Evolution, 42(2), 1988, pp. 227-238

A MOLECULAR REEXAMINATION OF INTROGRESSION BETWEEN HELIANTHUS ANNUUS AND H. BOLANDERI (COMPOSITAE)

loREN H. RIESEBERG, I DOUGLAS E. SOLTIS, Department ofBotany, Washington State University, Pullman, WA 99164

AND

JEFFREY D. PALMER Department ofBiology, University ofMichigan, Ann Arbor, MI48109

Abstract. -Heiser (1949) hypothesized that a weedy race of Helianthus bolanderi had originated by the introgression of genes from H. annuus into a serpentine race of H. bolanderi. Although Heiser's investigation of these species is frequently cited as one ofthe best examples ofintrogression in plants, definitive evidence of gene exchange is lacking (Heiser, 1973). To determine whether the weedy race of H. bolanderi actually originated via introgression, we analyzed allozyme, chlo­ roplast-DNA (cpDNA), and nuclear-ribosomal-DNA (rDNA) variation. Evidence from enzyme electrophoresisdid not support the proposed introgressive origin ofweedy H. bolanderi. We detected a total of 37 low-frequency alleles distinguishing the serpentine race of H. bolanderi from H. annuus. Weedy H. bolanderi possessed only four of the 37 marker alleles. Further analysis demonstrated that serpentine H. bolanderi combined seven of the 35 alleles distinguishing H. annuus from weedy H. bolanderi, indicating that serpentine H. bolanderi shares three more marker alleles with H. annuus than does weedy H. bolanderi. These results are similar to expectations for race divergence from a single common ancestor and suggest that, ifintrogression occurred, the majority of marker alleles were rapidly lost following the initial hybridization event. Even more compelling evidence opposing Heiser's (1949) hypothesis, however, was from re­ striction-fragment analysis of cpDNA and nuclear rDNA. We detected a total of 17 cpDNA and five rDNA restriction-site mutations among the 19 populations examined. No parallel or back mutations were observed in phylogenetic trees constructed using eithercpDNAor rDNA mutations, and both phylogenies were completely congruent regarding the alignment of all three taxa. In addition, the weedy race ofH. bolanderi possessed a unique cpDNA, which was outside the range of variation observed among populations of either ofthe presumed parental species. Mean sequence divergences between the cpDNAs of weedy H. bolanderi and those ofserpentine H. bolanderi and H. annuus were 0.30% and 0.35%, respectively. These estimates are comparable to sequence­ divergence values observed between closely related species in other plant groups. Given the lack of parallel or convergent mutations in the cpDNA and rDNA phylogenetic trees, the complete congruence of these trees with flavonoid- and allozyme-variation patterns, and the presence of a unique and divergent chloroplast genome in the weedy race of H. bolanderi, we suggest that the weedy race of H. bolanderi was not derived recently through introgression, as hypothesized, but is relatively ancient in origin.

Received June 9,1987. Accepted October 27,1987

For the past 50 years, interspecific gene permanent addition of genes from one flow via hybridization and subsequent in­ species into another, resulting in the cre­ trogression has been implicated as one of ation and establishment ofa new type, has the primary factors bringing about evolu­ rarely been demonstrated (Heiser, 1973). tionary change in many plant groups (An­ Nevertheless, numerous studies have im­ derson, 1949; Stebbins, 1959, 1969; Grant, plicated introgression in the origin and es­ 1981). However, the frequency of occur­ tablishment ofnew races, varieties, subspe­ rence and importance ofintrogression in the cies, and species (e.g., Heiser, 1949, 1951b; evolution of plants is unknown. Although Grant, 1950; Levin, 1963; Ornduff, 1967; the occurrence of backcrossed individuals Bloom, 1976; Davis, 1985). Although in­ in hybrid swarms is well documented (e.g., trogression may have occurred in these ex­ Heiser, 1949; Alston and Turner, 1963, amples, they are based primarily on mor­ Levin, 1975; Soltis and Soltis, 1986), the phological data and historical inference, making conclusive documentation of in­ trogression difficult. Other equally plausible

I Present address: Rancho Santa Ana Botanic Gar­ hypotheses are convergent evolution, reten­ den, 1500 N. College, Claremont, CA 91711. tion ofancestral characters, and phenotypic 227 228 L. H. RIESEBERG ET AL. plasticity. However, the use of precise ge­ sive hybrid swarms in areas ofcontact. Us­ netic markers provided by allozyme varia­ ing morphological and cytological data, tion and restriction-site mutations in spe­ Heiser (1949) hypothesized that the weedy cific DNA sequences may resolve ambiguous race ofH. bolanderi developed in California questions of introgression. through introgression ofgenes from the re­ The annual species of Helianthus (sun­ cently introduced H. annuus into the ser­ flowers) are particularly favorable for study­ pentine race of H. bolanderi. Although H. ing introgression (Stebbins and Daly, 1961; annuus and H. bolanderi do hybridize, and Heiser et aI., 1969; Levin, 1975). They are although the weedy race ofH. bolanderi ap­ widespread, self-incompatible, and weedy, proaches H. annuus in a number of mor­ often colonizing disturbed habitats. Heiser phological characters, no evidence of gene (1947,1949, 1951a, 1951b, 1954, 1961)has exchange away from hybrid swarms has been shown that hybridization occurs frequently documented. In fact, even Heiser (1973) and that hybrids, although nearly sterile, can considered his earlier morphological find­ produce some offspring by backcrossing with ings to be less than definitive and envi­ the parental species. Heiser (1965) suggest­ sioned that new biochemical and genetic ed that ecological amplitude and genetic tools might help solve ambiguous questions variation in some of these species has been of introgression. increased by this process. Finally, this group Although allozyme variation has been has been the subject ofextensive cytological used successfully to investigate introgres­ and morphological studies (Heiser, 1947, sion in animal species (e.g., Selander et aI., 1949, 1951a, 1951b, 1954,1961), and 1969; Hunt and Selander, 1973; Avise and members of this group are often cited as Smith, 1974; Avise and Saunders, 1984), outstanding examples of introgression. the use of electrophoretic markers to study The most frequently cited example ofin­ introgression in plants has been limited trogression in Helianthus, indeed one ofthe (Levin, 1975; Bell and Lester, 1978; Soltis classic cases of introgression in plants, in­ and Soltis, 1986; Doebley et aI., 1984; Sol­ volves the two annual sunflowers of Cali­ tis, 1985). fornia, Helianthus annuus and H. bolanderi Comparison of cytoplasmic and nuclear (Heiser, 1949). Helianthus annuus is a com­ genetic markers has produced perhaps the mon roadside weed in California, occurring most suggestive evidence for introgression frequently in the central valley and in south­ in nature, in both plants and animals. Palm­ ern California. Helianthus annuus was al­ er et al. (1983) demonstrated a discrepancy ready in the state when the first botanical between biparental nuclear and maternal­ collections were made and was used by the cpDNA phylogenies in Brassica and native Americans for various purposes suggested the occurrence of introgression. (Heiser, 1949). Since it does not now occur Similarly, comparison of karyotypes and in natural sites in California, it was very cpDNAs in Pisum (Palmer et aI., 1985) sug­ likely introduced quite recently by native gested one possible instance in which cy­ Americans (Heiser, 1949; Stebbins and toplasms may have been exchanged be­ Daly, 1961). Helianthus bolanderi, in con­ tween related species. Studies of mitochon­ trast, is native to California and was shown drial-DNA variation in deer (Carr et aI., by Heiser(1949) to consist oftwo races. One 1986), Drosophila (Powell 1983; Solignac of these races is restricted to the foothill and Monnerot, 1986), mice (Ferris et aI., regions of California and grows chiefly on 1983), sunfish (Avise and Saunders, 1984), serpentine-derived soils. This race is re­ tree frogs (Lamb and Avise, 1986), trout ferred to as the serpentine race ofH. bolan­ (Gyllensten et aI., 1985), and water frogs den (Heiser, 1949). The other race is weedy (Spolsky and Uzzell, 1984) have also un­ and invades fields, ditches, and roadsides covered possible cases of introgression. in the central valley of California; it is re­ Here we examine allozyme, chloroplast­ ferred to as the weedy race of H. bolanderi DNA, and nuclear-ribosomal-DNA varia­ (Heiser, 1949). The weedy race of H. bo­ tion among populations of H. annuus and landeri is sympatric with H. annuus in the H. bolanderi. As part of an exhaustive ex­ central valley, and the two taxa form exten- amination of introgression in these taxa INTROGRESSION IN HELIANTHUS 229

