Environment and host genotype determine the outcome of a plant–virus interaction: from antagonism to mutualism
Summary
- It has been hypothesized that plant–virus interactions vary between antagonism and conditional mutualism according to environmental conditions. This hypothesis is based on scant experimental evidence, and to test it we examined the effect of abiotic factors on the Arabidopsis thaliana–Cucumber mosaic virus (CMV) interaction.
- Four Arabidopsis genotypes clustering into two allometric groups were grown under six environments defined by three temperature and two light-intensity conditions. Plants were either CMV-infected or mock-inoculated, and the effects of environment and infection on temporal and resource allocation life-history traits were quantified.
- Life-history traits significantly differed between allometric groups over all environments, with group 1 plants tolerating abiotic stress better than those of group 2. The effect of CMV infection on host fitness (virulence) differed between genotypes, being lower in group 1 genotypes. Tolerance to abiotic stress and to infection was similarly achieved through life-history trait responses, which resulted in resource reallocation from growth to reproduction. Effects of infection varied according to plant genotype and environment from detrimental to beneficial for host fitness.
- These results are highly relevant and demonstrate that plant viruses can be pleiotropic parasites along the antagonism–mutualism continuum, which should be considered in analyses of the evolution of plant–virus interactions.
Introduction
It is widely accepted that viruses are important pathogens of plants (Anderson et al., 2004). This concept derives primarily from the study of viruses that cause diseases in crops, which may set a bias towards antagonistic plant–virus interactions. Studies of plant–virus interactions in wild ecosystems are still limited (Cooper & Jones, 2006; Alexander et al., 2014; Roossinck & García-Arenal, 2015) and although there are well documented examples of viruses causing obvious diseases in wild plants, affecting plant population sizes and plant ecosystem composition (Malmstrom et al., 2005; Power et al., 2011; Rúa et al., 2011; Rodelo-Urrego et al., 2013; Prendeville et al., 2014), most virus infections in wild plants are asymptomatic (Pagán et al., 2010; Prendeville et al., 2012; Roossinck, 2012; Stobbe & Roossinck, 2014). Indeed, it has been proposed that plant viruses would most often be commensals, or even mutualists of plants, and that pathogenic virus infections may be, at least in part, a result of the specific conditions of agricultural ecosystems (Gibbs, 1980; Wren et al., 2006; Xu et al., 2008; Roossinck, 2011). Thus, the pathogenicity of viruses for plants is an ongoing subject of debate, and it is important to understand under which conditions viruses will be virulent parasites of plants.
Virulence has been defined as the negative impact of parasite infection on the host fitness (Read, 1994; Little et al., 2010). Virulence results from pathology of the host, and is thus determined by traits of both the host and parasite (Little et al., 2010). Host defenses may decrease virulence, either through resistance, which results in less effective parasite multiplication and in lesser parasite load, or through tolerance, which specifically decreases virulence regardless of the amount of parasite multiplication (Little et al., 2010; Råberg, 2014). While resistance of plants to viruses, and its various mechanisms, has been extensively analyzed (Csorba et al., 2009; Truniger & Aranda, 2009; de Ronde et al., 2014), plant tolerance to virus has been comparatively neglected (Jeger et al., 2006). Virus multiplication and pathogenesis, as well as plant defense (Pagán et al., 2009; Roden & Ingle, 2009; Cheng et al., 2013), are known to be modulated by environmental conditions, suggesting that virulence or, more generally, the outcome of virus infection will be environment-dependent. This topic is now the object of a renewed interest under the present conditions of global climate change.
The purpose of this work is to analyze how the environment modulates the outcome of plant–virus interactions. We focus on the wild plant Arabidopsis thaliana (Brassicaceae) and the generalist virus Cucumber mosaic virus (CMV, Bromoviridae). A. thaliana (from here on Arabidopsis) is increasingly used as model for analyses of the evolutionary ecology of plant–parasite interactions (Salvaudon et al., 2005; Goss & Bergelson, 2006; Kover et al., 2009; Atwell et al., 2010; Pagán et al., 2010; Karasov et al., 2014a,b). CMV has a three-partite, single-stranded, messenger sense, RNA genome encapsidated in isometric particles. CMV presents an extremely broad host range, infecting over 1200 plant species, and is efficiently transmitted by > 75 aphid species in a nonpersistent manner (Jacquemond, 2012). CMV is also transmitted through the seed, with efficiencies in Arabidopsis of 2–8%, according to the host and virus genotypes (Hily et al., 2014; Pagán et al., 2014). Analyses from wild Arabidopsis populations in central Spain have shown that CMV had a high prevalence of up to 70% according to population and year (Pagán et al., 2010), indicating that the Arabidopsis–CMV interaction is significant in nature. In this plant–virus system, virulence and virus multiplication are uncoupled as a result of virus genotype × host genotype-specific tolerance (Pagán et al., 2007). Under controlled conditions, tolerance varied over Arabidopsis genotypes as a quantitative trait with moderate to high heritability, and long-lived genotypes with a low seed production to total biomass ratio (group 1 genotypes) were more tolerant than short-lived genotypes that had a high seed to biomass ratio (group 2 genotypes) (Pagán et al., 2007). It was also shown that tolerance in group 1 genotypes was attained through plastic modifications of life-history traits upon CMV infection, mainly the reallocation of resources from growth to reproduction and, to a lesser degree, by a delay of flowering time (Pagán et al., 2008). Life-history theory predicts that resource investment by organisms to different fitness components, such as growth and reproduction, will be adjusted to maximize fitness in different environments, including parasite infection (van Noordwijk & de Jong, 1986; Michalakis & Hochberg, 1994; Koella & Agnew, 1999; Agnew et al., 2000). Because tradeoffs limit resource allocation, adjustment to different environmental conditions will affect tolerance. Indeed, this was the case in Arabidopsis, as plant density was a determinant of resource allocation and of tolerance to CMV (Pagán et al., 2009).
