| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Ecology andEvolutionary Biology, University of California, Irvine, California92697-2525
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDiscussionREFERENCES |
|---|
Key Words: Alsinidendron • Caryophyllaceae • dioecy • insectpollination • pollen:ovuleratios • pollensize • Schiedea • windpollination
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDiscussionREFERENCES |
|---|
Although adaptations for biotic and abiotic pollination are oftenreadily recognized, there is little information available for ecologicalfactors promoting shifts between these pollination systems. Comparisonsare particularly difficult when distantly related species arecontrasted, because many different factors, including evolutionaryhistory, may play important roles in determining pollination syndromes.Comparisons of closely related species, especially when phylogeneticinformation is available, may be useful for identifying conditions thatfavor shifts in pollination vectors, and the consequences of thoseshifts for breeding system evolution (Cox,1990; Chase and Hills, 1992;Armbruster, 1993; Cox and Humphries, 1993; Hodges and Arnold, 1994, 1995; Goldblatt,Manning, and Bernhardt, 1995; Bruneau,1997; Melendez-Ackerman,1997; Campbell, Waser, andMeléndez-Ackerman, 1997).
The evolution of wind pollination and the relationship of this shiftto the evolution of separate sexes were investigated inSchiedea and Alsinidendron (Caryophyllaceae:Alsinoideae), a monophyletic assemblage of species endemic to theHawaiian Islands. The 25 species of Schiedea and four speciesof Alsinidendron occupy a wide range of habitats and possessdiverse morphology and breeding systems (Wagner,Weller, and Sakai, 1995; Weller,Wagner, and Sakai, 1995). Species occurring in wet or mesicforest (all Alsinidendron species and many Schiedeaspecies) are large-leaved hermaphroditic vines, herbs, or shrubs. Species of Schiedea occupying dry habitats are oftennarrow-leaved woody shrubs and are typically gynodioecious (female[pistillate] and hermaphroditic individuals present inpopulations), subdioecious (female, male [staminate], andhermaphroditic individuals in populations), or dioecious (female andmale individuals in populations). These reproductive systems arecollectively termed dimorphic breeding systems. Phylogenetic analyses(Wagner, Weller, and Sakai, 1995;Weller, Wagner, and Sakai, 1995;Sakai et al., 1997) using morphologicaland molecular approaches suggest that species of Schiedeaoccurring in dry habitats have evolved from ancestors found in mesicforests. Mapping breeding systems on a phylogenetic tree based onmorphological data indicates that dimorphism has evolved on twooccasions in this lineage (Fig.1; number of transitions to dimorphism is dependent on theassumptions underlying character coding, see Weller, Wagner, and Sakai, 1995, fordetails).
|
The first goal of this research was to characterize pollinationsystems in Alsinidendron and Schiedea. A wind tunnelwas used to provide a direct measure of the propensity for winddispersal of pollen in five species of Schiedea. For speciesthat were not tested in the wind tunnel, a number of morphologicalmeasures were used to predict potential wind pollination. Pollen:ovuleratios were calculated because wind-pollinated species typically havehigh values, and because of the general utility of pollen:ovule ratiosfor detecting differences in breeding systems (Cruden, 1977), particularly among closelyrelated species or in monophyletic lineages (Preston, 1986; Mione andAnderson, 1992). Pollen size was measured becausewind-pollinated species usually produce relatively small pollen grains(Proctor, Yeo, and Lack, 1996). Anther volume was measured to determine the association of anther volumeand pollen production. The degree of inflorescence condensation wascalculated because wind-pollinated species typically produce highlycondensed inflorescences that disperse and trap pollen more effectivelythan diffuse inflorescences (Niklas,1985, 1987). Morphologicalfeatures of species that dispersed pollen in the wind tunnel were usedto determine which of the remaining species were wind pollinated. Differences in nectar volume and sugar concentration were alsoinvestigated to determine whether they varied in patterns consistentwith the distribution of wind pollination, biotic pollination, andautogamy.
The second goal of this research was to use phylogenetic informationto determine whether wind pollination is necessary for the evolution ofdimorphic breeding systems. If wind pollination is essential forfemales to increase in frequency in populations, wind pollination shouldbe found in all dimorphic species, including those gynodioecious specieswith low frequencies of females. Alternatively, if the shift to windpollination occurs prior to the evolution of dimorphism, then windpollination should be found in hermaphroditic species occurring inhabitats similar to those of dimorphic species.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDiscussionREFERENCES |
|---|
Wind tunnel experiment
A wind tunnel was used to investigate the degree of wind transport ofpollen in five Schiedea species. These species included onewhere wind pollination seemed certain (S. globosa), one wherebiotic pollination seemed certain (S. nuttallii), two specieswhere it was not possible to predict whether pollination was biotic orabiotic (S. menziesii and S. salicaria), and S.lydgatei, where previous field studies indicated a combination ofwind and biotic pollination. Species used in the wind tunnel experimentwere representative of the range of inflorescence types found in thegenus (Fig. 2). To provide anadequate pollen source, three genetically distinct individuals of eachspecies were used for each dispersal measurement. A false floor wasused in the wind tunnel to minimize turbulence around the pots (J.LaRue, University of California at Irvine, personal communication). Pots containing adult plants were inserted into holes arranged in anequilateral triangle at the midpoint of the false floor. The center ofthis triangle was used to approximate the point source for pollenrelease. Plants were placed in the wind tunnel prior to antherdehiscence, which occurs between 1700–1800. Measurements ofdispersal were carried out the following morning. In order to preventloss of pollen, a single species was tested each day. Pollen dispersalfor each set of three individuals per species was tested on fivedifferent days to provide replication. For each measurement, the totalnumber of inflorescences and the total number of open flowers perindividual were counted. Counts were used to calculate the averagenumber of flowers per inflorescence for each plant used in the windtunnel experiment.
