Teacher track · about 100 minutes · 14 steps

Who says? Reading the evidence behind a food web

Students interrogate the Cooloola web of life as a set of claims: what each relationship asserts, what evidence carries it, and what it would take to know more.

Most food webs handed to a class are finished drawings: every arrow confident, no working shown. This one shows its working. Every relationship in the Cooloola web of life carries an evidence tier, and the overwhelming majority say “expected” rather than “seen here”. That is not a weakness to apologise for; it is the lesson. This web is a model of how the sand country works, assembled almost entirely from what is known about these kinds of plants, animals and fungi rather than from local sightings, and it says so on every card it draws. A student asking what would it take to move one of these up a tier, and would it be worth anyone’s money is analysing scientific claims against their evidence, which is precisely the hardest thing the Year 7 curriculum asks of them.

The walk runs as two lessons, or one double. Steps 1 to 7 are lesson one, the web as a model: what it claims, how far its chains run, and the one trade in it that was measured in this sand. Steps 8 to 14 are lesson two, the claims interrogated directly: tiers ranked, traps sprung, a prediction checked against cited evidence, and a written argument to finish. Each web step is a plain link that opens the web already set to the right view, so you can also paste any step straight into your class notes or slides.

For the teacher

Curriculum

The spine is AC9S7U02 (use models, including food webs, to represent matter and energy flow and predict the impact of changing factors) and AC9S7I06 (analyse methods, conclusions and claims for assumptions, sources of error, conflicting evidence and unanswered questions), with AC9S7I01, I05, I07 and H01 in support. In a Queensland assessment plan the achievement-standard sentences this serves are: they represent flows of matter and energy in ecosystems and predict the effects of environmental changes, and they analyse methods, conclusions and claims. Every code above was verified against the published curriculum before it was tagged, and only the pages this walk routes through carry tags.

Running it

One student per device suits lessons like these, but every web step also projects well. The app checks understanding along the way; the assessable thinking lands on the printed sheet, which is what you collect. Each student leaves with one artefact: a completed evidence sheet holding a prediction and its correction, a drawn chain with its warrant written underneath, a three-tier ranking, and a half-page argument that cites a source. Nothing a student does here is stored anywhere but their own device, and only if they have turned learning tracking on; there is no class dashboard and no completion report, so the paper is the evidence, and you walk the room as usual. Shared devices share whatever progress exists; the Learn page has a reset.

Two honest footnotes to give the class

First, the evidence sort in step 8 runs on three tiers, because this web currently contains no relationship at its fourth tier, “our reading”; the tier exists for signed editorial interpretation and nothing in this web has needed one yet. Second, the frogs’ connection in step 4 displays a cap: it was authored from a local measurement, but the build holds it at “expected” until an open sourcing flag on that page is resolved, and the drawer says so in as many words. Both are the machinery working in the open, and both are worth thirty seconds of class discussion.

On First Nations knowledge

The Kabi Kabi (Gubbi Gubbi) people are the Traditional Owners and continuing custodians of this country, and have read it far longer than any source in this web. This release presents the published scientific record only, and this activity claims no cultural-perspectives curriculum content, because knowledge of Country enters a resource like this through collaboration and cultural protocol or not at all. That collaboration has not yet happened. The absence is deliberate and stated, here and in the product.

What good looks like

For step 3: a correctly arrowed chain, and any honest version of “the word is expected, which claims this is how these animals are known to feed and not that anyone has recorded these meals in this place”. The best answers go on to notice that the chain’s length says nothing about its warrant, since the longest chain on the web runs at the same tier as the shortest. For step 14, a strong answer names the claim and its tier, then reasons about risk: a land manager can act on “expected” if the cost of being wrong is small, but should want “seen here” before spending money moving water. A weak answer treats the tier as a mark out of four instead of a kind of knowledge.

What they will get wrong, predictably

Four things, and each is worth catching aloud. Students read the tiers as grades of truth, so “expected” becomes “probably false”; the correction is that it is a different kind of knowledge, inferred from the group rather than checked here, and the drawer says who inferred it and from what. They cite the tier word as if it were the source (“because it says expected”); the source is the reference in the drawer, and the tier word is only a summary of it. In step 3 they treat a long chain as a strong one, because length looks like knowledge; the web’s longest chain and its shortest carry the same tier word, which is what the writing under the drawing is for. And some will read the size of the web as the size of the evidence, taking a page this full to mean that many local records stand behind it. It means close to the opposite: the web is full because it was written from what is known of these kinds of organisms, and the tier column is where it keeps that honest, which is what step 8 exists to make them argue.

