Agricultural and Forest Entomology (2013), 15, 135–145 DOI: 10.1111/afe.

12008

Influence of sampling effort on saproxylic beetle diversity

assessment: implications for insect monitoring studies in European
temperate forests

Guilhem Parmain∗†‡ , Marc Dufrêne§¶ , Antoine Brin∗∗ and Christophe Bouget‡

∗ NationalLaboratory of Forest Entomology, National Forest Office (ONF), 2 rue Charles Péguy, F-11500 Quillan, France, † Natural Patrimony
Department, National Museum of Natural History, 36 rue Geoffroy St Hilaire, CP 41 75 231 Paris Cedex 05, France, ‡ ‘Forest Ecosystems’ Research
Unit, National Research Institute of Science and Technology for Environment and Agriculture (IRSTEA), Domaine des Barres, F-45290
Nogent-sur-Vernisson, France, § Department of Natural and Rural Environnement Monitoring (SPW/DGARNE/DEMNA), Avenue Maréchal Juin 23,
B-5030 Gembloux, Belgium, ¶ Liege University, Gembloux Agro Bio Tech (GxABT) Forests, Nature, Landscape Department Passage des Déportés, 2,
B 5030 Gembloux, Belgium, and ∗∗ Purpan Engineering School, University of Toulouse, UMR INPT/INRA 1201 Dynafor, 75 voie du T.O.E.C., BP
57611, F-31076, Toulouse Cedex 03, France

Abstract 1 Saproxylic beetle diversity monitoring provides a tool for estimating the efficiency
of forest conservation measures. Flight interception traps are commonly employed
to monitor beetle assemblages, although little explicit knowledge of the efficiency
of this trapping method is available.
2 The present study investigated how slight changes in sampling effort can influence
species richness and species composition of assemblages in data sets from standard
window-flight traps.
3 At both trap and plot levels, an additional year or an additional trap provided a
50% increase in the number of species detected (a 75% increase for rare species)
and resulted in a different estimated composition of the assemblages. Adding 2 or
3 years of sampling gave twice as many species and resulted in assemblages that
were 50% dissimilar. Increases in the detection of species and the dissimilarity of
assemblages were similarly affected along a gradient of forest conditions, suggesting
that changes in sampling effort were not affected by forest condition.
4 At the forest level, year or trap replication provided smaller increases in species
richness (31% and 25%, respectively). Within sites, distance measures in species
composition between traps did not differ significantly when based on 1 or 2 years
of data. Using two traps per plot compared with one trap influenced comparisons
between stand types, based on species richness, in 25% of the cases.
5 Species detection was similarly increased by either year replication or trap
replication. The results of the present study highlight the significant role played
by finescale patterns of habitat structure and inter-annual variation with respect to
determining catch size and assemblages of saproxylic species.

Keywords Biodiversity, dissimilarity, flight-interception trap, replication, species

richness.

Introduction biodiversity associated with dead wood was recommended

by the European Council in 1988 (Comité des Ministres,
Saproxylic organisms, comprising a functional group that
1988a, b). They are also used as a tool for estimating the
depends on dead or dying wood (Alexander, 2008), have been
used in Europe (as indicators of forest biodiversity (Nieto efficiency of forest conservation measures in several countries
& Alexander, 2010) ever since the preservation of forest around the world (Grove, 2002b; Hammond et al ., 2004;
Lachat et al ., 2006; Ohsawa, 2007). Approximately 30% of
Correspondence: Christophe Bouget; Tel.: (00-33) 23 895 0542; fax: European species that depend on forest habitats need dead
(00-33) 23 895 0359; e-mail: christophe.bouget@irstea.fr wood to some extent (Stokland et al ., 2004). Globally, the

© 2013 The Royal Entomological Society

136 G. Parmain et al.

saproxylic biota is species rich (Grove, 2002a), although many Materials and methods
species are threatened by loss and fragmentation of habitats
The window trap dataset
with sufficient dead wood and veteran trees.
Beetles account for a large proportion of saproxylic biodi- In the present study, we used datasets compiled using saprox-
versity [e.g. approximately 25% of the saproxylic species in ylic beetles obtained from several biodiversity surveys and
Scandinavia (Stokland et al ., 2004), second to fungi]. Foresters ecological studies carried out from 1999 to 2010 by different
and conservationists are paying more attention to them than to French organizations National Research Institute of Science
saproxylic fungi or Diptera for both practical and ecological and Technology for Environment and Agriculture (IRSTEA),
reasons. Many beetle species have high conservation value; National Forest Office (ONF), University of Toulouse-Purpan
11% of species are considered as threatened at the European Engineering School (EIP), Office for Insects and their Envi-
level (Nieto & Alexander, 2010) and they are assumed to ronment (OPIE)] and DEMNA (Departement of Natural and
provide valuable information on the quality and continuity of Rural Environnement Monitoring) in Belgium.
woodland habitats (Grove, 2002b). We only compiled data originating from unbaited or ethanol-
If saproxylic beetle diversity is to be used effectively as a baited (methylated spirit, 20%) window traps, suspended
management tool in forestry, more explicit knowledge about approximately 1.5 m above the ground. The trap was the
the efficiency of trapping strategies is needed. A sound beetle basic sampling unit; at most, two traps, located approximately
sampling strategy should focus on: (i) the choice of an efficient 20–60 m apart, were grouped to represent captures from the
and standardized method, (ii) the timing of samples; and (iii) same plot (i.e. the same forest stand). Plots were grouped in
the spatial framework. Regarding the first point, window (flight sites, which were forests or a cluster of close forests dedicated
interception) traps are widely employed for catching active to the same research project. When several trapping years
flying saproxylic beetles (Økland, 1996; Wikars et al ., 2005; were available for a given plot, we included only data from
Alinvi et al ., 2006) because they are easy to replicate and consecutive years.
standardize, and are assumed to represent local saproxylic We divided the overall dataset into three subsets to
beetle communities that could only be obtained with much analyze the effects of replication on saproxylic beetle diversity
more effort using active or extraction methods such as bark assessments (species richness and assemblage composition)
peeling, dead wood beating and emergence trapping (Siitonen, after aggregating the data at three spatial scales (trap, plot and
1994; Økland, 1996; McIntosh et al ., 2001; Alinvi et al ., 2006; forest): (i) the Multi-Year-Trap set (MYT) at the trap level,
Hyvärinen et al ., 2006). to study the effects of year replication (one trap sampled over
In most studies of saproxylic beetles, species richness (SR) several years), (ii) the Multi-Trap-Plot set (MTP) at the plot
estimates are commonly compared based on data from only level, to study the effects of trap replication (two traps; i.e. one
on a single trapping year, although little is known about the additional trap located near the first, and sampled one single
errors involved. Martikainen and Kouki (2003) emphasized year) and (iii) the Multi-Trap-Multi-Year-Plot set (MTMYP),
the importance of having large sample sizes (more than 200 at the trap and plot levels, to compare the relative effects of
species) when studying threatened species. Larger samples can trap and year replications. We also analyzed the consistency
be obtained by increasing the number of traps, by sampling of the effects of trap or year replication over spatial scales,
for several years or by combining these two approaches. Using by upscaling from the trap/plot to the forest level on selected
a variety of existing data from entomological surveys based well-replicated sites.
In the MYT subset, we selected sites in which plots had been
on multiple-trap plots in France and Belgium, we assessed the
sampled at the same place for two or three consecutive years.
variation in species richness and species composition (evaluated
The MYT dataset contained 72 plots, for a total of 299 traps
in terms of Sorensen dissimilarity) of the saproxylic beetle
in 19 sites (Table 1). Six sites (n traps ≥ 10), with 239 traps in
assemblages caught with standard window traps (Brustel, 2004)
50 plots were selected for analyses at the forest level (at least
when traps or years of sampling were added. The available
10 traps cumulated over the same forest; Table 2).
data were limited in range (3 years, two traps per plot at most),
In the second data subset (MTP), a basic plot consisted of
although they covered a wide range of forest conditions. The
two replicate traps, separated by about 20 m (Bouget & Brustel,
present study aimed to determine:
2009) or 60 m (in the ORLEANS and BELGWAL datasets).
The MTP dataset included 14 sites for 294 plots and 588 traps
• How does an increase in local sampling effort (increasing (Table 1). Eight sites (n traps ≥ 10), with 257 plots and 514
the number of traps or yearly replication per plot) affect the traps, were selected for analyses at the forest level (Table 3).
assessment of species richness and assemblage composition In the BELGWAL set, we considered only the first two traps
at the trap, plot and forest level? in each plot, although the data provided by one of them during
• Does the influence of sampling effort on the quality of the second sampling year were analyzed as a new replicate.
biodiversity data vary with forest conditions? An independent analysis of trap replication from one to eight
• What are the contributions of trap replication exclusively, traps using the Belgian set only would be too idiosyncratic,
year replication exclusively and the combination of trap and and weakened by the small sample size (22 plots only). At
year replication to variation in estimates of specie richness? the multiple-plot forest level, we also studied whether trap
• Does an increased local sampling effort affect the results replication influenced the significance, magnitude and direction
of ecological comparisons between stand types at the forest of the faunistic differences between stand types. Environmental
level? variables describing the stand type and required to answer a

