Biological Conservation 143 (2010) 2559–2569

Contents lists available at ScienceDirect

Biological Conservation
journal homepage: www.elsevier.com/locate/biocon

Learning from a ‘‘benign neglect strategy” in a national park: Response

of saproxylic beetles to dead wood accumulation
Jörg Müller a,*, Reed F. Noss b,**, Heinz Bussler c, Roland Brandl d
a
Bavarian Forest National Park, Freyunger Str. 2, 94481 Grafenau, Germany
b
Department of Biology, University of Central Florida, Orlando, FL 32816-2368, USA
c
Bavarian State Institute for Forestry, Am Hochanger 11, 85354 Freising, Germany
d
Department of Ecology, Animal Ecology, Philipps-Universität Marburg, Karl-von-Frisch-Str. 8, 35043 Marburg, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Increasing demands for firewood owing to rising energy costs have accelerated discussions about the
Received 8 July 2009 amount of dead wood needed for conservation. A sharp increase in dead wood caused by bark beetles
Received in revised form 30 May 2010 in a German national park provides lessons for management of commercial and protected forests. We
Accepted 24 June 2010
investigated the effects of dead wood due to bark beetle infestation as well as tree senility on abundance
Available online 14 July 2010
and richness of saproxylic species of beetles. Increasing amounts of spruce dead wood and opening of the
canopy by bark beetles had positive effects on the abundance of host-generalist, conifer-specialist, and
Keywords:
red-listed saproxylic beetles. Broadleaf specialists were positively associated with the amount of broad-
Salvage logging
Rewilding
leaf dead wood and negatively associated with canopy openness. Gradient analysis of beetle assemblages
Threatened species revealed two major environmental axes: canopy openness and amount of dead wood. We found a thresh-
Ips typographus old for community divergence at a canopy openness of 23% (confidence interval CI: 11–49) and at an
Bark beetle amount of dead wood of 64 m3 ha 1 (CI: 35–160). Critically endangered species served as indicators of
Bavarian Forest National Park dense and open forests, but only when the amount of dead wood was high. Our results suggest that,
Forest dynamics to maintain saproxylic beetle assemblages, the amount of dead wood in commercial montane forests
(at present 15 m3 ha 1) needs to be tripled, with a focus on broadleaf wood in dense stands and spruce
wood in open stands. For large protected areas in Europe, our data suggest that bark beetle infestation
and senescence without active forest management improves habitat conditions for saproxylic beetles.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction roles in the ecosystem as decomposers, predators, or vectors for

fungi (Paine et al., 1997). Only a minority of saproxylic species
Approximately 25% of all species in European forests (mainly are able to kill living trees (i.e., 34 of the 122 scolytids on trees
fungi and beetles) are involved in the decomposition of dead wood in Central Europe; Grodzki et al., 2006; Lieutier et al., 2004; Parker
(Jonsson et al., 2005). Human activities have reduced the amount et al., 2006). In forests with no economic constraints on manage-
of dead wood in managed forests considerably, such that many ment, these ‘‘pest” species can play a positive role as ecosystem
species associated with dead wood appear on Red Lists (Grove, engineers (Jones et al., 1994) by controlling the availability of re-
2002). To support viable populations of saproxylic species, forest sources to other species (Buse et al., 2008).
authorities are considering various strategies for increasing dead Over the last 20 years, bark beetle outbreaks occurred in the
wood in managed forests (Davies et al., 2008; Wikberg et al., USA (mountain pine beetle; Raffa et al., 2008) and Central Europe
2009). Recently, however, rising energy costs have increased the (European spruce bark beetle; Schelhaas et al., 2003), (Baláž,
price of firewood. This economic pressure has refreshed discus- 2009; Grodzki, 1998; Jonášová and Pracha, 2004; McFarlane
sions about the amount of dead wood in managed forests (Jonsell, et al., 2006). Furthermore, a spread of such outbreaks to the boreal
2007). and alpine zones is predicted by models incorporating global
Saproxylic beetles are defined as those that depend on dying warming (Jönsson et al., 2009; Williams and Liebhold, 2002).
and/or dead wood as well as on wood-inhabiting fungi during Therefore, the ecological and economical importance of such
some part of their life cycle (Speight, 1989). They play important large-scale disturbances in forests might become even larger in
the future (Schroeder, 2007). Up to now, however, discussion about
the ‘‘engineering” role of bark beetles has focused mainly on re-
* Corresponding author.
** Corresponding author. sponses of mammals and birds (Koprowski et al., 2005; Martin
E-mail address: [email protected] (J. Müller). et al., 2006).

0006-3207/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biocon.2010.06.024
2560 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569

