Contributed Paper
Conservation Value of Sites of Hybridization in
Peripheral Populations of Rare Plant Species
JOHN D. THOMPSON,∗ § MYRIAM GAUDEUL,†‡ AND MAX DEBUSSCHE∗
∗
UMR 5175 Centre d’Ecologie Fonctionnelle et Evolutive, CNRS, 1919 route de Mende, 34293 Montpellier cedex 5, France
†Laboratoire d’Ecologie Alpine (LECA), UMR 5553 CNRS—Université J. Fourier, BP 53, 38041 Grenoble cedex, France
‡UMR 7205 CNRS—Muséum National d’Histoire Naturelle, 16 rue Buffon, 75005 Paris, France
Abstract: Populations at the periphery of a species’ range are of interest to conservation biologists because
they can show marked genetic differentiation from populations at the center of a range and because of potential hybridization among rare and common species. We examined two closely related Cyclamen species. One
is a narrow endemic, and the other is more geographically widespread (both protected by law in continental
southern France). We used floral traits and genetic variability to test for hybridization among the species in
peripheral populations of the rare species. The species co-occurred on Corsica in a disjunct, peripheral part of
the distribution of the endemic species and in an ecologically marginal area for the widespread species. The
two species have hybridized and the endemic species showed high levels of introgression with its widespread
congener. Genetic and floral variability in sites with both species was markedly higher than in sites with a
single species. Our results highlight the need for a conservation strategy that integrates hybrid populations
because they represent a source of novel diversity that may have adaptive potential.
Keywords: conservation, Cyclamen, endemism, hybridization, Mediterranean, marginal population, peripheral
population
Valor de Conservación de Sitios de Hibridación en Poblaciones Periféricas de Especies de Plantas Raras
Resumen: Las poblaciones en la periferia del rango de distribución de una especie son de interés para los
biólogos de la conservación porque pueden mostrar marcadas diferencias genéticas entre las poblaciones del
centro del rango y por la potencial hibridación entre especies raras y comunes. Examinamos dos especies
de Cyclamen cercanamente relacionadas. Una es endémica y la otra tiene una distribución geográfica más
amplia (ambas están protegidas por ley en el sur de Francia). Utilizamos caracterı́sticas florales y variabilidad
genética para probar la hibridación entre las especies en poblaciones periféricas de la especie rara. Las especies
coexistieron en Córcega en una parte disyunta, periférica de la distribución de la especie endémica y en un
área ecológicamente marginal para la especie de distribución amplia. Las dos especies han hibridizado y
la especie endémica mostró altos niveles de introgresión con su congénere más distribuido. La variabilidad
genética y floral en sitios con ambas especies fue marcadamente mayor que en sitios con una sola especie.
Nuestros resultados acentúan la necesidad de una estrategia de conservación que integre poblaciones hı́bridas
porque representan una novedosa fuente de diversidad que puede tener potencial adaptativo.
Palabras Clave: conservación, Cyclamen, endemismo, hibridación, Mediterráneo, población marginal,
población periférica
Introduction
Populations of widespread species that occur at the
species’ range limit (i.e., peripheral populations) have
long caught the attention of ecologists and population
geneticists because they provide insights into the factors
that shape distribution limits and the roles of adaptation
and drift in population divergence and speciation (Levin
1970; Bantock & Price 1974). Peripheral populations are
often small, spatially isolated, and occur in ecologically
§email [email protected]
Paper submitted September 30, 2008; revised manuscript accepted May 5, 2009.
236
Conservation Biology, Volume 24, No. 1, 236–245
C 2009 Society for Conservation Biology
DOI: 10.1111/j.1523-1739.2009.01304.x
Thompson et al.
marginal conditions (i.e., in different or extreme habitats
compared with those of central populations). They are
also often associated with low diversity and increased
differentiation compared with populations at the center
of a species’ range (e.g., Hamrick et al. 1989; Linhart &
Premoli 1994; Lönn & Prentice 2002).
Peripheral populations may also hybridize with related
species because of their location at the limits of a species’
distribution. Such hybridization may cause introgression
of genes from widespread species into populations of narrow endemic species (Brochmann 1984; Rieseberg et al.
1989; Petit et al. 1997), where the numerical disadvantage of the latter may cause local extinction of their populations (Levin et al. 1996; Rhymer & Simberloff 1996).
Where introduced species come into contact with rare
congeners as a result of human activities, such introgression may constitute a major conservation problem (Levin
et al. 1996; Rhymer & Simberloff 1996). Nevertheless,
natural hybridization may enhance genetic variation in
peripheral populations and set the template for adaptive evolution and hybrid speciation, a major evolutionary
force in the diversification of plants (e.g., Stebbins 1950;
Lewontin & Birch 1966; Rieseberg et al. 2003). Hence,
hybridization represents the “double-edged sword of conservation biology” (Haig & Allendorf 2006) and is of
primary importance in distinguishing natural hybridization from anthropogenic hybridization (Allendorf et al.
2001).
Natural hybridization of peripheral populations may
occur in spring flowering Mediterranean Cyclamen
species. This group contains two protected species in
southern France, C. balearicum, a species endemic to
the Balearic Islands and southern France, and C. repandum, which has a more widespread distribution. C.
balearicum may also occur on Corsica, which is part of
the French territory, where it may hybridize with its more
widespread congener (Debussche & Thompson 2000).
