American Journal of Botany 86(1): 124±130. 1999.

GENETIC CONSEQUENCES OF A SINGLE-FOUNDER POPULATION BOTTLENECK IN TRIFOLIUM AMOENUM ()1

ERIC E. KNAPP2,4 AND PETER G. CONNORS3

2Department of Agronomy and Range Science and Center for Population Biology, University of Davis, One Shields Avenue, Davis, California 95616-8515; and 3University of California Bodega Marine Laboratory, Bodega Bay, California 94923

We investigated the genetic consequences of a single-founder bottleneck in a population of showy Indian (Trifolium amoenum), a species presumed to be extinct until rediscovered near Occidental, California, in 1993. Electrophoretic variation was evaluated in the bottlenecked population and in a larger population (Dillon Beach) discovered during the course of this study, as well as in populations of two closely related species, T. albopurpureum var. dichotomum and T. macraei.We found a surprisingly high amount of polymorphism in the single-founder T. amoenum population from Occidental (15% of loci polymorphic; an average of 1.1 alleles per locus). However, this represents a 53% reduction in number of polymorphic loci and a 20% reduction in average number of alleles per locus compared to three Trifolium populations with putatively similar mating systems (the Dillon Beach T. amoenum population and both populations of T. albopurpureum var. dichoto- mum). Expanding the genetic base of the Occidental T. amoenum population is a priority due to concerns about loss of evolutionary potential and the possibility of deleterious effects associated with inbreeding. However, using seed from the Dillon Beach T. amoenum population may not be bene®cial due to distinct, presumably adaptive differences between from the two populations and concerns about outbreeding depression.

Key words: endangered species; Fabaceae; genetic variation; population bottleneck; Trifolium albopurpureum; Trifo- lium amoenum; Trifolium macraei.

Trifolium amoenum E. Greene (Fabaceae) is a robust amoenum is capable of self-pollination (Connors, unpub- and showy native annual clover that at one time could be lished data), and we conclude that the single wild found in grasslands in several counties north of San Fran- discovered in 1993 must have produced seed through cisco Bay, California. By the middle of this century it self-pollination. Seeds collected from plants in subse- had become rare, due to a combination of factors, in- quent generations of seed multiplication may be the out- cluding habitat loss, competition from introduced species, come of either self- or cross-pollination, as they were and the pressures of livestock grazing (Connors, 1994). produced in groups of several plants that were visited by Botanists failed to locate any plants during the 1970s and bumble bees. The resultant collection of seeds is excep- 1980s, and in 1984 T. amoenum was listed as ``presumed tional in one genetic respect, however: the entire popu- extinct'' by the California Native Plant Society (Smith lation has gone through a single-founder bottleneck. This and York, 1984). The species reappeared in 1993 with raises the concern that genetic variation might be severely the discovery of a single individual at the edge of a dirt reduced compared with former levels, a condition that road near Occidental in Sonoma County, California (Con- could adversely affect future attempts to reestablish wild nors, 1994). Despite extensive searches, no additional populations. plants were located at the site. Because the location was To assess genetic consequences of the bottleneck, we threatened by development, most of the seeds produced evaluated allozyme variation within the Occidental T. by this plant were collected. We germinated 18 of the 92 amoenum population, intending initially to compare it collected seeds in the greenhouse and placed the remain- with allozyme variation in natural populations of two ing seeds in storage. Ten of the seedlings, when trans- closely related native annual species, T. macraei and T. planted to outside gardens, produced several thousand seeds in 1994. A second generation of seed multiplication albopurpureum var. dichotomum. Trifolium macraei is a has since increased the number of available seeds to primarily coastal species of northern and central Califor- Ͼ50 000. nia, but also grows in South America. Trifolium albo- Flower bagging experiments have demonstrated that T. purpureum var. dichotomum occupies coastal to interior grasslands from central California north to Washington. In the process of collecting material of these two species, 1 Manuscript received 29 January 1998; revision accepted 16 June 1998. we discovered a previously unknown natural T. amoenum The authors thank the Genetic Resources Conservation Program at population of ϳ225 plants near Dillon Beach in Marin the University of California Davis and the Center for Plant Conservation County, California, 16 km distant from the single indi- for funding support, Kevin J. Rice for assistance with sampling plant vidual found in 1993. This presented the unexpected op- material in the ®eld and for helpful comments on an earlier version of portunity to extend our planned among-species compar- the manuscript, Kara O'Keefe for assistance with electrophoresis, and Tess Lispi for the Fig. 1 illustration. isons to include a within-species comparison of the sin- 4 Author for correspondence (Tel.: 530-752-1701; e-mail: gle-founder population and the larger Dillon Beach pop- [email protected]). ulation. 124 January 1999] KNAPP AND CONNORSÐPOPULATION BOTTLENECK IN TRIFOLIUM AMOENUM 125

TABLE 1. Buffer system and number of loci scored for all stains used in the study. Buffer systems correspond to numbers 1, 3, and 5 of Wendel and Weeden (1989).

