Climate-Adapted Seed Tool

    –>  Launch the Climate-Adapted Seed Tool  <–


The Climate-Adapted Seed Tool (CAST) helps ecosystem managers identify seed sources best adapted to local climate conditions at their planting sites. CAST currently serves five western US states (CA, OR, WA, ID, & NV), though some features are limited to California.

The climate is changing faster than trees can evolve, resulting in climate adaptation mismatch (CAM) wherein trees grow more slowly and suffer higher mortality because they are not well adapted to the climate they now grow in. Reforestation efforts on public and private lands provide opportunities to plant trees that are better adapted to warming climate conditions. By planting seeds that are pre-adapted to the climates of our planting sites we can grow healthier forests, sequester more carbon, grow more lumber, and reduce wildfire risk (Aitken and Bemmels 2015).

The focal metric used in CAST is percent decline in productivity (%DP). %DP is the percent decline in tree volume or biomass that managers should expect due to using a seed source that is not well adapted to the climate the trees will grow in. %DP incorporates the effects of climate on both growth rate and survival. To optimize tree growth and survival managers should seek to minimize the %DP of their chosen seed source. CAST also estimates percent decline in survival (%DS), CO2 sequestered, and lumber produced (see settings).



Provenance Tests

The effects of climate adaptation mismatch (CAM) are measured with provenance test experiments, where seeds from different climates are planted at common locations and the trees are measured over time. Most provenance trials use robust randomized block designs, but trials planted for demonstration purposes sometimes plant trees in rows ordered by the climate of the seed origin. These demonstrations make the effects of CAM obvious to the casual observer (Fig 1).

The climate adaptations of trees are plain to see to the casual observer.

Fig 1. A demonstration provenance test with the rows arranged from the warmest seed sources in the foreground to the coldest in the distance. Trees that are not adapted to the climate of the planting site grow more slowly and suffer greater mortality. All trees are lodgepole pine. Photo courtesy Greg O’Neill.

The transfer functions currently used in the tool are fit to lodgepole pine data from the Illingworth Trial and Douglas-fir data from the Munger and Trinity trials. The Illingworth trial established over 70,000 lodgepole pine trees from 184 seed source locations at 60 planting locations and measured them for up to 32 years of growth. Together, the Munger and Trinity trials, planted over 25,000 Douglas-fir trees from 77 seed source locations at 6 planting locations and measured them for up to 100 years. We have similar data, albeit sometimes with smaller sample sizes, from 12 other Western-North-American tree species. Preliminary analyses of data for these species suggest the way lodgepole pine and Douglas-fir respond to climatic transfer of seeds is a reasonable first approximation of how most other species respond (see also Aitken and Bemmels 2015). However, species do vary in their responses and transfer functions tailored to other species are in development.

Statistical Models

CAST uses two types of statistical models: climate transfer functions and growth and yield functions. Climate transfer functions use provenance test data to estimate the effects of seed source on tree growth and survival (e.g. %DP, %DS). Growth and yield functions use Forest Inventory and Analysis (FIA) data to estimate the effects of site climate on CO2 sequestration and lumber production in the absence of CAM. The two models are integrated to estimate the effects of seed source on growth and yield.

The transfer functions in CAST fit multidimensional asymmetric bell-shaped curves (Fig 2) to provenance test data and four climate variables. The climate variables are mean annual temperature, mean cold-month temperature, mean annual precipitation, and temperature differential (also known as continentality, the difference between summer and winter temperatures). Growth and yield functions (CO2 and lumber production) fit modified Monod curves to FIA data. They currently use two climate variables: mean annual temperature and mean annual precipitation. Model fitting was conducted with Stan by Dr. Joe Stewart and in collaboration with Dr. Jessica Wright.

Fig 2. Mean-annual-temperature (MAT) climate transfer functions for lodgepole pine reveal different levels of asymmetry between left and right slopes at different measurement ages. Peak performance tends to occur at negative MAT transfer distance, suggesting that populations were already adapted to colder climates than they occupied at the time of the experiment was conducted (1974-2005). Transfer functions become steeper over time and may begin to stabilize as the trees age. At younger ages CAM primarily influenced tree growth rate. Mortality from CAM became more severe at older ages. Ribbons show 95% CI. Data from other species show similar results.


Age and species-specific models available in CAST are currently limited to those described in Table 1. Climate transfer functions for other species and tree ages are in development. The lodgepole pine transfer function is currently used as a stand-in for species that do not yet have their own transfer functions in CAST.

Table 1. Data used to fit models for each taxon.

Climate Data

CAST uses downscaled historical PRISM climate surfaces and LOCA future climate surfaces. PRISM surfaces were statistically downscaled to 300-m resolution by Dr. Joe Stewart in collaboration with Dr. Robert Hijmans. The future climate data currently used in the tool is the ensemble of 10 GCMs selected by the California Department of Water Resources Climate Change Technical Advisory Group based on their historical performance at the global scale, across the Southwestern United States, and for specific needs of California water resources planning (Lynn et al. 2015). The emissions scenario used by the tool is Representative Concentration Pathway 8.5 (RCP 8.5). RCP 8.5 is the trajectory that most closely matches recent historical (2005-2020) emissions (Schwalm et al. 2020) and is a good choice for near-term (e.g. 2021-2050) climate projections. Medium-term (e.g. 2051-2100) projections are more influenced by human behavior and this uncertainty is not yet accounted for in the tool.

Seedbank Inventory

To facilitate finding the best seed for your project CAST integrates seedbank inventory data. The data currently used in the tool is a static snapshot of CALFIRE and USFS seedbank inventory from December 2020. Seedbank inventory is not available in the public version of the tool. For access to the tool with seedbank inventory please contact JoeStewart :at:

Using CAST with Other Tools

Recommended Postfire Reforestation Workflow for Natural Lands:

    1. Use PostCRPT to prioritize areas in need of conifer restoration. Use local knowledge and site surveys to thoughtfully assess and potentially adjust model predictions.

    2. Use PReSET to prioritize areas where restoration is likely to succeed. Use local knowledge and site surveys to thoughtfully assess and potentially adjust model predictions. 

    3. Use CAST (this tool) to select native seeds that are likely to be pre-adapted to the changing climate conditions they will grow in. Use local knowledge and other data (e.g. soils, hydrology, sun exposure) to thoughtfully assess and potentially adjust model predictions.

Recommended Timber Production Workflow:

    • Use CAST (this tool) to select native seeds that are likely to be pre-adapted to the changing climate conditions they will grow in. Use local knowledge and other data (e.g. soils, hydrology, sun exposure) to thoughtfully assess and potentially adjust model predictions.


CAST is developed by Joe Stewart in collaboration with Jessica Wright. Greg O’Neill has been a crucial advisor. 


This project is supported primarily by CAL FIRE’s L.A. Moran Reforestation Center and Fire Resource Assessment Program with supplemental funding from the USFS Pacific Southwest Research Station.

Disclaimer: CAST is actively under development and is provided for information purposes only. Please consult with a forestry professional before planting seeds.