WWW.DISSERS.RU


...
    !

Pages:     | 1 || 3 |

Van Putten, Wilhelmus Frederik On host race differentiation in smut fungi / Wilhelmus Frederik van Putten. - Utrecht: ...

-- [ 2 ] --

dioica-dominating patch (Table 1). This might explain why we did not find higher values for FST and RST in chapter 2. When adding the 2001 dataset, we expect to gain more statistical power, and clarify the effects of local host spatial structure both on the microsatellite loci and on the SCC marker allele frequencies. Also, host sex can then be included as a factor to be analyzed as well. Pollinators are differentially attracted to Silene plants of different sex (Shykoff and Bucheli 1995;

Van Putten et al. chapter 5).

Hence, spatial structuring of smut populations might be further enhanced due to gender and/or gender disease interactions (Real and McElhany 1996).

Assuming a long-term divergence of host races in allopatric host populations, as argued in chapter 2, both the microsatellite data and the SCC marker show signs of fungal gene flow in this sympatric host population. The differentiation between host species, or between patches of host plants can often be a starting point of host specialization, but may here represent a different situation. Presumably, the two host races have come into secondary contact with each other in this population (Van Putten et al. chapter 2). Whereas gene flow opposes the historically evolved differentiation between the host races, and acts to homogenize the genetic diversity, the local deme structure of both host species in this population will favor its maintenance (cf. Mopper 1996). Habitat, or host choice is found to be a crucial factor in theoretical models on host specialization (Fry 1996, Kawecki, 1997;

1998). Being the actors of host choice for this fungus, the vector/pollinator guilds of these host species are expected to play a dominant role in the process of maintaining the host-related differentiation among fungal isolates (Van Putten et al. chapter 5).

T I with Arjen Biere and Jos van Damme submitted to Evolution We have studied intraspecific competition and assortative mating between strains of the anther smut Microbotryum violaceum from two of its host species Silene latifolia and S. dioica.

Host differentiation between strains from these two host species is maintained in sympatric host populations despite the presence of gene flow. We studied whether higher competitive ability of strains on their native host species and/or positive assortative mating between host races occurs, which could contribute to the maintenance of such host differentiation.

In general, strains isolated from S. latifolia outcompeted strains isolated from S. dioica on both host species, but in female hosts, heterotypic dikaryons had the largest competition success. Furthermore, latency period was significantly shorter in infections that contained strains from S. latifolia, compared to homokaryotic infections with a S. dioica origin.

The frequencies of conjugation between strains originating from S. latifolia were much higher than conjugation frequencies between strains from S. dioica. A significant positive correlation was detected between the relative success rate of strains in competition and in conjugation, suggesting that success of a strain in competition might be partly determined by its swiftness of mating. In addition, reciprocal differences between homotypic and heterotypic crosses revealed a significant effect of fungal mating type, with mating type a1 being the main determinant of mating pace.

The observed differences in infection success, conjugation rate and latency period, in favor of strains from S. latifolia relative to strains from S. dioica on both host species are discussed in an evolutionary context of the maintenance of host race differentiation in sympatric populations of hosts.

I T TI Plant parasites can often employ more than one host species, and show intraspecific variation in host use. How such variation originates or is maintained, and under which conditions this may lead to host race formation and speciation are central questions in evolutionary biology. Positive assortative mating and differences in performance of pathogens on different host species are expected to play an active role in processes of divergence between strains, or in the maintenance of host-related differentiation. Indeed, host associated fitness trade-offs, i.e. by antagonistic pleiotropy of genes, would lead to increased divergence between strains, and host race formation (Jeanike 1990). Empirical evidence for such genetic correlations across different host species is often found to be ambiguous, or non-negative in studies of phytophagous insects (cf. Fry 1996;

but see Via et al. 2000), and is scant for other organisms including the group of phytopathogenic fungi. From an evolutionary point of view, competition among strains from different host species may be important as well, especially in cases where different physiological races have evolved and occur in sympatry (Day 1980). In such cases competition may enhance differences in the ability of host resource exploitation among strains. Depending on the amount of gene flow between the races, one race might outcompete the other, which could lead to extinction, both races could merge into one race by introgression, or host races might coexist. Empirical examples of phytopathogenic fungi that share the same host species and have to compete for its resources intraspecificly, are scarce (cf. Shearer 1995), and often involve saprophytic fungi rather than biotrophic fungi (but see Day 1980;

Newton et al. 1997;

1999). Besides differences in performance, strains from different host species may show host-specific mating preferences that affect the process of differentiation. In the group of heterothallic basidiomycetes, i.e. fungi that reproduce sexually with physiologically different strains (= mating types), mating involves processes of recognition and conjugation between cells of opposite mating type in order to produce an infectious dikaryon, and precedes entering a host. Positive assortative mating with respect to host species due to faster recognition, or conjugation will promote the divergence between strains, and may lead to (sympatric) host race formation and eventually speciation (Kondrashov and Shpak 1998). Such positive assortative mating in sympatry may be expected between races that have diverged considerably, to avoid maladaptive hybridization between strains, a process that is known as reinforcement (cf. Noor 1999).

The anther smut Microbotryum violaceum, obligate parasite of a wide host range within the Caryophyllaceae, is a well studied example of a pathogenic fungus, for which different physiological strains have been described on a number of host species (Zillig 1921). Recently, fungal strains from a number of host species have been characterized for a number of molecular markers, revealing strong differentiation (e.g. Perlin 1996;

Shykoff et al. 1999;

Bucheli et al. 2000). Two closely related host species, Silene latifolia and S. dioica, common roadside herbs, frequently occur in sympatry in areas where their preferred habitats overlap (Goulson and Jerrim 1997).

Anther smuts from these host species proved to be genetically differentiated in allopatric populations of hosts (Bucheli 2001;

Van Putten et al. chapter 2), but are still able to cross infect the other host species (Zillig 1921;

Baker 1947;

Biere and Honders 1996a). In natural sympatric populations of these host species, fungal isolates of anther smut from different host species show significant differentiation (Van Putten et al. chapter 3) as well. Gene flow between the host species in these sympatric populations is evidenced by the occurrence of interspecific hybrids, reported to constitute more than 6% of the population (Biere and Honders 1996b). Since this fungus is vectored by the natural pollinators of their hosts (Jennersten 1983), this raises the question how host races remain differentiated in the presence of fungal gene flow (Van Putten et al. chapter 2), especially since infection success is not necessarily higher on conspecifics of the host of origin (Biere and Honders 1996a). One factor that could contribute to the maintenance of host differentiation in sympatry is the finding that spore production in male hosts tends to be higher on conspecifics of the host of origin than on heterospecifics in these cross inoculation experiments. Other possibilities that also could play a role include host fidelity of vectors (Van Putten et al. chapter 5), a higher competitive ability of strains on their native host species, and/or sufficient host-related positive assortative mating between fungal isolates from the same host species.

The aim of this chapter is to study intraspecific competition and assortative mating between the host races from S. latifolia and S. dioica of M. violaceum in more detail. Specifically, we will address the following questions: (1) Is the success of strains from a host race in competition with strains from a different host race higher on the host species from which they originate, i.e. have strains adapted to their own host species with respect to success in competition? (2) Are there positive assortative mating patterns, i.e. is conjugation between strains of the same host race more frequent and/or faster than between strains of different host races? (3) Do mating frequencies of intra- and interracial crosses depend on the host species (host extract) on (in) which mating occurs? (4) How much of the variation in success of strains in competition can be explained by their mating success?

From the observed divergence between these host races in allopatric host populations, that is observed to some degree in sympatry as well (Van Putten et al.

chapter 2;

chapter 3), we hypothesize that strains isolated from allopatric S. latifolia and S. dioica hosts;

(1) have a higher infection success on conspecific hosts than on heterospecific hosts. Heterosis might counteract any host adaptations, but it is difficult to predict the infection success of the heterospecific dikaryons (heterokaryons) relative to conspecific dikaryons in competition;

(2) mate assortatively with respect to host species of origin, i.e. conjugate in higher frequencies with strains that were isolated from conspecifics than with strains that were isolated from heterospecific hosts;

and (3) that mating (conjugation) frequency is higher in the presence of the native host extract than in water and/or in extracts of the non-native host species.

T I T The anther smut fungus Microbotryum violaceum (Pers.: Pers) Deml & Oberw.

(=Ustilago violacea (Pers.) Fuckel) (Ustilaginaceae) (Vnky 1994) is a well-studied example of a heterobasidiomycete fungus that obligatory parasitises susceptible members of the Caryophyllaceae plant family to complete its sexual lifecycle (Thrall et al. 1993). To achieve a better understanding of the processes involved in competition and conjugation between haploid cells of this fungus, this sexual life cycle will be shortly reviewed here. Starting with diploid teliospores arriving on a healthy host plant, the sexual life cycle commences with germination. Germination of smut spores by meiosis results in four-celled promycelia from which four haploid cells (sporidia) of two mating types (designated a1 and a2) bud off. In principle, the two mating types are produced in a 1:1 ratio. Haploid cells of opposite mating types can then mate on the surface of (flowering) plants. The mating process and recognition between cells is influenced by pheromones (Blker and Kahmann 1993;

Snetselaar et al. 1996), which promote the formation of fungal fimbriae, i.e. microscopic hair-like structures composed of collagen, carbohydrates and RNA (Poon and Day 1975 but see Celerin and Day 1998). Hereafter, a conjugation tube between these cells of different mating type is formed (Poon et al. 1974), marking the start of the dikaryotic phase in which the infectious hyphae are produced. Critical factors in the mating process are low temperature and low nutrient level, and the presence of oxygen and salts (Cummins and Day 1977). The development of the fungus from this point on is regulated by the presence of host plant chemicals (Day et al. 1981;

Kokontis and Ruddat 1986;

1989), of which -tocopherol (= vitamin E) has been identified as one of the major factors stimulating hyphal growth (Castle and Day 1984). Infectious hyphae grow intercellularly (Spencer and White 1951;

Audran and Batcho 1981) and grow along with the plants apical meristemic regions (Day 1980). When the dikaryotic parasitic mycelium grows into the stamens of a flower, anthers produce teliospores instead of pollen (Thrall et al. 1993). As spores mature in the anther sacs, karyogamy marks the start of the diploid phase (Day and Garber 1988). In dioecious host species, an infection of female plants causes a morphological sex change;

ovaries are aborted and staminal rudiments develop into stamens that bear spore-filled anthers (Day and Garber 1988;

Thrall et al. 1993), a process that is induced by the fungus (Audran and Batcho 1981;

Scutt et al. 1997). The teliospores are dispersed by the natural insect visitors of the host plant that serve the dual role of pollinators of healthy plants and vectors of this sexually transmitted disease (Jennersten 1983). M. violaceum is found to be highly selfing (Baird and Garber 1979), resulting in strong homozygosity in several host races (Bucheli et al. 2000;

Van Putten et al. chapter 2). Automixis (mating among products of different meioses from the same diploid origin, i.e.

between haploid sporidia from different teliospores of the same strain) as well as autogamy (mating among products of a single meiosis, i.e. within a single basidium) are more likely to occur (Hood and Antonovics 1998;

2000) than outcrossing, limiting the opportunities for outcrossing.

Fungal isolates, hereafter referred to as strains (sensu Staley and Krieg 1984;

a strain is made up of the descendants of a single isolation in pure culture, and usually is made up of a succession of cultures ultimately derived from a single colony), were selected out of a collection containing haploid sporidia of single mating type. The original diploid teliospores were collected from several natural allopatric populations of hosts in the Netherlands in 1992 and 1993, and then cultured to separate mating types. Since then, sporidia have been stored at 20C. Strains were chosen in such a way that both mating types of the original single flower teliospores were present. Wild type (+) strains of M. violaceum produce pink colonies (Sporidial Colony Color;

SCC) when growing on standard medium due to the formation of lycopene. The yellow colored mutant converts this lycopene into -carotene through a cyclase that is inactive in the wild type (Garber et al. 1975). Single sporidia colonies that originated from allopatric S. latifolia host populations were almost fixed for the pink allele at one of the SCC loci, whereas single sporidia colonies that originated from allopatric S. dioica hosts were almost fixed for the yellow allele (cf. Biere and Honders 1996a;

Van Putten et al. chapter 3).

T Silene latifolia Poiret (= S. alba (Miller) Krause (Caryophyllaceae), the white campion is a dioecious short-lived perennial from open, disturbed habitats and borders of arable land. Silene dioica (L.) Clairv. (Caryophyllaceae), the red campion is a closely related dioecious perennial that mainly occurs in more shady humid habitats as woodland borders. In areas where habitats overlap, both species frequently occur in sympatry and hybridization is a common phenomenon (Baker 1947;

Goulson and Jerrim 1997).

Seeds of S. latifolia and S. dioica were collected from several geographically spread natural allopatric populations in the Netherlands in 1997 and 1998. Per species, seeds were bulked and then thoroughly mixed to randomize. Seeds were germinated in petridishes on demi-water moistened filter paper at a density of approximately seeds per petridish in a growth cabinet (16/8h light/dark, 21/15C day/night temperature) after a vernalization period of three days at 4C that has proven to be sufficient to increase the proportion of flowering plants in previous experiments.

Nearly all seeds had germinated after a week.

T I For the competition experiment, 24 strains (12 of each mating type) that were isolated from 12 different infected hosts from six allopatric S. latifolia populations, and similarly 24 strains that were isolated from 12 different infected hosts from seven allopatric S. dioica populations, were used. With these 48 (2x12x2) strains, different inoculation mixtures (one-ninth of all possible combinations) were prepared containing randomly selected strains from each of the four different sporidial types (S.

latifolia mating type a1 L1;

a2 L2;

S. dioica mating type a1 D1;

a2 D2). The number of times that an individual strain participated in one of the mixtures, ranged between and 12 with median 5 (expected mean = 5.3) and mode 3. Combinations containing two strains originating from a single teliospore were excluded from the experiment.

