Some plants genuinely change the ground for the ones beside them - and one will kill your tomatoes. But the real effects look nothing like the companion chart.
The companion planting guide makes a fairly blunt case: the famous companion charts are folklore, and pairs of plants don't have secret relationships that the chart's author discovered.
That leaves an obvious question, and it deserves a straight answer rather than a shrug. Do plants affect the plants next to them? Yes. Several do, the mechanisms are known, and one of them is serious enough that the planner refuses the planting outright. But the real effects look nothing like the chart - they're mostly about chemistry in the soil and physics at the soil surface, they often outlast the plant that caused them, and the honest ones come with wildly different confidence.
Here they are in order of how sure anyone is, which is the ordering that matters and the one you'll never find on a chart.
Certain, and serious: black walnut
Well establishedDo not plant a juglone-sensitive species within the root zone of Juglans. The sensitive list is the hazard's own, and it is wider than the nightshades - it also carries apple and cabbage.
fatal if ignored → block
Mechanism: Juglans roots exude hydrojuglone, oxidized to juglone, a respiration inhibitor. nightshades are highly sensitive: wilt and death.
Evidence: confirmed against 2 sources in the research-based literature.
the 2 sources
UW-Madison / Wisconsin Horticulture (Joy, Hudelson & Jull, rev. 2024-02-28), 'Black Walnut Toxicity', hort.extension.wisc.edu/articles/black-walnut-toxicity (read 2026-07-13): 'Vegetables such as tomato, potato, eggplant and pepper ... are particularly sensitive to juglone'; 'The toxic effects of a mature black walnut tree can extend 50 to 80 feet from the trunk of the tree, with the greatest toxicity occurring within the tree's dripline.'
Iowa State Univ. Extension & Outreach, Yard and Garden FAQ, 'What plants are sensitive to the juglone produced by black walnuts?', yardandgarden.extension.iastate.edu (read 2026-07-13): juglone-sensitive plants include 'members of the tomato family (tomato, potato, eggplant, pepper).'
remedy
Plant it outside the walnut's root zone — the whole drip line and beyond. There is no soil treatment for juglone.
Read the full rule → · open in the planner ↗
Start here, because it's the one that can actually cost you a season, and because it's the clearest proof that plant-to-plant chemistry is a real category and not a hedge.
Walnut roots give off a compound that turns into juglone, which chokes off respiration. Tomatoes, potatoes, eggplant and peppers are highly sensitive: not stunted, not sulky - they wilt and die. And the range is much larger than people expect. The toxic zone reaches across the whole drip line and well beyond, which routinely means a neighbour's tree you didn't plant and can't remove.
Three things make this different from every companion-chart claim:
- It's well established, and it blocks. The planner doesn't advise around this one; it refuses the bed outright.
- There's no treatment. You can't amend juglone out of soil. The only answer is distance, or containers with fresh soil above the root zone.
- It's not reciprocal and not a "pairing" - it's one plant changing the chemistry of a volume of ground. That's what a real interaction looks like, and it's nothing like "plant basil near tomatoes."
If you've got a mature walnut anywhere near the garden, this outranks every other siting question you were weighing.
Real but modest: sunflower residue
PromisingSunflower residue suppresses germination of some small-seeded species.
costly if ignored → warn
Mechanism: Sunflower tissue and roots release allelochemicals that persist in the soil and inhibit the germination and early growth of sensitive plants. Small-seeded, direct-sown neighbours are the ones at risk: a small seed has little reserve to push through the inhibition, whereas a transplant is already past the germination window and largely escapes.
Evidence: confirmed against 1 source in the research-based literature.
the source
'The Allelopathic Activity of Aqueous Extracts of Helianthus annuus L. ... on Germination, Development, and Physiological Indices of Pisum sativum L.', PMC10180669, pmc.ncbi.nlm.nih.gov/articles/PMC10180669 (read 2026-07-11): 'If a sensitive test plant is exposed to allelopathic compounds, its seed germination and seedling growth are reduced'; 'Aqueous extracts of various organs of sunflower significantly suppress the growth of weeds ... and agricultural plants.' Effect was concentration-dependent (stimulatory at low concentration, inhibitory at 75%), consistent with the 'modest' effect_size. NOTE: this study used pea (large-seeded); the small-seeded/direct-sown scoping is the corpus's operationalization (small seeds have less reserve to overcome inhibition; transplants bypass the germination window), not a claim from this source.
remedy
Keep small-seeded direct sowings out of last year's sunflower ground, or set transplants into it instead of sowing seed.
Read the full rule → · open in the planner ↗
Sunflowers leave chemicals behind in their residue and root exudate, and those can hold back germination in small-seeded crops direct-sown into that ground afterwards.
Notice two things that promising is carefully carrying. First, this is a residue effect - it's about what grew there last year, so it belongs with rotation thinking rather than with neighbour thinking. Second, the remedy names a genuine way out: transplants bypass it. The suppression acts on germination, so a seedling set into that ground with its roots already built isn't running the gauntlet the seed would.
The corpus is unusually candid about the edge of this one. The study behind it tested pea - a large-seeded crop - and the "small-seeded, direct-sown" scoping is the corpus's own reasoning about who's most vulnerable (small seeds carry less reserve to push through), not something the paper claims. That reasoning is sound and it's still reasoning, which is why this warns rather than blocks.
