Oecologia (2013) 172:417425

DOI 10.1007/s00442-013-2620-0

BEHAVIORAL ECOLOGY - ORIGINAL RESEARCH

Acquisition of species-specific perfume blends: influence

of habitat-dependent compound availability on odour choices
of male orchid bees (Euglossa spp.)
T. Pokorny M. Hannibal J. J. G. Quezada-Euan
E. Hedenstrom N. Sjoberg J. Bang T. Eltz

Received: 26 November 2012 / Accepted: 6 February 2013 / Published online: 27 February 2013
Springer-Verlag Berlin Heidelberg 2013

Abstract Male orchid bees (Euglossini, Apidae, Hymenoptera) expose species-specific blends of volatile chemicals (perfume bouquets) during their courtship display. The
perfumes are acquired by collecting fragrant substances
from environmental sources, which are then accumulated
in specialised hind leg pouches. To balance the perfume
composition, the males need to find and collect the required
substances in specific relative amounts while facing seasonal, local or habitat-dependent differences in compound
availability. Experience-dependent choice of odours, i.e.
learned avoidance of recently collected components, has
been proposed as the mechanism that mediates the accumulation of the stereotypical compound ratios. In the
present study, we used the presence of certain compounds
in male hind leg pouches as proxy for the respective
local compound availability, and investigated whether

differences in content are correlated with differences in

chemical choice assays. Our results suggest that volatile
availability differs between localities (n = 16) as well as
habitats (n = 2; coastal vs. inland) across the Yucatan
peninsula, Mexico, for both studied species. Male Euglossa
dilemma showed a pronounced preference for benzyl
benzoate and eugenol at locations where those compounds
were rare in hind leg extracts, as predicted by the learned
avoidance model. No equivalent correlations were found
for Euglossa viridissima. This is the first study to combine
chemical analyses of perfumes with bioassays of odour
choice. It strengthens the view that negative feedback from
collected odours modifies future chemical choice and helps
males to acquire specific perfume blends.
Keywords Chemical preference
Euglossini
Hind-tibial
bouquet
Learned avoidance
Pheromone analogue

Communicated by Richard Karban.

Electronic supplementary material The online version of this

article (doi:10.1007/s00442-013-2620-0) contains supplementary
material, which is available to authorized users.
T. Pokorny (&)
T. Eltz
Department of Animal Ecology, Evolution and Biodiversity,
Ruhr University Bochum, 44780 Bochum, Germany
e-mail: tamara.pokorny@ruhr-uni-bochum.de
T. Pokorny
M. Hannibal
Institute of Sensory Ecology, Heinrich-Heine University
Dusseldorf, 40225 Dusseldorf, Germany
J. J. G. Quezada-Euan
Departamento de Apicultura, Universidad Autonoma
de Yucatan, Merida, Yucatan 97100, Mexico
E. Hedenstrom
N. Sjoberg
J. Bang
Department of Applied Science and Design, Mid Sweden
University, 851 70 Sundsvall, Sweden

Introduction
Most insects use semiochemicals for sexual communication (see El-Sayed 2012). Predominantly, these are of
endogenous origin, while some species ingest precursory
substances from their environment, which are then metabolised to create the final pheromone (Schneider et al.
1975; Schulz et al. 2004; Honda et al. 2006). In this
respect, males of the Neotropical orchid bees (Euglossini,
Apidae, Hymenoptera) are exceptional, as they use exogenous volatiles without further modification to compose
their hind-tibial perfumes (Eltz et al. 2005a, 2007; Zimmermann et al. 2009). The various volatile compounds that
make up those perfumes are collected from a range of floral
and non-floral sources, and accumulated in specialised hind
leg pouches (Vogel 1966; Dodson et al. 1969; Dressler

123

418

1982; Williams and Whitten 1983; Whitten et al. 1993;

