Tanja Koskela, Department of Biological and Environmental Science, University of Jyväskylä, PO Box 35, FIN-40351 Jyväskylä, Finland. Tel.: +358 14 2602244; fax: +358 14 2602321; e-mail: firstname.lastname@example.org
If parasites decrease the fitness of their hosts one could expect selection for host traits (e.g. resistance and tolerance) that decrease the negative effects of parasitic infection. To study selection caused by parasitism, we used a novel study system: we grew host plants (Urtica dioica) that originated from previously parasitized and unparasitized natural populations (four of each) with or without a holoparasitic plant (Cuscuta europaea). Infectivity of the parasite (i.e. qualitative resistance of the host) did not differ between the two host types. Parasites grown with hosts from parasitized populations had lower performance than parasites grown with hosts from unparasitized populations, indicating host resistance in terms of parasite’s performance (i.e. quantitative resistance). However, our results suggest that the tolerance of parasitic infection was lower in hosts from parasitized populations compared with hosts from unparasitized populations as indicated by the lower above-ground vegetative biomass of the infected host plants from previously parasitized populations.
There is some evidence of resistance against parasitic plants from crop populations (e.g. Lane & Bailey, 1992; Lane et al., 1993; Cubero et al., 1994), but studies from natural populations of parasitic plants and host plants are scant. Selection exerted by parasitic plants in crop populations that often originate from a single or a few strains and are grown in repeated monoculture practices, may be very different from that exerted in natural populations. Resistance to biological enemies (e.g. herbivores, parasites) has often been measured as host traits that completely prevent the infection (i.e. qualitative resistance). Resistance has also been measured in terms of performance of the enemy (i.e. quantitative resistance), for example, as decreased growth or reproduction of herbivores or parasites on resistant hosts (e.g. Lane et al., 1993; Karban & Baldwin, 1997; Kuiper et al., 1998). Resistance mechanisms that lead to unsuccessful establishment of a parasitic plant may include mechanical barriers (e.g. localized necrosis of host tissue around the penetration site of the parasite) as well as different defense chemicals (e.g. flavonoids) (Lane et al., 1993; Parker & Riches, 1993; Riopel & Timko, 1995; and references therein). Host species have been shown to vary in their ability to tolerate parasitic plants and in their capacity to support parasite growth and reproduction (e.g. Atsatt & Strong, 1970; Clay et al., 1985; Gibson & Watkinson, 1989; Kelly, 1990; Marvier, 1996). Further, a single parasitic species may have differential effects on different host species (Gibson & Watkinson, 1991; Marvier, 1996; Matthies, 1996, 1997). To our knowledge, there are no previous studies that have taken the approach we use in this study, that is, compared the performance of both a host plant and a parasitic plant, when plants from formerly unparasitized (potentially susceptible and/or intolerant) and parasitized (potentially resistant and/or tolerant) natural populations of the same host species are infected with the parasitic plant.
We conducted a greenhouse pot experiment with the holoparasitic plant, greater dodder (Cuscuta europaea) and its dioecious host plant, the stinging nettle (Urtica dioica) that originated from natural populations. We examined infectivity and performance of parasites (originating from the parasitized populations) on host plants from previously parasitized and unparasitized populations. We also compared performance with and without parasites of host plants with different histories in parasitism to assess trade-offs between host resistance and reproduction.
We addressed two specific questions. First, are there differences in host traits, and in the infectivity and performance of the parasite between the two host types, that is, has there been selection for resistance or tolerance in the host? We measured resistance as infectivity (qualitative resistance), and performance of the parasite (quantitative resistance), and tolerance as biomass accumulation of the host. Second, is there evidence of costs of resistance in terms of reproduction of the host?