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
 Saproxylic beetle monitoring: sampling effects 137

Table 1 Summary of the dataset used for analyses

Number of sampling Number of Number of Number of

Dataset Site years species traps plots

MYT Ballons-Comtois∗ 3 135 12 6 (6)

MYT Bannes∗ 2 101 4 2 (2)
MYT BelgWal 2 116 176 22
MYT Chalmessin∗ 3 106 4 2 (2)
MYT Chaumes∗ 2 47 4 2 (2)
MYT Courneuve 2 85 10 2
MYT Fontbleau-Opie∗ 2 77 2 1 (1)
MYT Haute-Meurthe∗ 3 134 4 2 (2)
MYT Hauts-de-Seine∗ 2 148 16 5 (2)
MYT Jujols∗ 2 78 2 1 (1)
MYT Kertoff∗ 3 82 2 1 (1)
MYT Larchant-Marais∗ 2 102 4 2 (2)
MYT Lozere∗ 2 201 16 10 (6)
MYT Mantet∗ 2 38 4 2 (2)
MYT Rnva∗ 3 85 2 1 (1)
MYT Sausset 2 105 25 5
MYT Tourbiere-Charmes∗ 2 62 4 2 (2)
MYT Troncais-Onf∗ 3 162 6 3 (3)
MYT Vauhalaise∗ 2 47 2 1 (1)
MTP Auberive 1 146 48 24
MTP Belg-Wal 2 81 44 22
MTP Brie 1 112 28 14
MTP Caylus 1 93 4 2
MTP Chaux-Regix 1 57 6 3
MTP Coppices 1 210 58 29
MTP Fontainebleau 1 188 50 25
MTP Landes 1 210 104 52
MTP Orleans 1 125 42 21
MTP Orleans-Regix 1 95 6 3
MTP Rambouillet 1 265 120 60
MTP Tronçais-CEM 1 190 62 31
MTP Ventron 1 52 16 8
∗
Denotes the sites used to compare the number of additional species collected by a second trap in 1-year plots or by a second year of running one
trap in 2-year plots. The number of plots used for multi-year comparisons is given in parenthesis in the ‘Plot’ column.
MTP, multi-trap plots; MYT, multi-year traps.

transversal ecological question (e.g. dead wood poor versus ‘conifer’, ‘deciduous’ and ‘mixed’), altitudinal group (two
dead wood rich) were available on eight sites only in the levels: ‘highland’ and ‘lowland’, with the reference altitude
MTP set. We used these eight sites to compare managed versus distinguishing the levels being 1000 m above the sea level) and
unmanaged stands (Auberive, Fontainebleau), dead wood-poor climatic (or biogeographical) domain [four levels according
versus dead wood-rich stands (Rambouillet, BelgWal Year1, to the ETCB (European Topic Centre on Biological Diver-
BelgWal Year2, Landes) and overmature versus mature stands sity) (2006): ‘alpine’, ‘atlantic’, ‘continental’, ‘continental-
(Tronçais-CEM, Coppices). Mediterranean’]. Data from alpine or Mediterranean regions
In the third data subset (MTMYP), we selected two-trap 2- were insufficient to provide rigorous tests. The use of bait in
year plots from the MYT dataset. We excluded the third year the trap (methylated spirit, 20%) was the only methodological
for some sites because a third trap per plot was not available factor considered (two levels: ‘ethanol-baited’ and ‘unbaited’;
(exept for BELGWAL). This set (i.e. MTMYP) included 16 Table 3).
sites, 36 plots and 72 traps (Table 1). Samples available for
this analysis were well distributed over the ecological forest
gradients.
Beetle data
The beetle records from different sets first had to be harmo-
nized, both with respect to nomenclature and saproxylic status.
Environmental data
We choosed to follow the French database FRISBEE developed
Three environmental factors and one methodological fac- by Bouget et al . (2008). Only those records from families for
tor were used to describe trap features. The environmental which beetles were identified to the species level were used for
variables qualifying trap location were: forest type (three levels: the present analysis. These included Alleculidae; Anobiidae;

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
138 G. Parmain et al.

