Fixation Indices and Genetic Diversity in Hermaphroditic and Gynodioecious Populations of Japanese Chionographis (Liliaceae)

Heredity 68 (1992) 329—336 Received 15Apr11 1991 Genetical Society of Great Britain

Fixation indices and genetic diversity in hermaphroditic and gynodioecious populations of Japanese Chionographis (Liliaceae)

MASAYUKI MAKI Department of Biology, College of Arts and Sciences, University of Tokyo, Komaba, Meguro-ku, Tokyo 153, Japan

InJapanese Chionographis, populations are either hermaphroditic or gynodioecious, and diploid or tetraploid. The average level of inbreeding was compared among these populations by estimating Wright's fixation index. The estimates were significant in all gynodioecious populations irrespective of ploidy level, and hermaphrodites in gynodioecious populations are considered to be selfing. In hermaphroditic populations, f values were not significant in diploid taxa but were significant in tetraploid taxa. These results suggest that gynodioecy evolved only in selfing populations and support a traditional view that the outcrossing advantage of females plays an important role in the evolution of gynodioecy. Estimates of various genetic diversity measures suggest that tetraploid hermaphroditic populations experienced population bottlenecks, and females may have been lost historically.

Keywords:Chionographis,fixation index, genetic diversity, gynodioecy, polyploidy, self-compati- bility.

