Saproxylic beetle (Coleoptera) communities and forest management practices in coniferous stands in southwestern Nova Scotia, Canada

Abstract Old-growth forests in Nova Scotia typically exhibit an uneven-aged, multi-layered stand structure and contain signifi cant amounts of coarse woody debris. Many forest species, including invertebrates, depend in various ways on deadwood substrates. Th e objective of this study was to investigate relationships between forest stand age, silvicultural treatment, dead wood, and invertebrate biodiversity, using saproxylic beetles as an indicator group. Saproxylic beetle communities were also compared in the context of other studies in Nova Scotia. Beetles were gathered using four collection techniques: pitfall traps, funnel traps, sweep-netting, and manual searching. Results show that both stand age and harvest treatment had an eff ect on species richness and species composition. Younger stands had lower species richness and hosted a signifi cantly diff erent suite of species than medium-aged or older ones. Similarly, harvested stands had lower species richness and were host to a signifi cantly diff erent suite of species than unharvested stands. Th e results from the investigation of stand age are of particular interest. Forest management that disregards the dependence of diff erent suites of beetles on forest stands of various ages and compositions, emphasizing even-aged single-species stands, may be harmful to the species diversity of Nova Scotia’s forest ecosystems.


Introduction
Many indicators have been developed for monitoring forest health and human impacts on forest ecosystems (Noss 1999).Communities of invertebrates are especially useful for monitoring environmental change.Several attributes make them particularly useful indicators including, ease of collection, functional importance, high site specifi city, and known taxonomy (Langor and Spence 2006;Oliver and Beattie 1996a).A frequently suggested indicator is the diversity of saproxylic beetles.
Saproxylic beetles are a functional group of Coleoptera that depend, at some point in their life cycle, on dead or decaying wood or fungi associated with deadwood (Speight 1989;Økland et al. 1996).Not only do these insects comprise a large proportion of total forest species richness, but they also play an important role in decomposition and nutrient cycling in forest ecosystems (Siitonen 2001;Grove 2002b).Saproxylic beetles are considered pioneers as they are often the fi rst to colonize dead wood.Early colonization by wood-boring species is thought to precondition the wood for succeeding species (Hammond et al. 2001).Saproxylic beetles are, in large part, responsible for the mechanical breakdown of coarse woody debris (CWD) (Hickin 1963).Th ese beetles also demonstrate sensitivity to timber-harvest practices (Simila et al. 2002).
Nova Scotia forests have been subjected to a long history of human activity (Lynds 1989).Th is ranges from land clearing by early settlers, to forest "highgrading" between the 17th and 19th centuries, and fi nally clearcutting by the timber industry (Lynds 1989).Due to intensive management, 91% of the forested landscape is made up of young (less than 100 years) even-aged stands (Stewart et al. 2003).Few examples of really old forests with canopy trees of 250-300+ years old still exist.Although 73% of the land base is forested, no more than 0.6% of that land is comprised of old-growth forests (McMahon 1989;Loo and Ives 2003).
Nova Scotian old-growth forests (defi ned herein as being dominated by canopy trees over 120 years old) are typically comprised of uneven-aged, multi-layered stand structures and contain large quantities of CWD (Stewart et al. 2003).Th ompson (2004) conducted an analysis of the CWD in southwestern Nova Scotia and found that mean stand volumes of CWD were relatively higher in old stands as compared to young and mid-aged stands.CWD volumes were higher in partially-harvested stands than in the unharvested stands of the same age class, but only in young stands.
Although old growth stands are rare in the Acadian forest region, their contribution to biodiversity may be signifi cant (Loo and Ives 2003;Stewart et al. 2003;McMullin et al. 2008).Th e structural heterogeneity provided by CWD in forest ecosystems also gives rise to a wide range of ecological niches at the small-scale level.CWD and other deadwood materials in forests provide a multitude of habitats for numerous plant and animal species (Speight 1989;Franklin 1990;Grove 2002a).Although the study of saproxylic insects in the Maritime Provinces is relatively young, and old-growth forests have been little investigated in this regard, there are already preliminary indications that these same principals apply to forests in this region.Majka and Pollock (2006) reported the results of four studies of forest beetles that found between 54 and 76% of forest species were saproxylic.Majka (2007b) examined 14 families, subfamilies, and tribes of saproxylic beetles and found 59 apparently rare species that comprise 33% of the 178 species within these groups -a large proportion of the saproxylic fauna.Majka (2007b) proposed that this apparent scarcity might be due to the history of forest management practices in the region.Th ese preliminary indications of the importance of saproxylic beetles, the scarcity of many species, and the very low fraction of old-growth forests in the region, together suggest that the virtual disappearance of microhabitats found in old growth forests may have aff ected a substantial proportion of Nova Scotia's native saproxylic species.
Th ere are few studies of saproxylic invertebrates in eastern Canada compared to many areas of northern Europe (Jonsell et al. 1998;Kaila et al. 1997;Kolmstrom and Lumatjarvi 2000;Kouki et al. 2000;Martikainen 2001;Martikainen and Kouki 2003;Martikainen et al. 2000;Muona 1999;Økland et al. 1996;Rainius and Jansson 2002;Siitonen 1994Siitonen , 2001;;Simila et al. 2002Simila et al. , 2003;;Sverdrup-Th ygeson and Ims 2002;Vaisanen et al. 1993).Only in the last decade have serious investigations of the Nova Scotia saproxylic fauna been undertaken.Bishop (1998) and Bishop et al. (in press) highlighted the relationship between forest disturbance and saproxylic beetles.Th e relationship between forest disturbance, whether anthropogenic or not, and saproxylic beetle diversity is of growing importance.Th e principal goal of this study was to determine how saproxylic beetle communities vary with forest stand age and silvicultural treatment in mature conifer stands in southwestern Nova Scotia.Th ese relationships were studied indirectly through the relationship between dead wood and beetle diversity.Other objectives were to defi ne habitat for saproxylic species, mainly by qualitative and quantitative examination of CWD, and to contribute to a baseline understanding of the composition of forest-beetle communities in the Maritime Provinces.
All 11 study sites fell within the Atlantic Interior theme region (Davis and Browne 1996).A closer examination shows that within the Atlantic Interior region, the sites fall within three distinct units.Sites 1-4, 7, and 9-11 are found in the Lake Rossignol sub-unit of the Mersey Meadows unit within the Quarzite Plains district.Th e Lake Rossignol sub-unit is characterized generally by moderate to coarsely textured soils.Th e area around the south of Lake Rossignol itself supports eastern hemlock and red spruce, with some shade-tolerant hardwoods such as yellow birch.Culturally, hunting, fi shing, and canoeing have been practiced in this area, as the Mersey River was a traditional transport route for the Mi'kmaq and the French.In the 1920s, Lake Rossignol was fl ooded for hydro power for pulp-and-paper companies, thus aff ecting the biota and the hunting and fi shing of the Mi'kmaq.
Sites 6 and 8 are found in the Alma Lake sub-unit of the Annapolis Drumlins unit within the Drumlins district.Th e Alma Lake sub-unit is characterized by large granite boulders and well-drained soils.Although these drumlins are made up of granite materials, which are not typical, they do provide very productive forest sites.A mixed forest is common, and include eastern hemlock, red spruce, white pine, sugar maple, yellow birch, as well as some red maple.Tourism is and has been an important activity in this area beginning in the 1870s with American sportsmen who sought out hunting and angling experiences.
Site 5 is found in the South Mountain sub-unit of the Granite Uplands unit within the Granite district.Th e South Mountain sub-unit is characterized by uniform topographic features, including granite terrain and many large surface boulders.Th e soil is coarse, well-drained sandy loam, which is usually quite shallow.Characteristic forest trees are red spruce, eastern hemlock, white pine, balsam fi r, and red maple with occasional red oak.Fire has played a major role in this area, and regeneration patterns suggest that the area is turning into a coniferous-dominated forest.Forestry activities are the dominant land use in this largely uninhabited area, but mining and tourism are also prevalent.