Materials and methods
Stinging nettle, Urtica dioica L. (Urticaceae), is a wind-pollinated dioecious perennial that is common in nutrient-rich habitats. It reproduces sexually by seed and vegetatively by rhizomes. Greater dodder, Cuscuta europaea L. (Cuscutaceae), is an annual rootless stem holoparasite that lacks chlorophyll (Machado & Zetsche, 1990; Parker & Riches, 1993). It thus completely depends on its host for resources. C. europaea extracts water, nutrients, and carbon predominantly from the phloem of its host via haustorial connections (Kuijt, 1969; Press et al., 1990; Stewart & Press, 1990; Parker & Riches, 1993).
The experiment was carried out in summer 1998 in an unheated greenhouse under natural light conditions at the Konnevesi Research Station, Central Finland (62°37′N, 26°21′E). The plant material was collected in the beginning of May from Turku, South-western Finland (60°N, 22°E). We collected small equally sized [length 8.2 ± 0.2 cm (mean ± SE)] rhizome-derived U. dioica plants from eight populations of which four were previously unparasitized and four previously parasitized by C. europaea. According to the collections of the Turku University Herbarium and information from local botanists (K. Syrjänen, personal communication), the parasite has been present in some of the parasitized populations for at least 80 years, and in all of the parasitized host populations the parasite has been present for at least 20 years. We believe that the unparasitized host populations have been parasite-free for at least the same length of time as the parasite is rather rare in Finland having a very patchy distribution, and thus almost all parasite populations are well known to the local botanists. Within each population, two U. dioica plants were collected from each of 15 separate patches in order to have several clones in the experiment. C. europaea seedlings were collected from the ground from the sites of the four parasitized U. dioica populations just after they had germinated and before they had attached to the host plants. The populations were similar with regard to habitat type, vegetation, moisture and soil nutrient levels (T. Koskela, unpublished observation). In the study sites, U. dioica was the predominant host for C. europaea and present in 97.5% of plots studied (T. Koskela, unpublished observation). In the parasitized populations, C. europaea was abundant, and present in 87.5% of plots in which U. dioica occurred (T. Koskela, unpublished observation). Distances among populations varied from 500 m to 20 km. Distances between unparasitized and parasitized populations were always at least 800 m.
The host plants were planted individually in 0.5 L plastic pots, using a mixture of sand and fertilized soil (1:1), and their stem length was measured. The original number of replicates was 15 per host population (altogether eight populations) per treatment (uninfected/infected), that is, altogether 240 plants. However, the final number of replicates is lower as a result of mortality and failed infection attempts. Most of the host mortality occurred very early during the experiment, and was random between uninfected and infected treatments (χ21=3.380, P=0.066) and between the two host types (χ21=1.725, P=0.189). Mortality was mostly caused by the fact that the host plants were established from parts of the rhizomes, which did not work out as well as we expected. In addition, some of the infection attempts failed, and altogether five host and two parasite individuals died during the actual experiment. The final number of replicates is presented in figures.
The experiment was conducted as a fully factorial design. Half of the host plants from each host population were randomly assigned to the infection treatment (i.e. one host plant from each of the patches from each of the host populations was grown uninfected, and one plant was grown infected with the parasitic plant). In the infection treatments, we used an equal proportion of parasites from the four different parasite populations to infect each of the eight host populations (i.e. one-fourth of the parasites used for each host population originated from each of the four parasite populations). Two C. europaea seedlings were planted per pot at the base of the host plant 3–4 days after planting of the host. If neither of the parasite seedlings attached to the host within a few days, new seedlings were planted. If both of the parasite seedlings attached to the host plant, one was immediately and carefully removed. Thus, in the infection treatments we had one host plant and one parasitic plant per pot.
The host plants were kept well watered. The pots were randomly arranged in the greenhouse, and rearranged regularly. The experiment was terminated 10 weeks after the planting of the host seedlings. At this time the first host individuals started to senesce, and most female host plants and parasitic plants had produced seeds. At the end of the experiment we recorded the sex of the host plants, counted the number of flowers of the parasitic plants, and collected the biomass of the host and the parasite. We collected the above-ground biomass of the host separately for vegetative (stems and leaves) and reproductive (flowers, seeds, and flower stalks) parts. We carefully washed the roots free of debris and collected the below-ground biomass of the host. All plant parts were oven-dried (80 °C, 24 h) and weighed.
The total biomass and the number of flowers of the parasitic plant were analysed with a two-way analysis of variance (ANOVA) with host type (previously unparasitized or parasitized population, fixed) and host population (random, nested within host type) as factors. We tested the infectivity of the parasite with a χ2 test with host type (previously unparasitized or parasitized) as a factor, and the success of infection on each host plant (infected or uninfected) as the response variable. The data on above-ground vegetative biomass and root biomass of the host were analysed with a three-way ANOVA with host type (previously unparasitized or parasitized population, fixed), host population (random, nested within host type) and treatment (uninfected or infected with the parasitic plant, fixed) as factors. As only two host plants in the infection treatment flowered, sex of the host plant was not included as a factor in the analysis described above. To examine differences in reproductive traits between the two host types and sexes in uninfected treatments (where all host plants flowered), sex of the host plant was also used as a factor in the analysis of variance. Four individuals that had both female and male flowers were excluded from this analysis.
When testing host traits, we used the stem length of the host plant in the beginning of the experiment as a covariate, excluding it when it was not significant. To fulfil the assumptions of parametric ANOVA, data on the above-ground vegetative, root and reproductive biomass of the host were log-transformed. Figures represent untransformed data. All statistical analyses were performed with SPSS for Windows (SPSS Inc., 1997).
Resistance: infectivity and performance of the parasite
The percentage of successful infections (i.e. infectivity) did not significantly differ between parasites grown with hosts from previously unparasitized and parasitized populations (78.1 and 76.1%, respectively, χ21=0.084, P=0.772). Parasites that grew with hosts from previously unparasitized populations had significantly higher total biomass and produced significantly more flowers than those grown with hosts from previously parasitized populations (Table 1, Fig. 1). The total biomass of the parasite explained a significant amount of variation in the flower number (linear regression: F=241.66, n=26, P > 0.0001, r2=0.91, parasites grown with hosts from previously parasitized populations were used in this analysis). These results indicate that the quantitative resistance of the host plant against parasitic infection was higher for hosts from parasitized populations compared with those from unparasitized populations; the lower the parasite performance, the higher the resistance of the host plant.
Table 1. ANOVAS for testing the effects of host type (previously unparasitized or parasitized population, fixed), and host population (random, nested within host type) on the total biomass and number of flowers of the parasitic plant Cuscuta europaea.
Host tolerance: effects of infection in relation to host type
Parasitic infection had a severe negative effect on the reproduction of the host. Only two of 62 infected host plants produced flowers whereas all of the 64 uninfected host individuals reproduced. Parasitic infection had a significant negative effect on the above-ground vegetative biomass and root biomass of the host (Table 2, Fig. 2). The biomass of infected host plants was on average 87.1 ± 1.4% (mean ± SE) lower than that of the uninfected plants. There were no significant differences in the above-ground vegetative or root biomass between host types or among host populations, neither were there any statistically significant interactions between parasitic infection and host type for root biomass (Table 2, Fig. 2). However, the interaction between parasitic infection and host type was significant for above-ground vegetative biomass (Table 2). Because of this significant interaction we also conducted the analysis of variance separately for uninfected and infected hosts. According to this analysis, host population did not have a significant effect on host’s above-ground vegetative biomass in either uninfected or infected hosts, and host type did not have a significant effect on host’s above-ground vegetative biomass in uninfected hosts (Table 2). When infected, however, host type had a significant main effect on the above-ground vegetative biomass of the host (Table 2). Infected host plants from previously parasitized populations had a lower above-ground vegetative biomass than infected hosts from previously unparasitized populations (0.184 ± 0.020 and 0.301 ± 0.038 g, respectively, Table 2, Fig. 2). To further test for tolerance differences between the host types we used a reaction norm approach that takes into account the effect of variation in parasite burden on host fitness (e.g. Simms & Triplett, 1994; Abrahamson & Weis, 1997). We analysed the host’s above-ground vegetative biomass with host type (fixed) and host population (random, nested within host type) as factors, and parasite biomass as a covariate, including the interactions between the covariate and the host type, and between the covariate and the host population in the model. The values of the uninfected plants (i.e. parasite burden equals to zero) were included in the analysis to represent host fitness in the absence of damage (Stowe, 1998). The effects of host type and host population on above-ground vegetative biomass were not significant (Table 3). However, the interaction between the covariate (parasite biomass) and host type was significant indicating that with a similar parasite burden, hosts from previously unparasitized populations had higher above-ground vegetative biomass than hosts from parasitized populations (Table 3, Fig. 3). These results suggest that the tolerance of parasitic infection in terms of above-ground vegetative biomass was lower for host plants from previously parasitized populations compared with hosts from unparasitized populations.
Table 2. ANOVAS for testing the effects of parasitic infection by Cuscuta europaea, host type (previously unparasitized or parasitized population, fixed), and host population (random, nested within host type) on above-ground vegetative biomass and root biomass of the host plant Urtica dioica. Stem length of the host in the beginning of the experiment was used as a covariate for root biomass. Because of the significant interaction between infection and host type, the analysis was also conducted separately for uninfected and infected host plants for above-ground vegetative biomass.
Table 3. ANOVAS for testing the effects of host type (previously unparasitized or parasitized population, fixed), host population (random, nested within host type), and the parasite burden (parasite biomass, used as a covariate) on above-ground vegetative biomass of the host plant Urtica dioica.
Costs of resistance: reproduction of the host
To examine whether selection for increased resistance had led to correlated changes in other host life-history traits, we compared differences in reproduction between the two host types and sexes of the uninfected hosts. There were no significant differences in the reproductive biomass between the two host types, between the sexes, or among host populations (Table 4, Fig. 4). Thus, these results do not indicate costs of resistance in terms of host’s reproduction.
Table 4. ANOVAS for testing the effects of sex (fixed), host type (previously unparasitized or parasitized population, fixed), and host population (random, nested within host type) on reproductive biomass of the host plant Urtica dioica. Stem length of the host in the beginning of the experiment was used as a covariate.
One could expect that hosts evolve to reduce the deleterious effects of a parasite (e.g. by resistance mechanisms), whereas parasites evolve to optimize host exploitation (e.g. May & Anderson, 1983; Ebert & Hamilton, 1996). Given the high virulence of C. europaea demonstrated in this and in our previous study (Koskela et al., 2000), and because of the abundance of C. europaea in the parasitized host populations used in this study (T. Koskela, unpublished observation), we expected that in populations affected by parasitism, selection should have favoured host traits that decrease the negative effects of the parasite, resulting in differences in resistance and/or tolerance against the parasitic plant between hosts from previously parasitized and unparasitized populations.
A central assumption in our study was that the host populations were similar in other aspects except for the presence of the parasite. This is supported by field data showing no differences among parasitized and unparasitized populations in, for example, vegetation and soil nutrient levels (T. Koskela, unpublished observation). In addition, we did not find significant differences in above-ground vegetative, reproductive or root biomass between the two host types or among host populations as indicated by the nonsignificant main effects of host type and host population. Therefore, we speculate that the differences between the two host types in traits related to resistance or tolerance to the parasitic infection reflect the effects of previous selection by the parasite. However, we realize that strong inferences about the role of parasites as a selective factor for host resistance can only be made when presence and absence of the parasite and changes in resistance levels of the host can reliably be documented (e.g. Rausher, 1996).
