Introduction

Most flowering plants offer different rewards to attract pollinators (Rathke 1988; Waser et al. 1996); however, a small number of Angiosperms are rewardless (Ackerman 1986; Renner 2006). These plant species have developed alternative strategies to ensure constant visits by pollinators and enhance their reproductive success. One of these is to take advantage of neighboring plant species that offer rewards (Laverty 1992; Johnson et al. 2003a; Renner 2006; Juillet et al. 2007), and another is deceptive pollination (Ackerman 1986; Johnson and Nilsson 1999; Jersáková et al. 2006; Renner 2006).

Pollination success in non-rewarding plants that flower in association with reward-producing plants will be enhanced (facilitation) or diminished (competition) depending on pollinator behavior (Laverty 1992; Lammi and Kuitunen 1995; Johnson et al. 2003a; Internicola et al. 2006, 2007). Pollination by facilitation is an amazing phenomenon that has been gaining the attention of ecologists who want to understand how the pollination mechanisms between reward and rewardless species work (Johnson et al. 2003a; Jersáková et al. 2006; Juillet et al. 2007). Several studies have recorded that animal-pollinated plants, especially those pollinated by insects, may have higher visitation rates when co-occurring in communities with species that attract pollinators (Thomson 1978; Feinsinger 1987; Feldman et al. 2004; Moeller 2004). This has been recorded as beneficial to non-rewarding species growing in the same plant community and close to individual species that offer a reward (Laverty 1992; Johnson et al. 2003a, Juillet et al. 2007). Few studies have demonstrated the intra- inter- or multi-species interactions of facilitation for pollinator attraction and flower pollination (Feldman et al. 2004; Moeller 2004; Hegland and Totland 2005). On the other hand, in a plant community it is unusual to find several places where both rewarding and non-rewarding species occur along with places where the rewardless species grows in monospecific clumps (Johnson 2000; Internicola et al. 2006, 2007). Understanding the relative importance of floral traits (morphology, color, display, and density) between rewarding and rewardless species in a system of pollination facilitation is crucial to further our understanding of reproductive ecology (Johnson et al. 2003a).

The Orchidaceae is one of the largest families of Angiosperms and its species exhibit some of these mechanisms of pollination by facilitation (Ackerman 1986; Jersáková et al. 2006). One third of the orchid species have rewardless flowers (van der Pijl and Dodson 1966; Ackerman 1986), so they obtain pollinator services by mimicking a rewarding model plant (flowers appear similar) (Johnson 1994, 2000), or by taking advantage of their proximity to a rewarding plant (flower similarity not necessary) (Johnson et al. 2003a; Juillet et al. 2007). Rewardless orchid species are a unique study model because pollinium (structure containing pollen grains) is not eaten by the pollinator, and it is deposited in different body places of the pollinators than pollen of the facilitating plant species (Neiland and Wilcock 1998; Johnson and Nilsson 1999).

Facilitation between rewarding and rewardless plant species that share pollinators will have one of two principal mechanisms: the magnetic species effect (Thomson 1978; Laverty 1992) or Batesian floral mimicry (Dafni and Ivri 1981a; Johnson 1994, 2000). In the magnetic species effect hypothesis, a rewarding species increases the pollination success of rewardless plants by increasing the local abundance of pollinators (Thomson 1978; Laverty 1992). Non-rewarding plant species profit from their proximity to plants that receive more pollinator visits (Laverty 1992; Johnson et al. 2003a). The dependence of a rewardless orchid on a rewarding plant species for pollination has been shown in a few convincing cases (Alexandersson and Ǻgren 1996; Johnson et al. 2003a; Juillet et al. 2007). However, other studies have shown that the presence of rewarding species did not improve the fitness of the rewardless orchids (Lammi and Kuitunen 1995; Alexandersson and Ǻgren 1996; Internicola et al. 2006). The evidence for the magnetic species effect as a positive pollination mechanism in non-rewarding orchids is both inconclusive and contradictory. For example, Lammi and Kuitunen (1995) reported increased fruit set in the terrestrial orchid Dactylorhiza incarnata in the absence of nectar-producing Viola flowers, whereas Alexandersson and Ǻgren (1996), Internicola et al. (2006), Johnson et al. (2003a), and Juillet et al. (2007) found that the presence of nectar-producing neighboring plants improves the pollination success and fruit set in the orchids Calypso bulbosa, Dactylorhiza sambucina, Traunsteinera globosa, and Anacamptis morio (respectively). Furthermore, the understanding of the magnetic species effect and the ecological and evolutionary forces that maintain this reproductive system remain unclear.

