Appendix A2: Definitions of acronyms

Contents - Previous - Next

Appendix A3: Estimating crop densities and yields

In FSD trial work measurement of crop densities and yields are often two of the most important activities. The following sections illustrate the complexity of such activities and, in doing so, also indicate the necessity of often having to adjust the methodology to the local situation, in this case to the semi-arid climate of Botswana.

A3.1 Introduction and use of density measurement

After grain yield, plant density is the most important direct measurement in on-farm crop trials. Plant populations vary greatly among planted areas in Botswana. Different technologies and other causes can influence the percent of seeds sown that emerge and become useful plants. In FSD, plant density measurements are used to estimate the percent field emergence.

The FSD researcher should measure plant density when most of the plants of the eventual crop stand are emerged and established. Establishment is a relative term, but with sorghum it is usually four to six weeks after first emergence. Established plants generally have sent roots into the sub-soil below the ploughing layer. The researcher often wishes to measure the percentage of seeds sown that became established as the response variable. Two ways to calculate percent field emergence are:

• When it is assumed that all the seeds sown are viable:

Percent field emergence = [100 x crop stand (plants/ha)] / number seeds (seeds/ha)

• When a seed viability test is conducted to test the percent of seed that can germinate:

Percent field emergence = [100 x crop stand (plants/ha) x % seed viable] / number seed (seeds/ha)

Note: Researchers not farmers, usually measure population density. When it is not possible or desirable to conduct a stand count for a whole plot, it is necessary to use some sampling procedure. FSD staff in Botswana frequently use the systematic quadrat sampling technique, In the following sub-sections, this method plus several of its variants is discussed,

A3.2 Measuring plant densities

A3.2. 1 Systematic Quadrat Sampling for Broadcast Planting

The procedure consists of the following steps:

• Sub-Sampling. Sub-sample the plot with a 2m by 2m quadrat. Take 20 quadrat sub samples in a systematic pattern. Usually, this pattern consists of two passes through the plot each with 10 equally spaced sub-samples. Preferably, two people carry the quadrat, count plants, and record data. The front carrier follows a sampling line and the rear carrier counts paces. The rear carrier places the quadrat on the ground, without bias, at the point of his or her toe,
• Data Collection. In each sub-sample, count and record the number of plants of the crop,
• Calculate Crop Stand. This is done as follows:

Crop stand (plants/ha) = [average number plants per quadrat x 10,000 quadrat length (metres)] / quadrat width (metres)

For example, if there is an average of 8.3 sorghum plants per sub-sample and a quadrat sub-sample of 2m by 2m, the:

Crop stand (plants/ha) = (8.3 x 10,000)/(2 x 2) = 20,750 plants/ha.

Note: When the shape of the plot is long and narrow or of another irregular shape, the sub-sampling pattern should be such that quadrat sub-samples are spaced as equally throughout the plot as is possible.

A3.2.2 Systematic Quadrat Sampling for Row Planting

• Sub-Sampling. Sub-sampling for row-planted plots is similar to that for broadcast planting. The pattern for rows consists of two passes, each with 10 equally spaced sub samples. Each pass follows rows near, but not on, the border, Therefore, select rows that are one-fourth of the plot width from the border.
• Data Collection. The procedure is similar to that for broadcast plots. The main difference is that for row planting, the number of rows in each sub- sample is recorded. The number of rows sampled should remain the same for all sub-samples in a plot. When row spacing is 0,75 metres, two rows should be included in each sub- sample,
• Calculate Crop Stand. This is done as follows:

Crop stand (plants/ha) = [average number plants per quadrat x 10,000] / [number rows in quadrat x quadrat length (metres) x row spacing (metres)]

For example, with an average of 7.4 sorghum plants per sub-sample, when each quadrat sub-sample is 2m x 2m, with an average row spacing of 0,75 metres, and two rows in the sub-sample, then the:

Crop stand (plants/ha) = (7,4 x 10,000)/(2 x 2 x ().75) = 24,667 plants/ha.