Insect outbreaks are one of several sources of disturbance in for- 3. Managed forests in the northern part of the national park and in
ests (Schelhaas et al., 2003). Like fire and windstorm events, insect the buffer zone around the park are subject to intensive salvage
outbreaks create substantial amounts of dead wood and open the logging of all bark-beetle-infested spruce trees.
canopy across large areas (Schroeder, 2007). The majority of stud-
ies investigating saproxylic beetles have concentrated on logging We established 293 plots of 0.1 ha, each along four transects
practices or small-scale disturbance, such as senescence of single (total length = 29.3 km) spanning the altitudinal gradient, plus an
trees or crown damage by windstorms in old-growth stands; only additional 500-m transect in an old-growth stand at higher alti-
a few studies have investigated the impact of medium- to large- tude to obtain data on a beech-fir old-growth stand at higher ele-
scale natural disturbances on saproxylic beetles (Table S1/S2). vation. We sampled beetles from 126 randomly selected plots with
The latter are biased toward fire and wind (Bouget and Duelli, a minimum distance of 100 m (21 in old-growth relicts, 52 in
2004; Grimbacher and Stork, 2009). To the best of our knowledge, unmanaged areas, 53 in managed areas; Bässler et al., 2008; Müller
only one study to date has investigated the effect of bark beetle and Brandl, 2009).
disturbance on the diversity of saproxylic beetles (Müller et al.,
2008a). 2.2. Beetle data
In Central Europe, the Bavarian Forest National Park was the
first protected forest affected on a large scale by bark beetles. Logistical constraints dictate the methods of beetle sampling,
Therefore, this park serves as a pilot study area for Central Europe, particularly in remote areas. We followed the suggestion of
from which management guidelines can be developed for commer- Martikainen and Kouki (2003) for sampling saproxylic beetles in
cial forests and strictly protected areas with a ‘‘benign neglect” boreal forests by combining various methods (Fig. S1). We used
strategy (Zahner, 1992). Such a strategy is not always appreciated three different methods at each plot: (1) in the center of each plot,
aesthetically by the public (Lindenmayer et al., 2004; Müller and we installed one flight-interception trap consisting of a crossed
Job, 2009; Stokstad, 2006). The objectives of our study are twofold: pair of transparent plastic shields (40  60 cm); (2) to sample
species that are flightless for most of their life cycle (e.g., genus
1. To analyze the relationship between features of natural dynam- Acalles), we installed one pitfall trap under each flight-interception
ics and the abundance, species richness, and community com- trap (for more details, see Müller and Brandl, 2009), and (3) to
position of saproxylic beetles of different levels of host avoid a bias of capturing only active individuals, an experienced
specialization. entomologist (H. Bussler) searched appropriate substrates in each
2. To determine the minimum amount of dead wood needed for 0.1-ha plot for 45 min in July 2007 over a period of 3 weeks, under
the conservation of comprehensive saproxylic beetle assem- sunny and warm weather conditions.
blages and the minimum level of canopy openness needed for All the sampled beetles were identified to the species level
species of open areas. (Freude et al., 1964–1983) and classified as saproxylic or non-
saproxylic according to Schmidl and Bußler (2004). Because the
2. Methods bark beetle I. typographus is the main agent for providing spruce
dead wood, we excluded this species from our analyses. We
2.1. Study area and study sites grouped all saproxylic species according to their host species dur-
ing larval development into the following guilds (Gibb et al., 2006):
To assess the effects of dead wood accumulation and canopy (1) specialists on broadleaf trees, (2) specialists on coniferous trees,
opening by bark beetle infestations and single-tree senility on and (3) generalists living on dead wood of broadleaf and coniferous
saproxylic beetles, we established sampling plots within trees (Koch, 1989–1992; Köhler, 2000). In addition, we quantified
24,000 ha of the Bavarian Forest National Park, founded in 1970 the number of red-listed species, summarizing the species of the
in the southeastern corner of Germany (for a map, see Müller IUCN classifications from vulnerable to regionally extinct using
and Brandl, 2009). The cool, temperate, montane forests of the park the most up-to-date Red List for Bavaria (Schmidl et al., 2003).
are dominated by spruce (Picea abies) at 1150–1430 m, and at low- Remaining species were classified as others. For each group, we to-
er elevations above 650 m by mixed stands of spruce, beech (Fagus taled the number of individuals as a surrogate of abundance and
sylvatica), and fir (Abies alba). The climate and acidic soils are rather the number of species per sampling plot. After visual inspection,
homogenous (Bässler et al., 2008). Although trees were logged in the abundance estimates, the number of species of broadleaf and
these mountains to produce charcoal for glass production and to conifer specialists and the number of red-listed species were
create pastures, human influence was low up to the mid 19th cen- log10(x + 1) transformed to approach a normal distribution. Use
tury. At that time, after large windstorms and subsequent damage of a linear model for all dependent variables enabled us to compare
by bark beetles, modern forestry began and changed tree species the explanatory power of all models using R2. We used the pres-
composition. Before long, old-growth stands remained only as ence/absence data of species, including only species with an occur-
small patches of ca. 100 ha. Coupled with these changes, some bee- rence of at least three plots, for indirect gradient analyses.
tle species associated with old forests (i.e., Peltis grossa and Lacon
lepidopterus) became extinct. Today, we can divide the park area 2.3. Environmental variables
into three categories of forest management:
We restricted our environmental sampling to 0.1-ha plots,
1. Old-growth relict forests are interspersed small stands domi- which are sufficiently large to avoid unrealistic high estimates
nated by veteran trees >400 years old, which have not been but small enough to capture small-scale variation (Müller et al.,
logged for >50 years; in these stands, dead wood of Silver Fir 2008b; Økland et al., 1996). From a comprehensive set of environ-
and European Beech increased during these 50 years because mental variables (Bässler et al., 2008), we selected four variables
of senility, crown damage, and single tree falls; but also spruce that characterize forest dynamics caused by bark beetle infestation
trees were killed by bark beetles. and single-tree senility. We calculated the amount of dead wood
2. Unmanaged forests in the southern part of the national park are (m3 ha 1) on an area of 1 ha of (1) Silver Fir, (2) Norway Spruce,
dominated by spruce, which have died back mainly because of and (3) broadleaf trees (consisting almost entirely of European
infestation by bark beetles (Ips typographus) across large areas. Beech, with <1% of Mountain Ash and maple). We included all
This process started in the early 1990s (Müller et al., 2008a). pieces of wood with a diameter P0.12 m (the standard minimum
 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569 2561

in dead wood inventories in Central Europe) by up-scaling from which does not provide adjusted R2. To test the significance of
0.1 ha in a circle around the traps. Volume equations of the stand- the influence of our independent fractions on beetle data, we ap-
ing dead trees are based on tree species and diameter at breast plied a permutation test with 1000 permutations using the func-
height; the volume of logs and snags was calculated assuming a tion anova in varpart. To determine the errors of our independent
cylinder and using the length and diameter at half length (Bässler fractions, we used 1000 bootstrapping samples, which allowed
et al., 2008). We considered Silver Fir and Norway Spruce sepa- the construction of 95% confidence intervals.
rately because fir is host to some typical coniferous beetle species To display variation in species composition among the study
as well as to some species that inhabit broadleaf trees. After visual plots, we used unconstrained correspondence analysis (CA).
inspection of the distribution of our data, we log10(x + 1) trans- Furthermore, we used the function adonis (analysis of variance
formed the amount of dead wood per plot, which resulted in an al- between groups of assemblages) in vegan with Bray-Curtis dissim-
most normal distribution for further analyses. A square-root ilarity to test for significant differences in species composition
transformation produced almost identical results. Following Siito- between the three management types (managed, unmanaged,
nen et al. (2000), we additionally calculated the diversity of dead old-growth relict) with a Bonferoni correction for multiple testing
wood (Fig. S3). (4) Canopy openness was measured by airborne la- (Anderson, 2001). To identify community shifts along a gradient of
ser scanning using the penetration rate of LiDAR impulses at 2 m increasing dead wood amount and penetration rate (Andersen
above ground (Müller and Brandl, 2009). Across our plots, the pen- et al., 2009), we applied a conditional inference tree using the func-
etration rate decreased with increasing bark beetle attack. More- tion ctree in the package party (Hothorn et al., 2006) with total
open stands may be important for many saproxylic organisms in amount of dead wood or penetration rate as predictor and site
montane areas because these stands provide warm and dry habi- scores of correspondence analysis as dependent variable. This tree
tats, as well as flowers (Bouget and Duelli, 2004). method is robust for all types of distribution of the independent
To control for effects independent of the forest dynamics in our and dependent variables (Hothorn et al., 2006). A confidence inter-
montane area, we used elevation (Körner, 2007) and the binary val for the threshold was calculated using 1000 bootstraps (Roff,
variable habitat continuity (Jonsson et al., 2005) as covariates; 2006). For testing of nestedness in our community, we used the
the 21 plots in old-growth relicts were scored as plots with long function nestedtemp (Rodríguez-Gironés and Santamaria, 2006) in
habitat continuity. Habitat continuity is difficult to quantify vegan. To identify species with a preference for low or high
(Sverdrup-Thygeson, 2001), but could be important because some amounts of dead wood, and dense and open canopy stands identi-
species of saproxylic beetles depend on a continuous availability of fied by conditional inference trees, we applied indicator species
old decaying trees (Jonsson et al., 2005; Speight, 1989). analysis for species (p < 0.05) with a minimum occurrence in three
plots and considered their abundance (Dufrêne and Legendre,
2.4. Statistical analyses 1997).

