Pollination in Epidendrum densiflorum Hook. (Orchidaceae: Laeliinae): Fraudulent Trap-Flowers, Self-Incompatibility, and a Possible New Type of Mimicry

The pollination and the breeding system of Epidendrum densiflorum (Orchidaceae: Laeliinae) were studied through fieldwork and controlled pollinations in cultivated plants. Pollination is exclusively promoted by males of diurnal Lepidoptera: five species of Arctiinae and four of Ithomiinae were recorded as pollinators. These male insects are known to obtain alkaloids (through the nectar) in flowers of Asteraceae and Boraginaceae. However, the flowers of E. densiflorum are nectarless, despite presenting a cuniculus (a likely nectariferous cavity). Pollinators insert their proboscides into the flowers and remove or deposit the pollinaria while searching for nectar. The floral tube is very narrow, and insects struggle for up to 75 min to get rid of the flowers. Plants are pollinator-dependent and nearly fully self-incompatible. Pollinarium removal, pollination, and fruiting success (2.85%) were very low; facts that are consistent with the patterns globally observed in deceptive (rewardless) orchids. Nilsson’s male efficiency factor (0.245) was also low, indicating pollen loss in the system. Based on our field observations, we suggest that the fragrance of E. densiflorum likely mimics these plants that are normally used as a source of alkaloids by male Lepidoptera, a hypothesis that we intend to test in the future.


Introduction
Epidendrum L. (Orchidaceae: Laeliinae) ranks among the biggest Neotropical orchid genera, embracing about 1500 described species and occurring from the Southern United States to Northern Argentina [1]. Phylogenetic studies suggest that Epidendrum is monophyletic [1]. The genus can be easily diagnosed by its characteristic flowers, with the labellum (median petal) fused to the lateral sides of the column, forming a narrow tube [1,2]. As a result, the flower cavity is tubulose and long. In addition, as in most other orchid genera of subtribe Laeliinae, Epidendrum flowers present a well-developed cuniculus (a putative nectariferous cavity) below the column and parallel to the ovary [1,2]. However, in most studied Epidendrum species, this cuniculus is not secretory [1]. Darwin [3] was the first researcher to speculate that Epidendrum flowers are morphologically suitable for pollination by Lepidoptera. Darwin [3] manipulated flowers and correctly speculated that the pollinarium of these flowers should glue onto the surface of the proboscides of their pollinators. Hágsater and Soto Arenas [1] (pp. 247-250) and Van der pijl and Dodson [4] (p. 185) list and review several brief observations of floral visitors in several species. However, these reports are essentially anecdotal and mostly consist of observations of insects at flowers with no evidence that these animals (bees, moths, flies, and butterflies) remove and deposit

Breeding System and Fruiting Success under Natural Conditions
The results of breeding system experiments are summarized in Table 1. No fruits were obtained through intact or emasculated flowers, clearly indicating that E. densiflorum is pollinator-dependent and, thus, unable to set fruit in absence of pollinators. As the data analyzed between treatments did not show the normal distribution in fruit production, the Kruskal-Wallis test was used (x 2 (1) = 87.51, df = 3, p-value = p < 0.001), which showed that there are significant differences between treatments and fruit generation: crosspollination (48 fruits, mean of 0.96 ± 0.19 per plant), self-pollination (1 fruit, mean of 0.02 ± 0.14 per plant), and intact flowers and emasculation, showing no fruit formation.
Under natural conditions (Table 2), only 28 fruits (over 969 flowers) were formed during the observation period. This represents ca. 2.88% of natural fructification. As a whole, during the observation period, 158 flowers acted as pollen donors and 41 acted as pollen receivers. Nilsson's male efficiency factor scored 0.259 (Table 2), indicating pollen loss in the system. This value indicates that 0.259 flowers were pollinated by pollinarium removal.

