• AFLP;
  • Colletotrichum lindemuthianum;
  • local adaptation;
  • Phaseolus vulgaris;
  • population structure;
  • recombination;
  • wild populations


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Local adaptation, which has been detected for several wild pathosystems is influenced by gene flow and recombination. In this study, we investigate local adaptation and population structure at a fine scale in wild populations of a plant-pathogen fungus. We sampled hierarchically strains of Colletotrichum lindemuthianum in a wild population of its host. The analysis of AFLP patterns obtained for 86 strains indicated that: (i) many different haplotypes can be discriminated, although occurrence of recombination could not be shown; (ii) migration between adjacent plants seemed rare during the season; and (iii) neutral diversity is structured according to groups of plants and individual host plants. Furthermore, we tested for the occurrence of local adaptation using a cross-inoculation experiment. Our results showed local adaptation at the scale of the individual host plant. These results indicate that fine-scale dynamics has evolutionary consequences in this pathosystem.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Studies of epidemics on crops have shown that plant parasites can quickly adapt to their host plant and overcome resistance genes used in crop protection. This process is thought to occur in wild pathosystems as well, where hosts and parasites show high levels of diversity structured into metapopulations (Burdon & Thrall, 2000).

Many studies have explained parts of the process driving this dynamic of reciprocal adaptation in mutualistic or parasitic symbioses but Thompson (1994) first proposed an integrative view of the evolution of traits involved in co-evolution among geographically structured populations of interacting species, referred to as the geographic mosaic theory of co-evolution. This theory emphasizes the importance of local selective forces that create those traits and of gene flow that reshuffles them at the species level. It points out the need to study geographic structure of wild pathosystems at different scales to detect the pertinent level of reciprocal selection, i.e. the mesh of the co-evolutionary web. To that end, the analysis of gene flow is usually achieved by using neutral molecular markers. Then, the population structure obtained with these markers can be compared with the population structure found for traits involved in the interaction (Thompson, 1997). However, the comparison between the structure observed for presumed neutral and selected markers can be obscured in asexual organisms because all loci are linked and thus submitted to the same evolutionary forces. Thompson's theory has proved to be useful as predicted patterns can be observed in nature (Burdon & Thrall, 2000; Koskela et al., 2000; Thompson & Cunningham, 2002), and are consistent with many observations in naturally occurring epidemics (Burdon & Thrall, 1999).

One of the most salient results of co-evolution is the emergence of a process of local adaptation, which provides an insight into the spatial scales at which coevolutionary dynamics occurs. It is encountered when parasites perform better when confronted with local partners than with partners from distant populations (Kaltz & Shykoff, 1998). When parasites evolve faster than their hosts, they have an evolutionary advantage because they can quickly track the changes of the local host population, leading to their local adaptation. Conversely, parasites should lag behind their host if the latter evolves faster (Barrett, 1988) and appear maladapted. Mathematical modelling has shown that the evolutionary advantage leading to local adaptation relies mainly on gene flow (Gandon et al., 1996), in conjunction with population size, mutation rate and generation time (Gandon & Michalakis, 2002). In particular, the partner who has the higher migration or mutation rate should be the locally adapted one. Generally, parasites are considered to have an evolutionary advantage over their hosts (Hamilton et al., 1990), although in some cases this is not clear (Mutikainen et al., 2000). Local adaptation has been recognized for numerous host/parasite interactions, like snail/trematode (Lively & Jokela, 1996), plant/fungus (Burdon & Thrall, 2000), crustacean/bacteria (Ebert, 1994), or plant/plant (Mutikainen et al., 2000) (see Kaltz & Shykoff, 1998, for a review). It has been mainly detected at the population (Ebert, 1994; Lively & Jokela, 1996; Koskela et al., 2000) or meta-population (Burdon & Thrall, 2000) levels, but some studies have shown that it can arise at the level of the individual host, particularly in insect/tree interactions where parasites have a low migration distance and a far shorter generation time than their hosts (Edmunds & Alstad, 1978; Karban, 1989; Hanks & Denno, 1994; Mopper et al. 2000).

