Thread by @MerKhaiBurch

ANOTHER paper in @ScienceMagazine came out claiming a 33% increase in soy yield. Since my last “yield” tweet took off - lets do it again!

From a (wanna be) plant breeder 5 years into my PhD, let me explain how this 33% increase in yield is misleading 🧵science.org/doi/10.1126/sc…

In this paper, the authors introduced three photosynthesis genes related to non-photochemical quenching - an essential process of protecting plants from high light intensity☀️ by avoiding the damaging effects of reactive oxygen species on the photosynthetic apparatus🌱 2/13

This is a cool paper on the physiology of photosynthesis& shading. In modern AG we control the amount of light that reaches the canopy, unlike in natural ecosystems that have to deal with layers of shading constantly. So there’s potential here for building on knowledge gaps 3/13

While this paper was cool, the yield trials lack sufficient replication to make any claims. The experiment was grown over 2 years (‘20, ‘21) in just 1 environment in Illinois. Solid yield trials are tested in 15-20 target location-year combos in numerous genetic backgrounds 4/13

Genotypes: they used an elite, transformable soybean line (Maverick) to test their constructs which is great. However, testing in more lines, especially elite ones backcrossed into un-transformed lines with blank constructs would be ideal, but this is a preliminary study 5/13

In 2020, all 9 lines were grown (WT untransformed, 8 versions of the constructs). Plants were grown in 20 blocks, with 1 line/block & 17 plants/line (n = 340). A quarter of a million soybean plants are typically grown, so 340 plants is REALLY small 6/13

Their yield estimates are ~5 t/ha for the WT and ~7 t/ha in their best line. This is where the 33% yield increase comes from, 1 environment in 1 year with 1 line.

The average increase was 24.5% (Fig 1B). Also, these results were only marginally significant (** = p < 0.05) 7/13

The average yield of soy in Illinois is ~4 t/ha so what they’re seeing is on the same level as state-wide trends (given their field design and ag management practices) 8/13

nass.usda.gov/Statistics_by_…

In 2021 they only grew 4 lines (WT+3 highest yielding lines from '20). Plants were grown in 16 blocks, 1 line/block, 17 plants/line (n = 272). Again not enough reps BUT a storm hit their field during grain-fill and caused many plants to lodge (fall over & not right themselves) 9/

This time all their high-yielding lines yielded LESS than their WT. Weather happens & they do mention this. However, the emphasis in their discussion & in the news don’t hype SFig 12.

Replication is key in science & unreplicated results make effects hard to believe 10/

This article has already taken off in the media (Altmetrics📈👀), with even the New York Times @nytimes writing a column on it in today’s print edition paper: 11/13

nytimes.com/2022/08/18/cli…

While these genes definitely hold some promise, as a plant science community we have to stop with “click-bait” yield science. We mislead the public, other scientists, and policymakers who allocate funding towards our research 12/13

In plant breeding, we need replicated field trials in numerous backgrounds in target growing environments.

Studies that do not replicate appropriately are not going to cut it.

13/13

Thanks everyone for reading through.

For those interested in a plant breeder's take on another @ScienceMagazine article about the "41-86% yield increase" in rice, check out my last thread:

https://twitter.com/MerKhaiBurch/status/1553072144426405889