Genetic Evidence of Hybridization Between Oenothera wolfii (Wolf's Evening Primrose) and O. glazioviana, a Garden Escape

2008·7.8 MB·English

by  Jennifer Dewoody

Madrono, Vol. 55, No. 2, pp. 132-142, 2008
GENETIC EVIDENCE OF HYBRIDIZATION BETWEEN OENOTHERA WOLFII
(WOLF'S EVENING PRIMROSE) AND O. GLAZIOVIANA, A GARDEN ESCAPE
Jennifer DeWoody*', Leonel Arguello^, David Imper\ Robert D. Westfall"^ and
Valerie D. Hipkins'
USDA
' Forest Service, Pacific Southwest Research Station, National Forest Genetics Lab,
2480 Carson Road, Placerville, CA 95667
-U.S. Department ofthe Interior, National Park Service, Redwood National and State Parks,
P.O. Box 7, Orick, CA 95555
'U.S. Fish and Wildlife Service, 1655 Heindon Road, Areata, CA 95521
USDA Forest Service, Pacific Southwest Research Station, P.O. Box 245, Berkeley,
CA
94701-0245
Abstract
Isozyme analysis ofthe rare Oenothera wolfii(Wolfs evening primrose) and the garden escape, O.
glazioviana, indicates that hybridization between these species may be more widespread than
morphological evidence indicates. Although both speciescontained low amounts ofgeneticvariation,
unique alleles were identified in both taxa. Analysis of22 populations, including pure populations of
each species, identified eight populations as containing putative hybrid individuals. Four of these
putative hybrid populations were considered pure O. wolfii based on morphological analysis. This
study confirms that the native O. wolfii may be at risk not only from habitat destruction, but
potentiallyfromgeneticswampingwhereitco-occurswith O. glazioviana. Theseresultscan beusedas
baseline information for future genetic monitoring efforts.
Key Words: complex heterozygote, genetic swamping, hybridization, isozymes, Oenothera.
Although habitat loss usually poses the great- completelydescribehybrid swarmsofindividuals,
est threat to a rare species' survival, there is particularlyifsecond-generation hybrids orback-
evidence that hybridization with widespread cross individuals occur frequently. Genetic infor-
related taxa poses an immediate threat to some mation may provide greater power to identify
species (Rhymer and Simberloff 1996). A rare hybrids if unique alleles occur in either or both
species may become functionally extinct through purespecies. Evenintheabsenceofuniquealleles,
genetic swamping after repeated hybridization given sufficient variation in neutral, bi-parentally
and backcrossing with a more common species inherited, genetic markers (for example, iso-
(Levin et al. 1996). Management efforts to zymes), statistical methods exist to identify not
minimize hybridization in order to protect a rare only first-generation hybrid individuals, but also
species may not be justified if hybridization second-generation hybrids and introgressed indi-
results from natural processes. By contrast, viduals resulting from backcrosses with either
artificial hybrid zones arising from human- parental species (Rannala and Mountain 1997;
mediated habitat modification or species intro- Rieseberg et al. 1998; Nason et al. 2002).
duction may require management action to Oenothera wolfii [Munz] Raven, W. Dietr.
minimize the potential loss of a rare species Stubbe (Onagraceae) (Wolfs evening primrose)
(Rhymer and Simberloff 1996; Allendorf et al. is a biennial to short-lived perennial native to the
2001). In orderto minimize the effects ofartificial coastal areas ofnorthern CaHfornia and southern
hybrid zones, managers must be able to distin- Oregon. Populations of this species are rare and
guish between pure populations ofa rare species patchy in distribution, found on moderately
and hybrid swarms where the two species coexist. disturbed sites, including the upper margin of
Frequently, hybrids display phenotypes interme- beach strand and coastal bluffs (Imper 1997).
diate to either parent species, although hybrid While disturbance resulting from continued
morphology may be extreme to either parent development and recreation along the coast have
(Schwarzback et al. 2001). Due to these varia- created new habitat for O. wolfii in some
tions, morphology alone may be insufficient to instances, the net effect of human encroachment
has been negative for existing populations (Imper
*Author for correspondence. Current address: Uni- 1997). As a result, O. wolfii is listed as threatened
versity ofSouthampton, School of Biological Sciences, by the state of Oregon, and both the CaUfornia
Southampton, SO16 7PX, United Kingdom. +44-(0)79- Native Plant Society and the Oregon Natural
8363-0570 (voice), [phone] (fax),j.dewoody@ Heritage Program list this species as endangered
soton.ac.uk. throughout its range (Imper 1997).
A
2008] DEWOODY ET AL.: HYBRIDIZATION IN OENOTHERA 133
While habitat loss is affecting O. \volfii\ of two distinct genomes. Multiple reciprocal
hybridization with a common congener, O. translocations across the genome have produced
glazioviana Micheli, may prove the more imme- a single linkage group consisting of both sets of
diate threat (Imper 1997). As a garden escape chromosomes at meiosis. As a result, the 14
(i.e., a horticukural species of hybrid origin that chromosomes in a diploid individual form a
has become established in natural areas), O. single ring instead of seven bivalents. This
glazioviana may be interfertile with O. wolfii, phenomenon persists through self-pollination
producing artificial hybrid zones where the coupled with balanced lethals, with gametophytic
species coexist. Several factors support this and sporophytic lethals persisting in the hetero-
hypothesis. First, introgression is common be- zygous state. Although self-fertilization may
tween many members of this genus. Greenhouse produce embryos homozygous for either genome,
experiments have shown that hybridization be- only heterozygous embryos survive to produce
tween O. wolfii and other members of the genus viable seed, since those embryos homozygous for
readily occurs (Wasmund and Stubbe 1986). one genome will also be homozygous for either
Second, individuals of hybrid origin have been the sporophytic or gametophytic lethal allele. As
identified at the California-Oregon border area a result, alleles are not independently assorted,
based on morphological traits (Carlson et al. and populations are not randomly mating. This
2001). Hybrids are fertile, vigorous, and display a mechanism explains the lack of genotypic diver-
greater fitness than either parent species (Imper sity and recombination observed in the data set
1997). Although genetic typing ofhybrid individ- (see Results), and prevents the use of statistical
uals indicates that hybrids tend to breed true, analyses typical ofco-dominant geneticdata (e.g.,
there is evidence ofhybrids back-crossing with O. admixture analyses or population assignment
wolfii (Imper 1997). Third, O. wolfii is potentially tests).
susceptible to genetic swamping by O. glazioviana Samples were collected from populations at 22
based on the mating systems of each species. O. sites whose taxonomy was determined by mor-
wolfii is self-compatible and produces the major- phological traits (Tables 1 and 3, Fig. 1). Typical
ity of its seed via self-pollination (Carlson et al. O. wolfii plants produce small (<5 cm) pale-
2001). This breeding system is a consequence of yellow corollas having petals that do not overlap,
O. wolfii'^ complex heterozygous genome, which with sepals covered in dense long-spreading
is maintained through self-fertilization and bal- pubescence, both villous and glandular pubes-
anced lethals, and results in approximately halfof cence occur on fruits, and plants have a reddish
the mature pollen grains being sterile (Wasmund upper stem (Imper 1997). By contrast, typical O.
and Stubbe 1986). In contrast, O. glazioviana is glazioviana plants produce larger (>8 cm) bright
an outcrossing species (Imper 1997). Given the yellow flowers having substantial overlap in the
asymmetry of available pollen between these petals, minimal pubescence on either sepals or
parent species asymmetric gene flow might occur fruit, green upper stems, and more wrinkled and
as O. glazioviana pollen swamps O. wolfii stigmas lighter green foliage than observed on O. wolfii
at sympatric sites. Together, this evidence sug- (Imper 1997). Field observations identified four
gests that hybridization likely occurs between this populations as O. glazioviana (nos. 1 to 4; the
rare endemic and the widespread garden escape. populations ofthe garden escape most proximate
This study reports an investigation into the to O. wolfii), 14populations as O. wolfii(nos. 6 to
extent and structure of hybrid zones between O. 19), and three populations as intermediates or
wolfii and O. glazioviana using isozymes, which putative hybrids (nos. 20 to 22). Field observa-
are putatively neutral, bi-parentally inherited, tions could not distinguish between O. glaziovi-
molecular markers. Three questions were ad- ana and O. data, a common congener at one site
dressed: First, does sufficient genetic variation (no. 5), and one population appeared to be O.
exist to discriminate between pure O. wolfii and wolfii, but occurred in a novel location (no. 15).
O. glazioviana populations? Second, can hybrid A single leafwas collected from between four and
populations be identified using these molecular 25 individuals in each population for subsequent
markers? Third, what is the frequency of hybrid genetic analyses.
individuals in natural populations of O. wolfiP. Tissue was prepared for isozyme analysis
Ultimately, these genetic findings provide greater following the liquid nitrogen procedure using
insight and guidelines for management plans and Gottleib (1981) extraction buffer, as described in
conservation objectives. NFGEL Standard Operating Procedures (USD
Forest Service 2003). Samples were frozen at
Methods —70°C until electrophoresis.
Electrophoresis took place on three buffer
O. wolfii is a complex heterozygote (Wasmund systems (adapted from Wendel and Weeden
and Stubbe 1986), or complex hybrid (Bussell et 1989): a tris-citric acid gel buffer (pH 8.3) with
al. 2002), where a diploid individual contains not a lithium hydroxide-boric acid tray buffer
two copies ofa single genome, but a single copy (pH 8.3; LB), a tris-citric acid gel buffer
W
MADRONO
134 [Vol. 55
Table 1. Population Number, Name, Location (Latitude, Longitude), and Species Compositionof 22
Sites Sampled for This Study. Species composition was determined by field observations and genetic analysis,
and is indicated by: WO = O. wolfii, GL = O. g/azioviana, HY = intermediate morphology potentially due to
hybridization, and UN = unknown taxonomy.
Number Name Location Species
21 CChraerslceesnttonC,ityC,ooDselCoN.o,rtOeRCo., CA 4431..37349876NN,, 112244..32300282WW GGLL
3 Manila, Humboldt Co., CA 40.8483N, 124.1650W GL
4 Trinidad, Humboldt Co., CA 41.0353N, 124.1058W GL
5 Junction City, Trinity Co., CA 40.7378N, 123.0575W UN
6 Port Orford City Park, Curry Co., OR 42.8320N, 124.5020W UN
7 Houda Point, Humboldt Co., CA 41.0359N, 124.1187W WO
8 Port Orford Beach, Curry Co., OR 42.7318N, 124.4825W UN
9 Port Orford Bridge, Curry Co., OR 42.7318N, 124.4825W UN
111102 PPLiuosfitfnoetlnhSRtoi.lvteGzre,,orHCguuerm,rbyoDelCldo.tN,oCrOot.R,e CCoA., CA 444211...207737571387NNN,,, 111222444...1422045407I5WW wWwoOo
13 Devil's Gate, Humboldt Co., CA 40.4055N, 124.3914W UN
14 Davis Creek, Humboldt Co., CA 40.3765N, 124.3725W wo
15 McKerricher State Park, Mendocino Co., CA 39.5146N, 123.7769W UN
16 Freshwater Spit, Humboldt Co., CA 41.2667N, 124.1058W WO
17 Crescent Beach, Del Norte Co., CA 41.7194N, 124.1447W wo
18 False Klamath Cove, Del Norte Co., CA 41.6027N, 124.1064W wo
19 Crescent Overlook, Del Norte Co., CA 41.7048N, 124.1447W wo
20 Klamath, Del Norte Co., CA 41.5151N, 124.0298W HY
21 Lucky Bear Casino, Del Norte Co., CA 41.9529N, 124.2022W HY
22 Fruit Station, Curry Co., OR 41.9984N, 124.2124W HY
(pH 8.8) with a sodium hydroxide-boric acid tray But, portions of the Oenothera isozyme data
buffer (pH 8.0; SB), and a citric acid-N-(3- violate these assumptions. Given the unique
aminopropyl)-morpholine gel and tray buffer nature of the genetic system of O. wolfii, and
(pH 8.0; MC8). A total of 15 loci were examined. the lack of statistical procedures available to
Four loci were resolved on the LB system: account for complex hybridity as a mode of
phosphoglucose isomerase (PGI2), phosphoglu- inheritance, an ad hoc approach involving two
comutase (PGMl), and two loci in leucine statistical analyses was employed to determine if
aminopeptidase (LAPl and LAP2). Four loci hybridization occurs between O. wolfii and O.
were also resolved on the SB system: aspartate glaziovianna: multivariate analyses over popula-
aminotransferase (AATl), superoxide dismutase tions and individuals, and a maximum likelihood
(SODl), triosephosphate isomerase (TPIl), and analysis over populations. While acknowledging
uridine diphosphoglucose pyrophosphorylase that neither approach is statistically ideal, we
(UGPPl). Seven loci were resolved on the MC8 contend that given the unique nature of the
system: two loci in esterase (ESTl and EST2), genome of Oenothera which may bias results ofa
fluorescent esterase (FESTl), isocitrate dehydro- single statistical test, combining statistical analy-
genase (IDH), malate dehydrogenase (MDH), ses with careful examination of the electropho-
and two loci in 6-phosphogluconate dehydroge- retic patterns provides an informative approach
nase (6PGD1 and 6PGD2). All stain recipes were to describe the genetic similarities and potential
adapted from Conkle et al. (1982). Banding for hybridization between these species.
