← Real World Protocols
Real World Ready  ·  Layer 1: Authentic Experience

Stream Macroinvertebrates

Biology  ·  Environmental Science  |  Years 1–13  |  Facilitator resource  ·  No specialist equipment needed
Every NZ school within walking distance of a stream has access to a world-class biology laboratory. Stream macroinvertebrates — the insects, snails, and larvae living among the pebbles — are direct indicators of water quality. Students who have knelt at a stream, scooped a net through muddy water, and watched something unexpected emerge arrive at the classroom with something that belongs to them. No AI can manufacture that starting point. This protocol gives you everything you need to make it happen, and introduces iNaturalist as the identification tool that connects your students to a global citizen science network.
Prepare
In the field
iNaturalist + AI as thinking partner
Trace and act
What You Need
  • White flat-bottomed containersDish tubs, bus trays, or large ice cream containers. White is essential — macroinvertebrates show clearly against a pale base. Avoid coloured containers.
  • Bug nets or pond netsInexpensive butterfly nets work well. One per pair is ideal. A flat-edged net scoops more efficiently than a round one.
  • Magnifying glassesOne per group minimum. A hand lens of 5x to 10x magnification is sufficient for identifying most common species.
  • Mobile phone with iNaturalist installedDownload the free iNaturalist app before you leave school. Enable location services. Students photograph specimens in the tray and submit observations from the stream or back in the classroom.
Tip: A white ice cream container is the single most useful piece of equipment. Bring one per group. The contrast between the pale base and the dark organisms makes identification straightforward even without a magnifying glass.
The Technique
1
Find a riffle

Shallow, fast-flowing water over pebbles is where macroinvertebrates are most abundant. Avoid deep, slow-moving pools for sampling.

2
Position the net

Hold the net downstream of where you will disturb the streambed, opening facing into the current, bottom edge touching the stream floor.

3
Kick sample

Agitate the streambed for 30 seconds just upstream of the net. The current carries dislodged creatures directly into the net.

4
Transfer to the tray

Empty the net into a white tray filled with stream water. Rinse the net thoroughly — small creatures cling to the mesh.

5
Photograph and observe

Use magnifying glasses to identify organisms. Tally each type found. Photograph specimens clearly against the white tray background for iNaturalist submission. Enable location before photographing.

6
Submit to iNaturalist

Submit photographs as observations from the stream if signal is available. Otherwise save as drafts and submit back at school. Your observation becomes part of the global biodiversity record.

7
Return everything

Release all specimens to the stream after observation. Return the tray water. Leave the site exactly as you found it.

What You Are Looking For

The variety and sensitivity of organisms found tells the story of stream health. A healthy stream supports a wide range of species, including sensitive ones that cannot survive pollution.

OrganismKey featuresIndicator
Stonefly larvaeFlat body, two tails, clings to stonesVery good
Mayfly larvaeThree tails, leaf-shaped gills along abdomenGood
Caddisfly larvaeOften inside a case of stones or sticks; free-living forms also presentGood
Freshwater snailsCoiled shell, slides on hard surfacesModerate
Water boatmenOval body, paddle-shaped back legsModerate
Midge larvaeThin, worm-like, often red in colourPoor
Aquatic wormsVery thin, writhing movementPoor
Reading the result: Finding stonefly or mayfly larvae indicates clean, well-oxygenated water. Finding only midge larvae and worms suggests the stream is under stress. A diverse sample with many types is the best possible result.
Health and safety: As with any activity outside the classroom, please ensure your school's own EOTC requirements and health and safety procedures are followed. Your staff will know what that looks like for your context.

iNaturalist: citizen science identification at the stream

iNaturalist is a free app and global platform that combines AI-assisted image recognition with verification by a worldwide community of expert naturalists. When your students photograph a specimen and submit it, two things happen simultaneously: they receive an identification, and their observation is added to a global biodiversity dataset used by researchers to track species distributions, behaviours, and change over time. That is meaningfully different from asking a general AI chatbot the same question.

How to use iNaturalist at the stream

  • Install the free iNaturalist app before leaving school
  • Enable location services — location is critical for accurate identification of NZ species
  • Photograph specimens in the white tray, well-lit, close up
  • Submit as an observation; the AI provides an initial identification
  • The community of expert identifiers will review and confirm or correct it, sometimes within hours
  • Check back on your observations: you may receive expert identifications and comments from specialists anywhere in the world

Why iNaturalist differs from a general AI chatbot

  • iNaturalist uses your location to narrow identification to species actually present in your region
  • Its AI is trained specifically on biodiversity image data, not general internet text
  • Identifications are verified by human experts, not generated from language patterns
  • Your observation contributes real scientific data to global research — it does not disappear after you close the app
  • Free-living caddisfly larvae, for example, are routinely misidentified by general AI chatbots but correctly identified by iNaturalist when location is provided
Using AI tools for species identification: a caution worth teaching. General AI tools can produce plausible-sounding but incorrect species identifications, particularly for invertebrates, juveniles, and species with regionally distinct forms. Conservation and biodiversity professionals have found AI image identification unreliable for specialist work. Teaching students to interrogate identification results — comparing iNaturalist (location-aware, expert-verified) with a general AI chatbot — turns this limitation into a genuine learning task about how different tools work and why tool choice matters.

