Short-term fitness consequences of parasitism depend on host genotype and within-host parasite community

Abstract

Multiparasite communities inhabiting individual hosts are common and often consist of parasites from multiple taxa. The effects of parasite community composition and complexity on host fitness are critical for understanding how host–parasite coevolution is affected by parasite diversity. To test how naturally occurring parasites affect host fitness of multiple host genotypes, we performed a common-garden experiment where we inoculated four genotypes of host plant Plantago lanceolata with six microbial parasite treatments: three single-parasite treatments, a fungal mixture, a viral mixture, and a cross-kingdom treatment. Seed production was affected by both host genotype and parasite treatment, and their interaction jointly determined the growth of the hosts. Fungal parasites had more consistent negative effects than viruses in both single- and mixed-parasite treatments. These results demonstrate that parasite communities have the potential to affect the evolution and ecology of host populations through their effects on host growth and reproduction. Moreover, the results highlight the importance of accounting for the diversity of parasites as well as host genotypes when aiming to predict the consequences of parasites for epidemics as the effects of multiparasitism are not necessarily additive of single-parasite effects, nor uniform across all host genotypes.

Introduction

In wild populations, individuals must grow and reproduce under constant parasite attack. While some of the variation in growth and reproduction among individuals is caused by genotypic differences and responses to abiotic conditions (Laine, 2004; Westerband et al., 2021), differences in the ability to cope with parasite pressure can also affect host growth and fitness (Johnson & Hoverman, 2012; Pagán et al., 2008; Wintermantel, 2005). Individuals are often exploited by multiple parasite species simultaneously (Cox, 2001; Rigaud et al., 2010; Telfer et al., 2010; Tollenaere et al., 2016), and theoretical and empirical research shows that coinfection can change parasite replication as well as the negative effects experienced by the host (Alizon et al., 2013; Al-Naimi et al., 2005; Anjos et al., 1992; Tollenaere & Brugidou, 2017; Wintermantel, 2005). Although multiparasitism is common (Cox, 2001; Rigaud et al., 2010; Telfer et al., 2010; Tollenaere et al., 2016), experiments investigating the effect of parasitism on host fitness rarely account for the diversity of parasites observed in the wild (Rigaud et al., 2010). Understanding how the diversity of within-host parasite communities affects host growth and reproduction are key steps in linking parasite diversity to the evolution and ecology of hosts and parasites, as well as understanding the causes of intraspecific variation in growth and reproduction.

The same host may be attacked—simultaneously or sequentially—by a multitude of parasites (Karvonen et al., 2019). The successfully establishing parasites form within-host parasite communities that utilize the same resource base to support their own growth and reproduction (Rigaud et al., 2010; Tollenaere et al., 2016). Theory predicts that coinfection favors high host exploitation rates, although more aggressive exploitation of the host could reduce the host’s life span (Alizon et al., 2013; Mideo, 2009). The effect of multiple parasites on host fitness may be nonadditive, and hence, the outcome may be difficult to predict (Abbate et al., 2018; Holt & Dobson, 2013). Altered growth and replication under coinfection may result from direct and indirect interactions such as resource competition or through host immune responses (Alizon et al., 2013; Mideo, 2009; Tollenaere et al., 2016). Experimental evidence shows that immune responses triggered by the first-arriving parasites may have both positive and negative effects on the establishment and growth of later-arriving parasites (Bush & Malenke, 2008; Hoverman et al., 2013). Existing knowledge of plant immune responses to parasites indicates that plant responses to cross-species and cross-kingdom parasite communities are complex and difficult to tease apart without controlled experiments comparing different combinations of parasites (Bari & Jones, 2009). Notably, we lack empirical research combining plant viruses with other plant parasites, despite the abundance of viruses in wild plant populations (Cooper & Jones, 2006; Kamitani et al., 2016; Prendeville et al., 2012), and their potential impact on host fitness in both agricultural and wild plant populations (Alexander et al., 2017; Cooper & Jones, 2006).

Host genotype may determine the fitness consequences of parasites as genotypes differ often in their ability to resist and cope with parasites (Alexander et al., 1993; Bergelson et al., 2001; Laine et al., 2011; Pagán et al., 2008; Susi & Laine, 2015). While variation in these traits among host genotypes are well demonstrated (e.g., Pagán et al., 2008; Susi & Laine, 2015), remarkably little is known about how host genotypes differ in their ability to cope with multiparasite communities. Empirical evidence shows that coinfection can be more detrimental for hosts than infection by a single parasite (e.g., Anjos et al., 1992; Ben-ami et al., 2008; Tollenaere & Brugidou, 2017), but parasite–parasite competition may also restrain parasite growth under coinfection resulting in lower host damage (Ben-ami et al., 2008; Mideo, 2009). Individuals have limited recourses, and because parasite resistance is costly, the allocation of resources toward growth and reproduction may change under parasite attack (Brown & Rant, 2013; Huot et al., 2014; Pagán et al., 2008; Susi & Laine, 2015). The extent to which variation in host growth and fitness is determined by host genotype, parasite pressure, or jointly by both, and how the composition of the parasite community affects the outcome, is key for understanding how natural parasites affect host populations.

