CropsNH Complete Guide

Step-by-step setup

Each crop has a list of liked biome tags from Forge's BiomeDictionary (e.g. PLAINS, HOT, NETHER). If your farm's biome has 1 matching tag, you get +14 nutrients. Two matching tags gives +28 nutrients. These are huge bonuses — the difference between a crop growing healthily and getting sick.

Nutrient formula

The Production Calculator handles all of this — just select your biome and crop and it tells you everything.

Key biome locations in GTNH

• Personal Dimension (Mystcraft): You can write the biome you want. Jungle at high humidity is extremely powerful for mid-game farms.
• Nether: Has both HOT and NETHER tags — gives +28 to Ardite, Cobalt, Nether Wart, Tearstalks, StickyCane.
• End: Has END and COLD tags — gives +28 to Enderbloom, Void Ore Berry, Diareed.

Basic breeding line

Minimum setup: Parent A — Cross-crop — Parent B. The cross-crop only checks the 4 cardinal directions (N/S/E/W) — diagonals don't count.

4-parent breeding (for 4-way recipes like Ardite Ore Berry)

Some mutations require 4 specific parent species. Place one parent on each cardinal side of the cross-crop:

Production grid

Once you have your desired crop with good stats, fill a flat grid with all the same crop on farmland. A Crop Manager harvests automatically when mature and refills with seeds from a buffer chest. 9×9 is a common size that fits within a single Crop Manager's range at higher tiers.

Stat-up layout

There's no special layout required by the code — the cross-crop stick simply checks its four cardinal neighbours (N/S/E/W) for crops that can cross or breed. What matters practically is that your target crop has a cross-crop stick adjacent, and that cross-crop stick has two copies of the same crop as neighbours. The traditional "checkerboard" pattern (crops on corners, cross-crop sticks in between) is a community convention that achieves this efficiently, but it's not anything the source enforces. Any layout where a crop has a double-cropstick adjacent and that stick can see two copies of the same crop works identically.

With identical-species neighbours, the spreading path picks one at random, returns that same species, and averages stats only from neighbours of the matching species. Fertilizer in all contributing neighbours triggers the no-drop guarantee. Two identical parents means both need fertilizer storage > 0.

Every crop stick sits on a soil block. Two distinct checks happen at different Y levels:

The soil check applies to the cross-crop stick (the output stick), not the parent sticks. A mutation result is discarded if the cross-crop's soil doesn't match — the entire breeding tick is wasted. No retry within that tick.

Rubyne and Diareed are listed in Stone in earlier drafts of this guide but are actually verified as Farmland and Stone respectively — soil assignments above are pulled directly from each crop's real entry, not retyped by hand. If you spot a discrepancy, check the crop directly in the calculator rather than trust this summary table blindly.

Stats range 1–31 (from Constants.java). When a new crop is bred, each stat is calculated as:

Gain formula for drop rounds: dropChance × 1.03^Gain. Gain 31 gives 2.5× more drop rounds than Gain 1. This is the most important stat for production farms.

Resistance has three separate effects, none of which touch the growth-speed formula itself — getGrowthRate() only takes nutrientPoints, tier, and Growth as inputs. First, every crop tick a crop fails to gain growth (insufficient nutrients for its tier), the game rolls Resistance against a random 0–30 — succeed and the crop just sits idle; fail and it goes sick. At exactly 31 that roll can never fail, so a permanently-undernourished crop stays dormant forever instead of turning sick. Second, that same roll governs seed return when breaking a crop — low Resistance means breaking often yields no seed. Third, Resistance 31 makes a crop stick immune to breakage from running entities — but falling trample ignores Resistance entirely at any value, including 31. To protect against falls, use a trample-immune soil (Garden Soil, Enchanted Earth, Fertilized Dirt). Target 31/31/31 for all three Resistance benefits at once.

How the biome bonus is actually calculated

The bonus isn't just "matched tag or not" — there are two separate values and the game takes whichever is larger:

This means a biome with rainfall ≥ 0.8 gives +14 to every crop unconditionally — even if the crop has zero liked biome matches. Jungle (0.9), Swampland (0.9), Forest (0.8), and Temperate Rainforest (1.2) all hit the cap. Birch Forest at 0.6 gives a partial +4. Desert at 0.0 gives nothing. Exact rainfall for any biome is shown in the Production Calculator when you select it.

