Skip to main content

Data from: Abundance, origin and phylogeny of plants do not predict community-level patterns of pathogen diversity and infection

Cite this dataset

Schmidt, Robin et al. (2021). Data from: Abundance, origin and phylogeny of plants do not predict community-level patterns of pathogen diversity and infection [Dataset]. Dryad.


Pathogens have the potential to shape plant community structure, and thus it is important to understand the factors that determine pathogen diversity and infection in communities. The abundance, origin and evolutionary relationships of plant hosts are all known to influence pathogen patterns, and are typically studied separately. We present an observational study that examined the influence of all three factors and their interactions on the diversity of and infection of several broad taxonomic groups of foliar, floral and stem pathogens across three sites in a temperate grassland in the central United States. Despite that pathogens are known to respond positively to increases in their host abundances in other systems, we found no relationship between host abundance and either pathogen diversity or infection. Native and exotic plants did not differ in their infection levels, but exotic plants hosted a more generalist pathogen community compared to native plants. There was no phylogenetic signal across plants in pathogen diversity or infection. The lack of evidence for a role of abundance, origin and evolutionary relationships in shaping patterns of pathogens in our study might be explained by the high generalization and global distributions of our focal pathogen community, as well as the high diversity of our plant host community. In general, the community-level patterns of above-ground pathogen infections have received less attention than below-ground pathogens, and our results suggest that their patterns might not be explained by the same drivers.


Study area

The field sites were located at Washington University’s Tyson Research Center (38°31’N, 90°33’W), located near St. Louis, Missouri, USA. The climate is temperate with 914 mm annual precipitation. The cherty, stony, and well-drained soils of the Gasconade-Clarksville-Menfro Association dominating this area were formed in chert free limestone residuum, cherty limestone residuum and deep loess (Benham 1982). These soils rest on bedrock primarily composed of Mississippian limestone (Burlington-Keokuk formation) and Devonian sandstones (Bushberg sandstone). The vegetation in this region mainly consists of Oak-Hickory forest, interspersed with patches of old fields, which are marked by the dominance of several species of perennial grasses and tall-growing Asteraceae (


Study design

For our study, we selected three sites separated from each other by at least 1.2 km. At each site, we documented plant community composition by collecting information on the identity and percent cover of all plant species in 20 1m2 plots placed in a grid design that spanned the site (plots separated by 3-10 m, depending on site size). From this species pool we sampled pathogens on all vascular plant species that contained at least 10 individuals at a site and at our time of sampling (July 2015, N=51 total plant species of which 40 are native and 11 exotic). A threshold of 10 individuals allowed adequate sample size for statistical power while also allowing relatively rare species to be included in our observations. Plant species abundance ranged from 0.02 to 11.4 percent cover on a particular site (see Appendix A1-3). Exotic species showed the same typical lognormal distribution in abundance as native species did, with a few very abundant species and many species with less than one percent cover.

Plant individuals were randomly chosen by blindly throwing a pencil in the surrounding vegetation and subsequently picking the individual of the respective species closest to the fallen pencil. We sampled 10-15 individuals per plant species at each site. For each individual, pathogen infection apparent on the surface of stems, leaves and flowers were assessed separately using a percentage-based nine-level rating scheme, according to Oberforster (2001), ranging from 1 = 0% infection to 9 ≥ 70% infection. Additionally, we estimated pathogen infection for all above-ground parts for each individual, using the same rating scheme. As floral infections were rare (N=2 infected individuals of Festuca subverticillata) and the results for foliar and stem surface infection rates were similar to those based on all above-ground parts, we only present results for all aboveground parts.

We assigned all pathogens observed on individual plants to broad pathogen groups following Rottstock et al. (2014): fungal leaf spot diseases, powdery mildews, rusts and downy mildews. While the ‘fungal leaf spot disease’ group is not a taxonomically defined group, species in this group produce morphologically similar disease symptoms, which allows for accurate attribution of infection signs. We recorded rust and downy mildew infection as presence of sporulation structures, powdery mildews as presence of mycelium, and fungal leaf spot diseases as presence of necrotic leaf lesions (Rottstock et al. 2014). We chose this as our measure of pathogen infection, because loss of photosynthetic tissue is a direct indicator of pathogen impact on host productivity and directly comparable across host species (Parker & Gilbert 2007; Parker et al. 2015). Although being a major pathogen group, we did not find any smut fungi on our plants. However, we also found a considerable amount of infection signs not relatable to the mentioned fungal, or fungal-like, pathogen groups, but which we could attribute to be caused by pathogenic bacteria or plant viruses (henceforth called ‘bacterial and viral diseases’). While infection patterns may differ broadly between bacteria and viruses and among groups of fungal pathogens (García-Guzmán and Heil 2014; Rodriguez-Moreno et al. 2018), our goal was to cover infection patterns over a broad range of pathogen groups, and disease of bacterial and viral origin constituted a considerable proportion of the visible infection signs in our community. Distinct morphological and life-history characteristics across these pathogen groups allow pathogen groups for each host plant to be documented in the field.

