Plant Syst. Evol. 232: 201–212 (2002)

Selfing facilitates reproductive isolation among three sympatric species of Pitcairnia (Bromeliaceae) T. Wendt1, M. B. F. Canela1, D. E. Klein1, and R. I. Rios2 1 2

Departamento de Botaˆnica, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil Departamento de Ecologia, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil

Received May 10, 2001 Accepted February 25, 2002

Abstract. The reproductive biology of three sympatric species Pitcairnia flammea, P. corcovadensis and P. albiflos (Bromeliaceae) was studied on Corcovado Mountain in the city of Rio de Janeiro, Brazil. Controlled pollination experiments were also conducted using plants transplanted to a greenhouse. The three species are self-compatible and interspecifically cross compatible. Pitcairnia corcovadensis was principally an autonomous self-pollinator, having scarce pollinator activity, anthers and stigmas at the same level, and absence of nectar. In P. flammea, many flowers opened per day, the stigma and anther were close to each other, and its nectar was regularly consumed by hummingbirds; all factors that promote geitonogamy in this species. Although fully self-compatible, P. albiflos always opened a few flowers per days showing an adaptation to cross-pollination. Observations during two flowering seasons showed that despite different times for peak flowering, blooming of these three species overlapped in April. They grow in mixed clumps that may facilitate promiscuous pollination between them but no intermediate forms were observed under natural conditions. The absence of temporal isolation, geographic isolation and isolation via post-pollination reproduction suggests that evolution toward selfing was important to avoid hybridization between these sympatric species.

Key words: Bromeliaceae, Pitcairnia, breeding systems, autogamy, interspecific crossing, phenology, pollination, reproductive isolation.

The comparison of self- and cross-fertilization is a central theme of floral biology (Lloyd and Schoen 1992). Following the discovery that cross-fertilization is advantageous because it produces superior progeny (Darwin 1872), there was a period of intense activity in pollination biology when botanists applied these principles to explain the many specialized features of angiosperm flowers as adaptations for the promotion of outcrossing. This tradition continues up to the present day (Faegri and Van der Pijl 1979). It was also acknowledged that some plants are adapted to regular self-fertilization and have floral syndromes that contrast strongly with those associated with outcrossing (Richards 1986). Self-fertilization provides reproductive assurance when cross-fertilization is inadequate or unreliable (Affre et al. 1995, Erhardt and Ja¨ggi 1995, Ortega Olivencia et al. 1995, Klips and Snow 1997). Repeated reductions in population size and/or pollinator abundance may favour the evolution of self-compatible and self-fertilizing forms (Stebbins 1957). Many studies support

202

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

the idea that species with limited distribution tend to be more self-compatible than their more widespread congeners (Karron 1991, Hamrick et al. 1991, Wyatt et al. 1992). In sympatric, congeneric species of flowering plants, autogamy could operate as a reproductive isolating barrier (Levin 1971). This study is part of ongoing research into the pollination ecology and variation in breeding systems among species of Pitcairnia (Bromeliaceae). Pitcairnia is the largest genus (260 spp.) in the subfamily Pitcairnioideae (Smith and Downs 1974). In Brazil, Pitcairnia species (38 spp.) form three distinct geographic groups in the Amazon, in central Brazil (Goia´s state), and in the Atlantic Forest (Wendt and Chamas 1997). In this paper, we focus our attention on three species (P. albiflos Herb., P. corcovadensis Wawra and P. flammea Lindl.), and provide an appraisal of the relative significance of outcrossing and selfing in these sympatric taxa. Pitcairnia albiflos and P. corcovadensis are rare herbs found in small isolated populations limited to a few localities in the state of Rio de Janeiro. Pitcairnia flammea is one of the morphologically most variable Pitcairnia species in Brazil and is widespread in the Atlantic Forest. It comprises a complex of six varieties. Pitcairnia corcovadensis was considered a variety of P. flammea, but was recently re-established as a species based on the fact that it occurs sympatrically with P. flammea in many locations, and it is morphologically distinct (Wendt et al. 2000a). Putative hybrids were also identified between P. albiflos and other congeneric species (Wendt et al. 2000b). In this study, we begin to verify which potential reproductive isolation mechanisms are effective in impeding gene flow among these congeneric herbs. The reproductive biology of most species of Pitcairnia is unknown. Studies of the reproductive biology of bromeliads are mostly descriptions of floral events and pollinator behaviour (Sazima et al. 1989, Arau´jo et al. 1994, Sazima et al. 1995, Sluys and Stotz 1995). Few studies have attempted to understand breeding systems in this family (e.g.

