Why the target is not a snapshot of the past, and what it is built from instead.
Two photographs, taken on nearby comparable ecological sites. The first records pre-treatment condition on former cropland under corn residue. The second records a seeded stand at year 2. Neither is a reference model.
A reference model describes a condition. Under restoration standards it sets out the range a system moves through while in good condition, rather than fixing it at a single point. A healthy ecosystem keeps moving, so a target built on one moment specifies something the site clears only occasionally, including in years when the restoration is going well.
Usage varies. The frameworks a single project answers to define the term two different ways, and each definition carries its own consequences. The difference decides what a recovery target can require, whether it is reachable, and what a monitoring report measures against.
What a reference model describes
The range is the point.
A reference model sets out the conditions a system occupies while it is in high-integrity condition, which means functioning across its full set of attributes rather than simply sitting undamaged. A grassland in a wet decade and that same grassland in a dry one are both the system working. Choose either one as the target and you have described half of it.
The model also carries time forward. It describes what a site supports under the conditions it will actually experience, with degradation stripped out and change left in. A model built this way stays usable as the climate shifts because it anchors to a condition rather than to a year.
Two frameworks, one term
The term travels, though. An Ecological Site Description also defines a reference state, one anchored to the site’s historical range of variability. That document classifies: it tells you what a given site is capable of producing and how it moves between states. A restoration reference model directs: it tells you where one project is headed. One maps potential. The other sets the course. A target stays both reachable and checkable when the two are kept apart.
What a site records
The case for a range also comes from the ground itself.
Inside restoration ecology, the single historical condition has been under challenge for years. The stronger position treats a reference as a succession of states rather than one idealized picture, on the grounds that a target built from a single moment is a target built from the thinnest evidence available.
Walk a site and you will see why. Remnant vegetation, soil profiles, and land use history each record a span rather than an instant. Soils hold the longest and slowest account. Remnants hold the current one, and they are the closest thing to a live reading you can get. Land use history fills the middle. Read together, they describe what the site sustains across a range of conditions.
A model assembled from converging lines of evidence is easier to defend than one assembled from a single photograph or somebody’s recollection.
Why the model gets written first
A reference describing a range has a weak point. Widen the range far enough and every outcome fits inside it.
The concern is well established in the field. A target that can flex after the fact is a target that can be challenged after the fact, and a challenged target is an expensive one. The same applies to a ceiling set low because a site looks past saving. If that judgment rests on cost rather than on ecology, it will be contested the first time someone tests it.
Both risks point in the same direction. A model that stretches to accommodate whatever the site produces is a description wearing a target’s job title.
Two practices hold the line.
Write the model before the outcome is known. A range set in advance constrains what can be claimed. A range set afterward only explains what already happened.
And write it as measurable attributes rather than as narrative: absence of threats, soil and physical conditions, species composition, vegetation structure, ecosystem function, connectivity and external exchanges. Each one carries a value. A narrative can absorb a disappointing year without much trouble. A number records it.
What the model is built from
Four lines of evidence, all measurable.
The ecological site potential of the soils present. Soils, landform, geology, and climate combine to define what a given piece of land produces and how it responds when someone manages or disturbs it. That combination is the ecological site, and it sets the outer bound on the answer before anyone opens a seed catalog. The soils on a property have already decided a great deal.
Remnant vegetation on or near the property. Whatever persists nearby shows what the site supports under the conditions it faces now.
High-integrity regional reference sites on comparable ecological sites. Plural, deliberately. Several observations describe a range. One describes a place.
An adjustment for projected climate.
Setting the ceiling
The model and the target do different jobs, and separating them changes how a project gets written.
The model describes a high-integrity state. The target is how close a given site can get to the model, once you account for what the land will be used for. A graded footprint under active operations sits under a lower ceiling than an intact remnant under long-term stewardship, even when both are measured against the same model. Setting that ceiling at the outset is what makes it defensible three years later.
Setting that ceiling when the plan is written keeps the specification honest. Set it when the monitoring report comes due and the number moves to meet the result.
The model is also the monitoring standard
The attributes that define the model are the same ones the monitoring measures. That is the practical payoff. The metrics were set when the plan was written, so year three measures against them rather than inventing a new set.
Each attribute gets a reading. How close the site is to the reference on species composition. On vegetation structure. On soil condition. Take the reading again a year later and it has moved, one direction or the other, by an amount someone can put in a table.
The year five report states three things. Where the site is. Which way it is going. How far is left. That is more work than writing a narrative about how the planting is coming along, and answers the question a client, a regulator, or a lender is going to ask: how far along is it?
Sources
Gann, G.D. et al. (2019) International Principles and Standards for the Practice of Ecological Restoration, 2nd edn. Society for Ecological Restoration.
Gann, G.D. et al. (2026) ‘International principles and standards for the practice of ecological restoration, third edition’, Restoration Ecology.
Interagency Ecological Site Handbook for Rangelands (2013). USDA Natural Resources Conservation Service, USDI Bureau of Land Management, and USDA Forest Service.
Balaguer, L. et al. (2014) ‘The historical reference in restoration ecology: re-defining a cornerstone concept’, Biological Conservation, 176, pp. 12–20.
Murcia, C. et al. (2014) ‘A critique of the “novel ecosystem” concept’, Trends in Ecology & Evolution, 29(10), pp. 548–553