Table 2 Effect of sampling effort per plot (number traps/years) on species richness and assemblage composition at particular sites

Comparison between 1 and 2 (years per trap or traps per plot)

Number of Mean Mantel statistics Mean Mantel statistics
Sites plots (traps) SR-Benefit r (1 versus 1 + 2) r (2 versus 1 + 2)

MYT Ballons Comtois 6 (12) 24.10% 0.78*** 0.47***

BelgWal 22 (176) 20.21% 0.59*** 0.57***
Courneuve 2 (10) 40.50% 0.60*** 0.28*
Hauts-de-Seine 5 (16) 35.16% 0.52*** 0.60***
Lozère 10 (16) 29.06% 0.86*** 0.89***
Sausset 5 (25) 40.94% 0.80*** 0.78***
Mean 31.66% 0.69 0.60
MTP Auberive 24 (48) 30.94% 0.64*** 0.55***
Belg-Wal 22 (44) 33.88% 0.58*** 0.62***
Brie 14 (28) 24.44% 0.79*** 0.57**
Coppices 29 (58) 23.89% 0.71*** 0.73***
Fontainebleau 25 (50) 24.50% 0.71*** 0.68***
Rambouillet 60 (120) 27.27% 0.67*** 0.65***
Landes 52 (104) 15.72% 0.81*** 0.84***
Troncais-CEM 31 (62) 19.87% 0.34*** 0.60***
Mean 25.06% 0.66 0.66

*P < 0.05, **P < 0.01, ***P < 0.001.

The mean species richness benefit index (SR-Benefit) (see text) between traps (multi-trap plot; MTP) or years (multi-year trap; MYT) was used to measure
the increase in species number caught by one additional trap or year, respectively. Mantel tests (999 permutations) assessed whether within-site
between-trap Sörensen distance matrices based, respectively, on single traps/years were correlated with data that included another trap or year,
respectively. The number of traps in parenthesis is the total number of traps per site.

Table 3 Number of traps in multi-year trap (MYT) and multi-trap plot Latridiidae, Mordellidae, Ptiliidae, Scirtidae, Scraptiidae,
(MTP) datasets for each ecological studied factor Scydmaenidae, Sphaeritidae, Staphylinidae and Throscidae. A
total of 643 saproxylic beetle species [507 common species
Environmental/methodological factors Number of traps
(79%) and 136 rare species (21%)] were present in the studied
Factor Category MYT MTP datas. They belonged to 42 families (or sub-families).
We characterized each species with conservation value (at
Forest type Conifer 11 89
Deciduous 250 459
the country level) either as ‘common’ (IP = 1 or 2) or ‘rare’
Mixed 38 40 species (IP = 3 or 4), in accordance with principles discussed
Altitude Highland 44 16 by Brustel (2001) and the database FRISBEE (Bouget et al .,
Lowland 255 572 2008). In this database, each species has a patrimoniality index
Climatic domain Alpinea 6 0 (i.e. conservation value; IP), in other words its degree of
Atlantic 57 360 geographical rarity in France, with four levels: (i) common
Continental 220 228 and widely distributed species; (ii) not abundant but widely
Continental/Mediterraneana 16 0 distributed species, or only locally abundant species; (iii) not
Bait Alcohol-baited 82 32
abundant and only locally distributed species; and (iv) very
No 217 556
rare species (known in less than five localities or in a single
a
This category was not considered as a result of its low number of ‘county’ in France). The ‘all species’ group contains both the
replicates. ‘common’ and the ‘rare’ species.

Anthribidae; Biphyllidae; Bostrichidae; Bothrideridae;

Buprestidae; Cerambycidae; Cerophytidae; Cerylonidae; Statistical analysis
Ciidae; Cleridae; Cucujidae; Curculionidae (Scolytinae only); Because the abundance of beetles was not always available,
Elateridae; Endomychidae; Erotylidae; Eucnemidae; His- we only considered beetle occurrence for our analyses.We
teridae; Laemophloeidae; Leiodidae; Lucanidae; Lycidae; calculated two major indices based strictly on presence-absence
Lymexylidae; Melandryidae; Monotomidae; Mycetophagidae; data: (i) the mean benefit of SR (SR-Benefit) and (ii) the mean
Nitidulidae; Nosodendridae; Oedemeridae; Phloeostichidae; assemblage dissimilarity between traps or years. We defined
Prostomidae; Pyrochroidae; Salpingidae; Scarabaeidae; Sil- the SR-Benefit as the percentage increase in species added by
vanidae; Sphindidae; Tenebrionidae; Tetratomidae; Trogidae; a second trap or year, as follows:
Trogossitidae; Zopheridae. Several beetle families not studied General formula:
in a majority of the sets were excluded from our analyses:
Aderidae, Alexiidae, Cantharidae, Clambidae, Corylophi- (SR(1+2) -SRi )
SR-Benefit = × 100 (1)
dae, Cryptophagidae, Dasytidae, Dermestidae, Eucinetidae, SRi

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
 Saproxylic beetle monitoring: sampling effects 139