|
Three-way analysis of variance was used for analysis of pollencounts, with species, distance, and wind speed viewed as fixedtreatments. Pollen counts were square-root transformed to improvenormalization. For the most part the same sets of plants were used ineach run of a species, leading to a lack of independence for runs, butbecause it was necessary to replace plants on several occasions, arepeated-measures ANOVA was not carried out. Similarly, results of theone-way ANOVA used to analyze inflorescence number and the number offlowers per inflorescence should be interpreted withcaution.
Morphologicalanalyses
All extant species of Schiedea and Alsinidendronwere included in the morphological analysis. Inflorescence condensationwas measured using field plants or herbarium specimens of plants fromthe field because of the tendency for substantial modification of thisfeature in cultivation; all other measurements were based on plantsgrown in the greenhouse. To calculate pollen:ovule ratios, pollenproduction was measured for five individuals per species by placing fourundehisced anthers from each of ten flowers per individual in a 2.0-mLmicrocentrifuge tube with 0.5-mL lactophenol. After maceration inlactophenol, anthers were physically disrupted with a dissecting needle,spun on a vortex mixer for 1 min, and samples were withdrawn and placedin the chambers of five haemocytometers (two chambers per haemacytometerfor a total of ten counts per individual). Individual means of eachplant were then used as replicates within species for comparison ofdifferent species for this and subsequent comparisons. Ovules werecounted for five flowers of each of five genotypes per species. Foreach species, the pollen:ovule ratio was calculated as the ratio of themean number of pollen grains per flower to the mean number of ovules perflower.
Pollen size was calculated by measuring the diameter of 50 pollengrains from each of five individuals per species. Pollen productionmight vary as a function of anther volume as well as pollen grain size.To test this possibility, anther volume was calculated as width xwidth x length. Five flowers from each of four individuals ofeach species were measured.
To calculate inflorescence condensation the total number of flowersfor an inflorescence was counted and divided by the total length of theinflorescence, measured from the most basal node of the inflorescenceproducing flowers to the most distal flower. Separate condensationindices were calculated for each sex and then averaged for dimorphicspecies.
Standing crop nectar volume was measured using microcapillary tubes(1–20 µL capacity depending on the species) between 0800 and1200. An additional sample of nectar was collected after 24 h. Thetotal nectar volume was measured to within 0.001 µL. Standing cropand 24-h samples were taken for three flowers from each of threeindividuals of each sex per species and averaged for dimorphic species. Flowers were sampled when stigmas of the protandrous flowers werereceptive. Removal of nectar from Schiedea flowers waspossible without contamination from pollen; for flowers of the fourspecies of Alsinidendron, which are facultatively or obligatelyautogamous, some contamination of nectar by pollen was unavoidable dueto the campanulate sepals and early dehiscence of anthers.
Concentrations of fructose and sucrose in nectar were determinedusing the anthrone-sulfuric acid spectrophotometric method of Van Handel(cited in Kearns and Inouye, 1993). Nectar volumes ranged from 0.031 to 78 µm, which necessitatedaddition of double-distilled water in some cases to bring the volume tothe minimum required for analysis.
Results from the wind tunnel, combined with morphologicalinformation, were used to predict whether remaining Schiedeaspecies were wind pollinated. For example, when wind tunnel resultsindicated that S. salicaria was wind pollinated, untestedspecies with greater pollen production, greater inflorescencecondensation, higher pollen:ovule ratios, and smaller pollen sizes werecategorized as wind pollinated. Analysis of variance or nonparametricKruskal-Wallis tests were used to compare features of wind-pollinatedspecies with the remaining biotically pollinated or autogamousAlsinidendron and Schiedea species. Where necessary,transformations were carried out to ensure that data were homoscedastic.Significance levels were adjusted using the sequential Bonferroni methodfor multiple contrasts (Sokal and Rohlf,1995). The frequency of females in populations was correlatedwith pollen:ovule ratios, pollen size, and inflorescence condensation todetermine whether higher female frequency was associated with morepronounced adaptations for wind pollination. Schiedeahaleakalensis was deleted from correlation analyses because of thelimited number of plants surveyed to establish the percent females inthe population. Alsinidendron species were excluded fromanalyses contrasting features associated with biotic and windpollination in Schiedea.
Stepwise discriminate function analysis (PROC STEPDISC in SAS, 1989) and discriminate function analysis (PROCDISCRIM) were used to determine which features of the reproductivesystem of Schiedea were the best predictors of insect vs. windpollination. Stepwise discriminate function analysis is used when apriori decisions have been made to establish categories; the purpose ofthe analysis is to determine which data best support the a prioridecision. For stepwise discriminate function analysis, 17 species ofSchiedea (seven dimorphic, ten hermaphroditic) with no missingdata were classified a priori as either insect or wind pollinated. Sixcharacters (ovule number, pollen production, pollen size, anther volume,flower number, and inflorescence length) were used in the stepwisediscriminate function analysis.
