Arbuscular mycorrhizal fungi influence host infection during epidemics in a wild plant pathosystem

Abstract

While pathogenic and mutualistic microbes are ubiquitous across ecosystems and often co-occur within hosts, how they interact to determine patterns of disease in genetically diverse wild populations is unknown. To test whether microbial mutualists provide protection against pathogens, and whether this varies among host genotypes, we conducted a field experiment in three naturally occurring epidemics of a fungal pathogen, Podosphaera plantaginis, infecting a host plant, Plantago lanceolata, in the Åland Islands, Finland. In each population, we collected epidemiological data on experimental plants from six allopatric populations that had been inoculated with a mixture of mutualistic arbuscular mycorrhizal fungi or a nonmycorrhizal control. Inoculation with arbuscular mycorrhizal fungi increased growth in plants from every population, but also increased host infection rate. Mycorrhizal effects on disease severity varied among host genotypes and strengthened over time during the epidemic. Host genotypes that were more susceptible to the pathogen received stronger protective effects from inoculation. Our results show that arbuscular mycorrhizal fungi introduce both benefits and risks to host plants, and shift patterns of infection in host populations under pathogen attack. Understanding how mutualists alter host susceptibility to disease will be important for predicting infection outcomes in ecological communities and in agriculture.

Keywords: Plantago lanceolata; Podosphaera plantaginis; mutualism; mycorrhizal fungi; plant disease; plant pathogen; powdery mildew; protective symbiont.

Fig. 1

Inoculation with arbuscular mycorrhizal fungi produced variable growth benefits in experimental plants from different genetic origins. Before exposure to Podosphaera plantaginis in field conditions, negative binomial generalized linear models showed that the magnitude of the growth benefits in experimental Plantago lanceolata plants (following inoculation with a mixture of three arbuscular mycorrhizal fungal species) varied among 30 host maternal genotypes (a, left; Table S1; mycorrhizal inoculation (MYC) × maternal genotype (GEN), P < 0.001, n = 287 plants). In the right panel of (a), colored dots represent seed origin populations and black dots represent field epidemic populations. After pathogen exposure in the field epidemic experiment, host growth continued to be linked to mycorrhizal inoculation (b, c; Tables S2, S3; MYC, P < 0.001), seed origin population (b; Table S2; seed origin population, P < 0.001), and maternal genotype (c; Table S3; GEN, P < 0.001), but growth benefits no longer varied among maternal genotypes. In (a), error bars represent a 95% confidence interval. In (b, c), black dots represent outlier individuals, box notches represent a 95% confidence interval for comparing medians, box hinges correspond to the 1^st and 3^rd quartiles, and box whiskers extend to the largest and smallest values no further than 1.5× the interquartile range from the hinges. Myc., mycorrhizal; Non‐Myc., nonmycorrhizal.

Fig. 2

Host infection rate was increased in mycorrhizal‐inoculated plants and varied among field epidemic populations. Infection rates by Podosphaera plantaginis in experimental Plantago lanceolata plants increased over time in each mycorrhizal treatment (a) and field epidemic population (b) until the peak of the epidemic, then fell near the end of the experiment. Generalized logistic regression models showed that though host infection rate was marginally increased in arbuscular mycorrhizal‐inoculated plants (AMF) relative to nonmycorrhizal plants (NM) at the peak of the epidemic (c; Tables S5, S6; mycorrhizal inoculation (MYC), P = 0.09, n = 287 plants); at the end of the experiment host infection rates were increased in AMF relative to NM plants in all three field epidemic populations (Tables S7, S8; MYC, P = 0.04, n = 286 plants). Host infection rates also varied among the three field epidemic populations throughout the experiment (c, d; Tables S5–S8; field epidemic population, P = 0.01 (peak) and P < 0.001 (end), n = 286 plants). In (b), missing data as a result of heavy rains on days 15–20 are indicated by light gray solid lines. In (c, d), predicted values resulting from generalized logistic models are plotted; black dots represent outlier individuals, box notches represent a 95% confidence interval for comparing group medians, box hinges correspond to the 1^st and 3^rd quartiles, and box whiskers extend to the largest and smallest value no further than 1.5× the interquartile range from the hinges. Myc., mycorrhizal; Non‐Myc., nonmycorrhizal.

Fig. 3

Among infected plants, disease severity varied among mycorrhizal treatments, seed origin populations, and field epidemic sites. Generalized logistic regression models show that previous inoculation with a three‐species mixture of arbuscular mycorrhizal fungi was linked to reductions in the proportion of leaves infected by Podosphaera plantaginis in infected individuals of Plantago lanceolata, both at the peak of the epidemic (a; Tables S9, S10; mycorrhizal inoculation (MYC), P < 0.001, n = 210 plants) and at the end of the experiment (b; Tables S11, S12; MYC, P = 0.003, n = 98 plants). Mycorrhizal inoculation and maternal genotype also interacted to determine the number of infected leaves on infected plants at the peak of the epidemic, with both negative and positive effects of mycorrhizas on disease (c; Table S13; MYC × maternal genotype, P = 0.05). At the end of the experiment, infected mycorrhizal‐inoculated plants had more infected leaves than did infected nonmycorrhizal plants (d; Table S16; MYC, P = 0.04). Seed origin population (a; Table S9; seed origin population, P = 0.004) and field epidemic population (d; Tables S13, S15, S16; field epidemic population, P < 0.02) also influenced host disease severity. In (a, b, d), predicted values resulting from generalized logistic models are plotted; black dots represent outlier individuals, box hinges correspond to the 1^st and 3^rd quartiles, and box whiskers extend to the largest and smallest values no further than 1.5× the interquartile range from the hinges. In (c), error bars represent 95% confidence interval. Myc., mycorrhizal; Non‐Myc., nonmycorrhizal.

Fig. 4

Growth and disease susceptibility are linked to defensive effects from arbuscular mycorrhizal fungi (AMF) in the host genotypes. (a) In a field epidemic experiment with Plantago lanceolata individuals from 30 maternal genotypes, linear regression models show that host genotypes that grew larger when inoculated with a three‐species mixture of AMF experienced marginally more negative infection outcomes from exposure to Podosphaera plantaginis (Table S17; P = 0.11, n = 30 genotypes). (b) In the same experiment, host genotypes that were more susceptible to infection (when not inoculated with mycorrhizas) received stronger disease protection effects from inoculation with mycorrhizal fungi (Table S18; P < 0.001, n = 30 genotypes). In (a, b), each point represents one maternal genotype, originating from one of six seed origin populations. The y‐axis represents defensive effects as a result of mycorrhizas, that is, the estimated change in the mean number of infected leaves in each genotype as a result of mycorrhizal inoculation. The shaded area represents a 95% confidence interval. In (a), the x‐axis represents growth effects as a result of mycorrhizas, that is, the estimated change in mean host size (leaf number) in each genotype as a result of mycorrhizal inoculation. Changes in host size were estimated before pathogen exposure in variable field conditions. In (b), the x‐axis represents host disease susceptibility, that is, the estimated mean number of infected leaves in infected nonmycorrhizal (NM) plants in each genotype at the peak of the epidemic: more positive x‐axis values indicate more severe infections in the absence of the mutualist.