Pull One Thread: Testing a Prairie Web for Its Keystone

Investigation question: When one part of a shortgrass prairie is removed or weakened, how many other populations change, and does a change to a living part or to a nonliving part send the bigger ripple? Does real survey data agree with what the model predicts?

Scientific purpose

Students build a fifteen-card interaction web of a shortgrass prairie in which strings stand for food, shelter, water and plant clipping, not just feeding. They run four trials with the same written rules: remove prairie dogs, remove bison, remove grasshoppers, and weaken the soil water with a drought. Working wave by wave, they record how many populations are lost, go down or go up. Comparing the counts shows which change reaches furthest, which species acts as a keystone, and whether a physical change can shift as many populations as a biological one. Groups then test one of the model's predictions against real survey counts of burrowing owls on active and abandoned prairie dog colonies.

Why this helps students with this chapter

The textbook tells students that prairie dogs matter to many species; this lab lets them measure how much, in a model where every link is visible and every effect must follow a stated rule. Counting the ripple from each removal turns 'keystone species' from a label into a result: the bison, the biggest animal on the board, changes one card, while the prairie dog changes twelve. Running a drought through the same web shows that a change to a nonliving part can reach every population. Checking the model against real colony surveys then makes the difference between a model's prediction and empirical evidence concrete, which is exactly the kind of evidence the performance expectation asks for.

Standards connections

MS-LS2-4 (assesses, the closing argument (q9), which uses real survey counts of burrowing owls on active and inactive prairie dog colonies; the string-web trials themselves are practice): Students first practice tracing how a change to a biological component (a species removed) or a physical component (less soil water) shifts populations in a model. They then construct a short argument from published survey data that the loss of prairie dogs affected burrowing owl populations, which is an argument supported by empirical evidence as the PE requires.

MS-LS2-1 (supports, resource availability as the rule behind every change in the model): Every rule in the model is a resource rule: a population goes down or is lost when its food, water or shelter shrinks or disappears. Students apply the resource-population relationship from Lesson 20 rather than producing new evidence for it.

Roles (rotate each session)

• Card keeper: Lays out the cards, lifts the removed or lost card at the start of each wave, and resets the board between trials.
• Rule reader: Reads each tugged card's needs aloud, applies the rules card, and says which token it gets and why.
• Token placer: Places the red, yellow and green tokens and flips LOST cards face down, but only after the group agrees.
• Recorder: Fills in the trial table, counts the tokens, and checks the totals against the board before the reset.

Materials

Estimated cost: 8 to 14 per class (an estimate; Estimate. Four skeins of acrylic yarn in different colors (about 3 dollars each), about 16 sheets of cardstock for eight card sets, and a few sheets of construction paper for tokens; masking tape and colored pencils are classroom stock. Cards and tokens are reusable.)

Safety

• Yarn is a tangling and tripping hazard. Keep the strings on the desk or floor area, never around a neck, wrist or finger, and coil them before anyone stands up.
• If you build the web on the floor, keep it out of walkways and away from doors; step around the web, not over it.
• Scissors, if used for tape or tokens: cut away from fingers. No other hazards beyond ordinary classroom materials.

Teacher preparation and setup

1. Print one 15-card set per group on cardstock and cut the cards apart. Print the rules callout large enough to read across a desk.
2. Pre-cut the yarn into 50 cm lengths and bundle each group's 32 strings by color: 22 green, 6 brown, 3 blue and 1 orange.
3. Cut the tokens, 15 of each color per group, and store them in an envelope with the cards.
4. Run the prairie dog trial once yourself with the rules so you can coach the wave-by-wave logic. The teacher notes in the answer key list the result of every possible single removal, which you need if groups run trial 5.
5. Post the string list where every group can see it, and decide whether groups may run a fifth trial of their own choice if time allows.

