Reproductive biology of - Springer Link

Plant Syst. Evol. 223:109-123 (2000)

Plant Systematics and Evolution © Springer-Verlag 2000 Printed in Austria

Reproductive biology of Wahlenbergia (Campanulaceae) endemic to Robinson Crusoe Island (Chile) G. J. Anderson 1, G. Bernardello 2, P. Lopez S. s, D. J. Crawford 4, and T. F. Stuessy 5 1Department of Ecology and Evolutionary Biology, Storrs, CT, USA 2Instituto Multidisciplinario de Biologia Vegetal, C6rdoba, Argentina 3Concepcidn, Chile 4Department of Evolution, Ecology and Organismal Biology, Ohio State University, Columbus, OH, USA Slnstitute of Higher Plant Systematics and Evolution, Institute of Botany, University of Vienna, Austria Received August 31, 1999 Accepted March 6, 2000

Abstract. The reproductive biology of W. berteroi, W. fernandeziana, and a putative hybrid between 147. fernandeziana and W. grahamiae, endemic to Robinson Crusoe Island (Juan Fernfindez archipelago, Chile) was studied. Flowers are hermaphroditic, protandrous, offer nectar, and exhibit secondary pollen presentation involving pollen collecting hairs on the style. These features imply allogamy and biotic pollination. However, male and female phases overlap and no effective pollinators were observed. Experimental data indicate these taxa are self-compatible and facultatively autogamous, a conclusion also suggested by the pollen/ovule ratios. Selfing is accomplished when the stigmatic lobes reflex and touch the style, except for W. berteroi where they do not reflex completely. Autogamy is accomplished in the latter when pollen grains deposited on the inner surface of the corolla throat by the "pollen brush" are gathered by stigmatic lobes when shaken by wind. The degree of autogamy, and perhaps self-compatibility, seems to be inconstant, as implied by the variable natural seed set (overall range 21-188 seeds per fruit). A mixed mating system - primarily outcrossing/entomophilous, but also autogamous must have been present in the continental ancestors of these taxa. Autogamy promoting self-fertilization is important now - on an island with scarce

pollinators - and in the past - when the first founders arrived.

Key words: Angiosperms, Campanulaceae, Waklenbergia-Breeding system, pollination, pollen collecting hairs, autogamy, self-compatibility, nectar, island biology, Juan Fernfindez Islands. The genus Wahlenbe~ia (subfam. Campanuloideae) includes 267 species primarily distributed in the southern hemisphere (Lammers 1996). It is ecologically extremely variable, occupying a variety of habitats from sea level t o 4,000 m in altitude. Several of the species in this genus are endemic to the Juan Fernfindez archipelago, located 667 k m W of continental Chile. The archipelago consists of three islands: Robinson Crusoe ( = M a s a t i e r r a ) and Santa Clara to the east, and Alexander Selkirk ( = M a s a f u e r a ) , 100 k m further west. These islands are biogeographically interesting because of their relative geologic youth (5.8 million years for Santa Clara Island, ca. 4 million years for Robinson Crusoe Island, 1-2 million years for Alexander Selkirk Island; Stuessy et al. 1984) and the high level of endemism

110

G.J. Anderson et al.: Reproductive biology of Wahlenbergia

among vascular plants (ca. 62.5% of the flora; Stuessy et al. 1992, Marticorena et al. 1998). Five species of Wahlenbergia, all endemic, are recognized (Lammers 1996): three on Robinson Crusoe Island (W. fernandeziana A. DC., W. grahamiae Hemsl., and W. berteroi Hook. & Arn., which also occurs on Santa Clara), and two on Alexander Selkirk Island (W. masafuerae (Phil.) Skottsb. and W. tuberosa Hook. f.). In addition, a putative hybrid between W. fernandeziana and W. grahamiae was reported as well (Lammers 1996). There is a living specimen of this putative hybrid cultivated in the National Forestry Corporation (CONAF) gardens in the single permanent human settlement (San Juan Bautista, Robinson Crusoe Island). Lammers (1996) provided evidence of monophyly of these species that are supported by data on chromosome number (Sanders et al. 1983, Spooner et al. 1987, Crawford et al. 1990, Sun et al. 1990), allozyme variation (Crawford et al. 1990), and morphology (Lammers 1996). The Wahlenbergia species from the Juan Fernfindez are proposed to have arisen autochthonously, perhaps originally from a species similar to W. scopulicola Smith, from northeastern Australia (Lammers 1996). Lammers (1996) proposed Robinson Crusoe Island as the probable site of first colonization and origin of the non-cormose clade (W. fernandeziana and W. grahamiae). Several traits suggest that W. fernandeziana is the ancestral species on the islands (Crawford et al. 1990). Dispersal of the non-cormose clade to Alexander Selkirk Island was followed by evolution of the cormose clade (W. tuberosa and W. masafuerae). The establishment of W. berteroi (also cormose) on Robinson Crusoe and Santa Clara Islands is suggested to have occurred as a tertiary event (Lammers 1996). The non-cormose clade and W. tuberosa are found at higher inland elevations, while W. berteroi and W. masafuerae are restricted to lower coastal elevations (Lammers 1996). Although vegetative propagation can occur in Wahlenbergia via stolons, suckers, and pieces of detached roots, reproductive events