Breeding System and Fruiting Success under Natural Conditions
The results of breeding system experiments are summarized in Table 1. No fruits were obtained through intact or emasculated flowers, clearly indicating that E. densiflorum is pollinator-dependent and, thus, unable to set fruit in absence of pollinators. As the data analyzed between treatments did not show the normal distribution in fruit production, the Kruskal-Wallis test was used (x 2 (1) = 87.51, df = 3, p-value = p < 0.001), which showed that there are significant differences between treatments and fruit generation: cross-pollination (48 fruits, mean of 0.96 ± 0.19 per plant), self-pollination (1 fruit, mean of 0.02 ± 0.14 per plant), and intact flowers and emasculation, showing no fruit formation. Under natural conditions (Table 2), only 28 fruits (over 969 flowers) were formed during the observation period. This represents ca. 2.88% of natural fructification. As a whole, during the observation period, 158 flowers acted as pollen donors and 41 acted as pollen receivers. Nilsson's male efficiency factor scored 0.259 (Table 2), indicating pollen loss in the system. This value indicates that 0.259 flowers were pollinated by pollinarium removal.     A total of 64 pollinator visits were recorded (Table 3). All recorded pollinators were males of diurnal Lepidoptera of subfamilies Arctiinae (Erebidae) and Ithomiinae (Nymphalidae) ( Table 3 and Figure 3). Arctiinae moths were responsible for 81.25% of the observed pollination events (52 visits and 84.7% of the observed pollinarium removals). We recorded five species of tiger moths pollinating E. densiflorum: Phoenicoprocta teda Walker (26.56% of the visits and 31% of pollinarium removals), Antichloris eriphia Fabricius (20.31% of visits and 20% of pollinarium removals), Philoros rubriceps Walker (17.18% of visits and 17% of pollinarium removals), Calodesma collaris Drury (14% of visits and 15% of pollinarium removals), and Cyanopepla jucunda Walker com (3.12% of visits and 2% of pollinarium removals). Among the Ithominae, we recorded four species: Methona themisto Hübner (7.81% of visits and 7% of pollinarium removals), Episcada hymenaea Prittwitz (6.25% of visits and 4% of pollinarium removals), Hypothyris euclea Doubleday (3.12% of visits and 2% of pollinarium removals), and Placidina euryanassa C. & R. Felder (1.5% of visits and 2% of pollinarium removals). A total of 64 pollinator visits were recorded (Table 3). All recorded pollinators were males of diurnal Lepidoptera of subfamilies Arctiinae (Erebidae) and Ithomiinae (Nymphalidae) ( Table 3 and Figure 3). Arctiinae moths were responsible for 81.25% of the observed pollination events (52 visits and 84.7% of the observed pollinarium removals). We recorded five species of tiger moths pollinating E. densiflorum: Phoenicoprocta teda Walker (26.56% of the visits and 31% of pollinarium removals), Antichloris eriphia Fabricius (20.31% of visits and 20% of pollinarium removals), Philoros rubriceps Walker (17.18% of visits and 17% of pollinarium removals), Calodesma collaris Drury (14% of visits and 15% of pollinarium removals), and Cyanopepla jucunda Walker com (3.12% of visits and 2% of pollinarium removals). Among the Ithominae, we recorded four species: Methona themisto Hübner (7.81% of visits and 7% of pollinarium removals), Episcada hymenaea Prittwitz (6.25% of visits and 4% of pollinarium removals), Hypothyris euclea Doubleday (3.12% of visits and 2% of pollinarium removals), and Placidina euryanassa C. & R. Felder (1.5% of visits and 2% of pollinarium removals). Table 3. Number of visits, sex of the specimen, number of pollinaria removed, and average time (measured in minutes) of permanence in the flowers, followed by SD = standard deviation and SE = standard error of each species of lepidopteran pollinator of Epidendrum densiflorum Hook. The mean values followed by the distinct letter (a-b) indicate a significant difference (p < 0.05) according to the paired Dunn test.   