Journal of Insect Conservation

, 13:183

What window traps can tell us: effect of placement, forest openness and beetle reproduction in retention trees

Authors

    • Norwegian Institute for Nature Research – NINA
  • Tone Birkemoe
    • Norwegian Institute of Public Health
Original Paper

DOI: 10.1007/s10841-008-9141-x

Cite this article as:
Sverdrup-Thygeson, A. & Birkemoe, T. J Insect Conserv (2009) 13: 183. doi:10.1007/s10841-008-9141-x

Abstract

The use of flight interception traps (window traps) has been criticized for catching too many species without affinity to the immediate surroundings. We study aspen retention trees left for conservation reasons in a boreal forest in south-eastern Norway, and investigate how placement of window traps affects the beetle species assemblage, abundance of habitat specialists, saproxylic species and vagrant species. We also test the correlation between beetle trappings and beetle exit holes in wood. The window traps clearly responded to the immediate surroundings of the trap. Traps located on tree trunks had a different species assemblage than traps hanging freely. Traps mounted on trees caught more aspen associated beetles and less vagrant species than their free-hanging counterparts. The differences were larger when trees were dead than alive. There was a significant positive correlation between presence of individuals in the trunk-window traps and presence of exit holes for three aspen associated species. Thus, the trapping results indicated successful reproduction, showing that aspen associated beetles are not only attracted to but also utilise aspen retention trees/high stumps left in clear-cuts. This indicates that this conservation measure in forest management can have positive, alleviating effects concerning the dead wood deficit in managed boreal forest.

Keywords

ColeopteraSaproxylic beetlesNorwayAspenBoreal forestForest managementRetention trees

Introduction

Knowledge about how given insect species are associated with specific environmental properties is important. Of special interest for applied science is knowledge about the association with properties focused in the guidelines for sustainable forestry, such as retention trees and high stumps. Several methods can be used for beetle sampling, including pitfall traps, fogging, sieving of bark samples, rearing (emergence traps) and different kinds of flight interception traps. Different traps vary in their bias and their ability to target species associated with specific forest types, substrates etc. Even though the most efficient method for ensuring large samples in forest is the window trap (Hyvarinen et al. 2006), the limitations in the use of these traps have been focused by several authors lately (Jonsell and Weslien 2003; Martikainen and Kouki 2003; Ranius and Jansson 2002; Wikars et al. 2005; Økland 1996).

As window traps measure the activity around the trap location where it is located, the discussion concerns to what extent the beetle species caught in these traps represent species actually utilizing the substrate, as opposed to representing only transient visitors. The challenge is to tease apart these components. Using other traps or sampling techniques can sometimes give additional insight, but they all have their disadvantages (reviewed in e.g. Alinvi et al. 2007; Hyvarinen et al. 2006; Ranius and Jansson 2002; Wikars et al. 2005; Økland 1996). Since using window traps is and probably still will be a widely used technique, there is a need to get a better grip on the interpretation of the trapping results.

In this study we focus on aspen trees and their related beetle fauna. Among the tree species in the boreal forest, the aspen is unique in having a large number of species associated with it. Even though aspen trees make up about 1.5% of the volume of living trees in Norway (Statistics Norway; ssb@ssb.no), approximately 6% of the beetle species on the Norwegian Redlist is associated with this tree species (as the only or one of several tree species) (Kålås et al. 2006). In forestry, aspen used to be considered a pest species, because of its ability to outgrow and slow down the growth of spruce in the early seral stages of succession. Use of herbicides against aspen regrowth was common up till about 15 years ago. Recently, there has been much focus on sustainable forestry (“new forestry”, see e.g. Hansson 1997; Hunter 1990) and the extent of herbicide use in Norwegian forestry has dropped from an area of 102 km2 in 1990 to only 5 km2 in 2006 (Statistics Norway; ssb@ssb.no). More important, instead of trying to get rid of deciduous trees, they are now left in the clear cuts. With forest certification introduced in Norway around 1998, 5–10 retention trees per hectare should be left at final felling. The trees left in clear-cuts in Norway have mostly been deciduous trees and contrary to Sweden, high stump cutting has not been a common practice (Sverdrup-Thygeson et al. 2005).