Each of these inoculation mixtures can lead to four different types of conjugates;

L1L (yielding teliospores with SCC pink phenotype after a successful infection), L1D2, or D1L2 (yielding teliospores with pink/yellow phenotype after a successful infection) and D1D2 (yielding teliospores with yellow phenotype after a successful infection).

Strains were cultured on a plate containing standard medium. Before the experiment each strain was checked for mating type using reference strains. A large loop of cells was scraped off a plate and suspended in 1.0ml of sterile milliQ water. A CoulterCounter Z1 (Coulter Electronics Ltd., Luton, England) was used to count the cells. Each sample was diluted to 5.0x107 cells/ml with sterilized demineralized water.

An inoculation mixture (inoculum) contained 2.5ml of each of the four standardized strain samples, and was thoroughly mixed overnight in a shaker at 14C.

Two ml of inoculum was added to petridishes with growing seedlings, which were potted three weeks after inoculation. To check for possible cross-infections, a subset of petridishes with growing seedlings received 2ml demi-water instead of inoculum, and served as control plants in the experiment.

552 S. latifolia seedlings (8 per mixture + 40 control plants) and 690 S. dioica seedlings (10 per mixture + 50 control plants) were potted and placed in a greenhouse which was kept below 25C and with a 16/8 hr day/night light regime. In the first week the seedlings were covered with cloth to protect from dehydration. In the second week plant/inoculation mixture combinations were randomized over the greenhouse tables. Plants that started flowering were removed instantly from the greenhouse compartment and were checked for disease status. From infected plants, spores from flowers were collected in separate eppendorf cups, and analyzed for SCC type.

Preferably, teliospores were taken from closed flower buds, and from different flowering stalks whenever possible.

Since strains from different host origins have different alleles at one of the SCC loci (cf. Garber et al. 1975), and strains with the pink allele and strains with the SCC yellow allele were mixed in a 1:1 ratio, the resulting teliospores can be either pink, pink/yellow, or yellow with the null hypothesis of a 1:2:1 segregation. To determine the SCC phenotype of each of the infected flowers, a large number of teliospores were transferred with a sterile inoculation loop from the eppendorf cup to a petridish containing standard medium. After one week of growth at 21C, plates that showed both pink and yellow colonies were scored as heterokaryotic. From plates that showed colonies of only one color, 16 single spore colonies were taken and replated on fresh medium. The sporidial colony color was determined after another week of growth by evaluating the color of these 16 colonies. If all 16 colonies were still of the same color, plates were scored as either homokaryotic pink, or homokaryotic yellow.

Note that, following this procedure, heterokaryons originating from a1 S. latifolia strains and a2 S. dioica strains are not distinguishable from the reciprocal heterokaryon type.

T For the conjugation experiment, 16 strains (8 of each mating type) that were isolated from 8 different infected hosts from five allopatric S. latifolia populations and 16 strains (8 of each mating type) that were isolated from 8 different infected hosts from four allopatric S. dioica populations were used. Eight of these 32 strains overlapped with those selected for the previous experiment. This procedure provided 256 (2x8x2x8) possible combinations for conjugation in a complete diallel that were all studied. The procedure that we used for studying the conjugation frequencies between different strains was similar to that described in Kaltz and Shykoff (1999), with the modification that a total volume of 8.0l was used instead of 50l, as our pilot studies showed that using smaller volumes leads to higher conjugation frequencies (data not shown). This way, putative differences in conjugation frequency between different combinations of strains may be more pronounced. Conjugation was studied in three different environments, host extract from S. latifolia, host extract from S. dioica, and sterilized milliQ water. Leaf extracts were used to mimic the different host environments. Studies by Day et al. (1981) and Kaltz and Shykoff (1999) indicate that such extracts evoke the same qualitative conjugation behavior as found when placing sporidia directly on the leaf surface, and that conjugation frequencies in host extracts correspond well with conjugation frequencies in more realistic situations where plant architecture is left intact. Host extracts of S. latifolia and S. dioica were prepared by grinding a few grams of fresh leaf material in a mortar with a pestle after adding 15.0ml/g milliQ water. Host extracts were filtered through a 0.2m fine mesh filter before use. MilliQ water was sterilized before use. Haploid cells of single mating type were cultured for one week on a plate containing standard medium at 21C.

Approximately 1x108 cells were scraped off and suspended in 100l sterile milliQ water. The suspension was mixed thoroughly and cells were counted on a CoulterCounter Z1 (Coulter Electronics Ltd., Luton, England). Samples were diluted to 2.0x108 cells/ml with sterile milliQ water. Samples were let to conjugate in eppendorf cups containing 2.0l a1 cell suspension, 2.0l a2 cell suspension, and 4.0l host extract of either host species, or a similar volume of sterilized milliQ water. The final cell density in the conjugation mixture was 1.0x108 cells/ml. Samples were placed at 14C for 24h. After coloring the samples with 8.0l cotton-blue, conjugation frequencies were determined by counting all conjugated and all non-conjugated cells in 20 small squares of a Cell-Vu disposable counting chamber (Norwell technologies inc., Marietta, GA USA), under a light microscope (at 400x magnification). The mean number of cells ( SE) that was counted (in 1536 samples) was 198.7 1.5. At two times during the experiment, the mating type of all 32 haploid strains was checked, and confirmed. All 256 possible mating combinations between the a1 and a2 strains were made twice (two replicates). For practical reasons, only 48 experimental crosses for all three extracts could be tested simultaneously. Therefore, the complete experiment was spread out over eleven blocks. Crosses and replicates were randomly divided over blocks with the restriction that replicates of the same cross were not allowed to occur within a single block.

T A small subset of combinations was selected for a time series experiment.

These were six mating combinations with the highest, and six combinations with the lowest conjugation frequencies after 24h (calculated over the three treatments combined). Conjugation frequencies in all three environments were determined after 24, 48, 72 and 168h. They were measured twice in two separate blocks.

T x The analyses are based on segregation of the SCC marker in successful infections of the 64 inoculation mixtures. Effects of mix, host species and host sex on the frequency of chromosome copies with S. latifolia origin (marked by the pink allele) per plant was analyzed in a generalized linear model using the SAS procedure GENMOD (SAS v8, The SAS Institute Inc., Cary NC USA). The frequency of heterokaryons (marked by the pink/yellow phenotype) per plant, and the frequency of multiple infections (number of different SCC phenotypes per plant) were analyzed the same way. Differences in latency period between different teliospore types (here, the time between inoculation and the release of teliospores in the first infected flower) were tested within host classes non-parametrically using the Kruskal-Wallis test (procedure NPAR1WAY in SASv8, The SAS Institute, Cary NC USA). To correct for multiple comparisons within a host class, significance levels were adjusted using a Bonferroni correction ( = 0.05 / k, with k=3 when comparing all combinations of homo- and heterotypic teliospore types;

Sokal and Rohlf 1995, p. 703).

T x The data were tested for normality (Shapiro-Wilk test for normality) and for heterogeneity of variances (Levenes test for homogeneity of variances). To improve a normal distribution of the data, and homogeneity in the variances among groups, the conjugation frequencies were angular transformed before data analysis. Transformed data were analyzed in a general linear model using SAS procedure GLM (SASv8, The SAS institute Inc., Cary NC USA). We performed two separate analyses;

in the first analysis the plant origins of the strains were accounted for, whereas in the second analysis crosses were classified into six cross types (outcross;

L1xL2, L1xD2, D1xL2, D1xD2, and self;

L1xL2, D1xD2). In both analyses, effects of host extract and cross type (second analysis and time series) were treated as fixed effects. Effects of strain origin (first analysis), and of replicates were treated as random effects.

The relative success rate of each strain was calculated for the different experiments in the following way;

for strains in the conjugation experiment, the conjugation frequencies of the crosses were first divided by their block means, because of a strong block effect (see results for details). The resulting values were averaged over the three extracts. Finally, in order to normalize the values the relative success rates of individual strains were derived by averaging the values over all combinations they participated in. Note that, due to the unbalanced design of the blocks in the experiment, these estimates of relative success rates have slightly different weights. For strains in the competition experiment, relative success rate was derived by calculating the infection success of individual strains, and dividing these values by 0.5 to normalize the values (0.5 is the initial chance for an individual strain to successfully infect a flower when competing with the other strain of the same mating type in any inoculation mixture). Four strains with a S. latifolia origin and four strains with a S. dioica origin were used in both the competition and the conjugation experiment, thereby directly linking the two experiments together.

T T x From all flowering S. latifolia, 49.7% (n=437) contracted an infection with M.

violaceum. A significantly higher proportion of the S. dioica (62.5% n=355;

p<0.001) was infected. None of the flowering plants in the control group (S. latifolia n=36;

S.

dioica n=38) became infected. Thus, cross infections with spores coming either from outside the greenhouse, or from relatively late detected infected plants within the experiment can safely be neglected.

Segregation of the 8.9 9. Sporidial Colony Color marker in germinated teliospores of the 21. 23. fungal pathogen Microbotryum violaceum from female and male 69.5 67. plants of Silene latifolia and S.

dioica in the competition N = 482 flowers N = 515 flowers experiment. The difference in (104 plants) (120 plants) frequency of the pink allele = Pink between host sexes was = Pink/Yellow significant (p<0.02). Differences between host species, as well as = Yellow 4. in the interaction between host 12. species and host sex, were marginally significant (p<0.06).

30. 42.6 The difference in the frequency 52. 56. of heterozygotes between host sexes was highly significant N = 484 flowers N = 421 flowers (p<0.001), but neither the effect (105 plants) (91 plants) of host species, nor the interaction was significant.

S. latifolia hosts S. dioica hosts Not all of the maximum of 5 flowers (actually 4.6 flowers per S. latifolia, and 4.4 flowers per S. dioica on average) that were taken from a single infected plant were of the same spore type. Hence, we detected multiple infections in single host plants.

Female hosts Male hosts 21.5% of the S. latifolia was infected with two different types and 0.5% with all three types. 17.5% of the S. dioica was infected with two different types and 0.5% with all three types. Since the heterokaryotic L1xD2 type could not be distinguished from the D1xL2 type, these values underestimated the true frequency of multiple infections.

Generalized linear model of the frequency of Silene latifolia-specific chromosome copies (L) and the frequency of heterokaryotic infections (H) of the fungal pathogen Microbotryum violaceum, as a function of inoculation mix, host species and host sex in the competition experiment.

2 Source df ) P-value ) P-value Inoculation mix 63 79.4 =0.08 62.1 n.s.

Host species 1 3.6 <0.06 0.1 n.s.

Host sex 1 8.5 <0.02 11.4 <0. Species * Sex 1 4.0 <0.06 0.2 n.s.

On average, copies of the strains with a S. latifolia origin were significantly more successful than strains with a S. dioica origin on both host species (testing the deviation from 50:50% that is expected under the null-hypothesis of a 1:2: segregation ratio of alleles (one tailed ), overall: 60.2% n=420 p<0.002;

in S. latifolia hosts: 62.6% n=209 p<0.007;

in S. dioica hosts: 57.7% n=211 p<0.057). Table shows the effect of mix, host species and host sex on the frequency of the SCC alleles in a generalized linear model, which is visualized in figure 1. Contrary to what we expected beforehand, there was hardly an effect of host species. Strains performed slightly better on conspecifics of the host species of origin, but the effect was only marginally significant (p<0.06). There was a much stronger effect of host sex. First, the frequency of S. latifolia strain copies was significantly (p<0.02) higher in male hosts than in female hosts. This difference was expressed in significantly (p<0.001) higher frequencies of heterokaryons in female hosts primarily at the expense of the homokaryotic L1L2 class rather than the homokaryotic D1D2 class. A G-test (Sokal and Rohlf 1995, p. 699) showed that in female hosts the ratio of heterokaryons to homokaryons carrying the pink allele was significantly higher than expected on basis of random mating and equal competition success (2:1), at least in S. latifolia hosts (S. latifolia 76.3: 23.7% n=95 p<0.05;

S. dioica 74.4: 25.6% n=108 p<0.10;

both hosts 75.3: 24.7% n=203 p<0.02). There was no significant effect of the randomly composed inoculation mix at the 95% significance level (Table 1, p=0.08). Latency period of the D1D2 teliospores was significantly longer than that of the other two teliospores types except in S. dioica females (Figure 2).

SCC phenotype Mean latency period (in = yellow *** days) of the three = pink/yellow different teliospore = pink SCC phenotypes in different host types.

* The significance of ** differences between ns the SCC phenotypes within a host class is shown (Kruskal Wallis test). Levels of significance after a Bonferroni correction are designated;

ns = not significant;

* = p<0.05;

** = p<0.01;

females males females males *** = p<0.001.

S. latifolia S. dioica Host type Mixed model analyses of variance of angular transformed conjugation frequencies in the fungal pathogen Microbotryum violaceum after 24h as a function of host extract and strain origin (analysis 1), and as a function of host extract and cross type (analysis 2).

Source of variation df (n, d) MS F P-value Analysis Block 10, 758 6188.8 167.9 <0. Origin of mt-a1 [O1] 15, 43 4057.3 21.8 <0. Origin of mt-a2 [O2] 15, 37 871.4 3.0 <0. Extract [E] 2, 48 21504.0 53.8 <0. O1 x O2 225, 471 61.8 2.1 <0. O1 x E 30, 450 159.8 5.4 <0. O2 x E 30, 450 269.5 9.1 <0. O1 x O2 x E 450, 758 29.5 0.8 n.s.

Total error 758 36.9 - Analysis Block 10, 758 6188.8 167.9 <0. Extract [E] 2, 504 21504.0 529.3 <0. Cross type [C] 3, 252 18321.0 139.3 <0. Sporidial combination within Cross type [S(C)] 252, 520 131.9 3.3 <0. E x C 6, 504 946.1 23.3 <0. E x S(C) 504, 758 40.6 1.1 n.s.