Roubik and Hanson 2004; Cappellari and Harter-Marques
2010). Eventually, the accumulated odours are exposed
during male courtship display (Bembe 2004; Eltz et al.
2005b), and may function as an attractant or courtship
signal to conspecific females (Eltz et al. 2003, 2005b). The
individual male perfumes are variable regarding complexity and concentration, with older males possessing the
most complex and rich bouquets (Eltz et al. 1999). Notwithstanding this intraspecific variability, the individual
differences in perfume composition are normally larger
among species than within species (Eltz et al. 1999,
2005a). For instance, in a comparative analysis covering 15
species of Euglossa in Central Panama, Zimmermann et al.
(2009) found that all species were chemically distinct,
partly by the presence of certain characteristic compounds
but also based on differing relative ratios of shared compounds. Thus, the individual compounds contribute both
quantitatively and qualitatively to the species-specificity of
fragrance blends. Given that male bees must acquire the
volatile compounds over time and from a range of unpredictable sources, they require a mechanism that compensates environmental variability in supply. Specifically, they
have to avoid excessive collection from locally/seasonally
abundant odour sources while at the same time maintaining
an attraction to rare ones. One mechanism proposed to
solve this problem is experience-dependent choice, i.e.
learned avoidance (Eltz et al. 2005a). When caged bees
were confronted with the same odour compound over four
consecutive days, they initially collected it in high rates,
but the attraction rapidly decreased. Subsequent additional
presentation of a different attractive compound, however,
elicited vigorous collecting behaviour for the new chemical. It was argued that an operant learning of odours during
collection and the consecutive avoidance of such odours
would prevent the bees from overloading their bouquet
with abundant substances (Eltz et al. 2005a). Whether this
model is applicable under natural conditions remains
unknown. Predictions arising from the learned avoidance
model include (1) geographical or seasonal differences in
the choice of odour compounds, and (2) a corresponding
reciprocal variation in the availability of these compounds.
Geographical and seasonal variation in the choice of synthetic chemicals has indeed been shown repeatedly in
baiting studies (Ackerman 1989; Abrahamczyk et al.
2012). However, it is not known whether the observed
variability in chemical choice was based on a differential
availability of volatiles in the different environments.
Naturally, the availability of odoriferous substances is
difficult to evaluate due to the large, complex environments
which the bees inhabit. Moreover, natural sources of the
respective volatiles, e.g. epiphytic euglossophilous orchids,
are very hard to find and most odour sources are not even

123

Oecologia (2013) 172:417425

known to science. Thus, indirect methods are required for

the assessment of compound availability.
In the present study, we used the ratio in which a given
chemical was present in the perfumes of males as an indicator for its availability at a given site and time. This
approach makes two critical assumptions. First, the males
that are captured at a site have actually foraged for most of
the volatiles of their perfume blend in the surrounding area.
Second, most males have incomplete perfume blends with
regard to the chemicals in question, so as to provide sufficient variability for among-site comparisons. At least the
second assumption is evidently valid, because many males
tend to have rather low quantities of perfume and lack
components that are otherwise known from the respective
species bouquets (Eltz et al. 1999, 2008). The study was
conducted on the Yucatan peninsula, Mexico, where we
focused on two orchid bee species, Euglossa dilemma
Bembe & Eltz and Euglossa viridissima Friese, that are
common and easily attracted to six synthetic chemical
compounds. E. dilemma has only recently been described
(Eltz et al. 2011), representing a tridentate (3D) sibling
species of E. viridissima (Eltz et al. 2008). It has previously
been designated as E. viridissima in Skov and Wiley (2005)
and as E. aff. viridissima in Ramrez et al. (2010). The six
chemicals are benzyl benzoate, 1,8-cineole, eugenol, methyl
salicylate, p-dimethoxybenzene and methyl cinnamate (Skov
and Wiley 2005; Eltz et al. 2008; Eltz, personal observation).
All these compounds can be found in the hind leg perfumes
of field-captured males of both species, and were therefore
chosen for our choice experiments (Fig. 1). In addition, the
perfumes of the two species are often characterised by
dominant semivolatile compounds, a different one in each
species, which we did not have at hand in sufficient quantities for bioassays. In E. dilemma, the compound in question
is 2-hydroxy-6-nona-1,3-dienyl-benzaldehyde in four stereoisomeric conformations (HNDB), which on average
comprises 67 % of the male perfume in this species (Eltz
et al. 2008). Male E. viridissima, in contrast, lack HNDB but
often possess large quantities of a fatty acid lactone derivative of linolenic acid (Hedenstrom, Sjoberg, Wallin and
Eltz, unpublished data), that still awaits complete structural
characterisation, and thus henceforth is referred to as L97
(mass spectrum of L97 is provided as Electronic Supplementary Material). Males of the correct species are
attracted to the respective semivolatile in field experiments,
indicating that the compounds are behaviourally active and
thus probable key components of male perfumes (HNDB:
Eltz et al. 2008; L97: see Materials and methods).
Therefore, HNDB and L97 were targeted by our chemical
analyses along with the six compounds used for baiting to
assess the extent of variation in local availability (Fig. 1).
We measured compound content/availability and compound choice at 16 sites grouped in two different habitat

Oecologia (2013) 172:417425

Fig. 1 Structures of the chemical compounds used for analyses of

compound presence in hind leg bouquets (asterisk) and baiting assays
(hash). Mass spectrum of L97 is provided as Electronic Supplementary Material

types (coastal and inland) in two seasons (dry and rainy).

We evaluated whether (1) there are differences in the
availability of specific volatiles between seasons, localities,
or habitats, and (2) whether bees show above-average
preferences for odour compounds that are rare in their
respective sites or habitats. In addition, we tested (3)
whether the complex bouquets of older bees reach speciesspecificity despite possibly existing ecological constraints
on compound availability.

