The experiment underlying the study provides a diversity gradient of 1–60 plant species, established in assemblages randomly chosen from a pool of species typical of Arrhenatheretum grasslands. Recently sown on fertile arable soil and maintained by weeding, this experiment is a highly artificial system that fails to meet the definition of semi-natural grasslands7. Four years after establishment, a management intensity gradient of one to four annual cuts and three fertilization levels was established in subplots randomly assigned to the 1–60-species plots. Data presented in this study were collected in the following year.
Intensive management was thus imposed on plant species typical of Arrhenaterethum meadows, a plant community characterized by two annual cuts8. The potential effect size of increased management intensity is thus underestimated by applying the management to a plant community not adapted to it. More importantly, it is unlikely that the species-richness of high-diversity plots could be maintained under increased management intensity over longer periods. In fact, 22% of these subplots managed at very high intensity had to be excluded for missing or insufficient yield after only two years, indicating that their species did not persist under high defoliation frequency and fertilizer levels, even when competitors were excluded by weeding.
While the discussion hardly addresses this crucial trade-off between management intensity and plant diversity, Schaub et al.6 do indicate that repeated resowing is likely to be necessary to maintain high diversity under increased management intensities. In contrast to permanent grasslands, whose species composition is shaped by site conditions and management, species selection in (re-)sown grasslands is a conscious choice. To be advantageous, mixtures have to show larger yields than the most productive monoculture, so-called transgressive overyielding. Transgressive overyielding is one of the reasons why mixtures, especially grass-clover mixtures, are frequently used in sown grasslands. A European-scale experiment demonstrated that four-species mixtures showed transgressive overyielding at a wide range of sites under intensive agricultural management9,10. Although Schaub et al.6 generally quantify the diversity effects in comparison to monocultures, they argue that grasslands with the high-diversity characteristic of semi-natural grasslands have benefits not only over monocultures but over low-diversity grasslands, such as the 1–8 species standard mixtures shown in Fig. 6 of their paper. However, their results fail to demonstrate that their high-diversity plots show any transgressive overyielding even over monocultures, not to speak of low-diversity mixtures. As species assemblages of the experiment are randomly drawn from the species pool, monocultures and low-diversity mixtures cannot be expected to include the most productive species or species combinations and thus cannot be used to assess transgressive overyielding. When transgressive overyielding was quantified for one- to eight-species plots of the same experiment under extensive management in 2003, it decreased with species number. While two-species mixtures showed a mean transgressive overyielding of 5%, eight-species mixtures were only 70% as productive as the corresponding best monoculture, on average11.
Accordingly, the experimental design fails to capture the real trade-offs faced by grassland managers, either in permanent or in sown grassland. It cannot answer if high levels of diversity and the associated biodiversity benefits can be maintained under intensive management for a longer period than just a few years. Neither can it show a productivity benefit of high-diversity grassland assemblages compared to species-poor mixtures, or even monocultures, when in practice the sown species are deliberately chosen rather than randomly drawn from a species pool. While the underlying biodiversity experiment has made valuable contributions to our fundamental understanding of plant diversity effects on ecosystem functioning, it thus cannot be used to derive direct management recommendations for managed grassland.
Source: Ecology - nature.com