The pollination mechanism is the same, irrespective of the insect species involved ( Figure 4 and Supplementary Material video S1): moths and butterflies insert the proboscis inside the floral tube looking for nectar. After a variable period (see below, Table  3) of struggling, these Lepidoptera withdraw the proboscis carrying the pollinarium adhered on its surface ( Figure 4). The pad-like viscidium of the pollinarium is responsible for its adherence to the proboscis. The pollinarium is removed with the anther cap, which falls after a few seconds ( Figure 4). Pollination takes place when a pollinarium-carrying insect visits another flower and the pollinia are arrested at the flower's hollow stigmatic The pollination mechanism is the same, irrespective of the insect species involved (Figure 4 and Supplementary Material Video S1): moths and butterflies insert the proboscis inside the floral tube looking for nectar. After a variable period (see below, Table 3) of struggling, these Lepidoptera withdraw the proboscis carrying the pollinarium adhered on its surface ( Figure 4). The pad-like viscidium of the pollinarium is responsible for its adherence to the proboscis. The pollinarium is removed with the anther cap, which falls after a few seconds (Figure 4). Pollination takes place when a pollinarium-carrying insect visits another flower and the pollinia are arrested at the flower's hollow stigmatic surface. Insects removing pollinaria from fresh flowers spend (depending on the species) 4-73 min struggling to get rid of the flowers. The mean time of permanence in the flowers is shown in Table 3. The results showed significant differences between the mean time of the visita- tion of the species of the Arctinae and Ithomiinae subfamilies, (Kruskal-Wallis X2 = 31.8, df = 8, p-value < 0.001), however, they did not present significant differences when comparing the values of the mean time of visitation of individuals of the same subfamily. Itomiinae visits are significantly faster, with an overall average visitation of (7.5 min), compared to 55.8 min of Arctiinae. During our observations, we only recorded Arctiinae moths of Antichloris eriphia and Phoenicoprocta teda laden with pollinaria, depositing them at the stigmatic cavities of flowers. surface. Insects removing pollinaria from fresh flowers spend (depending on the species) 4-73 min struggling to get rid of the flowers. The mean time of permanence in the flowers is shown in Table 3. The results showed significant differences between the mean time of the visitation of the species of the Arctinae and Ithomiinae subfamilies, (Kruskal-Wallis X2 = 31.8, df = 8, p-value < 0.001), however, they did not present significant differences when comparing the values of the mean time of visitation of individuals of the same subfamily. Itomiinae visits are significantly faster, with an overall average visitation of (7.5 min), compared to 55.8 min of Arctiinae. During our observations, we only recorded Arctiinae moths of Antichloris eriphia and Phoenicoprocta teda laden with pollinaria, depositing them at the stigmatic cavities of flowers.