Study design
Data collection was replicated in four young (40-80 years) stands, three middle-aged (80-120 years) stands and four old-growth (120+ years) stands.Stand age was determined by coring specimens of the dominant tree species.Within each age class, there were two replicates of each of two treatments, harvested and unharvested, with the exception of the middle-aged stand since it was not possible to fi nd two suitable harvested stands for this age class.Treatment for the harvested stands was by commercial thinning or by removal of approximately 30% of the canopy cover by uniform selection harvest and/or shelterwood harvest.All treated stands had been harvested 3-6 years prior to data collection.During the fi eld season, a thorough overstory analysis was completed for each stand.Measurements including age, species, overstory condition, and diameter at breast height (DBH).Relative density, relative dominance and relative frequency of tree species were then calculated to determine the importance value for each species (Roberts-Pichette and Gillespie 1999).
A quantitative analysis of CWD for each site was completed by Th ompson (2004).Th e sampling techniques used for measurement were adapted from the Nova Scotia Department of Natural Resources Forest Inventory Permanent Sample Plot Field Measurement Methods and Specifi cations (NSDNR 2002).In each of the sites, fi ve plots of 400 m 2 area were designated randomly for measurement.All snags in each plot were recorded by species, decay class, crown class (intact or broken), height (m) and DBH (cm).
Downed CWD was also measured by census, and for each piece species, diameter at mid-point and length were recorded.Each piece was assigned to one of fi ve decay classes ranging from freshly dead to thoroughly rotted [see Table 5 of Th ompson 2004, an adaptation of methods by Sollins (1982)].Total CWD volumes were calculated and averaged over the fi ve plots to obtain the mean volume per stand.

Beetle sampling methods
Between 15 May 2003 and11 August 2003, fi ve visits were made to each site approximately two weeks apart.In order to collect as wide a spectrum of species as possible, beetles were sampled using four collection techniques: pitfall traps, Lindgren traps, sweep-netting, and manual searching.Approximately four hours of search time was spent in each site for each of the fi ve visits between 08:00 and 16:00 hr.Timing of visits to the sites was rotational, therefore sites were visited at diff erent times of the day on each date.Collection did not take place during inclement weather, or on mornings following rough weather.An attempt was made to sample as many diff erent sizes and species of vegetation as possible to ensure that a maximum number of habitat types was examined.Sweeping covered both ground vegetation and tree branches within reach and was performed for approximately one hour per site.
For each site, fi ve pitfall traps were placed and visited fi ve times over the course of the fi eld season.Pitfall traps consisted of plastic cups buried in the ground and covered with a piece of CWD.Th e traps were fi lled 1/3 with a solution of one part ethylene glycol and two parts water.One Lindgren trap was set up at each site and was visited four times during the fi eld season.Th e eight-funnel Lindgren trap was suspended approximately 1.3 m off the ground in a relatively clear area within the site.Th e Lindgren traps were baited with a solution of half turpentine and half a 70% ethyl-alcohol solution.Th e collection jar at the bottom of the trap was fi lled 1/3 with water.Specimens were collected approximately every two weeks and were in good condition and not disarticulated so as to readily allow for identifi cation.