Martinelli 1997, Wendt et al. 2001). Although it is well known that bromeliads easily hybridize in cultivation (McWilliams 1974), little is known about natural hybridization in the field (Luther 1984, Gardner 1984, Wendt et al. 2000b). The aim of this paper is to (1) compare the reproductive biology of these species, (2) estimate the extent of autofertility, self-compatibility and interspecific compatibility; and (3) point out mechanisms of reproductive isolation between these species. Materials and methods The present study was carried out between January and August 1996, in January 1997 and in March 1998. The populations we studied are located on Corcovado Mountain in an urban tropical rain forest (National Park of Floresta da Tijuca) of the state of Rio de Janeiro, Brazil (22
57¢S, 43
13¢W; altitude c. 650 m). Corcovado Mountain is wellknown and attractive tourist site within the city of Rio de Janeiro. The climate is wet with a mean annual rainfall of 1500 mm and a mean annual temperature of 24
C (Niemer 1979). Trees dominate areas having a layer of soil, but granitic outcrops are colonized by saxicolous herbs. Bromeliads are important physiognomic elements of the outcrops. We studied the three saxicolous, closely related species of Pitcairnia that grow side by side, often in mixed clumps, on the slope alongside the road named Estrada do Redentor (Fig. 1). Pitcairnia flammea and P. corcovadensis were very abundant in this location, but P. albiflos was represented by only few individuals. All species showed marked vegetative reproduction by offshoots with each ramet blooming only once per lifetime. The flowers of the three species are hermaphroditic. They are borne on a single racemose inflorescence. The perianth is dichlamydeous, and consists of three free sepals and three free petals held close together similar to a corolla tube. Each has six stamens and a one-pistil gynoecium. The ovary is semi-inferior with three locules. Pitcairnia flammea and P. corcovadensis have red flowers. The petals remain erect during anthesis but converge upwards and overlap, giving the flower a zygomorphic symmetry. This pattern is frequently observed in the genus. Pitcairnia albiflos has white flowers with spiralled petals but maintains an actinomorphic symmetry during anthesis (Wendt 1994).

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

203

Fig. 1. Distribution of the three species of Pitcairnia along the road, Estrada do Redentor on Corcovado Mountain. The ellipses represent the clumps of the species formed by clonal growth. The different sizes of the ellipses are correlated with an estimation of the number of plant in each clump. Curves in the road are not shown in this figure

Phenological data were derived from two seasons of field observations. Phases of anthesis were observed in situ and in the greenhouse. Inflorescence and floral traits (Table 1) were quantified using fresh material. Nectar volume was measured by sampling previously bagged flowers using calibrated microcapillary tubes. The pollen grains from two anthers per flower were placed on separate microscope slides. One of the slides was stained with buffalo-black and the other with acetocarmine. For each anther, the viability of 500 pollen grains was recorded by counting the number of viable (stained) and malformed (unstained) pollen. Periodic observations of visits by pollinators were made for each species during flowering

(March to June, 1996 and 1997) to identify pollinators and their foraging behaviour. Observation periods were divided into 1 to 4 h periods from 07:00h to 17:00h and evenly distributed among patches of plants. Insects that visited flowers were collected for identification. Bird pollinators were photographed for identification. Nocturnal observations were not conduced because the area is unsafe at night. Controlled pollination experiments during the 1997 flowering season were conducted in a greenhouse using plants transplanted from the field. The plants were transplanted at the onset of inflorescence formation. Most of them flowered and fruited normally. Only the plants used for evaluation of natural pollination (controls) were left in the

Table 1. Reproductive features of three species of Pitcairnia from Corcovado Mountain, Rio de Janeiro, Brazil

Scape length (cm) Number of flowers per inflorescence (range) Number of opened flowers per day (range) Corolla length (cm) Diameter of corolla mouth (cm) Pistil length (cm) Stigma-anther separation (cm) Nectar volume (ll) Sugar concentration (%) Pollen viability buffalo-black carmine acetic

P. corcovadensis

P. flammea

P. albiflos

N

X ± SD

N

X ± SD

N

X ± SD

25 25

19 19

25 25 25 25 29 22

88.78 ± 17.12 58.41 ± 21.17 (33–118) 11.67 ± 5.8 (1–22) 5.93 ± 0.30 0.74 ± 0.10 5.90 ± 0.27 0.36 ± 0.19 14.28 ± 5.98 14.20 ± 3.48

10 10

47 47 47 47 24 44

31.92 ± 6.42 5.32 ± 1.62 (3–9) 2.04 ± 1.05 (1–4) 6.11 ± 0.36 0.86 ± 0.10 5.86 ± 0.36 0.04 ± 0.19 0.92 ± 1.72 13.06 ± 4.51

8 8 8 8 8 37

72.50 ± 6.89 18.10 ± 5.85 (15–30) 1.60 ± 0.51 (1–2) 5.62 ± 0.44 2.90 ± 0.40 6.35 ± 0.18 0.65 ± 0.18 7.06 ± 2.43 12.72 ± 0.80

