Skip to main content
Dryad logo

Environmental stochasticity increases extinction risk to a greater degree in pollination specialists than in generalists

Citation

Dumoulin, Christine; Armsworth, Paul (2022), Environmental stochasticity increases extinction risk to a greater degree in pollination specialists than in generalists, Dryad, Dataset, https://doi.org/10.5061/dryad.z612jm6fh

Abstract

Pollination sustains terrestrial food webs and agricultural systems and links the dynamics of interacting plant and pollinator species. Although environmental stochasticity is ubiquitous and can propagate through communities via species interactions in a way that increases extinction risk, it is unknown whether stochasticity affects species uniformly across pollination networks. In this paper, we introduce a stochastic dynamic model that makes novel use of the birth function and apply it to pollination networks of increasing size. We start with two- and four-species networks, in order to first illustrate the effects of stochasticity per se and then how those effects combine with specialization. We then describe the relationship between partner number and stochastic extinction risk in empirical networks with >20 species. In the 2-species network, increasing the variance of the stochastic term of the model increased the size of the region in parameter space where extinctions occur. In networks with 4 or more species, specialists were more vulnerable to extinction than generalists over a broad range of variances. Extinction risk in networks with >20 species declined nonlinearly with increasing mutualist partner number. Our results demonstrate the importance of including species interactions and stochasticity when using population-dynamic models to compare species’ extinction risk. While models that omit either of these factors are likely to underestimate extinction risk, they disproportionately underestimate the vulnerability of specialists.

Methods

Scripts generating model outputs and graphs.

Usage Notes

Data are in .csv and .mat formats. The latter can be opened using MATLAB or Octave.

All scripts are either .R or .m. The latter can be viewed with a text editor, and executed in either MATLAB or Octave.

Funding

National Science Foundation