This paper is part of a larger study in which the main purpose is to evaluate the importance of aspen trees as habitat for beetles in boreal forest, e.g. as retention trees in forest management. To achieve this we also wanted to know more about the sampling effects of different placements of window traps. We further wanted to study the beetles’ attraction to aspens and how this related to reproduction in the substrate. Therefore, we set up a study comparing catches in trunk-window traps mounted on aspen trees with catches in free-hanging window traps in the same forest stands. The comparisons were carried out over 3 years, with alive, 1 year dead and 2 years dead aspens in consecutive years. To address possible effects of major habitat differences, both open and in closed canopy forest were included.

We then ask the following questions:
  1. 1.

    To what degree does mounting the trap on a tree trunk as opposed to hanging the trap freely away from trees, affect the general species assemblage of beetles, the proportion of saproxylic beetles, the proportion of aspen associated beetles and the proportion of vagrant species?

     
  2. 2.

    How does openness around the trap (clear-cut versus closed canopy forest) affect the outcome of the two different trap placements?

     
  3. 3.

    How do exit holes in coarse woody debris (CWD) correlate with trapped samples of the same species?

     
  4. 4.

    What do the results tell us about the conservation value of retained aspen trees?

     

Materials and methods

Study area

The study was conducted in a 40 km2 study area in the boreonemoral vegetation zone (Moen 1998) in Østmarka, Akershus county, south eastern Norway. The bedrock in Østmarka is Prekambrian, dominated by gneisses, and covered with thin layers of glaciofluvial deposits (ngu.no/kart/losmasse). The study area is dominated by coniferous forest, mainly spruce (Picea abies) but with pine (Pinus sylvestris) on the ridges, with some admixture of deciduous trees like birch (D. pubescens) and aspen (Populus tremula). The climate is slightly suboceanic with annual mean precipitation of 760 mm and annual mean temperature of 4.1°C.

Within the study area, forest stands with large deciduous trees (according to the forest inventory database) were surveyed and all aspen trees exceeding 20 cm diameter and situated at least 10 m away from the stand border were mapped using a GPS. From the resulting database of 230 trees, we selected at random 15 trees in 2–4 years old clear-cuts and 15 trees in closed canopy forest (age 90–120 years), totalling 30 aspen trees. The minimum distance between trees had to be at least 100 m.

Beetle sampling and killing of trees

The beetle sampling commenced in spring 2001, when large (40 cm × 60 cm) trunk-window traps (Kaila et al. 1997) were mounted on the 30 trees. Each trunk-window trap (TWT) was mounted on a tree, facing south and with the lower edge of the window pane 1 m above ground. We twinned each trunk-window trap with a free-hanging window trap (WT) (hanging at the same height from a tripod of long sticks) placed within the same forest stand and at least 10 m away from stand border, but as far from any tree as possible. The average distance between the twinned traps was 32 m. We refer to “trap placement” in this paper as meaning the difference between free-hanging window trap (WT) and trunk-window traps (TWT). Trapping design is illustrated in Fig. 1
https://static-content.springer.com/image/art%3A10.1007%2Fs10841-008-9141-x/MediaObjects/10841_2008_9141_Fig1_HTML.gif
Fig. 1

Schematic drawing illustrating the study design and the traps. (a) Study design: grey areas denote closed canopy forest, white areas denote clear-cuts, tree symbols indicates trunk-window traps (TWTs, totalling 15 in clear-cuts and 15 in closed canopy forest) while circles mean free-hanging window traps (WTs, also totalling 15 in clear-cuts and 15 in closed canopy forest). Traps: (b) Trunk window traps TWT and (c) Window traps hanging free WT

.

In the late fall of 2001, all trees were cut approximately 4 m above ground using detonating chord. The resulting logs were left at the base of the high stump, to minimize the difference between the live tree (intact with upper bole and canopy) and the dead tree (upper bole and canopy changed to a log with branches at the base of a high stump). The beetle trapping continued in 2002 and 2003, all years operating from middle of May to middle of August. In addition, the TWTs were also operating in 2004; these data are used in one analysis.

