Predicting Douglas-ﬁr Sapling Mortality Following
Prescribed Fire in an Encroached Grassland
Eamon A. Engber1,2and J. Morgan Varner1
Tree encroachment in ﬁre-maintained woodlands and
grasslands is a major management concern, yet little infor-
mation exists regarding the mechanisms of small tree mor-
tality following prescribed burns. We sought to clarify the
relative importance of tree size and ﬁre-induced injury in
the post-ﬁre mortality of encroaching Douglas-ﬁr trees and
to compare results with an existing mortality model for
larger Douglas-ﬁr trees. Crown injury to small Douglas-
ﬁr trees was a signiﬁcant explanatory variable in post-ﬁre
mortality models, with results suggesting a 20% thresh-
old in crown scorch. Crown injury was strongly related
to bole injury, and delayed mortality was important as we
documented new mortality 20 months post-burn. Mortality
models for large Douglas-ﬁr tend to over-predict small tree
mortality, underscoring the need to better understand the
mechanisms of ﬁre-caused mortality for small, encroaching
Key words: crown scorch, grassland restoration, post-ﬁre
tree mortality, tree invasion.
Fire inﬂuences the structure and composition of savanna and
woodland ecosystems worldwide (Bond & Keeley 2005), and
the encroachment of woody species in these ecosystems in
the absence of ﬁre can proceed rapidly (Staver et al. 2011).
Encroachment results in reduced herbaceous diversity and fuel
mass (Devine et al. 2007; Engber et al. 2011), and crown
retreat and mortality of remnant savanna trees (Devine &
Harrington 2006; Spector & Putz 2006). Recent work in
North America has addressed mowing, burning, and herbicide
applications to restore plant community composition (Ansley
& Castellano 2006; Dunwiddie & Bakker 2011), yet gaps exist
in the understanding of treatment effects on small trees.
Prescribed ﬁre is a common restoration treatment in Paciﬁc
Northwest, U.S. grassland ecosystems encroached by Douglas-
ﬁr (Pseudotsuga menziesii ; Hamman et al. 2011), yet mecha-
nisms of ﬁre-caused mortality of small (<3 m tall, hereafter
saplings) saplings are poorly understood. Sapling size has been
the only factor evaluated in past Douglas-ﬁr encroachment
studies (Gruell et al. 1986; Sugihara & Reed 1987; Tveten &
Fonda 1999), with larger saplings surviving more than smaller
saplings. In contrast, mortality models for tree-size Douglas-
ﬁr have incorporated ﬁre-caused injury to the crown and bole
(e.g. Ryan & Reinhardt 1988; Hood 2010), thereby addressing
the underlying mechanisms of mortality. Further models for
1Department of Forestry, Box 9681, Mississippi State, MS 39762-9681, U.S.A.
2Address correspondence to E. A. Engber, e-mail email@example.com
©2012 Society for Ecological Restoration
saplings are needed given the frequency with which managers
use prescribed ﬁre in their treatment.
To address these needs, we evaluated the importance of
Douglas-ﬁr size and ﬁre-induced injury, patterns of delayed
mortality, and relationships between crown and bole injury fol-
lowing ﬁre. Our results were intended to increase understand-
ing of Douglas-ﬁr sapling mortality and provide a framework
for related restoration efforts.
The study site was a 40 ha grassland/woodland (hereafter
Eastside) within the Bald Hills of Redwood National Park.
Eastside is on an upper, NE-facing aspect (20.3% average
slope) at an elevation of circa 865 m. Surface fuel loading
averaged 5.6 (±2.2) Mg/ha and consisted primarily of
herbaceous fuels (81.3%), the remainder shrubs and ferns.
Restoration treatments were initiated by conifer harvest in
2002 (hand cutting, piling, and burning), yet subsequent
encroachment was substantial, with circa 250 saplings/ha
established by fall 2009. In response, managers ignited a
prescribed burn on 1 October 2009 using strip and spot
ignition patterns, with 0.3–2 m ﬂame lengths.
Three months preceding the prescribed burn, we tagged and
measured height and diameter at ground level (DAG) on 100
Douglas-ﬁr saplings (0.5–3 m tall) in a heavily encroached
portion of the burn unit. Bark thickness was measured
on 19 saplings adjacent to tagged saplings to establish a
relationship between bark thickness and DAG (r2=0.80)
for use in a mortality model for mature Douglas-ﬁr (Ryan
& Reinhardt 1988). Three weeks post-burn, percent crown
volume scorched (PCVS; discolored foliage and buds) and
Restoration Ecology 1
Post-Fire Mortality of Small Douglas-ﬁr
Figure 1. Differences in crown injury (PCVS and PCVC) and tree size (diameter at ground level and height) across living and dead Douglas-ﬁr saplings
5.5, 9, and 20 months following a prescribed burn.
percent crown volume consumed (PCVC; charred foliage
and buds) were estimated on all saplings. Twenty saplings
(from the original 100) were randomly selected for cambium
assessment; bark was peeled to reveal underlying cambium
injury [living or dead, from Wagener (1961)]. Maximum
height of cambium injury was recorded (CImax). Tree mortality
was assessed 5.5, 9, and 20 months post-burn.