Buffer No. loci Stain (enzyme commission number) system scored AconitaseÐACO (E.C. 4.2.1.3) 3 2 AmylaseÐAMY (E.C. 3.2.1.1) 5 1 Aspartate aminotransferaseÐAAT (E.C. 2.6.1.1) 5 2 Isocitrate dehydrogenaseÐIDH (E.C. 1.1.1.41) 3 3 Leucine aminopeptidaseÐLAP (E.C. 3.4.11.1) 5 2 Malate dehydrogenaseÐMDH (E.C. 1.1.1.37) 1 5 PhosphoglucoisomeraseÐPGI (E.C. 5.3.1.9) 5 2 PhosphoglucomutaseÐPGM (E.C. 5.4.2.2) 3 1 Phosphogluconate dehydrogenaseÐPGD (E.C. 1.1.1.44) 3 2

in a refrigerator at 4ЊC for 3 d. We inferred the genetic basis of banding patterns from known enzyme subunit structure and intercellular com- partmentalization for these enzymes (Kephart, 1990). Loci for each stain and alleles at each locus were numbered sequentially, from fastest to slowest migrating. Using BIOSYS-1 (Swofford and Selander, 1989), we calculated genetic variability statistics and performed a cluster analysis on the matrix of genetic distances (Nei, 1978) between populations with the unweighted pair-group method using arithmetic averages (UPGMA). We considered a locus polymorphic if the frequency of the most com- mon allele did not exceed 0.99.

RESULTS

Fig. 1. Map of the Paci®c coast near Bodega Bay, California, show- Allozyme variation within populationsÐDespite the ing locations of sampled Trifolium amoenum, T. albopurpureum var. single-founder bottleneck in the Occidental T. amoenum dichotomum, and T. macraei populations. population A, we found that three of the 20 loci evaluated were polymorphic (Table 2). Thus, the original plant was heterozygous at these three loci. Each polymorphic locus MATERIALS AND METHODS harbored two alleles with intermediate allele frequencies, We collected fresh leaf material from meristematic tips of 30 green- as expected in a population with only one founding dip- house-grown T. amoenum plants in May 1995. These 30 plants con- loid individual. Because we sampled 30 plants in this sisted of three open-pollinated offspring of each of the ten reproductive population, and the expected allele frequency for variant plants grown from seeds collected from the original plant found near alleles is 0.5, it is highly unlikely (P K 0.001) that any Occidental. On the same day we collected leaf material from 13 to 16 additional scorable variation at these loci went undetect- plants in each of T. macraei populations E, F, and G, growing along ed. the California coast north of San Francisco Bay (Fig. 1). We sampled The T. amoenum population B from Dillon Beach con- 20 plants in each of two inland T. albopurpureum var. dichotomum tained more polymorphic loci and a higher average num- populations (C and D) and from 18 to 20 plants from three additional ber of alleles per locus compared to the Occidental pop- T. macraei populations (H, I, and J) in May 1996 (Fig. 1). At site J, ulation A (30 vs. 15% and 1.4 vs. 1.1, respectively) (Ta- we discovered the Dillon Beach population of T. amoenum, and col- ble 3). Levels of allozyme variation within T. albopur- lected one lea¯et from each of 20 plants in this population. At least 5 pureum var. dichotomum populations C and D (30 and m separated all sampled plants, thereby reducing the chance of collect- 35% of loci polymorphic and 1.3 and 1.4 alleles per lo- ing material from closely related individuals. cus, respectively) were similar to that found in the Dillon Immediately after collection, we placed plant material on ice for Beach T. amoenum population. Populations of T. macraei transport to Davis, where enzymes were extracted from the fresh, un- possessed on average less allozyme variation than wild frozen tissue. Leaves were ground with a ¯at-bottomed Plexiglas rod, populations of the other two species. Percentage of poly- together with ®ve drops of a chilled extraction buffer (Gottlieb, 1981a). morphic loci in T. macraei populations ranged from 10 The extract was absorbed onto Whatman number 3 ®lter paper wicks, to 25%, while the average number of alleles per locus which were then placed in 0.5-mL microfuge tubes and stored at Ϫ80ЊC until used in electrophoresis. ranged from 1.1 to 1.4 (Table 3). We resolved loci with standard starch gel electrophoresis techniques, Observed heterozygosity was less than expected (as- using histidine-citrate, tris-citrate, or sodium-borate buffer systems (Ta- suming Hardy-Weinberg equilibrium, i.e., completely ble 1). Gels were made with 12% w/v Connaught brand starch. Staining random mating, and no migration, selection, or mutation) followed recipes from Wendel and Weeden (1989), with the exception in all populations, but varied dramatically among species of ACO, which was stained according to Morden, Doebley, and Schertz (Table 3). Individual plants of T. amoenum and T. albo- (1987), and IDH, which was stained according to Soltis et al. (1983). purpureum var. dichotomum showed considerable hetero- Amylase (AMY) appeared as translucent bands on the gel stained for zygosity at polymorphic loci (mean observed/expected ϭ LAP, and genotypes at this locus were scored after storing the gel slice 0.416 for T. amoenum; mean observed/expected ϭ 0.542 126 AMERICAN JOURNAL OF BOTANY [Vol. 86

TABLE 2. Allele frequencies at polymorphic loci for all Trifolium populations in the study.

Population T. amoenum T. albopurpureum T. macraei Locus Allele AB CD EFGHI J Aat-1 1 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.650 1.000 2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.350 0.000 Aco-1 1 1.000 1.000 0.975 0.850 1.000 1.000 1.000 1.000 1.000 1.000 2 0.000 0.000 0.025 0.150 0.000 0.000 0.000 0.000 0.000 0.000 Aco-2 1 1.000 0.969 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 2 0.000 0.031 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Amy-1 1 0.448 0.350 0.000 0.000 0.000 0.000 0.000 0.000 0.368 0.000 2 0.000 0.025 0.325 0.079 1.000 1.000 0.188 1.000 0.526 0.050 3 0.552 0.625 0.675 0.921 0.000 0.000 0.688 0.000 0.105 0.950 4 0.000 0.000 0.000 0.000 0.000 0.000 0.125 0.000 0.000 0.000 Idh-1 1 1.000 1.000 0.925 0.950 1.000 1.000 1.000 1.000 1.000 1.000 2 0.000 0.000 0.075 0.050 0.000 0.000 0.000 0.000 0.000 0.000 Idh-2 1 0.617 0.500 0.447 0.417 0.071 0.308 0.438 0.111 0.050 0.000 2 0.000 0.000 0.000 0.000 0.000 0.000 0.250 0.000 0.050 0.650 3 0.383 0.500 0.553 0.583 0.929 0.692 0.313 0.889 0.900 0.350 Idh-3 1 0.000 0.000 0.000 0.000 0.000 0.000 1.000 0.000 0.650 0.650 2 1.000 1.000 1.000 1.000 1.000 1.000 0.000 1.000 0.350 0.350 Lap-1 1 0.667 0.600 0.853 0.895 0.692 0.923 0.375 1.000 0.000 0.333 2 0.333 0.400 0.147 0.105 0.308 0.077 0.625 0.000 1.000 0.667 Lap-2 1 1.000 1.000 1.000 0.975 1.000 1.000 1.000 1.000 1.000 1.000 2 0.000 0.000 0.000 0.025 0.000 0.000 0.000 0.000 0.000 0.000 Pgi-2 1 0.000 0.950 0.643 0.538 0.281 0.900 0.900 0.000 0.342 0.056 2 1.000 0.050 0.357 0.462 0.469 0.100 0.100 0.842 0.658 0.944 3 0.000 0.000 0.000 0.000 0.250 0.000 0.000 0.000 0.000 0.000 4 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.132 0.000 0.000 5 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.026 0.000 0.000 Pgm-1 1 1.000 0.775 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 2 0.000 0.225 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 for T. albopurpureum var. dichotomum ), while individ- leles found in the Occidental T. amoenum population uals of T. macraei were almost entirely homozygous de- were a subset of alleles present in the Dillon Beach T. spite abundant polymorphism, with only one heterozy- amoenum population. The considerable allozyme differ- gote found (mean observed/expected ϭ 0.004). Genotype entiation between these two populations (FST ϭ 0.225) frequencies deviated signi®cantly from Hardy-Weinberg was due mainly to ®xation in the Occidental T. amoenum expectations for all polymorphic loci in all T. macraei population of a PGI allele which was rare in the Dillon populations, and for two of three polymorphic loci in the Beach T. amoenum population. (For comparison, the FST Occidental T. amoenum population A, four of six poly- between T. albopurpureum var. dichotomum populations morphic loci in the Dillon Beach T. amoenum population was 0.047, and the FST among T. macraei populations B, three of six polymorphic loci in the Cazadero T. al- was 0.483.) bopurpureum var. dichotomum population C, and two of seven polymorphic loci in the Ida Clayton Road T. al- Allozyme variation among speciesÐCluster analysis bopurpureum var. dichotomum population D. by the UPGMA method demonstrated that all three spe- cies possessed similar allozymes (Fig. 2), suggesting that Allozyme variation among populationsÐAllozyme al- the species are closely related to each other. The average

TABLE 3. Summary of genetic variation at 20 allozyme loci for all populations. Standard errors, where applicable, are in parentheses.

