Vegetative branches of three individuals of Passiflora alata and P. edulis were collected from plants growing in Centro de Pesquisas Químicas, Biológicas e Agrícolas

(CPQBA, Brazil). Voucher specimens were deposited in the Herbarium of Universidade Estadual de Campinas (P. alata: UEC 145620; P. edulis: UEC 111509).

Morphological aspects of individuals exposed to shade and sunlight conditions were registered. Two plants of each species were observed over four consecutive days during the beginning of the morning (4 - 7 am) in the months of December to March (2008/2009). Nectar feeding visitors were collected, preserved in 70% ethanol and sent for identification.

The base of branches was isolated with Tanglefood grease to avoid the removal of the exudate by insects. In the subsequent morning, exudate was tested with Glicophita Plus (Bayer S. A.) [14] and collected for chemical analysis.

Petiolar EFN at different developmental stages were fixed in Karnovsky [15]. The material was washed, dehydrated in a graded ethanol series, critical point dried and coated with a thin layer of gold. Observations were carried out on a JEOL JSM 5000 LV 6350 scanning electron microscope at an accelerating voltage of 20 kV.

Petiolar EFN were fixed in FAA [16] for 24 hours and in Lillie’s buffered neutral formalin [17] for 48 hours to better preserve hydro and lipophilic compounds, respectively. The material was isolated and dehydrated in a graded tertiary butanol series [16] and embedded in Paraplast. The serial sections were stained with Safranin and Astra blue [18], and Flemming’s triple staining [16]. The sections were also examined under polarized light to verify the occurrence of starch grains, crystals and tracheary elements.

Histochemical tests were performed in both fresh and fixed material. Treatments and target compounds are summarized in Table 1. Standard control procedures were carried out according to the respective techniques.

The exudate from EFN of both species were collected with a microcapillary, clustered and stored in eppendorff tubes at −10˚C. A 10 μl aliquot of exudate was dissolved at a concentration of 50 μg/ml in water:methanol (7:3, v: v) containing 1% formic acid. These solutions were analysed by direct infusion with a syringe pump (Harvard Apparatus) at a flow rate of 10 μL/min.

ESI-MS mass spectra were acquired in the positive ion mode using a hybrid high resolution Micromass Q-TOF mass spectrometer, under the following conditions: Capillary +3000 V, cone +40 V, dessolvation temperature 100˚C.

Table 1. Results of the histochemical tests performed on EFN of Passiflora alata Curtis and P. edulis Sims.

Notes: ^*excitation at 350 - 390 nm; +, positive reaction; −, negative reaction.

There is an asynchrony concerning the development and release of nectar from petiolar EFN of P. alata, where it all starts on glands close to the stem (basal glands, Figure 1(a)). During secretory phase, nectar accumulates on the crateriform region of the nectary and is exhibited in the apex of the gland (upper gland, Figure 1(b)).

Anthocyanins are visualized on young organs of P. edulis exposed to sunlight (Figure 1(e)). In this species, EFN presents a synchronous development and the release of nectar can be seen over the gland surface (Figures 1(f) and (g)).

The secretory phase of petiolar EFN is recognized since the second visible node in both species. Profuse exudate is registered only in plants whose base of branches was isolated with Tanglefoot grease (Figures 1(b), (f) and (g)). Crematogaster ants are observed collecting the nectar from both species (Figures 1(c) and (h)) and eventually can injure the secretory epidermis of EFN of P. edulis. Worker bees of Tetragonisca angustula also collect the nectar from P. alata (Figure 1(d)).

3.2. Micromorphological, Ontogenetic and Histochemical Data The meristematic phase of EFN is identified by the appearance of protuberances on the base of leaf primordium (Figures 2(a-c)). The presecretory phase of P. alata is characterized by the stretching of the peripheral portion until it acquires a stipitate-crateriform shape (Figures 3(a) and (b)). In P. edulis, the protuberances increase in volume until become flat in shape (Figure 2(d)), an indicative of the beginning of presecretory phase; the

morphology of the EFN at the end of this phase is a convex shape, as a result of the increase in volume of the gland.

The asynchrony in the development of EFN of P. alata is shown in Figures 3(a) and 4(a). Only the central region of EFN has nectariferous cells (Figure 4(b)); the peripheral region consists of ordinary epidermal cells (Figure 4(f)). Note in Figures 4(b) and 6(a) that pre-nectar can be seen inside nectariferous cells fixed in FAA. There is a differential preservation of cell content according to the fixative used.