All beetle individuals were identified and categorised as aspen associated (meaning beetles having aspen as a preferred host tree species in the study region, see Table 1), other saproxylic (wood living) beetles and non-saproxylic beetles according to the literature (Hansen et al. 1908–1965; Palm 1959), a database compiled by Dahlberg and Stokland (accessible at http://www.saproxylic.org/ “Go to database”, see also Dahlberg and Stokland 2004; Stokland and Meyke in press) and the Norwegian Red List Database (http://www.biodiversity.no/Article.aspx?m=39&amid=1864).
Table 1

Species categorised as aspen associatesa in the present study: wood status, life cycle, diet, occurrence in other tree species and red list status

Aspen associated beetles

Wood status

Life cycle (years)

Diet

Other tree species D: deciduous C: coniferous

Red list status

Literature

Living

Recently dead

Rotten

Aegomorphus clavipes (Schrank 1781) (Cerambycidae)

x

2

Xylophagous

D

 

1,2,3,4

Ampedus nigroflavus (Goeze 1777) (Elateridae)

x

x

2–4

Xylophagous and predator

D

NT

1,2,3

Cerylon deplanatum (Gyllenhal 1827) (Colydiidae)

x

x

1?

Predator and/or scavenger

D

 

1,2

Cerylon ferrugineum (Stephens 1830) (Colydiidae)

x

x

1?

Predator and/or scavenger

D, C

 

1,2

Cucujus cinnaberinus (Scopoli 1763) (Cucujidae)

x

>2

Xylophagous

D, C

VU

1,2

Cyphea curtula (Erichson 1837) (Staphylinidae)

x

Most likely scavenger

D

 

1,2

Endomychus coccineus (Linnaeus 1758) (Endomychidae)

x

x

Eats bark, wood or fungi

D

 

1,2

Homalota plana (Gyllenhal 1810) (Staphylinidae)

x

x

-

D, C

 

1,2

Mycetophagus fulvicollis (Fabricius 1792) (Mycetophagidae)

x

x

1?

Fungivorous

D, C

NT

1,2

Orchesia micans (Panzer 1794) (Melandryidae)

xb

xb

1

Fungivorous

D, C

 

1,2,3

Platysoma deplanatum (Gyllenhal 1808) (Histeridae)

x

x

Predator

D

 

1,2

Ptilinus fuscus (Geoffroy 1785) (Anobidae)

x

1–2

Xylophagous

D

 

1,2,3

Rhizophagus parvulus (Paykull 1800) (Monotomidae)

x

x

1

Sap eater, possibly predator and scavenger

D

 

1,2

Saperda perforata (Pallas 1773) (Cerambycidae)

x

1–2

Xylophagous

 

1,2,3,4

Trypophloeus bispinulus (Eggers 1927) (Curculionidae)

x

x

2?

Xylophagous

 

1,2,3

Xylotrechus rusticus (Linnaeus 1758) (Cerambycidae)

x

x

2–3c

Xylophagous

D

 

1,2,3,4

Literature: Palm (1951)1, Palm (1959)2, Ehnström and Axelsson (2002)3 and Bílý and Mehl (1989)4

aPreference for this wood quality by information sources used in Dahlberg and Stokland (2004)

bLive in fruiting bodies of bracket fungi

cIn our study, several exit holes observed in fall 2003 indicate that many larvae developed in only 1 year

All snags and accompanying logs were searched systematically for galleries and exit holes in the fall of 2004 (although in many cases these exit holes were recognised already in the fall of 2003). We counted the number of exit holes of Xylotrechus rusticus and the number of empty and intact (i.e not predated by wood-peckers) pupal chambers of Saperda perforata and Aegomorphus(Acanthoderes) clavipes.

To contrast living versus dead trees, we only use data from 2001 to 2003 in most of the analyses in this paper.

Statistical methods

Species assemblages were compared using DCA (Canoco for Windows 4.5). The data were log10-transformed to reduce the importance of common species relative to uncommon species. Species occurring only once in one particular year were not included in the analysis. The default option detrending by segments was used.

