University of Alberta

Retention Patch Characteristics and Ground Dwelling Beetle Diversity: Implications for Natural Disturbance-Based Management

by

Matthew Paull Pyper \Sv

A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfillment of the requirements for the degree of

Master of Science in Forest Biology and Management

Department of Renewable Resources

Edmonton, Alberta Spring 2009 Library and Archives Bibliotheque et 1*1 Canada Archives Canada Published Heritage Direction du Branch Patrimoine de I'edition

395 Wellington Street 395, rue Wellington Ottawa ON K1A 0N4 OttawaONK1A0N4 Canada Canada

Your file Votre reference ISBN: 978-0-494-54998-8 Our file Notre reference ISBN: 978-0-494-54998-8

NOTICE: AVIS:

The author has granted a non­ L'auteur a accorde une licence non exclusive exclusive license allowing Library and permettant a la Bibliotheque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par telecommunication ou par I'lnternet, preter, telecommunication or on the Internet, distribuer et vendre des theses partout dans le loan, distribute and sell theses monde, a des fins commerciales ou autres, sur worldwide, for commercial or non­ support microforme, papier, electronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats.

The author retains copyright L'auteur conserve la propriete du droit d'auteur ownership and moral rights in this et des droits moraux qui protege cette these. Ni thesis. Neither the thesis nor la these ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent etre imprimes ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author's permission.

In compliance with the Canadian Conformement a la loi canadienne sur la Privacy Act some supporting forms protection de la vie privee, quelques may have been removed from this formulaires secondaires ont ete enleves de thesis. cette these.

While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n'y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.

•+• Canada To my wife, and best friend Shelagh Marie Pyper, for supporting me in everything I do. Your love and patience has meant so much. Abstract

Scientific evaluation of natural disturbance-based forest management requires fine-scale studies

of biodiversity. I studied the role of retention patch size and isolation from the un-cut forest

matrix in preserving communities of ground beetles (Coleoptera: Carabidae) and rove beetles

(Coleoptera: Staphylinidae) in North-western Alberta. Patch size was an important driver of beetle community structure, with beetle assemblages in patches greater than two hectares most

similar to the mature forest. Patch isolation had little effect on common species but resulted in

decreased catches of rare species, emphasizing the need to assess rare species in forest management research. I also studied edge effects along forest-harvest edges and found narrow,

abrupt edge effects. Beetle assemblages of edges began to recover by 15 years post-harvest in

deciduous stands. My work suggests that increasing the frequency of large retention patches in

managed forests has clear benefits for biodiversity, and that large patches can be used to reduce

the negative influences of isolation. Acknowledgements

First and foremost I would like to thank my supervisors, Dr. John Spence and Dr. David Langor. To Dave, for his encouragement, unique perspectives, and extensive support during initiation of this project. To John, for taking a wildlife enthusiast under his wings, providing endless opportunities, and mentoring my passion for science communication; for this, I am truly grateful. I also thank my committee members Dr. Fangliang He and Dr. Heather Proctor for comments, advice, and encouragement throughout this project; as well as members of the examining committee. I have also been fortunate to have encountered many amazing people along the road to completing this thesis. First, to my field assistants: Tomona Morita, and Wallis Johnson, for showing grit and determination. To Wallis, I hope you too develop an appreciation of our amazing wolf encounter. To Jim Witiw, for the endless hours spent helping with field site selection, and continual interaction during this project. To the Spence lab (C.Wood, J. Pinzon, T. Cobb, C. Bergeron, J. Jacobs, E. Esch, S. Lee, S. Abele, S. Bourassa, E. Kamunya, M. Koivula and C. MacQuarrie) for influential discussions, statistical advice, daily humour, constructive criticism, and lifelong friendships; it has been a good time, but not a long enough time. I also thank L. Gray and J. Fitzpatrick for spurts of field assistance, and Brett Bodeux for his enthusiasm and ideas during countless conversations. To Dustin Hartley, for the rove beetle identifications in this thesis, your help was invaluable. Jason Edwards and Charlene Halm ran the best field station a student could ask for, thanks for your discussions and support. Special thanks to Shelagh Pyper, for the many hours committed to this thesis; whether in the wee hours of the night or on trains in Bolivia, your help was unwavering. To my parents, Brian and Kathy Pyper, for instilling in me a passion for hard work, a love of the outdoors, and a desire to make a difference. To Andrew Murphy and Susan Crump for support, encouragement, and advice throughout this thesis. Funding for this work was provided by the Sustainable Forest Management Network- Networks for Centres of Excellence, ACA Grants in Biodiversity- supported by the Alberta Conservation Association, National Science and Engineering Research Council, Daishowa Marubeni International Ltd., Canadian Forest Products Ltd., and the Canadian Forest Service Graduate Supplement program. Table of Contents

Chapter 1: Introduction 1 Background and Rationale 1 The Role of Community Ecology in Forest Management 2 Thesis Objectives 4 Literature Cited 5 Chapter 2: Aggregated retention patch size and beetle conservation in the boreal mixedwood forest of Alberta, Canada 8 Introduction 8 Materials and Methods 9 Study Sites 9 Study Design 10 Data Collection // Data Analyses 12 Results 14 Beetle Community 14 Single Species Comparisons 16 Indicator Species 16 Temperature 17 Discussion 17 Species Richness and Abundance 17 Impacts on the Beetle Community 18 Temperature 20 Indicator Species 21 Management Implications 22 Literature Cited 23 Appendices 33 Chapter 3: Ground and rove beetle responses to retention patch isolation: the value in considering rarity 37 Introduction 37 Materials and Methods 39 Study Design 39 Data Collection 40 Data Analyses 40 Results 43 General Results 43 Beetle Catch and Species Richness 43 Community Analyses 43 Single Species Comparisons 45 Discussion 45 Impact of Patch Isolation 46 Challenges of Analyzing Rarity 48 Conclusions 49 Literature Cited 49 Chapter 4: Spatial and temporal response of ground beetles to recent and regenerating harvest edges in the boreal forest of Alberta, Canada 60 Introduction 60 Methods 61 Study Site 61 Study Design 61 Beetle Sampling 62 Data Analyses 63 Results 65 Cover-type Effect 65 Edge Recovery 67 Discussion 68 Cover-type Effect 68 Indicator Species 70 Edge Age 71 Conclusions 72 Literature Cited 72 Chapter 5: Discussion 84 Main Results 84 Implications for Forest Management 86 Future Research and Limitations of the Dissertation 87 Literature Cited 89 List of Tables

Table: 3.1: Known habitat affinities of ground and rove beetles exhibiting strong (>0.2) correlation with canonical analysis of principle co-ordinates (CAP) axis 2 55

Table 4.1: Ground beetle indicator species that exhibited the highest Indicator Values to two forest zones, identified by previous analyses, and to the two cover-types studied (deciduous dominated and conifer dominated) 77 List of Figures

Figure 2.1: Map of study area showing relative proximity of harvest blocks 26 Figure 2.2: Mean (+1 S.E.) standardized catch of ground beetles and rove beetles within clear- cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha); within two forest cover-types. Bars with the same letters are not significantly different (a =0.05) 27 Figure 2.3: Rarefaction-estimated species richness of ground and rove beetles within clear-cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha). Two cover-types were sampled: a) deciduous dominated stands; b) conifer dominated stands. Arrows indicate sample size used to compare estimates across treatments... 27 Figure 2.4: Hierarchical clustering analysis of Bray-Curtis similarity coefficients for_ground beetles and rove beetles sampled in clear-cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha). Two cover-types were sampled: a) deciduous dominated stands; b) conifer dominated stands. Numbers 1-9 represent node labels for comparison to indicator species analysis (ISA) 28 Figure 2.5: Non-metric multidimensional scaling (NMDS) ordinations of: a) ground and rove beetles; b) rove beetles; and c) ground beetles, sampled from: clear-cuts; mature forest; and small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha) aggregated retention patches within two dominant forest cover-types of the boreal region 29

Figure 2.6: Standardized catch of the six most common species collected via standard pitfall trapping within clear-cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha); and in two of the dominant cover-types (deciduous dominated and conifer dominated) 30 Figure 2.7: Significant ground beetle and rove beetle indicator species (p<0.05) for each node of a hierarchical clustering analysis. Species are listed until they achieve their highest indicator value (bolded text). Samples collected by pitfall traps in clear-cuts, mature forest, and three sizes of retention patches: small (<1.8 ha); medium (2-4.4 ha); and large (>4.5 ha). Numbers 1-7 represent the corresponding hierarchical clustering node where values were derived 31

Figure 2.8: Comparison of temperatures collected via Thermochron ibuttons withinjnature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha);_and large (>4.5 ha); within two dominant cover types of the boreal forest. Bars with same_letters are not statistically different (a =0.05). Sample sizes indicated above bars 32 Figure 3.1: Standardized catch of rare and common ground beetles and rove beetles within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest. Rare species are those representing <5% of the total catch 52

Figure 3.2: Average species richness (+1S.E.) for ground beetles and rove beetles. Species richness estimates were generated via individual based rarefaction standardized to 373 individuals for clear-cuts, and mature forest, and three retention patch distance classes: near (10- 55 m); mid (65-125 m); and far (140-360 m) from the mature forest 53 Figure 3.3: Hierarchical clustering analysis using: a) log transformed abundances; and b) abundances relativized by site then species totals; for ground and rove beetles. Samples were collected within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m) from the mature forest 54

Figure 3.4: Non-metric multidimensional scaling ordination displaying a) log transformed abundances; and b) abundances relativized by site then species totals; for ground and rove beetles. Samples were collected within clear-cut, mature forest, and three distance classes of retention patches: near (10-55 m), mid (65-125 m), and far (140-360 m) from the mature forest. . 55

Figure 3.5: Canonical analysis of principal co-ordinates (CAP) ordinations displaying: a) log transformed abundances; and b) abundances relativized by site then species totals; for ground and rove beetles. Samples were collected within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest 56

Figure 3.6: Reletavized catch of the 6 species with the strongest positive correlation to canonical analysis of principal co-ordinates (CAP) axis 2. Ground beetles and rove beetles were sampled within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest 57

Figure 4.1: Schematic of edge transect design used to collect ground beetles in the summers of: a) 2006; and b) 2007 76 Figure 4.2: Individual-based rarefaction estimate of ground beetle species richness at different location along stand edges for: a) deciduous stands; and b) conifer stands, in boreal mixedwood forests of northwestern Alberta. Samples were grouped as: 'clear-cut' (at least 50 m from edge in the adjacent cut block); 'forest edge' (0-1 m from forest edge); 'forest mid' (30-45 m); and 'forest interior'(60-90 m) 76

Figure 4.3: Hierarchical clustering analysis of beetle community composition along: a) deciduous dominated edges; and b) conifer dominated edges. Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0m), and up to 90 m into the forest 78 Figure 4.4: Canonical analysis of principal co-ordinates (CAP) ordinations of: a) deciduous dominated edges; and b) conifer dominated edges. Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0 m) and up to 90 m into the forest 79 Figure 4.5: Percentage of standardized catch in each of two forest cover-types (decidous,conifer) of 10 indicator ground beetle species with significant indicator values (p<=0.05). Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0 m) and up to 90 m into the forest 80 Figure 4.6: Individual-based rarefaction estimates of ground beetle species richness at different location along stand edges 2, 8, and 15 years post-harvest in boreal mixedwood forests of northwestern Alberta. Samples were grouped as: 'clear-cut' (at least 50m from edge in the adjacent cut block); 'forest edge' (0-1 m from forest edge); 'forest mid' (30-45 m); and 'forest interior'(60-90 m) 81 Figure 4.7: Hierarchical clustering analysis of ground beetle community composition along edges: a) 8 years post-harvest; and b) 15 years post-harvest. Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0m), and up to 90 m into the forest 82

Figure 4.8: Percentage of standardized catch for each of four ground beetle species collected along forest edges. Transects consisted of traps placed in the clear-cut (CC), along the forest edge (0 m), and up to 90 m into the forest. Stands with edges of three different ages (2, 8, and 15 years post-harvest) were sampled 83 List of Appendices:

Appendix 2A: Location, size, and collection details of aggregated retention patches sampled within the boreal forest of North-western Alberta, Canada 33

Appendix 2B: Ground and rove beetles collected via pitfall trapping within deciduous and conifer stands 34

Appendix 3A: Ground and rove beetles collected within conifer stands and grouped by abundance class 58 Chapter 1: Introduction Background and Rationale The conservation of biological diversity within managed forests has emerged as a central issue in sustainable forest management (Spence 2001, Lindenmayer and Franklin 2002, Schmiegelow and Monkkonen 2002, Work et al. 2003). Not only has public concern for alternative forest values increased (MacLean 2007), so too has the recognition that conservation of biological diversity (hereafter referred to as 'biodiversity') is critical to economic success in the forest industry (Burton et al. 2006). As a result, biodiversity conservation is a target of adaptive management strategies that seek to use and increase scientific knowledge through application and assessment of outcomes (Holling 1979, Walters 1986, Kessler et al. 1992). Scientists play a critical role in this process and have produced innovative, informative, and relevant studies of the impacts of forest management alternatives on biodiversity (Kangas and Kuusipalo 1993, Spence et al. 1996, Schmiegelow et al. 1997, Lemieux and Lindgren 2004, Buddie et al. 2006, Macdonald 2007). These studies have in turn contributed to our understanding of biotic responses to industrial activities within the boreal forest, and support proactive solutions to biodiversity conservation in managed forests. The research generated in this context has promoted greater collaboration between ecologists and foresters, resulting in development and testing of new ideas such as basing forest management on the emulation of natural disturbances (Hunter 1993). The resulting natural disturbance model (NDM) aims to emulate natural processes, and retain associated structural features on managed landscapes (Lindenmayer and Franklin 2002). Such natural disturbance features in the boreal forest include fire frequency intervals, fire perimeter designs, and structural characteristics such as live tree retention. In general philosophy, the model attempts to foster a less anthropocentric approach to forest management. Adoption of natural disturbance guidelines has been accelerated in the Canadian boreal forest because the NDM presents an opportunity to balance ecological and economic objectives (Burton et al. 2006). However, a critical assumption of the NDM is that because species have evolved alongside natural disturbances, such as forest fires, by emulating the temporal, spatial and ecological characteristics of natural disturbances through harvest and regeneration practices, forest-dwelling species will be better able to adapt to forestry activities (Hunter 1993, Bergeron et al. 2002). This 'coarse filter' approach to biodiversity management remains largely untested

1 and there is little evidence that application of these broad scale management changes effectively conserve biodiversity, either at the stand or landscape level (Work et al. 2003). It is therefore essential that scientists work to test whether NDM actually achieves the results that we intend (Spence2001). One particular aspect of the NDM that requires investigation is the use of aggregated retention patches to conserve biodiversity. These retention patches are islands of live and dead trees left following harvesting, with the goal of emulating fire skips on a landscape (Franklin et al. 1997). While known to serve important roles in providing habitat for forest species (Gandhi et al. 2001), the patch characteristics associated with effective conservation have not yet been clearly identified. This is an important next step in testing the NDM, as environmental factors such as edge effects can greatly reduce the actual benefit of small retention applications (Bradshaw 1992, Matveinen-Huju et al. 2006). Thus, not only does such work directly evaluate the conservation potential of the NDM, it also introduces valuable biological information into natural disturbance modelling. This incorporation of research results into forest planning should support effective policy and make forest managers increasingly confident about resource management choices.

The Role of Community Ecology in Forest Management As scientists test the NDM they must consider how best to measure biodiversity responses. One approach is the use of indicator species. The concept of indicator species has received significant attention, both theoretically and in an applied framework. This concept assumes a correlation between the response of an indicator species and the overall response of biodiversity to environmental change (McGeoch 1998, Carignan and Villard 2002). The perceived efficiency of monitoring indicator species is undeniably appealing to conservationists, governments, and land managers alike (Carignan and Villard 2002). However, there is mounting empirical evidence that indicator species have limited ability to represent broader biodiversity responses to disturbance (Simberloff 1998, Andelman and Fagan 2000). Prendergast (1997) questioned their utility on the grounds that single species are highly variable, and exhibit spatially complex relationships within landscapes. Further limitations result from the fact that indicator species occupy only a narrow range of ecological conditions or habitat types (Koskimies 1989), and therefore cannot effectively 'indicate' the response of an entire ecosystem

2 (Simberloff 1998). The dependability of using a few indicator species is therefore highly questionable. Many of the identified problems that reduce the utility of indicator species are at the same time the major advantages of community level management. For example, a broader community level focus provides improved capacity to describe heterogeneity in biodiversity responses to ecosystem changes (Maleque et al. 2006). In addition, observing the responses of the whole community can provide better insight on specific drivers of ecosystem change. Community responses clearly correlate with broader ecosystem responses, and have the additional benefit of permitting cross-taxon comparisons to enhance our understanding of multiple species responses (Olden et al. 2006). In recognition of their utility, community analyses have been used to answer a broad range of forest management questions. For example, Schieck et al. (1995) used bird communities to determine the utility of retention patches in maintaining old-growth dependent species. Beetle and spider community responses have also been used to determine how well current forest management practices are emulating the biodiversity outcomes of natural disturbances (Buddie et al. 2006). Similarly, the impacts of variable retention harvesting on biodiversity have been assessed through a variety of community level analyses using beetles, plants, and birds as surrogate communities (Brais et al. 2004, Work et al. 2004, Harrison et al. 2005). The major benefit of using whole communities to monitor biodiversity responses to forest management practices is that it permits an evaluative process, whereas monitoring indicator species alone permits only a determination of effectiveness. The question of focus for indicator species is: 'how effective was our management in maintaining target species?' However, this approach presents little room to determine ways to improve overall management options because typically only presence/absence values for one or a few indicator species are collected. Evaluative monitoring however allows managers to assess their success at maintaining 'biodiversity' and, using these data, to determine possible ways to adapt their strategies in the future to better meet management objectives (Rempel et al. 2004). The value of having true measures of diversity and species richness presented from community monitoring therefore greatly enhances the utility of biodiversity studies.

3 In this thesis, I present results of my work on epigaeic beetle communities to evaluate aggregated retention applications in the boreal forest of Alberta, Canada. Epigaeic beetles were selected because they represent a broad range of sensitivities to ecosystem change (e.g., Rainio and Niemela 2003), are easily sampled (e.g., Spence and Niemela 1994) and are known to respond to forest management practices (Spence et al. 1996, Buddie et al. 2006, Pohl et al. 2007).

Thesis Objectives I aimed to provide a comprehensive evaluation of aggregated retention patches in preserving epigaeic beetle communities. More specifically, I tested whether patch characteristics, such as size and isolation, influence the conservation of mature forest communities within retention patches. I also evaluated the importance of edge effects for explaining diversity patterns observed in patches. By sampling industrially harvested blocks designed under the NDM, I sought to determine the success of the NDM in conserving biodiversity within a managed boreal land base. In chapter 2,1 explore the influence of retention patch size on conservation of mature forest communities. By exploring patch sizes greater than those previously studied, I am able to suggest potential size thresholds whereby communities representative of the mature forest are preserved. Based on this assessment of biotic and abiotic parameters, I propose a method of integrating my results into a NDM of forest management. In chapter 3,1 examine how retention patch proximity to an intact forest stand influences beetle community composition. My results also provide insight into the importance of considering rare species in community ecology. By contrasting the responses of rare and common species, the analyses facilitate a broader understanding of rarity, and emphasize the potential importance of considering patch isolation in forest management. Chapter 4 focuses on edge effects as an explanation of biodiversity patterns associated with retention patches of different sizes and degrees of isolation. By assessing beetle community variability across forest edges, I aimed to determine the degree to which recent edge effects contribute to fragmentation impacts. Through incorporation of edges of various ages into the experimental design, I was also able to assess changes in edge effects over time. In chapter 5,1 synthesize the research findings in light of their importance for sustainable forest management. I discuss the overall implications of the size thresholds suggested by my

4 work, and recommend that managers increase the frequency of larger patches on the landscape. Furthermore, I hypothesize ways of reducing isolation effects by leaving larger patches at greater distances from the interior forest. Finally, I discuss potential limitations of the dissertation, and present ideas for future research that could advance our understanding of the processes analyzed in this thesis.

Literature Cited Andelman, S. J. and W. F. Fagan. 2000. Umbrellas and flagships: Efficient conservation surrogates or expensive mistakes? Proceedings of the National Academy of Sciences of the United States of America 97:5954-5959. Bergeron, Y., A. Leduc, B. D. Harvey, and S. Gauthier. 2002. Natural fire regime: A guide for sustainable management of the Canadian boreal forest. Silva Fennica 36:81-95. Bradshaw, F. J. 1992. Quantifying edge effect and patch size for multiple-use silviculture - a discussion paper. Forest Ecology and Management 48:249-264. Brais, S., B. D. Harvey, Y. Bergeron, C. Messier, D. Greene, A. Belleau, and D. Pare. 2004. Testing forest ecosystem management in boreal mixedwoods of northwestern Quebec: initial response of aspen stands to different levels of harvesting. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:431-446. Buddie, C. M., D. W. Langor, G. R. Pohl, and J. R. Spence. 2006. Arthropod responses to harvesting and wildfire: Implications for emulation of natural disturbance in forest management. Biological Conservation 128:346-357. Burton, P. J., C. Messier, W. L. Adamowicz, and T. Kuuluvainen. 2006. Sustainable management of Canada's boreal forests: Progress and prospects. Ecoscience 13:234-248. Carignan, V. and M. A. Villard. 2002. Selecting indicator species to monitor ecological integrity: A review. Environmental Monitoring and Assessment 78:45-61. Franklin, J. F., D. R. Berg, D. A. Thornburgh, and J. C. Tappeiner. 1997. Alternative silviculatural approaches to timber harvesting: Variable retention harvest systems. Island Press, Washington, D.C. Gandhi, K. J. K., J. R. Spence, D. W. Langor, and L. E. Morgantini. 2001. Fire residuals as habitat reserves for epigaeic beetles (Coleoptera : Carabidae and Staphylinidae). Biological Conservation 102:131-141. Harrison, R. B., F. K. A. Schmiegelow, and R. Naidoo. 2005. Stand-level response of breeding forest songbirds to multiple levels of partial-cut harvest in four boreal forest types. Canadian Journal of Forest Research 35:1553-1567. Holling, C. S. 1979. Adaptive environmental assessment and management. John Wiley and Sons, New York. Hunter, M. L. 1993. Natural fire regimes as spatial models for managing boreal forests. Biological Conservation 65:115-120. Kangas, J. and J. Kuusipalo. 1993. Integrating biodiversity into forest management planning and decision-making. Forest Ecology and Management 61:1-15. Kessler, W. B., H. Salwasser, C. W. Cartwright, and J. A. Caplan. 1992. New perspectives for sustainable natural-resources management. Ecological Applications 2:221-225. Koskimies, P. 1989. Birds as a Tool in Environmental Monitoring. Annales Zoologici Fennici

5 26:153-166. Lemieux, J. P. and B. S. Lindgren. 2004. Ground beetle responses to patch retention harvesting in high elevation forests of British Columbia. Ecography 27:557-566. Lindenmayer, D. B. and J. F. Franklin. 2002. Conserving forest biodiversity: A comprehensive multi-scaled approach. Island Press, Washington, D.C. Macdonald, S. E. 2007. Effects of partial post-fire salvage harvesting on vegetation communities in the boreal mixedwood forest region of northeastern Alberta, Canada. Forest Ecology and Management 239:21-31. MacLean, D. A. 2007. Does the Canadian forest sector have a viable future? Is current forest management acceptable to the general public? Would you advise your kids to take forestry? Forestry Chronicle 83:54-60. Maleque, M. A., H. T. Ishii, and K. Maeto. 2006. The use of arthropods as indicators of ecosystem integrity in forest management. Journal of Forestry 104:113-117. Matveinen-Huju, K., J. Niemela, H. Rita, and R. B. O'Hara. 2006. Retention-tree groups in clear- cuts: Do they constitute 'life-boats' for spiders and carabids? Forest Ecology and Management 230:119-13 5. McGeoch, M 1998. The selection, testing and application of terrestrial insects as bioindicators. Biological Reviews 73: 181-201. Olden, J. D., M. K. Joy, and R. G. Death. 2006. Rediscovering the species in community-wide predictive modeling. Ecological Applications 16:1449-1460. Pohl, G. R., D. W. Langor, and J. R. Spence. 2007. Rove beetles and ground beetles (Coleoptera: Staphylinidae, Carabidae) as indicators of harvest and regeneration practices in western Canadian foothills forests. Biological Conservation 137:294-307. Prendergast, J. R. 1997. Species richness covariance in higher taxa: Empirical tests of the biodiversity indicator concept. Ecography 20:210-216. Rainio, J. and J. Niemela. 2003. Ground beetles (Coleoptera : Carabidae) as bioindicators. Biodiversity and Conservation 12:487-506. Rempel, R. S., D. W. Andison, and S. J. Harmon. 2004. Guiding principles for developing an indicator and monitoring framework. Forestry Chronicle 80:82-90. Schieck, J., K. Lertzman, B. Nyberg, and R. Page. 1995. Effects of patch size on birds in old- growth montane forests. Conservation Biology 9:1072-1084. Schmiegelow, F. K. A., C. S. Machtans, and S. J. Hannon. 1997. Are boreal birds resilient to forest fragmentation? An experimental study of short-term community responses. Ecology 78:1914-1932. Schmiegelow, F. K. A. and M. Monkkonen. 2002. Habitat loss and fragmentation in dynamic landscapes: Avian perspectives from the boreal forest. Ecological Applications 12:375- 389. Simberloff, D. 1998. Flagships, umbrellas, and keystones: Is single-species management passe in the landscape era? Biological Conservation 83:247-257. Spence, J. R. 2001. The new boreal forestry: adjusting timber management to accommodate biodiversity. Trends in Ecology & Evolution 16:591-593. Spence, J. R., D. W. Langor, J. Niemela, H. A. Carcamo, and C. R. Currie. 1996. Northern forestry and carabids: The case for concern about old-growth species. Annales Zoologici Fennici 33:302-302. Spence, J. R. and J. K. Niemela. 1994. Sampling carabid assemblages with pitfall traps - the madness and the method. Canadian Entomologist 126:881-894.

