Raman Spectroscopy in Agriculture: An Interview with Dmitry Kurouski

Dmitry Kurouski, an associate professor of biomedical engineering in the department of Biochemistry and Biophysics at Texas A&M University in College Station, Texas, USA, recently spoke to Spectroscopy about Raman spectroscopy’s role in determining crop yield of key food items as the world population continues to increase. His paper on the subject, “Raman Spectroscopy and Machine Learning for Agricultural Applications: Chemometric Assessment of Spectroscopic Signatures of Plants as the Essential Step Toward Digital Farming,” co-authored by student Charles Farber, appeared in the journal Frontiers in Plant Science in April 2022 (1). The following is transcribed from an online video interview between Spectroscopy and Kurouski in January 2024 with the assistance of artificial intelligence (AI) software, and has been lightly edited for length and clarity.

Please briefly summarize the problem you were examining in this study and its implications in real-world applications. At its heart, the concern here is the issue of crop yield, as the world’s population continues to increase and demand more of the food supply. Is that correct?

This is completely correct. The big project that we pursue here at Texas A&M is we want to develop sensors that can be used directly in a field for non-invasive and non-destructive analyses of plant health. And, developing the sensors, we aim to detect and identify plant diseases, as well as so-called abiotic stresses such as drought and salinity, stress, and so on. And we understand that if we detect them in a timely manner before the manifestation spreads, we can first reduce the amount of pesticides, herbicides, et cetera, that are used to treat the problem. And there’s a direct economic yield here. But ultimately, we want to save the crop yield, because if we can seize the spread of fungal infection or viral infection, then we can save fields in the entire crop yield in the entire field and the country. Ultimately what we found, working on this topic for the past five years in my lab, is that we get very good signal-to-noise spectra from plants, and we observe vibrational bands that correspond to plant biological molecules, such as carotenoids and phenylpropanoids. And we have done a very important and foundational study to understand what changes in plants, biochemically. The big piece that remains to be covered is biological variability in plants, because plants grow in different environments and different geographic areas, different soils, et cetera. So it’s very important to demonstrate that the sensing that we develop can work on different plants and different areas in the U.S. and elsewhere. And the way to address this problem is to develop chemometric methods, to develop statistics that can separate spectra from healthy plants and from infected plants. So the point of this paper was to describe to the audience what we worked on for the past five years and really summarize the advantages of many chemometric approaches that people can use to answer these questions.

These experiments were carried out primarily on roses, is that correct?

Well, that’s a good point. So for simplicity, we focused on roses as a plant model. But in reality, for the past five years, we worked on a large number of different plants. We worked on corn, we worked on wheat, on rice, and we still very actively work on rice, we work on banana, we work on even more exotic things, again, because we target the entire farming sector of the United States.

Many of these studies do kinds of comparisons between different techniques or approaches. What was the advantage of Raman here?

The advantage of Raman is, first and foremost, it’s non-invasive and non-destructive, and here it beats PCR and ELISA, as well as more exotic methods such as flow cytometry, that are commonly used in plant pathology. Imagine that all we do is we shine laser light on a plant. We don’t have to sample it, we don’t have to grind it, do lots of gel, et cetera, and ultimately we use no chemicals. So essentially, it’s big because it saves direct cost of analysis, but also Raman probes the chemical composition of the plant. And it’s very important because there are other imaging techniques, such as thermography, for example, and RGB imaging that were first developed for satellites. And now people use drones to do this type of imaging, but with this type of thermography, RGB imaging, people don’t probe the structure of the plant, and they only rely on change in the color. But obviously the plant can get yellow due to about five different reasons. It can be drought or bacterial infection, and this is the weakness of this method. So Raman is advantageous because it’s non-invasive, non-destructive, and because it is sensitive to the chemical structure, but also it is portable, so we can bring it directly in the field, and we can also have Raman telescopes that can be operated on vehicles that will travel through the field fully autonomously.

You mentioned that this work took place over the course of five years, and it’s understandable that studies sometimes may take a long time until you find either something close to what you were looking for, or something totally opposite from what you may have been looking for when you started. As this progressed, what were any challenges or surprises that you found contrary to what you thought you would be exploring initially?