Here, we test the hypothesis that parasite community composition and host genotype jointly affect host fitness. Despite both empirical and theoretical work suggesting the importance of host genotype and parasite community for host fitness, studies explicitly testing the consequences of different combinations of host genotypes and parasite combinations are rare. We conducted a common-garden experiment where four genotypes of host plant, Plantago lanceolata, were inoculated with microbial parasite treatments that differ in diversity and taxonomic composition. We used the fungal pathogens Podosphaera plantaginis and Phomopsis subordinaria, as well as Plantago lanceolata latent virus (PlLV), and a virus bulk inoculum (collected from symptomatic wild plants), to produce six parasite community treatments: three single-parasite treatments, a fungal mixture, a virus mixture, and a cross-kingdom mixture of fungi and viruses. We also included a mock-inoculation treatment to compare the effects of the parasite treatments to the average of each host genotype in the absence of parasites, and to control for possible side effects of the inoculation itself. We compared several metrics of short-term host fitness: host growth rate, number of flowers and number of seeds, and the ratio of the number of flowers to relative growth rate, and the number of seeds to relative growth rate, among the treatments and host genotypes over the growing season. Our results show that both parasite community composition and host genotype can affect host fitness as the growth rate is influenced by the interaction between treatment and genotype, while seed production is directly affected by both the plant genotype and parasite treatment.

Methods

The study system

This study focuses on the parasites of host plant, Plantago lanceolata, in the Åland Islands (SE Finland, 60° 11ʹ 58.38″ N, 20° 22ʹ 16.22″ E), where data on a mapped network of ~4,000 populations have been collected yearly since the early 1990s (Ojanen et al., 2013). Plantago lanceolata is a self-incompatible perennial herb that reproduces via seeds that drop near the maternal plant and via side rosettes (Sagar & Harper, 1964). Interactions with foliar fungal pathogen Podosphaera plantaginis and P. lanceolata have been studied since 2001 in the Åland Islands. Recently studies of P. lanceolata microbial parasites have been extended to include the fungal parasite Phomopsis subordinaria and a diverse virus community (Jousimo et al., 2014; Laine, 2003; Sallinen et al., 2020; Susi et al., 2017a, 2019, 2022).

Powdery mildew P. plantaginis is an obligate foliar fungal pathogen (family Ascomycota, order Erysiphales) and is host specialist on P. lanceolata. The spores of P. plantaginis are wind dispersed, but they typically travel short distances with most spores being deposited in the immediate proximity of the infection source (Tack et al., 2014). Infection reduces the fitness of individual plants (Susi & Laine, 2015) and can reduce P. lanceolata population growth (Laine, 2004; Penczykowski et al., 2014). The interaction between P. plantaginis and P. lanceolata is strain specific (Laine, 2004).

Phomopsis subordinaria (Desm.) Trav. (telemorph Diaporthe adunca (Rob.) Niessl.) is an Ascomycota fungus that causes a stalk disease that is mainly found infecting P. lanceolata (de Nooij & van der Aa, 1987; Laine, 2003). It is splash dispersed, but needs a wound to enter the host (de Nooij & van der Aa, 1987; Linders et al., 1996). Infection typically begins below the inflorescence and spreads toward the rosette while killing the host tissue and producing spore structures on the dead tissue (de Nooij & van der Aa, 1987). Infection can have severe consequences for the host, as the development of infected flowers is halted, and seeds suffer from desiccation. Ultimately, infection can lead to mortality of the host plant (de Nooij & van der Aa, 1987).

The P. lanceolata populations in the Åland Islands support a diverse community of viruses which occur both singly and in coinfections (Susi et al., 2017a, 2019). The five most common viruses include the Plantago lanceolata latent virus PlLV, which occurs in low frequencies in P. lanceolata populations and can cause conspicuous yellowing of leaves of host individuals (Susi et al., 2019). PlLV currently has no other known host species and its vector dynamics and the distribution outside of the Åland Islands remain unknown. Plantago lanceolata genotypes vary in their likelihood to host viruses (Sallinen et al., 2020).

Currently, there is limited data testing how parasites included in this study respond to coinfection with each other. Coinfection with PlLV has been shown to slow down the spread of P. subordinaria symptoms and increase spore density of P. subordinaria (Susi et al., 2022), and coinfection with P. plantaginis increases the transmission of P. subordinaria (Sallinen et al., 2022). While P. subordinaria has been shown to induce production of iridoid glycosides in host plants (Marak et al., 2002) and P. plantaginis induces various plant signaling pathways involved in resistance, including plant hormones salicylic acid, jasmonic acid, and abscisic acid (Safdari et al., 2021), the effect of these hormones on the other parasites is not clear and could vary among host and parasite species to contrasting directions (Alazem & Lin, 2015).

Parasite material collection from wild populations

To test how parasite communities of varying diversity and taxonomical composition affect host fitness, we collected fungal material of the fungal parasites P. plantaginis and P. subordinaria, as well as virus material from wild host populations in the Åland Islands for inoculations. For treatments including P. plantaginis, a single strain collected from the Åland Islands (population 9021, Supplementary Figure S1) was used. The strain was purified in the laboratory by sequential single-colony inoculations of greenhouse-grown P. lanceolata leaves. Podosphaera plantaginis material was amplified over multiple rounds of inoculation before the experiment and kept on Petri dishes on moist filter paper in climate chambers with 21 °C and 16:8 hr light–dark cycle. For treatments including P. subordinaria, a single strain (ID: P29) collected from the Åland Islands in 2017 (population 1720, Supplementary Figure S1) was used. The strain was purified on sequential inoculation on oat agar plates (de Nooij & van der Aa, 1987) and further maintained in living P. lanceolata plants in climate chambers until the experiment.

For treatments with viruses, we included the PlLV (genus Capulaviridae), one of the five most common viruses in the P. lanceolata populations in the Åland Islands (Susi et al., 2017a, 2019), in the experiment as a single virus treatment. The virus was collected from the Åland Islands in 2017 by placing sentinel plants into wild P. plantaginis populations, and the plants were brought to University of Helsinki for upkeeping inside growth cabinets with continuous conditions. The plants were tested with virus-specific PCR primers for the five most common viruses occurring in the P. lanceolata populations in the Åland Island (Susi et al., 2017a) and were only found infected with PlLV. The plants were regularly tested for PlLV infection with the specific primers (Susi et al., 2017a). To amplify PlLV material for our experiment, we conducted virus inoculations of multiple additional individuals of quinoa (Chenopodium quinoa), which has been found susceptible to PlLV (Susi et al., 2019). Both the field-collected P. lanceolata plants and the inoculated quinoa plants were transported to the Lammi biological station for the inoculations.

To incorporate a virus diversity treatment that includes as high a diversity of viruses as possible, we pooled virotic looking plants collected from the Åland Islands for a “virus bulk inoculum.” To produce the virus bulk inoculum, 44 wild P. lanceolata plants showing typical virus infection symptoms (curly leaves, yellow or red color; Susi et al., 2019) were collected from ten populations in the Åland Islands (Supplementary Table S2). The plants were dug out with some soil, placed in pots with their soil, and transported to a greenhouse in Lammi biological station for the inoculation.

Field experiment measuring host fitness under varying parasite communities

To test how the composition and complexity of the parasite community affect host growth and reproduction, we set up a common-garden experiment at the Lammi Biological station in S. Finland (61°05ʹ28″N, 25°03ʹ90″E), where we inoculated plants with varying parasite combinations. Replicates of four P. lanceolata genotypes were produced for the experiment by cloning four maternal plants (see Supplementary Methods), each originating from single population in the Åland Islands and different from the populations from which the parasites were collected (Supplementary Figure S1). The maternal plants were collected as seeds and were planted and grown in a greenhouse at the University of Helsinki (Finland). They represent maternal lines that were confirmed to be free of the five most common viruses infecting P. lanceolata in the Åland Islands using PCR detection primers (Sallinen et al., 2020; Susi et al., 2017a, 2019). The genotypes represent different resistance genotypes against P. plantaginis, which has been observed during upkeeping of different P. plantaginis strains. As an obligate biotroph, P. plantaginis must be maintained on living host tissue and the four genotypes have been used as material when keeping strains alive. Resistance of the genotypes against viruses is unknown although they differ in their likelihood of getting infected by viruses (Sallinen et al., 2020). Plant genotypes are typically susceptible to P. subordinaria but may differ in the speed at which symptoms appear (Marak et al., 2002). The experimental plants were grown in 30:70 sand and potting soil mixture after the cloning in an insect-free greenhouse for two months before being moved to the field experiment site in the last week of June 2018 (Figure 1). Due to the variable success in cloning, a different number of each genotype were transplanted (75 of genotype 1, 90 of genotype 2, 135 of genotype 3, 120 of genotype 4).

In the common garden, the plants were transplanted into plots placed in a grid with 1.5-m distance between each plot. Each plot contained a 13-cm-deep 40 cm × 50 cm sand soil base, separated from the ground with a polypropylene root barrier (Hortex, Finland), and each plot was enclosed inside an insect fabric cage (Hortex, Finland) to protect the plants from herbivores and obstruct vector transmitted parasite movement. In each plot, five plants were transplanted into the sand with their potting soil. Each plot contained a mixture of genotypes so that majority of the plots included one of each genotype and an additional plant of genotype 3 or 4. Ten plots that did not contain genotype 1, but a mixture of the three other genotypes, each received a different set of parasite inoculations.

To produce parasite communities of varying composition and taxonomical complexity, the experimental plants were inoculated with 15 combinations of four types of parasite material (Figure 1), later pooled into six types of parasite community treatments. The four types of parasite materials were P. subordinaria alone, P. plantaginis alone, PlLV alone, a bulk virus inoculum alone, and all possible pairs and three-way combinations as well as all four treatments (Figure 1). All plants in a plot received the same inoculations and five plots were used to replicate each of the 15 inoculation treatments (Figure 1). A mock-inoculated control was also included, and there were five additional plots with four plants in the parasite-free mock-inoculated control treatment. Hence, there were in total 420 plants at the beginning of the experiment in 85 plots. Because infection status was not confirmed for all fungal and viral treatments (see Results), we analyzed the data using inoculation treatment groups as an explanatory variable. The treatments were pooled into six parasite community treatments, to which we refer as “parasite treatments,” to allow for more generalized interpretation of the results: P. subordinaria alone, P. plantaginis alone, PlLV alone, a fungal mixture of both P. subordinaria and P. plantaginis, a virus mixture of bulk virus inoculum or both bulk virus inoculum and PlLV, and a cross-kingdom mixture of fungi and viruses (Figure 1). The cross-kingdom mixture included any combination of one or both fungi, and one or both of the PlLV and virus bulk inoculation treatments. For the analysis of data collected in the first measurement (Figure 1), when P. subordinaria was not yet inoculated, the plants were divided based on the inoculations they had already received and were divided into P. plantaginis alone, PlLV alone, a virus mixture, and a cross-kingdom mixture.

