Open data for improved cropland nutrient budgets and nutrient use efficiency estimations

Cameron Ludemann; Researcher, Wageningen University and Research, Netherlands Author

11/13/2023 Added

11 Plays

Description

Cameron will highlight new field experiment databases and prediction models his team are developing to improve estimates of crop harvest index and nutrient concentrations of crop products and residues with an aim to improve nutrient budgets and nutrient use efficiency measures at a local and global scale.

Searchable Transcript

Search:

[00:00:00.780]The following presentation
[00:00:02.220]is part of the Agronomy and Horticulture Seminar Series
[00:00:05.790]at the University of Nebraska-Lincoln.
[00:00:08.318]All right, good afternoon.
[00:00:09.151]Thank you everybody for attending our seminar series.
[00:00:12.397]Today, I have the great pleasure to introduce
[00:00:15.300]Cameron Ludemann to you.
[00:00:17.033]Cameron is originally from New Zealand.
[00:00:19.680]Before going to do his PhD,
[00:00:22.371]he works on a consulting firm
[00:00:24.572]related to a (indistinct) genetics.
[00:00:28.050]Afterwards, he completed his PhD
[00:00:30.297]at the University of Melbourne in Australia,
[00:00:32.651]where he work out of the
[00:00:35.967](indistinctly)
[00:00:41.537]After University of Melbourne,
[00:00:43.257]he manage a research and development program
[00:00:45.388]in the New Zealand dairy industry,
[00:00:48.027]with the aim of improving pastures.
[00:00:50.130]Then change to other country,
[00:00:51.720]he passed nearly a year at the Netherlands
[00:00:53.310]where he took up a position the second year
[00:00:55.580]at the Wageningen University.
[00:00:57.913]And during the past four years or so,
[00:01:00.107]he has been working
[00:01:02.250]on global crop nutrient removal database,
[00:01:06.810]which has working collaboration
[00:01:08.267]with National Agricultural Association and with the FAO.
[00:01:13.140]So Cameron, it's a great pleasure to have you today with us.
[00:01:18.073]The floor is yours.
[00:01:19.260]Great.
[00:01:20.700]Thanks, Patricio, for that introduction.
[00:01:23.550]And like Patricio said,
[00:01:25.170]I'm working in Wageningen University in the Netherlands,
[00:01:30.180]and the main project I started with in Wageningen.
[00:01:34.740]And my postdoc was developing
[00:01:36.780]a global crop nutrient removal database.
[00:01:39.870]And that led on to working
[00:01:41.220]with the food and agricultural organization
[00:01:43.710]cropland nutrient budget,
[00:01:45.060]which has come out last year.
[00:01:47.160]And the presentation today will go through
[00:01:50.790]some of the data and findings from those projects.
[00:01:55.110]And several things that I'd like to go through today,
[00:02:00.270]include firstly, giving you a bit of an insight
[00:02:03.120]on what I think are
[00:02:05.520]some of the bits of value
[00:02:07.860]from making data available in science,
[00:02:11.430]as well as what open data are actually available,
[00:02:15.000]in terms of nutrient budgets and use efficiencies
[00:02:18.210]at a world level and at a more localized level.
[00:02:23.340]I will take you through some of what we're doing,
[00:02:25.590]in terms of trying to improve those estimates
[00:02:27.990]with some actual localized data.
[00:02:32.220]And if we have time,
[00:02:33.210]I'll go through some of the challenges
[00:02:35.310]to working in such an open data project.
[00:02:41.550]Okay, so, we've made a lot of progress in science,
[00:02:43.950]particularly since we started writing up our methods,
[00:02:47.820]and results, and discussions.
[00:02:50.340]We've been able to work on other people's work
[00:02:53.040]and stand on the shoulders of giants.
[00:02:55.980]But I feel like we're now coming to a stage
[00:02:58.920]where it's not just enough
[00:03:00.360]to be making your articles available online to everyone.
[00:03:04.860]Actually making the data that led you to get,
[00:03:09.150]do your analysis and your conclusions,
[00:03:12.210]is becoming much more important now.
[00:03:14.430]So I feel like,
[00:03:15.600]it's not enough just to publish your results,
[00:03:18.600]it's becoming more and more important
[00:03:20.010]to open up your data to the world.
[00:03:25.950]That's really important
[00:03:26.783]'cause of replicability and transparency.
[00:03:30.120]Now,
[00:03:32.790]particularly now,
[00:03:33.623]when we're working on more complicated statistical methods,
[00:03:36.810]we're not just looking at means and standard deviations,
[00:03:40.110]but we're looking at advanced statistical methods.
[00:03:43.380]We're looking at machine learning methods,
[00:03:46.018]and these can change
[00:03:48.510]depending on slight changes in your code.
[00:03:51.180]So, it becomes increasingly important
[00:03:53.850]to make sure that that code
[00:03:55.710]and the data that you've used to do that analysis
[00:03:58.350]is actually open for interrogation,
[00:04:01.740]to see exactly what was done to come to those conclusions.
[00:04:08.010]Another value, in terms of opening up data
[00:04:11.040]in the science projects
[00:04:13.620]is just opening up your data to a wider range of people
[00:04:18.210]from different sectors.
[00:04:20.220]Now a good example of this,
[00:04:22.350]is that NASA, back in, I think 2018,
[00:04:26.190]did a competition to see
[00:04:29.490]whether people can improve algorithms
[00:04:32.220]to understand what dark matter,
[00:04:36.570]what effect it's having
[00:04:38.010]on some of the images of the galaxy.
[00:04:42.810]And that was actually won by a glaciology student.
[00:04:46.890]So, it's a great way of increasing the scope of people
[00:04:51.600]that can actually do the analysis,
[00:04:55.260]and could also be done in agronomy as well.
[00:05:00.240]Combating misinformation.
[00:05:01.530]Well, that's another important thing
[00:05:05.640]to hopefully improve with open data.
[00:05:08.280]Now, we're not gonna stop people from...
[00:05:13.080]We're not gonna stop people in every case
[00:05:16.800]from creating misinformation.
[00:05:19.590]But we saw through the the COVID pandemic,
[00:05:22.860]how important it was to have real time
[00:05:25.050]and open data available,
[00:05:26.820]so that decision makers could make the right decisions
[00:05:29.850]and there could be an assessment of the benefits and costs
[00:05:34.320]of some of the decisions they make,
[00:05:36.660]likes of vaccines, which would change over time.
[00:05:42.420]And of course, when you open up those data,
[00:05:44.820]there's gonna be efficiencies.
[00:05:46.830]So if you imagine, you know,
[00:05:49.170]50 different people wanting to do a meta-analysis
[00:05:51.540]on the same sorts of topics or slightly different topics,
[00:05:55.440]but with the same data,
[00:05:56.970]you don't want to have to search through the literature
[00:05:59.970]or compile those data 50 times.
[00:06:04.050]And so, there can be a lot of efficiencies
[00:06:06.600]if it's actually made available,
[00:06:08.910]so that everyone can do those analysis
[00:06:11.220]in different ways that they want.
[00:06:13.980]Also makes meta-analysis easier.
[00:06:16.140]So, if you're doing your own trial,
[00:06:17.970]and you wanna see how it compares to a trial down the road
[00:06:21.840]or in a different country,
[00:06:23.430]having those data open allows you to do that.
[00:06:31.141]Now I mentioned COVID,
[00:06:32.280]which was a important thing
[00:06:34.230]that was quite personal to people.
[00:06:36.960]The data was really important for making it open
[00:06:40.020]for the decision making,
[00:06:41.550]but we're getting increasing awareness
[00:06:45.090]of what effect some nutrients are having at a global scale.
[00:06:50.340]So on this figure here,
[00:06:52.380]we're showing basically the planetary boundaries
[00:06:57.300]that are being crossed,
[00:06:58.860]and we've got a number of different aspects
[00:07:04.620]that have been assessed over time.
[00:07:07.410]And what you've seen is that nitrogen and phosphorus
[00:07:11.520]and their biochemical flows,
[00:07:13.620]have been pretty important
[00:07:16.590]and have become increasingly important,
[00:07:19.530]and shown to be well beyond the planetary boundaries.