Introduction females are maintained under various sets of parameter Gynodioecy,a sexual system in which both herma- values (Charlesworth, 1981 ; Delanny eta!., 1981). A phrodites and females occur regularly within a popula- combination of high selfing rate of hermaphrodites and tion, has been reported for various groups of strong inbreeding depression has been viewed as the angiosperms. The reproductive success of females is most plausible condition for the evolution of gynodi- achieved only through ovules, while hermaphrodites oecy (Lloyd, 1975; Charlesworth & Charlesworth, contribute genes to the next generation through both 1978), and thus most experimental work has examined ovules and pollen. Therefore, for females to be main- the selfing rate of hermaphrodites (Sun & Ganders, tained within a population, they must have an advan- 1986, 1988; Wolff et aL, 1988) or inbreeding depres- tage which compensates for a lack of male sion (Shykoff, 1988; Kohn, 1988; Jolls & Chenier, reproductive success. 1989; Sakai eta!., 1989). Recent theorelical studies on Several factors may give females an advantage. the joint evolution of selfing rate and inbreeding These include greater fecundity (Lewis, 1941), depression, however, reveal that inbreeding depression inbreeding depression in hermaphrodites (Lloyd, is expected to decrease with increased selfing rate 1975), and longer reproductive life (Van Damme and (Lande & Schemske, 1985; Charlesworth & Van Delden, 1984). These factors are, however, not Charlesworth, 1987). It is uncertain whether inbreed- mutually exclusive and can work simultaneously to ing depression is strong enough to maintain females in maintain gynodioecy (Charlesworth & Charlesworth, highly selfing populations. 1978; Charlesworth & Ganders, 1979). Theoretical Several theoretical works emphasize that the evo- studies on the joint effects of these factors suggest that lution of gynodioecy depends on the genetic control of male sterility (Charlesworth & Charlesworth, 1978; Correspondence: Masayuki Maki, Department of Biology, College of Arts and Sciences, University of Tokyo, Komaba, Meguro-ku, Charlesworth & Ganders, 1979; Charlesworth, Tokyo 153, Japan. 1981). Under nuclear control of male sterility, gyno- 329 330 M. MAKI dioecy is expected to evolve readily to dioecy, whereas Materials and methods gynodioecy is not maintained in equilibrium under cytoplasmic control of male sterility. With a cytoplas- Study organisms mic—genic inheritance of male sterility, gynodioecy may be stable in equilibrium under some conditions, and it Chionographisis a liliaceous genus and a relative may be maintained in non-equilibrium processes under of North American Chamaelirium luteum, an extens- dynamic interactions between nuclear and cytoplasmic ively studied dioecious species (Meagher & Antonov- factors (Frank, 1989; Belhassen et al., 1989). ics, 1982; Meagher,1984). All speciesof Because the above-mentioned factors combine in Chionographis are perennial herbs that occur in under- complicated ways to facilitate the evolution of gyno- stories of temperate forests. They do not reproduce dioecy, their relative contributions should be carefully vegetatively and individual plants are easily identified. evaluated. For this purpose, comparative studies They initiate flowering in late spring or early summer among conspecific populations with various female fre- and continue to flower for approximately 1 month. quencies are useful. Sun & Ganders (1986) estimated In Japan two Chionographis species occur; C. japon- selfing rates of hermaphrodites in eight gynodioecious ica (Willd.) Maxim. and C. koidzumiana (Ohwi Hara. populations of Hawaiian Bidens and found that the Chionographis japonica has five infraspecific taxa; var. selfing rate was highly correlated with the frequency of japonica, var. kurokamiana Hara, var. kurohimensis females. This result implies that the selfing rate of her- Ajima et Satomi, ssp. hisauchiana Okuyama, and ssp. maphrodites is an important factor which determines minoensis Hara (Hara, 1968, 1971; Ajima, 1976). the frequency of females. On the other hand, Beihassen These taxa are referred to without rank below. Japon- et a!. (1989) examined relationships between female ica, kurokamiana, and koidzumiana have a chromo- frequencies and population age in Thymus vulgaris and some number of 2n —24 and are considered to be showed that female frequency is the highest in the pop- diploid. Japonica and koidzumiana differ in their tepal ulation with an intermediate age. This finding suggests colour (white in japonica, greenish white or purplish in instability of female frequencies due to dynamic inter- koidzumiana) and the number of locules in anthers action between cytoplasmic male sterility and nuclear (two in japonica, one in koidzumiana). Kurokamiana restorer genes. is morphologically intermediate between these taxa. Chionographisjaponica (Wild.) Maxim., a liliaceous Hisauchiana and minoensis have