Podosphaera plantaginis was first inoculated in the same week of transplanting of the experimental plants in the last week of June and again 5 weeks later in the last week of July (Figure 1). Inoculation was conducted twice to ensure that plant immunity was challenged by the parasite. Spores propagated in the laboratory (see “Parasite material collection from wild populations”) were inoculated onto the leaves of experimental plants with a fine paint brush. Similar-sized lesions with similar spore densities were used, and the spores were brushed onto all leaves.

Virus inoculations (PlLV, virus bulk inoculum) were performed 1 week after transplanting of the experimental plants (first week of July) and again 5 weeks later (beginning of August). Inoculation was conducted twice to ensure that plant immunity was challenged by the viruses. Inoculations were performed using plant sap extracted from infected plants (see “Parasite material collection from wild populations”). To create the virus bulk inoculum, leaves of the plants collected from the wild populations were crushed in phosphate buffer inside plastic extraction bags (Bioreba, Switzerland), pooled together, and 500 µL of the solution was pressed into the leaf tissue of the experimental plants with a syringe. To create the PlLV inoculum, plant sap from both the infected field-collected plants and the quinoa was combined to produce sufficient PlLV material. Leaves of the infected plants were crushed in phosphate buffer inside plastic extraction bags (Bioreba, Switzerland), pooled together, and 500 µL of the solution was pressed into the leaf tissue with a syringe. Two leaves were inoculated on each plant. In treatments with both PlLV and bulk virus inoculums, two different leaves were inoculated, one with bulk virus inoculum and another one with PlLV. In treatments with only PlLV or virus bulk inoculum, one leaf was mock inoculated with phosphate buffer and the other leaf was inoculated with the virus material. In treatments without viruses, two leaves were mock inoculated with the buffer.

Phomopsis subordinaria was inoculated 6 weeks after transplanting the experimental plants (second week of August) once enough plants had produced flowers. Flowers were necessary because the infection begins from the flower stalk (de Nooij & van der Aa, 1987). Inoculation was conducted by pipetting 500 µL of water containing P. subordinaria spores below the ear of the flower, then puncturing a needle through the spore solution into the plant tissue. All flower stalks were inoculated. Plants that were not inoculated due to lack of flowers were removed from the analyses. Phomopsis subordinaria was inoculated only once because the success of P. subordinaria inoculation can be confirmed in the field based on symptoms shortly after inoculation, unlike with viruses and P. plantaginis.

To analyze how parasite community composition affects host fitness, we measured plant growth and flowering three times over the course of the experiment and collected seeds at the end of the experiment (Figure 1). First, to investigate the effects of parasite inoculation on growth rate over time, we took preinoculation measurements of the size of the plants (hereafter called initial size) in the second week of June in the greenhouse where the plants were maintained before transplanting into the common-garden (Figure 1). We performed the first measurement of the experiment 5 weeks after transplanting (beginning of August) and a second measurement 14 weeks after transplanting (September). Each time, we measured the length and width of the longest leaf, the number of leaves, and the number of flowers. We used the length and width of the longest leaf and the number of leaves (n_leaves) to calculate plant size as: A × n_leaves, where A = πab is the formula of the area of an ellipse. Here, a is the half axis of the width of the longest leaf, and b is the half axis of the length of the longest leaf. We used this size data to calculate relative growth rate: (ln (size₂) − ln (size₁))/t2 − t1, where size₂ is the size at the time of the first measurement or the second measurement, size₁ is the initial size, t₂ is the week when size₂ was recorded (7 weeks at the time of the first measurement and 14 weeks at the time of the second measurement), and t₁ is the initial time point in June (zero). In the end of the experiment (beginning of October), we collected and counted seeds for up to six inflorescences per plant.

Testing of virus infection in experimental plants

To confirm that the plants had received living virus particles in the inoculation, we tested a subset of plants that had been inoculated with PlLV for PlLV, and another subset among plants that were inoculated with virus bulk treatment for PlLV and four additional, most common viruses in the P. lanceolata populations in the Åland Islands, using specific PCR primers (Sallinen et al., 2020; Susi et al., 2019). Leaf samples were collected from the experimental plants 3 weeks after the second virus inoculations, in the later half of August. For detailed PCR protocols and information on primers, see Supplementary Methods. We tested 46 plants that had received PlLV inoculation and 30 plants in the bulk virus treatment for PlLV. Among the plants inoculated with the virus bulk inoculum, 30 plants were also tested for Caulimovirus, and 24 were tested for three additional viruses (Enamovirus, Closterovirus, and Betapartitivirus).

Statistical analysis

To test our hypotheses that the parasite treatment and host genotype interact to affect host growth rate, reproductive effort, and allocation ratios between the two, we fit a set of linear and generalized linear models with an interaction between the parasite treatment (7 categories) and plant genotype (4 categories) as response variables. All analyses were performed using R software (version 4.0.3; R Core Team, 2019).

