The Land Sector and Removals (LSR) Guidance Just Landed. Here's What It Means for Food and Beverage

How the inventory fits together

The basic architecture explains most of what follows. Land emissions break into three categories: land use change, land management biogenic CO2 emissions, and land management production emissions. Set apart from all three, in its own category, sits land management CO2 removals. Entirely outside the land sector sit fossil fuel and industrial emissions. These categories are reported separately and are never netted against one another.

Land use change covers the conversion of forest or another native ecosystem, assessed over a 20-year period for most crops (longer for a crop like palm oil, which stays in the ground longer). It also covers land occupation, a mandatory metric measured in hectares for both Scope 1 and Scope 3. The calculation method depends on data availability. There are three tiers: dLUC, where the specific farm is known and data has been collected directly; jdLUC, new in the final Standard, where the crop and region are known and a crop-specific map exists even without farm-level traceability; and sLUC, the fallback when no crop map exists and sourcing is only known at a jurisdictional or global level.

Biogenic CO2 emissions result from tillage, overgrazing, biomass loss, residue and irrigation changes, the draining of organic soils, and fire or storm losses, each representing a step down in carbon stock over time. Zero biogenic emissions can only be assumed under a narrow set of conditions, which most operations will not meet.

Production emissions occur at the farm gate: methane from cattle and rice, nitrous oxide from fertilizer and manure, fossil CO2 from lime and urea. These are reported gross, and sequestration does not net them down. On-farm fuel combustion sits outside this category for Scope 1 reporting, since burning fuel is classified as fossil fuel and industrial emissions. Further down the supply chain, where farm-level visibility doesn't exist, fuel use can be folded into land emissions, provided nothing is counted twice.

CO2 removals carry their own designation, since this is the first standard to properly address the category. Removals are always calculated independently, with no netting against emissions. Reporting them is optional, but they remain one of the largest available mitigation levers. And the category is data-intensive: high granularity is required, a meaningfully higher bar than most companies' existing sustainability data can currently clear.

‍

Traceability determines what can be claimed

The level of supply chain traceability a company has determines what it is permitted to claim. At one end sits global or jurisdictional sourcing, which limits reporting to sLUC and conservative averages. At the other sits farm- or parcel-level traceability, which unlocks the more specific dLUC and jdLUC methods and, eventually, removals claims. Physical traceability is required to make that jump. Mass balance is allowed, with guardrails. Identity-preserved sourcing is allowed. Book-and-claim is explicitly excluded.

That exclusion carries weight. The certificate-based approaches many companies have relied on for supply chain claims will not satisfy this guidance. Knowing that cocoa comes specifically from Ghana permits use of Ghana-specific values; without that origin data, the only option is a global default, typically a far less favorable number.

The practical shift is threefold. Suppliers will be asked for considerably more data, since generic global averages no longer suffice. Supply chain visibility starts paying off directly, because the way a product is tracked becomes the proof behind the reported numbers. And procurement and sourcing functions become far more central to sustainability reporting, since none of this can be produced in isolation from where the commodities actually come from.

‍

What counts as a removal

A land management removal is defined as a net CO2 removal resulting from increases to carbon stored in land-based pools, driven by ongoing land management practices. All such removals originate from biological sinks: carbon returning to the ground, not simply less carbon leaving a tailpipe or a fertilizer application.

The distinction between a removal and a reduction is a common point of confusion, and a concrete example clarifies it. Improved crop rotations or reduced tillage build organic matter and limit soil disruption. When that raises the carbon stock above baseline, the net increase, once quantified with uncertainty and monitored over time, qualifies as a removal. Compare that to applying fertilizer in smaller, split doses timed to crop needs rather than one large application. That practice reduces nitrous oxide emissions, which has real value, but it is a reduction, not a removal. Only the soil organic carbon gain qualifies as a removal; reductions are reported on a separate line entirely.

Because the bar for removals sits well above standard emissions accounting, the Guidance introduces three specific new requirements. First, an empirical data requirement: direct measurements, or measurements used to calibrate models and remote sensing, are required to quantify land carbon stock changes. A database lookup does not satisfy this. Second, quantified uncertainty: every removal value requires a confidence interval and justification that it does not overestimate the actual removal. Third, resampling at an interval of five years or less, either to remeasure directly or to keep a model properly calibrated.

‍

Three key ways to quantify soil carbon

Once a company commits to pursuing a removals claim, it still has to select a measurement approach. Three main paths exist, each with distinct tradeoffs.

Samples only involves sampling, then re-measuring within the same representative strata, where repeated measurement over time yields the net carbon stock change directly. It is the most direct form of measurement, which would hypothetically give  it the highest confidence, but the sampling burden is often so great that it is not feasible to use for large projects. It suits very small projects, where sampling the entire area is affordable and there is little need to scale or transfer the method elsewhere. A pilot project may be able to  absorb that cost; a large, multi-region supply shed generally cannot, since cost scales poorly as project size grows.

Biogeochemical modeling, uses a  process-based approach to simulate soil carbon dynamics over time using a model calibrated with soil samples alongside management and climate inputs. It requires meaningfully fewer samples than direct measurement, mainly for calibration and periodic recalibration, but it does depend on additional primary data like climate inputs and practice information sourced from growers, to run and keep the prediction's uncertainty in check. Costs range from low to moderate and scales with calibration effort and project size. The approach performs well where the geography, practices, and crop or land use in question already have a well-established, available model, but availability and performance can be a challenge when stepping into crop types or novel practice changes the model wasn’t built for.  

Digital soil mapping estimates soil carbon through a combination of soil samples, machine learning, and biogeochemically-informed predictors derived from remote sensing. Sample requirements are the lowest of the three approaches, because the digital soil sampling model can rapidly learn the relationships between carbon and other remotely sensed factors — enabling it to learn more from every sample and benefit from existing sample databases. Both sampling density and uncertainty decline as project area scales, and no additional primary data collection from growers is required beyond the project area and sample data itself. Cost is low at scale, thanks to the low sampling burden. The approach is outcome-based, practice-agnostic. It works across a wide range of practices, geographies, crops, and land uses, and is particularly well-suited for projects planning to scale or projects spanning complex, fragmented supply chains where multiple modeling partners would otherwise be needed. Digital soil mapping can also be leveraged to improve performance of other approaches — for example to inform more cost-efficient sampling design or as inputs to improve a biogeochemical model's performance, meaning the approaches are not mutually exclusive.

The process of running a digital soil mapping project follows a fairly consistent arc. First, define the project area to be quantified for removals. Then collect a targeted set of soil samples as ground truth. Next, calibrate and validate the model against that data, alongside existing soil datasets, to produce a soil organic carbon stock value with uncertainty across the full area. Finally, present results in a format suitable for whatever needs to be documented. As future measurement events occur or as a project scales, uncertainty and sampling density shrinks.

‍

Questions worth answering before starting a removals program

A few questions are worth working through honestly before committing resources; 

Which crops and geographies would actually have the largest impact on removals? Attempting removals everywhere at once is a fast way to spend budget without producing a defensible number; deep-rooting crops that can be grown with minimal tillage tend to offer the strongest return, both financially and in climate terms. In addition to looking at crops and practices, it’s also worth considering the sequestration potential of the land itself. (Perennial can help you determine the ideal region to start removals, given your unique supply chain.)   

Does the necessary contact and data collection relationship with suppliers or growers exist? These processes require far deeper visibility into a supply chain than most legacy sourcing relationships provide. 

What is the realistic budget, timeline, and level of organizational buy-in behind the effort? Plenty of removals programs launch with ambitious goals and no realistic resourcing plan, and those programs tend not to survive. Working through these questions early is what separates a program that produces a credible number from one that stalls. 

‍

Getting ready before 2027

The Guidance takes effect January 1, 2027, with additional supplements for forestry and specific crops still to come. A reasonable readiness checklist for the year ahead includes mapping traceability by commodity and volume, and identifying early which commodities could realistically support a removals claim. It also means checking existing certifications against what the Guidance actually accepts, requesting yield data from suppliers, since land use change reporting requires disclosing the yield factors used in the calculation, and establishing a baseline read on land use change risk.

None of this happens quickly, which is the point of building a roadmap. 2027 is meant to be the year companies bring their data up to the traceability level required, so that when 2027 emissions are reported in 2028, there is something credible behind the number.

‍

What granular data actually produces

It is worth grounding this in what the output looks like once the underlying data exists. Consider a hypothetical batch of climate-smart yellow corn grown under strong land management practices. Its cradle-to-farm-gate carbon removals might land around -0.25 kg CO2e per kilogram, roughly an 89 percent reduction against the gross footprint. That figure appears as its own line in a GHG inventory, positioned beneath the gross emissions subtotal rather than netted into it, and again as its own line in a target annex, where emissions and removals are reported against separate targets. Both figures remain visible, additive to the overall accounting, and clearly separated from the gross emissions number. That is the entire purpose of the structure described above: when farm-level practice data exists, a removal becomes visible, defensible, and something a company can actually book.

‍

What the new LSR Guidance signals for the future of carbon accounting in the F&B industry

The clearest way to read this moment is as a marker of maturity for the industry's accounting systems. There has been considerable anxiety about what the Guidance requires, but read closely, its structure resembles a choose-your-own-adventure more than a strict mandate. Without granular data, there is still a path forward; a company simply forgoes the ability to claim removals. Carbon accounting can still proceed. Targets can still be set. For companies willing to invest in the traceability and measurement rigor required, there is now an agreed, defensible path to claim removals as well.

That is the real shift. The industry now has a path that is agreed upon and defensible, not necessarily perfect, but built on rules that can be worked from at whatever depth a company chooses. Financial accounting did not standardize globally until the mid-twentieth century, and business existed for a long time before that. Carbon accounting remains early by comparison. More revisions will come, more clarity will be needed on edge cases, and open questions like the ones raised above will persist. But for the first time, food and beverage companies serious about land-based removals have a shared rulebook instead of a dozen competing interpretations.

‍