Atypical distyly in Psychotria goyazensis Mull. Arg. (Rubiaceae), an intramorph self-compatible species

Distyly is a genetically controlled fl oral dimorphism, characterized by the reciprocal positioning of pin and thrum morphs, a heteromorphic incompatibility system and a balanced morph ratio (isoplethy). Th is study aimed to investigate the phenology, features of distyly, reproductive biology, and fl oral visitors of Psychotria goyazensis Mull. Arg. (Rubiaceae), an intramorph self-compatible species, in a tropical semideciduous forest in the municipality of Catalão, located in the state of Goiás, Brazil. Th e phenology was recorded for a sample of individuals, the fl oral morphology being described using fresh and fi xed material. Th e incompatibility system was tested using controlled hand pollinations. Floral visitors were collected and identifi ed. Blooming began in November 2009 and ended in April 2010. Fruits developed from December 2009 to March 2010. Th e population presented atypical distyly, because only plants with pin fl owers were found. Th e intramorph and open pollinations set more fruits than did spontaneous self-pollinations. Th e main fl ower visitors were Systropus sp. (Diptera), as well as Hymenoptera and Lepidoptera species. Morphological variations in distyly can occur by breakdown/permutation of the gene that controls the polymorphism. However, some authors argue that, in Rubiaceae, genetic control of the incompatibility system can operate independently of the morphological factor, and species might respond in diff erent


Introduction
Heterostyly is genetically controlled fl oral polymorphism, characterized by reciprocal positioning of stigmas and anthers between morphs (reciprocal herkogamy), as well as by an intramorph self-incompatibility system (Ganders 1979). Th e most common manifestation of heterostyly is distyly, in which the two fl oral morphs, pin fl owers (stigma above the anthers) and thrum fl owers (stigma below the anthers), occur separately (Hamilton 1990). Th ese morphs are usually found in similar proportions within populations (isoplethy); these features are controlled by genetic and ecological mechanisms, such as the incompatibility system and intermorph pollen transfer (Ganders 1979).
In some botanical families, the incompatibility system and fl oral morphs are controlled by a single genetic component, usually referred to as a supergene, whereby the thrum morph presents the heterozygous genotype and the pin morph presents the recessive homozygous genotype (Barrett 1990). However, some authors argue that, in certain families (especially Rubiaceae), the morphological and physiological integrity of the supergene diff ers from that seen in other families (Richards 1986;Lewis & Jones 1992; Barrett & Shore 2008;Consolaro et al. 2011). Some Rubiaceae genera, such as Psychotria, Palicourea, Coccocypselum, Declieuxia, Sabicea, Guettarda and Manettia, are known to be typically distylous but present anomalies or breakdown of distyly (Sobrevila et al. 1983;Hamilton 1990;Richards & Koptur 1993;Passos & Sazima 1995;Faivre & McDade 2001, Coelho & Barbosa 2003Consolaro et al. 2009).
It is estimated that in Brazilian savanna (cerrado), distylous species represent approximately 41% of the Rubiaceae taxa (H. Consolaro, personal observation). In the Psychotrieae tribe, distyly seems to be a primitive feature and is taxonomically widespread, other reproductive systems in the tribe therefore being attributed to some evolutionary variation of distyly (Hamilton 1990). Because of diff erent evolutionary and ecological implications, disturbance in the plant-pollinator relationship and genetic changes in the distyly, the fl oral structures and the frequency of the morphs can vary, resulting in the loss of one of the morphs (monomorphism) or a rearrangement of the reproductive structures, which can be placed at the same height (homostyly, Ganders 1979).
Plants show many diff erent phenological strategies to avoid overlap between blooming and fruiting, as well as to compete for pollinators and seed dispersers, given that these adaptations can improve the fi tness and survival of these species (Monasterio & Sarmiento, 1976;Bullock 1985). In recent decades, the number of studies on heterostyly and phenology of heterostylous species in the cerrado has been increasing (Monteiro et al. 1991;Coelho & Barbosa 2003;Coelho & Barbosa 2004;Consolaro et al. 2005;Toledo et al. 2007;Lenza et al. 2008;Oliveira 2008;Consolaro et al. 2009;Machado et al. 2010;Consolaro et al. 2011) and could be compared to those conducted in the Atlantic Forest (Passos & Sazima 1995;Lopes & Buzato 2005;Castro & Oliveira 2002;Castro et al. 2004;Teixeira & Machado 2004a, 2004bRossi et al. 2005;Pereira et al. 2006;Mendonça & Anjos 2006;Virillo et al. 2007, Fonseca et al. 2008. To increase understanding of the heterostyly system in the cerrado, this study aimed to investigate the phenology, distylous features, reproductive biology and fl oral visitors of Psychotria goyazensis Mull. Arg. (Rubiaceae).