are usually sexual (Lothian 1947). The literature on the reproductive biology in the family is voluminous and dates back at least 250 years to Linnaeus (cf. Shetler 1979). However, for several genera including Wahlenbergia, there is little reproductive information. Beyond a few scattered data mainly on floral visitors (e.g. Thomson 1880, Mfiller 1883, Haviland 1884, Knuth 1909, Lothian 1947, Carolin 1960, Thulin 1975, Moeliono and Tuyn 1960), the only detailed study is of W. aIbomarginata Hook. from New Zealand (Lloyd and Yates 1982). Carl Skottsberg (1928) summarized observations on pollination and reproductive biology of the Juan Fern~mdez Wahlenbergia species as part of study of the flora of the islands. The only publication on reproductive biology since Skottsberg's are brief comments on W. berteroi (Crawford et al, 1990) and on the hybrid (Ricci and Eaton 1994, as W. larrainii). The reproductive biology of the Campanulaceae as a whole is complex. Protandry and entomophily are widespread and are combined with an unusual mechanism of secondary pollen presentation via pollen collecting hairs on the terminal 1/3 of the style (cf. Gadella 1964, Shetler 1979, Leins and Erbar 1990, Erbar and Leins 1995, Nyman 1993a, Yeo 1993). This syndrome can function with selfpollination to facilitate autogamy or with effective pollinators to promote allogamy (cf. Shetler 1979, Nyman 1993b). Given these complexities, detailed study of specific taxa are needed to clarify the reproductive biology. Furthermore, the knowledge of the reproductive biology, besides being fundamental for systematic and evolutionary studies (Anderson 1995), is essential to successful conservation efforts (Holsinger 1991, Bernardello etal. 1999), particularly of endangered species where there are very few populatiov/s - as is the case of these endemics. In the course of three recent expeditions to Robinson Crusoe Island, we studied the reproductive biology of W. fernandeziana and W. berteroi in the field. We chose to study

G. J. Anderson et al.: Reproductive biology of Wahlenbergia

W. fernandeziana and W. berteroi in particular because they represent different clades of the genus in the archipelago that are accessible. Unfortunately, W. grahamiae is extremely rare and known for a few specimens (Lammers 1996), so we were not able to study it. Both species are shrubby with flowers arranged in showy terminal, cymose inflorescences. Wahlenbergia berteroi grows on exposed rock outcrops, cliffs, and ledges near the sea (occasionally at elevations up to 300 m), whereas W. fernandeziana is more frequent at 100800 m above sea level, most often in crevices on steep or vertical rock escarpments (Lammers 1996). Populations of W. berteroi in particular are limited. We were able to locate only two populations, each of which included only 3-7 individuals. Although this species m a y be found elsewhere on the islands in small numbers, it certainly is not as c o m m o n as W. fernandeziana, which has some populations with nearly 100 individuals. In addition, we analyzed one specimen of the hybrid. This individual, planted in the C O N A F garden from seeds harvested in the field, is presumably the same as that studied by Ricci and Eaton (1994). We addressed the following questions: (1) What is the functional structure of the flowers? (2) Are there floral visitors? (3) Are the taxa self-compatible? (4) Are self-pollination and autogamy possible, and if so, is selfor cross-pollination the dominant mode? And (5) what is the implication o f these data for the future conservation of these endemic species? Material and methods Eight natural populations and one cultivated hybrid plant were studied on three field trips to Robinson Crusoe Island, Chile, during January 1991 and 1996, and December 1996/January 1997 (see Appendix I, vouchers are deposited at CONN). The taxonomic treatment of Lammers (1996) was followed in the delimitation of the taxa. Floral anatomy: Flowers were fixed in 70% ethanol, dehydrated in an ethyl alcohol-xylol series, and embedded in Paraplast. Serial cross- and longi-

111

sections were cut at 10 btm, mounted, stained with safranin-fast green-hematoxylin and observed with a compound microscope. Photomicrographs were taken with Kodak T-Max film and drawings made with the help of a camera lucida. Pollen/ovule (P/O) ratio and pollen viability: Pollen quantity was estimated using Anderson and Symon's (1989) modification of Lloyd's (1965) technique. Ovules were counted with a dissecting scope. Buds examined for P/O ratios were near anthesis, so pollen was mature but anthers had not dehisced. The mean P/O was generated from individual ratios of separate buds. A minimum of three buds per specimen was examined. Pollen viability was estimated with percent stainability of 100 grains from each of 10 flowers (a minimum of three specimens except in the case of the hybrid) using aniline blue in lactophenol (Hauser and Morrison 1964). Observations on pollen viability were done in greenhouse plants grown at the University of Connecticut produced from seeds collected in the field. Experimental crosses: Branches with tagged buds were bagged in the field with nylon mesh bags to exclude potential visitors. Flowers were emasculated in bud prior to anther dehiscence. Crosses were performed by applying pollen held among the style hairs (using the styles themselves as pollen applicators) to receptive stigmata. After 48 hours, flowers were collected and fixed in 70% ethanol. Natural autogamy was checked by bagging young tagged buds and collecting the unmanipulated flowers four days after opening. Open pollinated control flowers were randomly collected. Tanglefoot ® was applied in some cases to the treated branches in an attempt to exclude ants from the experiments. To count seeds from open pollinated flowers, mature (but not yet dehiscing) fruits were selected and crushed in a Petri dish. Mature seeds were counted under a dissecting microscope. Pollen tube growth: The gynoecia were dissected from the flowers and were softened with 8 N NaOH for one hour at 60 °C in a water bath, rinsed, and stained in aniline blue -0.1 N K3PO4 for two hours (Martin 1959), then flattened in glycerin on a slide. Pollen tubes were examined under a epifluorescence microscope. Floral visitors: Approximately 10 plants of W. berteroi and 30 of W. fernandeziana were observed in the field for more than 50 hours, during our three


112

field trips. Periods of observation ranged from 20 to 40 minutes all during daylight hours (from 08:00 to 17:00 hr). Visitors were recorded and some were captured for identification. A similar procedure was followed with the single cultivated specimen of the hybrid.

Results Flower structure. All taxa share the same basic floral structure (Figs. 1, 2). Flowers are hermaphroditic and actinomorphic. The ovary is inferior with a shallow obconic hypanthium to which the calyx is also fused (Figs. 1A, C; 2A, C). The corolla is sympetalous with a long tube and five shorter lobes. The androecium has five stamens alternate with the petals; they are inserted at the upper part of the hypanthium and dehisce introrsely by longitudinal slits. The anthers are tetrasporangiate, dithecal, connivent in the bud but diverging after the flower opens. The ovary is 3-1ocular with numerous ovules on axile placentae. The style is hollow (Fig. 3C) and terminal, glabrous at

the base, while the upper portion is covered with long hispid hairs (Fig. 3A). The style in W. fernandeziana is constricted halfway down. The stigma is 3-fid (Figs. 1C, 2C), occasionally 2- or 4-rid, and has unicellular papillae (Fig. 3A, B). The major difference in floral structure between W. fernandeziana and W. berteroi was found in the corolla: in the former species it is campanulate (Fig. 1C), mainly white with purple veins, averaging 14 m m long x 11 m m wide, whereas in W. berteroi (Fig. 2C) it is more tubular and slightly urceolate, about 15 mm long x 7 m m wide, pink or rose at first, darkening to carmine later, sometimes with white streaks. The white corolla of the hybrid is similar to W. fernandeziana but lacks the purple veins. The flowers of these taxa offer nectar as a reward. The exudate is produced in small amounts by a disk-shaped nectary located above the inferior ovary at the bottom of the flower (Fig. 3D, E), and can be characterized as "type 9" according to Fahn's terminology

