2016. Pollination by sexual deception—It takes
chemistry to work. Curr. Opin. Plant Biol. 32:37–46.
Breitkopf, H., Onstein, R.E., Cafasso, D., Schl€
uter, P. M. &
Cozzolino, S. 2014. Multiple shifts to different pollinators
fuelled rapid diversification in sexually deceptive Ophrys
orchids. New Phytol. 270: 377–389.
Cortis, P., Vereecken, N.J., Schiestl, F.P., Barone Lumaga,
M.R., Scrugli, A. & Cozzolino, S. 2009. Pollinator conver-
gence and the nature of species’ boundaries in sympatric
Sardinian Ophrys (Orchidaceae). Ann. Bot. 104: 497–506.
Cresswell, J.E. 1998. Stabilizing selection and the structural
variability of flowers within species. Ann. Bot. 81: 463–473.
Cuervo, M., Rakosy, D., Martel, C., Schultz, S. & Ayasse, M.
2017. Sexual Deception in the Eucera-pollinated Ophrys leo-
chroma: a chemical intermediate between wasp- and
Andrena-pollinated species. J. Chem. Ecol. 43: 469–479.
De Jager, M.L. & Peakall, R. 2015. Does morphology matter?
An explicit assessment of floral morphology in sexual decep-
tion. Funct. Ecol. 30: 537–546.
Delforge, P. 2005. Orchids of Europe, North Africa and the Middle
East. A&C Black, London. 640 p. ISBN 071367525X.
van Der Cingel, N. 2001a. An atlas of Orchid Pollination: Euro-
pean Orchids. A. A. Balkema, Rotterdam.
van Der Cingel, N. 2001b. Atlas of Orchid Pollination: America,
Africa, Asia and Australia. A.A. Balkema, Rotterdam.
Devey, D.S., Bateman, R.M., Fay, M.F. & Hawkins, J.A. 2008.
Friends or relatives? Phylogenetics and species delimitation
in the controversial European orchid genus Ophrys.Ann. Bot.
101: 385–402.
Ellis, A.G. & Johnson, S.D. 2010. Floral mimicry enhances pol-
len export: the evolution of pollination by sexual deceit out-
side of the Orchidaceae. Am. Nat. 176: 143–151.
Francisco, A. & Ascensao, L. 2013. Structure of the osmophore
and labellum micromorphology in the sexually deceptive
orchids Ophrys bombyliflora and Ophrys tenthredinifera (Orchi-
daceae). Int. J. Plant Sci. 174: 619–636.
Gaskett, A.C. 2011. Orchid pollination by sexual deception:
pollinator perspectives. Biol. Rev. 86:33–75.
Gaskett, A.C. 2012. Floral shape mimicry and variation in sex-
ually deceptive orchids with a shared pollinator. Biol. J. Linn.
Soc. 106: 469–481.
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
14 D. RAKOSY ET AL.
G€
ogler, J., St€
okl, J., Sramkova, A., Twele, R., Francke, W.,
Cozzolino, S. et al. 2009. M
enage
A Trois—Two endemic
species of deceptive orchids and one pollinator species. Evo-
lution 63: 2222–2234.
G€
ogler, J., St€
okl, J., Cortis, P., Beyrle, H., Lumaga, M.R.B., Coz-
zolino, S. et al. 2015. Increased divergence in floral morphol-
ogy strongly reduces gene flow in sympatric sexually deceptive
orchids with the same pollinator. Evol. Ecol. 29: 703–717.
Hirth, M. & Paulus, H.F. 2016. Neue Best€
auer-Beobachtungen
in der Ophrys tenthredinifera-Artengruppe der
€
Ag€
ais mit der
Beschreibung von Ophrys lychnitis aus Paros. J. Eur. Orch. 48:
346–388.
Johnson, S.D. & Schiestl, F.P. 2016. Floral Mimicry. Oxford
University Press, Oxford. ISBN 9780198732709.
Joly, S., Lambert, F., Alexandre, H., L
eveill
e-Bourret,
E. &
Clark, J.L. 2016. Pollinator specialization imposes stronger
evolutionary constraints on flower shape. bioRxiv, 041533.
Kullenberg, B. 1961. Studies in Ophrys pollination. Zoologiska
bidrag fran Uppsala 34:1–340.
Kullenberg, B. 1973. New observations on the pollination of
Ophrys L. (Orchidaceae). Zoon, suppl 1:9–15.
Martel, C., Cairampoma, L., Stauffer, F.W. & Ayasse, M. 2016.
Telipogon peruvianus (Orchidaceae) flowers elicit pre-mating
behaviour in Eudejeania (Tachinideae) males for pollination.
PloS One 11: e0165896.
Nazarov, V.V. & Gerlach, G. 1997. The potential seed produc-
tivity of orchid flowers and peculiarities of their pollination
systems. Lindleyana 12: 188–204.
van der Niet, T., Zollikofer, C.P.E., Le
on, M.S.P.D., Johnson,
S.D. & Linder, H.P. 2010. Three-dimensional geometric mor-
phometrics for studying floral shape variation. Trends Plant
Sci. 15: 423–426.
Nilsson, L.A. 1992. Orchid pollination biology. Trends Ecol. Evol.
7: 255–259.
Paulus, H.F. 1988. Beobachtungen und Experimente zur Pseu-
dokopulation auf Ophrys Arten (Orchidaceae) Kretas 2: mit
einer Beschreibung von Ophrys sitiaca HF Paulus & CA Alib-
ertis nov. spec. aus dem Ophrys fusca-omegaifera-Formenkreis.
Mitteilungsblatt des Arbeitskreises Heimische Orchideen Baden-
Wurttemberg 20: 817–882.
