Heredity 64 (1990) 139—143 The Genetical Society of Great Britain Received 4 August 1989

Outcrossing rates and allozyme variation in rayed and rayless morphs of pilosa

M. Sun* and * Departmentof Agronomy and Range Science, Fred R. Ganderst University of California, Davis, CA, 95616, U.S.A. 1 Department of Botany, University of British Columbia, Vancouver, B.C., Canada V6T 2B1.

Thirty-four isozyme loci were assayed in 1048 from three Hawaiian populations of the autogamous introduced weed Bidens pilosa. Total isozyme gene diversity was very low, 0049. One population containing only rayless plants was completely monomorphic at all loci except PGI-3, and this locus was nearly fixed. In a large population polymorphic for rayed, rayless, and intermediate plants, two loci showed rare variants and PGI-3 was polymorphic. Allele frequencies at PGI-3 were not significantly different among the floral morphs, but outcrossing rates, measured by progeny tests using PGI-3 as a genetic marker, were significantly higher in the radiate morph (9 per cent) than in the rayless and intermediate morphs (5 per cent). Presumably radiate plants have higher outcrossing rates because they are more attractive to pollinators.

INTRODUCTION rayless plants should provide a better experimental system for testing the effects of ray florets on out- Bidenspilosa L. () is a cosmopolitan crossing rates than comparisons between , subtropical and tropical weed native originally to because other features of the mating system are North, Central, and South America (Ballard, less likely to vary between morphs than between 1986). Usually the flowering heads are discoid species. (without ray florets) and relatively inconspicuous, Several studies of populations of Senecio vu!- only 6-7 mm wide. However, a morph with 4-7 garis L. polymorphic for radiate and rayless plants white ray florets with ligules 5-7 mm long is also have shown that the radiate form has a significantly common throughout most of the range of the higher intermorph outcrossing rate than the rayless species. The flowering heads of this morph are form (Marshall and Abbott, 1982, 1984a). This, in 12—15 mm wide, which is still relatively small for part, has been attributed to the effect of the ray species of Bidens, but they are considerably more florets in pollinator attraction (Abbott and Irwin, conspicuous than the rayless morph. The radiate 1988). However, there are several complicating form is sometimes distinguished taxonomically as factors in this polymorphism in Senecio. The radi- var. minor, but appears to differ in no other ways ate morph was derived by introgressvie hybridisa- from the rayless form, and both frequently occur tion from the radiate species S. squalidus L. within intermixed in the same population. very recent times. The ray florets of S. vulgaris are The major function of ray florets in most mem- pistillate and cannot automatically self pollinate bers of the Asteraceae is usually assumed to be (Marshall and Abbott, 1984b). pollinator attraction, and many rayless species are It therefore seems desirable to test the thought to be predominantly self pollinating. Con- hypothesis that radiate morphs have higher out- sequently, one would predict that a radiate morph crossing rates than rayless morphs in other species would have a higher outcrossing rate than a rayless in the Asteraceae. We have estimated outcrossing morph. However, there is little quantitative data rates and genetic variability at isozyme loci in available to test whether the presence of ray florets populations of B. pilosa, to test this hypothesis and alone can affect pollinator visits or outcrossing to test our prediction that this species is highly rates. Populations polymorphic for radiate and self-pollinating. 140 M. SUN AND F. R. GANDERS