In this paper, we seek to complete the analysis of the co-evolutionary process between a fungal plant parasite, Colletotrichum lindemuthianum (Ascomycota) and its host, the common bean (Phaseolus vulgaris, Fabacea) in wild populations. This pathosystem is well suited for the study of co-evolution because the genetic basis of the interaction has been partly elucidated (Geffroy et al., 1999; Ferrier-Cana et al., 2003) and is thought to follow a gene-for-gene relationship (Flor, 1956). It has previously been investigated for diversity, population structure and local adaptation at different levels: interspecific level (Sicard et al., unpublished data), worldwide (Fabre, 1997), centres of host diversity (Balardin et al., 1997; Sicard et al., 1997b) and regional scales (Sicard et al., 1997a; Rodriguez-Guerra et al., 2003). These studies have evidenced significant population differentiation for neutral markers and local adaptation of the parasite has concomitantly been detected at several scales (Geffroy et al., 1999; Neema, unpublished data; Sicard, unpublished data). Therefore, local adaptation seems to be associated with a reduction of gene flow between geographic units. However, this pattern could be due either to broad meta-population dynamics acting at the regional scale or to a local intra-population dynamics. In order to know if local adaptation can arise at a finer scale and if it is related to a lack of gene flow, we searched for these patterns at a local level in a single host population. We used three hierarchical levels of analysis: groups of plants from a single population, individual plants within those groups and pods within plants. As the comparison between these results is impaired by the absence of recombination, we also tested our dataset for the presence of recombination. So far, C. lindemuthianum has been considered as asexual since no sexual stage has been observed in nature and scarcely in laboratory conditions (Tu, 1992). However, a study at a regional scale has measured a level of linkage disequilibrium consistent with recombination (Sicard et al., 1997b).

The principal objectives of our experiments were to determine:

  • 1
    the extent of gene flow at a local scale, by examining the population structure observed for neutral molecular markers;
  • 2
    the occurrence of local adaptation at this scale;
  • 3
    the extent of recombination in the population studied.

Material and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Fungus material

Wild common bean plants were collected in 2001, at Vaqueros site, situated in the province of Salta, in northwest Argentina. We sampled 13 plants from two groups, 500 m apart, of approximately 10 plants each, situated along a road: six plants from group 1 and seven plants from group 2. Each group of plant is located in an area of approximately 20 × 100 m and individual plants were sampled at least 3 m apart, in order to avoid re-sampling of the same plant. Fungal isolations were conducted as described in Sicard et al. (1997a) and 93 monoconidial strains were isolated. Strains were named according to their origin as follows: the first number refers to the group of plants, the second number to the plant of origin. The third number corresponds to the pod of origin and the letter to the lesion from which the strain was isolated. Thus, 11.3A and 11.2B are two strains isolated from two different pods collected on plant 11.

DNA extraction

Spores were multiplied on sterilized bean leaves placed on petri dishes filled with potato dextrose agar. After 20 days of incubation at 20 °C, the surface was flooded with sterile water and scraped to harvest spores. The solution was mixed with 10 mL of a potato dextrose broth containing 1% casamino acids and 1% yeast extract. One per cent penicillin G and streptomycin were added to prevent bacterial contaminations. After 72 h incubation at 20 °C, the mycelium was harvested, dried on sterile filter paper and stored at −20 °C.

The DNA was extracted from mycelium with a phenol/chloroform protocol as described in Villareal et al. (2002) and stored at −20 °C in TE 0.1 (10 mm Tris–HCl, pH 8; 0.1 mm EDTA).

The AFLP analysis

The AFLP reactions were performed as described in Justesen et al. (2002), except we used the enzyme TruI instead of MseI. The primer combinations used for selective amplifications were those in Justesen et al. (2002): P12/M24 + T, P11/M13 + T and P13/M24 + T. The amplified bands were stained with silver nitrate as described in Chalhoub et al. (1997). Each AFLP reaction was performed twice on the same DNA sample. As DNA extraction could lead to AFLP artefacts, 24 strains presenting rare bands were re-extracted and checked against previous data. Polymorphic bands were detected and scored visually.

Cross-inoculation experiments

Fifty-four strains previously genotyped were multiplied and suspensions of spores were prepared without any additional use of antibiotics. Spore suspensions were adjusted to 106 spore mL−1 for inoculation of bean leaflets.

Seeds of 12 of the 13 plants collected were sown in 10 cm diameter pots filled with commercial soil. The pots were then placed in a growth chamber at 20 °C with 12 h of light for 5 weeks, until the plants reached the 4-leaf stage.

Tests were performed on trifoliate leaves collected from each plant. Each of the three leaflets from the same trifoliate leaf was inoculated with a different fungal strain. The first leaflet was inoculated with a strain isolated from the same plant (interaction type A), the second with a strain isolated from the same plant group (interaction type B) and the third with a strain isolated from the other group (interaction type C) (Fig. 1). Some strains were inoculated several times on their plant of origin, to reach the same number of interactions for types A–C. A total of 320 leaves were inoculated, using 54 plants/strains combinations for type A interactions, 178 for type B and 163 for type C.


Figure 1. Design of the cross-inoculation experiment used to test local adaptation at the scale of individual hosts plant and groups of plants. All three leaflets were detached from the leaf and each one was inoculated with a strain from a different type of interaction. Each leaf has been inoculated with the three types of interaction. An example of repartition of interaction types on leaflets is presented for a leaf from the plant 12.