5 5

87.68 ± 6.03 88.40 ± 9.77

3 3

98.14 ± 0.42 98.31 ± 0.11

6 6

86.19 ± 6.48 86.54 ± 6.92

25

19

10

204

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

field throughout the experiments. Each species produced different numbers of flowers per inflorescence, therefore different numbers of plants and flowers were used for the controlled pollination tests: for P. corcovadensis, 142 flower buds (105 hand-manipulated) from 50 plants (41 in the greenhouse); for P. flammea, 492 flower buds (189 hand-manipulated) from 14 plants (9 in the greenhouse); and for P. albiflos, 222 flower buds (50 hand-manipulated) from 14 plants (5 in the greenhouse). Different treatments were often applied on the same inflorescence. Each flower received one of the following treatments: (1) Agamospermy – Flower buds were emasculated and the inflorescence enclosed in a paper bag to identify agamospermy. (2) Self-pollination – Inflorescences with flower buds were bagged, and when flowers opened they were hand-pollinated using fresh pollen obtained from the same flower to determine if they were selfpollinating. (3) Autonomous self-pollination – Inflorescences with flower buds were enclosed as above and left unmanipulated to determine the level of autonomous self-pollination. (4) Crosspollination – Flower buds were emasculated and the inflorescence bagged. When flowers opened they were hand-pollinated using fresh pollen obtained from another plant of the same species to determine if the level of cross-pollination. We made sure that the two plants were from different areas of the field to avoid using ramets of the same clone. (5) Interspecific cross-pollination – Flowers buds were manipulated as for cross-pollination except that flowers were hand-pollinated using fresh pollen obtained from flowers of the other species to determine if the species crosses interspecifically. (6) Natural pollination – Since pedicels remain on the inflorescence regardless of whether or not a fruit is produced, the number of pollinated and non-pollinated flowers can be quantified for each plant by simply counting the pedicels with or without fruits. The paper bags were removed to prevent accidental injury to inflorescences two days after the last flowers of each inflorescence opened. Each inflorescence in the greenhouse was labelled and the flower sepals were colour coded using acrylic paint. Fruit development was monitored periodically until maturity. Evaluation of fruiting success was based on counts of mature fruits. The fruits were collected at the onset of capsule dehiscence. The

number of seeds per fruit was determined for all mature fruits (except those that were preyed upon or dehisced before collection) from the six handpollination treatments, and for a sample from the field to represent natural pollination (33 fruits of P. corcovadensis, 18 fruits of P. flammea, and 18 fruits of P. albiflos). The seeds of Pitcairnia spp. are small. They were counted on a paper grid using a hand tally counter. We could easily discriminate between fully developed and aborted or unfertilized seeds. The total seed weights of both viable and aborted seeds per fruit were weighed on precision scales. Statistical analyses were performed to compare seed production success among the pollination treatments for each species. The mean numbers of viable and aborted seeds per fruit, and the means of total weight of seeds per fruit were compared with one-way ANOVAs. Differences between treatments were subsequently tested using Unequal NHSD multiple comparison tests. Statistical tests followed Statistica 4.2 procedures (Statsoft Inc., 1993). Indices of autofertility and self-compatibility were calculated for each species (Lloyd and Schoen 1992). Autofertility was calculated as percentage fruit set (AFIf) or mean viable seed per fruit (AFIs) from autonomous self-pollination versus crosspollination. These parameters measure the ability of flowers to self in the absence of pollinators. Each measurement of autofertility was estimated relative to the fruit or seed set after hand outcrossing. Two measurement of self-compatibility (SCI) were calculated based on percentage fruit set (SCIf) and mean viable seed per fruit (SCIs) from selfpollination relative to the values from cross-pollination. A value of one is interpreted as complete self-compatibility; a value less than 0.75 is interpreted as being due to at least partial self-incompatibility (Lloyd and Schoen 1992). Both fruit and seed set were used because they reflect self-incompatibility and inbreeding depression effects at different levels in the hierarchy of reproductive output (Jacquemart 1996). We also propose an ‘interspecific-compatibility index’ (ICI) to interpret the results of the interspecific crosses. Two measurements of interspecific compatibility were estimated, based on fruit and seed set from interspecific cross-pollination relative to those from cross-pollination. A value of one is interpreted as absence of interspecific incompatibility; and a value less than 0.75

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia suggests at least partial interspecifc incompatibility. We compared germination rates among treatments for species. For each pollination treatment, seeds from one mature fruit of each treatment were planted on a tree-fern trunk substrate in aluminium containers (21 · 16 · 5 cm) in a greenhouse. One container was used for each fruit. Each fruit contained c. 100–500 seeds per fruit. The containers were covered with plastic film to avoid seed loss by wind and promote high humidity at the onset of germination. A commercial plant insecticide was sprayed over each container before covering to prevent insect attack. The number of germinating seeds was estimated by counting the seedlings over a period of six months and was recorded as percentage germination of the total number of viable seeds planted from each fruit from each treatment.

Results Phenology and floral biology. The blooming period of the three Pitcairnia species lasted from late January to June during both 1996 and 1997. The peak time of flowering differed slightly for each species: mid-March to midApril for P. corcovadensis; April to mid-May for P. flammea; with only a few individuals of P. albiflos flowering in April and June 1996, and February and April 1997. In April all species flowered simultaneously. Diurnal anthesis occurred in P. corcovadensis and P. flammea from 6:00–7:00 h. Flowers lasted a single day. The red, scentless flowers of these species are nearly identical in structure but species differ in inflorescence size, number of flowers per inflorescence, and number of flowers opened per day (Table 1). The separation of stigma and anthers is practically absent in P. corcovadensis (0.04 ± 0.19 cm), while in P. flammea the stigma is slightly separate from the anthers (0.36 ± 0.16 cm). The proximity of stigmas and anthers in both species probably allows self-pollination. We also observed that as flowers wilted their petals twisted causing the anthers to touch the stigma; perhaps leading to delayed selfing. The sugar concentration in the

