and long-term impacts of fuels treatments and simulated wildfire on an old-forest species

19 Pages·2015·2.39 MB·English

Evaluating short- and long-term impacts of fuels treatments
and simulated wildfire on an old-forest species
DOUGLAS J. TEMPEL,1,(cid:2) R. J. GUTIE´RREZ,2 JOHN J. BATTLES,3 DANNY L. FRY,3 YANJUN SU,4 QINGHUA GUO,4
MATTHEW J. REETZ,1 SHEILA A. WHITMORE,1 GAVIN M. JONES,1 BRANDON M. COLLINS,5 SCOTT L. STEPHENS,3
MAGGI KELLY,3 WILLIAM J. BERIGAN,1 AND M. ZACHARIAH PEERY1
1DepartmentofForestandWildlifeEcology,1630LindenDrive,UniversityofWisconsin-Madison,Madison,Wisconsin53706USA
2DepartmentofFisheries,WildlifeandConservationBiology,1980FolwellAvenue,
UniversityofMinnesota,St.Paul,Minnesota55108USA
3EcosystemSciencesDivision,DepartmentofEnvironmentalScience,Policy,andManagement,130MulfordHall,
UniversityofCalifornia,Berkeley,California94720-3114USA
4SierraNevadaResearchInstitute,SchoolofEngineering,UniversityofCalifornia,
5200NorthLakeRoad,Merced,California95343USA
5USDAForestService,PacificSouthwestResearchStation,1731ResearchParkDrive,Davis,California95618USA
Citation:Tempel,D.J.,R.J.Gutie´rrez,J.J.Battles,D.L.Fry,Y.Su,Q.Guo,M.J.Reetz,S.A.Whitmore,G.M.Jones,B.M.
Collins,S.L.Stephens,M.Kelly,W.J.Berigan,andM.Z.Peery.2015.Evaluatingshort-andlong-termimpactsoffuels
treatmentsandsimulatedwildfireonanold-forestspecies.Ecosphere6(12):261.http://dx.doi.org/10.1890/ES15-00234.1
Abstract. Fuels-reductiontreatmentsarecommonlyimplementedinthewesternU.S.toreducetherisk
of high-severity fire, but they may have negative short-term impacts on species associated with older
forests. Therefore, we modeled the effects of a completed fuels-reduction project on fire behavior and
California Spotted Owl (Strix occidentalis occidentalis) habitat and demography in the Sierra Nevada to
assess the potential short- and long-term trade-offs. We combined field-collected vegetation data and
LiDARdatatodevelopdetailedmapsofforeststructureneededtoparameterizeourfireandforest-growth
models. We simulated wildfires under extreme weather conditions (both with and without fuels
treatments), then simulated forest growth 30 years into the future under four combinations of treatment
and fire: treated with fire, untreated with fire, treated without fire, and untreated without fire. We
compared spotted owl habitat and population parameters under the four scenarios using a habitat
suitability index developed from canopy cover and large-tree measurements at nest sites and from
previously derived statistical relationships between forest structure and fitness (k) and equilibrium
occupancyattheterritoryscale.Treatmentshadapositiveeffectonowlnestinghabitatanddemographic
ratesupto30yearsaftersimulatedfire,buttheyhadapersistentlynegativeeffectthroughoutthe30-year
period in the absence of fire. We conclude that fuels-reduction treatments in the Sierra Nevada may
providelong-termbenefitstospottedowlsiffireoccursunderextremeweatherconditions,butcanhave
long-term negativeeffects on owls iffiredoes not occur.However,weonlysimulated one fireunder the
treated anduntreated scenariosand therefore had nomeasures ofvariation anduncertainty.Inaddition,
the net benefits of fuels treatments on spotted owl habitat and demography depends on the future
probability that fire will occur under similar weather and ignition conditions, and such probabilities
remaindifficulttoquantify.Therefore,werecommendalandscapeapproachthatrestrictstimberharvest
withinterritorycoreareasofuse(;125hainsize)thatcontaincriticalowlnestingandroostinghabitatand
locates fuelstreatments inthesurrounding areastoreducethepotentialforhigh-severityfireinterritory
core areas.
Key words: California Spotted Owl; fuels treatment; habitat; Sierra Nevada; Strix occidentalis occidentalis; territory
fitness;territoryoccupancy;wildfire.
Received21April2015;revised16July2015;accepted17July2015;published11December2015.CorrespondingEditor:
D.P.C.Peters.
v www.esajournals.org 1 December2015 v Volume6(12) v Article261
TEMPELETAL.
Copyright:(cid:2)2015Tempeletal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the
originalauthorandsourcearecredited.http://creativecommons.org/licenses/by/3.0/
(cid:2)E-mail:[email protected]
INTRODUCTION populationsfromreachingcriticallysmallsizesis
widelyregardedasanimportantpolicyobjective
The management of fire-adapted forests in the (e.g., National Forest Management Act of 1976),
western U.S. is increasingly challenged by the inpartbecausemeetingthehabitatneedsofsuch
need to consider the ecological impacts of speciesmightprotect broader old-forest commu-
wildfire and fuels-reduction treatments intended nities (Temple 1997).
to modify wildfire behavior (Stephens et al. To reduce the potential for large patches of
2013). Historic fire regimes in many of these high-severity fire, forest managers are imple-
forests were characterized by fires burning at menting fuels-reduction treatments in many
intervals of less than 20 years and having western U.S. forests (e.g., USFS 2004). Fuels-
primarily low- and moderate-severity fire effects reduction treatments primarily remove duff,
butinterspersedwithsomeareasofhigh-severity downed wood, shrubs, and smaller trees (i.e.,
effects(Agee1993,SkinnerandChang1996),and surface and ladder fuels), and fire models
somehigh-severityfireapparentlyoccurredwith suggest that these treatments can reduce poten-
regularity (Collins and Stephens 2010, Hanson tial fire spread and intensity across landscapes
and Odion 2014). This type of fire regime (Ager et al. 2007, Finney et al. 2007). However,
resulted in highly heterogeneous landscapes in these treatments also reduce canopy cover and
which vegetation conditions were governed by vertical forest structure, which could have
complex interactions between topography, site negative short-term impacts on old-forest-associ-
productivity,anddisturbance(Collinsetal.2015, ated species such as the spotted owl (Strix
Stephens et al. 2015). However, decades of fire occidentalis).Hypothetically,suchshort-termneg-
exclusion have disrupted historic fire regimes, ative impacts would be outweighed by the
alteredforest structure and landscapevegetation longer-term benefits from reductions in the
patterns, increased forest fuel loads, and led to amount of habitat lost during future wildfires,
increases in the frequency of large fires (West- as has been suggested by previous simulations
erling et al. 2006), as well as increases in (Ageretal.2007,Roloffetal.2012).Similarly,the
proportions and patch sizes of high-severity fire currentmanagementplanforthenationalforests
(Miller et al. 2009, Miller and Safford 2012). In in the Sierra Nevada posits that fuels-reduction
addition, further increases in fire activity are treatments will result in long-term increases in
expected under most climate change scenarios theamountofsuitableCaliforniaSpottedOwl(S.
(Westerling and Bryant 2008, Liu et al. 2013). o. occidentalis) habitat while acknowledging the
High-severity fire effects (defined by .75% potential for short-term negative impacts (USFS
mortality of overstory trees) can impact ecosys- 2004). Indeed, a recent study found that fuels-
tem processes such as erosion rates, stream reduction treatments can negatively impact
sedimentation, and carbon sequestration (Bena- spottedowlpopulationsovershortertimeframes
vides-Solorio and MacDonald 2001, Breshears (,10 years) if they reduce the amount of high-
and Allen 2002), as well as modify forest canopy-cover ((cid:2)70%) forest dominated by trees
structure and wildlife habitat. While some (cid:2)30.5 cm diameter at breast height (dbh) within
wildlife species become more abundant after owl territories (Tempel et al. 2014a). However,
high-severity fire (Smucker et al. 2005, Fontaine whethershort-termimpactsoffuelstreatmentsto
and Kennedy 2012), other species, particularly spotted owls and their habitat in the Sierra
those associated with older forests, may be Nevada will be offset by long-term gains
negatively impacted by habitat loss resulting resulting from reductions in high-severity fire is
from large patches of high-severity fire (e.g., Lee unknown.
et al. 2013). Old-forest species with large home Here, we used fire and forest-growth models
ranges are typically rare and preventing their to simulate how fuels treatments might alter the
v www.esajournals.org 2 December2015 v Volume6(12) v Article261
TEMPELETAL.
effects of fire on spotted owl habitat and wassimulatedunderextremeweatherconditions
demographic rates at the ‘‘fireshed’’scale over a shortly after treatment implementation, we pre-
30-year period in the Sierra Nevada. Firesheds dicted the treatments to reduce the amount of
are contiguous areas with similar fire histories habitatlostduringthefireandtoresultingreater
and have been identified by the U.S. Forest habitat amounts 30 years post-fire because forest
Service as useful landscape units for fuels- growth would be insufficient to compensate for
treatmentplanningandeffectivefiresuppression the lossof overstory trees during this timeframe.