Download figure to PowerPoint

Leaflets were inoculated using a paintbrush and placed in 5.5 cm diameter Petri dishes lined with moistened filter paper. After inoculation the dishes were randomly placed in a growth chamber and high relative humidity (95%) was maintained by daily spraying water into the dishes. Symptoms were scored every 2 days by three different experimenters, from the 5th to 15th day after inoculation. An additional final reading was made on day 18. We used the following 5-point scale:

  • 1 = 
    no visible symptoms
  • 2 = 
    brown limited necrotic spots
  • 3 = 
    large lesions on leaf veins
  • 4 = 
    large brown lesions on the leaf blade
  • 5 = 
    maceration of more than the half of the leaflet area

This scale is consistent with the international scale used for seedlings inoculations (Pastor-Corrales, 1988). Therefore, notes equal to or above 3 correspond to compatible interactions, whereas notes 1 and 2 correspond to incompatible interactions.

Data analysis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

The AFLP genotyping

Data check

The AFLP patterns scored were checked for reproducibility with different replicates.

Diversity and spatial repartition of the strains on the plants

The mean number of plants colonized per haplotype, P, was computed for an estimation of the probability that strains infect different plants. The expected value is 1 if each strain is restricted to a single plant, and greater than 1 otherwise. A distribution of simulated indices corresponding to the null hypothesis of a random repartition of the strains on the plants was generated by 1000 random displacements of the strains on the plants, keeping the original number of strains isolated per plant. The P value of the test is given by the frequency of simulated indices smaller than or equal to the observed one.

Nei's diversity indices (for each hierarchical level analysis) were calculated using the following equation:

  • image

where pj is the frequency of the jth haplotype and n is the total number of haplotypes per hierarchical level i. The DAS distance (Jin & Chakraborty, 1994) implemented in the Populations software (Langella, 2002) was computed between all AFLP patterns. The distance matrix obtained was used to compute an UPGMA distance tree with Treeplot (Langella, 2003). This allowed an efficient detection of identical haplotypes; then one strain for each distinct haplotype was selected for subsequent analysis.

Population structure analysis

Genetic differentiation between geographic units (plants and groups of plants) was estimated using the Weir and Cockerham's theta (1984). The difference to zero of each calculated θ was tested using a permutation procedure: 1000 datasets were generated by a random displacement of the strains within the groups of strains, thus breaking the geographic associations. Then, the index was calculated for each random dataset to generate a distribution of simulated indices corresponding to the null hypothesis of no geographic association between alleles. The P value is given by the frequency of simulated indices superior or equal to the observed one. This procedure is implemented in the Multilocus software (Agapow & Burt, 2001).

In order to compare the population differentiation observed at each hierachical level, we used an analysis of molecular variance (amova). It allowed partitioning variability into its hierarchical components (variability between individuals within plants, among plants within groups and among groups). Fixation indices were derived from these components and tested against the null hypothesis of no geographic association obtained from a permutation procedure as described for θ. This analysis was done using the Arlequin software (Shneider et al., 2000).

Detection of recombination

The degree of association between alleles was estimated using a modified version of the index of association (Ia), which is the ratio between the observed variance of the number of differences between pairs of strains and the expected variance under the hypothesis of no linkage disequilibrium. Indeed, if some alleles are associated, some individuals will be closer or more distantly related together than to the rest of the population. Therefore, the observed variance will vary and the value of Ia will depart from zero (Maynard-Smith et al., 1993). We used a modified version of this index, inline image, corrected for the dependence to the number of loci used (Agapow & Burt, 2001). The inline image indices obtained were tested for difference to zero in the same way as for θ, except that alleles were randomly displaced between strains instead of strains between geographic units, to generate the distribution of expected indices under the null hypothesis of no linkage disequilibrium between loci. The P value of the test is given by the frequency of simulated indices superior or equal to the observed one. Significant departures from zero are interpreted as a lack or an absence of recombination. This procedure is implemented in the Multilocus software (Agapow & Burt, 2001).

Cross-inoculation experiment

Kendall's coefficient of concordance (W) was calculated on 154 leaflets scored at day 18, by three experimenters.

Effect of type of interaction

We assumed that local adaptation occurs at the individual plant scale when mean symptom scores for type A interactions were superior to either type B or C interactions. Similarly, we defined local adaptation at the group scale as mean symptoms scores being significantly higher in type B interactions than in type C interactions.

We tested the effect of the type of interaction on the mean symptom score using a repeated measure anova procedure, with leaf and type of interaction as main factors and day of scoring as error term. Comparison between levels of types of interactions was calculated using the Tukey's Honest Significant Differences test.