Investigation design

Student procedure

1. Lay out the 15 cards about 30 to 40 cm apart: the rain and soil water card along the bottom edge, the three plant cards above it, the plant eaters and the burrows card in the middle, and the predators along the top.
2. Tape one string for every pair in the string list: green for food, brown for shelter, blue for water, orange for clipping. Tape each end to the back of a card so the cards can still slide. Count the strings: there should be 32.
3. Read the rules card together. Each person predicts which card's removal will change the most other cards (q1); write the group's predicted number of changed cards for every trial in the trial table.
4. Trial 1: remove the prairie dogs. The Card keeper slowly lifts the prairie dog card about 20 cm. Every card that slides or tugs is directly linked; those cards make up wave 1. The Rule reader applies the rules to each wave-1 card, and the Token placer puts down tokens and flips any LOST card face down.
5. Next wave: lift each card that was just LOST and apply the rules to the cards it tugs. Also follow DOWN tokens to the cards that need the DOWN card (its eaters, or the users of a shelter or of water). UP tokens stop where they are. Keep going, wave by wave, until a wave adds no new tokens.
6. Record the trial: the number of wave-1 cards, the names of LOST cards, the numbers of DOWN, UP and MIXED cards, and the total number of cards that changed. Then remove every token, flip the cards face up and put the removed card back.
7. Trial 2: remove the bison. Trial 3: remove the grasshoppers. Repeat steps 4 to 6 for each.
8. Trial 4: drought. Do not remove the water card. Put a yellow DOWN token on it, follow its blue strings, and apply the rules wave by wave as before.
9. If time allows, run trial 5 on a card your group chooses, predicting first.
10. Graph the total number of cards changed and the number LOST for each trial. Then read the field evidence, answer q3, and compare it with your model's prediction for burrowing owls.
11. Answer the analysis questions; finish at home if needed. Untape the strings, bundle them by color, and return cards and tokens to the envelope.

Expected observations

• Trial 1 (prairie dogs): 8 cards change in wave 1 and 12 of the other 14 cards by the end. Burrows, black-footed ferrets and burrowing owls are LOST; cottontails, rattlesnakes, coyotes and hawks go DOWN; grasses, wildflowers, shrubs, grasshoppers and deer mice go UP. Only bison and the water card do not change.
• Trial 2 (bison): only the grasses change (UP). Trial 3 (grasshoppers): 7 cards change and none is LOST; grasses and wildflowers go UP, while deer mice, burrowing owls, rattlesnakes, coyotes and hawks go DOWN.
• Trial 4 (drought): all 14 other cards go DOWN over three waves, but nothing is LOST.
• Groups that misapply a rule usually let UP tokens travel, or forget that the burrows card needs prairie dogs and so miss the second wave of trial 1; comparing totals across groups catches both.

Likely difficulties

• Groups let UP tokens travel (for example, more grass making the bison go UP). The rule says UP stops; this is a simplification students should name in q8, not something to fix during the trial.
• Students forget that the burrows card is a nonliving part that depends on a living one. Ask: who digs and repairs the burrows?
• The board gets confusing after several trials. Insist on a full reset (tokens off, cards face up) before each trial, and have the Recorder check totals before the reset.
• Taping 32 strings takes longer than expected. Pre-bundled strings and the posted string list keep it to about 10 minutes; if time runs short, skip trial 5.

Structured debrief

1. Post every group's totals for trials 1 to 4. Where groups disagree, find the rule that was applied differently: the model is deterministic, so differences are rule errors, not data.
2. Ask which removal changed the most, and whether it was the biggest animal. Use the bison-versus-prairie-dog contrast to define a keystone species by the size of the ripple, not the size of the animal.
3. Contrast the drought with the removals: a physical change at the base of the web reached every population, which is why the performance expectation names both physical and biological components.
4. Put the owl table up. The model said LOST; the field said 69 percent of active colonies versus 11 percent of inactive ones. Discuss what the gap means: models make predictions, field data test them, and the mismatch tells us what the model left out.
5. Connect to slide 22: the owl survey has with-and-without comparisons and repetition across hundreds of colonies, but no before-and-after counts on the same colonies. What study would add that?

Recording sheet: what students fill in

Lab 2 recording sheet: Pull one thread

This is a model of a patch of shortgrass prairie on the Great Plains. Each card is one population or one nonliving part, and each string is a link: who eats whom, who shelters where, which plants need the soil water, and which plants the prairie dogs clip. A ★ marks a MUST HAVE: if it is lost, the population is lost from this patch.

Check the model against the field

Two more field findings. 1. Black-footed ferrets get almost all of their food from prairie dogs and live in prairie dog burrows. By 1987 the last 18 known wild ferrets had been captured for captive breeding, and the ferrets living in the wild today come from captive-bred animals released onto prairie dog colonies. 2. In a five-year experiment in a desert grassland, mesquite shrubs became three to five times more abundant in plots where prairie dogs had been removed than in plots where prairie dogs remained.