Such hybridization could lead to the genetic swamping
of C. balearicum; however, observations of marked floral variability in these sites suggest that the reproductive
interaction could also lead to enhanced diversity. A conservation strategy aimed at conserving evolutionary processes generating diversity and adaptive potential may
thus be more pertinent than species protection.
We sought to determine whether C. balearicum occurs on Corsica and whether there is evidence of hybridization with its more common congener C. repandum. We addressed four questions: Do floral-trait variation and genetic analyses confirm the presence of C.
balearicum on Corsica? Can the two species cross and
produce fertile offspring? Do plants that morphologically resemble C. balearicum on Corsica show significant genetic introgression with the more common C.
repandum? Is genetic introgression associated with enhanced genetic diversity in hybrid sites? We examined
the conservation significance and relevant conservation
237
policy for sites where natural hybridization among rare
and widespread species may occur.
Methods
Study Species
Cyclamen balearicum and C. repandum are two morphologically distinct but closely related species (Affre
& Thompson 1998; Gielly at al. 2001, Debussche &
Thompson 2002). They are thought to have allopatric
distributions. C. balearicum has a disjunct and narrow endemic distribution in the Balearic Islands and
Languedoc–Roussillon (Debussche et al. 1995). C. repandum is widespread across the Mediterranean from the
Tyrrhenian Islands across Italy and the Balkans, southern Greece, and Crete and Rhodes (different subspecies)
and occurs in isolated sites in southern France and Algeria (Debussche & Thompson 2002). C. balearicum is
probably derived from widespread C. repandum at its
western distribution limits (Gielly et al. 2001; Debussche
& Thompson 2002). C. balearicum is protected by
French law throughout its distribution in the Languedoc–
Roussillon region of southern France, but has no listed
status outside this region. Hence, it is not protected on
Corsica. Similarly, the widespread C. repandum is not
protected on Corsica, where it is very common, but is
protected in continental France at the single site where
it is known to occur (Provence region).
The two species differ in leaf design, floral traits, and
breeding system. Whereas C. repandum has large pink
flowers with a stigma protruding beyond the mouth of
the corolla, C. balearicum has small white flowers and a
stigma positioned within the corolla. C. repandum probably has a mixed outcrossing–selfing breeding system
(Affre & Thompson 1997) and is pollinated by bumble
bees on Corsica (L. Affre & J.D.T., unpublished data).
C. balearicum is a highly inbred species throughout its
range (Affre et al. 1997); its autonomous self-pollination
is facilitated by the close proximity of stigmas and anthers (Affre & Thompson 1999). Leaves of C. balearicum
are almost unlobed, grayish green, extensively marbled
with pale gray or silvery patches, whereas leaves of C.
repandum have a well-lobed margin, green overall with
grayish or grayish green hastate patterning.
The two species have little ecological overlap (Debussche & Thompson 2003). Whereas C. repandum is
primarily a species of forest and woodland (deciduous
or evergreen), establishes on a large range of bedrock
types, and occurs in habitats with deep litter rather than
on bare soil, C. balearicum occurs almost exclusively
in evergreen shrublands and open woodlands on rocky
limestone substrates. In three neighboring populations
on Corsica, outside the previously reported range of C.
balearicum, plants with a morphological resemblance to
Conservation Biology
Volume 24, No. 1, 2010
238
C. balearicum are intermingled with plants of C. repandum (Debussche & Thompson 2000). The three populations (hereafter, St. Florent populations) occur near the
town of St. Florent on a single limestone massif of <
5 km. They occur in shrubland and low-evergreen oak,
open woodland on rocky limestone outcrops, where ecological conditions are marginal for C. repandum and typical for C. balearicum habitats (Debussche & Thompson
2003). The sites contain two chloroplast DNA haplotypes
that characterize either C. repandum or C. balearicum
(Gielly et al. 2001).
Frequency of Floral Types on Corsica
We quantified the frequency of pink- and white-flowered
plants in 34 populations on the island of Corsica between
April 1992 and April 2000 by walking through populations and randomly sampling flowers (at least 1 m apart).
Altogether we sampled 8686 plants for flower color. Sample sizes ranged from 78 to approximately 500 (in very
large populations); most populations had 150–400 sampled plants. In the three St. Florent populations, we also
counted the relative number of plants with bicolored
flowers (i.e., corollas either white and pink or pink with
a marked contrast of hues).
Quantitative Floral-Trait Variation
We studied floral-trait variation in three populations of C.
balearicum from Majorca (Balearic Islands), three populations of C. repandum on Corsica with only pink flowers, and eight populations (including the three St. Florent populations) of C. repandum on Corsica with both
pink and white flowers of different frequency. Flower
size (corolla lobe length) and stigma–anther separation
(the difference between style and anther length) are important for taxon delimitation in spring-flowering Cyclamen (Affre & Thompson 1998; Debussche & Thompson
2002). Hence, we randomly sampled pink- and whiteflowered plants and used digital calipers to measure
corolla lobe length, style length, and anther length to
0.1 mm on one fully mature flower per plant.
For the three populations of C. repandum with only
pink flowers (populations 1–3), we measured 25 flowers
per population. The three populations of C. balearicum
(populations 12–14; data collected for a separate study
[Affre & Thompson 1998]) had 20, 18, and 20 flowers
measured per population, respectively. These populations provided a standard for each type of flower color
in populations with mixed pink- and white-colored flowers. For populations with two flower colors, sample sizes
were different because there were fewer white flowers.
