Host plant phylogeny predicts arbuscular mycorrhizal fungal communities, but plant life history and fungal genetic change predict feedback

Abstract

Symbioses exert strong influence on host phenotypes; however, benefits from symbionts can increase or degrade over time. Understanding the context-dependence of reinforcing or degrading dynamics is pivotal to predicting stability of symbiotic benefits. Host phylogenetic relationships and host life history traits are two candidate axes that have been proposed to structure symbioses. However, the relative influence of host evolutionary history and life history on symbiont composition, and whether changes in symbiont composition translate into stronger mutualistic benefits is unknown. We tested the influence of plant phylogenetic relationships and plant life history on the composition of arbuscular mycorrhizal (AM) fungi, perhaps the most ancestral and influential of plant symbionts, and then tested whether AM fungal differentiation resulted in improved mutualism as expected from coadaptation. We constructed mycobiomes composed of seven AM fungal isolates derived from tallgrass prairie and grew them for two growing seasons with 38 grassland plant species. We found that host phylogenetic structure was a significant predictor of the composition of AM fungal communities and the genetic composition of AM fungal species, patterns consistent with phylosymbiosis. However, the phylogenetic structure of AM fungi failed to translate to improved benefits to their host. While AM fungi generally improved plant growth and mycorrhizal feedback was generally positive, the strength of feedback was not predicted by plant phylogenetic distance. The composition of the AM fungal community and genetic composition within AM fungal species were also significantly influenced by plant life history and feedbacks between early and late successional species were generally positive. Interestingly, positive mycorrhizal feedback was predicted by changes in genetic composition of the two most abundant AM fungal species, not by changes in species composition. Positive mycorrhizal feedbacks across life history can mediate plant species turnover during succession and suggests that consideration of mycorrhizal dynamics could improve ecosystem restoration.

Fig 1. Conceptual figure of the experimental design.

A common AM fungal community composed of mixtures of characterized isolates was grown in four replicate mesocosms with 38 plant species for two growing seasons (a). Changes in relative abundance of AM fungi were monitored using amplicon sequencing (b) and used to test for plant phylogenetic and life history impacts on AM fungal species and genetic composition (c). The effect of the AM fungal inoculum strains used in this experiment on plant species was characterized using meta-analyses of prior studies (d). Correlations of AM fungal relative fitness and fungal impacts (e) on host phylogenetic and life history groupings were tested as one measure of mycorrhizal feedback [11]. Feedback was also measured directly by testing growth responses of different plant species to conditioned AM fungal communities (f). Pairwise feedbacks, as represented by interaction coefficients [39], were calculated and used to test patterns of strength of feedback across host phylogeny and life history (g). We also tested whether the observed changes in AM fungal species composition or changes in genetic composition within species predicted strength of mycorrhizal feedback (h).

Fig 2. AM fungal species relative abundances across plant phylogeny.

AM fungal species composition varied across plant phylogeny and life history. Relative abundances of Claroideoglomus lamellosum, Claroideoglomus claroideum, Entrophospora infrequens, Funneliformis mosseae, Racocetra fulgida, Cetraspora pellucida, Acaulospora spinosa are presented alongside estimated abundance (proportion of ASVs identified as inoculated over total AM fungal ASVs) and Shannon’s Diversity. Host plant species are arranged by phylogeny with phylogenetic group indicated by vertical colored bars and life history status indicated by colored triangles adjacent to each plant species name. Significance of phylogenetic and life history effects on individual metrics of AM fungal composition are presented for each year separately. For each plant species, Year 1 and Year 2 means of AM fungal metrics are presented as grouped columns with year one being the lighter shaded top bar. The data and code underlying this Figure can be found in https://doi.org/10.17605/OSF.IO/NAXMT.

Fig 3. AM fungal genetic composition across plant phylogeny.

AM fungal genetic composition (ASV relative abundances) varied with host phylogeny and life history. ASVs of Claroideoglomus lamellosum, Claroideoglomus claroideum, Entrophospora infrequens, Funneliformis mosseae, Racocetra fulgida, Cetraspora pellucida, Acaulospora spinosa are arranged according to the AM fungal phylogeny presented at the top of the figure, with ASVs belonging to different AM fungal species indicated by differently colored horizontal bars underneath the phylogeny. AM fungal species names are listed on the bottom of the figure. Host plant species are arranged as in Fig 2. For each host plant species, the proportion of ASVs for each individual AM fungal species is represented by color intensity on a log scale. The significance of phylogenetic and life history effects are presented for each year separately. For each AM fungal species, results are reported for the most abundant ASV to show a significant effect, with rank abundance of the ASV indicated by the superscript next to the significance symbols. The data and code underlying this Figure can be found in https://doi.org/10.17605/OSF.IO/NAXMT.

Fig 4. Predicted feedbacks between phylogenetic groups.

We tested the correlation of differential growth response of plants of different phylogenetic groups to AM fungal species (derived from meta-analysis of previous growth assays), with the measured differential AM fungal accumulation in host plant phylogenetic groups (Fig 2). Positive slopes of the regression lines in gray suggest a strengthening of host fitness with phylosymbiosis, though there were no significant correlations at the p ≤ 0.1 level. Panels are arranged by year at the top of the figure. Differential responses are calculated with plant response relativized to the grass family. R² and p values corresponded to linear correlations weighted by AM fungal relative abundance. AM fungal species are denoted by color and symbol type. Symbol size denotes the relative proportion of each AM fungal species. The data and code underlying this Figure can be found in https://doi.org/10.17605/OSF.IO/NAXMT.

Fig 5. Predicted feedbacks between life history categories.

The differential growth response of plants of different life histories to AM fungal species (derived from meta-analysis of previous growth assays) was negatively correlated with the differential AM fungal response to host plant life history category (Fig 2) in year 2 (p = 0.001), but not year 1. The negative correlation is consistent with changes in AM fungal composition generating negative feedback between life history categories. Panels are arranged by year at the top of the figure. Differential responses are calculated with plant response to early successional plant-associated fungal communities as the baseline. R² and p values correspond to linear correlations weighted by AM fungal relative abundance. Gray lines are the regression lines. The data and code underlying this Figure can be found in https://doi.org/10.17605/OSF.IO/NAXMT.

Fig 6. Measured feedback strength between AM fungi and plant hosts.

Mycorrhizal feedback, measured as pairwise interaction coefficients [39], are plotted on the Y-axis in all plots. Phylogenetic distance (A) had a marginally significant quadratic relationship with fitness (p = 0.085). (B) shows feedbacks broken out by pairs of early, pairs of late, and early vs. late host plant parings. Bars indicate mean feedback, which were significantly positive for early–early (p = 0.01) and early-late (p = 0.003) parings. Both panels show individual interaction coefficients plotted as shaded dots, with gray for insignificant and black for significant pairwise interactions (p > 0.05). The data and code underlying this Figure can be found in https://doi.org/10.17605/OSF.IO/NAXMT.

Fig 7. Mycorrhizal feedback strength predicted by genetic dissimilarity of AM fungi.

Regressions of strength of pairwise feedback against measures of AM fungal dissimilarity. AM fungal species composition, as measured by pairwise dissimilarity of total species counts (Aitchison distance) did not predict feedbacks (p = 0.59). Pairwise feedback was significantly predicted by genetic dissimilarity of two AM fungal species, Entrophospora infrequens (p < 0.001) and Claroideoglomus lamellosum (p < 0.001). The data and code underlying this Figure can be found in https://doi.org/10.17605/OSF.IO/NAXMT.