The number of individuals sampled in the different placement categories (WT in open forest, WT in closed forest, TWT in open forest, TWT in closed forest) was rather different and this would affect also the number of saproxylics and the number of aspen associated beetles. The data did mostly not conform to normality in Shapiro-Wilk tests, or to assumptions of homogeneity of variance between samples in Levene’s tests, and satisfactory transformation was not possible for several of the variables. As our interest here was not in absolute numbers, but in the relative distribution in each category, chi-square tests and non-parametric methods was therefore used instead of ANOVA. We used Wilcoxon signed ranks test for the proportions in the dependent samples of WTs versus TWTs, and Mann–Whitney U-test for the comparisons of the independent samples from open versus closed forest. To investigate the relationship between exit holes and trap catches, logistic regression and linear regression on transformed variables were used. The number of exit holes of Xylotrechus rusticus was categorized as an ordinal variable (0, 1–50, >50) in the analysis to handle non-normality.

Results

Altogether 19,819 individuals from 653 beetle species were sampled in the study. The trap data from the living aspens (2001-data) consisted of 6,469 beetles representing 454 species, while the data from the traps mounted on the aspens when they were dead contained 4,805 individuals of 389 species in 2002 and 6,224 individuals of 404 species in 2003. The TWTs in 2004 collected 2,321 individuals of 229 species.

Difference in species assemblage depending on trap placement

The trunk-window traps (TWTs) and free hanging window traps (WTs) showed a similar species composition in the closed forest in 2001 and 2003 (Fig. 2). In the open areas the TWTs and WTs was partly separated when the trees were alive in 2001, and totally separated in 2003 when decay had advanced in the high stumps. The species composition varied between the open and closed forest both years as they separate the samples into two groups along the first ordination axes.
https://static-content.springer.com/image/art%3A10.1007%2Fs10841-008-9141-x/MediaObjects/10841_2008_9141_Fig2_HTML.gif
Fig. 2

DCA ordination plot of beetle species in the four different trap placement categories in (a) 2001 and (b) 2003. Trunk-window trap (TWT) in forest (circles), window trap hanging free (WT) in forest (diamonds), TWT in open (squares) and WT in open (rectangles). Eigenvalues of the first two axis were 0.37 and 0.22 in 2001 (sum 5.8) and 0.35 and 0.25 in 2003 (sum 5.0). Percent variation explained was 10% in (a) and 12% in (b)

Proportion of singletons, aspen associates and saproxylics depending on trap placement

If beetles are attracted to the aspen trees, one would expect the proportion of vagrant species (“tourists”) to be less in the trunk traps than in the window traps. Assuming that many of the vagrants would occur in low numbers and typically be singletons, we evaluated the proportion of singletons in free hanging window traps (WT) and in trunk-window traps (TWT) when trees were alive (2001) and when trees were dead (2003). The results show that there were indeed higher proportions of singletons in the WTs, and these differences were statistically significant both in 2001 (41% vs. 36%, Wilcoxon signed ranks test, Z = −2.11, P = 0.04, N = 30) and in 2003 (44% vs. 35%, Wilcoxon signed ranks test, Z = −2.36, P = 0.02, N = 30).

If a substantial part of the beetles actually are attracted to the aspen trees, the proportion of aspen associated beetles on the other side should be higher in the trunk-window traps than in the traps hanging freely. We find that the traps mounted on the aspen trees seem to sample more aspen associates, even though they sampled fewer beetles altogether and fewer saproxylic beetles than the free hanging traps (Table 2). When the trees were alive, this difference was not statistically significant (Wilcoxon signed ranks test, Z = −1.65, P = 0.11, N = 60), but when the trees were dead, it was quite clear that the trunk-window traps had a higher proportion of aspen associated individuals than their free hanging counterparts (Wilcoxon signed ranks test, Z = −3.88, P < 0.0001, N = 60). In the dead trees, Xylotrechus rusticus forms an outlier with 158 individuals in TWT in open forest, and rather few individuals in the other categories. Therefore we repeated the analysis without this species, still resulting in a significant difference between the TWTs and the WTs (Z = −3.63, P < 0.0001, N = 60). Actually, more than 90% of the trapped beetle individuals belonging to one of the 16 species defined as aspen associates was trapped in trunk-window traps (Table 2).
Table 2

Abundance and species richness of aspen associates, other saproxylic beetles and non-saproxylic beetles in the study, in the four different trap placement types

 

Individuals

 

Species

 

Closed

Clear-cut

Closed

Clear-cut

TWT

WT

TWT

WT

Sum indiv.