Study site
Th e study was carried out from April 2009 to April 2010 in a forest fragment of approximately 26 ha in the municipality of Catalão, located in the state of Goiás, Brazil (18°9'36.595"S; 47°55'33,989"W). According to the Köppen classifi cation system (Köppen 1958), the climate of the study area is type Aw, with a dry season from May to September and a rainy season from October to April.

Phenology
We marked 45 plants, with at least 5 m between individuals. During the reproductive phase (fl owering and fruiting), observations were made on a weekly basis, whereas they were made once every two weeks during the vegetative phase. Th e development of fl ower buds, the appearance of open fl owers (blossoming) and fruiting were recorded. To evaluate the phenological phases, we employed the activity and intensity indices devised by Bencke & Morellato (2002). Th e activity index analyzes the presence or absence of each phenological phase (fl ower buds, fl owers and fruits), evaluating the synchrony of the phenological events among the plants. Th e intensity index uses the classes established by Fournier (1974), showing the peak intensity of each phenological event on a fi ve-point scale: 0 = 0%; 1 = 25%; 2 = 50%; 3 = 75%; and 4 = 100%. Phenological patterns were classifi ed using the categories described by Newstron et al. (1994).

Floral biology
Forty pre-anthesis buds were marked on 10 random plants (4 buds per plant). Th e fl oral longevity, the release of pollen by the anthers and the beginning of anthesis were recorded. Anthesis onset was defi ned as the beginning of the opening of corolla lobes, and the end of fl oral longevity was defi ned as a loss of brightness and a change in the color of the corolla. Th e release of pollen was determined by observation of loose pollen on the anthers. Stigma receptivity was determined by the separation of the stigma lobes (n = 16), which is usually suitable for studies of Rubiaceae (Coelho & Barbosa 2003, Consolaro et al. 2005, Consolaro et al. 2009, Consolaro et al. 2011.

Distyly and incompatibility system
Five fl owers per individual were collected from 20 plants. Th e lengths of the corolla tube, stamens, pistil and stigma lobes were measured under a stereomicroscope and are expressed as mean ± standard error of mean. Th e fruit set resulting from controlled hand pollination was used in order to determine the incompatibility system. In 27 plants, fl ower buds in pre-anthesis (n = 172) were isolated and bagged. When the fl ower buds opened, the following treatments were applied: self-pollination (exposure to pollen from the same fl ower, n = 41); intramorph cross-pollination (exposure to pollen from fl owers of diff erent plants of the same morph, n = 42); spontaneous self-pollination (fl owers tagged and bagged, n = 48); and control (fl owers marked and exposed to open pollination, n = 41). We did not expose fl owers to intermorph cross-pollination, because the studied population was strictly monomorphic. Th e index of self-incompatibility (ISI) was calculated as the ratio between the proportion of fruit production by selfi ng and that resulting from intramorph cross-pollinations. An ISI greater than 0.25 was considered indicative of self-compatibility (adapted from Bullock 1985).

Floral visitors
During four days at the blooming season, we observed the fl ower visitors of 14 plants. Observations were intercalated at the morning and at the aft ernoon (from 06:00h to 16:00h), with four hours of duration and a total of 24h of observation. Th e fl oral visitors were categorized as nectar robbers (those "robbing" nectar from outside the corolla, i.e., without making contact with the anthers or stigma) or pollinators (those taking nectar or pollen and making contact with the anthers and stigma). Th e frequency and type of approach to the fl owers were also recorded, and the visitors were collected and identifi ed. Th e reproductive effi ciency index was calculated as the ratio between the control (open pollination) fruit set and the intramorph cross-pollination fruit set (adapted from Ruiz & Arroyo 1978).

Statistical analysis
A two-tailed t-test was used in order to identify diff erences between the anthers and stigma heights, and a chi-square test was applied in order to compare the fruit-set frequency between the hand pollinations treatments. Data analysis was performed using BioEstat 3.0 soft ware (Ayres et al. 2003).

Phenology
Th e reproductive phenology of P. goyazensis showed an annual intermediate pattern (sensu Newstron et al. 1994): it presented only one fl owering cycle over a year with an approximate duration of four months (Fig. 1). Buds began to appear in November and continued to appear until the end of February, and fl owering began at the end of November 2009 and continued until April 2010, both phenological phases occurring during the rainy season. During the fl owering peak, we noted that the proportion of plants with open fl owers in the population (25.58%) was substantially lower than the observed proportion of plants showing fl oral buds (73.78%). As can be seen in Fig. 1, fruit production occurred concurrently with fl owering, beginning in December 2009, the fruits maturing in March 2010 (at the beginning of the dry season). At all stages observed, the synchrony between individuals was approximately 41.33%, although the proportion of plants with evidence of phenological phases ranged from 11.71% to 100%.