Total error 758 36.9 Contrasts between Cross types Between both homotypic cross types LxL vs DxD 1, 252 39526.0 300.5 <0. Between both heterotypic cross types LxD vs DxL 1, 252 13732.0 104.4 <0. Between homo- and heterotypic cross types LxL, DxD vs LxD, DxL 1, 252 1621.6 12.3 <0. Contrast error 252 131.5 - Mean latency period in days (SE) T x In the conjugation experiment, conjugation frequency after 24h was determined between strains from the two host species in different environments. Two separate analyses were performed, one in which strain origin was accounted for, and the other in which variation due to strain origin was repartitioned in cross type. In both the analyses, the effect of extract type was highly significant as is shown in table 2. Figure 3 shows the magnitude of this difference. Conjugation frequencies in S. dioica extracts after 24h was on average ten percent (absolute, or 30% relative) higher than conjugation frequency in S. latifolia extract, and 20 percent (absolute, or 79% relative) higher than in the water control. Effects of origins of the a1 and the a2 strains, and their interaction were also significant. Thus, the performance of individual strains did not only depend on their own origin, but also interacted with the origin of the strain of the opposite mating type. This effect was directly analyzed as cross type in the second analysis (Table 2). The selfing cross types (L1xL2 and D1xD2, crossing strains that originated from the same single flower) were not significantly different from the outcrossing L1xL2 and D1xD2 types, and were therefore pooled.

0. = S. dioica extract Mean = S. latifolia extract conjugation frequency of Microbotryum violaceum = Sterilized milliQ water a strains in Silene dioica ab leaf extract, S. latifolia 0.50 bc leaf extract, and in cd cd sterilized milliQ water d d for the different cross types in the conjugation e e experiment. Columns ef f 0.25 with different letters are significantly different from each other with g p<0.05 (Tukey HSD test).

0. L1xL2 L1xD2 D1xL2 D1xD Cross type Figure 3 shows the mean conjugation frequencies of each cross type for each extract.

In general, the L1xL2 cross type had significantly higher conjugation frequencies than other cross types after 24h, immediately followed by the L1xD2 cross type. The D1xL and D1xD2 cross types had significantly lower conjugation frequencies than the first two types in all extract types, but were only significantly different from each other in Mean conjugation frequency (SE) the water control (Figure 3). Analysis per strain (analysis 1 in Table 2) showed that significantly (p<0.0001) more cells from S. latifolia had conjugated after 24h than cells from S. dioica. This was true for cells of both mating types (overall means SE);

for mating type a1: S. latifolia 42.9 0.6%;

S. dioica 26.0 0.6 and for mating type a2: S. latifolia 36.6 0.7%;

S. dioica 32.3 0.7%).

1. S. dioica extract High selection Low selection S. latifolia extract Water control 0. a ab abc abc ab bcd abc 0. cd abcd bcd bcd d e 0. ef ef fg 0.2 gh fg gh h gh gh gh h 0. 0 24 48 72 96 120 144 168 Time (hours) Time course of mean conjugation frequencies of Microbotryum violaceum strains for six selected high (filled symbols), and six selected low (open symbols) combinations out of the combinations in the conjugation experiment. Squares represent samples with Silene dioica host extract, triangles represent samples with S. latifolia host extract, and circles represent the control samples with sterile milliQ water. Points with different letters are significantly different with p<0. (Tukey HSD test).

T Figure 4 shows the results of the time series experiment in which conjugation frequencies were examined after 24, 48, 72 and 168h, of six high and six low combinations that were selected from the previous experiment. Conjugation frequencies reached their maximum after approximately 72h. However, differences between high and low cross combinations were already present after 24h.

Significant differences between extracts in this subset of 12 crosses were only found after 72h between the water control and the host extract from S. dioica in the low selection (p<0.0001), and after 168h between the water control and both host extracts Conjugation frequency ( SE) in the low and the high selection (p<0.001), but at no point in time between the extract of S. latifolia and the extract of S. dioica (p>0.48).

Strain origins and relative success rate of strains of the fungal pathogen Microbotryum violaceum in the competition experiment (I), and in the conjugation experiment (II). See text for the definitions of relative success rates. Populations are located in the Netherlands, unless stated in bold face. = Strain was omitted from the experiment due to contamination. # = Reference strain.

Relative success rate per exp. and mating type Strain Origin of host Host Exp. I Exp. II species a1 a2 a1 a Mi24 Millingen S. latifolia 1.26 1.19 - - Mi30-1 Millingen S. latifolia 1.20 1.31 1.26 1. Mi33-1 Millingen S. latifolia 1.30 1.15 - - Mi44-1 Millingen S. latifolia 1.30 1.17 - - Mi28 Millingen S. latifolia 1.32 1.26 1.24 0. Bij1 Bijland S. latifolia 1.34 1.11 1.25 1. Bij2 Bijland S. latifolia 1.09 1.21 1.37 1. 023-a Wolfheze S. latifolia 0.96 0.82 - - 023-d Wolfheze S. latifolia - - 1.37 0. 016-4 Wolfheze S. latifolia - - 1.39 1. 097-a Goedereede S. latifolia 1.30 1.13 - - 097-b Goedereede S. latifolia 1.29 1.34 - - 104-a Sonniuswijk S. latifolia - - 1.21 1. 112-a Eefde S. latifolia 1.38 1.27 - - 125-a Varssel S. latifolia - - 1.13 0. Sp2-1 Spain S. latifolia 1.00 1.20 - - Means SE S. latifolia 1.23 0.04 1.18 0.04 1.28 0.03 1.07 0. Ca3-1 Castricum S. dioica - - 0.58 0. Ca5-1 Castricum S. dioica 0.76 0.92 0.53 0. Ca7-1 Castricum S. dioica 0.70 0.76 - - Om10-1 Oude Molen S. dioica 0.53 0.90 0.68 0. Om4-1 Oude Molen S. dioica - - 0.68 0. Om7-1 Oude Molen S. dioica 0.91 0.96 - - Pl1-3 Populierenlaan S. dioica - - 0.77 1. Pl2-2 Populierenlaan S. dioica 0.86 0.76 0.71 0. Pl5-3 Populierenlaan S. dioica 0.69 1.02 - - Bb16-1 Burgvallen S. dioica 0.76 0.85 - - Bb3-1 Burgvallen S. dioica 0.63 0.78 - - Bb8-1 Burgvallen S. dioica - - 1.08 0. Bb9-1 Burgvallen S. dioica 0.80 0.65 0.77 0. Vij1-3 Vijlen S. dioica - 0.66 - - Um23 Ume, Sweden S. dioica 0.87 0.78 - - Vs2a-1 USA # S. dioica 0.76 0.82 - - Means SE S. dioica 0.75 0.03 0.82 0.03 0.73 0.06 0.93 0. Table 3 shows the origins of all fungal strains used in the experiments, and the calculated relative success rate of each strain for both mating types. Since four of the strains from S. latifolia and four of the strains from S. dioica were used in both experiments, relative conjugation success after 24h and relative success in competition could be compared directly, as is presented in figure 5. There was a significantly positive correlation between the relative success of strains in the conjugation experiment, and the corresponding strains in the competition experiment.

1. a r2 = 0.70** a2 from S. latifolia a r2 = 0.75** a2 from S. dioica 1. 0. 0.5 1.0 1. Relative conjugation success after 24h Relative success rate of strains of the fungal pathogen Microbotryum violaceum of mating types a1 (filled symbols) and a2 (open symbols) that were used in both the conjugation experiment and the competition experiment. Strains in the upper right corner of the graph were sampled from Silene latifolia (triangles), strains in the lower left corner were sampled from S. dioica (circles). De slopes of the a1 and a2 regression lines do not significantly differ from each other (ANCOVA, p>0.33;

Sokal and Rohlf 1995, p. 499). The combined linear regression function of both mating types together is:

y=0.84x+0.16 (r2 = 0.71;

p<0.001;

not plotted in the figure).

Strains in the upper right corner of the graph perform better than average in both experiments. Likewise, strains in the lower left corner perform worse than average.

All eight strains from S. latifolia appear in the upper right group, while all eight strains from S. dioica are in the lower left group. The slopes between the regression lines for mating type a1 and mating type a2 were not significantly different (ANCOVA p>0.33;

Sokal and Rohlf 1995 p. 499). The regression line for the combined strains is y=0.84x+0.16 (r2=0.71;

p<0.001). This line is not significantly different from the identity line (y=x), which can be plotted within the 95% confidence interval of the regression line (data not shown).

Relative competition success I I Day (1980) was among the first authors to mention multiple infections of M.

violaceum in S. latifolia (before the taxonomical debates in the 1990s known as Ustilago violacea and S. alba). In the competition experiment, examining at maximum five flowers per infected host plant, we could detect that overall 20% of the plants were infected with more than one type of dikaryon, and 1% was infected with all three types. Since we could not distinguish between both heterokaryon types the overall value is most likely underestimated with circa 4% (by simple extrapolation, this is the average of the classes that could be detected). With the two-allele variation at one of the SCC loci in our strains, we could also not detect multiple infections of single flowers like Day occasionally observed in his classic study. The extent of multiple infections in natural populations is poorly studied. If they would occur in natural host populations as well, this would increase the opportunities for outcrossing, because it enhances the chance that non-related teliospores are picked up by a vector on the same host plant, and are transmitted simultaneously to a healthy susceptible host.

In contrast to studies of multiple infections of animals, mainly found in medical literature, ecological studies on multiple infections of single hosts by parasites are scarce, let alone examples involving plants and phytopathogenic fungi. Given the importance of the presence or absence of multiple infections in models of the evolution of virulence (e.g. Van Balen and Sabelis 1995), this is very unfortunate. The few empirical studies that do exist, have investigated mostly viral and bacterial pathogens, e.g. Wolbachia in European raspberry beetles (Malloch et al. 2000), or pathogenic nematodes in fig wasps (Herre 1993).

x In the conjugation experiment, the highest proportion of conjugates after 24h was observed in crosses between strains that both originated from S. latifolia in each of the host extracts. Significantly lower proportions were observed in the interracial crosses, and in crosses between strains from S. dioica, respectively. We conclude that strains from S. latifolia conjugate faster than strains from S. dioica. Furthermore, there was a significant difference between the reciprocal interspecific crosses, with the L1xD2 crosses showing higher conjugation frequencies than D1xL2 crosses. In host extract, L1xD2 resembles L1xL2, whereas D1xL2 resembles D1xD2, suggesting a predominant role of the a1 mating type in conjugation success. Conjugation tubes grow in a directed manner from cells with mating type a2 towards cells with mating type a1 (Day 1976), as a reaction to the more active signaling of mating type a1 (Day et al. 1981). Presumably, cells with mating type a1 are more active in signaling because they have a slower rate of degradation of pheromones that are produced by both mating types, at least in Ustilago maydis, a related smut from Maize (Snetselaar et al. 1996). Indeed, if mating type a1 is more active in signaling, this may cause reciprocal differences in crosses between strains, i.e. mating type a1 more strongly determining conjugation rate than mating type a2.

Generally, conjugation frequencies in the presence of host extracts were significantly higher than in water, but differences between cross types were almost independent of the type of medium used (Figure 3). Contrary to earlier studies (Ruddat and Kokontis 1988 and references therein), we did not observe any hyphal growth induced by the presence of host extract, not in the large 24h experiment, nor in the 168h time series using pre-selected crosses. Presumably our extracts were either more diluted, or other environmental conditions in our assay were simply unsuitable for hyphal growth. The extract from S. dioica leaves induced significantly more conjugated cells than the extract from S. latifolia leaves. Assuming, like other studies (Kaltz and Shykoff 1999), that host extracts indeed sufficiently mimic the relevant cues experienced during conjugation in vivo, this suggests that mating of these strains may be more stimulated by S. dioica than by S. latifolia (Figure 3).

The time series showed that conjugation of strains was consistent;

differences in conjugation frequency between the selected high and low combinations could be repeated. Moreover, the results showed that the conjugation process still progressed after 24h, until about 72h. At, and after this point in time, differences between crosses were even more pronounced. This is in contrast to a study by Poon and colleagues (Poon et al. 1974) on the kinetics of mating in this fungus, who observed that 70% of the cells had mated after 20h. In our study, this point is reached only in the selected high crosses, and not before 72h. This indicates that the conditions for conjugation in our assay were sub optimal. Being one of the most critical factors in mating (Cummins and Day 1977), supply of oxygen was probably sub optimal in the 8l volumes at the bottom of the eppendorf cups.

In general, strains from S. latifolia were more successful in producing infections than their S. dioica counterparts in the competition experiment in vivo. The conjugation experiment actually represents in detail the initial stages of the competition experiment, in which the infectious dikaryons were produced. From the combined success rates displayed in figure 5, it can be seen that a large part of the infection success can be explained by the swiftness of mating. Thus, although Day (1980) rightly suggests;

competition may be expressed at the level of growth rate and the ability to keep up with the meristem and/or in the ability to compete for the available space and nutrients in the meristem, the outcome of competition is initially highly dependent on the time it takes to produce an infectious dikaryon, giving some combinations a decisive head start over other combinations simply by conjugating faster. Interestingly, latency period of the infections involving S. latifolia strains was much shorter than infections with only S. dioica strains in the competition experiment (Figure 2), which is consistent with this conclusion, and was found in a cross inoculation study by Biere and Honders (1996a) as well. Contrary to our results however, heterotypic combinations in their study showed latency periods that were more comparable to the homotypic S. dioica isolates. With a flower-longevity of a few days at most, time for mating, conjugation, producing an infectious dikaryon and entering host tissue is short, especially in male hosts, and conjugation rate is probably an important aspect of infection success. Male plants of S. latifolia have been reported to actively drop flowers that contained spores, presumably to avoid becoming infected, shortening the longevity of these flowers with on average 10h (Kaltz and Shykoff 2001). In contrast, female flowers remain much longer on the plant when they are pollinated, in order to produce seed capsules and seeds.