Beetle analysis methods
All the specimens collected in the study were identifi ed to species by C.G. Majka with the assistance of taxonomic experts (listed in the acknowledgements) with the exception of: a) fi ve species of Aleocharinae (Staphylinidae) (5 individuals) for which determination was possible only to the level of genus or tribe; and b) one species of in the genus Medon (Staphylinidae: Paederinae) which was identifi ed as M. nr.rufi penne (Casey, 1905) by Lee Herman.In the case of some beetles in the genera Cyphon (Scirtidae: 3 species) and Atomaria (Cryptophagidae: 1 species), the species found in the Maritime Provinces are currently under taxonomic revision (Klausnitzer and Majka, and Johnson and Majka respectively).Present determinations of these species may change as a result of these taxonomic revisions.Th e species of Paratenetus (Tenebrionidae) represents an undescribed species currently being described (P.Bouchard and Y. Bousquet, in preparation).Th e taxonomy and nomenclature employed follows Arnett and Th omas (2000) and Arnett et al. (2002).Th e collection of Coleoptera has been deposited with the Nova Scotia Museum (NSMC).
Th e data collected in the present study were additionally analyzed in relation to those gathered by Bishop (1998), a comparable study of the same trophic group of beetles in similar forest environments in Nova Scotia.Saproxylic species (sensu lato) were selected according to the criteria specifi ed by Speight (1989), namely species which depend, at some point in their life cycle, on dead or decaying wood, or fungi associated with deadwood, or on other saproxylics.Th e inclusion of species in this category was made on a specifi c or, more commonly, generic basis, consulting a wide variety of published sources (commencing with Arnett and Th omas (2000) and Arnett et al. (2002), followed by family-specifi c treatments such as Larochelle and Larivière (2003)), or if such information was not readily available in the literature, by consulting with appropriate specialists.
Inclusion in the saproxylic category is always accompanied by some degree of uncertainty, particularly in relation to groups or species that have been little investigated in terms of their bionomics (such as some Elateridae or Aleocharinae).Furthermore, some species, for example large forest-fl oor predators such as many Carabidae and Staphylininae, while frequently found in or under decomposing wood or logs, and therefore predaceous on other saproxylics, also range widely in the forest fl oor environment, and consequently may sporadically exit the saproxylic system.
Trophic categories were assigned to all species as a result of information derived from the aforementioned sources.Th e trophic categories employed were: Bolitophagous: feeding on the fruiting bodies of fungi (i.e., mushrooms); Mycetophagous: feeding on fungal hyphae (i.e., mold and mildew feeders); Myxomycophagous: feeding on slime molds; Phloeophagous: feeding on phloem of woody vegetation (i.e., cambium feeders); Phytophagous: feeding on leafy vegetation; Pollen Feeder: feeding on pollen of vascular plants; Predaceous: feeding on invertebrates; Predaceous/Nectarivorous: both predaceous and feeding on nectar; Rhizophagous: feeding on roots (i.e., plant material growing underground); Rhizophagous/Predaceous: both rhizophagous and predaceous; Sap Feeder: feeding on sap; Saprophagous: feeding on decomposing animal material; Saprophytic: feeding on decomposing plant material; Xylophagous: feeding on xylem of woody vegetation.Th e categories listed above are general and there are species of Coleoptera that a) overlap between two or more categories, particularly in regard to the diff erent bionomics of adults and larvae; and b) are insuffi ciently well studied for assignation with acceptable certainty to a particular category.For example, it is not always clear if detritivores are deriving nutrition from decomposing material (i.e., are saprophagous) or from fungal hyphae growing in such a matrix (i.e., are mycetophagous), or a combination of both.Nonetheless, such categorizations are a useful fi rst-order approximation to gain insight into the trophic structure of forest-beetle communities.

Beetle data analysis: statistical techniques
Th e sampling methods of this study emphasized alpha diversity (species richness).For the purpose of analysis, all species collected, no matter the quantity, were only counted as present or absent in any given site.A two-way analysis of variance (ANOVA) was performed based on species richness (sum of species) because it allows the investigation of interactions.An ANOVA was performed for both age and harvest treatment.To facilitate comparison with the present study, the data from Bishop (1998) (which included quantitative abundance information) were converted to presence/absence data.All subsequent data treatments were identical for both studies.
To analyse the diff erences in species composition between sites, a SIMPER (similarity percentages) test (Clarke and Warwick 2001) was completed.Th is test was used to determine which species, from all dates in each site, contributed most to the dissimilarity between sites; those species that provided the strongest discrimination between two sites.Th e species selected for the modifi ed list were those whose similarity/ dissimilarity, divided by the standard deviation, had a value greater than or equal to 1.All subsequent tests were performed using the modifi ed species list.
Th e presence/absence of all species over all dates in each site was used to calculate the Bray-Curtis similarity measure.A one-way analysis of similarities (ANOSIM) between samples was performed for the factors of age and treatment within each site using the Bray-Curtis similarity measure (Clarke and Warwick 2001).Given the qualitative, presence/absence nature of the data collected, the SIMPER analysis was most applicable.ANOSIM was used for variables such as species richness and trophic category richness (trophic composition).Multi-dimensional scaling (MDS) plots were completed for each of the variables.Th e MDS plots give a two-dimensional ordination, illustrating relationships between sites.
A test of taxonomic distinctiveness (TAXD) was performed using the following taxonomic levels: species, genus, tribe, subfamily, family, superfamily, series, and suborder.Th e TAXD measures biodiversity based on the relatedness of species within a sample, thus adding additional statistical sampling properties (Clarke and Warwick 2001).
No standardisation or transformation was performed before analysis.Several qualitative comparisons were made between the two studies, including total number of species, diff erences in species origin, and trophic categories.A quantitative comparison of species richness between the two studies with diff ering species and specimen numbers was possible using the EstimateS rarefaction curve (Colwell 2004).

Relationships between beetle communities and CWD
Coarse woody debris volumes were calculated by Th ompson (2004) using only pieces where diameter > 7.0 cm.Stand volumes were analyzed with respect to stand age and harvest treatment using a general linear model (GLM) in SPSS 11.5 (SPSS 2002).Th e eff ects of age, treatment, and age and treatment combined were tested for their infl uence on the volume of CWD in each stand.Decay class and CWD diameter across age groups and harvest treatment were plotted to look for trends in the data.
Th e multivariate patterns arising from the CWD data were compared to those of beetle species data to determine the extent to which CWD aff ects beetle species diversity.Both an analysis of covariance (ANCOVA) and a correlation analysis were used to test whether CWD volumes were correlated with beetle species richness (Clarke and Warwick 2001).
CWD volume data were then used to assign each stand to one of three volume classes: 1-55 m 3 /ha (low), 56-110 m 3 /ha (medium) and 111-165 m 3 /ha (high).Th e CWD volume classes were paired with beetle data.CWD was qualifi ed as a factor and analyzed using ANOSIM.

Overview of stand data
Th e eleven study sites were mainly dominated by red spruce, white pine, and/or eastern hemlock (Fig. 2), ranging in age from 49 years to approximately 160 years (Table 1).Minimum mean diameter at breast height (DBH) was 10.5 cm at site 4 and maximum mean DBH was 24.6 cm at site 10.Th e mean DBH of dominant species ranged from 10.8 cm for the red spruce, white pine, and balsam fi r in site 4, to 28.4 cm for the eastern hemlock and red spruce in site 10.Stand density ranged from 510 stems/ha in site 9 to 4,320 stems/ha in site 4.
Using data collected by Th ompson ( 2004), volume of CWD for each decay class was plotted for each of the 11 forest stands.Volumes of CWD varied from > 160 m3 / ha to < 40 m 3 /ha.CWD total volumes were clearly higher in the oldest stands sampled (Fig. 3).Tree species composition also varied across the sites (Fig. 4), but with no clear pattern associated with stand age or treatment.Site 5 in particular had a large proportion of red maple, and site 8 had a large proportion of eastern hemlock.Sites 1, 2, 4, and 7 have no white birch CWD.

Overview of beetle data
Beetle sampling yielded 2,302 specimens of 346 species from 56 families (Appendix 1).
Of the 346 beetle species, 264 were determined to be either facultative or obligatory saproxylics.Fifteen species (4%) were common to all young stands, 25 species (7%) were common to all medium-aged stands, and 16 species (5%) were common to all old stands.Eleven species (3%) were common to all harvested stands and seven species (2%) were common to all unharvested stands.Fifty-one species were caught in only one of the 11 sampling sites.Only three species (Isomira quadristriata (Couper), Platydracus viridanus (Horn) and Rhagonycha mandibularis (Kirby)) were common to all 11 sites.Results from the SIMPER analysis identifi ed 97 beetles primarily accounting for species assemblage similarities and dissimilarities for factors of age and treatment.Th ese species were used for all subsequent analyses.In comparison, 387 species were collected by Bishop (1998), 76 of which were primarily responsible for species assemblage similarities and dissimilarities and were used in subsequent analysis.

Comparative analysis of beetle communities
Both the present study and Bishop (1998) found relatively similar numbers of beetle species (346 and 387 respectively).However, Bishop (1998) collected over four times the number of specimens collected in the present study (9,881 vs. 2,301 respectively).Th e larger number of individuals in Bishop's (1998) study refl ects a greater sampling eff ort (180 fl ight intercept traps which were continuously in place for 90 days during the spring/summer fi eld season).Similar numbers of beetle species were considered facultative or obligatory saproxylics (264 species in the present study; 297 in Bishop (1998)).One hundred and sixty-four species were common to both studies.Results from a rarefaction analysis (Fig. 5) indicate that, if similar numbers of specimens were collected in the Bishop (1998) study, the number of beetle species would not have been as high as in the present study (Fig. 5).It is also clear that the number of species in the present study has not yet approached an asymptote (Fig. 5).Th is indicates that the four collecting techniques we employed are sampling a much wider spectrum of the forest beetle community than the single technique employed by Bishop (1998).
For example the present found more species of forest fl oor beetles in the families Carabidae, Tenebrionidae, and Histeridae than did Bishop (1998) (Fig. 6).Although some of these species are macropterous and are capable of fl ight, it would appear that many of them fl y infrequently.Th e larger numbers of Curculionidae (Fig. 6) in the present study consist almost entirely of fl ightless species in the subfamily Entiminae, which are not sampled at all by fl ight intercept traps.Th e present study also collected substantial numbers of specimens in the Geotrupidae (Geotrupes horni Blanchard, a forest fl oor species) and Zopheridae (Phellopsis obcordata (Kirby), a largely fl ightless bolitophagous species found on polypore fungi) (Appendix 1), two families not represented in the material collected by Bishop (1998).A sizeable number of species were restricted to stands of specifi c ages: 47 were exclusively found in young stands, 52 species were found only in middle-aged stands and 69 species were exclusively found in old stands.Th ese 69 species are listed in Table 2 along with indications of their abundance and distribution within Nova Scotia.While such categorizations admittedly include a degree of subjective judgment, and are based on a continually evolving base of knowledge of the Coleoptera in the province, they do allow a mechanism for selecting potentially vulnerable species.Th ey include 23 rarely collected and locally distributed species in Nova Scotia (indicated in boldface in Table 2) that may be indicator species of old-growth forests conditions.

Eff ects of stand age
More species were present in middle-aged and old stands than in young stands (F = 22.511; d.f.= 2; p = 0.003).Signifi cantly more species were present in middle-aged and old stands than in young stands (Fig. 7).Species composition was signifi cantly aff ected by stand age.Th e Global R, ANOSIM was 0.442 with a signifi cance level of 0.7% (the sample statistic is similar to, yet not synonymous with, a p-value of 0.007).Both the young and mid-aged stands were signifi cantly diff erent from the old stands, but there was no signifi cant diff erence between the young and mid-aged stands.Stand age had a pronounced eff ect on species composition.
Trophic composition was also aff ected by stand age (Global R, ANOSIM of 0.503 with a signifi cance level of 0.1%).Th ere were signifi cant diff erences in the trophic composition of beetle communities between young and both mid-aged and old groups, but not between mid-aged and old groups.
Th e TAXD test did not show any signifi cant results for age, therefore indicating that there were no diff erences detected in the taxonomic distinctiveness of species between age groups of stands.