Here we analyze the effect of two major environmental factors that determine plant growth and development – light and temperature – on the outcome of the Arabidopsis–CMV interaction. Four Arabidopsis genotypes previously rated as differing in tolerance to CMV, were inoculated with CMV and grown in similar conditions, within a range of light intensity and temperature. Data showed that environment had a great impact on the developmental schedule and architecture of uninfected plants, as well as on the plant's response to virus infection. As a consequence, plant–virus interactions ranged from mutualistic to antagonistic, according to differential effects of environmental conditions on the assayed plant genotypes.
Materials and Methods
Arabidopsis thaliana genotypes and growing conditions
Four genotypes of Arabidopsis thaliana (L.) Heynh., rated by Pagán et al. (2007) as nontolerant (Columbia glabrata 1 (Colgb1) and Landsberg erecta 0 (Ler)) or as tolerant to CMV (Cumbres Mayores (Cum-0) and Llagostera (Ll-0)) were multiplied simultaneously under the same glasshouse conditions to minimize maternal effects. Before in vitro germination, Arabidopsis seeds were surface-sterilized, then plated on media containing half-strength Murashige and Skoog basal salt mix (Murashige & Skoog, 1962), and stratified in the dark at 4°C for 96 h before being transferred to a growth chamber at 22°C under long-day conditions (220–250 μmol m−2 s−1) with 65–70% relative humidity. Five-day-old seedlings were then transferred to 96-well trays containing a 3 : 1 peat : vermiculite substrate. After 10 d, plantlets were transferred into larger, 10-cm-diameter pots with the same substrate, in order to reduce resource limitation.
Virus isolation, inoculation and detection methods
LS-CMV, belonging to subgroup II of CMV strains, was derived from biologically active cDNA clones (Zhang et al., 1994). Virions were purified as in Lot et al. (1972) and viral RNA was extracted by virion disruption with phenol and sodium dodecyl sulfate. Plantlets were mechanically inoculated at stage 1.04–1.05 (Boyes et al., 2001) with either 15 μl of a 100 μg ml−1 suspension of LS-CMV RNA in 0.1 M Na2HPO4, or the buffer itself (mock-inoculated controls). Following inoculation, plants were placed in three Heraeus-VB1514 bioline growth chambers (Nordrhein-Westfalen, Germany) at 65–70% relative humidity, three different temperatures (17, 22, and 27°C) and two light intensities (high light (HL), 220–250 μmol m−2 s−1, and low light (LL), 45–60 μmol m−2 s−1; 16 : 8 h, light : dark), so that six environments (three temperatures for each light intensity) were assayed. These conditions were chosen on the basis of preliminary experiments in which the performance of the four assayed Arabidopsis genotypes under different conditions, including these, was evaluated (not shown). For each of the 12 treatments (mock-inoculated and CMV-infected for each environment), eight replicated plants were included.
Cucumber mosaic virus was detected and quantified as viral RNA accumulation via quantitative real-time reverse transcription-PCR. For each plant, three leaf discs, 4 mm in diameter, were randomly excised from systemically infected leaves at 10 d postinoculation (dpi) and total nucleic acids were extracted using Trizol® reagent (Life Technologies, Carlsbad, CA, USA). For each run, 0.5–3 ng of total RNA was added to the Brilliant III Ultra-Fast SYBR® Green qRT-PCR Master Mix according to the manufacturer's recommendations (Agilent Technologies) in a final volume of 10 μl. Each plant sample was assayed in triplicate on a Light Cycler 480 II real-time PCR system (Roche) and relative levels of gene expression were deduced from standards as previously described (Hily et al., 2014).
Evaluation of Arabidopsis life-history traits
Temporal parameters of Arabidopsis life cycle were evaluated: life span (LP) was measured as the number of d between the end of stratification and senescence (considered as the time at which 50% of siliques had shattered). LP was divided into the preflowering vegetative growth period (GP, time in d between the end of stratification and the opening of the first flower; stage 6.0 of Boyes et al., 2001) and the postflowering period (PFP, comprising the reproductive and silique maturation periods, i.e. the rest of the life span after stage 6.0 until senescence). Rosette relative growth was estimated as previously described (Hily et al., 2014), and the number of rosette leaves was determined at flowering time.
Above-ground plant structures were harvested at complete senescence, and dry weights were determined after incubation at 65°C for 2 wk. The weights (g) of the whole above-ground biomass (BM), rosettes (RW), inflorescence structures without seeds (IW), and seeds (SW) were obtained separately for each plant. RW represents the vegetative growth effort, and SW represents the reproductive effort; the ratio (SW : RW) was used as a proxy of resource allocation to reproduction relative to growth. In addition, single seed weight (SSW, in μg) was estimated after collecting all seeds from 11 to 13 siliques from three plants per treatment, which were then dried, weighed, and counted. After a 10 month dormancy period, the percentage of seed germination (%germ) was established, following the aforementioned germination protocol. Fitness was evaluated as the viable seed production (VS), which was estimated for each plant from total seed production, SSW and seed germination.
