One sheet with one row per area answers it, and the shape of that sheet is not a matter of taste. United States federal regulation defines a conservation plan as the document that “Describes the conservation system applicable to the highly erodible cropland and describes the decisions of the person with respect to location, land use, tillage systems, and conservation treatment measures and schedules”. Location, practice, and dates, set down as the decisions of a named person. That is a table, and every farm that conserves soil is already producing the content of it and losing it.
Somebody built the terraces on the place. The road at the bottom was moved after a rain that took out a culvert. One area kept its cover because nobody managed to get a machine into it in time, and it turned out to be the right thing. None of it was written as a decision. When a gully opens, the farm reacts, hires the machine and closes it; when no gully opens, the subject disappears for two years.
The terrace on the middle field was cut by a machine crossing it three years ago and nobody rebuilt it, because there was no list saying where the terraces are or who answers for them. Then a buyer whose contract carries a sustainability clause asks whether the property has a conservation plan, somebody assembles a document in two afternoons out of photographs of things that already existed, and the document names neither what will be done nor who does it.
Five columns close that gap: the field, the conservation work decided for it, the deadline with month and year, the name of the person answerable, and one word at the end, done or not done, written when the deadline expires and never before. Works already standing get their own rows, opening with what is there and the date somebody last walked it. Half a day in the field with whoever knows the place, plus an hour of writing, done before the rains rather than during them.
What three legal systems agree the page has to say
Area, practice, in writing. Three legal systems written in different decades landed on those same three, with Uruguay requiring the plan filed in advance for the production system as projected, Brazil requiring the protection works executed before the land is farmed and then described in writing, and the American definition naming a fourth item outright, the schedule. That the three converge on the same contents is the strongest evidence available that this is the shape of the document and not one country’s paperwork habit.
Uruguay put it in a decree that reads almost column for column like the sheet described above. Under Decreto N 405/008, regulating a 1981 law, the ministry will require a Plan de Uso y Manejo Responsable del Suelo to be filed, setting out that the production system as projected produces tolerable erosion, taking into account the soils of the property, the sequence of crops and the management practices. Soils of the property, sequence of crops, management practices, all in a paper filed beforehand. That is Uruguayan law and it binds nobody outside Uruguay.
Brazil got to the same place from the money side. A 1975 federal law provides that in regions the ministry has designated, land may only be cultivated, or exploited economically in any other way, after prior execution of soil protection and erosion control plans. That is Brazilian law, and it bites only where a region has been designated.
| Where | What the document has to state | Who has to sign it or receive it |
|---|---|---|
| United States, for land classified as highly erodible | Location, land use, tillage systems, treatment measures and schedules | The local soil conservation district, in consultation with the local committees and the Natural Resources Conservation Service |
| Uruguay | Soils of the property, sequence of crops, management practices, and that the projected system yields tolerable erosion | Filed with the Ministerio de Ganaderia, Agricultura y Pesca |
| Brazil, in designated regions | The soil protection and erosion control system actually used | An agronomist, whose certificate the lender then requires |
Read the middle column across and the right column down. The middle column is the same document three times. The right column is three unrelated answers, and that is exactly why the sheet this article asks for is not any of these three. It is the farm’s own list, which is where the middle column comes from in every one of the three countries, and it is the only one that exists on a day when nobody has asked for anything.
Why “we conserve here” is not the same as a row on a page
Because what people believe about the persistence of a practice sits well above what the ground shows, and somebody measured the gap. Ferraro and eight colleagues, reporting in PNAS in 2025, used a state survey in Indiana in which staff drive a fixed route and record the fields on both sides of the road, and matched it against what the state’s own experts predicted. Their finding, in their words, is that “persistence is low and contrasts sharply with the predictions made by Indiana conservation experts”.
The direction of the miss is worth carrying. More than a hundred Indiana conservation experts were asked how many of the fields that had cover crops in 2014 would still have them over the next five years, and the middle guess was that most of those fields would carry cover crops in at least three of the six years, while the field data showed that fewer than half did.
On the harder question, cover crops in every one of the years rather than in most of them, the miss is of another order, and it is a good part of what pulled the whole prediction up: the middle expert expected half of those fields to carry cover crops every single year, and the field data found one in twenty-five.
One state, one practice, and fields counted from a moving vehicle on a fixed route, so what was found there is Indiana’s and is nobody else’s. What travels is the direction of the error and where it comes from. The same paper checked the national form on which producers report their own acreage each year and found the same low persistence in the self-reports, gathered a completely different way. What runs high is the belief, and a belief is not a record. Neither is a form filled in once a year for somebody else’s file, because it carries the acreage and not the decision behind it. The row with a date on it carries both.
The column that is only filled once the deadline has passed
One word, done or not done, written on the day the deadline expires. Writing it earlier turns the sheet back into a plan, and a plan is the half of this that farms already have in their heads.
Technical guidance carries the same obligation to keep looking. The FAO Voluntary Guidelines, approved by the FAO Council in December 2016 and written to apply everywhere rather than in one country, open the section on minimizing soil erosion by repeating what the Status of the World’s Soil Resources report had identified, “soil erosion by water and wind as the most significant threat to global soils and the ecosystem services they provide”, and then list the measures as management choices rather than as legal duties.
In the measure for sloping land, the phrase that matters for a status column is “terrace construction and maintenance”, one item covering both. A sheet with no status column records the construction and loses the maintenance, which is precisely the half that a broken terrace on the middle field belongs to.
When the answer at the deadline is “not done”, the row does not get erased and does not get quietly moved. Write the date, one line saying what stopped it, the new deadline, and the name again. Three consecutive “not done” entries on the same row with the same name is information about the row, and it usually means the work was assigned to somebody who was never going to have a machine free in that month. That reading is only available to a farm that kept the failed entries.
The works already standing get a row, with the date somebody last walked them
Because “how many years has this been in use on this area” is an answerable question, and the answer lives in a record rather than in anybody’s memory. Rees, Ineichen, Finger and Grovermann built a dataset for Scientific Data in 2025 in which they surveyed thousands of Swiss arable farms in late 2023 and early 2024 and captured, for each holding, “detailed records of twelve arable soil management practices, including uptake extent, number of years used”.
Those are Swiss farms and the survey was built to become a research dataset, so nothing in the responses describes anybody else’s ground. What travels is the shape of the question. Somebody put that question to farm after farm and got answers back, which means it is answerable at farm level, by whoever holds the record. Inside one property it has exactly one place to live: the row for the work that already exists, carrying what is there and the date it was last inspected.
That inspection date is what turns an old terrace from history into a line item. It also feeds the rest of the field-level paperwork, since the areas named by a compaction check with a date on it are the first candidates for a row here, and the field names in the first column have to be taken from the same list that carries a declared sampling interval at the top of it. Two different lists of fields on one farm is how a document stops being checkable by anyone.
Who signs the sheet, and why it is not the plan an office asks for
Whoever answers for the work on that area, which is a person on the farm and is not the person who signs the official document in any of the three countries above. The United States requires the plan to be “approved by the local soil conservation district” in consultation with local committees and the Natural Resources Conservation Service. Uruguay requires it filed with the ministry. Brazil requires a certificate issued by an agronomist, and requires that the certificate state the specifications of the soil protection and erosion control system used by the interested party. Three jurisdictions, three different answers to who has to put a name on it, and none of them transfers across a border.
That divergence is the reason this article stops where it does. Which practice belongs on which area, how a terrace is sized, and how any of these documents gets filed are questions for the reader and for the professional who carries the technical responsibility in the reader’s own country. What does not change with the border is the column: somebody on the property is answerable for the work on that area, and if the sheet does not carry a name, the work is assigned to the weather.
The same reconciliation of decided against executed is already running on the input side of the farm, where the question is which product went out and the soil test the recommendation rests on says why. Conservation work is that operation on a longer clock, with the decision and the execution sometimes years apart, which is the only reason it needs its own page instead of a column on somebody else’s.
The day somebody outside asks
Money asks first, and it asks in writing. Brazilian law ties the two together without hedging: any application for crop or livestock finance on land where those plans are required may only be granted by a lender if accompanied by a certificate proving that execution. The document is not decoration on the loan file, it is a condition of the loan, and it certifies execution and not intention.
Requirements of this kind also arrive slowly and then all at once, which is the part farms misjudge. Uruguay’s ministry records the sequence plainly. A pilot stage began in September 2010, and then the mandatory filing phase started in April 2013, staged by area and by production system, with dairy operations brought in afterwards.
Staging is how a ministry rolls out a rule, and it says nothing about which farm needs a written record of its own decisions. Brazil’s version of the same requirement keys on the region and on the type of land, never on the size of whoever works it, and the reason is the one under this whole article: an area that erodes does not consult a title deed first.
Two afternoons of photographs is what a farm produces when the question arrives before the sheet does. A sheet kept from the year before produces something else, which is a list of decisions with dates, names and outcomes attached, and that list is worth more to the person who wrote it than to the person who asked for it. That is the check step of farm management applied to work that happens outdoors, over years, and the axis it sits in is the whole of production.
Where to start
Half a day walking the property with whoever has been there longest, plus an hour at the table. One page, one row per area, five columns, a date and a signature at the top.
The gap this sheet closes is not the one between a farm that conserves and a farm that does not. It is the one between a work that was decided and done and a work that was never decided at all, and once the grass has grown over both of them they look the same from any distance. The person who walks the property in five years will not be the person who wrote the rows, and on that day the rows are the only thing that will tell those two apart.