(Rieseberg, 1987), some of the populations (I.dh-l, 2), leucine aminopeptidase (Lap-I,2), employed in this study have also been ex­ malate dehydrogenase (Mdh-2,3), malic en­ amined using flavonoid chemistry and mor­ zyme (Me-I), phosphoglucoisomerase (Pgi­ phology. Two specific questions are ad­ 1,2), phosphoglucomutase (Pgm-I,3), dressed in this study: 1) are allozyme, 6-phosphogluconate dehydrogenase (Pgd-3), cpDNA, and rDNA data consistent with the and triosephosphate isomerase (Tpi-I,2,3). hypothesized origin ofthe weedy race ofH. All enzymes were resolved on 12.5% starch bolanderi through introgression? 2) are DNA gels. Gel and electrode buffer systems used restriction-site data consistent with evi­ are given in Rieseberg et al. (1988). dence from flavonoids and allozymes? DNA Isolation. -High-molecular-weight total DNA (nuclear, chloroplast, and mi­ MATERIALS AND METHODS tochondrial) was isolated for analysis of Plants. -Achenes for isozyme studies rDNA variation using the method ofDoyle were gathered from 43 populations through­ and Doyle (1987). Chloroplast DNAs from out North America, including seven pop­ pooled single-population samples ofH. an­ ulations ofthe serpentine race ofH. bolan­ nuus (AI) and serpentine H. bolanderi (Sl) deri, four populations of the weedy race of and from three individuals from the hybrid H. bolanderi, and 32 populations of H. an­ swarm (H) were prepared from fresh leaves nuus. Achenes were collected from mature following the protocol ofPalmer(1982). For heads of3o-l00 individuals per population, the remaining populations (S2-S5, Wl-W4, pooled, and used for enzyme electropho­ A2-A9, H, M), as well as for individual resis. A list of populations and localities is samples from populations Al and SI, chlo­ given in Rieseberg (1987). roplasts were isolated from fresh leaves of For cpDNA and rDNA variation, five individual plants using the sucrose-gradient populations ofthe serpentine race ofH. bo­ procedure described in Palmer (1982, 1986). landeri (Sl-S5), four populations of the Chloroplast DNAs were then isolated from weedy race of H. bolanderi (WI-W4), nine the purified chloroplast preparations fol­ populations of H. annuus (Al-A9), and a lowing the method of Doyle and Doyle hybrid swarm involving H. annuus and both (1987), starting with the addition ofhot ex­ races of H. bolanderi (H) were examined traction buffer to the cpDNA pellet. (see Rieseberg [1987] for locality data). In Restriction-Site Comparisons. - Restric­ addition, one population ofH. maxmiliani tion-endonuclease digestions, agarose-gel (M) was used as an outgroup to polarize the electrophoresis, bidirectional transfer of direction of restriction-site mutations. Be­ DNA from agarose gels to nitrocellulose fil­ cause relationships within Helianthus sec­ ters, preparation of32P-labeled probes, and tion Helianthus are unclear (Schilling and filter hybridizations were conducted follow­ Heiser, 1981; Chandler et aI., 1986), the ing the methods of Palmer (1986). Prepa­ outgroup was chosen from the most closely ration and detection ofbiotin-labeled probes related section in Helianthus, section Di­ followed the instructions of the manufac­ varicatii (Heiser et aI., 1969). We sampled turer (Bethesda Research Laboratories). 1-4 individuals per population for both Thirty-six restriction endonucleases were cpDNA and rDNA variation, with the ex­ used to examine cpDNA variation in He­ ception ofthe hybrid population, from which lianthus, including 30 endonucleases that 30 individuals were analyzed for cpDNA recognize six-bp (base pair) sequences, four variation only. endonucleases that recognize five-bp pair Enzyme Electrophoresis. -Sample prep­ sequences, and two endonucleases that rec­ aration, electrophoresis, and staining ofen­ ognize four-bp sequences. Restriction-site zymes followed the general methodology of mutations were detected by visualizing aga­ Soltis et al. (1983). The following enzymes rose gels stained with ethidium bromide un­ (loci in parentheses) were examined: alcohol der ultraviolet light. Restriction-site muta­ dehydrogenase (Adh-I,2), aldolase (Ald­ tions were then characterized by hybridizing 1,2,3), (3-galactosidase «(3 Gal), fructose-l ,6­ with clones containing Sac I fragments from diphosphatase (Fdp-I,2), glutamate dehy­ lettuce (Jansen and Palmer, 1987a, 1987b). drogenase (Gdh), isocitrate dehydrogenase The 15 clones utilized, representing 95% of 230 L. H. RIESEBERG ET AL.