MATERIALS AND METHODS phroditic (Ballard, 1986). Thus technically these radiate heads are disciform. However, in our Seedssamples were collected from individual sample the styles degenerate and/or the anthers plants of B. pilosa in two populations consisting do not dehisce to release pollen at anthesis, result- only of rayless plants and one large population ing in complete infertility of the ray florets. consisting of radiate, rayless, and intermediate Autofertility (seed set in the absence of pollinators) plants. One rayless population, sampled in 1984, averaged 93±75 per cent (range 69-100 per cent) occurred along highway 580, 16km west of the in rayless plants (sample size 16 plants, 40 heads, intersection with highway 56, near Wailua, Kauai, 1880 florets), and 88±71 per cent (range 75—97 and the other, sampled in 1985, was from the per cent) in radiate plants (sample size 2 plants, Pacific Tropical Botanical Garden (PTBG) at 9 heads, 344 florets). The florets which failed to Lawai, Kauai. The polymorphic population, set seed were probably limited by resources rather sampled in 1986, is located along the dirt road at than pollination, because the heads which flowered the base of the large cinder cone Ahumoa on the first usually set a higher percentage of seed than Island of Hawaii. later flowering heads on the same inflorescence. Seeds were germinated in vermiculite with a 12 No seed dormancy was observed, seeds germinat- hour photoperiod at a day/night temperature of ing in about four days in growth chambers or at 25/20°C. Young seedlings were analysed electro- room temperature. There was a difference in phoretically using an extraction buffer containing flowering phenology between populations. Plants 0.1 M tris, 0.2 M sucrose, 2 per cent PEG, 06 per from the low elevation population from the Pacific cent PVP, 0001 M EDTA, 01 per cent bovine Tropical Botanical Garden, Kauai, flowered three serum albumin, and 066 per cent mercaptoethanol months after transplanting, whereas those from the (pH 7.5). Two electrophoresis buffer systems, Ahumoa population (elevation about 2000 m) con- morpholine-citrate (Clayton and Tretiak, 1972) tinued vegetative growth for eight months after and tris-citrate (Pitel and Cheliak, 1984), were used transplanting under the same glasshouse condi- to resolve 12 enzyme systems using 14 per cent tions until a short day, cold treatment was applied. starch gels (Electrostarch, lot 392). Only the PGI-3 Progenies from rayless plants were all rayless and locus in the Ahumoa population was sufficiently progenies of radiate plants were all radiate, but polymorphic to be used as a genetic marker. Out- the genetics of the polymorphism are not yet crossing rates were estimated from progeny known. genotype arrays at this locus for the three different All attempted crosses between B. pilosa and B. flower morphs in this population using the method sandvicensis failed to set viable seed, whereas con- of Ritland and Jam(1981). Genetic identities trolled crosses within B. sandvicensis set normal between populations were calculated according to seed. This is not surprising as there is no evidence Nei (1974). Some seedlings were raised to maturity that any introduced species of Bidens in the to document floral morphology and autofertility, Hawaiian Islands ever hybridise with the endemic and to test whether B. pilosa would hybridize with species, although all of the 19 endemic species are the Hawaiian endemic species, B. sandvicensis interfertile with each other (Ganders and Nagata, Less., which occurs sympatrically with B. pilosa at 1984). the Wailua, Kauai, locality. A total of 12 enzyme systems were resolved, controlled by a minimum of 34 loci (table 1). The minimum number of loci was determined from the RESULTS number of isozyme bands observed, their staining intensities, their segregation or lack thereof in Thedisc florets of all three floral morphs are her- progenies of single plants, and the known quater- maphroditic, yellow, and range in number from nary structure and intracellular distribution of the 30—63 per head. Prior to anthesis, the style branches enzymes. For example, PGI is a dimeric enzyme have partially separated and trapped many pollen so that a single heterozygous locus, or two loci grains on their papillose surfaces as the style homozygous for different alleles, will both result extends through the anther tube. This mechanism in two homodimers and one heterodimer, giving appears to ensure self pollination in the absence three isozymes. However, cytosolic PGI and of pollinators. Radiate heads have 4-6 white ray organelle PGI are isolated within the cell and do florets. These are not "true" ray florets, which in not interact to form heterodimers. The minimum other species of Bidens are sterile and lack styles, number of PGI isozymes observed was six, indicat- because they are usually morphologically herma- ing either two heterozygous PGI loci, one organel- OUTCROSSING IN RAYED AND RAYLESS BIDENS 141

Table 1 Enzymes assayed, minimum number of loci recorded, and minimum number of isozymes of each enzyme

Enzyme, acronym, and code Loci Isozymes

Acid phosphatase (ACP) EC. 31.3.2 3 5 Aconitase (AGO) E.G. 4.2.1.3 4 4 Aspartate aminotransferase (AAT) E.C. 2.6.1.1 4 5 Esterase (Fluorescent; FLE) E.C. 3.1.1.1 1 1 Isocitrate dehydrogenase (IDH) E.C. 1.1.1.42 2 2 Leucine aminopeptidase (LAP) E.C.3.4.11.1 4 4 Malate dehydrogenase (MDH) E.C. 1.1.1.37 3 5 Malic enzyme (ME) E.C. 1.1.1.40 2 5 6-phosphogluconate dehydrogenase (6-PG) E.G. 1.1.1.44 2 4 Phosphoglucose isomerase (PGI) E.G. 5.3.1.9 4 6 Phosphoglucomutase (PGM) E.G. 2.7.5.1 3 4 Shikimate dehydrogenase (SKDH) E.G. 1.1.1.25 2 2

