Genetic structure of the poplar rust fungus Melampsora larici-populina:
Evidence for isolation by distance in Europe and recent
founder effects overseas
Benoı ˆt Barre `sa,1, Fabien Halketta, Cyril Dutechb, Axelle Andrieuxa, Jean Pinona, Pascal Freya,*
aINRA, Nancy-Universite ´, UMR1136, Interactions Arbres-Microorganismes, IFR 110, F-54280 Champenoux, France
bINRA, Universite ´ Bordeaux I, UMR1202 BioGeCo, F-33883 Villenave d’Ornon, France
Dispersal is a broad source of genetic variation in plant
pathogen populations. It leads to the foundation of new popula-
tions and can havea strong and rapid impact on populationgenetic
structure, hence influencing evolutionary processes. The aerial
spread of plant pathogens can occur in one of two dispersal modes,
each with very different epidemiological and evolutionary con-
sequences (Brown and Hovmøller, 2002). The first, often regarded
as the natural process of dispersal, results from the gradual spread
of the disease from an original source of inoculum. Often wind-
mediated, this gradual dispersal process is characterized by a rapid
decrease of the probability of dispersal with distance to the source.
It leads to a particular spatial genetic structure, called the isolation
by distance (IBD) pattern. Indirect estimates of dispersal, based on
1997; Smouse and Peakall, 1999; Hardy and Vekemans, 1999;
reviewed in Fenster et al., 2003), namely by testing the null
hypothesis of increasing population genetic differentiation with
geographic distance (Epperson et al., 1999). The second dispersal
mode involves the transport of spores over very long distances
(even between continents), often in a single step. These long
distance dispersal (LDD) events could result in devastating disease
outbreaks (Aylor, 2003). LDD events have had drastic conse-
quences for human well-beings because of the worldwide spread
of plant diseases (Brown and Hovmøller, 2002) such as the potato
late blight, caused by Phytophthora infestans (Fry et al., 1992). LDD
events are rare and highly stochastic, which makes them difficult
Infection, Genetics and Evolution 8 (2008) 577–587
A R T I C L EI N F O
Received 26 October 2007
Received in revised form 9 April 2008
Accepted 11 April 2008
Available online 20 April 2008
Isolation by distance
Long distance dispersal
A B S T R A C T
Dispersal has a great impact on the genetic structure of populations, but remains difficult to estimate by
direct measures. In particular, gradual and stochastic dispersal are often difficult to assess and to
distinguish, although they have different evolutionary consequences. Plant pathogens, especially rust
fungi, are suspected to display both dispersal modes, though on different spatial scales. In this study, we
inferred dispersal capacities of the poplar rust fungus Melampsora larici-populina by examining the
genetic diversity and structure of 13 populations from eight European and two overseas countries in the
Northern hemisphere. M. larici-populina was sampled from both cultivated hybrid poplars and on the
wild host, Populus nigra. The populations were analyzed with 11 microsatellite and 8 virulence markers.
Although isolates displayed different virulence profiles according to the host plant, neutral markers
revealed little population differentiation with respect to the type of host. This suggests an absence of
reproductive isolation between populations sampled from cultivated and wild poplars. Conversely,
studying the relationship between geographic and genetic structure allowed us to distinguish between
isolation by distance (IBD) patterns and long distance dispersal (LDD) events. The European populations
exhibited a significant IBD pattern, suggesting a regular and gradual dispersal of the pathogen over this
spatial scale. Nonetheless, the genetic differentiation between these populations was low, suggesting an
important gene flow on a continental scale. The two overseas populations from Iceland and Canada were
shown to result from rare LDD events, and exhibited signatures of strong founder effects. Furthermore,
the high genetic differentiation between both populations suggested that these two recent introductions
were independent. This study illustrated how the proper use of population genetics methods can enable
contrasted dispersal modes to be revealed.
? 2008 Elsevier B.V. All rights reserved.
* Corresponding author. Tel.: +33 383 394056; fax: +33 383 394069.
E-mail address: firstname.lastname@example.org (P. Frey).
1Present address: CIRAD, UMR 54 BGPI, CIRAD-INRA-SupAgro, TA A-54/K,
Campus International de Baillarguet, 34398 Montpellier, France.
Contents lists available at ScienceDirect
Infection, Genetics and Evolution
journal homepage: www.elsevier.com/locate/meegid
1567-1348/$ – see front matter ? 2008 Elsevier B.V. All rights reserved.
to study (Nathan et al., 2003). In this attempt, it is of primary
importance to disentangle LDD events that result in the foundation
of new populations,from anunderlying IBD pattern. In the absence
of selection, departure from the IBD pattern may indicate
stochastic dispersal processes resulting either in a panmictic unit,
where intense random gene flow tends to erase population
differentiation (Zeller et al., 2004), or in distinct populations
founded by LDD events. In the latter case, strong founder effects
can be expected (Slatkin, 1977), leading to larger population
differentiation, reduced genetic diversity, and genetic disequili-
brium. This is exemplified in populations of the banana leaf streak
fungus, Mycosphaerella fijiensis, which are located out of its original
distribution area (Carlier et al., 1996; Rivas et al., 2004). To our
knowledge, no population genetics analysis has yet assessed the
extent of gradual vs. stochastic dispersal processes in a plant
Rust fungiare ideal models to tackle thesequestions, because in
most species both gradual and stochastic dispersal processes likely
occur (Nagarajan and Singh, 1990; Brown and Hovmøller, 2002).
Indeed, rust fungi aremore likely thanother pathogensto be wind-
dispersed over very long distances because their spores are
comparatively resistant to environmental damage (Rotem et al.,
1985). The Eurasian poplar rust fungus Melampsora larici-populina
causes severe damage and economic losses in poplar cultivation
(Frey et al., 2005). This heteroecious macrocyclic rust fungus,
alternating on larches (Larix spp.), produces five spore stages
during its life cycle, three of them (basidiospores, aeciospores and
urediniospores) being wind-dispersed (Frey and Pinon, 2004). The
high dispersal capacities of the asexual spores (urediniospores) are
illustrated by the spread of a new virulent lineage, which covered
Western Europe in less than 5 years (Pinon and Frey, 2005).
Previous studies of M. larici-populina populations using neutral
markers showed frequent occurrence of recombination (Ge ´rard
et al., 2006), but did not demonstrate a relationship between
geographic and genetic distances, either in France (Ge ´rard et al.,
due to the insufficient scale of sampling compared to the dispersal
potential of this pathogen.
M. larici-populina is native to Eurasia and its distribution range
is supposed to encompass that of its natural host Populus nigra. In
the past century, the distribution range of the rust has expanded
greatly, now covering most poplar-growing regions worldwide.
Outbreaks of M. larici-populina on other continents were reported
from South America (Spegazzini, 1918; Fresa, 1936; Kern and
Thurston, 1954), Southern Africa (Lloyd, 1971; Gibson and Waller,
1972), and Australia (Walker et al., 1974). In 1973, M. larici-
populina was introduced to New Zealand from Australia, likely via
trans-Tasman air-currents over a 2000 km distance (Wilkinson
and Spiers, 1976). In North America, M. larici-populina was first
detected in 1991 in the US Pacific Northwest (Washington State
and Oregon) (Newcombe and Chastagner, 1993), and subsequently
in California (Pinon et al., 1994) and in Eastern Canada in 2002
(Innes et al., 2004). In 1999, M. larici-populina was also discovered
in Southern Iceland, infecting Populus trichocarpa clones originally
from Alaska (H. Sverrisson, personal communication).