Comparative eff ects of stand age
Species richness was aff ected by stand age in this study only.Young unharvested stands had the lowest species richness, and mid-aged harvested stands the highest.All other stand age-treatment combinations had similar species richness.Stand age had no signifi cant eff ect (F = 1.632431; d.f.= 2; p = 0.217) on species richness in Bishop (1998).Species composition and trophic composition were signifi cantly aff ected by stand age only in the present study.

Eff ects of harvest treatment
Species richness was aff ected by treatment type.Th e mean number of species present in harvested sites was signifi cantly higher than in unharvested sites (F = 13.395;d.f.= 1; p = 0.015) (Fig. 8).Th e interaction between harvest treatment and age was not signifi cant.Species composition was signifi cantly aff ected by harvest treatment (Global R, ANOSIM of 0.299 with a signifi cance level of 3.7%).Treatment did have a pronounced eff ect on species composition.
Species trophic composition was not signifi cantly aff ected by harvest treatment (Global R, ANOSIM of 0.155 with a signifi cance level of 13.4%).Th e diff erence was not statistically signifi cant.Th e TAXD test did not show any results for harvest treatment, thereby indicating that there was no diff erence in the taxonomic distinctiveness of species between harvested and unharvested sites.

Comparative eff ects of harvest treatment
Species richness was aff ected by treatment type in this study only.Harvest treatment had no eff ect on species richness in Bishop (1998).Species composition was significantly aff ected by harvest treatment in Bishop (1998) (Global R, ANOSIM of 0.254 with a signifi cance level of 0.1%).Treatment thus had a pronounced eff ect on species composition.Species trophic composition was not signifi cantly aff ected by harvest treatment in Bishop (1998).

Analysis of relationships
Th ere is a weak positive relationship between volume of CWD and species richness (Fig. 9).An increase in CWD volume is accompanied by an increase in beetle species richness, but the relationship is not statistically signifi cant (F = 2.48; d.f.= 9; p = 0.15).
Species composition was not aff ected by volume of CWD.Th is is true even with the use of the modifi ed species list developed through the SIMPER analysis.Th e Global R, ANOSIM value was 0.234 (p = 0.086).Th e greatest diff erence was seen between the group with the lowest volume of CWD (1-55 m3/ha) and the group with the highest volume of CWD (111-165 m3/ha) (R = 0.75, p = 0.10).

Beetle Sampling Methods
Th e methods used to collect beetles in the present study were quite diff erent from those employed by Bishop (1998).Flight intercept traps (FITs), employed by the latter study, were more easily standardized.Th e use of FITs is much less laborious in the fi eld compared to manual searching (Siitonen 1994).As well, there is ongoing collection even when the researcher is not present.Th ere are several shortcomings associated with the use of FITs or other passive collection methods.Some species or even families of beetles live inside decaying trunks for several generations and would therefore not likely be caught in fl ight (Siitonen 1994).Given the common trend of abundance of large, poorly dispersing specialists to decrease with increasing disturbance, and like- wise for small generalists to increase in abundance, the bias of FITs may be signifi cant (Rainio and Niemela 2003).If the presence of specialists goes undetected in disturbed forest ecosystems, an incorrect picture of beetle diversity would emerge for those sites.Additionally, of course, fl ightless species are not collected at all.
In studies conducted in the Oulanka National Park region of Finland (an area whose beetle fauna is very thoroughly documented) Muona (1999) found that FITs caught 44-48.3% of forest beetle species.A combination of pitfall traps and FITs caught 60.5% of species, and pitfall traps, FITs, and sweep nets taken together collected 91.4% of species.Only 55% of rare species were caught and only 25% of species designated as threatened were collected.Muona (1999) pointed out that the populations of many rare species are small and they may be patchily distributed, making them diffi cult to detect and sample.Martikainen and Kouki (2003) found that window traps were the most eff ective trap type in sampling threatened beetles.However, rarer species were not collected using any other method besides direct searching.Direct searching includes netting, peeling of bark, and other searching methods for beetles by entomologists themselves.As well, the effi cacy of the window traps depended largely on the location of the trap, and those located randomly were not as eff ective.
Manual searching is a more time-intensive process than passive methods like FITs, but the time saved by the latter approach has to be balanced against the time involved in subsequent sorting, pinning, and identifi cation.In the Bishop (1998) study, 4.3 times the number of specimens had to be processed.In designing sampling programs for environmental impact assessment or ecological monitoring in a fi scal climate when both human and fi nancial resources may be limited, such considerations may also have an important bearing on decision-making (Oliver and Beattie 1996b).
Although alpha diversity (species richness) carries less information than quantitative species abundance, it can be less time-and labour-intensive to collect and process such information.Competitive exclusion (sensu Hardin 1960) supports the view that measuring alpha diversity is indicative of the presence or absence of microhabitats occupied by respective species.Hence, examining species richness, particularly of hyper-diverse groups such as Coleoptera, allows for an examination of some dimensions of the environment as perceived through a fi ne ecological mesh (Majka and Bondrup-Nielsen 2006).
Although the number of specimens collected varied considerably between the two studies (2,302 specimens in the present study, and 9,881 by Bishop (1998)), the total number of beetle species was rather similar (346 versus 387, respectively).Rarefaction analysis (Fig. 5) indicates that the expected number of species collected in the present study would be much lower.Th is higher-than-expected species richness may be a function of using four collection methods as compared with one by Bishop (1998).It is possible that the combination of four collection methods was effi cient in sampling many microhabitats and avoiding the collection of high numbers of specimens of the same species.It has been suggested that to maximally sample the species present in a particular environment, several trapping methods should be employed (Økland 1996;Ranius and Jansson 2002;Martikainen and Kouki 2003).

