Five variables, one sample: what a stream survey can and cannot prove
Te awa o Te Rākau runs from a QEII-covenanted bush block, through a dairy farm, and into an estuary where the local hapū gathers pipi. A Year 11 Pūtaiao class sampled three sites once, between 10:00 and 11:00 am on 12 March, two days after 38 mm of rain:
| Site | Temp (°C) | Turbidity (NTU) | E. coli (cfu/100 mL) | Nitrate-N (mg/L) | MCI |
|---|---|---|---|---|---|
| A – above farm | 13.8 | 2.4 | 45 | 0.3 | 118 |
| B – farm drain | 16.1 | 18.6 | 980 | 4.9 | 72 |
| C – below farm | 15.2 | 14.2 | 610 | 3.6 | 81 |
(The MCI, or macroinvertebrate community index, rises with cleaner, cooler, more oxygenated water. The recreation guideline for E. coli is 540 cfu/100 mL.)
The class writes: 'The farm is polluting the stream, and the worst pollution is at site B.' The farm manager replies that the drain also carries runoff from a gravel track and a winter stand-off pad, and points out that the estuary is fed by a second, unmonitored tributary.
1. Which parts of the class's statement - 'The farm is polluting the stream, and the worst pollution is at site B' - are supported by these data, and which are not? Justify each judgement with specific values.
Read the worked answer
Supported
- On that day there is a consistent downstream gradient. Turbidity rises 2.4 → 18.6 NTU, nitrate-N 0.3 → 4.9 mg/L, E. coli 45 → 980 cfu/100 mL, and MCI falls 118 → 72. The size of the changes matters: nitrate is about 16× higher at B than at A and E. coli about 22× higher, which is well beyond plausible measurement noise.
- Site B (980 cfu/100 mL) is roughly double the 540 cfu/100 mL recreation guideline, so B was not safe for contact recreation on that day.
- The chemical and biological lines of evidence agree - a 46-point MCI drop is what you would expect where sediment and nutrients rise - so the pattern is not resting on one indicator.
Not supported
- Causation. One sampling occasion cannot tie the pattern to the farm. The manager names two other sources feeding the same drain (track runoff, stand-off pad), and the unmonitored tributary could be supplying much of what appears at C.
- Timing. Sampling two days after 38 mm of rain means recession flows, when sediment and faecal indicator bacteria are still elevated. Without a baseflow sample, the A–B difference may partly be a storm artefact rather than normal conditions.
- No replication, no flow. One sample per site gives no spread, so 4.9 versus 3.6 mg/L cannot be called a real difference. Without discharge (L/s), loads cannot be calculated, and 'dilution' versus 'extra inputs' at C stays unresolved.
- 'Worst at B' is defensible only for that hour on that day, not as a general claim.
2. Choose two changes to the sampling design that would most strengthen the evidence, and explain the reasoning behind each.
Read the worked answer
Any two of the following, provided the reasoning explains what alternative explanation the change removes.
Repeat over time and flow states. Sample the same three sites at least monthly for 12 months, plus targeted samples at baseflow and on the rising and falling limb of storms. Reasoning: a single wet-weather snapshot cannot separate routine discharges from storm runoff, and only repeated sampling shows whether the A → B → C pattern persists. A pattern that holds at baseflow is much harder to attribute to rain.
Sample the inputs directly, and measure flow. Add sites in the farm drain above the track and pad, above and below the unmonitored tributary confluence, and record discharge at every site. Reasoning: this is what allows source apportionment - concentration alone misleads, because a modest concentration carried in a large flow can be a bigger load (load = concentration × flow). Measuring flow also settles whether C is lower than B because of dilution.
Replicate. Three samples or three transects per site, reported as a mean with a range or 95% CI. Reasoning: without an estimate of variability there is no way to judge whether a difference is real; overlapping ranges would force a more cautious conclusion.
Control the sample handling for E. coli. Chilled containers, analysis within 6 hours, field blanks. Reasoning: 980 cfu/100 mL is only useful evidence if it is not a holding-time or contamination artefact.
3. Site B is only 1.3 °C warmer than site A, yet its MCI is far lower. Explain how temperature, nitrate and turbidity could combine to reduce macroinvertebrate numbers, and why the result matters for the hapū's pipi beds.
Read the worked answer
Mechanisms, linked rather than listed
- Temperature: warm water holds less dissolved oxygen, and 16.1 °C also raises the metabolic oxygen demand of animals. A 1.3 °C rise is modest on its own, but it acts on top of the other pressures - the effect is cumulative, not additive in a simple way.
- Nitrate: 4.9 mg/L fuels algal growth. When that algae dies, microbial decomposition consumes dissolved oxygen, so the oxygen minimum tends to occur at night and just downstream - exactly where sensitive taxa (mayflies, stoneflies, caddisflies) are lost first. That is consistent with an MCI of 72 versus 118.
- Turbidity: 18.6 NTU means suspended sediment. Sediment abrades gills, fills interstitial spaces in the bed that invertebrates use as refuge, and reduces light for submerged plants. Fine sediment also carries attached bacteria and phosphorus.
- Why the MCI drops so far: the taxa that disappear are the oxygen-sensitive ones; the community that remains is dominated by tolerant worms and snails, so the index falls sharply even though total animal numbers may not.
Why it matters for the pipi beds Pipi are filter-feeders, so they concentrate suspended sediment, bacteria and any contaminants bound to it. Elevated E. coli at 980 cfu/100 mL signals faecal contamination, which is a food-safety issue for a mahinga kai area, not only an ecological one. Sedimentation also changes the estuary bed itself, smothering the substrate pipi need. The data therefore support a precautionary message to the hapū about harvesting, while the source question stays open until the design is improved.