Data from: The ecological dynamics and consequences of phytoplasma in a spring ephemeral

Methods

In the field, we characterized the infection dynamics of phytoplasma, a vectored and sterilizing bacterial pathogen, in wild populations of white trillium. We identified the highest molecular matches for the local bacterial strain and putative vectors. We then leveraged iNaturalist data to reveal patterns and potential drivers of phytoplasma occurrence across the range of white trillium.

Key results: Within populations, we found that phytoplasma symptoms were more likely in low host density plots, where insects occupied a higher proportion of trillium, indicative of an encounter-dilution effect. Leafhoppers in the genus Empoasca were positive for phytoplasma, suggesting they are a candidate vector of disease between trilliums at our field site. The closest match to our local phytoplasma was related to ‘Candidatus Phytoplasma pruni’, ribosomal group 16SrIII-F. Phytoplasma symptoms were widely distributed across trillium’s range, even at range edges, and the probability of symptoms was significantly increased by proximity to cropland.

Conclusions: We studied the impacts of a sterilizing, vector-transmitted pathogen on host plant populations and patterns of infection across host range. Our results suggest sap-feeding insects do not occupy all hosts at high host densities, thereby reducing transmission and leading to higher likelihood of finding infection at low host densities, and a higher proportion of host plants infected at low host densities. At a range-wide scale, we found evidence that the pathogen was able to persist across both the core and edge of host ranges.

Methods

Study organisms and site Trillium grandiflorum (white trillium), is a spring ephemeral native to the deciduous woodlands of eastern North America. There are large populations of wild T. grandiflorum at our study site, the Koffler Scientific Reserve (KSR, www.ksr.utoronto.ca) in Ontario, Canada (44°01'48”N, 79°32'01”W). Symptoms of phytoplasma infection have been observed in trillium populations at KSR since at least the early 2000s (A. Weis, personal communication).

Phytoplasma infection is most apparent when the white trillium are flowering, as petals will range from having a streak of green to full virescence or phyllody. Effects of phytoplasma infection on Trillium grandiflorum growth and reproduction To investigate whether infection was associated with differences in plant size, we quantified total plant height, leaf length, and sepal length in symptomatic (n = 32) and asymptomatic (n = 30) trillium during peak flowering season (May 20-29, 2023) across 30 locations in an approximately 15-hectare section of mixed deciduous forest at KSR.

We determined infection status by visually inspecting petals for virescence; we suspect that some infected individuals have no visible symptoms, so this method likely underestimates the overall infection rate of the population. To quantify differences in seed production due to infection, we first characterized the relationship between leaf length and seed number in asymptomatic trilliums. We haphazardly sampled 100 individuals on 13 June 2024, aiming to maximize variation in plant size.

On these plants, we measured leaf length and dissected fruits to count seed set. We also measured leaf length (May 20-28, 2024) and counted the number of seeds (early to mid June 2024) produced by 81 haphazardly chosen symptomatic trillium. To estimate potential reproductive loss due to infection in a given environment, we identified the nearest asymptomatic trillium of similar size and estimated its predicted seed count from the relationship between leaf length and seed number.

Within-population infection dynamics We conducted surveys of trillium density and infection frequency in T. grandiflorum populations of KSR in mid-late May 2018 and 2023 (populations indicated in Figure S2). In both survey years, 1 m2 plots were chosen haphazardly by throwing a flag into trillium populations (Figure S1); all plots were >5 m apart and plot locations differed in the two sampling years. To account for spatial proximity of plots, we categorized general sampling locations based on geographic features (e.g., separate hill slopes).

In 2018, we sampled 210 plots; we counted the number of flowering T. grandiflorum in the plot and their infection status based on petal virescence. In 2023, we sampled 64 additional plots with the aim of including more high-density populations and surveying potential vectors. We counted the number of non-flowering and flowering T. grandiflorum individuals that had at least three leaves, noted whether phytophagous Hemiptera nymphs were present on the plants, and noted the infection status of flowering trillium.

