Midwest Agriculture Conference: Session 2

SPEAKER: Hello, everybody. We're ready to start up after break now. And we're going to have Bruce Erickson, who's a professor at Purdue University, talk about some crop inputs here. So take over, Bruce.
 
BRUCE ERICKSON: Talking and, of course, no one else is listening. The kids are paying no attention. My wife is ignoring me, so this is normal. OK? So thank you for the invitation. Happy to be here.

So what I want to talk about is digital agriculture and how it may affect crop inputs. And so I'm an agronomist, not an economist, but I've really enjoyed my time spent with economists. And I hang out with a lot of them at Purdue and other places. And when I'm in the Krannert building at Purdue, I feel special, because I'm not an economist.

But I go back to Lilly Hall and my own department, and I'm just run of the mill there. But if you take a look at my background-- I'm not going to dwell on this-- but I have a diverse-- oh, we had fire alarms before, and now we have construction. OK? So anyway, you can see a lot of diverse activity.

I've done my own consulting, started with Pioneer Hi-Bred there in Iowa, Iowa farm kid. And currently, I'm directing an academy of online courses at Purdue that's been very successful. And so you can see quite a mix there. And it provides me with a lot of background.

But I didn't start high tech. I just want to show here that it was pretty low tech on the farm, although you can see dad there on a hand-crank tractor with two row, front mount cultivator. And the only reason I show this, really, is I show it to students at Purdue to think if is the progression of technology in my lifetime, and I'm there in the diapers, just think of what it's going to be toward the end of your careers as students. And so we won't dwell on that a little bit anymore.

So a little context before I get into some more meaningful comments. So I'm pleased to be associated with this survey. I've been doing it for the last 10 years, but it's been running more than two decades. And so we track adoption of precision farming technologies. It's a retail survey.

What are the retailers using? What products and services are they offering to customers? What are their customers doing? All this kind of stuff. And it's always the cover story in CropLife Magazine in the summer.

Maybe you've seen it. And so I'll talk some about that as we go on. And so just for some additional context, what we're talking about with precision technologies are guidance, VRT, imagery, automation, all that kind of stuff that we've been tracking.

We also took this information and we spent many, many hours taking a look at other surveys on digital adoption around the world. And so we published in Agronomy Journal, and it's a fairly high impact journal. And we were pretty proud to say that we had the most downloads of any journal in 2021. And so if you take a look around the world, there's big differences in what farmers are doing in terms of technology adoption.

Big grain farms around the world typically are doing quite a bit. Little farms and specialty farms, more of the horticultural and vegetables and grapes and all that kind of stuff, are doing much less. And so this is open access if you want to take a look at that.
 
It's a little bit out of date, but we actually did an update of that and presented it at the Precision Ag conference this past summer in Kansas and worked with the RS on this. And so we looked at Brazil, South America, all parts of the world. Hungary, Denmark has some really good statistics. But it's very hard to get precision farming adoption statistics, anything that's a randomized study that has good statistics. A lot of studies are done that aren't very good.

In fact, one of the studies, I had to giggle a little bit. They were doing a precision farming adoption survey at a precision farming conference with farmers there. Well, what do you think is going to be skewed at that precision farming conference? You're going to have a lot more, obviously, in that sort of a scenario here.

So part of the reason that we talk about this is because we've had waves of technology mechanization, hybrid corn, pesticides, improved fertilizers. Digital agriculture is viewed as the future of innovation wave. Of course, all those others are still innovations occurring there, too. But we have great promise in what's going to be happening in terms of automation, AI, variable rate technology, all that kind of stuff.

And as this graphic shows here, in the past, we've been able to increase production by farming more land or pouring on more fertilizer and that kind of stuff, as is shown by the tan colored bars there. The green bars there show that in the future, we'll need to increase production more by doing a better job of things. And how do we do that? We use sensors to monitor things, we analyze the data, and we try to be more precise and more intelligent in the things that we're doing.

And a little dangerous here-- this is an economic concept here-- for an agronomist to be talking about this. But we know that technology adoption doesn't happen in a vacuum. When someone does something, there is an effect that happens with other people. And so these are inflation adjusted wheat, soybean, and corn prices. And as technology increases yields and reduces costs, we know that that affects the market.

And it's like when hybrid corn was introduced, if you weren't growing hybrid corn and your neighbors were, you were put at a competitive disadvantage. And so same thing with digital technologies. And one of our greatest worries is probably that, yes, big grain and oilseed farmers around the world are adopting a lot of this technology. But little farmers and oftentimes disadvantaged farmers are not adopting that. And so the haves seem to be growing in terms of their influence.

And the have-nots seem to be falling further behind. And there's a lot of need to take a look at digital agriculture. And the smallholder farmers all over the world really aren't doing anything. And I know this is probably more about American agriculture, but we also concern ourselves with the people around the world and how they're doing also.

OK. So the big picture, faster adoption of digital agriculture. And I'll get into more details as we go on. And like I've said, large grain and oilseed farms. There's a lot of GPS guidance and section controllers on those particular farms. Quite a bit of precision soil sampling, but we're not always continuing that on to variable rate technology, which is really interesting.

And I'll get into some of the reasons for that. And then quite a bit of yield monitoring. But in general, I'm sure you know that we aren't using those yield maps very much to make decisions. And we need to be doing that more on our farms.
 
OK? Slower adoption on the small farms, like I said before, on the specialty farms. Interestingly, we've had an ability to do remote sensing. Since about the 1960s, the satellites that went up, we've had aerial imagery there. Lately, we have a lot of use of drones for imagery here, but we're not doing very much with imagery on farms. And so we need to find uses for that.