the chloroplast genome, were divided into among the three taxa range from 0.96 be­ four groups, and each group of clones was tween either race of H. bolanderi and H. used as a single probe. Given the similarity annuus to 0.98 between serpentine H. bo­ of Helianthus cpDNAs « 1% sequence di­ landeri and weedy H. bolanderi. The mean vergence), it was not necessary to map re­ genetic identity for comparisons ofall pop­ striction-site mutations to interpret the ob­ ulations of H. annuus and H. bolanderi ex­ served fragment differences. amined is 0.97. To analyze rONA variation in Helian­ Although H. annuus and the two races of thus, 22 restriction endonucleases were used, H. bolanderi are very similarat genes coding including 17 endonucleases that recognize for enzymes, each taxon possesses unique six-bp sequences, and five endonucleases alleles, as well as alleles detected in only one that recognize five-bp sequences. Two of the other two taxa. The distribution of probes were used to detect homologous seg­ these "marker alleles" was used to examine ments of Helianthus rONA: 1) pHA-l, a the hypothesized origin of the weedy race plasmid containing a single l8S-25S rONA ofH. bolanderi through introgression. A to­ repeat from Pisum (Jorgensen et al., 1987) tal of 37 low-frequency alleles distinguish and 2) pGMr-l, a plasmid containing a sin­ the serpentine race of H. bolanderi from H. gle l8S-25S rONA repeat from Glycine max annuus. Weedy H. bolanderi combined only (Doyle and Beachy, 1985). Ribosomal four of the 37 alleles exclusive to the pu­ DNAs within and among populations of tative parental species. Serpentine H. bo­ Helianthus annuus and H. bolanderi were landeri, however, combined seven ofthe 35 very similar, and fragment differences be­ alleles differentiating weedy H. bolanderi tween mutated ONAs were easily inter­ from H. annuus, and H. annuus combined preted without detailed mapping of restric­ seven ofthe 16 alleles differentiating the two tion sites. races of H. bolanderi. Data Analysis. -Genetic-identity values Chloroplast-DNA Variation. - The chlo­ were computed following Nei (1972) for all roplast genome of Helianthus is similar in pairwise population comparisons using the size and organization to the vast majority BIOSYS-l program ofSwofford and Selan­ of angiosperms examined (Palmer, 1985). der (1981). Chloroplast-DNA sequence-di­ The only major difference is a 22-kb inver­ vergence values were computed using the sion in the large single-copy region of most equations of Brown et al. (1979). Ribo­ Compositae species, including species of somal-DNA sequence-divergence values Helianthus (Jansen and Palmer, 1987b). Our were not computed, because it is likely that size estimate of 151 kb for the Helianthus additional site mutations are present in the chloroplast genome is very similar to size nontranscribed intergenic spacer but were values obtained in previous studies ofCom­ not detected because ofnonhomology to the positae species (reviewed in Jansen and probes used. Thus, these values could bias Palmer [1987a]). estimates of sequence divergence. Phylo­ Variation among cpDNAs of Helianthus genetic analyses ofcpDNA and rONA data was analyzed by digesting each cpDNA with were conductedusing the computerprogram 36 restriction endonucleases and comparing "Phylogenetic Analysis Using Parsimony" 507 restriction sites (approximately 2,927 (PAUP, version 2.3) developed by Swofford nucleotide bases). Chloroplast-DNA mu­ (1983). For each data set, cladistic analysis tations encountered in H. annuus and H. was performed using a designated outgroup. bolanderi are summarized in Table 1. All 17 fragment-pattern differences detected RESULTS were diagnosed as restriction-site mutations Allozyme Similarity and the Distribution (point mutations). No length mutations or of Taxon-Specific Alleles. -Genetic-iden­ inversions were detected. Polarity ofrestric­ tity values within and among the three taxa tion-site mutations is given relative to H. under investigation are extremely high. maxmiliani. Mean genetic identities within each taxon Four restriction-site mutations differen­ range from 0.97 for H. annuus to 0.98 for tiate all populations ofH. bolanderi (Sl-S5 each race of H. bolanderi. Mean identities and Wl-W4) from all populations of H. INTROGRESSION IN HELIANTHUS 231

TABLE 1. Chloroplast-DNA restriction-fragment changes in Helianthus. Only well resolved fragments were scored. Therefore, fragments below I kb in size generally were not scored. Each restriction-site change is inferred to result from a single point mutation within a restriction site. The evolutionary direction of restriction-site mutations is given relative to H. maxmiliani. No point mutations were observed with Bel I (14 fragments), Bgi I (10 fragments), BstE II (9 fragments), BstN I (13 fragments), BstX I (14 fragments), Cfo I (16 fragments), EcoR V (20 fragments), Hae II (13 fragments), Hpa I (13 fragments), Hpa II (II fragments), Nco I (15 fragments), Nde I (16 fragments), Nhe I (5 fragments), Nru I (15 fragments), Pst I (9 fragments), Pvu I (9 fragments), Sac II (II fragments), Sal I (9 fragments), SnaB I (16 fragments), Ssp I (21 fragments), Tthlll I (10 fragments), or Xba I (14 fragments).

Changed fragments Number of Restriction enzyme fragments scored Loss Gains Populations with mutated DNAs Apa I 9 3.8 + 3.8 7.6 SI-S5 Ava I 22 3.7 + 7.3 10.0 SI-S5 BamH I 19 9.3 5.8 + 3.5 SI-S5 BglII 20 5.5 5.1 + 0.4 WI-W4 Cia I 22 3.7 + 0.4 4.1 WI-W4 Dra I 17 6.0 5.1 + 0.9 SI-S5 3.0 2.3 + 0.7 AI-A9 EcoR I 17 9.5 6.0 + 3.5 SI, S5 Hind III 15 9.0 8.3 + 0.7 SI-S5, WI-W4 Kpn I 8 14.7 + 3.1 17.8 AI-A2, A5-A7, A9 Nci I 13 3.2 + 0.5* 3.7 A8 Sac I 17 4.6 3.4 + 1.2 WI-W4 Spe I 15 21.0 + 4.2 25.2 SI-S5, WI-W4 Stu I II 5.5 3.7 + 1.8 SI-S5 16.3 15.4 + 0.9 AI-A9 Xmn I 19 3.7 3.0 + 0.7 AI-A7, A9 1.7 + 0.4* 2.1 A4 * Fragment too small to be visualized. annuus (Al-A9). Populations of H. bolan­ viduals from a hybrid swarm demonstrated deri share derived Hind III and Spe I re­ the presence of three different cpDNAs in striction sites, whereas all populations ofH. the population. Twenty-four of the 30 in­ annuus share derived Dra I and Stu I re­ dividuals assayed possessed a chloroplast striction sites (Table 1). The serpentine race genome identical to that ofpopulations Wl­ ofH. bolanderi (S1-S5) is characterized by W4, five plants possessed a chloroplast ge­ five unique restriction-site mutations (Apa nome identical to that of population A3, I, Ava I, BamH I, Dra I, and Stu I; Table and one plant had a chloroplast genome 1), and populations of weedy H. bolanderi identical to that of populations S2-S4. No (Wl-W4) can be differentiated from all oth­ novel cpDNAs were observed in the hybrid er populations by three restriction-site mu­ swarm. tations (Bgl II, CIa I, and Sac I; Table 1). Sequence-divergence values (Table 2) Intraspecific cpDNA variation was ob­ among Helianthus cpDNAs were calculated served in both H. bolanderi and H. annuus. using the 30 six-bp pair restriction enzymes. The two races ofH. bolanderi differ by eight Sequence-divergence values among popu­ or nine restriction-site mutations, depend­ lations ofweedy H. bolanderi, serpentine H. ing on the populations compared (Table 1). bolanderi, and H. annuus were 0.00%, However, no mutations were observed 0.02%, and 0.04%, respectively. Mean among populations of weedy H. bolanderi, cpDNA sequence divergence among these and only a single EcoR I restriction-site mu­ three taxa ranged from 0.30% between the tation was observed among populations of two races ofH. bolanderi to 0.40% between serpentine H. bolanderi. In H. annuus, three serpentine H. bolanderi and H. annuus. The restriction-site mutations serve to differ­ average sequence divergence was 0.35%. entiate populations. A phylogenetic tree based on the 17 Examination of the cpDNAs of 30 indi- cpDNA restriction-site mutations listed in 232 L. H. RIESEBERG ET AL.