6 Walters, C. 1986. Adaptive management of renewable resources. Macmillan, New York. Work, T. T., D. P. Shorthouse, J. R. Spence, W. J. A. Volney, and D. Langor. 2004. Stand composition and structure of the boreal mixedwood and epigaeic arthropods of the Ecosystem Management Emulating Natural Disturbance (EMEND) landbase in northwestern Alberta. Canadian Journal of Forest Research 34:417-430. Work, T. T., J. R. Spence, W. J. A. Volney, L. E. Morgantini, and J. L. Innes. 2003. Integrating biodiversity and forestry practices in western Canada. Forestry Chronicle 79:906-916.

7 Chapter 2: Aggregated retention patch size and beetle conservation in the boreal mixedwood forest of Alberta, Canada Introduction A central aim of forest harvesting under the natural disturbance model is to maintain structural features and habitat heterogeneity similar to the results of natural disturbance (Attiwill 1994, Bergeron and Harvey 1997, Franklin et al. 1997). The model suggests that these features contribute, among other things, to conservation of biodiversity within managed landscapes (Hunter 1993, Haila et al. 1994, Bergeron et al. 2002). In the boreal region, forest fires create habitat variability through features such as fire skips that are green-tree residual islands within a highly disturbed matrix (Eberhart and Woodard 1987). These residual islands provide critical habitats for maintaining biodiversity, thereby functioning as 'life boats' for species characteristic of older forests, enhancing landscape connectivity, and acting as source populations for the recolonization of disturbed habitat (Gandhi et al. 2001, Lindenmayer and Franklin 2002). The importance of such features is well recognized and has led to widespread application of aggregated retention, defined as isolated patches of live trees, within recent harvest blocks (Work et al. 2003). Such applications enable managers to better approximate harvest footprints to the disturbance patterns associated with wildfires. Size of retention patches that will best conserve biodiversity within harvest blocks has been of particular interest to forest managers and researchers (Schieck and Hobson 2000, Bradbury 2004, Gandhi et al. 2004, Pearce et al. 2005, Matveinen-Huju et al. 2006, Oliver et al. 2007). Attention to the influence of patch size on biodiversity is not new to ecology (Gleason 1922, Diamond 1976, Bender et al. 1998), and much work on this topic has flowed from the Equilibrium Theory of Island Biogeography (Macarthur and Wilson 1967). However, use of residual patches in forestry is a recent practice and important questions about appropriate retention patch size remain unanswered. Studies to date have typically focused on the role of aggregated retention sizes of < 2 hectares (ha) (but see Schmiegelow et al. 1997, Pearce et al. 2005), and conclude that small patches do not completely conserve biodiversity characteristic of continuous mature forests (Bradbury 2004, Matveinen-Huju et al. 2006). However, residual patches (i.e., fire skips) produced by wildfires are known to range in size from <1 ha up to at least 10 ha (DeLong and Tanner 1996), and increase in size as the overall disturbance size increases (Eberhart and Woodard 1987, Andison 2004). Thus, developing a greater understanding of how these larger residual patches contribute to biodiversity conservation is a

8 critical step in understanding the benefits of natural disturbance emulation, especially given the emphasis on increasing harvest block sizes under the natural disturbance model (Hunter 1993, DeLong and Tanner 1996). Terrestrial arthropods are excellent subjects for exploring questions about aggregated retention size, as they respond to variation in forest harvesting treatments (Spence et al. 1996, Gandhi et al. 2004, Buddie et al. 2006). Beetles are particularly appropriate for addressing forest management questions because they are easily sampled, well defined taxonomically, and have been widely studied in forest management research (Niemela et al. 1993, Haila et al. 1994, Lemieux and Lindgren 2004). Ground and rove beetles (Coleoptera: Carabidae and Staphylinidae) have especially been promoted as effective indicators of ecosystem intactness and recovery following disturbances (Pearce and Venier 2006, Pohl et al. 2007), and ground beetles have been widely adopted for use in bioindicator research (Rainio and Niemela 2003). Many species of these two families can also be collected using the same sampling technique (pitfall traps) thereby providing a cost effective way to study multiple taxonomic groups and strengthen conclusions about impacts of aggregated retention on biodiversity (Rainio and Niemela 2003). In the following chapter I explore the effects of aggregated retention size on epigaeic beetle assemblages within the boreal forest of north-western Alberta, Canada. I sampled aggregated retention patches ranging in size from 0.3-14.1 ha to test the hypothesis that larger patches will be more important for the conservation of mature forest species within a harvested matrix. In addition, I considered how variation in air temperature and cover-type could be related to changes in the beetle assemblages.

Materials and Methods Study Sites Study sites were located in north-western Alberta, Canada within the lower foothills forest region (Rowe 1972) of the boreal forest (Fig. 2.1). Patches were selected from within an industrial forest landscape ranging from 98 km west of Peace River (56.41°N 118.39°W) to approximately 25 km northwest of Manning (56.54°N 117.37°W). The maximum distance between study sites was 98 km with a mean distance of 35 km. Samples were collected within the two dominant merchantable forest cover-types of the mixedwood boreal region. Choice of sample sites for each cover-type was driven to some extent by a distinct east-west transition in over-story composition.

9 The western half of the study landscape was dominated by white spruce (Picea glauca (Moench)) with lodgepole pine (Pinus contorta Douglas), black spruce {Picea mariana (Miller)), trembling aspen {Populus tremuloides Michaux) and balsam poplar (Populus balsamifera L.) occurring as secondary canopy elements. The eastern half of the study landscape was dominated by trembling aspen and balsam poplar with white spruce and paper birch (Betula papyrifera Marshall) occurring as secondary species.

Study Design To assess the influence of aggregated retention size in preserving mature forest species, I sampled three size classes of retention patches within each cover-type: small (<1.4 ha), medium (1.8-4.4 ha) and large (4.5-14.1 ha) (Appendix 2A). Clear-cut and mature forest stands were also sampled to provide comparisons to aggregated retention sites. The two cover-types were categorized as conifer dominated (>70% of overstory canopy composed of conifer species) and deciduous dominated (>70% of overstory canopy composed of deciduous species). All treatments were replicated five times for a total of 25 sampling units per cover-type (Appendix 2A). Intact forest stands were within a continuous forest matrix of pyrogenic origin and were at least 108 years of age, with the exception of one deciduous stand that was only 70 years of age. Because of the relatively recent use of aggregated retention as a management option within this landscape, we were able to control for time since harvest only within cover-types thereby reducing the power of comparisons between cover-types. Deciduous harvest blocks were an average size of 155 ± 51 ha (85-307 ha), were naturally regenerated with no site preparation following harvesting, and were sampled 2-4 years post-harvest with the exception of one patch (R2-32-1) which was 7 years post harvest (Appendix 2A). Conifer harvest blocks were sampled 7 years post-harvest, with the exception of two patches (R2-16-1, R3-13-1) that were only 3 years post-harvest (Appendix 2A). Conifer harvest blocks were an average harvest size of 174 ± 73 ha (44-379 ha), and had been exposed to a variety of site preparation techniques, ranging from minor removal of coarse woody debris to scarification exposing mineral soils. Conifer sites had been planted with white spruce and lodgepole pine seedlings an average of one year following harvest.

10 Forest left in all aggregated retention patches was greater than 88 years of age, and composed of merchantable trees representative of the adjacent forest composition removed during harvesting. The patches were of pyrogenic origin with the exception of three patches (R3-9-1, R3-21-1, R1-9-1) which exhibited signs of previous harvest (Appendix 2A). These three patches were, however, similar in age to wildfire-origin patches (Appendix 2A), and, since large scale industrial harvesting has only occurred in this area for 37 years, likely experienced non-intensive harvesting. Thus, although an important consideration when interpreting results, this should not have had a major effect on dominant inferences in the data set.

Data Collection Epigaeic beetles were collected in all sites from May 15th to August 22nd 2007 (spring thaw to the start of overnight frosts) using pitfall traps filled with approximately 1.5 cm of silicate free ethylene glycol as a preservative (Spence and Niemela 1994). A subset of the deciduous sites was also sampled from May to August in 2006 (Appendix 2A). Pitfall traps were covered with an elevated plastic lid to reduce the amount of precipitation and debris collected, and the beetle catch was removed every three weeks throughout the sampling period. A total of four traps per sample unit (i.e., patch, forest, or clear-cut) were placed randomly with respect to microsites within the central core of small and medium patches, and along transects through the centre of large patches. The central core of patches was defined as the area within a 25-50 m radius from the centre of the patch; with all traps being a minimum of 15-20 m from the patch edge. Transects were used when sampling the mature forest and clear-cut treatments. All traps were a minimum of 25 m from each other in order to ensure independence (Digweed et al. 1995). Samples from the mature forest were collected a minimum of 60 m (ca. two tree lengths) from any disturbed edge to avoid the influence of edge effects (Spence et al. 1996, Pohl et al. 2007). Similarly, samples from clear-cuts were collected a minimum of 50 m from any forested edge. Sites studied in 2006 were sampled in the same way but with only three pitfall traps per sample unit. Samples were sorted to family and identified to species using Lindroth (1961-1969) for ground beetles, and Newton et al. (2001) and references therein for rove beetles. Larvae from both taxa were excluded from analyses because they were not identifiable below the family level. Specimens from the staphylinid sub-family Aleocharinae, with the exception of those belonging

11 to Lypoglossa franclemonti Hoebeke and Lypoglossa angularis (Maklin), were excluded from all analyses except in the case of relative abundance comparisons, as species-level identifications were not possible. Voucher specimens have been deposited at the Strickland Entomological Museum, and Northern Forestry Centre in Edmonton, Alberta, Canada. Ambient air temperature was recorded once every 90 minutes between July 28l and August 16th using Thermochron iButton data loggers (Maxim Dallas 2007). One logger was placed within each of four replicates of the small, medium, large and mature treatments of both cover-types. However, two of the loggers malfunctioned and thus temperature records are missing for these sites (Fig. 2.8). Data loggers were placed at a height of one metre from the ground and secured to the north-facing side of a tree. A plastic lid was placed over the logger to reduce the influences of direct sunlight.

Data Analyses a) Beetle Community For all analyses except rarefaction estimates, the beetle data were standardized to 100 trap-days to adjust for differences in sampling effort resulting from trap disturbance. For sites sampled in both 2006 and 2007, raw abundances were pooled between years and also standardized to 100 trap-days. Mean standardized catch was compared between treatments using two-way ANOVA in SPSS 15.0 (SPSS 2007) following a log transformation to meet assumptions of the test. Rarefaction was used to compare species richness between treatments because traditional diversity indices are sensitive to variations in sample size (i.e., number of individuals collected). Rarefaction avoids this problem by adjusting species richness estimates based on sample size, thus providing a standardized approach to compare species richness within and among treatments (Gotelli and Colwell 2001). Since rarefaction takes into account both species richness and abundance, it can also be interpreted as a measure of species diversity (Buddie et al. 2006). Rarefaction curves were generated from raw species abundance data using the Vegan package (Oksanen et al. 2005) in R 2.6 (R Development Core Team, 2007). Variation in beetle assemblage composition was compared across forest type and patch size by calculating Bray-Curtis percent similarities on standardized abundances. The data were log (x+1) transformed to reduce the influence of abundant species (McCune and Grace 2002), and analysed via a hierarchical clustering analysis using web-based software of Brzustowski

12 (1999). Distances between groups were assessed using unweighted arithmetic averages, defined as the average distance between a sample in group A, and a sample in group B. Non-metric multidimensional scaling ordination (NMDS) was conducted using the Bray- Curtis calculation on log (x+1) transformed data in PC-ORD Version 5 (McCune and Mefford 1999). NMDS was selected for ordinations because it avoids the assumption of linear relationships between variables, and efficiently handles zero abundance values within community datasets (McCune and Grace 2002). Ordinations were constructed for both taxa, and for ground and rove beetles separately to enable comparisons of family responses to treatments. All ordinations were developed using a random starting configuration, followed by 50 iterations of real data. The final ordination was permitted to increase in dimensionality provided that doing so reduced the stress by more than 5%. Significance of the ordination was assessed against 100 runs of randomized iterations of data using a Monte Carlo analysis. The significance of ordination groupings based on treatment was assessed using a multi- response permutation procedure (MRPP) in PC-ORD Version 5 (McCune and Mefford 1999), using the Bray-Curtis dissimilarity measure and the default weighting option. This analysis produces a chance corrected within group agreement (A), which defines the within-group homogeneity of the treatments. The A value ranges from 1 (maximum homogeneity) to 0 (random grouping), an A value of >0.3 indicates strong grouping within treatments (McCune and Grace 2002). The analysis also produces a p-value for testing the significance of the groupings. b) Indicator Species Analysis The extent of species specialization to sample groupings was determined using the Indicator Value technique proposed by Dufrene and Legendre (1997), with significance being tested via a Monte Carlo analysis in PC-ORD Version 5 (McCune and Mefford 1999). Firstly, indicator values were calculated separately for all 10 treatment/cover-type combinations (i.e., clear-cut, small, medium, large, mature). Second, to determine how robust these estimates were to the grouping of treatments, indicator values were re-calculated at all 9 nodes of the hierarchical clustering analysis by pooling subsequent treatments together. This permits assessment of whether species achieve a higher indicator value at a grouping level that is different from the assigned treatments and, therefore, increases understanding of the level at which the species is the best indicator. Species were displayed within the hierarchy until they achieved their highest

13 indicator value. The standardized catch of the six most common indicator species were then graphed to show variation in catch across all treatments and for both cover-types. c) Temperature Analysis Temperature is known to influence ground beetle activity and distribution (Lovei and Sunderland 1996, Magura et al. 2001, Magura 2002), thus average, minimum and maximum temperature, and its coefficient of variation across the sampling period were calculated for each sample unit in both cover-types. Values were then assessed by two-way ANOVA using SPSS 15.0 (SPSS 2007). A Tukey's post-hoc analysis was also used to assess which treatments differed significantly from each other.

Results Beetle Community A total of 16,467 beetles representing 92 species (60 Staphylinidae and 32 Carabidae) were collected (Appendix 2B). The most common rove beetles were Tachinus elongatus Gyllenhal (n=1916) and Tachinus frigidus Erichson (n=1396). The most common ground beetle was Pterostichus adstrictus Eschscholtz (n=1608). Thirty-six species were represented by fewer than 10 specimens, and 14 were singletons. Interestingly, Philonthus fulcinus Smetana was collected within large deciduous retention patches a total of 14 times; these are the first records for this species in Alberta and they document an extension of known range from northern Washington, USA. Average standardized catch was significantly higher in deciduous dominated sites than in conifer dominated sites (F=18.3, df=l,4,/?<0.001) and this pattern was evident for all classes (Fig. 2.2). Within both cover-types, mean standardized catch was significantly lower in clear- cuts than in medium and large retention patches, or mature forest controls (/?<0.05). This difference in catch increased with patch size (Fig. 2.2). Beetle catches from small patches, however, did not differ statistically from those of any other treatment. Rarefaction-estimated species richness showed similar patterns among treatments in both cover-types in that clear-cuts had the greatest species richness, followed by small aggregated retention patches (Fig. 2.3), when compared at the lowest observed catch to standardize for sample size. Among the other treatments in deciduous dominated stands, large patches had slightly greater species richness than the medium patches and mature forest (Fig. 2.3a). In the

14 conifer dominated stands, the large patches and mature forest were similar to each other and to small patches, but medium patches had the lowest estimated species richness (Fig. 2.3b). Hierarchical clustering analysis showed a strong influence of cover-type on beetle community structure (Fig. 2.4). Overall similarity values between retention classes and the mature forest were slightly lower in conifer stands than in deciduous stands. In both cover-types, clear-cuts diverged most from other treatments, especially so in the conifer stands. Among forested treatments, small patches were the least similar to the other treatments in both cover- types. The percent similarity among medium and large patches and mature forest varied by cover-type, with mature forest and medium patches most similar in conifer forests but mature forest and large patches most similar in deciduous forests. NMDS ordination produced two-dimensional solutions for all three comparisons: rove beetles, ground beetles and both groups combined (Fig. 2.5). The ordination for all taxa combined was most similar to that for rove beetles, indicating that rove beetles had the greatest influence on overall patterns. The ordinations achieved a final stress of 14.6 for both taxa, 15.4 for rove beetles, and 13.1 for ground beetles, all of which were significantly lower than random (Monte Carlo test, n=100, p<0.01). Total variance explained by the ordinations were 93.7% for both taxa, 91.6% for rove beetles, and 94.1% for ground beetles, with consistently greater proportions of the variance explained by axis 1 (Fig. 2.5). MRPP analyses determined that groupings within treatments were significantly stronger than would be expected by chance in all of the ordinations (p<0.00\). The greater proportion of variance explained along axis 1 in all of the ordinations reflected the strong cover-type effects on the beetle communities (Fig. 2.5). In all three ordinations, a gradient of disturbance levels was also observed along axis 2. Clear-cuts diverged most from other treatments, and the beetle assemblages of medium, large, and mature treatments converged (Fig. 2.5). The small patches consistently grouped closest to the clear-cuts of all the retention treatments in both stand types, with two of the smallest deciduous patches grouping most closely to deciduous clear-cuts (Fig. 2.5). In deciduous forests, large patches grouped more closely to mature forest than did medium patches, a similar result to that indicated by the clustering analysis (Fig. 2.4). Analyses of the ground beetles and rove beetles separately yielded results similar to those from analysing both families together. Rove beetles (Fig. 2.5b) apparently showed more

15 sensitivity to the retention treatments, as ground beetles showed somewhat more overlap among all the retention treatments and the mature forest (Fig. 2.5c). However, results for rove beetles supported earlier interpretations of a convergence between medium, large, and mature treatments, and again suggested that large patches were most similar to the mature forest in the deciduous cover-type.

Single Species Comparisons The six most abundant species responded strongly to both patch size and cover-type. Catches of T. elongatus, the most abundant species in the study, increased markedly with patch size in the deciduous stands (Fig. 2.6). Tachinus frigidus responded similarly in the conifer cover-type and was infrequently captured in the clear-cut and small patch treatments (Fig. 2.6). Captures of P. adstrictus peaked in the clear-cuts, and the species was more commonly collected in small deciduous patches than other aggregate treatments. The remaining species showed variable responses but the overall trend emphasizes the importance of larger patches in maintaining beetle abundances similar to those in the mature forest.

Indicator Species Among the 44 species identified as significant indicators (p<0.05), only 20 achieved their highest indicator value at the final level of clustering when all cover-type/treatment combinations were separated {i.e., clear-cut, small, medium, large, mature) (Fig. 2.7). A greater proportion of carabids (56%, 18 species) than rove beetles (43%, 26 species) were significant indicators of cover-type and patch size combinations. The greatest number of species achieved a maximum indicator value at the hierarchical clustering level of 3 groups (node 2, Fig. 2.7), including 'conifer clear-cut', 'all deciduous', and 'conifer with trees'. A total of 6 species, including Dinothenarus pleuralis (Leconte), and T. elongatus, reached their highest indicator value when the 'conifer with trees' and 'all deciduous' groups were clustered together (Node 1, Fig. 2.7). I also noted that many species that could be considered as specialists under traditional indicator criteria (i.e., IndVal>80) achieved this level only when a large number of treatments were grouped together (e.g., node 2: 'all deciduous'; 'conifer with trees'), or when both the 'all deciduous & conifer with trees' were distinguished from 'conifer clear-cut' (e.g., node 1) (Fig. 2.7). These results suggest that species such as

16 Platynus decentis (Say), and Calathus advena (Leconte) had a strong presence in a wide range of the treatments studied.

Temperature The temperature data consist of 360 temperatures recorded over the course of 20 days for each sample site. The average temperature throughout this sampling period did not vary among the forest harvest treatments (Fig. 2.8), but was significantly lower in the conifer cover-type (F=64.6, df=l, 3,/?<0.001). The average maximum temperature was however significantly higher in small patches of both cover-types (F=3.54, df=l, 3,/?=0.031), and the average minimum temperature was significantly lower in the small patches of both cover-types than any other treatment (F=6.17, df=l, 3, /?=0.003 ). The coefficient of variation for the temperatures was also significantly greater within the small retention treatments of both cover-types (F= 3.65, df=l,3,Jp=0.028)(Fig. 2.8).

Discussion Species Richness and Abundance Both beetle catch and species richness tended to increase as patch size decreased. The significant reduction in pitfall catches and increased species richness in clear-cuts has also been observed in past studies in Alberta (Niemela et al. 1993, Buddie et al. 2006, Pohl et al. 2007), and underscores that harvesting can significantly alter biodiversity. The statistically similar catches found in the small patches and the clear-cuts indicate that small patches may experience disturbance effects similar to clear-cuts (Pearce et al. 2005, Matveinen-Huju et al. 2006). In contrast, the potential for larger patches to serve as 'lifeboats' for invertebrates was highlighted by the increasing catches as patch size increased. The higher catches accumulated within larger patches corroborates that they can serve as long term source populations for recovery of biodiversity within harvested units (Franklin et al. 1997). Effects of patch size on species richness may also influence the long term function of aggregated retention as lifeboats. In this study, small patches had the highest species richness among all aggregated retention treatments within both cover-types. This reflected the higher numbers of generalist species within these patches (e.g., Pterostichus adstrictus) (Fig. 2.6), suggesting a predominance of edge effects (Spence et al. 1996). Others have found similar

17 effects of small patch size on species richness of vascular plants (Bradbury 2004), and arthropods (Gandhi et al. 2004), all of which suggest edge effects as the driver for this pattern.