It's surprisingly small. It's in the single digits or teens. And I don't know if I have a graphic on that. And also, what's very interesting, too, is variable rate technology is not being used as much as one would think.

Here's some ERS information that spells out a little bit of specifics in terms of this. And so this is the percent of acres in the US with autoguidance. And each of those dots is a crop, because ERS doesn't sample every crop every year. So I can't draw just a nice line across there. The yellow is corn.

The blue circles there are soybeans. I think the, yeah, the diamond, brown diamonds are wheat. But you can see in general there's been a relatively high adoption of guidance. This is a relatively simple technology that you put on your tractor or sprayer or combine, and it guides your implement through the field.

I just bought a brand new 2025 Toyota Camry, and I've got the lane guidance on it now. And I am in love with that technology. I even use it if I'm going just two miles down the road. It's a lot of fun.

OK. So contrast the guidance here with another precision farming technology, variable rate technology. And take a look at the left bar there in terms of the percentages, the x-- or excuse me, the y-axis. The last one was 80% was the highest. This is 40%. And again, yellow for corn, blue for soybeans, the diamonds for wheat, et cetera.

You can see that as opposed to the guidance one, which I'll flip back there, and now take a look at the variable rate technology, most farmers in the US, after two decades of having an ability to do variable rate technology, are still not doing variable rate technology. They're still putting the same amount of seeds, the same amount of fertilizers, the same amount of pesticides on every acre. Most farmers. Again, there's a minority that are. But I'm going to get into some of the reasons why this potentially is.

OK. In terms of inputs, we know that guidance has helped us with the efficiency. Most precision farming now has been more about saving costs than increasing yields. And back 20 years ago, when I left my career job with Pioneer Hi-Bred and went to Purdue because I was so excited about the future of technology, we thought maybe yields would increase too. But they haven't so much. It's been more about cutting costs.

And so with a full width tillage implement like you're seeing there being pulled on the right, instead of guessing where you're going to come together there and instead of rubbernecking back the whole day, you've got a guidance system that will direct you. And you're saving fuel. You're saving time. You can operate in foggy conditions, dust, all that kind of stuff. And it's helpful.

Also, all the spin-off technologies. Instead of over spraying into the end rows, as you all well know, and with planter controls and section controllers on sprayers, that's where the efficiency can come in. So where we seem to be somewhat stuck in digital agriculture is this area of understanding the variation in fields, characterizing it, analyzing it, and then deciding what to do and come back. So this is the farm just to the north of my home farm in Iowa, in Story County, not far from Iowa State University. And you could look on Google Maps and you can look at all kinds of places.
 
My son and I-- I've noticed my son does the same thing. He's inherited this maps gene or whatever for me, but we really like to spend a lot of time on Google Earth or Google Maps and take a look at fields and forests and everything from above. But take a look just in terms of soil color that you see on this particular piece of land. Wouldn't it make sense that the farmer that's farming this, who I know very well, would do something different on those light areas of the field versus the dark areas of the field? And this is just one example of things that vary on a farm.

We could lower costs. We could increase the yields potentially. We can reduce risks. You know as economists all of those things where you can increase your net revenues. But we've been struggling with how to put together the story of fields and to understand the cause and effect in many cases.

Like I said before, yield maps. We have good stories of how things end up. Every year, we put together a mix of hybrids and pests and fertilizers and all that stuff. And at the end, we have a yield map that tells us how every part of the field did. But we've been struggling to understand the yield maps and how to use those.

OK. Back to variable rate adoption. Most so far from ourCropLife survey that we have been doing has been related to fertilizers, not so much with seeds. And so the red line there is VRT fertilizer applications. This is the dealers on that survey telling us what their farmer customers were doing, the percent of the acres in their areas over the course of the time of the CropLife survey there. The black line is variable down pressure on the planter.

I won't even get into that today. The green line going across is the percent of acres-- this is somewhat Midwest biased-- the farmers that are doing variable rate seeding. And you can see it's around 20% are doing that. And the purple line going across is the percentage of acres that are doing variable rate pesticide applications. So it seems to be that pesticide applications are probably the best opportunity for us to save on costs in terms of inputs.

So here's the meat of my presentation. Let's talk about fertilizers, seeds, and pesticides and look at variable rate technology and where the opportunities might be. OK? In general, for P, K, and lime, it's pretty much like you would suspect. The plants need their nutrition. And it's going to be pretty hard to skimp on that over the course of time.
For nitrogen, it's a lot more complicated than phosphorus, potassium, and soil pH. I'm sure you all know that type of situation. For variable rate seeding, you need plants out there, corn plants, soybean plants, wheat plants, whatever in order to produce the yield. And so cutting back typically doesn't provide much opportunity. It seems to be the biggest opportunity is with pesticides, because that seems to be an area where we're putting on pesticides on parts of fields that maybe don't need pesticides on a year-to-year basis.

OK. So a little bit more detail in terms of fertilizers. So if you take a look at the amounts of phosphorus and potassium and liming that we put on fields, yes, there are some opportunities in fields where the soil tests are higher, where we can cut back some, where the soil tests are lower, where we need to put on a little bit more. But in general, we're going to need to continue to put on those basic amounts of those fertilizers. The possible black swan in this is that there's a bunch of new start-up companies that are taking a look at nutrients in the soil and trying to figure out is there a way that we can unlock some of that.
 
If you take a look at the amount of nitrogen, phosphorus, and potassium in soils, there's way, way more in those soils that's available to plants that they can't access. And I think I have an example here. Total soil phosphorus from this one Extension publication says it's around 800 pounds per acre.

But in the soil solution where the plants get it from, that little bit of water around each of those soil particles, the concentrations are very, very low. So if there could be some biological product or something that would unlock that nutrient in the soil, whether it be phosphorus, potassium, sulfur, whatever, this is a potential black swan that could sail into our nutrient management scheme. But so far, what I've seen, there haven't been revolutionary things happen in this area.

So I know the person that's the agronomist for A&L Great Lakes Lab in Fort Wayne, and they are due I think he said around a million soil test samples per year there. And they cover large parts of Michigan, Illinois, Indiana, Ohio. And so they've compiled all of their soil test results in that graphic in the left there.

And I'm showing just phosphorus. But you can go to their website. It's publicly available. And you can see the trend in soil tests for phosphorus, for potassium, for soil carbon, and other things that they test for. And in general, most things are gradually going down.

And so in this particular graphic, it looks like about 20% of the soil tests are in the low category, just 20%. And you can see that in general, they have been going from an average of around 50 parts per million of a Bray-1 test down to around 40 parts per million. OK. So my point here is that above 40 parts per million is typically where our university recommendations would say don't need any additional phosphorus fertilization. OK?

So we're right on the cusp of that soil test level that says that on average that you don't need to put on any fertilizer, which just really makes sense. That means we've probably been managing our soil nutrients very well, if that's what the average is, is to be right on that cusp. And you'll notice here that this is the Bray-1 test. Maybe getting a little too agronomic on you here. But most of these are adjusted for Bray-1.

Most of our phosphorus tests now are what are called Mehlich. It's a different chemical extractant to pull off the phosphorus from the soil particles. And typically, the Mehlich readings are a little higher than the Bray-1. But you can see the general-- the tests there in general

OK. So probably not a lot of availability to cut phosphorus or potassium nutrient inputs, because in general, we are right on that cusp of having enough nutrients anyway. And I threw in a couple of things related to fertilizer here, too, that I thought were interesting, in that with agro dealers, their fertilizer sales and that enterprise, part of their agronomy sales, is typically their most profitable. We ask them every year, what are your most profitable enterprises? Is it yield map analysis? Is it your fertilizer?

Is it your soil sampling? All that kind of stuff. And typically, as you can see over the years there, the red line, VRT fertilizer applications, the gray line there, grid zone soil sampling, those are consistently profitable for most dealers, whereas their satellite and aerial imagery and yield monitor analysis, most dealers say that the-- or a minority say that they are profitable. So I find that interesting. Over the course of time, that seem to be consistent.
 
What about decisions guided by data? Again, fertilizers dominate, with the red line P and K decisions, the yellow line there liming decisions. Where are they using data to make decisions here? And it tends to be on the things that are fertilizer related, not so much on those things that are seed related or pesticide related. Just to show you the big picture.
And all this, we have the last 20 or so years of this precisionCropLife survey posted on a website at Purdue. And these are 30 to 40 page reports where we comment on things. And so you can go in and take a look at all these things.

So that was fertilizers. Let's take a look at seeds. How profitable or how advantageous would it be to use variable rate technology for seeds? Fairly small. And that's maybe why only a third of acres are using variable rate technology.

That's putting a little higher population here when you seed a little lower over here based on yield, previous yield, all that kind of stuff. This graphic from Pioneer shows that at higher yield environments, you can see the yield scale on the right there versus the optimum seeding rate. You can see that higher yield environments typically have a higher optimum seeding rate. And so if you have a field that has very high yielding areas and also very low yielding areas, it might pay to do variable rate seeding.

But again, the differences there aren't distinct, because most fields, most of that yield variation is going to be from field to field and not within field, if you know what I mean. And so doing a variable rate approach would have marginal impact. Where the biggest impact probably could be-- and Brian mentioned the see and spray. I think that was Brian, wasn't it, that mentioned that earlier. And we know that a lot of pesticide applications-- I know-- I remember when my wife and I first got married, we lived on a rented farmhouse in Iowa.

And those farmers in that area were just crazy about making sure their weeds were-- their fields were perfectly clean of weeds, even if it really didn't matter. And so some of this is cosmetic. We all know. The thing with pesticides is that even though a lot of pesticides may not be needed, it's that it's difficult to characterize in a field that is mostly weed free but there's a weed patch here and a weed patch here, in a field that's mainly insect free but there's an insect problem over here, but not over here.

It's very difficult to characterize those types of things and to make a decision as to what you should do. And so an automated thing like see and spray is probably one of the best things that we can do. Although the weed scientists at Purdue say that don't do just see and spray. You need a soil applied pre or type of herbicide before you attempt this. If you do only see and spray, you're probably going to have a weedy mess or more escapes than you're really going to need. OK?

So OK. A few things to end up with here. And how am I doing on time? What time do I need to finish? OK. I should be able to do it.

We asked the dealers on ourCropLife survey what percent of you are currently using a drone to put on crop inputs, what percent of you are currently doing a see and spray type thing in the red line. And you can see that about 11% of them were doing the see and spray currently, but about 25% of them were planning to do that in three years. In terms of crop inputs applied with the drone, the top line there, about a third of them are currently doing it, and about half of them said that they would be doing it in three years. So we always find that interesting. And the robotics, you can take a look at that and see where that's going.
 
There's quite a bit of work with robotic, on-the-ground type sprayers, too. And we have a company in Indiana that's working on this. And the Keystone Cooperative tells me that they spend enormous amounts of money with their ground spraying rigs, and so they're looking for ways to cut costs. And so here's one possible thing, as a robot that's out there can go 24 hours a day other than fill-ups, and you don't have to have a person on it. And those are some ways that you could potentially cut costs.

Just for fun, we asked the retailers, if you're doing drone spraying, what are some of the costs and numbers associated with this? And I really don't have time now to go into this, but about a quarter of them are offering these services in-house. Another quarter are contracting with a company.

But in general, if they're doing drone spraying services, they're using about two people per crew. They typically have about two drones per crew, and they're spending about $62,000 on an average to set up one of these crews. So I'd never seen this type of information before. I hope you'll find it useful.

We also asked those same retailers, how is AI going to affect your business? How is automation going to affect your business? And this question, I think, was somewhat problematic because does the typical dealer know really what AI is? I'm not sure if I really know what AI is. I've really struggled with that.

And so the general thing, and you can take a look at theCropLife on your own time there, but in general, I'll just say that they think that it's going to improve the accuracy of their work, but they don't think it's going to save them any labor. And that's always the danger with things like that, is that we always worry that, oh, is my job going to be replaced by a robot sort of a thing. And they don't think so. So kind of interesting.

If you take a look at the last 10 years of ourCropLife survey, the biggest issues that the dealer were saying that were customer related, as you can see, the first two bullet points there, were economic related. Farmers don't have enough money to invest. I don't see the value in doing this type of thing was-- these were always rated the highest. In terms of the dealers and their enterprise, the biggest thing is we can't find employees to do this type of thing, which Brian, I think, mentioned also.

It's really hard to find students that are-- it's hard to find employees. And at Purdue, we're trying to produce the students that will be able to go out to these ag retail locations and to do these types of things. But there certainly is a definite deficit in terms of their digital tech understandings.

So a couple of things to end up with. If you're interested in a basic agronomy course that's online or precision ag course, I spend most of my time with the elearning academy here. And there's an example of this. And we've had over 3,000 course completions. We've had lots of people from John Deere and AGCO and Bayer and Beck's Hybrids. And all kinds of corporate people have sent their employees to us.

And then also, we're really ramping up our digital offerings at Purdue. We're launching this month in about a week from now our new Institute there, where we've always had site specific management center, a digital ag center, all those types of things. But we're putting increased emphasis on it. We've been working on our website. And so take a look at that sometime if you want more information on what we're doing there at Purdue.

So I need to acknowledge that part of my time comes from a grant that Purdue got from an NSF funded institute on artificial intelligence called AI-Climate. So I get a sub grant from the University of Minnesota to do some of this work. So questions, comments, critiques, criticisms? OK? I'd love to hear it all.
 
And I had an interesting discussion with a good friend of mine. And some of your best friends you probably are those that take the time to criticize you. And I had a good friend that did just that here last week. And it hurt a little bit, but I sure needed it. And I'm very thankful that he took the time to do it. So OK. Yes.

AUDIENCE: So I'm curious on your VRT for nutrients and fertilizer. You talked about how farmers don't like it as much because they still need a base level. Is there anything that can give an instantaneous reading on your pH and nutrients that the farmer can actually see in the field what he needs to add at that point in time? Because that's how he's going to see real life cost benefit, kind of like with fertilizer, how it makes your plants greener.
 
BRUCE ERICKSON: So you're talking like a-- instead of a see and spray, you're talking like a see the nutrient and do something about it in the moment. Correct?
 
AUDIENCE: Yes.

BRUCE ERICKSON: And so actually, the green secret technology was invented at Oklahoma State back over a decade ago. And we thought that would probably revolutionize. And if you're not familiar with it, it's optical sensors that go through the field and they take a look at the plants. And you know that nitrogen deficiency is very visually apparent. If it's nitrogen rich, the leaves are green. If it's nitrogen deficient, they're yellow.

And so why not send this sensor through the field and then make an instant decision on that and do just what you're talking about? And so unfortunately, that technology has mostly been a flop. And not because it didn't work. It's because that farmers just haven't used it. And I think one of the main reasons is is that by the time your crop gets up to that point in time and is showing the yellowing, you've already been hurt a lot.

And so it's good to try to fix that ahead of time. And then the other thing that my colleagues at Oklahoma State told me, too, is that in Oklahoma, there's a law that when you sell fertilizer, you have to bill it when you're going to the farm with that fertilizer on your rig. And so they can't have the opportunity to cut back on fertilizer rates, because the farmer has to be billed, by some archaic law, as to the amount of fertilizer that's going to that field. And they can't bring back any.

I don't know that I fully understand it. But the thing that we've always wanted, so soil testing for phosphorus and potassium and soil pH now, is done the way it's been done for years. And that-- well, I mean, it's typically done with an all terrain vehicle. And we're doing some grid and zone sampling. But you send a person out on an ATV.

They get off of the ATV. They walk a circle around that and collect several sub samples. But the economic problem with this is that it has to be sent to a lab analyzed, and then there's a time delay. We don't have any on- the-go type of way to do phosphorus, and potassium. Now, there was an on-the-go soil pH sensor that's being used by various technologies.

And Kansas has developed this based on the work that was done at Purdue. And it does offer some availability to do that. But again, it's not being used to much, too. So yes, we need some on-the-go type. I think that would revolutionize production agriculture more than about anything if we had some instantaneous, on-the-go type of soil test system.
But we just don't have it, because it's wet chemistry. But I mean, we're making progress all the time. I mean, I have high blood sugar. And so, I mean, I should get one of those glucose monitors to wear. Those are new.
 
And so that's an on-the-go sort of a thing instead of the finger prick. So I think it's probably coming sometime in the future, but you raise a very good point. One more question here.

AUDIENCE: To what extent are soil genomics being considered in the overall variable rate tech technology?

BRUCE ERICKSON: You mean the microbial populations?
 
AUDIENCE: Exactly. Understanding the microbial ability to promote uptake of phosphorus.