Note: Researchers should avoid sampling in rows where the plant stand is unusual due to a cause other than the treatment. The most common example is blockage of a planter during one pass through the plot. When this problem is observed, a neighbouring, but not adjacent, row is sampled.

A3.2.3 Row Segment Measurement for Row Planting

• Sub-Sampling. Sub-sampling is on segments of rows. This method is similar to the systematic quadrat method for row planting The main difference is the use of a measuring stick or measuring string to mark the row segment instead of a square quadrat.
• Data Collection. Data collection is also similar to that of the systematic quadrat method for row planting. The segment length is recorded as the length of the stick or other measuring device used.
• Calculate Crop Stand. This is done as follows:

Crop stand (plants/ha) = [average number plants per segment x 10,000] / [segment length (metres) x row spacing (metres)]

For example, with an average of 21.4 sorghum plants per segment sample, each segment having a length of 1() metres, and average row spacing of 0.75 metres, then the:

Crop stand (plants/ha) = (21.4 x 10,000)/(10 x 0.75) = 28,533 plants/ha.

If more than one row in the segment sub-sample is included, then:

Crop stand (plants/ha) = [average number plants per segment x 10,000] / [segment length (metres) x row spacing (metres) x number of rows]

Note: Researchers who use this method must check that the measuring stick does not slide out of position when they make their plant counts.

A3.2.4 Methods for Mixed Cropping in a Broadcast Planting

Use the method described for systematic quadrat sub-sampling for broadcast planting (see Section A3.2.1). When crops are mixed, the researcher must record the number of plants separately for each crop that he or she identifies in a sub-sample. A crop stand for each crop in the mixture and for the combined mixture is calculated.

Note: Mixed cropping situations happen in many on-farm trials. Volunteer watermelons, cowpeas, and other crops commonly establish in sorghum trials. In experiments controlled by the farmer, these volunteers should be left and counted. In other experiments, where agronomic data are more important, the researcher may wish to remove these plants. The researcher should not count the removed plants.

For example, if there is an average of 8,3 sorghum plants and an average of 0,9 watermelon plants per sub-sample and a quadrat sub-sample of 2m by 2m, the:

Sorghum stand (plants/ha) = (8,3 x 10,000)/(2 x 2) = 20,750 plants/ha

And the:

Watermelon stand (plants/ha) = (0,9 x 10,000)/(2 x 2) = 2,250 plants/ha

Therefore:

Intercrop stand (plants/ha) = 20,750 sorghum plants + 2,250 watermelon plants/ha.

A3.2.5 Methods for Mixed Cropping in a Row Planting (intercropping)

Use the method described for row segment measurement for row planting (see Section A3.2.3), The researcher sub-samples or counts each crop in the intercrop and records these data separately, Usually, the segment samples for one crop are paired with segment samples for the other crop. Record the proportion of intercrop rows occupied by each crop.

Stand for each crop (plants/ha) = [average number plants per segment x proportion of rows x 10,000] / [segment length (metres) x row spacing (metres)]

Stand for the intercrop (plants/ha) = ((average number plants of first crop per segment x proportion of rows) + (average number plants of second crop per segment x proportion of rows) x 10,000)/(segment length (metres) x row spacing (metres))

Take the example of a two-row sorghum to one-row cowpea intercrop. Research staff count an average of 18.4 sorghum plants per segment sample and an average of 12.6 cowpea plants per segment sample. Each segment has a length of 8 metres, The average row spacing is 0.82 metres.

The sorghum stand = (18.4 x 0.67 x 10,000)/(8 x 0,82) = 18,793 plants/ha

The cowpea stand = (12,6 x 0,33 x 10,000)/(8 x 0,82) = 6,338 plants/ha

The intercrop stand = [((18.4 x 0,67) + (12,6 x 0,33) x 10,000)] / (8,0 x 0,82) = 25,131 plants/ha

Note: The inter-crop stand equals the sorghum stand plus the cowpea stand.