Materials and methods

Study sites and baiting
Studies were conducted on the Yucatan peninsula, Mexico,
during the dry (March/April) and the rainy season (September/October) of 2010. Sampling took place at eight sites
along the coast and eight corresponding sites situated at
least 30 km inland. Coastal habitat is characterised by
mangroves and wetland vegetation (Trejo-Torres et al.

419

1993; Rejmankova et al. 1996), while the typical inland

vegetation consists of low stature deciduous secondary
forest which is interspersed with patches of agricultural
areas (Hartter et al. 2008). Sites were chosen on the basis of
being either known for high bee abundance or at least
featuring predominantly woody vegetation while being
located at a distance from agricultural areas, as this indicates suitable bee habitat. All sites were located along the
western and northern coastline of the peninsula (Fig. 2),
where E. dilemma and E. viridissima occur in sympatry.
Each site was sampled once per season using synthetic
benzyl benzoate (BB), 1,8-cineole (C), eugenol (E), methyl
salicylate (MS), p-dimethoxybenzene (PDMB) and methyl
cinnamate (TMC) as baiting substances. Standardised
amounts of each chemical were applied to scent-free tissue
papers, which were enclosed separately in stainless steel
mesh tea balls to prevent the bees from directly accessing
the bait chemicals. Twelve baits (two of each compound)
were exposed simultaneously at a given site for 3 h
between 0800 and 1200 hours. Placing was in random
order, with 1.52.5 m distance between adjoining baits
(baits containing the same compound were never adjoining). The baits were suspended from vegetation at a height
of 11.5 m above ground and refilled after 90 min to
compensate for evaporation. All baits were checked for
bees at regular intervals. Bees were captured with hand
nets and identified in the field. The species and choice of
compound was recorded for every captured bee, and all
individuals were marked to prevent pseudoreplication.
Additionally, the first 20 males of each species were collected in individual, clean Eppendorf vials, and kept on ice
for later perfume extraction.
Extraction and analyses
Bees were killed by freezing at -20 C, after which both
hind legs were removed and placed in 500 ll of hexane.
The resulting extracts were stored at -20 C prior to
chemical analysis in Bochum, Germany, with coupled gas
chromatography and mass spectrometry (GCMS) using
splitless injection on a DB-5 column (30 m, 0.25 lm film
thickness, 0.25 mm diameter). The oven of the HP5890 II
GC was programmed from 60 to 300 C at 10 per min.
A HP5972 MS served as detector. Aliphatic cuticular
hydrocarbons and substances originating from the bees
labial glands were excluded from analyses because they are
not of exogenous origin (Eltz et al. 2007; Ramrez et al.
2010). The total amount of exogenous substances in an
extract was determined by adding up the remaining peak
areas (integrated ion currents) of volatile substances.
Standardisation of the single peak areas by total yielded the
relative compound contribution (in %). Compound identification was accomplished either by direct comparison of

123

420

Oecologia (2013) 172:417425

viridissima approached the filter paper, landed on it, and

performed typical collection behaviour. At the same time,
males of E. dilemma were present nearby at other chemical
baits, but none were attracted to L97.

Results
The total numbers of bees attracted during a single 3-hour
sampling period were highly variable across sites and
seasons, and ranged from 0 to 318 for E. dilemma, and 0 to
100 for E. viridissima. Sites with less than 10 bees were
excluded from seasonal comparisons within sites for the
respective species. Analyses focussing on among-site
variations and habitat type, for which the data from both
seasons for each site were pooled, nevertheless included
these sites when the total number of bees at the given site
resulted in a minimum of 10.
Compound presence in hind leg bouquets
Fig. 2 Baiting sites on the Yucatan peninsula, Mexico. Sites 18 are
situated in coastal habitat, sites 916 at least 30 km inland of the
coastline

mass spectra and retention times with those of reference

samples or by using commercially available spectral
libraries (Wiley 275; Adams 2001). Where no matches
could be found, compounds were numbered and added to a
custom-made spectral library. All chromatograms were
analysed with regard to the presence or absence of the six
bait chemicals as well as HNDB and L97 (see Introduction). Additionally, two of the most highly concentrated
extracts (in terms of total perfume compound peak area as
well as in the number of different compounds) per site and
species were selected for detailed analyses of overall perfume composition. BrayCurtis similarities were calculated
between pairs of extracts based on the similarity/dissimilarity of relative compound contributions to the total perfume peak area, and the resulting similarity matrix was
ordinated in two dimensions by non-metric multidimensional scaling (MDS) with the software Primer v.6 (Clarke
1993; Clarke and Gorley 2006). Statistical analyses were
done in R v.2.12.1 (R Development Core Team 2010).
Isolation and field testing of L97
The compound L97 used in the field tests was isolated from
an n-hexane extract of hind legs of male E. viridissima by
solid phase micro-extraction. The compound was obtained
in high purity ([98 % chemical purity according to GC).
Aliquots were exposed on filter paper pinned to small trees
in the understory of the inland site Xmatkuil on March 14
in 2010, from 1015 to 1145 hours. Seven males of E.

