Dimensions of Host Specificity in Foliar Fungal Endophytes

Abstract

Foliar fungal endophytes (FFE) colonized the phyllosphere at least 400 million years ago and have since diversified across every terrestrial ecosystem that supports plant life. Understanding how these complex symbiotic associations are generated, distributed and maintained is a challenging task that requires an understanding of host specificity. We propose a conceptual framework that outlines four ‘dimensions’ of host specificity that account for the geographic, phylogenetic or sampling scale under consideration. These ‘dimensions’ quantify FFE abundance and evenness (structural specificity), interaction strength (network specificity), evolutionary relationships (phylogenetic specificity) and the spatial or temporal consistency of the interaction (beta-specificity). We present one case study that quantifies and compares structural, network and phylogenetic specificity across FFE communities partitioned by taxonomy (Ascomycota vs. Basidiomycota). We focus on the effects of rare FFE species, approximated as Operational Taxonomic Units (OTUs), as a key methodological consideration for communities surveyed with next-generation sequencing (NGS) because the statistical nature of rarity confounds the quantification of host specificity. The exclusion of rare FFE OTUs consistently changed ecological inference by decreasing host specificity averages and increasing variances within FFE phyla. To evaluate the degree to which rare FFE OTUs affect statistical power, we compared our empirical community to that of randomized communities. Excluding rare FFE OTUs (>10% of total sequences in the case community removed) may lead to spurious host specificity metrics that are not statistically significant from that of randomized communities. Therefore, rare FFE OTU removal should be done with explicit rationale. We propose conceptualizing FFE host specificity with a multidimensional framework that will allow future studies to use quantitative, comparable and theory-driven metrics that can scale towards more meaningful estimates of global fungal biodiversity.

Appendices

Appendices

Supplementary material and reproducible code for bioinformatic processing, sequence removal and host specificity measurements can be accessed on Github at https://github.com/austenapigo.

Appendix A

Bioinformatic pipeline using USEARCH (version 9.2.64; Edgar 2010), BLAST+ (Camacho et al. 2009), MEGAN (Huson et al. 2007) and QIIME (Caporaso et al. 2010).

Appendix B

OTU tables used for structural, network and phylogenetic specificity with fungal taxonomic annotations.

Appendix C

R code for (1) sequence removal using Multivariate Cutoff Level Analysis (MultiCoLA; Gobet et al. 2010), (2) structural (vegan; Oksanen et al. 2017), network (bipartite; Dormann et al. 2008; Dormann 2011) and phylogenetic (picante; Kembel et al. 2010) specificity, and (3) randomized community analysis using the ‘nullmodel’ function (bipartite; Dormann et al. 2008) for structural and network specificity and ‘ses.mpd’ function (picante; Kembel et al. 2010) for phylogenetic specificity.

Appendix D

Host species phylogenetic tree used to calculate phylogenetic specificity in Newick format.

Appendix E

Table of p-values for Figs. 2, 3 and 6.