Wildfire evidence · Coast redwood · Research reviewed October 2026
Coast redwoods can survive fire and produce new growth after severe damage. That does not mean every burned tree survives, or that a forest has fully recovered. The useful question is what survived: the original stem, its crown, the connected living plant, or a particular forest function.
Can coast redwoods survive wildfire?
The USDA Forest Service describes coast redwood, Sequoia sempervirens, producing shoots from root crowns and stumps, and from dormant buds along a damaged trunk. These are different routes for vegetative renewal. Its silvics account also describes crown reconstruction after fire damage. This is established species biology, rather than a survival percentage for the 2020 CZU fire. [3]
Keep coast redwood separate from giant sequoia. The latter is Sequoiadendron giganteum. The Forest Service says older giant sequoias normally do not sprout from stumps or roots, although it documents exceptions and other sprouting responses. A finding about coast-redwood basal sprouts should therefore stay attached to that species. [4]
Even within one species, the word “survived” needs a definition. A report about living basal shoots answers a different question from a report about intact canopy. Both can be valuable observations. Combining them into a single undifferentiated success rate makes the evidence less useful.
Five outcomes that a recovery photograph cannot collapse into one
The following is Tecnobosque’s reading framework. It organizes questions; it is not a new classification of the trees in either research paper. Use the authors’ actual definitions when extracting their results.
| Outcome | Question to ask | Evidence still needed |
|---|---|---|
| Original stem | Is the previously standing stem still living? | The study’s living/dead rule and observation date |
| Crown or trunk foliage | Has foliage remained or developed on the original tree? | Location and amount of growth; repeat observations |
| Connected plant | Does living basal growth persist despite stem damage? | The connection being assessed and how shoots were assigned |
| Recruitment | Are new plants establishing and persisting? | Seedlings distinguished from sprouts; later survival |
| Forest function | Has the particular function being discussed returned? | A direct measure of that function, rather than greenness alone |
Genet refers to a genetic individual that may have multiple connected stems; a stem is a narrower observation unit. For editorial purposes, record the unit before the count. Ten shoots do not automatically represent ten newly established genetic individuals. Nor does counting the living shoots establish how much of the old canopy remains.
A photograph can document visible growth at a particular place and time. It cannot supply all the denominators, species identities, stem histories and later outcomes required by the table. Repeated observations of identified subjects are more informative than a sequence of attractive images whose locations change.
What the August 2026 CZU redwood study adds
The 2026 study examined 77 focal coast redwoods across six parks two years after CZU. Selection spanned five bole-damage strata and dominant or codominant trees over 30 cm diameter. This sampled group is not a census. [1]
Bole-char ratio helped explain basal-resprout abundance. Higher climatic water deficit was associated with lower abundance, but its coefficient in the separate resprouting-presence model was not statistically significant (p = 0.73). The paper also uses an earlier, separate plot dataset for presence modeling. Keep the responses and datasets distinct. [1]
Interpretation: “more shoots” and “any shoots” are different outcomes. Imagine two groups where nearly every tree has at least one shoot, but one group has many more shoots per tree. A predictor can describe abundance differences without meaningfully separating presence from absence. This illustration explains the distinction; it does not reconstruct the researchers’ observations.
The same caution applies to a model’s performance. Explaining variation among sampled trees is a different achievement from predicting the fate of an unexamined tree. A useful report identifies the outcome, the sampled population and the test used before translating a model result into ordinary language.
Arithmetic check: the stated percentage and count do not agree
The Results section states “85.3% (61 of 77)” for focal trees with basal resprouts. That wording appears in both the publisher’s HTML and page 6 of the PDF checked on 13 October 2026. [1]
| Evidence object | Value | How this review uses it |
|---|---|---|
| Percentage stated by the paper | 85.3% | Retained as the authors’ published statement |
| Count stated by the paper | 61 of 77 | Retained separately; not silently altered |
| Tecnobosque arithmetic | 61 ÷ 77 × 100 = 79.2%, rounded | A consistency check, not a corrected field result |
The difference is about 6.1 percentage points. Arithmetic alone cannot tell us whether the percentage, numerator or denominator needs correction. We have not obtained the underlying focal-tree records or an author clarification. The inconsistency remains unresolved in this review.
This does not establish that the whole study is invalid. It does establish that this particular count and percentage should not be repeated as though they agreed. It also prevents us from turning either number into a percentage of all CZU redwoods that survived: basal-resprout presence in a selected sample is a narrower result.
What does bole-char ratio measure?
The metric divides char height along the stem by total tree height. A relative measure can distinguish trees with the same char height but different total heights. The following numbers are invented solely to explain that calculation.
Tree A: 12 ÷ 20 = 0.60, or 60%. Tree B: 12 ÷ 40 = 0.30, or 30%. The denominator changes the interpretation without changing the assumed 12-metre observation.
Use a metric for the quantity it defines. This ratio does not directly measure soil damage, remaining foliage or subsequent survival. A ratio of 60% does not mean 60% of the tree died. It also supplies no automatic conversion to a soil burn severity class. Linking those outcomes would require additional observations and an explicitly evaluated relationship.