(Bahro et al. 2007, North et al. 2015). Firesheds If no fire was simulated after treatment imple-
are commonly delineated by sub-watershed mentation, we expected that the amount of
boundaries and range in size from ;3,200 to habitat would initially decline, but that similar
16,200 ha within the Sierra Nevada (North et al. amounts would be present on treated and
2015). We chose this spatial scale because of its untreated landscapes after 30 years because of
management relevance (i.e., project planning) forest regrowth (Collins etal.2011). Becauseowl
andbecauseourfield-basedvegetationsampling demographic rates are strongly and positively
would not have been feasible at larger spatial correlatedwiththeamountofhigh-canopy-cover
scales. In contrast to previous studies that relied ((cid:2)70%)forestwithinowlterritories(Tempeletal.
upon simulated treatments (Ager et al. 2007, 2014a), wepredicted that owl demographic rates
Roloff et al. 2012; but see Stephens et al. 2014), wouldfollowsimilarpatternsashabitatamount.
our study involved actual fuels-reduction treat- Thus, we hypothesized that fuels treatments
ments implemented by the U.S. Forest Service would reduce territory fitness and occupancy in
and was intended to assess the efficacy of the short-term, but result in higher fitness and
existingmanagementguidelinesgoverningforest occupancy after 30 years in the event of
management at a bio-regional scale. We inten- simulated fire.
sively sampled the vegetation within field plots
beforeandaftertheimplementedtreatmentsand MATERIALS AND METHODS
coupled this fine-scale vegetation data with
LiDAR data to quantify changes in forest Study area
structure and parameterize fire and forest- Our13,482-haLastChanceStudyArea(LCSA)
growth models. Finally, we linked spotted owl was located within the Tahoe National Forest in
demographic rates to changes in vegetation the central Sierra Nevada, California (Fig. 1).
conditions resulting from fuels treatments and Elevations ranged from 600 to 2,200 m. The
wildfire using data from a long-term demogra- vegetation was primarily mixed-conifer forest
phy study, as few previous studies have simu- dominatedby white fir (Abies concolor),Douglas-
lated the short- versus long-term trade-offs of fir (Pseudotsuga menziesii), incense-cedar (Calo-
fuel treatments on wildlife population dynamics cedrus decurrens), sugar pine (Pinus lambertiana),
(however,seeSchelleretal.2011).Wespecifically ponderosa pine (Pinus ponderosa), and California
considered two demographic parameters at the blackoak(Quercuskelloggii),withlesseramounts
scale of an owl territory: (1) fitness, which we ofotherforesttypesandmontanechaparral.The
defined as the population growth rate (k) LCSA had a Mediterranean climate with an
conferred on resident owls by habitat conditions average of 1,182 mm of precipitation, most of
within the territory (Franklin et al.2000); and (2) which fell as snow, from 1990 to 2008 (Hell Hole
equilibrium occupancy, which is the level at Remote AutomatedWeatherStation).Thehistor-
which occupancy probability at a territory will ic fire regime in this region mainly consisted of
stabilize when colonization and extinction prob- frequent, low-to moderate-severity fire occurring
abilitiesremainconstant(MacKenzieetal.2006). every 5–15 years (Stephens and Collins 2004).
We hypothesized that fuels treatments would AspartoftheexperimentaldesignforSNAMP,
result in a short-term loss of owl habitat, but in the study area was composed of a central
the event of simulated wildfire, these treatments treatment fireshed (4,293 ha) and two adjacent
would result in a long-term increase in owl watersheds to the north and south that together
habitat (relative to the untreated landscape) by served as a control ‘fireshed’ (5,658 ha; Fig. 1).
reducing owl habitat loss to fire. Thus, if a fire We further expanded the study area by an
v www.esajournals.org 3 December2015 v Volume6(12) v Article261
TEMPELETAL.
Fig.1.MapoftheLastChanceStudyAreainthecentralSierraNevada,California,showingthedelineationof
treatment and control firesheds, the locations of four California Spotted Owl (Strix occidentalis occidentalis)
territories used in our demographic analyses, fuel treatments conducted in 2011–2012, the plot network for
ground-basedsamplingofforestvegetationbeforeandafterthefueltreatments,andanaerialoverviewofpartof
the treatedarea before and after the fueltreatments.
additional 3,531 ha of untreated landscape Treatmentswereimplementedon942ha(7.0%of
covered by the LiDAR footprint to incorporate the total study area) as follows: 561 ha of
additionalowlterritories(Fig.1).Fuels-reduction mechanical thinning (tractor and cable), 247 ha
treatments were implemented within the treat- of prescribed fire, and 134 ha of mastication of
ment fireshed by the U.S. Forest Service during shrubs and small trees. Although no treatments
2011–2012aspartoftheSierraNevadaAdaptive were implemented in the control fireshed, the
Management Project (SNAMP; Sierra Nevada 2008 Peavine Fire burned 268 ha within the
Adaptive Management Project 2014). The fuels southern unit of the control fireshed (Fig. 1).
treatments, also known as Strategically Placed Collins et al. (2011) modeled treatments and
Landscape Area Treatments (SPLATs), followed hazardous fire potential in the same study area,
the guidelines specified in the 2004 management but focused on the treatment fireshed only.
plan for national forests in the Sierra Nevada
(USFS 2004). The management plan specified Development of vegetation map
that no trees (cid:2)76.2 cm can be harvested, at least Wedeveloped a pre-treatment vegetation map
40% canopy cover must be retained, and at least using a combination of LiDAR, high-resolution
40% of a stand’s basal area must be retained. digitalcolor-infrared(CIR)aerialimagery,andan
v www.esajournals.org 4 December2015 v Volume6(12) v Article261
TEMPELETAL.
intensive network of field plots. First, we used of vegetation groups. We identified 8 vegetation
LiDARandCIRdatatocreateaninitialpolygon- types onour studyarea—lowshrub,high shrub,
basedmapwherethepolygonsrepresentedareas open true fir, pine forest, cedar forest, young
of homogeneous vegetation in terms of species, mixed-conifer forest, and mature mixed-conifer
vertical structure, basal area, and canopy cover. forest. The dominant vegetation type on the
The mean polygon size was 9.4 ha (range¼0.9– studyareawasmixed-coniferforest(56%mature,
72.6 ha). We collected the LiDAR and CIR data 19% young); the other forest types were present
before the fuels-reduction treatments, and we in lesser amounts (13% cedar, 7% pine, 4% open
sampled vegetation at the field plots before and true fir). Chaparral (low and high shrubs)
after treatment. We then used the field-plot data covered only 1% of the study area. Post-treat-
to impute detailed attributes (e.g., tree lists and mentLiDARwascollectedin2013andwasused
fuelsmodels)foreachpolygon.Thus,wederived to delineate actual treatment areas based on a
twodifferentmaps(withandwithouttreatment), change-detection algorithm to identify where
which we used in fire and forest-growth model- forest structure noticeably changed between the
ing. We used field-plot data to assess the two LiDAR acquisitions. This approach was
accuracyofthepre-treatmentandpost-treatment employed because there can be inconsistencies
maps in terms of canopy cover and large tree between agency-generated treatment polygons
density, which were the variables we used to and actual treatment extent on the ground.
identify spotted owl habitat (see Materials and Wesampledforestvegetationatfieldplotsthat
methods:Assessingeffectsoffuelstreatmentsandfire were spaced at 500-m intervals across the LCSA,
on spotted owl habitat). We found that values for except the southwest corner of the LCSA where
percentcanopycoverandlargetreedensitywere extremetopographyprecludedsampling(Fig.1).
similar for field plots and their associated map We sampled more intensively at 125- and 250-m
polygon, although on average the field plot spacing around instrument locations for a sepa-
values were slightly lower than the map values. rate hydrological study. In August 2008, we also
For percent canopy cover, the average difference intensively sampled the area burned by the
for the pre-treatment map was (cid:3)5.42% canopy Peavine Fire. In total, we sampled 408 plots in
cover(SE¼1.12),andforthepost-treatmentmap 2007–2008 (pre-treatment) and 369 plots in 2013
the average difference was(cid:3)2.05% canopy cover (post-treatment). We briefly summarize the
(SE¼1.21; Appendix A). For large tree density, vegetation sampling here, but refer the reader
the average difference for the pre-treatment map to Collins et al. (2011) for greater detail. We
was ¼(cid:3)7.05 large trees per hectare (SE ¼ 1.51), sampled within 0.05-ha circular plots and re-
and for the post-treatment map the average corded information on individual trees using
difference was ¼ (cid:3)4.30 large trees per hectare three different sampling intensities based on tree
(SE¼1.55; Appendix A). size:(1)throughouttheentireplotfortrees(cid:2)19.5
We contracted with the National Center for cmdbh;(2)withinarandomone-thirdoftheplot
Airborne LiDAR Mapping (National Center for (167 m2) for trees 5.0–19.4 cm dbh; and (3) along
Airborne Laser Mapping 2011) to collect small- a random belt transect (76 m2) for trees ,5.0 cm
footprint, multiple-return airborne LiDAR data dbh. We recorded tree species, vigor, crown
with a point density of 6–10 points/m2 in position, dbh, total height, and height to live
September 2008, and we obtained 1 3 1 m2 crown base (live trees only) for all trees in the
resolution CIR data collected by the National upper two size classes, and species and dbh for
Agriculture Imagery Program (NAIP) in 2005. trees in the smallest size class. In addition, we
After initial processing of the LiDAR and CIR sampled downed wood, litter, duff fuels, and
data, we used an object-based segmentation woody shrub cover on three randomly chosen
approach to delineate polygons of homogeneous transects within each plot. We used the line-
vegetation types. We then applied an unsuper- interceptmethodtosampledownedwoodyfuels
vised classification strategy to label the different (van Wagner 1968, Brown 1974), and we record-
vegetation types based on the Bayesian Informa- ed percent cover and average height for woody
tion Criterion algorithm, which is used to shrubs intersecting each transect.
automatically determine the optimized number We then used the field-plot data to impute
v www.esajournals.org 5 December2015 v Volume6(12) v Article261
TEMPELETAL.
detailedvegetationattributesforeachpolygonof fire behavior parameters over a complex land-
the vegetation map for use in the fire and forest- scape. Topographic inputs such as slope, aspect,
growth modeling. We developed an imputation and elevation were obtained from the LiDAR-
proceduretoassignthreefieldplotstoeachmap derived surface elevation model at 30-m resolu-
polygon based on their similarity in ‘‘gradient tion. We derived forest structure map layers for
space’’ (Ohmann and Gregory 2002). We per- canopy cover, canopy bulk density, canopy base
formedamultivariateanalysisoftheplotdatato height, and canopy height using the imputation
define the gradient space. The definition of the procedure previously described. We calculated
gradient nearest neighbors for each polygon fuel-model assignments using a selection logic
(sensu Ohmann and Gregory 2002) included based on surface fuels and forest structure
topographic variables (e.g., slope, aspect, eleva- measured at the plots (Collins et al. 2011, 2013).
tion), canopy structure (percent canopy cover This approach has proven sufficient at assigning
and an index of large tree density), and vegeta- fuel models based on actual fuel loads rather
tion type. To maintain some of the fine-scale than relying on the Forest Vegetation Simulator
heterogeneityobservedinthefield,weidentified (FVS; Dixon 2002), which has been shown to use
allplotsinthe95thpercentileintermsofnearest fuel models that underestimate fire behavior
neighbor distance for each stand and then (Collins et al. 2013). This approach for assigning
randomly assigned three of those plots to the fuel models was different for treated and
stand. Our pre-treatment map represented con- untreated stands. For untreated stands, we used
ditions after the Peavine Fire occurred (see a regression tree analysis with several response
Materials and methods: Study area) because we variables representing surface fuels: shrub cover,
collected the remotely sensed data and sampled litter,1-to 100-hour woodyfuels, and 1000-hour
additionalfieldplotswithintheburnedareaafter woody fuels. Forest structure and stand vegeta-
thefire.Thetreatment scenario differed from the tion classification were used as independent
no treatment scenario in what field-plot data variables. Model fits were moderate (R2 ¼ 0.3–
were used to impute vegetation attributes for 0.6), but given the known variability in surface
polygons where treatments occurred. For the fuels in mixed-conifer forests (Lydersen et al.
treatment scenario, we used post-treatment tree 2015), we deemed the assignments to be suffi-
listsfrom treated plots (n¼49) forpolygons that cient in describing the generalized fuel condi-
experienced noticeable structural change based tions represented by surface fuel models(Collins
on LiDAR change detection or were confirmed et al. 2011, 2013). For treated stands, post-
on-the-ground to have been burned by pre- treatment fuel models were based on treatment
scribed fire. type and post-treatment fuel measurements.
Prescribed-burn plots were assigned a moder-
Modeling forest dynamics and fire ate-load timber-litter fuel model (Scott and
We considered four scenarios when modeling Burgan 2005). We assigned a low-load, timber-
forestdynamicsandwildfire:(1)withtreatments understory model to initial post-treatment mas-
and with fire; (2) without treatments and with ticated stands based on observed fire behavior
fire;(3)withtreatmentsandwithoutfire;and(4) fromKnappetal.(2011).Thereweretwotypesof
without treatments and without fire. For the tree-harvest treatments: thinning and cable log-
‘‘with fire’’scenarios, we used FARSITE (Finney ging. Prescriptions for the cable-logging units
1998)tosimulateasinglefireforboththetreated indicatedthat the slash was to remain on site, so
and untreated landscape based on the weather we used a moderate-load, timber-slash model
conditions during the 2001 Star Fire, which followedbytimber-understorymodels.Thinning
burned 6,817 ha, including 314 ha on the treatments used whole-tree removal in which
northeast edge of our study area (Fig. 2). slash typically was removed, so we used the
Approximately 39% of this fire burned at high same selection logic for these treatments that we
severity(www.mtbs.gov;accessedon4February used for untreated stands.
2015). FARSITE is a spatially explicit fire-growth We obtained weather information from the
model that uses several topographic, forest- Duncan Remote Automatic Weather Station,
structure, and fuel-model map layers to project limited to the active burning period of the Star
v www.esajournals.org 6 December2015 v Volume6(12) v Article261
TEMPELETAL.
Fig.2.Burn-severitymapofthe2001StarFireand2014KingFirethatburnedneartheLastChanceStudyArea
inthecentralSierraNevada,California.Theburn-severitymapswerecreatedbytheU.S.ForestService(USFS)as
detailedinFincoetal.(2012).WealsoshowthespottedowlProtectedActivityCenters(PACs)thatwereaffected
by thesefires, and fuels-reduction treatments that wereimplemented bytheUSFS from 2006to2014.
Fire (August–September 2001), which served as average of 8 km h(cid:3)1 (range ¼ 0–15 km h(cid:3)1).
the basis of our fire modeling. Moisture content Our ignition location was established in the
for live and dead woody fuels and live herba- northeastcornerofthestudyareawheretheStar
ceousfuelsusedinthemodelwereequivalent to Fireperimeteroverlappedourstudyareabound-
97th percentile weather conditions. Winds were ary. There were other wildfires that burned into
generally easterly in the morning, switching to thestudyarea(2008PeavineFire,2013American
southwest to west during the day, with an Fire), but the 2001 Star Fire location and the
v www.esajournals.org 7 December2015 v Volume6(12) v Article261
TEMPELETAL.
conditions it burned under yielded the highest normal distribution at a ¼ 0.05 (Shapiro-Wilk
potential to burn a large portion of our study test, p ¼ 0.052), so we estimated the standard
area, and in doing so impact more known owl deviationofthe101nesttreediametersandused
sites. The simulation duration was set to allow the10%quantilevalueofanormaldistributionto
the fire perimeter to expand through the entire identify the minimum size of a large tree as 71.3
study area. cmdbh.Thus,90%ofowlnesttreesonourstudy
Weusedthetreelistdatabasesassociatedwith areawereexpectedtobe(cid:2)71.3cmdbh.Wethen
the 2008 pre-treatment field plots when simulat- performed a logistic regression (Hosmer et al.
ing fire under the ‘‘no treatment’’ scenario, and 2013) of owl nesting habitat as a function of
weusedthe2013post-treatmentfieldplotswhen canopy cover and large tree density using data
simulating fire under the ‘‘treatment’’ scenario. collected by Bond et al. (2004) within 0.02-ha
Stand average flame lengths and proportion plots at 25 nest trees and 36 random locations
burned by fire type (surface fire, conditional withinpotentiallysuitableowlnestinghabitaton
crownfire,andactivecrownfire)werecalculated the EDSA (Fig. 3). We identified the following
for both scenarios and used as inputs for fire logistic regression equation for canopy cover
effectssimulationusingthekeywordSIMFIREin (CC; percent) and large tree density (LT; ha(cid:3)1)
FVS with the Fire and Fuels Extension (FFE; using SAS (SAS Institute, Cary, North Carolina,
Reinhardt and Crookston 2003). USA):
For all four scenarios, we then simulated 30
logitðPr½nesting habitat(cid:4)Þ ¼(cid:3)4:141
years of forest growth on the study area in 10-
þ 0:0263 CC
year time steps using FVS with FFE. The
þ 0:052 3 LT: ð1Þ
simulationswereperformedusingtheintegrated
platformArcFuels(Ageretal.2006,Vaillantetal. Theparameterestimateforlargetreedensitywas
2011), which runs FVS-FFE to produce the forest statisticallysignificantata¼0.05(p,0.01,Wald
structure inputs needed for FARSITE. We used chi-squared test statistic [Q ]¼8.71, df¼1), but
W
the western Sierra variant of FVS to simulate theparameterestimate forcanopycoverwasnot
forest dynamics over the simulation periods. (p¼0.19,Q ¼1.72,df¼1).However,weelected
W
Because this variant does not include a ‘‘full- to include CC in the model given that canopy
establishmentmodel,’’usersmustsetparameters coverisknowntobeanimportantcomponentof
for tree regeneration by identifying number, spotted owl nesting habitat (Bias and Gutie´rrez
species, and frequency of establishment. Follow- 1992, Moen and Gutie´rrez 1997).