Batesian floral mimicry is another pollination facilitation mechanism in non-rewarding orchid species. This system is operating if the reproductive fitness of the mimic is higher when it occurs sympatrically with the rewarding model species, than in the absence of the model, and when the effect is constant over time and space (Roy and Widmer 1999; Johnson 2000). In Batesian floral mimicry, additional evidence must be provided to conclude that an orchid is exploiting an adaptive resemblance to a particular plant species (Roy and Widmer 1999). Such evidence might include matching spectral reflectance and behavioral experiments to establish that pollinators are unable to distinguish between the mimic and the model plants (Dafni and Ivri 1981a; Johnson 1994, 2000; Jersáková et al. 2006). A few studies support the theory that the presence of a model is necessary for reproduction by the mimic (Dafni and Ivri 1981a; Johnson 1994, 2000; Gigord et al. 2002), but there is no experimental evidence. On the other hand, Bierzychudek (1981) reported that monospecific clumps of the rewardless Epidendrum radicans and heterospecific clumps of this orchid with Lantana camara and Asclepias curassavica had similar fruit production, suggesting that the fitness of the mimic is similar in the presence or the absence of the model.

As far as we know, few natural and artificial manipulated experiments have demonstrated that the reproductive success of a rewardless orchid species is greater in the presence of rewarding species than in that of monospecific rewardless natural and artificial populations (Gigord et al. 2002; Johnson et al. 2003a; Anderson and Johnson 2006; Internicola et al. 2007). Furthermore, very little is known about the constancy of this pattern over time and space in the monospecific rewardless and heterospecific populations (Roy and Widmer 1999; Johnson et al. 2003a).

In this study, we evaluate the reproductive success of the rewardless O. cosymbephorum (Orchidaceae) in the presence of the rewarding M. glabra (Malpighiaceae). The orchid’s flowers resemble those of M. glabra and both are pollinated by the female Centris bees (Carmona-Díaz 2001). We experimentally test whether the presence of M. glabra increases the reproductive success of O. cosymbephorum and if that relationship is constant throughout time and across their distributions.

Methods

Study sites

This study was carried out between July 2000 and December 2003 at five sites in Veracruz, Mexico where O. cosymbephorum grows naturally isolated or in mixed clumps with M. glabra (hereafter referred to as monospecific and heterospecific groups): Agaltepec Island (monospecific and heterospecific groups), Pipiapan, Mimiahua, and Amamaloya (monospecific groups) in the Los Tuxtlas Biosphere Reserve, and at the La Mancha Coastal Research Centre of the Instituto de Ecología, A. C. (monospecific and heterospecific groups) (Fig. 1).

Agaltepec (18°24′ N, 95°05′ W; area 8.3 ha; altitude 430 m a.s.l.), Amamaloya (19°26′ N, 95°31′ W; area 30 ha; altitude 250 m a.s.l.), and La Mancha (19°36′ N, 96°22′ W; area 30 ha; altitude 30 m a.s.l.) have subdeciduous tropical forest as dominant plant community (Castillo-Campos and Medina 2005; Serio-Silva et al. 2002; Castillo-Campos and Laborde 2004). Their climate is highly seasonal, with a rainy season (July–October) and dry season (March–May). The mean annual temperature and precipitation are 26°C and 1,980–2,000 mm, respectively. Pipiapan (18°25′ N, 95°18′ W; area 200 ha; altitude 500 m a.s.l.) and Mimiahua (18°24′ N, 95°10′ W; area 500 ha; altitude 400 m a.s.l.) have tropical rain forest as dominant plant community (Castillo-Campos and Laborde 2004; Cristobal-Azkarate et al. 2005). The rainy season is from June to November, and the dry season from March to May. The mean annual temperature and precipitation are 25–26°C and 2,000–2,100 mm, respectively.