Data analyses
The values of resource allocation life-history traits (BM, RW, SW etc.) and the various ratios between them were normally distributed according to a goodness-of-fit test, but were not homoscedastic according to Levene's test of equality of error variances. On the other hand, the distribution of the data of temporal life-history traits (life span, vegetative growth period, reproductive period, etc.) was not normal but was homoscedastic. Data of virus accumulation were both normal and homoscedastic. Consequently, all life-history traits were analyzed by parametric (ANOVA) and nonparametric (Kruskal–Wallis) tests. Both Kruskal–Wallis and ANOVA tests gave similar results when the effect of the different factors (plant genotype, infection status, light, and temperature) on these traits was analyzed. Thus, since ANOVA is robust to the partial violation of its assumptions, and allows the analysis of factor interactions, while Kruskal–Wallis doesn't, for simplicity, only ANOVA analyses are presented. Full factorial ANOVA models were used to analyze data of life-history traits, and of virus accumulation, according to allometry group, genotype nested within group, light and temperature, considering these four factors as fixed effects.
The effect of infection in host life-history traits was analyzed by similar full factorial nested ANOVA models, using as variables the ratio between the trait values in infected and mock-inoculated plants, computed by dividing the life-history trait value of each infected plant (i) by the mean value of the eight replicas of mock-inoculated controls (m) of the same genotype and treatment (traiti : traitm). Significance of differences among classes within each factor was determined by least significant difference (LSD) analyses.
Values of trait variables of mock-inoculated and infected plants were used also in a principal component analysis (PCA), with loading scores on the PC1 axis representing a continuous range of variables predicting group/genotype, while loading scores on the PC2 axis represented light and status factors.
Analyses were performed using the software packages Statgraphics Centurion 15.1.02 (Stat Point technologies, Inc., Warrenton, VA, USA) and SPSS v. 21 (SPSS Inc. Chicago, IL, USA).
Results
Developmental schedule and resource allocation to different life-history traits of Arabidopsis genotypes
The four Arabidopsis genotypes used in this study were classified by Pagán et al. (2007, 2008) as belonging to two different groups according to allometric relations between seed production and growth: Cum-0 and Ll-0 belonged to group 1, characterized by a lower SW : RW ratio and a longer LP, while Colgb1 and Ler belonged to group 2, characterized by a higher SW : RW ratio and a shorter LP. Our first aim was to analyze in mock-inoculated plants if reported differences in allometry and developmental schedule held across six environmental conditions assayed, that is, three temperatures (17, 22 and 27°C) and two light intensities (see the 2 section) for each temperature. Full factorial ANOVA models were used to analyze data of life-history traits according to allometry group, genotype nested within group, light, and temperature.
Neither group nor genotype nested within group had significant effects on LP (F1,159 = 4.59, P = 0.165 for group; F2,159 = 4.75, P = 0.189 for genotype), while light and temperature did (F1,159 = 46.50, P = 0.021 for light; F2,159 = 503.98, P < 0.001 for temperature). In addition, the interactions group × temperature and group × light × temperature had significant effects on LP (F2,159 = 40.74, P = 0.002; and F4,159 = 31.82, P = 0.003, respectively). For each of the six environmental conditions tested, the LP of group 1 plants was significantly longer than that of group 2 plants (Fig. 1a). Similar results were obtained when comparing the vegetative growth (GP) and reproductive period (RP) (not shown). It is worth noting that group had a significant effect on the ratio GP : LP (F1,159 = 316.03, P < 0.001). Group 1 plants invested half of their LP in vegetative growth, whereas group 2 plants invested only one-third (Fig. 1b; Table 1).
| Group 1 | Group 2 | |||
|---|---|---|---|---|
| n | X ± SE | n | X ± SE | |
| Temporal traits | ||||
| GP : LP | 94 | 0.36 ± 0.01 | 89 | 0.49 ± 0.01 |
| Morphological and physiological traits | ||||
| BM (g) | 94 | 3.74 ± 0.15 | 89 | 1.44 ± 0.13 |
| SW (g) | 94 | 0.85 ± 0.04 | 89 | 0.50 ± 0.04 |
| SW : RW | 94 | 2.40 ± 0.56 | 89 | 14.80 ± 0.59 |
| Rosette leaf number | 94 | 83.55 ± 5.33 | 89 | 10.66 ± 0.45 |
| Rosette relative growth (cm d−1) | 60 | 0.18 ± 0.01 | 60 | 0.10 ± 0.01 |
| SSW (μg) | 35 | 20.17 ± 0.76 | 34 | 19.98 ± 0.62 |
| Germination percentage | 66 | 68.89 ± 2.02 | 62 | 72.19 ± 2.45 |
- a Data are means ± SE of n individuals for each trait, considering the pooled data from six different environmental conditions. Traits are: GP, vegetative growth period; LP, life span; BM, biomass; SW, seed weight; RW, rosette weight; SSW, single seed weight.