Beetle communities
Beetle communities were similar in both studies in terms of species richness, and proportions of species from diff erent zoogeographic origins (Nearctic, Holarctic, adventive Palaearctic).Th ese similarities support the contention that both approaches are sound with respect to producing accurate (albeit incomplete) representations of forest beetle communities.Not all groups of beetles are thoroughly sampled by either of these approaches.For instance, both studies recorded few species of Buprestidae, even though 39 species of these saproxylic beetles have been recorded in Nova Scotia (C.G.Majka, unpublished data), and some forest-litter species such as those in the Ptiliidae were poorly represented in both studies since litter sifting was not a technique employed in either.Surprisingly, macropterous, litter-dwelling species in the Pselaphinae and Scydmaenidae were well represented in both studies (collected by fl ight-intercept and funnel traps) (Appendix 1).
Furthermore the combination of techniques employed in the present study better samples forest fl oor species in the Carabidae, Tenebrionidae, Entiminae, Geotrupidae, and Zopheridae than fl ight-intercept traps do (Fig. 6).Nevertheless, these approaches, while off ering an incomplete representation of forest beetle communities, do provide an apparently consistent index of these communities, something of utility in terms of comparing forests of diff erent ages and compositions, and also in monitoring changes in forest communities over time.Martikainen and Kouki (2003) suggested that the measurement of the number of species has several advantages compared to other estimates.Advantages include availability and/or ease of measurement and consistency over large geographic areas.Th ey also indicated that the number of trapped species should be over 300-400, corresponding to a number greater than 4,000 individuals, in order to include rare species (Martikainen and Kouki 2003).
Th e proportion of introduced species of beetles in an environment can frequently be an indicator of disturbance, with anthropogenic or disturbed habitats typically exhibiting much higher proportions of adventive species (Spence and Spence 1988).In the present study, 22 of 346 saproxylic species (6.4%) were introduced, while in Bishop's (1998) work, 20 of 387 species (5.2%) were introduced.In contrast, in a study of Coleoptera of Point Pleasant Park, Nova Scotia, a highly disturbed early successional forest, C.G. Majka found 17.7% of species to be introduced (unpublished data).Th is compares to the Nova Scotia provincial fauna as a whole which consists of 15.3% of introduced species (C.G.Majka, unpublished data).
Th e number of species previously unrecorded for Nova Scotia in both studies indicates that knowledge of the baseline biodiversity of saproxylic beetles in Nova Scotia is far from complete.Th e present study found 135 species of beetles unrecorded from Nova Scotia by Bousquet (1991).Both studies thus contribute to ongoing programs to document the distribution, abundance, origins, and trophic categories of the beetle fauna of the province which are essential for many ecological, zoogeographic, and environmental monitoring studies (McCorquodale et al. 2005).Some specimens derived from this study have already contributed to recent surveys of the Coleoptera fauna of the region (Bousquet and Laplante 2006;Majka 2006aMajka , 2006bMajka , 2007aMajka , 2007bMajka , 2007cMajka , 2008a;;Majka andCline 2006a, 2006b;Majka and Johnson 2008;Majka and Pollock 2006;Majka et al. 2006Majka et al. , 2007aMajka et al. , 2007a, in press), in press).

Stand age
In the present study, stand age was shown to aff ect both species richness as well as species composition.Th is agrees with other studies in both Canada and Europe (Kaila et al. 1997, Hammond et al. 2004).In general, as stand age increases, so does species richness.Th is high level of beetle species richness in old forests is likely related to habitat heterogeneity, often characteristic of old-growth forests (Heinrichs 1983).Old-growth forests typically are more structurally diverse, and have higher concentrations of large-diameter CWD, therefore supporting a wider array of microhabitats suitable for a larger diversity of species (Duchesne 1994).In both the present study and in Bishop (1998) old stands had a lower species richness then medium-aged stands (Fig. 7).Indeed in Bishop (1998) the old stands had a slightly lower species richness than young stands (Fig. 7).Th ese results are counterintuitive.Further investigations should be undertaken in similar northern-temperate forests with extensive disturbance histories to ascertain if the present results are anomalous, or if there are reasons why the species-richness in older forest stands is being underestimated and inadequately sampled by present collection techniques (see the further discussion on this subject in the section below).
Species assemblages of beetles of younger stands do not appear to be entirely similar to those present in older forest stands.Not only were species compositions diff erent, but the trophic category compositions also varied slightly between young and old stands.Th is may be due to diff erences in decay class and diameter class of the CWD, and therefore to diff erences in food and habitat availability present in the diff erentaged forest stands.A study by Hammond et al. (2004) saw an increase in trophic complexity in older stands.