In this study, the percentage of successful infections (i.e. infectivity) was high and did not differ between the two host types. That is, there were no differences in qualitative resistance between the two host types. Other measures used as expression of resistance in crop plants include, for example, poor growth and development of the parasitic plant on resistant hosts (Lane et al., 1993; Kuiper et al., 1998), and a low number of emerged parasitic plant shoots per host plant or per unit area (Cubero et al., 1994 and references therein). In this study, the parasite performed better (measured as biomass and reproduction) when grown on hosts from unparasitized populations compared with hosts from parasitized populations. Thus, our results suggest that previous parasitism had selected for quantitative resistance, as indicated by the lower fitness of the parasites on hosts originating from previously parasitized populations. In a previous study, we found differences among host populations in their resistance when resistance was measured as parasite’s performance (Koskela et al., 2000). The benefits of this kind of resistance to the host may be explained by the correlation between parasite biomass and damage level, by the effects of quantitative resistance on the population dynamics of the annual parasite, and by the perenniality of the host. In another experiment, where we studied parasite performance on several host families and compared fitness of infected and uninfected plants representing the same families, we found a significant positive correlation between parasite performance (resistance) and the fitness reduction caused by the infection (T. Koskela, S. Puustinen, V. Salonen and P. Mutikainen, unpublished observation). In other words, the higher the parasite biomass, the higher is the reduction in host fitness the parasite induces. Taken together, this suggests that parasite performance can be used as an indicator of resistance. Further, because of the lower growth and reproduction of the parasite on resistant hosts, there will be fewer parasites attacking the same host clone and its offspring in subsequent year(s). However, the mechanistic basis of the observed resistance requires further studies. Altogether, the mechanistic and genetic basis of plant resistance to parasitic plants is generally not well understood, and often it is not clear which host traits are related to resistance.
Results from two different statistical analyses suggest that the tolerance to the parasitic infection was lower for hosts from previously parasitized populations. In the first analysis, in which the main effect of parasitic infection was also included, hosts from previously parasitized populations had a lower above-ground vegetative biomass compared with hosts from previously unparasitized populations when infected. In the second analysis, we included the effect of parasite burden (parasite biomass used as a covariate) on the biomass accumulation of the host, as well as tested if the effect of parasite burden on host biomass differs between the two host types. The results of this analysis suggest tolerance differences between the two host types; with the same parasite burden, hosts from previously unparasitized populations were able to maintain the same or even higher fitness (measured as host’s above-ground vegetative biomass) compared with the hosts from previously parasitized populations.
Together with the results on parasite performance, these results suggest that selection for increased parasite resistance (indicated by the lower parasite performance) may have led to a correlated decrease in tolerance of the parasitic infection. In other words, in previously parasitized host populations, selection would have favoured either resistant or tolerant genotypes, but would have acted against genotypes that are both resistant and tolerant. Such a trade-off between resistance and tolerance has been reported by Fineblum & Rausher (1995) and by Stowe (1998) for a host plant–herbivore interaction. It is clear that further studies are needed to examine potential genetic correlations between resistance (whether qualitative or quantitative resistance) and tolerance in U. dioica against infection by C. europaea.
Parasitism by C. europaea may have selected for quantitative resistance in U. dioica as indicated by the lower parasite performance on hosts from previously parasitized natural populations. However, the tolerance of parasitic infection was lower in hosts from previously parasitized populations compared with hosts from previously unparasitized populations. One possible explanation for this may be a trade-off between resistance and tolerance, but this needs further studies. We are currently examining within-population genetic variation in resistance and tolerance of the host plant to infection by a parasitic plant, as well as genetic correlations among resistance and tolerance and other fitness-related traits with the same pair of species.
We thank J. Jokela, S. Puustinen, K. Clay, and A.E. Weis for helpful comments on the manuscript. K. Syrjänen is acknowledged for helping to locate the C. europaea populations and for assistance in the field. We thank S. Rantala and A.-M. Koivisto for technical assistance, and Konnevesi Research Station for greenhouse facilities and help in the greenhouse. This study was funded by the Academy of Finland.