The four Arabidopsis genotypes were also split into two groups according to plant architecture and to allometry between plant parts. Group, but not genotype nested within group, had a significant effect on plant BM (F1,159 = 16.70, P = 0.055 for group; F2,159 = 5.95, P = 0.100 for genotype), light and temperature (F1,159 = 51.25, P = 0.019 for light; F2,159 = 87.62, P < 0.001 for temperature) and the triple interaction group × light × temperature had also significant effects on biomass (F4,159 = 30.01, P = 0.003). Similar results were obtained when analysing SW (not shown). Over all conditions, group 1 plants had a higher biomass and produced more seeds (mean ± SE, 3.74 ± 0.15 and 0.85 ± 0.04 g, respectively) than group 2 plants (for which BM and SW were of 1.44 ± 0.13 and 0.50 ± 0.04 g, respectively), and this held for each specific environment (Fig. 1c,d; Table 1). However, RW depended on genotype but not on group (F1,159 = 5.81, P = 0.137 for group; F2,159 = 7.86, P = 0.027 for genotype), and on light but not on temperature (F1,159 = 23.78, P = 0.040 for light; F2,159 = 0.482, P = 0.649 for temperature), with a significant interaction group × light (F1,159 = 16.59, P = 0.055). Other evaluated traits were the number of rosette leaves at flowering and rosette growth over a 12 d period. Both group and genotype nested within group had significant effects on the number of rosette leaves (F1,159 = 1406.20, P = 0.001 for group; F2,159 = 543.42, P = 0.001 for genotype), light, temperature, and the interactions of these factors with group also having significant effects (F > 3.90, P < 0.022), and this trait showed higher values for group 1 than for group 2 plants in all conditions (Fig. 1f; Table 1). Similar results were obtained for rosette growth rate (F1,159 = 77.33, P = 0.001 for group; F2,159 = 18.00, P = 0.001 for genotype), which was faster for group 1 plants in all conditions (Fig. 1g; Table 1). Last, genotype nested within group, but not group, had an effect on SSW (F1,46 = 0.13, P = 0.725 for group; F2,46 = 10.74, P = 0.001 for genotype), and both group and genotype nested within group had effects on the percentage of seed germination (F1,104 = 3.91, P = 0.051 for group; F2,104 = 10.66, P = 0.001 for genotype). However, when the effect of group was reassessed by comparing the mean values of the two genotypes within each group (two, rather than 16, values per group), the germination rates did not differ between groups (Fig. 1h; Table 1). Interactions of group or genotype with light or temperature did not affect these traits (F < 2.79, P > 0.103). As for the allometric relationship between SW and RW, group had a more significant effect than genotype (F1,159 = 17.46, P = 0.053 for group; F2,159 = 7.25, P = 0.069 for genotype), temperature (F2,159 = 12.69, P = 0.018) and the triple interaction group × light × temperature (F4,159 = 19.09, P = 0.007) also having significant effects. The SW : RW ratio was significantly lower in group 1 than in group 2 plants under all conditions (overall values 2.40 ± 0.56 vs 14.8 ± 0.59; Fig. 1h; Table 1), indicating that group 1 plants allocate significantly more resources to growth than to reproduction, as compared with group 2 plants.
Thus, the split of Arabidopsis genotypes into two allometric groups proposed by Pagán et al. (2007, 2008) held over a range of environmental conditions. Moreover, differences in life-history traits between group 1 and group 2 plants indicated a fundamental rift in their life-history strategy and, potentially, in their responses to environmental changes.
Effect of the environmental conditions on life-history traits of plants of group 1 and group 2
As significant effects of group, light, and temperature were found for most life-history traits analyzed, we studied the impact of abiotic factors on these traits. We compared six environmental conditions, as defined earlier and in the 2 section, by three-way ANOVA, with group, light, and temperature as factors. Analyses were restricted to traits in which the ANOVA presented in the previous section showed significant effects of group and/or the interactions involving group.
Environmental conditions had a similar effect on LP regardless of whether plants belonged to group 1 or group 2 (Fig. 1a; Supporting Information Table S1). LP was significantly reduced upon the increase in light intensity (HL < LL, F1,171 = 427.43, P < 1 × 10−4) and temperature (17°C > 22°C > 27°C, F2,171 = 258.30, P < 1 × 10−4).
Light and temperature also affected the architecture and allometry of plants according to a similar pattern for both group 1 and group 2 plants, but effects were larger on group 2 plants (Figs 1c,d, S1; Table S1). Plants performed worse under LL than under HL conditions (F1,171 = 208.62, P < 1 × 10−4), with an average loss in BM of 39.8% for plants belonging to group 1 and close to 61% for group 2 plants (Fig. 1c; Table S1). For both groups, BM decreased with increasing temperature (F2,171 = 98.47, P < 1 × 10−4), with a maximum loss of 32.7% for group 1 and of 77.5% for group 2 plants. Plants from both groups performed best at 17°C HL and worst at 27°C LL, with a drastic loss of BM between those two conditions, reaching 69.4% for group 1 and 96.4% for group 2 plants (Fig. 1c; Table S1). Interestingly, the second best environments for group 2 plants (17°C LL and 22°C HL), presented a BM loss > 50% compared with 17°C HL, while for group 1 plants a BM decrease of 50% was only reached in the worst environment, 27°C LL. Thus, BM was less affected by environmental conditions for group 1 than for group 2 plants. The same trends were found for SW (Fig. 1d; Table S1).
Environmental conditions differentially affected the number of rosette leaves, and rosette growth, of plants from each group (Fig. 1f,g; Table S1). While temperature did not affect the number of leaves of group 1 plants (F2,88 = 1.72, P = 0.185), low light resulted in a reduced number of leaves (101.47 ± 7.49 and 67.52 ± 6.58 leaves under HL and LL, respectively; F1,88 = 9.53, P = 0.003). On the other hand, rosette leaf number of group 2 plants was not affected by light (F1,83 = 1.84, P = 0.179), but was reduced as temperature increased (12.31 ± 0.69, 10.84 ± 0.71 and 8.13 ± 0.79 leaves at 17, 22 and 27°C, respectively; F2,83 = 7.97, P = 0.001). Light had the greatest impact on SSW and percentage seed germination (Fig. 1h; Table S1). A decrease in light intensity was associated with a decrease in SSW (F1,29 = 26.34, P < 1 × 10−4 and F1,28 = 58.53, P < 1 × 10−4 for group 1 and group 2, respectively) and seed germination (F1,60 = 6.27, P = 0.015 and F1,56 = 11.21, P = 0.001 for group 1 and group 2, respectively).