205

nectar of P. corcovadensis (13.06 ± 4.51%) and P. flammea (14.20 ± 3.48%) was similar, but the volumes were very different. Flowers of P. corcovadensis frequently had no or very little nectar (0.92 ± 1.75 ll) but flowers of P. flammea had more nectar per flower (14.28 ± 5.98 ll). The disparity in the quantity of nectar between these species is even greater if one considers that few flowers open per day in P. corcovadensis (2.04 ± 1.05) when compared to P. flammea (11.67 ± 5.80). Pitcairnia albiflos differed from the other two species in flower and inflorescence features (Table 1). It showed nocturnal anthesis (18:00– 23:00 h) and white, sweet-scented flowers that last for a single day and night. This species is herkogamous with larger separation between stigmas and anther than the other species (0.65 ± 0.18 cm). As flowers wilt, the stigmas and anther remain separated in this species. It also produced nectar (7.06 ± 2.43 ll) with concentrations of sugar similar to the other two species (12.72 ± 0.80%). Pollen was dehisced from all three species as soon as their flowers opened. The stigmas became turgid shortly after anthesis and remained so during the following day. Pollen viability was high for all three species (Table 1). Flower visitors. Three different hummingbirds, Thalurania glaucopis, Phaethornis cf. petrei petrei, and an unidentified species were observed to visit P. flammea. Hummingbird males of T. glaucopis were the most aggressive species in defending nectar sources against intruders. The behaviour of T. glaucopis was observed for a total of 12 h over a four-day period. This hummingbird systematically visited each inflorescence every 30 min during the entire day (7:00–17:00). It was slightly more active in the early hours of the day when visiting intervals were shorter between visits (20 min). During these visits, stigmas and anthers would touch the top of head of the bird while the bird hovered to feed on nectar from different flowers of the same inflorescence. A few visits by halictidaen bees were observed on P. flammea, collecting pollen and also touching the stigma.

206

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

Pitcairnia corcovadensis was observed for 30 h over a period of seven days (7:00–17:00). Despite similarity in flower colour and shape between P. corcovadensis and P. flammea, we observed only a single brief hummingbird visit to three individuals of P. corcovadensis. Bee visitation did not occur in this species. The small size of the P. albiflos population and scantiness of blooming plants in the study area hindered field observations. The bee, Trigona spinipes, was a commonly observed visitor. Groups of trigonas (3–8 bees) visited flowers of P. albiflos, took pollen and nectar, and entered the corolla tube, but were not observed to touch the stigmatic surface during visits. In another study of a population of P. albiflos, trigonas were also the most common diurnal visitors with night-time visits to the flowers by hawk moths (Sphingidae) and bats (Glossophaga, Lonchophylla; Wendt et al. 2001). Breeding systems. Table 2 shows the fruiting percentages for each taxon resulting from different pollination treatments. The results from emasculated and isolated buds suggest that agamospermy can occur in P. corcovadensis (32%) and in P. flammea (10.5%), but does not occur in P. albiflos (0%). Fruiting

percentages for self-pollinated and interspecific cross-pollination treatments were high for all species (>70%). Fruit production through autonomous self-pollination was higher in P. corcovadensis (68.6%), when compared to P. flammea (35.5%) and P. albiflos (28.5%). P. corcovadensis (89.1%) and P. flammea (82.2%) showed relatively high fruit production under natural pollination, with lower values for P. albiflos (34.3%). Conversely when cross-pollinated, P. albiflos (100%) had far higher fruit set than P. corcovadensis (43.7%) and P. flammea (46.6%). Table 3 summarizes the success of seed production of the seven pollination treatments for each species. Pitcairnia corcovadensis and P. flammea showed a similar pattern: the higher the number of viable seeds, the greater the seed weight and the lower the number of aborted seeds. For both species, the highest seed production resulted from natural pollination, followed by cross- and self-pollination treatments. The mean number of aborted seeds exceeded that of viable seeds for both species, under agamospermy, autonomous selfing and interspecific crossing with P. albiflos. Agamospermy resulted in the lowest viable seed production. Pitcairnia albiflos showed a

Table 2. Percentage of fruit set after pollination treatments within and between species. The number of fruits formed and the number of flowers tested are shown in parentheses (fruits/flowers). The number of plants indicates how many different inflorescences were available for each treatment. The species named in first column indicates the pollen donor for the interspecific pollination Treatment

Agamospermy Self-pollination Autonomous selfpollination Cross-pollination Interspecific crossing P. corcovadensis P. flammea P. albiflos Natural pollination

P. corcovadensis

P. flammea

P. albiflos

No. of plants

No. of plants

Fruit set

No. of plants

Fruit set

Fruit set

15 9 27

32.0 (8/25) 80.0 (8/10) 68.6 (35/51)

7 7 4

10.5 (4/38) 73.3 (44/60) 35.5 (11/31)

2 5 1

14

43.7 (7/16)

4

46.6 (7/15)

2

100 (4/4)

–

– 83.3 (5/6) 100 (3/3) 89.1 (33/37)

3 – 2 5

70.9 (22/31) – 92.8 (13/14) 82.2 (249/303)

2 3 – 9

100 (3/3) 85.7 (6/7) – 34.3 (59/172)

6 1 9

0 (0/10) 77.7 (14/18) 28.5 (2/7)

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

207

Table 3. Pollination treatments in P. corcovadensis, P. flammea, and P. albiflos. Average number of viable seeds per fruit, aborted seeds per fruit, and total seed weight per fruit (means ± 1 SD) were recorded after pollination treatments within and between species. Tests of analysis of variance (ANOVA, Unequal N HSD Test) were performed between treatments within each taxa. Values followed by different superscript letters within a column for each species are significantly different Treatment P. corcovadensis Agamospermy Self Autonomous self Crossing · P. flammea · P. albiflos Natural