The distribution of traits associated with mode of pollination wasinterpreted in a phylogenetic context, by comparing the distribution ofthese traits on phylogenetic trees with shifts in breeding systems. Comparisons of sister taxa with differing breeding systems andpollination biology were also used to understand the evolutionaryrelationship of shifts in pollination biology to breeding systemevolution.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDiscussionREFERENCES |
|---|
|
|
Pollen andovule production
Pollen production per flower was nearly twice as great for dimorphicSchiedea species as hermaphroditic species (16212 vs. 9322grains per flower; Table2), a significant difference (F = 10.6, df= 1,20; P = 0.0048; P < 0.05 usingthe sequential Bonferroni adjustment). Mean pollen production inAlsinidendron was similar to dimorphic Schiedeaspecies (Table 2), buthighly variable. Pollen production in Schiedea showed a strongpositive association with the percentage females in populations(r = 0.714, df = 19; P < 0.01). There were no significant differences in ovule production betweenhermaphroditic and dimorphic Schiedea species (Kruskal-Wallistest, P = 0.594, df = 1), nor were theredifferences in ovule production related to percentage of females inpopulations (r = -0.219, df = 1,18, ns). Ovulenumber in Alsinidendron is more that four times the ovuleproduction in Schiedea species.
|
|
Anthervolume
Anther volume in Schiedea and Alsinidendron variedby an order of magnitude among hermaphroditic species, ranging from 0.31mm in S. pubescens to 2.7 mm in S.verticillata. Anther volume in dimorphic Schiedea speciesranged from 0.37 mm in S. ligustrina to 0.84mm in S. globosa. There was no significantdifference in anther volume between hermaphroditic and dimorphicSchiedea species (Kruskal-Wallis test: P =0.78, df = 1).
Flowernumber
Among species of Schiedea, flower number varied widely, from<20 flowers per inflorescence in hermaphroditic S. hookerito >200 flowers per inflorescence in subdioecious S.globosa. Although the highest numbers of flowers per inflorescencewere found among dimorphic taxa, several hermaphroditic species (e.g.,S. membranacea and S. pubescens) also produced manyflowers, and differences between dimorphic and hermaphroditic specieswere not significant (F = 2.26, df = 1,16;P = 0.152). There was a positive relationship of flowernumber with the frequency of females in populations (r =0.515; df = 16; P < 0.05).
Inflorescence condensation
Using the number of flowers divided by the inflorescence length as ameasure of the degree of compactness of the inflorescence (thecondensation index), there was a striking relationship with breedingsystem in Schiedea. The average condensation index fordimorphic species was 19.5 vs. 2.86 for hermaphroditic Schiedeaspecies, a highly significant difference (Kruskal-Wallis test;P = 0.0007; P < 0.05 with the Bonferroniadjustment). Overlap in the degree of condensation betweenhermaphroditic and dimorphic species occurred for S. menziesii,a hermaphroditic species inhabiting dry habitats, S. nuttallii,a hermaphroditic species occurring in mesic forest, and gynodioeciousS. salicaria, which has a small number of females inpopulations (Fig. 2). Specieswith more highly condensed inflorescences (e.g., S. globosa;Fig. 2) had higher frequenciesof females in populations (r = 0.650, df = 16;P < 0.01; Fig. 4C).Differences in condensation were due in large part to variation ininflorescence length between hermaphroditic and dimorphic species(Fig. 2); dimorphicSchiedea species had short inflorescences compared tohermaphroditic species (5.51 vs. 22.7 cm), a highly significantdifference (Kruskal-Wallis test, P = 0.0025; df =1, P < 0.05, Bonferroni adjustment).
Nectar production
Species of Alsinidendron produced far more nectar thanSchiedea species (Table2). Nectar variation among Schiedea species wassubstantial, with the highest values recorded for S.verticillata, also the species with the largest flowers (Weller, Wagner, and Sakai, 1995). ForSchiedea species, there were no obvious differences in nectarproduction between hermaphroditic and dimorphic species for eitherstanding crop or the amount of nectar produced after 24 hr (for standingcrop, P = 0.211, df = 1, Kruskal-Wallis test; for24 h production, F = 0.75, df = 1,13, P= 0.4021). Nectars are strongly hexose dominant; dividing theaverage micrograms per microlitre of fructose by the combined values forfructose and sucrose, values were 91.4% forAlsinidendron, 71.3% for hermaphroditicSchiedea, and 87.5% for dimorphicSchiedea. These values ignore the contribution ofglucose, which is not measured using the anthrone-sulfuric acid methodfor analysis of nectar sugars. Alsinidendron had lowerconcentrations of sucrose than all species of Schiedea, andlower concentrations of fructose than dimorphic species ofSchiedea (P < 0.05; Tukey's test). Hermaphroditic and dimorphic species of Schiedea differedsignificantly in sucrose concentration (P < 0.05,Tukey's test), although not infructose.
Categorization of modes of pollination and stepwisediscriminate function analysis
The two dimorphic species included in the wind tunnel showed clearevidence of wind pollination (Tables1, 2); becausethe remaining dimorphic species had combinations of either higher pollenproduction, higher pollen:ovule ratios, greater inflorescencecondensation, or smaller pollen size compared to most hermaphroditicspecies (Table 2), theywere characterized as wind pollinated. None of the hermaphroditicspecies included in the wind tunnel showed evidence of wind dispersal ofpollen; therefore, remaining hermaphroditic Schiedea specieswith combinations of greater pollen size, lower pollen production, lowerpollen:ovule ratios, or more diffuse inflorescences were classified aspossessing biotic vectors, or possibly autogamy in one case.
Results from stepwise discriminate analysis indicated that pollensize (F = 24.2, df = 1,15; P <0.0002) and pollen production (F = 8.08, df =1,14; P < 0.013) were most important in categorizingSchiedea species as wind pollinated (Fig. 5). Ovule number, anther volume,flower number, and inflorescence length were not significant in thestepwise analysis. Discriminate analysis used to cross-validate resultsfrom the stepwise analysis indicated that pollen size and pollenproduction provided reasonably accurate discrimination of insect vs.wind pollination. Using pollen size, insect-pollinated species werecorrectly classified in nine of ten cases, and wind-pollinated specieswere correctly classified in six of seven cases. Pollen production wasslightly less accurate in predicting pollination status: althoughinsect-pollinated species were again correctly classified in nine of tencases, misclassifications occurred for two of the seven wind-pollinatedspecies.
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDiscussionREFERENCES |
|---|
Floral morphology in Alsinidendron, which occurs in mesic orvery wet forests, is strikingly different from Schiedea(Wagner, Weller, and Sakai, 1995;Weller, Wagner, and Sakai, 1995). Species of Alsinidendron have pendant flowers, petaloid,campanulate calyces, and large, flap-like extensions of the nectaries. In contrast, flowers of Schiedea have narrow, green sepals andtubular nectary extensions. Alsinidendron flowers usuallyproduce an order of magnitude more nectar than Schiedea species(Table 2). In view oftheir substantial nectar production, species of Alsinidendronmay have been pollinated by native honeyeaters (Meliphagidae; allHawaiian species are now extinct) or honeycreepers (Drepanidinae;species often rare or in some cases extinct; Weller, Wagner, and Sakai, 1995). Hexose-dominant nectars are characteristic of species pollinated bypasserine birds (Baker and Baker, 1983;Lammers and Freeman, 1986), includingthe native honeycreepers and honeyeaters of Hawaii, further suggestingthat the hexose-dominant Alsinidendron species are adapted forbird pollination. Paradoxically, despite the abundant nectar productionof Alsinidendron species, they are either facultatively orobligately autogamous (Weller, Wagner, and Sakai,1995). Autogamy is consistent with the low pollen:ovuleratios for these species. Nectar produced by Schiedea speciesis also strongly hexose-dominant. The only native pollinators that havebeen observed visiting Schiedea are pyralid moths (Norman, Weller, and Sakai, 1997), whose nectarpreferences are unknown.
Direct evidence from the wind tunnel and evidence based onmorphological correlates of wind pollination indicate that ten of the 13Schiedea species occurring in dry habitats are wind pollinated(Fig. 1). All wind-pollinatedspecies have dimorphic breeding systems. The three Schiedeaspecies that occur in dry habitats and have no obvious adaptations forwind pollination (S. lydgatei, S. menziesii, andS. verticillata) possess hermaphroditic breeding systems. Mosthermaphroditic species of Schiedea occurring in mesic or wethabitats are probably insect pollinated (Fig. 1). Without artificialpollination, all but one autogamous Schiedea species found inwet forests fail to produce capsules in the greenhouse (Weller andSakai, unpublished data).
Schiedea salicaria and S. globosa, both verified aswind pollinated in the wind tunnel experiment, represent the endpointsof morphological variation associated with wind pollination. Morphological traits used to infer wind pollination include high pollenproduction per flower, high pollen:ovule ratios, small pollen size,large flower number, short inflorescence axis, and highly condensedinflorescences. High pollen production, and in particular highpollen:ovule ratios, are often associated with outcrossing and windpollination (Cruden, 1977; Proctor, Yeo, and Lack, 1996). Althoughabsolute values for pollen:ovule ratios vary for breeding systemsdistributed among different families and genera, relative differences,particularly within monophyletic lineages, appear to be usefulindicators of breeding systems (Mione andAnderson, 1992). Pollen:ovule ratios for hermaphroditicSchiedea fall between Cruden's(1977) values for facultative selfing and facultativeoutcrossing categories; dimorphic Schiedea are comparable tohis facultative outcrossing category, but much lower than the mean valueof 5859 for the obligately outcrossing category. Preston (1986) noted that Cruden's limitedsample of diverse species was unlikely to have predictive power, butfound that within the Brassicaceae, outcrossing and selfing species weresignificantly different, and standards could be established to predictbreeding systems for other species in the family.
In Schiedea, high pollen:ovule ratios result from increasedpollen production as well as reduced ovule number. Stepwisediscriminate analysis showed that pollen production, although not ovulenumber, was important for discriminating between wind pollinated speciesand those that are presumably insect pollinated or autogamous (Fig. 5). Increased pollen productionmay result primarily from selection for wind pollination, while changesin ovule production are likely to be the result of many factors,including differences in habitat that favor seed sizes (and consequentlyovule number and size) adapted to differing moisture regimes (Baker, 1972; Weller,1985, 1989).
Among species used in the wind tunnel experiment, wind-pollinatedspecies had significantly more open flowers per inflorescence thanspecies that were not wind pollinated. Because there were no detectabledifferences in total flower number for wind-pollinated and bioticallypollinated species when all species of Schiedea wereconsidered, this result suggests either that inflorescences of windpollinated species disperse pollen over a shorter time period, or thatflowers open in synchrony, perhaps when environmental cues indicateconditions favorable to wind dispersal of pollen. In either case, ifthe inflorescence is the functional unit involved in dispersal of pollen(Niklas 1985), the significantdifference among species in the number of open flowers, as well asdifferences in pollen production per flower, may be related toproduction of sufficient pollen to achieve wind pollination.