For group 1 plants, both light and temperature affected the SW : RW ratio (F ≥ 4.91, P ≤ 0.026), with plants grown under LL conditions and/or low temperatures allocating significantly more resources toward reproduction than growth (Fig. 1e; Table S1). On the other hand, while temperature influenced SW : RW for group 2 plants (F2,83 = 5.55, P = 0.005), light intensity did not (F1,83 = 0.31, P = 0.580) (Fig. 1e; Table S1). This result indicates that group 1 exhibits a higher phenotypic plasticity than group 2, resulting in the reallocation of resources towards reproduction under unfavorable or limiting conditions, such as low light. In all the environments tested, SW : RW ratios were significantly different between group 1 and group 2 plants (F ≥ 34.04, P < 1 × 10−4), showing that while changes in resource allocation occurred across environmental conditions, they were always such that plants never switched allometric group (Figs 1e, S2e). Indeed, a PCA including all the temporal and morphological traits, identified two principal components, PC1 and PC2, that accounted for > 76% of the variability of the data (not shown), and that clearly differentiated group 1 and group 2 plants into the positive and negative regions of PC1, respectively (Fig. S3).
Effect of the environmental conditions on CMV multiplication and on the effect of infection on host plant fitness
Cucumber mosaic virus multiplication, estimated as CMV RNA accumulation in systemically infected tissues, was used to evaluate host resistance/susceptibility to infection. Neither group nor genotype nested within group had significant effects on CMV multiplication (F1,168 = 2.32, P = 0.130 for group; F2,168 = 0.27, P = 0.766 for genotype). Over all assayed conditions, group 1 and group 2 plants were similarly susceptible to CMV, with similar levels of RNA accumulation (6.03 ± 0.73 and 7.35 ± 0.74 pg of CMV RNA ng−1 total RNA for group 1 and group 2, respectively; Table 2). Light had the highest impact on virus multiplication (F1,168 = 59.36, P < 0.001), which was 3.5 and 2.6 times higher in LL than in HL in group 1 and group 2 plants, respectively. Also, a significant decrease (F2,168 = 10.92, P < 0.001) of virus accumulation was observed with increasing temperature, with 8.88 ± 0.76, 7.55 ± 0.74 and 3.89 ± 0.74 pg ng−1 of total RNA at 17, 22 and 27°C, respectively. Environmental conditions greatly affected virus multiplication (Table 2), and to a similar extent for both allometric groups: no interactions between groups and any of the environmental factors were found (F ≤ 1.08, P ≥ 0.342), and no significant differences in virus accumulation were observed between groups in any specific environmental condition (F2,168 = 1.38, P = 0.255 for the interaction group × light × temperature). These results indicate that potential differences observed in the response to CMV infection between allometric groups were not a result of genotype-specific limitation of virus multiplication, that is, were not a result of resistance.
| Environmental conditionsb | Group 1 | Group 2 | ||
|---|---|---|---|---|
| n | X ± SE | n | X ± SE | |
| Overall | 94 | 6.15 ± 0.73 | 96 | 7.35 ± 0.74 |
| 17°C HL | 15 | 3.61 ± 1.18 | 16 | 5.24 ± 1.25 |
| 17°C LL | 15 | 14.91 ± 2.17 | 16 | 11.88 ± 1.97 |
| 22°C HL | 16 | 2.77 ± 0.21 | 16 | 4.26 ± 0.95 |
| 22°C LL | 16 | 9.35 ± 1.99 | 16 | 13.81 ± 2.38 |
| 27°C HL | 16 | 1.83 ± 0.24 | 16 | 2.59 ± 0.72 |
| 27°C LL | 16 | 4.84 ± 1.37 | 16 | 6.30 ± 1.15 |
- a RNA accumulation is expressed as pg of viral RNA ng−1 of total RNA. Data are means ± SE of n individuals for each treatment.
- b Environmental conditions indicate the growth temperature and the degree (HL, high; LL, low) of light intensity. Overall indicates ratio from the analysis of data pooled over all conditions.
The effect of CMV infection on progeny production was estimated as the effect on total number of VS produced per plant, calculated from SW, SSW and %germ data, as a proxy to fitness. The effect of infection on every life-history trait was quantified by comparing trait values of virus-infected plants (i) with the mean value obtained from the eight mock-inoculated control replicas (m). It has been reported that group 1 genotypes were more tolerant to CMV infection, that is, had a higher SWi : SWm ratio, than group 2 plants, which was associated with resource reallocation from growth to reproduction, as shown by a higher SW : RW ratio in CMV-infected than in mock-inoculated group 1 plants (Pagán et al., 2007, 2008). In the present study, we found a trend towards higher tolerance of group 1 plants: the VSi : VSm ratio, was higher for group 1 than for group 2 plants (0.97 ± 0.06 and 0.94 ± 0.05, respectively). However, group had no significant effect on VSi : VSm, while genotype nested within group did (F1,166 = 0.38, P = 0.537 for group; F2,166 = 8.54, P < 0.001 for genotype). Note, however, that group had an effect under specific conditions. For instance, at 22°C HL, group but not genotype had a significant effect on VSi : VSm (F1,28 = 13.93, P = 0.001 for group; F2,28 = 2.41, P = 0.108 for genotype). As group had no significant effect on VSi : VSm, the effects of genotype, light, and temperature on this trait were analyzed. Genotype had a significant effect (F3,166 = 5.72, P = 0.001) and, except in the highly stressfull 27°C LL condition, the effect of infection on plant fitness was lower in group 1 genotypes than in group 2 genotypes. In addition, group, but not genotype nested within group, had a significant effect on the (SW : RW)i : (SW : RW)m ratio (F1,166 = 9.13, P = 0.003 for group; F2,166 = 2.11, P = 0.124 for genotype), neither light nor temperature having effects on this trait (F1,166 = 0.17, P = 0.680 for light; F2,166 = 1.10, P = 0.335 for temperature), indicating reallocation of resources from growth to reproduction specifically in group 1 ((SW : RW)i : (SW : RW)m = 1.72 ± 0.24) but not in group 2 plants ((SW : RW)i : (SW : RW)m = 1.00 ± 0.05) (Fig. 2), regardless of environmental conditions.