[00:07:24.630]So this is something that's gonna become
[00:07:26.430]really important in the future,
[00:07:27.870]and we need to have open data
[00:07:29.700]to support any decisions that are made,
[00:07:32.730]much like has been done
[00:07:34.350]with greenhouse gas emissions and many other things.
[00:07:43.620]So, it's not surprising then
[00:07:45.697]that at a world level, at the United Nations,
[00:07:49.080]they've been trying to improve the availability of data
[00:07:52.500]for nutrient budgets.
[00:07:54.840]So the ability to see what sort of surplus,
[00:07:58.710]which is the nutrient inputs minus the nutrient outputs,
[00:08:02.610]there are in different parts of the world.
[00:08:06.150]So last year,
[00:08:07.650]the UN FAO developed the Cropland Nutrient Budget,
[00:08:12.120]and this is an open data set that anyone can use,
[00:08:16.020]which have all the different components
[00:08:17.940]that it may make up the overall budget,
[00:08:20.670]but also the actual balances.
[00:08:23.430]So the surpluses and deficits
[00:08:25.020]and the cropland nutrient use efficiencies
[00:08:28.230]of those nutrients.
[00:08:29.700]This is done by country, so it's quite high level.
[00:08:32.940]And the latest data is up to 2020.
[00:08:36.630]But the latest new data should be coming available
[00:08:40.680]in November this year.
[00:08:44.817]And when you look at the balance from the FAO,
[00:08:48.660]they include nutrient inputs from synthetic fertilizers.
[00:08:54.000]Now,
[00:08:56.700]they get data at a world level
[00:08:59.220]and they source their fertilizer information
[00:09:02.700]across all agriculture.
[00:09:04.920]And so, some assumptions have to be made
[00:09:07.140]of what fraction of those synthetic fertilizers
[00:09:09.420]that should go to Cropland,
[00:09:11.370]to get an idea of the amount of inputs
[00:09:13.350]of synthetic fertilizers.
[00:09:16.710]Of course there's manure applied to soils as well.
[00:09:19.530]And that's again, at a country level,
[00:09:22.020]looking at total numbers of livestock
[00:09:24.000]and some of the excretion factors using IPCC,
[00:09:27.720]the Inter-governmental Planner
[00:09:29.490]of Climate change Coefficients,
[00:09:32.550]and other inputs like atmosphere at nitrogen deposition,
[00:09:36.960]using experts who have developed world models
[00:09:41.430]of nitrogen deposition,
[00:09:43.140]to get an estimate of how much nitrogen's been deposited
[00:09:45.750]on a per hectare basis.
[00:09:48.720]And of course, biological fixation.
[00:09:50.970]Again, experts who have been looking through the literature
[00:09:53.790]and seeing what amount of nitrogen is being fixed
[00:09:56.850]for some of the different legume species.
[00:09:59.910]And so, you've got all those nutrient inputs,
[00:10:02.820]that have the nutrient outputs,
[00:10:05.130]which at this stage just include crop nutrient removal
[00:10:09.690]to get the overall balance.
[00:10:15.450]And with those data,
[00:10:17.520]this is the kind of thing that you can do with it.
[00:10:20.160]So, we've got three figures showing up here
[00:10:23.340]for nitrogen, phosphorus, and potassium.
[00:10:26.430]And these show those balances
[00:10:30.120]either as a surplus or a deficit.
[00:10:33.720]Now, you see,
[00:10:37.140]in some cases,
[00:10:38.160]like say nitrogen phosphorus,
[00:10:40.170]what's important from these maps
[00:10:41.910]is that it can give you an idea
[00:10:43.440]of some of the hotspot locations
[00:10:45.390]where too much nutrients are being applied,
[00:10:47.700]and in this case that can have adverse effects
[00:10:50.280]on the environment.
[00:10:51.600]Nutrification of waterways, leaching, runoff,
[00:10:54.780]nitrogen oxide emissions.
[00:10:58.320]On the other hand, you also see with the light green color,
[00:11:02.460]that there's some places
[00:11:03.870]where we're getting annual deficits in nutrients.
[00:11:07.050]So you see, for nitrogen in parts of Africa,
[00:11:10.860]and some quite green areas
[00:11:12.870]indicating that there's very low balances or surpluses,
[00:11:17.850]which could mean that there's mining of the soils.
[00:11:21.450]You also see for potassium at the bottom there.
[00:11:24.840]In South America you're getting negative values,
[00:11:27.900]indicating there's mining of the soils there,
[00:11:31.080]which can obviously can have negative effects
[00:11:35.190]on plant production, but also livestock production as well.
[00:11:42.510]Looking at some of the major countries,
[00:11:45.480]if you look at the top figure here,
[00:11:47.880]at the nutrient balance in millions of tons of nutrient
[00:11:51.150]for nitrogen, phosphorus, and potassium,
[00:11:53.190]in the red, green, and blue,
[00:11:56.250]what sticks out is that China and India
[00:12:00.360]are major players when it comes to nutrient surfaces,
[00:12:05.670]in terms of nitrogen.
[00:12:08.130]Also see, USA also having quite substantial
[00:12:11.790]nutrient surpluses.
[00:12:14.910]But in some of those major countries,
[00:12:17.850]we're actually seeing a deficit.
[00:12:21.660]On a per hectare basis,
[00:12:23.280]on the figure below that,
[00:12:25.857]we see that China and India, France, Pakistan
[00:12:31.320]are having very high nutrient surpluses for nitrogen.
[00:12:37.410]But again, there are some countries
[00:12:39.810]where there's some deficits like so the USA with phosphorus.
[00:12:48.450]When we look over time,
[00:12:50.820]you can see why the planetary boundaries of nitrogen
[00:12:54.540]are being exceeded.
[00:12:56.670]Also for phosphorus,
[00:12:57.720]but it's really striking here
[00:13:00.150]when you look at the nutrient balance
[00:13:03.030]in total millions of tons per region,
[00:13:08.400]how much of an increase there's been over time
[00:13:11.370]since about 1961, when these data were first available.
[00:13:16.860]You see that sharp increase in the surplus.
[00:13:19.770]So this is an annual surplus,
[00:13:21.840]so it can accumulate over time.
[00:13:23.910]And depending on which nutrient you're talking about,
[00:13:26.490]and there has been a bit of a plateauing you could say,
[00:13:30.360]or it hasn't increased to the same extent.
[00:13:33.450]But when at a world level,
[00:13:35.220]but when you look at the regions,
[00:13:37.830]what's quite important is,
[00:13:39.864]is that's been driven a lot
[00:13:41.910]by the amount of surplus in Asia,
[00:13:45.750]particularly in China and India.
[00:13:50.010]For Europe, we saw quite an increase up until about 1990,
[00:13:55.650]and then quite a sharp reduction over time.
[00:14:02.610]Another thing to consider isn't just the surplus
[00:14:06.900]and how big they may be,
[00:14:09.780]but we're also seeing these annual deficits over time
[00:14:13.290]that are happening to some of the regions.
[00:14:15.900]Now this is also important.
[00:14:17.070]So with Africa,
[00:14:18.927]that's indicating there's a mining of nutrients over time,
[00:14:23.820]which can obviously erode the quality of soils.
[00:14:27.060]So important to understand what's going on over time
[00:14:32.250]and what's happening in each of the countries
[00:14:34.500]and regions of the world.
[00:14:38.940]Now, when trying to do these big maps
[00:14:42.180]of hotspot locations for nutrient deficits or surpluses,
[00:14:46.230]it's sometimes hard to find data
[00:14:48.390]for every country, or sub region, or for the world.
[00:14:52.380]So, what some sometimes people do,
[00:14:55.710]which is based on the Intergovernmental Panel
[00:14:57.990]of Climate Change with greenhouse gas emissions,
[00:15:00.300]is you can use a tiered approach.
[00:15:03.600]So, if you have no data available for your country,
[00:15:07.020]you can use what we call this world average values,
[00:15:10.470]and that's what the current FAO cropland budget,
[00:15:14.010]nutrient budget uses at the moment.
[00:15:18.210]Or if you have more specific data from surveys
[00:15:22.650]for a regional country,
[00:15:23.850]you can use those,
[00:15:24.990]and we call that a tier two approach.
[00:15:28.560]Or if you're lucky and you have field experiment data
[00:15:31.830]or really detailed within country survey data,
[00:15:35.220]then you can use those at the tier three level.