Flower Features
Overall, recorded flower features are in agreement with those already mentioned in the literature for other Epidendrum species [6][7][8] or Laeliinae orchids as well [2]. As in most Laeliinae orchids, the flowers of Epidendrum densiflorum present a well-developed cuniculus [2], a structure that has been interpreted as a nectariferous cavity. However, researchers have often reported that, despite the presence of the cuniculus, no free nectar is found in Laeliinae flowers [1,13,14] and, in agreement with the absence of nectar, pollination events are rare [6,13,14], as seen below. Based on anatomical features alone, Cardoso-Gustavson et al. [15] have challenged this idea and have proposed that many Epidendrum species (including E. densiflorum) are nectar-secreting. Conversely, during our observations, we did not find any free nectar in the flowers of E. densiflorum and, in agreement with this, we recorded a low visitation rate, low rates of pollinarium removal and deposition, and a low fruiting success (see below), which are all consistent features among deceptive orchids [16,17]. When compared to rewardless orchids, rewarding ones almost score double fruiting success [16]. Pansarin [7] also did not find free nectar at the cuniculus of the closely-related E. paniculatum. Some researchers have speculated [4] that the cells within the cuniculus may be thin-walled and may easily break, exposing a liquid content when contacted by the mouthparts of the pollinators. However, this has not been proven yet, and the values of reproductive success (see below) in Epidendrum species studied so far are consistent with those of nectarless/deceptive orchid species [16,17]. We

Flower Features
Overall, recorded flower features are in agreement with those already mentioned in the literature for other Epidendrum species [6][7][8] or Laeliinae orchids as well [2]. As in most Laeliinae orchids, the flowers of Epidendrum densiflorum present a well-developed cuniculus [2], a structure that has been interpreted as a nectariferous cavity. However, researchers have often reported that, despite the presence of the cuniculus, no free nectar is found in Laeliinae flowers [1,13,14] and, in agreement with the absence of nectar, pollination events are rare [6,13,14], as seen below. Based on anatomical features alone, Cardoso-Gustavson et al. [15] have challenged this idea and have proposed that many Epidendrum species (including E. densiflorum) are nectar-secreting. Conversely, during our observations, we did not find any free nectar in the flowers of E. densiflorum and, in agreement with this, we recorded a low visitation rate, low rates of pollinarium removal and deposition, and a low fruiting success (see below), which are all consistent features among deceptive orchids [16,17]. When compared to rewardless orchids, rewarding ones almost score double fruiting success [16]. Pansarin [7] also did not find free nectar at the cuniculus of the closely-related E. paniculatum. Some researchers have speculated [4] that the cells within the cuniculus may be thin-walled and may easily break, exposing a liquid content when contacted by the mouthparts of the pollinators. However, this has not been proven yet, and the values of reproductive success (see below) in Epidendrum species studied so far are consistent with those of nectarless/deceptive orchid species [16,17]. We would like to stress that, whereas Cardoso-Gustavson et al. [15] made their proposal of nectar secretion based on detailed anatomical studies, they did not give any details of nectar features, such as volume and concentration, that are frequently mentioned in the literature [18,19]. Orchids pollinated by Lepidoptera normally present a nectar column inside their nectar spurs or nectariferous cavities, and the properties of nectar (volume, concentration) are quantifiable [18].

Breeding System and Fruiting Success under Natural Conditions
Most Laeliinae orchids are self-compatible [14,[20][21][22][23][24], being able to set fruit following self-pollination. However, E. secundum [8], E. fulgens [6], and E. tridactylum [9] are selfcompatible, but also pollinator-dependent, being unable to set fruit in absence of pollinators. In our study, flowers of E. densiflorum also behaved as pollinator-dependent and were almost completely self-incompatible, aborting most (98%) of the hand self-pollinations. Conversely, these plants had a high fruit development through cross-pollination (95%). Very similar results were found by Pansarin [7] while studying the breeding system of the closely-related E. paniculatum. Self-incompatibility was also found in the nectar-secreting E. avicula [10]. The latter species is very morphologically different from E. paniculatum and E. densiflorum, a fact that suggests that self-incompatibility may have evolved more than once in the genus. It is important to point out, however, that, even in self-compatible Laeliinae orchids, the number of viable seeds obtained from cross-pollination can be significantly higher when compared with the results of self-pollination [14,22,23].
Under natural conditions, pollinarium removal, deposition, and fructification were low. These results concur with those obtained by Pansarin [7] for E. paniculatum. Nilsson's male efficiency factor was also low (0.25). Overall, this value indicates that one in four dislodged pollinaria reached a stigmatic cavity. The low fruiting success (less than 3%) is, in our opinion, explained by the following factors: (1) low pollinarium removal and deposition and (2) self-incompatibility. By comparing male (16.3% of available flowers) and female functions (4.2%), it is possible to notice that roughly one-fourth of the pollinated flowers turned into fruits. This, in our opinion, can be partially explained by selfincompatibility: possibly, some pollinarium-laden butterflies returned to the plants from where they dislodged the pollinaria and promoted some insect-mediated self-pollinations. It is well known that males of Ithomiinae butterflies are territorial [25], thus, it seems possible that such behavior (in addition to self-incompatibility) could prompt insect-mediated self-pollinations and, ultimately, abortions. Analog scenarios have already been proposed for other Epidendroids orchids with self-incompatibility and low natural fruit sets [26,27].