Plants used in the wind tunnel were grown in the same-sized pots, anddifferences that developed during growth and production ofinflorescences were probably the result of inherent differences ingrowth rates among species. Whether or not such differences reflectthose typically found under field conditions, or whether differences inplant size are related to wind pollination is not known. Habitat andsurrounding vegetation may have important influences on overall plantstature in relation to mode of pollination. The shortest species(<0.5 m) included in the wind tunnel experiment was S.globosa; this species may nonetheless be able to disperse polleneffectively because it occurs on rocky cliffs with relatively sparse,low vegetation.
Pollen size is also a very strong indicator of pollination system inSchiedea, as indicated by stepwise discriminate analysis(Fig. 5). With the exceptionof S. salicaria, wind-pollinated species all produced smallerpollen grains than insect pollinated or autogamous species. Thenegative correlation of pollen size with the frequency of females inpopulations indicates that the greater efficiency of wind pollinationdue to changes in pollen size may have favored increases in thefrequency of females in populations.
Changes in anther volume have played no consistent role in modifyingtotal pollen production. Although anther volumes differ by nearly anorder of magnitude, these differences appear unrelated to pollinationsyndrome or breeding system. The effects of anther size on total pollenproduction may be complex because of differences in pollen size. Forexample, in S. globosa, pollen grains are relatively largeamong wind-pollinated species (Fig.4), but high pollen production is nonetheless achieved byproduction of larger anthers than in most other species ofSchiedea (Table2). Conversely, S. ligustrina has the smallestanthers of any wind pollinated species, but also produces the smallestpollen grains within Schiedea (Fig. 4), and again achieves substantialpollen production.
A striking feature of wind-pollinated species is a trend towardcompaction of the inflorescence (Fig.2). Highly condensed inflorescences are in general associatedwith wind pollination (Niklas 1985,1987) and increase the likelihood ofboth pollen dispersal and receipt as harmonic oscillations areestablished. Niklas, (1987) comparedgrass species with condensed and open panicles, and demonstrated thatspecies with more condensed inflorescences had greater efficiency ofwind pollination as wind speeds increased. Among Schiedeaspecies, the inflorescence condensation index was most strongly affectedby changes in inflorescence length rather than flower number. Thecorrelation of flower number and the frequency of females in populationssuggests, however, that selection for greater numbers of flowers perinflorescence may be associated with wind pollination.
All wind-pollinated Schiedea species produce uprightinflorescences at the tips of branches, an arrangement likely tofacilitate wind transport of pollen (Proctor,Yeo, and Lack, 1996). In S. globosa, inflorescencesare produced on elongated stems with reduced leaf size, decreasing thelikelihood that foliage will serve as a pollen filter (Fig. 2). Selection for more effectivewind pollination may be driving the shift in inflorescence structureseen among dimorphic species.
Pollinator limitation has been hypothesized as a factor leading tothe presence of females in species occurring on dry, windswept ridgeswhere pollinators may be rare or absent (Wellerand Sakai, 1990). This hypothesis depends on the capacity forsome self-pollination in these species, which leads to the expression ofinbreeding depression. The small amount of pollen dispersed in the windtunnel for all species may indicate that under natural conditions thereis sufficient pollen transfer to account for limited seed productionduring the transition from biotic to wind pollination. With theexpression of sufficient inbreeding depression, the progeny ofoutcrossed females may be favored, but only if traits favoring efficientwind pollination evolve. In the absence of biotic pollination orwell-developed adaptations for wind pollination, females may be slow toestablish in populations. This might explain the close correspondenceof wind pollination and dimorphism in Schiedea. Alternatively,in windy habitats where there are few pollinators, selection for theevolution of wind pollination could occur prior to the appearance offemales. In a hermaphroditic species, increased selfing and theexpression of inbreeding depression would be expected to favor malesterility mutations causing females to appear in populations. In eitherscenario, the occurrence of females is closely tied to the evolution ofwind pollination.
The analyses presented here assume that wind pollination has evolvedindependently in each species where it occurs. Statistical independenceis unlikely, however, because features associated with dimorphism andpollination biology probably evolved in common ancestors of currentspecies. In this likely scenario, closely related species sharefeatures through common ancestry, rather than because of the independentevolution of traits, and appropriate analyses would involve estimatingwhether shifts in pollination syndromes and breeding systems occurredeach time species colonized dry habitats (Donoghue, 1989; Weller,Wagner, and Sakai, 1995). Based on character mapping(Weller, Wagner, and Sakai, 1995),dimorphism has evolved within Schiedea in two of the four majorclades identified using morphological and molecular characters (Fig. 1; Wagner,Weller, and Sakai, 1995; Weller,Wagner, and Sakai, 1995; Soltis et al.,1996; Sakai et al., 1997). The exact number of transitions from hermaphroditism to dimorphismdepends on the details of the mapping procedure, how characters arecoded, and the degree of confidence in the phylogeny. The greatestuncertainty about the number of transitions to dimorphism occurs in theS. globosa clade (Fig.1), where hermaphroditism may represent the basal condition iftransitions are delayed (Weller, Wagner, andSakai, 1995), or be the result of reversals from dimorphicancestors if transitions are accelerated. In the second clade withdimorphic species, the S. adamantis clade (Fig. 1), all methods for characterreconstruction indicate that gynodioecy evolved at the base of theclade. Because of the tight correspondence between dimorphism andcharacters associated with wind pollination (Fig. 1), it is likely that transitionsto wind pollination occurred at those points on the phylogeny whereshifts to dimorphism occurred. Precise mapping of the shift to windpollination would require greater ability to discriminate stages in theevolution of this trait.
Contrasts between sister taxa, especially where phylogeneticrelationships are strongly supported, provide insights into theevolution of pollination systems. For example, S. salicariaand S. lydgatei are sister taxa based on morphological andmolecular data, but differ in breeding systems and pollination biology. Schiedea lydgatei is a hermaphroditic species pollinated bynative moths, and to a lesser extent, wind (Norman, Weller, and Sakai, 1997). Phylogeneticanalyses provide clear evidence that hermaphroditism in S.lydgatei represents a reversal from a dimorphic ancestor (Weller, Wagner, and Sakai, 1995), probably as aresult of colonization of dry shrubland on the leeward side of Moloka`i,habitat where native moths serve as pollinators and result in highoutcrossing rates (Norman, Weller, and Sakai,1997). With high outcrossing rates, the substantialinbreeding depression characteristic of S. lydgatei (Norman et al., 1995) is not expressed among theprogeny of hermaphrodites, and there is no selection for male sterility. Females have been detected in progeny raised from field-collected seedsof this species (Norman, Weller, and Sakai,1997), although they are not present in naturalpopulations.
In contrast to S. lydgatei, gynodioecious S.salicaria is found on steep windward ridges of West Maui. Thisspecies is wind pollinated, even though females constitute only12% of the population. Clearly, wind pollination is associatedwith dimorphism at a very early stage in the evolution of this breedingsystem, and it seems likely that females would not have reached thisfrequency unless wind served as a successful pollen vector. Amongdimorphic species, S. salicaria has larger pollen, a lowerpollen:ovule ratio, and inflorescences with little evidence ofcondensation, all indications that this species has few of theadaptations characteristic of species with well-developed windpollination. These features may indicate that wind pollination hasevolved recently in S. salicaria. Niklas (1987) suggested that compact panicles inhermaphroditic, wind-pollinated grasses might increase the probabilityof self-pollination. If selection for wind pollination resulted in theevolution of more compact inflorescences in S. salicaria,higher selfing rates of hermaphrodites would result in the expression ofthe strong inbreeding depression present in this species (Sakai, Karoly, and Weller, 1989), and favorfemales in populations. The contrast between S. lydgatei andS. salicaria suggests how shifts in habitat, the availabilityof pollinators, and the expression of inbreeding depression may playcritical roles in the evolution of both reproductive systems andpollination biology in Schiedea.
In hermaphroditic S. menziesii, results from the wind tunnelexperiment indicate that this species is not wind pollinated, butneither is there evidence for biotic pollination under field conditions(S. Weller, personal observations). Absence of adequate biotic orabiotic pollination may have prevented the establishment of females inS. menziesii, even though they have been observed among progenyraised in the greenhouse, and inbreeding depression and selfing ratesare high enough to favor the establishment of these females in the field(Weller et al., unpublished data). The condensation index for S.menziesii is in the range for wind-pollinated, dimorphic species(Table 2), suggesting thatselection for more effective wind pollination may have occurred withoutthe appearance of females. The slightly condensed inflorescences ofS. menziesii might also result from the dimorphic ancestry ofthis species (Fig. 1; ACCTRANoptimization). In either case, this modification of inflorescencestructure did not enhance pollen dispersal in the wind tunnelexperiment. Slight differences in the degree of wind pollination maynot have been detectable in our study, since we saw no differences inthe magnitude of wind pollination, even when we expected thatdifferences would be substantial. For example, differences in themagnitude of wind pollination for S. salicaria and S.globosa were not evident, despite the pronounced adaptations forwind-pollination is S. globosa, and the apparent absence ofthese adaptations in S. salicaria.Species of Schiedeawith transitional pollination systems may obscure detection of featuresimportant in biotic versus wind pollination using discriminate functionanalysis. For example, differences in inflorescence condensation forhermaphroditic and dimorphic species were highly significant usingANOVA, although discriminate function analysis did not identify thistrait as important in separating hermaphroditic and dimorphicspecies.
Comparisons of species that appear to be transitional betweenhermaphroditism and dimorphism indicate that the evolution of windpollination may be critical for the presence of females in populations. In species where wind pollination is detectable, either in the windtunnel or through observation of morphological correlates of windpollination, females have established successfully in populations,presumably because wind pollination circumvents pollinator limitation. In contrast, the absence of outcrossing, hermaphroditic species withwell-developed wind pollination indicates that females appear prior towind pollination. Transitional species thus provide insights in to theorder of events associated with the evolutionary transition fromhermaphroditism and biotic pollination to dimorphism and windpollination. An important point is that lack of strict specializationin pollination (e.g., Waser et al.,1996) may be critical during the transition fromspecialization to biotic vs. wind pollination. Otherwise, stages arehypothesized where species have no effective pollen vectors. Studies ofpotential pollinator limitation for transitional species would be ofparticular interest and reveal the extent to which shifts in pollenvectors entail a loss of reproductivecapacity.
|
FOOTNOTES |
|---|
3 Current address: Department of Botany, Duke
University, Durham, NC 27706.
4 Current address: Environmental Management Agency,
County of Orange, 10852 Douglass Rd., Anaheim, CA
92702-4048.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDiscussionREFERENCES |
|---|
Allison, T. D. 1990 Pollenproduction and plant density affect pollination and seed production inTaxus canadensis. Ecology 71: 516–522.
Armbruster, W.S. 1993 Evolution of plant pollination systems:hypotheses and tests with the neotropical vine Dalechampia.Evolution 47: 1480–1505. [CrossRef][ISI]
Baker, H.G. 1963 Evolutionary mechanisms in pollination biology.Science 139: 877–883.
———. 1972 Seed weightin relation to environmental conditions in California. Ecology53: 997–1010. [CrossRef][ISI]
———, and I.Baker. 1983 Floral nectar sugar constituents inrelation to pollinator type. In C. E. Jones and R. J. Little[eds.], Handbook of experimental pollination biology,117–141. Scientific and Academic Editions, New York,NY.
Berry, P. E., and R. N.Calvo. 1989 Wind pollination, self-incompatibility, andaltitudinal shifts in pollination systems in the high Andean genusEspeletia (Asteraceae). American Journal of Botany 76:1602–1614.[CrossRef][ISI]
Bruneau, A. 1997 Evolutionand homology of bird pollination syndromes in Erythrina(Leguminosae). American Journal of Botany 84:54–71.[Abstract]
Campbell, D. R., N. M. Waser, and E. J.Meléndez-ackerman. 1997 Analyzingpollinator-mediated selection in a plant hybrid zone: hummingbirdvisitation patterns on three spatial scales. AmericanNaturalist 149: 295–315. [CrossRef][ISI]
Charlesworth,D. 1993 Why are unisexual flowers associated with windpollination and unspecialized pollinators? American Naturalist141: 481–490. [CrossRef][ISI]
Chase, M. W., and H. G.Hills. 1992 Orchid phylogeny, flower sexuality, andfragrance-seeking. BioScience 42: 43–49. [CrossRef][ISI]
Cox, P. A. 1990 Pollinationand the evolution of breeding systems in Pandanaceae. Annals of theMissouri Botanical Garden 77: 816–840. [CrossRef][ISI]
———, and C. J.Humphries. 1993 Hydrophilous pollination and breedingsystem evolution in seagrasses: a phylogenetic approach to theevolutionary ecology of Cymodoceaceae. Botanical Journal of theLinnean Society 113: 217–226. [CrossRef]
Cruden, R.W. 1977 Pollen:ovule ratios: a conservative indicatorof breeding systems in flowering plants. Evolution 31:32–46.
Donoghue, M.J. 1989 Phylogenies and the analysis of evolutionarysequences, with examples from seed plants. Evolution 43:1137–1156.[CrossRef][ISI]
Goldblatt, P., J. C. Manning, and P.Bernhardt. 1995 Pollination biology ofLapeirousia subgenus Lapeirousia (Iridaceae) insouthern Africa; floral divergence and adaptation for long-tongued flypollination. Annals of the Missouri Botanical Garden 82:517–534.[CrossRef][ISI]
Hesse,M. 1979 Ultrastructure and distribution of pollenkittin the insect- and wind-pollinated genus Acer (Aceraceae).Plant Systematics and Evolution 131: 277–289. [CrossRef][ISI]
Hodges, S. A., and M. L.Arnold. 1994 Floral and ecological isolation betweenAquilegia formosa and Aquilegiapubescens. Proceedings of the National Academy ofSciences, USA 91: 2493–2496.
———, and———. 1995 Spurring plant diversification:are floral spurs a key innovation? Proceedings of the Royal SocietyLondon B 262: 343–348.
Kaplan, S. M., and D. L.Mulcahy. 1971 Modes of pollination and floral sexualityin Thalictrum. Evolution 25:659–668.[CrossRef][ISI]
Kearns, C. A., and D. W.Inouye. 1993 Techniques for pollination biologists.University of Colorado Press, Niwot, CO.
Lammers, T. G., and C. E.Freeman. 1986 Ornithophily among the HawaiianLobelioideae (Campanulaceae): evidence from floral nectar sugarcompositions. American Journal of Botany 73:1613–1619.[CrossRef][ISI]
Lemen, C. 1980 Allocation ofreproductive effort of the male and female strategies in wind pollinatedplants. Oecologia 45: 156–159. [CrossRef][ISI]
Levin, D. A., and H.Kerster. 1974 Gene flow in seed plants.Evolutionary Biology 7: 139–220.
Meléndez-Ackerman, E.J. 1997 Patterns of color and nectar variation acrossan Ipomopsis (Polemoniaceae) hybrid zone. American Journalof Botany 84: 41–47. [Abstract]
Midgley, J. J., and W. J.Bond. 1991 Ecological aspects of the rise of theangiosperms: A challenge to the reproductive superiority hypotheses.Biological Journal of the Linnean Society 44:81–92.
Mione, T., and G. J.Anderson. 1992 Pollen:ovule ratios and breeding systemevolution in Solanum section Basarthrum (Solanaceae).American Journal of Botany 79: 279–287. [CrossRef][ISI]
Niklas, K.J. 1982 Simulated and empiric wind pollination patternsof conifer ovulate cones. Proceedings of the National Academy ofSciences, USA 79: 510–514.