I We found no (positive) assortative mating between the host races. On the contrary, if the supposed gene flow would be large enough, based on its low competitive ability, we would expect the host race of S. dioica to decline or go extinct in sympatric populations of these host species. Since both host races do occur in natural sympatric populations of hosts (Biere and Honders 1996b;

Van Putten et al.

chapter 2;

chapter 3), the existence of large differences in conjugation rate, infection success and latency period between both host races, presumably representing large fitness differences, are even more puzzling.

Two factors could contribute to the maintenance of host race differentiation in sympatric host populations. First, we observed a highly significant effect of host sex on dikaryon type. Heterotypic dikaryons were found to be the better competitors in female host plants, at least in S. latifolia. The fact that this apparent overdominance at one of the SCC loci is primarily found in female hosts might be explained by the more complex requirements for successful infection of female hosts as compared to male hosts. In dioecious Silene sex is genetically determined (Ono 1939;

Westergaard 1940). The infection of a female host requires a morphological sex change by fungal induction of male specific gene expression in the developing flower (Scutt et al.

1997). Subsequently, the development of pollen grains has to be prevented by blocking the formation of microspores (Audran and Batcho 1981). Being more complex and presumably more costly to the fungus, complementation at heterozygote loci may be more important in the infection of female hosts. If this is the case, female hosts could provide a possible safe haven for strains from S. dioica in sympatric populations of hosts that might otherwise go extinct. However, the predominantly selfing nature of this fungus (Baird and Garber 1979), resulting in strong observed homozygosity (Bucheli and Shykoff 2001;

Van Putten et al. chapter 2) opposes this thought. Interestingly in this respect, the production of teliospores per plant in male hosts was found to be significantly higher in conspecifics than in heterospecifics of the host of origin in a cross inoculation experiment, but not in female hosts (Biere and Honders 1996a), which held true both in S. dioica and S. latifolia.

Second, the natural pollinators of the host plants transmit the fungal spores, and the existence of different pollinator guilds for these two host species (e.g. Jrgens et al. 1996;

Goulson and Jerrim 1997) could thus reduce fungal gene flow. If pollinators that serve as vectors of smut spores, can prevent, or strongly limit the possibilities for host competition and outcrossing simply by being choosy in host sympatry, both host races might coexist. We have studied the behavior of vectors in artificial, completely mixed plots of S. latifolia and S. dioica in detail using fluorescent dyes as spore mimic (Van Putten et al. chapter 5). Indeed, results showed that the different pollinator guilds are choosy in their visitation patterns with respect to host species.

However, in these experiments there was still a considerable amount of interspecific visitation. Nevertheless, since natural sympatric populations of hosts are spatially (Goulson and Jerrim 1997;

Van Putten et al. chapter 3) and temporally (Biere and Honders 1996b) much more heterogeneous, the exchange of fluorescent dye between host species that we observed (approximately 30% of total visits), is likely to be an estimate of the absolute maximum exchange possible. Fungal gene flow in natural sympatric populations of both host species is expected to be considerably lower.

Therefore, we expect that host fidelity of vectors will strongly contribute to the reproductive isolation needed to explain the observed host-related differentiation between fungal isolates from different host species.

The authors wish to thank Oliver Kaltz for useful hints and discussions on methodological issues concerning the conjugation experiment, Hans Peter Koelewijn for help with data analysis, and Sonja Honders for greenhouse assistance. This study was financially supported by the Earth and Life Science Foundation of the Netherlands Organization for Scientific Research (NWO-ALW);

grant 805 36-391.

T with Arjen Biere, Jelmer Elzinga and Jos van Damme submitted to Oecologia We have studied conspecific and heterospecific visitation patterns of the pollinators of Silene dioica and S. latifolia in experimental, fully mixed plots of these plant species, using fluorescent dyes. Dye particles were used as mimics of teliospores of the pollinator-transmitted fungal pathogen Microbotryum violaceum, to estimate the amount of gene flow between the different host races of this fungus from these host species in complete host sympatry.

The two host species were visited by different pollinator guilds, with bumblebees preferentially visiting S. dioica diurnally and noctuid moths preferentially visiting S. latifolia nocturnally. After 24h, we observed a mean rate of interspecific transfer of 26% from S.

latifolia to S. dioica. From S. dioica to S. dioica interspecific transfer was 34%. These estimates probably represent the absolute maximum of interspecific visitation between these host species, since natural sympatric populations of these host species have found to be spatially and temporally more heterogeneous. Therefore, the observed visitation pattern of pollinators/vectors, in combination with spatial and temporal separation of the host species, might contribute to the maintenance of genetically differentiated host races of the anther smut M. violaceum as observed in sympatric and parapatric populations of these host species. Male hosts were found to be preferentially visited over female hosts.

In addition, non-linear regression analysis suggested that the range in which the teliospores can be transmitted is probably much larger (20-50+m) than the actual infection range (not much larger than 12-13m) of this venereal disease within a single flowering season.

I T TI Host fidelity can play an important role in sympatric host race formation by providing a mechanism for prezygotic reproductive isolation, for instance in phytophagous insects (cf. Berlocher 1998a). In addition to this, to achieve or maintain host races, it has been argued that host-related fitness trade-offs, i.e. by antagonistic pleiotropy of genes involved in pathogen performance, are needed to overcome any leakiness in host fidelity (Feder 1998). However, empirical evidence for such genetic correlations across different host species is often found to be ambiguous, or non-negative in studies of phytophagous insects (cf. Fry 1996;

but see Via et al.

2000), and is scant for other organisms, including the group of phytopathogenic fungi.

Although fungi are usually dispersed by wind, many phytopathogenic fungi are vectored by insects (e.g. by bark beetles in the Dutch elm disease: Ingold 1971;

by shining flower beetles in the floral smut Anthracoidea: Ingvarsson and Ericson 1998), and some of them can even manipulate their host plants to attract insect vectors that promote their own dispersal (Roy 1994;

Pfunder and Roy 2000). In these cases, host fidelity is regulated by the vectors rather than by the pathogen itself, which obviously decouples possible direct linkages between genetic trade-offs of pathogen performance on different host species and host fidelity.

A well-studied example of a pollinator-transmitted disease is the anther smut Microbotryum violaceum, which has a wide host range within the Caryophyllaceae (Thrall et al. 1993). Teliospores of anther smuts, the stage in which this fungus is dispersed (see Piepenbring et al. 1998 for a review) are produced in the anthers of host flowers and are transmitted by the insect visitors of their host plants that serve the dual role as pollinators and vectors of this disease (Baker 1947;

Hassan and MacDonald 1971;

Jennersten 1983;

Alexander and Antonovics 1995). Two of its host species, Silene dioica and S. latifolia, frequently meet in sympatry (Goulson and Jerrim 1997).

S. latifolia, the white campion, is a dioecious, short-lived perennial weed from open, disturbed habitats such as roadsides and arable land with a typical moth-pollination syndrome (Baker 1961;

Baker and Hurd 1968), including heavily scented white flowers that open at dusk (Jrgens et al. 2001). S. dioica, the red campion, is a closely related dioecious perennial that mainly occurs in open woodland, and that is primarily pollinated by bumblebees (Kay et al. 1984). Flowers open at dawn and remain open during the day (personal observation). Also, flowers of diseased plants of S. latifolia emit scent in the evening, and they are still visited by moths (Baker 1947), although they may be less attractive (Shykoff and Bucheli 1995;

Altizer et al. 1998). Visitors of S. dioica also visit infected flowers, and have found to be carrying teliospores (Hassan and MacDonald 1971).

The presence of interspecific hybrids in areas where their preferred habitats are adjacent or intermixed indicates that there is gene flow between these plant host species, although the extent of this has never been studied in much detail (but see Goulson and Jerrim 1997). Likewise, the amount of fungal gene flow between anther smuts from different host species in sympatric populations of hosts is yet unknown.

Early cross-inoculation studies of this fungus have shown that anther smuts from different host species can be grouped into distinct host races (Zillig 1921). Recent studies have demonstrated that smut isolates from allopatric populations of S. dioica and S. latifolia were differentiated for a number of genetic markers;

karyotypes (Perlin 1996;

Perlin et al. 1997);

a sporidial colony color marker (SCC (cf. Garber et al. 1975) Biere and Honders 1996a;

Van Putten et al. chapter 3);

random amplified polymorphic DNA (RAPDs;

Biere and Honders, unpublished results);

and microsatellite loci (Bucheli et al. 2001;

Van Putten et al. chapter 2). Microsatellite analysis of smut isolates from sympatric and parapatric populations of S. latifolia and S. dioica showed that fungal isolates were genetically differentiated in a host-specific manner in parapatric and parapatric/sympatric populations, but not in one true sympatric population where both host species grew truly intermingled (Van Putten et al. chapter 2). This suggested that the degree of sympatry is important in maintaining the genetic differentiation. However, even in this true sympatric population, the alleles from two of the four loci were not distributed homogeneously over the population (Van Putten et al. chapter 3). At the same time, there was significant host-related differentiation in a sporidial colony color marker, suggesting that there could be strong selection on this locus, and/or that the patchy local structure of hosts in this population could severely limit the amount of gene flow between the host races. A question that arises is how such host-related genetic variation could be maintained in sympatry. One option is that there are trade-offs in performance between different fungal isolates on the different host species that contribute to the maintenance of host related genetic variation. In a cross inoculation experiment by Biere and Honders (1996a), they showed that the production of teliospores per plant was significantly higher in conspecifics than in heterospecifics of the host of origin, but only in male hosts. This trade-off in performance in male hosts will indeed contribute to some extent to the maintenance of host-specific variation. However, since they found no host-related differences in virulence (sensu Jarosz and Davelos 1995) between fungal isolates from the two host species, this trade-off is presumably not strong enough by itself to entirely explain the observed host-specific genetic variation in sympatry.

Also, since a competition experiment has shown that the host race of S. latifolia outcompetes the host race from S. dioica, presumably due to its higher conjugation rates (Van Putten et al. chapter 4), we need to look for other explanatory mechanisms.

As this fungus is vectored by pollinators of their host, gene flow between the two host races could highly depend on the visitation behavior of pollinators that visit both host species in order to accomplish substantial interspecific transfer of smut spores. This raises the question, whether host fidelity of the pollinator guilds of S. dioica and S.

latifolia can be an additional factor contributing to the maintenance of these host races in sympatry.

In this chapter we study patterns of spore transfer between and within the host species S. latifolia and S. dioica. For this purpose, we use fluorescent dyes as traceable teliospore surrogates to explore the visitation patterns of the pollinator community in an artificial sympatric setup of S. latifolia and S. dioica host plants. Since we expected that different pollinator guilds would be active during day- and nighttime (e.g.

Groman and Pellmyr 1999), different experiments were carried out, covering either 24h visitation or daytime only. Furthermore, since patch size may also be an important factor (e.g. Sowig 1989), experiments with different patch sizes were carried out as well. Additionally, to connect the dispersal of fluorescent dye directly to the dispersal of the anther smut disease, the infection rate in relation to the linear distance to a true teliospore source was examined for one of the host species, S. latifolia. Specific questions that will be addressed here are: (1) Do different guilds of pollinators discriminate between S. dioica and S. latifolia in an artificial, completely mixed setup, and show host fidelity? (2) What is the frequency of interspecific visitation between both host species, i.e. to what extent would an assortative visitation pattern with respect to host species provide a basis for the maintenance of host races of M.

violaceum in sympatric populations of hosts? (3) What is the range of teliospore (fluorescent dye) dispersal after 24h, and of realized infections of M. violaceum from a true smut source within a single flowering season?

T I T Silene latifolia Poiret (= S. alba (Miller) Krause (Caryophyllaceae), the white campion, is a dioecious short-lived perennial from open, disturbed habitats and borders of arable land. Silene dioica (L.) Clairv. (Caryophyllaceae), the red campion, is a closely related dioecious perennial that mainly occurs in more shady humid habitats as woodland borders. In areas where habitats overlap, both species frequently occur in sympatry and hybridization is a common phenomenon (Baker 1947;

Goulson and Jerrim 1997).

The anther smut fungus M. violaceum (Pers.: Pers) Deml & Oberw. (=Ustilago violacea (Pers.) Fuckel) (Ustilaginaceae) (Vnky 1994) is a vector-borne venereal disease that sterilizes its Caryophyllaceous hosts (Thrall et al. 1993). Smut spores are produced in the anthers of the host and are transmitted by the natural pollinators of their hosts (Jennersten 1983). In dioecious host species, the ovaries of female plants are reduced and staminal rudiments develop into stamens and anthers that also contain smut spores instead of pollen.

Seeds of S. latifolia and S. dioica were collected from several geographically spread natural allopatric populations in the Netherlands in 1997 and 1998. Seeds were germinated in petridishes on demi-water moistened filter paper at a density of approximately 25 seeds per petridish in a growth cabinet (16/8h light/dark, 21/15C day/night temperature) after a vernalization period of three days at 4C that has proven to be sufficient to increase the proportion of flowering plants in previous experiments. Nearly all seed had germinated after a week. For experiments 1-4, in total 750 S. dioica and 500 S. latifolia were potted after two weeks from vernalization in round containers (12cm), and grown in a acclimatized greenhouse at 21C and with 16/8h light/dark until flowering. For experiment 5, an additional 1550 S. latifolia were potted in 18 cm square containers. From these plants, 200 were inoculated with a suspension of M. violaceum conjugates in the seedling stage prior to potting. These conjugates were derived by mixing haploid sporidia of both mating types in an aqueous suspension at 14C for several hours (Cummins and Day 1977).