Since we did not find free nectar at the flower's cuniculus, we suggested that E. densiflorum is a deceptive orchid. The observed low fruiting success is in full agreement with this proposal. Fruiting success in Orchidaceae has been reviewed by Tremblay et al. [16] and Neiland and Wilcock [17]. As a whole, fruiting success in Orchidaceae roughly surpasses 17%, but is especially low in deceptive (rewardless) orchids, ranging from less than 1 to 7% [16,17]. Moreover, the obtained value for Nilsson's male efficiency factor indicates that only a fraction of the available flowers participated in reproduction and that there was pollen loss in the system.

Pollinators and Pollinator Behavior
This contribution confirms the importance of Lepidoptera as Epidendrum pollinators [4,7,8,10]. The characteristic flower structure, with the lateral sides of the column fused to the labellum, makes Epidendrum flowers particularly suitable for Lepidoptera, since the narrow floral tube only allows the entrance of their proboscises [2,4]. During this study, we found males of Arctiinae and Ithomiinae as pollinators, and this is in full agreement with preceding reports on closely-related species, such as Epidendrum floribundum and E. paniculatum [7,28,29], which also found pollinators of the same taxonomic groups. Despite being taxonomically distant, males of Ithomiinae and Arctiinae share an important ecological feature: males of both groups actively collect pyrrolizidine alkaloids that are used during mating [30]. Alkaloids are acquired through floral nectar and, according to Pliske [30], the main sources of these alkaloids involve species of the genera Heliotropium, Tournefortia, and Myosotis (Boraginaceae), as well as Eupatorium (Asteraceae). Early observations of Arctiinae and Ithomiinae in species related to E. densiflorum, such as E. floribundum and E. paniculatum [28], led researchers to propose that the flowers of these orchids should also be a source of pyrrolizidine alkaloids. However, alkaloids were not found in the flowers of these species [7]. According to our observations, flowers of E. densiflorum are devoid of free nectar, as the related E. paniculatum [7]. Since alkaloids are acquired through nectar, and the latter is absent in the flowers of E. densiflorum, there is no way for the male Lepidoptera to obtain such resources in the flowers of the orchid under study. Flowers used for breeding system experiments were enclosed with white tule (see Methods). During the whole flowering season of 2021 and 2022, the enclosed flowers were regularly visited by Arctiinae and Ithomiinae males ( Figure 5). This behavior is similar to that described by De Vries and Stiles [28], which noticed males of Arctiinae and Ithomiinae attempting to reach enclosed inflorescences of E. paniculatum. This highlights the importance of fragrance features (chemical compounds) in this pollination strategy. Since the used covering is white, flower color and shape features are hidden. Thus, we assume for now that the observed attraction is mostly or mainly mediated by fragrance volatiles, a common feature in deceptive orchids [31]. Finally, we would like to propose a new hypothesis for the pollination strategy of E. densiflorum: instead of offering alkaloids, we propose that the flowers of E. densiflorum (and related species) mimic the fragrances of plants that are the actual alkaloid sources for these Lepidoptera. If this hypothesis is correct, it may constitute a specialized, novel subtype of pharmaco-food mimicry in Orchidaceae, since the flowers may explore a narrow, specialized, ecological niche. ecological feature: males of both groups actively collect pyrrolizidine alkaloids that are used during mating [30]. Alkaloids are