Short of time

The full walk is two lessons of about 50 minutes. In a single 40-minute lesson, run steps 1 to 7 with sheet questions 1 to 4; that half stands on its own as “the web as a model”, and steps 8 to 14 keep as a second lesson or a structured homework, since every step is a plain URL. You do not need to be a biology specialist to run it: the drawers carry the content, the sheet’s answers are in “What good looks like” above, and the printed page carries each check’s answer under the card.

Who says? The evidence sheet

Name: ______________________ Class: ____________

Before touching anything. The resident I predict has the most connections: ______________________ After looking, the actual busiest resident: ______________________
The longest chain the web allows. Draw it, arrows pointing at the eater, each organism labelled producer or consumer. Every arrow in your chain carries the same tier word. What is it claiming, and what is it not?
The frogs and the acid water. What did someone actually measure, and why does the label still say “expected”?
The measured trade. The one connection between two residents measured in this sand is between ______________________ and ______________________. What did the measurement show?
Three connections, three tiers. Write one connection at each tier, then rank 1 to 3 by strength of warrant.
Seen here: ____________________________________________ rank ___
Seen elsewhere: ____________________________________________ rank ___
Expected: ____________________________________________ rank ___
What would move your “expected” connection up a tier?
Predict, then check. Years of heavy groundwater pumping under this sand mass. What changes at the surface, and why does the first dry season look deceptively normal?
The argument. One connection: state the claim, name its tier, and argue whether a land manager should act on it. Cite at least one source from its evidence drawer.

Teacher quick-start

Walk it yourself first, then project it or paste step links into your class notes: every step below is a plain URL.
Each check is formative retrieval, not summative marking; written work lands on the printed sheet, which is what you collect.
Nothing a student does here leaves their device, and nothing is reported back to anyone.
Printing works in plain black and white on A4, two ways. Print this page as-is for the teacher pack: these notes, the steps, and every check with its answer printed under it, with the student sheet on a page of its own to photocopy. The pack shows answers, so it is yours, not a class set. For the student sheet alone, follow this link to the sheet first, then print: with the sheet targeted, the printout is the sheet by itself.

Begin the pathway →

Before you touch anything, predict: which resident has the most connections? Sheet, question 1. · opens the web of life, set to this step; its bar brings you back
Read the banksia's pollen connections. Who are the partners, and what does the tier word say? · opens the web of life, set to this step; its bar brings you back
Draw the longest food chain the web's connections let you draw, arrows pointing at the eater. Then write: every arrow in it carries the same tier word. What is that word claiming about your chain, and what is it not claiming?
The frogs' connection with the acid water. Open 'How sure we are' and read the whole drawer, cap and all. · opens the web of life, set to this step; its bar brings you back
Find the one connection between two residents that was measured in this sand itself. What did the measurement show? · opens the web of life, set to this step; its bar brings you back
Only one organism-to-organism relationship in the whole Cooloola web is labelled 'seen here': the scribbly gum's trade with its root fungi. What earned that one relationship the top label?

What earns 'seen here' is a local, published measurement of the interaction itself, and this relationship has one: the scribbly gum grown on rich soil and on wallum sand, with its dependence on ectomycorrhizal fungi heavy on the poor ground and slight on the rich (Schmidt et al. 2006). Each distractor is a weaker kind of evidence wearing the top label's clothes. Living in the same soil is co-occurrence, which documents a pattern and says nothing about a mechanism; it is where the question starts, not where it ends. Textbook biology applied to a new place is exactly what the 'expected' tier is for, and most of this web honestly sits there. And healthy trees have many possible explanations, of which fungal help is only one. Worth holding onto the honest remainder: the trade is measured, but no wallum fungus doing the trading has yet been named in the record, so even the web's best-documented relationship has an open end. (Ch 15; Ch 9.)
The cleanup crew is named here, and every one of its connections sits at the same tier. Read why, and note what a scientist would have to publish to move one of them up. · opens the web of life, set to this step; its bar brings you back
Pick three connections at three different tiers and rank them by warrant. The wallum's own page carries all three side by side. · opens the web of life, set to this step; its bar brings you back
The web labels 'the heath's honeyeaters and night mammals move the banksia's pollen' as 'expected'. What single new piece of evidence would move it up to 'seen here'?