C

13 '~'"

0.40 m m

.5 m m

Fig. 1. Wahlenbergiafernandeziana flower stages. A-C Flower longi-sections, A flower just beginning to open, B one-day old flower, C three-day old flower. D top view of stigma of a flower at stage depicted in A. E detail of style and stigma from C

G. J. Anderson et al.: Reproductive biology of WahIenbergia

113

2~

[

" . :: 5:"-.~ : ". ~ ' : ": . ~

"~.~i ~ .i, S"~''~ ,! 0.35 rnrn

Fig. 2. Wahlenbergiaberteroi flower stages. A-C Flower longi-sections, A flower just beginning to open, B oneday old flower, C three-day old flower. D top view of stigma of a flower at stage depicted in A. E detail of style and stigma from C

(1979). The nectary is distinguished macroscopically by its boron yellow color. The secretory tissue is composed of intensely stained isodiametrical cells that have dense cytoplasm and small vacuoles (Fig. 3E). The nectary does not have special vascular supplies; rather, it is served by the ovarian bundles that are located directly below it. Flower function. Flowers in all taxa are protandrous. In bud, the introse anthers form a closed cylinder around the immature style (Figs. IA, 2A). As they dehisce in the bud stage, their pollen is discharged and deposited onto the dense brush of unicellular collecting hairs, located around the upper third of the style and the external surface of the still unexpanded stigma (Figs. 1A, B; 2A, B; 3A, B). Close examination of both species reveals that the longest hairs are positioned at an angle so such that adjacent hairs from above and below touch at their apices, forming a

series of "ridges" that extend the entire length of the hirsute region of the style (Figs. 1A, B, 2A, B). There are 10 such ridges, evenly distributed around the style. The three branches of the stigma remain unexpanded but are fully formed (Figs. 1D, 2D). Prior to, and just after opening, the flower is in the male phase. A rapid elongation of the lower part of the style causes the collecting hairs at the distal end to sweep up the pollen as it grows past them. The anthers then wither and remain inconspicuous at the lower part of the flower (Figs. 1C, 2C). Thus, pollen is presented secondarily (i.e. not by the anthers themselves) to potential visitors as a cylindrical layer on the style in the center of the flower, positioned near the throat of the corolla-tube. Next, the hairs in the two species retract into large, spherical basal cavities (Fig. 3A, B). However, examination of styles from the hybrid showed that hair retraction did not occur.

114

A


B


115

Table 1. Mean m standard deviation of pollen and ovule numbers and pollen/ovule (P/O) ratios of Wahlenbergia from Robinson Crusoe Island. N = 10 flowers in each case Taxa

Pollen number

Ovule number

P/O ratio

Pollen viability (%)

W. berteroi FV.fernandeziana W. fernandeziana x W. grahamiae

33,609 + 8,893 42,889 ± 10,547 31,786 ± 10,392

363 ± 126 447 ± 157 342 ± 48

92 96 93

92 97 90

The flowers remain in male phase for about two days. The onset of the female phase occurs when the terminal 1/6 of the style splits into three parts and reflexes, exposing the papillate stigmatic surfaces (Figs. 1C, E; 2C, E). However, as some pollen continues to be present on the style below the stigma, male and female functions overlap at this stage (see below). We did not conduct receptivity tests, but obviously the folded, and thus unexposed, stigmatic surfaces prevent pollen from being deposited or germinating prior to this stage. The exp a n d e d stigma assumes a horizontal position at about the level of the corolla-tube throat (Figs. ] C, 2C). The degree to which the stigma branches recurve is variable. In W. fernandez# ana and the hybrid, they approach the style, or even touch it, recurving almost 360 ° (Fig. 1E). In W. berteroi, they reflex about 200 ° and do not come in contact with the style itself (Fig. 2E). Complete recurvature of the branches would allow a u t o g a m o u s pollen deposition (if still present on the style), as was the situation in the hybrid. However, in both W. fernandeziana and W. berteroi, the hairs have almost completely retracted into the basal cavities by the time the flower reaches the female phase. Nevertheless, in both species, cross sections of styles showed pollen in the basal cavities of the hairs (range 4-8 grains,

n = 3 flowers). Presumably, in W. fernandez# ana, these grains come into contact with the stigmatic branches when they touch the style.

P/O ratio, pollen viability, and breeding system. Pollen and ovule numbers and P/O ratios were determined (Table 1). In all taxa, the P/Os, at less than 100, were in the low end of the range of facultative a u t o g a m o u s species (Cruden 1977). Pollen viability (Table 1) was high for all taxa. Observations of greenhouse grown plants of W. berteroi at the University of Connecticut revealed that pollen grains were not present, or if so did not germinate when temperatures were high ( > 30 °C). Over 100 manipulations were successfully performed to characterize the breeding system of these species. We carried out at least twice as m a n y crosses and other manipulations, but m a n y were discarded because either the style was damaged or the stigmata did not mature. Compatibility. We can be certain that W. fernandeziana and the hybrid are selfcompatible because isolated plants in each had pollen tubes reaching the ovules (hereafter referred to as "positive" results) in open pollinated flowers (4/9 m a r k e d flowers for W. fernandeziana, 9/9 for the hybrid). In addition, flowers of W. fernandeziana selfpollinated by h a n d were positive (2/2, using pollen from other flowers from the same

Fig. 3. Wahlenbergiafernandeziana flower anatomy. A Longi-section of stigma and upper part of style of flower just opened, arrow indicates one unicellular hair, B cross section of a stigma similar to the one depicted in A, C cross section of a style similar to the one depicted in A, D detail of flower longi-section, showing nectary position, E detail of D showed by an arrow. Abbreviations: C corolla, K calyx, F filament, L locule from the ovary, N nectary, NT nectariferous tissue, VT vascular tissue. The bar equals 0.20 mm for A, B, and D, 0.15 mm for C, and 50 gm for E

116


plant). The 4/7 positive results for tests of autogamy (which might be tests of geitonogamy because introduced ants were able to get into even tightly sealed bags; of course, geitonogamy tests self-compatibility), as well as the emasculated, open pollinated flowers tests (4/4 positive) are also evidence of self-compatibility in the hybrid. The latter are likely autogamous, or may be geitonogamous. They cannot be allogamous because there is only one plant of the hybrid known, and this single plant is separated by very long distances from any other members of the genus. Similarly, several of the emasculated, open pollinated flowers of W.fernandeziana with pollen tubes (7 of the 10 positive results), were from an isolated plant. Many manipulations were carried out with W. berteroi, but these suffered an inordinate amount of damage. However, two of the three flowers tested for autogamy (buds bagged and harvested subsequent to stigmata maturation) were positive in a setting with plants separated sufficiently that ants could not carry pollen between plants, and, as below, no other regular floral visitors were observed. Allogamy. There is only one living hybrid plant known; thus, there is no allogamy possible. The successful tests for open pollination on W. fernandeziana all involved an isolated plant, so these do not offer any insight to inter-plant pollen transport. However, the tests on emasculated, open pollinated flowers of this species also included some individuals from larger populations. In these, only 3 out of 29 flowers included pollen tubes, but, of course, any of these could have been the result of geitonogamy as well. More open pollinated flowers of W. berteroi were positive (14/20), and these involved plants in small populations (lighthouse = 3 plants, #3069; near the landing strip = 7 plants, #3095; Appendix 1), so this could be evidence of inter-plant transfer of pollen. As we discuss below, however, this is not likely given the virtual absence of floral visitors other than introduced ants. Thus, these few positive results might be a manifestation of autogamy, or possibly, geitonogamous transfer by ants.

Visitors. Although we observed that individuals of all taxa bloom profusely and regularly produce nectar, visitation by insects is very rare. As there are few pollen-collecting visitors, pollen may remain in the older flowers in female phase, attached to the style hairs, as we observed in the hybrid. In more than 50 hours of observation, we detected one fly on W. berteroi, one moth on W. fernandeziana and the hybrid, and two halictid bees on flowers of the cultivated hybrid (Table 2). Note that the only places we observed the presumably recently introduced bees (Engel, in press) was near the only settlement, village of San Juan Bautista. Two of the three captured bees had pollen on their bodies, indicating they could act as effective pollinators - except for their lack of abundance. And, our focus here is on the reproductive biology of the native plants and pollinators; thus, the bee and ants (see below) are considered irrelevant. Ants were also observed to visit the flowers of the hybrid (at the CONAF station), of one population of W. berteroi, and two of W. fernandeziana (Table 2). The ants were identified by E. O. Wilson as Linepithema humile (Mayr), native to Argentina, but common world-wide invasives. Given their introduced status, they aie not considered of consequence in the evolution of the reproductive system. However, like the introduced halictid bee, they could be of consequence for conservation - i.e., positively or negatively, affecting sexual reproduction. The comments that follow are relevant in that context. Thus, overall, we have no good evidence for any native biotic pollinators, but the nature of pollen (sticky), the secondary pollen presentation (the stylar brush), and the shape of the flowers (bells), make abiotic (wind) pollination unlikely as well. Seed set. The fruit is a capsule that bears light small, smooth-coated seeds (0.7-0.9 mm long) probably wind dispersed. If we are correct in assuming facultative autogamy, the data on seed numbers indicate that the level of autogamy is variable. For instance, in W. berteroi, we obtained the following seed


117

Table 2. Insect floral visitors to Wahlenbergiaon Robinson Crusoe Island Plant taxa

Location

Date

Order

Family

Pollen presence

W. berteroi

Lighthouse Landing strip Lighthouse

1/17/96 12/31/96 1/2/97

Hymenoptera Diptera Hymenoptera

Formicidae 1 No Undetermined Not collected Formicidae Not collected

W. femandeziana

Lighthouse Lighthouse Lighthouse Lighthouse Pirfimide

1/13/96 1/17/96 1/17/96 1/2/97 1/3/97

Hymenoptera Formicidae Lepidoptera Pyralidae Hymenoptera Formicidae~ Hymenoptera Formicidae Hymenoptera Formicidae

Not collected No No Not collected Not collected

W. fernandeziana x W. grahamiae

CONAF CONAF CONAF CONAF CONAF

1/12/96 1/20/96 1/23/96 1/24/96 1/10/97

Hymenoptera Halictidae2 Hymenoptera Halictidae2 Hymenoptera Halictidae2 Lepidoptera Moth Hymenoptera Formicidae

No Yes (many) Yes (few grains) Not collected Not collected

~Identified as same species under the genus Lassioglossum (subgen. Dialictus) 2 Identified as same species: Linepithema humile

set per fruit (n = 5) in open pollinated flowers: 188, 90, 29, 27, 22, as well as in the hybrid: 125, 54, 44, 28, and 21, and in W.fernandeziana: 98, 74, 51, 42, and 23. Discussion

The basic floral morphology of the three taxa we studied is concordant with previous observations (Lammers 1996), and is shared generally by Wahlenbergia (e.g. Lothian 1947, Moeliono and Tuyn 1960, Thulin 1975, Rosatti 1986). Interestingly, the association between corolla type and style constriction observed by Smith (1992) in some Australian species is manifest in the Juan Fernfindez species too. That is, styles from narrower corollas are not constricted (e.g.W. berteroi), whereas those from broader corollas (e.g. IV. fernandeziana) are constricted. The location of the nectary above the ovary is typical of several other families with inferior ovaries as well (e.g., Myrtaceae and Cucurbitaceae, Fahn 1979). This nectary position has been cited for Campanula (Fahn 1979), but this is the first report for Wahlenbergia. Protandry is widespread in the family (e.g. Shetler 1979, Nyman

1992, Inoue et al. 1996) and has been reported for other Wahlenbergia as well (e.g. Haviland 1884, Lothian 1947, Moeliono and Tuyn 1960, Thulin 1975, Lloyd and Yates 1982), but is not as pronounced in the Fernandezian species, because male and female phases overlap. The pollen collecting hairs in Wahlenbergia are typical of the Campanulaceae (cf. Uphof o 1962; Shetler 1979, 1982; Leins and Erbar 1990; Nyman 1993a, b). This unusual mechanism of secondary pollen presentation is analogous to the well-known type in Campanula (Carolin 1960; Gadella 1964; Shetler 1979, 1982; Erbar and Leins 1989; Yeo 1993; Nyman 1993a, b), first described by Sprengel in 1793 and observed by Lanais in 1738 (Shetler 1979). These hairs not only serve as pollen collectors, but also in Campanula have been shown to serve as a mechanism controlling stigma maturation and promoting cross-pollination (Nyman 1993b). The length of the male phase is significantly shortened when the hairs are stimulated by contact (Richardson and Stephenson 1989, Nyman 1993b). Such stimulation can cause the hairs to retract (Shetler 1979, 1982) making the pollen unavailable for transfer, be it by subsequent pollinators or by

118


selfing via the recurvature of the style (Nyman 1993b). Different species display variation in this mechanism, perhaps reflecting differences in mating systems. Hair retraction in some selfincompatible species is fixed, thereby dramatically reducing transfer of self-pollen to the stigma. In self-compatible species, hairs may not retract unless visited, acting as some sort of fertility insurance, promoting autogamous pollination if allogamy fails (Nyman 1993a). Hair retraction was observed in both W. fernandeziana and W. berteroi and thus appears not to be dependent on contact, as it occurred even without floral visitors. Of particular interest is that pollen production can be almost nil, especially when temperatures are very high (greenhouse experiments), and yet hair retraction still occurs. This may indicate that in these species, retraction is also not associated with pollen production. Hairs on mature styles of the hybrid did not retract; this could reflect the syndrome of the very rare W. grahamiae, the other parent or perhaps be a manifestation of hybrid breakdown. Wahlenbergia fernandeziana, W. berteroi, and the studied hybrid, are self-compatible. This conclusion agrees with that previously reported for W. berteroi and the hybrid (Crawford et al. 1990, Ricci and Eaton 1994, respectively), as well as for some other species in the genus (Lothian 1947, Moeliono and Tuyn 1960, Lloyd and Yates 1982). Given the species are self-compatible, autogamy is possible, as reported for several other Wahlenbergia (Mtiller 1883, Knuth 1909, Lothian 1947, Moeliono and Tuyn 1960, Thulin 1975). Our data support facultative autogamy. Thus, the variable seed set numbers could indicate variance in the degree of self-compatibility, or perhaps they may simply reflect differences in plant vigor, water availability, etc. Greenhouse experiments to test the duration of pollen viability in W. berteroi proved unsuccessful, as pollen production was low and plants failed to produce mature seed. This is likely due to unfavorable environmental conditions, given that in other taxa, high temperatures cause aberrant meiosis (e.g. Impatiens balsamina L., Nakamura 1935).

Most features of the breeding system protandry, nectar as reward, secondary pollen presentation - reduce self-pollination or suggest allogamy and biotic pollinators. In spite of these many features, autogamy prevails in several Wahlenbergia species (Lothian 1947, Thulin 1975, Petterson 1997), as well as W. fernandeziana and the hybrid. Autogamy is effected by self-compatibility, partial overlapping of male and female phases, and recurvature of stigmatic branches to the style. In the hybrid, as the hairs do not retract, pollen is readily available for self-pollination. And in W. fernandeziana, although the hairs retract, the pollen grains that stay in the hair bases may come in contact with the stigma when it recurves, allowing selfing. In W. berteroi, although the same degree of stigmatic recurvature does not take place and this type of autogamy is not possible, most of the flowers observed have pollen on the stigmatic surfaces. We suggest an unusual method of autogamy, as follows. Pollen grains adhere to the corolla-tube throat, usually at the level where the mature stigma is located. As anthers dehisce in the bud stage, the corolla throat closely surrounds the anther cylinder which in turn, surrounds the style (Fig. 2B). Following anther dehiscence, both corolla and style elongate, and some pollen is deposited on the inner surface of the narrow urceolate corolla-throat (Fig. 2C). The ring of pollen in the throat is easy to see because its white color stands out against the pink corolla. As pollen stays viable throughout the receptive stage of the stigma (as shown in our bagging experiments), transfer from corolla to stigmata could occur when flowers are shaken by the everpresent wind, particularly on the exposed sea cliffs where W. berteroi grows. This mechanism circumvents the likely fixed hair retraction on the styles of this species as well. This autogamous wind-aided mechanism of pollination, although unusual, appears effective in this case. A similar situation was reported in Iris versicolor L. from Kent Island, New Brunswick, by Zink and Wheelright (1997). They observed that facultative self-pollination was

G. J. Anderson et al.: Reproductive biology of Wahlenbergia promoted by wind that forced reflexed stigmas to brush against petals of the same flowers where pollen had settled. Even though the reproductive mechanisms of W. fernandeziana and W. berteroi are close, phylogenetic analysis of these species indicates that W. fernandeziana is ancestral to W. berteroi (Lammers 1996). The narrowing of the floral tube in W. berteroi may have evolved to promote selfing in the presence of a nonrecurving stigma that shows less recurvature. Given the virtual lack of dedicated pollinating insects (Skottsberg 1928, Bernardello et al. 1999, Anderson et al. in press), and in particular, specialists that would respond to color differences, the darker purple variants of this species must be explained as co-variants linked to some other feature. In general, Wahlenbergia has been considered insect-pollinated (Thulin 1975, Lloyd and Yates 1982, Petterson 1997). The nectar chemical composition of all the taxa studied here is hexose dominant (Bernardello et al. in press). Such nectar is often associated with exposed nectaries and visits by short-tongued bees and flies (Baker and Baker 1983). However, that is not possible on the Juan Fernfindez where insect pollinators are extremely uncommon (Skottsberg 1928, Bernardello et al. 1999, Anderson et al. in press). The paradox of the majority of floral features suited to entomophfly, and a lack of native pollinating insects arises frequently in the Juan Fern/mdez flora (Bernardello et al. 1999, Anderson et al. in press). Such a syndrome may well characterize many other archipelagos as well where the insect fauna is limited (Bernardello and Anderson in prep.). Hummingbirds might visit the flowers for nectar (Colwell 1989), but they have never been recorded on any of these species, and the nectar quantity available is small as well. A primarily outcrossing entomophilous system may well have been present in the continental ancestors of these evolutionary young island species (Thulin 1975, Lloyd and Yates 1982, Petterson 1997) - perhaps in addition to self-compatibility and facultative autogamy. These reproductive options, pro-

119

moting self-fertilization, would obviously be important to the first Wahlenbergia colonists on the islands, but remain so today - on islands with virtually no insect pollinators. Data on genetic diversity of three populations of W. berteroi showed no allozyme variation, as is the situation in other highly selfing species on the archipelago (Crawford et al. 1990). Of course, there are also few plants of this species still extant. The comparatively much more abundant W. fernandeziana (Crawford et al. 1990) exhibited relatively high enzyme diversity at both the species and population levels. This fact is difficult to explain from the available data that indicate a primarily autogamous system. A similar situation was observed in Dendroseris litoralis Skottsb. (Asteraceae) that has quite high allozyme diversity (Crawford et al. 1998) despite the fact that it is very rare (Stuessy et al. 1998), is self-compatible, and facultatively autogamous (unpubl. data). The degree of autogamy and allogamy can vary greatly within a single plant species, depending of factors such as pollinator abundance, degree of dichogamy, or duration of pollen viability (e.g. Horowitz and Harding 1972, Schaal and Leverich 1980, Wyatt 1984, Nyman 1991). In some species, evolution has encouraged a strategy of mixed mating systems, combining abilities for outcrossing as well as selfing (Richards 1997). The stability of mixed mating systems can promote fitness benefits in offspring generated through outcrossing as well as selfing (Holsinger 1992). By the time of arrival on this island, Wahlenbergia ancestors probably had a mixed mating system, but the absence of pollinators may have forced the autogamous system present today. Conservation. The seed set produced by these taxa is quite variable, with some fruits bearing comparatively few seeds, implying that pollination may be inefficient to some extent and/or that inbreeding depression is being expressed in terms of limiting seed set. This is understandable on the basis of a breeding system with many allogamous features that is now largely autogamous. Furthermore, we have not seen seedlings in the wild or individ-

G.J. Anderson et al.: Reproductive biology of WahIenbergia

120

uals propagating vegetatively. The marginally successful breeding system and the permanently windy and cool to cold eroded habitat where these plants live makes survival difficult. In addition, a new pressure has been added: we noted that introduced locusts virtually destroyed a population of W. berteroi (#3095, near the landing strip), eating stems, leaves, fruits, and seeds. Both environmental and reproductive constraints seem to be reducing the fitness of these taxa. Wahlenbergia grahamiae is already extremely rare on the island (Lammers 1996), the hybrid is known from only one cultivated plant, and there are very few, and then small, known populations of W. berteroi. F r o m the genetic point of view, there are differences between the species as well: the variation of W. berteroi is zero, whereas that of W. fernandeziana is comparatively higher (Crawford et al. 1990). In order to conserve these plants, programs must involve a combination of environmental and reproductive measures. The populations of introduced insects and weeds must be controlled. And, experimentally produced allogamous seeds would enhance diversity in a restoration program. However, one recent

insect introduction might be valuable. Our field work, and that of others (pers. comm. of H. Toro) verify that there is now a known bee species on Robinson Crusoe Island (Anderson and Bernardello in prep.). This small halictid bee could prove to be of benefit for promotion of allogamy with species like those of Wahlenbergia. And, given the lack of any significant native insect fauna, it does not pose a danger at least to Wahlenbergia (it could compete for nectar with the two hummingbird species). We thank the National Science Foundation, University of Connecticut Research Foundation and the Department of Ecology and Evolutionary Biology, and CONICET, CONICOR and SECYTUniversidad Nacional de C6rdoba (Argentina) for support, Mary J. Spring for preparing figures 1 and 2, Jane O'Donnell for moth and fly identification, Edward O. Wilson for the ant identification, Michael S. Engel for the bee identification, Thomas G. Lammers for helpful comments on the manuscript, Richard Jensen for statistical comments, Maryke Schlehofer for helping in many ways, and Adriana P6rez for slide preparation. The CONAF (Corporaci6n Nacional Forestal, Chile) is sincerely acknowledged for help, specially from Iv/in Leiva (Head). Assistance was provided by the Servicio Meteoroldgico from Robinson Crusoe Island.

Appendix I

Wahlenbergia specimens studied. All collected on Robinson Crusoe Island, Ai;chipelago Juan Fernfindez, Chile, by G. J. Anderson et al. Abbreviations: M morphology, F fluorescence microscopy, P/O pollen/ ovule count, PV pollen viability, A anatomy, N nectar. Elevation in meters above sea level Species

