Occurrence and abundance of fungus – dwelling beetles ( Ciidae ) in boreal forests and clearcuts : habitat associations at two spatial scales

Occurrence and abundance of fungus–dwelling beetles (Ciidae) in boreal forests and clearcuts: habitat associations at two spatial scales.— Insect material (> 30,000 individuals) reared from the fruiting bodies of wood–decaying Trametes fungi was compared between old–growth boreal forests and adjacent clearcuts in Finland. Sulcacis affinis and Cis hispidus occurred more frequently and were, on average, more abundant in the clearcuts. Interestingly, Octotemnus glabriculus and Cis boleti had a slightly higher frequency of occurrence in the forests, despite lower resource availability. The former also showed a higher average abundance. On average, the cluster size of Trametes fruiting bodies occurring on woody debris was higher in the clearcuts than in the forests and had a positive effect on species occurrence and abundance in these clusters. The independent effect of the macrohabitat (forest or clearcut) underscores the importance of the macrohabitat where specific resources occur, and this may override the positive effects of resource availability.


Introduction
Ecological research on habitat fragmentation has often metaphorically viewed suitable habitats as "islands" in a hostile "sea" (Haila, 2002).Because there is a low degree of deforestation in the boreal forest, landscapes rarely appear simply as a black-and-white contrast between habitat and non-habitat, rather the many shades of grey reflect differential habitat suitability (Mönkkönen & Reunanen, 1999).Forestry practices in boreal forests create highly dynamic landscapes which remain forested while undergoing spatial and temporal changes in structure and dynamics (Kouki et al., 2001;Schmiegelow & Mönkkönen, 2002).The structural heterogeneity in the forest landscape is manifested in the amount and quality of living and dead wood (Siitonen, 2001), forest-stand age structure and naturalness (Uotila et al., 2002), and in the distribution of different stand types in the landscape (Löfman & Kouki, 2003).
Dead wood is a key resource for thousands of organisms in boreal forests (Esseen et al., 1997;Siitonen, 2001).Dead wood is scarce in managed forests and many dead-wood dependent organisms are consequently absent or occur in low numbers.This has contributed to the misconception that the majority of these species would require natural forests.However, many dead-wood dependent species can persist in clearcuts if critical resources are left in adequate densities and qualities in management operations (Kaila et al., 1997;Jonsell et al., 2001;Jonsson et al., 2001;Martikainen, 2001;Sverdrup-Thygeson & Ims, 2002).However, the importance of the macrohabitat and landscape context where these resources occur is poorly understood, particularly for species that have high habitat specificity and limited powers of dispersal, such as fungus-dwelling insects (Jonsell et al., 1999;Jonsson et al., 2001;Komonen et al., 2000).It is important to disentangle the large-and small-scale effects of forestry on habitat suitability to fully understand the limiting factors for species occurrences and abundances (Mönkkönen & Reunanen, 1999).
Trametes fruiting bodies occur in a variety of forest environments where deciduous dead wood is available.Nevertheless, it is not known how Trametes-dwelling beetles respond to the different forest surroundings.There are generally marked ecological differences among the fungus-dwelling insects in dispersal ability (Jonsell et al., 1999;Jonsson, 2003), habitat requirements (Nilsson, 1997;Guevara et al., 2000a;Thunes et al., 2000;Jonsell et al., 2001) and host-fungus specificity (Lawrence, 1973;Økland, 1995;Fossli & Andersen, 1998;Guevara et al., 2000b;Komonen, 2001).In this paper, the occurrence and abundance of four fungivorous beetle species (Ciidae) co-occurring in Trametes fruiting bodies is investigated at two spatial scales.The small-scale effects of fungal-cluster size and the large-scale effects of macrohabitat (forest-clearcut) are tested.
Table 1.Study site characteristics.Area of forest refers to the area (ha) with > 60 m 3 fallen woody debris (diameter at breast height ≥ 7 cm) ha -1 : As.Area sampled; Af.Area of the forest; Ac.Area of the clearcut, refers to the area (ha) logged at the same time and consists of interconnected openings rather than a single large opening; CWD.Number of stumps, snags, logs and branches (diameter ≥ 10 cm) of birch and aspen ha -1 ; Yl.Year of logging; MAl.Mean age at the time of logging, refers to the mean age of trees belonging to the dominant canopy storey; Vl.Volume at logging, refers to the volume of living spruce, pine and birch at the time of logging.

Study sites
The study area is located in North Karelia, eastern Finland, in the middle boreal zone (63 o 00' -30' N, 30 o -31 o E).First, 12 spruce-dominated oldgrowth stands were visited.They were rich in aspen (Kouki et al., 2004) and thus assumed to have a sufficiently high density of Trametes to provide adequate sample size.Only four stands were adjacent to clearcuts and these were selected for the study: 1. Ruunavaara; 2. Pieni

Study system
Four species of Trametes (formerly included in Coriolus Quél.) occur in the study region in eastern Finland.Of these, Trametes ochracea (Pers.)Gilb.& Ryvarden is the most common species inhabiting dead deciduous trees, mainly aspen (Populus tremula L.) and birch (Betula spp.) (Niemelä, 2001).Over 95% of our samples were T. ochracea (K.Junninen det.), but as the fruiting bodies of all Trametes species are very similar in physical structure (Ryvarden & Gilbertson, 1993) it was impossible to identify some of the heavilyconsumed samples with certainty.As far as it is documented, there are no great differences in the Ciidae fauna associated with the different species of Trametes found in Fennoscandian boreal forests (Fossli & Andersen, 1998;Selonen, 2004).Trametes fruiting bodies are annual and typically occur in a relatively early phase of decay succession (3-7 yrs; Hintikka, 1993).They are common in woody debris and stumps, and typically form clusters of fruiting bodies.Although the fruiting bodies of Trametes are annual, dead fruiting bodies can remain attached to wood for one to two years and become entirely consumed by insects, mainly larvae and adults of Ciidae.Hovinvaara; 3. Haapahasianvaara; 4. Vankonvaara (table 1).As not all stands had large enough (≥ 2 ha) adjacent clearcuts, two of the sampling quadrats (see below) in site 4, and one in site 2 were located further from the studied forest, yet within 1-km distance and adjacent to other patches of oldgrowth.All the clearcut areas had once been part of the larger old-growth forest but had been logged 7 or 8 years earlier; part of the clearcut in site two had been logged 11 years earlier (data from Forest and Park Service, Lieksa).This oldest clearcut area was the only one with 2-3 m tall birch trees, the others were fully open.The small difference between the mean fungal density in site two results from this difference (fig.1).

Sampling
Fungal samples were collected between 22 IX and 6 X 01, the optimal time given the species phenology.One hectare study quadrats were randomly positioned and marked in the field; in the forests, quadrats were at least 50 m from the forest-clearcut edge.An equal area was sampled in a given oldgrowth forest and the adjacent clearcut (table 1).
The sampled areas varied in size among forestclearcut pairs, because larger forest stands allowed larger sampling coverage, thus increasing sample size and statistical power.However, the area sampled does not consistently follow the area of forests due to the discrepancy between the forest area and the area and shape of the adjacent clearcuts.Due to the shape and size of the clearcut in site 4, it was impossible to establish all the sample units as 1-ha quadrats; instead we used 200 m x 50 m strips.In sites 3 and 4, two 50 m x 50 m quadrats were used for sampling the area of 0.5 ha.
In all study quadrats we examined all woody debris (dbh ≥ 10 cm) of deciduous trees for Trametes fruiting bodies.As Trametes fruiting bodies often occur in tight clusters, insect larvae could potentially move from one fruiting body to another.Thus, all the fruiting bodies in a cluster were considered one sample and carefully removed.If the fungal clusters were located in separate parts of the same individual tree, for example in a stump and a trunk not attached to each other, these were considered separate samples.Samples were transferred to meshnet covered plastic boxes and kept in outdoor conditions.On 8 II 02, all the samples were transferred to room temperature and weighed after two weeks, a period which was considered adequate for excess water to evaporate and make the weight of samples collected under different daily weather conditions more comparable.The fungal cluster size was determined by weighing, reflecting both the number and size (g) of fruiting bodies.All the fruiting bodies were carefully dissected and the adult insect individuals were removed and identified.

Statistical analyses
Generalized Linear Models (GLM; McCullagh & Nelder, 1989) were used to analyze the data.In all models, site was introduced as a random effect and management category (forest or clearcut) as

Patterns of beetle occupancy
A total of 32,193 insect individuals were removed from the sampled fruiting bodies.These included 12 species of Ciidae, the four most common species of which (96% of all insect individuals) were used in the analyses (table 2).Fungal-cluster weight had a positive effect of similar magnitude on the probability of a cluster being occupied for all four species (tables 3, 4).Sulcacis affinis and Cis hispidus had a very consistent pattern of occurrence in all sites, in comparison with Octotemnus glabriculus and Cis boleti, in that they were more frequent in clearcuts (fig.2, table 4).Interestingly, O. glabriculus and C. boleti were tributed errors were assumed.The significance of each term was evaluated based on the increase in deviance when the term was dropped from the full model containing all explanatory parameters.Fungal-cluster weight and the number of individuals were log 10 (x+1)-transformed in all analyses, unless otherwise stated.

Host availability
Altogether, 106 and 245 pieces of woody debris occupied by dead fruiting bodies of Trametes were recorded and sampled from forests and clearcuts, respectively.The density of woody debris occupied by Trametes was consistently lower in the forest sites (mean ha -1 ± SE = 7.34 ± 0.63) than in the clearcut sites (16.00 ± 2.97; fig.1A).

Cis hispidus Cis boleti
more likely to be found from forests, as indicated by signs of parameter estimates (table 4), despite lower fungal availability (fig.1).For the two "forest species", the predicted probability of occurrence increased more slowly as a function of fungalcluster weight in the clearcut clusters than in the forest clusters (fig.3).The opposite pattern was observed for the two "clearcut species".

Discussion
Findings from this study demonstrate that the macrohabitat where specific resources occur is important for ciid beetles in Trametes fruiting bodies (c.f.Thunes & Willassen, 1997;Jonsell et al., 2001;  Table 5. GLM results on the mean abundance of ciid beetles in fungal clusters.The change in deviance indicates the model improvement when a given term is included in the model.The interaction term between management category and weight was non-significant for each species (F < 2.63, P > 0.1), and thus it was excluded from the final models: η 2 .Proportion of explained variance; Mng.Management nested within site; W. Weight; Rd.Residual deviance; * d.f. is only 3 because one site did not include any occupied fungal clusters).
Tabla 5. Resultados GLM de la abundancia media de coleópteros cíidos en las masas fúngicas.El cambio de la desvianza indica la mejora del modelo cuando se incluye en éste un término dado.El término de interacción entre la categoría de gestión y el peso para cada especie (F < 2,63, P > 0,1) no fue significativo, por lo que se excluyó de los modelos finales: η 2 .Proporción de varianza explicada; Mng.Manejo anidado dentro del área; W. Peso; Rd.Desvianza residual; * d.f.es sólo de 3 debido a que uno de los lugares no incluía ninguna masa fúngica ocupada).et al., 2001).Most interestingly, some ciids seem to prefer forests over clearcuts despite lower fungal density and smaller size of fungal clusters in the former.In the absence of experiments, however, we cannot make explicit causal inferences.We therefore discuss two potential ecological factors that could contribute to the difference observed in beetle species occurrence and abundance between forests and clearcuts.

Small-scale effects of fungal cluster
For all ciid species of the present study, fungalcluster weight contributed positively to the prob-ability of a sample being occupied, both in the forest and clearcut (see also Midtgaard et al., 1998).The higher frequency of occurrence of ciids can be explained by the greater probability of detecting larger fungal clusters (actively or by chance only), as well as by a longer expected persistence time of the local beetle population in larger clusters.There is evidence that both walking and flying ciids are attracted to the volatile compounds of their host fungus rather than finding the fungi accidentally (Jonsell & Nordlander, 1995;Guevara et al., 2000aGuevara et al., , 2000b)).If larger fungal clusters emit greater amounts of volatile compounds, then they would also attract more specific ecological requirements.In larger clusters there is also more potential for resource partitioning, which could facilitate coexistence by reducing interspecific competition between ciid species (Guevara et al., 2000a).Of the species recorded in this study, only S. affinis is known to inhabit -in great numbers-fruiting bodies of fungal species other than Trametes, namely Pycnoporus cinnabarinus (Jacq.: Fr.) P. Karsten (Økland, 1995).The fruiting bodies of this fungus occur exclusively in warm, open areas (Niemelä, 2001) and were also abundant in the clearcuts of this study.This supplementary host fungus could contribute to the very high frequency of occurrence of S. affinis in the clearcuts.Larger fungal clusters inherently support larger ciid populations (Midtgaard et al., 1998), which in turn could affect local occurrence patterns via increased number of dispersing individuals.Evidence on the dispersal ability of ciid beetles is scarce and indirect, making it impossible to assess the dispersal rate  In this study it is shown that "forest species" have a higher probability of occurrence in a fungal cluster of a given size in forests (optimal macrohabitat) vs clearcuts (suboptimal macrohabitat), and "clearcut species" show an opposite pattern.These results suggest that species may compensate for adverse environmental conditions in the suboptimal macrohabitat by utilizing larger fungal clusters.As the measure of fungal cluster weight used here incorporates both the number and the weight of fruiting bodies, it is difficult to say much about the possible mechanisms that cause the observed difference in species incidence.The observed pattern may result from a higher colonization rate or from a longer ciid population persistence time in larger fungal clusters.Besides, larger clusters probably have greater variation in the quality of fruiting bodies, thus increasing the likelihood that a given beetle species meets its between fungal clusters.Jonsell et al. (1999) demonstrated that some ciid species were absent from fruiting bodies placed out in forest fragments 350-2,000 m from the natural forest.Similarly, Rukke (2000) found out that the incidence of ciid species in fruiting bodies was negatively affected by increased isolation at the scale of individual trees, 15 to over 500 m depending on the species.Given the high density of Trametes fruiting bodies it seems unlikely there would have been great difficulties for ciids in moving between clusters in clearcuts.However, restricted movement between clusters in forests and between adjacent forests and clearcuts is possible.