Mean heterozygosity Mean sample Mean no. % polymorphic Species Population size/locus alleles/locus loci Observed Expected T. amoenum A 29.3 1.1 (0.1) 15.0 0.042 (0.023) 0.073 (0.039) B 19.5 1.4 (0.1) 30.0 0.026 (0.016) 0.101 (0.043) T. albopurpureum C 19.8 1.3 (0.1) 30.0 0.047 (0.021) 0.094 (0.039) var. dichotomum D 19.5 1.4 (0.1) 35.0 0.045 (0.017) 0.077 (0.030) T. macraei E 13.9 1.1 (0.1) 15.0 0.000 (0.000) 0.053 (0.032) F 13.0 1.1 (0.1) 15.0 0.000 (0.000) 0.055 (0.033) G 16.0 1.4 (0.2) 20.0 0.003 (0.003) 0.115 (0.054) H 18.0 1.1 (0.1) 10.0 0.000 (0.000) 0.016 (0.011) I 19.7 1.4 (0.2) 25.0 0.000 (0.000) 0.095 (0.042) I 19.6 1.3 (0.1) 25.0 0.000 (0.000) 0.084 (0.038) January 1999] KNAPP AND CONNORSÐPOPULATION BOTTLENECK IN TRIFOLIUM AMOENUM 127