a chromosome num- species native to Japan, is particularly suitable for the ber of 2n =42,and kurohimensis has 2n =44.The lat- study of gynodioecy because it includes both herma- ter three taxa are considered to be polyploids (Hara & phroditic and gynodioecious populations, and, more- Kurosawa, 1962; Tanaka & Tanaka, 1979). Cytolo- over, the frequency of females varies among gical observation (Tanaka & Tanaka, 1980) suggests gynodioecious populations. In addition, this species that these taxa are allotetraploids. These tetraploid has diploid and tetraploid varieties which include her- taxa and japonica differ in the thickness and shine of maphroditic and gynodioecious populations (Tanaka & leaves and/or the number and length of tepals. Tanaka, 1977, 1979; N. Y. Tanaka, 1985, personal Tanaka (1985) showed that gynodioecy is common communication). Another species, C. koidzumiana, is in the polyploid taxa, kurohimensis, minoensis, and diploid (Tanaka & Tanaka, 1979) and hermaphroditic hisauchiana, but rare in the diploid taxa. Among 15 (Kitamura et al., 1964; Hara, 1968). By utilizing intra- populations of kurohimensis examined, nine were specific and interspecific variation, conditions that gynodioecious; the two populations of minoensis and favour gynodioecy can be examined in depth. hisauchiana examined were also gynodioecious. In As a first step in a series of comparative studies on diploid japonica, however, only two of 12 populations Japanese Chionographis, this paper reports on esti- examined were gynodioecious. No other gynodioe- mates of Wright's fixation index and various genetic cious population has been found despite extensive field diversity measures for hermaphroditic and gynodioe- surveys. The other two diploid taxa, kurokamiana and cious populations. Wright's fixation index reflects the koidzumiana, are hermaphroditic (Hara, 1971; average level of inbreeding in a population. Genetic Kitamura eta!., 1964). diversity within a population can provide evidence of population bottlenecks, thus it can clarify an aspect of history of the populations. By comparing estimates of Electrophoresis Wright's fixation index and genetic diversity between Plantmaterials for electrophoresis were collected from hermaphroditic and gynodioecious populations, it will natural populations representing large areas of the be shown that selling appears to play a critical role in range of C. japonica (Fig. 1). These collections the evolution of gynodioecy in Japanese Chionographis. included all infraspecific taxa of C. japonica. One pop- FIXATION INDICES IN GYNODIOECIOUS POPULATIONS 331 ulation of C. koidzumiana was also examined. Popu- mutase (PGM), superoxide dismutase (SOD), 6-phos- lation codes, localities, mean sample size per locus, and phogluconate dehydrogenase (6PGDH), and triose- frequency of females within populations examined are phosphate isomerase (TPI). AAT was resolved using summarized in Table 1. The frequency of females is the disc PAGE system described by Shiraishi (1988). largely from Tanaka (1985) and partly supplemented The remaining enzyme species were resolved on a 12.8 by the observations of this research. per cent starch gel. System 8 of Soltis et a!. (1983) Mature leaves were collected from individuals, modified by Haufler (1985) was used to resolve LAP, placed on ice and transported to the laboratory of the PGI, and TPI. The buffer system II of Crawford et a!. Botanical Gardens, University of Tokyo, Nikko. These (1988) was used to resolve [NADP] G3PDH, MDH, samples were kept in a refrigerator until electrophor- 6PGDH and PGM. Staining procedures followed Soltis esis was carried out. et al. (1983), except for [NADP] G3PDH which was Plant materials were ground in cold extraction buf- stained following Rieseberg et a!. (1987) and SOD fer, as described by Odrzykoski & Gottlieb (1984). which was visualized on the gel stained for TPI. The enzyme species examined were asparate amino- transferase (AAT), [NADP} glyceraldehyde-3-phos- Statistical phate dehydrogenase ([NADP] G3PDH), leucine analysis aminopeptidase (LAP), malate dehydrogenase (MDH), Wright's(1951) fixation index (f)wascomputed for all phosphoglucose isomerase (PGI), phosphoglucose polymorphic loci in each population, as f=1—(H/ 2pq), where H is observed heterozygosity, and p and q are the frequencies of the alleles. For loci with more than two alleles, the frequencies of the less frequent alleles were combined into a single class. The signifi- cance of fwasexamined by computing Chi-square values for each locus in a population as Nf2, where N is the number of individuals per population (Workman & Niswander, 1970). The proportion of polymorphic loci (F), the number of alleles per locus (A ),andthe gene diversity within a population (H5), were calculated for each population and averaged for each of the following four categories: hermaphroditic diploid and tetraploid populations, and gynodioecious diploid and tetraploid populations.