Paulus, H.F. 2007. Wie Insekten-M€
annchen von Orchi-
deenbl€
uten get€
auscht werden - Best€
aubungstricks und Evo-
lution in der mediterranen Ragwurzgattung Ophrys. In:
Evolution - Ph€
anomen Leben, vol. 20 (F. Gussenleitner, G.
Aubrecht, E. Aescht & M. Pfosser, eds), pp. 255–294. Biolo-
giezentrum der Ober€
osterreichischen Landesmuseen,
Denisia.
Paulus, H.F. & Gack, C. 1990. Pollinators as prepollinating iso-
lation factors: evolution and speciation in Ophrys (Orchi-
daceae). Isr. J. Bot. 39:43–79.
Paulus, H.F. & Hirth, M. 2012. Best€
aubungsbiologie und Sys-
tematik der Ophrys tenthredinifera-Artengruppe in der
Ost€
ag€
ais (Orchidaceae, Insecta). J. Eur. Orch. 44: 625–686.
Peakall, R. & Whitehead, M.R. 2014. Floral odour chemistry
defines species boundaries and underpins strong reproduc-
tive isolation in sexually deceptive orchids. Ann. Bot. 113:
341–355.
Phillips, R.D., Tingbao, Xu, Hutchinson, M.F., Dixon, K.W. &
Peakall, R. 2013. Convergent specialization—The sharing of
pollinators by sympatric genera of sexually-deceptive orch-
ids. J. Ecol. 101: 826–835.
Phillips, R.D., Scaccabarozzi, D., Retter, B.A., Hayes, C.,
Brown, G.R., Dixon, K.W. et al. 2014. Caught in the act: pol-
lination of sexually deceptive trap-flowers by fungus gnats
in Ptersostylis (Orchidaceae). Ann. Bot. 113: 629–641.
Schiestl, F.P. 2004. Floral evolution and pollinator mate choice
in a sexually deceptive orchid. J. Evol. Biol. 17:67–75.
Schiestl, F.P. 2005. On the success of a swindle: pollination by
deception in orchids. Naturwissenschaften 92: 255–264.
Schiestl, F.P. & Ayasse, M. 2001. Post-pollination emission of a
repellent compound in a sexually deceptive orchid: a new
mechanism for maximising reproductive success? Oecologia
126: 531–534.
Schiestl, F.P., Ayasse, M., Paulus, H.F., Lofstedt, C., Hansson,
B.S., Ibarra, F. et al. 1999. Orchid pollination by sexual
swindle. Nature 399: 421.
Schiestl, F.P., Ayasse, M., Paulus, H.F., Lofstedt, C., Hansson,
B.S., Ibarra, F. et al. 2000. Sex pheromone mimicry in the
early spider orchid (Ophrys sphegodes): patterns of hydrocar-
bons as the key mechanism for pollination by sexual decep-
tion. J. Comp. Physiol. A Sens. Neural Behav. Physiol. 186: 567–
574.
Sedeek, K.E.M., Scopece, G., Staedler, Y.M., Sch€
onenberger,
J., Cozzolino, S., Schiestl, F.P. et al. 2014. Genic rather
than genome-wide differences between sexually deceptive
Ophrys orchids with different pollinators. Mol. Ecol. 23:
6192–6205.
Soliva, M. & Widmer, A. 2003. Gene flow across species
boundaries in sympatric, sexually deceptive Ophrys (Orchi-
daceae) species. Evolution 57: 2252–2261.
St€
okl, J., Paulus, H., Dafni, A., Schulz, C., Francke, W. &
Ayasse, M. 2005. Pollinator attracting odour signals in sexu-
ally deceptive orchids of the Ophrys fusca group. Plant Syst.
Evol. 254: 105–120.
Tremblay, R.L., Ackerman, J.D., Zimmerman, J.K. & Calvo,
R.N. 2005. Variation in sexual reproduction in orchids and
its evolutionary consequences: a spasmodic journey to diver-
sification. Biol. J. Linn. Soc. 84:1–54.
Vogel, S. 1962. Duftdr€
usen im Dienste der Best€
aubung.
Abhandlungen der Mathematisch-Naturwissenschaftlichen
Klasse, Akademie der Wissenschaften und der Literatur in
Mainz 10: 599–763.
Weber, E. 2012. Pollinator attraction in long-horned bee polli-
nated Ophrys species on Crete. Master thesis. Ulm University.
Whittall, J.B. & Hodges, S.A. 2007. Pollinator shifts drive
increasingly long nectar spurs in columbine flowers. Nature
447: 706–709.
Xu, S., Schl€
uter, P.M., Scopece, G., Breitkopf, H., Gross, K.,
Cozzolino, S. et al. 2011. Floral isolation is the main repro-
ductive barrier among closely related sexually deceptive
orchids. Evolution 65: 2606–2620.
Supporting information
Additional Supporting Information may be found
online in the supporting information tab for this article:
Figure S1 Map of all the sampled O. leochroma popula-
tions (pink circles from left to right: Prina, Kritsa, Schi-
nokapsala and Orino) and the site at which
experiments were conducted with E. kullenbergi males
(black: Neapoli)
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Flower form affects pollination effectiveness 15
Video S1 Video recording of males approaching a con-
trol flower without landing or pseudocopulating
Video S2 Video recording of a male landing on a con-
trol flower without pseudocopulating
Video S3 Video recording of a male pseudocopulating
and removing pollinia on a control flower while
another male attempts to push him away from the per-
spective female
Data deposited at Dryad: https://doi.org/10.5061/dryad.q7p5j
Received 15 November 2016; revised 23 July 2017; accepted 29 July
2017
ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.13153
JOURNAL OF EVOLUTIONARY BIOLOGY ª2017 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
16 D. RAKOSY ET AL.
