Skip to content
NatureHQ

Ecologyphenomenon

Food webs

In a chain, removing the fox has one consequence. In a web, it depends on everything else.

Who eats whom, drawn as a network rather than a line. Useful, and simplified in specific ways: real webs are full of animals eating their own kind, changing levels as they grow, and feeding at several levels at once.

A food chain is a single path: grass, rabbit, fox. A food web is the network of all such paths, and the difference is not merely one of completeness — it changes what can be predicted. In a chain, removing the fox has one consequence. In a web, the rabbit has other predators, the fox has other prey, and whether anything happens depends on what else is connected. Most ecological surprises live in that gap. What is worth knowing is how much even a good web diagram leaves out, and this is unusually well documented. When somebody described one desert community in genuine detail — every invertebrate, every life stage, every dietary breadth — the result was far richer and more connected than any published web of the time, and several of the regularities that food-web theory had derived from those published webs turned out to be artefacts of how coarsely they had been drawn. Three things in particular vanish from a tidy diagram. Cannibalism, which is common. Ontogenetic shifts, where a fish is eaten by something as a larva and eats that same thing as an adult, so the arrow reverses over a lifetime. And omnivory across levels, where an animal eats plants and herbivores and other predators, which makes "trophic level" a statistical average rather than a position. There is also a whole half of most webs that popular versions omit entirely. Grass, rabbit, fox accounts for a minority of the energy in most systems: the majority of plant material is never eaten alive, and goes instead to fungi, bacteria and detritivores, supporting its own predators. A food web without the detrital half is not a simplification of the system so much as a picture of a small part of it.

Early coverage · 52% complete · reviewed 2026-09-04

What this page covers

Every organism is in one. The best-documented webs are from systems small enough to enumerate — streams, ponds, islands, and one exceptionally well-described desert.

Often confused with: A food chain, which is a single path through a web rather than the web; Trophic levels as fixed positions, when many animals occupy several; A diagram of a web being a record of one, when most are heavily simplified

Quick facts

Chain versus web
A chain is one path; a web is the network, and prediction needs the network
What diagrams omit
Cannibalism, life-stage reversals, and omnivory across several levels
The missing half
Most plant material is never eaten alive — it goes to the detrital web
Resolution matters
Several food-web "laws" turned out to be artefacts of coarse drawing

One path, and the network it belongs to

The difference decides what can be predicted.

Real food webs hold animals that eat their own kind, occupy different levels as juveniles and adults, and feed across several levels at once. A chain shows none of it — and nor does most published web data.

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

Detailed documentation of a desert community found species richness, connectance, omnivory, cannibalism and ontogenetic diet shifts far exceeding those in published food webs of the period, and showed several apparent regularities of food-web theory to be artefacts of coarse taxonomic resolution.

Who this applies to
Demonstrated in detail for one desert community; the resolution problem applies to food-web data generally.
Studied in
Animalia, Plantae
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

A direct comparison between a highly resolved web and the coarser webs that theory had been built on.

How far it can be extended

The resolution issue is a property of how webs are constructed rather than of any one system.

Caveats

  • This is not an argument against food-web diagrams, which remain the clearest way to show who eats whom. It is an argument for saying what they omit.
  • Web resolution has improved since, though rarely to the level that exposed the problem.

Still unanswered

  • How many current food-web generalisations would survive at the resolution of a fully documented community.

Last reviewed 2026-09-04

The evidence (2 studies)
What a chain shows, and what a web adds
FeatureIn a chainIn a real web
Paths from plant to top predatorOneMany, of differing strength
An animal’s trophic levelFixedOften an average across several
Eating its own kindAbsentCommon
Change with life stageAbsentFrequent, sometimes reversing an arrow
Dead materialUsually omittedOften the larger share of the energy

Diagram

The half that gets left out

The chain people draw, and the detrital web beneath it.