Results Thenine enzyme systems used in this study were inter- preted as being encoded by 16 loci in diploid taxa: japonica, kurokamiana, and koizumiana. By contrast, 24 presumed loci encoded these nine enzyme systems in tetraploid varieties: kurohimensis, minoensis, and hisauchiana. Additional loci observed in tetraploid varieties are considered to originate by gene duplications in the following eight loci (Aat-1, Pgm, Mdh-1, NAN Mdh-2, Mdh-4, 6pg-1, Tpi-1 and Tpi-2). Fixed heter- / ozygosities were observed in these loci in tetraploid varieties and thus disomic inheritance was assumed. In YA some of these duplicated loci, one of a pair was poly- SIR morphic. Fixation indices were calculated using allele Fig.1 Distribution of the populations examined. The abbre- frequencies in these polymorphic loci. viations of population names and their summarized data are The banding pattern of AAT is shown in Fig. 2 as a shown in Table 1. (A) Chionographisjaponica ssp. japonica representative of these duplicated genes. In the tetra- var.japonica.(D) Cf.ssp.j.var. kurokamiana.(•) C.j. ssp. I. var. kurohimensis. (.)C.j. ssp. minoensis. ()C.j. ssp. ploid taxa, interlocus heteromers between Aat-1 and hisauchiana. (*) C. koidzumiana. Aat-2 were expressed, and Aat-1 was polymorphic. 332 M. MAKI

Table 1 Population codes, localities, mean sample size per locus and frequency of females for populations examined in this study