Computations of Kendall's coefficient of concordance, ANOVA and Tukey's HSD tests were done under the R statistical environment (R Development Core Team, 2003). All tests were considered significant at the threshold α = 0.05.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References


Data used

Seven of the 93 strains were removed from the dataset either because they resulted from contaminations, or because more than three loci were not clearly reproducible. Therefore, data on 86 strains was analysed. The AFLP generated 138 identifiable markers, among which 59 monomorphic markers and 19 polymorphic markers were removed: 11, which were not reproducible and eight, which presented missing values. Some loci presented linkage disequilibrium values of 1. We kept only one locus from each linkage group because these groups can be due to AFLP artefacts and bias recombination tests toward asexuality. The resulting dataset consisted of 55 loci of which 42 had one rare allele (<10% in frequency, mean Nei's diversity index per locus = 0.14). These loci discriminated 50 haplotypes among the 86 strains analysed.

Diversity and spatial repartition of the strains

At the group scale, Nei's diversity index showed that each group consisted of many different strains. At the plant scale, each AFLP haplotype was observed only on one plant and in general, several haplotypes were identified on each plant (Table 1). The mean number of plants infected per haplotype is minimal, i.e. inline image = 1. The probability of obtaining such a value of inline image if plants are infected at random is P < 0.001. Each pod could also be colonized by different haplotypes (Tables 1 and 2). The most common haplotype was found for 13 strains isolated on six different pods of the same plant. Nei's indices of diversity indicated that diversity increased with increasing scale, however a high diversity (0.59) was also observed at the pod level (Table 2).

Table 1.  Structure of the dataset of C. lindemuthianum strains isolated from wild bean, used for AFLP analysis and cross-inoculation experiments.
Group*PlantPodStrainAFLP† haplotype
  1. *Groups of plants are separated by 500 m.

  2. †The AFLP haplotypes were determined using 50 loci.

Table 2.  Nei's indices of diversity for AFLP markers, calculated at four nested levels using a dataset comprising 86 strains of C. lindemuthianum isolated on wild common bean.
 H * AFLP haplotypes†
  1. *Nei's index of diversity + SE.

  2. †Mean numbers of AFLP haplotypes per level.

  3. ‡The SE evaluated by jacknifing of individuals.

Population0.96 ± 0.01‡50
Group0.96 ± 0.0325
Plant0.77 ± 0.082.94
Pod0.59 ± 0.091.44

Population structure

Diversity was analysed at three hierarchical levels: groups of plants, plants within groups and pods within plants. Most of the diversity was observed within plants and pods (Table 3) though there was significant differentiation among groups with a calculated θ of 0.047 (P < 0.005) between groups. Within groups of plants, most of the variability is at the intra-pod level (Table 4), with a significant variance part among plants and among pods. This is consistent with an overall θ of 0.15 (P < 0.01) among plants inside group 1. However, this pattern is not found in-group 2 (θ = 0.02, n.s.), indicating that the strains which infect different plants in the group 2 were similar.

Table 3.  The amova conducted on AFLP haplotypes for 86 strains of C. lindemuthianum in a population of wild common bean plants from Argentina divided in two geographically distant groups.
 Variance (%)P value
  1. P value was calculated from 1000 random permutations.

Among groups3.980.02
Among plants30.51<0.001
Within plants65.51<0.001
Table 4.  The amova conducted on AFLP haplotypes for 86 strains of C. lindemuthianum in two groups of wild common bean plants from Argentina.
 Group 1Group 2
Variance (%)P valueVariance (%)P value
  1. P values were calculated from 1000 random permutations.

Among plants25.57<0.00118.810.08
Among pods32.87<0.00121.000.4
Within pods41.55<0.00160.180.04

Recombination test

A significant deviation from random association of markers, which is predicted for recombination was detected using the full dataset (inline image = 0.036, P < 0.01). Because this pattern can be partly due to the structure observed among the groups, we performed the same calculation within each group again using the full dataset. No recombination signal was detected at this scale either (group 1: inline image = 0.033, P < 0.05; group 2: inline image = 0.01, P < 0.05). Too few haplotypes were available to study recombination at the plant scale.

Cross-inoculation experiments

Kendall's concordance coefficient between the three experimenters was W = 0.87 (P < 0.001).

Local adaptation

Symptoms increased during the course of the experiment (Fig. 2). Results from the anova indicated a significant effect of the type of interaction (Table 5). Tukey's HSD test showed that this could be mainly due to type A interaction symptoms scores being more important than types B and C mean scores (Table 6). Symptoms scores for B interaction types were not significantly higher than symptoms scores for C interaction types (difference = 0.057 ± 0.058, Table 6).