patterns were consistent with pubhshed protein For the multivariate analysis, we scored each
structure (Crawford 1989). allele as 1.0, 0.5, or 0.0 for homozygous,
As a consequence of complex hybridity in O. heterozygous, or homozygous for another allele,
wolfii, the isozyme data resolved in this study respectively (Westfall and Conkle 1992). These
violate two assumptions common to most statis- scored data were submitted to a canonical
We
tical analyses designed for genetic data: indepen- discriminant analysis. first ran the analysis
dent assortment of loci and random mating. on populations, without respect to species classi-
Multivariate analyses can often be used to fication to determine if populations grouped by
analyze co-dominant genetic data, but these species identity. We then contrasted these results
analyses require that errors (or residuals) are with a classification based on the apriori group-
normally distributed. This assumption is gener- ings. Based on the canonical discriminant plot,
ally satisfied under independent assortment and each population was classified as either "pure"
random mating, and experience shows that even (that is, a parental species) or "unknown" (that
binary-scored markers can fit the assumption. is, either putative hybrid or unknown taxonomy)
2008] DEWOODY ET AL.: HYBRIDIZATION IN OENOTHERA 135
1
MADRONO
136 [Vol. 55
Table 2. Isozyme Diversity Summary Statistics for 22 Populations Described in Table 1. Means
OVER SpeciesIncludeOnly "Pure" Populations. = numberofsamples, P = percent polymorphicloci, A =
meanallelesperlocus, Ap = meanallelesperpolymorphiclocus, //„ = observedheterozygosity,F= fixationindex.
Variance reported in parentheses.
T^or\iilc\i^^^t^ p /±p /17
Mean over species:
O. wo/fii 137 13.33 1.200 (0.293) 2.500 0.021 (0.005) -0.571
O. glazioviana 61 6.67 1.067 (0.062) 2.000 0.067 (0.062) -1.000
1 25 0.067 1.067 2.000 0.067 —1.000
2 17 0.067 1.067 2.000 0.067 —1.000
3 1 0.067 1.067 2.000 0.067 -1.000
4 8 0.067 1.067 2.000 0.067 -1.000
5 21 0.133 1.133 2.000 0.133 -1.000
6 6 0.133 1.133 2.000 0.078 -0.750
7 12 0.067 1.067 2.000 0.011 —0.048
-
8 9 0.067 1.067 2.000 0.067 1.000
9 9 0.067 1.067 2.000 0.067 -1.000
10 13 0.067 1.067 2.000 0.015 -0.091
11 25 0.000 1.000 n.a. 0.000 0.000
12 19 0.067 1.067 2.000 0.018 -0.125
13 10 0.200 1.200 2.000 0.073 0.214
14 9 0.133 1.133 2.000 0.067 -0.385
15 10 0.000 1.000 n.a. 0.000 0.000
16 25 0.000 1.000 n.a. 0.000 0.000
17 5 0.000 1.000 n.a. 0.000 0.000
18 25 0.067 1.067 2.000 0.067 -1.000
19 4 0.000 1.000 n.a. 0.000 0.000
20 10 0.067 1.067 2.000 0.067 -1.000
21 10 0.067 1.067 2.000 0.067 -1.000
22 5 0.067 1.067 2.000 0.067 -1.000
(Table 1). The allele frequencies observed over all given to those loci displaying alleles unique to at
pure populations of each species were then used least one parental species.
as the basis to classify each unknown population
using the methods described above. These anal- Results
yses were done in JMP (SAS Institute, Inc, 2004).
This software's canonical discriminant analysis is Six ofthe 15 loci examined were polymorphic:
based on Bayesianprobabihties, whereby, inwell- 6PGD2, AATl, UGPPl, FESTl, ESTl, and
differentiated species, individuals of one species EST2 (Appendix A). Four loci displayed varia-
will have a probability of 1.0, those of the other tion within or among populations of the pure
species, 0.0, and those of hybrid or backcross species, and two loci contained variation in
types will have probabilities between 1.0 and 0.0. unknown populations not observed in either pure
The second analysis estimates the frequency of species. Low levels of genetic variation were
six genealogical classes (each parental class, first- observed over all populations surveyed (Table 2).
and second-generation hybrids, and first genera- Based on the mean over species, O. wolfii
tion backcross to each parent species) in each contained higher levels ofpolymorphism (percent
population based on the maximum likelihood polymorphic loci, P = 13.3), alleles per locus {A
estimates ofthe multilocus genotypes observed in = 1.20), and alleles per polymorphic locus {Ap =
a population arising from the allele frequencies 2.50), but lowerlevels ofheterozygosity (observed
observed in each of the pure parental species proportion ofheterozygotes, //q = 0.021) than O.