If a crop has two matched liked tags (+28) and the biome also has high humidity (+14), the +28 wins — they don't add together. The Production Calculator handles all of this automatically from your selected biome.

Growth rate is added every 256 game ticks (12.8 seconds). Verified example: Ardite Ore Berry (tier 5, likes NETHER+HOT) in Hell (HOT, DRY, NETHER tags, rainfall 0.0) with Growth 31, full water/fertilizer, no sky bonus — matches both liked tags for +28, giving nutrient score 53 and a real time-to-mature of ~5.5 minutes. Run this yourself in the Production Calculator to confirm or to check a different crop/biome combination.

The game filters all neighbours to those passing canCross() (has crop, no weed, growth ≥ 80%), picks one at random, and returns that crop's species. It never creates a new species — only a clone of one parent.

Important fallback: If the 50% roll picks the spreading path but no parent passes canCross(), the spreading path is skipped and the breed path runs instead. The 50/50 split is not a hard wall.

Same species × 2 (e.g. Corium + Corium):

• Spreading (50% of crop ticks): Always returns Corium — cloning
• Breed path (50% of crop ticks): Deterministic check is skipped (deduplication makes parent list size < 2). Falls to pool mutations, which CAN produce Meatrose, MilkWart, Corpseplant, Flax, etc.

Deterministic Mutations

A specific recipe: e.g. Ferrofern ← StoneLily + BlackGraniteLily. If both parent species are present as neighbours, this recipe can fire. The parent list is deduplicated before lookup, so two of the same crop (Corium + Corium) will never match a deterministic recipe — the deduped list has only 1 entry, which is < 2 required.

If multiple deterministic recipes match your parent combination, one is chosen at random — each has equal probability.

Pool Mutations

Every crop is registered to 0–7 named pools (tags like METALLIC, EDIBLE, TENDRILLY). Pool mutations fire when no deterministic recipe matches. The parent list is NOT deduplicated for pool checks — so Corium + Corium counts as two Corium entries, and any pool containing Corium matches.

Selection: pick a random matching pool → pick a random member of that pool → return it. Crops in more pools have proportionally higher effective chance because they appear in more "lottery tickets."

The soil check happens AFTER getBreedingResult() picks a result. If the soil doesn't match, attemptBreeding() returns false — the crop tick is wasted, no retry, no fallback path.

This means your effective mutation probability is: P(result chosen) × P(result's soil matches cross-crop soil). Placing a soil that only one or two pool members require can dramatically focus your output.

Formula: Corpseplant is in the TENDRILLY pool (13 members, not 9 — verify against the Mutation Calculator directly rather than a hand count, since pool membership changes between versions). Corium is in 4 pools (COW, SILK, TENDRILLY), all 4 match (same-species breeding), one is picked at random each crop tick. P = (1/3) × 0.5 × (1/3 pools) × (1/13 members in TENDRILLY) = ~0.321%/tick. These numbers come directly from running the actual calculator engine on this exact pairing+soil combination — pull up Corium × Corium on Graveyard Soil yourself to verify.

All biome tags and rainfall values below come directly from GTNH's biome table (temperature, rainfall, BiomeDictionary tags per biome ID). Crop liked-biome tags come from the decompiled CropsNH jar. Every bonus number is the actual min(2, matchedTags) × 14 calculation — not approximated.

Recommended farm locations

Humidity bonus — the hidden free nutrient

Every biome has a rainfall value (0.0 to ~1.2 in GTNH's table). The formula is clamp((rainfall − 0.5) / 0.3, 0, 1) × 14 — so rainfall ≥ 0.8 gives the full +14 regardless of any liked-tag match, and rainfall ≤ 0.5 gives +0. This is a real, separate bonus from biome tags, calculated from GTNH's actual per-biome rainfall value, not a flat assumption. The Production Calculator looks this up automatically — type your exact biome name and it shows the real rainfall and resulting bonus.

The game takes whichever is larger: the humidity bonus or the liked-tag bonus. They do not stack. A crop already getting +28 from two matched tags gets nothing extra from humidity. But a crop with zero matched tags in a high-rainfall biome (e.g. Mega Taiga at rainfall 0.8, or Rainforest at rainfall 1.0) still gets +14 for free.

