Data & sources
Transparency is the product. Here's where every piece of data comes from and how we mark its reliability.
Authoritative sources
| USDA PLANTS Database | Taxonomy, accepted names & synonyms, native status, distribution, and trait characteristics where available. Public domain. |
|---|---|
| USDA Plant Hardiness Zone Map (2023) | ZIP-code → hardiness zone and average annual minimum temperature. Public domain. |
| Wikimedia Commons / GBIF / iNaturalist | Creative Commons plant photography. Each image keeps its photographer credit and license. |
What the Guild Builder is based on — and what it isn't
The Guild Builder assembles a layered planting from facts we can cite, not from folk pairings. It matches on structural layer (derived from growth habit and mature height), hardiness zone, and bloom overlap — that last one is a USDA "Bloom Period" claim about when a plant flowers, deliberately not a claim about pollinators.
It also refuses to suggest two kinds of plant, both flagged by USDA: those recorded as allelopathic (they release compounds that suppress neighbours — the opposite of a support plant) and those carrying an invasive or noxious status. If the plant you chose carries either flag, we say so rather than quietly changing your choice.
We do not publish pairwise companion-planting relationships. We looked hard for a source. Companion planting is overwhelmingly a vegetable-garden tradition, and USDA PLANTS — the dataset our trait coverage rests on — does not characterise cultivated vegetables at all: crops like basil, carrot, onion and lettuce return no trait data, while wild and native species return around eighty fields each. The open companion datasets we could license cover those same vegetables, so almost none of it applies to the native and wild plants in this catalogue. Rather than import advice we can't ground, we've left it out. If we find a source covering interactions between wild plants, we'll add it and cite it here.
How we label reliability
On every plant profile you'll see one of two badges next to each trait:
- USDA — taken directly from an authoritative public dataset.
- Derived — calculated deterministically from USDA columns (for example, a plant's structural layer in a guild is worked out from its growth habit and mature height). Not a guess: the same inputs always give the same answer, and the inputs are cited above.
We do not fill gaps with AI-generated trait data. If an authoritative source doesn't publish a value, we leave it blank and we don't offer a filter for it. That's why you won't find a "pollinator-friendly" filter here: no open dataset rates it reliably, and we'd rather ship a smaller honest tool than a bigger one you can't trust. When we find a credible source, we'll add the filter and cite it.
Spotted an error? Let us know.
What we're still building
Localized soil data, via the USDA Soil Data Access service, is the main piece still missing; today's matching is precise to your ZIP's hardiness zone but doesn't yet know your soil. We're also still looking for a credible open dataset on pollinator value and bloom-time wildlife use — until we find one, those filters stay off rather than being estimated.
Photography is now attached to 2,182 of our 2,186 plants, every image under a licence that permits commercial reuse, credited to its photographer.