51.55 bosomal-RNA genes in higher plants are

5 organized as arrays of tandemly repeated genes (reviewed in Appels and Honeycutt 0 52-54 [1986]). Ribosomal RNA is transcribed as 2 a single transcript and subsequently pro­ cessed into 18S, 5.8S, and 25S rRNAs. Each 3 Wl-W4 repeat unit contains both a ribosomal-RNA transcription unit and an intergenic spacer (IGS) sequence that separates the transcrip­ 1 Al-A2. tion units of adjacent repeats. The rRNA M A5-A7.A9 coding sequence is generally highly con­ served, whereas the spacer regions are often 1 0 A3 highly variable in both sequence and length within and among individuals, populations, and species (e.g., Doyle et aI., 1984; Polans 2 I A4 et aI., 1986; Sytsma and Schaal, 1985b; Sag­ hai-Maroof et aI., 1984; Appels and Dvo­ rak, 1982; Appels et aI., 1980). The IGS 1 AS sequence usually contains a region ofsmall tandem subrepeats, and length heteroge­ FIG. 1. Phylogenetic tree of 18 Helianthus popu­ lations based on 17 restriction-site mutations. The tree neity generally results from changes in copy is rooted relative to H. maxmiliani restriction sites. number of the subrepeats. Population descriptions are given at the ends of the No repeat-length heterogeneity was ob­ branches, and the number ofmutations is given above served within individuals, populations, or each branch. species of Helianthus. However, H. max­ miliani possessed a repeat of 10.5 kb, Table I is given in Figure l. The PAUP whereas H. annuus and the two races ofH. program found a single most parsimonious bolanderi possessed a 9.9-kb repeat. The tree, which required only 17 mutations to complete lack of rDNA repeat-length vari­ account for the observed distribution ofthe ation within Helianthus species contrasts seven cpDNA phenotypes. Thus, there is no sharply with the extensive variation ob­ need to postulate any convergent restric­ served within numerous other species (e.g., tion-site mutations. However, populations Saghai-Maroof et aI., 1984; Doyle et aI., [AI-A2, A5-A7, A9] form an unresolved 1984; Appels and Dvorak, 1982; Polans et trichotomy with populations A3 and A4 aI., 1986). This rDNA homogeneity greatly (Fig. I). It is likely that the observed tri­ simplified intra- and interspecific rDNA chotomy simply reflects a lack ofpotentially comparisons in Helianthus. resolving characters. However, in contrast to the lack ofrepeat­ Ribosomal-DNA Variation. -Nuclear ri- length heterogeneity within single individ-

TABLE 2. Sequence differences among Helianthus chloroplast DNAs. The number ofrestriction-site mutations between any two cpDNAs is given in the upper right-hand section ofthe matrix; values given in the lower left­ hand section are the percentages of sequence divergence (Brown et aI., 1979). Divergence estimates are based on 30 six-bp restriction enzymes. We compared a total of2,514 bp (419 six-bp restriction sites; Table I).

Populations [AI-A2, Populations [SI. S51 [S2-S41 [WI-W4} A5-A7, A91 [A3] [A4] [A8] [SI, S5] 8 11 10 11 9 [S2-S4] 0.04 7 10 9 10 8 [WI-W4] 0.32 0.28 9 8 9 7 [AI-A2, A5-A7, A9] 0.44 0.40 0.36 I 2 2 [A3] 0.40 0.36 0.32 0.04 I I [A4] 0.44 0.40 0.36 0.08 0.04 2 [A8] 0.36 0.32 0.28 0.08 0.04 0.08 INTROGRESSION IN HELlANTHUS 233 uals, a limited amount of restriction-site o 51-55 polymorphism was observed among repeat 1 units within single individuals in Helian­ I thus. These restriction-site polymorphisms Wl-W4 may be due to nucleotide substitutions, dif­ ferential methylation, and/or partial digests M 0 and were observed among repeat units using AI-AS

BamH I and EcoR I which recognize se­ 2 quences prone to methylation (Gerlach and 1 Bedbrook, 1979; Siegel and Kolacz, 1983). /llrA9 That rDNA restriction sites are polymor­ FIG. 2. Phylogenetic tree of 18 Helianthus popu­ phic within single plants may simply reflect lations based on five rDNA restriction-site mutations. that these species are of ancient tetraploid The tree is rooted relative to H. maxmiliani restriction (N = 17) origin (Jackson and Murray, 1983). sites. Population designations are given at the ends of branches, and the numbers of mutations are given above Isozyme number also reflects the polyploid the branches. status of these species (Rieseberg and Soltis, unpubl.). Additional isozymes per enzyme, relative to the number ofisozymes annuus share derived BamH I, and Nco I characteristic of diploid seed plants (Gott­ restriction sites (Table 3). Little intraspecific lieb, 1982), were observed for aldolase, rDNA variation was observed in H. annuus fructose-I,6-diphosphatase, malic enzyme, and H. bolanderi (Table 3). A Spe I restric­ phosphoglucomutase, 6-phosphogluconate tion-site mutation differentiates the weedy dehydrogenase, and triosephosphate isom­ and serpentine races of H. bolanderi, and a erase. Hinc II mutation divides H. annuus into Although no rDNA repeat-length heter­ two groups. ogeneity was observed, ribosomal-DNA re­ A phylogeny based on five rDNA restric­ striction-site variation was detected among tion-site mutations is given in Figure 2. The populations ofH. annuus and H. bolanderi PAUP program found a single most parsi­ (Table 3). We observed a total of five re­ monious tree that required only five mu­ striction-site mutations. Polarity of restric­ tations to account for the observed distri­ tion-site mutations is given relative to H. bution ofthe four rDNA phenotypes. Once maxmiliani. Three restriction-site muta­ again, there is no need to postulate any con­ tions differentiate all populations of H. bo­ vergent restriction-site mutations. landeri (Sl-S5 and Wl-W4) from all pop­ Chloroplast- and Ribosomal-DNA Phy­ ulations ofH. annuus (A l-A9). Populations logeny. -A phylogenetic tree was construct­ of H. bolanderi share derived Hinc II re­ ed from pooled rDNA andcpDNAdata (Fig. striction sites, whereas populations of H. 3). A single most parsimonious tree was

TABLE 3. Restriction-fragment changes of nuclear ribosomal DNA in Helianthus. Each restriction-site change is inferred to result from a single point mutation within a restriction site. The evolutionary direction ofrestriction­ site mutations is given relative to H. maxmiliani. No point mutations were observed with Ban I (3 fragments), BstE II (I fragment), Dra I (2 fragments), EcoR I (2 fragments), EcoR V (I fragment), Hind III (I fragment), Nde I (I fragment), Nhe I (I fragment), Sac I (3 fragments), Ssp I (I fragment), Tthlll I (3 fragments), Xba I (I fragment), or Xmn I (2 fragments). Bel I, BgllI, BstN I, Nci I, and Stu I apparently recognize no sites in the rDNA repeat. At least one additional mutation has occurred at Hinc II.