Note on biome names: GTNH's world generation mixes vanilla, Biomes O' Plenty, and Realistic World Generation biomes, and many vanilla Overworld biomes don't actually generate in a real GTNH world even though they exist in the biome table. If a biome name from this guide doesn't show up when you search the Production Calculator, check the in-game biome name directly (F3 debug screen or a Crop Lens) rather than assuming the name maps 1:1 to vanilla Minecraft.

Stats on a new crop = average(parentStats) + random(-2, +4) per trait. If every contributing parent has fertilizer storage > 0, the roll becomes average + random(0, +4) instead — can only improve, never drop. "Contributing parents" means whichever crops actually produced this result: on a 2-parent breed, that's both of them; on the cross/clone path with one matching neighbour, it's just that one.

The spreading path picks one parent species, filters all neighbours to just that species, then uses only those as stat contributors. A single parent is sufficient for stat-up via spreading — one high-stat parent next to an empty cross-crop stick, fertilizer in that parent, and the clone's stats can only improve. For a normal 2-parent breed producing a mutation, fertilizer must be in both parents for the no-drop guarantee to apply. Stat drops only happen on a breed-path result without full fertilizer coverage — not from WeedEx, not from being sick.

Efficient strategy:

1. Get one copy of your target crop with any stats
2. Place it as the sole parent next to a cross-crop stick, fertilizer in the parent
3. Collect clones, keep any that improved, discard the rest
4. Replace the parent with your current best and repeat
5. Each generation can only improve with fertilizer in the parent

Weeds spawn in empty crop sticks at a config-controlled rate. A mature crop spreads weeds to neighbours only if Growth > 24 AND Resistance < Growth. Having Resistance ≥ Growth stops outward weed spreading only when combined with WeedEx storage — without WeedEx, the crop will still spread weeds even with Resistance ≥ Growth. If a weed has no crop stick to spread onto in the direction it picks, it instead converts Dirt or Farmland directly beneath that spot into Grass and plants tall grass on top — left unchecked, weeds can eat into the terrain around your farm. Separately, on an empty cross-crop stick, weed-spawning and breeding compete for the same crop tick: if the breeding roll fails, the spot becomes eligible to spawn a weed instead.

*Bonsai growth duration is hardcoded to a flat 1200 regardless of tier — they're all tier 1 but take as long as a tier-2 crop would by the standard formula.

Potato isn't actually the fastest option — and "starter" means more than the five base crops

Wheat, Carrot, Potato, Dandelion, and Poppy are directly plantable because each one registers a real-world item as an alternate seed — Wheat Seeds, a Carrot, a Potato, and so on. That same mechanic exists on 85 other crops in the database, most of them gated behind a specific mod (Natura, Biomes O' Plenty, Twilight Forest) rather than vanilla Minecraft. Several of them grow far faster than any of the original five, and some bridge pools more usefully too. Treat the five base crops as a baseline, not the full set of starting options. Note (updated v2.0.91): Water Artichoke (Witchery) now also has an alternate seed — if you have Witchery artichokes growing nearby, you can plant them directly.

Among every alternate-seed crop, Torchberry (Twilight Forest) has the shortest growth duration in the entire database at 150 points — about 14.6 seconds to 80% maturity under a solid nutrient setup, versus Potato's ~75 seconds. It's also the most restrictive: it only grows at light level 10 or below, and its only pool (BERRY) doesn't touch EDIBLE or FLOWER at all, so it's a fast dead end rather than a bridge to the rest of the tree.

A cluster of tier-2, 200-growth-duration berries — Strawberry, Blueberry, Blackberry, Raspberry, Huckleberry, Maloberry, and BoP Berry — are all roughly four times faster than Potato (about 19.4 seconds to 80%) and all sit in EDIBLE, so any of them can fish the same 39-member pool Potato does. Maloberry (Natura) is the standout: it's the only one of these that's also in YELLOW, the same bridge pool Potato uses to reach Dandelion — meaning Maloberry alone covers everything Potato covers, at roughly a quarter of the wait. BoP Berry (Biomes O' Plenty) takes a different bridge instead, sitting in RED alongside Poppy directly, reaching the flower cluster from the other end.