Regression analyses were conducted to assess effects of
sapling size on crown injury and crown injury on bole
injury. Two-sample t-tests were used to compare sapling
size and crown injury across dead and living saplings 20
months post-burn. Variables differing signiﬁcantly between
dead and living trees were investigated further in a binary
logistic regression model. Appropriate transformations were
employed if assumptions of statistical tests were not met using
untransformed data (Zar 1999).
Binary logistic regression was conducted to assess the
effects of crown injury on mortality probability for the ﬁnal
mortality assessment. Logistic models were developed for each
of the three dates based on PCVS; probability curves for our
saplings were compared with a logistic mortality model for
mature Douglas-ﬁr (Ryan & Reinhardt 1988).
The prescribed ﬁre resulted in relatively high PCVS, averaging
81.7 (±23.9)% on the 100 saplings. Values for PCVC were
much lower, averaging 39.3%, with considerable variation
(coefﬁcient of variation =75%). Regression analyses revealed
weak, nonsigniﬁcant associations between tree size and crown
injury (all r≤0.20). CImax was moderately associated with
PCVS (r=0.67, p=0.001) but strongly associated with
PCVC (r=0.94, p<0.001).
Cumulative post-ﬁre mortality was 47, 74, and 94% at
5.5, 9, and 20 months post-burn, respectively. No signiﬁcant
differences in tree size were found between dead and living
trees 20 months post-burn (p>0.92), though small trees
died earlier (Fig. 1); therefore, sapling size was excluded from
logistic regression models. Crown injury differed signiﬁcantly
between dead and living saplings (p<0.002), with dead
saplings suffering circa 40– 50% (absolute) greater crown
injury. PCVS and PCVC were both strong mortality predictors
(Fig. 2), with the change in deviance slightly larger for PCVC
Although ﬁre is a common restoration treatment in tree-
invaded grasslands and savannas (Pyke et al. 2010), small trees
have been under-represented in post-ﬁre mortality studies. This
study highlights the efﬁcacy of prescribed ﬁre in halting the
encroachment of woody species in grasslands and points to
important variables that can increase accuracy of predictive
mortality models for small trees.
Post-Fire Mortality of Small Douglas-ﬁr
Figure 2. Post-burn mortality probabilities for Douglas-ﬁr saplings, by
PCVS and months post-burn. Mortality probability predictions from the
Ryan and Reinhardt (1988) model (developed for larger Douglas-ﬁr) are
Table 1. Logistic regression mortality parameters (β), change in
deviance, and p-values for term signiﬁcance (p) for Douglas-ﬁr saplings
based on ﬁre-caused crown scorch (PCVS; model 1) and consumption
(PCVC; model 2).
Model β0β1X1Deviance p
1−2.3419 5.1938 PCVS 12.49 <0.001
2 0.5960 6.9354 PCVC 14.23 <0.001
Both PCVS and PCVC were arcsine-square root transformed prior to analysis.
Crown injury is a well-recognized post-ﬁre mortality predic-
tor for many species (Hood 2010), and our work supports its
importance for small Douglas-ﬁr. If we consider 50% mortality
probability a cutoff for tree death, results suggest PCVS above
20% is sufﬁcient to kill trees, while any measurable crown
consumption results in mortality probabilities exceeding 50%.
Interestingly, Ryan and Reinhardt’s (1988) mortality model
predicts a much lower PCVS threshold for our trees (circa
10%). Future work could investigate whether this discrepancy
relates to differential injury tolerance between saplings and
mature Douglas-ﬁr trees. The weak correlation between crown
injury and sapling size was unexpected but may be explained
by limited sapling sizes and heterogeneity in ﬁre behavior.
Bole injury is a well-recognized predictor of post-ﬁre
Douglas-ﬁr large tree mortality (Hood et al. 2008), and the
substantial cambium injury observed on destructively sampled
saplings in this study may help explain ﬁre-caused mortality.
The strong correlation between crown and cambium injury
suggests that a single measurement may sufﬁce to estimate
post-burn mortality in these scenarios. Battaglia et al. (2009)
found that tree size, crown injury, and bole char all improve
mortality predictions for ponderosa pine (Pinus ponderosa)
saplings. These differences suggest that we may need to
reassess traditional tree mortality models and focus research on
small trees where their eradication is a common management
Implications for Practice
•Fire-induced crown injury to Douglas-ﬁr saplings is a
signiﬁcant mortality predictor and should be incorpo-
rated into models used in the planning of restoration
•Small tree mortality monitoring should occur for at least
20 months post-burn to capture delayed mortality and
increase accuracy of predictive models.
