Looking closely at their yield trials, they grew overexpressed OsDREB1C lines over 3 years in 3 environments, testing only 99-120 plants each time (Tables S1-S5). For yield trials, this is a VERY small # of plants to test for yield stability (it's usually of 1000s to millions) 4/

Zooming in on their yield, they report the 41-68% yield gain on a per-plot basis from 1 year in 1 environment where they yielded 910-1084s vs 644g.

Yield is typically measured in tons per hectare, so going from g to T/ha given their plot spacing, they get 1.16-1.39 T/ha. 5/

Math for the yield numbers:

20 rows*10 plants per row*.233 interspace rows *.167 interspace plants=7.78 meters^2/plot

1 hectare=(100 meters)^2=10000 meters^2

# of their plots that could fit in a ha: (10,000m^2)/(7.78 m^2/plot)=1284 plots/ha

1284 plots/ha*1084g=1.39 T/ha

6/

1.16-1.39 T/ha is an incredibly low yield. The average yield for Chinese rice is 6.8 T/ha; it can get up to 12 T/ha (see linked pub).

With the same vibes as their abstract, commercial varieties yield 389-763% more than lines overexpressing OsDREB1C 7/ frontiersin.org/articles/10.33…

There’s no Nature or Science paper for that 389-763% yield increase because it’s what plant breeders have routinely accomplished over decades of plant improvement & genetic gain 8/

The variety they used, Nipponbare, isn’t meant to be a high-yielding variety. This is probably why the authors overexpressed OsDREB1C in the elite cultivar Xiushui134. Their grain yield was ~7-8 T/ha. In other studies, Xiushui134 yields ~10T/ha (9/)

doi.org/10.1016/S2095-…

Flashy “Yield increase by [insert wild #] in X crop” studies are frequently published in big-name journals but rarely follow the rigorous QC steps needed to verify these yield claims.

Once the news gets out - articles like these spread like wildfire (Altmetric metrics👀) 10/

With other scientists & the public thinking plant breeders can “solve” yield with 1 gene, more $$ goes into studying unreliable single gene effects over the incremental, stable & repeatable yield gains over 1000s of small effect loci performed by industry & uni plant breeders 11/

Although large-effect single gene changes can drastically alter yield, let's not forget that evolution has had millions of opportunities (or more) to overexpress OsDREB1C over the past 10k years, have these changes become fixed, & then have it contribute to yield 12/

This won’t be the last time a paper like this is published. Keep an eye out for the sample size, yield estimates against commercial lines, and if these estimates are corrected for field-design/grain-moisture/etc…. 13/

My angry plant breeding student plea to future authors of “Yield increase by [insert wild number] in X crop” studies:

Please verify your yield estimates in multi-environment trials in elite backgrounds. Yield in lab-optimized, unimproved lines means very little 14/