Population code Locality Sample size Frequency of females japonica SUM Sumato-kyo, Sizuoka Pref. 28.2 0.21 MIK Mikawa-makihara, AichiPref. 30 0 INU Inuyama, Gifu Pref. 30 0 ISE Mt Asama, Mie Pref. 28.4 0 KOM KomakiDam, ToyamaPref. 28.5 0 HIR Mt Hira, Kyoto Pref. 30 0 SAN Sandan-kyo, Hiroshima Pref. 30 0 OSA Osaka path, Kochi Pref. 30 0 NAN Fukue Isi., Nagasaki Pref. 30 0.18 ASO Mt Aso, Kumamoto Pref. 28.2 0 YAK Yakushima Is!., Kagoshima Pref. 30 0 kurokamiana YAM Mt Kurokami, Saga Pref. 14 0 koidzumiana SIR Yakushima Is!., Kagoshima Pref. 28.2 0 kurohimensis TOK Mt Toko, Akita Pref. 30 0 HON Mt Yakushi, Akita Pref. 28 0 HAM Hamashimbo, Niigata Pref. 30.4 0.22 TAN Tanne, Niigata Pref. 30 0.18 JOE Mt Yakushi, Niigata Pref. 30 0 ITO Mt Kurohime, Niigata Pref. 14 0.28 minoensis NEO Neo Valley, Gift' Pref. 30 0.16 1131 Ibigawa, Gifu Pref. 28.8 0.10 hisauchiana KAW Naguri, Saitma Pref. 30 0.29

Population SUM of japonica and populations JOE and ITO of kurohimensis were monomorphic for all loci examined. Aat-1 Tables 2 and 3 summarize estimates of the fixation [ index (f) for polymorphic loci in each population together with the mean value for each population. In diploid taxa, fixation indices were not significantly dif- Interlocus ferent from Hardy—Weinberg expectations except in heteromer the only gynodioecious population (NAN). On the other hand, the fixation indices were significant in all tetraploid populations examined, irrespective of the Aat-2 sexual system. The mean values of f estimates over populations belonging to the four different categories Aat-3 (2x vs. 4x; hermaphroditic vs. gynodioecious) are pre- sented in Table 4. In the diploid populations, the fixa- Genotype FF FS ss tion index is higher in gynodioecious than in of Aat-1 hermaphroditic populations but the reverse trend is Fig.2 The banding pattern of AAT in the teraploid taxa. found in tetraploid populations. Table 5 summarizes estimates of various measures japonica. Population genetic diversity is lower in gyno- of genetic diversity over populations belonging to the dioecious populations thanin hermaphroditic above four categories. In diploid taxa, gynodioecy has populations. On the other hand, the genetic diversity of been found only in two peripheral populations of hermaphroditic tetraploid populations is slightly lower FIXATION INDICES IN GYNODIOECIOUS POPULATIONS 333

* than that of tetraploid gynodioecious populations N * — although not statistically significant; and genetic diver- c 'r C C\O Ir sity of tetraploid populations is lower than that of C C diploid populations, regardless of the sexual system. Hi I I I I I I