Large scale effects of macrohabitat
The forest management category had an independent effect on species frequency of occurrence, even less so than fungal-cluster weight.Similarly, the management category contributed positively to the abundance of S. affinis and C. hispidus in the clearcuts and O. glabriculus in the forests.In this study, it is difficult to explicitly distinguish between the effect of fungal cluster density and management category on the frequency of occurrence of the ciid beetles in the clusters, as the density was consistently higher in the clearcuts.The apparent inconsistent response of the two "forest species" to the management category may be due to opposing forces of microclimate and resource availability (Jonsell et al., 2001;Jonsson et al., 2001).However, a larger sample of forest and clearcut sites is needed to clarify whether such trade-off exists, or if the inconsistency results from inadequate sampling.
Microclimatic conditions are one of the most important abiotic differences between forests and clearcuts.Therefore, varying interspecific responses of ciid species to forest management may result from different microclimatic optima.Other studies have documented that increased sun-exposure and dryness of fruiting bodies increase the probability of occurrence of some fungus-dwelling insect species (Midtgaard et al., 1998;Rukke & Midtgaard, 1998), whilst other species occur more frequently in moist fruiting bodies or in shady conditions (Økland, 1996;Jonsell et al., 2001;Thunes et al., 2000).Sverdrup-Thygeson & Ims (2002) showed that among the beetle species in dead aspen there are clear preferences concerning the degree of sun-exposure.Their window-trap material also included O. glabriculus and, as in the present study, the species preferred shady conditions being more abundant in traps on shady aspen logs.Fossli & Andersen (1998) collected Trametes fruiting bodies from forests and, again, S. affinis was very rare.In Germany, S. affinis occurs readily in clearcuts, whereas O. glabriculus and C. boleti manage well in more shady conditions (Reibnitz, 1999).Microclimate can affect the abundance of ciids directly by speeding up the individual development and indirectly by affecting the quality of fruiting bodies.However, there are no studies linking population growth to the quality of fruiting bodies.

Conclusions
Many dead-wood dependent organisms can successfully occur in clearcuts if critical resources are left in adequate densities and qualities (Jonsell et al., 2001;Jonsson et al., 2001;Martikainen, 2001;Sverdrup-Thygeson & Ims, 2002).Prior to the extensive fire suppression in Finnish forests in the 1900s, many of these species may have favored open areas created by forest fires.Nevertheless, sweeping generalizations about species responses should be avoided even for common species, as demonstrated here.Despite higher Trametes density, clearcuts are more ephemeral environments for Trametes-dwelling insects in comparison with old-growth forests.The rationale is that most woody debris becomes unsuitable for Trametes over the course of years (3-7 years; Hintikka, 1993), after which it takes decades before new woody debris is available.Retaining green and dead deciduous trees in clear-cutting makes woody debris available for Trametes and many other more demanding species after the logging residues have become unsuitable.

Fig. 1 .
Fig. 1. A. The mean ± SE fungal density in the study sites (density is measured as the number of occupied pieces of woody debris ha -1 ); B. The mean ± SE fungal-cluster weight in the study sites, note that y-axis starts from 0.6: Black circles represent forests and open circles indicate clearcuts.

Fig. 2 .
Fig. 2. Percent of fungal clusters occupied by ciid species in the four study sites; black bars represent forests and grey bars show clearcuts.

Fig. 3 .
Fig. 3.The predicted probability of occurrence of the four ciid species in the fungal clusters as a function of cluster weight (untransformed), based on a logistic regression model.Figures give the estimated odds ratios and 95% CIs; black dots represent forests, grey dots clearcuts.

Fig. 4 .
Fig. 4. The mean ± SE population size per fungal cluster occupied by a given ciid species in the study sites; black circles represent forests and open circles clearcuts.

Table 2 .
Number of individuals and percent of samples occupied by ciid beetles in this study in forests and clearcuts.Tabla 2. Número de individuos y porcentaje de muestras ocupadas por escarabajos cíidos en este estudio, en los bosques y en los claros de tala.
a fixed effect nested within site.Fungal-cluster weight was included as a continuous covariate.For the distribution of a species (i.e. the presence or absence of a species in a cluster; a binary response) a binomial error distribution and a logit link-function were assumed.For the abundance of a species (the average number of individuals per cluster) an identity link-function and normally dis-

Table 3 .
GLM results on the fungal-cluster occupancy for four ciid beetles.The change in deviance indicates the model improvement when a given term is included in the model.

Table 4 .
Parameter estimates ± SE give the change in the log of the odds for a ciid species occurring or not occurring in a fungal cluster sampled from forests, holding fungal-cluster weight constant.Estimates are taken from the GLM in table 3: Mng.Management; W. Weight.(* Unstable estimate as this ciid species was absent from the forest in this site.)