Show Notes & Resources Mentioned:
For questions or comments, contact Melissa via email at email@example.com.
What is a podcast?
For those of you who are newer to the medium, a podcast is like a pre-recorded radio show. In the same way that you turn on a talk radio show, you have to turn on a podcast. The major difference is that while our cars are equipped to find radio frequencies, they are not built to accommodate direct access to podcasts. On your smartphone or computer with internet access (since the files tend to be on the larger side), you can discover podcast shows of any kind, in any field, on any topic.
Listed above are some of the most used podcast hosts. iTunes and the iTunes Podcast app are preinstalled on your iPhone and are the simplest tools to use. You simply search for “WSU Wheat Beat Podcast” in the search bar, hit “subscribe” and the download arrow, and listen whenever it’s convenient for you.
If you use an Android or use another type of smartphone, you will need to find a different podcasting app because those devices don’t come with a preinstalled app like Apple. If you don’t know which podcast app you’d like, simply hit the “Android” link above and it will show you to several Android podcast apps for you to choose from.
After you download an episode, you can listen without using data any time of day. Our goal is to post a new podcast every other Monday. Your podcast app should automatically load our new episodes and download them for you (on WiFi), hands-free if you choose that in the app settings.
If you have further questions about what a podcast is, which app is best for you or need more assistance with getting started with podcasts, don’t hesitate to contact us.
Drew Lyon: Hello. Welcome to the WSU Wheat Beat podcast. I’m your host, Drew Lyon, and I want to thank you for joining me as we explore the world of small grains production and research at Washington State University. In each episode, I speak with researchers from WSU and the USDA-ARS to provide you with insights into the latest research on wheat and barley production. If you enjoy the WSU Wheat Beat podcast, do us a favor and subscribe on iTunes or your favorite podcast app and leave us a review so others can find the show too.
[ Music ]
Drew Lyon: My guest today is Dr. Melissa LeTorneau. Melissa is a research soil scientist with the USDA-ARS Northwest Sustainable Agroecosystems Research Unit. She completed her Ph.D. in Soil Science at Washington State University in 2017 and holds additional degrees in Geology from Indiana University and Computer Science from Oregon State University. During her Ph.D., she studied bacterial biofilms, mineralogical transformations and bioavailability of nutrients in the root zone of wheat. In her postdoctoral research, she linked microbial communities with the severity of soilborne fungal disease. She started her current position in July 2021 and aims to integrate multiscale biophysical chemical and ecological data to enhance nutrient and water use efficiencies as well as crop yields throughout the Columbia Plateau. Hello, Melissa.
Dr. Melissa LeTorneau: Hey, Drew.
Drew Lyon: So could you describe for us a little bit your prior research and explain how it is set the stage for your current work?
Dr. Melissa LeTorneau: Sure thing. So I started a studying geology at Indiana University, and while I was there, I worked on some research projects looking at some field scale geophysical and hydrological processes. So after completing that degree, I took a little bit of time off, travel around, but really missed working in the natural sciences. And so I went back to study biology and organic chemistry at North Seattle Community College. And from there, I developed a really strong interest in plant physiology. And so I decided to pursue a Ph.D. in botany at Washington State University. And I started out there in Mechthild Tegeder’s lab working on nitrogen uptake and transport in Arabidopsis and a variety of other crop plants. But I really became more interested in those below ground processes. I felt that they were difficult to study, and maybe for that reason not quite as well studied as a lot of above ground processes. And so I ended up switching over to the soil science department at WSU where I worked with Jim Harsh and Linda Thomashow, and we also collaborated with some folks at Pacific Northwest National Lab and Argonne National Lab to study root associated biofilms. And so if you’re unfamiliar with the term biofilm, it’s a very dense bacterial colony. And often these colonies produce a kind of slime made up of biopolymers like carbohydrates and proteins and these things are very pervasive in the environment, especially in association with plant roots. Every blade of grass, every tree, every shrub. And yet this way of life is is kind of incomprehensible to us. It’s not the sort of thing that we would encounter in our day-to-day experience. And so studying these things is really very interesting. And I spent a lot of time actually looking up at them under some powerful microscopes. And that gives you a kind of very intimate feeling, you know, looking at this beautiful association between microbes and plant roots. But aside from just making some pretty pictures, we also studied the impacts of these biofilms on soil, mineralogy and nutrient bioavailability. And we felt we had this particularly interesting finding where one particular type of bacteria in producing biofilms in the in the root zone of wheat could