To analyze whether parasite community and plant genotype influenced the relative growth rate of the experimental plants, we fit two linear mixed-effects models (lmer, package nlme; Bates et al., 2015): one with growth rate from the initial measurement of the experimental plants to the first measurement as a response variable and another with growth rate from the initial measurement to the second measurement as response variable. Both models had an explanatory interaction variable of parasite treatment (7 levels) and plant genotype and included plot ID as a random effect, nested within parasite inoculation treatment (15 levels), because plants in each plot were all inoculated with the same inoculations. No other variables were included. We confirmed that our data met the assumptions of normality and homoscedasticity using residual diagnostics. Significance of the main effects was tested using Wald χ² tests (function “Anova” in package “car,” Fox & Weisberg, 2019) and predicted responses, and standard errors of the significant fixed effects were obtained using the “effects” package (Fox, 1987; Fox & Weisberg, 2019).

To analyze whether parasite community and plant genotype affect reproduction of the host plants, measured as the number of flowers and the number of seeds, we fit a set of generalized linear models. To test the effects on the number of flowers in the first and in the second measurement, we fit a negative-binomial generalized mixed-effects model to account for overdispersion and a generalized Poisson model to account for underdispersion, respectively (package glmmTB, Brooks et al., 2017). An interaction term between parasite treatment and plant genotype, the initial number of flowers, and the current size (size in the first measurement in the model of the number of flowers at the time of the first measurement, size at the time of the second measurement for the number of flowers at the time of the second measurement) were incorporated as fixed effects. Plot ID, nested within parasite inoculation treatment, was included as a random effect. Significance of the main effects was tested using Wald χ² tests (function “Anova” in package “car,” Fox & Weisberg, 2019) and predicted responses, and standard errors of the significant fixed effects were obtained using the “effects” package (Fox, 1987; Fox & Weisberg, 2019). The effects of parasite community and plant genotype on the number of seeds were analyzed with a zero-inflated negative-binomial mixed-effects model to account for overdispersion and zero-inflation (package glmmTB). In addition to an interaction term between parasite treatment and plant genotype, plant size and the number of flowers at the time of the second measurement were included as fixed effects. Plot ID, nested within parasite inoculation treatment, was included as random effect. Significance of the main effects was tested using Wald χ² tests (function “Anova” in package “car,” Fox & Weisberg, 2019) and predicted responses, and standard errors of the significant fixed effects were obtained using the “effects” package (Fox, 1987; Fox & Weisberg, 2019).

To test whether parasite community and plant genotype affect resource allocation, we ran two linear mixed-effects models (lmer, package nlme Bates et al., 2015). The ratio of the number of flowers to growth rate was used as the response variable in the first model and the ratio of the number of seeds to growth rate was used as the response variable in the second model. Both models included an interaction between parasite treatment and plant genotype. Plot ID, nested within treatment, was included as a random effect. Additionally, the model of seeds to growth rate included the size of the plant. Measurements taken at the end of the experiment were used.

When a significant main effect of parasite community treatment was found, we performed a Dunnett’s test (R-package DescTools, function DunnettTest; Andri et al., 2002) to compare the control with the parasite treatments. To disentangle the effects within the cross-kingdom treatment and of pairs of parasites, we also performed planned comparisons between single-parasite treatment pairs, and virus treatment with viruses with single fungus or viruses with both fungi. Both the Dunnett’s test and the comparisons were calculated with a simple ANOVA of the response variable explained by treatment and using base R commands.

For the analyses, we excluded the 90 plants that belonged to P. subordinaria treatments but were not inoculated due to the lack of flowers on the inoculation day, and did not get infected during the experiment. Hence, 330 plants remained in the analyses (Supplementary Table S1). For seed data, 12 plants were removed due to mislabeling of seed bags. In addition to these, an outlier in the data on the number of flowers and two outliers in the ratio of the number of flowers and growth rate affected model fit and were hence removed from the analysis (Supplementary Figure S6). Significance of the main effects was not affected by removing these outliers.

Results

Virus detection results and fungal infection rate

To confirm that the plants received living virus particles in the inoculations, we tested a subset of plants with virus-specific PCR primers. Of the 46 tested plants inoculated with PlLV, 26 were infected (57%). Among the plants inoculated with the bulk virus inoculum that we tested for five viruses, we found plants infected with four viruses. Of the 30 plants inoculated with bulk virus inoculum that we tested for PlLV and Caulimovirus, we found PlLV in 8 plants (27%) and Caulimovirus in 8 plants (27%). Among the 24 plants that were inoculated with bulk virus inoculum and that we tested for Closterovirus, Betapartitivirus, and Enamovirus, Closterovirus was found in 21 plants (88%), Betapartitivirus was found in 4 plants (17 %), and Enamovirus was not found. Fungal infections were recorded at the times of measurements but not followed in between. All of the P. subordinaria inoculated plants got infected. Of the plants inoculated with P. plantaginis, symptoms were observed on seven plants at the time of the measurements. This is likely to be an underestimation as signs of infection are challenging to document with the naked eye. Because infection status was not confirmed for all fungal and viral treatments, we analyzed the data using inoculation treatment groups as an explanatory variable.

