Cross-scale neutral ecology and the maintenance of biodiversity

James P. O’Dwyer,Stephen J. Cornell

doi:10.1038/s41598-018-27712-7

James P. O’Dwyer, Stephen J. Cornell

Open Access

https://doi.org/10.1038/s41598-018-27712-7

Copy DOI

Abstract

One of the first successes of neutral ecology was to predict realistically-broad distributions of rare and abundant species. However, it has remained an outstanding theoretical challenge to describe how this distribution of abundances changes with spatial scale, and this gap has hampered attempts to use observed species abundances as a way to quantify what non-neutral processes are needed to fully explain observed patterns. To address this, we introduce a new formulation of spatial neutral biodiversity theory and derive analytical predictions for the way abundance distributions change with scale. For tropical forest data where neutrality has been extensively tested before now, we apply this approach and identify an incompatibility between neutral fits at regional and local scales. We use this approach derive a sharp quantification of what remains to be explained by non-neutral processes at the local scale, setting a quantitative target for more general models for the maintenance of biodiversity.

Highlights

One of the first successes of neutral ecology was to predict realistically-broad distributions of rare and abundant species
Neutral theory has most often been formulated in a spatially-implicit way, so that local species abundance distributions depend on two free parameters characterizing the influx from a larger regional community[1,37]
It has been difficult to know the values of these two fitted parameters θ and m are biologically reasonable or not, even when the neutral theory successfully matches the distribution of species abundances in a local community

Summary

Application to Tropical Forest Communities

Armed with a spatially-explicit prediction for the species abundance distribution, we test whether the observed distribution of tree species abundances at the Barro Colorado Island 50 ha plot (BCI) is consistent with a neutral model where parameters are fixed independently of the plot-scale counts. It has already been shown[23,27] that this function takes the following form at large spatial separations: F(r) Using this result, and data from trees with diameter >10 cm in 34 1 ha plots in Panama (separated by values of r between ~0.5 and ~50 km), Condit et al.[24] obtained parameter fits of σ = 40.2 m and ν/b = 5.10−8. While speciation would be difficult to estimate independently of this fit, this value of σ is similar to those obtained from seed-trap data[42] With these parameters fixed, we can test whether these large-scale data are consistent with the observed distribution of species abundance at the 50 ha plot scale[43,44,45]. The formulation of spatial neutral theory we have considered here allows us to show that local abundances are not consistent with the parameters inferred from large-scale data

Discussion

Additional Information