Split-field fertiliser trial on grassland
On a 6 ha grassland field, a student compares two nitrogen fertilisers: treated urea (A) and calcium ammonium nitrate (B), both applied on 1 April at 30 kg N/ha. The field is split into two 3 ha blocks along a natural slope. Block A is on the upper, freely draining sandy loam; block B is on the lower, heavier clay loam that lies wetter. Five quadrats are cut in each block on 15 May and again on 15 June. Block B averages 0.4 t DM/ha more than block A. No soil samples were taken before the trial, and 45 mm of rain fell within three days of application. The student concludes that treated urea is clearly inferior and recommends it not be used on the farm. The teacher asks the class to review the trial design before any recommendation is made.
1. Identify the main factors that make this trial unable to support the student's conclusion. Explain how each factor could produce the observed result.
Read the worked answer
The trial confounds treatment with field position. Block A receives treated urea and is on free-draining sandy loam; block B receives CAN and is on heavier, wetter clay loam. Any difference in yield could be due to soil texture, drainage, slope, organic matter or background fertility rather than fertiliser type. The lower block may retain water and nitrogen better, while the sandy upper block may leach nitrate after the 45 mm of rain. There was no baseline soil test, so initial nitrogen, phosphorus, potassium and pH differences are unknown. There is no true replication of treatments: each product is applied to one large strip, and the five quadrats within a strip are subsamples, not independent replicates. This is pseudoreplication. The student also has no untreated control, so the size of the nitrogen response is unknown. Application method and exact timing within the day are not stated. A 0.4 t DM/ha difference is small enough to arise from soil variation and measurement error. The conclusion that treated urea is inferior is therefore unsupported; the trial has not isolated the fertiliser effect.
2. Outline a randomised and replicated design that would compare the two fertilisers fairly on the same farm. Include what should be measured and how many plots you would use, with a reason.
Read the worked answer
Use a randomised complete block design with blocks based on soil and drainage. For example, divide a uniform field into four blocks, each containing two plots of at least 0.25 ha, giving eight plots in total: four treated urea and four CAN. Randomise which product goes into each plot within every block, and leave a buffer strip between plots to prevent overlap. Apply both products at the same nitrogen rate, on the same date, with the same applicator and in similar weather. Include an untreated control if the aim is also to measure the nitrogen response. Record soil test results, soil moisture, sward height, dry matter yield, nitrogen content of the herbage, rainfall and any visible leaching or runoff. Cut quadrats from fixed, marked areas and keep them separate by plot. Four replicates per treatment allow the variation between plots to be estimated; randomisation prevents the farmer or student from unconsciously assigning the better ground to one product. A single season is still limited, so repeat the trial or treat the first year as preliminary.
3. The farmer wants a decision before next season. What should the student report, and what further evidence should be requested?
Read the worked answer
The student should report that the current comparison is inconclusive because fertiliser type is completely confounded with soil type and drainage, and because there is no replication or baseline soil data. The 0.4 t DM/ha difference cannot be attributed to treated urea being inferior. The report should describe the field layout, the rainfall event, the lack of soil tests and the pseudoreplication. Before changing farm practice, the student should request soil test results, application records, yield measurement method and weather data. The student should recommend a small replicated trial on uniform ground, with an untreated control and at least four replicates, rather than a farm-wide ban. If the farmer needs a decision immediately, the safest advice is to use both products where they suit the soil and weather, and to test the claim properly next season.