That’s a very good question. There were a lot. So first and foremost, when we started this project five years ago, no one believed it was possible. No one believed, not a single person believed that if we shine laser light on a plant, we can find what happens to the plant. To them, it was impossible, but we took a challenge, we said, we know it’s possible, and we made it work. And we first made it work on corn, but then it was very important to demonstrate that technique works on a large number of crops, because the way it works in the plant world is, if someone is a wheat breeder, that person is focused on wheat.
They don’t do corn. And if someone is a corn breeder, that person doesn’t want to hear what happened in rice, for example. So we understood that to make the entire community excited, we have to do lots of different plants and crops. So that was the first challenge, and we succeeded on all of them. The second biggest thing for us was to demonstrate work. What do we sense? What changes in plants? Because we knew carotenoids change. But the question is, what are these carotenoids? There are many of them. Same as for phenylpropanoids. What is changing? So over the past years, we did extensive HPLC analysis of the plant material to find what do we actually sense by Raman and convince the plant world that what we sense is actually what happens in plant biochemistry. So it’s not the technology. We actually know what we did for them. It was a game changer. After this foundational work, they understood what we see in a plan, because they use HPLC, so we have to use their method. Finally, and this is what the success came on, in the last year, the biggest thing was to demonstrate that we are pathogen-specific, that we are sensing the problem, that we can differentiate between drought and viral infection, et cetera, et cetera. And on two different crop systems, on wheat and on corn. And at this moment, we do it on rice. We demonstrate that Raman is specific to different stresses, and we can also detect and identify so-called combined stresses when, for example, plants experience drought and a fungal infection.

And when you’re talking about some of these crops, roses are nice to receive, or they’re a nice gift to give, but you’re talking about wheat, rice, corn—these are essential to our survival.

You’re completely correct. We are focused on so-called economically important plants. It’s first and foremost wheat and corn. But we also very actively work on rice, because in the U.S., we grow a lot of rice, primarily along the Mississippi River, and here in Texas it’s very big because it’s humid and the climate is quite suitable for rice production. But also, we didn’t overlook other important plants, like banana, for example. The big problem we have with banana is that genetically, they’re all very similar. They’re like twin brothers and sisters, and because of these, they are not resistant to diseases. And the big problem that we all face with banana is so-called fusarium wilt, or a fungus that destroys banana, and again, banana is so vulnerable that we cannot do breeding to make them stronger. And it becomes very important to have a technology that one can use to sense sick bananas and quickly remove them from the field. I also want to point out that we very actively work on citrus plants, and we were able to diagnose so-called citrus greening disease, or Huanglongbing; all trees in Florida are infected. But it’s not yet the case for Texas and California. So it’s very important to have the sensing technologies to save the citrus industry.

And these approaches, using these technologies that you have been, this is in line with what’s happening in the agricultural industry at large, right? For years, I’ve heard that this has kind of taken a lot of human element out of it—entire fields are planted by GPS mapping.

Precisely correct. It’s very active in all different areas of farming, especially soil thermometry. It’s quite well developed. So now farmers have sensors and soil, and they don’t show up in the field until temperature is perfect as well as many other things that you describe in terms of GPS navigation, et cetera. So our discoveries are quite in line with what is trending now, in terms of the field it’s in, general, we call it all digital farming, or digital agriculture. And this digital farming concept, everyone understands, is the only way to save the planet, because every year we have increasing population, and at this moment we have millions of people around the world suffering from different kinds of malnutrition. So obviously, to solve this problem, we would have to increase agricultural territories. But sooner or later, we’d turn the whole planet into farm field, which is not exciting. The alternative solution is to do it better, is to make farming better, and sensors are what allow us to do farming better and increase the yield without the need to increase agricultural territory.

Is there anything else that you think our audience would like to know about what you’ve been studying? I know that it continues, as you said, with rice currently, but anything up to now that you’d like to add?

I think we covered most of it. It was a really great opportunity to share this information with you. Of course, as usual, I would want to give the credit to my students. And at this moment, I have a very talented PhD student, Isaac Juarez, leading this work, and he works on this and he did a fantastic job already modeling several different stresses in rice. His work is focused on Arsenicum toxicity. It’s the biggest problem in Asia because soils have high concentration of Arsenicum 3 and Arsenicum 5, and these poisonous ions get accumulated by plants. And people, unfortunately, consume this bad rice. So what we try to do is to develop sensing that can be used to probe toxins in plants, these metal-based toxins. And of course, I want to give the credit to him, because he works on this day and night. And he is an absolutely gifted guy who wants to go and work for NASA. He used to be in a corps of cadets here, and definitely a very organized guy.

The original interview with Kurouski can be found at this link: https://www.spectroscopyonline.com/view/spectroscopy-agriculture-interview-dmitry-kurouski

Reference

(1) Farber, C.; Kurouski, D. Raman Spectroscopy and Machine Learning for Agricultural Applications: Chemometric Assessment of Spectroscopic Signatures of Plants as the Essential Step Toward Digital Farming. Front. Plant. Sci. 2022, 13, 887511. DOI: 10.3389/fpls.2022.887511

Dmitry Kurouski is an associate professor of Biochemistry & Biophysics and Biomedical Engineering at Texas A&M University in College Station, Texas.●