(Nason et al. 2002). While this method assumes glazioviana (P = 6.67, A = 1.07, Ap = 2.00, Ho =
both independent assortment of alleles and 0.067). Many populations displayed an excess of
random mating within each population, assump- heterozygotes, as indicated by fixation indices,
tions that are violated here, it includes all loci in which is consistent with complex hybridity
the data set in the maximum likelihood estima- (Table 2).
tions, without requiring unique alleles in each Multivariate analyses over populations indi-
parent species (Nason et al. 2002). cated sufficient genetic differentiation exists to
Finally, the conclusions from each statistical distinguish between O. wolfii and O. glazioviana
analysis were considered in the context of the (Fig. 2). In general, populations grouped by
alleles observed in each genotype in the nine species identity as defined in field observations,
unknown populations, with particular attention with the majority ofpopulations being separated
2008] DEWOODY ET AL.: HYBRIDIZATION IN OENOTHERA 137
all individuals as either pure O. wolfii or pure O.
HYno.20 glazioviana (Table 3); no intermediate probabili-
wo ties were observed. Samples from three popula-
UNno. 15 tions identified as pure O. wolfii based on
in - morphological observations were classified as O.
glazioviana (nos. 6, 8, and 9). Of the three
remaining populations of unknown taxonomy,
o - UNno.13 afliledinadsivO.idguallasziofvrioamnao(nneo.po5)p,uallaltfironomweanroethclearssais-
O. wolfii (no. 15) and the final population
o6U.N8.n9os. cboontthaipnuerde aspemciixetsur(neo.of13i)n.diOvfidtuhaelsthcrleaessipfoipeudlaa-s
tions classified as hybrid based on morphological
LO _ GL observations, genetic analyses classified samples
UHNYnnoo.s.521.22 from one as O. wolfii(no. 20), and those from the
Q) other two as O. glazioviana (nos. 21 and 22).
Genealogical class frequency estimates were
-30 -20 -10 0 also inconsistent with morphological predictions
CAN1 (Table 3). Unlike the Bayesian classifications,
however, some genotypes were identified as
Fig. 2. Distribution ofpopulationsalongthe first two consistent with hybrid origin. Ofthe populations
canonical variables produced by a discriminate coordi-
nate analysis. Populations classifications and numbers identified as pure O. wolfii a priori, three were
correspond to Table 1. Canl = first canonical variable, classified as hybrids (nos. 6, 8, and 9), and a
Can2 = second canonical variable. fourth (no. 13) was classified as a mixture of O.
wolfii and hybrids. Consistent with the Bayesian
by species identification along the second canon- classifications, two of the putative hybrid popu-
ical axis (Fig. 2). However, the first two canonical laantdion2s2)w,eraeltchloasusgihfietdheastOh.irdglawzaisovicalnaassi(fnioesd. 2a1s
coefficients revealed greater genetic differentia-
tion than predicted among populations classified backcross to O. wolfii (no. 20).
as O. wolfii from morphological characteristics.
Discussion
In particular, three populations from Oregon,
nos. 6, 8, and 9, were genetically distinct from the Does sufficient genetic variation exist to
other populations considered pure O. wolfii
discriminate between species?
(Fig. 2). Additionally, population 13 was inter-
mediate to the two parental species (Fig. 2). As a In genetic studies of hybridization, the genetic
result of these observations, these four outlying variation in each species is often defined by
populations were classified as unknown taxono- identifying ''pure" populations from morpholog-
my forthe remaining statistical tests, reducingthe ical observations and assaying each for genetic
number ofO. wolfii populations to those listed in markers. Alternatively, multivariate analyses
Table 1. such as the canonical discriminant analysis
No evidence of hybridization (or admixture) described above, can identify genetically similar
was found using the Bayesian classification or distinct populations without a priori classifi-
analysis of individuals. Bayesian tests classified cation. In this study, the canonical discriminant
Table 3. Classification of Nine Populations of Unknown or Hybrid Origin Based on 6 Variable
Isozyme Loci. See text for details of the Bayesian classifications and genealogical class freuqencies. Isozyme
phenotypes are classified by the presence ofalleles found to be unique to either parental species.
Field Bayesian Genealogical
Population Observations classification class frequency Isozyme phenotype
5 Unknown O. glazioviana Backcross to O. Neither species or Hybrid
glazioviana
6 O. wolfii O. glazioviana Hybrid Hybrid
8 O. wolfii O. glazioviana Hybrid Hybrid
9 O. wolfii O. glazioviana Hybrid Hybrid
13 O. wolfii Mix ofpure parental Mix of O. wolfii and Mix of O. wolfii and Hybrid
individuals Hybrid
15 O. wolfii O. wolfii O. wolfii O. wolfii
20 Hybrid O. wolfii Backcross to O. wolfii Hybrid
21 Hybrid O. glazioviana O. glazioviana O. glazioviana
22 Hybrid O. glazioviana O. glazioviana O. glazioviana
MADRONO
138 [Vol. 55
analysis indicated that four populations which data were collected from other Oenethera species,
were identified as O. wolfii in field observations we cannot test this alternate model.
