Yield gap analysis measures the difference between the yield you harvested and the yield your field's rainfall, soil and season could have supported, then splits that difference into causes. For a rainfed field the correct benchmark is the water-limited potential yield, not a neighbour's best year or a seed company's trial plot. Once the gap is measured against that benchmark, the causes can be ranked by size.
The ranking is the useful part, because it rarely matches the problems a farmer notices first. Weeds and pest damage are easy to see. Low soil fertility and early water stress spread their losses evenly, so they are easy to miss.
What is a yield gap, and which benchmark should you use?
A yield gap is the difference between a benchmark yield and the actual farm yield. Agronomists use three levels, defined in a 2009 review by Lobell, Cassman and Field in the Annual Review of Environment and Resources:
- Yield potential (Yp): the yield of an adapted variety with no limits from water, nutrients, pests or diseases. This is the benchmark for irrigated fields.
- Water-limited potential yield (Yw): the same, but under rainfed conditions. This is the benchmark for rainfed fields.
- Actual yield (Ya): what the farm harvested.
Average yields rarely reach full potential. The same review found that yields in most major irrigated rice, wheat and maize systems sit at or near 80% of yield potential, with no evidence that they have gone higher. Past that point, the extra inputs and care usually cost more than the extra grain is worth. The useful target is therefore the exploitable gap: the distance from your yield to 80% of Yw.
For many farms that gap is very large. A 2016 study in PNAS by van Ittersum and colleagues found that rainfed maize yields in ten sub-Saharan African countries averaged 1.2–2.2 t/ha between 2003 and 2012, only 15–27% of water-limited potential.
How do you estimate water-limited potential yield for one field?
You do not need a research station or a crop model licence. There are two practical routes.
Route 1: a water-use benchmark. This method multiplies the water a crop actually used by the best transpiration efficiency observed in field data. The classic rule for wheat comes from French and Schultz (1984), summarised in a GRDC update paper:
Potential yield (kg/ha) = (crop water use in mm − 110) × 20
The 110 mm stands for soil evaporation, and 20 kg of grain per mm is the upper boundary of efficiency. Sadras and Angus (2006, Australian Journal of Agricultural Research) tested this on 691 crops from four continents. They found a maximum of 22 kg/ha per mm, with average crops at only 32–44% of attainable efficiency. For maize, Grassini, Yang and Cassman (2009) set the upper limit at 28 kg of grain per hectare per mm in the US Western Corn Belt.
Crop water use is stored soil water at sowing, plus growing-season rainfall, minus the water left in the soil at harvest. Free datasets cover each term:
- Rainfall: CHIRPS daily rainfall from the Climate Hazards Center, at 0.05° resolution (about 5.5 km) from 1981, between 50°S and 50°N.
- Temperature and radiation: the NASA POWER agroclimatology data, with daily weather from 1981.
- Soil: SoilGrids 2.0 from ISRIC, which maps texture, organic carbon and bulk density at 250 m for six depths down to 200 cm. Texture and depth set how much water the soil can store.
Route 2: a published regional value. The Global Yield Gap Atlas gives Yp and Yw for up to 13 food crops in more than 70 countries, following a protocol van Ittersum and colleagues described in 2013 in Field Crops Research. Start with the Yw for your climate zone and adjust for soil depth.
| Benchmark method | Data you need | Best for | Main weakness |
|---|---|---|---|
| Water-use rule (20–22 kg/ha per mm for wheat, 28 for maize) | Rainfall, sowing soil water, harvest soil water | Rainfed cereals in dry seasons | A fixed evaporation figure fails on sandy soils and in very wet seasons |
| Global Yield Gap Atlas value | Your location and crop | Fast first estimate | Averages a whole climate zone, not your soil |
| Best farms in your own records | 30 or more field-seasons of yield data | Comparing fields on one farm | Only as high as the best farmer in the dataset |
A fixed 110 mm evaporation term can mislead, because evaporation changes with soil type and season. Treat the rule as a first estimate.
Worked example: a rainfed wheat field
The figures below are illustrative, chosen to show the method.