weather enough iron to account for all of the iron that was taken up by the by, say, a four week old wheat seedling, which doesn’t sound like a lot, but actually that’s several orders of magnitude more than what the bacteria need themselves. So as far as environmental engineering feats go, this is at least equivalent to humans building skyscrapers. And this is taking place in the root zone of wheat. So that was very interesting work. But we were only looking at one type of bacteria and there are tens to hundreds of thousands of types of bacteria and also fungi associated with plant roots. And so for my postdoctoral research, I went on to study microbial communities overall and how they can impact take-all disease severity. And in that work, we found some very interesting patterns related to soil type and to wheat cultivar selection — selection by specific cultivars. So we’re still probing those data and there are a few different groups of bacteria that might be good bio-control candidates for future studies of bacteria that might be able to suppress disease severity a little bit. So that’s my prior research. So given this broad background, geology, plant physiology, soil microbiology, and mineralogy, and also a little bit of computer science that I studied during my post-doc, I decided to pursue this new position with the Northwest Sustainable Agroecosystems Research Unit as the USDA-ARS Research Unit, led by Dave Huggins And my role in this unit will be cropping system modeling. But, really, I think my role in this unit is going to be collaboration more than anything. Because cropping systems, if I’ve learned anything and I think most of us know this, they’re exceedingly complex. There’s probably not a single discipline that doesn’t doesn’t have some bearing on outcomes and in cropping systems and so and even in the fields that I’ve studied, you know, you study a lot of things, you learn a little about them all. So there are plenty of people with much greater depth of knowledge than I have in some of these things. And so I’m really looking forward to that aspect of this work. Just bringing in a lot of ideas and expertise and practical knowledge as well to understand our cropping systems better here in the Pacific Northwest.
Drew Lyon: Okay. Well, when you say you’re going to do cropping systems modeling… What is a cropping system model?
Dr. Melissa LeTorneau: Cropping systems models, these are basically a synthesis of data that’s been collected over time and space. And I mean, usually you want quite a bit of data, lots of different types of measurements. And the utility of these things is that you can make predictions maybe about about a particular outcome in the future, such as yield, what’s the yield look like going to look like if we have this amount of rainfall by this date, you know, things along those lines, but it doesn’t stop there. It’s very important to compare the predictions that you’ve made with with actual observations to see, you know, how accurate your predictions are. And it’s okay if they’re not very accurate because you can use that information to identify sources of error that represent knowledge gaps and that can lead to future research, right? You can you can develop hypotheses based on some of those sources of error, do more detailed research. But on the other hand, if you have highly accurate predictions, that’s really great because then you can anticipate and adapt to change. You can select in advance maybe at least a little bit in advance, you know, management practice or rotational crop is going to be good for this for this particular soil type one under this set of climatic conditions. So that’s the real utility of these things. And I just want to give briefly a few examples. I’m not going to go into too much detail because I didn’t develop them, and I actually haven’t worked with them much yet. But but there’s one developed by — there’s one widely popular model called , CropSyst and this was developed by Claudio Stockle, Roger Nelson, and Armen Kemanian. And the core of this model, it’s been widely used, there are a lot of different sub-models and modules that have been built to interface with this thing. But the core of the model is really a series of crop physiological elements that are impacted by nitrogen supply and water status and temperature. So, you know, a fairly basic set of parameters at the end of the day, but of course, the the way that these are developed and understood is much more complex than that, and then as I mentioned, there are other factors that can be interfaced with this like climate. So based on CropSyst, a former Ph.D. student of David Huggins and John Reganold, Harsimran Kaur, came in and developed a flex cropping model that is sort of able to do some yield predictions for rotational crops based on locations within the agroecological classes that we see in Washington State. And those agroecological classes she actually delineated based on land use using actual cropland data. But also based on potential evapotranspiration. And so this is this is something that drives a lot of the variability in the Columbia Plateau. And then the last example I wanted to give is another model based. It’s an extension again, of CropSyst, and this was developed by Brian Carlson and Lynn Carpenter-Boggs to look at the carbon footprint of organic farming systems. So these are just a handful of examples to give you an idea of what cropping system models are for and what the point is.