were genetically distinct from the other O. wolfii Analyzing genetic data from these species and
populations (Table 3, Fig. 2). Given the striking conclusively identifying hybrid individuals is
genetic differences between these populations, the further complicated by the recombination system
four outliers were treated as "unknown" taxon- displayed by O. wolfii. As a complex hybrid,
omy for the remaining data analyses and putative diploid individuals contain not two
interpretation. copies of a single genome, but one copy each of
The multivariate analysis also indicates that two distinct genomes. Wasmund and Stubbe
sufficient genetic differentiation exists between (1986) showed that O. wolfii maintains this
the nine O. wolfii populations and four O. heterozygosity through self-fertilization coupled
glazioviana populations to discriminate between with balanced lethals. This recombination system
the parental species (Fig. 2). Although isozyme causes species to be functional apomicts, typically
markers revealed low levels of variation in O. displaying little genetic variation and heterozy-
wolfii and O. glazioviana (Table 2), greater gosity (Russell and Levin 1988). The low levels of
variation was observed in O. wolfii (0-20% allelic diversity and near lack of genotypic
polymorphic loci) than O. glazioviana (6.7% diversity observed in O. wolfii are consistent with
polymorphic loci), and all samples from "pure" these expectations. Although sufficient genetic
O. glazioviana populations (nos. 1^) shared a variation exists to allow differentiation of pure
common genotype: heterozygous at AATl, and species and identification of hybrid individuals,
monomorphic at all other loci. O. wolfii con- the lack of recombination and independent
tained a greater number ofalleles per locus (1.20 assortment at meiosis means that these data
compared to 1.07 in O. glazioviana), and dis- violate the assumptions common to most statis-
played greater levels offixation (—0.57 compared tical analyses. Thus, standard statistical methods
to —1.00 in O. glazioviana). of identifying and monitoring hybrid swarms
However, had four populations initially con- may not be applicable to O. wolfii. In order to
sidered O. wolfii (nos. 6, 8, 9, and 13) been appropriately interpret genetic data without
included in thedescription ofthe parental species, loosing information due to violations of model
the genetic differentiation would have been much assumptions, comparing the results of multiple
less pronounced. Specifically, analyzing these statistical analyses coupled with phenotypic
populations as O. wolfii would affect the distri- descriptions ofthe multilocus genotypes provides
bution of alleles at locus 6PGD2, making allele insight into the origin of unknown or putative
6PGD2-2 no long unique to O. glazioviana, but hybrid populations.
shared between the species. Allele AAT-1 would As a final caveat, interpretation ofthisdata set,
remain unique to O. glazioviana, however. This aswell as its applicationin future studies, must be
difference would have likely reduced but not considered in thecontext ofthe small sample sizes
removed the ability of the multivariate and at some populations and the small number of
genealogical class frequency analyses to distin- pure O. glazioviana populations sampled. Anal-
guish between pure and hybrid individuals. ysis of additional "pure" populations of O.
These analyses are compHcated, however, by glazioviana may identify additional unique alleles
the occurrence of alleles in several populations or reveal alleles thought to be unique to O. wolfii
that are not observed in either pure species to be shared by the two species. Either observa-
(Appendix A). There are three possible explana- tion could change the classification of unknown
tions for these observations. First, these alleles samples and the conclusions herein. Ultimately,
may be present in other populations ofeither or this data set represents a fraction of the
both parental species that were not sampled for Oenothera genome, and may not completely
this study. Second, considering populations 6, 8, represent the levels of variation or hybridization
and 9 as O. wolfii as per field observations would in these species.
make alleles ESTl-2, EST2-2, and FESTl-2
unique to O. wolfii. As population 20 has Can hybrid populations be identified
consistently been considered of putative hybrid using isozymes?
origin, such a change in classification of other
populations would not explain the origin ofallele The genetic differences observed between the
FESTl-3. Third, the model we are testing, that all nine populations of O. wolfii and the four
populations are either pure O. wolfii, pure O. populations of O. glazioviana are sufficient to
glazioviana, or an admixture ofthe two, may not allow identification of hybrid populations. Re-
explain the genetic structure observed. Past sults of the two statistical analyses are inconsis-
hybridization and introgression between O. wolfii tent, but indicate that hybridization may occur at
and a third, unidentified species (possibly O. a rate greater than that expected from morpho-
elata) may explain the high frequency ofalternate logical observations. The multivariate classifica-
alleles observed in some test populations. As no tion ofthe six unknown and threeputative hybrid
2008] DEWOODY ET AL.: HYBRIDIZATION IN OENOTHERA 139
populations identified each collection as either Despite violations of assumptions in each
parental species ora mixture ofthe two (Table 3). statistical analysis, this genetic study reveals
The genealogical class frequency estimates, by evidence of hybridization between the rare
contrast, only identified three populations as endemic O. wolfii and the garden escape O.