P. flammea Agamospermy Self Autonomous self Crossing · P. corcovadensis · P. albiflos Natural

P. albiflos Agamospermy Self Autonomous self Crossing · P. corcovadensis · P. flammea Natural

Number of fruits

Viable seeds per fruit

8 8 35 7 5 3 33

86.7 301.7 266.3 437.4 361.4 308.3 570.7

df = 92

F = 10.40***

3 42 10 7 22 13 18

112.0 450.8 176.5 514.0 396.3 274.3 632.9

df = 108

F = 12.39***

F = 11.29***

F = 13.40***

0 13 2 3 3 6 18

– 294.3 6.5 251.3 141.3 242.3 113.2

– 215.3 54.0 299.6 452.6 321.6 416.9

– 35.2 2.5 16.0 22.9 22.3 16.3

df = 39

F = 4.09

± ± ± ± ± ± ±

± ± ± ± ± ± ±

± ± ± ± ± ±

92.1a 174.0ab 144.3ac 274.0bc 147.6ab 174.6ab 249.6b

87.2ab 178.0b 193.9a 137.4bc 164.2ab 61.4ab 197.0c

152.3b 0.7ab 95.8ab 59.6ab 107.1ab 133.8a

Aborted seeds per fruit 334.0 264.1 350.2 212.4 280.6 350.3 139.8

± ± ± ± ± ± ±

116.3ab 144.5ab 170.9a 264.3ab 196.8ab 127.6ab 134.0e

F = 5.37*** 248.3 246.5 435.9 222.0 253.2 475.5 221.6

± ± ± ± ± ± ±

± ± ± ± ± ±

159.7ac 91.2a 142.8bc 148.5a 74.7a 87.5bc 170.7a

137.5a 11.3ab 32.1ab 220.9ab 153.8ab 126.8b

F = 5.49*

Total seed weight per fruit (mg) 9.32 22.5 22.4 40.2 31.2 16.4 42.5

± ± ± ± ± ± ±

8.6a 11.3ab 9.4ac 23.9bc 11.2ab 8.5ab 19.1b

F = 9.34*** 14.5 49.5 19.5 61.6 42.7 30.6 77.3

± ± ± ± ± ± ±

± ± ± ± ± ±

10.9ab 22.5b 17.9a 15.4bc 18.5ab 6.1ab 25.3c

29.2a 2.1a 10.1a 9.5a 9.1a 12.8a

F = 2.08

*** P < 0.0001, ** P < 0.001, * P < 0.01

more distinctive pattern of seed production with greater variation between treatments. The absence of pollinators (autonomous selfing) drastically affected seed production. For most treatments, the mean number of aborted seeds was higher than that of viable seeds except for self-pollination. One-way ANOVA showed significant differences between treatments for each species. For P. corcovadensis (Table 3), agamospermy and autonomous self-pollination were signifi-

cantly different from cross- and natural-pollination treatments in regard to viable seed and seed weight. The number of aborted seeds per fruit differed significantly only between autonomous self-pollination and natural pollination. For P. flammea (Table 3), natural pollination was significantly different from most treatments (agamospermy, self-, autonomous self-, and interspecific cross-pollination) in regard to viable seeds and total seed weight per fruit, but did not differ from cross-

208

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

Table 4. Autofertility, self-compatibility and interspecific-compatibility indices based on fruit set (f) and seed set (s) in autonomous self-, self-, cross- and interspecific cross-pollination Autofertility AFIf

P. corcovadensis P. flammea P. albiflos

1.52 0.76 0.28

AFIs

0.61 0.34 0.02

Self-compatibility Interspecific compatibility SCIf

1.82 1.57 0.77

SCIs

0.69 0.86 1.17

pollination. Autonomous self-pollination was significantly different from self-, cross- and natural pollination for all measured parameters. Pitcairnia albiflos (Table 3) showed significant differences between self- and natural pollination treatments in relation to viable seeds per fruit. Autonomous self-pollination showed a tendency to differ from all other treatments but was not statistically significant. This statistical insignificance could be related to the small number of replications of the autonomous self-pollination treatments. Measurements of autofertility based on fruit set (AFIf) and seed set (AFIs) showed that all three species were able to spontaneously set fruit and seeds in the absence of pollinators (Table 4). Pitcairnia corcovadensis showed the highest values for AFIf (1.52) and for AFIs (0.61). Pitcairnia flammea was intermediate (AFIf ¼ 0.76; AFIs ¼ 0.34). Autonomous self-pollination was lowest in P. albiflos (AFIf ¼ 0.28; AFIs ¼ 0.02). Indices of self-compatibility based on fruit set (SCIf) were greater than 1.0 for P. corcovadensis (1.82) and P. flammea (1.57), indicating that the two species are completely self-compatible. For P. albiflos, the SCIf (0.77) was only slightly above 0.75 that is interpreted as partial self-compatibility (Lloyd and Schoen 1992). Conversely, self-compatibility based on mean number of viable seeds (SCIs) was greater than 1.0 for P. albiflos (1.17), and less than 1.0 for both P. flammea (0.86) and P. corcovadensis (0.69), probably expressing an effect of inbreeding depression in the latter two species (Table 4).