Floral biology
Psychotria goyazensis has white tetramerous or pentamerous fl owers (Fig. 2), arranged in cymose infl orescences. Th ey lasted approximately 12 h, presenting nectar and pollen as fl oral resources. Inside the corolla, there is a ring of trichomes arranged in the middle portion of the tube, above which are the four stamens. Anthesis began at 05:00h fl owers typically being fully opened by 06:00h, at which time stigmas became receptive and pollen release began.

Distyly and isoplethy
Psychotria goyazensis presented a case of atypical distyly, because only the pin morph was found in the population (pin monomorphism). Th e pin morph fl owers had a mean corolla tube length of approximately 3.1 ± 0.33 mm, mean stamen height of 2.77 ± 0.02 mm, mean pistil height of 3.27 ± 0.03 mm and mean stigma lobe length of 1.9 ± 0.32 mm. Th e diff erence among the heights of the reproductive structures was signifi cant (t 2,93 =16.48, p<0.01), characterizing the pin morph. Th e mean anther-stigma separation was 0.5 ± 0.03 mm, with an apparent variation in herkogamy between fl owers. A few fl owers presented pistils at approximately the same level as or even slightly below the anthers (Fig. 3).

Incompatibility system
Psychotria goyazensis showed intramorph and self--compatibility, with an index of self-incompatibility of 0.96. As can be seen in Tab. 1, there was no signifi cant diff erence between self-pollination and intramorph cross-pollination in terms of fruit set (χ 2 =0.017, p>0.05). We also found that the fruit set resulting from spontaneous self-pollination did not diff er signifi cantly from that resulting from the cross-pollination and open pollination (control) treatments (reproductive efficiency index=1.19, χ 2 =0.38, p=0.61). Fruit set was observed 15 to 20 days aft er controlled hand pollinations.

Floral visitors
Only insects were observed visiting the fl owers of P. goyazensis. Most of the visits (52%, n = 60) were made by Diptera (Systropus sp., mimics of Hymenoptera, 43% of the visits being made by fl ies of the Conopidae and Syrphidae families), bees (Halictidae, 2 species), wasps (Polistinae, 2 species), and butterfl ies (Nymphalidae and Danainae). Th e visits were legitimately performed, meaning that the visitors apparently touched the anthers and stigmas with the front of their body and were primarily seeking nectar. We identifi ed no nectar robbers. Although visits were recorded throughout the day, they were more frequent at times when there was a higher incidence of sunlight on plants (between 9:00h and 11:00h). Pollinators visited many fl owers on the same plant and fl owers on plants that were closer to each other.