We analyzed the relationship of beetle abundance and species

richness to habitat factors using multiple regressions. To check 3. Results
the residuals of our models for spatial independence (Dormann
et al., 2007), we used cross-correlograms provided by the add-on Among managed forests, unmanaged forests, and old-growth
package ncf (Bjørnstad and Falck, 2001) in R (R Development Core relicts, we found clear differences in the amount and composition
Team, 2008). The residuals of our models showed no spatial auto- of dead wood (Fig. 1). The highest amount of dead wood, domi-
correlation even at very small distances (Fig. S2); therefore, we ig- nated by spruce and caused by bark beetle infestation, was found
nored space in subsequent analyses. To separate the effects of in unmanaged areas (Fig. 2, Table S3). The amount of dead wood
abundance (individuals of beetles) and species richness, we first in portions of the national park subjected to salvage logging was
concentrated our analysis on the number of individuals of each about one order of magnitude lower than in unmanaged areas
group; for the analysis of species richness, we used the log10 num-
ber of individuals as a covariate (Gotelli and Colwell, 2001). Total dead wood
Broadleaf dead wood
The amount of dead wood in our plots was highly correlated
Silver Fir
with the diversity of dead wood (r2 > 0.7; Fig. S3). This fact con-
Norway Spruce
strains any attempt to separate the two factors in observational
studies. Therefore, we concentrated in further analyses on the
amount of dead wood because it is easier to measure and to com-
municate to foresters and conservationists. For generalists, which Managed
can feed on both conifers and broadleaf trees, we used the total
amount of dead wood as the predictor; for specialists, both com-
mon and red-listed species, we used the amount of dead wood sep-
arated in our three tree species groups. Elevation and habitat Unmanaged
continuity were included as covariates in all models. We detected
no interactions between predictors.
To compare the explanatory power of large-scale disturbance
by bark beetles, with that of small-scale effects due to tree senility,
Old-growth relict
we built three predictor sets of environmental variables (Bark bee-
tle: dead wood of spruce, penetration rate; Senility: dead wood of
fir and beech; Control: elevation and habitat continuity). We parti-
0 100 200 300 400
tioned the variation of responses with respect to the three explan- 3 -1
atory variable sets using the function varpart in vegan (Oksanen Amount of dead wood m ha
et al., 2006). This application uses partial multiple regression anal-
Fig. 1. The amount of dead wood (mean and standard deviation) of the various tree
ysis. We calculated adjusted R2 values because this is the only species in the three management-type areas within the Bavarian Forest National
unbiased method (Peres-Neto and Jackson, 2001). The community Park. The shaded box shows the range of dead wood in mixed montane Norway
was analyzed with a partial canonical correspondence analysis, Spruce–Silver Fir–European Beech forests in Europe (Christensen et al., 2005).
2562 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569

50

40

Number of species
30

20

10

0
10 100 1000
Number of individuals

Fig. 3. Correlation of individuals and species of saproxylic beetles sampled with

direct search, flight-interception traps, and pitfall traps at 126 sites. Ips typgrog-
aphus, the main agent of spruce dead wood, is excluded.

After excluding the bark beetle I. typographus, our final data set
comprised 12,257 individuals of saproxylic beetle species belong-
ing to 280 species. Flight-interception traps yielded 244 species
(103 exclusive), pitfall traps 33 species (1 exclusive), and direct
searching 164 species (22 exclusive). Seventy-eight species were
red-listed species. Most saproxylic beetle species (113) were gen-
eralists, as judged by their larval hosts; 90 species inhabited dead
wood of coniferous trees, and 77 species inhabited dead wood of
broadleaf trees. The number of species was clearly correlated with
the number of individuals (Fig. 3).
The variance partitioning with three predictor sets (bark beetle,
senility, control) revealed that bark beetle had the highest indepen-
dent effect on the abundance of generalists, conifer specialists, and
red-listed species (Fig. 4). The abundance of common species and
broad-leaved specialists was more affected by single-tree senility
of broad-leaved and fir trees. Our linear models with abundance
as a dependent variable showed that amount of dead spruce had
a positive effect on red-listed species. The amount of dead wood
of broadleaf trees had a positive influence on the abundance of
broadleaf specialists, red-listed species, and common species; the
total amount of dead wood had a positive effect on the abundance
of generalists (Table 1). Conifer specialists were positively affected
only by increasing penetration rate, which was important for al-
most all dependent variables. Broadleaf specialists were the only
group negatively affected by increasing penetration rate. The mul-
tiple R2 showed that the model for red-listed species had the high-
est explanatory power, which decreased to the model for other
species (Table 1). After taking into account the log10(number of
individuals), the species richness of generalists was significantly
and positively influenced by the total amount of dead wood. Spe-
cies richness of conifer specialists increased, whereas species rich-
ness of red-listed species decreased with increasing penetration
rate (Table 1).
For community composition, the independent effects of the var-
iance partitioning were similar for the data sets capturing the ef-
fects of bark beetle, senility, and elevation (control). The analysis
of the composition of beetle assemblages using unconstrained
correspondence analysis supported the above findings. The first
ordination axis extracted by correspondence analysis was corre-
Fig. 2. The amount of dead wood in the Bavarian Forest National Park caused by lated to the penetration rate (Fig. 5; r2 = 0.83) and elevation
natural dynamics of large-scale bark beetle disturbance and small-scale tree senility
(Fig. 5; r2 = 0.76), whereas the second axis represented a gradient
varies greatly from 5 m3 ha 1 to 700 m3 ha 1.
of the amount of dead wood (from low to high resource availabil-
ity; Fig. 5; r2 = 0.65). Therefore, the second axis mirrors the
and was similar to that found in the adjacent, commercial forests increase in the amount of dead wood in unmanaged areas on the
(14 m3 ha 1; dashed line in Fig. 1). composition of beetle communities. A constrained canonical
 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569 2563

Generalists Red-listed species

80 ** 80
***
**
40 40 *

0 0

-40 -40

Conifer specialists Common species

80 80 ** **
**
Independent effects %

40 40

0 0

-40 -40

Broadleaf specialists Community

80 80

** ** ** **
40 40
**

0 0

-40 -40
BB SE CO BB SE CO

Fig. 4. Exclusive explanatory contribution of bark beetle infestation (BB), single-tree senility of broadleaf and Silver Fir (SE), and control variable sets (CO) with elevation and
habitat continuity, of the total explained variance for each dataset based on variance partitioning, using the function varpart in vegan (Oksanen et al., 2006) for the abundance
and a canonical correspondence analysis application for community. To test the significant influence of each independent fraction on the independent variables, a
permutation test was applied using the function anova in vegan; significant p-values are indicated by asterisks: p < 0.05, p < 0.01, p < 0.001. To display the variability of
our independent fractions, a bootstrapping of variance partitioning was additionally applied, and the 95% confidence interval is shown in the box plots by whiskers. Note that
the significance by permutation and the confidence intervals shown by the whiskers are not related to each other. Also note that the target variables differ significantly from
each other (vector, community matrix); they cannot be compared quantitatively among each other.

correspondence analysis using penetration rate as the single pre- growth relicts. However, the community composition of these
dictor revealed a constrained inertia of 0.29 (expected inertia two forest categories rich in dead wood were distinct (adonis post
assuming random distribution of species; 95% CI: 0.04–0.07). The hoc difference, p < 0.01). A simulation test of the reliability of the
total amount of dead wood as single predictor led to a constrained ordination axis and its correlation with penetration rate and the to-
inertia of 0.16. This was again significantly larger than expected tal amount of dead wood showed two important results (Fig. S4).
from a random distribution (95% CI: 0.04–0.07). Bootstrapping First, the comparatively low proportion of variance summarized
with 1000 replicates revealed that the constrained inertia of the by the first two axes of the CA is a result of the large number of
penetration rate was significantly (p < 0.001) larger than the value community samples. Secondly, when more than 70 communities
for dead wood. were sampled, the penetration rate was always significantly corre-
The unmanaged plots followed the direction of increasing lated with axis 1 and the amount of total dead wood was corre-
amount of dead wood along the second axis, similar to the old- lated with axis 2. The application of an NMDS or a direct
2564 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569

Table 1
Models estimated by multiple linear regressions 1 for saproxylic beetles grouped according to substratum host specification and level of endangerment (for classifications, see
Section 2). Estimates are based on mean-zero unit variance standardized independent variables; na = not analyzed.