and therefore contribute comparatively little to measures of population variance (Frankel and SouleÂ, 1981). How- ever, even alleles that are currently rare in a population may be of considerable evolutionary importance in the long term, particularly in a variable and changing selec- tive landscape (Millar and Libby, 1991; Lesica and Al- lendorf, 1995). The amount of genetic variation remaining following a bottleneck is a function of not just the severity of the bottleneck, but how rapidly the population size rebounds (Nei, Maruyama, and Chakraborty, 1975). However, pop- ulation growth will do little for restoring alleles lost dur- Fig. 2. Cluster analysis of populations based on Nei's (1978) ge- ing a bottleneck (Nei, Maruyama, and Chakraborty, 1975; netic distance. (amo ϭ T. amoenum, alb ϭ T. albopurpureum var. di- chotomum, and mac ϭ T. macraei) Barrett and Kohn, 1991). Over the longer term, rates of recovery of quantitative variation following a bottleneck may be much faster than recovery of allozyme variation, Nei's genetic distance between T. amoenum and T. ma- due to higher rates of spontaneous mutation for quanti- craei populations was 0.080, while an average Nei's ge- tative traits (Lande, 1980; Barrett and Kohn, 1991). How- netic distance of only 0.030 separated populations of T. ever, in this extreme case of a single-founder population, amoenum from populations of T. albopurpureum var. di- loss of even the most common allozyme and quantitative chotomum. The mating system similarities indicated by variation is a concern. the observed:expected heterozygosity ratios also suggest that these latter two species are the most closely related Genetic variation within the Occidental T. amoenum of the three. populationÐConsidering the severity of the bottleneck The cluster analysis of Nei's genetic distance values in the Occidental T. amoenum population, it is surprising (Fig. 2) presents an interesting geographical relationship that the amount of allozyme variation remains as high as among populations of T. macraei. Trifolium macraei pop- shown in this survey. We can infer that the single plant ulations E, F, and H cluster more closely with T. amoen- from which this population originated was heterozygous um and T. albopurpureum var. dichotomum than with T. at three allozyme loci. (By comparison, only one of the macraei populations G, J, and I. They also share a geo- 20 plants sampled from the Dillon beach T. amoenum graphic home with T. amoenum and T. albopurpureum population B was heterozygous at as many as three loci.) var. dichotomum on the American (Continental) land Genetic variability for morphological traits has apparent- plate, separated from the Paci®c land plate (T. macraei ly persisted in the Occidental T. amoenum population as populations G, J, and I) by the San Andreas Fault Zone. well. For example, ®rst-generation offspring grown to- In the area of these populations, the fault zone is delin- gether in a common garden environment had seed color eated over most of its length by bodies of water that and seed markings that varied among individuals but create a potential physical barrier to gene ¯ow. The clus- were consistent within individuals (Connors, unpublished ter analysis suggests that gene ¯ow may indeed be lim- observations). ited across the fault zone, a pattern that may hold clues Moderate levels of allozyme variation remaining in the to differentiation in this species complex. Occidental T. amoenum population provide empirical ev- idence that rapid population growth following a bottle- DISCUSSION neck may allow a considerable portion of the genetic var- iation to be maintained. It is also likely that the bottleneck Loci possessing alternate alleles that are in low fre- in this case was not long in duration. Due to the cumu- quency are the least likely to remain polymorphic through lative nature of the effects of genetic drift in small pop- an extreme bottleneck. In the current case, the only loci ulations, bottlenecks of this severity, if maintained for in the Occidental T. amoenum population expected to several generations, would likely result in complete or show polymorphism are those that were heterozygous in nearly complete loss of genetic variation. Our results are the single founding plant. Similarly, if the population consistent with a suggestion by Connors (1994) that the contained more than two alleles at any locus prior to the seed from which the 1993 wild plant grew may have been bottleneck, these additional alleles would have been lost. produced many years earlier when the population size Expected heterozygosity is less sensitive to population was larger. produce hard-coated seeds capable of bottlenecks than percentage of polymorphic loci or num- remaining viable for decades (Hull, 1973). Thus the 1993 ber of alleles per locus (Barrett and Kohn, 1991; Leberg, seed may have germinated from a long-dormant seed 1992). Bottlenecks might even increase levels of expect- bank when disturbed by road construction during the pre- ed heterozygosity in subsequent generations. For exam- vious year. ple, a population reduced to one individual will cause all variant alleles to be of intermediate frequency, at which Comparisons with other populationsÐA single indi- point expected heterozygosity for these loci is maxi- vidual is unlikely to harbor all of the genetic variation mized. that existed in the original population prior to the bottle- Measures of overall genetic variation may not change neck. Questions remain about how much variation was appreciably unless a bottleneck is severe because the al- lost, and how this loss of variation will impact strategies leles most likely to be lost are generally in low frequency for reintroduction. The best estimates of natural levels of 128 AMERICAN JOURNAL OF BOTANY [Vol. 86 background genetic variation come from comparisons An examination of leaf markings in the two T. amoen- with the Dillon Beach T. amoenum population, and from um populations suggests that genetic variation for traits comparisons with populations of closely related Trifolium other than allozymes may have been affected by the pop- species. We chose populations of T. albopurpureum var. ulation bottleneck as well. Different leaf markings (light dichotomum and T. macraei, two species with morphol- or dark bands, spots, or chevrons) are common in clovers, ogies, life histories, and geographic ranges similar to T. and the genetic basis of this variation has been demon- amoenum for the