The number of pink and white flowers in populations
4–11 were 30 and 5, 30 and 7, 30 and 9, 30 and 12, 30
and 20, 30 and 30, 30 and 41, and 49 and 42, respectively
(populations 9, 10, and 11 were St. Florent populations).
Conservation Biology
Volume 24, No. 1, 2010
Hybridization in Peripheral Populations
The locations of Corsican populations covered the geographic distribution of C. repandum on the island.
We compared floral traits with analysis of variance
(ANOVA) in PROC GLM (SAS 2001). We compared the
three populations of each species to confirm that they
exhibited the significant differences reported for a larger
number of populations (Affre & Thompson 1998). For
each flower color we conducted ANOVA among all
11 populations. We did a priori contrasts between the
three populations with a single flower color and all eight
mixed populations and the three St. Florent populations.
We also conducted a priori contrasts between the three
St. Florent populations and the five other mixed populations. Finally, we analyzed trait variation between flower
colors in each of the eight mixed populations. Because
we made four a priori contrast tests for each flower
color, we corrected the threshold significance level to
p = 0.05/4 = 0.012. We also performed a correlation
analysis to test whether floral traits differed relative to
the frequency of white-flowered plants in the eight mixed
populations.
In the three St. Florent populations, we found white
flowers with either an inserted or an exserted stigma and
bicolored flowers with an exserted stigma. The combination of traits typical of one or other of the two species
motivated us to include these floral types in the genetic
analyses. For bicolored, pink-flowered, and the two types
of white-flowered plants in the three St. Florent populations, we recorded relative number of plants with either a
stigma protruding beyond the corolla (typical of C. repandum) or a stigma inside the corolla close to the anthers
(typical of C. balearicum).
Hand Pollination within and between Species
To examine whether hybridization is possible and as
successful in producing fertile offspring as outcrossing
within species, we quantified seed set on outcross pollination within and among species and pollen viability of F 1 plants produced by crosses between the two
species. Cross-pollinations were performed on each of
the two species in an insect-free glasshouse at the CEFECNRS experimental gardens in Montpellier. The withinspecies crosses were part of a study by Affre and Thompson (1999). Crosses among species were done following
the same procedure, at the same time, and in the same
glasshouse (but not reported in the previously mentioned
study). We isolated plants in glasshouse and emasculated
two flowers on each maternal plant as flowers opened.
One flower on each plant was cross pollinated with a single pollen donor either of the same species or of the other
species by brushing pollen onto the stigma every day
for 3–4 days after flowers opened. For both species we
quantified fruit set and seed number per fruit for withinand between-species pollinations on 25 maternal parents.
We performed a Student t test on the number of seeds
Thompson et al.
produced on within- and between-species crosses for
each species.
We quantified pollen viability in the offspring of interspecific crosses by counting darkly stained pollen grains
in a solution of lactophenol aniline blue. We used a sample of 100 pollen grains taken from different flowers on
each of 10 flowering F 1 plants produced by betweenspecies crosses. Over three years, these plants had been
grown to flowering in pots at the CEFE-CNRS experimental gardens in Montpellier.
Genetic Variation
We compared amplified fragment length polymorphism
(AFLP) marker variation in the three St. Florent populations with that in three populations of each species
sampled either elsewhere on Corsica (C. repandum)
or from continental France and the Balearic islands (C.
balearicum). We sampled eight individuals per population in populations 1–3 of C. repandum on Corsica
(only pink-flowered plants) and population 15 in a habitat typical of C. repandum near the Teghime pass (approximately 10 km from the St. Florent populations),
which had only 0.2% white-flowered plants. Three populations of C. balearicum were sampled from the different
parts of the range of this species: Majorca, Minorca, and
Cévennes hills in southern France. In the St. Florent populations, we sampled eight individuals of each of four morphological types: pink flowers with an exserted stigma
typical of C. repandum, white flowers with a stigma inside the corolla typical of C. balearicum, white flowers
with an exerted stigma, and bicolored flowers with an
exerted stigma. We sampled leaves from 152 plants and
stored them at room temperature in silica gel. We compared between floral types and did not interpret differences among populations for a given floral type, other
than the global test among St. Florent populations and
other populations.
After extraction of total DNA, we conducted the AFLP
protocol (following Gaudeul et al. 2000). Amplified polymerase chain reaction (PCR) products were subjected
to electrophoresis on an ABI PRISM 3100 automated
sequencer (Applied Biosystems, Foster City, California).
Preliminary tests were performed to choose three primer
pairs on the basis of clarity and polymorphism in the AFLP
patterns. We used GeneScan Analysis 3.1 to score polymorphic AFLP markers between 50 and 500 bp long as
present or absent.
To examine whether AFLP markers discriminate between parental species and putative hybrid populations,
we used Jaccard distances between all pairs of samples in
a principal coordinates analysis in NTSYS (Rohlf 1990).
Samples were clustered with the Bayesian algorithm of
Structure (Pritchard et al. 2000). We computed five runs
(each with 50,000 burn-in runs followed by 200,000 iterations) for each number of clusters (K) ranging from
239
1 to 10. We assumed a mixed-ancestry model with correlated allele frequencies and used no prior population
information. For each run, Structure provides the probability of the data and the proportion of the genome of
each individual originating from each cluster (q i ’s). We
assigned an individual to a given cluster when the corresponding q i ≥ 0.90 (Vähä & Primmer 2006). In other
cases, individuals were considered admixed.
