The value of biodiversity experiments

Summary

Recent biodiversity experiments have investigated the relationship between diversity and ecosystem functioning by synthesizing plant communities from pools of species that have been experimentally manipulated to vary numbers and types of species present while holding abiotic factors constant. Biodiversity experiments therefore focus on a previously under-explored aspect of global change: the feedback from diversity to environment. Consequences of random manipulation of species communities may not correspond well to those of specific extinction sequences observed in the past in response to extinction drivers that cause highly non-random loss. However, random manipulation provides a good starting point given that existing communities could undergo many alternative orders of species loss in the future in response to a variety of different potential extinction drivers. Further, the effects of some extinction drivers are currently poorly understood and therefore difficult to predict (e.g. climate change) and it may be premature to dismiss the predictions of random scenarios as irrelevant to all real examples of species loss. The first generations of biodiversity experiments have provided valuable, and sometimes unexpected, discoveries about the general nature of the relationship between diversity and ecosystem functioning. These discoveries could not have been made using observational studies. We propose that different examples of extinction loss in the real or a potential future world form a continuum from situations where the results of the first-generation biodiversity experiments will be highly relevant to less relevant. At the one extreme are examples where the effects of biodiversity on ecosystem functioning will be overwhelmed by direct effects of the extinction driver on processes (e.g. chronic eutrophication). At the other extreme are situations where ecosystem processes are not strongly affected by direct effects of the extinction driver and where the effects of species loss on functioning may be more important (e.g. habitat fragmentation). Given the unprecedented uncertainty about the future of biodiversity and the functioning of ecosystems, a general approach with randomly varying species pools was the right place to start in order to provide a general foundation. The new challenge is to test for effects of biodiversity on functioning in real-world examples of species loss.

Zusammenfassung

Biodiversitätsversuche zeichnen sich dadurch aus, dass natürliche Artenpools experimentell reduziert werden und anschließend der Zusammenhang zwischen der Artenzahl und Ökosystemfunktionen unter konstanten abiotischen Umweltbedingungen untersucht wird. Dadurch unterscheiden sich Biodiversitätsexperimente grundsätzlich von anderen Versuchen, die die Biodiversität als Zielvariable behandeln und stattdessen die abiotische Umwelt manipulieren. Die Auswahl der Arten für die reduzierten Artenpools in Biodiversitätsexperimenten erfolgte bisher meist zufällig, während natürliche Aussterbefaktoren wie Eutrophierung nicht alle Arten gleichermassen gefährden. Für verschiedene Aussterbefaktoren ist aber so wenig bekannt, dass ein zufälliges Aussterbeszenario die beste gegenwärtig verfügbare Option ist. Dies trifft insbesondere für mögliche zukünftige Aussterbeprozesse zu, die durch globale Umweltveränderungen (Klima, biologische Invasionen) oder Habitatsfragmentierung ausgelöst werden könnten. Die erste Generation von Biodiversitätsexperimenten mit zufälligen Aussterbeszenarien hat wertvolle, teilweise unerwartete, generelle Zusammenhänge zwischen Artenzahl und Ökosystemfunktionen aufgedeckt. Diese Zusammenhänge ließen sich durch vergleichende Studien nicht erkennen. In Zukunft sollten Biodiversitätsexperimente dennoch vermehrt Aussterbeszenarien simulieren, die in der realen Umwelt mit größter Wahrscheinlichkeit auftreten.

Introduction

Despite the large number of biodiversity experiments reporting positive effects of plant species richness on ecosystem functioning (reviewed in Schläpfer & Schmid, 1999; Schmid, Joshi, & Schläpfer, 2002b; Hooper et al., 2004) there is still some debate about the relevance of these results to real examples of species loss (Lepš, 2004). The initial debate—which focused on the mechanisms generating the results—has been resolved by showing that more diverse communities generally have increased values of productivity and related ecosystem properties due to the inclusion of particular species with key traits or due to the presence of combinations of species with complementary niche differences (sampling/selection effect and complementarity effects—Loreau et al., 2001; Hooper et al., 2004). Progress has been made through the rejection of two hypotheses: that the results of biodiversity experiments can be entirely explained as sampling effects (Loreau & Hector, 2001; Hector, Bazeley–White, Loreau, Otway, & Schmid, 2002) or due to the fertilizing effects of legumes (van Ruijven & Berendse, 2003). Lepš (2004) raises a further criticism: that biodiversity experiments manipulate species richness in ways that do not mimic extinction sequences in the real world. Therefore, he argues, the experimental results cannot be used to predict potential consequences of species loss on ecosystem functioning in the real world. We agree that extinction in the real world will often be non-random, particularly in situations like the example given of nutrient enrichment of grasslands, and that future biodiversity experiments should consider realistic extinction scenarios. However, we believe that random extinction scenarios have enabled us to take the first steps in biodiversity and ecosystem functioning research and that they have provided some valuable, and sometimes unexpected, discoveries about the general nature of the relationship. Furthermore, we believe that some interpretations of the intentions of biodiversity experiments and assertions about potential alternative approaches need clarification. We take productivity as an example ecosystem process.