123

A total of 502 and 421 hind leg extracts were analysed for
E. dilemma and E. viridissima, respectively. The effect of
season on the presence or absence of any of the eight focal
compounds (BB, C, E, MS, PDMB, TMC, HNDB and L97)
was relatively weak and non-significant in the majority of
within-site comparisons for either species (Fishers exact
test; E. dilemma: n.s. in 66 out of 70 tests; E. viridissima:
n.s. in 54 out of 56 tests; each following Bonferroni correction). Therefore, we pooled data from both seasons for
each site for further analyses. In both species, there were
significant differences among sites in the presence/absence
ratio of all eight focal compounds (v2 test of independence;
for each compound: n (sites) = 14, df = 13, v2 [ 30,
P \ 0.005). Interestingly, at least some of these amongsite differences were related to habitat type. Extracts of
E. dilemma from coastal sites contained BB in significantly
greater proportions than those from inland sites (coastal
locations: n = 6, inland locations: n = 8, MannWhitney
U test, U = 48, P \ 0.001; Fig. 3a). The same pattern
was observed in E. viridissima (coastal locations: n = 8,
inland locations: n = 6, MannWhitney U test, U = 48,
P \ 0.005; Fig. 3b), although BB was generally much less
common in hind leg extracts of E. viridissima than in those
of E. dilemma. In addition, the semivolatile compound L97
was significantly more common in E. viridissima extracts
from coastal sites in comparison to inland ones (U = 47.5,
P \ 0.005; Fig. 3b).
Compound choice
The individual choice for one of the six baiting compounds
was recorded for a total of 2,165 E. dilemma (970

Oecologia (2013) 172:417425

421

Fig. 3 Proportion of individual

orchid bees containing focal
substances in their hind leg
bouquets and the relative choice
for one of the six odour
compounds for a E. dilemma
and b E. viridissima. Graphs
depict median, quartiles and
data range for each habitat type.
Coastal habitat depicted by grey
boxes, inland habitat by white
ones. a E. dilemma, significant
differences found for benzyl
benzoate (BB) in compound
presence (top row) as well as
odour choice (bottom
row) between habitats.
b E. viridissima, significant
differences in compound
presence between habitats found
for the fatty acid lactone
derivative of linolenic acid
(L97) and BB (top row). Odour
choice differed between habitat
types for methyl cinnamate
(TMC, bottom row).
**P \ 0.01, ***P \ 0.001

individuals at coastal locations, 1,195 at inland locations)

and 831 E. viridissima (557 individuals at coastal locations,
274 at inland locations). As was the case for compound
presence in hind leg extracts, seasonal effects on chemical
choice were mostly non-significant in both species (Fishers exact test; E. dilemma: n.s. in 55 out of 60 tests; E.
viridissima: n.s. in 47 out of 48 tests; each following
Bonferroni correction). Therefore, the compound choice
data from both seasons were pooled for each site. Compound choice within each species varied significantly
between sites (v2 test of independence; for each compound:
n (locations) = 14, df = 13, v2 [ 30, P \ 0.005). Compound choice at coastal and at inland sites was similar with
regard to most compounds (Fig. 3), but there were interesting exceptions. E. dilemma chose BB significantly more
often at inland locations than at coastal ones (coastal
locations: n = 6, inland locations: n = 8, MannWhitney

U test, U = 1, P \ 0.01), and tended to choose E more

often in coastal habitat (MannWhitney U test, U = 38,
P = 0.08; Fig. 3a). E. viridissima showed a significantly
higher proportional choice for TMC at inland locations
(coastal locations: n = 8, inland locations: n = 6, Mann
Whitney U test, U = 4, P \ 0.01; Fig. 3b). For both species, PDMB was clearly the most attractive substance,
followed by E and TMC. Bees were rarely attracted to MS,
and, contrary to the attractive effect of BB on E. dilemma,
only one individual of E. viridissima ever approached a
bait loaded with this substance (Fig. 3).
Correlation of compound presence and choice
To test whether odour choice is influenced by the respective odours availability in the bees habitat, we correlated
the relative choice for each chemical with the proportion of

