Recommendation sources

Upload a soil test and Zone Forge turns it into a plain-language report card — worst nutrients first, color- and word-coded, with variability hints and a fertility trend across your past uploads. Where your state's lab method and a published land-grant source line up, each nutrient gets a graded verdict in that university's own rating words; where they don't yet, we honestly show the measured value and say why — a wrong grade is worse than an honest "measured." Behind the card sits a recommendation-source library that maps a published source for every U.S. state. The three core tables below — Iowa State PM 1688, Tri-State 2020, and the Penn State Agronomy Guide — are the most deeply read tables in that library for the Corn Belt and Mid-Atlantic, and you can pick the one that matches your lab and your county.

Iowa State PM 1688

"A General Guide for Crop Nutrient and Limestone Recommendations in Iowa." Iowa State University Extension and Outreach.

Region. Iowa.

Phosphorus extractant. Bray-P1, Mehlich-3 colorimetric, Mehlich-3 ICP or Olsen.

Potassium extractant. Ammonium acetate or Mehlich-3, with Dried or Field-moist and slurry handling required.

Lime engine. Exact SMP/Sikora buffer-pH cells from PM 1688 Table 16, selected by the printed target and incorporation depth. Unprinted coordinates are refused; PM 1688 does not supply a soil-Mg threshold for choosing calcitic versus dolomitic material.

Crops. Corn, soybean, oats, wheat, sunflower, corn/sorghum silage, and the named hay/pasture tables.

Strengths. Fixed category cells from Tables 3-13; philosophy settings are disabled and do not change the rate. High/Very High is zero except Table 10's High-P seeding instruction. The printed Optimum cells stay fixed. Harvested-yield footnotes are disclosed where printed; Table 13 bluegrass cells are approximate estimates.

Tri-State 2020

"Tri-State Fertilizer Recommendations for Corn, Soybeans, Wheat, and Alfalfa," 2020 update. Joint publication of Ohio State University, Purdue University, and Michigan State University.

Region. Ohio, Indiana, Michigan.

Phosphorus extractant. Mehlich-3 or Bray-P1 (both calibrations published).

Potassium extractant. Mehlich-3.

Lime engine. Buffer-pH cross-walk table — Tri-State's print form uses pH+CEC, but buffer pH is widely interoperable when the lab reports it.

Crops. Corn, soybean, wheat, alfalfa.

Strengths. Most recent published update of the three. Refined buildup-and-maintain ranges, best multi-extractant flexibility. Use this if your lab reports either Bray or M3 and you're east of the Mississippi.

Penn State Agronomy Guide

"The Penn State Agronomy Guide," annually updated. Penn State Extension.

Region. Mid-Atlantic — Pennsylvania, Maryland, Delaware, New Jersey, Virginia.

Phosphorus extractant. Mehlich-3 only.

Potassium extractant. Mehlich-3.

Lime engine. Penn State Mehlich Buffer / Exchangeable Acidity — the current Penn State AASL method (it replaced the older Adams-Evans buffer for Pennsylvania; Adams-Evans is still used for the Coastal Plain states DE/NJ/SC/FL/AL).

Crops. Corn, soybean, wheat, hay, and a broader vegetable and forage palette than the corn-belt tables.

Strengths. Most extensive nutrient palette (S, Mg, Zn, B, more). Use this for the Mid-Atlantic; lime calibration is most direct here.

Why phosphorus extractant matters

Bray-P1 and Mehlich-3 measure different fractions of soil phosphorus. They're correlated but not interchangeable. Typical relationships:

• Mehlich-3 reads higher than Bray-P1 by 10–30% on most soils.
• Olsen reads lower than both, especially on acidic soils.
• The relationship varies with soil pH, calcium content, and clay type.

If you upload Bray-P1 values to a Mehlich-3-calibrated table without converting, the engine will read your soil as poorer than it is and over-recommend P. The reverse — Mehlich-3 values fed to a Bray table — under-recommends. The drift is in the 20–40 lb P₂O₅ per acre range — real money per acre in over- or under-applied phosphorus at today's fertilizer prices. Zone Forge guards against this: it won't quietly grade your report card or build a rate from a soil test run on a method the chosen table isn't calibrated to — it shows your measured value and asks you to confirm your lab method (or opt into a published conversion) first.

Cross-extractant conversions

Zone Forge ships two published conversions you can opt into when your lab and table don't match:

• Mallarino 1995 (Iowa State) — all three phosphorus pairs: Bray-P1 ↔ Mehlich-3, Bray-P1 ↔ Olsen, and Olsen ↔ Mehlich-3, regressed on 240 Iowa soils and banded by pH. The tests read about the same on acid and neutral soils and diverge most on high-pH soils, where Bray-P1 reads low — so above pH 7.45 we refuse to convert Bray-P1 rather than carry that underestimate into a rate.
• OSU ANR-75 (Ohio State Extension, Tri-State OH/IN/MI) — a single-factor Bray-P1 → Mehlich-3 conversion, plus ammonium acetate → Mehlich-3 for K, Ca and Mg. It carries no Olsen calibration, so every Olsen conversion we ship is Mallarino's. We use ANR-75 as the fallback when a sample has no pH to band on.

Both conversions are published, peer-reviewed equations, not guesses — but re-testing with the right method is always the better answer when feasible.

The three philosophies

Build-model sources use one of three philosophies to turn soil-test values into a rate. Fixed category sources such as Iowa PM 1688 use the printed crop-category cell directly, with philosophy controls disabled.

