We conducted a meta-analysis to test for an interaction between plant-soil feedbacks and herbivory, including effects on the magnitude and direction of feedbacks, herbivore consumption and herbivore growth.

 We identified 197 studies to address herbivore impacts on plant-soil feedbacks and 189 studies to address plant-soil impacts on herbivores. We calculated Hedge’s D values to assess three questions: 1) What is the plant-soil feedback value of plants exposed to herbivory or no herbivory? 2) What is the growth or biomass of herbivores feeding on plants exposed to home or away soils in plant-soil feedback studies? 3) How much plant tissue is consumed by herbivores on plants grown in home or away soils?

We found an overall significant weak negative effect of herbivory on plant-soil feedbacks that varied by plant functional type. In legumes herbivory drove plant-soil feedbacks from positive to negative, but herbivory on forbs further decreased negative feedbacks.  Herbivore consumption was generally greater on plants grown in away soils. However, herbivore consumption was greater in home soils conditioned by legumes but lower in home soils conditioned by forbs.

Therefore plant functional type determines the impact of conditioned soil on feedbacks, and herbivore consumption explains these results for legumes but not forbs.

In April of 2020 we conducted a search in Web of Science for “plant” AND “soil” AND “herbiv*” to identify papers published between 1994 (when the first paper on plant-soil feedbacks was published and 2020.  The search returned 8347 journal articles, abstracts, books, and data files.  In addition, we also examined references of articles cited by journal articles in our list to expand our search for suitable papers and contacted authors of several articles to enquire about data.  Papers used in the analysis were selected based on the following criteria: 1) plant species were grown in their own (home) and heterospecific (away) soil; 2) data was reported as biomass or growth means in each soil type (authors were contacted if data was in a different format); and 3) herbivory occurred in the feedback, not the conditioning, stage of the experiment.  We prioritized data collection for plants and insects by the type of data collected.  For plants, we prioritized total dry weight followed by aboveground dry weight followed by aboveground fresh weight.  For insects, we prioritized population size followed by biomass. To address potential publication bias we included an unpublished dataset. We identified: 7 published and 1 unpublished papers, however one published paper contained a subset of data from another paper so we excluded it. Within those papers we had 197 studies (datapoints) to address our first question about herbivore impacts on plant-soil feedbacks. We identified 4 papers but again excluded one as it contained the same data as another. Within these papers there was data from 189 studies allowing us to address our second question about plant-soil impacts on herbivores.

The columns of the database were grouped by sections: Publication (information about the journal article or study from which the data was collected), Data (the variable measured, units, means, errors, and replicates of each study), Home Plants (qualifiers describing information about the home or focal plant species in the study), Away Plants (qualifiers describing information about the away or heterospecific plant species in the study), Soil (qualifiers describing the soil used in the study), and Herbivores (qualifiers describing information about the herbivores used in the study). 

From each study within each identified paper we collected means, estimates of error, and sample sizes for plant and insect variables in home and away soil.  When raw data was not available we used WebPlotDigitizer (https://apps.automeris.io/wpd/) to collect means and estimates of error, and, when necessary, we calculated the standard deviation from the standard error and sample size.