Dependent variable R2 Individuals Elevation Habitat continuity Spruce Broadleaf Fir Total dead wood Penetration rate
Abundance
Generalists 0.37 na 0.10 0.15* na na na 0.36*** 0.30***
Broadleaf specialists 0.34 na 0.12 0.00 0.04 0.41 0.08 na 0.40
Conifer specialists 0.27 na 0.05 0.21 0.05 0.01 0.02 na 0.52***
Red-listed species 0.43 na 0.02 0.14 0.28*** 0.24*** 0.09 na 0.43***
Other species 0.17 na 0.01 0.15 0.02 0.34*** 0.07 na 0.29**
Species
Generalists 0.61 2.38*** 0.92*** 0.05 na na na 1.08*** 0.16
Broadleaf specialists 0.88 0.71*** 0.08* 0.01 0.01 0.01 0.00 na 0.01
Conifer specialists 0.76 0.44*** 0.01 0.00 0.01 0.03 0.02 na 0.11*
Red-listed species 0.71 0.51*** 0.02 0.04 0.02 0.04 0.02 na 0.10*
Other species 0.49 4.67*** 1.42 0.48 0.35 0.10 0.66 na 0.82
*
p < 0.05.
**
p < 0.01.
***
p < 0.001.

100
Indicator spec.

b d
Penetration rate (%)

25 Common
Red-listed
15 Individuals 1000
5

< 23 % > 23 %

49
100

23
11
10
0 -3 -2 -1 0 1 2 3
Second axis scores

3
a
Indicator spec.

Total dead wood 25 Common

Red-listed
c
Spruce DW 15
CA Axis 2: eigenvalue 0.27

< 64 m³/ha > 64 m³/ha

Penetration
rate Fir DW

Broadleaf DW
0 Elevation

Managed
Unmanaged
Old-growth relict

-3
-3 -2 -1 0 1 2 3 5 10 35 64 160 700
CA Axis 1: eigenvalue 0.37 Total amount of dead wood (m 3 ha-1)

Fig. 5. Unconstrained correspondence analysis of presence/absence data of 126 saproxylic beetle samplings with species occurring in at least three plots. The percentage of
explained variance is 5% for axis 1 and 3.3% for axis 2. The environmental variables were fitted on the plot using envfit in vegan. (a) Scores of each site in the three different
types of management areas. (b) Scatterplot showing the site scores of the first correspondence analysis axis versus penetration rate. The horizontal line indicates the
threshold value for the maximum difference in community with a 95% confidence interval (gray shading); the bar plot at the top shows the number of species with a
significant indicator value for dense and open forests. (c) Scatterplot showing the site scores of the second correspondence analysis axis versus dead wood amount. The
vertical and the gray area again shows the threshold with the 95% confidence interval; the bar plot at the top shows the number of indicator species. (d) Scatterplot of the
scores of the second axis plotted against the number of individuals.

ordination using canonical correspondence analysis supported our When we applied the conditional inference tree to the corre-
interpretation of the first two ordination axes (Fig. S5/S6). Taking spondence analysis scores (Fig. 5) of the first axis with penetration
both results into account, we conclude that our results are robust. rate as predictor, a split (=threshold) with a non-linear shift
Note also that our results are not due to differences in abundance (Fig. 5b) of community occurred at 23% penetration rate (95% CI:
between plots, because the analysis was restricted to the presence/ 11–49%). The conditional inference tree of the site scores of the
absence of species. second axis revealed a split at 64 m3 ha 1 of dead wood (95% CI:
 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569 2565

Table 2
Preference of single species [taxonomy according to Köhler and Klausnitzer (1998)] for dense and open canopies (threshold 23% canopy openness = CO), and for low and high
amount of dead wood (threshold 64 m3 ha 1 dead wood = DW) stands, using indicator species analysis (IV = indicator value) (Dufrêne and Legendre, 1997). Conservation status
according to the Red List of Bavaria (Schmidl et al., 2003): VU = vulnerable, EN = endangered, CR = critically endangered; substratum according to Schmidl and Bußler (2004):
fresh = fresh dead wood, decomposed = decomposed dead wood, fungi = wood-inhabiting fungi.

Species Host speciation Substratum Red List IV DW p DW IV CO p CO

Dense forest with low volume of dead wood
Salpingus ruficollis Broadleaf Fresh 41.2 <0.001 47.3 <0.001
Xyloterus domesticus Broadleaf Fresh 38.8 <0.001 39.5 <0.001
Xyleborus dispar Broadleaf Fresh 16.4 0.022 18.3 0.007
Dense forest with high volume of dead wood
Bolitophagus reticulatus Broadleaf Fungi VU 20.3 0.005 16.3 0.043
Ceruchus chrysomelinus Generalist Decomposed EN 11.1 0.040 15.4 0.005
Dense forest
Cychramus variegatus Generalist Fungi 50.4 <0.001
Melanotus castanipes Generalist Decomposed 47 0.026
Ernoporicus fagi Broadleaf Fresh 27.9 0.021
Quedius xanthopus Generalist Decomposed 27.1 <0.001
Micrambe abietis Conifers Fungi 22.5 0.022
Quedius plagiatus Generalist Decomposed VU 21.3 0.003
Acalles camelus Broadleaf Decomposed 17.1 0.004
Atrecus pilicornis Conifers Decomposed VU 15.9 0.012
Hedobia imperialis Broadleaf Decomposed 14.6 0.007
Acalles hypocrita Broadleaf Decomposed 14.1 0.002
Cis glabratus Generalist Fungi 11.9 0.021
Open forest with low volume dead wood
Ernobius abietis Conifers Decomposed 20.8 0.0114 19.5 0.03
Open forest with high volume dead wood
Anaspis rufilabris Generalist Decomposed 57.5 0.017 82.5 <0.001
Ampedus auripes Conifers Decomposed VU 28.2 0.0066 51.6 <0.001
Ampedus nigrinus Generalist Decomposed 46.3 <0.001 50.7 <0.001
Pityophthorus pityographus Conifers Fresh 52.9 <0.001 47.8 <0.001
Dasytes niger Generalist Decomposed 28.9 0.0024 44.8 <0.001
Ampedus aethiops Conifers Decomposed 28.8 0.0012 42.8 <0.001
Oxymirus cursor Generalist Decomposed 26.1 0.0326 40.5 <0.001
Judolia sexmaculata Conifers Decomposed EN 21.3 0.0032 32.3 <0.001
Pteryngium crenatum Generalist Fungi VU 39.4 0.0002 28.0 0.002
Corymbia rubra Conifers Decomposed 21 0.0058 26.6 <0.001
Thymalus limbatus Generalist Fungi VU 25.4 <0.001 26.5 <0.001
Dictyopterus aurora Generalist Decomposed 24.6 0.0066 24.6 0.005
Crypturgus cinereus Conifers Fresh 28.6 <0.001 22.9 0.008
Diacanthous undulatus Generalist Decomposed VU 17.8 0.001 19.4 <0.001
Crypturgus pusillus Conifers Fresh 13.1 0.0268 14.4 0.01
Tachyta nana Generalist Decomposed 11.7 0.0046 11.3 0.005
Hadreule elongatulum Conifers Fungi VU 8.3 0.0216 8.1 0.026
Open forest
Rhagium bifasciatum Generalist Decomposed 47 0.04
Hylastes cunicularius Conifers Fresh 42.4 0.012
Dryocoetes autographus Conifers Fresh 37.4 0.007
Polygraphus poligraphus Conifers Fresh 33.1 0.004
Rhizophagus ferrugineus Conifers Fresh 28.3 <0.001
Dasytes obscurus Conifers Decomposed 26.1 <0.001
Rhyncolus ater Generalist Decomposed 26 0.034
Stenurella melanura Generalist Decomposed 25.2 <0.001
Rhizophagus dispar Generalist Fresh 23 0.035
Leptusa fumida Generalist Decomposed 14.5 0.001
Nudobius lentus Generalist Fresh 11.3 0.008
Anostirus castaneus Generalist Decomposed 9.7 0.013
Hylurgops palliatus Conifers Fresh 9.7 0.011
Low volume of dead wood
Corticarina lambiana Conifers Fungi 34.4 0.006
Corticaria abietorum Conifers Fungi VU 26 0.041
Xylechinus pilosus Conifers Fresh 19.7 <0.001
Tetratoma ancora Broadleaf Fungi VU 15.9 0.003
Triplax russica Broadleaf Fungi VU 15 0.021
High volume of dead wood
Ampedus erythrogonus Generalist Decomposed VU 25.4 0.006
Gyrophaena boleti Generalist Fungi 24.7 0.002
Ostoma ferruginea Generalist Fungi VU 21.2 0.008
Rhagium mordax Generalist Fresh 18.9 0.038
Atrecus affinis Generalist Decomposed 18.5 0.002
Anisotoma humeralis Generalist Fungi 15 0.013
Ipidia binotata Generalist Decomposed CR 15 <0.001
Dorcatoma punctulata Generalist Fungi EN 13.3 0.003