Impacts on the Beetle Community Analyses of the beetle assemblages showed a strong convergence in species composition among the medium and large patches, and the mature forest sites. The greatest shift in composition occurred in the conifer clear-cut fauna. Within these sites, an influx of generalist species such as P. adstrictus and Mycetoporus americanus Erichson and a striking reduction of common mature forest species such as T. frigidus and Quedius velox Smetana were observed. The unique community composition created by harvesting was further amplified by the fact that 10 species achieved their highest indicator value within the conifer clear-cuts. A similar, although much reduced, shift in assemblage composition was observed in deciduous clear-cuts. This change was explained by increasing abundances of open habitat species including P. adstrictus, and less common species such as Cymindis cribricollis Dejean and Gabrius brevipennis (Horn). The reduced impact of harvesting observed within deciduous clear-cuts may be explained by the shorter time since harvest because activity of mature forest species can remain high immediately after clear-cutting (Niemela et al. 1993). However, similar responses have been observed in previous research on deciduous sites (Spence et al. 1996) and likely result from rapid regeneration of trembling aspen within these blocks. This rapid regeneration forms a 'pseudo-canopy' which limits colonization of the site by open habitat specialists (Holliday 1992). The beetle assemblages of small patches were less similar to those of mature forest than were assemblages from medium and large patches in both cover-types. Within the deciduous cover-type assemblages of the two smallest patches studied (0.3 ha) fell directly within the variability of the clear-cut treatments. This suggests the existence of a minimum size threshold below which patches provide little benefit for conservation of the ground-dwelling fauna. Similar results have been found in Finland (Matveinen-Huju et al. 2006, Matveinen-Huju and Koivula 2008), further supporting the concept of a lower size threshold in boreal forest systems. Beetle assemblages in the remainder of the small patches also more closely resembled those of the clear-cut treatments than any other retention treatments in the NMDS ordinations. This pattern was also clearly supported by the abundance data from the most common species.

18 Mature forest species such as T. frigidus and Q. velox were consistently captured in lower abundance in small patches than in other forested treatments, while higher catches of open habitat species such as P. adstrictus were recorded in these small patches. These findings further support previous studies on other taxa in suggesting that aggregated retention patches smaller than 1.5 hectares are unlikely to maintain assemblages similar to mature forests because of changes in habitat quality for mature forest species and higher abundances of open habitat species (Bradbury 2004, Pearce et al. 2005, Matveinen-Huju et al. 2006). The notably greater similarity of large patches to mature forest in the deciduous cover- type suggests a larger patch size requirement for maintaining mature forest species within deciduous stands than in conifer stands. Although this difference may reflect, to some extent, the larger patches studied in deciduous stands (Table 2.1), assemblages of medium sized deciduous patches were less similar to the mature forest than were medium conifer patches. This finding emphasizes the importance of studying multiple forest types, and raises questions about the factors driving this cover-type effect. Edge effects have been found to be similar in deciduous and conifer stands within the boreal forest (Chapter 4), suggesting that this cover-type effect may be driven by other ecological variables not analyzed in this study. To strengthen aggregated retention prescriptions future research is needed in exploring this cover-type effect. The stronger response of rove beetles to patch size compared to ground beetles also emphasizes the importance of studying multiple taxa for biodiversity assessment (Rainio and Niemela 2003). This sensitivity of rove beetles to forest disturbance underscores their utility for applied ecological studies (Pohl et al. 2007), and is likely attributed to the wide variety of trophic niches which rove beetles occupy, ranging from predation to fungivory (Newton et al. 2001). In addition, staphylinid assemblages may respond more strongly than carabids to forest harvesting because they are generally able to fly away from unsuitable habitat while some ground beetles are flightless (Pohl et al. 2007). Thus while ground beetles are common and valuable components of biodiversity surveys, use of rove beetles in applied forest biodiversity studies is strongly encouraged. The dominant assemblage-level patterns observed in response to retention patch size appear to be linked to a reduction in edge habitat within larger patches. For example, T. frigidus in particular is sensitive to edge effects (Pohl et al. 2007), and its catch patterns clearly emphasized the benefits of medium and large conifer patches. This result, combined with others

19 here, show that small patches suffer from increased penetration by open-habitat species while supporting fewer mature forest species (Spence et al. 1996). This demonstrates the importance of emulating patch size distributions found following wildfires (Eberhart and Woodard 1987, DeLong and Tanner 1996), and not simply deploying small patches; a technique that is commonly advocated (Coates and Steventon 1995, Lindenmayer and Franklin 2002). Results of the present study provide an ecological rationale for also including medium and large patches, and suggest they should become part of sustainable forest management planning that embraces conservation of biodiversity. Small patches will likely not sustain populations of mature forest species, even for arthropods, but do contribute to achieving other ecological objectives. These alternative objectives might include: coarse woody debris recruitment, enhancing stand structural variability, and promoting landscape connectivity (Franklin et al. 1997). Thus leaving the entire patch size distribution found following wildfires should more fully promote biodiversity conservation goals under the natural disturbance model.

Temperature Patch size strongly influenced the variability of ambient air temperatures within aggregated retention sites. Small patches (<1.4 ha) within both cover-types had significantly higher maximum temperatures, significantly lower minimum temperatures and significantly higher coefficients of variation than other sites. Heithecker and Halpern (2007) similarly found that temperature differed between one hectare aggregates and intact forest sites, but they could find no correlation between temperature and previously studied plant communities (Nelson and Halpern 2005). I observed a strong similarity between temperature and beetle response to patch size, suggesting a close relationship between these biotic and abiotic variables. In addition to the probable difference in response between plants and beetles, two other factors might further explain this discrepancy. First, Heithecker and Halpern (2007) used vegetation samples taken only 1-2 years following forest harvest, and thus could detect only short-term changes in vegetation composition. However, the beetle data used in the comparisons here ranged from 3-7 years post-harvest, and thus represent a broader temporal response to harvesting treatments. Second, Heithecker and Halpern (2007) used data from aggregated retention within relatively small (13 hectare) harvest blocks, whereas this study took place within harvest blocks ranging in size from 89 to 397 hectares. Although the influence of scale on conclusions drawn from

20 aggregated retention studies is not known, scale is an important ecological attribute (Peterson et al. 1998) and thus should be considered when interpreting aggregated retention results. Clearly, my study shows that aggregated retention size can directly influence temperature, and these effects, and similar effects on other abiotic factors, could mediate biotic responses.

Indicator Species A major assumption of the IndVal technique is that the treatment groupings used in the analysis are the variables that will most strongly affect performance of the study organisms. This assumption carries the risk of classifying species as indicators of the assigned treatment when in fact they may have stronger indicator weighting for a more general grouping of factors (Dufrene and Legendre 1997, McGeoch and Chown 1998). Nonetheless, analyses are frequently conducted simply at the level of single treatments (Boudreault et al. 2002, Sanchez-Martinez and Wagner 2002, Buddie et al. 2006, Macdonald 2007). To reduce the influence of such assumptions on the findings, I analyzed the indicator values for species at each node of a hierarchical clustering analysis, as originally demonstrated by Dufrene and Legendre (1997). By highlighting where the species achieves its highest rating, we may draw conclusions about the level of generalization or specificity of particular species (Dufrene and Legendre 1997). This process therefore facilitates broader meta-analyses of the role of particular species as indicators. In this study, a large proportion of species achieved maximum indicator values at a more general level of the hierarchy than the singular factors compared across the study {i.e., clear-cut, small, medium, large, mature), a result common to previous research (Dufrene and Legendre 1997, Pohl et al. 2007). In fact, the greatest number of species achieved their highest indicator value when the study factors were split into only 3 groups. For example, some species such as Calosoma frigidum Kirby and Quedius rusticus Smetana were significant indicators of mature forest when analyzed only at the single-factor level, however they respectively achieved a higher indicator value for the groups defined as 'all deciduous' and 'all deciduous & conifer with trees' (Fig. 2.7). This suggests that these species require forest conditions in general, as opposed to specifically mature forest conditions. In addition, species such as P. decentis and A. retractum achieved their highest indicator rating at the level that included both deciduous forests as well as deciduous clear-cuts. This emphasizes the role of these species as habitat generalists that use both open habitat clear-cuts and forested treatments, conclusions which agree with their known

21 habitat preferences (Lindroth 1961-1969). Further, my results illustrate that a broader understanding of species responses to environmental factors may be achieved by analyzing the strength of indication across the groups defined by cluster analysis of biotic assemblages. This approach also clarifies limitations of the indicator species analysis itself (Pohl et al. 2008).

Management Implications Halme and Niemela (1993) documented a strong contribution of edge related ground beetle species to patches < 3 hectares within a grassland matrix. Similarly, increasing patch size can contribute to preserving bird communities in mature forest communities (Schieck and Hobson 2000, Preston and Harestad 2007). The present study supports these conclusions and although I was unable to specifically test for a size threshold relevant for epigaeic beetles, the results suggest that these will likely be between 2 and 5 ha in conifer dominated sites, and between 3 and 6 ha in deciduous dominated sites. The study also suggests that these thresholds are correlated with effects on air temperatures. My results emphasize the importance of size variability in aggregated retention in the context of long term biodiversity health (Bergeron et al. 2002). Although small patches did not maintain beetle assemblages similar to the mature forest, they did sustain some indicator taxa and are thus likely to be important for management planning that includes biodiversity concerns. Smaller patches will enhance coarse woody debris recruitment, and function as critical stepping stones to provide connectivity on a landscape (Lindenmayer and Franklin 2002). This study highlights the significance of larger aggregated retention for preserving epigaeic beetle diversity, and provides biological support to natural range of variability guidelines based on wildfire patterns (Eberhart and Woodard 1987, DeLong and Tanner 1996). The results demonstrate that larger aggregated retention will provide enhanced benefits for biodiversity conservation through preservation of mature forest species. In addition, the study supplies further documentation that although small patches offer an intermediary habitat between clear-cuts and mature forests, they will suffer from invasion by open habitat species (Spence et al. 1996). Aggregated retention plans using natural disturbances as a baseline might be improved by increasing the frequency of larger patches to provide enhanced benefits for biodiversity conservation.

22 Literature Cited Andison, D. W. 2004. Island remnants on foothills and mountain landscapes of Alberta: Part 2 on residuals. Alberta Foothills Disturbance Ecology Research Series, Report No. 6, November 1994. Foothills Model Forest, Hinton, Alberta. Attiwill, P. M. 1994. The Disturbance of Forest Ecosystems - the Ecological Basis for Conservative Management. Forest Ecology and Management 63:247-300. Bender, D. J., T. A. Contreras, and L. Fahrig. 1998. Habitat loss and population decline: A meta­ analysis of the patch size effect. Ecology 79:517-533. Bergeron, Y. and B. Harvey. 1997. Basing silviculture on natural ecosystem dynamics: An approach applied to the southern boreal mixedwood forest of Quebec. Forest Ecology and Management 92:235-242. Bergeron, Y., A. Leduc, B. D. Harvey, and S. Gauthier. 2002. Natural fire regime: A guide for sustainable management of the Canadian boreal forest. Silva Fennica 36:81-95. Boudreault, C, Y. Bergeron, S. Gauthier, and P. Drapeau. 2002. Bryophyte and lichen communities in mature to old-growth stands in eastern boreal forests of Canada. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 32:1080-1093. Bradbury, S. 2004. Understorey plant communities in boreal cutblocks with different sizes and numbers of residual tree patches. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:1220-1227. Brzustowski, J. 1999. Brzustowaski free programs, http://www2.biology.ualberta.ca /jbrzusto/cluster.php. (Accessed 27/09/2008) Buddie, C. M., D. W. Langor, G. R. Pohl, and J. R. Spence. 2006. Arthropod responses to harvesting and wildfire: Implications for emulation of natural disturbance in forest management. Biological Conservation 128:346-357. Coates, D. K. and D. J. Steventon. 1995. Patch retention harvesting as a technique for maintaining stand level biodiversity in forests of north central British Columbia. Pages 102-106 in C. R. Bamsey, editor. Innovative silvicultural systems in boreal forests, Clear Lake. DeLong, S. C. and D. Tanner. 1996. Managing the pattern of forest harvest: Lessons from wildfire. Biodiversity and Conservation 5:1191-1205. Diamond, J. M. 1976. Island biogeography and conservation - strategy and limitations. Science 193:1027-1029. Digweed, S. C, C. R. Currie, H. A. Carcamo, and J. R. Spence. 1995. Digging out the "digging- in effect" of pitfall traps: Influences depletion and disturbance on catches of ground beetles (Coleoptera: Carabidae). Pedobiologia 39:561-576. Dufrene, M. and P. Legendre. 1997. Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecological Monographs 67:345-366. Eberhart, K. E. and P. M. Woodard. 1987. Distribution of residual vegetation associated with large fires in Alberta. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 17:1207-1212. Franklin, J. F., D. R. Berg, D. A. Thornburgh, and J. C. Tappeiner. 1997. Alternative silviculatural approaches to timber harvesting: Variable retention harvest systems. Island Press, Washington, D.C. Gandhi, K. J. K., J. R. Spence, D. W. Langor, and L. E. Morgantini. 2001. Fire residuals as

23 habitat reserves for epigaeic beetles (Coleoptera : Carabidae and Staphylinidae). Biological Conservation 102:131-141. Gandhi, K. J. K., J. R. Spence, D. W. Langor, L. E. Morgantini, and K. J. Cryer. 2004. Harvest retention patches are insufficient as stand analogues of fire residuals for litter-dwelling beetles in northern coniferous forests. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:1319-1331. Gleason, H. A. 1922. On the relation between species and area. Ecology 3:158-162. Gotelli, N. J. and R. K. Colwell. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4:379-391. Haila, Y., I. K. Hanski, J. Niemela, P. Punttila, S. Raivio, and H. Tukia. 1994. Forestry and the boreal fauna - matching management with natural forest dynamics. Annales Zoologici Fennici 31:187-202. Halme, E. and J. Niemela. 1993. Carabid beetles in fragments of coniferous forest. Annales Zoologici Fennici 30:17-30. Heithecker, T. D. and C. B. Halpern. 2007. Edge-related gradients in microclimate in forest aggregates following structural retention harvests in western Washington. Forest Ecology and Management 248:163-173. Holliday, N. J. 1992. The Carabid fauna (Coleoptera, Carabidae) during postfire regeneration of boreal forest - properties and dynamics of species assemblages. Canadian Journal of Zoology-Revue Canadienne De Zoologie 70:440-452. Hunter, M. L. 1993. Natural fire regimes as spatial models for managing boreal forests. Biological Conservation 65:115-120. Lemieux, J. P. and B. S. Lindgren. 2004. Ground beetle responses to patch retention harvesting in high elevation forests of British Columbia. Ecography 27:557-566. Lindenmayer, D. B. and J. F. Franklin. 2002. Conserving forest biodiversity: A comprehensive multi-scaled approach. Island Press, Washington, D.C. Lindroth, C. H. 1961-1969. The ground-beetles of Canada and Alaska. Opuscula Entomologica (Suppl. Nos. 24,29,33,34,35). Lovei, G. L. and K. D. Sunderland. 1996. Ecology and behavior of ground beetles (Coleoptera: Carabidae). Annual Review of Entomology 41:231-256. Magura, T. 2002. Carabids and forest edge: spatial pattern and edge effect. Forest Ecology and Management 157:23-37. Magura, T., V. Kodobocz, and B. Tofhmeresz. 2001. Effects of habitat fragmentation on carabids in forest patches. Journal of Biogeography 28:129-138. MacArthur, R. H. and E. O. Wilson. 1967. The theory of Island Biogeography. Princeton University Press, Princeton, New Jersey, USA. MacDonald, S. E. 2007. Effects of partial post-fire salvage harvesting on vegetation communities in the boreal mixedwood forest region of northeastern Alberta, Canada. Forest Ecology and Management 239:21 -31. Matveinen-Huju, K. and M. Koivula. 2008. Effects of alternative harvesting methods on boreal forest spider assemblages. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 38:782-794. Matveinen-Huju, K., J. Niemela, H. Rita, and R. B. O'Hara. 2006. Retention-tree groups in clear- cuts: Do they constitute 'life-boats' for spiders and carabids? Forest Ecology and Management 230:119-135. Maxin Dallas. 2007. Thermochron iButton DS1921G. Maxim Integrated Products, Dallas

24 Semiconductor. McCune, B. and J. B. Grace. 2002. Analysis of ecologcial communities, Glenedon Beach, Oregon, USA. McCune, B. and M. J. Mefford. 1999. Pc-Ord. Multivariate Analysis of Ecological Data, Version 5. MjM Software Design, Glendedon Beach, OR. USA. McGeoch, M. A. and S. L. Chown. 1998. Scaling up the value of bioindicators. Trends in Ecology & Evolution 13:46-47. Nelson, C. R. and C. B. Halpern. 2005. Edge-related responses of understory plants to aggregated retention harvest in the Pacific northwest. Ecological Applications 15:196- 209. Newton, A. F., M. K. Thayer, J. S. Ashe, and D. S. Chandler. 2001. Staphylinidae Latreille, 1802. Pages 272-418 in R. H. Arnett, Jr. and M. C. Thomas, editors. American Beetles. CRC Press, Boca Raton, Florida, USA. Niemela, J., D. Langor, and J. R. Spence. 1993. Effects of clear-cut harvesting on boreal ground-beetle assemblages (Coleoptera, Carabidae) in western Canada. Conservation Biology 7:551-561. Oksanen, J., R. Kindt, and B. O'Hara. 2005. Vegan: community ecology package. R package version 1.6-8. Oliver, I., H. Jones, and D. L. Schmoldt. 2007. Expert panel assessment of attributes for natural variability benchmarks for biodiversity. Austral Ecology 32:453-475. Pearce, J. L. and L. A. Venier. 2006. The use of ground beetles (Coleoptera : Carabidae) and spiders (Araneae) as bioindicators of sustainable forest management: A review. Ecological Indicators 6:780-793. Pearce, J. L., L. A. Venier, G. Eccles, J. Pedlar, and D. McKenney. 2005. Habitat islands, forest edge and spring-active invertebrate assemblages. Biodiversity and Conservation 14:2949- 2969. Peterson, G., C. R. Allen, and C. S. Holling. 1998. Ecological resilience, biodiversity, and scale. Ecosystems 1:6-18. Pohl, G. R., D. W. Langor, and J. R. Spence. 2007. Rove beetles and ground beetles (Coleoptera: Staphylinidae, Carabidae) as indicators of harvest and regeneration practices in western Canadian foothills forests. Biological Conservation 137:294-307. Preston, M. I. and A. S. Harestad. 2007. Community and species responses by birds to group retention in a coastal temperate forest on Vancouver Island, British Columbia. Forest Ecology and Management 243:156-167. R Development Core Team. 2007. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Rainio, J. and J. Niemela. 2003. Ground beetles (Coleoptera : Carabidae) as bioindicators. Biodiversity and Conservation 12:487-506. Rowe, J. S. 1972. Forest regions of Canada. Department of Fisheries and the Environment, Canadian Forestry Service. Sanchez-Martinez, G. and M. R. Wagner. 2002. Bark beetle community structure under four ponderosa pine forest stand conditions in northern Arizona. Forest Ecology and Management 170:145-160. Schieck, J. and K. A. Hobson. 2000. Bird communities associated with live residual tree patches within cut blocks and burned habitat in mixedwood boreal forests. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 30:1281-1295.

25 Schmiegelow, F. K. A., C. S. Machtans, and S. J. Hannon. 1997. Are boreal birds resilient to forest fragmentation? An experimental study of short-term community responses. Ecology 78:1914-1932. Spence, J. R., D. W. Langor, J. Niemela, H. A. Carcamo, and C. R. Currie. 1996. Northern forestry and carabids: The case for concern about old-growth species (vol 33, pg 173, 1996). Annales Zoologici Fennici 33:302-302. Spence, J. R. and J. K. Niemela. 1994. Sampling carabid assemblages with pitfall traps - the madness and the method. Canadian Entomologist 126:881-894. SPSS. 2007. SPSS for Windows. SPSS Inc., Chicago, Illinois, USA. Work, T. T., J. R. Spence, W. J. A. Volney, L. E. Morgantini, and J. L. Innes. 2003. Integrating biodiversity and forestry practices in western Canada. Forestry Chronicle 79:906-916.

Hi

• *W «,, Manning

jp" Peace River

Legend Municipality: Coniferous Site: Deciduous Site:

Figure 2.1: Map of study area showing relative proximity of harvest blocks.

26 160 n

••• Deciduous I , I Conifer 120 -\

O -o N

C

> <

Clear-cut Small Medium Large Mature Treatment Figure 2.2: Mean (+1 S.E.) standardized catch of ground beetles and rove beetles within clear- cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha); within two forest cover-types. Bars with the same letters are not significantly different (alpha =0.05).

a) Deciduous b) Conifer -A

—O" Clear-cut -H- Small —Q~ Medium —&- Large —Q- Mature fY\ 500 1000 1500 2000 500 1000 1500 2000 2500 Number of individuals Number of individuals

Figure 2.3: Rarefaction-estimated species richness of ground and rove beetles within clear-cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha). Two cover-types were sampled: a) deciduous dominated stands; b) conifer dominated stands. Arrows indicate sample size used to compare estimates across treatments.

27 Clear-cut Mature Medium Coniferous Large £ Small Clear-cut Small Medium Deciduous Large ^ Mature

40% 55% 70% 85% 100% Bray-Curtis Percent Similarity

Figure 2.4: Hierarchical clustering analysis of Bray-Curtis similarity coefficients for ground beetles and rove beetles sampled in clear-cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha). Two cover-types were sampled: a) deciduous dominated stands; b) conifer dominated stands. Numbers 1-9 represent node labels for comparison to indicator species analysis (ISA).

28 a) Ground and rove beetles

Conifer o Clear-cut D Mature X Small patch Medium patch A Large patch

Deciduous • Clear-cut Mature X• Small patch * Medium patch • Large patch

•0.5 0.0 0.5 1.5 Axis 1 (62.3%) b) Rove beetles c) Ground beetles 0 15 o^ o o o

1.0 • • t 0.5 - X X • X x • X 0,0 - x a A A • + x? >*' . A 0.5 - * % 1.0 - A

-1.0 0.5 1.0 1.5 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Axis 1 (79.9%) Axis 1 (47.3%)

Figure 2.5: Non-metric multidimensional scaling (NMDS) ordinations of: a) ground and rove beetles; b) rove beetles; and c) ground beetles, sampled from: clear-cuts; mature forest; and small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha); aggregated retention patches within two dominant forest cover-types of the boreal region.

29 36 45 Tachinus elongatus BB Deciduous Tachinus frigidus I i Conifer

24 i 30 I L In 1^ In I* Clear-cut Small Medium Large Mature Clear-cut Small Medium Large Mature

Clear-cut Small Medium Large Mature Clear-cut Small Medium Large Mature Treatment Treatment Figure 2.6: Standardized catch of the six most common species collected via standard pitfall trapping within clear-cuts, mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha); and in two of the dominant cover-types (deciduous dominated and conifer dominated).

30 All stands

Conifer Clear-cut All Deciduous & Conifer With Trees Patrobus foveocollis (70.8) Tachinus frigidus (91.1) Eucnecosum brunnescens (69.8) Quedius velox (87.6) Lathrobium Washington/ (68.1) Dinothenarus pleuralis (75.9) Trechus chalybeus (66.4) Quedius rusticus (75.6) Mycetoporus americanus (65.1) Quedius brunnipennis (73.1) Tachinus canadensis (58.7) Calathus advena (71.1) Pterostichus punctatissimus (49.5) Lypoglossa franclemonti (68.9) Agonum cupreum (40.0) Tachinus elongatus (64.3) Amara erratica (40.0) Acidota crenata (32.3)

All Deciduous Conifer With Trees Platynus decentis (88.4) Calosoma frigidum (54.7) Calathus advena (85.2) Agonum retractum (84.4) Lypoglossa fungicola (52.4) Lypoglossa angularis (74.4) Tachinus fumipennis (76.8) Synuchus impunctatus (51.2) Quedius brunnipennis (74.3) Quedius labradorensis (70.5) Pterostichus pensylvanicus (49.3) Stereocerus haematopus (70.7) Calathus ingratus (59.9) Philonthus cerambus (44) Micropeplus laticollis (60.8) Ontholestes cingulatus (59.7) Pseudopsis sagitta (41.1) Pterostichus brevicornis (54.6) Bolitobius horni (58.6) Cymindis cribricollis (32) Carabus chamissonis (57.1)

Deciduous Clear-cut Deciduous With Trees Small Large & Medium Pterostichus pensylvanicus (54.6) Tachinus fumipennis (90.2) Tachyporus borealis (40.2) £ Mature Lypoglossa fungicola (52.5) Carabus chamissonis (57.2) Gabrius brevipennis (47.5) Pseudopsis sagitta (50) Carphacis nepigonensis (40) Large Medium Trechus apicalis (33.5) Ischnosoma fimbriatum (29.7) Pterostichus adstrictus (37.4) Cymindis cribricollis (34.1)

Small Medium & Large Ontholestes cingulatus (35.9) & Mature

Figure 2.7: Significant ground beetle and rove beetle indicator species (p<0.05) for each node of a hierarchical clustering analysis. Species are listed until they achieve their highest indicator value (bolded text). Samples collected by pitfall traps in clear-cuts, mature forest, and three sizes of retention patches: small (<1.8 ha); medium (2-4.4 ha); and large (>4.5 ha). Numbers 1-7 represent the corresponding hierarchical clustering node where values were derived.