Because of the genotype vs group effect of virus infection on VS, for a detailed analysis of how environment modulates the effect of virus infection on the host plant, we focused on the genotypes that showed the most extreme phenotypes (Tables S1, S2), that is, Ll-0 in group 1 and Ler in group 2. For this, multifactorial ANOVA considering infection/noninfection status, light and temperature as fixed effect factors, was used to analyze data on the different traits for each genotype separately. The effect of CMV infection on the relevant life-history traits and its significance are shown in Table 3 and Table S3. Over conditions, virus infection did not affect seed viability, estimated as %germ, either in Ll-0 or in Ler plants (F1,48 = 0.93, P = 0.339; F1,61 = 0.17, P = 0.685, for Ll-0 and Ler, respectively). Virus infection did not affect SSW in Ler plants, although it reduced SSW in Ll-0 plants close to 10% (F1,24 = 22.38, P = 0.001). However, because infection had a negative effect on SW (F1,81 = 7.78, P = 0.007 for Ler; F1,83 = 3.65, P = 0.059 for Ll-0), which was more severe in Ler than in Ll-0 plants (SW was reduced to 81% in Ler and to 94% in Ll-0 infected plants as compared with mock-inoculated ones), the overall VS of Ll-0 plants was not affected by infection (VSi : VSm = 1.08 ± 0.07, F1,83 = 0.58, P = 0.450), while that of Ler plants was reduced by 20% (VSi : VSm = 0.79 ± 0.06, F1,81 = 5.82, P = 0.018). Focusing on the SW component of plant fitness, the effect of infection on Ll-0 plants differed importantly among conditions. SW was actually reduced in infected Ll-0 plants in three environments (17°C LL, 22°C LL and 27°C HL, with F ≥ 7.48, P ≤ 0.016), did not change in two (17°C HL and 27°C LL, with F ≤ 0.09, P ≥ 0.764) and increased at 22°C HL (F1,14 = 6.16, P = 0.026), relative to mock-inoculated controls. The negative impact of virus infection on SW of Ler plants was significant in only two environments (17°C LL and 22°C HL), and in all other conditions, a nonsignificant trend (P ≥ 0.222) towards SW reduction was observed. In summary, the impact of infection on the fitness of Ll-0 plants was smaller than on Ler plants, and such an impact was differentially modulated by the environment in each genotype.
| Traiti : traitm | Overall | Environmental conditions | |||||
|---|---|---|---|---|---|---|---|
| 17°C HL | 22°C HL | 27°C HL | 17°C LL | 22°C LL | 27°C LL | ||
| Ll-0 | |||||||
| LP | 1.11 | 1.02 | 1.13 | 1.25 | 1.17 | 1.09 | 1.00 |
| BM | 0.88 | 0.95 | 1.16 | 0.81 | 0.82 | 0.67 | 0.86 |
| RW | 0.74 | 0.78 | 0.90 | 0.97 | 0.62 | 0.66 | 0.50 |
| SW | 0.94 | 1.01 | 1.19 | 0.69 | 0.81 | 0.77 | 1.15 |
| SW : RW | 1.37 | 1.38 | 1.35 | 0.72 | 1.27 | 1.20 | 2.31 |
| Rosette leaf number | 0.87 | 0.89 | 1.00 | 1.01 | 0.76 | 0.69 | 0.89 |
| SSW | 0.91 | 0.93 | 0.87 | 0.87 | 0.84 | 0.95 | 1.01 |
| Germination percentage | 1.04 | 1.00 | 0.97 | 1.03 | 1.05 | 1.03 | 1.24 |
| Number of viable seeds produced | 1.08 | 1.10 | 1.34 | 0.82 | 1.03 | 0.84 | 1.36 |
| Ler | |||||||
| LP | 1.03 | 1.07 | 1.02 | 1.01 | 0.98 | 1.03 | 1.08 |
| BM | 0.84 | 0.98 | 0.73 | 1.11 | 0.46 | 0.72 | 1.03 |
| RW | 0.88 | 0.76 | 0.67 | 0.95 | 0.47 | 1.38 | 1.08 |
| SW | 0.81 | 1.08 | 0.65 | 1.18 | 0.39 | 0.72 | 0.82 |
| SW : RW | 1.04 | 1.09 | 1.09 | 1.43 | 1.16 | 0.61 | 0.83 |
| Rosette leaf number | 0.95 | 0.94 | 0.85 | 1.12 | 0.74 | 1.03 | 1.04 |
| SSW | 1.00 | 1.08 | 0.98 | 0.98 | 0.98 | 0.98 | 1.01 |
| Germination percentage | 0.96 | 0.97 | 1.03 | 1.01 | 0.80 | 0.88 | 1.11 |
| Number of viable seeds produced | 0.79 | 0.95 | 0.67 | 1.21 | 0.31 | 0.63 | 0.97 |
- a For each condition, data are the mean of the ratios of the trait value for eight infected plants divided by the mean value of eight repetitions of the mock-inoculated controls. Ratio values in bold type or underlined bold type are significantly bigger than 1 at P ≤ 0.05 and P ≤ 0.01 significance level, respectively. Ratio values in italics or underlined italics are significantly smaller than 1 at P ≤ 0.05 and P ≤ 0.01 significance level, respectively. Environmental conditions indicate the growth temperature and the degree (HL, high; LL, low) of light intensity. Overall indicates ratio from the analysis of data pooled over all conditions. Traits are: GP, vegetative growth period; LP, life span; BM, biomass; SW, seed weight; RW, rosette weight; SSW, single seed weight.