For a comparison, retain the measurement rule and timing. A survey taken immediately after burning and a survey years later may have access to different visible evidence. A shared label in two tables is not enough to establish that the underlying measurements are interchangeable.
What does the prescribed-fire comparison show?
Biblin, Russell and Wilkin’s study, published 10 March 2025, sampled 100 plots in Big Basin in 2021: 50 in three previously prescribed-burned sites and 50 in three sites without that history. The previous burns were in 1999, 2007 and 2011. This was a retrospective site comparison, not a randomized treatment experiment. [2]
The authors reported associations between prior prescribed fire and several post-wildfire outcomes, including higher canopy cover and lower mortality and topkill. Seedling counts were strongly concentrated in a few plots; the authors cautioned against treating those counts as assured long-term recruitment. Their post-fire plots approximated earlier locations rather than directly resampling physically marked plots. [2]
The interpretation should match the outcome and the design. The study provides evidence worth considering when evaluating forest management. It does not supply a guaranteed duration of treatment benefit, a randomized estimate for every coast-redwood forest, or a home-protection percentage.
Also distinguish the number of plots from independent treatment histories. Many observations inside a few sites do not create the same number of independently assigned management treatments. Comparisons should retain site structure, treatment history and the authors’ statistical approach rather than treating every observation as an unrelated experiment.
Neither selected study turns visible resprouting into proof that another severe fire would have no consequences. To test persistence through repeated disturbance, a new analysis would need subsequent exposure and repeated outcome measurements. A claim about resilience after one event should retain that event boundary.
How to assess a “redwood forest recovered” claim
- Name the species and unit. Coast redwood, original stem, basal shoot, seedling and plot are not interchangeable labels.
- Name the endpoint. Ask whether the claim concerns living tissue, canopy, recruitment or a directly measured forest function.
- Write the denominator. Selected damaged trees, every sampled tree and every tree in a fire perimeter describe different populations.
- Record the observation date. Publication in 2026 does not turn measurements made two years after a 2020 fire into a current forest census.
- Preserve the design. A site comparison, a fitted model and repeated field observations have different strengths and limits.
For example, a headline could accurately say that researchers documented basal regrowth in their sampled trees. Claiming that the same percentage of the entire forest recovered would introduce both a new population and a new outcome. Those two changes need evidence, not stronger adjectives.
The adjacent infrastructure story is also separate. Forest renewal cannot certify drinking-water quality or completion of rebuilding. Read the CZU Lightning Complex water-quality and recovery record for the dated evidence behind those questions.
| Date or interval | Evidence checkpoint |
|---|---|
| 2020 | CZU event underlying the selected studies |
| 2021 | Big Basin prescribed-fire comparison sampled |
| Two years after the fire | Focal-tree observations in the later resprouting study |
| 10 March 2025 / 15 August 2026 | Publication dates of the two research papers |
| 13 October 2026 | This review and the publisher HTML/PDF consistency check |
Common questions
Does a blackened trunk mean a redwood is dead?
A photograph of char alone does not establish the living/dead classification. Ask what living tissue or growth was examined, which stem or connected plant was assessed, and when. Do not assign a field classification to a tree you have not examined.
Is a basal sprout the same as a new seedling?
No. They describe different origins of growth. Keep those categories separate when counting establishment or discussing genetic individuals. Combining them could change the ecological meaning of an otherwise correctly totaled count.
Can the 2026 percentage predict a particular tree’s survival?
It cannot provide that prediction. The count and percentage contain an unresolved mismatch, the sample was selected for a particular research purpose, and basal-resprout presence is not the same endpoint as the future condition of an individual tree.
Primary sources, method and citation
- Carrasco, Kane and Tepley (2026): Prolific post-fire persistence: resilience and resprouting in the iconic coast redwood. Fire Ecology 22:108. Publisher PDF; Results, printed page 6. DOI 10.1186/s42408-026-00564-7.
- Biblin, Russell and Wilkin (2025): Long-term influence of prescribed burning on subsequent wildfire in an old-growth coast redwood forest. Fire Ecology 21:11. DOI 10.1186/s42408-025-00356-5.
- USDA Forest Service: Redwood, Silvics of North America. Olson, Roy and Walters; published 1990. Species biology, including vegetative regeneration; not a contemporary CZU inventory.
- USDA Forest Service: Giant Sequoia, Silvics of North America. Distinct species and vegetative regeneration; not a substitute for coast-redwood results.
Method: review of publisher articles, visual verification of the 2026 PDF statement, and independently labeled arithmetic. No raw-data reanalysis, field survey, causal effect estimate or current forest-health census was performed. The outcome framework and synthetic char-height example are Tecnobosque’s own explanatory work. The cited papers retain their authorship; both are published under Creative Commons Attribution 4.0.
Cite: Tecnobosque, Can Redwoods Survive Wildfire? CZU Evidence and Limits, TB-RESEARCH-US-2020-CZU-REDWOOD, version 1.0, reviewed 13 October 2026.