Changed fragments Number of Restriction enzyme fragments scored Losses Gains Mutated DNA, BamH I 4-5 3.1 + 1.8 4.9 AI-A9 Hinc II 3 4.5 + 0.2* 4.7 A6-A9 4.5 + 0.4* 4.9 SI-S5, WI-W4 Nco I 2-3 9.0 4.8 + 4.2 AI-A9 Spel 0-1 9.9 ** WI-W4 * Fragment not detected by probe; probably occurs in the highly diverged non transcribed spacer region. ** No cleavage sites. 234 L. H. RIESEBERG ET AL.

5\,55 DISCUSSION

5 Allozyme Variation and Introgression.­ 0 52-54 The high genetic identities observed be­ 3 tween serpentine H. bolanderi, weedy H. bo­ landeri, and H. annuus (l = 0.97) indicate 4 WI-W4 that, for genes encoding enzymes, little ge­ netic differentiation among taxa has oc­ curred. In fact, these values are more similar 0 Al-A2, to genetic identities reported for con specific AS I plant populations ofother plant groups than M to values comparing other congeneric plant I' A6-A7. species (Gottlieb, 1981; Crawford, 1983). A9 Because of the high level of allozymic 1 similarity, the electrophoretic data set can­ 0 A3 not be used to test the introgression hy­ pothesis formally. This is because only a few

4 ofthe 37 low-frequency marker alleles would 1 A4 be expected in the interspecific F 1 hybrids (due to the rarity of these alleles). Never­ theless, the observation that serpentine H. 2' A8 bolanderi shares three more marker alleles with H. annuus than does weedy H. bolan­ FIG. 3. Phylogenetic tree of 18 Helianthus popu­ deri certainly does not support the introgres­ lations using pooled cpDNA and rDNA data. The tree is rooted relative to H. maxmiliani restriction sites. sion model. Instead, these results are similar Population designations are given at the ends of to expectations for race divergence from a branches, and the numbers ofmutations are given above single common ancestor and suggest that, if the branches. Convergent mutations are designated by introgression occurred, the majority of stars. marker alleles were rapidly lost following the initial hybridization event. found for the eight DNA phenotypes. This In contrast to the allozyme data, much rooted tree (Fig. 3) required only one ad­ greater differentiation was observed be­ ditional mutation (24 steps) to account for tween the chloroplast and ribosomal DNAs the observed data. The consistency index ofthese species, making the latter two data was 0.96, and the rate of convergence was sets most useful for testing the introgression 4.2%. The single convergence occurred hypothesis. Ribosomal-DNA variation can among populations of H. annuus and was be used to detect hybridization and in­ observed only after combining the two data trogression, because rDNA should be ad­ sets. The apparent convergent mutation oc­ ditively combined in hybrids or introgres­ curred in population A8 and in populations sants. Chloroplast DNA can be particularly [A6-A7, A9]. Population A8 has a very dis­ useful for studying introgression, because tinct chloroplast genome relative to the oth­ cpDNAs do not recombine. A cpDNA from er three populations, but all four popula­ one oftwo putative parents should be pres­ tions have an identical rDNA phenotype. ent in hybrids or introgressants (not a Two hypotheses may account for the ap­ cpDNA unlike either of them). parent convergence: I) parallel restriction­ Chloroplast-DNA Sequence Diver­ site mutations occurred in two lineages of gence.- Mean sequence divergence ranged H. annuus; or 2) exchange of biparentally from 0.3 to 0.4% among the three Helian­ inherited rDNA genes occurred among pop­ thus taxa analyzed. These estimates are sim­ ulations ofH. annuus in the absence ofcy­ ilar to values obtained for closely related toplasmic gene flow. A trichotomy was ob­ species ofLycopersicon (0-0.7%; Palmerand served among the same populations of H. Zamir, 1982), Pennisetum (0.4%; Clegg et annuus that were involved in the trichoto­ aI., 1984), Pisum (0.1-0.8%; Palmer et aI., my associated with the cpDNA phyloge­ 1985), and Lisianthius (0-0.3%; Sytsma and netic tree. Schaal, 1985b), but generally lower than es- INTROGRESSION IN HELIANTHUS 235 timates obtained in Brassica (0.3-2.6%; gent cpDNAs (0.3-0.4% sequence diver­ Palmer et al., 1983), Clarkia (0.2-1.6%; gence). This result is in explicit contrast to Sytsma and Gottlieb, 1986), and Linum (0­ the situation in Lisianthius (Sytsma and 6%; Coates and Cullis, 1987). However, Schaal, 1985a, 1985b), where much lower mean sequence divergence between the two levels of cpDNA variation (0-0.3%) were races ofH. bolanderi is extremely high given observed relative to isozyme variation (i = the proposed recent divergence ofthese lin­ 0.62). The high degree ofisozyme similarity eages (Heiser [1949] suggests that the weedy in Helianthus could be attributed to selec­ race ofH. bolanderi may have been derived tive control ofisozymes (or linked genes) or through introgression as recently as during to gene flow and recombination. Gene flow, the past several hundred years). In fact, this however, is an unlikely mechanism for much cpDNA sequence divergence has nev­ maintaining homogeneity between popu­ er been observed within a plant species be­ lations ofH. annuus and H. bolanderi, given fore. Zurawski et al. (1984) estimated that their lengthy geographic isolation. the synonymous rate ofsequence change of Phylogenetic Relationships. -Analysis of the cpDNA-encoded rbcL gene among biparentally inherited rONA and mater­ grasses is 1 x 10-9 nucleotide substitutions nally inherited cpDNA (Tilney-Bassett, per year. IfH. bolanderi cpDNA is evolving 1978) provides a phylogenetic history of at a similar rate (however unrealistic this populations ofH. annuus and H. bolanderi. assumption may be), estimated time of di­ In phylogenetic trees derived from both vergence between the weedy and serpentine cpDNA and rONA data, the alignment of race is approximately 3,000,000 years. Thus, the two races ofH. bolanderi and H. annuus cpDNA-divergence estimates suggest that did not change (Figs. 1-3). The two races of the two races of H. bolanderi may be far H. bolanderi form a monophyletic lineage more ancient than was previously hypoth­ supported by three synapomorphies (two esized, and, on the basis of cpDNA data cpDNA and one rONA mutations). Like­ alone, these races could be viewed as sep­ wise, all populations of H. annuus form a arate species. Alternatively, cpDNA diver­ monophyletic lineage supported by four gence may have occurred rapidly in these synapomorphies (two cpDNA and two two lineages. If we postulate that the two rONA mutations). lineages diverged approximately 1,000 years Phylogenetic analysis of cpDNA and ago, a 3,000-fold increase in evolutionary rONA restriction-site variation also re­ rates (relative to the estimates of Zurawski solved populations within the serpentine et al., [1984]) would have to be postulated race of H. bolanderi and within H. annuus to account for the observed cpDNA diver­ (Fig. 3). Two DNA phenotypes were ob­ gence in H. bolanderi. It is possible that served in serpentine H. bolanderi. Popula­ interaction ofchloroplast and nuclear genes tions S1 and S5 correspond to the extreme from different parents in later generations "exilis" form of serpentine H. bolanderi of interspecific hybrids might result in un­ (Oliveri and Jain, 1977), whereas popula­ stable cpDNAs, increasing the rate of tions S2-S4 are the "foothills" form of ser­ cpDNA evolution in Helianthus. However, pentine H. bolanderi. In contrast, DNA the complete absence ofnovel cpDNA phe­ variation among populations of H. annuus notypes in 30 individuals from a 45-year­ does not correlate well with geographic dis­ old hybrid swarm containing F, and later­ tribution or morphological variation. The generation plants (Stebbins and Daly, 1961) poor correlation probably results from re­ suggests that interspecific hybridization and peated episodes of long-distance dispersal backcrossing do not produce rapid cpDNA by man (Heiser, 1949, 1954). However, changes in Helianthus. some insight into the evolutionary history Evidence from cpDNA and evidence from of this species can be gleaned. First, it is isozymes provide conflicting estimates of clear that population A8, which represents genetic divergence between the two races of H. annuus ssp. texanus (Heiser, 1954), was H. bolanderi and between H. bolanderi and derived early in the evolution ofH. annuus. H. annuus. The Helianthus populations The remaining eight populations corre­ compared were very similar at genes en­ spond to H. annuus ssp. lenticularis. Heiser coding enzymes (i = 0.97), but had diver- (1949, 1954) suggests that ssp. lenticularis 236 L. H. RIESEBERG ET AL.

