Previous Article |
Table of Contents
| Next Article
COMMENTARY
The "Goldilocks factor" in food webs
,
,

*University of California, Merced, Sierra Nevada Research Institute, Yosemite National Park, CA 95389;
Pacific Ecoinformatics and Computational Ecology Laboratory, 1604 McGee Avenue, Berkeley, CA 94703; and
Department of Biology, Darmstadt University of Technology, Schnittspahnstrasse 10, 64287 Darmstadt, Germany
In a well known children's tale, the little girl, Goldilocks, nearly gets herself eaten by bears by boldly choosing between the three bears' porridge bowls: one too hot and another too cold. In this issue of PNAS, the article "Size, foraging, and food web structure," by Petchey et al. (1), tells a similar story of food webs in which predator species choose between prey species that are too large or too small. In doing so, the authors show how an easily observed species trait—body size—can address long-standing ecological questions about who eats whom in complex natural communities. Although previous models explain well the general network properties of food webs, Petchey et al. go further by explaining how the particular links found between species within specific food webs can result from optimal foraging of predators on prey that are neither too large nor too small. Like Goldilocks, optimal predators choose food that is "just right."
In choosing their food, predators and other consumers create food webs—the feeding networks among species that sustain the life support systems on earth. Ecologists continue to discover fascinating regularities in the structure of this interdependent complexity (2, 3). Elucidating the mechanisms responsible for these regularities is a fundamental step in tackling long-standing questions in ecology that range from "What confers stability to complex ecosystems?" (4) to "How does species loss alter the abundance of other species?" (5). Surprisingly simple rule-based models successfully capture the overall structure of real food webs (2, 6, 7) and enable further theoretical exploration by stochastically generating webs that mimic the overall structure of real webs (6, 8–10). However, such models lack mechanistic explanations for their input parameters and poorly predict the actual links
To whom correspondence should be addressed. E-mail: eberlow@ucmerced.edu
![]()
CiteULike
Complore
Connotea
Del.icio.us
Digg What's this?
Companion article to this Commentary: