Fire-severity mitigation by prescribed burning assessed from fire-treatment encounters in maritime pine stands

Abstract

Maritime pine (Pinus pinaster Ait.) stands are prone to high-intensity fire. Fuel treatments lessen potential fire behaviour and severity, but evidence of their effectiveness when tested by wildfire is extremely scarce in Europe. We assess the longevity of prescribed burning in maritime pine plantations in decreasing fire severity. Heights of crown scorch and stem-bark char were measured in treated and untreated adjacent areas after fire-treatment encounters in Portugal, Italy, and Australia. Treatment effect was quantified as the log-transformed ratio between prescribed-burned and untreated fire-severity data. Linear mixed modelling indicated that for typical wildfire conditions, the effect of prescribed burning in crown scorch height lasts 2–6 years. The persistence of prescribed burning benefits is higher for fire control operations than for fire-severity mitigation. Regression tree analysis of data from one wildfire highlighted the roles of wind direction, topography, and stand height in explaining variability in fire severity. A 4-year interval between prescribed burning treatments in maritime pine stands is recommended in general, depending on site quality and stand age and structure. Improved fuel-consumption prescriptions and monitoring procedures are advisable to foster prescribed-burning effectiveness and its evaluation.

Full text

NOTE
Fire-severity mitigation by prescribed burning assessed from
fire-treatment encounters in maritime pine stands
Juncal Espinosa, Pedro Palheiro, Carlos Loureiro, Davide Ascoli, Assunta Esposito,
and Paulo M. Fernandes
Abstract: Maritime pine (Pinus pinaster Ait.) stands are prone to high-intensity fire. Fuel treatments lessen potential fire behav-
iour and severity, but evidence of their effectiveness when tested by wildfire is extremely scarce in Europe. We assess the
longevity of prescribed burning in maritime pine plantations in decreasing fire severity. Heights of crown scorch and stem-bark
char were measured in treated and untreated adjacent areas after fire-treatment encounters in Portugal, Italy, and Australia.
Treatment effect was quantified as the log-transformed ratio between prescribed-burned and untreated fire-severity data. Linear
mixed modelling indicated that for typical wildfire conditions, the effect of prescribed burning in crown scorch height lasts
2–6 years. The persistence of prescribed burning benefits is higher for fire control operations than for fire-severity mitigation.
Regression tree analysis of data from one wildfire highlighted the roles of wind direction, topography, and stand height in
explaining variability in fire severity. A 4-year interval between prescribed burning treatments in maritime pine stands is
recommended in general, depending on site quality and stand age and structure. Improved fuel-consumption prescriptions and
monitoring procedures are advisable to foster prescribed-burning effectiveness and its evaluation.
Key words: fuel treatment effectiveness, fire behaviour, fire management.
Résumé : Les peuplements de pin maritime (Pinus pinastri Ait.) sont sujets à des feux de forte intensité. Les traitements des
combustibles atténuent le comportement et la sévérité potentiels du feu mais la preuve de leur efficacité à la suite de feux de
forêt est extrêmement mince en Europe. Nous évaluons la durée de l’effet du brûlage dirigé pour réduire la sévérité du feu dans
des plantations de pin maritime. La hauteur du roussissement dans les cimes et de la carbonisation de l’écorce sur le tronc a été
mesurée dans des zones traitées et non traitées adjacentes après le passage d’incendies dans des zones traitées au Portugal, en
Italie et en Australie. L’effet du traitement a été quantifié au moyen du logarithme du rapport entre les données de sévérité du
feu avec et sans brûlage dirigé. Un modèle linéaire mixte a indiqué que dans des conditions typiques de feu de forêt, l’effet d’un
brûlage dirigé caractérisé par la hauteur du roussissement dans les cimes dure2à6ans. La persistance des bénéfices d’un brûlage
dirigé est plus longue lorsqu’on considère les opérations de lutte contre le feu plutôt que l’atténuation de la sévérité du feu.
L’analyse des données provenant d’un feu de forêt au moyen d’un arbre de régression a fait ressortir les rôles de la direction du
vent, de la topographie et de la hauteur du peuplement pour expliquer la variabilité de la sévérité du feu. En général, on
recommande un intervalle de 4 ans entre les brûlages dirigés dans les peuplements de pin maritime dépendemment de la qualité
de la station ainsi que de la structure et de l’âge du peuplement. L’amélioration des prescriptions de consommation des
combustibles et l’adoption de procédures de suivi sont souhaitables pour favoriser l’efficacité du brûlage dirigé et son évaluation.
[Traduit par la Rédaction]
Mots-clés : efficacité du traitement des combustibles, comportement du feu, gestion du feu.
1. Introduction
Maritime pine (Pinus pinaster Ait.) is a major western Mediterra-
nean Basin conifer, and its plantations are flammable and prone
to high-intensity fire owing to the nature, quantity, and structural
arrangement of fuels (Burrows et al. 2000;Fernandes and Rigolot
2007;Jiménez et al. 2016). In fact, conifer plantations often result
in rapid fuel buildup and homogeneous fuel structures requiring
fire-hazard reduction treatments (Keyes and O’Hara 2002;Cruz
et al. 2017;Zald and Dunn 2018). Changing the structure and re-
ducing the amount of forest fuels through treatments such as
prescribed underburning and mechanical thinning is expected to
moderate subsequent wildfire behaviour and mitigate its impacts
(Graham et al. 2004;Fernandes 2015;Cruz et al. 2017). Whether or
not a fuel treatment is effective when challenged by wildfire is a
function of several factors, both treatment-related (type, intensity,
Received 13 June 2018. Accepted 23 October 2018.
J. Espinosa. INIA, Forest Research Centre, Department of Silviculture and Forest Management, Forest Fire Laboratory, Ctra. Coruña Km 7.5, 28040 Madrid,
Spain.
P. Palheiro.* Gestão Integrada e Fomento Florestal Lda. Rua D. João Ribeiro Gaio, 9B – 1E, 4480-811 Vila do Conde, Portugal.
C. Loureiro and P.M. Fernandes. Centre for the Research and Technology of Agro-Environmental and Biological Sciences, University of Trás-os-Montes
e Alto Douro, Apartado 1013, 5001-801 Vila Real, Portugal.
D. Ascoli. Dipartimento of Agricultural Sciencies, Universita
`di Napoli Federico II, via Universita
`100, 80055 Portici, Italy.
A. Esposito. Department of Environmental Biological and Pharmaceutical Sciences and Technologies, University of Campania “Luigi Vanvitelli”, Via
Vivaldi 43, 81100 Caserta, Italy.
Corresponding author: Paulo M. Fernandes (email: pfern@utad.pt).
*Present address: Parks and Wildlife Service, Department of Biodiversity, Conservation and Attractions, P.O. Box 835, Karratha, Western Australia 6714,
Australia.
Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.
205
Can. J. For. Res. 49: 205–211 (2019) dx.doi.org/10.1139/cjfr-2018-0263 Published at www.nrcresearchpress.com/cjfr on 30 October 2018.
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.

spatial patterns, time since application) and contextual (vegetation
type, fire weather, terrain conditions, and fire-suppression capabil-
ity) (Cochrane et al. 2012).
Prescribed burning (PB) removes fuels beneath the forest can-
opy and thus reduces potential wildfire intensity and severity
(Agee and Skinner 2005). The interaction between fuel and
weather conditions determines to what extent and for how long
PB is effective in reducing fire growth and fire severity or assisting
with fire-suppression operations (Fernandes 2015). The hazard re-
duction effectiveness of PB is then dependent on its initial impact
and on fuel recovery rate, which define the longevity of the effect
(e.g., McCaw et al. 2012).
PB to reduce fire hazard has been practiced in southern Euro-
pean pine forests since the 1980s and is employed under environ-
mental conditions reconciling satisfactory removal of litter and
understory fuels with minimum tree damage and maintenance of
site quality (Fernandes and Rigolot 2007). Fuel consumption and
posttreatment fuel dynamics and their modelled effects on fire
behaviour have been reasonably studied (Fernandes 2018). Yet,
observational evidence of PB effectiveness is limited to the analy-
sis of experimental fire behaviour in maritime pine stands in
relation to fuel age, i.e., time since PB (Fernandes et al. 2004;
Fernandes 2009). Additional empirical-based quantification of
modified fire behaviour and severity in prescribed-burned stands
is needed to improve treatment guidelines and prescriptions
(Fernandes 2018), namely because of the shortcomings of fire
modelling systems in this regard (Cruz et al. 2014). Unforeseen
encounters between wildfires and treated areas offer opportuni-
ties to acquire valuable data, especially if fire behaviour or sever-
ity is documented in treated and untreated contiguous areas;
paired observations control for the influences of dissimilar
weather and terrain conditions, enabling objective assessment of
PB effectiveness (Martinson and Omi 2008).
We examine the extent to which PB mitigates fire severity in
maritime pine stands based on crown scorch and stem-bark char
data primarily collected after wildfires in Portugal, Italy, and Aus-
tralia. Specifically, we expect the PB effect to (i) decrease with time
since treatment, (ii) decrease with increasingly extreme fire
weather conditions, (iii) be affected by interactions between fire
spread and topography, and (iv) increase with stand height.
2. Methods
2.1. Study sites and data
We used data from maritime pine stands where adjacent un-
treated and treated (prescribed burned) areas had been exposed to
the same wildfire, presumably under very similar weather and
fuel moisture content conditions, and supplementary data from
experimental fires (Fernandes et al. 2004;Fernandes 2009). Eligi-
bility criteria of wildfire sites for sampling comprised knowledge
of burn date; time (months) since treatment, from forest service
records; undifferentiated stand characteristics and topography
within each untreated–treated pair; and the absence of poten-
tially confounding effects, e.g., fire-suppression activity (Pollet
and Omi 2002). Overall, we acquired 30 observations in the north
and centre of Portugal, distributed among 10 locations involving
12 wildfires and seven experimental fires (Fig. 1). Given the limited
extent of PB in pine forests in Portugal and Italy, data collection
spanned from 1992 to 2017 and included all fire-treatment en-
counters known to the authors, either directly or communicated
by forest managers. Also, PB units in pine forests in Portugal are
typically small and do not form large patches, which restricts
fire-treatment overlaps and the number of potential sample loca-
tions within a given fire. Each observation was defined by a fire ×
location within fire × time since PB combination. Additionally, we
collected data in the Vesuvio National Park, Italy, using the same
criteria and used data from the Gnangara plantation in Western
Australia (Burrows et al. 2000), structurally similar to Portuguese
coastal stands.