(continued on next page)

2566 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569

Table 2 (continued)

Species Host speciation Substratum Red List IV DW p DW IV CO p CO

Latridius consimilis Broadleaf Fungi CR 11.7 0.005
Anastrangalia dubia Conifers Decomposed 10 0.01
Anisotoma castanea Generalist Fungi 8.3 0.021
Bolitochara mulsanti Generalist Fungi 8.3 0.024
Cerylon histeroides Generalist Decomposed 8.3 0.021
Hexarthrum duplicatum Generalist Decomposed CR 8.3 0.023
Corticeus linearis Conifers Fresh 6.7 0.047
Xestobium austriacum Conifers Decomposed EN 6.7 0.046

35–160 m3 ha 1; Fig. 5c; see also Fig. S6). Nevertheless, the scores 4.2. Abundance versus species richness of beetles
of the second axis were slightly correlated with the number of
individuals but clearly correlated to the number of species In our study, the increase of dead wood and canopy openness
(r2 = 0.35, p < 0.001) (Fig. 5d). Our test for nestedness revealed a led to an increase in the abundance of saproxylic beetle species.
significant (p < 0.001) nestedness temperature of 23.6. Correcting the number of species by using individuals as covariate
The species indicator analysis revealed 16 species with a signif- resulted in weaker effects of environmental variables on species
icant indicator value for stands with a penetration rate <23%, three richness. This supports the ‘‘more individuals hypothesis”, i.e., sites
of which were also indicators for low amount of dead wood with more resources are inhabited by more individuals and there-
(<64 m3 ha 1) and two for high amount of dead wood (Table 2). fore also more species (Srivastava and Lawton, 1998). Although
The analysis revealed 31 species with a significant indicator value mostly ignored in previous studies, this hypothesis seems to be
for stands with a penetration rate >23%, 1 of which was also an an important explanation for patterns in diversity of saproxylic
indicator for low amount of dead wood and 17 for high amount beetles.
of dead wood. Five species had a significant indicator value for Many species respond to an increase in resources by a rapid in-
low amount of dead wood, and 16 species for high amount of dead crease in population, which also was true for the specialized and
wood (Table 2). Indicator species of the Red List class ‘‘critically red-listed species in our study. This explanation applies to the
endangered” were found only in categories with a high level of many studies that found an increase in rare species outside of pro-
dead wood. tected areas, when resources were created by management
(Hyvärinen et al., 2006; Wikars, 2004). We found a positive re-
sponse of saproxylic beetles to dead wood amount for both broad-
4. Discussion leaf and conifer specialists. An interesting example is Ostoma
ferruginea (Table 2), which was rare in the pre-disturbed forest of
Our study is among the first to quantify the effects of dead wood the national park (Apfelbacher and Geiß, 2006). An abundance of
accumulation and canopy opening by bark beetle infestation and >100 individuals of this species found in our samples has not been
tree senility on saproxylic beetle assemblages in a large protected reported in any standardized sampling in Germany within the last
area. Assessment of the effects of natural disturbance is often com- 100 years. This species is associated with Fomitopis pinicola, the
plicated by insufficient spatial or temporal replication and control major fungus on dead wood of spruce. Another example is the Ela-
(Harborne et al., 2008). We minimized this problem by establishing teridae, Ampedus auripes, which is rare across Europe (Wurst and
a large number of sampling sites along a wide gradient of increas- Kaupp, 1995). Populations of this species increased in our study
ing penetration rate and total amount of dead wood. Increasing area (to >300 individuals in our samples) after bark beetle infesta-
amounts of spruce dead wood and opening of the canopy both tion. Therefore, large-scale disturbance caused by bark beetles, if
by bark beetle infestation had positive effects on the abundance not followed by salvage logging, can restore population densities
of saproxylic beetles, including red-listed species, and on the com- of rare saproxylic species.
position of assemblages. This effect holds true when the analysis is
controlled for elevation and habitat continuity.
4.3. Differences in host specialization and conservation status

4.1. Amount versus diversity of dead wood The comparison of the explained variance for red-listed species
with common species indicates that the former are more closely
One key problem in most field studies of saproxylic species related to features of dead wood than the latter. Therefore, main-
based on survey data and using correlative analysis as in our study taining naturally disturbed protected areas with sufficient quanti-
is that the effects of amount and diversity of dead wood on the ties of dead wood is an important conservation strategy for these
abundance and richness of species cannot be separated because species. Species typical of habitats with low amounts of dead wood
the two factors are highly correlated (Økland et al., 1996; Siitonen in our study are mostly colonizers of thin dead trees (i.e., Salpingus
et al., 2000; Fig. S3). One earlier field study claimed to have found a ruficollis, Xylechinus pilosus), and some are considered pest species
greater influence of diversity than the amount of dead wood on the (Xyloterus domesticus, Xyleborus dispar). Furthermore, we found
number of saproxylic beetles (Similä et al., 2003). A critical evalu- only a few species that are indicators for low amounts of dead
ation of the results of this study, however, shows that both vari- wood. Our analysis of variance partitioning showed that our bark
ables were positively correlated with richness and had similar beetle predictor set had the highest positive effect on conifer spe-
correlation coefficients. Experiments following the standards of cialists and generalists, with canopy openness also being impor-
classic biodiversity studies are needed to separate the effect of tant, particularly for conifer specialists. This finding supports the
diversity and abundance (Hector et al., 1999). For practical pur- hypothesis that opening of the canopy by bark beetles across large
poses, however, knowledge that the amount of dead wood is a sur- areas leads to population increases of associated saproxylic spe-
rogate for the diversity of dead wood is useful, even if the cies. In general, beetle species typical of more-open forests, such
mechanisms behind the correlation are unclear. as dry pine or spruce, are often associated with open canopy con-
 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569 2567