[00:15:38.010]So in the future, with these global nutrient budgets,
[00:15:42.210]ideally we'd be able to see
[00:15:43.950]what countries have tier three data.
[00:15:47.010]Failing that, go to the tier two,
[00:15:49.350]or tier three coefficients.
[00:15:54.240]And we'd really like to enable researchers
[00:15:56.550]doing these global balances,
[00:15:58.500]to be able to use those mo more locally relevant estimates.
[00:16:05.010]So if you think of trying to get a nutrient coefficient
[00:16:08.520]for to represent the world,
[00:16:11.100]it can be hard to know exactly what source to use.
[00:16:16.080]Now just as an example,
[00:16:17.670]I have a figure here showing values for maize
[00:16:20.400]for some of the key coefficients
[00:16:22.710]that go into nutrient budgets.
[00:16:24.960]So on the far left, we have harvest index one,
[00:16:28.470]harvest index two.
[00:16:30.630]That indicates some of what people have purported
[00:16:33.420]to be average values for harvest index,
[00:16:35.790]for maize for the world, for tier one,
[00:16:38.370]that's harvest index one.
[00:16:40.200]Or at a more regional level, which is harvest index two.
[00:16:46.500]We see variation there between sources,
[00:16:49.890]and of course the same can be said about grain nitrogen
[00:16:53.850]at a tier one, which is grain N1,
[00:16:56.640]or grain nitrogen concentration at tier two.
[00:17:00.570]The same goes for grain and residue,
[00:17:03.390]nitrogen concentrations for phosphorus and potassium.
[00:17:08.100]In some cases, as you look here for potassium,
[00:17:12.900]there's actually a wide difference
[00:17:15.960]depending on which source you you choose.
[00:17:18.690]So depending on whether you choose this value down here
[00:17:22.980]or this value up here,
[00:17:24.630]it could almost double the estimates of potassium or more,
[00:17:29.100]that are being taken away,
[00:17:30.780]depending on just what source you're using.
[00:17:36.720]So, this then explains why
[00:17:39.030]there can be quite a lot of variation
[00:17:41.070]between different researchers
[00:17:42.780]when they come to do these estimates of nitrogen balance.
[00:17:46.560]At the top, so although a lot of the different researchers
[00:17:53.280]show similar trends over time,
[00:17:55.650]and nitrogen balance as a surplus,
[00:17:57.810]and millions of tons per year,
[00:18:01.470]there's a lot of variation there
[00:18:03.300]and we're talking about potentially
[00:18:06.300]tens of millions of tons different,
[00:18:09.000]at a world level.
[00:18:11.190]The latest results from the FAO are in bold and black.
[00:18:15.870]So you can see how that compares to the other estimates,
[00:18:20.595]and that makes sense
[00:18:21.428]considering they could be using
[00:18:22.410]different sources of information
[00:18:25.560]for their coefficients.
[00:18:27.390]The same goes with nitrogen use efficiency,
[00:18:30.570]okay, we're generally getting the same sort of trends
[00:18:34.110]with a reduction in nitrogen use efficiency
[00:18:36.690]from about 1961 to about the 1990s.
[00:18:41.250]But what's positive is that we are starting to see
[00:18:43.427]bit of an increase in trend over this period
[00:18:47.595]and it's looking like it's been increasing
[00:18:51.300]over the last sort of five years,
[00:18:53.370]which is positive to see.
[00:18:55.290]We do see a bit of a dip there,
[00:18:57.000]but we have to see the latest results to see what happens.
[00:19:04.140]Okay.
[00:19:14.940]Now, in terms of phosphorus,
[00:19:17.280]there haven't been as many data sets that have shown
[00:19:21.270]changes in nutrient balances over time,
[00:19:23.910]or nutrient use efficiency.
[00:19:26.250]But when we compare the latest FAO results,
[00:19:29.308]we again see similar trends
[00:19:32.040]but variation in the absolute values.
[00:19:34.650]And quite a lot of variation
[00:19:36.470]in the the actual use efficiencies over time.
[00:19:41.550]Again, people using different methods
[00:19:43.890]and you're getting different values.
[00:19:46.530]And we couldn't seek,
[00:19:47.490]we couldn't actually find
[00:19:48.450]the same kind of data to compare our data
[00:19:50.490]with four potassium.
[00:19:55.020]So, We are trying to develop models
[00:19:56.550]to improve the estimates of maize nutrient removal
[00:19:59.310]based on what we call widely available data
[00:20:02.820]at the tier three level.
[00:20:04.680]Now when I say widely available data,
[00:20:07.410]I'm talking about data that we can get
[00:20:09.240]at a world level, by country.
[00:20:12.930]So for instance, crop product yield data,
[00:20:15.150]you can get from the food and agricultural organization,
[00:20:18.300]fertilizer application rates,
[00:20:20.010]you can get from the International Fertilizer Association,
[00:20:23.220]or yield potential data,
[00:20:24.717]you can get from the yield gap atlas.
[00:20:28.440]What we're trying to do is develop those models
[00:20:30.210]to improve those tier three estimates
[00:20:32.550]of how much nutrients removed in crop products.
[00:20:36.990]And I'll show you how they compare
[00:20:39.330]when we just use those overall world values
[00:20:42.240]at the tier one level.
[00:20:46.110]Now, I'll be showing results,
[00:20:47.370]which are a partial nutrient budget.
[00:20:50.280]Now I call them partial
[00:20:51.480]because they don't include every component of a budget.
[00:20:58.230]We take away the nutrient outputs from the inputs,
[00:21:00.720]but the only inputs that we include are fertilizer applied,
[00:21:05.100]and the only outputs we include
[00:21:07.020]are nutrients taken away from the crop removed.
[00:21:09.900]And basically we're assuming all the other components
[00:21:12.150]stay the same,
[00:21:13.781]'cause we're really interested in what effect
[00:21:15.540]using the tier one versus three methods has.
[00:21:22.710]So we've got those inputs,
[00:21:24.180]and we can get data on total crop production, for say maize.
[00:21:29.550]But at a world level we don't have survey data
[00:21:33.270]for crop residues.
[00:21:34.860]And that's why it's important
[00:21:35.940]to get estimates of harvest index
[00:21:37.860]to do a back calculation,
[00:21:39.570]to estimate how much grain and residues are being produced,
[00:21:43.500]so they will get the total amount of outputs
[00:21:46.860]for those products
[00:21:49.470]and we can then create models for harvest index
[00:21:53.460]and nutrient concentrations of grain and residue.
[00:21:57.270]So we get the overall partial nutrient budget
[00:22:00.090]as a surplus or deficit.
[00:22:05.160]We include nitrogen, phosphorus, and potassium.
[00:22:10.230]And the models that we've used
[00:22:12.540]have been trained on data
[00:22:14.608]that's been taken from the literature.
[00:22:16.650]So we've literally gone through hundreds of journal articles
[00:22:20.940]and taken summary statistics from those pages,
[00:22:25.290]and turned them into machine readable format.
[00:22:29.760]The other set of data that we've trained the models on,
[00:22:34.050]include requests from organizations from around the world.
[00:22:38.010]So these could be individual researchers,
[00:22:40.723]commercial companies, universities,
[00:22:44.340]any organization who would share.
[00:22:49.620]Now, the data that came from the literature
[00:22:54.720]are available online in the Dryad repository.
[00:22:58.050]So feel free to check those out if you're interested,
[00:23:01.710]particularly for the tier one and tier two estimates
[00:23:04.170]I've just referred to in the budgets.
[00:23:08.490]And the data that have been sent in
[00:23:12.180]from different individuals or organizations
[00:23:14.910]are also freely available at this website here.
[00:23:18.180]So cropnutrientdata.net.
[00:23:21.420]Now anyone can sign up for this,
[00:23:23.700]and what's good about it
[00:23:26.610]is that once you've signed up to it,
[00:23:29.400]you can actually create your own workspaces
[00:23:31.500]to analyze the data.
[00:23:34.230]And you can save them so that they, as new data comes in,
[00:23:38.160]you can see what changes,
[00:23:39.990]saying aspects of yield and nutrient concentrations.
[00:23:44.640]You can also create different maps
[00:23:46.950]to see where the data are coming from.