Rewarding plant

Malpighia glabra L. (Malpighiaceae) is a deciduous shrub 1–5 m in height with a wide Neotropical distribution from Florida (United States of America) to northern Brazil (Anderson 1979). In Mexico, this species occurs on both the Atlantic and Pacific coasts. The flowering season in the coastal populations occurs from July to October and in the highlands, from September to December. Shrubs can produce 500 inflorescences (Carmona-Díaz 2001), and each inflorescence produces 1–4 magenta flowers (Fig. 2) that reward its pollinators—solitary female bees Centris ruthannae—with oil and pollen (Buchmann 1987; Vogel 1990). In the tropical rain forest and deciduous forest, M. glabra occurs in clusters of 5–15 individuals per 25 m2 (Carmona-Díaz 2001).

Fig. 1
figure 1

Study sites with monospecific clumps of Oncidium cosymbephorum (Orchidaceae) (▲) and heterospecific clumps with plants of O. cosymbephorum and Malpighia glabra (Malpighiaceae) (●) at Los Tuxtlas Biosphere Reserve, and at La Mancha on the coast of the Gulf of Mexico

Full size image
Fig. 2
figure 2

Flowers of rewarding Malpighia glabra (Malpighiaceae) (left) and rewardless Oncidium cosymbephorum (Orchidaceae) (right)

Full size image

Rewardless plant

Oncidium cosymbephorum Morren (Orchidaceae) is an endemic epiphytic orchid found in the states of Tamaulipas, Veracruz and Tabasco in Mexico (Jiménez 1993; Carmona-Díaz 2001). Inflorescences bear resupinate non-rewarding flowers (1–25) that are magenta in color (Fig. 2) (Jiménez 1993; Carmona-Díaz 2001; Chase et al. 2005). Flowers are not selfing, so pollinator visits are required for them to set fruit (Jiménez 1993; Carmona-Díaz 2001). The flowering season is from July to November depending on the region and environmental conditions. In central Veracruz, O. cosymbephorum blooms from July to mid-October and in southern Veracruz the species flowers from September to November. Individual plants typically produce 1–6 fruits, depending on inflorescence size (Carmona-Díaz 2001).

The pollinators

The pollinators of both M. glabra and O. cosymbephorum are solitary female bees: Centris ruthannae Snelling (Anthophoridae) (Buchmann 1987; Carmona-Díaz 2001). Centris ruthannae frequently visit M. glabra flowers, clinging to the petals and scratching the elaiophores with their fore and hind legs to collect oil. Pollen of M. glabra is deposited on the abdomen of the bees ventrally. When visiting O. cosymbephorum flowers, they also cling to the petals and exhibit the same behavior as to collect oil in M. glabra flowers (Carmona-Díaz 2001). Pollinium of O. cosymbephorum is deposited in front of the bee’s head. When bee visits a new flower the head touches the estigmata surface depositing the pollinium.

Experimental procedures

Natural and artificial monospecific and heterospecific clumps

In 2001 and 2002 we tested the reproductive success of O. cosymbephorum under monospecific and heterospecific growth conditions at Agaltepec Island and La Mancha. All clumps of O. cosymbephorum growing naturally with and without M. glabra plants in the neighborhood (heterospecific and monospecific groups, respectively) were recorded in both sites. Monospecific plant clumps: We chose three naturally monospecific clumps of O. cosymbephorum, and we made three artificial monospecific clumps of O. cosymbephorum by cutting the flowers, buds and/or branches of M. glabra individuals growing near the orchid individuals. Heterospecific plant clumps: We chose three heterospecific clumps that had intact individuals of both O. cosymbephorum and M. glabra, and we created three artificial heterospecific plant clumps by transplanting individuals of O. cosymbephorum close to naturally monospecific clumps of M. glabra. The experimental monospecific and heterospecific patches were separated by at least 100 m, and each was composed of 6–7 orchid plants and 10–14 M. glabra shrubs. Each individual orchid was tagged and monitored monthly until fruit production occurred. Fruit set (number of developing fruits per number of flowers produced per plant) was compared among the monospecific and heterospecific clumps as combinations of two fully crossed factors: presence or absence of M. glabra, and the type of plant clump (natural or artificially made). Natural and artificial clumps of O. cosymbephorum were considered to be monospecific when there were no flowers of M. glabra within a distance of 5 m, and heterospecific when both species were closer than 5 m.

