‘Smoothing and the environmental manifold’ published in Science Direct

November 15, 2021

This paper is about gaining a deeper understanding of the fundamental question of the relationship between the recorded occurrences of a species and the environmental factors influencing these occurrences, we introduce two concepts for this purpose: the smoothing function, and the environmental manifold. We began with studying data gathered from camera traps on the Sunda clouded leopard in Borneo, in which we numerically smoothed the occurrence data in order to uncover the patterns of occurrence along different environmental gradients (e.g. how does the abundance of the Sunda clouded leopard change as elevation increases?), and part of this paper is devoted to showing that numerically smoothing species occurrence data can be a simple and powerful alternative to other widely used species distribution models (like the generalised linear model, or random forest machine learning) for recovering the numerical relationship between a species and various environmental factors.

But in the process of uncovering these relationships, we were surprised by some of the counterintuitive patterns which emerged from the species occurrence data. We realised that, because different environmental conditions (e.g. elevation and forest cover) vary simultaneously as we move across different regions, we have to somehow understand the nature of this environmental co-variation in order to not be misled by the apparent relationship between a species and an environmental variable. This led us to formulating the concept of the environmental manifold, which, upon some reflection, reveals itself to be of central importance to understanding the Hutchinsonian niche of a species. The presentation of the environmental manifold in this paper is more on the theoretical side, but we explain in the appendix why it is of fundamental importance in conservation practice, and we are planning to write another piece which develops it as a helpful tool for empirical studies.

We conclude with a paragraph about the importance of acknowledging a multiplicity of ways of understanding and approaching the study of ecological phenomena: a quantitative lens can be very valuable, but it is best taken in balance with other forms of knowledge in order to have a deeper understanding, appreciation and love of the more-than-human world!

Kumar, S.U., Maini, P.K., Chiaverini, L., Hearn, A.J., Macdonald, D.W., Kaszta, Ż. and Cushman, S.A., 2021. Smoothing and the environmental manifold. Ecological Informatics, p.101472.