has been introduced into California on nu­ Given the complete agreement ofthese four merous occasions. This contention is clearly data sets, we conclude that the weedy race supported by the fact that all three Califor­ of H. bolanderi did not originate through nia populations analyzed have different introgression of genes from H. annuus. DNA phenotypes. Finally, the three popu­ An alternative hypothesis to account for lations from the Midwest have identical race formation in H. bolanderi may be iso­ DNA phenotypes (A6-A7 and A9). Given lation via discontinuities in geology and the abundance ofH. annuus in the Midwest, edaphics. The adaptation of certain popu­ these populations may possess the most lations of H. bolanderi to serpentine soils common DNA phenotype in H. annuus. and subsequent reduction in gene flow may Origin of the Weedy Race of H. bolan­ have initiated race formation in this species. deri.- Discrepancies between nuclear and We have no evidence in support ofthis hy­ cytoplasmic phylogenies have provided evi­ pothesis, other than the observed distribu­ dence suggesting the occurrence of in­ tion of the weedy and serpentine races in trogression in Brassica (Palmer et al., 1983), California. Pisum (Palmer et al., 1985), Drosophila In conclusion, primary reliance on mor­ (Powell, 1983; Solignac and Monnerot, phological data for studies of introgression 1986), and Mus (Ferris et al., 1983). How­ may be misleading. As we have demonstrat­ ever, with the exception ofa single conver­ ed in this study, precise genetic markers, gent mutation within H. annuus, no dis­ representing both nuclear and cytoplasmic crepancies were observed between genomes, are the most appropriate tools for cytoplasmic and nuclear phylogenetic trees the study of introgression. Given the pre­ in Helianthus. A hybridization or introgres­ sumed importance ofintrogression in plant sion event might result in reticulate patterns evolution and the ready availability of ap­ of evolution in a phylogenetic tree con­ propriate methodologies, unambiguous structed using biparentally inherited char­ documentation of introgression (or, con­ acters. However, no convergent or back mu­ versely, the documentation of the absence tations were observed in the Helianthus of introgression) in other plants is impera­ phylogenetic tree constructed using rDNA tive. Only with the accumulation of data mutations, nor were any additive rDNA from other plant groups will we be able to phenotypes observed in weedy H. bolanderi assess accurately the role ofintrogression in plants, although additive rDNA profiles the evolution of vascular plants. were observed for the majority of individ­ uals from the hybrid swarm. Finally, if in­ ACKNOWLEDGMENTS trogression were recent (as proposed for the Research was supported by NSF Grant origin of the weedy race of H. bolanderi), BSR 8315088 to D.E.S., NSF Grant BSR the presumed introgressant type would be 8415934 to J.D.P., a NSF Dissertation Im­ expected to possess the chloroplast genome provement Grant to L.H.R. and D.E.S., a ofone ofthe two parent species. Therefore, Sigma Xi Award to L.H.R., and a Wash­ the presence ofa unique and highly diverged ington State University Travel Grant to chloroplast genome in the weedy race ofH. L.H.R. We thank the following individuals bolanderi, when compared to cpDNAs of D. Crawford, J. Doyle, A. J. Gilmartin, L. serpentine H. bolanderi and H. annuus, sug­ Gottlieb, R. Jansen, P. Soltis, and J. gests that this race was not derived through Thompson for laboratory assistance, help­ introgression during the past several ful discussions, and critical reviews of the hundred years, as hypothesized, but is prob­ manuscript. ably relatively ancient in origin. Evidence from cpDNA and rDNA vari­ LITERATURE CITED ation is consistent with foliar and floral fla­ ALSTON, R. E., AND B. L. TURNER. 1963. Natural vonoid variation (Rieseberg and Soltis, hybridization among four species of Baptisia (Le­ 1988) and with allozymic variation. Al­ guminosae). Amer. J. Bot. 50: 159-173. ANDERSON, E. 1949. Introgressive Hybridization. though more than 60 taxon-specific markers Wiley, N.Y. were utilized in these studies, no evidence ApPELS, R., AND J. DVORAK. 1982. The wheat ribo­ of interspecific gene transfer was observed. somal spacer region. Its structure and variation in INTROGRESSION IN HELIANTHUS 237