Across both surveys, we estimated the proportion of trillium infected by dividing the number of symptomatic flowering plants by the total number of flowering trilliums (we used only flowering individuals in the denominator because infection was determined by visual petal symptoms). In addition, using data from our 2023 survey, we calculated the proportion of trillium with phytophagous Hemipterans present by dividing the number of trillium with Hemiptera by the number of trillium in the plot; we did this for total trilliums and also for only flowering trillium.

To determine how phytoplasma infection affected the age class distribution of KSR trillium populations, we calculated plot level ratios of vegetative to reproductive trillium and tested for correlations with infection frequency. In 2024, we sampled 0.5 m2 plots centered around the asymptomatic and symptomatic individuals measured for seed production, and recorded the number of one-leaved, three-leaved vegetative, and three-leaved flowering trillium.

Identification of vectors and phytoplasma strain To identify the phytoplasma present in trillium, we sampled a leaf from a symptomatic flowering plant and extracted DNA using QIAGEN’s DNeasy® PowerSoil® Pro Kit. To assess vector identity and prevalence, we deployed yellow sticky traps (15.24 cm x 25.4 cm) attached to bamboo poles just above trillium height at 0.5 m above ground (Figure S1). We placed 20 traps across the sampled KSR trillium regions on June 6th 2023, collected them after 6 days, and then stored traps at -20℃.

We counted and collected potential vectors from each trap, including leafhopper, psyllid, aphid, and planthopper morphotypes. We chose 110 insects across potential vector morphotypes and extracted their DNA using QIAGEN’s DNeasy® Blood & Tissue Kit, following the protocol for insect specimens. We detected and amplified the 16S rRNA region of phytoplasma in insect and plant samples with a nested polymerase chain reaction (PCR) using two universal phytoplasma primer pairs R16mF2/R1 and R16F2n/R2 (Gundersen & Lee 1996), and visualised the presence of phytoplasma using gel electrophoresis.

We amplified the mitochondrial cytochrome oxidase subunit 1 region of insect samples positive for phytoplasma using the lib220-1/LepR primers (Hajibabaei et al. 2006). Then, to identify the phytoplasma present in plants and the insects that vector it, we performed Sanger sequencing on PCR products (at the Center for Analysis of Genome Function and Evolution, University of Toronto) and looked for similar sequences using the National Center for Biotechnology Information's nucleotide basic local alignment search tool using the megablast algorithm (Altschul et al. 1990, Chen et al. 2015).

Metapopulation infection dynamics To characterize spatial patterns of phytoplasma infection of T. grandiflorum at a range-wide scale, we collected data from user observations on iNaturalist. First, we annotated all visually symptomatic T. grandiflorum observations with the observation field “Phytoplasma Infection?=Yes”. On March 14th, 2025, we downloaded all research-grade observations with their associated metadata and observation fields.

In total, this dataset included 43,037 observations between 1967-2024 with 846 infected individuals. We removed observations with missing latitudes or longitudes, observations with low or missing positional accuracy (< 10,000), and observations likely to be erroneous with both a latitude above 20°N and a longitude below 65°W, resulting in 32,649 observations. To reduce user bias in the dataset, we limited the maximum number of observations per user to 15, reducing the dataset to 29,412 observations.

Then, we used CoordinateCleaner (Zizka et al. 2019) to remove observations at country centroids or known biological institutions, observations with equal latitudes and longitudes, observations with any zeroes as coordinates, and observations with identical coordinates. After cleaning, the dataset had 28,390 total observations and 526 infected observations. Users contributed a mean 1.76 total observations and a median of 1 total observation.

The infected observations were observed by 433 users. The vast majority (82.4%) of all observations in the dataset occurred after January 1st, 2020. We reprojected the coordinates of observations using ESRI:102003, an equal area projection that is appropriate for T. grandiflorum’s range.