123

422

bees containing the respective substance in their hind leg

bouquet across sites. Two sites per species had to be
excluded from the analysis based on low bee numbers
(\10). In E. dilemma, there was a significant negative
correlation between the proportion of individuals containing BB and the proportional choice of BB (Spearman rank
correlation, n = 14, Rs = -0.8720, P \ 0.001; Fig. 4a).
Markedly, bees at inland sites, where BB appeared to be
particularly rare, chose this compound in significantly
higher proportions than those observed at coastal sites (see
above; Fig. 3a). Similarly, a negative correlation between
presence and choice was found for E in E. dilemma
(Rs = -0.7767, P \ 0.005; Fig. 4b), but not in the four
other bait chemicals. No significant correlation for any
compound was found for E. viridissima.
Compound accumulation and species average
The learned avoidance model for the acquisition of specific, complex blends of chemicals predicts that the proportion of a given chemical in individual blends is highly
variable in young bees and consecutively approaches a
certain target level in older, more perfume-rich males. To
determine whether this prediction holds across our sample
of males, we plotted relative compound contribution (% of
total peak area) as a function of the overall amount of
volatiles in individual extracts. We did this for the two

Oecologia (2013) 172:417425

most common compounds in each species, HNDB and BB

for E. dilemma, and L97 and E for E. viridissima. For each
of the depicted compounds, the relative contribution converged on the species median with increasing total amount
of volatiles (Fig. 5).
Species-specific perfume composition
For each species and site, two highly concentrated extracts
of complex composition were analysed in detail. Perfume
blends contained between 19 and 58 compounds for E.
dilemma, and between 8 and 56 compounds for E. viridissima. The species clusters resulting from MDS were
distinct and clearly separated from each other, with a stress
value below 0.15, indicating a good fit of the two-dimensional plot to the underlying pairwise dissimilarities
(Fig. 6a). While intraspecific variation within E. dilemma
was rather low, indicated by the tight arrangement of the
cluster, extracts of E. viridissima were scattered along the
vertical MDS axis. As visualised by the bubble overlay
(Fig. 6b), this variation was mainly caused by large differences in the relative amounts of compound L97, which
was present in many hind leg extracts from coastal sites,
while lacking in most extracts from inland sites (difference
between habitats: ANOSIM, Global R = 0.123, P \ 0.05;
L97 contributed 24.92 % (largest value for any compound)
to the dissimilarity between habitat groups, as calculated
by the PRIMER SIMPER algorithm).

Discussion

Fig. 4 Relationship between the proportion of male E. dilemma

containing a particular substance in their hind leg bouquets and
observed relative choice of the corresponding chemical bait at coastal
(dots) and inland locations (empty diamonds). a Benzyl benzoate
(BB), Spearman rank correlation Rs = -0.8720, P \ 0.001, b eugenol
(E), Rs = -0.7767, P \ 0.005

123

The results of this study emphasise that environmental

factors influence the availability of odour sources which
male orchid bees depend on for the acquisition of their
perfume. Locality and habitat seemed to affect odour
availability more than season in the case of E. dilemma and
E. viridissima on the Yucatan peninsula, despite pronounced seasonality of precipitation and flowering (Flores
1990). Low seasonal differences in availability of the eight
focal compounds could be explained by a predominance of
vegetative or microbial odour sources in the study species,
which are likely less seasonal than floral sources (Whitten
et al. 1993), by a temporally spaced phenology of alternative sources for the same substance (see Ramrez et al.
2010), or by the long adult life span of males (Roubik and
Hanson 2004), which probably compensates for some of
the seasonal shifts in odour source availability. The effect
of habitat type, coastal or inland, on tibial odour content is
particularly interesting, because it may explain differences
in habitat use of the two sibling species. E. viridissima
tends to be proportionately more abundant at coastal sites
than at inland sites (Eltz, personal observation), an

Oecologia (2013) 172:417425

423

Fig. 5 Relative contribution of

major chemical compounds to
the blend in relation to total
amount of perfume for each
species. Individual samples
from coastal sites represented
by dots, those of inland sites by
empty diamonds. The horizontal
line depicts the median
contribution. a E. dilemma,
2-hydroxy-6-nona-1,3-dienylbenzaldehyde (HNDB),
b E. viridissima, fatty acid
lactone derivative of linolenic
acid (L97), c E. dilemma, benzyl
benzoate (BB), d E. viridissima,
eugenol (E)
Fig. 6 a Two-dimensional
MDS representation of chemical
similarity between individual
samples of hind leg perfumes
based on standardised Bray
Curtis similarities, b The same
plot with the relative
contribution of the fatty acid
lactone derivative of linolenic
acid, L97, to perfumes overlaid
as grey bubbles of variable size.
Larger circles signify a higher
relative contribution of L97 to
the respective bouquet