populations and among species. Theoret. Appl. Ge­ FERRIS, S. D., R. D. SAGE, C.-M. HUANG, J. T. NIELSEN, net. 63:337-348. U. RITTE, AND A. C. WILSON. 1983. Flow ofmi­ ApPELS, R., W. L. GERLACH, E. E. DENNIS, H. SWIfT, tochondrial DNA across a species boundary. Proc. AND W. J. PEACOCK. 1980. Molecular and chro­ Nat. Acad. Sci. USA 80:2290--2294. mosomal organization of DNA sequences coding GERLACH, W. L., ANDJ. R. BEDBRooK. 1979. Cloning for ribosomal RNAs in cereals. Chromosoma 78: and characterization ofribosomal DNA genes from 293-311. wheat and barley. Nucl. Acids Res. 7:1869-1885. ApPELS, R., AND R. L. HONEYCUTT. 1986. rDNA evo­ GOTTLIEB, L. D. 1981. Electrophoretic evidence and lution over a billion years, pp. 81-135. In S. K. plant populations. Prog. Phytochem. 7:1-46. Dutta (ed.), DNA Systematics. CRC Press, Boca ---. 1982. Conservation and duplication of iso­ Raton, FL. zymes in plants. Science 216:373-380. AVISE, J. C., AND N. C. SAUNDERS. 1984. Hybridiza­ GRANT, V. 1950. Genetic and taxonomic studies in tion and introgression among species ofsunfish (Le­ Gila. Aliso 2:239-316. pomis): Analysis of mitochondrial DNA and allo­ ---. 1981. Plant Speciation. Columbia Univ. Press, zyme markers. Genetics 108:237-255. N.Y. AVISE, J. c., AND M. H. SMITH. 1974. Biochemical GYLLENSTEN, U., R. F. LEARY, F. W. ALLENDORF, AND characters of sunfish. I. Geographic variation and A. G. WILSON. 1985. Introgression between two subspecific intergradation in the bluegill, Lepomis cutthroat trout subspecies with substantial karyo­ macrochirus. Evolution 28:42-56. typic, nuclear and mitochondral genomic diver­ BELL, N. B., AND L. J. LESTER. 1978. Genetic and gence. Genetics 111:905-915. morphological detection ofintrogression in a c1inal HEISER, C. 8. 1947. Hybridization between the sun­ population of Sabatia section Campestria (Gen­ flower species Helianthus annuus and H. petiolaris. tianaceae). Syst. Bot. 3:87-104. Evolution 1:249-261. BLOOM, W. L. 1976. Multivariate analysis of the in­ ---. 1949. Study in the evolution ofthe sunflower trogressive replacement ofClarkia nitens by Clark­ species Helianthus annuus and H. bolanderi. Univ. ia speciosa polyantha (Onagraceae). Evolution 30: Calif. Publ. Bot. 23: 157-196. 412-424. ---. 1951a. Hybridization in the annual sunflow­ BROWN, W. M., M. GEORGE, JR., AND A. C. WILSON. ers: Helianthus annuus x H. argophyllous. Amer. 1979. Rapid evolution of animal mitochondrial Natur. 85:64-72. DNA. Proc. Nat. Acad. Sci. USA 76:1967-1971. ---. 1951b. Hybridization in the annual sunflow­ CARR, C. M., S. W. BALLINGER, J. N. DERR, L. H. ers: Helianthus annuus x H. debilis var. cu­ BLANKENSHIP, AND J. W. BICKHAM. 1986. Mito­ cumerifolius. Evolution 5:42-51. chondrial DNA analysis of hybridization between 1954. Variation and subspeciation in the sympatric white-tailed deer and mule deer in west common sunflower, Helianthus annuus. Amer. Texas. Proc. Nat. Acad. Sci. USA 83:9576-9580. MidI. Natur. 51:287-305. CHANDLER, J. M., C. JAN, AND B. H. BEARD. 1986. --. 1961. Morphological and cytological varia­ Chromosomal differentiation among the annual tion in Helianthus petiolaris with notes on related Helianthus species. Syst, Bot. 11:353-371. species. Evolution 15:247-258. CLEGG, M. T., J. R. Y. RAWSON, AND K. THOMAS. ---. 1965. Sunflowers, weeds and cultivated plants, 1984. Chloroplast DNA variation in pearl millet pp. 391-401. In H. G. Baker and G. L. Stebbins and related species. Genetics 106:449-461. (eds.), The Genetics of Colonizing Species. Aca­ COATES, D., AND A. C. CULLIS. 1987. Chloroplast demic Press, N.Y. DNA variability among Unum species. Amer. J. ---. 1973. Introgression re-examined. Bot. Rev. Bot. 74:260--268. 39:347-366. CRAWFORD, D. J. 1983. Phylogenetic and systematic HEISER, C. 8., D. M. SMITH, B. S. CLEVENGER, AND W. inferences from electrophoretic studies, pp. 257­ C. MARTIN. 1969. The North American Sunflow­ 287. In S. D. Tanksley and T. J. Orton (eds.), Iso­ ers (Helianthus). Mem. Torrey Bot. Club 22: 1-218. zymes in Plant Genetics and Breeding. Elsevier, HUNT,W. G., AND R. K. SELANDER. 1973. Biochem­ Amsterdam, Neth. ical genetics of hybridization in European house DAVIS, J. I. 1985. Introgression in Central American mice (Mus musculus). Heredity 31:11-33. Phytolacca(Phytolaccaceae). Amer. J. Bot. 72: 1944­ JACKSON, R. c., AND 8. G. MURRAY. 1983. Colchi­ 1953. cine induced quadrivalent formation in Helianthus: DOEBLEY, J. F., M. M. GOODMAN, AND C. W. STUBER. Evidence ofancient polyploidy. Theoret. Appl. Ge­ 1984. Isoenzymatic variation in Zea (Gramineae). net. 64:219-222. Syst. Bot. 9:203-218. JANSEN, R. K., ANDJ. D. PALMER. 1987a. Chloroplast DOYLE, J. J., AND R. N. BEACHY. 1985. Ribosomal DNA from lettuce and Barnadesia (Asteraceae): gene variation in soybean (Glycine) and its rela­ Structure, gene localization, and characterization of tives. Theoret. Appl. Genet. 70:369-376. a large inversion. Curro Genet. 1:553-564. DOYLE, J. J., R. N. BEACHY, AND W. H. LEWIS. 1984. ---. 1987b. A chloroplast DNA inversion marks Evolution of rDNA in Claytonia polyploid com­ an ancient evolutionary split in the sunflower fam­ plexes, pp. 321-341. In W. F. Grant (ed.), Plant ily (Asteraceae). Proc. Nat. Acad. Sci. USA 84:5818­ Biosystematics: 40 Years Later. Academic Press, 5822. Toronto, ON, Canada. JORGENSEN, R. A., R. E. CUELLAR, W. F. THOMPSON, DOYLE, J. J., AND J. L. DOYLE 1987. A rapid DNA AND T. A. KAVANAGH. 1987. Structure and vari­ isolation procedure for small amounts of fresh leaf ation in ribosomal RNA genes of pea. PI. Molec. tissue. Phytochem. Bull. 19:11-15. BioI. 8:3-12. 238 L. H. RIESEBERG ET AL.