Figure 2. Evolution of mean symptom scores throughout the cross-inoculation experiment for the three interaction types. Points are mean symptoms scores per day; bars show the SE of the mean. A, strain inoculated on its plant of origin; B, strain inoculated on a plant from the group of origin of the strain; C, strain inoculated on a plant from the other group.

Download figure to PowerPoint

Table 5.  The anova table of the test for local adaptation of C. lindemuthianum strains on common bean detached leaflets.
Effectd.f.Sum squareMean squareF valueP > F
  1. Type of interaction effect compares symptoms scores on bean leaflets inoculated either with a strain isolated on the plant (A), a strain isolated on the group of plant (B) or a strain isolated on the other group of plant (C).

  2. Calculations were computed using the symptom scores without transformation and with time as the error term.

Interaction type2140.370.2105.3<0.001
Table 6.  Tukey's HSD tests for differences between symptoms scores on detached leaflets for three interaction types of C. lindemuthianum strains infecting wild common bean plants. Tukey's HSD allows correction for multiple comparisons.
  1. Diff, difference in mean symptom scores for each level of interaction types; CI, Tukey's confidence intervals at the 95% level; Sign, indicates significance (yes if the 95% Tukey's CI does not contains zero, the null hypothesis of no difference between levels; no if otherwise).

A and B0.2760.218–0.334Yes
B and C0.057 −0.0004–0.116No


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Colletotrichum lindemuthianum strains showed a high intra-population diversity. Several haplotypes were found at all levels of analysis (group of plants, individual plants and pods), indicating that the diversity is created and/or maintained at a fine scale. This was consistent with previous studies, which have shown large amounts of diversity at broader geographical levels: centres of host diversity and regional scales (Balardin et al., 1997; Sicard et al., 1997a, b; Gonzales et al., 1998; Rodriguez-Guerra et al., 2003).

In the population studied, each haplotype was restricted to a single host plant whatever the number of strains sharing this haplotype. However, some haplotypes can be found on different pods of the same plant. This finding is consistent with the significant indices of differentiation observed between the two groups of plants, or between plants within the group 1. Moreover, a large part of the genetic diversity was found among plants (30% in total, 25% in group 1). This strongly supports the empirical hypothesis of very weak migration ability. Indeed, direct observations of dispersal under field conditions after a storm have evidenced a maximal distance of dispersal of a few meters (Tu, 1992). It is probable that dispersal of C. lindemuthianum during its parasitic phase on P. vulgaris mostly occurs between different parts of the same plant and occasionally between adjacent plants.

Local adaptation was also tested at the same scale using a cross-inoculation experiment. Type A interactions should always lead to compatible interactions, because they result from the inoculation of a leaflet with a strain isolated on the same plant genotype. However, at day 18, 17% of type A interactions were scored as incompatible (scores <3). This could reflect either the presence of opportunistic strains, which are not able to establish a successful infection on these plants, or a technical problem with the protocol, which may not reflect field conditions. However, the results of leaflets inoculation are well correlated with those obtained after seedlings inoculation (Dufresnes, unpublished data). Two measures are commonly used in local adaptation studies: first, parasites have on average better fitness gains on the hosts from their population of origin than on hosts from distant populations; second, local parasites are better than foreign parasites on their own host population. Our protocol was designed to use the second definition. So we chose to analyse the mean symptom scores without partitioning them into compatibility (0 or 1) and aggressiveness (symptom severity).

Results from the anova on mean symptom scores showed that P. vulgaris plants were more prone to C. lindemuthianum strains isolated on their plant of origin than to strains isolated on distant plants. Local strains also tended to perform better on their group of plant of origin than distant strains, although this effect is not significant.

Carius et al. (2001) already reported genotypic-specific adaptation in the Daphnia pulex/Pasteuria ramosa interaction. In this pathosystem, Pasteuria bacteria are endoparasites of a planktonic freshwater crustacean and are disseminated when the host dies. Adaptation to individual host has also been observed in insect/tree interactions, where the host is long-living and the parasite can complete many generations on the same individual tree (Edmunds & Alstad, 1978; Karban, 1989, Hanks & Denno, 1994; Mopper et al, 2000). In these examples, local interactions could be the shaping force of the co-evolution. To our knowledge, adaptation at the scale of the individual host plant has never been described in a fungal plant parasite.

A common trend of these systems is that host generations are longer than those of the parasite. The C. lindemuthianum/P. vulgaris pathosystem also meets this assumption: the reproducing cycle of the wild bean takes 5 months, whereas a spore of the fungus can produce a sporulating lesion in approximately 15 days. Therefore, the fungal generations are roughly 10 times shorter than those of the host.