The ladder climbs on records of the interaction, in the place, published where they can be checked. Expert agreement strengthens an expectation without changing its nature: it is still an expectation, which is all 'expected' has ever claimed (the current warrant is a plant guide's nectar note, Leiper et al. 2022, and the reasonable biology that a crowd of nectar-feeders carries pollen). A photograph of a bird near a flower is co-occurrence, the weakest read of all: near is not feeding, and feeding is not transferring pollen. And unpublished sightings, however many, cannot be followed and checked, which is the property the whole ladder is built on. Note the middle rung too: the same demonstration from a different heath would earn 'seen elsewhere', local demonstration being exactly what separates the top two tiers. This is how scientific knowledge changes, one relationship at a time: someone documents what everyone merely expected, and the label, and sometimes the biology, gets rewritten. (Ch 9.)
A respected plant field guide records that the paperbark's mass nectar flows draw flying-foxes to the trees at night. Yet the Cooloola web files 'the flying-foxes pollinate the paperbarks' at 'expected', not at 'seen here' or 'seen elsewhere'. Why hold the claim down?

Two honesty rules stack on this one card, and both are worth learning to spot. The first is that visitation is not pollination. What the record supports is bats arriving and drinking (Leiper et al. 2022); pollen riding a bat's fur from one tree's flowers to another's, and fertilising them, is the likely consequence, not the demonstrated fact, and sliding from one claim to the other is one of the easiest overclaims in ecology. The second rule is source competence: a plant field guide is a strong warrant for what a plant is and where it grows, which is exactly why the same guide carries 'seen elsewhere' on this web's plant-placement claims, and a weaker warrant for what animals do at night. Neither rule says the guide is wrong, and the second option overcorrects into dismissing a good source entirely. Both rules say the same careful thing: claim only what your source can carry, and file the rest one rung down, where new evidence can promote it.
Imagine a survey of some other heath, not this one, that finds two species always together: a sedge and a skink, and every plot with one has the other. A draft web of life labels this 'the skink depends on the sedge', tagged 'our reading'. What is wrong with that edge, and what evidence would earn the claim a real tier?

This hypothetical pair is the commonest trap in reading ecological data, laid out flat. Two species can share every plot for at least three reasons: one depends on the other; both depend on a third thing, a water table, a soil, a fire age; or a small survey got lucky. The survey documents a pattern, and does that honestly. 'Depends on' asserts a mechanism, which is a different claim, needing its own evidence: the interaction observed, or an absence tested. Raising the tag because 'the survey really happened' confuses the reality of the fieldwork with what the fieldwork was about; the survey saw locations, never the skink using the sedge. But deleting everything overcorrects, because co-occurrence data is far from worthless: it is where investigable questions come from, and the wallum has a worked example, occurrence surveys mapping where acid frogs persist (Shuker, Simpkins & Hero 2016) posing the question that physiological work then answered (Meyer, Franklin & Cramp 2021). Pattern first, then mechanism: that is the right order, and the trap is publishing the mechanism while holding only the pattern. The Cooloola web's own rule goes one step further than this question: it refuses to draw an edge from co-occurrence at all, which is why no relationship in it, at any tier, is warranted by 'found together'.
The swamp's connection with the water table is seen here. Predict on your sheet what years of heavy groundwater pumping would do at the surface. · opens the web of life, set to this step; its bar brings you back
A scheme draws hard on the groundwater beneath a wallum sand mass — heavy extraction and drainage — and lowers the water table. At the surface, through the first dry season, almost nothing looks different. Predict what happens over the following years to the wallum's window lakes, its perennial streams and its swamps — and why the calm surface is misleading.