Data collection was plot-based, with some variation in plot size
(100–600 m
2
) and shape (circular, quadrangular, rectangular) and
equal sampling effort (n= 1–4) among pairs of treated and un-
treated areas at each observation site. Fire severity in the PB area
would often decrease with increasing distance from the treatment
edge, as found by others (Safford et al. 2012). The PB plots were
thus established immediately after crown scorch height stabilized
and the edge effect was no longer discernible. Nonetheless, plots
were usually located within 100 m of either side of the untreated–
treated boundary.
The Cepões wildfire near Viseu burned 3074 ha under the most
extreme weather conditions in our sample. Four fuel ages, respec-
tively, 2, 3, 4, and 5 years since PB, were available at two distinct
locations within the wildfire perimeter. This provided an oppor-
tunity to further examine variability in fire severity through a
different sampling scheme. Plots (n= 30, each comprising a group
of five trees) were located as a function of combinations between
fuel age, slope position (lower or upper half), fire spread direction
(upslope or downslope), and whether the location was burned by
the heading, flanking, or backing sections of the fire.
As fire-severity metrics, we measured crown scorch height (h
s
)
and maximum (lee-side) stem-bark char height (h
c
) to the nearest
0.1 m on each individual tree within a plot. We did not distinguish
between crown scorch and crown combustion, hence burned
trees were equated to fully scorched trees.
Fire weather conditions for the day that the study sites burned
were generically described through the Fire Weather Index (FWI)
of the Canadian Forest Fire Weather Index System (Van Wagner
1987), the fire danger rating system adopted in Portugal, using
data from the nearest (usually within 50 km) weather station. The
FWI for the Western Australia wildfire was obtained in Gellie
(2005). The FWI is a relative numerical rating of fire-line intensity,
which in turn is a major determinant of h
s
(Van Wagner 1973).
2.2. Data analysis
Tree fire-severity metrics h
s
and h
c
were averaged by plot, and
the former was divided by stand height to obtain the scorch height
ratio (SHR). Inclusion of completely scorched and fully green trees
can underestimate mean h
s
(Martinson and Omi 2008). Hence, fully
scorched dominated trees did not contribute to the plot average; we
estimated the potential h
s
of green trees as 3.83 × h
c
, after regressing
the two variables, and used the resulting values to calculate mean h
s
.
The treated and untreated values for each observation resulted from
averaging the respective plot values; the 30 plots sampled within the
Cepões wildfire provided six observations.
Analysis of the effect of PB on fire severity was based on the
dimensionless extent of the mean difference between neighbour-
ing untreated and treated plots. Following Martinson and Omi
(2013), we adopted the log response ratio as the log-transformed
ratio between the treatment mean and the control mean (Hedges
et al. 1999), calculated for both h
s
and h
c
. Absolute log response
ratios < 0.20 indicate insignificant treatment effect (Cohen 1988).
We examined whether the sizes of PB effects were related to the
fire-severity descriptors. A linear mixed model was fitted to each
log response ratio, with fire as the random effect. The candidate-
independent models for inclusion in the model were the log-
transformed time since PB (in years), to approach the levelling-off
of fuel accumulation with time since fire, the FWI, and stand
height, plus their interactions.
We used the Cepões individual plot data to examine PB effec-
tiveness as affected by variations in tree size and in fire character-
istics along its perimeter by applying regression tree analysis. We
took h
c
as a fire-intensity proxy, but because crown damage de-
pends on the relative amount of scorched crown (Peterson and
Ryan 1986), we preferred SHR to h
s
to describe fire severity. Time
since PB, slope position, fire spread direction, location on the fire
206 Can. J. For. Res. Vol. 49, 2019
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.

perimeter, and tree height were used as putative independent
variables. Untreated plots were assigned a fuel age of 25 years, the
approximate age of the stand. The splitting process inherent to
regression tree analysis was carried out until the Akaiki’s infor-
mation criterion corrected for small samples (AICc) was mini-
mized. The influence of each variable on SHR and h
c
was evaluated
by its relative contribution (%) to the total amount of variability
explained by the model.
3. Results
The available PB–untreated pairs covered a substantial range in
time since PB, from 4 months to 13 years (Table 1), but 66% of the
observations pertained to fuel ages < 6 years (Fig. 2). Likewise,
variation in fire weather severity as represented by the FWI was
wide, ranging from low (FWI < 8.4) to extreme (FWI > 38.2) as per
Palheiro et al. (2006). The FWI was fairly regularly distributed but
with evident gaps (Fig. 2). Stand height was normally distributed
(Shapiro–Wilk Wtest, p> 0.05) and <15 m for half of the observa-
tions. Variation in fire-severity metrics (both in PB and in un-
treated plots) was wide, but data distribution patterns varied:
normal for h
s
, left-skewed for SHR, right-skewed for h
c
in PB plots,
and fairly homogeneous for h
c
in untreated plots. Full-canopy
scorch occurred in 56.3% of the untreated observations but only in
9.4% of the PB observations. On average, treatments h
s
and h
c
were
81% and 44%, respectively, of the corresponding measurements in
the untreated neighbouring area. This difference between fire-
severity metrics translated into a PB effect size that was larger by
a factor of three for h
c
in relation to h
s
.
Regression analysis indicated that the mitigating effect of PB on
h
s
decreased (p< 0.0001) with increasingly higher log-transformed
time since PB (Ln T) and increased nonsignificantly (p= 0.0936)
with higher FWI. We used FWI ≤ 23 and FWI ≥ 37, boundaries of
the first important discontinuity in the FWI distribution (Fig. 2)
that correspond approximately to low-to-high and to extreme fire
danger (Palheiro et al. 2006), to define two FWI classes, qualified
respectively, as lower and upper FWI ranges. This new variable
was significant (p= 0.0107) in the linear mixed model when added
to Ln T; there was a trend for the log response ratio of h
s
to
decrease with stand height, but its inclusion in the model was not
supported (p= 0.2565). Interactions between the independent
variables were not significant. Treatment h
c
increased signifi-
Fig. 1. Study sites location in Portugal and respective number of observations (map data courtesy of ©2018 Google). Numbers in parentheses
indicate the number of fires, whenever it is >1. Shaded dots correspond to wildfire locations; the black dot corresponds to experimental fires
(plus one wildfire observation). [Colour version available online.]
Espinosa et al. 207
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.

cantly with Ln T(p= 0.0252) but did not respond to FWI class or to
stand height.
Table 2 displays the coefficients of the resulting models for the
effect sizes of PB on h
s
and h
c
. Predictions and confidence intervals
for the models indicate that the PB effect on h
s
persists for 3 (2–6)
and 9 (5–15) years after treatment for the upper and lower FWI
ranges, respectively (Fig. 3). PB longevity in terms of h
c
is greater,
14 years as predicted by the fitted equation and 6 years for the
worst-case (lower confidence interval boundary) scenario.
Within the Cepões wildfire, mean SHR in PB plots was lower
than in the untreated plots by 9.4%. Regression tree analysis of fire
severity indicated that SHR was primarily determined by fire-
spread direction, averaging 0.72 and 0.87 for downslope and up-
slope spread, respectively (Table 3). Within the downslope partition,
SHR averaged 0.66 for fuel age ≤ 5 years, i.e., in PB plots, while in
untreated plots, it reached 0.82. Two major partitions were distin-
guished for upslope spread: fuel age < 5 years (SHR = 0.82) and fuel
age ≥ 5 years (SHR = 0.91). Within the later, SHR was higher for
stands lower than 14.6 m (0.96 versus 0.76).
Mean h
c
in the Cepões wildfire PB plots was two-thirds of the
corresponding value in the untreated areas. The first splitting
level was decided by fuel age, with an average h
c
of 3.3 m for fuel
age < 5 years, and 6.8 m for fuel age ≥ 5 years. Subsequent parti-
tions indicated lower h
c
in the backing and flanking sections of
the fire, as well as in the lower half of the hillsides burned by the
headfire, for fuel age < 5 years and lower h
c
for downslope fire
spread for fuel age ≥ 5 years.
4. Discussion
4.1. Prescribed-burning effects on wildfire severity and
treatment longevity
We found that PB decreases the impact of a subsequent fire on
maritime pine overstory and that the effect declines with time
since treatment. Under typical wildfire conditions, i.e., for the
upper fire danger range used in the h
s
effect-size model, h
s
is
reduced up to 2–6 years after PB. Fuel treatments may need to be
frequent (intervals of 2–10 years) in productive ecosystems, espe-
cially if understory regeneration is fast (Stephens et al. 2012). Such
is the case in this study’s plantations, which are grown under a
Mediterranean-type humid climate. Data from northern Florida
similarly indicates that PB in pine forest avoids the impacts of
high-severity wildfires for up to 5 years after treatment (Malone
et al. 2011). Sackett (1975) recommended a 3-year burn interval in
southeastern USA pine flatwoods.
As in other studies (Lydersen et al. 2014,2017;Storey et al. 2016),
fire-induced canopy injury increased with fire weather severity;
however, fuel treatments in conifer forests can be effective even
when tested by extreme fire weather (Finney et al. 2005;Safford
et al. 2012;Prichard and Kennedy 2014). In our case, recent
(<4 years) PB does not always warrant significant fire-severity
modification in relation to the contiguous untreated stand, as
58.3% of the observations in the upper fire danger range had a log
response ratio for h
s
higher than –0.20.
The longevity of the PB effect on potential fire behaviour and
severity is dependent not only on fuel consumption but also on
fuel dynamics (Fernandes 2015), which are a function of site pro-
ductivity, stand density and age, decomposition rate, and under-
story species response to fire. In particular, posttreatment needle
accretion can be a significant influence, depending on h
s
and canopy
base height, but the effects of PB on litterfall have seldom been
examined (Espinosa et al. 2018). Tools to assist in the evaluation of
the immediate impacts of PB on fuels are available (Fernandes et al.
2012), but the characterization of fuel dynamics and persistence of
Table 1. Mean (range) for the (putative) determinants and descriptors of fire severity (n= 32,
except stem-bark char metrics, n= 23).
Variable Units PB treatment Untreated
Time since PB years 5.4 (0.3–13) —
Stand height m 13.3 (7.5–22.9)
Fire Weather Index (FWI) adimensional 47 (5–87)
Crown scorch height (h
s
) m 10.2 (4.4–21.5) 12.6 (5.5–21.5)
Stem-bark char height (h
c
) m 2.9 (0.9–9.3) 6.6 (1.2–13.6)
Scorch height ratio (SHR) adimensional 0.77 (0.26–1) 0.93 (0.71–1)
PB effect, h
s
adimensional –0.23 (–1.36–0) —
PB effect, h
c
adimensional –0.77 (–2.43–0) —
Note: Prescribed-burning (PB) effect is the log response ratio, i.e., the PB mean to untreated mean
ratio after log transformation.
Fig. 2. Fire Weather Index and time since prescribed burn (PB)
histograms (n= 32). Count is the number of observations.
208 Can. J. For. Res. Vol. 49, 2019
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.

the hazard-reduction effect would benefit from more complete and
detailed monitoring procedures (Fernandes 2018).
4.2. Drivers of fire severity in the Cepões wildfire
Fuel age was a minor contributor to the explanation of variation
in SHR within the Cepões wildfire. Instead, fire spread pattern and
stand height were the major determinants, with lower crown
injury resulting from downslope spread or taller stands, as found
by others (Oliveras et al. 2009;Fernandes et al. 2010;Viedma et al.
2015), and showcasing spatial variation in treatment benefits. The
limited influence of fuel age variation on h
s
, as well as the small
SHR differences between downslope and upslope fire spread, may
be an outcome of the particularly extreme fire weather under
which the wildfire developed. Fuel age, however, was as impor-
tant in determining h
c
as the combined influences of fire spread
pattern, location on the fire perimeter, and topographic position.
Decreased fire intensity along the flanks and rear of a wildfire
(Catchpole et al. 1992) is expected to reduce fire impacts. Unequal
sampling did not favour detection of such effect, i.e., most of the
plots (25 out of 30) were burned by the head fire. Still, h
c
did
decrease in the backing and flanking sections of the fire.
4.3. Fire-severity mitigation and fire-suppression difficulty
While h
s
and SHR are of major interest to postfire tree survival
and growth, h
c
is more relevant from the fire-suppression perspec-
tive. The treatment effect size was substantially higher for h
c
in
comparison with h
s
. As an example, h
s
reached 9.5 m in a 10.1 m
tall stand 3 years after PB, signifying negligible treatment effect
(–0.13), even though it changed a crown fire into a surface fire. The
separate analysis of the Cepões fire data also supported higher PB
effectiveness regarding h
c
. Nonetheless, our findings that the ef-
fect of PB on h
c
persists for 14 years are likely to be inflated,
because changes in fuel accumulation and structure in maritime
pine stands in Portugal are usually minor beyond ⬃10 years after
PB (Fernandes and Rigolot 2007). Small sample size, confounded
effects with fire weather that the analysis could not resolve, and
influences of tree diameter and wind speed on the h
c
– flame size
relationship (Gutsell and Johnson 1996) might be implicated in
the result.
Treatment benefits are thus higher and extend longer in time
for fire control operations than for fire-severity mitigation. We see
three motives for this difference: crown scorch extent is depen-
dent on stand and canopy base heights; h
s
is mainly a function of
fire intensity, but wind speed and air temperature also play a role
(Van Wagner 1973); and postfrontal combustion of duff and
downed woody fuels may result in additional crown damage
(Alexander and Cruz 2012). Current burn prescriptions are conser-
vative regarding duff consumption (Fernandes and Rigolot 2007),
facilitating postfrontal combustion in a subsequent wildfire.
Detection of a significant PB treatment effect does not imply
success in avoiding substantial pine mortality and (or) bringing
fire behaviour to within control capacity. Maritime pine endures
low- to moderate-intensity fire, but the expression of fire adapta-
tion traits is variable, with prevalence of either thick bark or cone
serotiny (Fernandes and Rigolot 2007). Studies of maritime pine
mortality after wildfire show that the probability of mortality
is >90% when crown damage exceeds two-thirds of its volume
(Catry et al. 2010;Vega et al. 2011). However, those studies are with
respect to wildfires in non-treated areas. Maritime pine stands
thinned by consecutive short-return (<4 years) wildfire should be
a better analogue of the mitigating effect of PB on tree survival
(Fernandes et al. 2015). One year after the experimental summer
fire used in our analysis, with crown scorch ratios of 0.94 and 0.88,
tree mortality was 55% and 41%, respectively, where PB had been
applied 3 and 2 years before the summer fire (Fernandes et al.
2004). Avoidance of trunk basal girdling due to the shallow forest
floor in recently treated stands would explain this lower than
expected pine mortality level (Burrows et al. 2000).
Objective insight on the ability of recent (2–3 years) treatments
to assist with effective head fire suppression in maritime pine
stands is provided by the fire behaviour data in Burrows et al.
(2000) and Fernandes et al. (2004). In the former, flame heights
were in the range of 1–2 m in 4 t·ha
–1
fuels in comparison with
3–10 m in heavier fuels (9–23 t·ha
–1
). In the latter, a crown fire
changed to a surface fire with 1.2–2.5 m flame lengths when fuel
load decreased from 36–45 t·ha
–1
to 11–12 t·ha
–1
. In both cases, the
observed decreases are consistent with effective fire control by
direct attack (Hirsch and Martell 1996). It can be argued that more
severe fire weather conditions would override such effect. How-
ever, the head fire accounts for approximately one-third of the fire
perimeter length at most and for 10%–50% of the area burned, and
fire-line intensity varies 10-fold around the fire’s perimeter
Table 2. Coefficients (standard errors) for the linear mixed models for the effects of
prescribed-burning (PB) treatment on fire-severity descriptors.
Effect variable Intercept Ln TFWI class R
2
Crown scorch height (h
s
) –0.5783 (0.0739) 0.2358 (0.0441) –0.1047 (0.0406) 0.397
Stem-bark char height (h
c
) –1.1856 (0.2634) 0.3620 (0.1503) —0.655
Note: Ln Tis the log-transformed time since PB.
Fig. 3. Predicted effects of prescribed burning (PB, years) on crown
scorch height, h
s
, as a function of time since treatment and fire
weather as per Table 2. Increasingly higher PB effect (the log
response ratio) implies decreased effectiveness. The dotted and
dashed lines bound the 95% confidence intervals for the lower and
upper fire weather index (FWI) ranges, respectively.
Table 3. Relative influences (percentage; from regression tree analysis)
on fire-severity metrics for the Cepões wildfire (n= 30).
Independent variables h
s
(R
2
= 0.431), % h
c
(R
2
= 0.385), %
Time since PB 20.1 49.2
Fire spread direction 42.9 43.6
Topographic position 5.7 0.4
Location on the fire perimeter —6.9
Stand height 31.3 —
Note: h
s
, crown scorch height; h
c
, stem-bark char height; PB, prescribed burn.
Espinosa et al. 209
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.

(Catchpole et al. 1992). This variation in fire-line intensity en-
hances the chances of containing a wildfire through direct or
indirect attack along its flanks; decreased fuel loads in treated
stands further expand fire control strategies and tactics, hence
the opportunities for safe, fast, and effective fire suppression.
4.4. Limitations
This study sample included a wide range in fire weather condi-
tions and covered a representative range in fuel load and struc-
ture in the short term to midterm after PB (Fernandes and Rigolot
2007). However, limitations are readily apparent and are inherent
to the opportunistic and retrospective nature of the study.
Wildfire-treatment encounters cannot be anticipated, precluding
collection of data prior to and during the event. Crown scorch
ratio (length of crown scorch in relation to crown length) ex-
presses actual foliar damage and is a better indicator of tree mor-
tality likelihood than h
s
and SHR (Peterson 1985), but in most
cases, it could not be calculated due to lack of crown base height
data. Sample size was small, as was the number of fire-severity
variables and putative independent variables common to all ob-
servations.
Data analyses accounted for modest fractions of the observed
variability in treatment effect size and fire-severity metrics. To
some degree, this reflects the equalizing effect imposed by the
severe weather conditions experienced during wildfires, exacer-
bated by the young age and correspondingly relatively low height
of the stands (Zald and Dunn 2018). The moderate amount of
explained variability is also a natural outcome of the crude met-
rics used to evaluate fire behaviour potential, namely time since
treatment in lieu of local fuel characteristics and the FWI from a
fire weather station instead of the local wind speed and fuel mois-
ture content.
5. Conclusion
Fuel dynamics in flammable well-stocked conifer forests grown
for timber production challenges fire-severity mitigation through
surface fuel treatments. Despite the uncertainty in fuel and burn-
ing conditions inherent to its opportunistic nature, this study
offers empirical insight into the effectiveness of PB areas tested by
wildfire in Europe. The results highlight both the strengths and
limitations of PB: respectively, maintenance of fire-intensity lev-
els below the pretreatment situation for at least 10 years and
mitigation of fire-induced canopy injury restricted to a relatively
narrow period after treatment. While the former is consistent
with the dynamics of fuel load and fuel structure recovery in
maritime pine stands, the latter is a natural outcome of (young)
stand age and (small) tree size. Consequently, PB effectiveness
expectations depend on the perspective, either forest manage-
ment or fire suppression.
To comply with both fire-severity mitigation and effective fire
control operations, a generic interval of 4 years between consec-
utive treatments in maritime pine stands emerges from this
study, although benefits to suppression via facilitated contain-
ment of the flanks are expected to last longer. This recommenda-
tion is adjustable depending on site productivity and stand age
and structure. This also implies that lower crown-scorch levels are
more likely in taller and older stands, as well as in wind-protected
and moister topographic positions. Finally, avoiding marginal PB
conditions that decrease fuel consumption and adopting less con-
servative prescriptions in relation to duff retention can increase
treatment longevity. Further work would benefit from improved
documentation in the frame of the PB planning and monitoring
process.
Acknowledgements
This work is supported by POCI-01-0145-FEDER-006958 and by
FCT – Portuguese Foundation for Science and Technology, UID/
AGR/04033/2013, and co-financed by INIA (FPI-SGIT 2015) and the
European Social Fund through a grant awarded to J. Espinosa.
Fieldwork was partially funded by the European Commission FP6
project FIRE PARADOX (FP6-018505). António Salgueiro, André
Rebelo, Artur Costa, Hermínio Botelho, and Pedro Sal provided
information or assisted with identification and location of PB–
wildfire encounters. ICNF supplied the fire danger rating data.
Marty Alexander reviewed a draft version. Comments by Miguel
G. Cruz, one anonymous reviewer, and the Associate Editor re-
sulted in significant manuscript improvement.
References
Agee, J.K., and Skinner, C.N. 2005. Basic principles of forest fuel reduction treat-
ments. For. Ecol. Manage. 211: 83–96. doi:10.1016/j.foreco.2005.01.034.
Alexander, M.E., and Cruz, M.G. 2012. Interdependencies between flame length
and fireline intensity in predicting crown fire initiation and crown scorch
height. Int. J. Wildland Fire, 21: 95–113. doi:10.1071/WF11001.
Burrows, N., Ward, B., and Robinson, A. 2000. Behaviour and some impacts of a
large wildfire in the Gnangara maritime pine (Pinus pinaster) plantation,
Western Australia. CALMScience, 3: 251–260.
Catchpole, E.A., Alexander, M.E., and Gill, A.M. 1992. Elliptical-fire perimeter-
and area-intensity distributions. Can. J. For. Res. 22(7): 968–972. doi:10.1139/
x92-129.