Discussion
The blooming pattern of Rubiaceae and Psychotria species in the cerrado often seems to be linked to the rainy season in this ecosystem (Martin-Gajardo & Morellato 2003, Lopez & Buzato 2005, Consolaro et al. 2011. However, the bloom period and the fructification of many species can result not only from abiotic selective pressures, such as the climate system, but also as a way to enhance survival strategies when the species will be able to disperse and germinate its seeds (Sarmiento & Monasterio 1983). The synchrony among the studied plants followed the intensity of occurrence of the phenological phases observed, a pattern found in other Psychotria species (Augspurger 1983;Coelho & Barbosa 2004). The timing of a phenological phase, such as flowering and fruiting, is closely related to endogenous factors and abiotic factors that influence the physiology of plants (Bencke & Morellato 2002). Nevertheless, in an ecological context, this fact is important because it contributes to the reproductive process of the species, increasing the power of attraction and the maintenance of pollinators and seed dispersers in the habitat.
In the study area, we recorded fi ve other Psychotria species that bloom simultaneously with P. goyazensis, to which they are also similar in terms of morphology and fl oral biology: P. defl exa, P. prunifolia, P. hoff mannseggiana, P. trichophoroides and P. capitata (H. Consolaro, personal observation). Th e coexistence of plants whose fl owering times overlap could have a facilitating eff ect on pollination, increasing the attraction of pollinators to neighboring spe-cies (Feldman et al. 2004). However, it can also expose these species to interspecifi c competition for pollinators, having a negative impact on pollen transfer and fruit production (Pleasants 1980).
Because the number of studies on the reproductive biology of Psychotria species is increasing, it is oft en possible to fi nd the occurrence of monomorphic populations, especially pin species (Ganders 1979, Sakai & Wright 2008Consolaro et al. 2011). Th e same pin monomorphism found in P. goyazensis has been also reported in P. carthagenensis (Con-  solaro et al. 2011) and P. gracilifl ora, P. hoff mannseggiana, P. micrantha and P. tenuifolia (Sakai Wright & 2008). Four of the fi ve Psychotria species in the study area are typically distylous (P. hoff mannseggiana, P. capitata, P. defl exa and P. trichophoroides), although P. prunifolia is also pin monomorphic (E. Rodrigues, personal observation). A recent phylogenetic analysis showed that such unbalanced morph ratios can arise independently (Sakai & Wright 2008). Such ratios can also be observed among populations of the same species (Koch et al. 2010;Consolaro et al. 2011;Farias et al. 2012), as well as in congeneric species (Consolaro et al. 2009), although other species are relatively stable (e.g. Machado et al. 2010). In P. goyazensis, we were unable to determine how widespread pin monomorphism is among its populations.
Th ere is no concrete information about how the genetic control of distyly in Rubiaceae works, but studies of other families indicate that a diallelic system organized as a supergene controls the morphological and incompatibility features of distyly (Richards 1986;Lewis & Jones 1992;Barrett & Shore 2008). When a breakdown of distyly occurs by a recombination in the supergene (Lewis & Jones 1992), the most common result would be the formation of homostylous self-compatible fl owers, with reproductive structures at the same height (Ganders 1979;Consolaro et al. 2009). Th e breakdown of distyly can also be correlated with the ploidy level in some Rubiaceae species (Naiki &Nagamasu 2004;Naiki 2012). However, it is known that a gain in compatibility can occur without rearrangement of the reproductive structures (Sobrevila et al. 1983;Barrett 1988;Consolaro et al. 2011), which indicates that the action of the supergene in Rubiaceae is not very clear. Th e occurrence of pin monomorphism and self-compatibility in the studied population of P. goyazensis, as shown in P. carthagenensis (Consolaro et al. 2011), supports the idea that in Rubiaceae the control of the incompatibility system could act independently of the morphological factor.
Th e development of an autogamous reproductive mechanism becomes advantageous when pollination service and compatible pollen transfer is lost and seed production is defi cient (Schoen et al. 1996). Distylous species rely on an accurate intermorph pollen transfer due to the incompatibility system, so that the loss or laxity of this system becomes advantageous in case of disturbance of the plant-pollination relationship (Ganders 1979).
In the present study, P. goyazensis fl owers were mainly visited by fl ies, a rare situation for this genus that has bees as its main pollinators (Bawa & Beach 1983;Castro & Oliveira 2002). Variations in the type of pollinator among congeneric species might have a negative impact on intermorph pollen transfer, because the fl oral architecture might not favor contact with fl ower visitors (Consolaro et al. 2009), especially Syrphidae, which have a smaller body size and distinct behavior (Proctor et al. 1996). In addition, even when Diptera species visit the fl owers, the visits happen during specifi c periods of the day (between 08:00h and 12:00h), many Syrphidae species having been shown to present such behavior (Weems 1953). With visits of pollinators to P. goyazensis fl owers restricted to specifi c times of day, the pollination service could have become limited and insuffi cient. Th e irregular frequency of visits and instability in the pollen transfer might have favored the loss of incompatibility in the studied population.
Because there was a comparatively normal fruit set aft er spontaneous self-pollination treatment, we cannot exclude the possibility of apomixis in this population. However, in tiny fl owers, such as those of P. goyazensis, the anther-stigma separation might be insuffi cient to avoid spontaneous self--pollination. Th eir size also renders emasculation and agamospermy treatments impractical. Even herkogamy, which results in signifi cant diff erences in height of the reproductive structures, might not be suffi cient to prevent deposition of self-pollen on the pistil. Dai & Galloway (2011) suggested that there are cases in which herkogamy does not function as a system that prevents self-pollination but only a way to increase pollen transfer and outcrossing, which could be the case among fl owers of P. goyazensis.
Th e overlap between P. goyazensis and the fi ve other Psychotria species, in terms of blooming time; together with the fact that P. goyazensis has small Diptera as its main pollinators and a limited time for pollination as a function of the foraging activity of the fl oral visitors, as well as having tiny fl owers, might have contributed to the occurrence of variations in the reproductive system of the studied population. However, it was not possible to establish pin monomorphism or self-compatibility as features of the studied species, because the anomalies might be restricted to this population. Pin and thrum fl owers of P. goyazensis have been described from herbarium material (Delprete 2010), so it is possible that isoplethic and truly distylous populations occur elsewhere. For Rubiaceae, especially in studies of Psychotria, the genetic control of the incompatibility system and the fl oral morphology by a supergene does not seem to be well-defi ned. Th erefore, further specifi c studies in this family are needed in order to improve understanding of distyly and its variations in the Rubiaceae family, which presents a larger number of distylous species than does any other family of angiosperms.