with i = year 1 or 2 (MYT subset), or trap 1 or 2 (MTP subset). in richness observed in the MTMYP dataset into ‘exclusive’
Mean percentages were calculated over plots, forests or years, (species contacted by one mode of replication only) and
depending on the comparison. ‘interactive’ (species contacted by both modes of replication)
Mean assemblage dissimilarity was used to interpret the effects of trap or year replication by a method adapted from
significance of the additional captures for understanding the Alatalo and Alatalo (1977). We used relativized calculations
assemblages.The assemblage dissimilarity between plots or (i.e. the relative number of additional species compared with
years was calculated as the Jaccard–Dice–Sörensen index data from single traps and years).
(Oksanen et al ., 2011). At the forest level, we compared species richness (only the
To test the influence of environmental characteristics and ‘all species’ group) between two stand types A and B, estimated
the use of bait in the traps on the species richness benefit, with one or two traps per plot. Species richness was assessed
we fit generalized linear mixed models (GLMER) (Bolker using the order-1 Chao richness estimator corrected for bias
et al ., 2009), assuming a binomial distribution, with site and (Colwell, 1997) with 100 sample randomizations to calculate
plot as random factors and including an observation-specific SD. The species richness difference was:
random intercept to account for possible overdispersion (Elston
et al ., 2001).We tested the significance of effects by comparing (RSB -RSA )
Species richness difference = × 100 (2)
factorial models and a null model with a likelihood ratio test RSA
(LRT). A Tukey’s multiple-comparison test was performed to
with RSA and RSB being order-1 Chao estimators of the species
identify where the differences occurred. We set the significant
richness in the A and B stand types, respectively. We observed
value of the LRT at 0.01% to limit type II errors.
whether the A-B dissimilarity values (± SD) computed for one-
In the MYT within-site between-trap level analyses, we
trap or two-trap plots overlapped.
tested the effects of different combinations of 1, 2 and 3-year All statistical analyses were conducted using estimates
sampling designs on SR-Benefit and dissimilarity: a second (Colwell, 1997) and r (R Development Core Team, 2010) with
sampling year after the first one (SR-Benefit A), a third the lme4 (Bates et al ., 2011), mgcv (Wood, 2008), mvtnorm
sampling year after two first consecutive sampling years (SR- (Genz et al ., 2011), multcomp (Hothorn et al ., 2008) and vegan
Benefit B) and two additional sampling years after a single first (Oksanen et al ., 2011) packages.
one (SR-Benefit C).
In the same way, the dissimilarity value was calculated
among first-year and second-year or third-year samples.
Dissimilarity analyses were always conducted with assemblages Results
composed of all species; a potential more restricted analysis of Effects of year replication on beetle diversity assessments
rare species assemblages was not useful as a result of the small at trap and forest levels
proportion of rare species in our data. The same testing strategy
The MYT trap level dataset included 517 species [417 common
as that used for SR-Benefit was applied for dissimilarity. At the
species (81%) and 100 rare species (19%)]. At the trap level,
forest level, we only considered the first 2 years of sampling to
adding a second year of sampling gave a mean SR-Benefit value
calculate the difference in species richness between one and
of +53% (Fig. 1) and the mean dissimilarity of assemblages
two sampling years (see general formula): with SR = Specific
Richness and y(i) = year of sampling 1 or 2.
We used Mantel tests (method = Spearman, 999 permuta-
tions) to test whether within-site between-trap distance matrices
based, respectively, on 1- or 2-year data were correlated.
We compared the effect of additional traps within sites in
terms of SR-Benefit and dissimilarity values during single
years between one- and two-trap plots [SR = Specific Richness;
t(n) = trap number] [possible combinations for each plot:
SRt 1 ∼ SRt(1 + 2); SRt 2 ∼ SRt(1 + 2)].
The effects of methodological and environmental factors
were tested with a GLMER, assuming a Gaussian distribution
with a log + 1 transformation of the raw data. The model
was fit with site as a random factor. The dissimilarity value
was computed between one-trap and two-trap plots. The same Figure 1 Mean mean benefit of species richness (SR-Benefit) values
testing strategy as that used for SR-Benefit was applied for dis- between traps (multi-trap plot; MTP) or years (multi-year trap; MYT) for
‘all species’, ‘common species’ and ‘rare species’. SR-Benefit is the
similarity [see the general formula above, with SR = Specific
increase in species number caught by one supplementary trap or year,
Richness and t(i) = trap number 1 or 2]. We used Mantel statis-
which are compared as a percentage with one single trap or one single
tics on one-trap or two-trap data to test whether the distance year, respectively. In the MYT between year analyses, we tested the
measures in species composition between traps (i.e. assem- effects of different combinations of 1-, 2- and 3-year sampling designs
blage dissimilarity) was influenced by the number of traps on SR-Benefit: a second sampling year after the first one (SR-Benefit
per plot. A), a third sampling year after two first consecutive sampling years
To evaluate the contribution of each replication mode (trap (SR-Benefit B), two supplementary sampling years after a single first one
or year) to total species richness, we partitioned the increase (SR-Benefit C). Error bars represent the 95% confidence intervals.

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
140 G. Parmain et al.

using one trap/plot (Fig. 1). This value was similar for analysis
of common species only (+46%) but was much higher for
data about rare species (+78%). Mean assemblage dissimilarity
between designs with paired one-trap and two-trap plots was
33% (Fig. 2). At the forest level, two-trap plots provided 25%
more species, on average, than one-trap plots. Nonetheless,
the mean Mantel correlation value between the within-site
distance matrices of one- and two-trap plots was 66% and was
consistently significant (Table 2).
Similar to the analysis of MYTs, we did not find any
Figure 2 Mean Sörensen dissimilarity between traps (multi-trap
relationship between SR-Benefit or assemblage dissimilarity
plot; MTP) or years (multi-year trap; MYT) for ‘all species’. The that could be related to forest type, climatic domain or
mean dissimilarity is the difference in species composition between altitudinal group, or related to trap bait.
assemblages caught by one single trap or one single year and
assemblages caught by two traps or additional years. In the
MYT between year analyses, we assessed the dissimilarity between
assemblages caught with different combinations of 1-, 2- and 3-year Comparative effects of trap and year replication on beetle
sampling designs: a second sampling year after the first one (dissimA), a diversity assessments at trap and forest levels
third sampling year after two first consecutive sampling years (dissimB),
On average, sampling designs with two traps per plot or two
two supplementary sampling years after a single first one (dissimC). In
sampling years returned more species and the effects of an
the MTP between trap analyses, ‘dissim’ is defined as the dissimilarity
between assemblages caught by one or two traps. Error bars represent
additional trap or an additional year were similar (Fig. 3).
the 95% confidence intervals. The relative increase in richness as a result of trap replication
exclusively was approximately 48%, whereas the increase as
a result of year replication exclusively was 53%. However,
between paired 1-year and 2-year designs was 36% (Fig. 2).
the increase reflected in both approaches to replication was
At the forest level, using a second year of sampling increased
much lower for common species (mean of 17%). These effects
species richness by +31%. The mean Mantel correlation
were caused mostly by additions of rare species in the catches
between the within-site distance matrices of 1- and 2-year
(Fig. 3); the increase as a result of the addition of a single
data was nonetheless 65%, and significant in all cases studied.
trap was 43.8% and the increase as a result of a second
Within-site between-trap distance matrices based, respectively,
year of data was similar at approximately 44.7%. By contrast
on 1-year or 2-year data did not differ (Table 2).
to the results reported above for common species, increases
Including year-to-year variation led to notable increases in
in rare species were more commonly seen in both kinds of
understanding of biodiversity. Overall, the number of species
replication (40.8%) (Fig. 3). The increase was explainable by
detected after 3 years of sampling was almost twice as large as
trap replication exclusively, by year replication exclusively
the number of species after trapping only for 1 year (+88%)
and by both replication modes redundantly. However, the
(Fig. 1). The 3-year assemblages were almost half as dissimilar
relative increase in the number of rare species was highly
as the 1-year assemblages (D = 47%; Fig. 2). At the trap level,
adding a third year after 2 years of sampling provided only a variable.
mean SR-Benefit value of +27% (Fig. 1). Assemblages based
on 3 years of data were only 20% dissimilar to those from
2 years of collecting. Effect of trap replication on ecological comparisons of
The SR-Benefit values for common species were similar to stand types
those calculated for the whole assemblage. However, these were
In all datasets, assemblages from the stand types com-
much higher for the group of rare species only: +63% from a
pared (i.e. managed/unmanaged, dead wood poor/rich,
1-year design to a 2-year design and even +112% from a 1-year
mature/overmature) were less dissimilar with two traps (mean
design to a 3-year design (Fig. 1). Benefit values were much
of 68%) compared with one trap per plot (73%); however,
more variable for rare species only (the confidence interval was
these dissimilarity values (± SD) always overlapped. On
wider; Fig. 1).
average, over the eight cases studied, the difference in species
At the trap level, we did not observe any effect of
richness between the two stand types was similar using one-
forest type, climatic domain, altitudinal group or baiting
trap or two-trap plots (approximately 20% as absolute values
status on of SR-Benefit or assemblage dissimilarity in any
analysis. in both cases). The magnitude of this difference between
two- and one-trap plots depended on the case. No significant
changes in the direction (A > B or B > A) of the difference
between stand types was observed using one-trap or two-trap
Effects of trap replication on beetle diversity assessments at
plots. However, in terms of estimated species richness, two
trap and forest levels
comparisons gave significant A–B differences with two-trap
The MTP plot level dataset included 511 species [417 common plots only (Table 4). The only significant A–B difference
species (82%) and 94 rare species (18%)]. Using two traps/plot found with one-trap plots remained significant using data from
provided a mean SR-Benefit value of +48% compared with two-trap plots.