ICIf (ICIs) P. corcovadensis

P. flammea

P. albiflos

– 1.52 (0.77) 1.00 (0.56)

1.90 (0.82) – 0.85 (0.96)

2.29 (0.70) 1.99 (0.53) –

Interspecific crosses were possible between these species. Indices of interspecific-compatibility based on fruit set (ICIf) were greater or equal to 1.0 in all but one case (P. albiflos vs. P. flammea; Table 4). Most indices based on seed number (ICIs) were lower than those based on fruit-set (ICIf); indicating postzygotic inviability in interspecific crosses. Most of these values were less than 0.75 which we interpreted as being due to at least partial interpecific incompatibility. Seed germination and viability. Table 5 shows germination percentages of seeds from different pollination treatments for each species. All available treatments produced viable seeds. Percentage germination was low (5%) for seeds produced by agamospermy in P. flammea. A high percentage germination (72.1%) was observed for seeds resulting from P. albiflos vs. P. flammea. Discussion Reproductive biology and breeding systems. Self-pollination occurs in several ways, which Lloyd and Schoen (1992) described as ‘‘modes of self-pollination’’. Hand pollination trials conducted in the greenhouse showed that the three species studied are all self-compatible and autofertile. Yet, index values showed that self-pollination occurs differently in these species. Based on our field observations on pollinators and floral morphology, P. corcovadensis showed the highest values of selfcompatibility and autofertility indices. During observation of this species, practically no

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

209

Table 5. Percentage of seed germination (number of seedlings formed/number of seeds used) for each species, in different pollination treatments Percentage of seed germination (seedlings/seeds)

Agamospermy Self-pollination Autonomous self-pollination Cross-pollination Interspecific cross-pollination P. corcovadensis P. flammea P. albiflos Natural pollination

P. corcovadensis

P. flammea

P. albiflos

52.8 35.0 21.3 24.7

5.0 23.5 53.5 73.3

(6/119) (130/552) (75/140) (508/693)

– 60.2 (115/191) not available not available

42.2 (82/194) – 37.7 (180/477) 24.7 (185/747)

53.7 (86/160) 72.1 (70/97) – 66.5 (137/206)

(101/191) (41/117) (74/346) (172/696)

– 70.9 (110/155) 53.3 (398/746) 33.1 (289/871)

pollinating visitors appeared on the flowers, even though many individuals flowered at the same time in the population. This absence of pollinator activity did not interfere with high fruit production under natural conditions. The tiny stigma-anther separation, and high percentage of fruit set on bagged flowers, suggests that this species exhibits autonomous self-pollination. These data reinforce the viewpoint that, under natural conditions, P. corcovadensis is more frequently selfing than the other two species. However, P. corcovadensis had the highest self-compatibility index based on fruit set and lowest value when the index was based on seed production. Pronounced inbreeding depression during seed development may explain these latter observations (Charlesworth and Charlesworth 1987, Lloyd 1992). Self-compatibility indices of P. flammea were also high. This species was systematically visited by the hummingbird Thalurania glaucopis whose behaviour promotes self-pollination and reduces the possibility of cross-pollination. Occasional visits by other hummingbird species were often prevented by the aggressive behaviour of Thalurania, but these visitors may also promote cross-pollination. The flower morphology of P. flammea suggests that delayed selfing could ensure some seed production. The autofertility indices of P. flammea were approximately one-half of the indices observed for P. corcovadensis suggest-

ing a greater dependence of the former on pollinator activity. The self-compatibility index of P. albiflos based on fruit set identifies it as a partially selfincompatible species. But when the index of seed production is used, the species appears to be self-compatible. Individuals of this species had few opened flowers per day. Anthers and stigma were separated during both anthesis and senescence of the flowers. Autonomous selfing should be rare and was confirmed by the low autofertility indices. Thus, P. albiflos is adapted to avoiding self-pollination. The small number of individuals and the separation between individuals within the population (that usually do not flower simultaneously) pose barriers to successful outcrossing. Individuals of P. albiflos must increase selfing in the absence of cross-pollen to compete with self-pollen. Thus, selfing may probably occur under unfavourable pollination conditions (Lloyd and Schoen 1992). Pitcairnia corcovadensis and P. flammea showed a low degree of asexual reproduction by agamospermy as evidenced by fruit set. Although agamospermy has recently been reported in Pitcairnia (Wendt et al. 2001), the possibility of contamination of stigmatic surfaces by pollen prior to flower opening cannot be discounted because of the proximity of stigmas and anthers in unopened flowers. Embryological studies are necessary to test rigorously for the occurrence of agamospermy