A3.2.6 Percent Ground Cover as an Alternate Measurement

Percent ground cover can be used as an alternative to plant counts in some situations. It is often preferable to estimate weed growth by percent ground cover than by plant counts. This is because weed plants differ enormously in size per plant, Spreading crops such as watermelon, pumpkin, and indeterminate cowpea also might be measured as ground cover instead of plant number.

On-farm researchers in Botswana use percent ground cover to measure weed growth before ploughing, weed growth at weeding time, weed growth late in the season, and watermelon growth in a sorghum-melon mix. Researchers use two different cover-estimation methods:

• Whole Plot Method. FSD researchers estimate visually the percent of ground covered by vegetation. The estimate is from 0 percent (no ground cover) to 100 percent (when vegetation cover equals the ground area). The researcher stands at one end of the plot to make the estimate. If the plot is long, the researcher divides the length into equal parts and estimates percent cover for each division of the plot. The plot estimate is the average of all division estimates.
• Quadrat Sub-Sampling Method. The ground cover is estimated using two quadrat sub sampling methods. For each method, quadrats are used that measure 2m by 2m:

• Method 1: The quadrats remain at fixed positions in the plot. The positions represent different parts of the plot, This method is used when repeated measures of cover are needed. Because of the time and material costs of maintenance, work is usually to confined to a very few fixed quadrats (e.g., two to four per plot). Again, a visual estimate (i.e., 0 to 100 percent) of the percent of ground covered by vegetation is usually made.
• Method 2: Quadrats are placed for one measurement only. Sub-sampling is similar to that for systematic quadrat sub-sampling for broadcast planting, (Section A3.2. 1 ). The same visual estimate of the percent of ground cover is usually made.

Additional points to note are:

• Contrary to conventional wisdom, estimates of a percent cover are better than a simpler rating of cover, even if precision of the percentage seems suspect. Compare a 0 to 10 rating system with a 0 to 100 percent system. Some researchers may never differentiate better than between ratings of six or seven. Even then, nothing would be lost in using percent estimates, because true differentiation is still between 60 and 70 percent. Other researchers are able to differentiate better than between a six or seven rating particularly near the zero or 10 extremes. For these researchers, the 0 to 100 percent estimate is more precise.
• Accuracy, precision, and consistency between researchers will be improved with simple training on estimating percent ground cover. Even after training, several different staff members are asked to cross-check estimates on some of the same plots, Experience shows that a core of staff can be identified to give repeatable estimates of percent ground cover,

The following situation occurs commonly in on-farm research, Suppose there is a need to know the sorghum plant density as well as the ground cover provided by a secondary intercrop or by weeds. Using systematic quadrat sampling (see Section A3.2.1), the following were found: an average of 8,3 sorghum plants and an average watermelon plant cover (i.e., Method 2 above) of 62% per sub-sample, the:

Intercrop stand = 20,750 sorghum plants/ha, + 62% watermelon ground cover,

A3.3 Introduction to crop yield estimation

Yield is the most important direct measurement in crop experiments, In this section, methods that are used to measure crop yield in on-farm research are discussed, Over the years, these methods have been used and modified to suit the needs of work in Botswana. The methods vary, because requirements of experiments differ. Researchers should thoughtfully select the appropriate method for each experiment in their programme.

Once a method is selected, it should be used throughout the trial or experiment. This can ensure that differences in the data are a result of treatments and not of a change in research methods. If circumstances require a change, the change should take place between replications and not between treatments. Treatments in a replication must be handled in a uniform manner.

Not only methods, but personnel and equipment, cause bias in a yield measurement, if one treatment is favoured. Farmers or staff sometimes have a preferred treatment. When this happens, the data become biased and possibly invalid. Training of staff and farmers helps avoid bias caused by uneven use of measuring methods,

The researcher will choose a method for several reasons: speed, ease, cost, information needed, precision needed, nature of crop or crop mixture, and design of the trial, The researcher must also remember that yield measurement, whether by researcher or farmer, should be handled in a way that does not greatly inconvenience the farmer.

Contents - Previous - Next