observation that had originally inspired us to select habitat

type as one of the possible explanatory variables in our
analyses. This trend was confirmed by the present baiting
study, although the difference was not significant (median
proportion of E. viridissima bees at coastal sites 46 vs.
14 % at inland sites; MannWhitney U test, U = 50,
P = 0.065). If there is indeed a preference for the coastal
habitat, it could be due to a differential occurrence of odour
sources, in particular, the source of L97, which was much
more common in E. viridissima extracts from coastal sites.
The source of L97, as indeed most euglossine odour
sources, is unknown, and it is therefore difficult to independently test this notion. The vegetation of coastal areas
on the Yucatan peninsula includes mangroves and wetlands
(Trejo-Torres et al. 1993; Rejmankova et al. 1996),
whereas inland habitats are dominated by low stature
deciduous secondary forest interspersed with patches of
agricultural areas (Hartter et al. 2008). Thus, the composition of potential odour sources, i.e. plants or microbes, is
conceivably different around coastal and inland sites.
Generally, the importance of microbes as euglossine odour
sources should not be underestimated (see Eltz et al. 1999).
Here, habitat-related differences in the presence of microhabitats, e.g. in association with springs, streams and
canals (Wilson and Williams 1987), could lead to a

differential distribution or growth of certain microbial

sources, e.g. the fungi present in decaying wood and other
plant matter.
Previous baiting studies have demonstrated shifts in
odour choices within species between localities or seasons.
It was argued that a change of choice might be based on
spatial or temporal variation in the presence of odour
sources (Ackerman 1989; Abrahamczyk et al. 2012), but
this had not been tested. The present study is the first to
relate a measure of odour availability (compound presence
in male hind tibia) to compound choice. Compound choice
was negatively correlated with the respective availability in
two cases in E. dilemma. Males preferred to collect BB and
E in places where these compounds rarely appeared in hind
tibiae, which we infer to also mean rare in the environment.
Furthermore, for BB, this relationship was related to habitat, with combinations of lowest availability/highest relative choice occurring at inland locations. Why were similar
effects not detected in other compounds, and no effects at
all in E. viridissima? Partly, the lack of availability/choice
correlations may be due to high natural variability of the
measured variables in combination with insufficient sampling, i.e. more spatial replicates (sites) and repeated bioassays may have yielded significant correlations for
additional compounds. Also, it is possible that correlations

123

424

are easier to detect for major perfume compounds such as

HNDB and L97, which were unfortunately not available in
sufficient amounts for bioassays. Nevertheless, the negative
correlation between odour availability and odour choice
observed in two compounds confirms that negative feedback from collected odours can modify odour choice, and
is thus in agreement with the learned avoidance model
(Eltz et al. 2005a). However, learned avoidance is not the
only possible mechanism. For example, male bees could
use their sense of smell to directly obtain information on
the contents of their hind legs, independent of experience
gained during the process of collection. However, this was
not the case in Panamanian Euglossa imperialis Cockerell.
Caged males that had one of two attractive chemicals
injected directly into their hind tibial pouches did not
change their choice of odours, suggesting that males do not
evaluate their loads directly. Instead, they only changed
their choice behaviour when they had actively collected the
respective compound, i.e. when they had learned to avoid
it (Eltz et al. 2005a). Thus, unless Mexican species are
different in this respect, learned avoidance, i.e. the negative
interaction between experience and an innate template of
odour preferences, is the most likely mechanism to account
for the observed variations in odour choice in wild E.
dilemma and E. viridissima. The underlying neural mechanisms of learned odour avoidance are unknown (but see
the discussion in Eltz et al. 2005a), and it is unclear how
the avoidance of a certain odour will be modified by consecutive collecting experience. It is assumed that male
orchid bees continue to forage for chemicals for most of
their lives, adding substances to their bouquet, and over
time increasing its quantity and complexity (Eltz et al.
1999). In the simplest case, memory capacity may limit the
number and permanence of learned associations, which
might lead to the progressive loss of odour-specific
avoidance and result in a dynamic cycling of odour preferences (Eltz et al. 2005a). In fact, dynamic avoidance
could be responsible for the fine-tuning of quantitative
proportions of components in accumulative perfume loads,
which was observed when we related the relative compound contribution to the total amount of perfume across
individuals. The relative proportion of an odour compound
was nearest to the species median in the quantitatively
richest blends, while it was highly variable for those with
low total perfume quantities.
The detailed analyses of the richest and most complex
bouquets resulted in clearly species-specific clusters after
MDS, which further indicates that males reach a speciesspecific blend eventually, though they by no means all
become identical. Individual Euglossa perfume bouquets
vary to some extent within populations (Eltz et al. 1999;
Zimmermann et al. 2009) and among geographic regions,
as has already been shown for E. dilemma (Ramrez et al.

