Joshua G. Harrison, Casey Philbin, Zach Gompert, Glen Forister, Leonardo Hernandez-Espinoza, Benjamin W. Sullivan, Ian Wallace, Lyra Beltran, Craig Dodson, Jacob S. Francis, Amie Schlageter, Oren Shelef, Su’ad A. Yoon, Matthew L. Forister
There are approximately 400,000 described plant species, and best estimates for insect diversity are around 6,000,000 species, many of which associate with plants. Scientists are just beginning to glimpse the diversity of fungi, but there may also be millions of plant-associated fungal species. As these numbers suggest, the interactions between plants, microbes, and insects are tremendously important for how ecosystems work. Yet we still have much to learn regarding how variation among individual plants affects associated organisms.
We took a comprehensive approach to characterizing variation among 50 plants within an unmanaged patch of alfalfa and linked that variation to differences in the arthropods and fungi associated with each plant. We measured many physiological traits, as well as the plant’s genetic makeup, and variation in the concentrations of over 750 chemicals found within leaves. We also collected insects and spiders from each plant using nets and a modified leaf blower that functioned as a vacuum. Over just three sampling events, we collected 18,306 arthropods from 157 different species! Additionally, through the use of DNA sequencing, we found 78 different fungi were associated with the leaves of our focal plants.
Next, we used an algorithm called “random forest” to determine which plant traits were the best predictors of variation in arthropod and fungal richness and abundance. This analytical technique let us identify threshold effects of certain traits and complex interactions among those traits. For instance, plants that were less drought stressed had a much richer arthropod community than plants that were more stressed or that were average for the field. We found nonlinear effects of many of the influential plant traits, including plant size and certain chemicals. These results suggest that nonlinear and threshold effects could be important components of how plants affect the animals and fungi associated with them. In general, traits that were indicative of plant condition and reproductive stage (e.g. degree of flowering) were the best predictors of arthropod richness and abundance.
Our study highlights how new analytical techniques coupled with measurements of many plant traits can shed light on the complexity of plant-insect-microbe communities.