Effect of the environmental conditions on life-history trait responses to CMV infection
In the previous section we showed that the effect of CMV infection on the host fitness, that is, virulence, depended on the genotype and the environmental conditions. Here we analyze if virulence is modulated by plant responses to infection, that is, by tolerance mechanisms. For these analyses we focus again on Ll-0 and Ler.
Overall, virus infection resulted in an increase of LP for Ll-0 plants (F1,83 = 50.35, P < 1 × 10−4, Tables 3, S2, S3) that was the result of an increase of both vegetative GP and PFP (F1,83 ≥ 9.52, P ≤ 0.010). No such increase in LP, or in GP and PFP, was observed for Ler plants (F1,81 ≤ 2.23, P ≥ 0.139). Interestingly, the triple interaction between infection status, light, and temperature affected the LP of Ll-0 plants (F2,83 = 13.68, P < 1 × 10−4), indicating that temporal life-history traits were specifically modified in response to virus infection in each environment. On the other hand, this interaction was not significant for Ler plants (F2,81 ≥ 2.35, P ≤ 0.102), showing a limitation in temporal life-history trait response to infection.
Upon infection, both Ll-0 and Ler plants presented an overall reduction in RW and BM (F1,83 ≥ 9.40, P ≤ 0.003, and F1,81 ≥ 5.01, P ≤ 0.028, for Ll-0 and Ler, respectively) (Tables 3, S2, S3). Infection significantly increased SW : RW in Ll-0 plants ((SW : RW)i : (SW : RW)m = 1.37 ± 0.15, F1,83 = 11.71, P = 0.001; Tables 3, S2, S3), but not in Ler plants ((SW : RW)i : (SW : RW)m = 1.04 ± 0.09, F1,81 = 0.36, P = 0.552). This result indicates that upon CMV infection Ll-0 plants reallocated more resources towards reproduction than to growth. Nonetheless, resource reallocation in Ll-0 plants depended on the environment: the (SW : RW)i : (SW : RW)m ratio varied between 0.72 (27°C HL) and 2.31 (27°C LL) (Tables 3, S2, S3). In Ler, (SW : RW)i : (SW : RW)m varied over conditions within a much narrower range, never being significantly different from 1 (Tables 3, S3), indicating no resource reallocation upon infection.
These findings indicate that the lower virulence of CMV infection in Ll-0 plants is a result, at least in part, of environment-modulated responses in life-history traits, leading to resource reallocation from growth to reproduction, that is, a result of tolerance responses. On the other hand, there is no evidence of resource reallocation-based tolerance responses in Ler.
Discussion
In this work we analyze the effects of abiotic factors on virulence and on the outcome of host–parasite interactions, which is a major question in disease ecology and evolution (Lively et al., 2014). We focused on the interaction of the RNA virus CMV and four Arabidopsis genotypes that cluster into two groups according to the allometry of seed production (estimated as SW) to vegetative growth (estimated as RW). Arabidopsis–CMV interactions were assayed under a range of environmental conditions defined by differences in temperature and light intensity. Results show that environmental conditions differentially modulate the outcome of the Arabidopsis–CMV interaction according to host genotype, so that virus infection may be detrimental, neutral, or beneficial to the host plant in terms of viable seed production.
The results also contribute to understanding the processes that determine this variety of outcomes of the plant–virus interaction, which can be related to the interplay of virulence and tolerance (Little et al., 2010). In the Arabidopsis–CMV interaction, tolerance is a result, in part, of responses in life-history traits leading to resource reallocation from growth to reproduction (Pagán et al., 2008). Life-history theory predicts that the evolution of resource investment by organisms will be conditioned by tradeoffs between resource allocation to different fitness components, and the optimal resource allocation will be modified according to environmental conditions (Stearns, 1976). Theory also posits that environmental conditions affecting mortality rates, such as abiotic stress or parasitism, will modify optimal resource allocation to maximize fitness. Thus, models predict that under parasitism, hosts will allocate more resources to reproduction in detriment of resource allocation to growth (Williams, 1966; Minchella, 1985; Forbes, 1993; Perrin et al., 1996). In addition, infection by highly virulent parasites will result in shorter host pre-reproductive periods, so that progeny is produced before resource depletion, while infection by low virulent parasites will delay host reproduction, allowing for compensation of parasite damage (Hochberg et al., 1992; Gandon et al., 2002). Our results, as with previous work (Pagán et al., 2008), largely agree with these predictions, thus providing experimental support for theoretical analyses. It is important to underscore that in the present work this support derives from experiments in which plants were grown under a large range of temperature and light conditions. In most conditions, genotypes of group 1 were more tolerant to CMV infection than those of group 2, which was associated with the capacity of plants to reallocate resources from growth to reproduction. Tolerance to CMV was also associated with an increase in both LP and GP upon infection, which did not occur in the genotypes in which CMV virulence was higher. Note that tolerance was unrelated to virus multiplication, which was affected by environment but did not significantly differ according to genotype or allometric group. Note also that the limited virulence of CMV infection in group 2 genotypes under some environments (see later) cannot be explained by phenotypic plasticity responses, strongly suggesting that there is not a tolerance defense reaction in group 2 plants (Pagán et al., 2008). Plant tolerance to parasitism or herbivory has most often been explained by phenotypic plasticity resulting in resource reallocation to reproduction (Strauss & Agrawal, 1999; Agrawal, 2000; Fellous & Salvaudon, 2009), as reported here. Specifically, Arabidopsis tolerance to apical meristem herbivory was explained by increased resource allocation to reproduction resulting in higher branch and fruit numbers, and by longer life span (Weinig et al., 2003), and tolerance to the oomycete Hyaloperonospora arabidopsidis, which was expressed during the vegetative growth period, compensated for rosette biomass loss (Salvaudon & Shykoff, 2013). Hence, there are common features in Arabidopsis tolerance to herbivory, an oomycete and a virus.