LAMB, T., AND J. C. AVISE. 1986. Directional in­ generic classification of Helianthus (Compositae). trogression ofmitochondrial DNA in a hybrid pop­ Taxon 30:393-403. ulation of tree frogs: The influence of mating be­ SELANDER, R. K., W. G. HUNT, AND S. Y. YANG. 1969. havior. Proc. Nat. Acad. Sci. USA 83:2526-2530. Protein polymorphism and genic heterozygosity in LEVIN, D. A. 1963. Natural hybridization between two European subspecies ofthe house mouse. Evo­ Phlox maculata and P. glaberrima and its evolu­ lution 23:379-390. tionary significance. Amer. J. Bot. 50:714-719. SIEGEL, A., AND K. KOLACZ. 1983. Heterogeneity of ---. 1975. Interspecific hybridization, heterozy­ pumpkin ribosomal DNA. PI. Physiol. 72:166-171. gosity and gene exchange in Phlox. Evolution 29: SOLIGNAC, M., AND M. MONNEROT. 1986. Race for­ 37-51. mation, speciation, and introgression within Dro­ NEI,M. 1972. Genetic distance between populations. sophila simulans, D. mauritiana, and D. sechellia Amer. Natur. 106:283-292. inferred from mitochondrial DNA analysis. Evo­ OLIVERI, A. M., AND S. K. JAIN. 1977. Variation in lution 40:531-539. the Helianthus exilis-bolanderi complex. Madroiio SOLTIS, D. E., C. H. HAUFLER, D. C. DARROW, AND G. 24:177-189. J. GASTONY. 1983. Starch gel electrophoresis of ORNDUFF, R. 1967. Hybridization and regional vari­ ferns: A compilation of grinding buffers, gel and ation in Pacific Northwestern Impatiens (Balsa­ electrode buffers,and staining schedules. Amer. Fern minaceae). Brittonia 19:122-128. J. 73:9-27. PALMER, J. D. 1982. Physical and gene mapping of SOLTIS, D. E., AND P. S. SOLTIS. 1986. Intergeneric chloroplast DNA from Atriplex triangularis and hybridization between Conimitella williamsii and Cucumis sativa. Nucl. Acids Res. 10:1593-1605. Mitella stauropetala (Saxifragaceae). Syst. Bot. II: --. 1985. Evolution of cpDNA and mtDNA in 293-297. plants and algae, pp. 131-240. In R. J. MacIntyre SOLTIS, P. S. 1985. Studies ofgenetic variation in an (ed.), Molecular Evolutionary Genetics. Plenum, introgressive complex in Clarkia (Onagraceae). N.Y. Ph.D. Diss., Univ. Kansas, Lawrence. 1986. Isolation and structural analysis of SPOLSKY, c., AND T. UZZELL. 1984. Natural inter­ chloroplast DNA, pp. 167-186. In A. Weissbach species transfer ofmitochondrial DNA in amphib­ and H. Weissbach (eds.), Methods in Enzymology. ians. Proc. Nat. Acad. Sci. USA 81:5802-5805. Academic Press, Orlando, FL. STEBBINS, G. L. 1959. The role of hybridization in PALMER, J. D., R. A. JORGENSON, AND W. F. THOMPSON. evolution. Proc. Amer. Phil. Soc. 103:231-251. 1985. Chloroplast DNA variation and evolution ---. 1969. The significance of hybridization for in Pisum: Patterns ofchange and phylogenetic anal­ plant taxonomy and evolution. Taxon 18:26-35. ysis. Genetics 109:195-213. STEBBINS, G. L., AND K. DALY. 1961. Changes in the PALMER, J. D., C. R. SHIELDS, D. B. COHEN, AND T. J. variation ofa hybrid population ofHelianthus over ORTEN. 1983. Chloroplast DNA evolution and an eight-year period. Evolution 15:60-71. the origin of amphidiploid Brassica species. The­ SWOFFORD, D. 1983. PAUP, Phylogenetic Analysis oret. Appl. Genet. 65:181-189. Using Parsimony. Illinois Nat. Hist. Surv., Urbana. PALMER, J. D., AND D. ZAMIR. 1982. Chloroplast DNA SWOFFORD, D. L., AND R. B. SELANDER. 1981. BIO­ evolution and phylogenetic relationships in Lyco­ SYS-1. Univ. Illinois, Urbana. persicon. Proc. Nat. Acad. Sci. USA 79:5006-50 10. SYTSMA, K. J., AND B. A. SCHAAL. 1985a. Genetic POLANS, N. 0., N. F. WEEDEN, AND W. F. THOMPSON. variation, differentiation, and evolution in a species 1986. Distribution, inheritance and linkage rela­ complex of tropical shrubs based on isozyme data. tionships of ribosomal spacer length variants in pea. Evolution 39:582-593. Theoret. Appl. Genet. 72:289-295. --. 1985b. Phylogenetics ofthe Lisianthius skin­ POWELL, J. R. 1983. Interspecific cytoplasmic gene neri (Gentianaceae) species complex in Panama uti­ flow in the absence of nuclear gene flow: Evidence lizing DNA restriction fragment analysis. Evolution from Drosophila. Proc. Nat. Acad. Sci. USA 80: 39:594-608. 492-495. SYTSMA, K. J., AND L. D. GOTTLIEB. 1986. Chloro­ RIESEBERG, L. H. 1987. A re-examination of in­ plast DNA evidence for the origin of the genus trogression in Helianthus. Ph.D. Diss., Washington Heterogaura from a species of Clarkia. Proc. Nat. State Univ., Pullman. Acad. Sci. USA 83:5554-5557. RIESEBERG, L. H., AND D. E. SOLTIS. 1988. Origin of TILNEy-BASSETT, R. A. E. 1978. The inheritance and a weedy race of H. bolanderi: Biochemical evi­ genetic behavior of plastids, pp. 251-254. In J. T. dence. Biochem. Syst. Ecol. In press. O. Kirk and R. A. E. Tilney-Bassett (eds.), The SAGHAI-MAROOF, M. A., K. M. SOLIMAN, R. A. JOR­ Plastids. Elsevier, Amsterdam, Neth. GENSEN, AND R. W. ALLARD. 1984. Ribosomal ZURAWSKI, G., M. T. CLEGG, AND A. H. D. BROWN. DNA spacer length polymorphism in barley: Men­ 1984. The nature of nucleotide sequence diver­ delian inheritance, chromosomal location and pop­ gence between barley and maize chloroplast DNA. ulation dynamics. Proc. Nat. Acad. Sci. USA 81: Genetics 106:735-749. 8014-8018. SCHILLING, E. E., AND C. B. HEISER. 1981. An infra- Corresponding Editor: S. N. Handel