Interactions between pollinators and plants can be affected by the presence of plant pathogens that substitute their infectious propagules for pollen in flowers and rely on pollinators for transmission to new hosts. However, it is largely unknown how pollinators integrate cues from diseased plants such as altered floral rewards and floral traits, and how their behavior changes afterward. Understanding pollinator responses to diseased plants is crucial for predicting both pathogen transmission and pollen dispersal in diseased plant populations. In this study, we investigated pollinator responses to contact with plants of Dianthus carthusianorum diseased with anther smut (Microbotryum carthusianorum). We combined three approaches: 1) observation of individual pollinators foraging in experimental arrays of pre-grown potted plants; 2) measurements of floral rewards and floral traits of healthy and diseased plants; and 3) quantification of pollen/spore loads of pollinator functional groups. We found that pollinators showed only weak preferences for visiting healthy over diseased plants, but after landing on plants, they probed fewer flowers on the diseased ones. Since diseased flowers offered lower nectar and no pollen rewards, this behavior is consistent with the prediction of optimal foraging models that pollinators should spend less time exploring less rewarding patches or plants. Furthermore, pollen-foraging solitary bees and hoverflies responded to diseased plants more negatively than nectar-feeding butterflies did. Lastly, based on group-specific behavior and typical pollen/spore loads, we suggest that solitary bees and hoverflies contribute to both pollen and pathogen spore dispersal mainly over short distances, while butterfly visits are most important for long-distance dispersal.

For the complete description of materials and methods, see the paper Pollinators adjust their behavior to presence of pollinator-transmitted pathogen in plant population published in Behavioral Ecology. 

The study system for the experiments was the herbaceous perennial plant Dianthus carthusianorum together with the plant's specialized anther smut pathogen Microbotryum carthusianorum and the plant's pollinators from three broadly defined functional groups (butterflies, hoverflies, solitary bees). The study site was a dry grassland on shallow soil in the Czech Republic (N 49°38'10.9", E 14°13'35.7") with a natural population of the plant and long-term occurrence of the pathogen.

Experiment 1 – Pollinator foraging behavior

We observed pollinator foraging behavior in experimental arrays of potted pre-grown plants in order to examine pollinator response to diseased plants. We set up the arrays at the study site. Each array consisted of 36 plants divided into four clusters; one third of the plants was diseased with anther smut and two thirds were healthy. We set up four different types of arrays to reflect variable spatial aggregation of plants in natural populations (within-cluster spacing was dense or loose, between-cluster spacing was near or far, for details see the published paper).

We observed the arrays in 15-min observation blocks from ca 9:00 a.m. to 17:00 p.m. For each observed pollinator, we wrote down the sequence of visited plants and afterward, we constructed three variables that might reflect pollinator response to diseased plants. 1) We estimated each pollinator’s avoidance of diseased plants as the proportion of visited diseased plants out of the total visited by that pollinator. 2) We estimated the plant visit duration using the number of visited flowers on that plant. 3) We calculated the flight distance travelled between visited plants from the positions of the visited plants.

Experiment 2 – Floral traits and nectar rewards of healthy and diseased plants

We measured several floral traits of healthy plants and plants diseased with anther smut that could explain the pollinator behavior observed in Experiment 1. We used both potted pre-grown plants and plants naturally occurring at the study site.

1) We measured flower size as corolla diameter (the average of two measurements per flower). 2) We measured nectar volume in flowers using 0.5-µl microcapillary tubes (Hirschmann Laborgeräte, Eberstadt, Germany); we measured nectar standing crop in unmanipulated flowers from ca 8:00 a.m. to 17:00 p.m. and we also measured nectar production in flowers bagged for 12 hours prior to the measurements. 3) We measured reflectance of flowers between 300 and 700 nm on an AvaSpec-ULS2048CL-EVO spectrometer calibrated against a WS-2 diffuse reflectance standard (Avantes, Apeldoorn, Netherlands) using a BSD 50W lamp (Lucky Reptile, Waldkirch, Germany) as the light source.

Experiment 3 – Pollen and spore loads of the main pollinator functional groups

We estimated pollen and spore loads carried on pollinator bodies in order to evaluate potential contribution of the main pollinator functional groups to pollination and pathogen transmission at the study site. We captured pollinators leaving diseased or healthy flowers using a hand net and we dabbed their bodies with a cube of fuchsin-stained gelatin. The gelatin cubes were then melted and mounted on microscope slides.

We counted pollen grains and spores at 400× magnification. We used a 1×1 mm square grid and scored the numbers in every other grid cell on a semiquantitative scale (0, 1, 2-10, 11-25, 26-50 pollen grains/spores). For further calculations, we used the mid values of the ranges. Next, the first 200 pollen grains on each slide were identified to species or to the lowest taxonomical level possible. We thus obtained the proportion of D. carthusianorum grains, which could be used to extrapolate the numbers to the whole sample (if the sample contained more than 200 pollen grains).

Each data file (numbered 1-8) contains data for one of the published analyses. The numbering of the data files correspond to the order in which analyses are presented in the published paper. The data files can also be divided into three groups according to the three separate experiments in the published paper: Experiment 1 (files 1-3), Experiment 2 (files 4-7), and Experiment 3 (file 8). There is also the additional file0 that contains primary data for Experiment 1 prior to constructing response variables for analyses.

Handling of the data is explained in the "read_me" files, either "read_me.txt" or "read_me.r" with working examples of  R code.