IRRI Agronomy Challenge 2: What’s the yield?

Author // Dr. Achim Dobermann Categories // Achim Dobermann's blog

30 April

We’ve harvested a great crop this year because we did almost everything right and the weather was good.

Measuring the components that determine grain yield.

Hybrid rice harvest.

Complete video playlist in this series.

That is what I wanted to write after we harvested our field, but now that the numbers are in we must report another disappointment of not being able to reach our yield goals. Let’s just concentrate on the naked facts and come back in a couple of weeks with a deeper analysis. We had two beautiful harvest days and using a modern rice combine made it very easy to harvest the hybrid variety on April 5, and the inbred variety one week later. No lodging this time and the combine generally did a good job in terms of producing clean grain. So, things seemed alright, but driving through on the combine it already dawned on me that we just didn’t seem to have enough grain out there for a high yield. There were many good areas, but also many poor patches. In a nutshell, measured as total grain weight for the whole plot and including all gaps or empty patches, the inbred yielded 6.57 tons/ha (expressed at 14% moisture content) whereas the hybrid yielded only 5.62 tons/ha. In transplanted rice fields grain yield usually has a coefficient of variation of about 15-20% and follows a normal frequency distribution. So, what that means is that we actually achieved our target yield of 7 t/ha on about one third of the inbred field area.  In the case of the hybrid, however, the yield target of 7.5 t/ha was only achieved in a few places, less than 5% of the field.

Achim Dobermann on IRRI Agronomy Challenge

Dr. Achim Dobermann examines the grains.

Measuring the components of yield provides some indications for what may have caused a high or low yield. With help from our crop physiology group we collected 10-hill samples at physiological maturity stage, about a week before the combine harvest. That is the time when grain filling has been mostly completed. In the video, Leny Bueno explains how it was done through pinpointing 10 random sampling locations for each variety half and then collecting 10 hills along a single row at each location. Those samples were then processed in the lab to determine the three components that make up the final yield:

yield = no. of panicles per m2 × no. of filled grains per panicle × average weight of a grain

The latter is usually expressed as “1000-grain weight”, to make it easier. One can measure many other interesting things on such 10-hill samples. You can find the entire data set here [D7-2013DS Yield components.xls]. Let’s just look at a basic summary of the 10-hill sampling:

Yield and components of yield (10-hillsamples)



t-test at 5% alpha

Grain yield (t/ha, 14% moisture)




Harvest index (grain dry matter/total dry matter)




Number of panicles per m2




Number of filled grains per panicle




Dry weight of 1000 filled grains (g)








Number of filled grains per m2




Number of unfilled grains per m2




Filling efficiency, ratio of filled grains to total grain number




Tillering efficiency at harvest, reflects unproductive tiller




 ªns = nonsignificant.

So what does this tell us? First, you will notice that the yield measured from the 10-hill samples collected at 10 locations in each field is greater than the total grain yield measured with the combine. That is because the latter includes the whole plot and thus all gaps and areas with poorer growth, whereas the 10-hill samples may not be fully representative of that. It’s a common problem and a good reminder that one cannot measure the real crop yield per area using such small samples. Small sampling areas are good for analyzing yield components, but not for measuring productivity as a whole.

The mean yield collected from those 10-hill samples was also higher in the inbred than in the hybrid, but, due to the larger variability associated with such small samples, it was not statistically different. But, there were significant differences in several yield components. Generally speaking, the hybrid had fewer filled grains per panicle, more unfilled grains, and more unproductive tillers than the inbred. It seems very clear that we simply didn’t have enough biomass to produce a large sink size for a higher yield. Machine planting at 30-cm spacing resulted in a low hill density of only 16–17 hills per m2 (compared with 25 hills if you plant at 20 × 20 cm). Moreover, the lower yield of the hybrid was probably also because it had one week less growth duration (less time to fill grains) and it had a large proportion of unfilled and immature grains due to the rat damage we reported earlier, which didn’t affect the inbred

Read more items from Achim's blog.


About the Author

Dr. Achim Dobermann

Dr. Achim Dobermann

Achim is a soil scientist and agronomist with 25 years experience working in Asia, North America and Europe. He is recognized internationally as an authority on science and technology for food security and sustainable management of the world's major cereal cropping systems. He has authored or co-authored over 250 scientific papers and two books on nutrients in rice and has received numerous awards from various academic, government and industry organizations. He is a Fellow of the American Society of Agronomy and a Fellow of the Soil Science Society of America.

Read his profile | more blog articles