210

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

in the three studied species. In addition, our results show that selfing is the most important mechanism of reproduction in P. flammea, P. corcovadensis, and P. albiflos. Although outcrossing cannot be ruled out, it is probably of little importance in these populations given the behaviour of pollinators of P. flammea, the scarcity of pollinating visitors to flowers of P. corcovadensis, and unfavourable pollination conditions of small population size and scantiness of blooming individuals for P. albiflos. Reproductive isolation mechanisms. Isolating mechanisms in plants can be grouped into three main classes: 1) geographic isolation, 2) ecological isolation, and 3) reproductive isolation which can be divided into ‘external ’ (premating) and ÔinternalÕ (post-mating) processes (Grant 1981). The species studied in this report coexist in the same geographic area. Pitcairnia corcovadensis co-occurs at all known localities with P. flammea (Wendt et al. 2000a). Pitcairnia albiflos frequently co-occurs with these species or other congeners (Wendt et al. 2000b). These species are adapted to saxicolous growth on south-facing rocky outcrops. Geographic and ecological isolation are not isolating mechanisms for these species. Competition for the services of the same pollinators tends to occur in related species because they have similar floral signals and flowering periods (Levin 1971). Seasonal differences in flowering periods among sympatric species often contribute significantly to their temporal isolation. Slightly different peak flowering times were observed in P. corcovadensis and P. flammea, but all species had an overlapping flowering period. Sympatric taxa may achieve a high degree of isolation by opening their flowers at different times of the day. Diurnal differences are often associated with differences in pollinators. However, for diurnal isolation among sympatric congeners to be effective, periodicity must occur in stigma receptivity (Levin 1971). Pitcairnia corcovadensis and P. flammea have diurnal anthesis, but P. albiflos is nocturnal. Differences in the period of stigma receptivity apparently do not

occur among these species because all hand pollinations were conducted following a similar schedule. Ethological isolation is based upon the ability of pollinators to differentiate between floral signals (Levin 1971). Floral characteristics that attract specific groups of pollinators can act as a major barrier to gene flow between closely related taxa (Grant 1994). Pitcairnia flammea, P. corcovadensis and P. albiflos show different floral signals and do not share the same pollinators. Their pollination syndromes probably limit the formation of hybrids under natural conditions. The species are interfertile though, and interspecific hybridization might also occur because of non-specific pollinators. The incorporation of autogamy into the breeding system has a pronounced effect on parameters effecting gene dispersal. The magnitude of this effect depends on the balance between outcrossing and selfing. Autogamy may serve as a very effective mechanism to isolate two interfertile and sympatric populations (Levin 1971, Grant 1981). The three species studied in this report were self-compatible which will restrict gene flow and hybridization among the populations. It is known that self-compatible species can frequently hybridize as females with related selfincompatible species; however, the reciprocal cross is often impeded by inhibition of pollen tube growth in the style (Levin 1971, Arnold 1997). Self-incompatibility responses in plants are cytologically similar to hetero-incompatibility reactions. These similarities may be due to a common genetic control. Although, selfcompatibility prevents interspecific crossings (pre-mating), it involves the loss of heteroincompatibility (post-mating). The evolution of selfing could be driven by a) competition among sympatric species for pollinators, b) the paucity or unreliability of pollinators, or c) by events of dispersion and fragmentation common in rare plants. It could also be explained in terms of co-evolution with other congener species (Levin 1971, Solbrig and Rollins 1976). In summary, the three studied species of Pitcairnia were mainly self-

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

breeders. Self-breeding probably assures their reproductive isolation because post-mating isolation, and geographic, ecological, and temporal mechanisms for genetic isolation do not occur in these three sympatric species. We thank F. R. Scarano and D. S. Arau´jo for comments and linguistic advice; D. D. Biesboer for review of the English; J. E. Morrey-Jones, E. A. Almeida and M. C. Sampaio for fieldwork assistance; N. P. L. Paz and D. L. Gabriel for seed counting assistance; L. P. Gonzaga for hummingbird identification. This paper is part of a doctoral thesis undertaken at the Post-Graduate Programme in Ecology of the Universidade Federal do Rio de Janeiro, by the first author. The research was partly supported by the Brazilian Branch of Margaret Mee Amazon Trust and by Pronex/Finep.

References Affre L., Thompson J. D., Debussche M. (1995) The reproductive biology of the Mediterranean endemic Cyclamen balearicum Willk. (Primulaceae). Bot. J. Linn. Soc. 118: 309–330. Arau´jo A. C., Fischer E. A., Sazima M. (1994) Florac¸a˜o sequencial e polinizac¸a˜o de treˆs espe´cies de Vriesea (Bromeliaceae) na regia˜o de Jure´ia, sudeste do Brasil. Revta. Brasil. Bot. 17: 113–118. Arnold M. L. (1997) Natural Hybridization and Evolution. Oxford University Press, New York. Charlesworth D., Charlesworth B. (1987) Inbreeding depression and its evolutionary consequences. Ann. Rev. Ecol. Syst. 18: 237–268. Darwin C. (1872) On the origin of species by means of natural selection, or the preservation of favoured races in the struggle life. 6th edn. John Murray, London. Erhardt A., Ja¨ggi B. (1995) From pollination by Lepidoptera to selfing: the case of Dianthus glacialis (Caryophyllaceae). Plant Syst. Evol. 195: 67–76. Faegri K., Van der Pijl L. (1979) The principles of pollination ecology. 3rd edn. Pergamon Press, Oxford. Gardner C. S. (1984) Natural hybridization in Tillandsia subgenus Tillandsia. Selbyana 7: 380– 393. Grant V. (1981) Plant speciation. Columbia University Press, New York.