Fitness measure results

Relative growth rate of the experimental plants from the initial size to the time of the first measurement (Figure 2A) was significantly affected by an interaction between plant genotype and treatment (χ² = 22.53, df = 12, p = .032, Supplementary Tables S5 and S6). Relative growth rate from initial size to the time of the second measurement (Figure 2B) was also significantly affected by an interaction between plant genotype and treatment (χ² = 36.06, df = 18, p = .007, Supplementary Tables S7 and S8). At both time points, the magnitude and direction of the effect of parasite community on relative growth rate was variable among genotypes (Figure 2, Supplementary Tables S5 and S7). At the second time point, growth rates were higher when the plants of genotype 1 were singly inoculated with either the fungi or the virus mixture, and lower when singly inoculated with PlLV or with fungus–virus mixture (Figure 2B, Supplementary Table S7). In genotype 2, control plants had the lowest growth rate compared with the inoculated plants (Figure 2B, Supplementary Table S7). In genotype 3, growth rates were lower when the plants were singly inoculated with P. subordinaria, fungus mixture, or virus mixture but higher with fungus–virus mixture (Figure 2B, Supplementary Table S7). In genotype 4, growth rate was lowest in plants inoculated with P. plantaginis and second lowest in the control plants (Figure 2B, Supplementary Table S7). In comparison, at the time of the first measurement, all parasite treatments reduced the growth rate of plants of genotype 1 and the direction of the effect of the treatments also changed direction within other genotypes between the first and second measurements (Figure 2, Supplementary Tables S5 and S7). Hence, the effect of parasite treatments on growth rates may shift during the season, indicating that the effects of parasite communities may change during the host’s life span.

The number of flowers produced by the host plants at the time of the first measurement (Supplementary Figure S2, Supplementary Tables S11 and S12) was affected by plant genotype (χ² = 117.78, df = 3, p < .001), the initial number of flowers (χ² = 7.62, df = 1, p = .006), and host size (χ² = 68.42, df = 1, p < .001), but was not affected by the main effect of treatment (χ² = 4.69, df = 4, p = .32) nor the interaction between plant genotype and parasite treatment (χ² = 17.64, df = 12, p = .13). The number of flowers produced by the host at the time of the second measurement (Supplementary Figure S3, Supplementary Tables S9 and S10) was affected by plant genotype (χ² = 280.36, df = 3, p < .001), initial number of flowers (χ² = 14.86, df = 1, p < .001), and current size (χ² = 130.88, df = 1, p < .001). Parasite treatment did not affect flower production (χ² = 6.39, df = 6, p = .38), and there was no significant interaction between plant genotype and parasite treatment (χ² = 17.62, df = 18, p = .48). Number of seeds produced by the experimental plants by the time of the second measurement (Figure 3, Supplementary Tables S13 and S14) was not significantly affected by the interaction between plant genotype and parasite treatment (χ² = 7.85, df = 18, p = .98).

However, parasite treatment had a significant main effect on seed production (χ² = 42.14, df = 6, p < .001). Seed production was also significantly affected by plant genotype (χ² = 39.69, df = 3, p < .001) and the number of flowers (χ² = 15.76, df = 1, p < .001), but not by size (χ² = 0.03, df = 1, p = .85). Viruses increased seed production: Experimental plants that received a virus mixture produced on average 33.07% more seeds than mock-inoculated control plants and 17.32% more if they received the PlLV alone treatment. The other treatments decreased seed production so that with P. plantaginis, seed production was on average 11.94% lower; with P. subordinaria, it was 52.59% lower, with a fungal mixture treatment, it was 28.97% lower, and with the cross-kingdom mixture, it was 41.36% lower than in the mock-inoculated control. While plants that received the fungal mixture had higher seed production than plants in sole P. subordinaria treatment, they had lower seed production than plants in the P. plantaginis treatment. While both virus treatments increased seed production, combined with fungi into a cross-kingdom mixture, the seed production was again lower (Figure 3, Supplementary Table S13). In the Dunnett’s test calculated based on a simple ANOVA on the seed numbers explained by parasite treatments comparing the treatments to the control suggests that out of the parasite treatments, significantly differed P. subordinaria (difference = −85.96, p = .02), virus mixture (difference = 74.15, p > .01) and the cross-kingdom mixture (difference −84.98, p > .01). In the planned contrasts between pairs of parasite treatments, there were two significant pairs. Plants inoculated with both P. subordinaria with PlLV produced less seeds than plants inoculated with P. subordinaria only (estimate −41.56, t-value = −2.02, p = .4). Also, plants inoculated with PlLV only produced more seeds than plants inoculated with PlLV and P. subordinaria (estimate = 51.55, t-value = 2.17, p = .03). To disentangle the effects within the cross-kingdom treatment, we also compared plants that received viruses to those that received viruses with one fungus or two fungi. Both of these comparisons were significant, so that the plants that only received virus produced more seeds (estimate 42.28, t-value = 5.8, p > .01 virus vs. virus with a single fungus; estimate 56.05, t-value = 6.7, p > .01).