Discussion E C N C C NrC Thereappeared to be a tendency in Japanese Chiono- C CCC I graphis for the diploid populations to be hermaphro- I II I ditic, outcrossing, and have high gene diversity, N C r NN Ntr N — whereas the tetraploid populations are frequently C C CC C gynodioecious, highly inbreeding and have low gene III diversity. This result suggests that either selling or 00 trC000 N polyploidy or both seem to play an important role in — — C—— r r the evolution of gynodioecy. C d— C 00 z Although rare, gynodioecy has evolved in the diploid. Of two known gynodioecious populations, * * * * * ** * * SUM was monomorphic for 16 loci examined and z N — — N S0'- . 00 NAN showed large f values (0.659). The data suggest z IdIlIdICIC I C that these populations are highly selfing. On the other 00 hand, 11 hermaphroditic populations of the diploid — C 0 C — .-iCC were all polymorphic, had high gene diversity and cC C 00 dC C dCC I showed f values not significantly different from zero C II I I I I (Hardy—Weinberg expectation). These populations are NN 00 N ——NV NC-.r--00 N S considered to be highly outcrossing. This finding coin- CC CC C cides with an empirical fact that the diploid plant of I II I I I Chionographis is usually self-incompatible (N. Y. V 00 C CS r Tanaka, personal communication). Thus, gynodioecy qC QC C evolved only in association with selfing in diploid Chi- C CC CC 11111111 onographis. I'll A contrasting case was reported for Hawaiian * tfl 0 C Bidens. Although high selfing rates of hermaphrodites N tr N in gynodioecious populations were also reported in C C some populations of Hawaiian Bidens, low levels of 11111111111II selling were found in other populations (Sun & * * Ganders, 1986, 1988). This suggests that gynodioecy CNIrCrNC\C can be maintained not only in selling populations but NCC-.- C also in outcrossing populations in Hawaiian Bidens. 00 In the tetraploid populations, gynodioecy is a more N C N 'I common sexual system than in the diploid. Of nine '- N NC—I populations examined, six were gynodioecious and the C C other three were hermaphroditic. All nine populations I I I I I were monomorphic or had large f values and are sug- N C.-l — CC— gested to be selfing. This result is consistent with the d factthat hermaphrodites of tetraploid Chionographis 11111II are self-compatible as far as examined to date (N. Y. Tanaka, personal communication; M. Maki, unpublished data). Two explanations can be given for the evolution of gynodioecy in tetraploid Chionographis. Firstly, E gynodioecy may have evolved in association with selfing as in diploid Chionographis. Alternatively, poly- N ploidy may have played a more important role than selling in the evolution of gynodioecy. The latter possi- 2 bility is discussed in more detail below. 334 M. MAKI

Table 3 Fixation indices for individual loci and all polymorphic loci for populations of the tetraploid taxa, i.e. kurohimensis, minoensis and hisauchiana

Population

kurohimensis minoensis hisauchiana

TOK HON HAM TAN JOE ITO NEO IBI KAW

Sexual system H H G G H G G G G Locus 6pg-1 — — — 0.533*** — — 0.659* Pgi-2 — — — — — O.6O0 . Tpi-3 1.000' 1.000 — 0.627*** Pgm-2 — — — 0.773*** Mdh-1 — — 0.345 — — — — — — Mdh-3 — — — — — — — 0.509 Mdh-6 — — 0.515*** O.509** Aat-3 1.000 — 0.466** 0.506** All 1.000*** 1.000° 0.458" 0.576*** 0.600 0.659* 0.509** * <0.5;**<0.01; *** <0.005.

Table 4 Mean values of fixation index for the populations patibility system. Once lost, the self-incompatibility belonging to the different categories system might have been difficult to recover in these populations because accumulations of many mutations Ploidy level are needed for such a system to work. Thus ploidy 2X 4X level plays a more important role than selfing. Theoretical studies have shown that high inbreeding Sexual system H G H G depression is required for the evolution of gynodioecy f 0.031 0.659 1.000 0.560 in addition to a high selfing rate in hermaphrodites Number of populations 11 1 2 5 (Lloyd, 1975; Charlesworth & Charlesworth, 1978). If examined inbreeding depression is mainly caused by multiplica- tive effects of deleterious mutations, the amount of inbreeding depression is expected to decrease with an Genetic diversities are very low in the gynodioecious increased level of selling (Lande & Schemske, 1985). populations of Chionographis, regardless of ploidy However, recent theoretical studies on a mildly dele- level. The data suggest that these populations experi- terious mutation load predict that high selfing popula- enced bottlenecks in the past and that these historical tions still have considerable levels of inbreeding events triggered the evolution of gynodioecy. Self- depression (Charlesworth & Charlesworth, 1987; incompatibility is inherent in most diploid populations, Charlesworth et a!., 1990). Recent experimental stu- and this could prevent selfing of individuals and inva- dies on highly self ing or genetically monomorphic sion of male sterile mutants. Under an ordinary level of populations support this prediction (Holtsford & gene flow in an outcrossing population, the breakdown Ellstrand, 1990; Yahara, 1991; Weller, personal of self-incompatibility through population bottlenecks communication). An estimation of inbreeding depres- might be transient because S-alleles could be supplied sion in gynodioecious populations of Chionographis is from self-incompatible populations. In the diploid taxa, needed and studies are in progress. only peripheral populations are mainly selling, and this In addition to the outcrossing advantage, increased may be explained by restricted gene flow into these fertility due to the compensation effect can give populations. On the other hand, self-incompatibility is females an advantage. An increased fertility of females unknown in the tetrapoloid taxa and thus it may have can reduce the critical level of inbreeding depression been broken down through tetraploidization. As the required for the evolution of gynodioecy and relaxes tetraploid taxa should have originated from a small theconditionsneeded tomaintain females number of individuals, it is plausible that they lost the (Charlesworth & Charlesworth, 1978). In a gynodi- S-allele polymorphism responsible for the self-incom- oecious population of kurohimensis, seed set is actually FIXATION INDICES IN GYNODICECIOUS POPULATIONS 335

