1. Decline in species richness as well as changes in community evenness or functional diversity have been hypothesized to jointly affect ecosystem functioning. However, disentangling the relative effects of these changes in community structure is hard as these different aspects often covary with species richness in real-world ecosystems. In this study, we investigated the individual and interactive effects of functional diversity and community evenness of predators on the level of control of herbivorous prey. 
2. Using a highly-replicated mesocosm experiment, we crossed three levels of functional diversity of arthropod predators with two levels of community evenness while controlling for the effect of species richness. Using this experimental setting, we hypothesized that the effect size of functional diversity of predators depends on community evenness. We expected a positive effect of functional diversity of predators on top-down control at a high level of community evenness while we thought that species identity and their associated traits should drive most of the effect on top-down control at a low level of community evenness. 
3. Our results did not provide any evidence for an interaction between functional diversity and community evenness nor any beneficial effect of increased functional diversity overall on predation rates of herbivorous prey. In addition, our results revealed that species and functional identity drives most of the effects of predator community composition on top-down control of their prey in our study system. Assemblages composed of active hunters with low handling time and no starvation ability tended to have the highest impacts on prey biomass.
4. By indicating that top-down control of herbivorous prey by arthropod predators is mainly driven by species and functional identity and not by functional diversity, our study provides insights into the consequences of ongoing species loss on ecosystem functioning. Future research should now explore the predictability of trophic interactions based on functional traits of predator and herbivorous prey to anticipate the consequences of changes in species composition on ecosystem functioning.

To test how functional diversity modulates the effect of evenness in predator communities on herbivore consumption rates, we set up a mesocosm experiment that crossed three levels of functional diversity of predators with two levels of community evenness. We selected the aphid Acyrtosiphon pisum and the leafhopper Euscelidius variegatus as the two main herbivorous species for the experiment. Eleven predator species that are known to prey on aphids and leafhoppers and that are common predators found in grassland or faba bean fields were selected to define predator assemblages used in the experiment. In order to assemble three-species communities that represent three contrasted levels of functional richness, we first characterized these species using six functional traits involved in the predation. We collected information about body size (mm), circadian activity (diurnal, nocturnal), habitat domain (soil, foliage), hunting mode (ambush hunters, active hunters), averaged handling time (low, high) and starvation ability (yes, no). We selected three assemblages among the potential 165 assemblages of three species, each one in each of the third quartiles of the functional richness distribution. This led to combining A. bipunctata, A. nemoralis and X. kochi in the low functional richness level; A. bipunctata, C. carnea and P. mirabilis in the medium functional richness level; and C. carnea, P. opilio and P. mirabilis in the high functional richness level. A. bipunctata, A. nemoralis and C. carnea were bought at Biobest Company, while P. mirabilis, P. opilio and X. kochi were collected in the fields and maintained in the laboratory.

The two levels of community evenness were designed to have a low evenness modality (Pielou index = 0.79) and a high evenness modality (Pielou index = 1) for all levels of functional richness. The total number of predators in each mesocosm was six individuals so that the low level of evenness consisted of three-species communities composed of four individuals of the dominant species and one individual of the two remaining species, while the high level of evenness consisted of two individuals of each species.

In addition, for the low evenness modalities, we considered each situation of species dominance among assemblages of three species to control for the effect of predator identity. This resulted in four different treatments of evenness for a given level of functional richness: three low evenness treatments with each species dominating, and one high evenness treatment. As we considered three levels of functional richness, we ended up with 12 different treatments crossing each level of functional richness with each evenness treatment (four evenness treatments crossed with three functional richness levels). For a given round of the experiment, we replicated each treatment at least three times (one round had four replicates) and considered six mesocosm controls with herbivores only. One round of the whole experiment therefore consisted of 42 to 54 mesocosms. To ensure robustness of the results we replicated the whole experiment four times between 2016 and 2018 leading to a total of 180 mesocosms surveyed.

In each mesocosm, we introduced 150 aphids and 10 leafhoppers 24 hours before the introduction of predators based on realistic densities of these phytophagous species in fields (Ammann et al., 2022; Bosco et al., 1997; Erb et al., 2010). We assessed the number of aphids and leafhoppers after five days of predation exposure using a mouth aspirator and careful visual inspections in each mesocosm. The number of prey collected was converted to biomass using average weights of 0.51 mg and 3.53 mg for Acyrtosiphon pisum and Euscelidius variegatus respectively (Lamb et al., 1987; Purcell & Suslow, 1987). Mesocosms were kept in a greenhouse at ambient temperature and water was provided to plants every two days. We did not need any permission or appropriate license to perform the experiment.

The data attached correspond to the number of collected aphids or leafhoppers at the end of the experiment for each mesocosm and their given experimental modalities (e.g., functional richness, evenness and identity of dominant species). In addition, we attached the traits used to compute the functional richness metrics and explore how identity of predator species was affecting prey consumption.