later comparisons. The allozyme data strated in crossing experiments with T. repens (Davies, indicate a strong genetic similarity among these three 1963). At least four distinct leaf marking patterns oc- species. While reviews of the literature by Gottlieb curred among 34 plants grown in 1997 at the University (1981b) and Crawford (1983) have shown that, on av- of California Bodega Marine Reserve from seeds col- erage, congeneric species are separated by a Nei's genetic lected from the Dillon Beach T. amoenum population. distance of ϳ0.33, the greatest Nei's distance among Uniform green (with no red or white markings) was the these three species, averaged across populations, was most common pattern, but even the rarest of the four 0.08 (T. amoenum and T. macraei). The average Nei's patterns occurred at a frequency of 0.09. In contrast, genetic distance between populations of T. amoenum and leaves of 48 T. amoenum plants grown at the same site T. albopurpureum var. dichotomum was only 0.03. from seed of the Occidental population showed only the Levels of within-population genetic variation are also uniform green color pattern (Connors, unpublished data). best compared among species that share the same mating It is possible that the less common marking patterns were system (Barrett and Kohn, 1991). Species in which seeds also present in the Occidental T. amoenum population are produced primarily through self-pollination tend to prior to the bottleneck and have been lost. contain less allozyme variation at the within-population level than outcrossers. For example, data from previous Implications for reintroducing T. amoenumÐPopu- studies on plants, compiled by Hamrick and Godt (1990), lations with limited genetic variation are most vulnerable showed that an average of 20.0% of loci were polymor- to extinction, due to reduced potential for evolution in phic in populations of sel®ng species, while 38.7% of response to environmental changes (Beardmore, 1983; loci were polymorphic in populations of species with Huenneke, 1991). The presence of adequate genetic var- mixed or outcrossing mating systems. Hamrick and Godt iation might be especially critical to T. amoenum popu- (1990) also calculated that populations of sel®ng species lations used for reintroduction, because these populations contained an average of 1.31 alleles per locus, whereas will likely need to evolve and adapt to the major changes populations of species with mixed or outcrossing mating that have occurred within California grassland plant com- systems contained an average of 1.60 alleles per locus. munities. The biotic environment has, in many cases, Several lines of evidence point to signi®cant outcross- been greatly altered by competition with exotic species ing within populations of T. amoenum. The high level of as well as herbivory by cattle and other grazers. heterozygosity that apparently existed in the original Oc- We demonstrated that the Occidental T. amoenum pop- cidental T. amoenum plant would not have been expected ulation contains less allozyme variation than a larger pop- in a predominantly self-pollinating species. In addition, a ulation of the same species and populations of a closely mixed (mixture of sel®ng and outcrossing) mating system related species. Assuming that this loss of allozyme var- is suggested by the ratio of observed heterozygosity to iation resulted from the documented population bottle- expected heterozygosity within both T. amoenum popu- neck, a reduction in genetic variation for other traits of lations (Table 3). Populations of T. albopurpureum var. potentially greater adaptive importance would be expect- dichotomum showed ratios of observed heterozygosity to ed as well. Losses of this often quantitative variation expected heterozygosity which were similar to those in should be proportional to losses of allozyme variation, T. amoenum. In contrast, populations of T. macraei were because genetic drift is the dominant evolutionary force nearly devoid of heterozygosity, indicating a mating sys- in small populations (Barrett and Kohn, 1991), and ge- tem of a high degree of sel®ng. Amount of prebottleneck netic drift affects all genetic variation, allozyme or quan- allozyme variation in the Occidental T. amoenum popu- titative, in a similar fashion. This has been shown exper- lation is therefore best estimated through comparisons to imentally by Polans and Allard (1989) in populations of the Dillon Beach T. amoenum population and the two ryegrass (Lolium multi¯orum), where restrictions in size populations of T. albopurpureum var. dichotomum. Lev- of populations led to reductions in levels of allozyme els of allozyme variation within these latter three popu- variation and also to deleterious effects for quantitative lations, B, C, and D, were very similar (Table 3). The traits, thought to be due to this loss of genetic variation. Occidental T. amoenum population A contained half the However, low levels of genetic variation do not nec- number of polymorphic loci and 21% fewer alleles per essarily mean that reintroduction attempts could not suc- locus compared with the Dillon Beach T. amoenum pop- ceed. For example, Schwaegerle and Schaal (1979) de- ulation B. Considering populations B, C, and D together, scribed a thriving population of over 100 000 pitcher the bottlenecked Occidental T. amoenum population con- plants (Sarracenia purpurea) originating from a single tained 53% fewer polymorphic loci and 20% fewer al- translocated individual. Success of this introduction oc- leles per locus. Still, because we do not know how much curred in spite of the apparently reduced allozyme vari- allozyme variation occurred in the Occidental T. amoen- ation that accompanied the population bottleneck. um population (A) prior to the bottleneck, or the evolu- In addition to loss of evolutionary potential, inbreeding tionary history of populations B, C, and D, such com- depression may be a factor within the Occidental T. parisons can provide only a rough estimate of the amount amoenum population. Inbreeding increases the probabil- of variation that was lost. ity that deleterious recessive alleles will become ®xed, January 1999] KNAPP AND CONNORSÐPOPULATION BOTTLENECK IN TRIFOLIUM AMOENUM 129 and it is often most severe in populations of plants that cidental population, is appropriate because it is appar- normally outcross (Barrett and Kohn, 1991). However, ently more similar to historical populations (as indicated even relatively low rates of self-pollination may purge by comparisons with herbarium specimens) and thus has much of the genetic load from a population (Lande and the potential for reintroduction over a much wider geo- Schemske, 1985; Charlesworth and Charlesworth, 1987). graphic range than the coastal form. Therefore, if the natural mating system of the Occidental A strategy of controlled introgression might provide a T. amoenum population included some degree of self- compromise between maintaining local adaptive variation pollination, the likelihood that inbreeding will have sig- and promoting adequate levels of within-population ge- ni®cant negative effects is reduced. Indeed, plants pro- netic variation. Introgression would involve mixing a duced from these seeds so far appear quite robust, but no small proportion of the nonlocal Dillon Beach source into formal tests of inbreeding depression have been con- the Occidental T. amoenum population over time. Rein- ducted. troducing T. amoenum populations varying in the pro- Both the loss of genetic variation and inbreeding de- portion of the local source and thus presumably differing pression might be ameliorated by expanding the genetic in amount of genetic variation could be conducted in an base of the Occidental T. amoenum population before or experimental context (Barrett and Kohn, 1991; Guerrant, during the process of reintroduction. Numerous searches 1996). This would increase our understanding of the for additional remnant plants in the wild have been con- trade-offs involved. Mixing populations would not be the ducted since 1993 without success until our 1996 discov- method of choice if cross-pollinations between popula- ery at the Dillon Beach site. Using seeds from this more tions result in progeny exhibiting outbreeding depression. variable coastal T. amoenum population to expand the Although outbreeding depression has been reported in genetic base of the inland Occidental population might several plant species, how widespread this phenomenon now be an option, particularly if reintroductions of the is and the degree to which it should be considered when inland population are unsuccessful, and lack of popula- making decisions about whether to mix populations for tion persistence or reductions in the ®tness of individual conservation purposes, are still subject to debate (Fenster plants can be attributed to genetic limitations associated and Dudash, 1994). with reduced variation. Two generations of seed increase have resulted in the However, genetic mixtures may be problematic be- production of Ͼ50 000 seeds of the Occidental T. amoen- cause plants from the two T. amoenum populations differ um population. These seeds remain in storage at various in aspects of plant architecture that may indicate local locations since suitable sites for reintroduction have not adaptation. The original wild Occidental plant and its off- yet been identi®ed. This is partially due to the lack of spring all have an erect growth form, generally taller than knowledge about interspeci®c competition and herbivory broad. This matches the architecture of 28 herbarium and how these variables impact the species, which may specimens we have examined from historic populations make certain sites better candidates than others. Efforts at other inland sites. Before discovery of the Dillon are currently underway to investigate more completely Beach T. amoenum population, all reported populations the adaptive differences between the two T. amoenum and all herbarium specimens of this species, like the Oc- populations and to determine reasons for the species' de- cidental plant, occurred at inland locations. In contrast, cline, in order to better understand site requirements for plants of the Dillon Beach population, growing in the future reintroduction attempts. windy environment of a coastal bluff, have an almost prostrate growth form. These phenotypic differences have LITERATURE CITED a genetic basis. When plants from both populations were grown in a common garden at an inland location near BARRETT,S.C.H.,AND J. R. KOHN. 1991. Genetic and evolutionary Occidental, the differences in plant architecture were consequences of small population size in plants: implications for maintained (ratio of maximum height to maximum width conservation. In D. A. Falk and K. E. Holsinger [eds.], Genetics was 1.13 Ϯ 0.19 [N ϭ 7] for plants from the Occidental and conservation of rare plants, 3±30. Oxford University Press, population and 0.37 Ϯ 0.10 [N ϭ 10] for plants from the New York, NY. Dillon Beach population) (Connors and J. L. Maron, un- BEARDMORE, J. A. 1983. Extinction, survival, and genetic variation. In published data). C. M. Schoenwald-Cox, S. M. Chambers, B. MacBryde, and L. Thomas [eds.], Genetics and conservation, 125±151. Benjamin- Coastal bluff environments are cooler, moister, and Cummings, Menlo Park, CA. much windier than inland grasslands. The prostrate CHARLESWORTH, D., AND B. CHARLESWORTH. 1987. Inbreeding depres- growth form of plants from the Dillon Beach T. amoenum sion and its evolutionary consequences. Annual Review of Ecology population may be an adaptation to strong winds and is and Systematics 18: 237±268. a characteristic shared by most members of the coastal CONNORS, P. G. 1994. Rediscovery of showy Indian clover. Fremontia bluff plant community. Among other species growing in 22: 3±7. CRAWFORD, D. J. 1983. Phylogenetic and systematic inferences from both communities, grasses such as Bromus carinatus and electrophoretic studies. In S. O. Tanksley and T. J. Orton [eds.], Hordeum brachyantherum also have a low, decumbent Isozymes in plant genetics and breeding, part A, 257±287. Elsevier, form on the coastal bluffs and a tall, erect form in inland Amsterdam. grasslands. DAVIES, W. E. 1963. Leaf markings in Trifolium repens. In C. D. Dar- This suggestion of local adaptation argues for caution lington and A. D. Bradshaw [eds.], Teaching genetics in school and in mixing inland (Occidental) and coastal (Dillon Beach) university, 94±98. Oliver and Boyd, Edinburgh, UK. FENSTER, C. B., AND M. R. DUDASH. 1994. Genetic considerations for genetic stocks and leaves us with the question of how plant population restoration and conservation. In M. L. Bowles and best to reestablish the inland form. A focus on reestab- C. J. Whelan [eds.], Restoration of endangered species, 34±62. lishing this form, represented by the single-founder Oc- Cambridge University Press, Cambridge, UK. 130 AMERICAN JOURNAL OF BOTANY [Vol. 86

FRANKEL,O.H.,AND M. E. SOULEÂ . 1981. Conservation and evolution. MILLAR, C. I., AND W. J. LIBBY. 1991. Strategies for conserving clinal, Cambridge University Press, Cambridge, UK. ecotypic, and disjunct population diversity in widespread species. GOTTLIEB, L. D. 1981a. Gene number in species of Asteraceae that In D. A. Falk and K. E. Holsinger [eds.], Genetics and conservation have different chromosome numbers. Proceedings of the National of rare plants, 149±170. Oxford University Press, New York, NY. Academy of Sciences, USA 78: 3726±3729. MORDEN, C. W., J. DOEBLEY, AND K. W. SCHERTZ. 1987. A manual of ÐÐÐ. 1981b. Electrophoretic evidence and plant populations. Prog- techniques for starch gel electrophoresis of Sorghum isozymes. ress in Phytochemistry 7: 1±46. Texas Agricultural Experiment Station, MP1635, College Station, GUERRANT,E.O.JR. 1996. Designing populations: demographic, ge- TX. netic, and horticultural dimensions. In D. A. Falk, C. I. Millar, and NEI, M. 1978. Estimation of average heterozygosity and genetic dis- M. Olwell [eds.], Restoring diversity: strategies for reintroduction tance from a small number of individuals. Genetics 89: 583±590. of endangered plants, 171±207. Island Press, Covelo, CA. ÐÐÐ, T. MARUYAMA, AND R. CHAKRABORTY. 1975. The bottleneck HAMRICK,J.L.,AND M. J. W. GODT. 1990. Allozyme diversity in plant effect and genetic variability in populations. Evolution 29: 1±10. species. In A. H. D. Brown, M. T. Clegg, A. L. Kahler, and B. S. POLANS,N.O.,AND R. W. ALLARD. 1989. An experimental evaluation Weir [eds.], Plant population genetics, breeding, and genetic re- of the recovery potential of ryegrass populations from genetic sources, 43±63. Sinauer, Sunderland, MA. stress resulting from restrictions of population size. Evolution 43: HUENNEKE, L. F. 1991. Ecological implications of genetic variation in 1320±1324. plant populations. In D. A. Falk and K. E. Holsinger [eds.], Ge- SCHWAEGERLE, K. E., AND B. A. SCHAAL. 1979. Genetic variability and netics and conservation of rare plants, 31±44. Oxford University founder effect in the pitcher plant Sarracenia purpurea L. Evolu- Press, New York, NY. tion 33: 1210±1218. HULL, A. C. 1973. Germination of range plant seeds after long periods SMITH,J.P.,AND R. YORK. 1984. Inventory of rare and endangered of uncontrolled storage. Journal of Range Management 26: 198± plants of California. Special publication, California Native Plant 200. Society, Berkeley, CA. KEPHART, S. R. 1990. Starch gel electrophoresis of plant isozymes: a SOLTIS, D. E., C. H. HAUFLER,D.C.DARROW, AND G. J. GASTONY. 1983. comparative analysis of techniques. American Journal of Botany 77: 693±712. Starch gel electrophoresis of ferns: a compilation of grinding buff- LANDE, R. 1980. Genetic variation and phenotypic evolution during ers, gel and electrode buffers, and staining schedules. American allopatric speciation. American Naturalist 116: 463±479. Fern Journal 73: 9±27. ÐÐÐ, AND D. W. SCHEMSKE. 1985. The evolution of self-fertilization SWOFFORD,D.L.,AND R. B. SELANDER. 1989. Biosys-1: a computer and inbreeding in plants. I. Genetic models. Evolution 39: 24±40. program for analysis of allelic variation in population genetics and LEBERG, P. L. 1992. Effects of population bottlenecks on genetic di- biochemical statistics. Illinois Natural History Survey, Champaign, versity as measured by allozyme electrophoresis. Evolution 46: IL. 477±494. WENDEL,J.F.,AND N. F. WEEDEN. 1989. Visualization and interpreta- LESICA,P.,AND F. W. A LLENDORF. 1995. When are peripheral popula- tion of plant isozymes. In D. E. Soltis and P. S. Soltis [eds.], Iso- tions valuable for conservation? Conservation Biology 9: 753±760. zymes in plant biology, 5±45. Dioscorides Press, Portland OR.