The withheld cue is the point: the surface stays calm because the sand's freshwater sponge keeps dribbling water out long after the last rain, so damage runs underground for years before it shows. Predict from the dependence. Window lakes are the water table showing through, so they fall and shrink with it; the streams stay 'perennial' only because groundwater seeps into them through the dry, so a lowered table lets them dry up; and the swamps' zonation is set by hydroperiod, so fewer flooded days shrink the bands and dry out the breeding sites. The confident near-miss — 'the wallum is drought-hardy, it'll cope' — mistakes the toughness for independence: that toughness *is* the sponge, the plants plumbed into the hidden store, and lowering it removes the very thing their hardiness borrows from. Water quality won't 'improve' (losing the steady acid, tannin-stained supply degrades the specialist habitat), and it is not only the perched lakes at risk — those, rain-fed on a sealed floor, are the one part largely independent of the table, the opposite of the distractor's claim. (Ch 5; Ch 10; Dyring et al. 2025; Hofmann et al. 2020.)
The argument, half a page on your sheet: choose one connection from this web. State the claim, name its evidence tier, and argue whether a land manager should act on it, citing at least one source from the evidence drawer.

The end-check — read both ends of it

You’ve walked the steps. Now put them together: answer these from what the pathway taught, not from the pages. Getting them here is what tells you it stuck.

In the Cooloola web, 'expected' is the commonest evidence label by a wide margin. What does that mostly tell you?

A distribution of evidence labels is data about the literature, not about the heath. Cooloola is internationally studied ground: its giant podzols were measured here (Thompson 1981), its frogs' acid physiology was measured here (Meyer, Franklin & Cramp 2021), the scribbly gum's fungal trade was measured here (Schmidt et al. 2006). But almost none of that work is about species-to-species interactions, so when the web asks 'who is recorded eating, sheltering or pollinating whom in this place', the published record mostly goes quiet, and the labels say so. An 'expected' relationship is not 'probably false'; it is well-documented biology reasoned onto this place, likely true and honestly unproven here. And the last option misreads how science works: labels move when someone does the local study, and for many of these relationships no one may ever fund it. A web that admits which of its links are records and which are expectations is more useful to a student and a land manager than a bigger web that bluffs, because it shows exactly where new evidence would change the page. (Ch 9; Ch 4.)

The ground parrot's card says the bird needs heath of middle age since fire, not too fresh, not too rank, and the parrot's page cites a burning experiment that ran for decades. The web still labels that relationship 'expected', not 'seen here'. Why?

Two questions get asked of every source, and they come apart here. The first is whether the evidence has the right shape for the claim, and this one does: 'the parrot needs heath of intermediate age since fire' is a claim about a cycle that takes decades to turn, so it needs evidence at least as long as itself. A long-term burning experiment in south-east Queensland's heathy country, fire frequencies held apart for decades and the vegetation tracked throughout (Dooley et al. 2023), is exactly the design that can show frequency setting the structure a ground-dwelling bird lives in, and one visit could never do it. The second question is whether the evidence is about this place, which the top tier asks and this study cannot answer: it was run on other country, and it recorded no ground parrots at all. A superb study of the wrong ground cannot warrant a local record. So the row was moved down to 'expected' when someone checked what its key actually covered, which is the machinery doing its job rather than failing at it, and the parrot's fire-age need now sits at the same tier as its diet: two claims on one bird, both well-founded ground-parrot ecology reasoned onto this heath, neither of them a local sighting. Notice what did not happen. The claim was not deleted, and the source was not discarded; it still carries what it can carry, which is what fire frequency does to heath. (Ch 9.)

Most wallum plants trade with root fungi for their nutrients. The banksias mostly sit that market out. How do they get phosphorus from sand that has almost none left?

The banksia's answer to poverty is chemistry of its own: cluster roots, dense brushes that acidify the sand around them and free the locked-up phosphorus directly, no fungal middleman required. The strategy is documented as a principle on ancient nutrient-starved sands closely comparable to these (Zemunik et al. 2015, from Jurien Bay in Western Australia, a principle rather than a Cooloola measurement, which the guide page says plainly). The distractors each fail on the ground's own facts. There is no rich layer waiting below: the sand mass is leached from top to bottom, which is what made the poverty (Walker & Syers 1976 for why old ground only gets poorer). Paying fungi extra is the market the banksia conspicuously left, the very thing that makes it the exception on this trading floor. And nothing converts nitrogen into phosphorus: they are different elements, and no organism transmutes one into another, which is why phosphorus, once gone, is gone. One more thing worth noticing: the Cooloola web draws no edge for cluster-root mining, because its relationship vocabulary cannot yet say 'mines the soil directly' without overclaiming, so the story lives on the banksia's own page instead. A true, well-sourced strategy can still be missing from a diagram; absence of an edge is not absence of the fact. (Ch 9; Ch 4.)