Harvest treatment
Harvest treatment was shown to aff ect both beetle species richness and composition.Th e results from this study diverge from many other studies in that species diversity was higher in the partially harvested sites than in the unharvested sites.Kaila et al. (1997) found that median numbers of species caught did not diff er between closed forests and clearcuts, but species assemblages varied greatly.In other European studies, such as those by Økland et al. (1996) and Simila et al. (2003), species richness was signifi cantly higher in semi-natural or unharvested forests than in managed forests.Managed forests in the Scandinavian setting are, however, quite diff erent than those in Canada (Kimmins 1997).However, a Canadian study by Klimaszewski et al. (2003) measured arthropod abundance and found that clearcut and thinned forest stands have lower beetle abundance than unmanaged stands.
Sverdrup-Th ygeson and Ims (2002) suggested that one explanation for higherthan-expected species richness in harvested sites is that there is a possible bias towards collecting higher numbers of individuals in cleared, sun-exposed sites.Beetles tend to be more active and evident in warmer temperatures.Th ere is also the probable preference of saproxylic beetles for sun-exposed CWD, in that both visual cues and wind dispersal favour more open areas for searching out CWD (Jonsell et al. 1998;Kouki et al. 2001;Martikainen 2001).Another possible explanation is that there appears to be a peak in abundance and richness of some families of beetles approximately fi ve years after a disturbance to the forest ecosystem (Wermelinger et al. 2002).In one study, bird species richness also showed a sharp increase 2-6 years after clearcut harvest (Keller et al. 2002).Conversely, it may be that conventional wisdom of equating higher species diversity with older-aged stands requires some refi nement, particularly in northerntemperate and boreal forests with extensive disturbance histories.
In the present study, beetle species composition was signifi cantly diff erent between harvested and unharvested sites.Th ere were variations in tree species as well as general stand characteristics between diff erent sites.For example, site 5 had a considerably larger quantity of red maple CWD and much thinner overstory than any other site.Th e results show that site 5 is responsible for much of the variation among the harvested sites.It is also possible that the diff erences in species assemblages were related to the proportions of decay classes or sizes of CWD present in harvested versus unharvested sites.
Although freshly killed wood has a lower diversity of habitats for saproxylic beetles than old dead wood, there is a specifi c trophic group (phloeophagous species) of beetles that feed on the former (Hammond et al. 2004).Kruys and Jonsson (1999) found that fi ne woody debris was important for species richness of particular taxa of cryptogams in managed boreal spruce forests in northern Sweden.Th ese two aspects could account for the signifi cant diff erence in species composition between harvested and unharvested sites.Th ey may also be related to diff erences (although in the present study this measure is not statistically signifi cant) in trophic categories present in harvested and unharvested sites.
Despite a higher number of beetle species in the harvested sites of the present study, both studies showed statistically signifi cant diff erences in species composition of harvested versus unharvested sites.Many studies have indicated that invertebrates are often more sensitive to environmental change than vascular plants or vertebrates and will therefore respond more quickly to such changes (Rosenberg et al. 1986;Erhardt and Th omas 1991;Moore 1991;Ehrlich 1992;Kremen et al. 1993).In the case of the two present studies, the response to environmental change was more obvious in the diff erences in species composition.

Analysis of relationships between CWD and beetle communities
A positive correlation between mean volume of CWD and beetle species richness would not have been new or unexpected (Martikainen et al. 2000;Yee et al. 2001;Grove 2002a).Th is relationship is often highlighted because forest practices have "resulted in a progressive simplifi cation of stand structure and a loss of mature timber habitat" (Grove 2002b).In this study, however, the relationship was not statistically signifi cant.Th is may be due to sample size, as both CWD volumes and beetle species richness were signifi cant when measured individually for eff ects of stand age.However, as noted above these relationships may also require some refi nement in the context of northern-temperate and boreal forests with extensive disturbance histories.
Many other studies have shown a positive correlation between dead trees of large diameter and high numbers of beetle species, including many rare species (Vaisanen et al. 1993;Kolstrom and Lumatjarvi 2000;Siitonen 2001).In Great Britain, Alexander (2004) identifi ed 180 saproxylic species (of a total of 694 species in Great Britain) that are indicators of ecological continuity (an inverse of disturbance) and hence are characteristic of undisturbed forests.One hundred and sixty-one of these are variously listed as endangered, vulnerable, rare, or scarce in Great Britain.Indeed, of the full 694 species, 354 species (51%) are in some measure endangered, vulnerable, rare, or scarce (Alexander 2004).
In Canada comparatively less attention has been paid to saproxylic fauna and so it is often diffi cult to distinguish between genuinely rare species, and those that have simply been rarely collected.Majka (2007b) compiled a list of 59 "apparently rare" species of saproxylic beetles (defi ned as those species constituting ≤ 0.005% of specimens examined from the region) in the Maritime Provinces of Canada from 14 families, subfamilies and tribes of Coleoptera.Th ese 59 represented 33% of the 178 species found in the region within these groups.Majka (2006aMajka ( , 2006bMajka ( , 2007) ) and Majka and Pollock (2006) have all proposed that this apparent scarcity may be due to the history of forest management practices in the region that have greatly diminished old-growth habitatpractices which have dramatically decreased the amount of large diameter CWD.
In the present study a sizeable number of species were restricted to specifi c stand ages: 47 species were exclusively found in young stands, 52 species were only found in middle-aged stands and 69 species were exclusively found in old stands.In the latter category, 23 of these 69 species are rarely collected and locally distributed species in Nova Scotia (Table 2).Th ese are candidates for species associated with old-growth forest conditions.Nineteen of them are saproxylic species.In relation to species such as Stenichus badipes (Casey), Bibloporus bicanalis (Casey), Batrisodes lineaticollis (Aubé), Quedius densiventris (Casey), Ampedus protervus (LeConte), Dorcatoma falli White, Rhizophagus remotus LeConte, Silvanus bidentatus (Fabricius), Mycetophagus serrulatus Casey, and Mordellistena fuscipennis (Melsheimer), it is noteworthy that Alexander (2004) identifi es closely related congeners (or in the case of S. bidentatus, the same species) in Great Britain as indicator species of ecological continuity, in other words species of beetles associated with old-growth forest conditions.
In several previous studies, it could not be discerned whether stand age or largediameter CWD was the explanation for higher diversity in older forest stands (Økland et al. 1996;Hammond et al. 2004).Simila et al. (2003) emphasized the importance of considering the eff ects of diversity and volume of dead wood on survival and richness of saproxylic species.
In the present study, based on mean volumes of CWD for each stand, there appeared to be no signifi cant diff erences in beetle species composition.It is possible, however, that an examination of decay classes and diameter sizes of CWD would have a diff erent eff ect.As CWD passes through diff erent stages of decay, it is colonized by a succession of diff erent beetle assemblages (Speight 1989;Grove 2002a).Beetle communities depend on an array of factors including stage of decomposition, tree species, and type of rot (Grove 2002a).