Catry, F.X., Rego, F., Moreira, F., Fernandes, P.M., and Pausas, J.G. 2010. Post-fire
tree mortality in mixed forests of central Portugal. For. Ecol. Manage. 260:
1184–1192. doi:10.1016/j.foreco.2010.07.010.
Cochrane, M.A., Moran, C.J., Wimberly, M.C., Baer, A.D., Finney, M.A.,
Beckendorf, K.L., Eidenshink, J., and Zhu, Z. 2012. Estimation of wildfire size
and risk changes due to fuels treatments. Int. J. Wildland Fire, 21: 357–367.
doi:10.1071/WF11079.
Cohen, J. 1988. Statistical power analysis for the behavioral sciences. 2nd ed.
Erlbaum, Hillsdale, N.J.
Cruz, M.G., Alexander, M.E., and Dam, J.E. 2014. Using modeled surface and
crown fire behavior characteristics to evaluate fuel treatment effectiveness:
a caution. For. Sci. 60: 1000–1004. doi:10.5849/forsci.13-719.
Cruz, M.G., Alexander, M.E., and Plucinski, M.P. 2017. The effect of silvicultural
treatments on fire behaviour potential in radiata pine plantations of South
Australia. For. Ecol. Manage. 397: 27–38. doi:10.1016/j.foreco.2017.04.028.
Espinosa, J., Madrigal, J., De La Cruz, A.C., Guijarro, M., Jimenez, E., and
Hernando, C. 2018. Short-term effects of prescribed burning on litterfall
biomass in mixed stands of Pinus nigra and Pinus pinaster and pure stands of
Pinus nigra in the Cuenca Mountains (Central-Eastern Spain). Sci. Total Envi-
ron. 618: 941–951. doi:10.1016/j.scitotenv.2017.08.291. PMID:29146076.
Fernandes, P.M. 2009. Examining fuel treatment longevity through experimen-
tal and simulated surface fire behaviour: a maritime pine case study. Can. J.
For. Res. 39(12): 2529–2535. doi:10.1139/X09-145.
Fernandes, P.M. 2015. Empirical support for the use of prescribed burning as a
fuel treatment. Curr. For. Rep. 1: 118–127. doi:10.1007/s40725-015-0010-z.
Fernandes, P.M. 2018. Scientific support to prescribed underburning in southern
Europe: what do we know? Sci. Total Environ. 630: 340–348. doi:10.1016/j.
scitotenv.2018.02.214. PMID:29482142.
Fernandes, P.M., and Rigolot, E. 2007. The fire ecology and management of
maritime pine (Pinus pinaster Ait.). For. Ecol. Manage. 241: 1–13. doi:10.1016/j.
foreco.2007.01.010.
Fernandes, P.M., Loureiro, C., and Botelho, H. 2004. Fire behaviour and severity
in a maritime pine stand under differing fuel conditions. Ann. For. Sci. 61:
537–544. doi:10.1051/forest:2004048.
Fernandes, P.M., Luz, A., and Loureiro, C. 2010. Changes in wildfire severity from
maritime pine woodland to contiguous forest types in the mountains of
northwestern Portugal. For. Ecol. Manage. 260: 883–892. doi:10.1016/j.foreco.
2010.06.008.
Fernandes, P.M., Loureiro, C., and Botelho, H. 2012. PiroPinus: a spreadsheet
application to guide prescribed burning operations in maritime pine forest.
Comput. Electron. Agric. 81: 58–61. doi:10.1016/j.compag.2011.11.005.
Fernandes, P.M., Fernandes, M., and Loureiro, C. 2015. Post-fire live residuals of
maritime pine plantations in Portugal: structure, burn severity, and fire
recurrence. For. Ecol. Manage. 347: 170–179. doi:10.1016/j.foreco.2015.03.023.
Finney, M.A., McHugh, C.W., and Grenfell, I.C. 2005. Stand- and landscape-level
effects of prescribed burning on two Arizona wildfires. Can. J. For. Res. 35(7):
1714–1722. doi:10.1139/x05-090.
Gellie, N. 2005. Effectiveness of fire management measures in pine plantations.
Report for the ACT Department of Urban Services, Canberra, Australia.
Graham, R.T., McCaffrey, S., and Jain, T.B. 2004. Science basis for changing forest
structure to modify wildfire behavior and severity. USDA Forest Service, Gen.
Tech. Rep. RMRS-GTR-120. doi:10.2737/RMRS-GTR-120.
Gutsell, S.L., and Johnson, E.A. 1996. How fire scars are formed: coupling a
disturbance process to its ecological effect. Can. J. For. Res. 26(2): 166–174.
doi:10.1139/x26-020.
Hedges, L.V., Gurevitch, J., and Curtis, P.S. 1999. The meta-analysis of response
ratios in experimental ecology. Ecology, 80: 1150–1156. doi:10.1890/0012-
9658(1999)080[1150:TMAORR]2.0.CO;2.
210 Can. J. For. Res. Vol. 49, 2019
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.

Hirsch, K.G., and Martell, D.L. 1996. A review of initial attack fire crew produc-
tivity and effectiveness. Int. J. Wildland Fire, 6: 199–215. doi:10.1071/
WF9960199.
Jiménez, E., Vega-Nieva, D., Rey, E., Fernández, C., and Vega, J.A. 2016. Midterm
fuel structure recovery and potential fire behaviour in a Pinus pinaster Ait.
forest in northern central Spain after thinning and mastication. Eur. J. For.
Res. 135: 675–686. doi:10.1007/s10342-016-0963-x.
Keyes, C.R., and O’Hara, K.L. 2002. Quantifying stand targets for silvicultural
prevention of crown fires. West. J. Appl. For. 17: 101–109.
Lydersen, J.M., North, M.P., and Collins, B.M. 2014. Severity of an uncharacteris-
tically large wildfire, the Rim Fire, in forests with relatively restored frequent
fire regimes. For. Ecol. Manage. 328: 326–334. doi:10.1016/j.foreco.2014.06.
005.
Lydersen, J.M., Collins, B.M., Brooks, M.L., Matchett, J.R., Shive, K.L., Povak, N.A.,
Kane, V.R., and Smith, D.F. 2017. Evidence of fuels management and fire
weather influencing fire severity in an extreme fire event. Ecol. Appl. 27:
2013–2030. doi:10.1002/eap.1586. PMID:28644577.
Malone, S.L., Kobziar, L.N., Staudhammer, C.L., and Abd-Elrahman, A. 2011. Mod-
eling relationships among 217 fires using remote sensing of burn severity in
southern pine forests. Remote Sens. 3: 2005–2028. doi:10.3390/rs3092005.
Martinson, E.J., and Omi, P.N. 2008. Assessing mitigation of wildfire severity by
fuel treatments —an example from the Coastal Plain of Mississippi. Int. J.
Wildland Fire, 17: 415–420. doi:10.1071/WF06067.
Martinson, E.J., and Omi, P.N. 2013. Fuel treatments and fire severity: a meta-
analysis. USDA Forest Service Res. Pap. RMRS-RP-103WWW.
McCaw, W.L., Gould, J.S., Cheney, N.P., Ellis, P.F.M., and Anderson, W.R. 2012.
Changes in behaviour of fire in dry eucalypt forest as fuel increases with age.
For. Ecol. Manage. 271: 170–181. doi:10.1016/j.foreco.2012.02.003.
Oliveras, I., Gracia, M., Moré, G., and Retana, J. 2009. Factors influencing the
pattern of fire severities in a large wildfire under extreme meteorological
conditions in the Mediterranean basin. Int. J. Wildland Fire, 18: 755–764.
doi:10.1071/WF08070.
Palheiro, P.M., Fernandes, P., and Cruz, M.G. 2006. A fire behaviour-based fire
danger classification for maritime pine stands: comparison of two ap-
proaches. For. Ecol. Manage. 234(Suppl.): S54. doi:10.1016/j.foreco.2006.08.
075.
Peterson, D.L. 1985. Crown scorch volume and scorch height: estimates of post-
fire tree condition. Can. J. For. Res. 15(3): 596–598. doi:10.1139/x85-097.
Peterson, D.L., and Ryan, K.C. 1986. Modeling postfire conifer mortality for long-
range planning. Environ. Manage. 10: 797–808. doi:10.1007/BF01867732.
Pollet, J., and Omi, P.N. 2002. Effect of thinning and prescribed burning on
crown fire severity in ponderosa pine forests. Int. J. Wildland Fire, 11: 1–10.
doi:10.1071/WF01045.
Prichard, S.J., and Kennedy, M.C. 2014. Fuel treatments and landform modify
landscape patterns of burn severity in an extreme fire event. Ecol. Appl. 24:
571–590. doi:10.1890/13-0343.1. PMID:24834742.
Sackett, S.S. 1975. Scheduling prescribed burns for hazard reduction in the
southeast. J. For. 73: 143–147.
Safford, H.D., Stevens, J.T., Merriam, K., Meyer, M.D., and Latimer, A.M. 2012.
Fuel treatment effectiveness in California yellow pine and mixed conifer
forests. For. Ecol. Manage. 274: 17–28. doi:10.1016/j.foreco.2012.02.013.
Stephens, S.L., Collins, B.M., and Roller, G. 2012. Fuel treatment longevity in a
Sierra Nevada mixed conifer forest. For. Ecol. Manage. 285: 204–212. doi:10.
1016/j.foreco.2012.08.030.
Storey, M., Price, O., and Tasker, E. 2016. The role of weather, past fire and
topography in crown fire occurrence in eastern Australia. Int. J. Wildland
Fire, 25: 1048–1060. doi:10.1071/WF15171.
Van Wagner, C.E. 1973. Height of crown scorch in forest fires. Can. J. For. Res.
3(3): 373–378. doi:10.1139/x73-055.
Van Wagner, C.E. 1987. Development and structure of the Canadian Forest Fire
Weather Index System. Can. For. Serv. For. Tech. Rep. 35.
Vega, J., Jimenez, E., Vega, D., Ortiz, L., and Pérez, J.R. 2011. Pinus pinaster Ait. tree
mortality following wildfire in Spain. For. Ecol. Manage. 261: 2232–2242.
doi:10.1016/j.foreco.2010.10.019.
Viedma, O., Quesada, J., Torres, I., De Santis, A., and Moreno, J.M. 2015. Fire
severity in a large fire in a Pinus pinaster forest is highly predictable from
burning conditions, stand structure, and topography. Ecosystems, 18: 237–
250. doi:10.1007/s10021-014-9824-y.
Zald, H.S.J., and Dunn, C.J. 2018. Severe fire weather and intensive forest man-
agement increase fire severity in a multi-ownership landscape. Ecol. Appl.
28: 1068–1080. doi:10.1002/eap.1710. PMID:29698575.
Espinosa et al. 211
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by pfern@utad.pt on 02/07/19
For personal use only.