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
 Saproxylic beetle monitoring: sampling effects 141

Figure 3 Exclusive and interactive effects of trap or year replication on the total species richness in two-trap and 2-year plot designs, for all, common
and rare species (multi-trap-multi-year-plot; MTMYP dataset). The increase in species richness was partitioned into three components: as a result of
trap replication exclusively (from one to two traps) (I), as a result of year replication exclusively (from 1 to 2 years) (II) and as a result of both replication
modes redundantly (III). Error bars are the SD.

Table 4 Effects of sampling effort per plot (number of traps) on faunistic comparisons of different stand types

Comparison between A and B stand types

One trap per plot Two traps per plot
Number of Species richness Assemblage (A–B) Species richness Assemblage (A–B)
Stand type Forests plots (traps) difference dissimilarity(%) difference dissimilarity(%)

A = Managed Auberive 24 (42) 10%NS 71.30 13%NS 65.40

B = Reserve Fontainebleau 25 (50) −25%NS 78.80 −27%∗ 73.12
A = dead Wood-poor Rambouillet 60 (120) 23%NS 66.97 6%NS 61.41
Landes 52 (104) −1%NS 77.96 −30%∗ 73.27
B = dead Wood-rich BelgWal Year1 22 (44) 34%NS 75.28 10%NS 70.33
BelgWal Year2 22 (44) −12%NS 82.15 23%NS 76.52
A = mature Coppices 29(58) 36%∗ 64.73 41%∗ 62.79
B = overmature Troncais-CEM 31(62) −20%NS 68.49 6%NS 62.49
Mean |20.1%| 73 |19.5%| 68
∗
If order-1 Chao estimators of species richness in forest categories A and B did not overlap, not significant (NS) if they overlapped; Sörensen dissimilarity
values (+/− SD) did overlap in all comparisons of forest categories A and B with 1 or 2 traps per plot. The difference in species richness (order-1
Chao estimators) was calculated as the percentage of supplementary species in the B compared with the A stand type. Mean values of species
richness difference were based on absolute values (|mean value|). Plots were considered to be dead wood-rich, using the thresholds: 30 m3 /ha in the
Rambouillet oak forest and in the Belgian oak-beech forests, and 20 m3 /ha in the French Landes pine forest. Mature high forests were 150–175 years
old; overmature high forests were more than 200 years old (Troncais); mature coppices were 25–30 years old, whereas overmature coppices were
70–80 years old (Coppices). The number of traps between brackets is the total number of traps per site.

Discussion The impact was more striking for rare species with a 75%
increase in the number of species. On average, assemblages
Replication and species richness estimates
based on fewer traps and years were 35% dissimilar to those
Adding both traps and years to studies of saproxylic beetle with more extensive samples. At the forest level, either year or
assemblages dramatically increased the number of beetle trap replication provided a lesser increase in species richness
species collected at either the plot or forest level. On average, at (31% and 25%, respectively). Species detection was similarly
the plot level, adding both an additional year and an additional increased by either year replication or trap replication (one to
trap provided a 50% increase in the number of detected species. two traps).

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
142 G. Parmain et al.