The tolerance of group 1, but not of group 2, genotypes to CMV, however, was not expressed in all environments. To understand the environmental variation of the plant–virus interaction from detrimental to beneficial, life-history trait responses to both abiotic stress and parasitism need be jointly considered. The 17°C HL conditions are optimal for BM production as well as for reproduction (SW) of both group 1 and group 2 plants (Fig. 1). All other conditions resulted in different degrees of stress, increasing with higher temperature and lower light intensity. Arabidopsis genotypes differed sharply in how they responded to abiotic stresses, in agreement with other reports (Pigliucci & Kolodynska, 2002, 2006; Cookson & Granier, 2006; Mishra et al., 2012). Abiotic stress was more severe to group 2 than to group 1 plants, and all weight traits of group 2 plants were positively correlated with each other in any environmental conditions (r ≥ 0.71, P ≤ 0.004, data not shown), showing no changes in allometry and thus a lack of phenotypic plasticity over conditions. Conversely, group 1 plants were less affected by abiotic stress and showed a high phenotypic plasticity, with changes in allometry indicating resource reallocation from growth to reproduction (Figs 1, 2; Table 1). Thus, group 1 plants responded similarly to virus infection and to unfavorable abiotic conditions by reallocating resources from growth to reproduction, which resulted in tolerance to both types of stress.
An interesting result of this study is the demonstration that some Arabidopsis genotypes may benefit from CMV infection under specific environmental conditions. The increase in seed production of group 1 genotypes after CMV infection at 22°C HL indicates overcompensation. Overcompensation, defined as a higher fitness of plants parasitized or damaged by herbivory compared with unparasitized or undamaged plants, has been most studied in relation to herbivory, in which case it is related to the plant's ability to hold back resources for reproduction, that is, to resource reallocation (Strauss & Agrawal, 1999; Agrawal, 2000). Overcompensation has been reported in Arabidopsis under herbivory (Weinig et al., 2003) and after infection by Hyaloperonospora parasitica (Salvaudon et al., 2008; Salvaudon & Shykoff, 2013). Common traits of overcompensation in these two instances and after CMV infection are that it depends on the plant genotype and that it occurs in tolerant genotypes, suggesting that it is related to an extreme expression of the mechanisms leading to tolerance.
When overcompensation is environmentally dependent, it can be viewed as a conditional mutualism (Bronstein, 1994; Agrawal, 2000). Conditional mutualism, or pleiotropic parasitism (Michalakis et al., 1992), has been demonstrated in a broad range of symbiotic interactions, including those of viruses and their animal and plant hosts (Roossinck, 2011). Studies with plants and viruses, however, are rare. Thus, it has been shown that infection by Kennedya yellow mosaic tymovirus is detrimental or beneficial to its host, Kennedya rubicunda, depending on the presence or absence of herbivores (Gibbs, 1980), and that infection of different plants species by four viruses is detrimental when water is not a limiting factor, but beneficial under water-stress conditions (Xu et al., 2008). A major mechanism leading to conditional mutualism is host life-history trait modification by parasitism (Fellous & Salvaudon, 2009), which may alter the interplay of the direct costs of parasitism in the host fitness, and the indirect costs resulting from the differential performance of parasitized vs nonparasitized hosts under different ecological constraints. If in some environments the direct and indirect effects of parasitism act in opposite directions, the fitness of infected individuals may be greater than that of noninfected ones (Thomas et al., 2000). Thus, the overcompensation described in this study may be the result of the interaction of similar life-history trait responses to CMV infection and to abiotic stress at the suboptimal conditions of 22°C HL. However, at conditions of more severe abiotic stress, the direct and indirect effects of infection would both be negative. Changes in the direction of direct and indirect effects of CMV infection in Arabidopsis according to plant density and infection prevalence have been reported to determine host fitness (Pagán et al., 2009).
The results of this work are highly relevant as they contribute significantly to demonstrating that plant viruses can be pleiotropic parasites, according to recent hypotheses that consider plant–virus interactions along the parasitism–mutualism continuum (Wren et al., 2006; Roossinck, 2011). If viruses can be detrimental or beneficial to their host plants according to the environment, this will determine the selection pressures exerted on their hosts, which should be considered for understanding the evolution of plant defenses and of plant–virus interactions.
Acknowledgements
We are grateful for the excellent technical assistance of Antolín López-Quirós. This research was in part supported by a Marie Curie contract co-funded by the European Union and Universidad Politécnica de Madrid (FP7) awarded to J-M.H., by a European Union Marie Curie contract (E120050-150, ERVIR) awarded to N.P., by a Ramón y Cajal contract (RYC-2011-08574) awarded to I.P., and by grant CGL2013-44952-R (Plan Estatal de I+D+i, Spain) to F.G-A.
Author contributions
J-M.H. and F.G-A. planned and designed the research; J-M.H., N.P. and M-A.M. performed the experiments; J-M.H. and I.P. analyzed the data; and J-M.H., N.P., I.P. and F.G-A. wrote the manuscript.
