Food webs: Ordering species according to body size yields high degree of intervality

Alexander Zook, Anna Eklöf, Ute Jacob, Stefano AllesinaJournal of Theoretical Biology 2010 · DOI

TL;DR

This paper investigates whether a single species trait can explain the structure of ecological food webs, a property known as "intervality." They found that body size is a surprisingly good, though not perfect, predictor of this intervality across many food webs, outperforming traditional trophic level measures. This has implications for building simpler, more realistic models of ecological interactions.

What Problem Does This Paper Address?

Ecological food webs, which map out who eats whom in an ecosystem, are incredibly complex. Understanding their structure is a major goal in ecology. A key concept is the "niche," which describes a species' role and interactions. While niches are often thought of as multi-dimensional, recent research suggests that many food webs are "nearly interval." This means that if you could order all the species along a single axis (like a trait), most of a predator's prey would fall into a contiguous block on that axis. This "intervality" implies that a single trait might be sufficient to explain a significant portion of food web structure. However, the question remains: which trait best captures this ordering? Body size is a commonly proposed candidate, but its effectiveness hasn't been rigorously tested against other potential ordering traits, like trophic level, across a diverse set of real-world food webs. This paper aims to fill that gap by systematically comparing body size and trophic level as proxies for food web intervality.

Methods

The core of the study involves quantifying "intervality" for 16 empirical food webs. Intervality is measured by counting "gaps" – instances where a predator's prey are not contiguous in a given species ordering. A lower gap count indicates higher intervality.

  1. Ordering Species: The researchers ordered species within each food web based on four traits:
  2. Body Size (BS): Species are ordered from smallest to largest.
  3. Trophic Level (TL): Three different measures of trophic level were used: shortest-chain (TLsc), prey-averaged (TLpa), and short-weighted (TLsw). These are standard ecological metrics that estimate a species' position in the food chain relative to primary producers.
  4. Acyclic Versions: For trophic level measures, they also analyzed acyclic versions of the food webs to mitigate potential inflation caused by feeding cycles.
  5. Measuring Intervality:
    • Predator-Centric Intervality: This is the standard definition, where a predator's prey should ideally fall within a contiguous range.
  6. Prey-Centric Intervality: An inverse measure where a prey's predators should ideally fall within a contiguous range.
  7. Statistical Comparison: To assess the significance of intervality for each ordering, they compared the gap counts to those generated by 10⁶ random orderings of species. A low number of gaps compared to random orderings indicates that the trait provides a good, interval-like structure.
  8. Maximizing Intervality: They also employed a Metropolis-Coupled Markov Chain Monte Carlo (MC³) algorithm to find the absolute minimum number of gaps possible for each food web, representing the theoretical maximum intervality. This serves as a benchmark against which the trait-based orderings are compared.
  9. "Wiggling" Species: To understand how sensitive the body size ordering is to minor inaccuracies, they simulated small perturbations (moving one or two species) in the body size order and re-evaluated intervality.

Datasets

The study analyzed 16 empirical food webs, including:

  • Benguela Pelagic
  • Broadstone Stream
  • Scotch Broom
  • Carpinteria Salt Marsh
  • Caricaie Lakes
  • Coachella Valley
  • EcoWEB 41
  • EcoWEB 60
  • Grasslands
  • Mill Stream
  • Sierra Lakes
  • Skipwith Pond
  • Small Reef
  • Tuesday Lake
  • Ythan Estuary
  • Weddell Sea

These datasets consist of binary adjacency matrices representing consumer-resource interactions and the typical adult body size for each species. No new benchmarks were introduced; rather, existing empirical food web datasets were used for evaluation.

Key Insights and Ideas

  • Body Size as a Dominant Predictor: The most striking finding is that body size consistently yields a higher degree of intervality (fewer gaps) than trophic level measures across most food webs, and this difference is statistically significant in many cases. This suggests that the physical size of organisms plays a fundamental role in structuring ecological interactions.
  • Trophic Level's Limitations: Despite being a common ecological metric, trophic level measures performed relatively poorly in producing intervality. This might be due to the complexity of food chains, the presence of omnivory, or the way trophic levels are calculated from binary data.
  • Predator vs. Prey Intervality: The study found that predator-centric intervality (how a predator's prey are ordered) generally shows more intervality than prey-centric intervality (how a prey's predators are ordered). This could imply that predators have more control over their diet composition than prey have over who preys on them.
  • Body Size is Good, Not Perfect: While body size is a strong predictor, it rarely achieves the maximum possible intervality for a given food web. This indicates that other factors beyond body size are also crucial for fully explaining food web structure.
  • Sensitivity of Body Size Ordering: Small changes (mutations) in the body size ordering do reduce the number of gaps, but not to the extent of the theoretically minimal gap count. This reinforces that body size is a good proxy but not a complete descriptor.

Key Results

  • Body size ordering resulted in statistically significant intervality (p < 0.001) for the majority of food webs analyzed, for both predator-centric and prey-centric intervality.
  • All three trophic level measures (TLsc, TLpa, TLsw) consistently produced more gaps (lower intervality) and fewer statistically significant results compared to body size.
  • For predator-centric intervality, body size ordering yielded p < 0.001 for all webs except Broadstone Stream, Coachella Valley, Grasslands, and Skipwith Pond.
  • For prey-centric intervality, body size ordering yielded p < 0.001 for all webs except Broadstone Stream, Grasslands, Sierra Lakes, Skipwith Pond, and Small Reef.
  • The minimum number of gaps found by the MC³ algorithm was consistently lower than that achieved by body size ordering, highlighting the existence of potentially more interval structures.

Limitations and Future Work

  • Body Size is Not Perfect: The authors acknowledge that body size, while a good proxy, does not perfectly describe food web intervality. Other traits, such as spatial distribution or species phylogeny, may also be important and could explain intervality in webs where body size alone is insufficient (e.g., webs with many parasite-parasitoid interactions).
  • Data Resolution: The study used binary food web data. More detailed, quantitative data on diet composition could potentially improve the performance of trophic level measures and provide a more nuanced understanding of food web structure.
  • Predator vs. Prey Asymmetry: The observed difference between predator-centric and prey-centric intervality warrants further investigation into the underlying ecological mechanisms and potential biases in food web data collection.
  • Future Directions: The authors suggest exploring the combined effects of multiple traits, investigating the role of phylogenetic constraints, and developing more sophisticated measures of trophic levels that account for quantitative diet data. They also propose that future work should focus on rigorously testing whether observed patterns are due to biologically relevant mechanisms or are artifacts of data collection.

Summary updated

BibTeX
@article{zook2010food,
  author = {Alexander Zook and Anna Eklöf and Ute Jacob and Stefano Allesina},
  title = {{Food webs: Ordering species according to body size yields high degree of intervality}},
  journal = {Journal of Theoretical Biology},
  year = {2010},
  doi = {10.1016/j.jtbi.2010.11.045},
}
Figure