31 > <

Small Medium Large Small Medium Large Mature Deciduous Treatment Conifer Treatment 0.45 B A B n=4 n=4 n=4 n=3 B •SI AB B C n=4 AB CD T T n=4 n=3 T CD _L n=4 Hn=4 D B i n=4 I n^4 .ra 0.30

!£ 0.15 \

> < Medium Large Small Medium Large Mature

Treatment Treatment Figure 2.8: Comparison of temperatures collected via Thermochron ibuttons within mature forest, and three sizes of retention patches: small (<1.4 ha); medium (1.8-4.4 ha); and large (>4.5 ha); within two dominant cover types of the boreal forest. Bars with same letters are not statistically different {alpha =0.05). Sample sizes indicated above bars.

32 Appendices Appendix 2A: Location, size, and collection details of aggregated retention patches sampled within the boreal forest of North-western Alberta, Canada. ID Code Cover Type Treatment Size (ha) Origin Harvest Size (ha) Year Harvested Block Location Year Sampled Rl-21-1 Deciduous Small 0.3 1900 307 2003 57.70°N 117,42°W 2007 R1-94-2 Deciduous Small 0.3 1880 85 2003 56.31°N 118.15°W 2007 R1-9-1 Deciduous Small 064 1890* 118 2004 56.20°N 118.22°W 2006/2007 R1-94-4 Deciduous Small 0.85 1900 85 2003 56.31°N I18.15"W 2006/2007 R1-94-3 Deciduous Small 1.2 1900 85 2003 56.311M I18.15°W 2006/2007 R2-32-1 Deciduous Medium 2.4 1890 379 2000 56.41°N 1I8.39°W 2007 R2-9-I Deciduous Medium 2.4 1900 118 2004 56,20"N I18.22°W 2006/2007 R2-2I-3 Deciduous Medium 2.55 1880 307 2002 57 70°N II7,42°W 2007 R2-21-2 Deciduous Medium 3.97 1880 307 2003 57.70"N I17.42°W 2006/2007 R2-21-1 Deciduous Medium 4.4 1880 307 2003 57.70*N 1I7.42°W 2006/2007 R3-2I-1 Deciduous Large 5,98 1880* 307 2003 57.70°N 117.42"W 2007 R3-9-I Deciduous Large 6.4 1890 118 2004 56.20°N I18.22"W 2006/2007 R3-21-2 Deciduous Large 7.53 1920* 307 2002 57.70°N 117.42°W 2007 R3-10-1 Deciduous Large 11.5 1900 110 2004 56.20°N II8.22°W 2006/2007 R3-94-1 Deciduous Large 14.1 1880 85 2003 56.31°N U8.15°W 2006/2007 R1-79-1 Conifer Small 0.54 1850 44 2000 56.4 ]°N 118.36°W 2007 Rl-14-1 Conifer Small 0.68 1900 105 2000 56.42°N 1I8.46°W 2007 Rl-31-2 Conifer Small 0.79 1850 379 2000 56.41°N 118.39°W 2007 Rl-31-1 Conifer Small 0.94 1840 379 2000 56.41°N U8.39°W 2007 Rl-14-2 Conifer Small 1.36 1900 105 2000 56.42°N I18.46°W 2007 R2-31-1 Conifer Medium 1.81 1870 379 2000 56.41°N 118.39°W 2007 R2-79-2 Conifer Medium 2.29 1870 44 2000 56.41°N 118.36"W 2007 R2-16-I Conifer Medium 2.8 1920 80 2004 56.38°N I18.40°W 2007 R2-79-I Conifer Medium 2.97 1840 44 2000 56.41"N II8 36"W 2007 R2-14-1 Conifer Medium 3.34 1850 105 2000 56.42°N 1I8.46°W 2007 R3-3I-2 Conifer Large 4.5 1840 379 2000 56.4I°N 118,39"W 2007 R3-14-1 Conifer Large 45 1900 105 2000 56,42°N 118,46°W 2007 R3-31-3 Conifer Large 4.6 1900 379 2000 56.41"N I18,39°W 2007 R3-31-I Conifer Large 5.07 1840 379 2000 56.41°N 118,39"W 2007 R3-13-I Conifer Large 5.8 1890 264 2004 56.46"N 1I8,43"W 2007 Appendix 2B: Ground and rove beetles collected via pitfall trapping within deciduous and coniferous stands. Deciduous Stands Coniferous Stands Family Species Clear-cut Small Medium Large Mature Clear-cut Small Medium Large Mature Total Carabidae Agonum cupreum Dejean 0 0 0 0 0 2 0 0 0 0 2 Agonum retractum LeConte 29 11 31 34 23 0 2 1 2 0 133 Agonum sordens Kirby 0 0 0 1 0 0 0 0 0 0 1 Amara erratica (Duftschmid) 0 0 0 0 0 8 0 0 0 0 8 Amara laevipennis Kirby 0 0 0 0 0 7 0 0 0 0 7 Amara lunicollis Schiodte 1 0 1 0 0 4 0 0 0 0 6 Amara torrida (Panzer) 0 0 0 0 0 1 0 0 0 0 1 Bembidion grapii Gyllenhal 0 0 0 1 0 1 0 0 0 0 2 Bembidion mutatum Gemminger & Harold 1 0 0 0 1 0 0 0 0 0 2 Bembidion rupicola (Kirby) 1 0 0 0 0 0 0 0 0 0 1 Calathus advena (LeConte) 2 65 10 17 35 0 252 393 259 105 1138 Calathus ingratus Dejean 174 117 156 137 121 29 6 5 26 5 776 Calosoma frigidum Kirby 2 23 1 37 63 0 0 1 0 0 127 Carabus chamissonis Fischer von Waldheim 4 69 9 13 15 1 12 1 3 1 128 Cymindis cribricollis Dejean 4 2 6 0 6 0 0 0 0 0 18 Harpalus fulvilabris Mannerheim 2 1 0 0 0 2 0 0 0 0 5 Harpalus somnulentus Dejean 1 0 0 0 0 0 0 0 1 0 2 Loricera pilicornis (Fabricius) 0 0 0 0 1 0 0 0 0 0 1 Patrobusfoveocollis (Eschscholtz) 8 13 29 17 5 19 1 1 2 1 96 Platynus decentis (Say) 48 136 188 222 154 0 0 12 6 2 768 Platynus mannerheimii (Dejean) 0 0 0 0 0 0 1 0 1 0 2 Pterostichus adstrictus Eschscholtz 577 296 161 201 124 172 16 17 18 26 1608 Pterostichus brevicornis (Kirby) 2 8 3 6 0 0 13 17 3 5 57 Pterostichus pensylvanicus LeConte 23 14 4 13 24 2 1 0 0 0 81 Pterostichus punctatissimus (Randal 1) 0 0 0 0 0 4 4 3 0 1 12 Pterostichus riparius (Dejean) 0 0 0 0 0 0 1 0 0 0 1 Stereocerus haematopus (Dejean) 3 44 4 4 11 6 95 76 57 30 330 Synuchus impunctatus (Say) 95 55 61 40 57 9 0 0 1 0 318 Trechus apicalis Motschulsky 1 6 12 1 1 10 8 4 15 4 62 Trechus chalybeus Dejean 5 20 14 10 4 28 28 12 37 12 170 Trichocellus cognatus (Gyllenhal) 1 0 0 0 0 2 0 0 0 0 3 Staphylinidae Acidota crenata (Fabricius) 2 0 2 1 1 3 1 0 0 0 10 Acidota quadrata (Zetterstedt) 18 63 22 9 17 10 43 78 49 15 324 Anotylus sobrinus (LeConte) 0 3 0 1 0 2 0 0 0 0 6

4^ Deciduous Stands Coniferous Stands Family Species Clear-cut Small Medium Large Mature Clear-cut Small Medium Large Mature Total Bisnius tereus Smetana 0 0 0 0 0 1 0 0 0 0 1 Bolitobius horni Campbell 6 16 17 9 10 0 3 1 8 4 74 Bryoporus rufescens LeConte 0 0 0 0 0 1 0 0 0 0 1 Carphacis nepigonensis (Bernhauer) 2 0 0 0 0 0 0 0 0 0 2 Dinothenarus pleuralis (Leconte) 85 156 115 295 188 11 22 24 41 27 964 Eucnecosum brunnescens (J.Sahlberg) 0 0 2 0 0 21 5 3 11 2 44 Gabrius brevipennis (Horn) 18 12 6 3 6 4 5 1 3 2 60 Gabrius picipennis (Maklin) 1 1 0 0 0 1 1 0 1 0 5 Ischnosoma fimbriatum Campbell 3 3 0 2 0 5 2 11 3 1 30 Ischnosoma splendidum (Gravenhorst) 43 43 58 42 32 16 12 13 7 10 276 Lathrobiumfauveli Duvivier 1 0 0 1 3 0 0 0 0 0 5 Lathrobium washingtoni Casey 4 2 4 3 1 11 1 1 2 0 29 Lordithon fungicola Campbell 29 6 3 11 6 1 2 0 0 1 59 Lypoglossa angular is (Maklin) 1 9 1 1 5 0 39 99 157 34 346 Lypoglossa franclemonti Hoebeke 68 55 86 86 46 0 3 8 9 8 369 Megarthrus angulicollis Maklin 1 0 0 0 3 0 2 1 0 3 10 Micropeplus laticollis Maklin 3 6 3 1 0 7 35 79 43 27 204 Mycetoporus americanus Erichson 39 11 36 18 16 39 22 5 15 7 208 Mycetoporus neotomae Fall 0 0 0 0 1 0 0 0 0 0 1 Mycetoporus smetanai Campbell 2 1 0 0 0 3 2 2 8 3 21 Olophrum consimile (Gyllenhal) 0 0 0 0 0 0 0 0 0 1 1 Olophrum rotundicollis (Say) 0 1 0 0 0 0 0 0 1 0 2 Ontholestes cingulatus (Gravenhorst) 7 76 22 71 63 3 0 0 1 0 243 Oxyporus occipitalis Fauvel 0 2 0 0 0 0 0 0 0 0 2 Oxytelusfuscipennis Mannerheim 0 0 3 6 1 0 0 0 1 1 12 Philonthus cerambus Smetana 3 24 0 30 40 0 0 0 0 0 97 Philonthus cyanipennis (Fabricius) 0 1 0 1 1 0 0 0 0 0 3 Philonthus fulcinius Smetana 0 0 0 14 0 0 0 0 0 0 14 Philonthus spiniformis Hatch 0 2 0 2 1 0 1 0 0 0 6 Philonthus varians Paykull 0 1 0 0 0 1 0 0 0 0 2 Phyllodrepa spp. 0 0 0 0 0 0 0 0 0 0 0 Pseudopsis sagitta Herman 1 24 11 11 16 1 3 2 25 1 95 Quedius brunnipennis Mannerheim 2 55 23 10 14 2 57 91 64 49 367 Deciduous Stands Coniferous Stands Family Species Clear-cut Small Medium Large Mature Clear-cut Small Medium Large Mature Total Quedius caseyi Scheerpeltz 3 9 2 0 0 0 9 7 8 0 38 Que dius fell man i (Zetterstedt) 0 0 0 0 0 0 0 0 1 0 1 Quedius frigidus Smetana 0 0 0 0 0 6 4 6 9 6 31 Quedius fulvicollis (Stephens) 4 27 21 33 17 5 8 16 22 14 167 Quedius labradorensis Blair 50 37 81 86 33 8 6 1 1 3 306 Quedius plagiatus Mannerheim 0 0 0 0 0 0 0 0 0 0 0 Quedius rusticus Smetana 8 69 42 85 86 0 2 18 32 23 365 Quedius simulator Smetana 2 0 1 2 1 2 0 0 2 0 10 Quedius velox Smetana 20 50 67 68 53 2 80 125 140 87 692 Scaphium castanipes Kirby 0 4 1 15 0 0 1 0 0 0 21 Staphylinus capitatus Bland 3 0 1 2 4 0 0 0 1 0 11 Stenocrepis mammops Casey 0 0 0 0 0 0 0 0 1 0 1 Stenus austini Casey 2 2 2 2 0 1 5 3 4 1 22 Stenus sibiricus J.Sahlberg 0 0 0 0 0 0 0 0 1 0 1 Tachinus basalis Erichson 4 21 6 14 8 0 14 3 28 1 99 Tachinus contortus Hatch 1 1 0 2 0 0 0 0 0 2 6 Tachinus elongatus Gyllenhal 217 260 299 461 454 25 38 55 78 29 1916 Tachinus frigidus Erichson 11 35 119 164 116 0 37 262 410 242 1396 Tachinus fulvicollis 0 0 0 0 4 0 0 0 0 0 4 Tachinus fumipennis (Say) 1 30 113 173 195 0 0 1 8 2 523 Tachinus nigricornis Mannerheim 1 1 0 1 2 0 0 0 1 0 6 Tachinus quebecensis Robert 0 1 4 11 9 1 1 1 11 3 42 Tachinus tachyporoides Horn 0 0 3 0 0 1 1 0 2 1 8 Tachyporus borealis Campbell 11 6 9 7 4 28 40 2 14 9 130 Tachyporus canadensis Campbell 2 0 0 0 0 8 0 0 0 0 10 Chapter 3: Ground and rove beetle responses to retention patch isolation: the value in considering rarity

Introduction Maintaining habitat connectivity as a tactic to enhance population recovery is a central goal of recent forest management strategies (Franklin et al. 1997, Lindenmayer and Franklin 2002). Natural disturbance management (NDM) aims to assist recovery by emulating the patterns and processes of natural disturbances on a landscape (Hunter 1993, Bergeron et al. 2002). One element of NDM is leaving aggregated retention patches (i.e., isolated patches of live trees), which are designed to emulate fire skips within a disturbed area (DeLong and Tanner 1996, Gandhi et al. 2004). Such retention patches are hypothesized to increase connectivity among residual stands by functioning as habitat islands, and may also provide source populations to recovering harvest blocks (Franklin et al. 1997, Gandhi et al. 2001). However, little is known about the effect of spatial patterns, such as degree of isolation of retention patches from interior forests, on the long-term viability of biotic assemblages within these patches. Connectivity among forest patches is particularly important for rare species, which are known to be sensitive to increased fragmentation (Davies et al. 2000, Henle et al. 2004). This sensitivity may be linked to limited dispersal abilities (Kunin and Gaston 1993), low background population numbers (Henle et al. 2004), or a host of other factors that may affect rare species over small spatial scales (Kunin and Gaston 1993). Because of this, Hannon et al. (2004) noted that rare species are the most likely to be overlooked by coarse-filter management approaches such as NDM. Unfortunately, rare species are also known to represent the dominant component of species richness in most taxa (Magurran and Henderson 2003). Combined, these statements raise concern about long term persistence of rare species within harvested landscapes. Despite their well-known sensitivity to fragmentation, few studies have effectively evaluated the response of rare species to NDM. In fact, authors often remove rare species from analyses to eliminate excess 'noise' in the data set (McCune and Grace 2002), and to take out seemingly 'redundant' species (Marchant 1999). These practices may enhance the ability of multivariate statistics to show patterns, but they are questionable on biological grounds (Slocomb and Dickson 1978). Removal of rare species makes three major assumptions critical to conservation of overall biodiversity: 1) rare species, because of their scarcity, are largely irrelevant to the outcomes of an analysis; 2) the biological traits and responses of rare species are

37 correlated to those of more common species and, therefore, do not require consideration; and 3) rare species are unimportant ecologically and thereby of little conservation concern. The first assumption has some purely statistical basis (McCune and Grace 2002). The second assumption is at odds with previous ecological literature. In fact, it is well understood that common and rare species may exhibit very different biological traits (Kunin and Gaston 1993), and respond differently to habitat fragmentation (Davies et al. 2000, Henle et al. 2004). In addition, the responses of common species can often be correlated to each other and not to rarer species (Anderson and Robinson 2003, Anderson 2004). The third assumption simply asserts the unwarranted view that whole-system biodiversity is of little value. Therefore, despite the challenges in working with rare species, it would appear that more focus on them is required when analyzing ecological responses to forest management alternatives. Application of such methods may increase our understanding of whole community responses to NDM, and of processes associated with rarity, and facilitate robust meta-analyses of studies in the future. An additional motivation for excluding rare species in multivariate analysis is the potential biases resulting from sampling (Legendre and Gallagher 2001). While these challenges stem from biological characteristics such as aggregated distributions (Kunin and Gaston 1993), or limitations of sampling methods themselves, there is still ample potential to explore rare species within ecological studies. One option is to vary the degree of influence of rare species in community analyses through data transformations and relativizations, and compare outcomes of analyses with low and high weightings for rare species (Kunin and Gaston 1993). Subsequent comparisons of the biology of rare species most influential in these community analyses could then be used to ameliorate concerns of sampling biases. I sampled ground beetle (Coleoptera: Carabidae) and rove beetle (Coleoptera: Staphylinidae) communities to study the role of retention patch isolation on community composition of common and rare species in the boreal forest of Alberta, Canada. Ground and rove beetles respond directly to forest harvesting practices (Spence et al. 1996, Buddie et al. 2006, Pohl et al. 2007), and comprise an ecologically heterogeneous group of flying and flightless species (Lindroth 1961-1969, Newton et al. 2001) that are highly suitable for studies of connectivity. Species abundance curves of ground and rove beetles also follow classic logarithmic distributions, and thus provide ample opportunity to contrast responses of rare and common species to retention patch isolation. While controlling for retention patch size and

38 dominant tree species, I aimed to answer the following three questions: 1) Does distance of retention patches from the mature, continuous forest alter their constituent beetle assemblages?; 2) Are rare species more sensitive to patch isolation than common species?; and 3) Can increased weighting of rare species in community analyses be justified on biological grounds?

Materials and Methods Study Design This study was conducted in north-western Alberta, Canada within the lower foothills ecoregion of the boreal forest, approximately 98 km north-west of Peace River (56.41°N 118.39°W). The forest overstory was dominated by white spruce (Picea glauca (Moench)), and included lodgepole pine (Pinus contorta Douglas), black spruce {Picea mariana (Miller)), and trembling aspen (Populus tremuloides Michaux) as more minor elements. Understory vegetation was mainly moss species, with green alder (Alnus crispa (Aiton)), river alder (Alnus tenufolia Nutt.) and prickly rose (Rosa acicularis Lindl.) present in most of the patches. The patches available for study occurred predominantly within a single clear-cut of 397 hectares (ha), but one additional patch was selected from a 105 ha harvest block located approximately five kilometres west of the main study site. Although this lack of replication at the harvest block level may constrain generalization of findings, it was the best comparison possible in this landscape, where adoption of aggregated retention in clear-cuts is recent. On the other hand, the close geographic proximity of almost all patches provided an opportunity to control variables such as weather and surrounding landscape diversity while attempting to elucidate patterns associated with degree of patch isolation. Beetles were sampled in patches located at three distance classes from a mature forest matrix: near (10-55 m), mid (65-125 m) and far (140-360 m). Because patch size influences structure of beetle assemblages, I selected patches ranging between 0.7-1.7 ha, a size previously identified as insufficient for preserving mature forest assemblages (Chapter 2). Patches in this size range should be sensitive to isolation patterns so that early isolation effects on mature forest assemblages become more apparent. Beetle samples were also collected from the surrounding clear-cut and mature forest to compare with those from retention patches. Four replicates were sampled for each of the patch distance classes and for clear-cut and mature forest.

39 Retention patches were composed of merchantable trees more than 122 years old and were similar in species composition to the surrounding forest removed at harvest. Mature forest stands of pyrogenic origin and with trees > 117 years old were selected from the forest matrix surrounding the harvested area. The cut blocks were seven years post harvest when sampled, and had been planted with white spruce and lodgepole pine seedlings within one year of harvest.

Data Collection Beetles were collected with continuous pitfall trapping during the snow free season from May 15th to August 22nd, 2007. Traps contained approximately 30 mL of silicate-free ethylene glycol as a preservative, and were covered with an elevated plastic lid to reduce accumulation of debris and rainfall in the traps (Spence and Niemela 1994). Four traps were placed randomly with respect to microsites in all sample locations, and were spaced a minimum of 15 m apart to reduce depletion effects on trap catches (Digweed et al. 1995). Traps in retention patches were placed in the central core and kept at least 20 m from the edge. Traps in the mature forest and clear-cuts were placed along linear transects and were more than 60 m (ca. two tree lengths) from any disturbed or forested edge. Beetles were identified to species using Lindroth (1961-1969) for ground beetles, and Newton et al. (2001) and references therein for rove beetles. Immature specimens were excluded from analyses as identification beyond the family level was not possible. Within the diverse rove beetle sub-family Aleocharinae, only Lypoglossa franclemonti Hoebeke and Lypoglossa angularis (Maklin) could be reliably identified, and these species were included in all analyses. The remaining Aleocharinae data were only included in analyses of overall catch because reliable species level identification was not possible.

Data Analyses Data for all four traps within a site (patch, clear-cut, forest) were pooled across collection dates before being subjected to analyses. As traps experienced various degrees of disturbance resulting in occasional sample loss, the seasonal catch at each site was standardized to 100 trap days before analyses, with the exception that rarefaction analyses used non-standardized data. Total standardized catch of beetles was compared among the isolation classes, mature forest sites, and clear-cuts using a one-way ANOVA in SPSS 15.0 (SPSS 2007). Because rare species may be more sensitive to patch isolation (Davies et al. 2000), standardized catches were

40 graphically displayed by showing both the proportion of rare and common species in each class. I defined a rare species as one for which the standardized total catch across all traps and sites was <5% of the overall catch (McCune and Grace 2002). Unidentified Aleocharinae were included in the rare category as their combined catch was <5% of the overall catch. Species richness was estimated for each site using individual-based rarefaction in the Vegan package (Oksanen et al. 2005) in R 2.6 (R Development Core Team 2007). Rarefaction serves to standardize species richness comparisons according to the number of individuals sampled at a site (Gotelli and Colwell 2001). In addition, by taking into account both species richness and abundance, rarefaction estimates can be interpreted as a diversity measure. Richness estimates were compared at the highest observed catch common to all sites (373 individuals) to standardize for sample size. Community analyses were used to determine the relative influence of patch isolation when rare species were given both low and high weighting. To conduct these analyses, standardized abundances were first log-transformed which constituted the Tow' weighting for rare species. Although log transformation does increase the relative contribution of rare species, these species still represent a comparatively low influence on the analysis results because the general abundance patterns are preserved (McCune and Grace 2002). The 'high' weighting of rare species was conducted by relativizing the data by site totals and then by species totals, a process similar to the inverse of the commonly used Wisconsin double standardization (Bray and Curtis 1957). To achieve this, sites were first rescaled to produce a total site sum of one, and each species was then also subjected to the same rescaling to produce a total species sum of one. This resulted in total catch of each rare and common species being equal; however, because rare species occurred in only a few sites their relativized catch at each of these sites was greater than that of the common species that were present in many sites. This relativization eliminated the imbalance between abundance of rare and common species and facilitated a test of the influence of isolation on the rare species. Because incidental catches might lead to non-resident species being classified as rare species, I deleted any species that were present in fewer than four of the 20 sample sites (i.e., <20%) prior to relativizing the data. This removed 23 species representing approximately <0.2% of the overall catch and constrained the analysis to the remaining 39 species. All multivariate analyses were then conducted using both the transformed and relativized data sets to directly compare the responses of rare and common species.

41 Hierarchical clustering analyses were first used to assess the general differences between rare and common beetle species to patch isolation. The analyses were based on a Bray-Curtis similarity matrix and were constructed using web-based software of Brzustowski (1999). Distances between groups were assessed using unweighted arithmetic averages, defined as the average distance between a sample in group A, and a sample in group B. Non-metric multidimensional scaling (NMDS) ordinations were used as an indirect gradient analysis and were constructed using the Bray-Curtis index on both data sets. NMDS was selected as it provides strong visualization of the variability between sites and works to maximise the variance explained by each axis, while minimizing the stress value of the overall ordination (McCune and Grace 2002). This stress value in turn determines how well the ordination picture represents the original data set (McCune and Grace 2002). The ordination was constructed using a random starting configuration, followed by 50 iterations of real data. Significance of the ordination was assessed against 100 randomized iterations of data using a Monte Carlo analysis. Analyses were conducted using PC-Ord Version 5 (McCune and Mefford 1999). Canonical analysis of principal co-ordinates (CAP) provided a direct gradient analysis to test the hypothesis that patch isolation influenced beetle assemblages. The technique aims to find the maximum separation between these pre-defined classes (i.e., the isolation classes). Thus, the relative distances between classes and not the overall variability within classes is the point of interpretation (Anderson and Willis 2003). In addition, the CAP analysis is less sensitive to correlations between common species, and instead provides greater weight to those species which present unique patterns in the data set. CAP analyses were conducted using a FORTRAN program (Anderson and Willis 2003, Anderson 2004), and the Bray-Curtis similarity index. The number of axes used in the canonical analysis (m-value) was selected by the program and ordination significance was evaluated by a permutation test using 9999 random permutations. The CAP analysis also facilitates the interpretation of single species patterns by analyzing the correlation of each species with the CAP axis (Anderson and Willis 2003). Because the pattern of patch isolation was most evident on the second axis of the relativized data, species with a positive correlation >0.20 with this axis were selected for comparison of biological traits and habitat affinities. Relativized catch of the 6 species with the strongest correlations were then

42 graphed to visualize the individual species responses to patch isolation. All analyses were conducted strictly at the species level (see Spence et al. 2008).

Results General Results A total of 3,516 individuals representing 62 species (44 staphylinids, 18 carabids) were collected during the summer of 2007 (Appendix 3A). The most common ground beetle was Calathus advena (LeConte) (n=873), while the most common rove beetles were Tachinus frigidus Erichson (n=324), and Quedius velox Smetana (n=300). Thirty-one species were represented by fewer than 10 specimens, and among these 12 were represented by only a single individual.

Beetle Catch and Species Richness Total average standardized catch of beetles was lower in the clear-cuts than in forested sites (Fig. 3.1), a difference which was only marginally not significant (F=2.7, df=4,/?=0.071). Although the total catch in retention patches showed little sensitivity to isolation, a slight decrease in rare species catch was observed within the far patches (Fig. 3.1). Interestingly, while the catch of rare species in far patches approached the level of clear-cuts, the catch of common species in the far patches did not decrease similarly. Rarefaction-estimated species richness standardized to 373 individuals was similar between the mature forest and retention patch classes but was significantly higher in clear-cuts (Fig. 3.2). Despite overall similarity between the retention patches and the mature forest, species richness decreased slightly in the far patches, with an average of 1.3 species fewer than in the other forested classes. Species richness in the clear-cut class was an average of 10 species greater than the patch and mature forest classes (Fig. 3.2).

Community Analyses Hierarchical clustering analyses for the transformed and relativized data sets indicated a high influence of rare species weighting on final results. The analysis of log-transformed abundances suggested high similarity (i.e., >85%) between all retention classes and the mature forest, while clear-cuts had the lowest similarity to the mature forest (Fig. 3.3a). In contrast, when the weight of rare species was increased using relativization, the assemblages of far patches were much less similar to the mature forest at 72% (Fig. 3.3b), compared to ca. 85% for the transformed data

43 (Fig. 3.3a). Meanwhile, near patches and medium patches maintained 81% and 77% similarity to the mature forest, respectively, in the analyses of relativized data (Fig. 3.3b). The clear-cuts again had the lowest similarity to the mature forest class. NMDS ordination produced a two dimensional solution for the transformed data and a three dimensional solution for the relativized data. However, to facilitate direct comparisons, only two dimensional solutions were used. The ordinations achieved a final stress of 11.0 for the transformed data and 19.2 for the relativized data, both of which were significantly lower than expected from random data (Monte Carlo test, n=T00,/?=0.009). Total variance explained by the ordinations was 94.3% for the transformed data, and 77.6% for the relativized data. Consistently greater proportions of the variance were explained by axis 1 (Fig. 3.4). MRPP analyses determined that groupings within classes were significantly stronger than would be expected by chance for both ordinations (pO.OOl). Results of the MRPP analyses were driven by the clear- cut class which grouped out strongly from all other classes in both ordinations (Fig. 3.4). There was little observed effect of isolation on community structure in the NMDS ordination of log-transformed abundances, with a high degree of overlap in each of the retention patch distance classes (Fig. 3.4a). The influence of isolation on community patterns, however, became a little more apparent in the ordination of relativized abundances, despite an increased amount of variability between points (Fig. 3.4b). In this analysis, the mature forest and near distance points clustered mostly to the top right side of the ordination, whereas the mid and far distance classes clustered predominantly to the bottom and centre of the ordination picture (Fig. 3.4b). The direct gradient analysis provided by the CAP ordination gave further support to the influence of isolation on rare species. The ordination of log-transformed abundances had a successful classification rating of 30% using the leave-one-out technique (Anderson and Robinson 2003), and was highly significant (/?<0.001). The ordination of relativized abundances had a successful classification rating of 40%, and was also highly significant (/?=0.0017). Similar to the NMDS results, the mature forest and retention classes overlapped considerably in the CAP ordination on log-transformed abundances (Fig. 3.5a). The influence of isolation, however, was clearly evident along axis 2 in the CAP analysis of relativized abundances as was evidenced by the difference in placement between the far patches and other forested classes (Fig. 3.5b). Similar to the clustering analysis, the CAP ordination suggested a close resemblance

44 between the mature forest and mid and near patches, while the far patches clustered away from these classes (Fig. 3.5b).

Single Species Comparisons Nine species had positive correlations >0.2 with CAP axis 2 (i.e., displaying reduced catch in the far patch class) (Table 3.1). Despite the challenges of categorizing rare species, most species appeared to be associated with forest attributes expected to have special characteristics in mature forests, such as forest floor debris (Table 3.1). The six species most highly correlated with the CAP axis 2 showed reduced catch in the far patch class, and were all rare (i.e., <5% of catch) in this study. Quedius caseyi Scheerpeltz is associated with forest floor debris and was strongly associated with near and mid patches (Fig. 3.6). Bolitobius horni Campbell, commonly found on fungi and in forest floor debris, had a reduced catch in the far patches despite having abundances similar to the mature forest in both the mid and near patches (Fig. 3.6). Quedius fulvicollis (Stephens), an epigaeic species, and Carabus chamissonis Fischer, an epigaeic, predatory species, both showed decreased catches in the far patch class with C. chamissonis being completely absent from that class. Tachinus quebecensis Robert, a possible fungivore, showed an increased sensitivity to patch distance with lower catch in the mid and far classes, while Pseudopsis sagitta Herman, which is associated with forest floor debris, was completely absent from the mid and far classes.

Discussion There is no denying that there is uncertainty with respect to identifying rare species that are associated with particular habitats, as opposed to reflecting incidental catches, and this has complicated analysis of truly rare taxa. However, without taking steps to better understand patterns in distribution of rare species, they will remain generally ignored in biodiversity analyses despite the fact that they represent the lion's share of most invertebrate assemblages (Spence et al. 2008). By increasing the weight of rare species in multivariate analyses, I found a stronger response to patch isolation than would have been observed using only traditional analytical methods. These responses lend support to previous conclusions that rare species are especially sensitive to landscape fragmentation (Davies et al. 2000, Henle et al. 2004), and suggests that rare species are important subjects for community analyses.

45 Impact of Patch Isolation The impacts of disturbance on species catch, such as those observed in clear-cut areas, can have long term consequences for species recovery in harvested landscapes (Buddie et al. 2006). Such impacts also affect communities in small retention patches within the clear-cuts (Bradshaw 1992), and it is expected that impacts would increase with increasing patch isolation with respect to the surrounding mature forest (Burke and Goulet 1998). In the current study, rare species showed this response with reduced catch in the far {i.e., most isolated) patches; however, common species did not exhibit this sensitivity to isolation. This difference in degree of response to disturbance by rare and common species is also reported from previous studies on habitat fragmentation (Davies et al. 2000, Henle et al. 2004). The slight reduction in species richness in far patches compared to other forested sites is noteworthy, especially considering that retention patches are embedded in extensive clear-cut areas where on average species richness is 10 species higher than in the patches. In chapter 2,1 found that assemblages of small patches were the most similar to those of clear-cuts, suggesting that increased disturbance leads to increased invasion of species from the clear-cut habitat. Therefore, it was expected that the small, isolated far patches would suffer from a similar invasion of open-habitat species; however, this phenomenon was not observed, and, in contrast, a reduction in species richness was documented. This pattern, although not significant, suggests that species are lost in patches of increasing isolation from mature forest source populations (Burke and Goulet 1998, Jennings and Tallamy 2006). Community analyses of the beetle assemblages also demonstrated increased sensitivity of rare species to patch isolation. By increasing the weight of rare species in cluster analyses, isolation became more influential in the far patches, and similarity of assemblages in far patches fell to 72% of the mature forest. This pattern appeared to be driven by abundance patterns of rare species such as T. quebecensis and P. sagitta, as well as more common species like Q. brunnipennis. Thus, the increased weight of species with low and intermediate abundances enabled a greater detection of isolation impacts. Without these relativizations, these species patterns would have been masked by those of the common species which exhibited little response to patch isolation. Shifts in similarity between far patches and the mature forest were also observed in the ordinations with high weightings of rare species. Despite increased variability in the NMDS

46 ordination of relativized data, there was a greater separation between the mature forest and the mid and far patches. In addition, there was an increased similarity between the far patches and the clear-cuts, a pattern driven by rare species such as Ontholestes cingulatus (Gravenhorst) which was only present in the far and clear-cut classes. Such evidence suggests there may be a low level of species invasion occurring within the far patches (Spence et al. 1996). The CAP analysis, by being more sensitive to species which exhibited unique abundance patterns, allowed further analysis of species patterns which otherwise may have been overshadowed by the most common species. As a result there was a strong separation of the far patches from the mature forest, and other retention classes in the analysis of relativized data. These patterns provide further evidence that rare species are more sensitive to patch isolation than common species. Despite the challenges of analyzing data about rare species, the known habitat affinities of the less common species justified their increased weighting in the multivariate analyses. Quedius brunnipennis is strongly associated with forest floor debris, was associated with conifer forests in my previous study (Chapter 2), and exhibited a strong response to the isolation gradient with reduced catch in the far patches. A similar response was shown by C. chamissonis which has also previously been collected within conifer forests (Jacobs et al. 2008) and is associated with forest litter (Table 3.1). Furthermore, five of the rare species (I. fimbriatum, P. sagitta, Q. caseyi, Q. julvicollis, and T. quebecensis) also exhibit habitat affinities, such as litter dwelling, which make them conducive to pitfall trapping (Table 3.1), and appear to be associated with conifer forests (Chapter 2, Jacobs et al. 2008). These species were also found to be rare or non­ existent in a nearby study which used window traps to sample beetle diversity (Jacobs et al. 2007), supporting the idea that the observed rarity patterns are not a sampling artefact. From current knowledge, it is plausible that the remaining two species (B. horni, and I. splendidum) may be associated with deciduous forests (Chapter 2, Jacobs et al. 2007). Thus, their rarity in this study may be an artefact of the cover-type studied. However, considering the heterogeneous nature of the mixedwood boreal forest, it is still important to consider their responses in the context of retention patches, which are intended to serve as landscape elements to provide source populations in clear-cut recovery. Thus in general, it appears the approaches used to increase the weighting of rare species are justified based on the known biological characteristics of species considered here.

47 The results documented here suggest that forest managers should consider retention patch isolation when designing harvest blocks, particularly for the conservation of rare species. Furthermore, 'species relaxation' (Saunders et al. 1991), suggests that the relative impact of fragmentation, such as patch isolation, may increase over time as vulnerable species continue to drop out of the system. Because one of the roles of retention patches on a landscape is to function as long-term source populations for clear-cut recovery, the consideration of species relaxation is critical. In addition, because rare species themselves are characterized by patchy distributions, and/or limited dispersal abilities (Kunin and Gaston 1993), it is important to maximize their conservation within retention patches and ensure their continued persistence through future forest rotations. One possible option for ameliorating isolation effects would be to use larger patches when retention is required at distances greater than 140 metres, the distance observed to influence species in this study, because large (> 2-3 ha) patches better preserve epigaeic beetle assemblages of mature forest communities (Chapter 2), and may be less sensitive to disturbances (Saunders et al. 1991). However it is important to note that connectivity may be greatly enhanced by variable placement of large patches, and thus large patches should not simply be used in the most isolated locations within a clear-cut, but rather applied with some variability.

Challenges of Analyzing Rarity This study presents a clear example of how ignoring rare species in forest management research produces a limited perspective of ecosystem responses to harvesting. Although the approach taken in this study of relativizing the data matrix requires assumptions about the true rarity of the upweighted species (i.e., that the species are actually rare and not just rarely sampled), these actions appear justified by both the careful process of species removal employed, and the corresponding biological affinities of the most influential species. In addition, the relativization procedure enabled a true analysis of patterns exhibited by rare species that were otherwise masked by the common species. The assumptions also seem less restrictive than those made in other community ecology studies, which commonly remove species representing <5% of the total catch and assume that the species removed are well represented by the most common species. In this study for example, this would have meant analyzing the response of just 5 of 62 species collected. Similarly in chapter 2, this would have amounted to the analysis of just 5 out

48 of 91 species; three being the same as this study despite different research questions and a different range of cover-types. The increased variability, higher stress, and lower total variance explained in the NMDS ordination of the rare species data is a clear example of the challenges community ecologists face when analyzing rarity. Such challenges are attributed to the increased influence of species with a high number of zero abundance values (McCune and Grace 2002). However, this increased variability seems justified by the broader understanding of biotic responses, and increased grounding in biological reality which it facilitates. The analyses presented here are not advocated as a singular solution to challenges of analyzing rarity, but are meant to shed light on the potential biological signals which may be overlooked when analyzing only the most common species in a data set.

Conclusions Isolation appears to be an important consideration for forest managers wishing to conserve biodiversity under the principles of natural disturbance-based management. The patterns observed in rare species responses to isolation suggest that patches greater than 140 m from the intact forest experience isolation impacts. Such challenges can therefore be ameliorated through the use of larger retention patches, and spatial designs which maximize connectivity in a clear- cut. This study also highlights the potential importance of rare species in evaluations of natural disturbance management. It is acknowledged throughout the ecological literature that rare species may respond differently to fragmentation than common species, and this possibility has been underscored by this study. Ecologists should consider the relative contributions of rare species and work towards their inclusion in community level studies. Rare species present one of the largest challenges to community ecologists, but only once we take strides towards a broader understanding of their influences in ecosystems, may we begin to truly understand their importance. And, of course, rare species contribute significantly to the biodiversity that we seek to conserve.

Literature Cited Anderson, M. J. 2004. CAP: a FORTRAN computer program for canonical analysis of principal coordinates: User Guide. Department of Statistics, University of Auckland, New Zealand.

49 Anderson, M. J. and J. Robinson. 2003. Generalized discriminant analysis based on distances. Australian & New Zealand Journal of Statistics 45:301-318. Anderson, M. J. and T. J. Willis. 2003. Canonical analysis of principal coordinates: A useful method of constrained ordination for ecology. Ecology 84:511-525. Bergeron, Y., A. Leduc, B. D. Harvey, and S. Gauthier. 2002. Natural fire regime: A guide for sustainable management of the Canadian boreal forest. Silva Fennica 36:81-95. Bradshaw, F. J. 1992. Quantifying edge effect and patch size for multiple-use silviculture - a discussion paper. Forest Ecology and Management 48:249-264. Bray, J. R. and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27:326-349. Brzustowski, J. 1999. Brzustowaski Free Programs, http://www2.biology.ualberta.ca /jbrzusto/cluster.php. (Accessed 27/09/2008) Buddie, C. M., D. W. Langor, G. R. Pohl, and J. R. Spence. 2006. Arthropod responses to harvesting and wildfire: Implications for emulation of natural disturbance in forest management. Biological Conservation 128:346-357. Burke, D. and H. Goulet. 1998. Landscape and area effects on beetle assemblages in Ontario. Ecography 21:472-479. Campbell, J. M. 1991. A revision of the genera Mycetoporus Mannerheim and Ischnosoma Stephens (Coleoptera: Staphylinidae: Tachyporinae) of North and Central America. Memoirs of the Entomological Society of Canada 156:80-107. Davies, K. F., C. R. Margules, and K. F. Lawrence. 2000. Which traits of species predict population declines in experimental forest fragments? Ecology 81:1450-1461. DeLong, S. C. and D. Tanner. 1996. Managing the pattern of forest harvest: Lessons from wildfire. Biodiversity and Conservation 5:1191-1205. Digweed, S. C, C. R. Currie, H. A. Carcamo, and J. R. Spence. 1995. Digging out the "digging- in effect" of pitfall traps: Influences depletion and disturbance on catches of ground beetles (Coleoptera: Carabidae). Pedobiologia 39:561-576. Franklin, J. F., D. R. Berg, D. A. Thornburgh, and J. C. Tappeiner. 1997. Alternative silviculatural approaches to timber harvesting: Variable retention harvest systems. Island Press, Washington, D.C. Gandhi, K. J. K., J. R. Spence, D. W. Langor, and L. E. Morgantini. 2001. Fire residuals as habitat reserves for epigaeic beetles (Coleoptera : Carabidae and Staphylinidae). Biological Conservation 102:131-141. Gandhi, K. J. K., J. R. Spence, D. W. Langor, L. E. Morgantini, and K. J. Cryer. 2004. Harvest retention patches are insufficient as stand analogues of fire residuals for litter-dwelling beetles in northern coniferous forests. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:1319-1331. Gotelli, N. J. and R. K. Colwell. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4:379-391. Hannon, S. J., S. E. Cotterill, and F. K. A. Schmiegelow. 2004. Identifying rare species of songbirds in managed forests: application of an ecoregional template to a boreal mixedwood system. Forest Ecology and Management 191:157-170. Henle, K., K. F. Davies, M. Kleyer, C. Margules, and J. Settele. 2004. Predictors of species sensitivity to fragmentation. Biodiversity and Conservation 13:207-251. Hunter, M. L. 1993. Natural Fire Regimes as Spatial Models for Managing Boreal Forests. Biological Conservation 65:115-120.

50 Jacobs, J. J., T. T. Work, and J. R. Spence. 2008. Correcting for detection biases in the pitfall trapping of ground beetles (Coleopetera: Carabidae). Pages 425-450 in Back to the Roots or Back to the Future? Towards a New Synthesis between Taxonomic, Ecological and Biogeographical Approaches in Carabidology, Proceedings of the 13th European Carabidologists Meeting, Blagoevgrad, Bulgaria, 20-24 August 2007. . Jacobs, J. M, J. R. Spence, and D. W. Langor. 2007. Influence of boreal forest succession and dead wood qualities on saproxylic beetles. Agricultural and Forest Entomology 9:3-16. Jennings, V. H. and D. W. Tallamy. 2006. Composition and abundance of ground-dwelling Coleoptera in a fragmented and continuous forest. Environmental Entomology 35:1550- 1560. Kunin, W. E. and K. J. Gaston. 1993. The biology of rarity - patterns, causes and consequences. Trends in Ecology & Evolution 8:298-301. Legendre, P. and E. D. Gallagher. 2001. Ecologically meaningful transformations for ordination of species data. Oecologia 129:271-280. Lindenmayer, D. B. and J. F. Franklin. 2002. Conserving forest biodiversity: A comprehensive multi-scaled approach. Island Press, Washington, D.C. Lindroth, C. H. 1961-1969. The Ground-beetles of Canada and Alaska. Opuscula Entomologica (Suppl. Nos. 24,29,33,34,35). Magurran, A. E. and P. A. Henderson. 2003. Explaining the excess of rare species in natural species abundance distributions. Nature 422:714-716. Marchant, R. 1999. How important are rare species in aquatic community ecology and bioassessment? A comment on the conclusions of Cao et al. Limnology and Oceanography 44:1840-1841. McCune, B. and J. B. Grace. 2002. Analysis of ecologcial communities, Glenedon Beach, Oregon, USA. McCune, B. and M. J. Mefford. 1999. Pc-Ord. Multivariate Analysis of Ecological Data, Version 5. MjM Software Design, Glendedon Beach, OR. USA. Newton, A. F., M. K. Thayer, J. S. Ashe, and D. S. Chandler. 2001. Staphylinidae Latreille, 1802. Pages 272-418 in R. H. Arnett, Jr. and M. C. Thomas, editors. American Beetles. CRC Press, Boca Raton, Florida, USA. Oksanen, J., R. Kindt, and B. O'Hara. 2005. Vegan: community ecology package. R package version 1.6-8. Pohl, G. R., D. W. Langor, and J. R. Spence. 2007. Rove beetles and ground beetles (Coleoptera: Staphylinidae, Carabidae) as indicators of harvest and regeneration practices in western Canadian foothills forests. Biological Conservation 137:294-307. R Development Core Team. 2007. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Saunders, D. A., R. J. Hobbs, and C. R. Margules. 1991. Biological consequences of ecosystem fragmentation - a Review. Conservation Biology 5:18-32. Slocomb, J. and K. L. Dickson. 1978. Estimating the total number of species in a biological community. Pages 3-52 in J. K. Dickson, J. Cairns, and R. J. Livingston, editors. Water pollution assessment: Quantitative and statistical analyses. American society for testing and materials. Smetana, A. 1971. Revision of the tribe Quediini of America north of Mexico (Coleoptera: Staphylinidae). Memoirs of the Entomological Society of Canada 79:1-303. Spence, J. R., D. W. Langor, J. M. Jacobs, T. T. Work, and W. J. A. Volney. 2008. Conservation

51 of forest-dwelling arthropod species: simultaneous management of many small and heterogeneous risks. Canadian Entomologist 140:510-525. Spence, J. R., D. W. Langor, J. Niemela, H. A. Carcamo, and C. R. Currie. 1996. Northern forestry and carabids: The case for concern about old-growth species (vol 33, pg 173, 1996). Annales Zoologici Fennici 33:302-302. Spence, J. R. and J. K. Niemela. 1994. Sampling Carabid assemblages with pitfall traps - the madness and the method. Canadian Entomologist 126:881-894. SPSS. 2007. SPSS for Windows. SPSS Inc., Chicago, Illinois, USA. Wheeler, Q. and M. Blackwell, editors. 1984. Fungus-insect relationships: Perspectives in ecology and evolution. Columbia University Press.

+ 2.0 A o o -o N

o

Mature Clear-cut Treatment class Figure 3.1: Standardized catch (+ 1 S.E.) of rare and common ground beetles and rove beetles within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest. Rare species are those representing <5% of the total catch.

52 J\J w C/3 ^ 45 - T ^-^ r- o richnes s

"§ 35 - OH T T J.^ •:'-•'••:. t/2 Ti *

o • *. Estimate d T<\ - Zj i i • i i i Mature Near Mid Far Clear-cut Forest harvest class Figure 3.2: Average species richness (+1 S.E.) for ground beetles and rove beetles. Species richness estimates were generated via individual based rarefaction standardized to 373 individuals for clear-cuts, and mature forest, and three retention patch distance classes: near (1 55 m); mid (65-125 m); and far (140-360 m); from the mature forest. Bray-Curtis Percent Similarity

100% 75% 50% 25% i i i

a) Log transformed Clear-cut Far Mid Near h Mature

b) Relativized Clear-cut Far Mid Near Mature P

Figure 3.3: Hierarchical clustering analysis using: a) log transformed abundances; and b) abundances relativized by site then species totals; for ground and rove beetles. Samples were collected within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest.