either parental species, and the remaining popu- glazioviana. A number of genotypes contain
lations as some hybrid origin (Table 3). This lack alleles found to be unique to each pure species
of consensus between analyses may be a conse- (Appendix B), an observation most easily ex-
quence ofthe violations ofthe statistical assump- plained as evidence ofhybrid origin. These results
tions these data present. Three general conclu- indicate hybridization may occur at a greater rate
sions can be made when the field observations than expected based on morphological observa-
and statistical analyses are considered together. tions alone. Although the genetic structure of
First, populations 6, 8, 9, 13, and 20 are distinct population 20, a putative hybrid population
from either parental species. Second, populations found to contain an intermediate genotype,
5, 21, and 22 moreclosely resemble O. glazioviana demonstrates that not all hybridization events
than O. wolfii. Third, population 15 is consistent will lead to the genetic swamping of the rare
with being O. wolfii both morphologically and species, timely removal of O. glazioviana plants
genetically. from sympatric sites may be warranted to prevent
further loss ofthe endemic genotype.
What is the frequency ofhybrid individuals in uniWqhuieletotheeacihsopzayrmeentloscpiecuiseesd, ahenrdetphruosvtihdeeaablilleilteys
natural populations?
to identify hybrid individuals, the direction of
Although population-level analyses to detect hybridization and introgression cannot be as-
hybridization produced inconsistent results, care- sessed due to their bi-parental inheritance.
ful consideration of the multilocus genotypes Assaying these species for variation at maternal-
demonstrates that plants from six populations ly-inherited markers (e.g., chloroplast haplo-
display alleles unique to both parent species, and types), and combining data from those markers
are thus consistent with hybrid origin (Appen- with data from isozyme or other nuclear markers
dix B). Samples from four populations consid- may provide the power to determine which
ered O. wolfii from field observations (nos. 6, 8, species is serving as the seed donor in each
9, and 13) contained one allele unique to O. hybridization, and thus determine if O. wolfii
glazioviana (6PGD2-2) as well as one unique to flowers are being swamped by O. glazioviana
O. wolfii(UGPPl-2). Had these populations been pollen. In addition, including O. elata in future
considered pure O. wolfii for the classification studies would be prudent given the high rates of
tests, and allele 6PGD2-2 would consequently be hybridization between many members of this
shared between the parental species. However, genus.
populations 6, 8, and 9 also displayed three alleles Acknowledgments
not observed in either parental species, ESTI-2,
EST2-2, and FESTl-2. These alleles may be The authors would like to thank J. D. Nason for
unique to either parental species but not detected insightful discussion about this study as well as R. K.
in the pure populations, or it may be the result of Dumroese and colleagues and two anonymous review-
introgression with another Oenothera species (e.g. ers for thoughtful comments on the manuscript.
O. elata). As no other species was included in this
Literature Cited
study, no conclusions can be made regarding the
origin ofthese alleles from these data. Allendorf, F. W., R. F. Leary, P. Spruell, and
The genotype observed in population 5 is J. K. Wenburg. 2001. The probTleRmEsNwDitSh hybrids:
consistent with hybrid origin irrespective of the setting conservation guidelines. in Ecol-
ogy and Evolution 16:613-622.
ccloanstsaiifniscattihoen aolftepronpautleatailloenlse u6,ni8q,ueantdo 9O,. agslai-t BussSE.LHL., JJ.amDe.s,.M20.02W.aMyocloetcutl,arJ.phAy.loCghenaeptpiicllan,alaysnids
zioviana, AATl-2, as well as the allele unique to of the evolution of complex hybridity in Isotoma
O. wolfii, UGPPl-2. Similarly, the genotype petraea. Evolution 56:1296-1302.
observed in population 20 is also consistent with Carlson, M. L., R. J. Meinke, and A. Wierck.
a hybrid origin, containing the O. glazioviana 2001. Wolfs Evening Primrose {Oenothera wolfii)
allele AATl-2 as well as the O. wolfii allele hybridization, reproductive ecology, seed germina-
6PGD2-1. However, the genotype in population tion, and cultivation. Oregon Department of
20 also contains two alleles observed in popula- Agriculture, Plant Conservation Biology Program,
tAaignoanisanl,l6e,lge8i,vuaennnidtqhu9ee(aEbtSsoeTni1tc-se2 poaofnpdtuhlEeaSsteTi2ao-ln2l)el,(eFsaEsiSnTweleli-lt3h)ae.sr CONKASrLeaNsEliDe,sm,SMo..fOCR.Tc..o,nHiupf.nertDe.sre.HeOdsD1:G98S2Ka.ISlSSat,baorrcLa.htoBg.relyNeUmlNeMcNatAnrLuoLpahYlo,.-
parental species and the lack of other Oenothera General Technical Report PSW-64. USDA Forest
species in this study, the origin of these alleles Service, Pacific Southwest Forest and Range
cannot be determined. Experiment Station, Berkeley, CA.