The half that gets left outThe chain people drawPlantHerbivorePredatorThe half usually omittedDead materialFungi, bacteriaDetritivoresTheir predatorsIn most systems, most plant material is never eaten alive.Not shown anywhere: cannibalism, diets that change with age, feeding at several levels.
The same explanation in words

Two rows. The first shows the chain people draw: plant, herbivore, predator. The second, marked out as the half usually omitted, shows dead material leading to fungi and bacteria, then detritivores, then their predators. A line beneath records that in most systems most plant material is never eaten alive. A final note lists what appears in neither row: cannibalism, diets that change with age, and animals feeding at several levels at once.

The last row is the one that most changes the picture. In most systems the majority of plant production is never eaten by a herbivore at all: it dies and is processed by fungi, bacteria and detritivores, which support their own predators. A web drawn only as grass-to-rabbit-to-fox is not a simplified system; it is a small corner of one.

What the diagram on this page does not show

Stated here rather than in a footnote.

Every food-web figure, including the ones on this site, makes the same three omissions. Cannibalism rarely appears, because an arrow from a box to itself looks like an error. Life-stage changes are collapsed, so a species that is prey when small and predator when large gets one position. And omnivory is smoothed, because drawing an animal at three levels at once makes the figure unreadable.

This matters beyond tidiness. When food-web data were compared against an exceptionally detailed description of a single desert community, several apparent regularities — properties that theory had treated as general features of webs — turned out to be consequences of how coarsely the webs had been resolved. The lesson is not that diagrams are useless. It is that a web diagram is a claim about what was worth including, and the omissions are chosen rather than discovered.

Related

Pushed from above, or fed from below

Both, in proportions that differ by system.

Two ways of explaining how much of anything there is. Bottom-up control means the amount of nutrient and light sets plant growth, which sets herbivore numbers, which sets predator numbers — the system is limited by what enters at the base. Top-down control means predators limit herbivores, which releases plants, and the structure is imposed from above. Cascades are the top-down case made visible.

Framing this as a choice between the two is the error. Every system has both operating, and the useful question is their relative strength here, in this system, now — which changes with productivity, with food-chain length and with how many alternative pathways exist. Adding nutrients to a lake and removing its top predator both change the algae, and neither result establishes that one mechanism governs.

The research behind this page

8 studies, newest first. Each one has a page explaining what it found and what it could not show.

2010Ecology

Are wolves saving Yellowstone’s aspen? A landscape-level test of a behaviorally mediated trophic cascade

Aspen were not recovering preferentially in high-risk areas.

2004Ecology Letters

Detritus, trophic dynamics and biodiversity

Detritus supports food webs whose dynamics differ from grazing webs in important respects: the resource does not respond to its consumers, inputs arrive independently of consumption, and the resulting couplings can stabilise or destabilise the wider community depending on structure.

2004Science

Ecological linkages between aboveground and belowground biota

Above-ground and below-ground communities are reciprocally linked.

1999The American Naturalist

Patterns in the fate of production in plant communities

The share of plant production eaten alive varies enormously between ecosystem types, from a small minority in forests to a large majority in phytoplankton systems.

1991The American Naturalist

Complex trophic interactions in deserts: an empirical critique of food-web theory

The real web was vastly more species-rich and more connected than published webs suggested, with widespread omnivory, cannibalism, feeding relationships that change with life stage, and species occupying several trophic levels at once.

1974Science

Sea otters: their role in structuring nearshore communities

Islands with otters had few and small urchins and extensive kelp.

1966The American Naturalist

Food web complexity and species diversity

On the removal plot, mussels expanded and progressively monopolised the rock.

1927Sidgwick & Jackson

Animal Ecology

Animal communities can be described by who eats whom, with characteristic patterns in the numbers and sizes of organisms at successive feeding levels.

Where to go from here

Each of these follows from something on this page — a relationship in the evidence, a claim people ask about, or the next mechanism along.

How complete this page is, and what it is still missing

NatureHQ publishes its own gaps. This page is at 52% completeness against what we would call a finished subject, and was last reviewed on 2026-09-04. It carries 5 claims and answers 3 mapped search questions.

  • no research from the last few years is attached — check for newer work
  • more experiments could be explained in plain English
  • no popular claim about this subject has been checked yet
  • Quantitative food-web properties — connectance, interaction strength distributions — are described rather than covered.
  • Energy flow and ecological efficiency are not treated in their own right.
  • Microbial loops in aquatic systems are not covered.