54 Table: 3.1: Known habitat affinities of ground and rove beetles exhibiting strong (>0.2) correlation with canonical analysis of principle co-ordinates (CAP) axis 2. Taxa Species Correlation with Known Habitat Affinities CAP Axis 2 Rove Quedius caseyi 0.621 Forest litter, dung (a) Rove Bolitobius horni 0.439 Fungi, leaf litter, possibly deciduous forests Rove Quedius fulvicollis 0.347 Forest leaf litter, (a) Ground Carabus chamissonis 0.344 Predatory, forested habitats (b) Rove Tachinus quebecensis 0.336 Decaying fungi (c) Rove Pseudopsis sagitta 0.331 Forest leaf litter, dung (d) Rove Ischnosoma fimbriatum 0.313 Forest leaf litter, mosses (e) Rove Ischnosoma splendidum 0.277 Forest leaf litter, coniferous and deciduous forests (e) Rove Quedius brunnipennis 0.261 Moss, forest leaf litter (a) a. Smetana, 1971 b. Lindroth 1961-1969 c. Wheeler and Blackwell, 1984 d. Newton. 2001 e. Campbell, 1991

a) Log transformed b) Relativized 1.2 • • A 0.6 • • Clear-cut Far T O A o O T • Mid 0,0 • T A A Near • O • • Mature O T -0.6 -0.6 • • A O

-1.2 -0.9 0.0 0.9 -1 0 I Axis 1 (84.0%) Axis 1 (63.1%)

Figure 3.4: Non-metric multidimensional scaling ordination displaying a) log transformed abundances; and b) abundances relativized by site then species totals; for ground and rove beetles. Samples were collected within clear-cut, mature forest, and three distance classes of retention patches: near (10-55 m), mid (65-125 m), and far (140-360 m) from the mature forest.

55 a) Log transformed b) Relativized 0.3 & • Clear-cut T o Far A 0.1 A • • Mid • • A Near T • Of" • Mature -0.1 - • 3 O • •

-0.3 o

0 -0.5 -0.6 -0.4 -0.2 0.0 0.2 0.4 Axis 1

Figure 3.5: Canonical analysis of principal co-ordinates (CAP) ordination displaying: a) log transformed abundances; and b) abundances relativized by site then species totals; for ground and rove beetles. Samples were collected within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest.

56 Quedius caseyi T Bolitobius horni JL 0.10 - i

, • 0.05 - "•' 'l i i 1- i i Mature Near Mid Far Clear-cut Mature Near Mid Far C ear-cut 0.21 - Qu edius fulvicollis Carabus chamissonis

0.14 - J3 u o -a u 0.07 N T

-2 n no ui£E^ Mature Near Mid Far Clear-cut Mature Near Mid Far Clear-cut 0.21 Tachinus quebecensisl Pseudopsls sagitta

0.14 \

0,07 1 f

0,00 0.00 Mature Near Mid Far Clear-cut Mature Near Mid Far Clear-cut

Treatment class Treatment class Figure 3.6: Reletavized catch (+1 S.E.) of the 6 species with the strongest positive correlation to canonical analysis of principal co-ordinates (CAP) axis 2. Ground beetles and rove beetles were sampled within clear-cuts, mature forest, and three distance classes of retention patches: near (10-55 m); mid (65-125 m); and far (140-360 m); from the mature forest.

57 Appendix 3A: Ground and rove beetles collected within coniferous stands and grouped by abundance class. Clear-cut Far Mid Near Mature Total Family Abundance Class Species Mean S.E. Mean S.E. Mean S.E. Mean S.E. Mean S.E. Carabidac Common Calathus advena (LeConte) 0 0.00 220 16.10 219 15.58 352 39.48 82 7.31 873 Stereocerus haematopus (Dejean) 4 0.71 67 6.25 64 3.67 97 4.03 23 1.03 255 Pterostichus adstrictus Eschscholtz 133 3.90 19 0.48 13 0.63 24 4.02 25 2.36 214 Rare* Calathus ingratus Dejean 27 2.87 4 1.00 7 0.85 14 2.84 4 1.00 56 Trechus chalybeus Dejean 16 1.22 11 1.75 19 1.65 3 0.25 7 0.75 56 Pterostichus brevicornis (Kirby) 0 0.00 5 0.63 10 1.32 10 1.04 4 0.71 29 Trechus apicalis Motschulsky 7 0.85 5 1.25 5 0.63 2 0.29 0 0.00 19 Carabus chamissonis Fischer von Waldheim 1 0.25 0 0.00 8 2.00 5 1.25 1 0.25 15 Pterostichus punctatissimus (Randall) 3 0.48 2 0.29 3 0.75 1 0.25 1 0.25 10 Synuchus impunctatus (Say) 9 1.03 0 0.00 1 0.25 0 0.00 0 0.00 10 Patrobusfoveocollis (Eschscholtz) 6 0.50 0 0.00 1 0.25 0 0.00 0 0.00 7 Removed' Harpalus fulvilabris Mannerheim 2 0.50 0 0.00 0 0.00 0 0.00 0 0.00 2 Platynus decentis (Say) 0 0.00 0 0.00 0 0.00 1 0.25 1 0.25 2 Agonum cupreum Dejean 1 0.25 0 0.00 0 0.00 0 0.00 0 0.00 Agonum retractum LeConte 0 0.00 0 0.00 1 0.25 0 0.00 0 0.00 Amara erratica (Duftschmid) 1 0.25 0 0.00 0 0.00 0 0.00 0 0.00 Bembidion grapii Gyllenhal 1 0.25 0 0.00 0 0.00 0 0.00 0 0.00 Pterostichus pensyhanicus LeConte 1 0.25 0 0.00 0 0.00 0 0.00 0 0.00 Staphylinidae Common Tachinus frigidus Erichson 0 0.00 52 3.24 81 13.99 57 8.87 134 13.58 324 Quedius velox Smetana 2 0.50 96 5.67 67 3.04 73 2.87 62 2.18 300 Rare* Quedius brunnipennis Mannerheim 1 0.25 25 2.56 42 3.07 47 4.27 37 2.43 152 Tachinus elongatus Gyllenhal 20 1.29 25 2.93 27 2.25 25 4.25 29 4.61 126 Dinothenarus pleuralis (Leconte) 8 0.91 27 0.85 26 0.96 35 7.76 25 3.20 121 Tachyporus borealis Campbell 21 1.89 11 0.48 38 3.88 14 2.36 9 1.65 93 Acidota quadrata (Zetterstedt) 7 0.85 22 2.06 16 1.63 30 4.63 10 1.55 85 Lypoglossa angularis (Maklin) 0 0.00 17 2.02 26 2.60 17 1.93 19 4.42 79 Micropeplus laticollis Maklin 5 0.63 25 2.56 18 3.52 8 1.08 15 3.12 71 Mycetoporus americanus Erichson 23 2.29 13 1.70 16 1.22 6 0.87 7 0.63 65 Ischnosoma splendidum (Gravenhorst) 13 1.80 4 0.58 7 0.85 9 0.95 10 1.85 43 Quediusfulvicollis (Stephens) 5 0.48 5 0.63 9 0.63 7 0.25 11 1.03 37 Quedius labradorensis Blair 7 1.11 4 0.41 9 1.60 7 1.75 3 0.75 30 Eucnecosum brunnescens (J.Sahlberg) 11 1.49 6 0.96 1 0.25 5 0.95 2 0.29 25 Quedius frigidus Smetana 5 0.75 4 0.71 3 0.75 5 0.75 5 0.63 22 Quedius rusticus Smetana 0 0.00 4 0.41 1 0.25 3 0.75 13 1.11 21

oo Clear-cut Far Mid Near Mature Total Family Abundance Class Species Mean S.E. Mean S.E. Mean S.E. Mean S.E. Mean S.E. Lypoglossafranclemonti Hoebeke 0 0.00 0 0.00 2 0.50 8 2.00 8 1.35 18 Tachinus quebecensis Robert 1 0.25 1 0.25 1 0.25 9 1.31 3 0.75 15 Gabrius brevipennis (Horn) 1 0.25 2 0.29 4 0.71 4 1.00 2 0.29 13 Bolitobius horni Campbell 0 0.00 0 0.00 5 0.25 3 0.48 4 0.71 12 Ischnosoma fimbriatum Campbell 1 0.25 3 0.48 3 0.25 4 0.71 1 0.25 12 Quedius caseyi Scheerpeltz 0 0.00 0 0.00 3 0.48 7 0.85 0 0.00 10 Mycetoporus smetanai Campbell 2 0.29 1 0.25 1 0.25 3 0.75 2 0.50 9 Stenus austini Casey 1 0.25 2 0.50 1 0.25 4 0.71 1 0.25 9 Ontholestes cingulatus (Gravenhorst) 1 0.25 6 0.87 0 0.00 0 0.00 0 0.00 7 Lathrobium washingtoni Casey 5 0.48 1 0.25 0 0.00 0 0.00 0 0.00 6 Pseudopsis sagitta Herman 1 0.25 0 0.00 0 0.00 3 0.25 1 0.25 5 Gabrius picipennis (Maklin) 1 0.25 1 0.25 2 0.29 0 0.00 0 0.00 4 Tachyporus canadensis Campbell 8 1.08 0 0.00 0 0.00 0 0.00 0 0.00 8 Tachinus basalis Erichson 0 0.00 3 0.48 0 0.00 3 0.75 0 0.00 6 Megarthrus angulicollis Maklin 0 0.00 0 0.00 1 0.25 0 0.00 3 0.48 4 Acidota crenata (Fabricius) 3 0.48 0 0.00 0 0.00 0 0.00 0 0.00 3 Lordithon fungicola Campbell 1 0.25 0 0.00 1 0.25 1 0.25 0 0.00 3 Quedius simulator Smetana 2 0.50 0 0.00 1 0.25 0 0.00 0 0.00 3 Tachinus tachyporoides Horn 1 0.25 1 0.25 0 0.00 0 0.00 1 0.25 3 Anotylus sobrinus (LeConte) 2 0.50 0 0.00 0 0.00 0 0.00 0 0.00 2 Oxytelus fuscipennis Mannerheim 0 0.00 0 0.00 0 0.00 1 0.25 1 0.25 2 Bisnius tereus Smetana 1 0.25 0 0.00 0 0.00 0 0.00 0 0.00 Bryoporus rufescens LeConte 1 0.25 0 0.00 0 0.00 0 0.00 0 0.00 Carphacis nepigonensis (Bernhauer) 0 0.00 1 0.25 0 0.00 0 0.00 0 0.00 Phyllodrepa spp. 0 0.00 1 0.25 0 0.00 0 0.00 0 0.00 Quedius plagiatus Mannerheim 0 0.00 1 0.25 0 0.00 0 0.00 0 0.00 Scaphium castanipes Kirby 0 0.00 0 0.00 1 0.25 0 0.00 0 0.00 Tachinus fumipennis (Say) 0 0.00 0 0.00 0 0.00 0 0.00 1 0.25 * Rare species are defined as those representing <5% of the total catch. A Removed species represent those not indluded in analyses as they were present in <20% of all sites. Chapter 4: Spatial and temporal response of ground beetles to recent and regenerating harvest edges in the boreal forest of Alberta, Canada

Introduction Forest harvesting is widely recognized as a major contributor to landscape fragmentation in the boreal forest (Koivula et al. 2002, Schmiegelow and Monkkonen 2002, Schneider 2002). Such fragmentation is not only produced directly by the harvest block, but also indirectly through the creation of forest edges (Murcia 1995). These forest edges develop into ecotones in which changes in habitat quality (Harper and Macdonald 2002) and resource availability (Spence et al. 1996) can alter biotic composition. The extent of such edge effects are poorly defined for particular taxa, but changes in biodiversity can extend an average of 50 m into the residual forest matrix (Murcia 1995), resulting in a reduction of interior forest habitat. Study of edge effects is therefore critical for understanding fragmentation impacts on biodiversity, and to promote effective conservation within managed forests. Despite progress in understanding edge effects, there are still significant information gaps that limit a generalized understanding of edge processes (Murcia 1995). This is particularly evident for taxa such as arthropods which have rarely been studied extensively in the context of edge effects. For example, arthropod studies in the boreal region typically have low replication levels and are limited to a single forest cover-type (Spence et al. 1996, Heliola et al. 2001, Pohl et al. 2007). Knowledge of ground beetle response to edges in the circumpolar boreal forest, in particular, is almost completely limited to conifer forests (Spence et al. 1996, Heliola et al. 2001). Thus, work on cover-type as a driver of edge effect patterns, and in particular recovery from edge effects with forest regeneration, is needed to facilitate more robust management, particularly in the mixedwood boreal forest of Canada. Although biodiversity recovery following the direct impacts of forest harvesting has been a topic of extensive research (Work et al. 2004, Buddie et al. 2006, Work et al. 2009), recovery along forest edges has received little attention. Thus, our ability to model landscape level change in biodiversity over time is limited (but see Matlack 1994, Harper and Macdonald 2002), despite its central place in conservation planning. Assessment of edge effect influences on arthropods is fundamental to forest management as arthropods represent the majority of faunal diversity in the boreal forest (Langor and Spence 2006). Furthermore, arthropods contribute to nutrient cycles and serve as food sources for larger

60 taxa, thus contributing significantly to ecological function in forests (Mattson and Addy 1975, Maleque et al. 2006, Negro et al. 2007, Spence et al. 2008, Cobb et al. 2009). Arthropods have also been recognized as useful biological indicators as they are abundant, species rich, and easily sampled (Taylor and Doran 2002). Ground beetles in particular are highly sensitive to changes in habitat quality (McCullough et al. 1998), especially in forest systems (Haila et al. 1994, Spence et al. 1996, Heliola et al. 2001, Niemela et al. 2007), highlighting their utility as ecological indicators in forest ecosystems. I studied edge effects in both deciduous and conifer stands, and used a chronosequence design in deciduous stands, in pursuit of the following objectives: 1) assess the distance to which edge effects extend into residual forest stands; 2) determine whether the extent of edge effects vary by cover-type; 3) determine whether any ground beetle species specialize in edge or interior forest habitats; and 4) assess whether edge effects diminish over time.

Methods Study Site Study sites were located in the boreal mixedwood forest of northwestern Alberta, Canada in the lower foothills forest region (Rowe 1972), ranging from 50 km northwest of Grimshaw (56.2°N 118.2°W) to approximately 12 km northwest of Hotchkiss (57.7°N 117.4°W). Sample plots were in or adjacent to industrial harvest blocks within the land base managed by Daishowa Marubeni International Ltd.. Samples were collected within two dominant merchantable forest types of the boreal mixedwood forest. Deciduous dominated sites (>70% of overstory canopy is deciduous species) were primarily composed of trembling aspen (Populus tremuloides Michaux) and balsam poplar {Populus balsamifera L.). Conifer dominated sites (>70% of overstory canopy is conifer species) were primarily composed of white spruce (Picea glauca (Moench)) with lodgepole pine {Pinus contorta Douglas) and balsam fir {Abies balsamea (L.) Mill.) occurring as secondary canopy elements. The stands are part of a continuous forest matrix of wildfire origin, ranging from 70-160 years in age and exposed to the first industrial harvests within the past 15 years.

Study Design Two transects, spaced a minimum of 100 m apart, were established perpendicular to the forest edge along the east side of each harvest block, as this was the most consistent edge orientation

61 available across sites. Transects extended from the forest edge (0 m) to the forest interior, with additional traps located in the adjacent clear-cut (Fig. 4.1). In 2006, traps were placed every 15 metres along the transect, to a maximum of 90 metres into the forest interior (Fig. 4.1a). In 2007, the sampling design was adjusted based on findings from 2006; beetles were sampled at the edge and at 1 m, 5 m, 15 m, 30 m, 45 m and 60 m into the forest (Fig. 4.1b). To ensure independence among trap catches (Digweed et al. 1995), traps in each transect were kept a minimum of 25 metres apart through the use of two paired parallel trap lines (see Fig. 4.1). All forest interior traps were located at least 100 m away from any disturbed edge (e.g., seismic lines, large windthrow, harvest edges, etc.) other than the focal edge, and traps in clear-cuts were located at least 50 m from any forested edge. To determine the effects of cover-type on extent of edge effect, forest edges were sampled within each of three deciduous dominated and three conifer dominated stands. In 2006, one of the conifer replicates was destroyed by a forest fire and, therefore, only two replicates were sampled. A third replicate was added in 2007. As edge age (i.e., time since disturbance of the adjacent stand) is known to influence edge effects (Matlack 1994), I controlled for edge age effects by sampling the forest edges adjacent to 2-3 year-old harvest blocks. Deciduous harvest blocks were naturally regenerated with no site preparation, whereas conifer blocks had been scarified and planted to white spruce within 1 year of harvest. The temporal influence on the extent of edge effects was also studied, but only in deciduous stands. For this study, stands adjacent to eight and 15 year-old harvest blocks were sampled in addition to the 2-3 year-old edges. Edges older than 15 years were not available within this landscape. Two replicates of the eight and 15 year-old edges were sampled in 2006, and three replicates of each in 2007. Samples from the clear-cut were also collected from these older harvest blocks.

Beetle Sampling Ground beetles were sampled via continuous pitfall trapping from May 15th to August 27th in 2006 and 2007, dates which encompass >90% of beetle activity in this area. Traps were 1 L plastic containers with a smaller plastic insert, filled with approximately 1.5 cm of silicate-free ethylene glycol as a preservative, and covered with an elevated plastic lid to limit the amount of precipitation and debris within the traps (Spence and Niemela 1994). Beetles were collected

62 from the traps every three weeks throughout the sampling periods, sorted to family and subsequently ground beetles were identified to species using Lindroth (1961-1969). Larvae were not included in any analyses as identification below the family level was not possible. Voucher specimens have been deposited at the Strickland Entomological Museum, and the Northern Forestry Centre arthropod museum in Edmonton, Alberta, Canada.

Data Analyses The two transects sampled within each harvest block were pooled for analyses to provide a single replicate per block. In addition, for all analyses with the exception of rarefaction estimates, pooled trap catches were standardized to 100 trap-days to adjust for variable sampling effort as a result of trap disturbance and uneven replication. Species richness was compared along transects by pooling traps into four 'transect zones' (Heliola et al. 2001). The 'clear-cut' zone contained traps from the harvested block; 'forest edge' zone contained traps located at 0m and lm from the forest edge; 'forest mid' zone contained traps at 30 m and 45 m from the forest edge; and 'forest interior' zone contained traps located at 60 m, 75 m, and 90 m from the forest edge. The purpose of the groupings was not to assess specific distances at which edge effects penetrated, but rather to determine if there were any general changes in species richness driven by the edge creation, and how this might vary by cover-type. Traps located at 5 m and 15 m from the forest edge were left out of either the 'forest edge' or the 'forest mid' zones to limit the a-priori assumptions of distance of edge penetration on the ground beetle community. Traps 30 m and 45 m from the forest edge were clustered as 'forest mid' to investigate possible secondary edge effects which have been documented in other taxa (Harper and Macdonald 2002). Individual based rarefaction was used to generate the estimated species richness within each of these transect zones. Rarefaction is a robust method for comparing species richness between sites because it adjusts for variable sampling effort between treatments (Gotelli and Colwell 2001). In addition, since rarefaction takes into account both species abundance and richness, it can be interpreted as a diversity measure (Buddie et al. 2005). Rarefaction curves were generated from raw species abundance data using the Vegan package (Oksanen et al. 2005) in R 2.6 (R Development Core Team, 2007). To assess the influence of edge effects on community composition, hierarchical clustering analyses were conducted on the deciduous and conifer cover-types separately. The

63 analysis was based on a Bray-Curtis similarity matrix and calculated using web-based software of Brzustowski (1999). Distances between groups were assessed using unweighted arithmetic averages, defined as the average distance between a sample in group A, and a sample in group B. Subsequent analyses were also conducted on the 8 and 15 year-old deciduous data sets to analyze the influence of edge age on beetle community composition. Canonical analysis of principal co-ordinates (CAP) was selected as an ordination technique because it enables a direct test of hypotheses (Anderson and Willis 2003). Such an analysis is highly relevant for studies on edge effects as it facilitates understanding of distance class effects on assemblage composition. In addition, unlike non-metric multidimensional scaling, the CAP technique aims to find the maximum separation between pre-defined classes (i.e., the distance classes along the transect) (Anderson and Willis 2003), thereby increasing the sensitivity of the analysis. CAP analyses were conducted using a FORTRAN program (Anderson and Willis 2003, Anderson 2004), and the Bray-Curtis similarity index. The number of axes used in the canonical analysis (m-value) was selected by the program and ordination significance was evaluated via a trace statistic using 9999 random permutations (Anderson and Robinson 2003). Analyses were conducted for the deciduous and conifer cover-types separately, as well as the deciduous edges adjacent to 8 and 15 year-old harvest blocks. Indicator species analysis (ISA, Dufrene and Legendre 1997) was conducted on the most recent deciduous and conifer forest edges to assess species specialization to two forest zones as defined by the previous clustering and ordination analyses: 'disturbed' (Clear-cut, 0m, lm, 5m), and 'forest interior' (15 m-90 m). The analysis calculates an indicator value (IndVal) for each species based on its abundance and frequency within the pre-defined classes. Thus, a species occurring in only one zone, and all of the replicates would be given an indicator value of 100, whereas a species occurring in all sites at equal proportions would receive a value of zero. Indicator species analyses were conducted in PC-Ord Version 5 (McCune and Mefford 1999), with significance tested via Monte Carlo analysis. Percentage of total catch along the recent deciduous and conifer edges was compared for the 10 species with significant indicator values (p<0.05). Percentage of total catch was calculated as the standardized catch within each distance category divided by the total standardized catch within the corresponding cover-type. This increased the clarity of species patterns by removing variability caused by differences in total overall catch at each trap, and

64 produced a dominance value for each species considered. Values were graphed to display these catch patterns along the edge transect. In addition, percentage of total catch values for the 4 most common species in the 8 and 15 year-old edge transects were graphed to compare recovery patterns amongst these species.

Results A total of 6,902 individuals representing 35 species were collected during the two year study. The most abundant species were Calathus ingratus Dejean (rc=1375), Pterostichus adstrictus Eschscholtz (rc=1207), and Platynus decentis (Say) (n=833). Of the remaining species collected, 17 were represented by fewer than 10 individuals. Eleven species were collected exclusively in deciduous sites, seven species exclusively in conifer sites, and 17 species were common to both cover-types.

Cover-type Effect Rarefaction-estimated species richness varied consistently among cover-types, with the clear-cut samples having the highest species richness, mid and interior samples the lowest, and edge samples somewhat intermediate. This pattern was more pronounced in conifer transects (Fig. 4.2b) than in deciduous transects (Fig. 4.2a). In both stand types the 'forest mid' and 'forest interior' zones approached an asymptote in species accumulation, indicating that the fauna of these zones was well sampled. However, the more steeply ascending curves for the 'forest edge' and 'clear-cut' zones suggests that their constituent fauna was not as completely sampled (Fig. 4.2). Hierarchical clustering analyses showed a high amount of variability along the deciduous edges, and patterns did not indicate evidence of a clear edge effect as traps from the edge often showed high faunal similarity to those in the interior (Fig. 4.3a). The clear-cut assemblages were highly distinct from any of the distance classes (i.e., 0 m; 1 m; 5 m; 15 m; 30 m; etc.). In conifer forests there was a strong indication of an edge effect as traps from the clear-cut and closest to the forest edge (0 m, 1 m, and 5 m) clustered together and had a minimum similarity to more forest interior traps (15 m-90 m) of 65% (Fig. 4.3b). The CAP ordinations further supported the patterns evident from the hierarchical clustering analyses. In deciduous forests (Fig. 4.4a) there was higher overall variability and higher variability among replicates of the distance classes than was evident in conifer forests

65 (Fig. 4.4b). This high variability in the analysis of data from deciduous forests resulted in the detection of no significant differences between the distance classes {Trace Statistic= 1.85, jt?=0.53). Although traps at 0 m and 15 m distance classes mostly clustered close to those from clear-cuts, traps at 1 m and 5 m did not (Fig. 4.4a). Thus, it is difficult to discern any evidence of edge effects along recent deciduous forest edges. The CAP ordination of the conifer sites produced a more distinct pattern with a significant difference between the distance classes {Trace Statistic- 1.38, /7=0.026). The clear-cut traps and forest edge traps (0 m-1 m) grouped within the left half of the ordination, suggesting high similarity between these groups. The 15 m traps grouped somewhat intermediately between the forest edge and forest interior traps. Despite the significance reported, it is important to note the CAP ordination for conifer fragments had a mis- classification error rate of 78.6%, providing an indication of high variability in the ordination. Indicator species analysis revealed 10 species that had significant (/?<0.05) indicator values greater than 27.5 (Table 4.1). In the deciduous cover-type, three species were indicative of the disturbed habitats (Clear-cut, 0-5 m), and four were indicative of the forest habitat (15-90 m). In the conifer cover-type, three species were indicative of the forest habitat, and no species were indicative of the disturbed habitat. Only five species achieved indicator values greater than 40, with Calathus advena (LeConte) achieving the highest indicator value at 59.3 (Table 4.1). In addition, among the identified indicator species, only Carabus chamissonis Fischer achieved a higher indicator value (37.7) when analyses were conducted to assess specialization to edge habitat (0-5 m). Comparison of catch along the trap transects of significant indicator species provided further insight into single species responses to edge effects (Fig. 4.5). Of the three species that were indicative of clear-cuts and edge, P. adstrictus exhibited the clearest trend with decreasing relative catch from the clear-cut to the interior, particularly in the conifer stand type. The catch of Synuchus impunctatus Say was highest from 0m up to 15m into the forest, while there was an abrupt drop in catch at 30 m that was maintained further into the forest. Catch of C. chamissonis was variable along the transects but was consistently high in traps from 0-15 m. Of the four species that were indicative of interior deciduous forest habitats, the trends in catch were not consistent across species. For C. ingratus, catch was similar for all traps except those at 75 m and 90 m which were higher. The catch of P. decentis did increase from the edge to 60 m, but dropped precipitously in traps 75-90 m from the edge. Catches of both Calosoma

66 frigidum Kirby and Agonum retractum Leconte fluctuated greatly along the length of the transect. Of the three species indicative of interior conifer forests, C. advena exhibited the clearest trend, with the lowest catch in clear-cuts and traps at 0-5 m into the forest, but much higher catch in traps 15-90 m into the forest. This species also exhibited a similar trend in deciduous stands, but it was far less abundant there. Pterostichus brevicornis (Kirby) had low catch levels along the entire transect, but interestingly peaked in traps located 75 m from the forest edge. Catch of Stereocerus haematopus (Dejean) decreased in the forest interior. Clearly the response of individual species to edge is highly variable.

Edge Recovery Rarefaction curves of species richness for different ages of deciduous forest edges showed variable patterns among transect zones (Fig. 4.6). In clear-cuts, species richness was highest eight years post-harvest, followed by the two year-old harvests and finally the 15 year-old harvests which had the lowest species richness (Fig. 4.6a). Within the forest edge zones both the 8 and 15 year-old edges had lower species richness than the two year-old edges (Fig. 4.6b). Carabid response to the forest mid zone varied with time since harvest as the eight year-old edges had the highest richness and exhibited a steep species accumulation, even at the maximum number of individuals sampled, while mid-zone catches from 15 year-old edges had the lowest species richness (Fig. 4.6c). Species richness estimates from the forest interior converged within all of the forest edge age classes (Fig. 4.6d). The curves were steepest in clear-cuts but became less steep with increasing distance from edge, particularly in the two year-old edges. Hierarchical clustering analyses indicated a strong edge effect within the 8 year-old edges, with signs of recovery suggested by the 15 year-old edge analysis (Fig. 4.7). In the eight year-old edge analysis, catches from the clear-cut and forest edge traps (0 m-5 m) clustered clearly together and with a similarity of only 67% to the remaining forest traps (15 m-90 m) (Fig. 4.7a). The 15 year-old edge analysis, however, showed a high degree of variability between edge traps. The 0 m and 15 m traps clustered together, but the clear-cut, 1 m and 5 m traps clustered amongst other forest traps (Fig. 4.7b). CAP ordination was also conducted; however, neither of the ordinations suggested significant differences between distance classes (8 year-old Trace Statistic= 1.83,p=0.73; 15 year-old Trace Statistic- 2.47,^=0.31).

67 Percentage catch comparisons for the four most abundant species also varied with edge recovery over time (Fig. 4.8). Catch of C. ingratus varied along the transect in all years, with its lowest catch in 0 m and 1 m traps along the eight years post-harvest edge. Pterostichus adstrictus, which had a strong presence in clear-cuts in the two year-old sites, had reduced catch in this zone to more background levels in both the 8 and 15 years post-harvest edges. Platynus decentis was most commonly caught within the 15-60 m traps in both the two and eight years post-harvest edges; however abundance of this species was more variable along the 15 years post-harvest edges. Catch of C. frigidum fluctuated greatly along all edge ages, but its catch was highest within the 15-45 m traps of the 15 years post-harvest edges.

Discussion Cover-type Effect The consistency in overall species richness patterns between cover-types, with increased richness in the clear-cut and forest edge zones, indicates ground beetles were clearly altered by harvesting and edge effects in this study. These patterns are particularly interesting when the rate of species accumulation is considered; the forest mid and forest interior zones appeared to reach an asymptote in species accumulation, whereas the forest edge and clear-cut zones did not. This suggests that, had more individuals been sampled, the difference in species richness between the forest edge and forest mid/interior zones would have increased. Variation in degree of edge impact between cover-type was evidenced by the greater separation between the forest interior and forest edge zones in the conifer cover-type compared to deciduous stands. This finding lends support to previous studies that have suggested that conifer stands are more sensitive to the impacts of forest harvesting (Work et al. 2009). The community level analyses revealed that edges are narrow and abrupt in both cover- types, despite the high variability observed in edge effects of deciduous forests. The clustering analysis of the deciduous edges grouped the 0 m and 15 m traps and the 1 m and 5 m traps together, respectively, suggesting some change in community composition occurred within the edge zone. The pattern seen in the CAP ordination (Fig. 4.4), although non-significant, suggested the beetles captured were similar between the 0m, 15m, and clear-cut traps. The single species responses of P. adstrictus and S. impunctatus also supported this conclusion with higher catches up to 15m into the forest. Thus despite the variability in the deciduous edge,

68 there appears to be an edge effect extending up to 15m into the residual forest bordering cut blocks 2-3 years old. This edge effect on carabids was more strongly evident eight years post- harvest, suggesting that responses of forest beetles to the 2-3 year-old deciduous edges were delayed. This sort of lag response to harvest has been noted in other studies from Alberta (Niemela et al. 1993, Pohl et al. 2007, Jacobs et al. 2008), and undoubtedly contributed to the variability observed in the current study. The clearer community response in the conifer cover-type resulted from an obvious change in beetle assemblage in traps 0 m-5 m from the forest edge. This separation of groups, clearly apparent in both the cluster analysis and ordination, suggests a change in beetle community composition in the edge zone, extending between 5 and 15 m into the forest. Spence et al. (1996) found similar results in a lodgepole pine dominated forest and concluded that edge effects for carabids penetrated between 5 and 10 m into residual forest from clear-cut blocks. Similarly, Heliola et al. (2001) documented an abrupt edge for carabids in Norway spruce forests. This consistently narrow, abrupt edge effect is quite different from that exhibited by other taxa. Edge effects on vegetation communities (Harper and Macdonald 2002), bird communities (Brand and George 2001) and other arthropods (Pohl et al. 2007, Larrivee et al. 2008) have been shown to extend further (20-140 m) into the forest, suggesting that these taxa have greater sensitivity to edge effects than do carabids (Koivula and Niemela 2002). The rapid transition between forest edge and interior, which appears to be a common pattern exhibited by carabid communities in boreal forests, is interesting, and suggests that variables close to the forest edge are driving this short, abrupt response. Generally, areas close to forest edges receive more direct sunlight, and microclimatic gradients are known to drive edge patterns (Matlack 1993, Chen et al. 1995, Murcia 1995). Although these microclimatic gradients may extend up to 240 m into the forest for some variables (Chen et al. 1995), Matlack (1993) documented that elevated temperature and light variables extend only 11-35 m into Piedmont oak forests. Chen et al. (1995) also noted that increases in soil temperature and short-wave radiation were concentrated closest to the forest edge in Douglas fir forests. Temperature and, indirectly, light are known to influence carabid activity and distribution (Lovei and Sunderland 1996, Magura et al. 2001, Magura 2002), therefore, it seems likely that these variables are driving the narrow, abrupt edge responses of these taxa. In addition, the increased persistence of generalist species such as P. adstrictus and open habitat species such as S. impunctatus into the

69 forest edge further support this conclusion, and demonstrate the fragmenting effects of these changes in microclimate.

Indicator Species Only one species, C. chamissonis, was found to have its highest indicator value when specialization to the edge (0-5 m) was assessed. However, its indicator value was not near values typically used to classify specialists {i.e., >80) (Dufrene and Legendre 1997), and catch patterns clearly suggest that it is more of a forest generalist, a conclusion supported by previous work (Lindroth 1961-1969, Niemela et al. 1992). Thus, it appears that no species sampled here depends on, or is favoured by, edge habitat created by harvesting. Specialization to edge habitats created by forest harvesting appears to be generally absent within boreal carabid communities (Spence et al. 1996, Heliola et al. 2001); however, edge specialists are evident in forest-grassland edges (Magura 2002), and edges created by forest fire (Gandhi 1999). Gandhi (1999) collected three rare carabid species only along forest edges adjacent to a 15 year-old fire, and Magura (2002) identified two carabid species to be indicative of edge habitat, one of which was also rare. From these results it seems probable that carabid species adapt to edges of two types: 1) at the intersection of two permanently different habitats, as in the case of forest-grassland edges; and 2) along edges created by natural disturbances with which species have co-evolved and adapted. Forest fires have long been a major natural driver of boreal forest heterogeneity (Hunter 1993), and in turn species are known to specialize on natural habitats (McCullough et al. 1998). Therefore, I conclude that edge habitats caused by forest harvesting have no natural analogue and provide little direct conservation value for ground beetles. Forest management should therefore aim to minimize the impact of forest edges on interior forest habitats and species. The general conflict between the indicator values and the catch patterns exhibited by the identified indicator species suggests the analysis has limited utility for elucidating patterns in beetle communities along gradients. Although I used categories which encompassed a broad range of distance classes {e.g., 15 m-90 m) the species identified as indicator species were not always supported by the corresponding catch patterns. For example, P. adstrictus was identified as an indicator of deciduous clear-cut and edge habitats, even though its abundance patterns were clearly stronger in the conifer cover-type. These types of inconsistencies are common to studies

70 using the indicator species analysis (Chapter 2, Pohl et al. 2007) and suggest that caution is in order when drawing specific management recommendations from the ISA approach.

Edge Age There was no notable effect of edge age on species richness in any of the four 'transect zones', however, an interesting pattern was suggested by changes in the slope of the species accumulation curves over time. In the two years post-harvest edges, the slope of the species accumulation curve was consistently lower in the transect zones farther from the forest edge. On the contrary, in the 15 year post-harvest edges the slope of accumulation curves were much more consistent between all of the clear-cut and transect zones. This suggests that there was a relaxation in impact within the 'clear-cut', 'forest edge', and 'forest mid' zones over time. The variability in slopes observed in the eight year-old edges support such a conclusion, and suggests that recovery was likely occurring between 8 and 15 years following edge creation. The community level analyses supported these patterns in species richness as traps from clear-cuts and edge (0-5 m) traps were clustered together even 8 years post-harvest. However, recovery was evident by 15 years post-harvest as traps along the forest edge grouped together with more forest interior traps. The single species patterns provided further support to these conclusions with no detectable response to the edge gradient in any of the most common species. Recovery rates for other taxa, such as plants, do not appear to be as rapid as observed for carabids. Harper and Macdonald (2002) found edge effects on understory vegetation to be strongest 16 years following harvesting, despite being in similar forests to those studied here. Similarly, Matlack (1994) noted that although edge effects in plant communities declined over time along deciduous edges, they were still persistent for some species up to 55 years after edge creation. The discrepancies between taxa may reflect the deeper edge penetration documented along recent edges in both of these studies (Matlack 1994, Harper and Macdonald 2002), however, it may also reflect a greater sensitivity of ground beetles to the microclimatic gradients along forest edges. Matlack (1993) demonstrated that both temperature and light were substantially lower along forest edges with a regenerating side canopy. The sites sampled in my study clearly experienced this rapid rate of side canopy formation with canopy's reaching heights of greater than 8 m within only 15 years. Thus similar to the apparent microclimatic responses

71 of ground beetles to recent edges, it appears that these same variables may be the main drivers of the recovery patterns documented here. Although edge recovery in deciduous stands appears to be underway by 15 years following harvesting, this result should not be extrapolated to conifer stands. Conifers in harvested blocks adjacent to the studied edges regenerate at a much slower rate than deciduous cut blocks. Thus, microclimatic gradients along conifer forest edges may indeed persist for a much longer period of time (Gandhi 1999). There is also evidence that carabid assemblages will require more time to recover in harvested conifer stands as a result of slow regeneration (Work et al. 2009). The edge recovery rate documented in the present study may also be affected by the degree of surrounding landscape fragmentation (Spence et al. 1996). The sites studied here, for example, were contained within a landscape composed of large residual forests which can function as source populations for recovery. Results may be very different in landscapes which suffer from high amounts of fragmentation, and reduced levels of interior forest habitat (Spence etal. 1996).

Conclusions I documented a narrow, but abrupt response of carabid populations to creation of edge through harvest in both conifer and deciduous stands. These edge effects appear to extend between 5 and 15 m into the residual forest matrix. This estimate of edge dimension is similar to that reported in previous studies on ground beetles in other forest types and thus supports a general conclusion that edge effects are limited in size and are abrupt for boreal ground beetles. Recovery of ground beetles in these edge habitats appears to begin in 15 year-old stands, and is likely driven by the height of the regenerating canopy. Thus, recovery of beetle populations in edges is likely to be substantially delayed in conifer harvests because of the regeneration delay.

Literature Cited Anderson, M. J. 2004. CAP: a FORTRAN computer program for canonical analysis of principal coordinates: User Guide. Department of Statistics, University of Auckland, New Zealand. Anderson, M. J. and J. Robinson. 2003. Generalized discriminant analysis based on distances. Australian & New Zealand Journal of Statistics 45:301-318. Anderson, M. J. and T. J. Willis. 2003. Canonical analysis of principal coordinates: A useful method of constrained ordination for ecology. Ecology 84:511-525. Brand, L. A. and T. L. George. 2001. Response of passerine birds to forest edge in coast redwood forest fragments. AUK 118:678-686.

72 Brzustowski, J. 1999. Brzustowaski Free Programs, http://www2.biology.ualberta.ca /jbrzusto/cluster.php. (Accessed 27/02/2009) Buddie, C. M., J. Beguin, E. Bolduc, A. Mercado, T. E. Sackett, R. D. Selby, H. Varady-Szabo, and R. M. Zeran. 2005. The importance and use of taxon sampling curves for comparative biodiversity research with forest arthropod assemblages. Canadian Entomologist 137:120-127. Buddie, C. M, D. W. Langor, G. R. Pohl, and J. R. Spence. 2006. Arthropod responses to harvesting and wildfire: Implications for emulation of natural disturbance in forest management. Biological Conservation 128:346-357. Chen, J., J. F. Franklin, and T. A. Spies. 1995. Growing-season microclimatic gradients from clear-cut edges into old-growth Douglas-fir forests. Ecological Applications 5:74-86. Cobb, T. P., K. P. Hamman, B. E. Kishchuk, S. A. Quideau, D. W. Langor, and J. R. Spence. 2009. Effects of post-fire salvage logging on Monochamus scutellatus (Coleoptera: Carabidae): implications for wood decomposition, nutrient cycling and forest succesion. Agriculture and Forest Entomology In press. Digweed, S. C, C. R. Currie, H. A. Carcamo, and J. R. Spence. 1995. Digging out the "digging- in effect" of pitfall traps: Influences depletion and disturbance on catches of ground beetles (Coleoptera: Carabidae). Pedobiologia 39:561-576. Dufrene, M. and P. Legendre. 1997. Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecological Monographs 67:345-366. Gandhi, K. J. K. 1999. The importance of fore-skips as biotic refugia and the influence of forest heterogeneity on epigaeic beetles in pyrogenic stands of the northern Rocky Mountains. University of Alberta, Edmonton. Gotelli, N. J. and R. K. Colwell. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4:379-391. Haila, Y., I. K. Hanski, J. Niemela, P. Punttila, S. Raivio, and H. Tukia. 1994. Forestry and the boreal fauna - matching management with natural forest dynamics. Annales Zoologici Fennici 31:187-202. Harper, K. A. and S. E. Macdonald. 2002. Structure and composition of edges next to regenerating clear-cuts in mixed-wood boreal forest. Journal of Vegetation Science 13:535-546. Heliola, J., M. Koivula, and J. Niemela. 2001. Distribution of carabid beetles (Coleoptera, Carabidae) across a boreal forest-clearcut ecotone. Conservation Biology 15:370-377. Hunter, M. L. 1993. Natural fire regimes as spatial models for managing boreal forests. Biological Conservation 65:115-120. Jacobs, J. J., T. T. Work, and J. R. Spence. 2008. Correcting for detection biases in the pitfall trapping of ground beetles (Coleopetera: Carabidae). Pages 425-450 in Back to the Roots or Back to the Future? Towards a New Synthesis between Taxonomic, Ecological and Biogeographical Approaches in Carabidology, Proceedings of the 13th European Carabidologists Meeting, Blagoevgrad, Bulgaria, 20-24 August 2007.. Koivula, M., J. Kukkonen, and J. Niemela. 2002. Boreal carabid-beetle (Coleoptera, Carabidae) assemblages along the clear-cut originated succession gradient. Biodiversity and Conservation 11:1269-1288. Koivula, M. and J. Niemela. 2002. Boreal carabid beetles (Coleoptera: Carabidae) in managed spruce forests - a summary of Finnish case studies. Silva Fennica 36:423-436. Langor, D. W. and J. R. Spence. 2006. Arthropods as ecological indicators of sustainability in

73 Canadian forests. Forestry Chronicle 82:344-350. Larrivee, M., P. Drapeau, and L. Fahrig. 2008. Edge effects created by wildfire and clear-cutting on boreal forest ground-dwelling spiders. Forest Ecology and Management 255:1434- 1445. Lindroth, C. H. 1961-1969. The Ground-beetles of Canada and Alaska. Opuscula Entomologica (Suppl. Nos. 24,29,33,34,35). Lovei, G. L. and K. D. Sunderland. 1996. Ecology and behavior of ground beetles (Coleoptera: Carabidae). Annual Review of Entomology 41:231-256. Magura, T. 2002. Carabids and forest edge: spatial pattern and edge effect. Forest Ecology and Management 157:23-37. Magura, T., V. Kodobocz, and B. Tothmeresz. 2001. Effects of habitat fragmentation on carabids in forest patches. Journal of Biogeography 28:129-138. Maleque, A. M., H. T. Ishii, and K. Maeto. 2006. The use of arthropods as indicators of ecosystem integrity in forest management. Journal of Forestry 104:113-117. Matlack, G. R. 1993. Microenvironment variation within and among forest edge sites in the eastern United-States. Biological Conservation 66:185-194. Matlack, G. R. 1994. Vegetation dynamics of the forest edge- trends in spaces and succesional time. Journal of Ecology 82:113-123. Mattson, W. J. and N. D. Addy. 1975. Phytophagous insects as regulators of forest primary production. Science 190:515-522. McCullough, D. G., R. A. Werner, and D. Neumann. 1998. Fire and insects in northern and boreal forest ecosystems of North America. Annual Review of Entomology 43:107-127. McCune, B. and M. J. Mefford. 1999. Pc-Ord. Multivariate Analysis of Ecological Data, Version 5. MjM Software Design, Glendedon Beach, OR. USA. Murcia, C. 1995. Edge effects in fragmented forests- implications for conservation. Trends in Ecology & Evolution 10:58-62. Negro, M., A. Casale, L. Migliore, C. Palestrini, and A. Rolando. 2007. The effect of local anthropogenic habitat heterogeneity on assemblages of carabids (Coleoptera: Carabidae) endemic to the Alps. Biodiversity and Conservation 16:3919-3932. Niemela, J., M. Koivula, and D. J. Kotze. 2007. The effects of forestry on carabid beetles (Coleoptera : Carabidae) in boreal forests. Journal of Insect Conservation 11:5-18. Niemela, J., D. Langor, and J. R. Spence. 1993. Effects of Clear-Cut Harvesting on Boreal Ground-Beetle Assemblages (Coleoptera, Carabidae) in Western Canada. Conservation Biology 7:551-561. Niemela, J., J. R. Spence, and D. H. Spence. 1992. Habitat associations and seasonal activity of ground-beetles (Coleoptera: Carabidae) in central Alberta. The Canadian Entomologist 124:521-540. Oksanen, J., R. Kindt, and B. O'Hara. 2005. Vegan: community ecology package. R package version 1.6-8. Pohl, G. R., D. W. Langor, and J. R. Spence. 2007. Rove beetles and ground beetles (Coleoptera: Staphylinidae, Carabidae) as indicators of harvest and regeneration practices in western Canadian foothills forests. Biological Conservation 137:294-307. R Development Core Team. 2007. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Rowe, J. S. 1972. Forest regions of Canada. Department of Fisheries and the Environment, Canadian Forestry Service.

74 Schmiegelow, F. K. A. and M. Monkkonen. 2002. Habitat loss and fragmentation in dynamic landscapes: Avian perspectives from the boreal forest. Ecological Applications 12:375- 389. Schneider, R. R. 2002. Alternative futures: Alberta's boreal forest at the crossroads. Federation of Alberta Naturalists and Alberta Centre for Boreal Research, Edmonton, Alberta. Spence, J. R., D. W. Langor, J. M. Jacobs, T. T. Work, and W. J. A. Volney. 2008. Conservation of forest-dwelling arthropod species: simultaneous management of many small and heterogeneous risks. Canadian Entomologist 140:510-525. Spence, J. R., D. W. Langor, J. Niemela, H. A. Carcamo, and C. R. Currie. 1996. Northern forestry and carabids: The case for concern about old-growth species (vol 33, pg 173, 1996). Annales Zoologici Fennici 33:302-302. Spence, J. R. and J. K. Niemela. 1994. Sampling Carabid assemblages with pitfall traps - the madness and the method. Canadian Entomologist 126:881-894. Taylor, R. J. and N. Doran. 2002. Use of terrestrial invertebrates as indicators of the ecological sustainability of forest management under the Montreal Process. Journal of Insect Conservation 5:221-231. Work, T. T., J. M. Jacobs, J. R. Spence, and W. J. Volney. 2009. High levels of green-tree retention are required to preserve ground beetle diversity in boreal mixedwood forests. Ecological Applications (Accepted, pending revisions). Work, T. T., D. P. Shorthouse, J. R. Spence, W. J. A. Volney, and D. Langor. 2004. Stand composition and structure of the boreal mixedwood and epigaeic arthropods of the Ecosystem Management Emulating Natural Disturbance (EMEND) landbase in northwestern Alberta. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:417-430.

75 a) 2006 Sampling Design

Forest Clear-cut

75m 45m 15m * % > 25m -50m •

9()m 60m 30m Om

b) 200" Sampling Design

Forest !m « Clear-cut 5 m 45m 15m •

\ -25m -50m * • • *• 60m 30m 0m

Figure 4.1: Schematic of edge transect design used to collect ground beetles in the summers of: a) 2006; and b) 2007.

a) Deciduous A /-<^' --• 15 - yS^— — ~~

• Q_ Clcar-Cut 10 - — A — e±- Forest Edge —#- — o- Forest Mid _ • «- -o Forest Interior C3 B 5 - CO

n - 0 100 200 300 400 500 600 700 100 150 200 250 300 350 Number of individuals Number of individuals

Figure 4.2: Individual-based rarefaction estimate of ground beetle species richness at different location along stand edges for: a) deciduous stands; and b) conifer stands, in boreal mixedwood forests of northwestern Alberta. Samples were grouped as: 'clear-cut' (at least 50 m from edge in the adjacent cut block); 'forest edge' (0-1 m from forest edge); 'forest mid' (30-45 m); and 'forest interior' (60-90 m).