123

Oecologia (2013) 172:417425

2010; the species was at that point still designated as E. aff.

viridissima). On the smaller geographic scale of the present
study, there was a remaining variation according to habitat
type, most conspicuously so in E. viridissima. Here, the
intraspecific variation seemed to be mainly due to differing
proportions of the major semivolatile, L97, which was
found predominantly in bouquets from coastal regions (see
above). It seems likely that this is due to a differential
availability of the major compound of the species and not
to subpopulations with differential preferences for L97.
Individual bees of E. viridissima could be attracted to L97
purified from extracts at an inland location. In addition, the
study of Zimmermann et al. (2011) suggests that population sizes of orchid bees, especially across the rather lowstructured landscape of the Yucatan peninsula, are fairly
large and panmictic, as the genetic differentiation of E.
dilemma in this area is surprisingly low with an overall
high genetic variability within locations. In the present
study, blends of E. dilemma were found to be rather similar
in the two habitats, also with regard to the presence of its
major semivolatile, HNDB. However, in Florida, where a
stable population has established itself despite the lack of
associated orchids (Skov and Wiley 2005; Pemberton and
Wheeler 2006; and see Ramrez et al. 2011), HNDB does
not seem to be available in large amounts (judged from
tibial contents) and the bees have evidently taken to collecting a new compound (the herbicide-derived triclopyr
2-butoxyethyl ester; Ramrez et al. 2010). Whether the
substance is perceived as similar to HNDB, i.e. represents
an actual replacement with regard to signal function, is not
known.
All this variability in perfumes suggests that, while
learned avoidance is successful in compensating for some
of the environmental variation, it cannot completely
equalise perfume bouquets across all possible odour
markets. The resulting geographical and habitat clines in
perfume variability may eventually result in local adaptations with regard to mate recognition and thus promote
population divergence and speciation. Generally, our study
emphasises the recognition function of male perfumes that
is believed to be the basis of the relative specificity in
chemical composition (Zimmermann et al. 2009). However, euglossine perfumes may also be ornamental in nature, with females favouring certain variants that deviate
from the population mean. This is an intriguing aspect, but
at present it is completely unknown to what extent sexual
selection is involved in shaping within-species perfume
variability in male orchid bees.
Acknowledgments Jorge Ramirez-Pech helped with sample collection. Funding was provided by the PROALMEX program (120989)
of the German Academic Exchange Service and the Consejo Nacional
de Ciencia y Tecnologa (103341) to T.E. and J.Q.E., and the German
Science Foundation (EL 249/6). Santiago Ramrez, Florian Leese, the

Oecologia (2013) 172:417425

Evolutionary Ecology journal club group (Bochum), and two anonymous reviewers helped to improve the manuscript. E.H., J.B. and
N.S. would like to thank the European Union European Regional
Development Fund and the County Administrative Board of Vasternorrland for financing. All experiments comply with the current laws
of Mexico and were performed with the necessary permits.
Conflict of interest The authors declare that they have no conflicts
of interest.

References
Abrahamczyk S, Gottleuber P, Kessler M (2012) Seasonal changes in
odour preferences by male euglossine bees (Hymenoptera:
Apidae) and their ecological implications. Apidologie 43:212
217
Ackerman JD (1989) Geographic and seasonal variation in fragrance
choices and preferences of male euglossine bees. Biotropica
21(4):340347
Adams RP (2001) Identification of essential oil components by gas
chromatography/quadrupole mass spectroscopy. Allured, Carol
Stream
Bembe B (2004) Functional morphology in male euglossine bees and
their ability to spray fragrances (Hymenoptera, Apidae, Euglossini). Apidologie 35:283291
Cappellari SC, Harter-Marques B (2010) First report of scent
collection by male orchid bees (Hymenoptera: Apidae: Euglossini) from terrestrial mushrooms. J Kans Entomol Soc 83(3):
264266
Clarke KR (1993) Non-parametric multivatiate analyses of changes in
community structure. Aust J Ecol 18:117143
Clarke KR, Gorley RN (2006) PRIMER v6: User Manual/Tutorial.
PRIMER-E, Plymouth
Dodson CH, Dressler RL, Hills HG, Adams RM, Williams NH (1969)
Biologically active compounds in orchid fragrances. Science
164:12431249
Dressler RL (1982) Biology of the orchid bees (Euglossini). Annu
Rev Ecol Syst 13:373394
El-Sayed AM (2012) The pherobase: database of pheromones and
semiochemicals. http://www.pherobase.com. Accessed 1 October 2012
Eltz T, Whitten MW, Roubik DW, Linsenmair KE (1999) Fragrance
collection, storage, and accumulation by individual male orchid
bees. J Chem Ecol 25(1):157176
Eltz T, Roubik DW, Whitten MW (2003) Fragrances, male display
and mating behaviour of Euglossa hemichlora: a flight cage
experiment. Physiol Entomol 28:251260
Eltz T, Roubik DW, Lunau K (2005a) Experience-dependent choices
ensure species-specific fragrance accumulation in male orchid
bees. Behav Ecol Sociobiol 59:149156
Eltz T, Sager A, Lunau K (2005b) Juggling with volatiles: exposure
of perfumes by displaying male orchid bees. J Comp Physiol A
191:575581
Eltz T, Zimmermann Y, Haftmann J, Twele R, Francke W, QuezadaEuan JJG, Lunau K (2007) Enfleurage, lipid recycling and the
origin of perfume collection in orchid bees. Proc R Soc Lond B
274:28432848
Eltz T, Zimmermann Y, Pfeiffer C, Ramirez Pech J, Twele R,
Francke W, Quezada-Euan JJG, Lunau K (2008) An olfactory
shift is associated with male perfume differentiation and species
divergence in orchid bees. Curr Biol 18:18441848
Eltz T, Fritzsch F, Ramrez Pech J, Zimmermann Y, Ramrez SR,
Quezada-Euan JJG, Bembe B (2011) Characterization of the
orchid bee Euglossa viridissima (Apidae: Euglossini) and a new