ing the methods of Collins et al. (2011, 2013), we We used Eq. 1 to estimate the probability that
used a random-number generator, within de- each forest stand (i.e., map polygon) on our
fined bounds, to set the number of seedlings at studyareacontainedsuitableowlnestinghabitat
each time step in FVS while regulating height- under each ofthe four treatment/wildfirescenar-
growth rates to simulate realistic conditions in a ios at four points in simulated time (years 0, 10,
mixed-conifer forest. 20, and 30). Using the values for canopy cover
and large tree density from each map polygon,
Assessing effects of fuels treatments we calculated the probability that the polygon
and fire on spotted owl habitat contained suitable nesting habitat and obtained
We identified canopy cover and large trees as an average probability (weighted by the area of
the most important predictors of spotted owl each map polygon) for the entire study area,
habitatbecausenestlocationswerecharacterized which we refer to hereafter as the habitat
by greater amounts of these elements in the suitability index. We also obtained separate
central Sierra Nevada (Bias and Gutie´rrez 1992, habitat suitability indices for the control and
Moen and Gutie´rrez 1997, Williams et al. 2011). treatment fireshed within the study area (see
Todetermineabiologicallymeaningfuldefinition Materials and methods: Study area) because we
ofalargetree,weexamined101spottedowlnest expected the direct and indirect (i.e., through
treesonthenearbyEldoradoDemographyStudy modification of fire behavior) effects of fuels
Area (EDSA). The size distribution of these nest treatments to be more pronounced near the
trees was not significantly different from a treatments.
v www.esajournals.org 8 December2015 v Volume6(12) v Article261
TEMPELETAL.
Fig. 3. Comparison of (a) canopy cover (percent) and (b) large tree density ((cid:2)71.3 cm dbh; ha(cid:3)1) for 25
California Spotted Owl (Strix occidentalis occidentalis) nest locations and 36 random locations in potentially
suitableowlnestinghabitatontheEldoradoNationalForestinthecentralSierraNevada,California.Thesedata
were used to developa logistic regression model to estimate the probabilityof forest being suitable as spotted
owl nesting habitat; best-fitlogistic regression linesare shown on each graph.
Assessing effects of fuels treatments bytrees(cid:2)30.5cmdbhbecausepreviousanalyses
and fire on spotted owl demography showed that this vegetation type had a strong
Under each of the four treatment/wildfire positive relation with k and w (Tempel et al.
Eq
scenarios, we projected how changes in owl 2014a). On the 2008 pre-treatment map, forest
habitat were expected to affect fitness and stands with (cid:2)70% canopy cover always con-
equilibrium occupancy (wEq) at the spatial scale tained a substantial number of trees (cid:2)30.5 cm
of a spotted owl territory. We defined an owl dbh (mean density ¼ 55.8 ha(cid:3)1, range ¼ 16.0–
territory as the area contained within a 1,128-m 130.7 ha(cid:3)1), so we considered all of these stands
radius (400-ha) circle around each owl territory to be dominated by trees (cid:2)30.5 cm dbh.
center;thisradiuswasequaltoone-halfthemean
Although our inferences were limited by the
nearest neighbor distance between owl territory
small sample size of four territories, the amount
centers on the EDSA (Tempel et al. 2014a). We
of high-canopy-cover forest within these territo-
estimated the territory center as the geometric
ries (mean ¼ 154 ha, range ¼ 102–207 ha) was
mean of the most informative owl location(s)
typicaloftheamountfoundin70otherterritories
from each year that the territory was occupied.
nearourstudyarea(mean¼132ha;Tempeletal.
We used a nest location if one was located that
2014a).
year; otherwise we used the mean of the roost
To assess how changes in the amount of high-
locationsforthatyear.Welocatednestsandroost
canopy-cover forest impacted fitness and w at
sites during surveys conducted annually from Eq
each of the four territories, we used the habitat
2007 to 2013 during the spotted owl breeding
maps developed under each scenario at four
season (April–August; see Tempel et al. [2014a]),
points in simulated time (years 0, 10, 20, and 30)
andwefoundnobarredowls(Strixvaria)onour
to quantify the proportion of each territory
study area. We limited this analysis to four
spotted owl territories that were largely within consisting of high-canopy-cover forest. We esti-
our study area ((cid:2)80% of the 400-ha territory). mated territory fitness and equilibrium occupan-
Three of the territories were occupied by an owl cyfollowingthemethodsofTempeletal.(2014a).
pair every year from 2007 to 2013, and the other For fitness, where we used a stage-based,
was occupied in all but one of those years. For Lefkovitch matrix model parameterized with
the demographic analyses, we defined owl fecundityandsurvivalratestorepresentchanges
habitat as high-canopy-cover forest dominated in the female population size
v www.esajournals.org 9 December2015 v Volume6(12) v Article261
TEMPELETAL.