A field had 60 mm of plant-available water at sowing, received 260 mm of growing-season rainfall according to CHIRPS, and had about 20 mm left at harvest. Crop water use was 60 + 260 − 20 = 300 mm.
- Water-limited potential: (300 − 110) × 20 = 3,800 kg/ha
- Actual yield: 2,100 kg/ha, or 55% of Yw
- Exploitable ceiling: 80% × 3,800 = 3,040 kg/ha
- Exploitable gap: 3,040 − 2,100 = 940 kg/ha
At an assumed wheat price of USD 230 per tonne, the exploitable gap is worth about USD 216 per hectare each season. The next step is to find which causes make up those 940 kg.
How do you rank the causes of a yield gap?
The most practical method on a farm is boundary-line analysis. It needs only your own records, collected across many fields and seasons.
- Collect the data. For each field-season, record yield and the factors you suspect: nitrogen rate, sowing date, previous crop, weed score, disease score, and soil pH from your soil test results. Aim for at least 30 field-seasons.
- Plot yield against each factor. Each point is one field-season.
- Draw the upper boundary. In each band of the factor, take the highest yields. The line through them shows the best yield reached at that level of the factor.
- Read the limit for each field. For each field-season, read the yield the boundary line allows for each factor. The factor with the lowest allowed yield is the one that limited that field.
- Add up the losses. Sum the gap attributed to each factor across all fields, then divide by the number of fields. The factor with the largest average loss goes first.
Fermont and colleagues used this method on 99 on-farm cassava trials in Kenya and Uganda (2009, Field Crops Research). Under farmers’ own management, yields averaged 8.6 t/ha. With improved varieties, better planting and fertilizer, they more than doubled to 20.8 t/ha. In fields that already had improved varieties and planting, poor soil fertility cut yields by 6.7 t/ha, early drought by 5.4 t/ha and weak weed control by 5.0 t/ha. Pests and diseases were “relatively unimportant.”
Australia's National Paddock Survey followed 250 paddocks in 2015 and 2016 and found average wheat yield gaps of 1.1–1.3 t/ha by region. After rainfall, which no farmer controls, nitrogen applied was the largest manageable cause in all three regions, ahead of previous crop and root health.
For irrigated rice, Silva and colleagues (2017, European Journal of Agronomy) split the gap in Central Luzon, Philippines, into efficiency, resource and technology parts. That method needs more data but shows whether inputs are misused or simply too low.
Why is the largest cause rarely the most visible one?
Visible problems are patchy. A weed patch or a disease focus is concentrated, so it looks severe. Nutrient shortage and early water stress lower yield a little on every square metre, so the field looks uniform.
In the cassava study, pests and diseases mattered little next to soil fertility, drought and weeds. In the Australian survey, nitrogen ranked above weeds and root health in every region. Ranking by what you see sends effort to the smaller causes.
Once the ranking is done, test the top cause before changing the whole farm. A simple on-farm trial with replicated strips shows whether fixing it closes the gap on your soil. If nitrogen ranks first, the guide on fertilizer use efficiency explains how to raise the share the crop takes up.
Frequently asked questions
What is a good yield gap?
A gap of about 20% of water-limited potential is close to the practical best. Yields above 80% of potential are rarely economic, because the extra inputs and labour cost more than the extra grain earns. Most rainfed smallholder fields reach far less, often only 15% to 55% of potential.
How many fields do I need for a boundary-line analysis?
Aim for at least 30 field-seasons, and more if you test many factors. The upper boundary needs enough points at each level of a factor to show the best yield actually reached. Pooling records with neighbours or a cooperative on similar soils reaches that number fastest.
Can satellite imagery estimate my yield gap?
Satellite data can estimate actual yield and map variation within a field, but it does not explain causes. It is useful as one input: rainfall and vegetation data help estimate water-limited potential, while your own records of nitrogen, sowing date and weeds are still needed to rank causes.
With Valora Earth, a farmer can record yield, sowing date and input use for each field by voice on WhatsApp, which builds the season-by-season records that a boundary-line analysis needs.