Table 5 Proportion of polymorphic loci (P), mean number of alleles per locus (A and average gene diversity within a population (Its) for the populations belonging to different categories. Values in parentheses indicate the range for each value

Ploidy level

2x 4x

Sexual system H G H G P 0.364 0.125 0.056 0.076 (0.188—0.500)(0.000—0.250)(0.000—0.083)(0.000—0.208) A 1.426 1.188 1.042 1.077 (1.125—1.875)(1.000—1.375)(1.000—1.083)(1.000—1.208) H 0.123 0.066 0.017 0.029 (0.071—0.211)(0.000—0.133)(0.000—0.042)(0.000—0.090) Numberof 11 2 3 5 populations examined higher in females than in hermaphrodites (M. Maki, in tam. Quantitative studies on these factors are needed to preparation). understand fully the conditions under which females The above, traditional, view assumes that equilib- are maintained. rium states are under nuclear or nuclear-cytoplasmic control of gynodioecy; however, some recent studies have emphasized the importance of the dynamic inter- Acknowlegements action between cytoplasmic male sterility genes and Iwish to express my cordial thanks to Dr N. Y. Tanaka nuclear restorer genes for the evolution of gynodioecy for his kind suggestions and helpful advice. I would (Couvett eta!., 1986; Belhassen eta!., 1989). Unfortu- also like to thank Drs S. G. Weller and D. J. Crawford nately, genetic control of male sterility in Japanese for fruitful comments on a draft of the manuscript; to Drs Chionographis is uncertain. In non-equilibrium pro- S. Ishizawa, A. Takahashi and T. Morita for providing cesses, male sterile mutants can be found in both self- information on localities of the study plants, and to Dr ing and outcrossing populations. In Chionographis,, T. Yahara for his encouragement and valuable advice however, gynodioecy is found only in selfing or mono- during the course of this study. Staff of the Botanical morphic (putatively selfing) populations. It is unlikely, Gardens of the University of Tokyo, Nikko, were kind therefore, that gynodioecy in Japanese Chionographis enough to cultivate experimental plants. Miss M. is a contingent state under non-equilibrium processes. Masuda provided technical assistance and I am There is a trend in tetraploid varieties of Chionogra- indebted to Professor K. Iwatsuki, Drs M. Kato, phis hermaphroditic populations to be more highly J. Murata, N. Murakami and my colleagues in the selfing and have lower genetic variability than gynodi- Botanical Gardens at the University of Tokyo for their oecious populations. The latter suggests that these valuable comments on this study. populations experience a strong genetic drift. If the male sterility gene is lost due to strong genetic drift References from the populations, the population remains hermaphroditic until the male sterility gene is supplied by AJIMA,T. 1976. A new variety of Chionographis japonica mutation or migrations, even if the condition for the Maxim. J. Geobot., 24, 45—48. evolution of gynodioecy is satisfied. Gene flow between BELHASSEN, E., TRABAND, L., COUVETF D. AND GOUYON, P. H. 1989. populations is known to be restricted in selfing popula- An example of nonequilibrium processes: Gynodioecy of tions, (Hamrick & Godt, 1990), this is also the case for Thymus vulgaris L. in burned habitats. Evolution, 43, 662—667. the tetraploid populations of Chionographis (M. Maki, CHARLESWORTH, D. 1981. A further study of the problem of the unpublished data). maintenance of female in gynodioecious species. Heredity, The results of this study suggest that selling is a criti- 46, 27-39. cal factor for the evolution of gynodioecy in Chiono- CHARLESWORTH, B. AND CHARLESWORTH, D. 1978. A model for graphis, but the relative roles of other factors such as the evolution of dioecy and gynodioecy. Am. Nat., 112, polyploidy, genetic control of male sterility, inbreeding 975—997. depression and reproductive compensation, are uncer- CHARLnSWORTH, D. AND CHARLESWORTH, B. 1987. Inbreeding 336 M. MAKI