In addition, we computed a maximum-likelihood hybrid index (h) for each individual with HINDEX (Buerkle
2005) on the basis of allele frequencies in C. balearicum
and C. repandum populations. This index is an estimate
of the proportion of alleles inherited from each of the two
parental species and ranges from zero (individuals of C.
balearicum) to one (individuals of C. repandum). Values
around 0.5 (e.g., 0.4 ≤ h ≤ 0.6) suggest individuals are F 1
hybrids, whereas values < 0.4 or > 0.6 are indicative of
F 2 backcrosses with parental species or later-generation
hybrids.
Finally, we used AFLPSurv (Vekemans 2002) to compute diversity indices (percentage of polymorphic markers and Nei heterozygosity) for each morph within each
population (n = 8 samples in all cases). To compute allele
frequencies, we used a Bayesian method with nonuniform prior that assumed Hardy-Weinberg equilibrium
within populations. We performed two t tests to assess
differences in genetic diversity among plants of each
species and among plants in the St. Florent populations
(n = 24 samples in all cases except for C. repandum for
which n = 32).
Results
Frequency of Floral Types on Corsica
Of the 34 surveyed populations, 21 had only pinkflowered plants, five had <1% white-flowered plants, four
had between 1% and 5% white-flowered plants, and four
had from 9% to 16.5% white-flowered plants. The three
St. Florent populations (populations 9–11) were the only
populations with bicolored flowers and had the highest
frequency of white flowers (9.6%, 14.6%, and 16.5%). In
these populations, the percentage of bicolored flowers
was 10%, 2%, and 5.5%, respectively, and the percentage
of plants with a floral color typical of C. repandum was
thus 80.4%, 83.4%, and 78%, respectively. Nearly all bicolored flowers (95%, 93%, and 94%, respectively) and pink
flowers (98%, 100%, and 96%, respectively) had a protruding stigma, whereas the proportion of white flowers
with a stigma within the corolla was always < 50% of
the white flowers in a population (44%, 42%, and 33%,
respectively).
Quantitative Floral-Trait Variation
The ANOVA confirmed that C. repandum had larger
flowers (F 1,4 = 60.1, p < 0.001) and stigma–anther
Conservation Biology
Volume 24, No. 1, 2010
Hybridization in Peripheral Populations
240
(a)
Petal length (mm)
24
20
*
16
12
8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Stigma-anther separation (mm)
(b)
3
2.5
2
* *
*
1.5
1
0.5
the 11 populations. There also were significant contrasts
in petal length and stigma–anther separation between
populations with a single flower color and all mixed populations (F 1,213 = 89.4, p < 0.001 and F 1,213 = 158, p <
0.012); the mixed populations not from St. Florent
(F 1,213 = 100, p < 0.001 and F 1,213 = 231, p < 0.001);
and populations from St. Florent (F 1,213 = 26.4, p <
0.001 and F 1,213 = 34.1, p < 0.012). These characteristics also differed significantly between mixed
populations not from St. Florent and populations from
St. Florent (F 1,213 = 43.2, p < 0.001 and F 1,213 = 209,
p < 0.001).
White flowers were significantly smaller than pink
flowers (F 7,89 = 11.7, p < 0.01) in population 11.
Stigma–anther separation was significantly less in whiteflowered plants in all St. Florent populations (population
9: F 1,69 = 47.8, p < 0.001; population 10: F 1,58 = 42.3, p <
0.001; population 11: F 1,89 = 92.6, p < 0.001), but not in
populations from elsewhere (Fig. 1). For white-flowered
plants, petal length and stigma–anther separation were
correlated significantly and negatively with the frequency
of white-flowered plants in a population (Fig. 2).
Hand Pollinations within and between Species
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Population
Figure 1. Petal length and stigma–anther separation
of C. repandum populations (only pink flowers,
populations 1–3; both pink and white flowers, populations 4–11; both pink and white flowers, populations 9–11 at St. Florent) and C. balearicum
populations (populations 12–14). Asterisks mark
populations with significant differences (p < 0.05)
between flower colors.
separation (F 1,4 = 60.5, p < 0.001) than C. balearicum.
Pink-flowered plants (Fig. 1) differed significantly in
petal length (F 10,323 = 9.42, p < 0.001) and stigma–
anther separation (F 10,323 = 12.51, p < 0.001) among
the 11 populations. There was a significant contrast
between populations with a single flower color and
all mixed populations (F 1,323 = 8.2, p < 0.012 and
F 1,323 = 9.1, p < 0.012, respectively) and St. Florent populations (F 1,323 = 41.4, p < 0.001 and F 1,323 = 34.1, p <
0.001). Although there was a significant contrast between
the St. Florent populations and five other mixed populations (F 1,323 = 56.2, p < 0.001 and F 1,323 = 37.5, p >
0.001), no significant contrast between populations with
a single flower color and the five non-St. Florent mixed
populations (F 1,323 = 0.02, p > 0.05 and F 1,323 = 0.53, p >
0.05) was detected.
For white-flowered plants, petal length (F 10,213 =
11.01, p < 0.001) and stigma–anther separation
(F 10,213 = 29.9, p < 0.001) differed significantly among
Conservation Biology
Volume 24, No. 1, 2010
Neither species showed a significant decline in seed set
on maternal plants subject to either within species or
between species outcrossing (C. balearicum, Student
t test = 2.09, df = 41, p = 0.043; C. repandum, t = 1.15,
df = 35, p = 0.26). Over 65% of pollen produced by
the 10 hybrid plants obtained from interspecific crosspollination was viable. Of these plants, three had flowers
typical of C. balearicum, two had flowers typical of C.
repandum, one had large white flowers with an exserted
stigma, one had large pale pink flowers with an exserted
stigma and three had bicolored corollas intermediate in
size between the two parents with an exserted stigma.
Genetic Variation
Three pairs of primers (EcoR1.AGT/Mse1.CAA, EcoR1.
AGT/Mse1.CTT, and EcoR1.AGT/Mse1.CTA) produced
34, 45, and 43 polymorphic markers, respectively, for
a total of 122 polymorphic markers. The AFLP patterns
for C. repandum and C. balearicum aligned with no difficulty, which suggests the two species are closely related.
Principal coordinates analyses (with NTSYS) showed a
distinct separation between the two species and a range
of gradation from C. repandum toward C. balearicum
on the two principal axes that explained 17% of the total
variation (Fig. 3). This separation was also clear from the
distribution of marker frequencies in the C. repandum,
C. balearicum, and St. Florent populations. Although C.
repandum and C. balearicum showed only one diagnostic marker each (marker that is fixed in a species
and absent in another), 16 markers displayed an allele
frequency difference > 0.80 between the two species. In
Thompson et al.
(a)
24
Petal length (mm)
241
20
20
16
16
12
12
R = -0.79 **
8
8
0
Stigma-anther separation (mm)
(b)
24
R = -0.72 ns
5
10
15
(c)
3
5
10
(d)
3
R = -0.64 ns
2.5
0
20
2
2
1.5
1
1
0.5
0.5
0
20
R = -0.88 **
2.5
1.5
15
0
0
5
10
15
20
0
5
White-flowered plants in a population (%)
all cases, these markers occurred at intermediate frequencies in the St. Florent populations, but with frequencies
closer to those of C. repandum than to C. balearicum in
14 cases out of 16.
The percentage of polymorphic loci and heterozygosity were significantly higher (t tests: p = 0.023 and
p = 0.0017, respectively) in the St. Florent populations than in C. balearicum or C. repandum (Fig. 4).
We also observed four extremely rare markers in C.
10
15
20
Figure 2. Relation between floral
traits in (a & c) pink-flowered
and (b & d) white-flowered plants
and the frequency of
white-flowered plants in the eight
mixed populations (populations
4–11) on Corsica.
repandum and C. balearicum populations, which were
not present in more than two individuals of either C.
balearicum (n = 24) or C. repandum (n = 32) and
which were more frequent in the St. Florent populations
(at least 2, 5, 6, and 16 occurrences in each sample of
32 plants). Similarly, the frequencies of nine markers
in the St. Florent populations differed markedly from
the frequencies observed in both C. repandum and
C. balearicum.
Axis 2
Axis 1
Figure 3. Principal coordinate analysis based on amplified fragment length polymorphism data: black symbols, C.
balearicum (diamonds, n = 24) and C. repandum (squares, n = 32); other symbols, plants from the three St. Florent
populations (open triangles, bicolored flowers [n = 24]; open diamonds, white-flowered plants such as C.
balearicum but with an exserted stigma [n = 24]; open circles, white flowers typical of C. balearicum [n = 24]; plus
sign, pink flowers typical of C. repandum [n = 24]).
Conservation Biology
Volume 24, No. 1, 2010
Hybridization in Peripheral Populations
Mean (SE) expected heterozygosity
242
0.4
(a)
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
B
BIC
BSE
BSI
RSE
R
BIC
BSE
BSI
RSE
R
(b)
100
Polymorphic loci (%)
90
80
70
60
50
40
30
20
10
0
B
Figure 4. (a) Mean expected heterozygosity and (b)
percent polymorphic loci derived from AFLP analyses
for C. balearicum (B) and C. repandum (R) and in
plants from the three St. Florent populations (BIC,
bicolored flowers; BSE, white flowers such as for C.
balearicum but with an exserted stigma; BSI, white
flowers typical of C. balearicum; RSE, pink flowers
typical of C. repandum). In all cases, n = 24 except for
C. repandum for which n = 32.
Using the bayesian algorithm of Structure, the probability of the data increased with increasing K values.
Nevertheless, the relative gain of probability was greatest between K = 1 and K = 2; therefore, we inferred
K = 2 was the most adequate number of clusters (Evanno
et al. 2005). Assuming K = 2, all C. balearicum samples
except one (out of 24) were unambiguously assigned to
a single cluster, whereas almost all other samples were
grouped together in a second cluster (Fig. 5). Nineteen
samples were admixed, but had predominant assignation
(0.58 ≤ q i < 0.90) to the C. repandum cluster in all cases.
With K = 3, one cluster (in white, Fig. 5) unambiguously
included all 24 C. balearicum samples except for two
that were slightly admixed with the third cluster. A second cluster (in black, Fig. 5) was composed of 14 (out of
32) C. repandum samples, and the third cluster (in grey,
Fig. 5) regrouped 54 (out of 96) St. Florent samples. The
remaining C. repandum and St. Florent samples were admixed, but with predominant assignation to the second
(black, Fig. 5) cluster for C. repandum (12 out of 18) and
Conservation Biology
Volume 24, No. 1, 2010
to the third (gray) cluster for St. Florent samples (35 out
of 42).
On average, 97% of the genome of C. balearicum individuals originated from the white cluster, 74% of the
genome of C. repandum individuals originated from the
black cluster and 25% from the gray cluster, and more
than 75% of the genome of St. Florent individuals originated from the gray cluster (89%, 84%, and 77% for populations 9, 10, and 11, respectively). There was no clear
difference in the proportion of assignation to clusters and
admixture levels among the three populations or among
the morphs.
All C. balearicum and C. repandum samples had hybrid values close to zero and one respectively, with only
four samples of C. repandum with 0.85 ≤ h ≤ 0.90
(Fig. 6). In the St. Florent populations, all hybrid values
were between 0.60 and 0.90, except for nine (out of 96)
individuals with 0.44 ≤ h ≤ 0.60 and two individuals with
h ≥ 0.90. We did not detect significant differences in hybrid indices among populations (mean h = 0.75, 0.71, and
0.75 in populations 9, 10, and 11, respectively) or phenotypes (mean h = 0.68, 0.73, 0.78, and 0.76 for bicolored
flowers with an exerted stigma, white flowers with an
exerted stigma, white flowers with a stigma inside the
corolla typical of C. balearicum, and pink flowers with
an exserted stigma typical of C. repandum, respectively).
The overall distribution of the h values was trimodal and
suggested that plants in the St. Florent populations were
either the result of back crosses between hybrids and C.
repandum or later-generation hybrids.
Discussion
Our results bring to the fore the importance of strategies and policy for the conservation of hybrid plants.
Our data confirm the historical presence of peripheral
populations of Cyclamen balearicum on a limestone
massif on Corsica (St. Florent populations), where the
species has hybridized with ecologically marginal populations of widespread C. repandum to produce a range of
introgressed forms and a high level of genetic diversity.
The two species flower simultaneously, and we found
that the two species can hybridize to produce fertile
offspring.
Allendorf et al. (2001) provided a framework to categorize the different situations under which hybridization
and introgression occur and the factors to consider in assessment of the conservation value of hybrids. A primary
distinction is made between hybrids that have a natural or
anthropogenic origin because hybridization among introduced and native species may lead to genetic swamping
in the latter case. For example, in western North America,
westslope cutthroat trout (Oncorhynchus clarki lewisi)
hybridize with introduced rainbow trout (O. mykiss),
Thompson et al.
243
Proportion
(a) K = 2
1.0
0.5
0
C. repandum
St. Florent populations
C. balearicum
St. Florent populations
C. balearicum
Proportion
(b) K = 3
1.0
0.5
0
C. repandum
Figure 5. Population structure of the three groups of Cyclamen populations on Corsica as estimated by the
Structure analysis. Each individual is represented by a thin vertical line, which is partitioned into K differently
shaded segments that represent the individual’s estimated membership fractions in each cluster. Individuals of St.
Florent populations are presented by four morphs in the following order (left to right): bicolored flowers, white
flowers such as C. balearicum but with an exserted stigma, white flowers typical of C. balearicum, and pink flowers
typical of C. repandum.
and Allendorf et al. (2004) recommend that only nonhybridizing populations be included in the unit listed under the U.S. Endangered Species Act because only they
are likely to contain the local adaptations necessary for
long-term persistence. On the basis of examples of endemic species subject to hybridization and introgression
with more widespread congeners, Levin et al. (1996) con-
clude that this process is detrimental and that “isolation
from cross-compatible congeners should be a key goal
in rare plant conservation programs.” Thus, conservation policy should reduce the impacts of anthropogenic
hybridization on biodiversity loss (Rhymer & Simberloff
1996). Nevertheless, in cases of complete admixture,
there may be no other solution than protection of the
1.0
0.9
0.8
Hybrid index value
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
C. repandum
C. balearicum
BIC
BSE
BSI
RSE
Figure 6. Maximum-likelihood hybrid index values (h) for individuals (each histogram) of C. balearicum, C.
repandum, and of the four floral phenotypes in the St. Florent populations (BIC, bicolored flowers; BSE, white
flowers such as C. balearicum but with an exserted stigma; BSI, white flowers typical of C. balearicum; RSE, pink
flowers typical of C. repandum). Within each morphology, individuals are ordered by ascending h value. Vertical
bars represent the upper and lower bounds of each estimated value.
Conservation Biology
Volume 24, No. 1, 2010
Hybridization in Peripheral Populations
244
hybrids (Allendorf et al. 2001). Conservation policy for
hybrids thus has to be flexible.
Although we detected admixture in the St. Florent populations, our study provides an example of natural hybridization (Allendorf et al. 2001) in a newly discovered
part of the range of a species that is listed in France.
In the Mediterranean flora, there is strong evidence that
climate changes and their impact on distribution have
at various times created the opportunity for contact and
hybrid speciation (Thompson 2005). In Mediterranean
orchids, for example, such processes are clearly ongoing. Sympatric zones of congeneric orchids are frequent
and represent sites where changes in the action of natural
selection can be observed (Cozzolino et al. 2006). These
changes may promote local genetic adaptation; hence,
sites of hybridization, rather than being a threat to orchid biodiversity, merit conservation status. Our results
provide an additional example of this and illustrate some
important issues associated with conservation policy for
natural hybrids.
The first issue concerns the phenotypic and genetic diversity of the hybrid population. The genetic variability of
the St. Florent populations was enhanced relative to the
parental species (presence of rare markers and enhanced
heterozygosity); thus, these populations merit attention
for conservation purposes (Allendorf et al. 2001). In addition, the floral variability detected in the St. Florent
populations covered the range of floral types in the different subspecies of C. repandum that occur on Corsica,
other parts of the western Mediterranean, the Peloponnese, and on Crete and Rhodes (Debussche & Thompson
2000, 2002). Thus, we contend there is a need for a conservation strategy for the hybrid populations, at least until
there is more information on their ecological persistence
and adaptive capacity. In addition, gene frequencies in
the St. Florent populations differed markedly from those
in C. repandum and C. balearicum, which is indicative of
divergent evolution relative to the ancestral parental populations. This divergence may be due to prior divergence
of disjunct populations of C. balearicum on Corsica and
retention of this differentiation in hybrids. The St. Florent
populations occur on rocky limestone in shrubland and
low, open evergreen oak woodland, a habitat typical of C.
balearicum (Debussche & Thompson 2003), whereas all
other populations of C. repandum in the vicinity of these
populations occur on intrusive or metamorphic rocks in
woodlands with deep nonbasic humifer soils. This ecological segregation may enhance the differentiation of
hybrid populations.
Analysis of the “hybrid” populations revealed that although introgression probably occurs in both directions,
the species that is rare on Corsica, C. balearicum, may
incur more introgression from C. repandum. There is
thus a possibility of complete genetic swamping of this
species in this part of its disjunct range. This is not surprising because C. repandum produces more pollen (Affre &
Conservation Biology
Volume 24, No. 1, 2010
Thompson 1998) and is more abundant in and around the
three sites containing hybrid populations. What is particularly important is that the floral phenotype in the St. Florent populations that resembles parental C. balearicum
also shows evidence of high genetic introgression, which
is evidence that selection may be acting to maintain the
ancestral floral phenotype, perhaps as a result of selection for autonomous self-pollination (Affre & Thompson
1999).
A final point is whether hybridization poses a threat
to other “pure” populations. The greater the threat,
the lower the conservation value of hybrid populations.
Linked to this concept is the number and location of
pure populations that remain. For C. balearicum, the geographic extent of hybridization and introgression is likely
to remain limited because hybridization does not occur
in protected populations of C. balearicum in southern
continental France, where the closest C. repandum population is approximately 200 km to the east. The maintenance of C. balearicum as a listed species for protection
in continental France is thus not problematic.
We conclude that a strategy aimed at conserving the
evolutionary process of hybridization requires elaboration (Ennos et al. 2005). In some situations it may be
appropriate to protect hybrid populations either by their
inclusion as part of the taxonomic unit eligible for protected status (Allendorf et al. 2001) or by selecting sites
where they occur for habitat protection (Thompson
2005). This could be achieved through acquisition of the
land for designation as a reserve or through a contractual
approach with landowners. A strategy that opts for the
protection of habitats where adaptive variation and new
species are evolving due to natural hybridization is a priority as the 21st century plunges biodiversity into an era
of rapid extinction.
Acknowledgments
We are grateful to Y. Linhart, L. Affre, G. Debussche,
and M.-A. Thompson for help with data collection and L.
Gielly for help with AFLP work. The Centre National de
la Recherche Scientifique provided financial support.
Literature Cited
Affre, L., and J. D. Thompson. 1997. Variation in the population genetic
structure of two Cyclamen species on the island of Corsica. Heredity
78:205–214.
Affre, L., and J. D. Thompson. 1998. Floral trait variation in four
Cyclamen (Primulaceae) species. Plant Systematics and Evolution
212:279–293.
Affre, L., and J. D. Thompson. 1999. Variation in self-fertility, inbreeding depression and levels of inbreeding in four Cyclamen species.
Journal of Evolutionary Biology 12:113–122.
Affre, L., J. D. Thompson, and M. Debussche. 1997. Genetic structure
of continental and island populations of the Mediterranean endemic
Cyclamen balearicum (Primulaceae). American Journal of Botany
84:437–451.
Thompson et al.
Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg. 2001. The
problems with hybrids: setting conservation guidelines. Trends in
Ecology & Evolution 16:613–622.
Allendorf, F. W., R. F. Leary, N. P. Hitt, K. Knudsen, L. L. Lundquist, and
P. Spruell. 2004. Intercrosses and the U.S. Endangered Species Act:
should hybridized populations be included as westslope cutthroat
trout? Conservation Biology 18:1203–1213.
Bantock, C. R., and D. J. Price. 1974. Marginal populations of Cepaea
nemoralis (L.) on the Brendon Hills, England. I. Ecology and ecogenetics. Evolution 29:267–277.
Brochmann, C. 1984. Hybridization and distribution of Argyranthemum coronopifolium (Asteraceae—Anthemideae) in the Canary
Islands. Nordic Journal of Botany 4:729–736.
Buerkle C. A. 2005. Maximum-likelihood estimation of a hybrid index based on molecular markers. Molecular Ecology Notes 5:684–
687.