N 0 u b u b u b N (MacKenzie et al. 2006). Again, based on
J;tþ1 S;t S;t S;t A;t A;t A;t J;t
N u 0 0 0 N previous analyses in Tempel et al. (2014a), we
½ S1;tþ1(cid:4)¼½ J;t (cid:4)½ S1;t(cid:4)
N 0 u 0 0 N estimated extinction probability at each territory
S2;tþ1 S;t S2;t
N 0 0 u u N as a linear function of the hectares of HCF
A;tþ1 S;t A;t A;t
where N , N , N , and N , were the logitðeÞ ¼ (cid:3)1:944 (cid:3) 0:0583ðHCF=10Þ; ð4Þ
J,t S1,t S2,t A,t
number of juvenile, first-year subadult, second-
and we estimated colonization as a function of
year subadult, and adult females at time t,
the logarithm of the hectares of HCF
respectively;u ,u ,andu weretheapparent
J,t S,t A,t
survival rates of juvenile, subadult, and adult logitðcÞ ¼ (cid:3)3:528
femalesfromtimettotþ1,respectively;andb þ 2:1493logð½HCF=10(cid:4) þ 1Þ: ð5Þ
S,t
andb werethefecundityratesforsubadultand
A,t
Aswedidwhenestimating survival,wedivided
adult females at time t, respectively. Fecundity
the amount of HCF by 10 to facilitate model
was the number of female offspring produced
fitting.
per female in the population, assuming a 50:50
sex ratio for fledged owls. Based on previous
RESULTS
analyses in Tempel et al. (2014a), we estimated
survival at each territory as a function of female Effects of fuels treatments on forest structure
age and the logarithm of the hectares of high-
We compared pre- and post-treatment mea-
canopy-cover forest (HCF)
surements at 49 field plots located within the
logitðuÞ ¼ (cid:3)0:005 þ 0:5573age fuels-treatment network (Table 1). Fuels treat-
þ 0:4973logð½HCF=10(cid:4) þ 1Þ ð2Þ ments reduced the mean canopy and woody
shrub cover by ;10% and reduced mean total
whereage¼0forsubadultsand1foradults,and tree density from 540.8 to 263.6 trees/ha. Mean
wedividedtheamountofHCFby10tofacilitate large tree density increased slightly from 20.8 to
model fitting. However, we estimated fecundity 22.8 trees/ha, perhaps because of tree growth
solely as a function of female age because high- during the five years that elapsed between pre-
canopy-cover forest was not a significant predic- and post-treatment measurements. Fuels treat-
tor of reproductive output, ments decreased the amount of 1–1000 hour
woody fuels from 31.5 to 24.9 Mg/ha, whereas
b¼ 0:153 þ 0:1783age ð3Þ
duff fuels increased from 64.2 to 67.2 Mg/ha.
where age ¼ 0 for subadults and 1 for adults.
Using the territory-specific estimates of survival Fire modeling
and fecundity, we then computed a territory- The simulated fire spread across nearly all of
specific fitness (i.e., k) as the dominant eigenval- the study area for both scenarios (with and
ue of the matrix. As noted in Tempel et al. without treatment) because of the prevailing
(2014a),weexpectedourestimatesoffitnesstobe winds (Fig. 4). Fuels treatments reduced the
biased low because (1) we did not incorporate intensity of the fire, as evidenced by the
immigration intothe projection matrix, and (2) if predicted flame lengths, with the greatest reduc-
an individual was not resighted for one or more tions occurring within treated areas. Overall,
years and was then resighted on a new territory, when fire occurred on the untreated landscape,
we removed the portion of its capture history at 70.2%, 16.6%, 9.3%, and 3.9% of the study area
the original territory (which lowered the esti- experienced flame lengths of ,2, 2–4, 4–8, and
mates of annual survival) to avoid making .8 m, respectively. In contrast, when fire
assumptions about the owl’s location during the occurred on the treated landscape, 76.2%,
intervening period. Nevertheless, differences in 14.3%, 6.8%, and 2.7% of the study area burned
fitness allowed us to evaluate the relative at these flame lengths. Collins et al. (2011) noted
simulatedeffectsoffuelstreatmentsandwildfire. that flame lengths .2 m often corresponded to
We calculated equilibrium occupancy (w ) areas with crown fire initiation (i.e., torching).
Eq
from the territory extinction (e) and colonization Differences in fire behavior between the two
(c) rates at each territory where w ¼ c/(c þ e) scenarios (i.e., a greater proportion of the fire
Eq
v www.esajournals.org 10 December2015 v Volume6(12) v Article261