MADRONO
140 [Vol. 55
Crawford, D. J. 1989. Enzyme electrophoresis and Rhymer, J. M. and D. Simberloff. 1996. Extinction
plant systematics. Pp. 146-164 in D. E. Soltis and byhybridizationandintrogression. AnnualReview
P. S. Soltis (eds.). Isozymes in plant biology. ofEcology and Systematics 27:83-109.
Dioscorides Press, Portland, OR. RiESEBERG, L. H., S. J. E. Baird, and a. M.
Grant, V. 1971. Plant Speciation. Columbia Univer- Desrochers. 1998. Patterns of mating in wild
sity Press, New York, NY. sunflower hybrid zones. Evolution 52:713-726.
Gottlieb, L. D. 1981. Gene number in species of Russell, J. and D. A. Levin. 1988. Competitive
Asteraceae that have different chromosome num- relationships of Oenothera species with different
bers. Proceedings of the National Academy of recombination systems. American Journal of Bot-
Sciences USA 78:3726-3729. any 75:1175-1180.
Imper, D. K. 1997. EcologyandconservationofWolfs SCHWARZBACK, A. E., L. A. DONOVAN, AND L. H.
evening primrose in northwestern California. Rieseberg. 2001. Transgressive character expres-
Pp. 34-^0 in T. N. Kaye, A. Liston, R. M. Love, sion in a hybrid sunflower species. American
D. L. Luoma, R. J. Meinke, and M. V. Wilson Journal ofBotany 88:270-277.
(eds.). Conservation and management of native USDA Forest Service. 2003. National Forest Genet-
plants and fungi. Native Plant Society ofOregon, ic Electrophoresis Laboratory standard operating
Corvallis, OR. procedures. USDA Forest Service Pacific South-
Levin, D. A., J. Francisco-Ortega, and R. K. west Research Station, Placerville, CA.
Jansen. 1996. Hybridization and the extinction of Wasmund, O. and W. Stubbe. 1986. Cytogenetic
rare plant species. Conservation Biology 10:10-16. investigations on Oenothera wo/fii (Onagraceae).
Nason, J. D., S. B. Heard, and F. R. Williams. Plant Systematics and Evolution 154:79-88.
2002. Host-associated genetic differentiation in the Wendel, J. F. ANDN. F. Weeden. 1989. Visualization
goldenrod elliptical-gall moth, Gnorimoschema and interpretation of plant isozymes. Pp. 5-45
gallaesolidaginis (Lepidoptera: Gelechiidae). Evo- in D. E. Soltis and P. S. Soltis (eds.), Isozymes
lution 56:1475-1488. in plant biology. Dioscorides Press, Portland,
Rannala, B. and J. L. Mountain. 1997. Detecting OR.
immigration by using multilocus genotypes. Pro- Westfall, R. D. and M. T. Conkle. 1992. Allozyme
ceedingsoftheNationalAcademyofSciencesUSA markers in breedingzonedesignation. New Forests
94:9197-9201. 6:279-309.
;
2008] DEWOODY ET AL.: HYBRIDIZATION IN OENOTHERA 141
o o o o o o o
o o o o ri o o o
in in in in in in in
o d d d d d d d d
o o o o o o o o o o o o o o o o o o o o o o
0iod0no oo oo oo oo oo oidn odin oo oidn oidn oo oo 0od0o odin odin oo oo oo odin oo oo oo oo
o o
o o
o o
oooooooooo
o o o o o o o oooooooooo o o
o o o o o o o oooooooooo o o
o o o o o o o o o
Oo -^H
ri —I
d d
o o
o o
o o
Oinn ooO ooo ooo ooo ooo ooo o0oi0on^0oo^0oo)ooo0doo0o0N—0oO—0oo0—oo0—oo0oo0— —oopooo—
^ ^ ' '
o o
o o
o o
ooOooO OOoOOoOOoOOOoOro<O-oOioOO oooooooooooooooooooooooooooooo ooo ooo
o o o o o o o o
o o o o o o o o
IT) IT) IT) in IT) in in in
d> (6 cS d> d>
oooooooooooooooooooooo
o o oooooooooooooooooooooo
oO o inmininmooooooooooooooiniriiT)
in
o o o o o o
oooooooooooooooooo oooooo ooo ooo
ooooooooooo
ooo ooooooo
ooo ooooooo
IT)
IT)
II
o o n ^ ^ r<-)':fin^r--ooONO'-^ri
in 00 ON