TWT

WT

TWT

WT

Sum species

2001

Aspen associated

2

4

24

2

32

2

3

7

1

8

Other saproxylic

441

808

1600

1710

4559

92

110

163

153

245

Non-saproxylic

421

399

433

625

1878

60

68

87

140

201

Total

864

1211

2057

2337

6469

154

181

257

294

454

2003

Aspen associated

9

2

267a

31

309

5

2

16

10

16

Other saproxylic

687

774

1496

1890

4847

94

113

153

148

226

Non-saproxylic

114

200

267

487

1068

36

62

63

115

162

Total

810

976

2030

2408

6224

135

177

232

273

404

TWT, traps on treetrunk; WT, traps hanging free in 2001 and 2003. N = 60

aXylotrechus rusticus 158 individuals

Concerning other saproxylics, the proportion of individuals was not different between WTs and TWTs in either year when excluding the outlier X. rusticus, but the proportion of other saproxylic species was higher in TWTs both when trees were alive (2001: Close to statistically significant; Wilcoxon signed ranks test, Z = −1.57, P = 0.06, N = 60) and when trees were dead (2003: Wilcoxon signed ranks test, Z = −3.67, P < 0.0001, N = 60) (Table 2).

Effect of openness around traps

The proportion of aspen associates was different in freehanging and trunk-mounted traps even when the exposure effect was accounted for, according to Table 2 and a three-way chi-square model testing the proportion of aspen associates in different traps in the dead aspens (2003-data). As earlier studies have shown that sun exposure affects abundance of beetles, we assumed the null hypothesis in this three-way chi-square model to be that beetle individuals are distributed differently in clear-cuts and forest, but evenly among the trap types. The three-way chi-square test results indicate that this hypothesis must be rejected (chi-square 104.5, df = 7, P < 0.001).

We also analysed the effect of openness for WTs and TWTs in separate analyses. For the WTs when the trees were alive, there was no difference in proportion of aspen associates between the open clear-cuts and the closed canopy forest (Mann–Whitney U, Z = −0.67, P = 0.34, N = 30), while there was significantly higher proportion of aspen associates in the open WTs than in the closed forest WTs when the trees were dead (Mann–Whitney U, Z = −2.26, P = 0.04, N = 30). For saproxylic individuals in general, the proportion in WT was not significantly higher in open forest in either of the years, but close to significant in 2001 (Mann–Whitney U; 2001: Z = −1.576, P = 0.06, N = 30 and 2003: Z = −0.89, P = 0.19, N = 30).

For the TWTs, there was significantly higher proportion of aspen associates in the traps in the open clear-cuts than in the closed canopy forest, both when the trees were alive (Mann–Whitney U, Z = −2.76, P = 0.01, N = 30) and when the trees were dead (Mann–Whitney U, Z = −4.31, P < 0.0001, N = 30). For saproxylic individuals in general, the proportion in TWT was significantly higher in open forest in 2003, but not in 2001 (Mann–Whitney U; 2001: Z = −4.09, P < 0.0001, N = 30 and 2003: Z = −0.87, P = 0.20, N = 30).

Development over years

We compared the abundance of the aspen associated species in the traps hanging free and the traps mounted on aspens, in 2001, 2002 and 2003. Two clear trends can be seen (Fig. 3): Firstly, the abundance in TWTs was always higher than the abundance in WT for a specific species. Secondly, for all aspen associated species except one (Orchesia micans), the abundance increased from 2001 to 2003.
https://static-content.springer.com/image/art%3A10.1007%2Fs10841-008-9141-x/MediaObjects/10841_2008_9141_Fig3_HTML.gif
Fig. 3

Abundance of the aspen associated species when trees were alive (2001), 1 year dead (2002) and 2 years dead (2003). White bars are traps hanging free (WT), hatched bars are traps mounted on the aspens (TWT). N = 60