Drew Lyon: Okay. In a former life, I was a dryland cropping systems specialist in western Nebraska. And what I found is I could only afford to do an experiment for two or three years and that was such a small sample size of all the possible years I could have and I turned to crop models that try to help me figure out dud I just pick the two years that this was going to work well or is this something that work over a period of time? I never got very good at really understanding the model without a modeler helping me, but it was a it was a very useful tool to try and get that. We have lots of different like 2020 was an extremely high yielding what year. 2021 was one of the worst droughts in history and if you happen to run your two experiments in those two years you’d have totally different answers. But you might have good data for a model to predict how it do in between those two extremes. So it can be a useful tool, but it’s they can get rather complex, especially if they’re process models trying to figure it out. So what are some of the problems you hope to solve with models or what are the first things you’re going to try to model in your in your program?
Dr. Melissa LeTorneau: Definitely soil acidification That’s a growing problem in this region. A lot of producers are starting to see issues with aluminum toxicity, especially with the pulse crop rotations. They tend to be particularly susceptible to acidification, and the other issues that lime isn’t an affordable liming options aren’t readily available in this region. So if this is going to be a problem going forward, we want to maybe develop some tools that could be used for precision liming, for instance, and in case that does become necessary. But we also want to see if there are different management practices or cropping systems that can help to mitigate acidification without having to resort to liming. And I don’t know how practical that’s actually going to be, but but it’s certainly worth looking into and also just being able to measure, understand, and predict the extent of the acidification in our region is important. So linked to soil acidification and also of interest is carbon and nitrogen cycling. So we’re not going to be able to look at one without the other, actually. And I’m fortunate because we’ve also brought on a scientist in our unit, Claire Philips, who is a specialist in this area of soil carbon and nitrogen cycling and greenhouse gas emissions. So I’ll be working very closely with her, I think, on this on this problem. The other thing I’m interested in continuing to work on is crop disease. And so probably I’d like to continue my collaborations with Tim Paulitz and others in the USDA-ARS Wheat Health, Genetics and Quality Research Unit. So I worked with David Weller in this unit on my postdoctoral research, and I’d like to continue those collaborations. And Tim in particular has some ideas about looking at distribution of disease inoculum, the environmental conditions that are amenable to to seeing severe disease outcomes and also looking for more bio-control kinds of organisms that we might be able to use in the fungicides, you know, because we still can’t use methyl bromide anymore [ laughter ] so that I don’t know that anyone has come up with anything quite as effective as that but that’s also, you know, absolutely terrible for the soil biology so if we can if we can find alternatives, that would be nice. And then we also have a new remote sensing scientist in our unit. So I’m hoping to work with — that’s Joaquin Casanova — and I’m hoping to work with him on remote sensing methods to detect crop nutrient status, so maybe aluminum toxicity and maybe disease symptoms. It’ll be interesting to see what it takes to tease all of these different stressors apart, you know, to identify just based on spectral data or what is the stressor. I think that’s an interesting problem. And then the other the other thing we want to look at: new cropping systems. So we have an excellent cropping systems agronomist in our unit, Garett Heineck, who’s interested in doing some intercropping, in particular pea-ola, and maybe also looking at some perennial systems like Kernza. And so he’s been talking with a lot of folks about this, but there aren’t any I don’t think there are any cropping systems models that are covering these kind of new systems. So so we’re going to have to develop those to inform management decisions. And finally, I really want to work on this preexisting flex cropping model, in particular to see if we can improve the yield predictions. Maybe we can add some more rotational crops as time goes on, but working with some of these preexisting tools and just refining them as much as we can. So it’s kind of a long list, I guess, but — [ laughter ]
Drew Lyon: You’re not going to be short of things to work on, are you? [ laughter ] What kind of improvements are going to need to make to these models to do some of these things you’re talking about, like the soil acidification and the flex cropping type models?