211

Grant V. (1994) Modes and origins of mechanical and ethological isolation in angiosperms. Proc. Natl. Acad. Sci. USA. 91: 3–10. Hamrick J. L., Godt M. J. W., Murawski D. A., Loveless M. D. (1991) Correlations between species traits and allozyme diversity: implication for conservation biology. In: Falk D. A. S., Holsinger K. E. (eds.) Genetics and conservation of rare plants. Oxford University Press, Oxford, pp. 75–86. Jacquemart A. L. (1996) Selfing in Narthecium ossifragum (Melanthiaceae). Plant Syst. Evol. 203: 99–110. Karron J. D. (1991) Patterns of genetic variation and breeding systems in rare plant species. In: Falk D. A. S., Holsinger K. E. (eds.) Genetics and conservation of rare plants. Oxford University Press, Oxford, pp. 87–98. Klips R. A., Snow A. A. (1997) Delayed autonomous self-pollination in Hibiscus laevis (Malvaceae). Amer. J. Bot. 84: 48–53. Levin D. A. (1971) The origin of reproductive isolating mechanisms in flowering plants. Taxon 20: 91–113. Lloyd D.G. (1992) Self- and cross-fertilization in plants. II. The selection of self-fertilization. Int. J. Pl. Sci. 153: 370–380. Lloyd D.G., Schoen D. J. (1992) Self- and crossfertilization in plants. I. Functional dimensions. Int. J. Pl. Sci. 153: 358–369. Luther H. E. (1984) A hybrid Pitcairnia from Western Ecuador. J. Bromeliad Soc. 34: 272–274. Martinelli G. (1997) Biologia reprodutiva de Bromeliaceae na Reserva Ecolo´gica de Macae´ de Cima. In: Lima H. C., Guedes-Bruni R. R. (eds.) Serra de Macae´ de Cima: Diversidade Florı´ stica e Conservac¸a˜o em Mata Atlaˆntica. Jardin Botaˆnico do Rio de Janeiro, Rio de Janeiro, pp. 213– 250. McWilliams E. L. (1974) Evolutionary ecology. In: Smith L. B., Downs R. J. (eds.) Bromeliaceae (Pitcairnioideae). Flora Neotropica Monograph No. 14. Hafner Press, New York, pp. 40–45. Niemer E. (1979) Climatologia do Brasil. IBGE, Rio de Janeiro. Ortega Olivencia A., Claver J. P. C., Alcaraz J. A. D. (1995) Floral and reproductive biology of Drosophyllum lusitanicum (L.) Link (Droseraceae). Bot. J. Linn. Soc. 118: 331–351. Richards A. J. (1986) Plant breeding systems. George Allan and Unwin, London.

212

T. Wendt et al.: Selfing and reproductive isolation in Pitcairnia

Sazima I., Vogel S., Sazima M. (1989) Bat pollination of Encholirium glaziovii, a terrestrial bromeliad. Plant Syst. Evol. 168: 167–179. Sazima M., Buzato S., Sazima I. (1995) Polinizac¸a˜o de Vriesea por morcegos no Sudeste brasileiro. Brome´lia 2: 29–37. Sluys M. V., Stotz D. F. (1995) Padro˜es de visitac¸a˜o a Vriesea neoglutinosa por beija-flores no Espı´ rito Santo, Sudeste do Brasil. Brome´lia 2: 27–35. Smith L. B., Downs R. J. (1974) Pitcairnioideae (Bromeliaceae). Flora Neotropica Monograph No. 14. Hafner Press, New York, pp. 1–658. Solbrig O. T., Rollins R. C. (1976) The evolution of autogamy in species of the mustard genus Leavenworthia. Evolution 31: 265–281. Stebbins G. L. (1957) Self-fertilization and population variability in higher plants. Amer. Nat. 861: 337–354. Wendt T. (1994) Pitcairnia L’He´ritier (Bromeliaceae) of Rio de Janeiro State, Brazil. Selbyana 15: 66–78. Wendt T., Chamas C. C. (1997) Pitcairnias do leste do Brasil - suas verdadeiras identidades. Brome´lia 4: 12–18. Wendt T., Canela M. B. F., Morrey-Jones J. E., Henriques A. B., Rios R. I. (2000a) Recognition

of Pitcairnia corcovadensis (Bromeliaceae) at the species level. Syst. Bot. 25: 389–398. Wendt T., Paz N. P. L., Rios R. I. (2000b) A morphometric analysis of a putative hybrid between Pitcairnia albiflos and Pitcairnia staminea (Bromeliaceae). Selbyana. 21: 132–136. Wendt T., Canela M. B. F., Faria A. P. G., Rios R. I. (2001) Reproductive biology and natural hybridization between two endemic species of Pitcairnia (Bromeliaceae). Amer. J. Bot. 88: 1760–1767. Wyatt R., Evans E. A., Sorenson J. C. (1992) The evolution of self-pollination in granite outcrop species of Arenaria (Caryophyllaceae). VI. Electrophoretically detectable genetic variation. Syst. Bot. 17: 201–209.

Addresses of the authors: Taˆnia Wendt (e-mail: [email protected]), Maria B. F. Canela, Denise E. Klein, Departamento de Botaˆnica, Universidade Federal do Rio de Janeiro, CCS, IB, cep 21941-590, Rio de Janeiro-RJ, Brazil. Ricardo I. Rios, Departamento de Ecologia, Universidade Federal do Rio de Janeiro, CCS, IB, Caixa Postal 68020, cep 21941-590, Rio de Janeiro-RJ, Brazil.