Cozzolino, S., A. M. Nardella, S. Impagliazzo, A. Widmer, and C. Lexer.
2006. Hybridization and conservation of Mediterranean orchids:
should we protect the orchid hybrids or the orchid hybrid zones?
Biological Conservation 129:14–23.
Debussche, M., and J. D. Thompson. 2000. Les populations à fleurs
blanches de Cyclamen repandum Sibth. & Sm. en Corse. Bulletin
de la Société Botanique du Centre-Ouest 31:1–8.
Debussche, M., and J. D. Thompson. 2002. Morphological differentiation among closely related species with disjunct distributions: a case
study of Mediterranean Cyclamen L. subgen. Psilanthum Schwarz
(Primulaceae). Botanical Journal of the Linnaean Society 132:133–
144.
Debussche, M., and J. D. Thompson. 2003. Habitat differentiation between two closely related Mediterranean plant species, the endemic
Cyclamen balearicum and the widespread C. repandum. Acta Oecologica 24:35–45.
Debussche, M., G. Debussche, and L. Affre. 1995. La distribution fragmentée de Cyclamen balearicum Willk. en France: analyse historique et conséquence des activités humaines. Acta Botanica Gallica 142:439–450.
Ennos, R. A., G. French, and P. Hollingsworth. 2005. Conserving taxonomic complexity. Trends in Ecology & Evolution 20:164–168.
Evanno, G., S. Regnaut, and J. Goudet. 2005. Detecting the number of
clusters of individuals using the software STRUCTURE: a simulation
study. Molecular Ecology 14:2611–2620.
Gaudeul, M., P. Taberlet, and I. Till-Bottraud. 2000. Genetic diversity
in an endangered alpine plant, Eryngium alpinum L. (Apiaceae),
inferred from AFLP markers. Molecular Ecology 9:1625–1637.
Gielly, L., M. Debussche, and J. D. Thompson. 2001. Geographic isolation and evolution of endemic Cyclamen in the Mediterranean: insights from chloroplast trnL (UAA) intron sequence variation. Plant
Systematics and Evolution 230:75–88.
Haig, S. M., and F. W. Allendorf. 2006. Hybrids and policy. Pages 150–
163 in J. M. Scott, D. D. Goble, and F. W. Davis, editors. The en-
245
dangered species act at thirty. Volume 2: conserving biodiversity in
human-dominated landscapes. Island Press, Washington, D.C.
Hamrick, J. L., H. M. Blanton, and K. J. Hamrick. 1989. Genetic structure
of geographically marginal populations of Ponderosa pine. American Journal of Botany 76:1559–1568.
Levin, D. A. 1970. Developmental instability and evolution in peripheral
isolates. The American Naturalist 104:343–353.
Levin, D. A., J. Francisco-Ortega, and R. K. Jansen. 1996. Hybridization and the extinction of rare plant species. Conservation Biology
10:10–16.
Lewontin, R. C., and R. C. Birch. 1966. Hybridization as a source of
variation for adaptation to new environments. Evolution 20:315–
336.
Linhart, Y. B., and A. C. Premoli. 1994. Genetic variation in central and
disjunct populations of Lilium parryi. Canadian Journal of Botany
72:79–85.
Lönn, M., and H. C. Prentice. 2002. Gene diversity and demographic
turnover in central and peripheral populations of the perennial herb
Gypsophila fastigiata. Oikos 99:489–498.
Petit, C., P. Lesbros, X. Ge, and J. D. Thompson. 1997. Variation in
flowering phenology and outcrossing rate across a contact zone
between diploid and tetraploid Arrhenatherum elatius (Poaceae).
Heredity 79:31–40.
Pritchard, J. K, M. Stephens, and P. J. Donnelly. 2000. Inference of
population structure using mutilocus genotype data. Genetics 155:
945–959.
Rhymer, J. M., and D. Simberloff. 1996. Extinction by hybridization and
introgression. Annual Reviews of Ecology and Systematics 27:83–
109.
Rieseberg, L. H., S. Zona, L. Aberbom, and T. Martin. 1989. Hybridization
in the island endemic, Catalina mahogany. Conservation Biology
3:52–58.
Rieseberg, L. H., O. Raymond, D. M. Rosenthal, Z. Lai, K. Livingstone,
T. Nakazato, J. L. Durphy, A. E. Schwarbach, L. A. Donovan, and C.
Lexer. 2003. Major ecological transitions in wild sunflowers facilitated by hybridization. Science 301:1211–1216.
Rohlf, F. 1990. NTSYS-PC. Numerical taxonomy and multivariate analysis system. Version 2.02. Exeter Software, Setauket, New York.
SAS. 2001. SAS/STAT Users guide. SAS Institute, Inc., Cary, North Carolina.
Stebbins, G. L. 1950. Variation and evolution in plants. Columbia University Press, New York.
Thompson, J. D. 2005. Plant evolution in the Mediterranean. Oxford
University Press, Oxford, United Kingdom.
Vähä, J.-P., and C. R. Primmer. 2006. Efficiency of model-based Bayesian
methods for detecting hybrid individuals under different hybridization scenarios and with different numbers of loci. Molecular Ecology
15:63–72.
Vekemans, X. 2002. AFLP-Surv. Version 1.0. Laboratoire de Génétique
et Ecologie Végétale, Université Libre de Bruxelles, Belgium.
Conservation Biology
Volume 24, No. 1, 2010