All of these — Torchberry, Maloberry, BoP Berry, and the rest of the berry cluster — also have a real deterministic breeding recipe (e.g. Maloberry from OrangeTulip + Blueberry), which doesn't matter here since the alternate-seed path skips breeding entirely: any of these can be planted straight from the real in-world item with no breeding step at all. BoP Berry's source item is confirmed directly in source as a Biomes O' Plenty food drop, and generic Natura/Twilight Forest documentation describes their berry bushes as common, naturally-generating Overworld features — but GTNH heavily customizes world generation (Realistic World Gen, restricted biome lists), so whether any specific bush spawns at the rate or frequency you'd expect isn't something this guide can confirm from the decompiled crop code alone. If you can find one of these growing wild, planting it directly is faster than growing a Potato from scratch — if not, breeding the deterministic recipe or starting from Potato both still work.

The actual mechanic: same-species breeding still rolls for mutations

It's tempting to assume two copies of the same crop can only ever clone each other. They can't — MutationPool.isMatch() counts each parent slot against pool membership without removing duplicates first, so a list of two identical parents still registers two valid hits against any pool that crop belongs to. The 50/50 coin flip between spreading and the breed path happens regardless of whether the two parents are the same species; landing on the breed path with two copies of the same crop still checks every pool it belongs to for a pool-mutation result exactly the way two different crops would. Whatever comes out inherits the parents' stats, already maxed, with no separate stat-up step required.

Why this needs two farms, not one

The catch is soil. Maloberry, Potato, Carrot, and Wheat all need Farmland. Dandelion and Poppy need Dirt or Grass. A mutation result gets rejected outright if the soil under the cross-crop stick doesn't match what that specific result needs — so a Farmland setup can roll Carrot or Wheat all day but will simply waste every crop tick that would've landed on Dandelion. Swapping the soil itself isn't a workaround either: the moment the block under an already-planted crop changes to something invalid for it, the game runs a dedicated invalid-soil handler on that crop, rather than just leaving it idle. Plan for two separate cross-crop setups:

1. Farmland setup: max one Maloberry via the cross/clone path (single parent, fertilizer present), then pair it with a second maxed, fertilized Maloberry on the breed path. Fish EDIBLE and YELLOW here for Carrot, Wheat, and anything else those pools contain — all come out already maxed.
2. Dirt/Grass setup: once a maxed Dandelion drops from the Farmland setup, move (or re-grow) it onto Dirt or Grass, then pair it with a second maxed, fertilized Dandelion. Fish FLOWER (29 members) here for Poppy.

Everything else that happens to drop from EDIBLE (36 other members), YELLOW (7 other members), or FLOWER (27 other members) along the way comes out maxed too, as a free side effect of running either loop — there's no need to chase those individually.

How the biome locations were picked

Every crop's liked-biome tags were checked against every real GTNH biome's actual tag list (not vanilla Forge defaults — GTNH runs mostly BoP/RWG biomes, and most vanilla Overworld biomes aren't even generated). For each biome, a coverage score was computed: min(2, matchedTags) × 14, summed across all 179 crops. Canyon won outright among single biomes. For a second location, the top 30 biomes were paired against each other, scoring each crop by whichever side of the border gave it the better bonus — Canyon + Fungi Forest won because their tag sets don't overlap at all, so together they cover two completely different crop families instead of doubling up on the same one.

*"Best served here" means this location gives that crop a better bonus than any of the other three — many crops get a smaller, incidental bonus at more than one location (e.g. a DRY-liking crop picks up +14 at both Canyon and Crag), so these four columns don't sum to a simple total. Counting any crop that gets any bonus at all, Canyon alone covers 93 of 176, and Canyon + Fungi Forest together cover 152 of 176. Canyon's MOUNTAIN tag also overlaps Sacred Springs, Highland, and Rainforest if you'd rather substitute one of those for Fungi Forest. Run any candidate biome name through the Production Calculator to confirm its real tags and rainfall before committing a farm to it.

Past the starters

A graph search over every deterministic recipe in the database found only 13 of 179 crops reachable through pure deterministic chains — the other 166 require at least one pool-mutation step. Stone Lily (reached from Vine + Pumpkin, both downstream of the food cluster) is the deterministic root of nearly the entire metal and stone-crop branch — Ferrofern, Coppon, Tine, every Ore Berry, and every other Lily variant all trace back to it directly. Max Stone Lily once it appears, and the rest of that branch starts from a maxed baseline. Past that, there's no single fixed sequence — pick whatever's reachable from your current maxed crops' pools, breed toward it, max it, and move outward. The Mutation Calculator's "Find Path to Target" mode shows the real route to any specific crop you're still missing.