Quantifying direct and indirect effects of early-season herbivory on reproduction across four brassicaceous plant species

Insect herbivores can directly affect plant reproduction by feeding on reproductive tissues, or indirectly by feeding on vegetative tissues for which plants are unable to compensate. Additionally, early-arriving herbivores may have cascading effects on plant reproduction by altering the later-arriving community. However, the dynamic interplay between plant development and the assembly of herbivore communities remains underexplored. Hence, it is unclear whether non-outbreak levels of ambient herbivory early in the development of plants can impact plant fitness and to what extent these effects are mediated through changes in plant development and subsequent herbivory. By excluding the herbivore community in an exclosure experiment and by manipulating early-season herbivory in a common garden field experiment replicated across four Brassicaceae species and two years, we tested whether early-season herbivory by caterpillars (Pieris rapae) or aphids (Myzus persicae) affected development, reproduction, and the herbivore communities associated with individual plants. In addition, we tested a causal hypothesis to assess the relative importance and temporal interplay between variation in herbivore communities and variation in plant development in determining plant reproduction. Early-season herbivory affected plant reproduction in the exclosure experiment, with effects being highly dependent on the plant species, the herbivore species, and the year. However, we found no such effects in the field experiment. The exploratory path analysis indicated that variation in plant reproduction is best predicted by variation in plant development, explaining 80% of the total effect on seed production. This suggests that early-season herbivory had limited effects on later plant development, and plants were able to attenuate the impact of early-season herbivory. However, no clear compensatory mechanism could be identified. While early-season herbivory has the potential to affect plant reproduction through changes in plant development or the subsequent development of the associated community, these effects were small and varied across closely related species. This suggests that plant species may be exposed to different levels of natural selection by early-season herbivores through plant- or community-mediated effects on reproduction.

Methods
Herbivore_Community

We recorded the development of herbivore communities on the focal plants planted in monoculture plots by monitoring individual plants from seedling to seed set. Recording of the herbivore communities started two days after the experiment was installed. We monitored the development of herbivore communities on individual plants by weekly counts early in the season and by biweekly counts later in the season. Insects were identified in situ to species or family level. If accurate identification was not possible, we included the observations as morphospecies in our data.

Plant_Parameters

In addition to observations of the herbivore community, we recorded a set of plant parameters as proxies for plant biomass and development: plant height (measured from the ground to the top of the plant), diameter (measured as the distance between the two most distal leaves), the number of true leaves, and the number of flowering and seed-carrying branches (aggregated as reproductive branches). The height and diameter of plants were used to derive the volume of a cone, representing plant biomass as a single volume parameter in the subsequent analyses.

Seedset_Field

We harvested plants after the start of plant senescence but before the siliques started losing seeds. From the harvested plants we collected the total number of seeds produced as a proxy for plant fitness. For each plant, the total number of seeds was estimated by extrapolating the weight of 100 seeds to the total seed biomass and rounding to the nearest natural number.

SeedSet_Tents

The same protocol as described above for the field experiment.

Analysis

All analyses and statistical approaches are detailed in the Functional Ecology manuscript. In summary: We applied linear regression to test for effects on the different growth and reproduction parameters and followed up on this analysis with piecewise structural equation modelling where we test a causal hypothesis on how effects of herbivory can cascade to affect seed production. Finally, we use multivariate analysis to describe differences between plant species in terms of the herbivore communities with which they interact.