ditions, for example, those created by fire or wind (Duelli et al., Such increases of dead wood in logged forests in both situations
2002; Hyvärinen et al., 2006; Moretti and Barbalat, 2004). Interest- would support more complete saproxylic communities, including
ingly, it has been shown that high canopy openness is favorable generally all species associated with high amounts of dead wood
even for the establishment and growth of regenerated Norway (Table 2).
Spruce (Nilsson et al., 2002). In contrast, the specialists of broadleaf For large protected areas, such as national parks, our study
trees were negatively affected by canopy openness in general, highlights the importance of bark beetle dynamics in the restora-
which can be explained by significant higher proportions of species tion of habitats that support many saproxylic species (Table 1,
feeding on fungi (Chi2 < 0.001; Table 2; Fig. S7). For wood-rotting Fig. S8). The positive responses of abundance and composition of
fungi, diversity is negatively affected by a rapid opening of the can- beetle assemblages to dead wood produced by bark beetle infesta-
opy (Bässler et al., 2010), which may be explained by drier wood tion suggest that saproxylic beetles benefited from natural dynam-
conditions. ics in the national park over the last 20 years. Similar diversifying
effects of natural disturbance have been shown for fire and wind-
4.4. Thresholds for canopy openness and amount of dead wood storms (Agee, 1993; Bouget and Duelli, 2004; Hyvärinen et al.,
2006; Noss et al., 2006). In contrast to logging, which also creates
Our analysis discovered two splits for ordination scores: at 23% canopy openness, natural disturbances do not reduce the basic re-
penetration rate and at 64 m3 ha 1 amount of dead wood. The high source for saproxylic organisms: dead wood. Therefore, the distur-
number of indicator species for open (31 species) and dense forests bance effect of logging is short lived rather than self perpetuating
(16 species) supports the view that two distinct communities exist (Hyvärinen et al., 2006).
with respect to openness: one community in open habitats with Forest dynamics have evoked debates about ‘‘dead forests” and
species associated with sunny conditions, e.g., Judolia sexmacula- prompted calls for salvage logging (Noss and Lindenmayer, 2006).
ta, and one community occurring in dense stands, e.g., Hedobia With salvage logging, the amount of dead wood would decrease,
imperialis (Möller, 2009; Table 2). For the amount of dead wood, leading to ‘‘incomplete” assemblages of saproxylic species. There-
however, the correlation of the second ordination axis with species fore, we recommend that nature be allowed to take its course in
numbers, the significant nestedness, and the rather low number of strictly protected areas such as national parks, with forests of the
indicator species for low levels of dead wood suggest that the com- type studied here, and that the amount of dead wood within these
munities of saproxylic species become impoverished with low protected areas be allowed to increase. Moreover, our results dem-
amounts of dead wood, but increase in species richness as the onstrate that unmanaged areas or management by ‘‘benign ne-
amount of dead wood increases. Thus, communities containing glect” (Zahner, 1992) can be quite successful in restoring natural
all species, including such specialists as Ceruchus chrysomelinus forest dynamics, habitats, and the organisms dependent on forest
(Möller, 2009), need a minimum amount of dead wood. structures. Such management is considerably less expensive than
Both thresholds with their confidence intervals are informative intensive management and has the additional advantage of creat-
for developing guidelines for conservation. However, the calcula- ing the forest wilderness aesthetics that is of great interest to many
tion of ecological thresholds has been criticized from a theoretical human visitors (Müller and Job, 2009).
point of view. A disadvantage of specifying thresholds is that indi-
vidual species responses may be continuous without a sharp Acknowledgements
breakpoints (Lindenmayer and Luck, 2005; Müller et al., 2009a; Ra-
nius and Fahrig, 2006). Furthermore, different thresholds have We are grateful to all those who helped in the field and labora-
been found for different species and taxonomic groups (Moning tory sorting the material. We thank Boris Büche for identification
and Müller, 2009; Økland et al., 1996; Fig. S7) and for the same of some of the species. The study was partly supported by funding
species in different regions (Bütler et al., 2004). Independent of from the German Federal Environment Foundation (DBU).
whether real ecological thresholds exist, the recognition of splits
is useful to set targets for nature conservation. Such values are eas-
ily understood as benchmarks (Müller and Hothorn, 2004; Müller Appendix A. Supplementary material
et al., 2009b; Villard, 2009). In our study, a ‘‘complete” community
of saproxylic beetles is associated with clearly higher amounts of Supplementary data associated with this article can be found, in
dead wood than occur in commercial forests (see Fig. 1). Further- the online version, at doi:10.1016/j.biocon.2010.06.024.
more, as pointed out by Drapeau et al. (2009), if a variable in a sys-
tem falls below the defined threshold, this may be an important References
warning signal for managers.
Agee, J.K., 1993. Fire Ecology of Pacific Northwest Forests. Island Press, Washington,
4.5. Implications for managed and unmanaged forests DC.
Andersen, T., Carstensen, J., Hernández-García, E., Duarte, C.M., 2009. Ecological
thresholds and regime shifts: approaches to identification. Trends in Ecology
Disturbances can occur on large scales unexpectedly. The bark and Evolution 24, 49–57.
beetle outbreak in our national park was unanticipated, which ex- Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of
variance. Australian Ecology 26, 32–46.
plains the lack of a before-after-control impact design in our study. Apfelbacher, F., Geiß, G., 2006. Liste der Käfer des östlichen Bayerischen Waldes. Der
Therefore, our study should be regarded as a case study. Neverthe- Bayerische Wald 20, 3–43.
less, our results highlight general implications for montane habi- Baláž, 2009. Tichá Valley – wild nature under pressure in the Tatra National Park. In:
Bavarian Forest National park, Europés wild heart, pp. 64–65.
tats in Central Europe. For conservation-oriented management of Bässler, C., Förster, B., Moning, C., Müller, J., 2008. The BIOKLIM-Project: biodiversity
commercial mixed montane forests, our data suggest two rather research between climate change and wilding in a temperate montane forest –
contrasting strategies of importance. In dense stands with less than the conceptual framework. Waldökologie, Landschaftsforschung und
Naturschutz 7, 21–33.
around 20% canopy openness, dead wood of beech and fir, in par-
Bässler, C., Müller, J., Dziock, F., Brandl, R., 2010. Effects of resource availability and
ticular, should be increased to at least 30–60 m3 ha 1, which would climate on the diversity of wood-decaying fungi. Journal of Ecology.
improve habitat quality for beetles specialized on broadleaf trees doi:10.1111/j.1365-2745.2010.01669.x.
or associated with wood-decaying fungi. In open stands with more Bjørnstad, O.N., Falck, W., 2001. Nonparametric spatial covariance functions:
estimation and testing. Environmental and Ecological Statistics 8, 53–70.
than 20–50% canopy openness, management should aim at Bouget, C., Duelli, P., 2004. The effects of windthrow on forest insect communities: a
increasing dead wood, particularly of conifers, to similar amounts. literature review. Biological Conservation 118, 281–299.
2568 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569

