Summary 1.Litter decomposition recycles nutrients and causes large fluxes of carbon dioxide into the atmosphere. It is typically assumed that climate, litter quality and decomposer communities determine litter decay rates, yet few comparative studies have examined their relative contributions in tropical forests. 2. We used a short-term litterbag experiment to quantify the effects of litter quality, placement and mesofaunal exclusion on decomposition in 23 tropical forests in 14 countries. Annual precipitation varied among sites (760-5797 mm). At each site, two standard substrates ( Raphia farinifera and Laurus nobilis ) were decomposed in fine-and coarse-mesh litterbags both above and below ground for approximately 1 year. 3. Decomposition was rapid, with >95% mass loss within a year at most sites. Litter quality, placement and mesofaunal exclusion all independently affected decomposition, but the magnitude depended upon site. Both the average decomposition rate at each site and the ratio of above-to below-ground decay increased linearly with annual precipitation, explaining 60-65% of among-site variation. Excluding mesofauna had the largest impact on decomposition, reducing decomposition rates by half on average, but the magnitude of decrease was largely independent of climate. This suggests that the decomposer community might play an important role in explaining patterns of decomposition among sites. Which litter type decomposed fastest varied by site, but was not related to climate. 4. Synthesis . A key goal of ecology is to identify general patterns across ecological communities, as well as relevant site-specific details to understand local dynamics. Our pan-tropical study shows that certain aspects of decomposition, including average decomposition rates and the ratio of above-to below-ground decomposition are highly correlated with a simple climatic index: mean annual precipitation. However, we found no relationship between precipitation and effects of mesofaunal exclusion or litter type, suggesting that site-specific details may also be required to understand how these factors affect decomposition at local scales.

High biodiversity of forests is not predicted by traditional models, and evidence for trade‐offs those models require is limited. High‐dimensional regulation (e.g., N factors to regulate N species) has long been recognized as a possible alternative explanation, but it has not be been seriously pursued, because only a few limiting resources are evident for trees, and analysis of multiple interactions is challenging. We develop a hierarchical model that allows us to synthesize data from long‐term, experimental, data sets with processes that control growth, maturation, fecundity, and survival. We allow for uncertainty at all stages and variation among 26 000 individuals and over time, including 268 000 tree years, for dozens of tree species. We estimate population‐level parameters that apply at the species level and the interactions among latent states, i.e., the demographic rates for each individual, every year. The former show that the traditional trade‐offs used to explain diversity are not present. Demographic rates overlap among species, and they do not show trends consistent with maintenance of diversity by simple mechanisms (negative correlations and limiting similarity). However, estimates of latent states at the level of individuals and years demonstrate that species partition environmental variation. Correlations between responses to variation in time are high for individuals of the same species, but not for individuals of different species. We demonstrate that these relationships are pervasive, providing strong evidence that high‐dimensional regulation is critical for biodiversity regulation.

As ecological data are usually analysed at a scale different from the one at which the process of interest operates, interpretations can be confusing and controversial. For example, hypothesised differences between species do not operate at the species level, but concern individuals responding to environmental variation, including competition with neighbours. Aggregated data from many individuals subject to spatio-temporal variation are used to produce species-level averages, which marginalise away the relevant (process-level) scale. Paradoxically, the higher the dimensionality, the more ways there are to differ, yet the more species appear the same. The aggregate becomes increasingly irrelevant and misleading. Standard analyses can make species look the same, reverse species rankings along niche axes, make the surprising prediction that a species decreases in abundance when a competitor is removed from a model, or simply preclude parameter estimation. Aggregation explains why niche differences hidden at the species level become apparent upon disaggregation to the individual level, why models suggest that individual-level variation has a minor impact on diversity when disaggregation shows it to be important, and why literature-based synthesis can be unfruitful. We show how to identify when aggregation is the problem, where it has caused controversy, and propose three ways to address it.

There is currently a lack of in-situ environmental data for the calibration and validation of remotely sensed products and for the development and verification of models. Crowdsourcing is increasingly being seen as one potentially powerful way of increasing the supply of in-situ data but there are a number of concerns over the subsequent use of the data, in particular over data quality. This paper examined crowdsourced data from the Geo-Wiki crowdsourcing tool for land cover validation to determine whether there were significant differences in quality between the answers provided by experts and non-experts in the domain of remote sensing and therefore the extent to which crowdsourced data describing human impact and land cover can be used in further scientific research. The results showed that there was little difference between experts and non-experts in identifying human impact although results varied by land cover while experts were better than non-experts in identifying the land cover type. This suggests the need to create training materials with more examples in those areas where difficulties in identification were encountered, and to offer some method for contributors to reflect on the information they contribute, perhaps by feeding back the evaluations of their contributed data or by making additional training materials available. Accuracies were also found to be higher when the volunteers were more consistent in their responses at a given location and when they indicated higher confidence, which suggests that these additional pieces of information could be used in the development of robust measures of quality in the future.