The secretory phase is recognized by cuticle detachment in the central region of the petiolar EFN (Figures 3(c) and (e)-(g)). Initially, small elevations are observed and then they protrude, rising in the convex region (Figures 3(e)-(g)). The subcuticular space is shown in the transversal section of EFN (Figures 3(g), 4(c) and 6(c)). The anatomical features of EFN on secretory phase are presented ahead.

Nectariferous epidermis the multiple epidermis is covered with a thin cuticle (Figures 5(b) and (c), 6(b, e and f)) and cuticular flanges extend up to the inner periclinal wall of the first cell layer (Figures 5(c), 6(b and f)). The release of the nectar through nectariferous epidermis promotes the cuticle detachment. Nectar is confined to a subcuticular space, situated between the cuticle and the outer periclinal wall of epidermal cells (Figures 5(c) and 6(e)). Cuticle disruption was observed only after ants collect nectar from EFN of P. edulis.

Nectariferous parenchyma and subnectariferous parenchyma—are distinct in both species. Both tissues are separated from each other by vascular terminations (Figures 4(c) and 6(c)). EFN of P. alata possess secretory

idioblasts scattered through the nectariferous parenchyma (Figure 4(c)). The subnectariferous parenchyma presents few starch grains (Figures 4(d) and (e)) and crystals close to the vascular bundles (Figure 5(c)). Secretory idioblasts and starch grains are not visualized on EFN of P. edulis (Figures 6(a)-(f)).

Vascular tissue—the main trace comes from the petiole and it branches at the base of the nectary (Figure 4(b)). Vascular terminations consist of phloem only, and are embedded in the nectariferous parenchyma (Figures 4(c) and 6(c)).

The post-secretory phase is characterized by the establishment of fungal hyphae over the EFN secretory region in both species and by flattening the EFN edges of P. alata (Figures 3(d) and 4(g)).

The histochemical results are summarized in Table 1. The tests indicate the presence of terpenoids, phenolic compounds and alkaloids inside idioblasts and some nectariferous cells of P. alata (Figures 5(d) and (g)). Fresh section of P. edulis EFN shows anthocyanin restricted to non-secretory epidermis (Figure 6(d).

Glucose, fructose and sucrose were detected in nectar of both species (Figures 7 and 8). The hexoses (mass 180 Da) were detected mainly as their sodium (m/z 203) and potassium (m/z 219) adducts. Glucose (mass341 Da) was observed as its protonated (m/z 342) form as well as sodium (m/z 365) and potassium (m/z 381) adducts.

Among the secondary metabolites, apigenin (flavonoid) was detected in both species (m/z 271) Harmine, an alkaloid, was found only in nectar of P. edulis (Figure 9) as its sodium adduct (m/z 251).We searched for alkaloids and flavonoids commonly found in these species [24].

The visual selection conditioned many diversifications of leaf morphology on Passiflora and other tropical plants

taxa [8]. Behavioral experiments show that Heliconius butterflies present color vision as a possible answer to selection of the discriminative ability of oviposition sites by females [26-29], although the chemical stimuli are not excluded [6].

The sensibility spectra of color vision of Heliconiinae butterflies is 300 to 700 nm, where the peak of maximum sensibility is about 450 - 560 nm, corresponding to blue and yellow colors, respectively [28]. Agraulis vanillae, Eueides isabella and Heliconius erato stand out among heliconian species that consume P. alata leaves [30]. The description of their eggs in the surveyed literature [31] shows morphological similarities with the petiolar EFN of P. alata. Butterflies of these species oviposit only on young organs of the plant that do not possess previous eggs [5,31]. The presence of egg mimics prevents the oviposition of heliconian females, avoiding the feeding of plants by caterpillars. These species lay their eggs isolatedly or in groups of up to three on E. isabella. The selective pressure from those heliconian species could explain the variation among the number of EFN that exists in the same individual of P. alata, simulating distinct ovipositions on each petiole.

Plants of P. edulis exposed to sunlight present anthocyanins on the epidermis of young organs and are less consumed in comparison to shaded individuals. Anthocyanins stand out as constitutive defenses that act over herbivorous and are modulated by plants [32]. The most accepted hypothesis for anthocyanins as legitimate antiherbivory defenses refers to the aposematic color observed in plants that possess this pigment, inhibiting herbivores that do not recognize the red color [33]. Heliconius species do not show sensibility to the wavelengths related to this color [25] and cannot recognize the young organs preferential to oviposition, which could justify the lower consumption of plants under sunlight. However, parasitism and predation of this species could increase the incidence of herbivory: those two events can overwhelm lianas present in sunlight areas and explain why some butterfly species are not common in those places [5,34].