Despite large differences in species detection, ecological has also been demonstrated (Sverdrup-Thygeson & Birkemoe,
studies that ask functional questions about the general effects 2008). Our data showed significant assemblage dissimilarity
of various treatments or management strategies may not be between catches of two traps located only approximately 20 m
deeply affected, although the magnitude of differences may be apart in the same stand. The results obtained in the present
considerably underestimated. However, the problem remains study therefore strengthen the hypothesis that finescale patterns
for those who aim to monitor biodiversity as a conservation of habitat structure could play an important role in trap catches.
measure. They are limited by the efficiency of sampling Although traps may be located close together, data will differ
schemes and the extent to which possible approaches provide depending on whether or not they are in flight corridors, near
sufficient data (especially on rare species). rich microhabitats, or in open or closed spots.
At a larger spatial scale (i.e. a forest), the SR-Benefit
associated with trap replication appears to decrease; its value at
Temporal consistency and yearly variations the forest level is halved compared with the value at the local
plot level (site = 25%; plot = 50%). Nonetheless, the results
Our results support the findings of White et al . (2006) of research projects at the forest level may be affected to
concerning the importance of considering the yearly variation some extent by trap replication. For example, trap replication
in species assemblages when estimating species richness and strengthened some previously insignificant trends in the present
assemblage characteristics. study. Simply doubling the number traps per plot changed the
Increases in species number or contrasts in assemblage results for ecological comparisons of species richness in 25%
composition were driven more by rare than by common species. of the cases studied. However, the comparison of assemblages
Similarly, Martikainen and Kaila (2004) showed that rare in the selected stand types did not differ significantly among
species richness varies greatly between years and does not one-trap or two-trap plots.
vary synchronously among forests. During a 10-year study,
they observed a low between-year dissimilarity for common
species (approximately 20–30%) but a higher between-year Sampling rare species
dissimilarity for rare species. They showed that most of the
common species observed over a 10-year sampling period Sampling rare species is especially challenging because they
had already been sampled in the first 3 years. In the present represent only a small part of the total number of species caught
study, successively adding a second or a third sampling year (McArdle, 1990), approximately 20% in our data. Unlike some
(compared with 1 year only) gave 50% dissimilar assemblages, studies (Niemelä et al ., 1990; Novotny & Basset, 2000; Grove,
twice the number of species and 112% more rare species at the 2002b) that define rare species as those poorly represented
trap level. Even at the forest level, a 1-year replication provided in their samples, we followed Martikainen and Kaila (2004)
a 31% increase in species richness. and a priori defined as rare those species listed as such in
Inter-annual variation of saproxylic beetle assemblages is reliable databases (i.e. the French FRISBEE database in our
driven by several processes: beetle density and flight activ- case; Bouget et al ., 2008).
ity (Nageleisen & Bouget, 2009), meteorological variations For the results obtained in the present study, at the plot level,
(Williams, 1940; Rink & Sinsch, 2007), multi-year develop- all SR-Benefits associated with year replication were signifi-
mental cycles, variation in mean reproductive activity and the cantly higher for rare than for common species. Moreover, the
proportion of reproducing individuals driven by food avail- annual SR-Benefit remained high (+73% of rare species from
ability and/or weather factors, and, finally, yearly variations 1- to 2-year replicates, +38% of rare species from 2- to 3-year
in predator effects on prey populations (Turchin et al ., 1999). replicates) throughout a 3-year sampling period. Furthermore,
These sources of variation are well appreciated for ground bee- Martikainen and Kaila (2004) demonstrated that the annual
tles (Klenner, 1989; Niemelä et al ., 1992; Heyborne et al ., number of detected rare species is constant throughout a
2003; Irmler, 2003; Scott & Anderson, 2003) and also for 10-year sampling period. A multi-year study would therefore
saproxylic beetles (Ranius, 2001; Martikainen & Kouki, 2003). be particularly valuable to detect a large amount of rare
These variations lead to a ‘time-dependent species accumu- species. Martikainen and Kouki (2003) and Martikainen and
lation’. A multiple-year sampling strategy reduces the influ- Kaila (2004) observed that catches of rare species in small
ence of between-year variations on data quality (Martikainen samples are random and that between-site comparisons based
& Kaila, 2004). on such limited data do not provide very useful results.
In the present study, year or trap replication provided an
equivalent +75% increase in the number of detected rare
species at the plot level. Hedgren and Weslien (2008) showed
Between-trap within-plot variations
that selective trap placement (near well-known rich microhab-
Small-scale variation in microclimatic conditions, habitat and itats) was a more efficient way of catching rare species than
microhabitat distribution patterns among plots may lead to random trap placement. In the data obtained in the present
between-trap variation in beetle catches. The influence of study, even if adding a second trap per plot is assumed to sam-
small-scale heterogeneity in beetle habitats on trap catches has ple a wider range of microhabitats at the plot scale, the relative
already been shown in pitfall trap data for carabid beetles and net increase in rare species detection with an additional
(Niemelä et al ., 1986; Desender & Pollet, 1988; Niemelä & trap was not higher than that with an additional sampling year.
Spence, 1994; Brose, 2002). The importance of the immediate Data from a second sampling year accounts for between-year
surroundings on catches of freely hanging flight intercept traps variation in rare beetle species density and activity.

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
 Saproxylic beetle monitoring: sampling effects 143

Practical recommendations for saproxylic beetle diversity data (longer time series, denser sampling plots) will be useful
surveys for suggesting practical guidelines with respect to the sampling
strategies used in monitoring schemes. A longer time frame
Given the high between-trap variation in species number
for studies explicitly designed to support this type of analysis
and composition within plots, we recommend that ecological
would facilitate the better analysis of time-dependent species
comparisons in species richness should be made at the plot
accumulation rates. In addition, long-term studies would allow
level and not at the trap level.
us to better understand inter-annual fluctuations in assemblage
Our efforts to partition the effects on increase in species
composition (Kozlov et al ., 2010) and the impacts of global
richness suggest that an extra trap had a similar effect to
patterns of increasing or decreasing populations (Conrad et al .,
an extra year. However, yearly replication will accommodate
2004; Salama et al ., 2007), especially under the influence of
mainly inter-annual variation in species occurrences, and trap
climate change.
replication will probably accommodate microhabitat variation
Unfortunately, long-term, large-scale intensive insect sam-
(Hedgren & Weslien, 2008). In our analysis, the additional
pling designs are scarce despite their obvious relevance to
species differed between spatial and temporal replication
effective biological conservation and efficient biodiversity mon-
modes. For common species, the gross effect of sampling
itoring. The collection of such data is currently limited by
replication (both trap and year) was significantly lower than
financial constraints, a lack of qualified personnel or by insti-
the trap or the year replication effect. In other words, the
tutional changes in research orientations (Jackson & Füreder,
specific effect on catches of either yearly variation or small-
2006). We hope that the findings of the present study, aiming
scale habitat heterogeneity was stronger than a raw replication
to better understand the sampling methods for saproxylic bee-
effect (whatever the mode). For rare species, however, the
tles, provide or improve existing tools and aid in the design of
interactive effect of trap and year replication on the increase in
cost-effective biodiversity monitoring schemes.
species richness was as important as the exclusive effects of trap
or year replication. As previously suggested by Martikainen and
Kaila (2004), the raw effect of replication therefore appears to
be more important for rare species. Acknowledgements
A complete comparison of relative benefits of these two We thank Frederic Gosselin for his help with the data analysis
replication approaches should take costs into account. On and for helpful comments on the first drafts of the paper; Pavel
average, field work accounts for only 20% of the working Sebek for his contribution to the data compiling; and Bruno
time for data collection in a monitoring or research programme, Meriguet, Olivier Rose, Thomas Barnouin, Fabien Soldati,
whereas the remaining 80% is sorting and identification work Hervé Brustel, Thierry Noblecourt, Lionel Valladares, Benoit
in the laboratory (Bouget, 2009). However, this feature depends Nusillard and Carl Moliard for help with data collection. We
strongly on the spatial extent of the programme because field are indebted to John Spence and two anonymous rewiers for
costs indeed grow higher as the spatial scale of programmes their useful comments on an earlier version of the manuscript.
increases. Thus, trap replication is recommended in large- We also thank Vicki Moore for checking and improving our
scale programmes, mainly for economic reasons. The required written English.
sampling strategy should obviously take into account space
and time constraints dependent on the objectives of the
sampling programme (analysis of environment–biodiversity References
relationships, long-term monitoring, intensive inventory, etc.).
Alatalo, R.V. & Alatalo, R.H. (1977) Components of diversity:
Power analyses are needed to better define the minimum
multivariate analysis with interaction. Ecology, 58, 900–906.
number of traps per plot required to be able to detect at least 5%
Alexander, K. (2008) Tree biology and saproxylic coleoptera: issues of
differences between two groups of plots. Similarly, at the forest definitions and conservation language. Revue d’Ecologie (Terre Vie),
level, it would be useful to better understand the minimum 63, 1–7.
number of plots required to compare two groups of sites. To Alinvi, O., Ball, J.P., Danell, K., Hjältén, J. & Pettersson, R.B.
detect most common species in a site, Martikainen and Kaila (2006) Sampling saproxylic beetle assemblages in dead wood logs:
(2004) suggested using at least 20 traps during one single comparing window and eclector traps to traditional bark sieving and
year. Plots containing a larger number of traps are required to a refinement. Journal of Insect Conservation, 11, 99–112.
properly study the sample-dependent species accumulation rate. Bates, D., Maechler, M. & Bolker, B. (2011) lme4: Linear Mixed-
effects Models Using S4 Classes. R package version 0.999375-39
[WWW document]. URL http://CRAN.R-project.org/package=lme4
[accessed on 1 February 2012].
Conclusions and perspectives Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R.,
Our analyses were based on existing data obtained from Stevens, M.H.H. & White, J.-S.S. (2009) Generalized linear mixed
models: a practical guide for ecology and evolution. Trends in
France and Belgium after compiling them in a way that
Ecology & Evolution, 24, 127–135.
permitted comparison. Significant benefits of replication were
Bouget, C. (2009) Levels of taxonomic resolution in forest entomology
demonstrated despite a narrow range of year or trap replication. studies: needs, contingencies and perspectives. Mémoires de la
Slight variation in sampling effort (adding trap or year) deeply Société Entomologique de France, 8, 75–80 (in French).
affected the quality of data. Bouget, C. & Brustel, H. (2009) Chapter 4: target insect groupes in
Further studies about the relationships between sampling temperate forests, II – Saproxylic Coleoptera. Forest Insect Studies:
effort and catch characteristics based on a broader range of raw Methods and Techniques. Key Considerations for Standardisation. An

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
144 G. Parmain et al.

Overview of the Reflections of the ‘Entomological Forest Inventories’ Hothorn, T., Bretz, F. & Westfall, P. (2008) Simultaneous inference in
Working Group (Inv.Ent.For). Vol. 19: Les Dossiers Forestiers (ed. general parametric models. Biometrical Journal , 50, 346–363.
by L. M. Nageleisen and C. Bouget), pp. 100–111. Office National Hyvärinen, E., Kouki, J. & Martikainen, P. (2006) A comparison of
des Forêts, France. three trapping methods used to survey forest-dwelling Coleoptera.
Bouget, C., Brustel, H. & Zagatti, P. (2008) The French information Eurasian Journal of Entomology, 103, 397–407.
system on saproxylic beetle ecology (FRISBEE): an ecological Irmler, U. (2003) The spatial and temporal pattern of carabid beetles
and taxonomical database to help with the assessment of forest on arable fields in northern Germany (Schleswig-Holstein) and
conservation status. Revue d’Ecologie (suite de La Terre et la their value as ecological indicators. Agriculture, Ecosystems and
Vie), 10 (Suppl), 33–36 [WWW document]. URL http://frisbee. Environment, 98, 141–151.
nogent.cemagref.fr/index.php/en [accessed on 1 February 2012]. Jackson, J.K. & Füreder, L. (2006) Long-term studies of freshwater
Brose, U. (2002) Estimating species richness of pitfall catches by non- macroinvertebrates: a review of the frequency, duration and ecolog-
parametric estimators. Pedobiologia, 46, 101–107. ical significance. Freshwater Biology, 51, 591–603.
Brustel, H. (2001) Coléoptères saproxyliques et valeur biologique des Klenner, M. (1989) Überlebenstrategien einer stenotopen Waldart :
forêts françaises. PhD Dissertation, University of Toulouse, Institut Untersuchungen zur Dynalik einer westfälischen Carabus auronitens
National Polytechnique. Population (Coleoptera, Carabidae). Verhandlungen der Gesellschaft
Brustel, H. (2004) ‘PolytrapTM ’ a window flight trap for saproxylic für Ökologie, 18, 781–791.
beetles. Third Symposium and Workshop on the Conservation of Kozlov, M.V., Hunter, M.D., Koponen, S., Kouki, J., Niemelä, J. &
Saproxylic Beetles, Lativa. Price, P.W. (2010) Diverse population trajectories among coexisting
Colwell, R.K. (1997) EstimateS: Statistical Estimation of Species species of subarctic forest moths. Population Ecology, 52, 295–305.
Richness and Shared Species from Samples. Version 8. User’s Guide Lachat, T., Nagel, P., Cakpo, Y., Attignon, S., Goergen, G., Sinsin, B.
and Application. University of Connecticut, Storrs, Connecticut & Peveling, R. (2006) Dead wood and saproxylic beetle assemblages
[WWW document]. URL http://viceroy.eeb.uconn.edu/EstimateS in a semi-deciduous forest in Southern Benin. Forest Ecology and
[accessed on 1 February 2012]. Management, 225, 27–38.
Comité des Ministres (1988a) Recommandation N◦ R (88) 10 du Comité Martikainen, P. & Kaila, L. (2004) Sampling saproxylic beetles: lessons
des Ministres aux états membres pour la conservation des organismes from a 10-year monitoring study. Biological Conservation, 120,
saproxyliques et leurs biotopes (adoptée par le Comité des Ministres 175–185.
le 13 juin 1988, lors de la 418e réunion des Délégués des Ministres).
Martikainen, P. & Kouki, J. (2003) Sampling the rarest: threatened
Comité des Ministres (1988b) Recommandation N◦ R (88) 11 du Comité
beetles in boreal forest biodiversity inventories. Biodiversity and
des Ministres aux états membres pour la conservation des organismes
Conservation, 12, 1815–1831.
saproxyliques et leurs biotopes (adoptée par le Comité des Ministres
McArdle, B.A. (1990) When are rare species not there? Oikos, 57,
le 13 juin 1988 lors de la 418e réunion des Délégués des Ministres).
276–277.
Conrad, K.F., Woiwod, I.P., Parsons, M., Fox, R. & Warren, M.S.
McIntosh, R.L., Katinic, P.J., Allison, J.D., Borde, J.H. & Downey,
(2004) Long-term population trends in widespread British moths.
D.L. (2001) Comparative efficacy of five types of trap for woodbor-
Journal of Insect Conservation, 8, 119–136.
ers in the Cerambycidae, Buprestidae and Siricidae. Agricultural and
Desender, K. & Pollet, M. (1988) Sampling pasture carabids with
Forest Entomology, 3, 113–120.
pitfalls: evaluation of species richness and precision. Mededelingen
Nageleisen, L.M. & Bouget, C. (2009) Forest Insect Studies: Meth-
van de Faculteit Landbouwwetenschappen van de Rijksuniversiteit
ods and Techniques. Key Considerations for Standardisation. An
Gent, 53, 1109–1117.
Overview of the Reflexions of the ‘Entomological Forest Inventories’
Elston, D.A., Moss, R., Boulinier, T., Arrowsmith, C. & Lambin, X.
Working Group (Inv.Ent.For). Les dossiers forestiers n◦ 19 . Office
(2001) Analysis of aggregation, a worked example: numbers of ticks
on red grouse chick. Parasitology, 122, 563–569. National des Forêts, France.
ETCB (European Topic Centre on Biological Diversity) (2006) The Niemelä, J.K. & Spence, J.R. (1994) Distribution of forest dwelling
Indicative Map of European Biogeographical Regions: Methodology carabids (Coleoptera): spatial scale and the concept of communities.
and Development. Muséum National d’Histoire Naturelle, France. Ecography, 17, 166–175.
Genz, A., Bretz, F., Miwa, T., Mi, X., Leisch, F., Scheipl, F. & Hothorn, Niemelä, J., Halme, E., Pajunen, T. & Haila, Y. (1986) Sampling
T. (2011) mvtnorm: Multivariate Normal and t Distributions. R spiders and carabid beetles with pitfall traps: the effect of increased
Package Version 0.9-96 [WWW document]. URL http://CRAN.R- sampling effort. Annales Entomologici Fennici , 52, 109–111.
project.org/package=mvtnorm [accessed on 1 February 2012]. Niemelä, J., Halme, E. & Haila, Y. (1990) Balancing sampling effort
Grove, S.J. (2002a) Saproxylic insect ecology and the sustainable in pitfall trapping of carabid beetles. Entomologica Fennica, 1,
management of forests. Annual Review of Ecology and Systematics, 233–238.
33, 1–23. Niemelä, J., Spence, J.R. & Spence, D.H. (1992) Habitat associations
Grove, S.J. (2002b) The influence of forest management history on and seasonal activity of ground-beetles (Coleoptera, Carabidae) in
the integrity of the saproxylic beetle fauna in an Australian lowland central Alberta. Canadian Entomologist, 124, 521–540.
tropical rainforest. Biological Conservation, 104, 149–171. Nieto, A. & Alexander, K.N.A. (2010) European Red List of Saproxylic
Hammond, J.H.E., Langor, D.W. & Spence, J.R. (2004) Saproxylic Beetles. Publications Office of the European Union, Luxembourg.
beetles (Coleoptera) using Populus in boreal aspen stands of western Novotny, V. & Basset, Y. (2000) Rare species in communities of
Canada: spatiotemporal variation and conservation of assemblages. tropical insect herbivores: pondering the mystery of singletons.
Canadian Journal of Forest Research, 34, 1–19. Oikos, 89, 564–572.
Hedgren, O. & Weslien, J. (2008) Detecting rare species with random Ohsawa, M. (2007) The role of isolated old oak trees in maintaining
or subjective sampling: a case study of red-listed saproxylic beetles beetle diversity within larch plantations in the central mountainous
in boreal Sweden. Conservation Biology, 22, 212–215. region of Japan. Forest Ecology and Management, 250, 215–226.
Heyborne, W.H., Miller, J.C. & Parsons, G.L. (2003) Ground dwelling Økland, B. (1996) A comparison of three methods of trapping
beetles and forest vegetation change over a 17-year-period, in saproxylic beetles. Eurasian Journal of Entomology, 9, 195–209.
western Oregon, USA. Forest Ecology and Management, 179, Oksanen, J., Blanchet, F.G., Kindt, R. et al . (2011) vegan: Community
123–134. Ecology Package. R Package Version 2.0-2 [WWW document]. URL