Most pathogens can not be directly identified to species in the field. However, to attain an entire pathogen species list for each host plant species across all sites, we took one sample of each occuring sign of pathogen infection for each plant species and site, which we dried and pressed to preserve them for further determination via digital microscopy (VHX-2000, Keyence Corp., Osaka, Japan). Digital microscopic imaging extends conventional microscopy by combining the power of optical imaging, electronic detection and computerized analysis (Chen et al. 2011), thereby providing the opportunity to efficiently conduct pathogen identification on a single device. Taxonomy and determination of pathogens to species followed Farr et al. (1989), Brandenburger (1985), Klenke and Scholler (2015) and MycoBank, the online database of the International Mycological Association (Crous et al.


Pathogen diversity and infection

At each site, we quantified pathogen diversity with two metrics: the total number of pathogen groups for each host species (‘pathogen groups per species’) and the mean absolute number of pathogen groups per plant individual (‘number of pathogen groups per plant individual’). For each site and pathogen group, we quantified pathogen infection with three metrics: the percentage of infected individuals per plant species (‘incidence’), the mean percentage of infected plant tissue per plant species (‘severity’) and the product of both (‘overall infection’). Moreover, we calculated the incidence, severity and overall infection across all pathogen groups to estimate total pathogen load for each host species (‘total incidence’, ‘total severity’ and ‘total overall infection’, see Table 1, also see Rottstock et al. (2014)).


Plant phylogeny

We used a dated molecular phylogeny of vascular plants (Zanne et al. 2014) to create a phylogeny for our 51 focal plant species (Appendix C). Our focal species that were missing from this tree were bound into the phylogeny at the genus level using the function congeneric.merge in the pez package of R (Pearse et al. 2015).


Pathogen specialization and distribution

For each pathogen identified to species level, we calculated the mean number of host genera and families, as a measure of host range, by extracting the number of known host genera and families from the USDA Fungus-Host distribution database ( We restricted our metric to these higher taxonomic levels, since the number of known host species is unavailable for some of our pathogen species. We used the same database to extract the geographic distribution for each pathogen.


Statistical analysis

We used Blombergs K (Blomberg et al. 2003) to test for phylogenetic signal of pathogen diversity, incidence, severity and overall infection across plant species. For plant species occurring on multiple sites, values of pathogen diversity and infection were averaged across sites. Blombergs K was computed along with an associated p-value by comparing the real association between each pathogen response variable and phylogeny to a null distribution obtained by random permutations of the data. P-values less than 0.05 indicate non-random patterns (i.e. phylogenetic signal). We found no evidence of a phylogenetic signal in any pathogen response variable (see Results), and thus did not consider phylogeny in any other statistical analyses.

Statistical analysis on the effects of host abundance and origin on pathogen diversity and infection were performed using generalized linear mixed models in SAS® Release 9.4 (procedure GLIMMIX). We analyzed the total number of pathogen groups using a model with Poisson error distribution and a logarithmic link-function. The dataset for the number of pathogen groups per individual was square root-transformed (Ahrens et al. 1990) and analyzed using models with gaussian error distribution and identity link-function. For the response variables incidence and total incidence we used models with binomial error distribution and a logit link-function. The percentage data for severity, total severity, overall infection and total overall infection was logit-transformed (Warton and Hui 2011) and analyzed using models with Gaussian error distribution and identity link-function. Plant species occurring on more than one site were treated separately for each site. In each analysis, ‘site’ (N=3 sites) was considered as a random factor and ‘host abundance’ (N=18 species for the Perilla site, N=23 species for the Carduus site and N=20 species for each site) and ‘host origin’ (N=2 categories), as well as their interaction, were treated as fixed factors. We considered ‘pathogen group’, as well as its interactions with ‘host abundance’ and ‘host origin’, as a fixed factor in the analyses examining incidence, severity and overall infection.

To test for differences in pathogen specialization between exotic and native hosts we used a Mann-Whitney-U-test with host origin (‘exotic’ and ‘native’) as predictor variable and the mean number of host genera (and families, respectively) for each pathogen species as response variable (procedure NPAR1WAY). To compare the geographic distribution of pathogens (‘restricted to North America’ vs. ‘distributed on more than one continent’) between exotic and native host plants, we performed a Chi-square-test for independence (procedure FREQ).

Usage notes

Data are presented as means per species. For the phylogenetic analysis, data for species that occur at multiple sites have to be averaged.


Deutsche Forschungsgemeinschaft, Award: FZT 118