Method
How this works
Dryland pairs plants by rule, not by opinion and not by AI. Pick a plant, and the tool checks it against the other 172 in the catalog, discards anything that can’t share a bed with it, scores what’s left, and shows you the results in two tiers.
The whole method is on this page, including the weights. There’s nothing proprietary here. If you disagree with a weighting — and a working designer probably will, somewhere — you can see exactly which one and decide how much to trust the result.
What it is not
It doesn’t know your soil, your slope, your microclimate, your deer, or your client. It doesn’t design anything. It narrows 173 plants to a defensible short list and tells you why, which is the tedious part. The judgment is still yours.
Every plant in the catalog is commercially available in the Intermountain West.
Step one: the disqualifiers
Three conditions. A plant fails any one of them and it’s out — no score, no appearance, no override.
USDA zone overlap
The two plants’ hardiness ranges must intersect.
Water use within one step
Water needs are Very Low, Low, or Moderate. Adjacent steps can share a hydrozone. Very Low and Moderate cannot, ever. Putting a sagebrush on the same valve as something that wants regular water kills one of them, and which one depends on who’s running the controller.
Sun exposure overlap
The two plants must share at least one viable light condition.
These are deliberately strict. Loosening them would produce longer lists and worse recommendations. Of the 29,756 possible pairings in the catalog, 20,516 clear all three — about 69%. The rest are never shown.
Step two: the score
Everything surviving the filters starts at zero and accumulates points. Higher is a better design pairing, not a healthier plant.
| Points | Awarded for | Why |
|---|---|---|
| +3 | Different structural category | Layering. A shrub under a tree does something two shrubs don’t. |
| +1 | Height difference of 2 ft or more | Reinforces the layer when the categories already differ. |
| +1.5 | Each bloom season the companion adds | Sequence. A fall bloomer next to a spring bloomer extends the season. |
| +1.5 | Silver or blue foliage contrast | The most reliable foliage contrast in a dryland palette. |
| +1 | Evergreen structure | Something to look at in February. |
| +1 | Grass texture | Movement and fine texture against woody mass. |
| +1 | Wildlife or pollinator value | Tagged for birds, butterflies, or pollinators. |
| +0.5 | Fire resistance | Counted whenever the companion has it, even if the primary does too. Fire resistance is not an aesthetic contrast — it’s just good in every quantity. |
| +0.5 | Both plants native | A mild thumb on the scale, not a mandate. |
| +1 | Matched water use | Same water step, not merely adjacent. Simpler irrigation. |
Scores in practice run from 4.0 to 10.5. The average companion shown in the top tier scores 8.6.
On the weights
These numbers encode a landscape architect’s priorities. They are not empirical findings and nobody measured them. Layering is worth more than fire resistance here because in this climate, at this scale, that’s the call I’d make on a drawing. Someone else would weight them differently and produce a defensibly different tool.
The fire-resistance weight deserves more than a shrug, though.
On fire
In this region fire-wise design is a baseline responsibility; however, broadly speaking the profession hasn’t fully caught up. There is no mandatory national standard for wildfire-resilient planting design, and as of July 2026 landscape architecture licensure does not currently test it as a competency. I expect that to change. Until it does, the burden sits with the individual practitioner.
What exists is model documents, not mandates: the 2024 International Wildland-Urban Interface Code, and NFPA 1140, Standard for Wildland Fire Protection, which absorbed NFPA 1144, the structure-ignition standard. Neither carries force until a jurisdiction adopts it. In Oregon, SB 83 repealed the statewide wildfire hazard map in July 2025 along with the defensible-space and R327 construction mandates tied to it. The State Fire Marshal issued a model defensible space code in May 2026 that cities, counties, and fire districts may adopt, or not. So the operative question on any given site is what the authority having jurisdiction has actually adopted. My practice is to read the model documents, overlay them on the site, and design to them whether or not anyone is enforcing them.
What this tool does not do. The half-point fire-resistance bonus nudges toward species that tend to burn less readily. It is not a defensible-space analysis. Defensible space is zones, spacing, ladder fuels, irrigation, and maintenance — placement and upkeep, not species. A fire-resistant plant left unmaintained against a wall in the first five feet is still a hazard. Use the code for the plan. Use this for the plant list.
Reviewed August 2026
Step three: the diversity constraint
The top six are not simply the six highest-scoring companions. This is the one place the tool deliberately departs from its own scoring, and it’s worth explaining.
Certain plants are compatible with almost everything. Left alone, the scoring surfaced the same handful in most palettes — one groundcover appeared in 68% of all top-six lists. Each of those pairings was individually correct and collectively useless. A tool that recommends the same plant two times out of three isn’t recommending anything.
So the ranking applies a penalty proportional to how often a plant already appears across the whole catalog. A companion in 30% of other lists is ranked as though it scored 0.6 lower; one appearing rarely is barely touched.
Three things this does not do:
- — It doesn’t change the score you see. Displayed scores and match labels are always the unpenalized value.
- — It doesn’t override the disqualifiers. Nothing incompatible is ever promoted.
- — It doesn’t hide anything. A plant pushed out of the top six still appears in the full list below it, at its true score.
The effect: that 68% groundcover now appears in 20% of top-six lists, and the number of plants that surface in anyone’s top six went from 53 to 78.
You can regard this as a thumb on the scale. It is one. The alternative was a tool that gave everyone the same six plants.
The two tiers
Best companions — the top six, one balanced palette, capped at two plants per structural category so you don’t get six perennials.
More that work — everything else that cleared the filters and scored 4.0 or better, grouped by category. Averages 119 plants. It is genuinely long, because the answer to “what can I plant with this” is usually “quite a lot.”
Match labels
| Label | Score | Meaning |
|---|---|---|
| Excellent match | 10.0+ | Checks nearly every box. A confident pairing. |
| Strong match | 8.5–9.99 | A reliable, well-suited combination. |
| Good match | 6.5–8.49 | Works well together. |
| Worth considering | 4.0–6.49 | Compatible on water, sun, and zone. Whether the combination reads well is your call. |
The bottom label is not a rejection. It means the plants will live together and the design question is open.
Where the data comes from
This part matters more than the algorithm, and it’s the part with a blemish on it.
The authoritative source is OSU Extension EM 9136, Waterwise Gardening in Central Oregon (Amy Jo Detweiler, peer-reviewed, revised 2025). 106 of the 173 plants carry water, sun, zone, and size values from that publication.
The remaining 67 do not, and here’s the history. The catalog began as a machine-generated list. When OSU’s data was merged in, the two sources overlapped on 35 plants — and on every one of those 35, the generated values were drier and sunnier than the published values. A consistent bias in a single direction, not scatter. OSU’s numbers replaced them in all 35 cases.
The 67 plants still without an Extension source are the ones OSU doesn’t cover: the sagebrush-rabbitbrush-Apache Plume-penstemon core of a true xeriscape palette. Assume those 67 carry the same bias — that they’re described as drier and sunnier than they actually are. They’re flagged in the data, and they’re on the list to verify against OSU Extension EC 1623 and the OSU Landscape Plants database. The original plan was to check them against Colorado Springs Utilities. That list is good, but it states water use as an irrigation schedule for a Front Range climate that gets summer rain, and Central Oregon doesn’t. Verifying against it would have swapped one bias for another, so it stays below as a reference rather than a source.
45 further entries were dropped outright for being genus- or family-level (“Juniperus species,” “Cactaceae family”). You can’t score what you can’t specify.
Carbon. Dryland does not calculate carbon sequestration. There are no defensible per-species sequestration figures for high-desert shrubs and perennials, and the available urban forestry models are driven by tree diameter, which tells you nothing about a three-gallon rabbitbrush. Where landscape carbon accounting is warranted, the profession already has a better tool — Climate Positive Design’s Pathfinder, developed through a Landscape Architecture Foundation fellowship and used to implement the ASLA Climate Action Plan. The honest carbon argument for xeriscape is avoided emissions anyway: no mowing, no fertilizer, less pumped water. Not sequestration.
21 plants carry a zone caveat where hardiness varies by cultivar.
Bloom seasons were completed in July 2026. Each value records its own source: taken from EM 9136, derived from conventions already established in the catalog, or drawn from standard horticultural reference. Conifers are recorded as having no ornamental bloom, which is a decision rather than a gap.
The plant illustrations are AI-generated botanical plates in a consistent style. They’re meant to convey habit and character. They are not identification keys and shouldn’t be used as such.
What this doesn’t account for
Soil chemistry beyond a general descriptor. Aspect and slope. Cold-air drainage and frost pockets. Allelopathy. Mature root competition. Deer, rodent and insect pressures, which is local and merciless. Availability at your particular nursery this particular spring. Whether the client will actually water it for the first two years.
A short list is not a planting plan.
Corrections
If a value here is wrong — and across 173 plants some will be — I want to know. Specifics are more useful than impressions: the plant, the field, the correct value, and where it’s published.
Built by Jay Battleson, PLA. Licensed landscape architect, JBattleson|Design, Bend, Oregon. jbattlesondesign.com
Sources
- Detweiler, A.J. Waterwise Gardening in Central Oregon. OSU Extension Service, EM 9136, rev. 2025.
- OSU Landscape Plants database, Oregon State University Department of Horticulture.
- Colorado Springs Utilities, Water-Wise Plants database.
- OSU Extension EC 1623, native plant list.
- Climate Positive Design — Pathfinder. Free life-cycle carbon, water and biodiversity tool for landscape projects. Developed with the Landscape Architecture Foundation; supports the ASLA Climate Action Plan.
About this data
Every plant record in this tool carries a data_verified status. Plants sourced from OSU Extension EM 9136 are marked verified. A subset of earlier records were compiled from public sources and have not been individually verified against a published authority. Treat those entries as a starting point and confirm details before specifying a plant for a project.