In some areas, where the alternate host (Larix sp.) is absent, M.
larici-populina can survive only asexually (Walker et al., 1974). In
contrast, if larch is present, the fungus can reproduce sexually by
producing aeciospores on larch needles. For instance, following its
outbreaks on poplars, aeciospores of M. larici-populina were found
on larch needles (Larix spp.) in New Zealand (Wilkinson and Spiers,
1976), Canada (Grondin et al., 2005) and Iceland (H. Sverrisson,
personal communication). This implies that M. larici-populina has
completed its life cycle, proceeded to sexual recombination, and
has potentially established durably in these new areas. There is
actually no evidence that the recent populations still display the
hallmark of founder effects, such as a larger population differ-
entiation, a reduced genetic diversity and a genetic disequilibrium.
In the present study, our aim was to apply population genetics
methods to decipher the population structure of M. larici-populina
on both continental and global scales. The first goal was to use
recently developed microsatellite markers (Barre `s et al., 2006) to
characterize the genetic diversity and structure of M. larici-
populina populations on the European continental scale. Because
sampling both wild and cultivated poplars could bias the
estimation of the genetic structure by selecting host-adapted
individuals, additional typing for virulence characters was
performed. This allowed for comparison with results obtained
using selected markers (Pinon and Frey, 2005). We paid particular
attention to the relationship between geographic and genetic
distances in order to determine whether or not there was an IBD
pattern. The second aspect of this study was to document the
population genetic structure and characteristics of two recently
founded populations in Canada and Iceland. In particular, we
assessed whether or not (i) they display evidence for founder
effects and (ii) they display an IBD pattern with the European
populations. This sampling design, which includes isolates from
both the pathogen’s native range and from recently colonized
areas, will help assess whether population genetics tools can help
to distinguish between the different modes of dispersal.
2. Materials and methods
2.1. Sampling strategy
Poplar leaves infected with M. larici-populina were collected by
the authors and collaborative researchers in 10 countries of the
Northern hemisphere (Fig. 1, Table 1). Eleven populations were
sampled from European countries: Bosnia and Herzegovina (BIH),
Czech Republic (CZE), France (FRA-1, FRA-2, FRA-3, and FRA-4),
Germany (DEU), Italy (ITA), the Netherlands (NLD), Poland (POL),
and the United Kingdom (GBR). Two overseas populations were
sampled from Iceland (ISL) and Canada (CAN). Samples were
collected in the summer and autumn of 2003, except for FRA-2,
FRA-3 and FRA-4 populations which were sampled in autumn of
2004. In each country, approximately 30–100 rust-infected leaves
were harvested from poplar trees, over a total area ranging from 50
to 400 m2. Whenever possible, leaves were sampled from distinct
poplar trees, in order to minimize the effect of clonality. Seven
populations were sampled from natural P. nigra stands, whereas
the Icelandic population was sampled from P. trichocarpa. The five
remaining populations were sampled from different hybrid
P. ? euramericana
P. ? interamericana
2.2. Pathotype identification
One single uredinium (sporulating lesion producing uredinios-
ofP. ? euramericana‘Robusta’incontrolledconditionsasdescribed
by Ge ´rard et al. (2006). Briefly, a 10-ml droplet of water agar
(0.1 g l?1) was deposited onto each selected uredinium with a
micropipette and urediniospores were dispersed within the
droplet with the micropipette. The resulting spore suspension
was applied as 1-ml droplets on 12-mm-diameter leaf disks of
‘Robusta’. Leaf disks were incubated floating on deionized water in
24-well polycarbonate cell culture plates, abaxial surface upper-
most, at 20 8C under continuous illumination (fluorescent light,
25 mmol s?1m?2). After an incubation period of 8–10 days, the
sporulating disks were harvested and the urediniospores were
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
suspended in 100 ml of water agar (0.1 g l?1). Each mono-uredinial
leaf disks of a differential set containing eight poplar clones with
different qualitative resistance genes and incubated as described
above. This differential set allowed us to detect the presence of
eight virulence factors in the pathogen (Pinon and Frey, 2005). In
this gene-for-gene system (Flor, 1971), a virulence factor is defined
as the ability for the pathogen to cause disease on a given poplar
clone, and a pathotype as a combination of virulence factors. The
poplar differential set contained: P. ? euramericana ‘Ogy’ (Vir1),
P. ? jackii ‘Aurora’ (Vir2), P. ? euramericana ‘Brabantica’ (Vir3),
P. ? interamericana ‘Unal’ (Vir4), P. ? interamericana ‘Rap’ (Vir5), P.
deltoides ‘87B12’ (Vir6), P. ? interamericana ‘Beaupre ´’ (Vir7), and
P. ? interamericana ‘Hoogvorst’ (Vir8). The formation of uredinia
on the leaf disks was checked daily between the 7th and the 14th
day postinoculation, and the resulting pathotype (combination of
virulences) was determined for each isolate.
2.3. DNA analysis
After pathotype identification, approximately 2 mg of uredi-
niospores were collected for each isolate and stored at ?20 8C in
Eppendorf tubes until DNA extraction. DNA was extracted using
DNeasy196 Plant Kit (Qiagen). We followed the Fresh Leaves
protocol (DNeasy196 Plant Handbook, September 2002), except
that samples were disrupted with one tungsten carbide bead and
suspended in 200 ml of extraction buffer during 2 ? 1 min, instead
of 2 ? 1.5 min, at 30 Hz. DNA was eluted in a final volume of
200 ml. Individuals were genotyped using 11 microsatellite
markers (Barre `s et al., 2006). PCRs were performed individually
in a PTC-200 Peltier thermal cycler (MJ Research) using conditions
previously described (Barre `s et al., 2006). An exception was locus
mMLP31, where the PCR mix was modified as follows: 15 ng
template DNA, 2 ml of 10? reaction buffer, 3 mM MgCl2, 0.7 mg/ml
BSA (Sigma), 0.2 mM dNTP, 0.5 U Taq polymerase (Sigma), and
0.2 mM forward and reverse primers in a 20 ml final reaction
volume. To allow size and dye multiplexing, forward primers were
labeled with three different dyes (Proligo): D2 for mMLP13,
mMLP22, mMLP27, and mMLP37; D3 for mMLP20, mMLP28, and
mMLP30; and D4 for mMLP09, mMLP12, mMLP31, and mMLP36.
PCR products were separated, sized, and analyzed on a CEQTM8000
Genetic Analysis System (Beckman Coulter). In order to reduce the
number of analyses, PCR products were pooled in two sets of loci.
Set A was made up of mMLP09, mMLP13, mMLP27, mMLP30, and
mMLP36 loci with volumes of 2, 3, 4, 4, and 4 ml, respectively, in a
34 ml final volume. Set B contained mMLP12, mMLP20, mMLP22,
mMLP28, mMLP31, and mMLP37 loci, with volumes of 3, 4, 7, 4, 2,
and 8 ml, respectively, in a 61 ml final volume. Internal size
Characteristics of the collection sites of M. larici-populina populations
CountryPopulation IDLocationHostLatitude LongitudeCollector
Bosnia and Herzegovina
La Quincy, Picardie
B. Barre `s
Ziltendorfer Niederung, Brandenburg
Warta River, Poznan
Castlearchdale, Northern Ireland
Lotbinie `re, Que ´bec
Fig. 1. Origin of the 13 Melampsora larici-populina populations collected for the study.
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
standards of 400 and 600 bp (Beckman Coulter), labeled with D1
dye, were used to genotype Set A and Set B, respectively, in a
mixture containing 30 ml of Sample Loading Solution (SLS,
Beckman Coulter), 0.5 ml of internal size standard, and 1 ml of
each of the marker sets. When chromatograms were of poor
quality, or when a locus failed to amplify, PCRs of the entire set
were performed again. If the analysis failed again, the individual
was considered as missing data. It should be noted thatmMLP22 is
a mitochondrial microsatellite locus (Barre `s et al., 2006). Each
allele at this locus was therefore considered as a different
mitotype. The loci mMLP12 and mMLP13 are physically linked
but are not redundant (Barre `s et al., 2006). We tried to keep all the
information carried by these two loci by (i) reconstructing the two
haplotypes within an individual using the PHASE program
(Stephens et al., 2001) and (ii) summing both allele sizes, which
compound motif. All further analyses were made using this
compound locus, named mMLP38, instead of the linked locus pair
mMLP12 and mMLP13.
2.4. Data analyses
From pathotype frequency in each population, we calculated an
evenness index (E) and a richness index (Shannon relative index,
HSR) to illustrate both relative abundance of pathotypes and mean
pathotype diversity, respectively.Theevennessindex (E) isderived
from the Simpson’s index corrected for sample size after Fager
(1972). We first calculated the complement of Simpson’s index (D)
for each population as D ¼ 1 ?P
isolates analyzed for virulence in the population. Then we
calculated E, which is not influenced by sample size, as
E = (D ? Dmin)/(Dmax? Dmin)
[Np(Np? 1)] and Dmax= [(np? 1)Np]/[np(Np? 1)], where npis the
number of pathotypes found in the population. The richness index
HSRis derived from the Shannon’s index, corrected for sample size
and calculated as follows: HSR¼ ?P
as the mean number of virulences carried by a single isolate
(Andrivon and de Vallavieille-Pope, 1995).
We subsequently used the eight virulence markers to estimate a
pathotypic distance matrix between populations. Pathotypic
distance was computed using DARWIN 5.0 software (Perrier et al.,
2003). We chose the Manhattan distance, which is commonly used
for continuous variables and is less sensitive to large differences
than is the Euclidian distance. A matrix of pairwise geographical
distances was built using CIRCE´software (http://www.ign.fr). The
correlation between the pathotypic distance matrix and the
geographic/genetic distance matrices was assessed from Mantel
tests with the ZT program (developed by E. Bonnet and Y. Van de
Peer; url: http://www.psb.ugent.be/?erbon/mantel/), assuming
Identical multilocus genotypes were identified using GIMLET
version 1.3.3 (Valie `re, 2002) on a dataset pooling all individuals
from the different populations. Potential insufficient power of our
molecular markers could lead to scoring individuals as having the
to identify multilocus genotypes that are statistically over-
represented (assuming panmixia), and that could thus be
considered as belonging to the same clonal lineage, the method
described by Halkett et al. (2005) was used. The probability of
observing n times a multilocus genotype in a population was
computed using MLGSIM software (Stenberg et al., 2003). Using a
Monte Carlo simulation method, the program determines the
significance threshold for the probability values, indicating the
i½kiðki? 1Þ=NpðNp? 1Þ?, where ki
is the number of isolates of pathotype i and Npis the number of
Dmin= [(np? 1)(2Np? np)]/
We also computed an index of pathotype complexity (Ci), defined
multiple copies of the same multilocus genotype that did not occur
by chance (true clones). Calculation was done for each population,
taking into account sample size and allele frequencies. The
significance level was set to 0.01. Hence, a clone-corrected dataset
was built, keeping a single individual per identical multilocus
genotype. These individuals were considered as clones for each
population. We called ngthe number of multilocus genotypes in a
given population after this clone correction. Relative genotypic
diversity (ng/Ng) was computed for each population in order to
estimate the impact of asexual reproduction, where Ng is the
number of sampled isolates from a given population that were
genotyped. The clone-corrected dataset was used for all further
Genotypic linkage disequilibrium and deviation from Hardy–
Weinberg equilibrium were computed using GENEPOP 3.4
(Raymond and Rousset, 1995) and FSTAT (Goudet, 1995),
respectively. Significant levels were subsequently adjusted using
the sequential Bonferroni correction method (Rice, 1989). Allelic
richness (Ar), gene diversity (HE), and inbreeding coefficients (FIS)
were estimated for each population using FSTAT 22.214.171.124 (Goudet,
A test to detect recent founder effects was conducted using
BOTTLENECK version 1.2 (Piry et al., 1999) on the Canadian and the
Icelandic populations. This test is based on the assumption that
populations that have experienced a recent reduction of their
effective size exhibit a higher reduction of their allele number than
of their gene diversity. The program computes the Wilcoxon’s test
for gene diversity excess after estimating the expected gene
diversity at mutation-drift equilibrium, using three mutation
models: infinite allele model, stepwise mutation model and two-
phase model. Permutation tests were carried out using FSTAT in
order to test whether allelic richness and gene diversity were
significantly different between M. larici-populina populations
collected from their native distribution area (i.e. European
populations) vs. recently founded populations (i.e. Canadian and
Icelandic populations), and also between European populations
sampled from hybrid poplars (FRA-1, FRA-2, FRA-3, GBR) vs.
European populations sampled from P. nigra (BIH, CZE, DEU, FRA-4,
ITA, NLD, POL). In order to determine the putative source
population of Canadian and Icelandic individuals, assignment
tests were performed with a frequencies-based method (Paetkau
et al., 1995) using GENECLASS2 software (Piry et al., 2004). The
probability that an individual belongs to a reference population
was computedusing theresamplingmethoddevelopedbyPaetkau
et al. (2004).
Significance of genotypic differentiation was assessed from
exact tests conducted using GENEPOP 3.4. Pairwise FSTvalues were
estimated using the method of Weir and Cockerham (1984).
Shared allele distances (DAS, Bowcock et al., 1994) were computed
between the 13 populations of M. larici-populina with the 9 nuclear
microsatellite markers using the POPULATION program (O.
Langella, http://bioinformatics.org/project/?group_id=84). Princi-
pal component analyses were performed from pairwise matrix
distances (both genotypic and pathotypic distances) using NUEES
(O. Langella, http://bioinformatics.org/project/?group_id=84).
Spatial analyses were performed using two methods. First, we
tested the hypothesis of isolation by distance by plotting pairwise
FST/(1 ? FST) ratios against log-transformed geographic distances
according to Rousset’s method (1997). Geographic distances were
log-transformed because M. larici-populina populations evolve in a
two-dimensional space. Significance of the correlation between
FST/(1 ? FST) ratios and log-transformed geographic distance
matrices was assessed through a Mantel test using the ISOLD
program, implemented in GENEPOP (Raymond and Rousset, 1995).
Second, we performed an autocorrelation analysis (Smouse and
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
Peakall, 1999). We used GENALEX 6 (Peakall and Smouse, 2005) to
generate a spatial autocorrelogram which figures the decrease in
average genetic correlation (r) between pairs of individuals taken
within a given distance interval as a function of log-transformed
computed, starting from the logarithm of distance equaling five
(approximately 150 km) and with a 0.5 log interval. To test for
statistical significance, we performed 1000 random permutations
of individuals among distance classes, thus estimating the 95%
confidence interval of r about the null hypothesis of no spatial
genetic structure (Smouse and Peakall, 1999).
3.1. Pathotypic diversity
A total of 791 M. larici-populina isolates were analyzed for
virulence (Table 2), among which 57 distinct pathotypes were
found. The evenness index (E) showed a large disparity between
populations: the POL population was characterized by the
predominance of a single pathotype, whereas the BIH, CZE, FRA-
1, FRA-2, FRA-3, ITA and CAN populations displayed balanced
pathotype frequencies. Pathotype richness (HSR) was moderately
high, with the highest value found for the NLD population. Above
all, pathotype complexity (Ci) discriminated between two popula-
tion groups: FRA-1, FRA-2, FRA-3, and GBR were characterized by a
mean number of around four virulences per isolate, and the
remaining populations displayed a mean number of one to two
virulences per isolate (Table 2).
3.2. Genetic diversity and equilibrium tests
A total of 313 M. larici-populina isolates were genotyped in this
study (Table 3). Among these, 224 distinct multilocus genotypes
were identified. Very few identical multilocus genotypes were
detected in different populations, except two pairs (FRA-1/FRA-2
and FRA-3/POL). The mean expected heterozygosity over all loci
and all sites was moderately high (HE= 0.352 ? 0.239).All loci were
found to be polymorphic and exhibited numbers of alleles ranging
from two to 16, formMLP37 andmMLP38 loci, respectively. No pair of
loci was significantly linked after Bonferroni correction for multiple
Most identical multilocus genotypes were considered as non-
clonal according to the analysis performed using MLGSIM, except
in the BIH population, where 21 isolates were found to result from
asexual reproduction (Table 3). As a result, the relative genotypic
diversity was high in all the populations (0.78–1.00), except in the
BIH population (0.32). The subsequent analyses were performed
using the clone-corrected dataset.
All the European populations exhibited a high number of
polymorphic loci (8 or 9 out of 9), except the GBR population (5 out
of 9). Allelic richness was moderately high (Ar= 2.62 ? 0.33,
mean ? S.D.), ranging from 1.78 to 3.05, in the GBR and ITA
populations, respectively. Gene diversity also was relatively high
(HE= 0.345 ? 0.059, mean ? S.D.), the lowest and highest values also
being found in GBR (HE= 0.236) and ITA (HE= 0.431) populations,
respectively. None of the European populations was found to deviate
significantly from Hardy–Weinberg proportions. Mean allelic rich-
ness and mean gene diversity were not significantly different
between populations collected from P. nigra and on hybrid poplars
(P = 0.14 and 0.17, respectively).
Whereas European populations appeared to be consistent with
Hardy–Weinberg proportions, the CAN population did not
(FIS= 0.771, P < 0.001). Nonetheless, the ISL population did not
significantly deviate from
(FIS= ?0.007, P = 0.52). Only three and five loci were found to be
polymorphic in the Canadian and Icelandic populations, respec-
tively. The mean allelic richness (Ar= 1.43 ? 0.19, mean ? S.D.) and
mean gene diversity (HE= 0.131 ? 0.091, mean ? S.D.) were signifi-
cantly lower in these two overseas populations compared to the
European populations (P < 0.05). Besides, all the alleles found in the
CAN and ISL populations were found in the European populations.
Two to four mitotypes were found in the European populations,
whereas only mitotype D was identified in the CAN and ISL
populations(Table 3). Therefore,the genetic diversityof the Canadian
and Icelandic populations represents only a subset of the extant
diversity in the European populations. Regardless of the mutation
model assumed, the tests performed with BOTTLENECK were not
significant for both CAN and ISL populations.
3.3. Population differentiation
Pathotypic distances between populations were well explained
by the first axis of the principal components analysis (PCA)
(Fig. 2A), which accounted for more than 68% of the total inertia.
The second axis accounted for only 15% of the pathotypic distance
inertia. The first axis sorted the populations into complex and
simple pathotypes. Indeed, the increase in pathotype complexity
from the right to the left of the diagram matched a great variation
in vir7 frequency between FRA-1, FRA-2, FRA-3, and GBR
Pathotypic characteristics of the M. larici-populina populations
Vir1Vir2Vir3Vir4 Vir5Vir6 Vir7Vir8
aNumber of isolates analyzed for virulence.
bNumber of pathotypes found.
cPathotype evenness index (Simpson’s index corrected for sample size).
dPathotype richness index (Shannon’s index corrected for sample size).
ePathotype complexity (mean number of virulences per individual).
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
populations (in which most isolates carried this virulence) and the
remaining populations (Table 2). It is worth noting that popula-
tions carrying the virulence 7 were collected from hybrid poplars,
which either carried the corresponding resistance gene or were
surrounded by such resistant cultivars. This was not the case for
the hybrid poplars from Canada.
The PCA performed on genetic distances showed a very
different pattern (Fig. 2B). Based on the genetic distance analysis,
all the European populations were clumped together, regardless of
the original hosts. The discrepancy between the genetic and
pathotypic PCA analyses is further highlighted by the lack of
correlation between genetic and pathotypic distance matrices
(r = ?0.051, P = 0.47). In the PCA analysis performed on genetic
distances, the first and second axis accounted for approximately
54% and 32% of total inertia, and separated the Canadian and
Icelandic populations from the European cluster. This is further
supported by pairwise population differentiation analyses: pair-
wise FST values between the European and the two recently
founded populations were all very high and significant (P < 0.001),
ranging from 0.198 to 0.653 (Table 4). The genetic differentiation
between the CAN and ISL populations was also very high
(FST= 0.653, P < 0.001). Albeit much weaker, a significant genetic
differentiation was detected among the European populations
(global FST= 0.057, P < 0.001). Pairwise FSTranged from 0 to 0.162,
differentiation values within Europe were found between GBR and
continental populations. According to the assignment tests
performed with GENECLASS2 software, none of the Canadian or
Icelandic individuals could be clearly assigned to one of the
3.4. Spatial genetic structure
We found no evidence for a spatial structure of pathotype
diversity, as the correlation between pathotypic and geographic
distances was weak and non-significant (either taking all popula-
tions into account, r = 0.006, P = 0.39, or excluding the ISL and CAN
populations, r = 0.062, P = 0.36).
In contrast, we detected a significantly positive correlation
between the FST/(1 ? FST) ratios and the log-transformed geo-
graphic distances when performed on all populations (dashed
regression line, P < 0.001, Fig. 3). This overall IBD pattern seemed,
however, disrupted by an inflection point occurring when the
Genetic characteristics of the M. larici-populina populations
B, C, D, E
B, C, D
B, C, D
A, B, C, D
C, D, E
B, C, D
aNumber of isolates genotyped.
bNumber of genotypes after clone-correction.
cRelative genotypic diversity.
fInbreeding coefficient. Ar, HEand FISwere calculated on the clone-corrected dataset.
gSignificant departure of FISfrom 0.
Fig. 2. Principal component analysis based on (A) the pathotypic distance matrix
and (B) the genetic distance matrix between all population pairs. Pathotypic
distances between populations were estimated using the Manhattan method
reduced by the expected range. Shared allele distances (DAS) were used for genetic
distances. Black circles represent European populations; open triangles represent
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
logarithm of distance equals 7.5 (around 2000 km). This inflection
point corresponds to the overseas populations (CAN and ISL) being
included in the spatial regression analysis. Partial regression
performed on only European populations yielded a weaker
correlation, albeit still significant, between genetic and log-
transformed geographic distances (full regression line, P < 0.01,
Fig. 3). This drop in statistical confidence may be due to the large
decrease in the value of the slope of the regression line (from 0.31
to 0.04). The departure of overseas populations from the IBD
pattern was best figured by means of an autocorrelogram (Fig. 4).
The autocorrelation index r yielded positive and significant values
for 5.5–6.5 log distance classes, with an x-intercept at roughly 7.6
(approximately 2000 km). These five distance classes, which
corresponded to within-Europe distance intervals, determined
the extent of non-random genetic structure, i.e., an IBD pattern, as
Pairwise FSTestimated with Weir and Cockerham’s f (1984) on the clone-corrected dataset between M. larici-populina populations (above diagonal)a, and pairwise geographic distances between locations in kilometers (below
aUnbiased estimate of the P-value of log-likelihood-based exact test of genotypic distribution using a Markov chain method (dememorization = 5000, batches = 50, iterations = 2000) in GENEPOP version 3.4.
bPairwise geographical distances were computed using CIRCE´software.
cSignificance levels are indicated by stars (*, ** and *** for P < 0.05, P < 0.01 and P < 0.001, respectively).
Fig. 3. Regression plot between FST/(1 ? FST) ratios and log-transformed geographic
distances. P values were obtained by a Mantel test (10,000 permutations). Black
circles represent pairs of European populations; open triangles represent pairs with
at least one overseas (Canada or Iceland) population.
Fig. 4. Correlogram showing the genetic autocorrelation index r as a function of log-
transformed geographic distance classes. A 95% confidence interval around the null
hypothesis of the random distribution of Melampsora larici-populina individuals
(dotted line) was determined by 1000 permutations of the dataset. Black circles
represent within-Europe distance classes. Open triangles represent distance classes
including at least one overseas country (Canada or Iceland). The regression line
between positive values of r (within-Europe distance classes) and log-transformed
geographic distances was plotted (dashed line after it crosses the x-axis).
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
the corresponding values of r (black circles) well aligned along a
regression line (Hardy and Vekemans, 1999). The two other
distance classes (open triangles), which included all overseas
distance intervals, showed significant negative values of r (which
reflects that CAN and ISL are highly differentiated from European
populations and from each other). These two points did not fit the
expected decrease in r values when extrapolating the IBD pattern
(dashed line). Whereas gene flows across all the European
populations exhibited an IBD pattern, the overseas populations
displayed evidence for stochastic LDD events.
4.1. Evidence for both IBD pattern and LDD events in M. larici-
The main outcome of this study was that both an IBD pattern
and stochastic LDD events occurred among M. larici-populina
populations. Interestingly, Mantel tests performed according to
Rousset’s method (1997) were all significant, whether taking
overseas populations into account, or not. From these tests, it was
therefore not possible to determine whether or not the IBD pattern
extended overseas. Evidence for a break in the dispersal regime
occurring between European and overseas populations was better
assessed through the autocorrelogram analysis. According to this
method, a non-random spatial genetic structure was observed
when only the European populations were included. The points
corresponding to intra-Europe distance intervals aligned well and
showed a regular decrease of the relationship coefficient (r) as
geographic distances increased (Fig. 4). This is considered to be a
signature of an IBD (Hardy and Vekemans, 1999). The two last
points of the autocorrelogram – those that include overseas
populations in the analysis – did not fit with this regression line,
indicating that these populations separated from the European
populations through stochastic differentiation processes.
Interestingly, the IBD pattern and LDD events did not overlap.
The inflection point figured in the graphical representation of the
Mantel test enabled the geographic range for gradual dispersal to
be clearly delineated. In the autocorrelogram, this switch point
matched the intersection between the regression line and the x-
axis. However, it is important to notice that this coincidencehas no
biological meaning; first, because the interception point is largely
dependent on the sampling design, and second, because the extent
of the IBD pattern should not be regarded as an estimation of the
distance dispersal of an organism (Hardy and Vekemans, 1999;
Vekemans and Hardy, 2004). In M. larici-populina, it is likely that
gradual dispersal occurs across the whole native range of the
pathogen, and that the IBD pattern might have been larger,
provided we had sampled locations eastward within Europe.
Conversely, the distance at which we observed the shift from
gradual to stochastic dispersal might reflect the sufficient distance
interval – characterizedbya fulllengthof unsuitablehabitat– for a
stochastic genetic differentiation to appear; that is, a sea distance
of 2000 km was sufficient to create stochastic dispersal of M. larici-
In principle, the mean dispersal distance has to be estimated
through the slope of the regression line between the genetic and
geographic distances (Rousset, 1997; Hardy and Vekemans, 1999).
Here, it is obvious that including LDD would have biased the
estimate of the dispersal distance by artificially inflating the value
of the slope of the regression line. Significant IBD does not
necessary mean a spatially homogeneous gene flow (e.g. Garnier
et al., 2004). This raises the problem of how to distinguish LDD
events from an IBD pattern when these two processes occur within
the same geographic range. To our knowledge, there is no specific
methodology to deal with this situation. Recent advances in
Bayesian computing in population genetics enable genetic
discontinuities to be easily revealed (Pritchard et al., 2000; Falush
et al., 2003). However, these methods are more likely to detect
physical barriers to gene flow when clustering populations are at
Hardy–Weinberg equilibrium (Manel et al., 2003), than to
distinguish between IBD and LDD, the latter being often far from
genetic equilibrium (see below). The most promising method
(described in Garnier et al., 2004) is to perform successive
resampling of populations,test for IBD (e.g. with Rousset’s method,
1997), and then exclude the populations that greatly increased the
slope of the regression line. This method was successfully applied
to the ground beetle Carabus soleri to confirm that a geographic
barrier (the Alps) prevented secondary contacts during postglacial
recolonization from southern France and Italy (Garnier et al.,
2004). However, a standardized procedure is still missing, which
would aim to provide a statistically based exclusion rule.
4.2. Evidence for isolation by distance within Europe
An IBD pattern was found among all M. larici-populina
populations sampled across the European continent (including
the GBR population from Northern Ireland). This correlation
between geographic and genetic distances results from the regular
and gradual dispersal of the pathogen over the spatial scale
covering the sampled regions. The IBD pattern proved that there is
no ubiquitous mixing in M. larici-populina across Europe, but
instead that gene flow decreased over geographic distances.
Nonetheless, the genetic differentiation between European popu-
lations of M. larici-populina was moderately low, which resulted in
a weak value of the slope of the regression line. Therefore, even if
dispersal is limited over Europe, an important gene flow occurred
at this spatial scale.
The detection of an IBD pattern greatly depends on the spatial
scale studied (Rousset, 1997; Castric and Bernatchez, 2003), but
the appropriate sampling scale to reveal an IBD pattern for plant
pathogenic fungi is not obvious at first sight. This may account for
the limited number of studies that detected an IBD on plant
pathogens. Considering the high dispersal ability of airborne
pathogens (Brown and Hovmøller, 2002), it is not surprising that
most IBD patterns were found only at the continental scale (e.g.
Cronartium ribicola in Eastern Canada, Et-touil et al., 1999;
Rhynchosporium secalis in Australia, McDonald et al., 1999; or
Plasmopara viticola in Central Europe, Gobbin et al., 2006), or even
at the worldwide scale for Rhynchosporium secalis (Zaffarano et al.,
2006). Sampling at smaller spatial scales often resulted in the
failure to detect IBD, except for the particular example of
Microbotryum violaceum, for which no IBD was found between
populations (Delmotte et al., 1999; Giraud, 2004) but an IBD
pattern was detected within populations (Giraud, 2004). For some
airborne plant pathogenic fungi, IBD can be difficult to detect even
at the largest spatial scale, because of a high gene flow (Brown and
Hovmøller, 2002; Zeller et al., 2004). In a few cases, no genetic
structure was observed, which implies that all the populations
behaved as a single panmictic unit (Et-touil et al., 1999).
Nonetheless, it is also quite likely that at these large spatial scales
(world or continent), some barriers to gene flow produce a large
and stochastic genetic differentiation that would impede the
detection of IBD. Many physical barriers – mountains, seas or any
vast area without suitable hosts – can disrupt the natural dispersal
of the pathogen, leading to reproductive isolation. In these cases,
significant population differentiation is expected (Manel et al.,
2005). The effects of drift, together with selection, will generate
this genetic differentiation; it may also be strengthened by
founder effects, as observed in Mycosphaerella fijiensis during the
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
worldwide spread of the black leaf streak disease of banana.
Founder effects lead to highly stochastic differentiation processes
and a reduction of population genetic diversity (Rivas et al., 2004).
Conversely, the effect of physicalbarriers to gene flow canbe much
weaker, as observed in Plasmopara viticola, the causal agent of
grapevinedownymildew,for whichthestrengthofthe IBDpattern
across Europe was only slightly reduced when taking into account
the Greek populations (Gobbin et al., 2006). In the present study,
the English Channel, the North Sea and the Irish Sea could have
acted as physical barriers limiting spore dispersal of M. larici-
populina from the continent to Northern Ireland. Although the GBR
population showed a lower level of genetic diversity and a large
genetic differentiation with the remaining European populations,
this population nonetheless fell under the IBD pattern.
Another explanation for a lack of correlation between genetic
and geographic distances among pathogen populations is the
existence of selection driven by the host. Such selection could
indeed lead to local adaptation, which would interfere with an
underlying IBD pattern by selecting for the individuals most
adapted to their host plants (e.g. the bean anthracnose fungus
Colletotrichum lindemuthianum, Capelle and Neema, 2005). This
would particularly be the case if sexual reproduction occurs on the
same host plant where the selection acts, leading to reproductive
isolation (Giraud et al., 2006). Hence, the population genetic
structure of the pathogen would be more conditioned by the
distribution of the different host plants than by geographic
distances between pathogen populations. Local adaptation pro-
cesses are usually measured using selected markers such as
virulences. A well-known example of such studies is given by the
flax rust fungus, Melampsora lini, an autoecious rust fungus which
completes its whole life cycle on Linum marginale (Thrall and
Burdon, 2002). Patterns of metapopulation dynamics driven by
local adaptation were also shown in the Plantago lanceolata–
Podosphaera plantaginis interaction (Laine, 2005; Laine and Hanski,
2006). In the present study, we showed that the host type also
conditioned the virulence profile of the M. larici-populina isolates.
This was highlighted by the higher pathotype complexity of the
isolates collected from the cultivated poplars, and the large
pathotype differentiation between populations from cultivated vs.
wild poplars, as has already been shown in a previous study
(Ge ´rard et al., 2006). Indeed, the extensive use of cultivars carrying
specific resistance genes toward M. larici-populina in poplar
cultivation has strongly affected the pathogen populations (Pinon
andFrey,2005).For example,the massiveplantationofthecultivar
‘Beaupre ´’ carrying the resistance factor 7 had a great impact on the
population dynamics and virulence profile of the poplar rust
(Ge ´rard et al., 2006). Nonetheless, we found no correlation
between selected (virulence) and neutral (microsatellite) markers.
In addition, microsatellite markers revealed little (albeit signifi-
cant) population structure with respect to the sampled host plant
(FST= 0.025, P < 0.001). We can therefore argue that there is little
reproductive isolation between the populations sampled from
cultivated and wild poplars. It is noteworthy that M. larici-populina
has to alternate on larch, the aecial host, to reproduce sexually. In
accordance with previous results (Ge ´rard et al., 2006), the genetic
characteristics of M. larici-populina populations strongly suggest
that sexual reproduction occurs widely and frequently: (i) all the
European M. larici-populina populations were at Hardy–Weinberg
equilibrium; (ii) none of the loci pairs were linked in any of the
studied populations; (iii) there were very few clones, except in
Bosnia and Herzegovina. This is consistent with the wide
distribution of the European larch (Larix decidua). Although native
to the mountains of Central Europe (Alps, Carpathians, Sudetes,
Tatras) and lowlands in Northern Poland, it is now well distributed
in lowlands throughout Europe because it has been widely planted
for timber and ornamental purposes. Because it promotes genetic
intermixing, the need to alternate on larch is likely to explain the
lack of reproductive isolation of M. larici-populina populations.
4.3. Evidence for founder effects overseas
Introduction of M. larici-populina in Eastern Canada and Iceland
are recent events (Innes et al., 2004; H. Sverrisson, personal
communication). Such long distance dispersal events may have
originated with a limited number of individuals. Here we report
several lines of evidence that Canadian and Icelandic populations
exhibited signatures of strong founder effects. First, we observed a
reduction of genetic diversity compared to populations from the
native range. In the Canadian and Icelandic populations, most of
the loci were monomorphic, and allelic richness and gene diversity
were significantly lower than in the European populations. Second,
Cornuet and Luikart (1996) showed that recent founder effects
should result in a gene diversity excess at selectively neutral loci.
However, we did not detect such a genetic disequilibrium in the
Canadian and Icelandic populations. This may be explained by the
lack of power of the tests (Luikart et al., 1998). In addition, the
Canadian population showed an important heterozygote defi-
ciency. This large deviation from Hardy–Weinberg equilibrium
may result from the sexual reproduction among a very restricted
number of individuals (even clone mates), therefore resulting in a
high rate of selfing (Giraud, 2004; Raboin et al., 2006). Third, high
genetic differentiation was observed between the two newly
founded populations and the European populations. Indeed,
founder effects often result in high genetic differentiation between
populations because the reduced effective number of individuals
favors a rapid divergence of gene frequencies through drift
(Boileau et al., 1992; Carlier et al., 1996; Rivas et al., 2004).
Moreover, great differentiation was observed between Canadian
and Icelandic populations, suggesting the independence of the two
introductions, and illustrating again the stochastic character of
populations could not be determined from the assignment tests,
possibly because of too large population differentiation between
European and overseas populations.
4.4. Putative mechanisms of long distance dispersal
For plant pathogenic fungi, long distance dispersal events can
either result from passive transport by wind or from human
activity (Brown and Hovmøller, 2002). Although there is no direct
evidence, the migration from Europe to Iceland is likely to have
resulted from an airborne LDD event, as shown for other taxa.
Several insect species, especially Lepidopteran, were shown to be
transported by wind from Europe to Iceland (Downes, 1988). In
addition, large amounts of birch pollen originating from con-
tinental Europe were detected in Iceland in May 2006, several
weeks before the flowering of local birch trees (M. Hallsdo ´ttir,
personal communication). Similar airborne LDD events have
already been reported for M. larici-populina between Australia
and New Zealand via trans-Tasman air-currents (Wilkinson and
Spiers, 1976), and also for other rust fungi such as coffee leaf rust
(Hemileia vastatrix, Bowden et al., 1971), sugarcane rust (Puccinia
melanocephala, Purdy et al., 1985), and soybean rust (Phakopsora
pachyrhizi, Pan et al., 2006). The fact that poplars remained free
from any rust fungus for decades in Iceland reflects the very low
probability of viable spores reaching susceptible poplar leaves.
Nevertheless, the development of poplar cultivation in Iceland
during the 1990’s increased the net trapping effect, and severe
poplar rust epidemics in Western Europe during the period
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
1996–2000 certainly strengthened the inoculum pressure (Lons-
dale and Tabbush, 2002; Pinon and Frey, 2005).
The wind dispersal hypothesis seems less realistic for
explaining the introduction of M. larici-populina in Canada,
considering (i) the higher distance and (ii) the direction of
prevailing winds in the Northern hemisphere. Moreover, most of
the Populus species in North America are susceptible to M. larici-
populina. Therefore, if wind-dispersed migration could have
occurred in the past, M. larici-populina should have been
discovered far sooner than its first report in North America
(Newcombe and Chastagner, 1993). Wilkinson and Spiers (1976)
suggested the possible spread of M. larici-populina by infected
plant material to explain the introduction of the pathogen in
urediniospores attached to poplar buds, or as mycelium in buds,
has been reported so far. Hence, the introduction of M. larici-
populina in Canada seems more likely to be due to human
transport (e.g. by urediniospores carried on clothes).
It appears that M. larici-populina has become durably estab-
lished in Canada and Iceland. Indeed, larches (Larix spp.) are
present in both countries and aecia of M. larici-populina are
observed each spring on larch needles in the vicinity of poplars in
Canada (Grondin et al., 2005) and Iceland (H. Sverrisson, personal
communication). The lack of significant linkage disequilibrium
between microsatellite loci is in agreement with the occurrence of
sexual reproduction in the Canadian and Icelandic populations.
quarantine regulations in order to avoid the introduction of exotic
pathogens to healthy areas, as these pathogens can become
durably established under favorable conditions. Hybridization
(Brasier, 2001). Hybridization between M. larici-populina and M.
medusae f. sp. deltoidae, a North American poplar rust fungus, has
already been reported in New Zealand (Spiers and Hopcroft, 1994)
and South Africa (Frey et al., 2005). The hybrid taxon was shown to
exhibit the cumulative host range of both parental species, and
could thus be a potential threat to poplar cultivation.
This work was supported by INRA, ECOGER (InterPopGer
Programme), and a fellowship from the Re ´gion Lorraine to B.B. We
thank all the colleagues listed in Table 1 for collecting and sending
us rust-infected poplar leaves, and Alistair McCraken (DARD,
Belfast, UK) and Pierre Pe ´rinet (MRNFP, Que ´bec, Canada) for access
to poplar trials. We thank also Christine Ge ´hin and Be ´ranger Bertin
for their excellent technical help, and Tatiana Giraud for a critical
reading of the manuscript. The manuscript was proofread for
English language by American Journal Experts (http://www.jour-
poplar leaves from outside the EU was obtained from the French
Plant Protection Service, Permit No. JDB/03/388.
Andrivon, D., de Vallavieille-Pope, C., 1995. Race diversity and complexity in
selected populations of fungal biotrophic pathogens of cereals. Phytopathology
Aylor, D.E., 2003. Spread of plant disease on a continental scale: role of aerial
dispersal of pathogens. Ecology 84, 1989–1997.
Barre `s, B., Dutech, C., Andrieux, A., Caron, H., Pinon, J., Frey, P., 2006. Isolation and
characterization of 15 microsatellite loci in the poplar rust fungus, Melampsora
larici-populina, and cross-amplification in related species. Mol. Ecol. Notes 6,
Boileau, M.G., Hebert, P.D.N., Schwartz, S.S., 1992. Non-equilibrium gene fre-
quency divergence: persistent founder effects in natural populations. J. Evol.
Biol. 5, 25–39.
Bowcock, A.M., Ruiz-Linares, A., Tomfohrde, J., Minch, E., Kidd, J.R., Cavalli-Sforza,
L.L., 1994. High resolution of human evolutionary trees with polymorphic
microsatellites. Nature 368, 455–457.
Bowden, J., Gregory, P.H., Johnson, C.G., 1971. Possible wind transport of coffee leaf
rust across the Atlantic Ocean. Nature 229, 500.
Brasier, C.M., 2001. Rapid evolution of introduced plant pathogens via interspecific
hybridization. Bioscience 51, 123–133.
continental scales and its impact on plant disease. Science 297, 537–541.
Capelle, J., Neema, C., 2005. Local adaptation and population structure at a micro-
geographical scale of a fungal parasite on its host plant. J. Evol. Biol. 18, 1445–
Carlier, J., Lebrun, M.H., Zapater, M.F., Dubois, C., Mourichon, X., 1996. Genetic
structure of the global population of banana leaf streak fungus Mycosphaerella
fijiensis. Mol. Ecol. 5, 499–510.
Castric, V., Bernatchez, L., 2003. The rise and fall of isolation by distance in the
anadromous brook charr (Salvelinus fontinalis Mitchill). Genetics 163, 983–996.
Cornuet, J.M., Luikart, G., 1996. Description and power analysis of two tests for
detecting recent population bottlenecks from allele frequency data. Genetics
Delmotte, F., Bucheli, E., Shykoff, J.A., 1999. Host and parasite population structure
in a natural plant-pathogen system. Heredity 82, 300–308.
Downes, J.A., 1988. Thepost-glacial colonizationof the NorthAtlantic Islands. Mem.
Entomol. Soc. Can. 144, 55–92.
Epperson, B.K., Huang, Z., Li, T.Q., 1999. Measures of spatial structure in samples of
genotypes for multiallelic loci. Genetica 73, 251–261.
Et-touil, K., Bernier, L., Beaulieu, J., Be ´rude ´, J.A., Hopkin, A., Hamelin, R.C., 1999.
Genetic structure of Cronartium ribicola populations in eastern Canada. Phyto-
pathology 89, 915–919.
Fager, E.W., 1972. Diversity: a sampling study. Am. Nat. 106, 293–310.
multilocus genotype data: linked loci and correlated allele frequencies. Genet-
ics 164, 1567–1587.
Fenster, C.B., Vekemans, X., Hardy, O.J., 2003. Quantifying gene flow from spatial
genetic structure data in a metapopulation of Chamaecrista fasciculate (Legu-
minosae). Evolution 57, 995–1007.
Flor, H.H., 1971. Current status of the gene-for-gene concept. Annu. Rev. Phyto-
pathol. 9, 275–296.
Fresa, R., 1936. Argentine Republic: Melampsora larici-populina in the Delta of
Parana. Int. Bull. Plant Protec. 10, 145–146.
Frey, P., Ge ´rard, P., Feau, N., Husson, C., Pinon, J., 2005. Variability and population
biology of Melampsora rusts on poplars. In: Pei, M.H.,McCracken, A.R. (Eds.),Rust
Diseases of Willow and Poplar. CAB International, Wallingford, UK, pp. 63–72.
Frey, P., Pinon, J., 2004. La rouille du peuplier: un pathosyste `me mode `le. Biofutur
Fry, W.E., Goodwin, S.B., Matuszak, J.M., Spielman, L.J., Milgroom, M.G., Drenth, A.,
1992. Population genetics and intercontinental migrations of Phytophthora
infestans. Annu. Rev. Phytopathol. 30, 107–130.
Garnier, S., Alibert, P., Audiot, P., Prieur, B., Rasplus, J.Y., 2004. Isolation by
distance and sharp discontinuities in gene frequencies: implications for
the phylogeography of an alpine insect species, Carabus solieri. Mol. Ecol.
Ge ´rard, P., Husson, C., Pinon, J., Frey, P., 2006. Comparison of genetic and virulence
diversity of Melampsora larici-populina populations on wild and cultivated
poplar and influence of the alternate host. Phytopathology 96, 1027–1036.
Gibson, I.A.S., Waller, J.M., 1972. Notes from ODA plant pathology liaison officers.
Pest Art. News Sum. 18, 336–337.
Giraud, T., 2004. Patterns of within population dispersal and mating of the fungus
Microbotryum violaceum parasitising the plant Silene latifolia. Heredity 93,
Giraud, T., Jonot, O., Shykoff, J.A., 2006. Common sex-linked deleterious alleles in a
plant parasitic fungus alter infection success but show no pleitropic advantage.
J. Evol. Biol. 19, 970–980.
Gobbin, D., Rumbo, A., Linde, C.C., Gessler, C., 2006. Population genetic structure of
Plasmopara viticola after 125 years of colonization in European vineyards. Mol.
Plant Pathol. 7, 519–531.
Goudet, J., 1995. FSTAT (version 12): a computer program to calculate F-statistics. J.
Hered. 86, 485–486.
Grondin, J., Bourassa, M., Hamelin, R.C., 2005. First report of the aecial state of
Melampsora larici-populina on Larix spp. in North America. Plant Dis. 89, 1242.
Halkett, F., Plantegenest, M., Prunier-Leterme, N., Mieuzet, L., Delmotte, F., Simon,
J.C., 2005. Admixed sexual and facultatively asexual aphid lineages at mating
sites. Mol. Ecol. 14, 325–336.
Hardy, O.J., Vekemans, X., 1999. Isolation by distance in a continuous population:
reconciliation between spatial autocorrelation and population genetics models.
Heredity 83, 145–154.
Innes, L., Marchand, M., Frey, P., Bourassa, M., Hamelin, R.C., 2004. First report of
Melampsora larici-populina on Populus spp. in Eastern North America. Plant Dis.
Kern, F.D., Thurston, H.W., 1954. Additional species of Uredinales from Colombia.
Mycologia 46, 354–357.
Laine, A.L., 2005. Spatial scale of local adaptation in a plant-pathogen metapopula-
tion. J. Evol. Biol. 18, 930–938.
Laine, A.L., Hanski, I., 2006. Large-scale spatial dynamics of a specialist plant
pathogen in a fragmented landscape. J. Ecol. 94, 217–226.
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587
Lloyd, H.L., 1971. New records of Rhodesian fungi. 1. Melampsora larici-populina
Klebahn. Kirkia 8, 1–2.
Lonsdale, D., Tabbush, P., 2002. Poplar rust and its recent impact in Great Britain.
Information Note 7 (Revised) Forestry Commission, Edinburgh, UK.
Luikart, G., Allendorf, F.W., Cornuet, J.M., Sherwin, W.B., 1998. Distortion of allele
frequency distributions provides a test for recent population bottlenecks. J.
Hered. 89, 238–247.
Manel, S., Schwartz, M.K., Luikart, G., Taberlet, P., 2003. Landscape genetics:
combining landscape ecology and population genetics. Trends Ecol. Evol. 18,
Manel, S., Caggiotti, O.E., Waples, R.S., 2005. Assignment methods: matching
biological questions with appropriate techniques. Trends Ecol. Evol. 20, 136–
McDonald, B.A., Zhan, J., Burdon, J.J., 1999. Genetic structure of Rhynchosporium
secalis in Australia. Phytopathology 89, 639–645.
Nagarajan, S., Singh, D.V., 1990. Long-distance dispersal of rusts pathogens. Annu.
Rev. Phytopathol. 28, 139–153.
Nathan, R., Perry, G., Cronin, J.T., Strand, A.E., Cain, M.L., 2003. Methods for
estimating long-distance dispersal. Oikos 103, 261–273.
Newcombe, G., Chastagner, G.A., 1993. First report of the Eurasian poplar leaf
rust fungus, Melampsora larici-populina, in North-America. Plant Dis. 77, 532–
Pan, Z., Yang, X.B., Pivonia, S., Xue, L., Pasken, R., Roads, J., 2006. Long-term
prediction of soybean rust entry into the continental United States. Plant
Dis. 90, 840–846.
Paetkau, D., Calvert, W., Stirling, I., Strobeck, C., 1995. Microsatellite analysis of
population structure in Canadian polar bears. Mol. Ecol. 4, 347–354.
Paetkau, D., Slade, R., Burden, M., Estoup, A., 2004. Genetic assignment methods for
the direct, real-time estimation of migration rate: a simulation-based explora-
tion of accuracy and power. Mol. Ecol. 13, 55–65.
Peakall, R., Smouse, P.E., 2005. GenAlEx 6: genetic analysis in Excel. Population
genetic software for teaching and research. Mol. Ecol. Notes 6, 288–295.
Pei, M.H., Bayon, C., Ruiz, C., Tubby, I., 2007. Population structure of poplar rust
Melampsora larici-populina in the UK inferred from AFLP. Plant Pathol. 56, 472–
Perrier, X., Flori, A., Bonnot, F., 2003. Data analysis methods. In: Hamon, P., Seguin,
M., Perrier, X., Glaszmann, J.C. (Eds.), Genetic Diversity of Cultivated Tropical
Plants. Enfield Science Publishers, Montpellier, France, pp. 43–76.
Pinon, J., Frey, P., 2005. Interactions between poplar clones and Melampsora
populations and their implications for breeding for durable resistance. In:
Pei, M.H., McCracken, A.R. (Eds.), Rust Diseases of Willow and Poplar. CAB
International, Wallingford, UK, pp. 139–154.
Pinon, J., Newcombe, G., Chastagner, G.A., 1994. Identification of races of Melamp-
sora larici-populina, the Eurasian rust fungus, on Populus species in California
and Washington. Plant Dis. 78, 101.
Piry, S., Luikart, G., Cornuet, J.M., 1999. BOTTLENECK: a computer program for
detecting recent reductions in the effective population size using allele fre-
quency data. J. Hered. 90, 502–503.
Piry, S., Alapetite, A., Cornuet, J.-M., Paetkau, D., Baudoin, L., Estoup, A., 2004.
GeneClass2: a software for genetic assignment and first generation migrants
detection. J. Hered. 95, 536–539.
Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure
using multilocus genotype data. Genetics 155, 945–959.
Purdy, L.H., Krupa, S.V., Dean, J.L., 1985. Introduction of sugarcane rust into the
Americas and its spread to Florida. Plant Dis. 69, 689–693.
Raboin, L.M., Athiappan, S., Miranda Oliviera, K., Paulet, F., Calatayud, C., Zapater,
M.F., Brottier, P., Luzaran, R., Garsmeur, O., Carlier, J., D’Hont, A., 2006. Evidence
for the dispersal of a unique lineage from Asia to America and Africa in the
sugarcane fungal pathogen Ustilago scitaminea. Fungal Genet. Biol. 44, 64–76.
Raymond,M.,Rousset,F.,1995.Genepop(version 12):populationgenetics software
for exact tests and ecumenism. J. Hered. 86, 248–249.
Rice, W.R., 1989. Analyzing tables of statistical tests. Evolution 43, 223–225.
Rivas, G.G., Zapater, M.F., Abadie, C., Carlier, J., 2004. Founder effects and stochastic
dispersal at the continental scale of the fungal pathogen of bananas Myco-
sphaerella fijiensis. Mol. Ecol. 13, 471–482.
Rotem, J., Wooding, B., Aylor, D.E., 1985. The role of solar radiation, especially
ultraviolet, in the mortality of fungal spores. Phytopathology 75, 510–514.
Rousset, F., 1997. Genetic differentiation and estimation of gene flow from F-
statistics under isolation by distance. Genetics 145, 1219–1228.
Slatkin, M., 1977. Gene flow and genetic drift in a species subject to frequent local
extinctions. Theor. Popul. Biol. 12, 253–262.
Smouse, P.E., Peakall, R., 1999. Spatial autocorrelation analysis of individual multi-
allele and multilocus genetic structure. Heredity 82, 561–573.
Spegazzini, C., 1918. Cuarta contribucio ´n a la Micologı ´a chilena. Rev. Chil. Hist. Nat.
Spiers, A.G., Hopcroft, D.H., 1994. Comparative studies of the poplar rusts Melamp-
sora medusae, M. larici-populina and their interspecific hybrid M. medusae-
populina. Mycol. Res. 98, 889–903.
Stenberg, P., Lundmark, M., Saura, A., 2003. MLGSIM: a program for detecting clones
using a simulation approach. Mol. Ecol. Notes 3, 329–331.
Stephens, M., Smith, N.J., Donnelly, P., 2001. A new statistical method for haplotype
reconstruction from population data. Am. J. Hum. Genet. 68, 978–989.
Thrall, P.H., Burdon, J.J., 2002. Evolution of gene-for-gene systems in metapopula-
tions: the effect of spatial scale of host and pathogen dispersal. Plant Pathol. 51,
Valie `re, N., 2002. GIMLET: a computer program for analysing genetic individual
identification data. Mol. Ecol. Notes 2, 377–379.
Vekemans, X., Hardy, O.J., 2004. New insights from fine-scale spatial genetic
structure analyses in plant populations. Mol. Ecol. 13, 921–935.
Walker, J., Hartigan, D., Bertus, A.L., 1974. Poplar rusts in Australia with comments
on potential conifer rusts. Eur. J. Forest Pathol. 4, 100–118.
Weir, B.S., Cockerham, C.C., 1984. Estimating F-statistics for the analysis of popula-
tion structure. Evolution 38, 1358–1370.
Wilkinson, A.G., Spiers, A.G., 1976. Introduction of the poplar rusts Melampsora
larici-populina and M. medusae to New Zealand and their subsequent distribu-
tion. New Zeal. J. Sci. 19, 195–198.
Zaffarano, P.L., McDonald, B.A., Zala, M., Linde, C.C., 2006. Global hierarchical gene
diversity analysis suggests the fertile crescent is not the center of origin of the
barley scald pathogen Rhynchosporium secalis. Phytopathology 96, 941–950.
Zeller, K.A., Bowden, R.L., Leslie, J.F., 2004. Population differentiation and recombi-
nation in wheat scab populations of Gibberella zeae from the United States. Mol.
Ecol. 13, 563–571.
B. Barre `s et al./Infection, Genetics and Evolution 8 (2008) 577–587