Comparative study of the most tested hypotheses on relationships between biodiversity, productivity, light and nutrients

Many alternative hypotheses have been proposed about the mechanisms governing plant community structure and how plant species compete for resources. Seeking general conclusions requires identifying those hypotheses, describing them and comparing the results of the studies that tested them in a diversity of scenarios. However, comprehensive reviews and comparative studies on the hypotheses that have been actually tested rather than theoretically proposed are lacking. In this study, we focused on the relationships between four variables of interest to plant community and resource competition research: biodiversity, productivity, light and nutrients. We performed a novel type of systematic literature search in which we selected studies that had intended to quantify all four variables. From the shortlisted papers (88 out of 1557 initial), we identified causal interactions (effects) by doing a "backwards inference" from the statistical analyses, and assembled them as causal networks. We refer to each particular network as a tested hypothesis. The tested hypotheses that showed the same directionality of effects between the four variables of interest were simplified and grouped as compatible. We identified 43 hypotheses that could be classified into groups based on compatibility, with varying degrees of complexity in the number and directionality of relationships tested between variables. Only seven of these hypotheses had been tested in more than one study. The most tested hypothesis (47 times) has been one considering changes in nutrient amount (fertilization) as a sole effect variable driving changes in species richness, biomass and light penetration through canopy. Arguably, this is the simplest hypothesis and associated experimental design that one could think of for this system. Although more complex hypotheses that include indirect effect such as those of light penetration on biomass, would presumably better represent the higher complexity of the natural system, they have been tested only a handful of times each (2-21 times). Our results depict the landscape of hypotheses that have been tested for this system, a comparative description of their degrees of complexity, and information on available evidence. This information would be relevant to evaluate whether it is possible to proceed with more rigorous generalizations about mechanisms driving plant community structure. Moreover, our method facilitates the identification of knowledge gaps and mismatches between hypotheses, study designs and statistical tests being performed. (C) 2021 The Author(s). Published by Elsevier GmbH on behalf of Gesellschaft fur Okologie.