Projects per year
Most research that demonstrates enhancement and stabilization of ecosystem functioning due to biodiversity is based on biodiversity manipulations within one trophic level and measuring changes in ecosystem functions provided by that same trophic level. However, it is less understood whether and how modifications of biodiversity at one trophic level propagate vertically to affect those functions supplied by connected trophic levels or by the whole ecosystem. Moreover, most experimental designs in biodiversity-ecosystem functioning research assume random species loss, which may be of little relevance to non-randomly assembled communities. Here, we used data from a published ecotoxicological experiment in which an insecticide gradient was applied as an environmental filter to shape consumer biodiversity. We tested how non-random consumer diversity loss affected gross primary production (an ecosystem function provided by producers) and respiration (an ecosystem function provided by the ecosystem as whole) in species-rich multitrophic freshwater communities (total of 128 macroinvertebrate and 59 zooplankton species across treatments). The insecticide decreased and destabilized macroinvertebrate and, to a lesser extent, zooplankton diversity. However, these effects on biodiversity neither affected nor destabilized any of the two studied ecosystem functions. The main reason for this result was that species susceptible to environmental filtering were different from those most strongly contributing to ecosystem functioning. The insecticide negatively affected the most abundant species, whereas much less abundant species had the strongest effects on ecosystem functioning. The latter finding may be explained by differences in body size and feeding guild membership. Our results indicate that biodiversity modifications within one trophic level induced by non-random species loss do not necessarily translate into changes in ecosystem functioning supported by other trophic levels or by the whole community in the case of limited overlap between sensitivity and functionality.