Real mechanism, modest confidence: living mulch
A good hunchLiving mulch suppresses weeds and conserves soil moisture.
suboptimal if ignored → advise
Mechanism: A living cover keeps a leaf canopy over the ground: it intercepts light at the soil surface, so weed seedlings are shaded out before they establish, and it slows surface evaporation, so the soil holds more water. In a Three Sisters planting this is the squash's role — its broad leaves shade the bed.
Evidence: confirmed against 2 sources in the research-based literature.
the 2 sources
UMN Extension, 'Mulching for soil and garden health', extension.umn.edu/managing-soil-and-nutrients/mulching-soil-and-garden-health (read 2026-07-11): mulch 'Suppresses weeds'; 'Studies show that mulch reduces evaporation from the soil surface and helps soil hold water.'
Cornell Cooperative Extension (Warren Co. Master Gardener), 'Creating a Three Sisters Garden', warren.cce.cornell.edu/gardening-landscape/warren-county-master-gardener-articles/creating-a-three-sisters-garden (read 2026-07-11): 'The large, prickly squash leaves shade the soil, preventing weed growth' — the living-mulch role in Three Sisters.
Read the full rule → · open in the planner ↗
This is the other direction entirely - not chemistry but physics, and it's the one genuine mechanism underneath a lot of intercropping enthusiasm.
A plant with big low leaves covers the soil. Covered soil grows fewer weeds, because weed seeds need light, and it loses less water, because less evaporates from a shaded surface. That's the whole mechanism, and it isn't mysterious - it's what mulch does, done by something alive.
It's also the actual reason the squash is in a Three Sisters mound. Not because squash and corn enjoy each other's company, but because those big prickly leaves shade the ground the corn is standing in. Together with the trellis role, that's what makes the arrangement work - and pointedly not the famous nitrogen story, which the corpus refutes.
It's filed as a good hunch, and it's worth understanding why it isn't rated more strongly: the mechanism isn't in doubt, but how much weed suppression and water saving you get from a particular living mulch in a particular garden isn't a number anybody can hand you. A confident effect with an unmeasured size doesn't get promoted here.
Plausible, and honestly unmeasured: fennel
A good hunchFennel is allelopathic; a suppressive effect on neighbouring vegetables is plausible but poorly characterized (evidence is weed-directed, in vitro).
suboptimal if ignored → advise
Mechanism: Reported allelopathic exudates. Effect size poorly characterized.
Evidence: confirmed against 3 sources in the research-based literature.
the 3 sources
Nourimand, Mohsenzadeh, Teixeira da Silva & Saharkhiz (2011), 'Allelopathic Potential of Fennel (Foeniculum vulgare Mill.)', Medicinal and Aromatic Plant Science and Biotechnology 5(1):54-57 (read 2026-07-13): fennel-seed ethanolic extract inhibited germination and seedling growth of FOUR WEEDS (perennial ryegrass, wild barley, oat, dandelion) in vitro at 2.5-10%; no crop or vegetable was tested.
Tanase, Istrate & Stoleru (2026), 'Allelopathic Interactions in Vegetable Production Systems', Horticulturae 12(4):438, doi:10.3390/horticulturae12040438 (read 2026-07-13): lists fennel among aromatic vegetables (with dill, parsley, carrot, onion) whose extracts can have 'both inhibitory and stimulatory effects on plant germination and growth, depending on the species, concentration, and application conditions'; no fennel-specific suppression of a neighbouring vegetable is reported.
Gap: no study tests fennel suppressing a neighbouring VEGETABLE in situ; documented activity is weed-directed and in vitro. The claim was reworded 2026-07-13 to say exactly that, and the maintainer read both papers 2026-07-25 and promoted it. What is verified here is the HEDGE - that the effect is plausible and poorly characterized - not a suppression effect. A field trial of fennel against a neighbouring vegetable would change this rule whichever way it came out.
Read the full rule → · open in the planner ↗
Fennel is the one to look at if you want to see what this project does when the evidence runs out.
It has a real reputation as a bad neighbour, and there's genuine chemical activity behind it - fennel extract holds back germination in laboratory tests. But read the corpus's own account of the gap: the documented work is weed-directed and in the lab, and no study tests fennel suppressing a neighbouring vegetable in the ground. A review that covers fennel puts it among aromatics whose extracts can be either inhibitory or stimulatory depending on the species and the concentration.
So what's verified here is the hedge itself - that the effect is plausible and poorly characterised - not a suppression effect. The rule was deliberately reworded to say exactly that.
The practical upshot is small and honest: if you have fennel, giving it its own corner costs you nothing, and that's about the strength of recommendation the evidence supports. What you shouldn't do is repeat "fennel suppresses everything" as though it were established, because it isn't.
What this adds up to
Plants do affect their neighbours. The effects that survive scrutiny share a shape:
- They run through soil chemistry or surface physics, not through some pairwise affinity.
- They're often about time, not adjacency - what grew here last year, not what's beside it now.
- They come with wildly different confidence, from a well-established fatal block to a good-hunch shrug, and that spread is information rather than an embarrassment.
The reason the companion chart feels more useful than this page is that it never says "we don't know." Everything on it arrives at the same confident volume. That uniformity is the tell: real findings don't all come out equally certain, and a source that never varies its confidence isn't reporting evidence.