BRUCE ERICKSON: Well, that's one of those things that I said could be the black swan. That would be. And so there's a company. I'm actually met with them at the Tech Hub LIVE conference in Des Moines this past summer, where I was a speaker there. And a friend of mine from college works for this company. And you send your soil into a lab, and they do this whole analysis of what microbes are there and what's their capability of doing all kinds of various things.

And the only thing I can say is that's a possibility and it's being done. But I don't know if it's being done on a lot of acres. But I think that's the direction that we're going. Because you know, like with nitrogen, you can lose-- if you have denitrification, you can lose a huge percentage of your nitrogen in just a couple of days. You can have leaching.
You can have volatilization. If you've seen that nitrogen cycle, it's got holes in it everywhere. Phosphorus and potassium, not so much. But it's a complicated thing. I think there's maybe one more question. OK.
 
AUDIENCE: Hi. I was wondering if you were seeing any application of digitization and automation in calculating carbon intensity, which I understand will be necessary for upcoming tax credits.
 
BRUCE ERICKSON: Yeah. And what was your question regarding the carbonization?
 
AUDIENCE: So if people are using digital applications or automated applications in calculating one's own carbon intensity, for example, for biofuel feedstocks.
 
BRUCE ERICKSON: Yeah. So I'm on another project, the RCPP, if anyone is familiar with that, where we're working with Cardinal Ethanol in Winchester, Indiana. And what we're trying to do is to show that you can produce corn that goes into ethanol sustainably, where you can potentially increase the amount of carbon in the soil by using cover crops and reducing tillage and all that kind of stuff. And so we're really intent on making sure that we do the appropriate sampling with this. And those measurements are pretty reliable, like carbon. Carbon is pretty reliable.

Where our big challenge is is that how do we transform those farming management practices and how do we keep the carbon levels where they need to be. And I think everyone-- you saw the A&L lab where the phosphorus levels were gradually going down over time. In the Eastern Corn Belt, the soil carbon levels are ever so gradually going down over time. And so we certainly need to reverse that trend, because we need to be putting more carbon in the soil, not less.
 
But I hear the soil testing lab in Omaha, which is a sister lab to that one in Fort Wayne, if you take a look at soil carbon in Nebraska, northwest Iowa, southwest Minnesota and South Dakota, their soil carbon levels are very gradually increasing out to the west. And so that's really encouraging to see that type of approach. So OK, I think I've worn out my welcome.

DAVID: Thanks, and now we're going to have Victor Cabrera from the University of Wisconsin.
 
VICTOR CABRERA: Can you hear me OK? Great. Well, I'm very glad to be here. Thanks, David, for the invitation and for the opportunity to share what we are doing in Wisconsin. And we're going to talk about dairy farming and how we can enhance economic efficiency and control input costs basically using digital dairy farming and precision livestock agriculture in general.

My name is Victor Cabrera. I'm with the Department of Animal and Dairy Science in Wisconsin. And my work is in extension and research, and what we have tried to do over the years is help dairy producers to improve efficiency in general and profitability and sustainability of the dairy farm systems.

So I think in order to talk with this topic, it's important to identify the main income sources and cost sources in dairy farms. And as you can see here, milk sales and cattle sales are the most important ones, but if you see there by the difference, milk sales are more than 90% normally in a dairy farm of the income source. So that's important to keep in mind.

And if you see the second one there, it's the cattle sales, and basically it's the replacement animals that are being sold for the slaughter houses. And if we look, similarly, to the production costs, the most important, for sure, is feed. But here-- and that's why I ask Kate early on in the morning-- we are considering only the feed for the adult animals. And in the third item there, Raising Replacements, is very important as well after labor.

There are opportunities to also improve labor. But I'm going to talk today about how we can improve or decrease feed costs and how we can make more efficient or more income or cost decrease in the replacement arena, in the raising of the replacements on the herd. And this raising replacements will include all the costs, and in those costs will be also include the feed costs for those young stocks.

So the opportunities of improvement I want to bring today to consideration and I think are two things that are happening as we talk in the dairy industry is to leverage what I mentioned in the previous slides-- feed costs, milk sales, cattle sales, and raising replacements.

And we're going to talk about, I think, one thing that's important, and we can help producers to improve the nutritional efficiency, the nutritional accuracy by a simple management strategy they can do of nutritional grouping. I'm going to explain that as we move on.

And the second thing I'm going to discuss is about the use of crossbred beef calves, the production of crossbred calves. So we have the opportunity nowadays to breed the dairy animals to beef semen, beef meat production animals. So we have a crossbred animal that will be destined to the beef industry, and that's important because the price of beef is increasing, has been increasing, and remains high in the US and, actually, across the world.

But in order to do that, I think it's important to keep in mind that we're talking about digital agriculture precision livestock, so data is critical. And I'm glad to be part and actually leading this project we call the Dairy Brain. And the main goal of the Dairy Brain is to add value to the data dairy farms already have in their farms.
 
So it's enhancing their decision making and providing better insights out of the data they already have. What happens in the dairy industry-- and I think that happens also in the crop area but more in the dairy and livestock industry-- is we have silos of data, that they don't talk to each other.

We have the milking parlor data that doesn't connect with the feed management software, and it doesn't connect with the management software and doesn't connect with the health or the genetic information out of the animals. So the Dairy Brain is trying to connect in real time this data as it's being generated.

And nowadays we have a lot of data from sensors as well, like activity sensors, rumination sensors. And all this should and could be connected in real time, and we are trying to use as much as we can the new developments in AI to either connect the data and also provide better insights from the data.

OK, so data integration is critical. We have a platform to integrate data. We are doing this nowadays successfully. We have a number of papers that describe this process. It may not be the same in every single dairy farm, but the main idea is try to overcome the challenges of accessing the data from different systems, decoding the data, cleaning the data-- dairy farms have a very messy and noisy data-- then homogenizing the data and, at the end, integrating. So no matter from where is the source the data is coming, it should be readily available for decision making in similar fashion, regardless of the farm.

So based on that and keeping that in mind, let's talk about nutritional grouping. And the main idea here is waste less nutrients on dairy farms. What we do in dairy farms basically is what you see here. We did a survey a few years back, and 58% of the farms feed only one diet to all the adult milking cows.

So what that implies or what that brings up is the problem that we are giving either more or less nutrients to a large number of animals. Every time we feed more than one cow, obviously one cow will receive more or less than what she needs. That's logic.

So what we try to do in this idea of nutritional grouping is bring the cows together that are more homogeneous in their requirements and provide a diet that's more closer to their requirements. If we have 58% of the farms that feed the same diet to all the cows, normally what happens is most of the nutritionists are going to try to give a diet that is for the top animals.

So we assure those top animals, top-producing animals will keep producing high. But that means that a large proportion, 83% or more-- that's kind of normal-- of the animals will be overfed. And so that implies not only that those cows are wasting some nutrients when they are fed, but they will also excrete more nutrients, obviously, but also implies some problems for health of the animals because they will become overconditioned with more nutrients than what they need.

So that's what I was saying. Normally, farmers will prefer-- or nutritionists working with dairy farmers-- highest- performing cows for formulating the diet. So there is a lot of overfeeding and unnecessary feed input costs on farms. So if we do one TMR-- that's called one total mixed ration-- you can see in the figure there a representation of what happens in reality.

You can see the very thin line there is when we provide only one diet, which is the normal thing, as I mentioned before, and we compare that with two diets, two TMRs, or three diets, three TMRs, in green. And you can see clearly after days postpartum, which is the normal cycle of the animals, we will be feeding much closer to the requirements if we give three diets compared to one diet, for example.
 
That implies a little more management, obviously, but I think there is a lot of value, economic value for farmers and a lot of saving costs in feeds by doing so. And I'm going to show an example of that. We did a research a few years back-- this is published-- and we show that the gains could be substantial economically. You can see here three or four diets will imply about $46, $47 per cow per year of extra net return.

And in this case, most of that net return was because of saving in costs of feeds. It will also imply a little improved productivity, and actually, over time, we will expect that the productivity will be even much higher, which, at the moment, we don't have ways to measure.

I mentioned this. Also, when we feed more closely to the requirements, the body condition score, which is a good way to measure the health of the animals on a herd, will be more logically distributed like a normal distribution, like you see there in three TMRs, three diets, compared to one diet, for example.

And this will also-- if we give three diets versus one or four diets versus one-- and you can see here the example in three different or five different herds in Wisconsin-- we can demonstrate here that our feed efficiency related to the nitrogen consumed versus the nitrogen produced in the milk increases. The percentage is very small, but every percentage point counts because this will imply that less nitrogen is being released in the manure at the end.

OK, so now we can do precision nutrition. Since we have all this data connected, we can automatize the system and facilitate the implementation because we think one of the reasons farmers don't implement this-- because it requires management. It requires to recalculate the diet, the nutrient requirements of the animals. As they move through their lactation curve, it changes. It is very dynamic. So if we provide and facilitate this system, we can decrease the errors and foster these more accurate diets.

So it needs to be farm-specific. We can connect the data of different systems in a farm because one farm will have one different milking parlor, a brand. The brand will have a different software, so extraction of data is different than other systems. And we have published this as well, so there is plenty of information about that.

And I just want to give you this example that I think will make the case very clear of what we are trying to do here. This is one farm in Wisconsin. It's a large farm, almost 3,000 cows in total. But we're going to talk about 960 cows, and these 960 cows happen to be in the peak lactation. They are in the middle of the lactation.

And half of them are multiparous, so second and later lactation, and half of them are primiparous, in the first lactation. And there are about 480 cows multiparous and 480 primiparous. And what happens in this specific farm is they are split in three groups, each one of these primiparous or multiparous, and each one of the groups has 160 cows.
What happens is in this farm, because they receive one diet, the multiparous, and another diet, the primiparous, they don't care which animal goes to each one of these pens. And the diet is being given the same for all these animals. So our proposal is very simple, actually. Actually, this is what's happening.

Before I show you what we are proposing, if we actually-- now, with all the data being collected in real time, we can calculate specifically every animal, how much energy it requires and how much protein, in this case metabolizable protein, it requires. And we can calculate all the fiber and all the nutrients that every cow requires.
 
But in this case, the example is energy on the x-axis and metabolizable protein on the y-axis. And you can see every dot in different colors there is a cow, and here we are talking about only the peak multiparous cows. So there are 480 cows there. At the moment, they receive only one diet, and that diet is normally formulated for the 83 percentile cow.

And you can see the solid dot in the figure there. That's the diet they are receiving, these 480 cows, at the moment. Our proposal is very simple. We just split in three groups. We use clusters to group the animals that are closer together in the requirements of energy and protein, and we give a different diet for each one of these groups.

And even though one of the diets, probably the red one, will have a higher level of energy and protein, the other two diets will be much lower, and still the cows will be meeting the requirements in the same fashion as before. But we will be saving a lot of nutrients by doing so.

And this represents, at the end in dollar value, an overall gain of more than $200 per cow per year. And if you multiply this for thousands of cows, in this case around 1,000 cows, so it becomes a large amount of money that is very encouraging for the dairy farmer to adopt this. And obviously, this will include all the additional costs that could imply by doing this grouping, nutritional grouping. Basically, these cases move a little bit different the cows to these pens in the farm and tweak a little bit the diets to have the nutrients more accordingly to the cows that end up in each one of those groups.

But also, doing this, we will save nutrients, and in this case, if we look only to the nitrogen, we will save about 75 kilos less nitrogen being emitted for each cow in a year. The same would be with phosphorus. We didn't calculate in this research greenhouse gas emissions, but they will decrease, obviously, by doing this.

So brief highlights about this, nutritional grouping-- I think it is an important strategy. We are trying to promote more farmers to do that. It will increase feed efficiency. It will decrease feed costs, so the inputs will be decreased. Eventually, in the long term, it will increase productivity, and it will increase and enhance the cows' health.

The interesting thing-- and if anyone is interested, go to More Details. We have plenty of information, including papers and, actually, tools. Everything we do, we try to put decision support tools available online that are openly for anyone who wants to use and test what's going to happen in their different systems. And the second thing I wanted to discuss with you, changing gears a little bit here, is about the management and the strategy of producing crossbred beef calves.

So the dairy situation at the moment-- this has been changing a little bit in the very last two years, I guess, but one thing that's clear, the reproduction or reproduction efficiency in dairy farms has improved. So that means we have more replacements. We produce more replacements. That has become a good problem, actually, in the last 10 years. Reproductive physiologies have been very effective on having cows more fertile and producing more replacements.

And therefore, that gave an opportunity to do options, to either select more the animals or use a different semen, beef semen, and produce a different kind of product there. But also, we have a new technology 15 years ago, sex semen. It's very adopted now in the industry. So we can produce-- on those animals we like and we select, we can produce with 90% chance or more female calves. That's what we like as replacements.
 
And on the right, what you can see in the graph there, actually, is just a representation of the use of different semen in the dairy industry in the US. And you can see there in red, for example, is the beef semen increasing in different levels of service numbers-- service are the breedings that we do and different lactations-- and you can see how it has been increasing by years. And also, you can see it is increasing the use of sex semen.

So the conventional semen is shrinking in its use, and the sex semen and beef semen are increasing. The other thing important of using sex semen and beef semen together that normally go together is that we can enhance genetic improvement because our intergenerational interval decreases, and our selection intensity increases.

So there are a number of good benefits of this idea. It brings more money, obviously, a new income source for the industry-- and there are attractive beef prices, and it seems that they're going to continue being attractive in the foreseeable future-- and, as I mentioned before, genetic improvement.

Again, you can see on the graph there from the USDA data showing how the dairy conventional semen is decreasing in amount and the sex semen and beef semen are increasing in their use. And this trend seems to continue. I know a number of farms in Wisconsin and in other parts that are using exclusively sex semen and beef semen and not conventional semen anymore. So that seems to be the trend.

So based on those ideas, we kind of coined this term, "Income from Calves Over Semen Cost." In the dairy industry, you have heard, probably, "Income Over Feed Cost." That's very well known. And so similarly, we call this ICOSC, Income from Calves Over Semen Cost, and the idea is to see, what is the level that we should use semen of different types, beef, conventional, and sex semen, and produce crossbred calves and still, hopefully, have enough replacements from within our herd to maintain the herd size?

OK, so we create this framework, and there is a tool, again, on this. So every form individually can enter the data and do their own analysis because it will depend on their own conditions. And here, basically what we do is try to find out what should be the best level, economically efficient and still producing enough replacement balance.

You can see there in the green box, it will calculate this, what's the income from cups over semen cost, but also, we'll calculate what's the replacement balance, how much we are producing in every strategy of using different type of cement at different breedings, and what we need in order to maintain our herd size, which is important.

And based on that, we can do analysis like the one you see here. This is published relatively recently, a couple of years ago. And basically, I know it's a three dimensions there, but I think it's giving quite important information because on the y-axis, on the vertical axis, is the income from calves over semen cost. So that's what we're trying to maximize here.

In addition, if we do business as usual, it will be 0. So now, if we use semen, beef and sex semen, we can hopefully increase that value above 0. We don't want the negatives, obviously. On the x-axis there, well, on one of the horizontal axes, it's showing the use of sex semen, from not used, and NS, and all the way to 2C.

That means even using sex semen in the first breeding of the second lactation animal, so from not using all the way to using, quite a bit, not in all the animals but at least in the first breeding on the first lactation and second lactation animals in addition to the heifers. Normally, heifers are more heavily used in sex semen.
 
And then on the other axis, the beef semen cows, it's the use of beef semen in percentage. So there is from not using anything all the way to 100% of use. So basically, here we have different options. We are trying to maximize the income from calves of semen cost, and we can do that even if we don't produce enough animals to replace our herd.

And those are the cubes that are outside of the red circle. So we can have a higher value, but that will imply that we need to buy replacements from outside the farm. And that's normally an unlikely situation because most of the farmers would like to have a replacement from within. So more likely the red circle inside the graph represents where we need to operate.

And within that circle, we want the maximum income from calves over semen cost, and then if you find there the bar that represents that, basically in this case implies use of sex semen all the way to the second lactation, first lactation and about 75% of use of beef semen in the rest of the animals and a little bit of use of conventional semen as well.
But that will depend on the economic situation. Actually, I rerun this with the prices of today. As I mentioned, this research was about two years ago. So now our price of crossbred is much higher than then, at that time was about $200 or $250. Now the crossbred price currently is about $700, which is much higher than a heifer calf from Holstein. That's $225.

And if we put those numbers today, our values will be much higher. There will be no negatives, and in my opinion, it would be no-brainer to use beef at certain extent. In every farm, it will be different, depending on the reproductive performance and depending on their own cost and market opportunities. But in general, it is an opportunity that-- I think it's here to stay.

And actually, looking also to the data from USDA-- and this is from the Food and Agricultural Policy Research Institute in Missouri-- they actually track the cattle prices from fed steer and feeder steer prices, the two lines that you see there in the graph. And they project to the next 10 years, actually, which is very interesting.

You can see the prices are high now. They may even go higher, and they, after five years or so, maybe decline a little bit. But the important thing here-- they will remain higher for a good period of time. So it seems like the opportunity is here to stay for a while and probably to continue.

So some highlights of this, just finishing up here, there is an opportunity here. The opportunity will be higher. The higher is the reproductive performance on the farm, and that has to do with the management. The better the management of the farm, probably they will have better reproductive performance. And we measure the pregnancy rate. We need to improve the pregnancy rate. The better the pregnancy rate, the better the opportunity of using beef semen and sex semen together.

These crossbred calves prices are high and will remain high. Probably the opportunity will continue there. And always better opportunity would be used in combination with sex semen. It will make us open a space to use more crossbred breedings. There would be more benefits if the farm or farmers are willing to or have the opportunity to buy or sell dairy calves outside of the farm, which-- I mentioned before it's unlikely, at least in the situation we are, but maybe there are other situations.
 
And the final thing I want to mention here-- there is a tool available openly to anyone to use at my website, which is dairymgt.info. And with that, I will be very glad to hear your opinion and open for questions, comments, anything. Thank you very much.

[APPLAUSE]

Who wants to start? I'm expecting questions. Yes?

SPEAKER 1: So my question, being in the dairy industry, going back to your feeding different groups of animals, you talked about the feed cost, but did you factor in your labor costs of possibly having to move animals to different groups based on their stage of lactation or production level and then maybe the additional labor that's needed to have more groups of animals, maybe the additional labor to mix more loads of feed per day for those groups of animals, and also the social aspect of moving first-lactation animals with multilactation animals, just the social aspect of the animal housing?

VICTOR CABRERA: Thanks. Thanks for the questions, and that allows me to clarify a little bit more the research we did there. Those are very good points. As a short answer, we pondered everything except the social interaction, and I will comment on that.

But certainly, if we put in context the savings on the feet, that will pay off very easily all the rest of the costs, the cost of making an extra batch of feed or having an extra person or extra hours of a person to do a little more movement of cows. And if we automatize the system, relatively these costs will be lower. So if we put in context with-- that's why I put at the beginning this Pareto graph, how much it represents the feed cost in the farm. So really, it will pay off. They are included.

Now, the social aspect is very important because there is some evidence-- I should recognize that when you move cows, there is an effect, a stress effect on the animal that may produce a decrease in the production temporarily. We have found literature that indicates that maybe that effect may fade out, depending on the size of the group.

Actually, the latest research indicates that if the groups are more than 60 cows, it's very difficult that that social hierarchy is being maintained, and the cow will recognize-- it has such a big community that the cow-- the social interaction will be diminished. But now, if we consider in the research, actually, there, we actually put some sensitivity analysis-- if the social interaction or hierarchy will be part of the analysis, there will be a detriment in productivity.

So we use the best knowledge to find out what detriment would be. And yeah, there could be some level of decrease. I believe the number were 5 pounds per day during four days or five days. And we pondered that in value. Again, it's very small compared to the big picture. Thank you.
 
DAVID: And would using a robotic dairy system change this at all? Is that factored in?

VICTOR CABRERA: Great point, David, yes. The way how in the automatic milking systems cows are being fed normally is they have a portion of the feed, which is called the partial mixed ration, outside of the robot. And when they go, they visit the robot, they will have an amount being fed to each cow individually.
 
So there is the opportunity to do even a better precision feeding there because every animal, depending on how much they milk, what their stage of the lactation is, the parity, you can give a little more or less of that feed, even more in the newer systems. You can not only give more or less of the concentrate, but you can formulate on the fly when the cow goes there.

So some cows will require more energy or more protein that can be adjusted. But still, a large group of the animals will receive the same PMR, the partial mixed ration, and there is where that grouping could play an important role. So still there could be adjust to different groups of animals, and then there will be completed in the automatic milking system, in the robot.

In my opinion, there they have a much better opportunity and would be much easier to do. They have more data, actually. An important thing of the robots is that they have more data available for them and to do this analysis in a more automatic way.

DAVID: I think there's time for one more question. And we'll be heading to lunch next, so maybe no one wants to-- oh, we have one?
 
VICTOR CABRERA: Oh, there is a question.
 
DAVID: All right.

SPEAKER 2: [INAUDIBLE] to males?

VICTOR CABRERA: Correct.
 
SPEAKER 2: And the idea is to have a greater female population because they become milking cows rather than males that serve the industry?
 
VICTOR CABRERA: Correct. So let me-- thanks for that clarification. So sex semen, this technology that they select the semen at the chromosomal level in which you can breed the animal and have a higher proportion of females-- that's what you want in the dairy industry. You will still have a small proportion of males.

What happens with the males in the dairy farm-- the males will be sold for a different purpose, so they are not part of the dairy business. You don't want too many. Normally, if you give conventional semen, you will have half and a half males and females. Actually, indeed the number is 53% males and 47% females. That's what normal is.

With the sex semen, we consider normally 90%, 91% of females and the rest males. But then when we do that-- and the point I try to make is when we have-- we can produce more females of our breedings using sex semen. That gives us space to breed more animals with beef semen.

And when we breed beef semen, whether it's male or female, it's sold to the beef industry. Now, there is also sex beef semen, in which you can produce more males, beef. In that case, you prefer the males on the beef side.
 
DAVID: Well, thanks again, Victor-- for a--
 
VICTOR CABRERA: Thank you.
 
DAVID: --great presentation.