Buse, J., Ranius, T., Assmann, T., 2008. An endangered longhorn beetle associated Koprowski, J.L., Alanen, M.I., Lynch, A.M., 2005. Nowhere to run and nowhere to
with old oaks and its possible role as an ecosystem engineer. Conservation hide: response of endemic Mt. Graham red squirrels to catastrophic forest
Biology 22, 329–337. damage. Biological Conservation 126, 491–498.
Bütler, R., Angelstam, P., Ekelund, P., Schlaepfer, R., 2004. Dead wood threshold Körner, C., 2007. The use of ‘altitude’ in ecological research. Trends in Ecology and
values for the three-toed woodpecker presence in boreal and sub-Alpine forest. Evolution 22, 569–574.
Biological Conservation 119, 305–318. Lieutier, F., Battisti, A., Grégoire, J.-C., Evans, H.F., 2004. Bark and Wood Boring
Christensen, M., Hahn, K., Mountford, E.P., Ódor, P., Standóvar, T., Rozenbergar, D., Insects in Living Trees in Europe, A Synthesis. Kluwer Academic Publisher.
Diaci, J., Wijdeven, S., Meyer, P., Winter, S., Vrska, T., 2005. Dead wood in Lindenmayer, D.B., Foster, D.R., Franklin, J.F., Hunter, M., Noss, R.F., Schmiegelow, F.,
European beech (Fagus sylvatica) forest reserves. Forest Ecology and Perry, D.A., 2004. Salvage harvesting policies after natural disturbance. Science
Management 210, 267–282. 303, 1303.
Davies, Z.G., Tyler, C., Stewart, G.B., Pullin, A.S., 2008. Are current management Lindenmayer, D.B., Luck, G., 2005. Synthesis: thresholds in conservation and
recommendations for conserving saproxylic invertebrates effective? management. Biological Conservation 124, 351–354.
Biodiversity and Conservation 17, 209–234. Martikainen, P., Kouki, J., 2003. Sampling the rarest: threatened beetles in boreal
Dormann, C.F., McPherson, J.M., Araújo, M.B., Bivand, R., Bolliger, J., Carl, G., Davis, forest biodiversity inventories. Biodiversity and Conservation 12, 1815–1831.
R.g., Hirzel, A., Jetz, W., Kissling, W.D., Kühn, I., Ohlemüller, R., Peres-Neto, P.R., Martin, K., Norris, A., Drever, M., 2006. Effects of bark beetle outbreaks on avian
Reineking, B., Schröder, B., Schurr, F.M., Wilson, R., 2007. Methods to account for biodiversity in the British Columbia interior: implications for critical habitat
spatial autocorrelation in the analysis of species distributional data: a review. management. BC Journal of Ecosystems and Management 7, 10–24.
Ecography 30, 609–628. McFarlane, B., Craig, R., Stumpf-Allen, G., Watson, D.O., 2006. Public perceptions of
Drapeau, P., Leduc, A., Bergeron, Y., 2009. Bridging ecosystem and multiple species natural disturbance in Canada’s national parks: the case of the mountain pine
approaches. In: Villard, M.-A., Jonsson, B.G. (Eds.), Setting Conservation Targets beetle (Dendroctonus ponderosae Hopkins). Biological Conservation 130, 340–
for Managed Forest Landscapes. Cambridge University Press, Cambridge, pp. 348.
129–160. Möller, G., 2009. Struktur- und Substratbindung holzbewohnender Insekten,
Duelli, P., Obrist, M.K., Wermelinger, B., 2002. Windthrow-induced changes in Schwerpunkt Coleoptera – Käfer. Dissertation Freien Universität Berlin.
faunistic biodiversity in alpine spruce forest. Forest Snow and Landscape Moning, C., Müller, J., 2009. Critical forest age thresholds for diversity of lichens,
Research 77, 117–131. molluscs and birds in temperate beech (Fagus sylvatica L.) plant communities.
Dufrêne, M., Legendre, P., 1997. Species assemblages and indicator species: Ecological Indicators 9, 922–932.
the need for a flexible asymmetrical approach. Ecological Monographs 67, Moretti, M., Barbalat, S., 2004. The effects of wildfires on wood-eating beetle in
345–366. deciduous forests on the slope of the Swiss Alps. Forest Ecology and
Freude, H., Harde, K., Lohse, G.A., 1964–1983. Die Käfer Mitteleuropas. Goecke & Management 187, 85–103.
Evers, Krefeld. Müller, J., Hothorn, T., 2004. Maximally selected two-sample statistics as a new tool
Gibb, H., Pettersson, R.B., Hjältén, J., Hilszczanski, J., Ball, J.P., Johansson, T., Atlegrim, for the identification and assessment of habitat factors with an application to
O., Danell, K., 2006. Conservation-oriented forestry and early successional breeding-bird communities in oak forests. European Journal of Forest Research
saproxylic beetles: response of functional groups to manipulated dead wood 123, 219–228.
substrates. Biological Conservation 129, 437–450. Müller, J., Bussler, H., Goßner, M., Rettelbach, T., Duelli, P., 2008a. The European
Gotelli, N., Colwell, R.K., 2001. Quantifying biodiversity: procedures and pitfalls in spruce bark beetle Ips typographus (L.) in a national park – from pest to keystone
the measurement and comparison of species richness. Ecology Letters 4, 379– species. Biodiversity and Conservation 17, 2979–3001.
391. Müller, J., Bussler, H., Kneib, T., 2008b. Saproxylic beetle assemblages related to
Grimbacher, P.S., Stork, N.E., 2009. How do beetle assemblages respond to cyclonic silvicultural management intensity and stand structures in a beech forest in
disturbance of fragmented tropical rainforest landscape? Oecologia 161, 591– Southern Germany. European Journal of Insect Conservation 12, 107–124.
599. Müller, J., Brandl, R., 2009. Assessing biodiversity by remote sensing and ground
Grodzki, W., 1998. Bark beetles in the national parks of the Polish mountains. In: survey in montainous terrain: the potential of LiDAR to predict forest beetle
Jubilee Conference Forests and Forestry Research for the Third Millennium, 11– assemblages. Journal of Applied Ecology 46, 897–905.
14.10.1998, vol. I. Forest Research Institute Zvolen, Zvolen (Slovakia), 1998, pp. Müller, M., Job, H., 2009. Managing natural disturbance in protected areas: tourists’
197–202. attitude towards the bark beetle in a German national park. Biological
Grodzki, W., Jakus, R., Lajzova, E., Sitkova, Z., Maczka, T., Skvarenina, J., 2006. Effects Conservation 142, 375–383.
of intensive versus no management strategies during an outbreak of the bark Müller, J., Bässler, C., Strätz, C., Klöcking, B., Brandl, R., 2009a. Molluscs and climate
beetle Ips typographus (L.) (Col.: Curculionidae, Solytinae) in the Tatra Mts. in warming in a low mountain range national park. Malacologia 51, 133–153.
Polands and Slovakia. Annals of Forest Science 63, 55–61. Müller, J., Pöllath, J., Moshammer, R., Schröder, B., 2009b. Predicting the occurrence
Grove, S.J., 2002. Saproxylic insect ecology and the sustainable management of of Middle Spotted Woodpecker Dendrocopos medius on a regional scale, using
forests. Annual Review Ecological Systematics 33, 1–23. forest inventory data. Forest Ecology and Management 257, 502–509.
Harborne, A.R., Mumby, P.J., Kappel, C.V., Dahlgren, C., Micheli, F., Holmes, K., Nilsson, U., Gemmel, P., Johannsson, U., Karlsson, M., Welander, T., 2002. Natural
Sanchirico, J.N., Broad, K., Elliott, I.A., Brumbaugh, D.R., 2008. Reserve effects and regeneration of Norway spruce, Scots pine and birch under Norway spruce
natural variation in coral reef communities. Journal of Applied Ecology 45, shelterwoods of varying densities on a mesic-dry site in southern Sweden.
1010–1018. Forest Ecology and Management 161, 133–145.
Hector, A., Schmid, B., Beierkuhnlein, C., Caldeira, M.C., Diemer, M., Noss, R.F., Lindenmayer, D.B., 2006. Special section: the ecological effects of salvage
Dimitrakopoulos, P.G., Finn, J.A., Freitas, H., Giller, P.S., Good, J., Harris, R., logging after natural disturbance. Conservation Biology 20, 946–948.
Högberg, P., Huss-Danell, K., Joshi, J., Jumpponen, A., Körner, C., Leadley, P.W., Noss, R.F., Beier, P., Covington, W.W., Grumbine, R.E., Lindenmayer, D.B., Prather,
Loreau, M., Minns, A., Mulder, C.P.H., ÒDonovan, G., Otway, S.J., Pareira, J.S., J.W., Schmiegelow, F., Sisk, T.D., Vosick, D., 2006. Recommendations for
Prinz, A., Read, D.J., Scherer-Lorenzen, M., Schulze, E.-D., Siamantziouras, A.-S.D., integrating restoration ecology and conservation biology in ponderosa pine
Spehn, E., Terry, A.C., Troumbis, A.Y., Woodward, F.I., Yachi, S., Lawton, J.H., forests of the southwestern United States. Restoration Ecology 14, 4–10.
1999. Plant diversity and productivity experiments in European grasslands. Økland, B., Bakke, A., Hågvar, S., Kvamme, T., 1996. What factors influence the
Science 286, 1123–1127. diversity of saproxylic beetles? A multiscaled study from a spruce forest in
Hothorn, T., Hornik, K., Zeileis, A., 2006. Unbiased recursive partitioning: a southern Norway. Biodiversity and Conservation 5, 75–100.
conditional inference framework. Journal of Computational and Graphical Oksanen, J., Kindt, R., Legendre, P., ÒHara, B., 2006. The vegan package. <http://
Statistics 15, 651–674. cc.oulo.fi/~jarioksa>.
Hyvärinen, E., Kouki, J., Martikainen, P., 2006. Fire and green-tree retention in Paine, T.D., Raffa, K., Harrington, T.C., 1997. Interactions among scolytid bark
conservation of red-listed and rare deadwood-dependent beetles in Finnish beetles, their associated fungi, and live host conifers. Annual Review of
boreal forests. Conservation Biology 20, 1711–1719. Entomology 42, 179–206.
Jonášová, M., Pracha, K., 2004. Central-European mountain spruce (Picea abies (L.) Parker, T.J., Clancy, K.M., Mathiasen, R.L., 2006. Interactions among fire, insects and
Karst.) forests: regeneration of tree species after a bark beetle outbreak. pathogens in coniferous forests of the interior western United States and
Ecological Engineering 23, 15–27. Canada. Agricultural and Forest Entomology 8, 167–189.
Jones, C.G., Lawton, J.H., Shachak, M., 1994. Organisms as ecosystem engineers. Peres-Neto, P.R., Jackson, D.A., 2001. How well do multivariate data sets match? The
Oikos 69, 373–386. advantages of a Procrustean superimposition approach over the Mantel test.
Jonsell, M., 2007. Effects on biodiversity of forest fuel extraction, governed by Oecologia 129, 169–178.
processes working on a large scale. Biomass and Bioenergy 31, 726–732. R Development Core Team, 2008. R: A Language and Environment for Statistical
Jönsson, A.M., Appelberg, G., Harding, S., Bärring, L., 2009. Spatio-temporal impact of Computing. R Foundation for Statistical Computing, Vienna, Austria.
climate change on the activity and voltinism of the spruce bark beetle, Ips Raffa, K., Aukema, B., Bentz, B.J., Carroll, A.L., Hicke, J.A., Turner, M.G., Romme, W.H.,
typographus. Global Change Biology 15, 486–499. 2008. Cross-scale drivers of natural disturbances prone to anthropogenic
Jonsson, B.G., Kruys, N., Ranius, T., 2005. Ecology of species living on dead wood – amplification: the dynamics of bark beetle eruptions. Bioscience 58, 501–
lessons for dead wood management. Silva Fennica 39, 289–309. 517.
Koch, K., 1989–1992. Die Käfer Mitteleuropas. Ökologie, Bd. 2-3. Goecke & Evers, Ranius, T., Fahrig, L., 2006. Targets for maintenance of dead wood for biodiversity
Krefeld. conservation based on extinction thresholds. Scandinavian Journal of Forest
Köhler, F., 2000. Totholzkäfer in Naturwaldzellen des nördlichen Rheinlands. Research 21, 201–208.
Landesanstalt für Ökologie, Bodenordnung und Forsten/Landesamt für Rodríguez-Gironés, M.A., Santamaria, L., 2006. A new algorithm to calculate the
Agrarordnung Nordrhein-Westfalen, Recklinghausen. nestedness temperature of presence–absence matrices. Journal of Biogeography
Köhler, F., Klausnitzer, B., 1998. Verzeichnis der Käfer Deutschlands, Dresden. 33, 924–935.
 J. Müller et al. / Biological Conservation 143 (2010) 2559–2569 2569