Back in the classroom: AI as thinking partner (Real World Ready Layer 2)

These prompts build on what students observed, collected, and photographed at the stream. iNaturalist observations are the starting point for most prompts. Where gen AI is used alongside iNaturalist, the comparison between the two outputs is the learning task — not the AI output itself.

Years 0–6
What did iNaturalist say?

Look at your iNaturalist observation. What did the app identify your creature as? Ask a general AI chatbot the same question using your photograph. Did they agree? Which answer do you trust more, and why?

What will it grow into?

Ask iNaturalist and a gen AI chatbot: "What does a mayfly larva grow into?" or "What does a caddisfly larva look like as an adult?" Compare the answers. Ask the AI to show you the life cycle in simple steps.

Why does location matter?

Try submitting a photo to iNaturalist with location turned off. Then try the same photo with location on. Does the identification change? Ask a gen AI chatbot why location helps identify living things.

Tell the stream's story

Tell a gen AI chatbot everything your group found: "We found [list of organisms] in our stream near [place name]. What does this tell us about the water?" Then check its answer against the indicator table from the field.

Years 7–10
Compare the identifications

Submit your most interesting or ambiguous specimen to iNaturalist. Then describe it to a gen AI chatbot without the photograph. Where do the identifications agree? Where do they differ? What does the difference tell you about how each tool works?

Build the pollution story

Choose one organism from your sample that is sensitive to pollution — stonefly or mayfly larvae are good candidates. Ask a gen AI chatbot: "What pollutants most threaten this organism, and what land uses upstream produce them?" Then check the claims against your iNaturalist observation and any NIWA resources your teacher provides.

Your observation in context

Find your iNaturalist observation online and look at where else in NZ this organism has been recorded. Ask a gen AI chatbot: "Why might freshwater scientists track where these organisms appear and disappear over time?" What would a change in distribution tell researchers?

Design a monitoring question

Based on what you found, ask a gen AI chatbot to help you design a monitoring question: "If I wanted to know whether this stream's health changed between seasons, what would I measure, how often, and how would I know if something had changed?"

Years 11–13
Interrogate the identification tools

For an ambiguous specimen, compare: iNaturalist with location, iNaturalist without location, and a gen AI chatbot with a photograph description. Document the results. What does this comparison reveal about the training data, design assumptions, and appropriate use cases for each tool?

MCI calculation and verification

Use a gen AI chatbot to explain the NIWA Macroinvertebrate Community Index methodology. Then apply it to your field data. Verify each step against the NIWA source. Where does the AI explanation hold up? Where does it introduce imprecision or error?

Catchment analysis

Ask a gen AI chatbot: "Given a community of [your organisms], what land uses upstream would predict this result, and what changes to those land uses would most improve stream health?" Evaluate the response against your knowledge of the actual catchment. Is the AI's causal chain physically defensible?

Citizen science value

Research what makes iNaturalist observations scientifically useful: sampling protocol, location accuracy, expert verification, and data accessibility. Ask a gen AI chatbot the same question. Write a short comparison: what can citizen science data contribute that neither professional surveys nor AI-generated responses can replicate?

EXPERIENCE TRACE SCALE — STREAM MACROINVERTEBRATES
Level Years 0–6 Years 7–10 Years 11–13
1 Student names at least one organism found in the stream sample and can point to it in a photograph or the identification table. Understands the organisms came from a real stream, not a tank or illustration. Student identifies organisms from the sample by name, notes their indicator status, and makes a basic claim about stream health based on what was found. Student identifies organisms to the level required for an MCI calculation, notes the indicator weighting of each, and produces an initial water quality assessment from the field data.
2 Student links the organisms found to a water quality claim: "We found stonefly larvae, which means the water is clean" or equivalent. Can explain in simple terms why some organisms indicate good water and others do not. Student explains the mechanism connecting organism sensitivity to water quality — dissolved oxygen, pollutant tolerance, habitat requirements — and links their specific findings to specific conditions in the stream. Student constructs a causal account connecting land use in the catchment to in-stream conditions to organism community composition, using their field data as the evidence base.
3 Student compares what iNaturalist said about a specimen with what a gen AI chatbot said, and can explain in simple terms why the two tools gave different answers. Student documents a systematic comparison between iNaturalist and gen AI identification for at least one specimen, identifies where they agreed and differed, and explains what the difference reveals about how each tool works. Student analyses the identification comparison across multiple specimens, draws conclusions about the conditions under which each tool is reliable, and evaluates the implications for using AI in biodiversity and conservation contexts.
4 Student explains what being at the stream added that a photograph, video, or AI explanation could not: the kick sampling technique, the surprise of what emerged, the feel of the water, the smell of the streambed. Student articulates what direct field observation provides that secondary sources cannot — independently collected, location-specific, time-stamped data — and explains why that matters for a water quality claim. Student reflects on the epistemological difference between field-collected data, citizen science observation, and AI-generated explanation: what each can and cannot constitute as evidence in an environmental science context.
5 Student submits at least one iNaturalist observation from the field visit and can explain that it will be seen by experts and used by researchers. Generates one question they would like to investigate at the same stream in a different season. Student submits iNaturalist observations, checks back for expert identifications, and formulates a testable monitoring question: what would need to change in the data to conclude that the stream's health had improved or declined? Student designs a monitoring protocol suitable for repeat sampling: specifies the sampling method, site location, frequency, species to record, MCI threshold for action, and a hypothesis about what future change would look like and why.