Dr. Melissa LeTorneau: So there there are definitely a number of these. Number one on my list is actually site specificity. And I feel that this has sort of broad implications for modeling. Yeah, so we need to look in at specific sites in great detail, but we also need to look at sites with a real variety of growing conditions, and that’s really important. So luckily for me, our unit is part of the USDA-ARS Long Term Agroecosystem Research Network, which is a network of 18 sites that are set up for, you know, just collecting all kinds of measurements, you know, yield parameters but also soil data, crop physiology data, remote sensing data, all kinds of different data at these sites. And they’ve been doing it for quite some time. And so our site here is the Cook Agronomy Farm, but actually we’re hoping to extend the reach of our LTAR site to include the whole Columbia Plateau. And so we’re really hoping to collaborate with WSU Extension and look at data from their sites. We’re really excited that the WSU Soil Health Initiative is setting up their network of long term research sites, and we’re hoping to tap into that. And then Garett Heineck again is has been going around and cultivating some relationships so that we can do on-farm research as well. But the main thing is we’re hoping to have some long term research sites representing each of the agroecological classes in Washington so that we can do both some some very detailed modeling work, but also capture a wide range of conditions. And then the last piece on this near and dear to my heart, based on my past work, is really incorporating more root zone microbiology and chemistry. I feel that this tends to be overlooked a lot in favor of, you know, crop above-ground traits and also a field scale processes but the root zone is where all of the nutrient uptake happens. And so I suspect that these processes play a major role in nutrient use efficiencies and also acidification, actually, because the other thing to keep in mind is the root zone has a lot of carbon, a lot of energy going in that the bulk soil just doesn’t have. And so there’s a lot of options for chemical catalysis there by all these microorganisms.
Drew Lyon: Okay. So I think you hit on it earlier. Working under the soil is a lot more work [ laughter ] and maybe sometimes why it is and studied as much. So it’s a primary tool to continue to work in, I believe. What approaches are used to develop models from? So you collect all this data, how to use it to develop new models or to improve models you already have?
Dr. Melissa LeTorneau: Yeah. So there are there are two kind of families of approaches that are in use these days. One is sort of in the statistics camp And so in these cases, you have some hypothesis about how different factors are variables in the system are related to one another. And so you do what you can to take the data that you have to fit and try to fit that data to a function that you think expresses that hypothesized relationship. An example might be to say a probability distribution or maybe some kind of linear relationship. And you can use these kind of analysis to identify actual cause and effect relationships, which is handy. And the other the other nice thing about statistics is that you can differentiate between your actual factor effects and so-called random effects. And actually, random isn’t a real thing. What random really means is these are a bunch of effects that we haven’t accounted for. Maybe it’s something to do with the way that the data was collected. Or maybe it’s some variable that we just hadn’t thought about. You know, maybe we collected data on potassium, but we don’t have any data on magnesium. So that can be really important for making predictions because if you don’t know what the sources of errors and then those predictions might not work under under different scenario. So that’s important in terms of statistics. So the other group of approaches that are used are these machine learning approaches. So these are best for situations where you have a lot of different factors. Maybe there are a lot of complex relationships among these factors and maybe we have kind of limited understanding at this stage and, you know, perfect case in point are microbial communities. They have this incredible diversity. We haven’t been able to study every single microorganism out there so we don’t necessarily know what they do. So this is these kind of data, a great candidate, I think, for applying machine learning. So you can’t develop a clear hypothesis if you you know, if you if you don’t have a good understanding and if there’s too much complexity. Well, so what we do is we throw a massive data set and a