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145
 Saproxylic beetle monitoring: sampling effects 145

http://CRAN.R-project.org/package=vegan [accessed on 1 February November 2003, Italy (ed. by M. Marchetti), pp. 207–226. European
2012]. Forest Institute, Finland.
Ranius, T. (2001) Constancy and asynchrony of Osmoderma eremita Sverdrup-Thygeson, A. & Birkemoe, T. (2008) What window traps can
populations in tree hollows. Oecologia, 126, 208–215. tell us: effect of placement, forest openness and beetle reproduction
R Development Core Team (2010) R: A Language and Environment for in retention trees. Journal of Insect Conservation, 13, 183–191.
Statistical Computing. R Foundation for Statistical Computing, Aus- Turchin, P., Taylor, A.D. & Reeve, J.D. (1999) Population cycles of a
tria. [WWW document]. URL http://www.R-project.org [accessed on forest insect: an experimental test. Science, 285, 1068–1071.
1 February 2012]. White, E.P., Adler, P.B., Lauenroth, W.K., et al . (2006) A comparison
Rink, M. & Sinsch, U. (2007) Radio-telemetric monitoring of of the species-time relationship across ecosystems and taxonomic
dispersing stag beetles: implications for conservation. Journal of groups. Oikos, 112, 185–195.
Zoology, 272, 235–243. Wikars, L.O., Sahlin, E. & Ranius, T. (2005) A comparison of
Salama, N.K.G., Knowler, J.T. & Adams, C.E. (2007) Increasing three methods to estimate species richness of saproxylic beetles
abundance and diversity in the moth assemblage of east Loch (Coleoptera) in logs and high stumps of Norway spruce. Canadian
Lomondside, Scotland over a 35 year period. Journal of Insect Entomologist, 137, 304–324.
Conservation, 11, 151–156. Williams, C. (1940) An analysis of four years captures of insects
Scott, W.A. & Anderson, R. (2003) Temporal and spatial variation in a light trap. Part 2. The effect of weather conditions on insect
in carabid assemblages from the United Kingdom Environmental activity; and the estimation and forecasting of changes in the
Change Network. Biological Conservation, 110, 197–210. insect population. Transactions of the Royal Entomological Society
Siitonen, J. (1994) Decaying wood and saproxylic Coleoptera in two of London, 90, 227–306.
old spruce forests: a comparison based on two sampling methods. Wood, S.N. (2008) Fast stable direct fitting and smoothness selection
Annales Zoologici Fennici , 31, 89–95. for generalized additive models. Journal of the Royal Statistical
Stokland, J., Tomter, S. & Söderberg, U. (2004) Development of Society B , 70, 495–518.
dead wood indicators for biodiversity monitoring: experiences
from Scandinavia. Monitoring and Indicators of Forest Biodiversity Accepted 22 December 2012
in Europe – From Ideas to Operationality, EFI Workshop, 12–15 First published online 1 March 2013

© 2013 The Royal Entomological Society, Agricultural and Forest Entomology, 15, 135–145