•Commonly used mortality models for large Douglas-ﬁr
trees tend to over-predict sapling mortality, so managers
should consider using alternative models in the planning
of restoration treatments.
Managers at Redwood National Park (J. McClelland and R.
Young) provided site history and access. We are grateful to
M. Battaglia, L. Arguello, C. Edgar, N. Sugihara, D. Sarr, and
two Restoration Ecology reviewers for comments on an earlier
version of the manuscript. Field assistance was provided by
L. Quinn-Davidson, M. Cocking, E. Banwell, and P. Cigan.
Funding was provided by the National Park Service Klamath
Inventory & Monitoring Network and USDA McIntire-Stennis
Cooperative Forestry Research Program.
Ansley, R. J., and M. J. Castellano. 2006. Strategies for savanna restoration
in the southern Great Plains: effects of ﬁre and herbicides. Restoration
Ecology 14:420– 428.
Battaglia, M., F. W. Smith, and W. D. Shepperd. 2009. Predicting mortality
of ponderosa pine regeneration after a prescribed ﬁre in the Black Hills,
South Dakota, USA. International Journal of Wildland Fire 18:176– 190.
Bond, W. J., and J. E. Keeley. 2005. Fire as a global herbivore: the ecology
and evolution of ﬂammable ecosystems. Trends in Ecology & Evolution
Devine, W. D., and C. A. Harrington. 2006. Changes in Oregon white oak
(Quercus garryana Dougl. ex Hook.) following release from overtopping
conifers. Trees 20:747– 756.
Devine, W. D., C. A. Harrington, and D. H. Peter. 2007. Oak woodland
restoration: understory response to removal of encroaching conifers.
Ecological Restoration 25:247– 255.
Dunwiddie, P. W., and J. D. Bakker. 2011. The future of restoration
and management of prairie-oak ecosystems in the Paciﬁc Northwest.
Northwest Science 85:83– 92.
Engber, E. A., J. M. Varner, L. A. Arguello, and N. G. Sugihara. 2011. The
effects of conifer encroachment and overstory structure on fuels and ﬁre
in an oak woodland landscape. Fire Ecology 7:32– 50.
Gruell, G. E., J. K. Brown, and C. L. Bushey. 1986. Prescribed ﬁre opportunities
in grasslands invaded by Douglas-ﬁr: state-of-the-art-guidelines. General
Restoration Ecology 3
Post-Fire Mortality of Small Douglas-ﬁr
Technical Report INT-GTR-198, USDA Forest Service Intermountain
Research Station, Ogden, Utah. 20 Pages.
Hamman, S. T., P. W. Dunwiddie, J. L. Nuckols, and M. McKinley.
2011. Fire as a restoration tool in Paciﬁc Northwest prairies and
oak woodlands: challenges, successes, and future directions. Northwest
Science 85:317– 328.
Hood, S. M., D. R. Cluck, S. L. Smith, and K. C. Ryan. 2008. Using bark char
codes to predict post-ﬁre cambium mortality. Fire Ecology 4:57– 73.
Hood, S. M. 2010. Mitigating old tree mortality in long-unburned, ﬁre-
dependent forests: a synthesis. General Technical Report RMRS-GTR-
238, USDA Forest Service, Fort Collins, Colorado. 71 Pages.
Pyke, D. A., M. L. Brooks, and C. D’Antonio. 2010. Fire as a restoration tool:
a decision framework for predicting the control or enhancement of plants
using ﬁre. Restoration Ecology 18:272– 284.
Ryan, K. C., and E. D. Reinhardt. 1988. Predicting postﬁre mortality
of seven western conifers. Canadian Journal of Forest Research 18:
Spector, T., and F. E. Putz. 2006. Crown retreat of open-grown Southern live
oaks (Quercus virginiana) due to canopy encroachment in Florida, USA.
Forest Ecology and Management 228:168– 176.
Staver, A. C., S. Archibald, and S. A. Levin. 2011. The global extent and
determinants of savanna and forest as alternative biome states. Science
Sugihara, N.G. and L. J. Reed. 1987. Vegetation ecology of the Bald Hills oak
woodlands of Redwood National Park. Technical Report 21, Redwood
National Park Research and Development, Orick, California. 78 Pages.
Tveten, R. K., and R. W. Fonda. 1999. Fire effects on prairies and oak
woodlands on Fort Lewis, Washington. Northwest Science 73:145–158.
Wagener, W.W. 1961. Guidelines for estimating the survival of ﬁre-damaged
trees in California. PSW-Miscellaneous Paper 60, USDA Forest Service
Paciﬁc Southwest Forest and Range Experimental Station, Berkeley,
California. 11 Pages.
Zar, J. H. 1999. Biostatistical analysis. Prentice-Hall, Upper Saddle River, New
Jersey. 663 Pages.