Presence of exit holes of Xylotrechus rusticus and exit holes from empty, non-predated pupal chambers of Saperda perforata and Aegomorpus clavipes were correlated to pooled trap data from 2002 to 2004 of the same species in a logistic regression (excluding 2001, assuming that individuals of X. rusticus trapped that year would not be able to reproduce, as the trees were alive). For all three species, there was a significant positive correlation between presence of trapped individuals and presence of exit holes, indicating that presence of a species in the traps either resulted in or was a result of reproduction in the same tree (Table 3).
Table 3

Logistic regression on the presence of exit holes and the presence of individuals in the trunk-window trap of each species

Species

R2

Likelihood ratio chi square

P-value

N

Xylotrechus rusticus

0.55

19.8

<0.0001

30

Saperda perforata

0.15

5.9

0.02

30

Aegomorphus clavipes

0.11

4.5

0.03

30

We further related the abundance of exit holes of each species to the abundance of individuals in the traps, excluding the trees where no production had taken place (no exit holes present). For X. rusticus, the analysis yielded a significant positive correlation between number of exit holes and production (Logistic regression on the ordinal variable, R2 = 0.42, P = 0.04, N = 20), but there was a substantial numerical difference between the number of exit holes and the number of trapped individuals (Fig. 4). Data on S. perforata and A. clavipes were pooled together, as the abundance data for each species separately was hard to transform to normality. The analysis of S. perforata and A. clavipes pooled showed no significant correlation between number of exit holes and production (Regression on Log-transformed data, R2 = 0.08, P = 0.29, N = 16).
https://static-content.springer.com/image/art%3A10.1007%2Fs10841-008-9141-x/MediaObjects/10841_2008_9141_Fig4_HTML.gif
Fig. 4

The relationship between number of exit holes in the high stumps and the accompanying logs, and the number of individuals trapped in the trunk-window traps (TWT). (a) Xylotrechus rusticus (b) Saperda perforata (c) Aegomorphus clavipes

Discussion

In general, we found that window traps do not catch beetles indiscriminately, but that the samples respond to the immediate surroundings of the trap. We found that trap placement affected species assemblage, in that traps mounted on tree trunks had a different species assemblage than traps hanging freely in the open forest (clear-cuts) and partially in closed forest. Some of this difference in species composition may be related to the fact that traps mounted on trees caught more aspen associated beetles and less vagrant species than their free-hanging counterparts.

It is interesting that these differences in species assemblage, proportion of aspen associated beetles and proportion of singletons (beetle species occurring only once) become more pronounced when the trees are dead (2003) than alive (2001). Even though one might argue that age of snag is not independent of year of collection in our design, and that the removal of the upper bole and canopy from the live trees (left at the base of the high stump) could influence on our results, it is not likely that these factors would produce patterns like an increase in aspen associated beetles. Also, data from more years following the death of the trees (Fig. 3 and unpublished data) shows that this contrast is not a result of yearly random variation in the profile of beetles flying in 2001 as opposed to 2003, but rather indicates a trend. As the sampled trees in open locations were situated on clear-cuts already 2–4 years old at the outset of the experiment, it seems that this must be an effect of the change in the trees themselves and not a confounding effect of the clear-cutting process.

This indicates that the trunk-window traps indeed intercept some tree-level effect on species attraction, contrary to the results of Saint-Germain et al. (2006), who found very high degree of similarity in beetle assemblages, not only between snags of different ages (=decay stages) but also between snags of different tree species and stovepipe controls. These differences might result from the rather different study design of Saint-Germain et al.; they studied snags in a closed canopy forest, while half of our snags were located in clear-cuts. Both the difference we detect in our study between clear-cuts and closed forest, and the results of Saint-Germain et al. might indicate that a tree-level effect on species attraction is more pronounced in clear-cuts than in closed forest.

Earlier studies have emphasized the importance of sun exposure on abundance and species richness of beetles, especially for beetles in deciduous trees (Lindhe et al. 2005; Sverdrup-Thygeson and Ims 2002), although there are also studies on spruce which find no effect of sun exposure on species richness (Wikars et al. 2005). In this study, the results show that openness around the traps not only affected abundance and thereby species richness, but also the relative abundance of aspen associates, other saproxylic beetles and non-saproxylic beetles. The response to trap placement was not similar in clear-cuts and closed forest. When comparing living trees (2001) and dead trees (2003), the effect of the trees dying was more pronounced on clear-cut than in closed forest. One reason for this could be that the difference between the two trap placements was larger in clear-cuts than in closed canopy forest; free-hanging traps are necessarily closer to other trees in the forest than on the clear-cut.