Both analyzed allocation ratios were affected by plant genotype (Figure 4), but not by treatment. The ratio of the number of flowers to growth rate (Supplementary Figure S4, Supplementary Tables S15 and S16) was significantly affected by plant genotype (Figure 4, χ² = 163.71, df = 3, p < .001), while parasite treatment did not have a significant effect (χ² = 4.10, df = 6, p = .66), nor was there an interaction between plant genotype and parasite treatment (χ² = 17.77, df = 18, p = .47). The ratio of the number of seeds to growth rate (Supplementary Figure S5, Supplementary Tables S17 and S18) was affected by plant genotype (Figure 4, χ² = 64.87, df = 3, p < .001) and the number of flowers (χ² = 10.25, df = 1, p = .001) but not by treatment (χ² = 4.83, df = 6, p = .16), nor by an interaction between plant genotype and treatment (χ² = 17.48, df = 18, p = .49).

Discussion

Our understanding of how much the complexity and composition of within-host parasite communities as well as genotypic differences among hosts contribute to host growth and reproduction remains poor. In particular, while viruses are among the most diverse and abundant parasites in natural communities, remarkably little is known about their effects on hosts—alone or embedded in parasite communities. We inoculated cloned replicates of four plant genotypes with microbial parasite communities that varied in diversity and taxonomical composition and found that parasite treatment altered seed production independently of genotype, while the effects of parasite treatment on host growth varied among the genotypes. Host genotype was an important determinant of all measures of host fitness. Our results show that within-host parasite communities have the potential to affect the evolution and ecology of host populations through both genotype-specific and direct effects on host growth and seed production, and highlight the importance of accounting for variation among host genotypes when investigating the effects of parasite communities. Furthermore, our results indicate that an attack by a single parasite may sometimes be more harmful for the host than an attack by multiple parasites, demonstrating that experimental testing of multiparasite scenarios is necessary to untangle the effects of parasite diversity in nature.

While coinfection can be more detrimental to the host (Anjos et al., 1992; Ben-ami et al., 2008; Wintermantel, 2005), we did not observe a consistent trend of more complex parasite communities having consistently more negative or positive effects on host fitness. When multiple parasites attack a host, the negative effects may be reduced due to upregulation of host defense responses by first-arriving parasites (Cox, 2001; Halliday et al., 2018) and due to competition between parasites (Mideo, 2009). We find some evidence of a multiparasite attack being less harmful than single parasites. For example, plants of genotype 4 with the fungal mixture treatment had higher growth rate than plants in either of the fungal inoculation treatments, and the negative effect of P. subordinaria on seed production was weaker when P. subordinaria was combined with other parasites compared to the P. subordinaria treatment alone. However, because we only confirmed infections in subsets of plants to make sure that, at the very least, the plant’s immunity was challenged by the parasites and use parasite inoculations as the treatment instead of measuring fungal sporulation and virus titer, it is unclear whether the mechanism behind our results is related to actual coinfection or immunity responses of the host. The added stress of an immunity response triggered by the inoculation of parasites could also explain increased negative effects in multiparasite inoculation treatments. Hence, we conclude that experimental testing of multiparasite scenarios is necessary to untangle the effects and mechanisms of multiparasitism.

We found that parasite treatment influenced seed production in the experimental host individuals. Seed production consistently reduced relative to the healthy control when P. subordinaria was present and consistently increased when the parasite treatment comprised of viruses only. Phomopsis subordinaria reduced seed production by 53% compared with control treatment. However, the negative effect of P. subordinaria was weaker when viruses were also inoculated into the same host. While the negative effect of P. subordinaria was expected based on the biology of the fungus, in particular because it kills infected flower stalk tissue (de Nooij & van der Aa, 1987), the positive effect of viruses was more surprising. Plant viruses have been hypothesized to move along a continuum from mutualism to antagonism (Fraile & García-Arenal, 2016; Roossinck, 2011) and some experiments have shown that the effect of a virus on host growth and reproduction may vary according to abiotic conditions (Aguilar et al., 2017; Hily et al., 2016). Although examples of negative effects of virus infection are more common (Alexander et al., 2017; Malmstrom et al., 2005), limited examples show that a parasite may increase reproduction of the host, at least in the short term (Pagán et al., 2008; Susi et al., 2017b), especially if the plant has evolved tolerance mechanisms with which it is able to sustain reproduction despite parasite replication (Hily et al., 2016; Pagán et al., 2008). Theoretical models and empirical test have also shown that high tolerance could lead to overcompensation where infected individuals have higher reproductive performance than healthy ones (Hily et al., 2016; Mauricio et al., 1997). An increase in reproduction could also result from a shift in allocation of resources from growth to reproduction, as suggested by the life-history theory (Stearns, 1992), but we did not find support for such a shift in our analysis on the allocation ratios. Parasites may however have long-term effects (Alexander et al., 2017; Maskell et al., 1999; Susi & Laine, 2015) and especially for perennial plants, such as P. lanceolata, parasite effects may accumulate over time (Alexander et al., 2017; Susi & Laine, 2015). Therefore, while our results suggest that viruses have a positive effect on seed production, we cannot rule out the possibility that virus infection may have negative effects under more stressful conditions or on future survival and reproduction not captured by our single-season study. In addition, we only detected virus infections in a subset of plants, and while this could be due to only testing the plants once, it is also possible that some of the plants were able to resist the inoculated viruses and hence avoid negative effects of infection.

We observed a strong and consistent effect of plant genotype on host growth and reproduction. The strong genotypic effect was not surprising given that previous research has shown genotype to be a strong determinant of growth in plants and for these fitness metrics in P. lanceolata (Susi & Laine, 2015). We also found that the four P. lanceolata genotypes differed significantly in their allocation between growth, flowering, and seed production. Individuals harbor limited resources and are hence expected to face a trade-off in allocation toward, for example, reproduction or growth, and resistance against stressors (Brown & Rant, 2013; Huot et al., 2014; Stearns, 1992). Since individuals vary in their ability to cope with parasites, parasites may alter the allocation of resources differentially in different host genotypes (Pagán et al., 2008; Susi & Laine, 2015). However, the differences we observed in resource allocation were not affected by the parasite treatments, as indicated by the lack of main effects of treatments and interactions between treatments and genotype. These results highlight the importance of considering intraspecific genotypic variation when measuring host fitness.

Among the fitness measurements we took, growth rate was the only one affected by the interaction between parasite treatment and plant genotype. Growth rate either decreased or increased under parasite attack, and the direction and magnitude of change depended on the host genotype and type of parasite treatment. We measured growth over two time intervals because younger plants may be more affected by parasites (Kus et al., 2002; Raftoyannis & Dick, 2002; Whalen, 2005), and hence, parasite effects may be more pronounced earlier in the season. However, none of the parasite treatments had a consistent negative or positive effect on host growth across the genotypes over the two time intervals. Surprisingly, growth rates were typically higher when the plants were inoculated with parasites compared to the healthy controls. Our analysis of the allocation ratios indicates that the higher growth rate was not caused by a change in resource allocation from reproduction toward growth, but rather by other mechanisms. One such mechanism could be overcompensation, where regrowth induced by damage leads to higher growth rate or reproduction rate compared to undamaged plants (Belsky et al., 1993; Hily et al., 2016). During this experiment, the plants were watered, grown in fertile and ample soil, and isolated from herbivores. Hence, it is possible that the abundant resources helped the plants to buffer negative effects of parasitism. The effects of parasitism may be highly context dependent (Hily et al., 2016; Laine, 2007). Using a single experimental field and rather controlled growing conditions, we may have excluded potentially important factors that could alter the effect of these parasites in the wild. Our experiment also only lasted for one growing season and hence covers only one set of environmental conditions. Further studies would be required to gain more in-depth understanding of whether the results hold under a broader range of environmental conditions. Testing the robustness of the observed effects under varying abiotic and biotic stressors is an interesting avenue for future research.

Increasing number of experiments has shown how arrival order can change the outcome of parasite–parasite interactions (Clay et al., 2020; Halliday et al., 2020; Hoverman et al., 2013). In this experiment, the parasites were inoculated in a specific order with P. plantaginis first, followed by the viruses, and finally with P. subordinaria. While P. subordinaria can naturally only infect hosts after they produce flowers and hence later in the growing season than viruses or P. plantaginis, the order of arrival could vary for P. plantaginis and viruses in the wild. Limited evidence shows that preoccurring virus infection can slow the sporulation of fungal parasites (Bassanezi et al., 1998; Marte et al., 1993; Omar et al., 1986; Potter, 1980; Susi et al., 2022) and hence potentially affect the ecology and evolution of hosts and parasites via altering not only disease severity but also parasite transmission. Previous research in this system indicates that coinfection of the parasites may affect transmission; the spread and sporulation of P. subordinaria in coinfection with PlLV and P. plantaginis (Sallinen et al., 2022; Susi et al., 2022). Further experiments testing different arrival orders and measuring virus titer and fungal symptoms would help build a more complete picture of the host–parasite community dynamics in this system.

In conclusion, our results demonstrate that parasite diversity can have varying effects on host growth and reproduction, and help fill the knowledge gap of the effects of parasite communities including plant viruses with other plant parasites. The notion that the effect of parasites can change magnitude and direction when additional parasites are included demonstrates that to understand the role of parasites for host populations, multiparasite communities need to be accounted for. Multiparasite communities are common in natural environments and affect most organisms in both wild and managed habitats, including humans (Dobson et al., 2008). Understanding how the diversity of within-host parasite communities affects host growth and reproduction are key steps in linking parasite diversity to the evolution and ecology of hosts and parasites.

Data availability

The data are archived and publicly available in Dryad repository https://doi.org/10.5061/dryad.rjdfn2zhc.

Author contributions

S.S. and A.-L.L. designed the study. S.S. collected and analyzed the data. S.S. and A-L.L wrote the manuscript.

Conflict of interest: The authors declares no conflict of interest.

Acknowledgments

We thank Krista Raveala, Pauliina Hyttinen, Alma Oksanen, Oskari Lehtinen, Laura Häkkinen, Elise Schuurman, and the staff in Lammi Biological station for their help with the experiment. We thank Jenalle Eck and Elina Kaarlejärvi for their insightful feedback on the manuscript. We thank Seraina Cappelli for helpful discussions and Eetu Mykkänen for help with a figure. This work was funded by the European Research Council (4100097 RESISTANCE) and Academy of Finland (334276) to A.-L.L. and LUOVA doctoral programme of University of Helsinki.

References