425
cryptic sibling species, by morphological, chemical, and genetic
characters. Zool J Linn Soc 163:10641076
Flores JS (1990) The flowering periods of Leguminosae in the
Yucatan peninsula in relation to honey flows. J Apic Res
29:8288
Hartter J, Lucas C, Gaughan AE, Aranda LL (2008) Detecting
tropical dry forest succession in a shifting cultivation mosaic of
the Yucatan Peninsula, Mexico. Appl Geogr 28:134149
mura H (2006) Major components in the
Honda Y, Honda K, O
hairpencil secretion of a butterfly, Euploea mulciber (Lepidoptera, Danaidae): their origins and male behavioral responses to
pyrrolizidine alkaloids. J Insect Physiol 52:10431053
Pemberton RW, Wheeler GS (2006) Orchid bees dont need orchids:
evidence from the naturalization of an orchid bee in Florida.
Ecology 87(8):19952001
Ramrez SR, Eltz T, Fritzsch F, Pemberton R, Pringle EG, Tsutsui ND
(2010) Intraspecific geographic variation of fragrances acquired
by orchid bees in native and introduced populations. J Chem
Ecol 36(8):873884
Ramrez SR, Eltz T, Fujiwara MK, Gerlach G, Goldman-Huertas B,
Tsutsui ND, Pierce NE (2011) Asynchronous diversification in a
specialized plantpollinator mutualism. Science 333:742746
Rejmankova E, Pope KO, Post R, Maltby E (1996) Herbaceous
wetlands of the Yucatan peninsula: communities at extreme ends
of environmental gradients. Int Rev Gesamten Hydrobiol
Hydrogr 81(2):223252
R Development Core Team (2010) R: A language and environment
for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. http://www.Rproject.org. Accessed 6 September 2012
Roubik DW, Hanson PE (2004) Orchid bees of tropical America.
Instituto Nacional de Biodiversidad (INBio), Heredia, Costa
Rica
Schneider D, Boppre M, Schneider H, Thompson WR, Boriack CJ,
Petty RL, Meinwald J (1975) A pheromone precursor and its uptake
in male Danaus butterflies. J Comp Physiol A 97(3):245256
Schulz S, Beccaloni G, Brown KS Jr, Boppre M, Freitas AVL,
Ockenfels P, Trigo JR (2004) Semiochemicals derived from
pyrrolizidine alkaloids in male ithomiine butterflies (Lepidoptera: Nymphalidae: Ithomiinae). Biochem Syst Ecol 32:699713
Skov C, Wiley J (2005) Establishment of the neotropical orchid bee
Euglossa viridissima (Hymenoptera: Apidae) in Florida. Fla
Entomol 88(2):225227
Trejo-Torres JC, Duran R, Olmsted I (1993) Manglares de la
pennsula de Yucatan. In: Salazar-Vallejo S, Gonzalez NE (eds)
Biodiversidad marina y costera de Mexico. Comision Nacional
para el Conocimiento y Aprovechamiento de la Biodiversidad,
Centro de Investigaciones de Quintana Roo, pp 660672
Vogel S (1966) Parfumsammelnde Bienen als Bestauber von
sterr bot Z 113:302361
Orchidaceen und Gloxinia. O
Whitten MW, Young AM, Stern DL (1993) Nonfloral sources of
chemicals that attract male euglossine bees (Apidae: Euglossini).
J Chem Ecol 19(12):30173027
Williams NH, Whitten MW (1983) Orchid floral fragrances and male
euglossine bees: methods and advances in the last sesquidecade.
Biol Bull 164:355395
Wilson EM, Williams A (1987) A coastal ecosystem in northwestern
Yucatan. sites.maxwell.syr.ecu/clag/yearbook1987/wilson.htm.
Accessed 12 October 2012
Zimmermann Y, Ramrez SR, Eltz T (2009) Chemical niche
differentiation among sympatric species of orchid bees. Ecology
90:29943008
Zimmermann Y, Schorkopf DL, Moritz RFA, Pemberton RW,
Quezada-Euan JJG, Eltz T (2011) Population genetic structure
of orchid bees (Euglossini) in anthropogenically altered landscapes. Conserv Genet 12:11831194

123