supercomputer, you know, high performance computing clusters, and we have them execute these trial and error processes to try to identify patterns that are linked to a specific outcome. You know, maybe that’s a certain yield target or maybe that’s a certain precipitation pattern or maximum temperature. There could, you know, the the target could be any number of things. And what this allows us to do is oftentimes to make some very highly accurate predictions. But the drawbacks of these, you know, because you’re including a lot of factors and a lot of complex relationships, you’re not just fitting one simple model. But the drawback of these kind of techniques is, for one thing, you can be very difficult to identify a cause and effect relationships because these models are so complex. So you can’t necessarily go back in and actually understand what happened. You can still make your predictions and work with those, but but that doesn’t mean that’s going to necessarily give you a better overall understanding of the system. You know, you might be able to dig a few hypotheses out of there. It’s it’s a tricky business, though, and actually a pretty hot topic in the research in the machine learning research world. And the other side is your predictions might reflect some random effects because you haven’t been able to test the impact of your factors. Versus random effects like you have in a statistical model. So and that’s often referred to as overfitting. And so the problem with that is, again, if you take the predictions that you’ve made based on one dataset and you try to take that to a new system, it may not perform quite as well if there’s a lot of random effects captured in those predictions. And so in both cases, the important thing to keep in mind is you’re using an abstraction of your input data. And so the scope of your predictions, the accuracy of those predictions, is going to be limited by the scope of that input data. So for example, you can collect a broad range of input conditions with fairly low resolution, and maybe you can get reasonable accuracy over a large area. You know, maybe you can even get it so that you can address both the Corn Belt and the Columbia Plateau. [ laughter ] I don’t know that anyone’s actually accomplished that, but you’re not going to get very good precision at a single site. You know, the you know, for instance, if you wanted to capture the topographic variations at the Cook Agronomy Farm, you’re not going to be able to do that with a regional scale data set. You’re going to have to go into finer resolution, and then on the flip side, maybe you have some very detailed site specific conditions, like all of the linkages to topography on the Cook Agronomy Farm. So you can maybe make very precise predictions for that one site, but you’re not necessarily going to take that model and try to apply it to, say, the dryland research station at land, you know?
Drew Lyon: Right.
Dr. Melissa LeTorneau: So that’s the tradeoff and that’s why it’s so important to collect both regionally and site specific data. And that’s why I’m so excited about these long-term research networks and the on-farm — the producer collaborations that we have to do on-farm research. I think there’s a lot of potential in that network to work on kind of both directions.
Drew Lyon: I don’t think there’s going to be any shortage of things for you to work on in your new position here. [ Melissa laughs ] There are lots of interesting things, and yeah, you have those fundamental basic things you need to learn, but also these models should help make applied decisions for people, I would think. So very interesting. Thank you for taking some time to to visit with us about your past and what your future may look like.
Dr. Melissa LeTorneau: Yeah, thanks very much for having me. I’m really excited to be here and thank you for listening.
Drew Lyon: Thank you, Melissa.
[ Music ]
Drew Lyon: Thanks for joining us and listening to the WSU Wheat Beat podcast. If you like what you hear don’t forget to subscribe and leave a review on iTunes or your favorite podcast app. If you have questions or topics, you’d like to hear on future episodes please email me at drew.lyon — that’s firstname.lastname@example.org — (email@example.com). You can find us online at smallgrains.wsu.edu and on Facebook and Twitter @WSUSmallGrains. The WSU Wheat Beat podcast is a production of CAHNRS Communications and the College of Agricultural, Human and Natural Resource Sciences at Washington State University. I’m Drew Lyon, we’ll see you next time.
The views, thoughts, and opinions expressed by guests of this podcast are their own and does not imply Washington State University’s endorsement.