Analysis questions: answer key

Individual evidence of learning

Each student keeps their own trial table and graph and answers q1 to q9. The trial counts with the explanations in q4 to q7 show that the student can trace direct and indirect effects of physical and biological changes; q9, the argument from the real owl survey, is the individual evidence toward MS-LS2-4.

Assessment criteria

Accessibility adaptations

• Standing-web option for students who need movement: each student holds three or four cards on a clipboard in a circle, the strings run between clipboards, and a removal is felt as a tug.
• For students who need more support, tape the strings ahead of time so the group starts with the rules; provide the rules as a flowchart with icons.
• Color-blind access: tag each string with a letter (F food, S shelter, W water, C clipping) as well as its color, and print L, D and U on the tokens.
• Provide a bilingual glossary (population, shelter, predator, prey, drought, burrow, colony) for multilingual learners, and allow q9 to be drafted orally first.
• For students with fine-motor difficulties, use a printed web map with the links drawn as lines and place tokens on the map instead of taping yarn.

Low-cost alternative

Skip the yarn: print the web as a map with the 32 links drawn as colored lines, and run the trials by placing tokens, or writing L, D and U in pencil, on the map. The reasoning and the counts are identical; only the tug is lost, and one printed map per group costs a few cents.

Model limitations

The model is far simpler than a prairie. It has 15 cards where a real shortgrass prairie has hundreds of species, and it treats every effect as the same size: losing one food out of three counts as the same DOWN as a near-total collapse. UP effects stop where they land, so the model cannot show more grass feeding more bison, and DOWN effects travel only to eaters and users, never to prey. It leaves out competition between plants, other burrow diggers such as ground squirrels and badgers, animals moving in from nearby colonies, time lags and recovery. The ★ links are deliberately strict, which is why the model predicts that burrowing owls are lost when the survey shows a large drop instead. The trial counts are model outputs, not measurements; only the owl survey and the two field findings are empirical evidence. This lab has not been physically tested with students, and the timing is an estimate.

Teacher notes

Results of every single change under the rules (total of the other 14 cards that change; LOST cards named). Prairie dogs: 12 (burrows, ferrets, owls LOST; cottontails, rattlesnakes, coyotes, hawks DOWN; grasses, wildflowers, shrubs, grasshoppers, deer mice UP). Burrows: 12 (prairie dogs, ferrets, owls LOST; the rest as for prairie dogs). Grasses: 11 (bison LOST; the other animals and the burrows DOWN). Wildflowers: 10 (all DOWN: burrows and every animal except bison). Grasshoppers: 7 (grasses and wildflowers UP; deer mice, owls, rattlesnakes, coyotes, hawks DOWN). Deer mice: 7 (grasses, wildflowers, grasshoppers UP; owls, rattlesnakes, coyotes, hawks DOWN). Cottontails: 5 (grasses, wildflowers UP; rattlesnakes, coyotes, hawks DOWN). Shrubs: 4 (cottontails, rattlesnakes, coyotes, hawks DOWN). Rattlesnakes, coyotes or hawks: 3 each (prairie dogs, cottontails, deer mice UP). Owls: 2 (grasshoppers, deer mice UP). Ferrets: 1 (prairie dogs UP). Bison: 1 (grasses UP). Drought (water DOWN): all 14 DOWN, none lost. No single removal produces a MIXED card under these rules; MIXED appears only if students add a helper link, as suggested in q8. Sources for the field section: Sidle, J. G., and others (2001), 'Occurrence of burrowing owls in black-tailed prairie dog colonies on Great Plains National Grasslands,' Journal of Raptor Research 35(4): 316-321 (582 colonies surveyed in 1998; owls on 307 of 444 active and 15 of 138 inactive colonies); U.S. Fish and Wildlife Service and National Park Service accounts of black-footed ferrets (prairie dogs make up most of their diet; the last 18 known wild ferrets were captured by 1987); Ponce-Guevara and others (2016), PLoS ONE, a five-year experiment in which mesquite was three to five times more abundant where prairie dogs were removed. The card details are simplified: real ferrets and owls are also eaten by coyotes, owls and badgers, and prairie dogs eat some shrubs and forbs as well as clipping them.