depression and its evolutionary consequences. Ann. Rev. in dioecious plant populations: A case study of Chamaelir- Ecol. Syst., 18,237—268. ium luteum. In: Dingle, H. and Hegmann, J. (eds) Evolu- CHARLESWORTH, D. AND GANDERS, F. R. 1979. The population (ion and Genetics of Life Histories. Springer-Verlag. New genetics of gynodioecy with cytoplasmic-genic male steril- York, pp. 139—154. ity. Heredity, 43, 213—218. ODRZYKOSKY, I. J. AND GOTFLIEB, L. D. 1984. Duplications of CHARLESWORTH, D., MORGAN, M. T. AND cFIARLESw0RTH, B. 1990. genes coding 6-phosphogluconate dehydrogenase in Clar- Inbreeding depression, genetic load, and the evolution of kia (Onagraceae) and their phylogenetic implications. Syst. outcrossing rates in a multilocus system with no linkage. Bot., 9,479—486. Evolution, 44, 1469—1489. RIESEBERG, L. H., PETERSON, P.. M., SOLTIS D. E. AND ANNABLE, CR. COUVETL D., BONNEMAISON, F. AND GOUYON, t H. 1986. The 1987. Genetic divergence and isozyme number variation maintenance of females among hermaphrodites: The among four varieties of A ilium douglasii (Liliaceae). Am. importance of nuclear—cytoplasmic interactions. Heredity, J. Bot., 74, 1614—1624. 57, 325—330. SAKAL, A. K., KAROLY, K. AND WELLER, S. G. 1989. Inbreeding CRAWFORD, D. J., POST, B. J. AND WITKUS, R. 1988. Allozyme varia- depression in Schiedea globosa and S. salicaria (Caryo- tion within and between populations of Coreopsis latifolia phylaceae), subdioecious and gynodioecious Hawaiian (Asteraceae). P1. Spec. Biol., 3, 1—5. species. Am. J. Bot., 76,437—444. DELANDY, X., GOUYON. P. H. AND VALDEYLON, G. 1981. Mathema- SI-IIRAI5HI, s. 1988. Inheritance of isozyme variations in tical study of the evolution of gynodioecy with cytoplasmic Japanese Black Pine, Pinus thunbergii Parl. silvae Gene- inheritance under the effect of a nuclear restorer gene. tica, 37, 93—100. Genetics, 99, 169—18 1. SI-IYKOFF, J. A. 1988. Maintenance of gynodioecy in Silene FRANK, S. A. 1989. The evolutionary dynamics of cytoplasmic acaulis (Caryophylaceae): stage-specific fecundity and via- male sterility. Am. Nat., 133,345—376. bility selection. Am. J. Bot., 75, 844—850. !-IAMRICK, J. L. AND GODT, M. .w.1990. Allozyme diversity in SOLTIS, D. E., HAUFLER, C. FL, DARROW, D. C. AND GASTONY, G. J. plant species. In: Brown, A. H. D., Clegg, M. T., Kahier, A. 1983. Starch gel electrophoresis of ferns: a compilation of L. and Weir, B. S. (eds) Plant Population Genetics, Breed- grinding buffers, gel and electrode buffers, and staining ing, and Genetic Resourses. Sinauer Associates Inc., Sun- schedules. Am. Fern. J., 73, 9—27. derland, pp. 43—63. SUN, M. AND GANDER5, F. R. 1986. Female frequencies in gynodi- HARA, H. 1968. A revision of the genus Chionographis (Lilia- oecious populations correlated with selfing rates in her- ceae). J. Jap. Bot., 43, 257—267. maphrodites.Am. .1. Bot., 73,1645—1648. HARA, H. 1971. Chionographis japonica var. kurokamiana SUN, M. AND GANDERS, F. R. 1988. Mixed mating systems in Hara. J. Jap. Bot., 46,282. Hawaiian Bidens (Asteraceae). Evolution, 42,516—527. HARA, H. AND KUROSAWA, s. 1962. Morphological and chromo- TANAKA, N. .1985.Variation of sexual system in Chionogra- somal variations in Chionographis japonica Maxim. Acta phi,s species. Shuseibutugaku-kenkyu, 8, 11—19 (in Phytotax. Geobot., 20, 34—3 8 (in Japanese with English Japanese). summary). TANAKA, N. Y. AND TANAKA, N. 1977. Chromosome studies in I-IAUFLER, C. H. 1985. Enzyme variability and modes of evolu- Chionographis (Liiaceae). I. On the holokinetic nature of tion in the fern genus Bommeria. Syst. Bot., 10, 92—104. chromosomes in Chionographis japonica Maxim. Cytolo- I-1OLTSFORD, T. P. AND ELLSTRAND, N. C. 1990. Inbreeding effects gia, 42, 75 3—763. in Clarkia tembroriensis (Onagraceae) populations with TANAKA, N. Y. AND TANAKA, N. 1979. Chromosome studies in different natural outcrossing rates. Evolution 44, Chionographis (Liliaceae). II. Morphological charac- 203 1—2046. teristics of the somatic chromosomes of four Japanese JOLLS, C. L. AND CHENIER, T. c. 1989. Gynodioecy in Silene vu!- members. Cytologia, 44, 935—949. garis (Caryophylaceae): progeny success, experimental TANAKA, N. Y. AND TANAKA, N. 1980. Chromosome studies in design, and maternal effects. Am. J. Bot., 76, 1360—1367. Chionographis (Liliaceae). III. The mode of meiosis. Cyto- KITAMURA, S., MURATA, G. AND KOYAMA, T. 1964. Coloured Illu- logia, 45, 809—817. strations of Herbaceous Plants of Japan (Monocotyledo- VAN DAMME, J. M. M. AND VAN DELDEN, w. 1984. Gynodioecy in neae). Hoikusha, Osaka. Plantago lanceolata L. IV. Fitness components of sex KOHN, i. R. 1988. Why be female? Nature, 335, 431—433. types in different life cycle stages. Evolution, 38, LANDE, R. AND SCHEMSKE, D. w. 1985. The evolution of self-fer- 1326—1336. tilization and inbreeding depression in plants. I. Genetic WOLFF, K., FRISO, B, AND VAN DAMME, J. M. M. 1988. Outcrossing models. Evolution, 39,24-40. rates and male sterility in natural populations of Plantago LEwis, D. G. 1941. Male sterility in natural populations of her- coronopus. Theor. Appl. Genet., 76, 190-196. maphroditic plants. New Phytol. ,40,56—63. WORKMAN, P. L AND NISWANDER, J. D. 1970. Population studies LLOYD, D. a 1975. The maintenance of gynodioecy and on southwestern Indian tribes. II. Local genetic differen- androdioecy in angiosperms. Genetica, 45, 325—339. tiation in the Papago. Am. J. Hum. Genet., 22, 24-49. MEAGHER, T. R. 1984. Sexual dimorphism and ecological dif- wRIGH'r, s. 1951. The gentical structure of populations. Ann. ferentiation of male and female plants. Ann. Mo. Bot. Eugen., 15, 323—354. Gard, 71, 254—264. YAHARA, T. 1991. Graphical analysis of evolutionarily stable MEAGHER, T. R. AND ANTONOvIC5, j.1982.Life history variation mating systems in plants. Evolution 1991 (in press).