Total 12 34 47

lar and the other cytosolic, or four homozygous (G5, =055) indicates geographical differentiation loci, two organellar and two cytosolic. Because of populations. The two rayless populations on there was no segregation in families with six isozy- Kauai were almost identical, but the rare allele at mes, the loci could not have been heterozygous, PGI-3 in the Kauai populations was the more so that there must be a minimum of four loci for common allele in the Ahumoa population on the PGI. We observed a larger number of isozymes of island of Hawaii. Average Nei's genetic identity each enzyme than those reported for diploid among the populations was 0945. species (Gottlieb, 1982), but that is not surprising Outcrossing rates at the PGI-3 locus were because B. pilosa is a polyploid (Ballard, 1986). estimated for the three morphs in the Ahumoa All loci except three were monomorphic for population (table 3). The radiate morph had an the same allele in all populations (table 2). The outcrossing rate of 88 per cent, significantly higher rayless population from the Pacific Tropical than the outcrossing rates of the rayless (4.8 per Botanical Garden was completely monomorphic. cent) and intermediate morph (51 per cent) (t test, The rayless Wailua population was nearly P <0.001). Observed heterozygosity at the PGI-3 monomorphic, with variation at only the PGI-3 locus was twice as high in the radiate and inter- locus, and the frequency of the rare allele at this mediate morphs than in the rayless morph. There locus was only 3 per cent. The Ahumoa population, was no significant heterogeneity in allele frequen- polymorphic for rayed, rayless, and intermediate cies among maternal morphs in the population plants, showed variation at three loci, but only (X2] =1769, p>O.3O). Wright's observed fixation — PGI-3 was very polymorphic, and only two alleles index (F0) was calculated as F0 =1 (H0/2pq) were detected. Total gene diversity (H1) in the three where H0 is observed heterozygosity, and the populations was very low, only 0049, which is the inbreeding coefficient Fe as Fe =(1—t)/(1 + t), sum of within population diversity (H5 =0.022) where t is the outcrossing rate. The difference and between population diversity (D51 =0027).A between F0 and Fe, (zF) measures the effect of high value of the coefficient of gene differentiation evolutionary factors other than selfing on homozy-

Table 2 Genetic variability at 34 loci in three populations of Bidens pilosa. N =samplesize, PLP =% loci polymorphic, k =averagenumber of alleles per locus, H0 =averageobserved heterozygosity per locus (%),p,q =allelefrequencies at PGI-3 PGI-3 N N Population families progeny PLP k H0 p q

PTBG 19 71 0 l00 0 100 0 Wailua 30 133 3 103 005 097 003 Ahumoa 52 844 9 109 34 032 068 142 M. SUN AND F. R. GANDERS

Table 3Common allele frequency in maternal morph (qm) and progeny (qf), outcrossing rate (I), average observed percent heterozygosity in progeny (H0), Wright's fixation index (F0), inbreeding coefficient (Fe), and F at the PGI-3 locus in the three flower morphs in the Ahumoa population. N =samplesize, sd =standarddeviation

Morph

Radiate Intermediate Rayless

q,,, (sd) 0750 (0083) 0588 (0.095) 0633 (0120) q1 (sd) 0748 (0017) 0624 (0.021) 0646 (0.022) t (sd) 0088 (0039) 0051 (0.030) 0048 (0.024) H0 137 144 74 F0 064 069 084 F 084 090 091 F —020 —021 —007 N families 20 17 15 N progeny 329 273 242