The number of aspen associates in the trunk-window traps was higher in the open clear-cuts than in closed forest, maybe because several of the aspen associates prefer sunny locations for breeding (Ehnström and Axelsson 2002; Lindhe et al. 2005). The proportion of aspen associates in free hanging traps differed significantly between clear-cuts and closed forest only when the trees were dead. This might indicate that, in addition to the tree level response, there is also some response on a stand level to the increased number of aspen associates attracted to and/or produced in the dead aspens.

An important question is to what degree the samples in trunk-window traps are reflecting beetle attraction to the tree, and to what degree it is reflecting actual production of individuals in that tree. We evaluated the relative importance of these processes by comparing beetle trappings with beetle exit holes from the same species. We assumed that a correlation between emergence holes and beetle captures indicated local reproduction as there were practically not a single exit holes in the boles of the living trees, while there were lots of exit holes to be found here in the fall of 2003. For the three relevant species there was a significant positive correlation between presence of individuals in the TWTs and presence of exit holes, indicating that for these three aspen associated species, trapping results can be trusted as indicating successful reproduction. We further compared the actual abundance of individuals of these three species with the abundance of exit holes, and found that for the rather common beetle X. rusticus the trapping results could also be used to indicate the quantity of production, while this was not the case for the pooled data of the much more uncommon beetles S. perforata and A. clavipes. This last result could be an artefact of the small samples of the rarer beetles. Still, it is clear (Fig. 4) that the trunk-window trap often catches only a small proportion of the individuals produced in the trunk.

Effect of attraction versus production can also be evaluated by considering the yearly abundance of different aspen associates (Fig. 3). Since the trees were turned into high stumps and an accompanying log in the late fall of 2001, the first season for colonisation by beetles dependent on dead aspen would be 2002. Most of the aspen associates in this study are species utilising the cambium of recently dead trees (Table 1). One would expect these species to start arriving soon after the tree’s death, while the maximum number of trapped aspen associated individuals would be expected later, as the larvae developing in the aspens hatch and add to the attracted individuals. This is exactly the pattern we find in Fig. 3, which shows a surprisingly clear pattern across the species. Unpublished trap data from the following 2 years (2004 and 2005) from the same aspen trunks, indicate that yearly trap catches of the cambium living species decline after 2003, while late-successional species such as Ptilinus fuscus increase. A factor that could confound this pattern would be if the trees attract a sharply increasing number of individuals of a certain species as the decay advances, but it is not likely that this dominates the pattern for all the aspen associates in the study. The increase in number of aspen associated species seen in Fig. 4 is therefore most likely a result of reproduction in the aspens substrate.

The use of passive techniques such as window traps have been criticized in some studies as they collect too many transient species having no affinity with the immediate surroundings (Saint-Germain et al. 2006), although other authors have recognised the substantial “baiting effect” of mounting window traps on tree trunks (Hyvarinen et al. 2006 and references therein). The results of this study supports the view that beetle catches in window traps indeed respond both to trap placement (close to host substrate or not) and to the decay stage of the substrate (living trees or dead). Also, for some specialised beetles using aspens as main host, presence in trunk-window catches is shown to reflect successful reproduction in the same dead tree. This indicates that leaving retention trees gives aspen associates an added opportunity for breeding compared to not leaving retention trees, although we are not able to evaluate the importance of this relative to all aspen substrate in the forest. Understanding how window traps work is useful knowledge both when evaluating the effect of different conservation measures in forestry such as retention trees and high stumps, and when interpreting samples from trunk-window traps in general.

Acknowledgements

We are grateful to Espen Wandås for assistance in field, to Sindre Ligård for identification of beetles and to Steve Coulson, Erik Framstad and Frode Ødegaard for useful comments on an earlier version of the manuscript. The study was possible due to grants from the Norwegian Research Council (Part of NFR project no. 140161/110 and 163230).

Copyright information

© Springer Science+Business Media B.V. 2008