Individuals of M. glabra close to monospecific orchid plants that had been isolated

A second experiment was carried out to test the effects of rewarding M. glabra’s presence on the reproductive success of isolated monospecific O. cosymbephorum plants. We chose naturally monospecific orchid clumps, three clumps each at Pipiapan and Amamaloya, and two each at Mimiahua and Agaltepec Island. The fruit set of clumps containing six to ten individual orchid plants was recorded in 2000 and 2001. In 2002, a total of 120 shrubs of M. glabra were taken from Agaltepec Island and potted individually. When the shrubs bloomed, ten plants were put near to each of the previously monitored O. cosymbephorum clumps. A set of monospecific orchid patches monitored in 2000 and 2001 was used as a control at each site. Fruit set for each O. cosymbephorum individual with or without M. glabra plants nearby was recorded. In 2003 all the orchid clumps were transformed into monospecific groups again (all containers with M. glabra plants were excluded), and the orchid fruit set was recorded. The number of orchid plants was different between years because not all the individuals flowered in each year of the study.

Data analysis

The fruit set produced by natural and artificial clumps of O. cosymbephorum was analyzed using a factorial ANOVA. Factors were site, year, condition (presence or absence of M. glabra), and type of plant clump (natural or artificial). The response variable was fruit set (Zar 1986). We included the type of plant clump as a factor because artificial clumps might have an independent effect on fruit set. The effect of the nearby rewarding M. glabra plants on the reproductive success of O. cosymbephorum plants that had been isolated was tested as follows. In order to determine if the orchid flower display (flower number) was similar throughout the study, we performed a Kruskal-Wallis test with the factors, site and year. The 2002 fruit set was compared between sites using an ANOVA. Fruit set data were arcsine transformed before analyses, and all analyses were done with STATISTICA software V.7. Post-hoc analyses were developed in order to identify the significant differences.

Results

Natural and artificial monospecific and heterospecific plant clumps

At both the sites studied (Agaltepec Island and La Mancha) and for both years (2001 and 2002) O. cosymbephorum individuals in the presence of M. glabra shrubs had a larger fruit set than plants separated from the rewarding plants in both natural and artificial groups (Table 1). There were significant differences between sites, years, condition (presence or absence of M. glabra), and for the site × condition, and year × condition interactions. The interaction between the four factors analyzed was not significant (Table 2) (one way ANOVA, Tukey test at P < 0.05 level).

Table 1 Mean (±S.E.) number of flowers and fruits produced by Oncidium cosymbephorum (Orchidaceae) individuals in monospecific (M) and heterospecific (H) clumps (natural = N and artificial = A) at Agaltepec and La Mancha recorded for 2 years
Full size table
Table 2 Comparison of the fruit set produced by monospecific and heterospecific clumps (natural and artificial) of Oncidium cosymbephorum (Orchidaceae) using a factorial ANOVA, with site (La Mancha and Agaltepec Island), year (2000 and 2001), condition (monospecific or heterospecific clump), and type (natural or experimental) as factors, and fruit set as the response variable
Full size table

Individuals of M. glabra close to monospecific orchid plants isolated

In 2000 and 2001, isolated individuals of O. cosymbephorum did not produce any fruit at the Mimiahua, Amamaloya, or Agaltepec Island sites. At Pipiapan O. cosymbephorum plants produced only two fruits in 2000 and none in 2001 (Table 3). In 2002, O. cosymbephorum individuals in the proximity of cultivated M. glabra individuals had a larger fruit set than orchids separated from the rewarding species at all the sites studied (Table 3). In 2003, when all plants of M. glabra were excluded, O. cosymbephorum individuals did not produce any fruit set again. Sites differed significantly in fruit set (F 3,88 = 4.01, P < 0.01; Tukey test P < 0.05) and flower production (H = 58.48 P < 0.001), but there were no differences in flower production between years (H = 7.38, P < 0.11).

