Acta Biológica Colombiana

Study area and species

The studies were conducted from July 2013 to September 2014 at the Station of Research, Environmental Training and Education Mata do Paraíso, (hereafter, Mata do Paraíso; 20°48' S and 42°51' W, 690 m.a.s.l.), the largest forest fragment in Viçosa municipality, Minas Gerais state, southeastern Brazil. The fragment has Atlantic Forest vegetation (Oliveira-Filho and Fontes, 2000), being classified as montane seasonal semideciduous forest (Veloso et al., 1991).

According to Köppen's classification, climate in Viçosa is type Cwa, i.e., mesothermal with hot wet summers and cold dry winters. Annual means of rainfall, relative humidity and temperature are, respectively, 1221.4 mm, 81 %, and 19.4 °C. The dry cold season extends from April to September while the warm rainy season begins in October and lasts until March.

Piper caldense is a 2-3 m high shrub. The hanging spikes are 3 to 5 cm long. Flowers are protected by a triangular subpeltate bract with fringed margins; the androecium consists of four stamens; the gynoecium has a superior ovary, short style and three stigmas; and the fruit is a drupe (Carvalho-Okano and Alves, 1998). A voucher specimen was deposited in the VIC Herbarium (A Valentin-Silva 82, DL Vargas-Rojas).

At Mata do Paraíso, P. caldense plants occur in moist and partially shaded areas of the understory. All analyzed plants had spikes with flower buds throughout the study period. These spikes remained latent for up to 12 months and were approximately 0.5 cm long. Spikes with flowers at anthesis were observed in the dry season, between August and September 2013 and 2014.

Sex expression and floral morphology and biology

To define the sex expression of plants and flowers, we analyzed flowers on spikes of 50 plants with a hand lens (30 x). Additionally, spike length and diameter (central portion) were measured in 49 spikes (14 with bisexual flowers and 35 with unisexual ones) having flowers at late anthesis. To assess the average number of flowers per spike, the number of flowers per inflorescence was counted in ten spikes for each flower type.

Fresh spikes were stored in 70 % ethanol and examined under a stereomicroscope (model EZ4D, Leica Microsystems, Deerfield, Illinois, USA) to verify the presence of androecium and/or gynoecium. In addition, stigma receptivity, which was characterized by the exposure of turgid stigmatic papillae (Valentin-Silva et al., 2015), was monitored in bisexual flowers of 15 spikes from different plants in the field, for 10 consecutive days, with a hand lens. The dynamics of pollen release from the four stamens of both bisexual and unisexual flowers was observed. Pollen grain viability was tested with acetic carmine (Radford et al., 1974). To this end, 40 spikes (20 per floral type) were fixed in FAA₅₀ (formaldehyde, acetic acid, and 50 % ethanol, 1:1:18 v/v/v) for 48 h and then stored in 70 % ethanol (Johansen, 1940). Completely exposed stamens (not protected by the floral bract) were used. On each slide, five anthers of flowers from the same spike were squashed in a drop of acetic carmine. A total 200 pollen grains were counted per slide, being then separated into viable and non-viable (Kearns and Inouye, 1993).

To obtain further information on floral morphology and biology, morphoanatomical studies were performed at the Laboratory of Plant Anatomy of Federal University of Viçosa, and scanning electron microscope analyses were conducted at the Center for Microscopy and Microanalysis at the same institution. We used samples of spikes having bisexual and staminate flowers at different stages, previously fixed in FAA₅₀ and stored in 70 % ethanol (Johansen, 1940). The adopted techniques were similar to those described by Valentin-Silva et al. (2015).

Breeding system and flower visitors

Following the methods proposed by Dafni et al. (2005), two tests were conducted in spikes with bisexual flowers: spontaneous selfing-15 spikes were bagged at pre-anthesis until fruit formation or inflorescence abortion; and open pollination-15 spikes were labelled and left exposed to pollinators. Spikes were isolated with cloth (fully closed mesh) bags. The percentage of fruiting in these tests was computed, considering the mean number of flowers per spike.

We recorded the visitation frequency on five non-consecutive days in August 2014, from 7:00 AM to 5:00 PM, totalizing a sampling effort of 40 h. Floral visitors were captured to localize the pollen deposition sites on their bodies. Voucher specimens were deposited in the Entomology Museum of Federal University of Viçosa.

Statistical analysis

The above-mentioned measurements, as well as those of pollen grain viability, were subjected to analysis of variance (ANOVA) with software R, version 3.0.1 (R Development Core Team, 2010). We assessed the number of flowers with a means test (t test), using the same software.

Sex expression and floral morphology

Piper caldense has bisexual and unisexual (staminate) flowers, each flower type being located on separate spikes. In 15 of the 50 analyzed plants, all spikes having flowers still at the bud stage were aborted, which rendered flower sex identification impossible. Of the other 35 plants, 14 had spikes with staminate flowers as well as spikes with bisexual flowers, while the other 21 plants had only spikes with staminate flowers. Therefore, the number of plants having only staminate flowers was 1.5 times higher than that of plants having both flower types. There were size differences between spikes with bisexual flowers and spikes with staminate flowers (Table 1), with statistical differences being detected in both spike length (F = 16.7, p = 3.6 e-⁶) and diameter (F = 33.2, p = 1.18 e-⁹). Moreover, spikes with bisexual flowers had on average more flowers than spikes with staminate flowers: 252.8 ± 27.2 and 198.3 ± 27.5, respectively. These averages differed from each other (t =-4.5, p = 0.003).

Table 1: Size (mean ± standard deviation) of Piper caldense spikes having both bisexual and staminate flowers and viability (mean ± standard deviation) of pollen grains from all four stamens. Values in a same column followed by the same letter do not differ statistically.

Bisexual flowers have a floral bract which, at the bud stage, fully protects all four stamens; white, dithecous anthers (Fig. 1A); a pistil with trifid stigma; and a superior, unilocular, uniovular ovary with basal placentation. Staminate flowers also have a floral bract that fully protects all four stamens, having the same characteristics of the androecium of bisexual flowers and also a pistillode (nonfunctional, reduced pistil; Fig. 1B).

Figure 1: Piper caldense flowers. A) bisexual flower, with two of the four stamens and the pistil; note the abscission tissue in the filament (arrow). B) staminate flower, with two of the four stamens and the pistillode. C) bisexual flower with partially exposed anthers of stamens 1, 2 and 3 and upright, juxtaposed stigmas. D) bisexual flower with three stamens; note the filament scar of the first stamen that released pollen. E) bisexual flower with stamens 3 and 4. F) bisexual flower with stamen 4. G) staminate flower with partially exposed anthers of three stamens (1, 2 and 3). H) staminate flower with stamens 3 and 4; note the filament scar of stamens 1 and 2. I) staminate flower with exposed anther of stamen 4. The bract apex is directed toward the distal region of the spike. br = bract, fs = filament scar, ov = ovary, pi = pistillode, st = stigma, s1 = stamen 1, s2 = stamen 2, s3= stamen 3, s4 = stamen 4. Scale bars: 400 mm.

Biology of bisexual flowers

In bisexual flowers, there was a partial opening of the bract and exposure of the anthers of three stamens (stamens 1-3, Fig. 1C) for two to three days (stamens were numbered according to the position of the floral bract; Fig. 1C). Subsequently, for up to three additional days, the upright, juxtaposed stigmas were visible among those three stamens, as was the apical portion of the anther of stamen 4 in some flowers (Fig. 1C), due to the continuous bract opening. Thus, these changes occurred in the flowers for three to six days before anthesis.

Anthesis of bisexual flowers began with stigma receptivity, being characterized by the exposure of turgid papillae at the distal portion; stamens remained as shown in Fig. 1C, with indehiscent anthers. The occurrence of stigma receptivity in the period prior to pollen release indicates protogynous dichogamy. The pistillate flower phase lasted an average two days.

The bisexual phase began with pollen being released from flowers while stigmas were still receptive, a characteristic of incomplete protogyny. Either stamen 1 or 2 was the first one to release pollen. One of these stamens was higher than the other, due to complete exposure (filament and an ther), and released pollen first, beginning in late morning but occurring mainly in the afternoon, at around 2:00 PM (same time of release of the other stamens). Thereat, the distance among stigmas increased, almost completely exposing the still-turgid stigmatic papillae.

On the second day of the bisexual phase, the second stamen to undergo anther dehiscence (either stamen 1 or 2; Fig. 1D) released pollen. By then, the stamen that had released pollen the day before may either still have residual pollen, in which case such stamen has characteristics that indicate senescence, such as anther darkening and wilting; or it may have already been aborted (note the filament scar of the stamen that first released pollen, in Fig. 1D). At this stage, the bract is fully open, further exposing the anther of stamen 4. Furthermore, in some flowers, the distal portion of stigmas was darkened and had plasmolyzed papillae, indicating the onset of cell senescence.