acquired through floral nectar and, according to Pliske [30], the main sources of these alkaloids involve species of the genera Heliotropium, Tournefortia, and Myosotis (Boraginaceae), as well as Eupatorium (Asteraceae). Early observations of Arctiinae and Ithomiinae in species related to E. densiflorum, such as E. floribundum and E. paniculatum [28], led researchers to propose that the flowers of these orchids should also be a source of pyrrolizidine alkaloids. However, alkaloids were not found in the flowers of these species [7]. According to our observations, flowers of E. densiflorum are devoid of free nectar, as the related E. paniculatum [7]. Since alkaloids are acquired through nectar, and the latter is absent in the flowers of E. densiflorum, there is no way for the male Lepidoptera to obtain such resources in the flowers of the orchid under study. Flowers used for breeding system experiments were enclosed with white tule (see Methods). During the whole flowering season of 2021 and 2022, the enclosed flowers were regularly visited by Arctiinae and Ithomiinae males ( Figure 5). This behavior is similar to that described by De Vries and Stiles [28], which noticed males of Arctiinae and Ithomiinae attempting to reach enclosed inflorescences of E. paniculatum. This highlights the importance of fragrance features (chemical compounds) in this pollination strategy. Since the used covering is white, flower color and shape features are hidden. Thus, we assume for now that the observed attraction is mostly or mainly mediated by fragrance volatiles, a common feature in deceptive orchids [31]. Finally, we would like to propose a new hypothesis for the pollination strategy of E. densiflorum: instead of offering alkaloids, we propose that the flowers of E. densiflorum (and related species) mimic the fragrances of plants that are the actual alkaloid sources for these Lepidoptera. If this hypothesis is correct, it may constitute a specialized, novel subtype of pharmaco-food mimicry in Orchidaceae, since the flowers may explore a narrow, specialized, ecological niche. Whereas pollination by Lepidopterans was already documented in a few Epidendrum species [7,8,10], the fact that these insects are temporarily trapped by the flowers was Whereas pollination by Lepidopterans was already documented in a few Epidendrum species [7,8,10], the fact that these insects are temporarily trapped by the flowers was largely overlooked. According to our observations, insects can remain on the flower for more than 75 min, depending on the species. Our video record indicates that, after struggling to leave the flower, pollinarium-laden insects fly away without visiting another flower of the inflorescence. Such behavior may increase the chances of cross-pollination. However, it is important to point out that some small Arctiinae moths died at the flowers, resembling what happens with the "moth catcher" vine, Araujia sericifera (Apocynaceae) [32]. This last observation suggested that the flowers of E. densiflorum can be "trap-flowers". The presence of trap-flowers in Orchidaceae has already been documented in the subfamily Cypripedioideae (reviewed by [33]) and in the pseudocopulatory genera Pterostylis (Orchidoideae) [34] and Trigonidium (Epidendroideae: Maxillariinae) [35]. In none of the above cases, however, the trapping involves nectar-seeking insects, such as in E. densiflorum. Flowers of Cypripedioideae are food or oviposition site mimics (reviewed by [33]). The flowers of Pterostylis and Trigonidium attract insect males (of Sciariidae flies and Meliponina bees, respectively) that attempt copulation with the median petal (labellum) and are then temporarily trapped in the flower cavity during the process [34,35].