Sugars present in the exudate are the main source of this substance to mutual Crematogaster ants [35]. The observation of profuse nectar only after the isolation of branches with innocuous grease shows that the exudate is constantly collected. The asynchrony of EFN from P. alata can provide nectar for a long period, while the synchrony of P. edulis may offer a higher volume of this source. The test with Glicophita Plus indicated the presence of glucose in the exudate, ensuring that those glands are EFN. According to MS analysis, the nectar contains glucose, fructose and sucrose. These mono and disaccharides seem to be important food sources to ants [36].

Besides ants, other insects as bees and wasps can also collect nectar from EFN [37-39]. This is the first register of Tetragonisca angustula collecting nectar from an EFN. These insects do not present an aggressive behavior. In this context, they are analogous to robber insects, in the same sense used for flower visitants that do not pollinate.

The secretory phase of petiolar EFN occurs on young organs of both species. Feeding preference experiments show that Heliconius larvae select young leaves of Passiflora [40]. The presence of glands at the secretory phase limited to young portions of the plant—as already observed in other species of Passiflora [41]—can be related to the preference of heliconian females to lay eggs on young organs of plants [6].

The ontogenetic development of EFN from P. alata and P. edulis is consistent with other Passifloraceae species: the glands arise from the activity of the protoderm, procambium and ground meristem [41,42]. It also occurs with foliar EFN from Turneraceae species—Passifloraceae sensu lato [43]—and these glands are considered homologous structures [39,41,42]. The variation in the number of petiolar EFN, as found on P. alata and P. amethystina [41], suggests the occurrence of serial homology. The selective pressure responsible for this variation can be related to the coevolution with heliconian.

The release of nectar through cuticle ruptures was observed on species that do not present the expansion of the EFN non-secretory region [42]. However, the rupture founded in some EFN of P. edulis was associated with ant aggressive behavior during nectar collection. Although we found nectar accumulation in the subcuticular space of P. alata, there was no observation of cuticle rupture. Even though there is a report of the inexistence of pores on the cuticle of the Passiflora species [42], we strongly suggest more accurate studies about cuticle structure during the secretory process.

The crateriform morphology of EFN from P. alata can provide a microenvironment to confine terpenoids synthesized by the idioblasts, analogously to the corona in the flowers of Passiflora [44]. Terpenoids are the major volatile compounds issued by plants as an answer to herbivory, signalling to carnivorous arthropods either to prey or to parasitize herbivorous [45].

According to histochemical tests and the analysis on mass spectrometry, secondary metabolites are retained inside secretory cells (idioblasts and nectariferous cells) of the EFN from P. alata. A possible function of this retention would be the decrease in herbivory of these glands.

Alkaloids and flavonoids do not accumulate in the petiolar EFN from P. edulis, but are present in trace amount in nectar, suggesting that the concentrations of these compounds both in nectar and inside secretory cells possess a pleiotropic effect on the phloem [46]. The presence of secondary metabolites in nectar can be related to the attraction of specific consumers, since the deterrent or toxic nectar to some collector may not affect others [47]. Alkaloids, flavonoids and terpenoids are antifungical [48-50]. These metabolites can inhibit fungi proliferation, since hyphae are only found on the nonsecretory region of EFN of P. edulis and over petiolar EFN of both species at post-secretory phase.

Although histochemical tests and analysis of the nectar on MS provide qualitative results, it is possible to establish a relationship between the relative concentrations of secondary metabolites inside the plant and released by them in the nectar through the sensibility of both techniques. The MS sensitivity is femtogram-order, allowing the detection of trace compounds, as those present in the nectar. Histochemical tests present a lower sensitivity of microgram-order, detecting compounds in a higher concentration. Therefore, alkaloids and flavonoids are more concentrated inside the idioblasts than in the nectar of P. alata. The relationship between the concentration of secondary metabolites inside and outside the plant and the commitment of the development of herbivores and/or nectar collectors can be plotted on a growth versus concentration diagram, as shown in Figure 10.

The secondary metabolites present in higher concentration in the EFN of P. alata show deterrent function against herbivores. Those who are not able to detoxify these compounds undertake their development as these substances are accumulated in their organism. The low concentration of alkaloids and flavonoids in nectar is insufficient to affect the metabolism of collectors that establish a mutualistic relationship with P. alata and P. edulis. The analysis of the diagram suggests that the

modulation of the concentration of secondary metabolites allows the plant to relate differently to herbivores and mutualistic consumers, highlighting the importance of secretory tissues (idioblasts and nectariferous cells) in this process.