[00:23:51.180]Scatter plots, histograms, you name it, you can do it.
[00:23:56.220]So once you've had a look through those data,
[00:23:57.930]you can actually filter through them
[00:24:00.300]and you can download them as a CSV file.
[00:24:03.120]That's all been standardized.
[00:24:08.040]Sometimes, quite an issue with data
[00:24:11.250]is that you don't exactly know what each data point is.
[00:24:15.300]And we've spent a lot of time making sure
[00:24:17.910]that in this database,
[00:24:20.760]there's good anthology to show you exactly
[00:24:24.210]what you mean by the area of of field, for instance.
[00:24:28.140]Every parameter has many layers,
[00:24:30.300]a hierarchical layer
[00:24:31.260]of exactly what units you're talking about.
[00:24:34.008]And as you see here, the field of study,
[00:24:38.280]so that you know exactly what that data is,
[00:24:40.530]what those data are.
[00:24:43.140]And that enables it to be much more useful for others
[00:24:46.830]who want to interrogate those data.
[00:24:51.600]At this stage, we have data from all around the world,
[00:24:55.620]and this map shows where we have data for maize,
[00:25:00.750]rice, soybeans, and wheat.
[00:25:03.480]Now while we have good global coverage,
[00:25:06.990]there are obviously some areas we're lacking.
[00:25:10.980]So particularly in Africa and Russia,
[00:25:14.340]there's places where we are pretty limited
[00:25:17.880]at what data we have.
[00:25:22.650]The models that we've used with those data
[00:25:25.320]include random forest regression models
[00:25:28.860]and mixed effects regression models.
[00:25:34.380]So we use those what we called
[00:25:36.240]widely available prediction variables
[00:25:39.030]in the random forest and mixed-effects models
[00:25:42.840]to predict the nutrient concentrations
[00:25:44.247]of the grain and stover and maize,
[00:25:46.650]and the harvest indices.
[00:25:48.270]And we pulled those, all those data together,
[00:25:51.030]and set aside 80% of it for training the models
[00:25:55.290]and 20% of it for testing the prediction accuracies.
[00:26:02.190]If you wanna get right down to the details,
[00:26:05.820]we can talk about it later,
[00:26:06.960]but we tested the linear mixed-effects models
[00:26:11.010]to make sure that there wasn't too much collinearity
[00:26:14.070]between some of the variables.
[00:26:15.960]We tested them for normality and visually
[00:26:18.690]with quantile plots.
[00:26:21.690]We also tested some of the linear mixed-effects models
[00:26:24.990]for explanatory power, how well they explain variation,
[00:26:29.220]in say the harvest index or nutrient concentrations
[00:26:32.880]using Akaike Information Criteria
[00:26:36.720]and Nakagawa's R squared values,
[00:26:39.450]which are basically an R squared
[00:26:42.090]that's useful for mixed-effects models.
[00:26:46.260]We could then accept or reject our hypothesis,
[00:26:48.720]which is could we explain variation in these variables
[00:26:53.880]based on just a limited set of widely available variables.
[00:26:59.640]And then the best models from linear mixed-effects models
[00:27:04.320]were then tested in their prediction accuracy
[00:27:07.710]against the random forest regression models.
[00:27:12.630]We could then use the best models in our estimates
[00:27:16.080]of nutrient balances.
[00:27:20.790]Now on your right,
[00:27:23.100]you'll see those prediction accuracies
[00:27:26.220]for the mixed-effects models
[00:27:29.880]in the column on the left,
[00:27:33.060]and the random forest models.
[00:27:36.360]And you can see that I've included them
[00:27:38.670]for harvest index and grain nitrogen concentration,
[00:27:42.450]and residue nitrogen concentration for maze.
[00:27:47.580]We also did it for phosphorus and potassium,
[00:27:50.250]but for brevity, I won't go into those results today.
[00:27:55.140]What's most striking
[00:27:56.425]is that we have typically greater prediction accuracy
[00:28:01.740]with the random forest regression models.
[00:28:05.760]And so, we actually use the random forest models
[00:28:08.610]in our predictions.
[00:28:14.460]When we look,
[00:28:15.840]when we use those predictions
[00:28:18.210]in maize nutrient removal estimates
[00:28:21.390]for some of the major countries,
[00:28:22.890]so this table shows some of the major countries,
[00:28:25.530]including Argentina and Brazil, China, India, and the US
[00:28:29.790]using the tier one and the tier three approach
[00:28:34.860]for nitrogen, phosphorus, and potassium.
[00:28:40.230]What's important to take away from this table
[00:28:43.710]is that typically when we are using the tier three approach,
[00:28:47.160]we're actually getting lower estimates
[00:28:49.440]of nutrients being taken away
[00:28:51.480]as nitrogen, phosphorus, or potassium.
[00:28:54.660]This can be pretty substantial,
[00:28:57.150]as you see for instance, potassium in China.
[00:29:02.250]There's about a 50% difference in estimates
[00:29:05.130]of the amount of nutrients being removed.
[00:29:08.040]So that can have quite an effect
[00:29:09.660]on your overall nutrient balance.
[00:29:16.500]So why is this happening?
[00:29:20.262]Well, what we think is that the tier three maize,
[00:29:25.260]tier three estimates are based on,
[00:29:30.060]sorry, the tier one estimates
[00:29:32.130]are typically based on values
[00:29:34.800]that have come from field experiments from North America.
[00:29:38.610]When I talk to some of the people
[00:29:40.080]who know what trials were included in those,
[00:29:43.830]they were often high nutrient input experiments.
[00:29:47.160]So they may not even be relatable
[00:29:51.390]to what is practiced on farm,
[00:29:53.790]even in North America,
[00:29:55.230]let alone in different parts of the world.
[00:29:58.200]But those were the data that were actually available
[00:30:01.317]and that was used in the first tier one estimates
[00:30:04.890]of nutrient concentrations.
[00:30:09.720]So, when we use those different,
[00:30:13.380]or those predicted estimates of crop product
[00:30:17.010]and crop residue removal,
[00:30:19.170]and then plug them into the overall nutrient budgets
[00:30:22.290]to get the surpluses or deficits,
[00:30:25.920]as you can imagine,
[00:30:26.850]if you've got the same amount of nutrient being applied
[00:30:29.700]but not as much nutrient being taken away
[00:30:32.010]in the crop products and residues,
[00:30:34.170]you'll end up with a greater nutrient budget surplus.
[00:30:39.180]So, here we have some of the major countries,
[00:30:42.240]and for nitrogen, phosphorus and potassium,
[00:30:44.940]the difference in nutrient surplus
[00:30:49.050]between the tier three estimates and the tier one estimates.
[00:30:53.700]What you need to know here,
[00:30:55.320]is that any values for these nutrients above zero,
[00:31:00.060]indicate that there's a greater nutrient surplus
[00:31:03.780]using tier three coefficients.
[00:31:07.350]And you can see that for the most part,
[00:31:10.200]we're actually having greater nutrient surpluses across all,
[00:31:15.180]nearly all these major countries
[00:31:17.520]for nearly all those nutrients.
[00:31:22.200]Again, it's probably going back to those tier one estimates
[00:31:25.890]and where they're coming from, what they represent,
[00:31:29.520]it could be high input field experiments.
[00:31:35.850]So what we've seen,
[00:31:37.020]is these tier three estimates
[00:31:38.880]are indicating greater maize nutrient budget surpluses
[00:31:42.060]compared to the tier one.
[00:31:44.280]Now, this shows promise to hopefully improve estimates
[00:31:48.330]of crop nutrient removal
[00:31:50.040]at or using some of these global nutrient budgets,
[00:31:53.880]like what the FAO have with their crop and nutrient budget
[00:31:58.710]'cause they only use tier one estimates
[00:32:01.050]and would really like
[00:32:02.280]to take those global budgets
[00:32:04.890]from a tier one to a tier three level.
[00:32:10.980]The problem is that at the moment,
[00:32:12.750]we've tested the prediction accuracies of our models
[00:32:17.070]based on replicate field experiment data,
[00:32:20.280]and we're really wanting to test our models
[00:32:23.640]with what's happening out on farm.
[00:32:25.800]And so that's why we're now having a drive
[00:32:27.960]to get more data from actual on-farm experiments
[00:32:32.460]that show some of the nutrient concentrations
[00:32:36.000]and harvest indices
[00:32:37.440]before we incorporate these
[00:32:40.800]into some of these global nutrient budgets.
[00:32:46.590]We also need to incorporate better data
[00:32:48.630]into those tier one and two estimates.
[00:32:51.000]Now, more and more data's becoming available,
[00:32:54.960]and we'll want to improve those,
[00:32:57.420]so that if we don't have tier three estimates
[00:33:00.060]for certain countries or regions,
[00:33:02.010]we'll have hopefully better estimates at the world level.
[00:33:09.870]So that was some of the work that we've been doing
[00:33:12.390]to improve some of the estimates
[00:33:13.770]of nutrient removal and nutrient budgets.
[00:33:17.340]Now, just onto some of the challenges
[00:33:19.830]of working in an open project like this.
[00:33:24.480]Now, firstly,
[00:33:25.920]with my experience over the last three and a half years,
[00:33:29.010]one of the big challenges of the project
[00:33:31.560]has been to convince people to share data in a open fashion.
[00:33:37.698]Is obviously with researchers,
[00:33:39.930]the time poor, that can be an issue with sharing data.
[00:33:43.550]It can take a while to put it into a certain format
[00:33:48.120]that they can share it with someone else.
[00:33:50.430]Now we try and overcome that issue
[00:33:52.590]by working with a company
[00:33:54.870]who specialize in standardizing data.
[00:33:57.660]So we basically say, "Okay, just send it to us as is.
[00:34:01.260]As long as it's relatively clear what the data is,
[00:34:04.470]we can deal with various formats."
[00:34:07.650]It can also be legal,
[00:34:09.000]in that some universities, by default,
[00:34:14.250]own the data from their PhD students
[00:34:16.740]and that adds another hurdle to students or researchers
[00:34:21.030]to sharing their data
[00:34:22.410]'cause they have to get approval for that.
[00:34:26.400]Ambiguities in the datasets are always occurring.
[00:34:31.890]You think of say, grain yield seems simple enough,
[00:34:36.060]but have they shown it on a kilograms of dry matter basis,
[00:34:41.340]kilograms of fresh weight basis?
[00:34:43.110]Has it been changed?
[00:34:45.630]So, it's all on a 13% moisture level harvest index.
[00:34:49.710]How have they estimated that?
[00:34:51.630]All these sort of things can be a bit ambiguous,
[00:34:55.200]and so we have to talk to the sources of the data
[00:34:58.650]who actually measured it,
[00:34:59.483]but sometimes those people aren't available,
[00:35:01.800]which makes it a bit harder to really understand
[00:35:05.160]and categorize these different variables.
[00:35:08.250]We try and do as best as we can.
[00:35:13.350]I mentioned before
[00:35:14.250]that we really want to be able
[00:35:15.870]to evaluate our prediction accuracies,
[00:35:18.240]not just in these replicated field experiments,
[00:35:20.790]but also out on farm.
[00:35:22.470]But if you're talking about getting data on farm
[00:35:26.130]with nutrient concentrations,
[00:35:27.510]this is quite difficult.
[00:35:29.850]There are some datasets out there,
[00:35:34.410]but they can be hard to get
[00:35:36.660]and then use for our evaluation.
[00:35:39.300]But that's something that we're working on.
[00:35:43.710]Another thing when starting up an open data project,
[00:35:48.900]is that you really need to start thinking
[00:35:51.000]right from the start,
[00:35:51.900]well, how are we gonna share our data and our codes
[00:35:55.410]so that it's easily updateable?
[00:35:57.510]So in a journal sort of situation,
[00:35:59.790]you publish your article and there it is, it's static.
[00:36:03.450]But with these live databases, things can change over time.
[00:36:07.320]You also might find mistakes
[00:36:09.360]and you've got to be open to that.
[00:36:11.790]And it can be a challenge to think,
[00:36:13.597]"Well how are we gonna do this?
[00:36:14.910]How are we gonna make it clear
[00:36:16.710]that there's been version control over time?"
[00:36:19.230]That's why I quite like the Dryad data repository
[00:36:22.410]and there's others out there
[00:36:23.460]that can show you what happens at each version.
[00:36:30.900]And of course,
[00:36:32.430]long-term investment in these projects can be difficult
[00:36:35.580]'cause funders often want to see something new.
[00:36:39.914]They don't want to just fund a continuance of say,
[00:36:42.570]a long-term data set.
[00:36:44.610]So that's also a challenge.
[00:36:48.960]But the way we're trying to bring together
[00:36:52.920]these data with the metadata,
[00:36:57.450]it means it can allow people from different angles
[00:37:01.620]and wanting to answer different research questions
[00:37:03.990]to be interested in the data,
[00:37:06.570]which can hopefully increase the value
[00:37:09.750]that people place on continuing to fund
[00:37:12.720]these sorts of projects.
[00:37:16.770]But on a positive note, we now have over 60 contributors
[00:37:22.260]from more than 40 organizations
[00:37:24.540]that have already contributed,
[00:37:26.670]replicate field experiment data from all around the world,
[00:37:30.000]which is fantastic.
[00:37:32.280]And another 80 who have promised to share data,
[00:37:35.550]which is great to hear.
[00:37:38.220]That does take time to contact people
[00:37:41.910]and take it from a promise
[00:37:43.907]to actually data in the hand.
[00:37:49.800]Now, this work that I've presented today,
[00:37:53.970]firstly the cropland nutrient budget data from the FAO
[00:37:58.110]has been part of a broader group,
[00:38:00.840]which you see the steering group members listed above.
[00:38:06.090]And work on improving estimates of nutrient removal
[00:38:11.850]have been part of a
[00:38:13.770]the Global Crop Nutrient Removal Database team,
[00:38:16.530]which again are listed below here.
[00:38:20.580]This work wouldn't have been possible
[00:38:23.220]for the nutrient removal database
[00:38:26.730]without the funding
[00:38:27.930]of the International Fertilizer Association.
[00:38:33.570]So with that,
[00:38:34.403]I'd like to put out a call
[00:38:35.730]if you do have any field experiment data or on-farm data
[00:38:39.240]that you might think could create a legacy
[00:38:44.070]in this open database for the long term,
[00:38:47.670]feel free to get in contact with me,
[00:38:50.280]and also check out the cropnutrientdata.net database.
[00:38:54.780]Anyone can access it and spread the word.
[00:38:59.550]Thank you very much for your time.
[00:39:01.410]Any questions?
[00:39:02.880]Excellent, thank you very much, Cameron,
[00:39:04.470]for the presentation.
[00:39:05.700]And please reach out now for Q and A,
[00:39:08.391]who want to go first?
[00:39:12.900]Yeah.
[00:39:15.900]That was a very nice presentation.
[00:39:18.223]How do you integrate differences in prices
[00:39:21.990]for these nutrients between countries
[00:39:24.303]into your recommendations or analysis?
[00:39:28.440]Or do you?
[00:39:29.870]At this stage, we're not doing the economics,
[00:39:32.250]but certainly, economists could be quite interested
[00:39:35.460]in these data for looking at responses
[00:39:37.530]and then applying some of the economic factors to those.
[00:39:40.680]So, at this stage, I'm not working with the economics
[00:39:43.770]and putting prices on that.
[00:39:46.200]I think the International Fertilizer Association
[00:39:48.270]may be looking into how those sort of data
[00:39:50.610]could be collected,
[00:39:51.750]but it's outside the scope of this project.
[00:39:54.840]Yeah.
[00:39:57.196]The lady's first.
[00:40:00.729]You know, Department Head first.
[00:40:02.118](all laugh)
[00:40:04.620]Thanks for your presentation.
[00:40:07.067]In terms of your crop nutrient data,
[00:40:10.316]is your data access huge?
[00:40:13.257]When you look at regions,
[00:40:15.046]for example in Africa and Northern America for example.
[00:40:18.623]Can you share with us that?
[00:40:20.190]Thank you.
[00:40:21.690]Yeah, we started off with,
[00:40:26.220]maybe I'll go back to the...
[00:40:29.430]The map.
[00:40:34.855]Here.
[00:40:36.360]So yeah, you can see there's definitely a skew
[00:40:40.020]in North America for many of these crops.
[00:40:44.790]As you can imagine where some of those crops are for maize,
[00:40:47.597]as you imagine there's a lot of data around North America.
[00:40:54.420]A lot of data has come from India for many of the crops.
[00:40:59.610]As you see here, wheat,
[00:41:01.380]there's obviously a lot in Europe,
[00:41:04.196]but it's something that we have to think about.
[00:41:07.230]I mentioned the tier one estimates
[00:41:09.270]were very biased with North America,
[00:41:11.490]and we want to move towards
[00:41:14.460]a greater global representation.
[00:41:16.500]So it's something we need to work on.
[00:41:20.220]'Cause yeah, those first tier one estimates were,
[00:41:22.140]I think mainly USDA estimates.
[00:41:24.990]And as we're pulling more of those data sets,
[00:41:28.950]they're coming from different areas of the world.
[00:41:31.140]So hopefully, we'll be able to get a better balance.
[00:41:36.185]Here it goes.
[00:41:38.033]Enjoyed the presentation.
[00:41:40.770]I'm not sure I understood how you work
[00:41:44.730]with the nutrient removal from the crop residues.
[00:41:49.410]I'm really thinking about things like,
[00:41:51.180]how do you know how much of that residue
[00:41:52.770]has actually been removed
[00:41:54.000]from the field. Yeah.
[00:41:55.110]Versus that which remains and then gets recycled.
[00:41:58.500]Yes.
[00:41:59.940]So, we have data collected
[00:42:01.770]on the nutrient concentrations of the residues,
[00:42:05.070]but
[00:42:07.290]it has been hard to get
[00:42:09.510]good global estimates
[00:42:11.070]of what's happening to those residues.
[00:42:13.020]So it's a good question you asked
[00:42:14.430]that there's maybe one or two
[00:42:19.560]sources that have said,
[00:42:20.737]"Okay, this is the amount of residues
[00:42:23.160]that has been taken away from harvested area."
[00:42:26.520]And so that's a big gap in this,
[00:42:31.050]and we're working on,
[00:42:32.610]I think the International Fertilizer Association's
[00:42:34.470]working on actually improving that.
[00:42:36.120]So getting surveys of what's happening to those residues.
[00:42:39.450]So, it's an important limitation
[00:42:42.330]to think about in these these budgets, yeah.
[00:42:46.860]I have a working governance for you.
[00:42:50.220]You mentioned before about the extreme committee,
[00:42:52.770]in which we are serving to provide advice to FAO
[00:42:56.760]on how to refine their estimates.
[00:42:58.860]You know, we all tend to think about FAO
[00:43:00.930]like this big organization that moves very slow
[00:43:04.710]and sometimes a little bit inefficient.
[00:43:08.520]But I think that this was a quite successful story
[00:43:13.050]because within a quite short timeframe
[00:43:15.720]we were able to, you know,
[00:43:17.130]provide advice and suggestions
[00:43:19.020]that were readily uptaken by them
[00:43:21.180]and they're now being used for their new calculation.
[00:43:24.630]So can you talk a little bit,
[00:43:25.560]can you elaborate a little bit more
[00:43:26.640]about how this collaboration between universities,
[00:43:29.880]the International Fertilization Association and FAO
[00:43:33.390]has worked?
[00:43:34.560]Yeah, I think it's been great in that,
[00:43:38.190]the FAO have the statistical knowledge
[00:43:40.530]how the whole database works,
[00:43:42.300]but we've been able to pick people
[00:43:45.322]with various points of expertise.
[00:43:48.000]So Patricio, you've been on the steering group,
[00:43:50.760]we've had people that are specialized
[00:43:53.010]in the amount of atmospheric nitrogen deposition, you know,
[00:43:56.880]that's their life working on that aspect.
[00:43:59.640]And we could bring all those people together
[00:44:02.430]and my whole project's been about the nutrient removal.
[00:44:05.280]So, we've been able to get all those people together,
[00:44:08.940]get the experts together to hopefully come up
[00:44:11.130]with a consensus on what's best at this stage.
[00:44:15.930]Now, we had to sort of acknowledge
[00:44:17.301]there's gonna be iterations over time.
[00:44:21.600]And the first iteration that we had last year
[00:44:24.870]is gonna be improved on in future years.
[00:44:27.180]And that's why we continue to be
[00:44:29.280]in that steering group to improve it.
[00:44:30.960]So, I think it's good that the FAO can realize that
[00:44:37.557]we are not gonna have a perfect product in the first go,
[00:44:41.790]and then it's just a starting point that we can improve on.
[00:44:44.640]So go from that tier one to tier three level in the future.
[00:44:49.590]Yeah, so I think it's been quite successful,
[00:44:51.780]a lot of motivated people to get this out there.
[00:44:54.270]So I think it's worked quite well.
[00:45:02.190]All right.
[00:45:07.640]Thank you for the insightful presentation.
[00:45:09.150]I would like to ask you,
[00:45:10.680]in terms of building the tier one global nutrient maps,
[00:45:15.060]how do you consider the differences
[00:45:16.680]in agroclimatic zone soil condition,
[00:45:20.280]and also the management,
[00:45:22.170]and how far you can extrapolate the data that you see,
[00:45:25.440]like do field that have similarities or not?
[00:45:31.500]Thank you.
[00:45:32.333]Yeah, so at the tier one and two level,
[00:45:34.830]so tier one, I looked through the literature,
[00:45:37.620]and if someone said this represents the world,
[00:45:40.590]if I've used it to represent the world,
[00:45:42.270]then I used it as a value.
[00:45:44.700]And you could see the variation and values,
[00:45:47.340]and often it was quite hard to know exactly
[00:45:49.710]where they come up with that value.
[00:45:51.720]Sometimes they would refer to that same sources.
[00:45:55.110]And so when I was coming to collating those,
[00:45:57.564]it made it a bit difficult.
[00:45:58.440]So I really had to go back to maybe
[00:46:00.030]the secondary or tertiary source to see,
[00:46:02.820]oh, they're using the same source or slightly different.
[00:46:06.750]So at the tier one level,
[00:46:08.220]that was just purporting to represent the world
[00:46:12.090]and take it as it is,
[00:46:13.740]there's gonna be limitations in that,
[00:46:15.630]which I've said there's biases.
[00:46:17.790]At tier two level,
[00:46:19.860]they purported to represent a country or a region.
[00:46:23.730]So we're not getting down
[00:46:25.140]to agro ecological zones generally.
[00:46:27.450]There were maybe a couple that said
[00:46:29.490]this represents this general region.
[00:46:32.220]Then I'd have to say, "Okay, within this region,
[00:46:34.380]these countries are within that region."
[00:46:37.530]And then I'd get sort of averages and look at,
[00:46:40.680]okay, what values do we have from this country?
[00:46:43.710]And get averages at the tier two level.
[00:46:47.540]At the tier three level.
[00:46:49.080]I was then in the mixed-effects models.
[00:46:52.290]And the random forest models,
[00:46:54.810]I was using UN region as the random factor.
[00:46:59.670]So that's still quite high level,
[00:47:02.520]but it could be improved with agro ecological zones.
[00:47:05.790]But we're limited in the data that we have at this stage,
[00:47:09.780]getting it right down to a finer level, yeah.
[00:47:13.680]So Rana, I think about this.
[00:47:15.309]Before all this,
[00:47:16.500]we were using a physical efficient
[00:47:18.210]for the whole world, you know,
[00:47:19.805]but now we are moving into regional coefficient,
[00:47:21.660]so we are not even a country level,
[00:47:23.882]you know, so- Yeah.
[00:47:24.853]Your question I think is very rare, Rana.
[00:47:26.555]But it's probably, you know,
[00:47:28.972]an aspiration for let's say, I don't know,
[00:47:30.995]10 years then 15 years.
[00:47:32.305]Yeah.
[00:47:33.138]And that's where those prediction models hopefully come in.
[00:47:35.010]We can hopefully include accurate agro ecological zones,
[00:47:38.520]yield potential,
[00:47:39.630]which kind of encompasses that in those estimates.
[00:47:43.080]So, ideally once we get to the tier three level,
[00:47:46.320]we'll be accounting for some of those
[00:47:49.320]environmental aspects, yeah.
[00:47:51.330]Also genetics as well.
[00:47:52.500]We want to get down to maybe hybrids, non hybrids,
[00:47:56.040]functional groups, breed wheat, feed wheat,
[00:47:58.710]those sort of things.
[00:47:59.880]But at this stage, it's still pretty high level.
[00:48:02.820]But it's still far from perfection,
[00:48:05.267]but on the trajectory towards, okay?
[00:48:08.430]Yeah.
[00:48:10.140]Thank you, Cameron.
[00:48:11.133]It's a wonderful presentation.
[00:48:12.840]It was very insightful.
[00:48:15.120]I have a question, since you mentioned open data.
[00:48:18.750]From your point of view,
[00:48:19.770]what is going to be the reason
[00:48:21.540]for us as a scientific society
[00:48:24.240]to take that leap of faith
[00:48:25.560]and actually start sharing our data and codes?
[00:48:29.580]And how are we going to do,
[00:48:31.080]how are we gonna make that important step?
[00:48:33.270]And to be exact,
[00:48:34.770]I wanted to ask whether your codes for this,
[00:48:40.018]all these data that you gathered
[00:48:42.480]are available anywhere online?
[00:48:44.310]Thank you.
[00:48:45.480]Yeah, so, as I publish,
[00:48:48.360]I have a GitHub website that makes it available
[00:48:52.350]and they link to these Dryad repositories.
[00:48:55.740]So you can see the data that's been there.
[00:48:59.850]In terms of sharing the data,
[00:49:01.290]I mean, there's the legacy effective
[00:49:03.660]and the impact of your data.
[00:49:05.580]So beyond just the publication.
[00:49:08.610]And I accept that some people might not want to share that
[00:49:12.450]until they publish, that's okay.
[00:49:15.510]I've been sometimes waiting a few years
[00:49:18.090]for people to share their data for it to get published,
[00:49:21.090]is no problems with that.
[00:49:22.500]You've gotta be comfortable with sharing it.
[00:49:26.580]For the most part,
[00:49:27.413]I think people who have shared it have done so
[00:49:30.990]because they want to have that impact and legacy over time.
[00:49:34.950]'Cause otherwise, maybe it's gonna stay on the hard drive
[00:49:38.910]and will get lost in the future.
[00:49:41.580]And they see that often they've been paid
[00:49:44.940]through the public, through governments,
[00:49:47.160]or other institutions,
[00:49:48.780]and they see it as,
[00:49:50.820]in some countries it's a requirement,
[00:49:52.290]but also just a moral obligation
[00:49:54.600]to make that available to others.
[00:49:59.460]Yeah, and I suppose the benefit
[00:50:02.010]that you get from sharing it
[00:50:03.879]and not having to worry about the standardization process.
[00:50:08.250]So, some companies may not have the resources that we have
[00:50:12.960]with the company that we work with,
[00:50:14.050]to do the standardization.
[00:50:15.480]So you can send it in or standardize it,
[00:50:17.986]and it'll make it easier for others and yourself
[00:50:19.860]to actually work with it in the future.
[00:50:25.830]Thank you for your presentation.
[00:50:27.660]In terms of this repository for data sharing,
[00:50:32.100]are there conversations with other repositories, you know,
[00:50:35.700]in order to, you know,
[00:50:36.840]start also sharing the data among open repositories
[00:50:41.340]and in terms also of, you know,
[00:50:43.350]standards to record the data.
[00:50:45.983]And you know, if you can comment about?
[00:50:49.500]Yeah, so I have found data from some open repositories,
[00:50:53.910]and I make sure I ask them first and say,
[00:50:56.510]"Are you okay with it ending up in this repository?"
[00:50:58.950]And the important thing there is having attribution.
[00:51:01.440]So we have attribution back to who sent it,
[00:51:03.930]where it was published.
[00:51:05.040]You've got the DOI for the publication,
[00:51:08.100]so you're getting attribution that way.
[00:51:11.820]So that's been quite important.
[00:51:15.480]Yeah, and likes of, at the FAO,
[00:51:19.146]the articles that they described,
[00:51:23.010]the crop and nutrient budget,
[00:51:25.020]it shows where the data have come from.
[00:51:27.450]So it gets a bit of a summary,
[00:51:29.400]but it's a good point on understanding,
[00:51:31.504]yeah, where all these data are.
[00:51:34.380]Would like crop nutrient data,
[00:51:36.390]don't need to be a good place
[00:51:38.310]for everyone to put their data,
[00:51:41.070]but acknowledge that maybe for some people it's outside,
[00:51:43.800]a little bit outside the scope of what data they might have.
[00:51:49.320]Yeah, so, it's important we have good visibility
[00:51:52.410]of where all these data sets are.
[00:51:54.063]I've been mapping them to make it as easy as possible
[00:51:57.120]and sticking to the same kind of repository.
[00:51:59.880]So, I can just say look up my name in Dryad,
[00:52:03.030]and it makes it easier to see all the datasets that I have.
[00:52:07.680]Yeah.
[00:52:09.630]Zach, do we have any questions online?
[00:52:12.270]No questions online, okay.
[00:52:14.547]So, a quick question or comment.
[00:52:16.890]You show, Cameron, in one of your earlier slides,
[00:52:19.890]that the efficiencies for nitrogen and phosphorus
[00:52:22.770]at global level were going up over the past 20 years or so.
[00:52:26.970]I mean, if so, those are great news.
[00:52:29.490]Yeah. And how it's possible
[00:52:30.600]that every time I read the New York Times,
[00:52:32.580]there are bad news about agriculture?
[00:52:34.740]And I never seen any article talking about
[00:52:37.740]nutrients efficiency increasing globally?
[00:52:40.604]Why good news about agricultural don't show up
[00:52:43.680]in New York Times and in the media in general?
[00:52:46.553]What do you think?
[00:52:48.394]Well, bad news sells.
[00:52:49.440]That, but yeah, going back to at the start,
[00:52:52.680]some of the value of provoking data
[00:52:54.540]is that misinformation, we can,
[00:52:57.330]it's one way to counter it
[00:52:58.860]is having these open repositories to show over time.
[00:53:01.710]And hopefully, when we get this article published,
[00:53:04.890]so that's in submission at the moment.
[00:53:08.550]Yeah, we'll actually be able to show these trends over time,
[00:53:12.090]which is a positive news story.
[00:53:14.880]But these have been gradual changes over time
[00:53:17.340]and often those sort of things don't end up in the news.
[00:53:21.210]Hopefully it can make headlines.
[00:53:23.640]So that's not a good reason to share data,
[00:53:27.720]not just a citation, not just a development,
[00:53:32.040]but more about the vision of single CV,
[00:53:33.480]but also for justifying
[00:53:37.556]the importance of our discipline
[00:53:39.929]and promote, and also to make sure the good science
[00:53:44.460]lead to the right policy.
[00:53:45.870]So there are even major reasons why we should share our data
[00:53:49.320]that goes beyond our, you know, area around us.
[00:53:52.890]Yeah.
[00:53:53.723]The rest of questions there. Oh, okay.
[00:53:55.140]So, you will read them.
[00:53:57.810]So they ask, "How do you think
[00:53:59.880]that the projections you have worked on,
[00:54:01.830]as well as their availability to companies
[00:54:04.260]will influence the development of variable rate technology
[00:54:07.740]and inform the decision making
[00:54:09.150]for various sensing technologies
[00:54:10.890]to enhance nutrient use efficiency in the future?"
[00:54:23.100]Well, it will show people
[00:54:24.990]where some of the areas of the world are,
[00:54:28.030]where there isn't the same levels
[00:54:30.720]of nutrient use efficiency.
[00:54:32.670]So I showed the global averages starting to go up,
[00:54:36.660]but there could be others where it's not as great
[00:54:39.810]and that could allow people
[00:54:42.870]who are working in that technology space
[00:54:45.180]to invest in those areas,
[00:54:48.150]to use variable rate technologies
[00:54:50.910]to improve efficiencies there.
[00:54:53.040]So that's one way that could be market opportunities
[00:54:56.520]by looking at these hotspot locations
[00:54:58.710]where maybe nutrient use efficiency hasn't been increasing
[00:55:02.610]or it's been decreasing over time. Yeah.
