Tree farms, and planting against erosion#

2026-10-08. A plan, saved for later — nothing here is built yet. The third companion to plant-conditions.md and animals-bees-and-inputs.md. The species are in plant-list.csv, categories timber, Christmas tree, biomass, syrup, forest farming, erosion: … and invasive: do not plant (US).

Tree farms#

What makes trees different is the harvest date. A Christmas tree is cut after a few years, coppiced willow and poplar every few, pulpwood after one or two decades, hardwood sawtimber and veneer after many decades — typical rotations vary by region and product, and are to be sourced per species. A tree planted now must suit the climate of the decade it is cut in, not the one it is planted in. So for trees the climate-trend dimension moves from interesting to decisive:

• ~/dev/climate-analysis already projects the Mid-South to 2060;
• the USFS Climate Change Tree Atlas models current and future suitable habitat for 125 eastern US species on a 20 km grid;
• the USFS Seedlot Selection Tool matches a planting site's future climate to the current climate of seed sources — the right method, whether or not we use the tool;
• CHELSA's CMIP6 projections give the same anywhere.

Income while the trees grow is the usual obstacle, and it is where trees mix with the other lists: alley cropping (pecan or walnut rows with crops between), silvopasture (pines with cattle or sheep — the animal list), forest farming under the canopy (ginseng, goldenseal, ramps, black cohosh; shiitake on oak logs, a fungus and not on the plant list), maple and birch syrup, Christmas trees as the quicker return. Carbon credits and USDA cost-share for forest stewardship exist and have rules to look into.

Planting against erosion#

Erosion can be estimated before anything is planted. The USDA's Revised Universal Soil Loss Equation multiplies five factors: soil loss = R (rainfall erosivity) × K (soil erodibility) × LS (slope length and steepness) × C (cover) × P (practice).

So the useful output is a map of where on a farm R × K × LS is high, and which plantings bring C and P down there.

What to look up for each species — USDA PLANTS has most of it: root depth; vegetative spread rate (rhizomes knit soil); anaerobic tolerance (streambanks flood); propagated by cuttings (willows and dogwoods can be cut as stakes and pushed into a bank); resprout ability and coppice potential (recovery after floods or browsing); nitrogen fixation (alders, buffaloberry, New Jersey tea, false indigo).

Plantings that also yield. Hedgerows of elderberry, aronia, plums and hazelnut hold soil and produce fruit and nuts; windbreaks shelter livestock in cold wind and bees in spring; riparian buffers catch manure runoff before it reaches water — where the fertilizer list's phosphorus warning ends up.

Do not plant#

Some of the worst invasions in the United States were planted on purpose, for erosion control or forage: kudzu, crownvetch, sericea lespedeza, multiflora rose, autumn olive, Russian olive, Japanese honeysuckle, Chinese privet, saltcedar, giant reed, cogongrass, weeping lovegrass. They are on the list, marked, so a recommender recognises them by name and refuses them rather than not knowing them at all. Goats are one of the better tools for clearing several of them — a use for the animal list.

Invasiveness is regional, so the flag cannot be a property of the species alone. False indigo bush is native and useful in the eastern US and invasive in Europe; smooth cordgrass is native on the Atlantic coast and invasive on the Pacific. The US flag comes from USDA PLANTS' noxious and invasive status by state; worldwide, from GRIIS.

Sources#

Build order#

1. Add K and T to the tools/soil_survey.py query; slope from elevation.
2. An erosion-risk map per farm field: R × K × LS, against T. Built 2026-10-09: erosion_map.py runs RUSLE on every 10 m cell — slope from USGS 3DEP, K and T from the soil survey's own map-unit polygons, R from the site's rain or the county's official value — and writes GeoJSON for a GIS. On a 0.8 km square at McAllister Farm (Dickson, Mountview, Taft and Guthrie silt loams, slopes to 15%), clean-tilled row crops lose more than the soil's tolerable rate on 99% of it; no-till on 81%; hay on none. The slope length is one number for the whole field, and R from rain runs about a third above Tennessee's published values, so the map says where to look, not what to do.
3. Plantings per purpose, filtered by the site's conditions and by regional invasiveness.
4. For trees, check every recommendation against the climate of its harvest decade. The machinery now exists: modules/advise/advise.py --warm re-runs the whole match on the site's weather shifted to a later decade (site-match-tennessee.md, "The 2050s"; +1.0 °C for the 2050s and about +1.3 °C for the 2060s at the Tennessee farms, from ~/dev/climate-analysis). It is a uniform shift: species limited by summer nights or by drought need the seasonal and variability changes a CMIP6 downscale gives. For fruit trees, chill is now checked by cultivar for stone fruit, apples and pears, with bloom date and frost after bloom, and the 2050s run reports how each moves.