T In this study we assess the assortative visitation patterns of pollinator guilds visiting our sympatric plot of S. latifolia and S. dioica using fluorescent dyes. The occurrence of dye on a flower is a qualitative measurement of the footsteps of both pollinators and vectors, that has successfully been used to trace pollen movements across plants (e.g. Thomson 1981;

Waser and Price 1982;

Fenster et al. 1996;

Goulson and Jerrim, 1997), and has proven to be a good predictor of the dispersal of fungal spores as well (Shykoff and Bucheli 1995). However, as Thomson et al. (1986) rightly points out, using fluorescent dye as pollen analogue and simply scoring the presence/absence of dye on a flower surely would overestimate the extent of dispersal because stigmas are often more receptive to the much smaller dye particles than they are to pollen grains. Fortunately, whereas pollen of S. latifolia is size-ranged 35- m (Prentice et al. 1984), teliospores are much smaller as well, and range from 6-9m (Zogg 1985). Therefore, spores may remain longer on a pollinator than do pollen, as was found in the comparative study to the transfer of pollen and fluorescent dye (Thomson et al. 1986). Fluorescent dye particles are equal in size compared to M.

violaceum teliospores, and may therefore be better spore-analogues than they are pollen-analogues. However, dye particles are more irregular in shape than teliospores and may be more sticky (personal observation).

A: patch size = 1 B: patch size = S S O 1mO O O O O O O 1m O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O S O O O O S O O O O Schematic set-up of experiments 1-4, which represents plots of 8x8 large containers that each contains one female and one male plant of Silene latifolia (open circles), or S. dioica (filled circles). Left panel (A) shows the situation when patch size = 1 (DN-P1 and D-P1), right panel (B) represents the situation when patch size = 2 (DN-P2 and D-P2). S = Source of fluorescent dye. The triangles represent the distance between the center points of two adjacent containers, which is 1m. The North-South direction is top-to-bottom in the figure.

x 64 containers, each holding one male and one female flowering plant of the same species were placed in a grid of 8 by 8 (Figure 1), so that all neighboring containers were at a distance of 1m. The plants at both corners of one side of the plot served as source plants. A few milligrams of fluorescent dye (Radiant technologies Inc., Richmond CA, USA) was applied with a toothpick to ten open male flowers on the anthers and filaments, and to eight open female flowers on the pistils (because most female flowers bore less than ten flowers). Red fluorescent dye was applied to S.

dioica and used as M. violaceum teliospore mimic of S. dioica origin in one corner of the grid, while yellow fluorescent dye was applied to S. latifolia and used as M.

violaceum teliospore mimic of S. latifolia origin in the other corner. Both fluorescent dye colors were simultaneously present within a single replicate. Experiments 1- were carried out sequentially, separated in time throughout the natural flowering season of the plants. Each experiment was replicated four times (two true replicates, and two replicates with both sources mirrored in the north-south axis to correct for potential dominant wind influences). The experimental setup of experiments 1-2 is schematically represented in panel A of figure 1, and the experimental setup of experiments 3-4 in panel B of figure 1. The four experiments differed in their duration (diurnal versus diurnal plus nocturnal visitation) and in the patch size of host plants of the same species, and were set up sequentially during the season (see below).

x y 20 June-29 June 2000: Containers with S. latifolia and S. dioica plants were alternated in a checkerboard pattern (patch size = 1x1). Fluorescent dye was applied just before sunset. Dye transmissions were determined using a portable UV-light (UVP Inc., San Gabriel CA, USA) after approximately 24h (after sunset) by counting per plant all clean flowers, all flowers with only red or yellow dye, and all flowers with both red and yellow dye.

x y 18 July-31 July 2000: Containers with S. latifolia and S. dioica plants are placed in patches of 2x2 containers with the same plant species (alternating S. latifolia patches and S. dioica patches). Fluorescent dye was applied just before sunset. Dye transmissions were determined after approximately 24h (after sunset).

x y n y 31 August-13 September 2000: As experiment 1, but now with fluorescent dye applied just before sunrise. Source plants were removed at sunset of the same day.

Dye transmissions were determined after sunset. For this setup there are only two (mirrored) replicates due to unfavorable weather conditions at the end of the season.

x y n y 9 August-24 August 2000: As experiment 2, but now with fluorescent dye applied just before sunrise. Source plants were removed at sunset of the same day.

Dye transmissions were determined after sunset.

During experiments 1 and 2, on ten separate occasions from about 23:00 till about 1:00, all nightly insect visitors to flowers in the plot were caught, identified and checked for the presence of fluorescent dye, and released again afterwards. During experiments 3 and 4, on six separate occasions in total 215 minutes daily visitors to flowers in the plot were observed and identified. A few of these visitors were also caught and checked for the presence of fluorescent dye. For each experimental day (and night), day length was determined and mean day and night values for temperature, wind, sun radiation and rain were recorded.

0m Source of 180 inoculated S. latifolia [0-1m] Schematic set-up of experiment 5, representing seven blocks, of which six 2m consisted of 225 healthy (not inoculated) Silene latifolia and one consisted of 180 S. latifolia, 225 healthy S. latifolia [2-3m] that were inoculated with conjugates of Microbotryum violaceum in the seedling stage 4m prior to potting (of which more than 42% of the 225 healthy S. latifolia [4-5m] flowering plants became infected). The North South direction is left-to-right in the figure.

6m 225 healthy S. latifolia [6-7m] 8m 225 healthy S. latifolia [8-9m] 10m 225 healthy S. latifolia [10-11m] 12m 225 healthy S. latifolia [12-13m] 14m x In experiments 1-4, fluorescent dyes were used as mimic of teliospores. Since we were interested in the fungal disease rather than in fluorescent dyes, we wanted to link dispersal distance of dye in these experiments directly to the dispersal of a true infection with anther smut, using the pollinators guilds from the same area. Therefore, in addition to experiments 1-4, an experiment was carried out to investigate the infection rate as a function of distance to a true teliospore source, for one host species, S. latifolia. Containers were placed in seven rectangular blocks at regular distance intervals of 2m (figure 2). The first block contained 180 (placed in four rows of plants) inoculated S. latifolia hosts, serving as a source of teliospores. The other six blocks contained 225 (placed in five rows of 45 plants) S. latifolia hosts that were not inoculated, serving as recipient host plants. The experiment was set up near the experimental plots of experiments 1-4 (approx. 60m), and started at the end of June 2000. The bulk of plants started flowering at the beginning of August 2000. At the end of October 2000, all flowering plants were checked for sex and disease status.

Experiment 1-4. Each of the four experiments was analyzed separately.

Distances from the source of the fluorescent dye were calculated from the center of the source container to the centers of the receptive containers, neglecting actual positions of plants and flowers inside the container. Analyzed were visitation frequencies (defined as the ratio of number of flowers with dye divided by total flowers per plant) in a logistic regression with backward elimination of the higher order interaction effects until the final model was reached (SAS v8 procedure LOGISTIC, The SAS Institute Inc. Cary, NC USA). Plant species (SP), plant sex (SX) and replicate (R) are categorical factors, distances to the red (Dr) and yellow (Dy) fluorescent dye sources were regarded as continuous covariates. The fitted models were corrected for overdispersion using the Williams correction (Williams 1982).

Overdispersion was tested for by dividing the Pearson 2, and the deviance by their degrees of freedom, which both proved to be significantly larger than 1 in the unscaled model (cf. Stokes et al. 2000). The effects of host species and host sex on the interspecific visitation rate after 24h, which is defined as the proportion of interspecific visits divided by total number of visits, were tested in a generalized linear model (SAS v8 procedure GENMOD). In experiment 5, the effects of block, i.e.

distance from smut source plants, and host sex on the frequency of infection were analyzed using a generalized linear model (SAS v8 procedure GENMOD). To gain statistical power for a contrast analysis, the six recipient blocks were merged into three new blocks: 2-5m, 6-9m and 10-13m. Contrasts were then made between each pair of adjacent blocks.

T x Figure 3 shows the mean frequency of flowers with fluorescent dye per plant in relation to the distance to the source plants where the dye was applied. The statistical support for effects that are displayed in this figure comes from the results of the logistic regression analyses shown in table 1. The lines through the replicate means were fitted using an exponential law y=ae-bx, which is a commonly used model to describe proportions y that decrease with increasing distance x (see Gregory 1968;

Jeger 1990), and details of the fitted functions are shown in table 2. An overall effect in all of the experiments is that the frequencies of fluorescent dye, both red and yellow, significantly decrease with increasing distance from source plants (p<0.0001), and hence the frequency of plants with no dye significantly increases with distance (p<0.0001). More interesting is the strong host species effect in experiments DN-P1, DN-P2 and D-P2, i.e. the frequency of red fluorescent dye originating from S. dioica is significantly higher on S. dioica than on S. latifolia while the frequency of yellow fluorescent dye originating from S. latifolia is significantly higher on S. latifolia than on S. dioica (p<0.0001). This indicates that insect visitors coming from the S. dioica source plants tend to visit S. dioica more frequently than they visit S. latifolia, and vice versa. In other words, there is a significant host species-specific visitation pattern of pollinators/vectors in this artificially mixed sympatric plot. The effect of host species seems to hold true across the whole length of the plot, since the interactions between host species and distance to source were in most cases not significant. In experiment D-P1 however, the species effect is observed only for the yellow fluorescent dye (p<0.05), and since the interaction between host species and distance to source was highly significant in this case (p<0.0001), this also varied across the plot.

1. = S. dioica r2 = 0.69**** = S. dioica r2 = 0.82**** AB O = S. latifolia r2 = 0.84**** O = S. latifolia r2 = 0.78**** 0. 0. 0. 0. 1. = S. dioica r2 = 0.61*** = S. dioica r2 = 0.84**** C D O = S. latifolia r2 = 0.76**** O = S. latifolia r2 = 0.79**** 0. 0. 0. 0. 1. = S. dioica r2 = 0.58*** = S. dioica r2 = 0.93**** E F O = S. latifolia r2 = 0.78**** O = S. latifolia r2 = 0.63*** 0. 0. 0. 0. 1. = S. dioica r2 = 0.73**** = S. dioica r2 = 0.94**** GH O = S. latifolia r2 = 0.81**** O = S. latifolia r2 = 0.83**** 0. 0. 0. 0. 0 2.5 5.0 7.5 10 0 2.5 5.0 7.5 Distance to fluorescent dye source plants (m) S. dioica source S. latifolia source Inter- and intraspecific movements of fluorescent dye in experiment 1-4. Plotted are the mean frequencies of flowers with fluorescent dye per plant in relation to the distance to the source plant. Open circles represent Silene latifolia;

closed circles represent S. dioica. Left panels (A, C, E and G) represent red dye from a S. dioica source plant. Right panels (B, D, F and H) represent yellow dye from a S. latifolia source plant. Panels A-B, C-D, E-F and G-H represent experiments 1, 2, 3 and 4 respectively. Non-linear regression curves fit an exponential model of the form y = ae-bx (Gregory 1968;

Jeger 1990) where y is the dye frequency, x the distance to the source plant, and a & b are constants with a 1. All regression coefficients were highly significant with p<0.0001.

Patch size = 24h [DN-P1] Patch size = 24h [DN-P2] daytime [D-P1] Patch size = daytime [D-P2] Patch size = Frequency of flowers per plant with fluorescent dye Logistic regression model of the frequencies of flowers per plant with either red, yellow, both, or no fluorescent dye in experiments 1-4, with backwards elimination of higher order interactions until the final model was reached. Host species (SP), host sex (SX) and replicate were categorical factors, distances to the red (Dr) and yellow (Dy) fluorescent dye sources were taken as continuous covariates. Rows with eliminated (E) and/or non-significant effects only, as well as all interactions with replicate were omitted from the table for reasons of clarity, but did appear in some of the models. A (-) indicates that these factors were not in the final model. The indicated levels of significance are: # = p<0.10;

* = p<0.05;

** = p<0.005;

*** = p<0.0005;

**** = p<0.0001.

per analysis of dye frequency Exp. Source of variation df Red Yellow Red and No dye Yellow 1 Host species [SP] 1 19.6 **** 27.2 **** 0.0 26.3 **** Day- and Host sex [SX] 1 8.9 ** 2.8 # 5.4 * 4.7 * nighttime Distance red source [Dr] 1 57.7 **** - 0.0 21.2 **** visitation Dist. yellow source [Dy] 1 - 69.5 **** 11.0 ** 21.3 **** (24 h) [SP]*[Dr] 1 0.1 - 6.2 * E Patch size [SP]*[Dy] 1 - E 4.1 * E = 1 [SX]*[Dr] 1 3.0 # - E E [DN-P1] [SX]*[Dy] 1 - 3.2 # E E Replicate 3 12.0 ** 8.6 * 15.8 ** 9.8 * 2 Host species [SP] 1 22.2 **** 23.0 **** 1.8 0. Day- and Host sex [SX] 1 0.1 66.0 **** 39.9 **** 109.3 **** nighttime Distance red source [Dr] 1 88.7 **** - 0.0 48.9 **** visitation Dist. yellow source [Dy] 1 - 95.3 **** 8.9 ** 14.7 *** (24 h) [SP]*[SX] 1 1.1 3.2 # E 17.2 **** Patch size [SP]*[Dr] 1 4.0 * - 14.6 *** E = 2 [SP]*[Dy] 1 - 0.8 31.7 **** 3.9 * [DN-P2] [SX]*[Dr] 1 5.2 * - E E Replicate 3 2.8 7.1 # 69.9 **** 3. 3 Host species [SP] 1 0.0 5.4 * 1.2 1. Daytime Host sex [SX] 1 6.4 * 2.9 # 0.2 13.5 *** visitation Distance red source [Dr] 1 43.2 **** - 1.2 32.8 **** ( 13h) Dist. yellow source [Dy] 1 - 98.3 **** 35.6 **** 11.9 ** Patch size [SP]*[SX] 1 5.2 * E E 7.3 * = 1 [SP]*[Dy] 1 - 15.4 **** 4.7 * 3.2 # [D-P1] [SX]*[Dr] 1 3.5 # - E 7.4 * Replicate 1 1.2 0.4 13.8 *** 0. 4 Host species [SP] 1 29.4 **** 40.0 **** 0.9 4.1 * Daytime Host sex [SX] 1 1.9 3.5 # 5.8 * 1. visitation Distance red source [Dr] 1 130.3 **** - 5.3 * 103.2 **** ( 14 h) Dist. yellow source [Dy] 1 - 127.6 **** 38.0 **** 83.4 **** Patch size [SP]*[SX] 1 0.0 2.9 # E E = 2 [SP]*[Dy] 1 - E E 2.9 # [D-P2] [SX]*[Dr] 1 3.4 # - E 4.8 * Replicate 3 27.3 **** 70.0 **** 25.0 **** 12.7 * Table 1 shows that there is a frequently significant effect of host sex, which is most apparent for red dye in experiment DN-P1 (p<0.005), and for yellow dye, and red + yellow dye in DN-P2 (both highly significant p<0.0001). Male flowers have a higher chance to receive dye than female flowers (overall a relative higher chance of approx.