Roff, D.A., 2006. Introduction to Computer-intensive Methods of Data Analysis in Stokstad, E., 2006. Salvage logging research continues to generate sparks. Science
Biology. Cambridge University Press, Cambridge. 311, 761.
Schelhaas, M.J., Nabuurs, G.J., Schuck, A., 2003. Natural disturbances in the Sverdrup-Thygeson, A., 2001. Can ‘continuity indicator species’ predict species
European forests in the 19th and 20th centuries. Global Change Biology 9, richness or red-listed species of saproxylic beetles? Biodiversity and
1620–1633. Conservation 10, 815–832.
Schmidl, J., Bußler, H., 2004. Ökologische Gilden xylobionter Käfer Deutschlands. Villard, M.-A., 2009. Designing studies to develop conservation targets: a review of
Naturschutz und Landschaftsplanung 36, 202–218. the challenges. In: Villard, M.-A., Jonsson, B.G. (Eds.), Setting Conservation
Schmidl, J., Bußler, H., Lorenz, W., 2003. Die Rote Liste gefährdeter Käfer Bayerns im Targets for Managed Forest Landscapes. Cambridge University Press,
Überblick. In: B.L.f.U. (LfU) (Ed.), Schriftenreihe Bayer. Landesamt für Cambridge, pp. 30–49.
Umweltschutz, Augsburg, p. 384. Wikars, L.-O., 2004. Habitat requirements of the pine wood-living beetle Tragosoma
Schroeder, L.M., 2007. Retention or salvage logging of standing trees killed by the depsarium (Coleoptera: Cerambycidae) at log, stand and landscape scale.
spruce bark beetle Ips typographus: consequences for dead wood dynamics and Ecological Bulletins 51, 287–294.
biodiversity. Scandinavian Journal of Forest Research 22, 524–530. Wikberg, S., Perhans, K., Kindstrand, C., Djupström, L.B., Boman, M., Mattsson, L.,
Siitonen, J., Martikainen, P., Punttila, P., Rauh, J., 2000. Coarse woody debris and Schroeder, L.M., Weslien, J., Gustafsson, L., 2009. Cost-effectiveness of
stand characteristics in mature managed and old-growth boreal mesic forests in conservation strategies implemented in boreal forests: the area selection
southern Finland. Forest Ecology and Management 128, 211–225. process. Biological Conservation 142, 614–624.
Similä, M., Kouki, J., Martikainen, P., 2003. Saproxylic beetles in managed and Williams, W.D., Liebhold, A.M., 2002. Climate change and the outbreak ranges of
seminatural Scots pine forests: quality of dead wood matters. Forest Ecology two North American bark beetles. Agricultural and Forest Entomology 2002,
and Management 174, 365–381. 87–99.
Speight, M.C.D., 1989. Saproxylic invertebrates and their conservation. Council of Wurst, C., Kaupp, A., 1995. Ampedus auripes (Reitter) – ein verkannter
Europe, Nature and Environment Series 42, 1–79. mitteleuropäischer Schnellkäfer. Mitteilungen Internationaler Verein
Srivastava, D.S., Lawton, J.H., 1998. Why more productive sites have more species: Frankfurt 20, 79–86.
an experimental test of theory using tree-hole communities. American Zahner, R., 1992. Benign neglect management: an old model for restoring health to
Naturalist 152, 510–529. the southern Appalachian national forests. Wild Earth 2, 43–46.