gosity in populations (Brown, 1979). Negative been attributed to selection favouring heterozy- values of L?F in all three morphs indicate that gotes, but direct evidence is scanty or absent. heterozygotes were more frequent than expected Yearly fluctuations in outcrossing rates may also under the assumption of inbreeding equilibrium produce inbreeding disequilibrium. We have no in the population (table 3). evidence as to the cause of inbreeding disequili- brium in B. pilosa. Homozygosity and genetic monomorphism are DISCUSSION characteristic of all other rayless species of Bidens which have been investigated electrophoretically, Floralmorphology and the high level of auto- presumably because rayless species are all highly fertility suggest that B. pilosa is a predominantly autogamous. Roberts (1983) analysed nine enzy- self pollinating species, although we have observed mes coded by 16 loci in eight natural populations butterflies visiting both morphs in Hawaiian popu- of B. discoidea (Torr. & Gray) Britton, a species lations. Our estimates of 5 per cent outcrossing for with small, inconspicuous heads and no ray florets. the rayless morph and 9% outcrossing for the Seven of the 16 loci showed variability within the radiate morph confirm that both are highly but not species, but all populations except two were exclusively self fertilising. In contrast, the 20 popu- monomorphic at all loci. Two populations showed lations of nine self-compatible, yellow rayed taxa two alleles at each of two loci, but heterozygotes of Bidens in which outcrossing rates have been were found in only one population at a frequency measured all show mixed mating systems ranging of 13 per cent. He concluded that the genetic from 35-88 per cent outcrossing, and averaging 63 uniformity and extreme homozygosity within per cent outcrossing (Ritland and Ganders, 1985; populations of B. discoidea were due to high rates Sun and Ganders, 1986, 1988). The absence of of self pollination and founder effects. Helenurm allozyme frequency differences between radiate and Ganders (1985) examined isozymes within one and rayless forms in the Ahumoa population population each of the autogamous species B.fron- strongly suggests that these two morphs really rep- dosa L., B. tripartita L., and B. cynapifo1ia H. B. resent a simple genetic polymorphism rather than K., and found all populations completely two taxonomic varieties growing sympatrically. monomorphic. The populations of B.frondosa and We found very low levels of allozyme variabil- B. tripartita were rayless. B. cynapiifolia usually has ity in all three populations. Total gene diversity 3-4 small yellowish rays, 5-7 mm long, but some- was less than 5 per cent, and within population times the heads are rayless. gene diversity was only 2 per cent. Genetic iden- Marshall and Abbott (1982, 1984a) found that tities between populations were high, averaging 95 the rayless morph of S. vulgaris averaged only 2 per cent. This is not surprising in an autogamous, per cent outcrossing, with a range of 02-15 per weedy, colonising species such as B. pilosa. The cent, while the radiate form averaged 13 per cent, excess heterozygosity indicated by negative LF with a range of 3-36 per cent in natural polymor- values is also a commonly reported phenomenon phic populations in the United Kingdom. Because in selfing species (Brown 1979). This has usually they used the ray locus itself as a marker gene, this OUTCROSSING IN RAYED AND RAYLESS BIDENS 143 estimate represents only intermorph crossing. Acknowledgements Seed collection was supported by a grant Some of the difference was accounted for by the from the Natural Sciences and Engineering Research Council outcrossing rate of the pistillate ray florets them- of Canada to F.R.G. Electrophoresis was done at the Depart- ment of Forest Sciences, University of Alberta, Edmonton, selves (Marshall and Abbott, 1984b), which cannot Canada, and we thank Drs Francis Yeh and Bruce Dancik for self pollinate automatically and make the flower facilities there. heads functionally protogynous. Some of the apparent outcrossing in S. vulgaris could also be the result of linkage disequilibrium caused by the origin of the radiate morph by introgression from S. squalidus L. The polymorphism in S. vulgaris REFERENCES has only been widespread in the U.K. for two or three decades, so that it is likely that linkage equili- ABB0Tr, R.J. AND IRWIN, J. A. 1988. Pollinator movements brium has