Study Species and Flower Features
Throughout this contribution, overall orchid morphological terms follow Dressler [2]. Taxonomic delimitation of Brazilian Epidendrum species follows Pessoa [11]. Epidendrum densiflorum locally occurs both as an epiphytic or rupicolous species, always near river courses and streams. The cane-like stems may reach up to 1 m high [12]. The inflorescences are terminal and may reach up to 40 cm in length. The species has a flowering peak in January to April. A plant voucher is deposited at the ICN Herbarium (R. B. Singer s.n. 19/05/2019). Plant and overall flower features were documented through photos. To locate nectar, ten buds were isolated with tulle and checked for nectar 24 h after opening. A microsyringe CG model 701 RN (5 µL volume) was inserted in the floral cuniculus to pump the nectar (if present). Since no free nectar was found (see Results, Section 2), no further analyses (volume, concentration) were possible. Complementary, ten additional fresh, intact flowers were obtained from two specimens and dissected under a stereomicroscope to locate nectar. Pollinators and their behaviors at flowers were documented through photos and videos. The video record was useful to record pollinator behavior as well as to confirm which insects were pollinators (see below) and to quantify the time spent by the insects at the flowers. For the purposes of this contribution, and following Castro et al. [27], only animals that were observed removing and/or depositing pollinaria were considered pollinators. During the whole observation period, we recorded the percentages of pollinarium removals (male success) and depositions (female success) over the total of produced flowers. The average pollinator visitation time was compared using the Krukal-Wallis test, a non-parametric alternative, since the data did not meet the assumptions of normality and homoscedasticity of variances required by analysis of variance. Dunn's test was performed to differentiate a posteriori means with Holm's correction [37]. The ggstatsplot and ggplot2 packages were used for the analyses and performed in the R 4.0.5 software [38]. In addition, fruiting success was recorded in 10 plants bearing 437 flowers, by counting the number of fruits formed in their inflorescences 15 days after the end of the pollination observations. Additionally, we calculated Nilsson's male efficiency factor as the ratio between the percentages of pollinated flowers divided by the percentage of flowers that acted as pollen donors [39]. Pollinators were identified, and vouchers were deposited at the Museu de Ciências Naturais do Jardim Botânico de Porto Alegre (MCN, SEMA, RS).

Breeding System
The breeding system of E. densiflorum was studied by employing ten individuals cultivated at the Orchidarium of the Porto Alegre Botanical Garden, following the procedures detailed by Castro et al. [27]. Inflorescences were covered with tule to avoid pollinators and/or other insects. Four treatments were applied (Table 1): (1) intact flowers (test for autonomous self-pollination), (2) emasculation (test for apomixis), (3) hand self-pollination (test for self-compatibility), and (4) cross-pollination. The four treatments were applied to all the individuals under study. A total of 5 replications of each treatment were performed in each specimen, totaling 50 replications per treatment ( Table 1). The numbers of fruits were compared between treatments using the Krukal-Wallis test, a non-parametric alternative, since the data did not meet the assumptions of normality and homoscedasticity of variances required by analysis of variance (Anova). Dunn's test was performed to differentiate a posteriori means with Holm's correction [37]. The ggstatsplot and ggplot2 packages were used for the analyses and performed in the R 4.0.5 software [38].

Conclusions
At the end of the Introduction, we proposed the following hypotheses: (1) that E. densiflorum may be pollinator-dependent, (2) that pollinators may be diurnal Lepidoptera, (3) that, owing to the rareness of fruits under natural conditions, E. densiflorum may be self-incompatible, and (4) that fruit rareness may be caused by either absence of pollinators or by self-incompatibility. According to the data gathered in this contribution, hypotheses (1-3) were fully corroborated and hypothesis (4) receives at least partial support. Pollinators were present and the plant was, indeed, self-incompatible. However, further observations are needed to fully address the extent of insect-mediated self-pollinations (and, consequently, abortions mediated by passive or territorial insects) are necessary. Unlike other Epidendroid orchids recently studied (e.g., [27]), the pollinators of E. densiflorum tend to visit a single flower at the inflorescences. The exhaustion after removing the pollinarium from the trap-flowers of E. densiflorum makes the insects quickly leave the inflorescences. However, it cannot be discarded that insects could return later. This is especially true for territorial Lepidoptera, such as the Ithomiinae. As already commented by other authors [27], pollinator-dependent, self-incompatible orchids (such as E. densiflorum) may be particularly fragile in the context of habitat fragmentation, since pollen flux between conspecific individuals is mandatory. Thus, studies such as those presented herein are important if these orchids and their pollinators are to be conserved and correctly managed.