Many studies have demonstrated the occurrence of local adaptation by comparing performances of parasites on hosts from distinct populations (Kaltz & Shykoff, 1998). These studies often involve transplant experiments, where either hosts or parasites are transplanted from one population to another. A problem with some studies is that environmental conditions are confounded with genetic composition of the population (see Hanks & Denno, 1994 for an example of local adaptation of Pseudaulacaspis insects to the phenotype of their host tree). In our study, we used controlled experimental conditions but the seeds used were collected in natural populations. Cases of phenotypic transmission of resistance to disease or predator have been documented by Agrawal et al. (1999). They showed a greater defensive ability of the descendants to the disease encountered by the parents, contrary to the local adaptation situation where the descendants appear to be more prone to the disease experienced by the mother plant. However, we cannot exclude a potential effect of the mother plant on seed quality.

In insect/tree interactions, local adaptation to individual host can depend on gene flow between plants. Hanks & Denno (1994) found that Pseudaulacaspis insects were more adapted to their host of origin when compared with distant trees, but not with nearby trees, probably because of gene flow between the tree of origin of the insect and the nearest trees, indicating that gene flow prevents local adaptation to individual plants. Conversely, Mopper et al. (2000) found that local adaptation of Stilbosis lepidopteran on newly colonized oaks trees increases after several years although differentiation between demes diminishes because of insect migration between plants. In our study, ‘long distance’ (A vs. C) and ‘low distance’ comparisons (A vs. B) were significantly differentiated as well. This could be explained by a weak gene flow between strains infecting close plants, in accordance with the significant index of differentiation found for plants of the group 1. However, comparisons between near and distant plants only are not significant, contrary to the weak differentiation index observed between groups of plants. Thus, in our study, patterns of adaptation to the host do not correspond exactly to the population structure. This indicates that the correlation observed between local adaptation and neutral structure at broader scales may not exist at the intra-population level. However, the differentiation between groups is weak (θ = 0.047, P < 0.01) and strains of C. lindemuthianum tend to be locally adapted at this scale. Thus, we cannot rule out the possibility that our test was not powerful enough to detect a weak effect. If this is true, the occurrence of local adaptation and population structure in this pathosystem might be regarded as two concomitant effects of the same mechanism, probably involving a weak dispersal ability of the fungus facing a slow evolving host.

However, a second hypothesis can explain our findings. Recombination is rare or absent in the population studied, in accordance with field observations (Tu, 1992) and genome analysis, which has shown a high level of aneuploidy (O'Sullivan et al., 1998) compatible with an absence of meiosis (Kistler & Miao, 1992). This indicates that neutral and selected markers are probably associated. Therefore, the population structure observed might reflect strain selection by the plants rather than the history of neutral markers, and we cannot choose between these hypotheses.

Mathematical models at the metapopulation level have shown that migration ability is a key parameter in understanding local adaptation. In particular, the partner that migrates the most should be the locally adapted one (Gandon et al., 1996). Therefore, our conclusion of a very weak dispersal ability of C. lindemuthianum and have a lack of local adaptation at the group level apparently contradicts the patterns of local adaptation observed at broader geographic scales. Three arguments can give insight into this apparent contradiction. First, local adaptation at the metapopulation level is influenced by recombination ability, mutation rates population size and generation time of the species in interaction (Gandon & Michalakis, 2002). In particular, C. lindemuthianum generations are short and its population size is probably higher than that of its host. Therefore, C. lindemuthianum could have a greater evolutionary potential than P. vulgaris. Second, migration ability of the parasite has to be compared with migration ability of the host. Unfortunately, these traits are rarely measured during the same experiment for both species and it is difficult to conclude if local adaptation patterns are consistent with a difference in dispersal ability. Propagules of wild P. vulgaris are heavy seeds (approximately 80 mg) without any dispersive structure and are disseminated primarily by explosive dehiscence of the pods, projecting the seeds a few meters away from the mother plant. This argues for a weak dispersal for the host too. In addition, the mode of pollination of P. vulgaris is predominantly selfing (Gepts et al., 1998), preventing pollen migration. Finally, the existence of two dynamics taking place at a local and global scale is plausible. This has already been observed in a snail/trematode interaction where parasites are locally adapted to hosts genotypes living at different depths in a single lake (Lively & Jokela, 1996). When populations of parasites from different lakes are compared, they also appear to be locally adapted (Lively & Dybdhal, 2000).