76 Table 4.1: Ground beetle indicator species that exhibited the highest Indicator values to two forest zones identified by previous analyses and to the two cover types studied (deciduous dominated and coniferous dominated), CC=Clear-cut

Cover Type Forest Zone Species IndVal P Deciduous CC-5m Carabus chamissonis 33.9 0.01 Pterostichus adstrictus 38.4 0.002 Synuchus impunctatus 30.7 0.034

15-90m Agonum retractum 49.8 0.002 Calosoma frigidum 38.6 0.006 Calathus ingratus 43.6 0.002 Platynus decentis 49.5 0.002

Conifer CC-5m none

15-90m Calathus advena 59.3 0.002 Pterostichus brevicornis 28.8 0.05 Stereocerus haematopus 41.6 0.004

77 Bray-Curtis Percent Similarity 100% 75% 50% 25% -j a) Deciduous 0m 15m 75m 90m V 30m 45m 60m. lm 5m CC

b) Coniferous 0m lm CC 5m 60m 45m 30m 15m 90m 75m }

Figure 4.3: Hierarchical clustering analysis of beetle community composition along: a) deciduous dominated edges; and b) conifer dominated edges. Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0 m), and up to 90 m into the forest.

78 a) Deciduous b) Conifer • v • cc o 0m 0 D X Sm A^ A O % 15m A ° •**£& • o u 30m 5 oo • 45m • • 60m < 9• 75m V Wm •

-0.4 -0.3 -0 2 -0 1 0.0 0.1 0 2 0.3 0 4

Axis 1 Axis I

Figure 4.4: Canonical analysis of principal co-ordinates (CAP) ordination of: a) deciduous dominated edges; and b) conifer dominated edges. Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0 m) and up to 90 m into the forest.

79 Plerosiichus ads/rictus so ('ulalhus advena

60 ^^H Deciduous P 1 Conifer

0 40

20

11ll In In 1 In In In l 1 • 1 o CC 0 1 5 15 30 45 60 75 90 CC 0 1 5 5 30 45 60 75 90 O Synuchus impimctatu.s

"*-c O

30 uc

15 1 lllll 5 15 30 45 60 75 90 Slereocercus hatimatopus

5 15 30 45 60 75 90 CC 0 I 5 15 30 45 60 75 90 c3 20 O Calosoma frigidum\ Pretostichus hrevicomis o 60 15

u (X

.,11 -,n JL 5 15 30 45 60 75 90 CC 3 5 15 30 45 60 75 90

Agonum relracliim Carabus chamissonis 6

4

2

\s 1 1 1 1 II. CC 0 1 5 15 30 45 60 75 90 CC 0 1 5 15 30 45 60 75 90 Trap location (m) Trap location (m)

Figure 4.5: Percentage of standardized catch in each of two forest cover-types (deciduous, conifer) of 10 indicator ground beetle species with significant indicator values (p<=0.05). Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0 m) and up to 90 m into the forest.

80 20

2 Years Post-Harvest 8 Years Post-Harvest 15 Years Post-Harvest

100 200 300 400 500 600

d) Forest Interior

/^^•fw m

—B 2 Years Post-Harvest —£— 6 Years Post-Harvest f — ^— 15 Years Post-Harvest

0 100 200 300 400 500 600 700 100 200 300 400 500 600 700 Number of individuals Number of individuals

Figure 4.6: Individual-based rarefaction estimates of ground beetle species richness at different location along stand edges 2, 8, and 15 years post-harvest in boreal mixedwood forests of northwestern Alberta. Samples were grouped as: 'clear-cut' (at least 50 m from edge in the adjacent cut block); 'forest edge' (0-1 m from forest edge); 'forest mid' (30-45 m); and 'forest interior' (60-90 m).

81 Bray-Curtis Percent Similarity 100 75 50 _2_5 a) 8 Years Post-Harvest 0m CC , lm • 5m • 30m- 60m 45m. 15m- 90m- 75m- b) 15 Years Post-Harvest Om 15m h 30m 45m lm 60m 5m CC 90m 75m

Figure 4.7: Hierarchical clustering analysis of ground beetle community composition along edges: a) 8 years post-harvest; and b) 15 years post-harvest. Sampling transects consisted of traps placed in the clear-cut (CC), at the forest edge (0 m), and up to 90 m into the forest.

82 60 Pterostichus adstrictus J 2 Years Post-Harvest J 8 Years Post-Harvest I 15 Years Post-Harvest 45

30

15 LiWl CC 0 ] 5 15 30 45 60 75 90 CC 0 1 5 15 30 45 60 75 90

40 X. Calosoma frigidum I I 2 Years Post-Harvest H^ 8 Years Post-Harvest •• 15 Years Post-Harvest 30

20

10

l|l I iFjll CC 0 1 5 15 30 45 60 75 90 CC 0 I 5 15 30 45 60 75 90

Trap location (m) Trap location (m)

Figure 4.8: Percentage of standardized catch for each of four ground beetle species collected along forest edges. Transects consisted of traps placed in the clear-cut (CC), along the forest edge (0 m), and up to 90 m into the forest. Stands with edges of three different ages (2, 8, and 15 years post-harvest) were sampled.

83 Chapter 5: Discussion Main Results My research supports the argument that the study of insect assemblages is useful for evaluating forest management practices with respect to biodiversity conservation. I believe that the forgoing dissertation has supported this notion in several ways. I showed that two dominant epigaeic taxa, ground beetles (Coleoptera: Carabidae) and rove beetles (Coleoptera: Staphylinidae), that inhabit boreal forests responded strongly to the forest harvesting practices studied here. Thus, my work supports previous studies that recommend use of these taxa in research relevant to improving forest management (Niemela et al. 1993, Spence et al. 1996, Gandhi et al. 2001, Heliola et al. 2001, Pohl et al. 2007, Spence et al. 2008). In this general discussion, I aim to summarize and synthesize my findings and relate them to a broader discussion about sustainable management of the boreal forest. In chapter 2,1 demonstrated that retention patch size directly affects ground and rove beetle diversity in two of the principal cover-types, deciduous dominated and conifer dominated, of the mixedwood boreal forest. I also showed how the effect might be at least partly accounted for by variation in ambient air temperature across patch size. Beetle responses to retention patch size, in both conifer and deciduous forests, were strong, and suggested 2-6 hectares as a threshold size for measurable impact. In conifer stands, even small retention patches (i.e., 0.5- 1.5 ha) provided some conservation benefit to beetles; however, beetle assemblages that closely resembled those of the mature forest were found only in patches >1.8 ha. Within deciduous stands, small patches (i.e., <0.6 ha) provided little benefit for conservation of mature forest beetle species, but patches between 3-6 ha seemed sufficient to preserve species representative of the mature forest. The identification of such thresholds not only supports patch size requirements previously hypothesized for other taxa (Bradshaw 1992, Schieck and Hobson 2000), but it also provides forest managers with a tractable management tool. I focused Chapter 3 on comparison of the responses of common and rare species to retention patch isolation and demonstrated that rare species were more sensitive to isolation than common species. Unlike most previous studies which disregard data about rare species as 'noise', I explicitly compared responses between rare and common species. Although no influence of isolation could be detected on common species, analyses using increased weightings for rare species indicated that patch isolation was associated with diminished presence and

84 capture rates of rare species. These results were further strengthened by examining the biological attributes of the rare species most influential in the analyses. In general, these species were found to be: 1) associated with forest floor characteristics; 2) rare in other studies of ground dwelling fauna in Alberta; and 3) rare also in studies that utilized trapping techniques other than pitfall traps (e.g., window traps). Thus, although analysis of rarity patterns can be challenging, my analytical approach appears to be robust, and clearly highlights the advantages of including rare species in analyses. From chapter 3,1 conclude that patch isolation is an important consideration in assessing impacts of forest fragmentation, and should be considered in harvesting designs. In chapter 4,1 documented the existence of a narrow, abrupt edge effect on the boreal ground beetle community and provided a possible explanation of ecological processes driving the beetle response patterns observed in Chapters 2 and 3. I also showed that edge effects disappear over time along deciduous forest edges as the forest regenerates. This work provides a useful connection to ecological processes and contributes to broader understanding of how cover-type and time since harvest influence the degree of edge penetration. The results indicated that edge effects extend between 5 and 15 m in both deciduous and conifer cover-types, although the patterns were much clearer in conifer stands. My findings support previous studies of edge effects in the boreal forest (Spence et al. 1996, Heliola et al. 2001), and suggest a generalized response of ground beetles to recent forest harvest edges. The abrupt response that I observed over a relatively short distance appears to be driven by microclimatic gradients along the forest floor. The comparisons of older deciduous edges in this chapter also provided an interesting perspective about recovery from edge effects over time. Edge effects do not persist forever, and my results suggest that post-harvest edges begin to recover in deciduous stands at about 15 years post-harvest. This pattern again appears to be related to microclimatic changes occurring along the forest edge over time. The results from this chapter suggest: 1) there is no conservation benefit of edge effects for ground beetles in the boreal forest; 2) that edges should be minimized by leaving retention patches of sufficient size to preserve populations of mature forest species; and 3) that edge effects in deciduous forests disappear over time.

85 Implications for Forest Management The research presented in this dissertation provides timely, relevant, and useful information for forest managers. First and foremost, the research supports the notion that aggregated retention patches, as inspired by natural disturbance, can serve a critical role in sustainable forest management and biodiversity conservation (Gandhi et al. 2001, Lindenmayer and Franklin 2002, Gandhi et al. 2004). Thus, forest companies may benefit from closer consideration of retention patch size, and location (i.e., isolation) within harvest block designs. However, in Alberta, only a few companies presently recognize the value of larger retention patches and take full advantage of them in their harvest planning (e.g., Daishowa-Marubeni International Ltd. 2008). This void has, in part, been driven by a lack of information about the size threshold of effective retention patches (Bradshaw 1992, Schieck and Hobson 2000, Matveinen-Huju et al. 2006). This previous lack of information, is further amplified by the fact that the Alberta government currently has only vague guidelines about dimensions of retention patches on public lands (Alberta Sustainable Resource Development 2008). For example, companies are encouraged to utilize Targe' and 'small' retention patches to varying degrees depending on harvest block size (Alberta Sustainable Resource Development 2008), but what exactly Targe' and 'small' denote is not clear. Evidence from this study, and supported by previous predictions (Bradshaw 1992, Schieck and Hobson 2000, Pearce et al. 2005), suggests that the Targe' patches should be greater than 2-5 ha, and that small patches should be greater than 0.5 ha. The highly variable nature of patch size distributions following wildfire (Eberhart and Woodard 1987, DeLong and Tanner 1996), however, also suggests that managers following the natural disturbance model should embrace variation in patch sizes across harvested landscape, at least until the implications of patch size are more completely understood. While the topic of patch size has garnered much scientific interest (Bradshaw 1992, Schieck and Hobson 2000, Gandhi et al. 2001, Gandhi et al. 2004, Lemieux and Lindgren 2004, Matveinen-Huju et al. 2006), few studies have explored the role of patch isolation in conserving biodiversity of mature forests. The potential significance of patch isolation flows from the Theory of Island Biogeography (Macarthur and Wilson 1967), and such considerations have been important in broader scale studies of landscape fragmentation (Watling and Donnelly 2006 and references therin). Thus, there is a clear need to mitigate the impacts of patch isolation on biodiversity. My results suggest that retention patches isolated > 140 metres from the mature

86 forest will suffer loss of rare species characteristic of the mature forest. Forest managers designing large harvest blocks can utilize this information to ensure harvest block designs acknowledge isolation stresses on biodiversity. One option available is to use larger retention patches at greater distances from the mature forest. In addition to including larger populations to ensure local persistence, this would also provide larger source habitats for these rare species at greater distances from the mature forest (Bender et al. 1998), thereby increasing their landscape persistence over time (Ouborg 1993). However, fires are not selective of where large and small patches occur (Andison 2004) and so there may still be unexplored and undefined functional reasons to distribute larger patches more randomly on harvested landscapes with respect to distances from the mature forest. Rare species are more sensitive to landscape fragmentation (Davies et al. 2000, Henle et al. 2004), and this has been linked to limited dispersal abilities (Kunin and Gaston 1993), low population levels (Henle et al. 2004), as well as other factors which affect rare species at fine scales (Kunin and Gaston 1993). My findings, which illustrate the sensitivity of rare species, highlight the importance of monitoring, and understanding rare species within managed landscapes. In addition, rare species represent the lions share of species diversity on a landscape (Magurran and Henderson 2003), further emphasizing the need to manage for them. However, there has been little attention to rarity in forest management research (but see Hannon et al. 2004), or in broad scale monitoring programs (Alberta Biodiversity Monitoring Institute 2007). Managers and scientists together must recognize the value in considering rare species in such initiatives, and design research and monitoring programs to reflect the importance of rare species on a landscape (Spence et al. 2008).

Future Research and Limitations of the Dissertation As is the nature of science, this dissertation has generated a number of ideas for future research in the focal area addressed. In particular, research attention to the following five issues partially explored in my thesis would advance our understanding of the utility and effective deployment of retention patches as a component of sustainable forest harvest practices. 1) There would be much practical value in improving understanding of how patch size influences organisms of other taxonomic groups. Although such research has been conducted on a few other groups in boreal forests (Bradshaw 1992, Schieck and Hobson 2000, Chan-McLeod and Moy 2007), few studies have considered the size distribution of patches in the detail that I

87 did in this thesis. Colleagues here at the University of Alberta are currently working towards this broader understanding on additional arthropod taxa (B. Bodeux, S. Lee, personal communication) using the more specific study site adopted in Chapter 3. Collectively, such work is likely to produce results that improve our understanding of patch requirements for biodiversity. In addition to work with arthropods there is a specific need to also understand vertebrate responses to retention patch size. Bird communities in particular have been used in the past (Schieck and Hobson 2000), and should be a focal taxon for the study of patch sizes in the range presented in this thesis. Birds are particularly relevant as studies could incorporate analysis of fecundity and mortality within retention patches (e.g., Porneluzi and Faaborg 1999, Donovan and Thompson 2001), an important variable in long term conservation. 2) More detailed data about stand structure and environmental variables within retention patches should be collected in work with all taxa. Unfortunately, logistical restrictions constrained my ability to collect detailed information about variables such as stand structure, litter depth, shrub cover, and others which can be important for explaining beetle distributions (Lovei and Sunderland 1996, Magura 2002). However, collection and analysis of these variables may facilitate a better understanding of the microsite processes influencing beetle distributions around pitfall traps (but see Niemela and Spence 1994), and may provide insight into specific habitat features of high conservation value. This work will compliment a new wave of effort (e.g., Work et al. 2003, Montigny and MacLean 2005, Spence et al. 2008) attempting to use ecosystem classification as a broad management-oriented surrogate for biodiversity. 3) Given that retention patches are meant to emulate fire skips on a landscape, there is useful potential to examine how well current retention patches emulate these fire skips. Gandhi et al. (2001, 2004) attempted such a comparison but were unable to control for disturbance age due to logistical limitations on the landbase. Therefore, studies which directly compare fire origin and harvest origin retention, while controlling more completely for variables such as disturbance age, and stand structure would greatly increase understanding of natural disturbance emulation through harvesting. Those pursuing such studies may wish to compare fire skips on both lowland and upland ecosites, if available, to generate a broader understanding of fire processes in a diversity of habitats. 4) We need more information about processes that generate isolation effects, although these are admittedly hard to study, particularly at a landscape scale. Fortunately I had a study

88 site that permitted some analysis of these processes; however, the conclusions in this thesis are based on results within a single harvest block and not easily generalized across a landscape. I recognize that connectivity and isolation processes are much more complex than illuminated in this thesis, and thus a more thorough analysis of processes would be beneficial. Again, avian taxa, although operating at a scale much different than arthropods, may prove beneficial for such comparisons as movement between patches and influences of connectivity can more easily be sampled (e.g., Norris and Stutchbury 2001). Studies such as this could prove extremely beneficial for not only evaluating the impacts of patch distance from mature forest, but also in relation to how proximity to other patches affects persistence of taxa on an industrial landscape. Such information, would surely help inform design of improved retention patterns. 5) The topic of edge effect recovery remains of great importance for developing a broader understanding of edge processes, and additional study of these effects could yield new results with significant practical implications. Although my results were limited to deciduous stands, I did observe recovery from edge effects in this cover-type, a feature that is not often factored into landscape biodiversity models. Nonetheless, the need to understand recovery patterns along conifer forest edges remains high germane to effective mixedwood management. For full understanding, such work should be linked to paired measurement of microclimatic variables. My conclusions were limited to suggesting associations between beetle responses and microclimatic variables based on correlation data. A detailed understanding of this connection would be a useful outcome of future studies. In the forgoing chapters I have clearly demonstrated how studies in insect community ecology can facilitate a broader understanding of processes central to sustainable forest management. The results presented in this dissertation serve to increase scientific understanding of patch size dynamics on biodiversity, and to provide practical information to improve forest management applications. Therefore, this work contributes to the adaptive management process and to the goal of advancing sustainable forest management in Canada.

Literature Cited Alberta Biodiversity Monitoring Institute. 2007. Scope and application of the ABMI's data and information. Alberta Biodiversity Monitoring Institute. Alberta Sustainable Resource Development. 2008. Alberta timber harvest planning and operating ground rules framework for renewal. Public Lands and Forests Division, Forest Management Branch.

89 Andison, D. W. 2004. Island remnants on foothills and mountain landscapes of Alberta: Part 2 on residuals. Alberta Foothills Disturbance Ecology Research Series, Report No. 6, November 1994. Foothills Model Forest, Hinton, Alberta. Bender, D. J., T. A. Contreras, and L. Fahrig. 1998. Habitat loss and population decline: A meta­ analysis of the patch size effect. Ecology 79:517-533. Bradshaw, F. J. 1992. Quantifying edge effect and patch size for multiple-use silviculture - a discussion paper. Forest Ecology and Management 48:249-264. Chan-McLeod, A. C. A. and A. Moy. 2007. Evaluating residual tree patches as stepping stones and short-term refugia for red-legged frogs. Journal of Wildlife Management 71:1836- 1844. Daishowa-Marubeni International Ltd. 2008. Summary documentation: Detailed forest management plan-revision 2007. Daishowa-Marubeni International Ltd., Peace River Pulp Division. Davies, K. F., C. R. Margules, and K. F. Lawrence. 2000. Which traits of species predict population declines in experimental forest fragments? Ecology 81:1450-1461. DeLong, S. C. and D. Tanner. 1996. Managing the pattern of forest harvest: Lessons from wildfire. Biodiversity and Conservation 5:1191-1205. Donovan, T. M. and F. R. Thompson. 2001. Modeling the ecological trap hypothesis: A habitat and demographic analysis for migrant songbirds. Ecological Applications 11:871-882. Eberhart, K. E. and P. M. Woodard. 1987. Distribution of Residual Vegetation Associated with Large Fires in Alberta. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 17:1207-1212. Gandhi, K. J. K., J. R. Spence, D. W. Langor, and L. E. Morgantini. 2001. Fire residuals as habitat reserves for epigaeic beetles (Coleoptera : Carabidae and Staphylinidae). Biological Conservation 102:131-141. Gandhi, K. J. K., J. R. Spence, D. W. Langor, L. E. Morgantini, and K. J. Cryer. 2004. Harvest retention patches are insufficient as stand analogues of fire residuals for litter-dwelling beetles in northern coniferous forests. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:1319-1331. Hannon, S. J., S. E. Cotterill, and F. K. A. Schmiegelow. 2004. Identifying rare species of songbirds in managed forests: application of an ecoregional template to a boreal mixedwood system. Forest Ecology and Management 191:157-170. Heliola, J., M. Koivula, and J. Niemela. 2001. Distribution of carabid beetles (Coleoptera, Carabidae) across a boreal forest-clearcut ecotone. Conservation Biology 15:370-377. Henle, K., K. F. Davies, M. Kleyer, C. Margules, and J. Settele. 2004. Predictors of species sensitivity to fragmentation. Biodiversity and Conservation 13:207-251. Kunin, W. E. and K. J. Gaston. 1993. The biology of rarity - patterns, causes and consequences. Trends in Ecology & Evolution 8:298-301. Lemieux, J. P. and B. S. Lindgren. 2004. Ground beetle responses to patch retention harvesting in high elevation forests of British Columbia. Ecography 27:557-566. Lindenmayer, D. B. and J. F. Franklin. 2002. Conserving forest biodiversity: A comprehensive multi-scaled approach. Island Press, Washington, D.C. Lovei, G. L. and K. D. Sunderland. 1996. Ecology and behavior of ground beetles (Coleoptera: Carabidae). Annual Review of Entomology 41:231-256. Macarthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, New Jersey, USA.

90 Magura, T. 2002. Carabids and forest edge: spatial pattern and edge effect. Forest Ecology and Management 157:23-37. Magurran, A. E. and P. A. Henderson. 2003. Explaining the excess of rare species in natural species abundance distributions. Nature 422:714-716. Matveinen-Huju, K., J. Niemela, H. Rita, and R. B. O'Hara. 2006. Retention-tree groups in clear- cuts: Do they constitute 'life-boats' for spiders and carabids? Forest Ecology and Management 230:119-135. Montigny, M. K. and D. A. MacLean. 2005. Using heterogeneity and representation of ecosite criteria to select forest reserves in an intensively managed industrial forest. Biological Conservation 125: 237-248, Niemela, J., D. Langor, and J. R. Spence. 1993. Effects of clear-cut harvesting on boreal Ground- beetle assemblages (Coleoptera, Carabidae) in western Canada. Conservation Biology 7:551-561. Niemela, J. K. and J. R. Spence. 1994. Distribution of forest dwelling carabids (Coleoptera)- spatial scale and the concept of communities. Ecography 17:166-175. Norris, D. R. and B. J. M. Stutchbury. 2001. Extraterritorial movements of a forest songbird in a fragmented landscape. Conservation Biology 15:729-736. Ouborg, N. J. 1993. Isolation, population-size and extinction - the classical and metapopulation approaches applied to vascular plants along the Dutch Rhine-System. Oikos 66:298-308. Pearce, J. L., L. A. Venier, G. Eccles, J. Pedlar, and D. McKenney. 2005. Habitat islands, forest edge and spring-active invertebrate assemblages. Biodiversity and Conservation 14:2949- 2969. Pohl, G. R., D. W. Langor, and J. R. Spence. 2007. Rove beetles and ground beetles (Coleoptera: Staphylinidae, Carabidae) as indicators of harvest and regeneration practices in western Canadian foothills forests. Biological Conservation 137:294-307. Porneluzi, P. A. and J. Faaborg. 1999. Season-long fecundity, survival, and viability of Ovenbirds in fragmented and unfragmented landscapes. Conservation Biology 13:1151- 1161. Schieck, J. and K. A. Hobson. 2000. Bird communities associated with live residual tree patches within cut blocks and burned habitat in mixedwood boreal forests. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 30:1281-1295. Spence, J. R., D. W. Langor, J. M. Jacobs, T. T. Work, and W. J. A. Volney. 2008. Conservation of forest-dwelling arthropod species: simultaneous management of many small and heterogeneous risks. Canadian Entomologist 140:510-525. Spence, J. R., D. W. Langor, J. Niemela, H. A. Carcamo, and C. R. Currie. 1996. Northern forestry and carabids: The case for concern about old-growth species (vol 33, pg 173, 1996). Annales Zoologici Fennici 33:302-302. Watling, J. I. and M. A. Donnelly. 2006. Fragments as islands: a synthesis of faunal responses to habitat patchiness. Conservation Biology 20:1016-1025. Work, T. T., J. R. Spence, W. J. A. Volney, L. E. Morgantini and J. L. Innes. 2003. Integrating biodiversity and forestry practices in western Canada. Forestry Chronicle 79: 906-917.

91