Industries / Food, farming and forestry / Agriculture and farming / Ranching and livestock
Grazing and feed
Finding feed for animals that live on grass: fencing and sharing range, planting and resting pasture, and measuring how much forage there is.
6
Milestones achieved, from 1874 to 2018.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1874
Barbed wire closes the range
The US Patent Office issued Joseph Glidden patent No. 157,124 for barbed wire on 24 November 1874. Cheap wire fences let farmers enclose land that cattlemen had used as open, shared range, and the US National Archives says barbed wire "meant the end of both the open range and the freedom of the rancher and cowboy".
LessonOpen range had no limit on numbers: overgrazing on the remaining public land, then the harsh winter of 1886–1887, made large open-range cattle operations disappear (National Archives).
Credit
Joseph Glidden
1934
Rules for the public range
At the request of western ranchers, Congress passed the Taylor Grazing Act of 1934, which set up grazing districts where the government divides up and regulates grazing. The Bureau of Land Management (BLM) now manages grazing on 155 million acres (about 63 million ha) under nearly 18,000 permits and leases. In 2026 it charges $1.69 per animal unit month (AUM): the forage one cow and her calf, one horse, or five sheep or goats eat in a month.
Credit
Representative Edward Taylor of Colorado; administered by the BLM
1984
Marandu grass for Brazil
Embrapa, Brazil's public farm research company, released Marandu, a variety of the African grass Brachiaria brizantha, from its beef cattle research center in Campo Grande in 1984. By 2018 it was the most widely grown pasture grass in Brazil, according to an article on Embrapa's library site.
LessonSown pasture runs down without upkeep: Embrapa says about 60% of Brazil's 164 million ha of pasture (MapBiomas, 2024) shows some degradation.
infoteca.cnptia.embrapa.brinfoteca.cnptia.embrapa.brinfoteca.cnptia.embrapa.br
Credit
Embrapa (EMBRAPA-CNPGC, now Embrapa Gado de Corte); release paper by S. G. Nunes, A. Boock, M. I. de O. Penteado and D. T. Gomes
2008
Rotational grazing put to test
Rotational grazing moves a herd through a series of fenced paddocks so each rests between grazings, and it was widely promoted as the only sound way to graze. David Briske and eight co-authors reviewed the rangeland experiments and found plant production equal or greater under continuous grazing in 87% (20 of 23) of them, and animal production per head equal or greater in 92% (35 of 38).
LessonThe authors concluded that advocacy for rotation rested on "perception and anecdotal interpretations", and that results depend on how well a system is managed rather than on the system itself.
Credit
Briske (Texas A&M University) and colleagues, Rangeland Ecology & Management
2017
Fences drawn on a phone
A virtual fence is a GPS collar that plays a sound when an animal nears a boundary drawn in a phone app, then gives a mild electric pulse if it keeps going. By the company's account, Norway's Food Safety Authority accepted the Nofence collar for goats at the end of 2017; it is now approved there for sheep, goats and cattle, and Euronews reported more than 7,000 farms using the collars in 2025. Outside researchers have found no sign of extra stress: a 2022 Swedish study found no rise in the stress hormone cortisol compared with physical fences, and a 2025 doctoral study of sheep at the Norwegian University of Life Sciences judged the system no more stressful than an electric fence once animals learn the sound.
LessonThe collars depend on batteries, GPS and mobile signal, which can drop out in remote country (Lotten Valund, RISE Research Institutes of Sweden); Sweden banned virtual fences in 2018 over welfare concerns and said in October 2025 it would allow them for cattle and sheep.
Credit
Oscar Hovde Berntsen, who founded Nofence in 2011
2018
Rangeland mapped every year
Matthew Jones, Brady Allred, David Naugle and colleagues combined Landsat satellite images back to 1984 with more than 30,000 field plots to map, every year and at 30 m detail, how much ground grasses, shrubs, trees and bare soil cover across the western US. They call it the first such annual map at that detail, and the Rangeland Analysis Platform gives it away free. A 2022 analysis of every BLM grazing allotment with these maps found that annual plants now outweigh perennial ones on nearly 52 million acres (about 21 million ha), which the BLM says points to more frequent and intense wildfires.
LessonThe maps depend on ground truth: the BLM calls new field data "a precious commodity" for training and checking them.
Credit
Jones, Allred, Naugle and colleagues (Ecosphere, 2018); the 2022 trend study by BLM, USDA Agricultural Research Service and University of Montana scientists
Watch for
Reviving Brazil's worn-out pastures
Brazil's Caminho Verde (Green Path) program aims to restore 40 million ha of degraded pasture in ten years. At COP30 in November 2025, agriculture minister Fávaro said 3 million ha had already been recovered with loans from the Renovagro credit line, while Embrapa says statistics on pasture degradation are "highly variable and inconsistent".
Achieved whenan independent survey, from satellite images checked against field visits, confirms the recovered area and that it stays productive.
Claimed by
Brazil's Ministry of Agriculture, November 2025
Collars that rarely shock
Nofence says animals learn its warning sound within a few days, and that after training up to 96% of their boundary encounters end with the sound alone, without an electric pulse.
Achieved whenan independent study of commercial herds publishes the share of boundary events that end in a pulse, across several farms and seasons.
Claimed by
Nofence, October 2025 (reported by Euronews)
Open
01
Permitted grazing against forage grown
An open table and map that sets, for each BLM grazing allotment, the grazing the BLM authorizes (in AUMs) beside the forage the land grew each year, as estimated by the Rangeland Analysis Platform, from 1991 to 2025. Publish the conversion from forage to AUMs and every assumption, such as the share of forage animals may take, so others can change them and rerun the table. Work at allotment level and leave out ranchers' names.
The briefCredit · unclaimed02
Virtual fence brands compared
A season-long test of two or more commercial virtual fence systems on the same ranch, with similar groups of cattle on similar pasture, where the systems are legal. Log how often each collar reports its position, gaps in signal, battery failures, lost collars, animals that cross a boundary, and pulses per animal, and publish the raw logs.
The briefCredit · unclaimed
Open 01 · the brief
Permitted grazing against forage grown
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open table and map that sets, for each BLM grazing allotment, the grazing the BLM authorizes (in AUMs) beside the forage the land grew each year, as estimated by the Rangeland Analysis Platform, from 1991 to 2025. Publish the conversion from forage to AUMs and every assumption, such as the share of forage animals may take, so others can change them and rerun the table. Work at allotment level and leave out ranchers' names.
Why it is still open
The 2022 study that mapped vegetation trends on all BLM allotments says "Grazing management was not analyzed in this paper; there are no findings or interpretations provided." Searches found no allotment-by-allotment comparison of authorized grazing with satellite forage estimates.
Built with what exists
the BLM's public Rangeland Administration System reports (Allotment Master and Authorization Use by Allotment), which give each allotment's authorized forage use; Rangeland Analysis Platform production data through Google Earth Engine or its {rapr} R package; the open code and cleaned allotment boundaries from the 2022 trends study (Andrew Kleinhesselink's BLM_rangelands repository on GitHub), which extract production by allotment but do not compare it with grazing; Python or R for the joins.
First step
Download the Allotment Master report for one BLM field office and that office's production data from the Rangeland Analysis Platform, and compute the ratio for a single year.
Done looks like
A public, reproducible table covering at least 90% of the BLM's roughly 21,000 allotments, with the code. For at least 50 allotments, the table's ratio is compared with the BLM's own rangeland health findings, and the share of high-ratio allotments that failed health standards is reported beside the share of low-ratio ones.
Open 02 · the brief
Virtual fence brands compared
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A season-long test of two or more commercial virtual fence systems on the same ranch, with similar groups of cattle on similar pasture, where the systems are legal. Log how often each collar reports its position, gaps in signal, battery failures, lost collars, animals that cross a boundary, and pulses per animal, and publish the raw logs.
Why it is still open
Kansas State University's August 2026 guide compares five vendors sold in the US (eShepherd, Halter, Monil, Nofence and Vence) only from their specifications, and tells ranchers to contact the companies. A researcher at RISE in Sweden warns that collars depend on batteries, GPS and mobile signal, which can drop out in remote areas. The University of Arizona's June 2026 comparison sheet of virtual fence vendors for cattle also lists only specifications and prices, and says components from different vendors are generally not interchangeable. Published studies found test one system at a time; no same-pasture comparison of brands was found.
Built with what exists
collars bought or leased from two vendors whose minimum herd sizes suit a small trial (K-State lists Nofence at five or more animals, Monil at six or more and eShepherd at 20 or more); each vendor's data export; a phone signal survey app; Zenodo to publish the logs.
First step
Ask two vendors for a research lease or permission to publish raw collar logs, and map phone signal across the candidate pastures.
Done looks like
Two systems, at least 20 cattle each, at least 90 days on the same ranch. The report gives, for each system, the share of expected position reports received, boundary crossings per 1,000 animal-days, pulses per animal per week, and collar failures, with all raw logs public.