In conclusion, this analysis showed that an evolutionary dynamics could take place in this plant/fungus interaction at a very fine scale. However, the origin of these patterns is not clear and could involve the low migration ability, the greater evolutionary potential of the fungus and/or a selection pressure by the host. A better understanding would require more study, especially on the host population.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

We thank J. Enjalbert for his kind coaching with AFLPs and very useful discussions, D. Sicard for her advices and comments about inoculations and G. Damour and N. Arrouy for their help with cross-inoculation experiments. M. A. Sélosse, J. Shykoff, Y. Mickalakis, J. de Meaux and two anonymous reviewers improved this study and previous drafts. Prospections in Argentina was supported by a Dufresnoy grant from the Ffrench Academy of Agriculture. We also thank the international GNU project community for providing us with the free software used in this study.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Data analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References
  • Agapow, P.-M. & Burt, A. 2001. Indices of multilocus linkage disequilibrium. Mol. Ecol. Notes 1: 101102.
  • Agrawal, A.A., Laforsch, C. & Tollrian, R. 1999. Transgenerational induction of defences in animals and plants. Nature 401: 6063.
  • Balardin, R.S., Jarosz, A.M. & Kelly, J.D. 1997. Virulence and molecular diversity in Colletotrichum lindemuthianum from south, central and North America. Phytopathology 87: 11841191.
  • Barrett, J.A. 1988. Frequency-dependent selection in plant-fungal interactions. Phil. Trans. R. Soc. Lond. B 319: 473483.
  • Burdon, J.J. & Thrall, P.H. 1999. Spatial and temporal patterns in coevolving plant and pathogen associations. Am. Nat. 153: S15S33.
  • Burdon, J.J. & Thrall, P.T. 2000. Coevolution at multiple spatial scales: Linum marginale-Melampsora lini-from the individual to the species. Evol. Ecol. 14: 261281.
  • Carius, H.J., Little, T.J. & Ebert, D. 2001. Genetic variation in a host-parasite association: potential for coevolution and frequency-dependent selection. Evolution 55: 11361145.
  • Chalhoub, B.A., Thibault, S., Laucou, V., Rameau, C., Hofte, H. & Cousin, R. 1997. Silver staining and recovery of AFLP amplification products on large denaturing polyacrylamide gels. BioTechniques 22: 216220.
  • Ebert, D. 1994. Virulence and local adaptation of a horizontaly transmitted parasite. Science 265: 10841086.
  • Edmunds, G.F. & Alstad, D.N. 1978. Coevolution in insect herbivores and conifers. Science 199: 941945.
  • Fabre, J.V. 1997. Etude de la diversité des populations de Colletotrichum lindemuthianum, agent responsable de l'anthracnose du haricot, à l’échelle mondiale. PhD thesis, Université Paris 11, Orsay, France, p. 134.
  • Ferrier-Cana, E., Geffroy, V., Macadre, C., Creusot, F., Imbert-Bollore, P., Sevignac, M. & Langin, T. 2003. Characterization of expressed NBS–LRR resistance gene candidates from common bean. Theor. Appl. Genet. 1062: 251261.
  • Flor, H. 1956. The complementary genic systems in flax and flax rust. Adv. Genet. 8: 2954.
  • Gandon, S., Capowiez, Y., Dubois, Y., Michalakis, Y. & Olivieri, I. 1996. Local adaptation and gene-for-gene coevolution in a metapopulation model. Proc. R. Soc. Lond. B 263: 10031009.
  • Gandon, S. & Michalakis, Y. 2002. Local adaptation, evolutionary potential and host parasite coevolution: interactions between migration, mutation, population size and generation time. J. Evol. Biol. 15: 451462.
  • Geffroy, V., Sicard, D., de Oliveira, J.C.F., Sèvignac, M., Cohen, S., Gepts, P., Neema, C., Langin, T. & Dron, M. 1999. Identification of an ancestral resistance gene cluster involved in the coevolution process between Phaseolus vulgaris and its fungal pathogen Colletotrichum lindemuthianum. Mol. Plant Microbe. Interact. 9: 774784.
  • Gepts, P., Papa, R., Gonzales, A., Acosta, J. & Salinas, A.D. 1998. Human effects on Phaseolus vulgaris adaptation during and after domestication. In: Proceedings of the Seventh IOPB Symposium, Evolution in Man-Made Habitats (L.Van Raamsdonk, ed.), pp. 1115. Hugo de Vries Laboratory, Amsterdam, The Netherlands.
  • Gonzales, M., Rodriguez, R., Zaval, M.E., Jacobo, J.L., Hernandez, F., Acosta, J., Martinez, O. & Simpson, J. 1998. Characterization of mexican isolates of Colletotrichum lindemuthianum by using differential cultivars and molecular markers. Phytopathology 88: 292299.