———. 1985 Theaerodynamics of wind pollination. Botanical Review 51:328–386.[ISI]
———. 1987 Pollencapture and wind-induced movement of compact and diffuse grass panicles:implications for pollination efficiency. American Journal ofBotany 74: 74–89. [CrossRef][ISI]
———, and K. Tha PawU. 1982 Pollination and airflow patterns around coniferovulate cones. Science 217: 442–444.
Norman, J. K., A. K. Sakai, S. G. Weller,and T. E. Dawson. 1995 Inbreeding depression inmorphological and physiological traits of Schiedea lydgatei(Caryophyllaceae) in two environments. Evolution 49:297–306.
———, S. G. Weller, and A.K. Sakai. 1997 Pollination biology and outcrossingrates in hermaphroditic Schiedea lydgatei(Caryophyllaceae). American Journal of Botany 84:641–648.[Abstract]
Preston, R.E. 1986 Pollen:ovule ratios in the Cruciferae.American Journal of Botany 73: 1732–1740. [CrossRef][ISI]
Proctor, M., P. Yeo, and A.Lack. 1996 The natural history of pollination. HarperCollins, London.
Regal, P.J. 1982 Pollination by wind and animals: ecology ofgeographic patterns. Annual Review of Ecology and Systematics13: 497–524. [CrossRef][ISI]
Renner, S. S., and J. P.Feil. 1993 Pollinators of tropical dioeciousangiosperms. American Journal of Botany 80:1100–1107.[CrossRef][ISI]
Sacchi, C. F., and P. W.Price. 1988 Pollination of the arroyo willow, Salixlasiolepis: role of insects and wind. American Journal ofBotany 75: 1387–1393. [CrossRef][ISI]
Sakai, A. K., K. K. Karoly, and S. G.Weller. 1989 Inbreeding depression in Schiedeaglobosa and Schiedea salicaria (Caryophyllaceae),subdioecious and gynodioecious Hawaiian species. American Journalof Botany 76: 437–444. [CrossRef][ISI]
———, S. G. Weller, W. L.Wagner, P. S. Soltis, and D. E. Soltis. 1997 Adaptiveradiation in the endemic Hawaiian genera Schiedea andAlsinidendron (Caryophyllaceae: Alsinoideae): phylogeneticinsights into the evolution of dioecy. In T. Givnish and K.Sytsma [eds.], Molecular evolution and adaptive radiation,455–473. Cambridge University, Cambridge.
SAS. 1988 SAS/STAT user'sguide, 6.03 ed. SAS Institute, Cary, NC.
Sokal, R. R., and F. J.Rohlf. 1995 Biometry. W. H. Freeman, New York,NY.
Soltis, P. E., D. E. Soltis, S. G. Weller,A. K. Sakai, and W. L. Wagner. 1996 Molecularphylogenetic analysis of the Hawaiian endemics Schiedea andAlsinidendron (Caryophyllaceae). Systematic Botany21: 365–379. [CrossRef][ISI]
Stebbins, G. L. 1970. Adaptiveradiation in reproductive characteristics in angiosperms. I: Pollinationmechanisms. Annual Review of Ecology and Systematics 1:307–326
Wagner, W. L., S. G. Weller, and A. K.Sakai. 1995 Phylogeny and biogeography inSchiedea and Alsinidendron (Caryophyllaceae).In W. L. Wagner and V. A. Funk [eds.], Hawaiianbiogeography: evolution on a hot spot archipelago, 221–258. Smithsonian Press, Washington, DC.
Waser, N. M., L. Chittka, M. V. Price, N. M.Williams, and J. Ollerton. 1996 Generalization inpollination systems, and why it matters. Ecology 77:1043–1060.[CrossRef]
Weller, S. G. 1985 The lifehistory of Lithospermum caroliniense, a long-lived herbaceoussand dune species. Ecological Monographs 55:49–67.[CrossRef]
———. 1989 The effectof disturbance scale on sand dune colonization by Lithosperumcaroliniense. Ecology 70: 1244–1251.
———, and A. K.Sakai. 1990 The evolution of dicliny inSchiedea (Caryophyllaceae), an endemic Hawaiian genus.Plant Species Biology. 5: 83–95.
———, W. L. Wagner, and A.K. Sakai. 1995 A phylogenetic analysis ofSchiedea and Alsinidendron (Caryophyllaceae:Alsinoideae): implications for the evolution of breeding systems.Systematic Botany 20:315–337.[CrossRef][ISI]
This article has been cited by other articles:
|
|
 |
|
  S. G. Weller, A. K. Sakai, T. M. Culley, D. R. Campbell, P. Ngo, and A. K. Dunbar-Wallis Sexually dimorphic inflorescence traits in a wind-pollinated species: heritabilities and genetic correlations in Schiedea adamantis (Caryophyllaceae) Am. J. Botany, October 1, 2007; 94(10): 1716 - 1725. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Friedman and L. D. Harder Functional associations of floret and inflorescence traits among grass species Am. J. Botany, November 1, 2005; 92(11): 1862 - 1870. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Golonka, A. K. Sakai, and S. G. Weller Wind pollination, sexual dimorphism, and changes in floral traits of Schiedea (Caryophyllaceae) Am. J. Botany, September 1, 2005; 92(9): 1492 - 1502. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. J. Givnish, R. A. Montgomery, and G. Goldstein Adaptive radiation of photosynthetic physiology in the Hawaiian lobeliads: light regimes, static light responses, and whole-plant compensation points Am. J. Botany, February 1, 2004; 91(2): 228 - 246. [Abstract] [Full Text] [PDF]
|
|
|
|