12%). However, in experiment D-P1 the effect is reversed, and female hosts receive significantly (p<0.05) more red dye than male hosts.

Proportion of fluorescent dye per plant (y) as a function of distance (x) across the experimental plot. Extrapolated are the visitation frequency at the border of the plot (at 10m) and the distance from the source at which the visitation frequency becomes smaller than 1% (in m) and their 95% confidence intervals. R= red;

Y= yellow.

Exp. Dye Plant f(x): y = ae-bx at 10m (in %) < 1% (in m) color Species (95% CI) (95% CI) a (SE) b (SE) 1 R S. dioica 0.90 (0.09) 0.18 (0.02) 14.5 (6.9 - 28.6) 24.7 (18.3 -35.3) 24h R S. latifolia 0.88 (0.08) 0.46 (0.05) 0.1 (0.0 - 0.3) 9.7 (7.5 - 13.1) Patch 1 Y S. dioica 0.87 (0.08) 0.41 (0.05) 1.5 (0.5 - 4.7) 11.0 (8.5 - 15.0) [DN-P1] Y S. latifolia 0.87 (0.08) 0.26 (0.03) 6.4 (2.9 -13.7) 17.2 (13.3 -23.0) 2 R S. dioica 0.86 (0.06) 0.08 (0.01) 38.4 (25.1 - 57.8) 55.7 (39.9 - 88.2) 24h R S. latifolia 0.88 (0.08) 0.28 (0.03) 5.3 (2.4 - 11.5) 16.0 (12.6 - 21.1) Patch 2 Y S. dioica 0.90 (0.07) 0.37 (0.03) 2.3 (1.0 - 5.0) 12.3 (10.0 - 15.4) [DN-P2] Y S. latifolia 0.87 (0.06) 0.14 (0.01) 22.0 (14.1 - 34.0) 32.6 (25.9 - 43.4) 3 R S. dioica 0.88 (0.09) 0.15 (0.02) 19.5 (9.3 - 39.5) 29.9 (21.2 - 47.3) Daytime R S. latifolia 1.00 (0.08) 0.18 (0.02) 17.1 (9.3 - 30.7) 26.2 (20.2 - 35.8) Patch 1 Y S. dioica 1.00 (0.06) 0.36 (0.03) 2.8 (1.4 - 5.5) 12.9 (10.8 - 15.7) [D-P1] Y S. latifolia 0.87 (0.10) 0.21 (0.03) 10.7 (4.2 - 25.7) 21.4 (15.3 - 32.8) 4 R S. dioica 0.97 (0.08) 0.16 (0.02) 19.5 (10.5 - 35.1) 28.5 (21.5 - 40.2) Daytime R S. latifolia 0.91 (0.07) 0.21 (0.02) 10.7 (6.0 - 18.9) 21.1 (17.0 - 27.2) Patch 2 Y S. dioica 0.99 (0.05) 0.35 (0.02) 3.1 (1.9 - 4.8) 13.3 (11.7 - 15.1) [D-P2] Y S. latifolia 0.94 (0.07) 0.22 (0.02) 10.8 (6.0 - 19.0) 21.1 (17.0 - 27.1) Seasonal variations in weather conditions during the replicates of experiments 1-4. Day length (D, i.e. the time interval between sunset and sundown), mean temperatures (T), mean wind speed (W), mean sun irradiance (S), and rainfall (R) for each day and night of observation.

Date D T (C) W (ms-1) S (Wm-2) R Exp. Rep. in (h:m) day night day night day night (+/-) 1 1 20 Jun. 16:20 31.7 22.9 1.6 0.0 362 12 - [DN-P1] 2 21 Jun. 16:20 24.2 22.0 2.4 0.0 233 13 + 3 28 Jun. 16:17 16.4 9.6 1.9 0.3 307 19 - 4 29 Jun. 16:16 16.9 10.9 2.2 0.5 308 19 - 2 1 18 Jul. 15:47 16.5 13.1 1.9 0.4 244 16 + [DN-P2] 2 20 Jul. 15:42 21.0 12.9 1.7 0.0 330 18 - 3 26 Jul. 15:27 18.9 15.4 1.2 1.0 157 16 - 4 1 Aug. 15:09 26.6 17.3 1.5 0.2 302 15 + 3 1 31 Aug. 13:20 20.2 12.4 0.8 0.5 344 16 - [D-P1] 2 13 Sep. 12:51 19.4 15.3 1.6 0.5 189 14 - 4 1 9 Aug. 14:42 23.2 13.9 0.9 0.0 310 18 - [D-P2] 2 16 Aug. 14:16 22.2 15.1 0.3 0.3 258 16 + 3 22 Aug. 13:53 20.7 14.4 0.4 0.4 280 15 - 4 24 Aug. 13:46 22.2 13.0 1.1 1.1 350 15 - The last overall effect in experiments 1-4 is the frequent occurrence of significant differences between replicates (Table 1), and occasional significant interactions with replicate (interactions with replicate were included in the regression models, but were omitted from table 1 for reasons of clarity). This suggests that the strong variation in weather conditions between consecutive days and nights of observation within an experiment (shown in Table 3) might have a strong impact on the observed visitation patterns. Especially in experiments DN-P1 and D-P2 the main effect of replicate is significant. Table 3 shows that during experiment DN-P1 the average day and night temperatures are varying considerably across replicates. In experiment D-P2 the day length between replicate 1 and 4 has already shortened almost a full hour. Due to bad weather conditions mid August the replicates could not be performed within a few days as was intended.

Mean interspecific transfer of fluorescent dye after 24h. Shown is both the red dye source towards Silene latifolia and the yellow dye source towards S. dioica, for patch sizes 1 and 2.

Designated are the significance levels of reciprocal differences between the two host species: n.s. = not significant;

**** = p<0.0001.

Exp. Dye N Interspecific transfer ( SE) color 1 Red S. latifolia 194 0.346 ( 0.027) n.s.

[DN-P1] Yellow S. dioica 203 0.300 ( 0.023) 2 Red S. latifolia 238 0.339 ( 0.017) **** [DN-P2] Yellow S. dioica 243 0.226 ( 0.014) Table 4 shows that the rate of interspecific transfer of fluorescent dye after 24h was high in these mixed plots, approximately 30%, and turned out to be higher from S.

dioica to S. latifolia than in the other direction, although this difference was only significant in experiment DN-P2 (p<0.0001).

Table 5 shows the results of the observations of insect visitors to the plants in the experimental plot. The pattern that shows up, is that S. latifolia is visited mainly during the night by nocturnal moth species, in particular Hadena bicruris and Autographa gamma that is also active during daytime. S. dioica is mainly visited during daytime by several species of bumblebees (mainly Bombus terrestris, B.

hortorum and B. pascuorum), and hover flies (mainly Rhingia campestris). Nearly a third of the insects that were caught contained huge loads of fluorescent dye. Yellow fluorescent dye was detected on five out of 28 caught nocturnal visitors, while none of these insects contained red dye. Red fluorescent dye was detected on eight out of caught diurnal insect visitors to plant in the plot, of which one insect also contained yellow dye.

Results of direct observations of night- and daytime insect visitors to Silene latifolia and S.

dioica in the plots of experiments 1-4, and the presence/absence of fluorescent dye on caught individuals.

Time of Monitored vector species Visits to Dye on vectors observations (total # caught) Group Latin name N S. latifolia S. dioica 23:001:00 Noctuid Hadena 18 18 - 5 yellow in total moths bicruris (18) 15-20 hours Autographa 9 7 2 spread over gamma (9) 10 nights Plusia sp. 1 1 - - (1) 12:0017:00 Bumblebees Bombus spp. 17 3 87 1 red in total (2) 215 Hoverflies Rhingia 14 6 31 7 red, 1 yellow minutes campestris (10) spread over Episyrphus sp. 5 2 3 - 5 days (1) Syrphus sp. 1 - 1 - (-) Satyrid Satiridae sp. 1 - 2 - butterflies (-) Parasitoid Microplitis 1 - 1 - wasps tristis (1) Bees Apis sp. 1 - 1 - (-) Infection rate per plant in relation to distance from an inoculum source of the fungal pathogen Microbotryum violaceum in male and female plants of Silene latifolia (experiment 5).

Block7 represents the original 7 blocks (figure 2 and 4), block4 represents the original 7 blocks, but with the data from all recipient blocks merged into 3 new blocks, 2-5m, 6-9m and 10-13m, to gain statistical power for the contrast analysis. Designated significance levels;

** = p<0.01;

**** = p<0.0001.

Source of variation in infection rate df Block7 6 172.2 **** Host sex 1 35.8 **** Block7 * Host sex 6 18.2 ** Contrasts between blocks4 within sex Male hosts Source vs 2-5m 1 29.8 **** 2-5m vs 6-9m 1 8.2 ** 6-9m vs 10-13m 1 0. Female hosts Source vs 2-5m 1 60.9 **** 2-5m vs 6-9m 1 0. 6-9m vs 10-13m 1 0. 0. S. latifolia 0. Males Females 0. 0. 0. 0. Source 0. plants 0. 0. 0. 0. 0.01 0. 0.0 0.0 0.0 0. 0-1m 2-3m 4-5m 6-7m 8-9m 10-11m 12-13m (n = 153) (n = 158) (n = 179) (n = 155) (n = 163) (n = 136) (n = 124) Decreasing infection rate of flowering host plants of Silene latifolia with increasing distance from source plants (experiment 5). Hatched bars represent male plants;

open bars represent female plants. See figure 1 for details of the experimental setup.

Infection rate I Figure 4 shows a decrease in anther smut infection rate of flowering S. latifolia with increasing distance from the infected plants. Furthermore, both figure 4 and table 6 show that mean infection rate in male plants is significantly higher than in female plants across all blocks, including the source. For the contrast analysis, the six recipient blocks were merged two by two into three new blocks, 2-5m, 6-9m and 10 13m to gain statistical power. The contrast analysis between source and the three merged blocks showed a significant decrease of infection rate of male plants between the source and the blocks at 2-5m (p<0.0001), and between the blocks at 2-5m and the blocks at 6-9m (p<0.005), but not between the blocks at 6-9m and the blocks at 10 13m. For female plants, infection rate was different between the source and the three blocks (p<0.0001), but not between any other of the merged blocks. This indicates that the decrease of infection rate with increasing distance from a teliospore source was steep in female hosts, and shallower in male hosts. Moreover, results suggest that an infection of healthy S. latifolia within a single flowering season is not likely to occur when the nearest teliospore source is much further away than 13m.

I I Our data clearly show that visitation of S. latifolia and S. dioica is highly assortative with respect to host species, even when these host species are placed in an artificial, fully mixed sympatric setup. However, results were influenced by weather conditions that varied between experimental days (Table 3), and this might have influenced both composition and behavior of the pollinator community (e.g. Brantjes 1981;

Herrera 1995). Nevertheless, table 2 shows that the dyes are dispersed over a greater distance intraspecificly than by interspecific vectoring, which holds true both for red dye via S. dioica and yellow dye via S. latifolia, and indicates host fidelity of vectors. A likely explanation for host fidelity could well be the existence of different pollinator guilds for these plant species. This idea is supported by direct observations to both diurnal and nocturnal insect visitors in this study (Table 5). Moreover, the pollinator guilds of S. dioica (Kay et al. 1984;

Westerbergh and Saura 1994;

Carlsson Granr et al. 1998) and S. latifolia (Brantjes 1976a;

1976b;

Shykoff and Bucheli 1995;

Altizer et al. 1998) have been described in allopatry, and in sympatry (Baker 1947;

Biere and Honders 1998;

Jrgens et al 1996;

Goulson and Jerrim 1997), pointing consistently to the same genera as the main pollinators, that are different for these two Silene species. S. dioica is mainly visited during daytime by several bumblebee species and hoverflies of the Syrphidae family, whereas S. latifolia is mainly visited nocturnally, by nocturnal moth species of the Noctuidae family. However, interspecific visitation, i.e. pollinators visiting both species, has been described for moths, bumblebees and hoverflies (Goulson and Jerrim 1997). Indeed, interspecific transfer of fluorescent dye after 24h was found to be substantial in both host species.

Interestingly, relatively more red dye was transmitted to S. latifolia than yellow dye to S. dioica, which was only significant in experiment DN-P2, suggesting that interspecific visitation was higher in the direction from S. dioica to S. latifolia than opposite. This was supported by the direct observations, in which nocturnal moths seemed to be more choosy, and visited more exclusively S. latifolia than S. dioica, and is consistent with a comparative study by Wirooks and Plassmann (1999) who found that the number of eggs and caterpillars of seven noctuid moth species on S. latifolia plants was roughly twenty-fold higher than on S. dioica.

S. latifolia has a typical moth pollination syndrome (Baker 1961;

Baker and Hurd 1968), with flowers that open at dusk and emit an intense scenting fragrance during the night, thereby being more attractive to nightly visitors such as noctuid moths (Brantjes 1978), than S. dioica which lacks this phenomenon. Since flowers of S. latifolia are often closed during daytime (Jrgens et al. 1996) they are less attractive to diurnal pollinators. From the diurnal visitors, that mainly visited S. dioica but were recorded to pay occasional visits to S. latifolia as well, bumblebees were by far the most abundant. Bumblebees might prefer S. dioica over S. latifolia for two reasons.

First, flowers of S. dioica are smaller than S. latifolia, specifically the length of the calyx (Jrgens et al. 1996). Flowers of S. dioica are small enough to make the nectar available to long-tongued bumblebees (B. hortorum and B. pascuorum) that cannot gain access to the nectar of S. latifolia flowers. B. terrestris, the third bumblebee species that we observed, has a smaller tongue (Jrgens et al. 1996), and can reach nectar of neither S. latifolia, nor S. dioica flowers. Instead, it pierces tiny holes in the sepal to gain access to the nectar resources of a flower, which is known as nectar robbing (cf. Heinrich 1976). We observed such bumblebee-inflicted holes frequently in S. dioica flowers, as well as in S. latifolia flowers. Second, floral display size of S.

dioica is much larger than of S. latifolia in the field, mainly due to the significantly larger number of flowering stalks (which was up to two times larger for healthy female hosts, and up to 2.5 times larger for healthy male hosts;

A. Biere, unpublished results). Foraging bumblebees are more attracted to large floral displays (e.g.

Klinkhamer et al. 1989;

Goulson et al. 1998), trying to minimize flight and search times within a foraging bout. Furthermore, bumblebees exhibit flower constancy, i.e.

the tendency of experienced pollinators to visit the same plants species or type of flowers regardless of the presence of other potential rewarding flowers nearby (Levin and Anderson 1970;

Oster and Heinrich 1976;

Waser 1986), promoting a species specific visitation pattern in a mix of different flower types. Another very active diurnal vector in our study is the hoverfly Rhingia campestris. This species has been recorded to discriminate between different colors, favoring violet and blue colored artificial flowers over white ones (Haslett 1989). This is consistent with the Rhingia campestris in our study, which favors the pink flowers of S. dioica over the white flowers of S. latifolia. Flower constancy has been reported for related Syrphid hoverflies as well (Goulson and Wright 1998), which would again strengthen the assortative effect.

In natural sympatric populations of S. latifolia and S. dioica, species are rarely as mixed as in our experimental plot. Due to habitat differences (Goulson and Jerrim 1997) and differential adaptation to light intensity (Willmot and Moore 1973) populations with co-occurring S. latifolia and S. dioica are often patchy. This, and the fact that interplant distances between S. latifolia and S. dioica in the field are usually much larger than 1m, suggests that the interspecific visitation rates as found in this study are absolute maximum estimates of this parameter. Therefore, we expect the host species-specific visitation pattern that we observed in the experimental plot to be much stronger in natural sympatric populations. Nevertheless, the occurrence of hybrids in natural sympatric populations (Baker 1947;

Goulson and Jerrim 1997), which has been reported to constitute more than 6% of the sympatric population of Norg (Biere and Honders 1996b), is a silent witness of interspecific visitation of pollinators, and might provide an estimate of exchange between host species in natural sympatry. Moreover, it shows that interspecific visitation in natural field populations is ecologically significant.

x We found in experiments DN-P1, DN-P2 and D-P2, but not in D-P1, that male plants of these dioecious Silene species are visited more frequently than female plants, i.e. the frequencies of flowers per plant with either red dye, yellow dye, or both dyes were significantly higher in male plants than in female plants in most setups. Favoring male hosts over female hosts of pollinators is expected for the following reasons.

First, foraging pollinators favor large floral display sizes over smaller ones, as was argued for differences between host species in the previous paragraph. Male plants of S. latifolia (e.g. Gross and Soule 1981;

Meagher 1992;

Delph and Meagher 1995;

Carroll and Delph 1996) and S. dioica (Hemborg 1998;

Hemborg and Karlsson 1999) bear much more flowers than female plants. Second, male flowers of S. dioica (Kay et al. 1984;

Hemborg 1998) and S. latifolia (Shykoff and Bucheli 1995;

Biere and Honders 1996a;

Shykoff and Kaltz, 1998) contain higher concentrations of nectar than female flowers. In a number of studies, male plants of S. latifolia (Shykoff and Bucheli 1995;

Altizer et al. 1998) and S. dioica (Carlsson-Granr 1998) have been found to be preferentially visited over female plants by pollinators. Actually, active discrimination against female flowers is suggested to be a common phenomenon in dioecious plants (Bierzychudek 1987 and references therein). These differential sex preferences of pollinators may also contribute to the higher infection rate of the S.

latifolia males in experiment 5. Indeed, male-biased infection rates, i.e. the number of males that become infected in a flowering season, have frequently been documented in literature (Alexander 1989;

Thrall and Jarosz 1994;

Alexander and Antonovics 1995;

Biere and Antonovics 1996;

Biere and Honders 1998). Male plants flower earlier and longer than female plants, which increases the risk of getting infected in their first season (Thrall and Jarosz 1994;

Biere and Antonovics 1996). However, disease incidences, i.e. the number of diseased plants in a population at a certain point in time, often do not show biases between host sexes in other studies of these Silene species (Zillig 1921;

Alexander 1990), or are even female-biased (Lee 1981;

Kaltz and Shykoff 2001). This discrepancy between rate of infection and disease incidence might be explained by sexual differences in other parameters that affect disease incidence, such as pre-floral infection rates, recovery and disease-induced mortality (Biere and Honders, unpublished results).

Figure 1 showed that the dispersal of dye after 24h (the upper four panels in the figure) has the potential to go beyond the artificial 10m limits of the experimental plot. Indeed, the fitted equations in table 2 clearly show that at the border of the plot dye frequencies are still high, up to 38%. Moreover, if we look at the distances from the source at which the visitation frequency become smaller than 1%, by extrapolating the dataset (Table 2), this yields nearly 56m for red dye on S. dioica and nearly 33m for yellow dye on S. latifolia in DN-P2. The estimates for interspecific visitation distances, i.e. red dye on S. latifolia or yellow dye on S. dioica are significantly smaller. This strongly suggests that teliospores could be dispersed by their insect vectors much further than 10m, especially between hosts of the same species.

However, receiving teliospores does not guarantee becoming infected. Empirical studies have shown that the distance range at which infection occurs from a certain inoculum source might be more limited. In a study of by Alexander (1990) spores were transmitted at least 10m from a teliospore source, and infections up to 6m from this source in the following year. Roche et al. (1995) detected floral infections of S.

latifolia up to 11.2m, which was the farthest distance possible in their experiment. The results from experiment 5 are more or less consistent with these studies, and show that the distance of infection is in the same order of magnitude as in Roches study, with an infection rate of the most distant block (12-13m) still more than 4% of the male hosts. However, we must keep in mind that our teliospore source was much larger, and the density of our plants was much higher than in their studies. We detected no infected female hosts at these distances. Roche and colleagues suggested that the dispersal of teliospores, and the resulting infection of S. latifolia are limited to distances close to 12m (Roche et al. 1995). In addition to this, our results confirmed that the chances of healthy S. latifolia becoming infected with M. violaceum within a single flowering season are small at distances much further away then 12-13m from a teliospore source. In contrast to this, studies on metapopulations of the Silene Microbotryum system that examined recolonization rates of the fungus have shown that dispersal of infection further than 40m is likely to occur within a single year (Thrall and Antonovics 1995). Since our experiments were also limited to distances below 13m, we suggest that spatially larger experimental studies are needed to investigate the true dispersal capacity of M. violaceum and its infection.

I Summarizing our results, we have found a host species-specific visitation pattern of pollinators/vectors in an artificial, fully mixed sympatric plot of S. latifolia and S. dioica. Red campions were visited mostly during daytime by a diurnal guild consisting of several bumblebee species and hoverflies, and white campions were mainly visited during the evening and night by a nocturnal guild consisting of several noctuid moths species. Experiment 5 showed that the range of teliospore dispersal and first season infection was likely to be not much larger than 12-13m. Furthermore, infection rate of male hosts was significantly higher, and occurred at larger distances from the teliospore source than of female hosts, which could reflect a male-biased visitation pattern, like was found in experiments 1-4. Knowing that natural sympatric populations of S. latifolia and S. dioica are both spatially and temporally more heterogeneous (cf. Van Putten et al. chapter 2) than our artificially mixed plot due to habitat differences (Willmot and Moore 1973;

Goulson and Jerrim 1996) and differential flowering phenology of both host species (Biere and Honders 1996), we expect the interspecific visitation rates to be much lower in a natural situation than what was estimated in experiments 1-4. Moreover, pollinators/vectors have been recorded to discriminate against M. violaceum infected plants, with healthy inflorescences being visited up to three times more frequent than infected ones (Jennersten 1988). This would diminish the amount of teliospore exchange between the host races even further. Therefore, outcrossing between the host races in natural sympatric populations of the two host species is expected to be low, and could provide a basis for the maintenance of the genetic host differentiation, which was observed in a microsatellite study of anther smuts in natural sympatric/parapatric populations of host species (Van Putten et al. chapter 2), and in a sporidial colony color marker in a true sympatric host population (Van Putten et al. chapter 3). The existence of strong fitness differences between the host races, that were found in a mating and competition experiment (Van Putten et al. chapter 4), implies that there must be a mechanism that ensures reproductive isolation between the host races to prevent the host race from S. dioica race from getting extinct, and thereby maintain the host related genetic diversity that was observed in these populations. The host fidelity of the different pollinator guilds of S. latifolia and S. dioica may well contribute to such a mechanism.

The authors wish to thank Manja Kwak for a useful discussion on pollinator behavior, Hans Peter Koelewijn for help with data analyses, and Quiny Schmers for help with plant handling in the greenhouse and at our experimental plot. This study was financially supported by the Earth and Life Science Foundation of the Netherlands Organization for Scientific Research (NWO-ALW);

grant 805 36-391.

T In the different chapters the obtained results already have been discussed extensively. In this final chapter the most important results will be summarized and put into perspective with an emphasis on differentiation between anther smuts from different host species in sympatric populations of two host species. First, the following questions will be addressed: Can we consider fungal isolates from different host species in sympatry to be separate host races? To what extent does the degree of sympatry influence the genetic divergence between the host races? And, what is the impact of local host spatial structure on this variation? Secondly, the fitness differences between fungal isolates from different host species that appear in mating and competition, and the implications of these differences for these host races in sympatry are discussed. Thirdly, the host fidelity of vectors in sympatry, and the impact of host spatial structure on the differentiation is discussed, putting forward the plausibility of a balance between gene flow and selection in sympatry that could contribute to the maintenance of host-specific differentiation in host sympatry. Also, some attention will be paid to the role of interspecific hybrid hosts on the amount of fungal gene flow, and the differences between host sexes in the process of maintenance of host-specific genetic variation. Finally, in some concluding remarks, the potential of this model system to investigate sympatric host race formation and speciation of this model system will be evaluated.

T I T T T In chapter 1, we observed host-specific microsatellite alleles in the allopatric and parapatric host populations of S. latifolia and S. dioica. For two of the loci, the alleles found in isolates from these two host species were separated in size by a gap of 7-9 repeats each. This was consistent with the work of Bucheli and co-workers (Bucheli et al. 2001). Together with the fact that this type of allelic variation was observed in allopatric host populations throughout Western Europe, this strongly suggests that the observed variation represents long-term divergence between these host races of anther smut, which presumably arose in allopatry. In that scenario, fungal isolates from these host species in sympatry will come into secondary contact with each other whenever there is (fungal) gene flow between isolates from the two host species. A first question that arises is, whether the divergence between strains from S. latifolia and S. dioica is large enough to consider both races as two sibling species rather than host races. Jeanike (1981) provides distinct criteria to distinguish between host races and sympatric host-associated sibling species, stating that if gene flow among two or more population is restricted solely, or primarily because of differential host preferences, this would constitute host races. Thus, without this basis for reproductive isolation being present, host races would, in an extreme case, fuse into a single panmictic population, whereas sibling species would maintain their separate genetic identities.

On the other side of the spectrum, when is variation among fungal isolates from different host species high enough to consider them as separate host races? The significantly lower genetic divergence in the more sympatric populations of both host species (Chapter 2 and Chapter 3) suggests that these host races could indeed fuse into panmictia, especially since it is known that both races can cross-inoculate each others host species without being a priori at a disadvantage (Biere and Honders 1996a;

Chapter 4). However, chapter 4 also showed that the race from S. latifolia often outcompetes the race from S. dioica, presumably due to faster conjugation. This indicates that, without a reproductive isolation mechanisms such as temporal (Biere and Honders 1996b) and spatial heterogeneity (Chapter 3), and/or host fidelity of vectors (Chapter 5), this might rather be invade and take over of the host race of S.

latifolia instead of a merger of host races. Nevertheless, the genetic divergence in allopatry remains rather high, and surely justifies their state of being separate host races.

1. SYMPATRY PARAPATRY ALLOPATRY Oxford 0.464*** 0. Abbertbos 0.264*** 0.367*** Kings Worthy Norg 0.046ns ? ?

0. 1 10 100 1000 Distance between sub populations of different host species (m) Hypothetical relationship between the degree of sympatry (approximated by the distance between sub populations of different host species within a metapopulation) and the genetic divergence between host races (estimated by FST values). The degree of sympatry was roughly estimated 1-10m for Norg, 10-100m for Abbertbos, 100-1000m for Oxford, with the population at Kings Worthy in between Abbertbos and Oxford. The boundaries for sympatry, parapatry and allopatry (sensu Kondrashov and Mina 1986) only represent a schematic indication;

the exact transition between two types is all but clear. FST values are based on the four microsatellite loci (Chapter 2).

T The degree of host sympatry influenced the genetic differentiation between host races (Chapter 2). Results suggested that increasing levels of host mixing increased the amount of gene flow between the host races. The average distance between two sub populations within a metapopulation, dominated each by a different host species, would be a reasonable predictor of genetic divergence between isolates from both host species, i.e. host races, in that fungal metapopulation, which is shown in figure 1. In reality, this picture does not present a one to one relationship with divergence at small spatial scales, as was shown in chapter 3. In that chapter, we showed the impact local of structure of hosts on the genetic population structure of the host races in a sympatric population of hosts. The fungal environment, represented by the host plants of the Norg population, was structured in the following three different ways: First and foremost, the host species belong to two different species. Second and ST Genetic divergence between host races (F ) thirdly, the host populations are both spatially (in distinct patches) and temporally (differential flowering phenology of both host species and sexes, Biere and Honders 1996b) heterogeneous for the pathogen. From this, we expected that the population of M. violaceum in Norg might not be panmictic (as the FST values of Chapter suggested), but actually consists of a number of demes that genetically structures the pathogen population, as has been found for many phytophagous insects (Mopper 1996). The heterogeneity of the environment is often a starting point in host specialization and host race formation (Berlocher 1998a), but may in this population represent a barrier that contributes to the maintenance of pre-existing divergence.

Whereas gene flow opposes the historically evolved differentiation between the host races, and acts to homogenize the genetic divergence, the local deme structure of both host species in this population will favor its maintenance (cf. Mopper 1996). Habitat, or host choice is found to be a crucial factor in theoretical models on host specialization (Fry 1996, Kawecki, 1997;

1998). Therefore, being the actors of host choice in this model system, a dominant role in the maintenance of the divergence was hypothesized for the pollinator/vector guilds of the host plants (Chapter 5). Among other possible mechanisms that also may contribute to this maintenance of host specific genetic variation is positive assortative mating between fungal isolates with respect to host species. This latter possibility was investigated and described in chapter 4, in which we performed mating and competition experiments between fungal strains that were isolated from both host species.

TI IT I T T T In chapter 4, it was shown that the conjugation frequency in vitro after 24h of strains from the host race of S. latifolia was significantly higher than strains from the S. dioica host race, suggesting that mating between strains from S. latifolia occurs faster than between strains from S. dioica. Furthermore, strains from S. latifolia performed better in competition in vivo on both host species, which was shown to be positively correlated to their mating behavior. Also, in this experiment it was shown that strains from the S. dioica race had a longer latency period, which was consistent with the study of Biere and Honders (1996a).

As would have been expected beforehand, we did not find (positive) assortative mating between the host races. If the supposed gene flow (Chapter 2) would have been large enough, we would expect the host race of S. dioica to decline or go extinct in sympatric populations of these host species, based on conjugation rate, infection success, latency period and competitive ability of strains. Since both host races do occur regularly in natural sympatric populations of these host species (Biere and Honders 1996b;

Chapter 2;

Chapter 3), other factors must be responsible for the maintenance of host-specific differentiation in sympatry. One factor may be the higher spore production on the native host species under non-competitive conditions, as observed on male plants (Biere and Honders 1996a). In addition, the deme structure due to spatial and temporal heterogeneity of host plants (Chapter 3), the role of vector behavior (Chapter 5), and heterosis, that was observed in female hosts in the competition experiment itself, could be among the other factors that contribute to the maintenance of strains from S. dioica and the differentiation in host sympatry (Chapter 4).

I I TI TI T I IT T Pollinators serve a dual role as vectors of anther smut spores (Jennersten 1983).

If these vectors can prevent, or strongly limit the amount of fungal gene flow, and thereby the possibilities for competition and outcrossing between fungal isolates from different host species simply by being choosy in host sympatry, host-related genetic differentiation between these host races might be maintained, i.e. both host races might coexist in sympatry. Indeed, the results of chapter 5 showed a significant host species-specific visitation pattern of pollinators/vectors in artificial, fully mixed sympatric plots of S. latifolia and S. dioica. Red campions were visited mostly during daytime by a diurnal guild consisting of several bumblebee species and hoverflies, and white campions were mainly visited during the evening and night by a nocturnal guild consisting of several noctuid moths species. This reduction of gene flow caused by the existence of different pollinator guilds for these two host species would surely contribute to the maintenance of host-specific genetic variation between the host races in sympatry. However, we estimated that the exchange in the experiments between both host species was still as large as approximately 30%, and occurred mainly from S. dioica towards S. latifolia, suggesting that the vector guild of S. latifolia (noctuid moths) was choosier than the guild of S. dioica (bumblebees). This is a considerable amount of interspecific visitation, possibly high enough to setup the doom scenario that strains of the S. dioica host race eventually will go to extinction if this would lead to competition for susceptible hosts with the strains of higher competitive ability from the S. latifolia host race. However, the natural sympatric populations of both host species that we have examined turned out to be quite different from the artificially mixed plots in the experiments of chapter 5.

I The results of chapter 3 showed that natural sympatric populations of S.

latifolia and S. dioica are spatially more heterogeneous than the fully mixed plots of chapter 5, possibly due to differences in habitat preference of both host species (see also Goulson and Jerrim 1996). Together with the temporal heterogeneity of differences in flowering phenology (Biere and Honders 1996b), we expect the interspecific visitation rates to be much lower in a natural situation than this estimate of 30%, that itself could represent an absolute maximum estimate of leakiness in host fidelity (cf. Feder 1998) in this model system. Also, pollinators/vectors have been recorded to discriminate against M. violaceum infected plants (Jennersten 1988), which would diminish the amount of teliospore exchange between host species even further. Therefore, outcrossing rates between fungal isolates from different host species in natural sympatric populations of these hosts are expected to be considerably lower, and are presumed to be low enough to hypothesize that host fidelity of vectors plays an important role in the maintenance of host-specific differentiation in sympatry.

The occurrence of interspecific hybrids, reported to constitute more than 6% of the sympatric population of Norg (Biere and Honders 1996b), is a silent witness of gene flow between these hosts, and might therefore represent a more accurate estimate of effective interspecific visitation in host sympatry in this population, and hence the exchange of teliospores between the host species. Note that hybrid hosts that are reported in the study of Biere and Honders (1996b) and in our study may include F1s as well as backcrosses with either parental host species.

Interspecific hybrids can play an interesting role in fungal differentiation, since they could form a hybrid bridge (cf. Floate and Witham 1993), i.e. easier shifting between host species via hybrid swarms of intermediate morphology, physiology, phenology or susceptibility. This could lead to more gene flow than in a situation without hybrid hosts, and would predict that, all else being equal (i.e. similar host composition, degree of sympatry, spatial structure, etc.), genetic divergence between fungal isolates is significantly lower in host populations with a lot of hybrid plants than in host populations that almost lack hybrids. Hybrids between S. latifolia and S.

dioica indeed have an intermediate morphology (Baker 1951), although most morphological characters are often not reliable for identification of hybrids (cf.

Goulson and Jerrim 1997). Furthermore, Biere and Honders (1996a) showed that interspecific hybrids did not differ in susceptibility from the mid-parent value of the pure parental host species, suggesting that a hybrid bridge may well exist, and is in principle accessible to fungal isolates from both host species. However, hybrid hosts were found to grow predominantly among patches of S. latifolia rather than among patches of S. dioica, either due to (unknown) habitat preferences of the hybrids or by strong differences in flowering phenology between host species and host sexes (see also Goulson and Jerrrim 1997, and Chapter 2 for a discussion on this issue). This may suggest that a hybrid bridge, if existent at all, is more accessible to fungal isolates from S. latifolia host race than for the S. dioica host race. Indeed, chapter 2 showed that fungal isolates from hybrid hosts from examined populations resembled the S.

latifolia host race, rather than the S. dioica host race. A rather wild speculation would then be that the S. latifolia host race might have pre-adapted to a more S. dioica host like environment via the hybrids that grew among them, giving them a head start over strains from S. dioica when confronted with the alien host. If this would be true, it could be part of the explanation why we observed strong fitness differences in competitive ability between strains from different host species, the strains from S.

latifolia having the natural advantage on its own host species, and be competitive on S. dioica at the same time. However, except for the microsatellite analysis of chapter 2, we did not include interspecific hybrids in any of the experiments in this study, hence we lack the additional data that would support such an idea.

T T I TI TI I TI T T In this thesis, we have found a number of differences between the host sexes that suggest that the balance between selection and migration of fungal isolates from the two host species may be different for male and female hosts. First, differential selection of fungal strains may occur on the different host sexes, and male and female hosts each may contribute to the maintenance of host-specific genetic variation in fungal isolates in a different way. In a cross inoculation experiment, the production of teliospores per plant was found to be significantly higher in conspecifics than in heterospecifics of the host of origin in male host plants, but not in female hosts (Biere and Honders 1996a). This held true for both host species, and provides an argument for the maintenance of host-specific divergence in sympatry by this trade-off in performance in male hosts. In female hosts, a different mechanism may contribute to the maintenance of host-specific genetic variation in fungal isolates. In the competition experiment of chapter 4, we observed a highly significant effect of host sex on the success of dikaryon types. Heterotypic dikaryons were found to be the better competitors in female host plants, at least in S. latifolia, which may contribute to the maintenance of the competitively inferior strains from S. dioica in host sympatry. The fact that heterosis was observed on female hosts might be explained by the more complex requirements to successfully infect female hosts as compared to male hosts. Infection of a female host requires a morphological sex change by fungal induction of male specific gene expression in the developing flower (Scutt et al. 1997) and prevention of pollen grain development (Audran and Batcho 1981) in dioecious Silene species. Therefore, being more complex and presumably more costly to the fungus, complementation processes in heterotypic dikaryons may be more important in female hosts than in male hosts. If this is the case, female hosts could contribute to the maintenance of host-specific variation by providing a safe haven for the competitive weaker strains from S. dioica in sympatric populations of these host species. However, a number of studies have shown the predominantly selfing nature of this fungus (e.g. Baird and Garber 1979), resulting in strong observed homozygosities in natural host populations even in highly polymorphic microsatellite loci (Bucheli et al. 2001;

Chapter 2), which could strongly limit the importance the heterosis effects in female hosts.

Second, differential migration rates of fungal strains may occur on different host sexes, since we found significant differences in interspecific host visitation pattern between the two host sexes in the vector experiments of chapter 5. Male hosts were preferentially visited over female hosts, which was consistent with other studies examining pollinator visitation in both these Silene species (Shykoff and Bucheli 1995;

Altizer et al. 1998;

Carlsson-Granr 1998). This has also been reported in other dioecious perennial herbs (e.g. Vaughton and Ramsey 1998) and has been proposed to be a common phenomenon in dioecious plants (Bierzychudek 1987 and references therein). Male biased visitation is expected due to a larger floral display size of males (S. latifolia e.g. Gross and Soule 1981;

Meagher 1992;

Delph and Meagher 1995;

Carroll and Delph 1996;

and S. dioica Hemborg 1998;

Hemborg and Karlsson 1999), and higher nectar concentrations of male flowers (S. latifolia Shykoff and Bucheli 1995;

Biere and Honders 1996a;

Shykoff and Kaltz, 1998;

S. dioica Kay et al. 1984;

Hemborg 1998). Also, male plants flower earlier and longer than female plants. Both the preferential visitation, the longer flowering, and the large floral display could lead to the male-biased disease incidences that were observed in a number of studies (Alexander 1989;

Thrall and Jarosz 1994;

Alexander and Antonovics 1995;

Biere and Antonovics 1996;

Biere and Honders 1998;

but see Kaltz and Shykoff 2001 for a discussion).

Adding up these differences between host sexes, the sex ratio of the host population of interest is likely to be an important factor in the overall balance of selection and migration in host sympatry in this model system. Sex-ratio in the genus Silene seems to be generally female-biased (e.g. Westergaard 1958;

Mulcahy 1967;

Lloyd 1974) which is, at least for S. latifolia, S. dioica and their hybrids, thought to be caused by the presence of sex-ratio distorters and restorers that are linked to the Y chromosome, which are found to be polymorphic in natural populations (Taylor 1993;

1994;

1999). Since in strongly female-biased host populations the heterotypic crosses between both host races will stand a higher chance to be selected (Chapter 4), the presence of such sex-ratio distorters could indirectly contribute to the maintenance of host-specific strains as well.

I T Going back to the four stages of sympatric host race formation from Berlocher (1998a) that were mentioned in chapter 1, it seems that our model system could fit a stage 2 model, in which the races are isolated only by host fidelity with allele frequency differences between the host races. We did detect host-specific alleles (in the microsatellite loci;

Chapter 2), yet these were most apparent in more parapatric and allopatric populations of hosts, and not in the true sympatric population. Since there was also no evidence for any mechanism of pre- and/or postzygotic reproductive isolation that is unrelated to host fidelity whatsoever, we can safely conclude that this system has not reached stage 3 yet. However, the most obvious critic to this would be that a much more likely explanation for the evolution of these host races is a more classic allopatric model, in which populations are geographically separated and evolve independent from each other. Especially the strong fitness differences between strains from different host species that we have observed in chapter 4 (cf. Figure 1B in Chapter 1), but also the observations that similar (selectively neutral) genetic variation between the host races is widespread (Chapter 2), and that true sympatric populations of these host species infected with this anther smut are probably not so widespread, makes a more geographical model of host race formation much more feasible (i.e.

sympatric host race formation much less feasible) in this model system.

Whether these host races in allopatry will evolve to become different sibling species is even less clear. Fact is that within the host range of Caryophyllaceous host species, the host species S. latifolia and S. dioica are closely related themselves, sharing a recent common ancestor (Prentice 1978;

Desfaux and Lejeune 1996).

Pages:     | 1 || 3 |



2011 www.dissers.ru -

, .
, , , , 1-2 .