  • Hamilton, W., Axelrod, R. & Tanese, R. 1990. Sexual reproduction as an adaptation to resist parasite a review. Proc. Natl. Acad. Sci. USA 87: 35663573.
  • Hanks, L. & Denno, R.F. 1994. Local adaptation in the armored scale insect Pseudaulacaspis pentagona homoptera: Diaspidae. Ecology 75: 23012310.
  • Jin, L. & Chakraborty, R. 1994. Estimation of genetic distance and coefficient of gene diversity from single-probe multilocus DNA fingerprinting data. Mol. Biol. Evol. 11: 120127.
  • Justesen, A.F., Ridout, C.J. & Hovmoller, M.S. 2002. The recent history of Puccinia striiformis f. sp. tritici in Denmark as revealed by disease incidence and AFLP markers. Plant Pathol. 51: 1323.
  • Kaltz, O. & Shykoff, J.A. 1998. Local adaptation in host-parasite systems. Heredity 81: 361370.
  • Karban, R. 1989. Fine-scale adaptation of herbivorous thrips to individual host plant. Nature 340: 6061.
  • Kistler, H.C. & Miao, V.P.W. 1992. New mode of genetic change in filamentous fungi. Annu. Rev. Phytopathol. 30: 131152.
  • Koskela, T., Salonen, V. & Mutikainen, P. 2000. Local adaptation of a holoparasitic plant, Cuscuta europea: variation among populations. J. Evol. Biol. 13: 749755.
  • Langella, O. 2002. Populations, a free population genetics software. URL
  • Langella, O. 2003. Treeplot, a phylogenetic tree drawing tool. URL
  • Lively, C.M. & Dybdahl, M.F. 2000. Parasite adaptation to locally common host genotypes. Nature 4056787: 679681.
  • Lively, C.M. & Jokela, J. 1996. Clinal variation for local adaptation in a host-parasite interaction. Proc. R. Soc. Lond. B 263: 891897.
  • Maynard-Smith, J., Smith, N.H., O'Rourke, M. & Spratt, B.G. 1993. How clonal are bacteria? Proc. Natl. Acad. Sci. 90: 43844388.
  • Mopper, S., Stiling, P., Landau, K., Simberlof, D. & Van Zandt, P. 2000. Spatio-temporal variation in leafminer population structure and adaptation to individual oak trees. Ecology 81: 15771587.
  • Mutikainen, P., Salonen, V., Puustinen, S. & Koskela, T. 2000. Local adaptation, resistance, and virulence in a hemiparasitic plant-host plant interaction. Evolution 542: 433440.
  • O'Sullivan, D., Tosi, P., Creusot, F., Cooke, B.M., Phan, T.H., Dron, M. & Langin, T. 1998. Variation in genome organization of the plant pathogenic fungus Colletotrichum lindemuthianum. Curr. Genet. 334: 291298.
  • Pastor-Corrales, M.A. 1988. Variación patogénica de Colletotrichum lindemuthianum, el agente causal dela antracnosis del fijol y una propuesta para su estandardización. In: La antracnosis del frijol común Phaseolus vulgaris (Ciat, ed.), pp. 212251. América Latina, Cali, Columbia.
  • R Development Core Team 2003. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-00-3.
  • Rodriguez-Guerra, R., Ramirez-Rueda, M.-T., Martinez de la Vega, O. & Simpson, J. 2003. Variation in genotype, pathotype and anastomosis groups of Colletotrichum lindemuthianum isolates from mexico. Plant Pathol. 52: 228235.
  • Shneider, S., Roessli, D. & Excoffier, L. 2000. Arlequin Version 2.000: A Software for Population Genetics Data Analysis. University of Geneva, Genetics and Biometry Laboratory, Switzerland.
  • Sicard, D., Buchet, S., Michalakis, Y. & Neema, C. 1997a. Genetic variability of Colletotrichum lindemuthianum in wild populations of common bean. Plant Pathol. 46: 355365.
  • Sicard, D., Michalakis, Y., Dron, M. & Neema, C. 1997b. Genetic diversity and pathogenic variation of Colletotrichum lindemuthianum in the three centers of diversity of its host, Phaseolus vulgaris. Phytopathology 87: 807813.
  • Thompson, J.N. 1994. The Coevolutionary Process. The University of Chicago Press, IL, USA.
  • Thompson, J.N. 1997. Evaluating the dynamics of coevolution among geographically structured populations. Ecology 78: 16191623.
  • Thompson, J.N. & Cunningham, B.M. 2002. Geographic structure and dynamics of coevolutionary selection. Nature 417: 735738.
  • Tu, J. 1992. Colletotrichum: biology, pathology and control. In: Colletotrichum lindemuthianum on bean: population dynamics of the pathogen and breeding for resistance, pp. 203224. CABI, Wallingford.
  • Villareal, L.M.M.A., Lannou, C., Vallavieille-Pope, C. & Neema, C. 2002. Genetic variability in Puccinia striiformis fsp. tritici populations sampled on a local scale during natural epidemics. Appl. Environ. Microbiol. 68: 61386145.
  • Weir, B. & Cockerham, C.C. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 13581370.