On the third day, stamen 3 was completely exposed and pollen was released from the anther (Fig. 1E). The stamen that had released pollen the day before may either have already been aborted or still have a senescent anther with residual pollen. In that occasion, the anther of stamen 4 remained indehiscent. Stigmas then were totally exposed (Fig. 1E); from this stage, the darkened portion of stigmatic surface was quite variable, but was thereat observed on the middle portion of the stigma towards its base.

On the fourth day, senescence- or abortion-related characteristics were observed on the anther of stamen 3 (Fig. 1F). Stamen 4 was completely exposed and its anther released pollen. Most flowers had completely darkened stigmas with collapsed papillae, indicating the end of stigma receptivity; however, thereat some flowers may have turgid papillae in the proximal third of stigmas.

Anthesis of bisexual flowers thus lasted six days. Thereafter, ovary size increased due to fruit formation in fertilized flowers.

Biology of staminate flowers

In staminate flowers, with the beginning of bract opening for two to three days, the anthers of three stamens (stamens 1-3; Fig. 1G) were partially exposed. At the end of this period, stamens 1 and 2 were more exposed than stamen 3. Flowers may remain at this stage for up to five additional days. Therefore, these changes occurred in the flower at three to eight days before anthesis.

Anthesis began with pollen release (see the filament scar of stamens that had already released pollen, in Fig. 1H-I), with similar dynamics to the one of bisexual flowers, for four consecutive days. In stamen 4, we observed that some flowers might have residual pollen for up to two days. Thereafter, spikes dried out and were aborted.

Pollen viability, breeding system and flower visitors

Pollen grain viability was similar between stamens of bisexual and staminate flowers, as well as among stamens of each flower (Table 1). Means did not differ statistically (F = 0.5, p = 0.861).

No fruiting was observed in the spontaneous self-pollination test, while a 63.3 % fruit set was observed in the open-pollination test, thus indicating the occurrence of self-incompatibility in P. caldense. The fact that the number of fruits/spike (=161.5 ± 93.1) was smaller than the one of flowers/spike (= 252.8 ± 27.2) indicates that not all flowers were pollinated and/or fertilized.

Spikes were visited by insects of two orders, each being represented by one family: Hymenoptera, Apidae (bees, Fig. 2) and Diptera, Syrphidae (flies) (Table 2). These insects collected pollen, the only floral resource available.

Figure 2: Bees on spikes of Piper caldense. A). Melipona mondury. B). Schwarziana quadripunctata.

Table 2: Visitation frequency (number and percentage of visits per insect) of flower visitors on Piper caldense flowers at Mata do Paraíso, Viçosa municipality, Minas Gerais state, southeastern Brazil.

Among bees, we observed eight species grouped into six genera (Table 2). These insects made 349 flower visits. Apis mellifera accounted for the highest visitation frequency, followed by Melipona quadrifasciata, M. bicolor, Trigona spinipes, M. mondury (Fig. 2A), and Swarziana quadripunctata (Fig. 2B); these bees altogether performed 79.9 % of all visits (Table 2). As for flies, we observed seven species grouped into four genera (Table 2). These insects made 67 flower visits. The highest visitation frequencies were of three Ocyptamus species; these flies altogether accounted for 14.6 % of all visits (Table 2).

Bees began visiting at around 8:00 AM, which lasted until 4:00 PM. Visits peaked between 9:00 and 11:00 AM, with another minor peak at 2:00 PM (Fig. 3). The timetables and visitation peaks of flies and bees were similar, differing only in number of visits (Fig. 4).

Figure 3: Number ofvisits ofthe most frequent bees on Piper caldense flowers, from 7:00 AM to 4:00 PM, at Mata do Paraíso, Viçosa municipality, Minas Gerais state, southeastern Brazil. • = Apis mellifera, ▲ = Melipona bicolor, ◊ = Melipona mondury, o = Melipona quadrifasciata, ■ = Schwarziana quadripunctata, Δ= Trigona spinipes.

Figure 4: Number of visits of the most frequent flies on Piper caldense flowers, from 7:00 AM to 4:00 PM, at Mata do Paraíso, Viçosa municipality, Minas Gerais state, southeastern Brazil. • = Ocyptamus sp.1, ▲ = Ocyptamus sp.2, ■ = Ocyptamus sp.3.

The insect visitation frequencies and deposition sites of pollen grains on insect bodies (in the ventral region and legs) both suggest that the two groups of flower visitors, especially bees, are effective pollinators of P. caldense.