Table 3 Fruit production of O. cosymbephorum recorded at four sites over 4 years
Full size table

Discussion

Ours findings are consistent with the magnet species effect hypothesis (Thomson 1978), O. cosymbephorum benefits from the presence of the oil-producing plant. The reproductive success in O. cosymbephorum was significantly enhanced by the presence of individuals of M. glabra even though the orchid offers no flower reward to female Centris bees. Therefore, the presence of M. glabra to the orchid appears to be crucial to attract pollinators. Pollination facilitation has been documented for other non-rewarding orchid species (Alexandersson and Ǻgren 1996; Johnson et al. 2003a; Juillet et al. 2007).

Facilitation of O. cosymbephorum by M. glabra, as an ecological force, occurred along the geographic range of their distribution as evidenced by the results from La Mancha and Agaltepec Island that have different biophysical characteristics and are 188 km apart. Facilitation also occurred throughout the 2-year period of both study sites. Similar results were observed at Pipiapan, Amamaloya, and Mimiahua sites where M. glabra’s presence facilitated the orchid pollination. These findings were consistent throughout the 4 years of systematic monitoring.

Natural habitat conditions and heterospecific or monospecific clumps affect fruit set in non-rewarding orchids (Sabat and Ackerman 1996; Neiland and Wilcock 1998; Johnson and Nilsson 1999). Several studies indicate that the frequency of rewarding and non-rewarding plants (e.g. Internicola et al. 2006), pollinator density (Johnson et al. 2003a, b), floral display (Juillet et al. 2007), and spatial distribution (Johnson et al. 2003a) directly affect the fruit production of a rewardless species. Thus, the greater number of fruits on Agaltepec Island relative to La Mancha likely results from these factors. On Agaltepec Island (8.3 ha), M. glabra forms larger clumps and is more abundant (1,400 individuals) than at La Mancha (30 ha) (341 individuals), so Agaltepec Island provides both more abundant and denser rewards to pollinators. A small habitat facilitates foraging on rewarding species by pollinators and, therefore, pollen or pollinarium flow (Johnson 1994; Roy and Widmer 1999). Under these circumstances, female Centris bees can visit more flowers of M. glabra and O. cosymbephorum per unit of time.

The hypothesis of Batesian floral mimicry offers an alternative explanation to the magnet species effect for our results (Dafni and Ivri 1981a; Johnson 1994, 2000). In a Batesian floral mimicry system, the fitness of the mimic should be higher in the presence of the model than in its absence (see Johnson 1994, 2000; Roy and Widmer 1999). Our data allow us to conclude that O. cosymbephorum depend upon M. glabra, but they do not allow to insure that one of the plant species is the mimic and the other is the model. The presence of M. glabra is a determinant of fruit production by the orchid on both spatial and temporal scales. We found that natural monospecific clumps of O. cosymbephorum that had not produced any fruit for two consecutive years, suddenly produced fruits when cultivated flowering individuals of M. glabra were in their vicinity. This suggests that some kind of particular interaction occurs between O. cosymbephorum and M. glabra. Oncidium cosymbephorum is sympatric with M. glabra, it has lower plant frequency, its flowering time overlaps that of M. glabra, it shares pollinators with M. glabra, and its flowers resemble those of M. glabra in both color and morphology (Carmona-Díaz 2001). When Centris bees visited M. glabra flowers, they included the flowers of O. cosymbephorum on their foraging route. Our results of the increase in fruit set, indirectly suggest that bees do mistake the flowers of O. cosymbephorum for those of M. glabra. Not all bee-visited flowers are expected to be successfully pollinated because fertilization only takes place if the pollen is cross-compatible (Carmona-Díaz 2001).

A few cases of floral mimicry have provided evidence that some rewardless orchids produce more fruits when grown adjacent to their model plant (Dafni and Ivri 1981a, b; Johnson 1994, 2000). It has been argued that in some Batesian floral mimicry systems, the resemblance between model and mimic is so strong that pollinators fail to distinguish between them (Johnson 1994, 2000). However, these studies did not explore spatial and temporal variation or constancy. Our results provide experimental evidence for pollination facilitation and suggest that the study system is a possible case of Batesian floral mimicry.

Future field studies and specific experiments are necessary to address whether the facilitation of pollination that we have documented here is an adaptive similarity of the rewardless plant species, or simply a general phenomenon that can occur even when the flowers of the orchid do not resemble those of the nearby rewarding species.