Sufficiency

Apply only what the crop will remove this season — no buildup, no maintenance beyond the year's needs. Best for short-tenure or rented ground where multi-year soil building doesn't pay back to you.

rate = yield_goal × removal_factor

Build-and-Maintain

Bring soil-test values up to target over a configurable horizon (default four years), then maintain at target with crop-removal rates. Best for owned acres with a long-term plan.

rate = yield_goal × removal + (target − soil) × buildup_factor / years

Hybrid Default

True 3-zone system: build below the maintenance band (deficit-scaled — adjacent variable-rate zones get genuinely different rates), maintain at pure crop removal in the band (default target ± 10%), draw down above the band at a configurable aggressiveness (none / light / moderate default / aggressive). Balanced economics with strong agronomic outcomes. Read the full decision tree in the calculations guide.

Per-nutrient defensibility

Not every number in every table has the same level of published support. Zone Forge tags every value with a confidence tier so you know what to scrutinize.

Strong = directly published in the source's main tables. Moderate = published in supplementary literature, well-supported. Limited = published but with narrow regional applicability. Cross-walk = pragmatic buffer-pH-keyed table used as a surrogate when the source's print form prescribes lime via a different method (SMP buffer, pH+CEC). Placeholder = inferred from industry practice; flagged in-app as verified: false.

The lime engine — separate from fertilizer recommendations

Lime is a different problem from fertilizer. The driver is buffer pH (a measured lab value reflecting reserve acidity), the output is tons CaCO₃-equivalent per acre, and product conversion uses ECCE (Effective Calcium Carbonate Equivalent) — not nutrient analysis percentages. Zone Forge runs lime through a parallel pipeline that never gets entangled with the build/maintain/drawdown engine.

Crop-specific target pH

Each Corn Belt / Mid-Atlantic reference profile (Tri-State 2020, Penn State) ships a per-crop target pH table. Those defaults are not universal across all 50 states. Iowa uses the crop and subsoil-specific goals printed in PM 1688 Table 16. Other live lime calculators (including California, Georgia, Montana, Pennsylvania, Texas, Vermont) use source-owned goals and crop scopes from their land-grant publications — for example UVM may publish 6.2/6.8 columns, Texas grasses use a minimum-rate pH×texture table, and Montana stages wheat at 6.0. Pending methods never inherit this map.

Reference-profile defaults for Tri-State and Penn State (Iowa does not use this table — PM 1688 Table 16 prints 6.0 / 6.5 / 6.9 by crop and subsoil, and refuses any other goal):

Resolution falls back through: explicit override → crop-specific target from the active source → that source's profile default → system default of 6.5 when allowed. Provenance flows through to the UI so a farmer can see which source chose the goal.

Calcitic vs dolomitic lime

When lime is required, most shipped profiles pick calcitic vs dolomitic from soil Mg signals. Iowa PM 1688 Table 16 does not use those thresholds — it supplies an ECCE amount and leaves material choice to the laboratory. The default Mg rule for profiles that do publish it:

• Mg ppm < 50 or Mg base saturation < 10% ⇒ dolomitic (corrects pH and Mg together).
• Otherwise ⇒ calcitic.
• No Mg data ⇒ calcitic (safer than loading Mg into already-balanced soils).

The thresholds live in each source's dolomitic_if JSON block and can be tuned per region.

Magnesium correction — separate paths, not coupled

The fertilizer engine and lime engine both touch Mg, but they don't share state. The fertilizer engine produces a Mg fertilizer rate via the standard build-maintain-drawdown logic; the lime engine uses Mg only to choose calcitic vs dolomitic. Default field-scale Mg sources are dolomitic lime (when liming is also required) and K-Mag (langbeinite). Epsom salt is tagged as specialty / advanced and is excluded from default recommendations — at field scale on row crops, it isn't the practical product.

Beyond the three core tables — all 50 states run their own method

The three tables above are reference profiles for the Corn Belt and Mid-Atlantic. They are no longer the whole engine: as of June 2026, nutrient recommendations compute via your own state's published land-grant method in all 50 states — categorical rating tables (Alabama, Arkansas, Kentucky, Wisconsin…), fertility-index lookups (Delaware, Maryland), yield-goal equations (South Dakota), reactive-aluminum tables (New England), region- and CEC-keyed classes (Illinois, Georgia, Idaho…) — not a neighboring state's calibration. Lime sources are also mapped state by state, with live methods and pending implementations labeled separately.

Every implemented state method has been re-checked against the publishing university's own document — worked examples, table cells, band boundaries — and every discrepancy we found was corrected against the publication. Where a state's method needs an input we can't responsibly assume (soil region, irrigation, CEC, a lab-reported index), we ask for it rather than guess; where a published number couldn't be verified, that path honestly declines rather than serving an unverifiable rate.

Nutrient grading — turning your soil test into a verdict in your state university's own rating words — is live for 26 states today. Examples: Arkansas's "Very Low → Above Optimum," Delaware's Fertility Index Value, Oklahoma's "Low / Sufficient." For the remaining states we show your measured value with a note on why we don't grade that lab format yet. The pH/lime rail appears alongside it with its own status shown first: live calculation, coming soon, or lime rarely needed for alkaline-soil guidance.

We never invent a grade or a rate. A soil test run on a lab method a state's publication doesn't accept gets an honest explanation, not a converted guess. Every rate traces to the published document your extension agronomist would pull off the shelf — and the public agronomy calculator runs the exact same engine, so you can check us before you ever sign in.

See the full coverage: browse every state, crop and source — with the published method, edition and citation behind each — in the Coverage Explorer.

Next: Calculations & cost model →