not been achieved (Marshall and and the polymorphism for outcrossing rate at the ray locus Abbott, 1984b). Other characters of S. squalidus in Groundsel, Senecio vulgaris L. Heredity, 60, 295-298. BALLARD, 1986. Bidens pilosa complex (Asteraceae) in North could be associated with the radiate morph of S. and Central America. Amer. J. Bot., 73, 1452-1465. vulgaris. Nevertheless, Abbott and Irwin (1988) BROWN,A.H. D. 1979. Enzyme polymorphism in popula- concluded that pollinators preferentially visit the tions. Theor. Pop. Biol., 15, 1-42. radiate morph in polymorphic populations, result- CLAYTON, J. W. AND TRETIAK, D. N. 1972. Amine-citrate buffers for pH control in starch gel electrophoresis. J. ing in a higher rate of outcrossing in the radiate Fisheries Res. Board Canada, 29, 1169—1172. morph. However, Warren (1988) found that the GANDERS, F. R. AND NAGATA, K. M. 1984. The role of hybridiz- outcrossing rate of the two morphs did not differ ation in the evolution of Hawaiian Bidens. In Grant, in garden experiments when populations were W. F. (ed.) Plant Biosystematics, Academic Press Canada, monomorphic. Only in polymorphic populations Don Mills, Ontario, pp. 179-194. did the radiate morph outcross more than the GOTTLIEB, L. D. 1982. Conservation and duplication of iso- zymes in plants. Science, 216, 373—380. rayless morph. He attributed this to competition HELENURM,K. AND GANDERS, F. R. 1985. Adaptive radiation between morphs for pollinators in polymorphic and genetic differentiation in Hawaiian Bidens. Evolution, 39, 753—765. populations, but adequate pollinator activity in MARSHALL, D. F. AND ABBOTT, R. i. 1982. Polymorphism for either type of monomorphic population. The outcrossing frequency at the ray fioret locus in Senecio positioning of plants in the garden experiments vulgaris L. 1. Evidence. Heredity, 48, 227—235. was not designed to be random, and pollinators MARSHALL, D. F. AND ABBOTT, R. .t. 1984a. Polymorphism for were apparently abundant, so it is possible that outcrossing frequency at the ray floret locus in Senecio these results could differ from what occurs in vulgaris L. II. Confirmation, Heredity, 52, 331-336. MARSHALL, B. F. AND ABBOTT, R. i. 1984b. Polymorphism for natural populations. outcrossing frequency at the ray floret locus in Senecio The difference in outcrossing rate between vulgaris L. III. Causes. Heredity, 53, 145-149. radiate and rayless plants in the population of B. NEI, M. 1974. A new measure of genetic distance. In Crow, pilosa we studied is not the result of pollination J. F. and Denniston, C. (eds) Genetic Distance, Plenum Press, New York and London, pp. 63-76. of the ray florets themselves, because they do not PITEL, J. A. AND CHELIAK, W. M. 1984. Effect of extraction set seed. Using the PGI-3 locus as a genetic marker buffers on characterization of isozymes from vegetative tissues measures outcrossing between plants of the same offive coni,fer species: a user's manual. Information Report morph as well as between morphs. Conceivably P1-X34, Petawawa National Forestry Institute. there could be assortative pollination by pol- RITLAND, K. AND GANDERS, F. R. 1985. Variation in the mating system of Bidens menziesii (Asteraceae) in relation to popu- linators in regard to floral morph, but because lation substructure Heredity, 55, 235—244. allele frequencies at PGI-3 were not significantly RITLAND, K. AND JAIN, S. K. 1981. A model for the estimation different among the morphs this should have no of outcrossing rate and gene frequencies based on n effect on the outcrossing estimate at this locus. The independent loci. Heredity, 47,37-54. ray polymorphism in B. pilosa is not the result of ROBERTS, M. L. 1983. Allozyme variation in Bidens discoidea (Compositae). Brittonia, 35, 239-247. introgression from another species, and B. pilosa SUN, M. AND GANDERS, F. R. 1986. Female frequencies in does not hybridise with other species in the gynodioecious popultions correlated with selfing rates in Hawaiian Islands. We think it is likely that the hermaphrodites. Amer. J. Bot., 73, 1645-1648. higher outcrossing rate of the radiate morph in B. SUN, M. AND GANDERS, F. R. 1988. Mixed mating systems in Hawaiian Bidens. Evolution, 42, 516-527. pilosa is solely due to the fact that the flower heads WARREN, J. M. 1988. Outcrossing frequencies within- and are more conspicuous and attract more pollinator between-capitulum morphs in groundsel Senecio vulgaris visits. L. Heredity, 61, 161—166.