Management implications
In Canada and the United States, intensive silvicultural treatment has not been as complete as in most parts of Europe (Kimmins 1997).Th e North American forest industry may not have aff ected the diversity of saproxylic beetles within their forests to the same extent as has occurred in many European countries, nevertheless, ongoing anthropogenic disturbances to Nova Scotian forest ecosystems have the potential for signifi cant impact on saproxylic insect communities.Commercial thinnings in softwood stands in south western Nova Scotia appear, however, not to have the pronounced negative eff ects that have been demonstrated in clearcuts elsewhere.
Particularly careful attention was paid in this study to species determinations.Informed conservation and management strategies must be based in accurate speciesand population-based data.Goldstein (1999) argued forcefully that any ecosystem approach that decouples species-and population-specifi c requirements from management strategies, risks compromising fundamental conservation objectives.Furthermore, as Wheeler (1995) pointed out, accurate taxonomic work is indispensable to conservation decisions."'Fast and dirty' taxonomy will not remove the taxonomic roadblock; it will simply add to it" (Wheeler 1995).
In terms of the conservation of beetle species diversity, the results of this study agree with those of Kaila et al. (1997).To conserve the highest proportion of the saproxylic beetle fauna, we should maintain a variety of habitats including both young and old forest stands.For example, in this study, 15 species (4%) were common to all young stands, 25 species (7%) were common to all medium-aged stands, and 16 species (5%) were common to all old stands.Th e 23 rarely collected and locally distributed species found in old-growth forests in the present study may be representative of a suite of beetles whose success and survival may depend on the presence of old-growth forests.Some of the 59 species of saproxylic beetles identifi ed by Majka (2007b) may represent species already reduced to "relict" populations in the region by the long history of forest management practices.
Management programs that ignore old-growth forests and the suite of insects dependant on them may impoverish the species diversity of Nova Scotia's forest ecosystems.Considering that saproxylic beetles, as important mechanical wood decomposers, are responsible for substantial amount nutrient cycling and decomposition in forest ecosystems, loss of this diversity may have adverse consequences.For instance, when both saproxylic invertebrates and fungi are present, in northern climates the decomposition phase of wood is in excess of 25 years (Ehnström 1979).An absence of saproxylic invertebrates causes the decomposition phase of wood (mediated solely by fungi) to be doubled in length (Dajoz 1980;Speight 1989).As a result, general forest health and sustainability of forest use, including product extraction, may be at risk.

Further research
Globally, patterns of biodiversity remain poorly documented (Mittermeier et al. 1999).Many studies have used taxonomic groups with large-bodied species, such as birds, mammals, and vascular plants, to infer general patterns (Myers 1988(Myers , 1990;;Myers et al. 2000).However, these groups make up only a tiny fraction of the world's species richness and although the estimates obtained are useful, it is doubtful that these groups truly refl ect the species richness of other groups, which are less well-studied but far more species-rich (May 1988;Heywood 1995;Lawton et al. 1998).
Th ere is an unquestionable need for lists of indicator species of forest health, especially as the demand for forest products increases and silviculture becomes more intense.Nilsson et al. (1995) suggested that forest continuity has important implications for forest condition.Forest continuity, however, is a property that may be diffi cult to construe from present forest structure (Nilsson et al. 1995).
Th e role of saproxylic beetles in forest ecosystems, the need for their conservation, and their possible use as indicators, are well documented in Europe (Speight 1989, Wermelinger et al. 2002Simila et al. 2003;Alexander 2004).However, studies of saproxylic beetles in the Acadian forests (Kehler et al. 2004;Bishop 1998)   Notes: Families, subfamilies, and species are listed in taxonomic order.†, adventive Palaearctic species; * Holarctic species; no symbol, Nearctic species.

Fig. 5 .
Fig. 5. Rarefaction curve demonstrating projected species richness for number of individuals based on Bishop (1998), and the present study (Dollin et al.) beetle collections.

Fig. 6 .
Fig.6.Th e number of species of beetles in the 22 most speciose families collected in the present study and inBishop (1998).

Fig. 7 .
Fig. 7. Mean species richness of beetles in diff erent forest stand age classes, including standard deviation from both the present study and Bishop (1998).

Fig. 8 .
Fig. 8. Mean species richness across harvest treatment, including standard deviation, for 11 stands in southwestern Nova Scotia.

Fig. 9 .
Fig. 9. Scatter plot demonstrating a non-signifi cant positive correlation between beetle species richness and CWD volume.Included are the regression equation and the R 2 value.

Table 1 .
Summary of characteristics of eleven study sites on Bowater land in southwestern Nova Scotia including mean age, mean height, mean diameter at breast height (DBH), dominant species and density.

Harvest treat- ment 1 Dominant overstory species 2 Age class (years) Mean Age (years) 3 Total Basal Area (m 2 /ha)
are still very few in number.todes), Donald S. Chandler (University of New Hampshire), Joyce Cook (Carleton University), Lee Herman (American Museum of Natural History), Jan Klimaszewski (Laurentian Forestry Centre), David McCorquodale (Cape Breton University), Sean O'Keefe (Morehead State University), Darren Pollock (Eastern New Mexico University), Wolfgang Rücker (Latridiidae.com),Quentin Wheeler (Arizona State University), and Adriano Zanetti (Museo Civico di Storia Naturale) for assistance with Coleoptera determinations.Th anks to Andrew Hebda of the Nova Scotia Museum for the use of many services and facilities.Jan Klimaszewski and David McCorquodale read earlier versions of this manuscript and made many constructive suggestions.And fi nally, sincere thanks to Delancy Bishop for the generous use of his data and unpublished study results.