Location

W. berteroi

Lighthouse Near landing strip Lighthouse

W. fernandeziana

Villagra Lighthouse Corrales de Molina Villagra Villagra Villagra Villagra

Elevation

Collection number

Date

Study

10

3069

1/28/96

F, P/O, M

0 10

3095 3096, 4006, 4007

12/31/96 1/1/97

F, S, A F, P/O

450 10 210

2026 4005 4014

1/14/96 1/2/97 1/3/97

P/O, M, A F, P/O F, P/O, PV, A

385 450 415 380

4025 4026, 4027 4033 4034

1/5/97 1/5/97 1/5/97 1/5/97

V F,P/O F F


121

Appendix (Continued) Species

Location

W. fernandeziana x W. grahamiae

CONAF CONAF

Elevation

Collection number

Date

Study

60

2008

1/12/96

P/O, PV, M, A

60

3097, 4002, 4003, 4004

1/2/97

F, P/O, S

References Anderson G. J. (1995) Systematics and reproductive biology. In: Hoch P. C., Stephenson A. G. (eds.) Experimental and molecular approaches to plant biosystematics, Monogr. Syst. Bot. 53, pp. 263-272. Missouri Botanical Garden, St. Louis. Anderson G. J., Bernardello G., Lopez S. P., Stuessy T. F., Crawford D. J. (2000) Dioecy and wind pollination in Pernettya rigida (Ericaceae) of the Juan Fernfindez Islands. Bot. J. Linn. Soc. (in press). Anderson G. J., Symon D. E. (1989) Functional dioecy and andromonoecy in Solanum. Evolution 43: 204-219. Baker H. G., Baker I. (1983) Floral nectar sugar constituents in relation to pollinator type. In: Jones C. E., Little R. J. (eds.) Handbook of experimental pollination biology. Van Nostrand Reinhold, New York, pp. 117-141. Bernardello G., Anderson G. J., Lopez S. P., Cleland M. A., Stuessy T. F., Crawford D. J. (1999) Reproductive biology of Lactoris fernandeziana (Lactoridaceae). Amer. J. Bot. 86: 829840. Bernardello G., Galetto L., Anderson G. J. (2000) Floral nectary structure and nectar chemical composition of some species from Robinson Crusoe Island (Chile). Canad. J. Bot. (in press). Carolin R. C. (1960) The structures involved in the presentation of pollen to visiting insects in the order Campanales. Proc. Linn. Soc. NSW 85: 197-207. Colwell R. K. (1989) Hummingbirds of the Juan Fernfindez Islands: natural history, evolution and population status. Ibis 131: 548-566. Crawford D. J., Sang T., Stuessy T. F., Kim S-C., Silva O. M. (1998) Dendroseris (Asteraceae: Lactuceae) and Robinsonia (Asteraceae: Senecioneae) on the Juan Fernandez Islands: similarities

and differences in biology and phylogeny. In: Stuessy T. F., Ono M. (eds.) Evolution and speciation of island plants. Cambridge University Press, Cambridge, pp. 97-119. Crawford D. J., Stuessy T. F., Lammers T. G., Silva O. M., Pacheco P. (1990) Allozyme variation and evolutionary relationships among three species of Wahlenbergia (Campanulaceae) in the Juan Fernandez Islands. Bot. Gaz. 151: 119-124. Cruden R. W. (1977) Pollen ovule ratios: a conservative indicator of breeding systems in flowering plants. Evolution 31: 32-46. Engel M. S. (2000) A new Lassioglossum from the Juan Fernfindez Islands (Hymenoptera: Halictidae). Revista Chilena Entomol. (in press). Erbar C., Leins P. (1989) On the early floral development and the mechanisms of secondary pollen presentation in Campanula, Jasione and Lobelia. Bot. Jahr. Syst. 11 I: 29-55. Erbar C., Leins P. (1995) Portioned pollen release and the syndromes of secondary pollen presentation in the Campanulales-Asterales-complex. Flora 190: 323-338. Fahn A. (1979) Secretory tissues in plants. Academic Press, London. Gadella T. W. J. (1964) Some cytological observations in the genus Campanula. II. Proc. Royal Netherl. Acad. Sci. Ser. C 66: 270-283. Hauser E. J. P., Morrison J. P. (1964) The cytochemical reduction of nitro blue tetrazolium as an index of pollen viability. Amer. J. Bot. 51: 748-752. Haviland E. (1884) Occasional notes on plants indigenous in the immediate neighborhood of Sydney. Proc. Linn. Soc. NSW 9: 1171-1174. Holsinger K. (1991) Conservation of genetic diversity in rare and endangered plants. In: Dudley E. C. (ed.) The unity of evolutionary biology: the proceedings of the fourth International Congress

122

G . J . Anderson et al.: Reproductive biology of Wahlenbergia

of systematic and evolutionary biology. Dioscorides Press, Portland, pp. 626-633. Holsinger K. (1992) Ecological models of plant mating systems and the evolutionary stability of mixed mating systems. In: Wyatt R. (ed.) Ecology and evolution of plant reproduction. Chapman and Hall, London New York, pp. 169-191. Horowitz A., Harding J. (1972) Genetics of Lupinus. V. Intraspecific variability for reproductive traits in Lupinus nanus. Bot. Gaz. 133: 155-165. Inoue K., Maki M., Masuda M. (1996) Evolution of Campanula flowers in relation to insect pollinators on islands. In: Lloyd D. G., Barrett S. C. H. (eds.) Floral biology: studies on floral evolution in animal-pollinated plants. Chapman and Hall, New York, pp. 377-400. Knuth P. (1909) Handbook of flower pollination, 3. University Press, Oxford. Lammers T. G. (1996) Phylogeny, biogeography, and systematics of the Wahlenbergia fernandez# ana complex (Campanulaceae: Campanuloideae). Syst. Bot. 21: 397-415. Leins P., Erbar C. (1990) On the mechanisms of secondary pollen presentation in the Campanulales-Asterales-complex. Bot. Acta 103: 87-92. Lloyd D. G. (1965) Evolution of self-compatibility and racial differentiation in Leavenworthia (Cruciferae). Contr. Gray Herb. Harvard Univ. 195: 1-134. Lloyd D. G., Yates J. M. A. (1982) Intrasexual selection and the segregation of pollen and stigmas in hermaphrodite plants, exemplified by Wahlenbergia albomarginata (Campanulaceae). Evolution 36: 903-913. Lothian N. (1947) Critical notes on the genus Wahlenbergia Schrader, with descriptions of new species in the Australian region. Proc. Linn. Soc. NSW 71: 201-235. Marticorena C., Stuessy T. F., Baeza C. M. (1998) Catalogue of the vascular flora of the Robinson Crusoe or Juan Fern/mdez Islands, Chile. Gayana Bot. 55: 187-211. Martin F. W. (1959) Staining and observing pollen tubes in the style by means of fluorescence. Stain Technol. 34: 125-128. Moeliono B., Tuyn P. (1960) Campanulaceae. In: Steenis C. G. G. J. Van (ed.) Flora Malesiana, Djakarta. 6: 107-141. Mfiller H. (1883) The fertilization of flowers. (Translated by Thompson, D'A. W.). Macmillan and Co, London.

Nakamura M. (1935) On the irregular meiosis of the pollen mother cells of I. balsamina Linn. by the effect of artificial high temperature. Japan. J. Bot. 11: 118-123. Nyman Y. (1991) Crossing experiments within the Campanula dichotoma group (Campanulaceae). Plant Syst. Evol. 177: 185-192. Nyman Y. (1992) Pollination mechanisms in six Campanula species (Campanulaceae). Plant Syst. Evol. 181: 97-108. Nyman Y. (1993a) The pollen-collecting hairs of Campanula (Campanulaceae). I. Morphological variation and the retractive mechanism. Amer. J. Bot. 80: 1427-1436. Nyman Y. (1993b) The pollen-collecting hairs of Campanula (Campanulaceae). II. Function and adaptive significance in relation to pollination. Amer. J. Bot. 80: 1437-1443. Petterson J. A. (1997) Revision of the genus Wahlenbergia (Campanulaceae) in New Zealand. N. Zeal. J. Bot. 35: 9-54. Ricci M., Eaton L. (1994) The rescue of Wahlenbergia larrainii in Robinson Crusoe Island, Chile. Biol. Conserv. 68: 89-93. Richards A. J. (1997) Plant breeding systems. 2nd edn. Chapman and Hall, London. Richardson T. E., Stephenson A. G. (1989) Pollen removal and pollen deposition affect the duration of the staminate and pistillate phases in Campanula rapunculoides. Amer. J. Bot. 76: 532538. Rosatti T. J. (1986) Sphenocleaceae and Campanulaceae. J. Arnold Arbor. 67: 38-45. Sanders R. W., Stuessy T. F., Rodriguez R. (1983) Chromosome numbers from the flora of the Juan Fernandez Islands. Amer. J. Bot. 70: 799810. Schaal B. A., Leverich W. J. (1980) Pollination and banner markings in Lupinus texensis (Leguminosae). Southw. Nat. 25: 280-282. Shetler S. G. (1962) Notes on the life history of Campanula americana, the tall bellflower. Mich. Bot. 1: 9-14. Shetler S. G. (1979) Pollen-collecting hairs of Campanula (Campanulaceae), I: Historical review. Taxon 28: 205-215. Shetler S. G. (1982) Variation and evolution of the Nearctic harebells (Campanula subsect. Heterophylla). J. Cramer, Vaduz. Skottsberg C. (1928) Pollinationsbiologie und Samenverbreitung auf den Juan Fernandez

G. J. Anderson et al.: Reproductive biology of Wahlenbergia Inseln. In: Skottsberg C. (ed.) The natural history of the Juan Fernandez and Easter Islands, 2, Botany, Almqvist & Wiskell, Uppsala, pp. 503547. Smith P. J. (1992) A revision of the genus Wahlenbergia (Campanulaceae) in Australia. Telopea 5: 91-175. Spooner D. M., Stuessy T. F., Crawford D. J., Silva O. M. (1987) Chromosome numbers from the flora of the Juan Fernandez Islands. II. Rhodora 89: 351-356. Stuessy T. F., Crawford D. J., Marticorena C. (1990) Patterns of phylogeny in the endemic vascular flora of the Juan Fernandez Islands, Chile. Syst. Bot. 15: 338-346. Stuessy T. F., Marticorena C., Rodriguez R., Crawford D. J., Silva O. M. (1992) Endemism in the vascular flora of the Juan Fernandez Islands. Aliso 13: 297-307. Stuessy T. F., Sanders R. W., Silva M. (1984) Phytogeography and evolution of the flora of the Juan Fernandez Islands: a progress report. In: Radovsky F. J., Raven P. H., Sohmer S. H. (eds.) Biogeography of the tropical Pacific. Assoc. Syst. Coll. and B.P. Bishop Museum, Lawrence, Kansas, pp. 55-69. Stuessy T. F., Swenson U., Crawford D. J., Anderson G., Silva O. M. (1998) Plant conservation in the Juan Fernandez Archipelago, Chile. Aliso 16: 89-101. Sun B. Y., Huessy T. D., Crawford D. J. (1990) Chromosome counts from the flora of the Juan Fernandez Islands, Chile. III. Pacific Science 44: 258-264.

123

Thomson G. M. (1880) On the fertilization, etc., of New Zealand flowering plants. Trans. N. Zeal. Inst. 13: 241-291. Thulin M. (1975) The genus Wahlenbergia s. lat. (Campanulaceae) in tropical Africa and Madagascar. Symb. Botanicae Upsal. 21: 1-223. Uphof J. C. (1962) Plant hairs. In: Zimmerman W., Ozenda P. G. (eds.) Encyclopedia of Plant Anatomy, 2nd edn., 4. Borntraeger, Berlin, pp. 1-206. Wyatt R. (1984) The evolution of self-pollination in granite outcrop species of Arenaria (Caryophyllaceae). I. Morphological correlates. Evolution 38:80d~816. Yeo P. F. (1993) Secondary pollen presentation. Plant Syst. Evol., Suppl. 6. Springer, New York, pp. 1-268. Zink R. A., Wheelwright N. T. (1997) Facultative self-pollination in island irises. Amer. Midl. Nat. 137: 72-78.

Addresses of the authors: Gregory J. Anderson, Department of Ecology and Evolutionary Biology, Storrs, CT 06269-3043, USA; Gabriel Bernardello, Instituto Multidisciplinario de Biologia Vegetal, C. C. 495, 5000 Cdrdoba, Argentina; Patricio Lopez S., Casilla 1340, Concepcidn, Chile; Daniel J. Crawford, Department of Evolution, Ecology, and Organismal Biology, Ohio State University, Columbus, OH 43210, USA; Tod F. Stuessy, Institute of Higher Plant Systematics and Evolution, Institute of Botany, University of Vienna, Rennweg 14, Vienna A-1030, Austria.