[00:55:09.120]Any other question?
[00:55:09.953]We still have five minutes or so.
[00:55:15.690]Congratulations.
[00:55:16.523]So, you come and you make people stay in the room
[00:55:18.600]on a Friday evening.
[00:55:19.680]So that speaks very highly about your presentation.
[00:55:23.820]Thank you, Cameron,
[00:55:24.930]for your presentation.
[00:55:26.310]I have a couple of question.
[00:55:27.660]One is more about the methods,
[00:55:30.090]so I don't know if I catch you,
[00:55:31.380]but the nutrient balance does not include the losses,
[00:55:36.120]like leaching volatilization, for example, in nitrogen,
[00:55:39.270]is that correct?
[00:55:40.560]Yes, that's correct.
[00:55:41.460]So, it was a partial.
[00:55:42.480]So, the overall FAO crop of nutrient budget doesn't either,
[00:55:46.530]it doesn't include leaching.
[00:55:47.730]It does show I think, some estimates of some components,
[00:55:52.230]but it's not actually included in the balance.
[00:55:54.600]Whereas when I was showing
[00:55:56.190]the effect of tier one versus tier three,
[00:55:58.710]it was a very simple nutrient inputs
[00:56:01.110]as fertilizer minus the amount of nutrients
[00:56:05.310]taken away as crop products and residues.
[00:56:08.070]So it's just basic to see what effect
[00:56:10.740]the tier one versus three had,
[00:56:13.050]assuming all the other components stayed the same.
[00:56:15.570]So it's only partial, yeah. Excellent.
[00:56:17.550]And then, you are showing per crop and then per country.
[00:56:23.610]So would it be a change
[00:56:25.890]if you do it by cropping system in each country
[00:56:28.860]because the surplus, for example, in one crop,
[00:56:31.590]could be different depending on the rotation
[00:56:33.750]that you have in each country?
[00:56:35.610]Yeah, so understanding the whole system,
[00:56:38.299]that's different approach.
[00:56:40.650]For the FAO, it's very broad,
[00:56:42.630]it's all cropland,
[00:56:43.950]and you have to look up the definition of cropland.
[00:56:46.140]So that includes all these annual crops
[00:56:48.090]and I think even some forages
[00:56:50.640]that are renewed every up to five years.
[00:56:54.390]So it's quite a broad definition of cropland.
[00:56:59.220]So, it will have an effect,
[00:57:01.050]but we haven't got to the point
[00:57:02.250]where we're understanding the systems,
[00:57:04.200]we're just looking at a crop by itself
[00:57:06.600]with the nutrient budget.
[00:57:09.270]Yeah, it is scalable down to a lower level,
[00:57:13.140]down to the field level.
[00:57:14.820]But you're right in that,
[00:57:17.760]we haven't accounted for legacy nutrients
[00:57:21.600]for the next crop or from the previous crop.
[00:57:24.150]So that's a limitation of what we're doing.
[00:57:26.910]Okay, I have the last one.
[00:57:28.560]So, in different places you have some surplus, right?
[00:57:34.650]Would you say that that surplus is because there is more,
[00:57:37.350]for example, a gas emission or more losses,
[00:57:40.410]or could it be some cases where that surplus
[00:57:44.010]has to be with building more,
[00:57:45.720]let's say nitrogen in the soil?
[00:57:51.750]So, the big driver of the surpluses,
[00:57:53.490]I think are nutrient inputs.
[00:57:55.590]So the fertilizers and manure, in some cases.
[00:57:59.070]So particularly from China,
[00:58:02.070]I can't comment on where that's going,
[00:58:04.350]but you imagine with nitrogen converting to nitrates,
[00:58:08.130]and that there's probably gonna be a lot of leaching
[00:58:10.740]and nitrate oxide emissions happening.
[00:58:13.650]So,
[00:58:15.900]yeah, but we haven't focused on estimating that.
[00:58:19.020]That's something that could be improved.
[00:58:20.610]We're really just working on a mass balance system.
[00:58:23.850]We're not getting down to the level of,
[00:58:25.927]"Okay, what's happening in the soils?
[00:58:28.950]Is there more phosphorous or potassium levels
[00:58:32.970]increasing the soils?"
[00:58:34.500]Hopefully that's something we could do in the future,
[00:58:36.390]but at this stage, it's just the mass balance,
[00:58:38.370]and we can interpolate from that
[00:58:41.083]a loss of some sort.
[00:58:43.686]What that loss is, we're not sure, yeah.
[00:58:47.550]Yep.
[00:58:49.200]Also, I want to add to that answer if I may,
[00:58:51.780]that you know, in a way those maps revise you
[00:58:54.060]with a nice roadmap about where to target interventions.
[00:58:57.540]Because it's not like the surplus is the same
[00:59:00.120]across all places, but rather, you know,
[00:59:01.530]you can, I mean you can see on that map
[00:59:04.290]that overall, you know,
[00:59:06.720]agriculture is doing pretty well
[00:59:07.950]and the hotspots are two or three countries.
[00:59:10.530]So, you don't really need to, you know,
[00:59:12.810]cut down natural, you know, all over the world.
[00:59:14.250]You really to focus on two or three places.
[00:59:15.960]And by doing so,
[00:59:18.240]you will get the greater return to your investment.
[00:59:20.670]So I think that at the end of the day,
[00:59:21.870]this provides with a nice roadmap about, you know,
[00:59:24.330]how you should prioritize intervention
[00:59:26.910]to do a much sound nutrient management.
[00:59:29.400]In some places you will need to add more,
[00:59:30.960]like in Africa.
[00:59:32.040]Perhaps in some, not perhaps for sure,
[00:59:35.580]while in other cases you will probably need to cut
[00:59:38.370]like it is the case of China.
[00:59:40.380]But anyway, I think it's about probably a roadmap.
[00:59:43.470]I don't know if you can, you know,
[00:59:45.360]I won't go too much deeper into the interpretation
[00:59:47.310]of the exact numbers
[00:59:48.143]because that that's a little bit, I would say risky,
[00:59:50.580]and requires a lot of finance expertise
[00:59:52.410]to make sense of those numbers.
[00:59:56.010]Would you agree with that?
[00:59:57.300]Yeah, yeah, Technically?
[00:59:58.320]Yeah.
[00:59:59.348]In light to prioritize those countries,
[01:00:00.990]I mean, China's a big thing.
[01:00:02.760]There's big uncertainties in some of the inputs though,
[01:00:06.660]that we have to be aware of.
[01:00:09.510]But okay, given the data that we have,
[01:00:13.860]there seems to be positive news emerging
[01:00:18.900]that maybe the surpluses aren't increasing
[01:00:22.020]as much as they have.
[01:00:22.920]Maybe they're plateauing.
[01:00:25.980]But we've gotta think about the uncertainties
[01:00:28.680]in those data, which are quite substantial.
[01:00:30.840]So,
[01:00:32.580]yeah, I'll be interested to see the latest results
[01:00:34.560]that'll be coming out
[01:00:35.730]and whether that trend sort of continues
[01:00:38.010]with the plateau.
[01:00:39.480]But you know, like Patricio said,
[01:00:42.720]in terms of global surpluses,
[01:00:45.720]we have to think about China basically.
[01:00:48.180]That's a big driver.
[01:00:50.070]Don't quote me. (Patricio and Cameron laugh)
[01:00:52.200]All right, and on a separate note,
[01:00:54.840]I want to say that Cameron drove
[01:00:56.340]all the way from St. Louis to here
[01:00:57.960]just to visit us.
[01:00:59.160]So, thank you very much, Cameron,
[01:01:01.110]to be willing to drive so many miles
[01:01:04.050]to make it here for seminar.
[01:01:05.280]We really appreciate it. Yeah.
[01:01:06.799]And I think that this is another great example
[01:01:08.910]about collaboration between UNL
[01:01:10.657]and other organizations,
[01:01:12.300]trying to provide good science that can help
[01:01:16.140]make this world a better place
[01:01:17.490]to live for us and for our kids.
[01:01:19.860]So again, thank you very much
[01:01:20.880]and let's give Cameron another round of applause.
[01:01:24.124]Thank you. (attendees applauding)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Open data for improved cropland nutrient budgets and nutrient use efficiency estimations

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments