INSECT COMMUNITIES AND MULTICOHORT STAND STRUCTURE IN BOREAL MIXEDWOOD FORESTS OF NORTHEASTERN ONTARIO

By

Erica Patricia Barkley

A thesis submitted in conformity with the requirements for the degree of Master of Science in Forestry, Faculty of Forestry University of Toronto

© Copyright by Erica Patricia Barkley 2009

Barkley, Erica Patricia. 2009. communities and multicohort stand structure in boreal mixedwood forests of northeastern Ontario. Master of Science in Forestry, Faculty of Forestry, University of Toronto.

ABSTRACT

Current forest management in boreal northeastern Ontario results in young, even-aged forests; however, fire history research has found old stands with multiple cohorts of trees are common, supporting the value of Multi-cohort Management. I investigated relationships between insect communities and stand live-tree diameter distribution, cohort class and structure. Results showed that variation in abundances of Carabidae, Diapriidae, Diptera and Hymenoptera were not strongly predicted by cohort class. The concept showed greater strength when parameters of live- tree diameter distributions were used. Forest structure, not age, was important for all communities, including heterogeneity of understory and/or overstory vegetation. Trap height was a strong predictor of aerial insect community structure, with insect abundance higher in the understory than in the canopy. In summary, a more diversified classification approach which includes important habitat features in addition to simple characterization of diameter distributions should be considered to better assess forest structural variation and management.

ii ACKNOWLEDGMENTS

I would like to thank my supervisors, Dr. Jay Malcolm and Dr. Sandy Smith and my committee members, Dr. Peter DeGroot and Dr. Chris Darling for your guidance and support throughout the process.

Thank-you to the people who have contributed greatly to the MCM project: Charlotte Sharkey, Kathleen Ryan, and Ben Kuttner. Special thank-you to Charlotte Sharkey, who worked hard on determining the MCM sites, and gave permission to use two figures from her thesis.

With regards to my taxonomic learning adventure, I would like to thank the following taxonomists for their assistance and guidance: Nurul Islam, Henri Goulet, Bill Crins, John Huber, and Jan Klimaszewski. Thank-you to Brad Hubley and Antonia Guidotti, at the Royal Ontario Museum for letting me come in to study the Carabidae collections.

For all their help in the lab, my work-study students, Dorothy Maguire and Natallia Revinskaya, deserve a large thank-you for putting the many hours of processing samples for my project.

To the field assistants, Dave Swinson and John Kearns, and also the Kapuskasing field crew: your efforts and support was imperative to the project’s successful completion.

For the many insightful discussions and support over the last two years, I would like to thank my peers in the Wildlife Ecology and Forest Entomology labs as well as my good friends and colleagues in the faculty. Especially, to all the CONFOR 2009 committee members: thank-you for your support, encouragement, and the good memories.

Lastly, I’d like to thank my family and friends, and my fiancé and fellow researcher, Mike Burrell. Your love and support helped make this experience a positive one.

FUNDING for the Multi-cohort Management project has been graciously provided by Forestry Futures Trust, Tembec Inc., Lake Abitibi Model Forest, Ontario Ministry of Natural Resources, National Science and Engineering Research Council of Canada, Faculty of Forestry, Edward H. Buckley and Grace B. Buckley Scholarship, and the Government of Ontario/Adam Zimmerman scholarship for students working in forest conservation.

iii TABLE OF CONTENTS

ABSTRACT...... i ACKNOWLEDGMENTS...... ii LIST OF TABLES ...... v-vi LIST OF FIGURES ...... vii-ix LIST OF APPENDICES ...... x GENERAL INTRODUCTION...... 1-5

CHAPTER 1- DOES MULTI-COHORT STAND STRUCTURE AFFECT GROUND (FAMILY: CARABIDAE) COMMUNITIES IN NORTHESTERN BOREAL FORESTS?...... 6-49 INTRODUCTION...... 6-10 METHODS...... 10-22 Study area...... 10 Study design...... 13-22 RESULTS...... 23-40 DISCUSSION...... 41-49

CHAPTER 2 – CHAPTER 2- VERTICAL STRATIFICATION AND EFFECTS OF MULTI-COHORT MANAGEMENT IN THE ABUNDANCE OF DIPTERAN AND HYMENOPTERAN FAMILIES IN BOREAL MIXEDWOOD FORESTS…………...50-105 INTRODUCTION...... 50-55 METHODS...... 55-61 RESULTS...... 62-97 DISCUSSION...... 98-105 CONCLUSIONS AND MANAGEMENT IMPLICATIONS...... 106-109 LITERATURE CITED...... 110-118 APPENDICES...... 119-125

iv LIST OF TABLES

Table 1.1- Study sites in upland mixedwood stands in northeastern Ontario, representing a range of cohort, age, and management histories. Modified from Sharkey (2008).

Table 1.2- Abundances of carabid beetle species in 2007 in boreal mixedwoods of northeast Ontario, standardized to 1000 trap-days, in structural cohort classes. Organization of species is based on ecological synopses in Lindroth (1961-1969) and Larochelle & Lariviere (2003).

Table 1.3 - Redundancy analysis significance (forward selection) examining correlations between carabid communities of boreal mixedwoods of northeastern Ontario and Weibull shape and scale of all stems >2.5cm dbh (Weibull 2.5) and Weibull shape and scale of all stems >10 cm dbh (Weibull 10). Shown are percent variance explained and P values (significant values in bold).

Table 1.4- Partial redundancy analysis examining the relative value of various habitat predictors in explaining carabid community composition from pitfall traps in boreal mixedwoods of northeastern Ontario in 2007 in comparison to Weibull parameters from the diameter distribution of stems >2.5 cm dbh.

Table 1.5- As Table 1.4 except for the stand level variables age, productivity (based on FEC herb richness) and percentage of deciduous composition.

Table 2.1- Mean abundance (standardized to 100 trap-days) and univariate test results according to cohort classes (1-4) and trap heights (canopy versus understory) for families of Diptera and Hymenoptera collected from understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007.

Table 2.2 - Significance of all canonical axes for the Diptera or Hymenoptera families correlation matrix (square-root transformed) in understory and canopy traps in boreal mixedwoods of northeastern Ontario and their relationship to cohort class and Weibull parameters of all stems (>2.5cm dbh).

Table 2.3- Partial redundancy analysis examining the relative value of various habitat predictors in explaining diptera and hymenopteran communities from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007 in comparison to Weibull parameters from the diameter distribution of stems >2.5 cm dbh.

Table 2.4- same as Table 2.3, except for canopy Malaise trap samples

Table 2.5- same as Table 2.3, except examining age, productivity and percentage of deciduous composition for both understory and canopy Malaise trap samples

Table 2.6- Mean abundance (standardized to 100 trap-days) and univariate test results of cohort class effects for Diapriidae morphospecies collected from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007. Description of each morphospecies can be found in Appendix 3.

v

Table 2.7- Partial redundancy analysis examining the relative value of various habitat predictors in explaining diapriid communities from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007 in comparison to Weibull parameters from the diameter distribution of stems >2.5 cm dbh.

Table 2.8- same as Table 2.7 but examining age, productivity and percentage of deciduous composition in comparison to Weibull parameters

vi LIST OF FIGURES

Figure 1.1 - Locations of 18 mixedwood study sites in the Gordon Cosens and Iroquois Falls forest management units in northeastern Ontario. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Black lines represent roads and highways; grey shading indicates forest management units (FMUs) and site codes are in Table 1.1. Modified from Sharkey (2008)

Figure 1.2- Weibull shape versus scale for boreal mixedwood sites in northeastern Ontario (clustered with k-means non-hierarchical clustering into four a priori groups): ♦ cohort 1; ▲ cohort 2; ● cohort 3; ■ cohort 4. Numbers in grey boxes indicate the mean for a cohort class. Live stem diameter distributions, with fitted Weibull curves, are shown for an example site for each cohort class. From Sharkey (2008).

Figure 1.3 - Principal Components Analysis (PCA) of structural variables in mixedwood sites in northeastern Ontario. ● cohort 1; □ cohort 2; ○ cohort 3; ■ cohort 4. The stand age vector was plotted passively. Grey triangles represent cohort class centroids. Codes for structural features can be found in Appendix 1. From Sharkey (2008).

Figure 1.4- Redundancy Analysis (RDA) for Carabidae in boreal mixedwood sites in northeastern Ontario in 2007, constrained by Weibull parameters of all stems >2.5cm dbh. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Significance of each parameter is shown (**= P<0.01, *= P<0.05). Species codes are in Appendix 2.

Figure 1.5- As Figure 1.4 except that Weibull parameters are for stems >10 cm dbh.

Figure 1.6- As Figure 1.4 except that data is constrained by cohort class.

Figure 1.7 - Principal Components Analysis (PCA) for Carabidae (correlation matrix) from pitfall traps in 2007 in boreal mixedwoods of northeastern Ontario with environmental variables plotted passively (only those with axis scores >0.40 shown). ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Significance of each variable is shown (**= P<0.01, *= P<0.05).

Figure 1.8. Carabid species richness plotted against abundance from pitfall traps in boreal mixedwoods of northeastern Ontario in 2007, with fitted regressions shown. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4.

Figure 1.9 – Principal Components Analysis (PCA) of carabid community metrics in of boreal mixedwood sites in northeastern Ontario in 2007, with environmental variables (only those with axis scores >0.40 shown) plotted passively. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Significance of each parameter is shown (**= P<0.01, *= P<0.05. += P<0.05 after rsvund or rvarcanht was selected). Variable acronyms are identified in Appendix 1 and 2.

Figure 2.1- Malaise trap set-up for canopy and understory malaise in boreal mixedwood forests of northeastern Ontario.

vii Figure 2.2- Abundance per 100 trap days of A. Diptera and B. Hymenoptera for cohort classes 1- 4 and understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007. Standard error bars shown. Note contrasting scale.

Figure 2.3- Family diversity (Shannon Wiener indices) of Diptera and Hymenoptera for cohort classes 1-4 and understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007. Standard error bars shown.

Figure 2.4- Family richness against unstandardized abundance (Number of individuals) for A. Diptera and B. Hymenoptera from understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007, with fitted regressions shown. Note contrasting scale. ●=understory □=canopy.

Figure 2.5– Principal Components Analysis (PCA) for families of A. Diptera and B. Hymenoptera found in boreal mixedwood sites in northeastern Ontario caught in understory and canopy Malaise traps. Only families found in >10% of sites shown. ABI0217 outlier removed (TEM01095 removed from Hymenoptera); □=canopy ●=understory. Variable acronyms are identified in appendix 2.

Figure 2.6 – Redundancy Analysis (RDA) of dipteran families in the A. understory and B. canopy Malaise traps constrained by Weibull parameters from boreal mixedwood sites in northeastern Ontario in 2007. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Only families found in >10% of sites shown. ABI0217 outlier removed. Variable acronyms are identified in appendix 1 and 2.

Figure 2.7– As Figure 2.6, except that ordinations are for hymenopteran families in the A. understory and B. canopy

Figure 2.8 – Principal Components Analysis (PCA) of dipteran families A. in understory traps and B. in canopy Malaise traps from boreal mixedwood sites in northeastern Ontario with environmental variables plotted passively (only those with axis scores >0.40 shown) in 2007. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Only families found in >10% of sites shown. ABI0217 outlier removed. Significance for each variable are shown (**= P<0.01, *= P<0.05, +=0.05 in forward selection). Acronyms for variables are found in Appendix 1 and 2.

Figure 2.9- as Figure 2.8 except that ordinations are for hymenopteran families A. in understory traps and B. in canopy Malaise traps.

Figure 2.10 – DCA ordination of diapriid morphospecies classified by cohort class from boreal mixedwood sites in northeastern Ontario in 2007 with environmental variables plotted passively (only those with axis scores >0.35 shown). ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Only families found in >10% of sites shown. *=P<0.05, **= P<0.01. ++= P<0.01 after rsvover was selected. Variable acronyms are identified in Appendix 1 and 2.

viii Figure 2.11- Mean morphospecies abundance per 100 trap days for Diapriidae from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007, shown in cohort classes 1-4. Standard error bars shown. No significant effects were detected.

Figure 2.12- Mean Shannon-Wiener diversity of morphospecies per 100 trap days for Diapriidae from understory Malaise traps in boreal mixedwoods of northeastern Ontario, shown in cohort classes 1-4. Standard error bars shown. No significant effects were detected.

Figure 2.13- Family richness against unstandardized abundance for diapriid morphospecies richness from understory Malaise Traps in boreal mixedwoods of northeastern Ontario in 2007, with fitted regressions shown. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4.

ix

LIST OF APPENDICES

Appendix 1. Ordination diagram codes for forest structure features collected in boreal mixedwoods of northeastern Ontario in 2007

Appendix 2. Ordination diagram species codes used for Carabidae species (pitfall traps), Diptera families and Hymenoptera families (understory and canopy malaise trapping) in boreal mixedwoods of northeastern Ontario in 2007

Appendix 3. Diapriidae male Belytinae morphospecies with their morphometric measurements from understory malaise trapping in Ontario’s northeastern boreal forest in 2007. All measurements are in mm, measured using micrometer at 12 X magnification in the dissecting microscope.

x 1

GENERAL INTRODUCTION

Biodiversity conservation is a key goal for sustainable forest management. In natural systems, periodic disturbances are important in providing habitat for a diverse array of species

(Loucks 1970); hence silvicultural techniques that attempt to emulate natural disturbance regimes have been suggested as coarse-filter strategies to maintain biodiversity (Attiwill 1994,

Bergeron and Harvey 1997, Hunter 1999). The goal in this approach is to emulate the size, frequency and severity of disturbances that species are adapted to, increasing their chances for persistence on the landscape (Franklin 1993, Hunter 1999).

In Ontario, timber harvesting must emulate natural disturbances within the boundaries of silviculture requirements (OMNR 2001). The main disturbances in Ontario’s boreal forest are fire and insect outbreaks (e.g., spruce budworm), both of which are often stand replacing

(Bergeron et al. 2001). Fire can influence tree species distributions, age-class distributions, productivity and wildlife habitat characteristics in forest landscapes (Bergeron et al. 2007). The generalization that boreal forests are subject to fairly short fire cycles has been used to support clear-cut silviculture on a relatively short (c. 100-year) rotation, resulting in even-aged, homogenous forests (Bergeron et al. 2001). However, such even-aged management does not produce a distribution of stand ages that would occur naturally. For example, if the probability of a burn is mainly independent of age (Johnson 1992), the age-class distribution of fire-origin stands follows a negative exponential curve, wherein 37% of the stands are older than the fire cycle (Johnson & VanWagner 1985, Bergeron et al. 1999). Indeed, in north-eastern Ontario’s boreal forest, the fire history estimated for the Lake Abitibi Model Forest revealed stands that had escaped fire for 200-300 years, resulting in 50% of the unmanaged forest mosaic composed of "over-mature" and old-growth stands (Bergeron et al. 2001). These relatively old stands are

2 dominated by a very different natural disturbance regime, one which is not represented in clear-cut silviculture (Bergeron et al. 2001). The net result is one where species associated with such stand conditions could be eliminated from the landscape if even-aged management continues to truncate the natural stand-age distribution (Bergeron 2004). On the other hand, if the annual allowable cut (AAC) is decreased to create longer harvest rotations, and thereby increase the proportion of mature and old-growth stands in the landscape, the reduction in timber volume could damage an already struggling forest industry (Bergeron et al. 1999).

A multi-cohort approach has been suggested as a solution to this problem (Bergeron et al.

1999). The idea is to use a combination of silvicultural practices to maintain the structure of stands in a variety of structural successional stages, including old growth conditions, without a drastic decrease in the AAC (Bergeron et al. 1999, Bergeron et. al. 2001, Harvey et. al. 2002).

Under Multi-cohort management (MCM), forests would not just contain single-cohort stands resulting from clear-cutting; they would also include multi-cohort stands with mixed tree age and size classes resulting from selection harvesting and partial cutting (Bergeron et al. 1999,

Bergeron et. al. 2001). Depending on forest structure and composition, a stand could be sent into stand-initiation via clear-cutting, or guided into a more multi-cohort structure by use of partial- cutting (Bergeron et al. 2002). The proportion of stands in certain cohorts and the particular harvesting treatments would depend upon the natural disturbance regime of a region (Bergeron et al. 2002).

At the same time, however, stand dynamics are variable: stands of the same age may not always have the same composition and structure, making it difficult to base cohort class on time since fire as many have done (e.g., Harvey et al. 2002). Sharkey (2008) explored the extent to which live-tree diameter distributions using Weibull curves and the cohort concept explained

3 forest structural and habitat development as had been suggested by other authors (Nyugen 2000,

Harvey et al. 2002, Boucher et al. 2003, Kuttner 2006). In keeping with the structural cohort concept, she found that diameter distribution parameters corresponded to the structural development of certain key habitat variables, including canopy height, overstory foliage thickness, shrub openness, vertical foliage complexity, live tree basal area, understory foliage thickness and volume of late decay-class downed woody debris (DWD) (Sharkey 2008). Weibull parameters were correlates of small mammal community structure in boreal mixedwoods, and had higher explanatory power than stand age, but were poor correlates of community structure in black spruce-dominated forests. Of particular interest, she found that variation among small mammal communities was correlated with tree diameter distributions as envisioned in the multi- cohort concept (Sharkey 2008). In addition, statistics derived from the distribution of stem diameters, which are key in defining cohort classes (Nyugen 2000), performed better at describing small mammal community composition than certain other coarse filter habitat variables, such as the variation in the quality and quantity of downed woody debris (Sharkey

2008).

Multi-cohort Management has the potential to improve on existing forest management strategies, but requires further research to guide its implementation. Aside from Sharkey's (2008) work on small mammals, information on wildlife communities as a function of stand cohort-type variation is lacking. Key questions include, for example, to what extent can the cohort structure of a stand describe habitat variation? Is stand age decoupled from stand cohort structure, that is, do different cohort structures describe different wildlife communities, even if stands are of the same approximate age? The latter is a key assumption of the multi-cohort approach, which supposes that silvicultural manipulation of stand cohort structure can be used to emulate forest

4 development normally associated with stand age. Sharkey’s (2008) findings give some insight on how wildlife may respond to cohort structure; however, it is important to recognize inherent differences in habitat associations among taxonomic groups.

Insect communities are of particular interest in this regard because of the diverse range of niches that they occupy in part due to their diverse feeding habits (e.g., plant feeders, predators, parasites, parasitoids, and saproxylics) and diverse microhabitat requirements (Thiele 1977,

Triplehorn & Johnson 2005). In Fennoscandia, where boreal forests have been structurally simplified due to long-term intensive harvesting, many forest species, including , are declining or extinct (Haila et al. 1994, Niemela et al. 1997). As a result, not only are insect communities suitable candidates to examine the effects of structural changes in forest stands, there also is a certain urgency to understand their relationships with forest management in

Canada, where we still have the possibility of designing management to conserve a presumably more-intact fauna.

To evaluate the impacts on biodiversity and insect communities, it is important to include assessments of both the understory and canopy fauna (Lowman & Wittman 1996, Su & Woods

2001). In considering only one stratum, one may not only risk inadequately assessing overall diversity, but variation in responses to forest management among heights also could be overlooked as a result of partial assessment of the complete forest profile. For example, Akutsu et al. (2007) found prominent effects of logging in the understory on flying insects, but a more obscure response at higher strata. Forest strata can have distinct insect communities that vary in their sensitivity to forest change (Schowalter 1995, Winchester & Ring 2006, Pucci 2008). For example, ground-level Carabidae and spiders are well known for their higher post-clear-cut abundances and lower abundance following canopy closure (Duschesne and McAlpine 1993,

5

Niemela et al. 1993, Spence et al. 1996, Beaudry 1997, Koivula et al. 2002, Buddle et al. 2006,

Niemela et al. 2007). Dolichopodidae in the understory can decrease with stand height and basal area retention (Deans et al. 2005), and herbivorous insects of the canopy can show differences between old-growth and managed forests. As a result of new and improved sampling techniques, the number of studies of insects in the forest canopies has increased dramatically (Lowman &

Wittman 1996). The majority of canopy studies so far have occurred in the tropics, and there is a lack of vertical stratification studies for northern forests, especially for boreal mixedwoods

(Winchester & Ring 1996, Vance 2002).

In chapter 1, I examine the relationships between insects and multi-cohort stand structure, focusing on ground-dwelling Carabidae due to their mature and natural history, their abundance in pitfall traps, and their suitability as a potential ‘indicator’ for finer-scale evaluations of forest change (Larochelle and Lariviere 2003, Rainio and Niemela 2003, Work et al. 2008). I examine the extent to which various forest structural features are correlates of

Carabidae community composition, and in particular focus on characteristics of the diameter distribution of live stems as envisioned in the multi-cohort management concept.

In chapter 2, I further examine the cohort concept, but in the context of vertical stratification of insect communities for Diptera and Hymenoptera (at the family level) and understory communities in one hymenopteran family, the Diapriidae (at the morphospecies level).

I used both understory and canopy Malaise traps to assess cohort effects and to examine strata associations of families and morphospecies.

6

CHAPTER 1- DOES MULTI-COHORT STAND STRUCTURE AFFECT (FAMILY: CARABIDAE) COMMUNITIES IN NORTHESTERN BOREAL FORESTS?

INTRODUCTION

Currently, Ontario’s northeastern boreal forests are managed under clear-cut silviculture on relatively short (c. 100-year) rotations, based on the assumption that stand-replacing fire cycles with relatively short return times dominate the landscape (Bergeron et al. 2001). However, in northeastern Ontario’s boreal forest, reconstruction of fire histories suggests much longer fire cycles, with many stands escaping fire for 200-300 years (Bergeron et al. 2001). As a result,

50% of the unmanaged forest mosaic is composed of "over-mature" and multi-cohort stands

(stands with multiple cohorts of trees), which are currently not being explicitly managed for under clear-cut silviculture (Bergeron et al. 2001).

As a potential solution to this problem, multi-cohort management (MCM) has been proposed, which uses a combination of clear-cut, selection and partial cutting to emulate a diversity of forest structures found in the landscape (Bergeron et al. 1999, Bergeron et. al. 2001,

Harvey et. al. 2002, Bergeron et. al. 2002). One of the primary goals of natural disturbance emulation is to manage for biodiversity, and thus an important component of understanding the appropriateness of MCM is to examine responses of wildlife communities as a function of variation in cohort structure (Attiwill 1994, Bergeron and Harvey 1997, Hunter 1999). It is possible that wildlife communities will show strong compositional variation by cohort class, which would support the need for MCM (Sharkey 2008). However, it is also possible that this coarse-filter strategy may not adequately reflect relevant variation in wildlife communities; for example, some insect taxa associated with post-fire stands can be missing in stands following a clear-cut that aims to emulate fire disturbance, and the response of species diversity and

7 abundance can differ greatly between post-fire and post-harvest stands (Buddle et al. 2006,

Holliday 1991).

The growing body of literature examining insect relationships to boreal forest management and disturbance has primarily focused on ground-dwelling taxa (e.g., Spence et al.

1996, Heliola et al. 2001, Vance and Nol 2003, Work et al. 2004, Moore et al. 2004, Buddle et al. 2006). Carabidae, also known as ground beetles, are ubiquitous, using a range of niches related to forest structure (Larochelle and Lariviere 2003) and are known to respond to forest change. Diversity and abundance of Carabidae tend to be high after clear-cutting, followed by a decrease with canopy closure as many open-habitat species are replaced by a smaller set of forest specialists (Spence et al. 1997, Niemela et al. 1993, Duschesne and McAlpine 1993, Beaudry

1997, Koivula et al. 2002, Buddle et al. 2006, Niemela et al. 2007). The majority of Carabidae can be found across a successional gradient in a closed forest, but several rare species are associated with old-growth stands (Niemela et al. 1996). The ability of ground beetles to recover to pre-disturbance levels following clear-cutting appears to be much greater than other ground dwelling taxa such as Staphylinidae, (Buddle et al. 2006). For carabid communities, it has been estimated that pre-disturbance conditions are obtained in some 30 years (Kovula et al. 2002,

Vance & Nol 2003, Buddle et al. 2006). However, one study in western Canada showed no recovery 30 years post harvest, and in many Fennoscandian forests, forest carabids are at low abundances even in relatively old forests due to structural simplification (Niemela et al. 1993,

Niemela 1997).

Forest structural variation influences microhabitat and microclimatic factors, such as temperature, humidity and light that may effect Carabidae behavior and distribution (Thiele 1977,

Work et al. 2004). High soil moisture, for instance, is often strongly related to Carabidae

8 communities under closed canopies (Niemela et al. 1992, Niemela 1997). Relationships with ground-level microhabitat features can be more apparent than stand-level forest structures

(Simila et al. 2002). The amount of exposed ground, moss cover and litter is important for forest dwelling spiders and Carabidae; even a higher proportion of aspen litter to surrounding needle litter can promote greater diversity of carabids in mixedwoods (Bultman & Uetz 1982, Niemela et al. 1992, Guillemain et al. 1997, Work et al. 2004, Cobb et al. 2007). Thicker leaf litter supports primarily forest species, and results in low species richness (Guillemain et al. 1997).

Loss of downed woody debris (DWD) can have a negative effect on saproxylic insects including

Carabidae (Niemela et al. 1992, Niemela 1997, Martikainen et al. 2000, Work et al. 2004, Cobb et al. 2007, Jacobs et al. 2007). Downed woody debris provides important larval sites for carabid species such as Agonum retractum (LeConte) and Pterostichus adstrictus (Eschscholtz) (Goulet

1974). In Canada, a few studies have suggested DWD is not yet a limiting factor in forests compared to extensively harvested stands in Fennoscandia, with stronger positive relationships between Carabidae and DWD found in clear-cut stands (Kotze et al. 2003, Pearce et al. 2003).

While focus is often on ground-level habitat structures and microclimatic factors for Carabidae, relationships between tree-associated structures and insects have been assessed in several studies.

A negative association between tree density and forest generalists has been found (Koivula et al.

2002, Work et al. 2004). Work et al. (2004) found that live crown ratio, basal area of black spruce trees, high-strata vegetation cover, and plant cover and richness were significantly associated with carabid abundance. In plantations, there can be a decrease in richness and diversity of carabids with increases in canopy cover (Humphrey et al. 1999). In hardwood stands, canopy openings due to drastic canopy changes (e.g., ice-storm damage) can result in a loss of species richness, diversity and abundance of Carabidae (Martel et al. 1991, Saint-Germain &

9

Mauffette 2001). Forest structure therefore appears to be important to Carabidae communities, suggesting the possibility that variation in stand cohort structure, and manipulation of such, can be used to conserve carabid communities. Forest specialists make up <10% of carabid species and are the most at risk (Niemela 1997). This has been seen in Fennoscandia, where certain specialists, such as Platynus mannerheimii (Dejean), do not persist in managed stands (Niemela

1997). Small-scale silviculture techniques used to manage multi-cohort stands may provide enough canopy closure and related forest structures to maintain forest carabids and staphylinids in stands (Koivula et al. 2002, Koivula & Niemela 2003, Niemela et al. 2007, Klimaszewski et al. 2008). The diverse habitat associations makes the Carabidae of particular interest in a multi- cohort context: to what extent can information on the distribution of live-tree diameters act as a coarse filter for this group, or do other habitat features perform (such as such as stand age, productivity, variation in the quality and quantity of DWD) better?

In this chapter, I examine Carabidae abundance, composition, and richness with respect to cohort class and the diameter distribution of live stems, the latter as measured by Weibull parameters providing a succinct descriptor of tree-cohort variation. I expect Carabidae diversity to decrease with cohort class as old-growth stands typically have only a few abundant species

(Niemela 1993). I expect that old-growth and forest specialists will be found more in stands with multiple tree cohorts, as opposed to more generalist species in young, even-aged stands. Based on findings by Sharkey (2008) and our knowledge of the importance fine-scale microhabitat conditions for the group, I also expect that diameter assessments based on a finer stem resolution

(measuring the diameter of trees >2.5cm dbh) will more effectively describe Carabidae community variation than the more commonly used limit of >10cm dbh since a higher resolution describes finer scale structures more important for carabids. I also determine whether certain

10 structural features have stronger influences on Carabidae community variation than others.

Structural attributes that dictate microclimate such as canopy cover and volume of woody debris are likely significant drivers of Carabidae communities based on the literature (e.g., Martikainen et al. 2000, Pearce et al. 2003, Work et al. 2004). I investigate whether Weibull parameters function better as predictors of Carabidae communities than these other important structural features, such as canopy height and downed woody debris, and other stand-level characteristics such as productivity, tree species composition and age.

METHODS

Study Area

The study was conducted in northeastern Ontario, Canada, within the northern clay region of the boreal forest (Rowe 1972; Fig. 1.1). This region is dominated by lacustrine clay soils with gentle topography and little exposed bedrock (Taylor et al. 2000). Mixedwood stands of trembling aspen (Populus tremuloides Michx.), balsam fir (Abies balsamea Mill.), and white spruce (Picea glauca Voss) are commonly found in areas of better drainage within a landscape dominated by large stands of black spruce (Picea mariana Mill) (Rowe 1972). Sites were located within two forest management units: Iroquois Falls Forest and Gordon Cosens Forest (OMNR

2009). Forest management in this region began at the beginning of the 20th century, with the

Iroquois Falls pulp mill opening in 1912 and the Spruce Falls mill in the 1920’s (Radforth 1987).

In 2007, temperatures ranged from -39.4°C to 32.4°C, and annual precipitation accumulation was 1206 mm (Weather Network 2009).

11

Table 1.1- Study sites in upland mixedwood stands in northeastern Ontario, representing a range of cohort, age, and management histories. Modified from Sharkey (2008).

Management Age since history Cohort disturbance (L = logged; Site code class (as of 2006) U = unlogged) Northing Easting KAT1C1 1 35 L 49.46300 -82.96391 LMW15 1 55 L 49.02921 -79.86804 TEM01094 1 50 L 49.38881 -83.10281 TEM01101 1 42 L 49.77533 -82.32050 CHAR01 2 59 L 49.38895 -82.37605 KAP00011 2 43 L 49.79258 -82.35767 TEM03006 2 71 L 49.42031 -82.16623 ABI0314 3 50 L 49.26096 -80.62276 TEM01083 3 54 L 49.45276 -82.27595 TEM01084 3 54 L 49.45327 -82.25859 TEM02098 3 61 L 49.64678 -82.30247 TEM0405 3 60 L 49.59431 -82.32026 ABI0310 4 66 L 49.39806 -80.47224 TEM01095 4 71 L 49.25575 -82.72356 TEM03040 4 44 L 49.69617 -82.80895 ABI0217 4 130 U 49.48577 -80.64098 GCF9411 4 110 U 49.12955 -82.31147 LISNC 4 115 U 48.82121 -82.53115

12

Figure 1.1 - Locations of 18 mixedwood study sites in the Gordon Cosens and Iroquois Falls forest management units in northeastern Ontario. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Black lines represent roads and highways; grey shading indicates forest management units (FMUs) and site codes are in Table 1.1. Modified from Sharkey (2008).

13

Study Design

Eighteen sites in upland mixedwood forests were selected to represent a range of stand structure and age since disturbance (Table 1.1; Fig. 1.1), with a particular effort to provide a similar range of stand ages for each cohort type (see below for the definition of cohort types, stand ages, and disturbance types). A range of logging histories also was sought, but there was a shortage of un-logged stands accessible by road, so only three unlogged stands (all in cohort type

4) were included. Forest composition was controlled to have a mix of deciduous and coniferous species; specifically, 23-71% hardwood and 29-77% conifer composition by basal area, with at least some poplar (1-63%) and spruce (1-52%) (Sharkey 2008). The stands are the same 18 mixedwood stands studied by Sharkey (2008).

Assignment of cohort classes

Sharkey (2008) fitted two-parameter Weibull probability density functions to live-tree diameter distributions and used the parameter estimates to classify cohorts (Nguyen 2000,

Boucher et al. 2003, Kuttner 2006, Sharkey 2008; Fig. 1.2). Here, I use the same cohort typing for the stands. A general description of the cohorts is as follows: (Cohort 1) even-aged stands with a high stocking density of small-diameter stems and a unimodal diameter distribution;

(Cohort 2) stands with a multi-modal diameter distribution, beginning to show uneven age structure; (Cohort 3) inverse-J distribution in stands with small diameter stems with some stems belonging to larger diameter classes and with a well-established multi-age structure and evident gap dynamics; (Cohort 4) inverse-J distribution with sparse tail as stands return to a more homogenous state. Because stems smaller than 2.5 cm diameter at breast height (dbh) were not measured, Sharkey (2008) subtracted 2.5 for the dbh of all stems before fitting with Weibull

14 curves, to deal with the ‘empty probability space’. To cluster the sites into cohorts, Sharkey

(2008) plotted the two parameters against each other and used cluster analysis to identify four site clusters representing four cohorts (the number of clusters was chosen to represent the cohort classes as envisioned by Kuttner (2006) and well characterized the spread of parameter values)

(Fig. 1.2). Furthermore, to verify cohort clustering, a Principal Components Analysis (PCA) of site structural variables was generated and showed four more-or-less distinct clusters of sites (Fig.

1.3).

In addition to the cohort types, I also used the two Weibull curve parameters (shape [c] and scale [σ]) as quantitative descriptors of live-tree distributions. Shape can be negative- exponential, log-normal, or normal (Sharkey 2008): a high value indicates a more normally distributed forest structure, such as in cohorts 1 and 2. Scale is the dbh size class at which 63.2% of stems have accumulated (Sharkey 2008). If scale is high, then there are more large trees, such as in cohorts 3 and 4.

15

Figure 1.2- Weibull shape versus scale for boreal mixedwood sites in northeastern Ontario (clustered with k-means non-hierarchical clustering into four a priori groups): ♦ cohort 1; ▲ cohort 2; ● cohort 3; ■ cohort 4. Numbers in grey boxes indicate the mean for a cohort class. Live stem diameter distributions, with fitted Weibull curves, are shown for an example site for each cohort class. From Sharkey (2008).

16

Figure 1.3 - Principal Components Analysis (PCA) of structural variables in mixedwood sites in northeastern Ontario. ● cohort 1; □ cohort 2; ○ cohort 3; ■ cohort 4. The stand age vector was plotted passively. Grey triangles represent cohort class centroids. Codes for structural features can be found in Appendix 1. From Sharkey (2008).

17

Habitat Sampling

At each site, a 125 by 125 m grid was established with 25 m station spacing. Structural habitat features were measured by Sharkey (2008), primarily at the 16 internal grid stations (i.e., not on the grid edges), and are summarized below.

Live trees: diameter distribution, basal area and composition

To generate live-tree diameter distributions and to calculate basal areas, data on the species and diameter at breast height (dbh) of all live stems > 2.5 cm dbh within 6 m of the 16 grid intersections was collected. Percent composition of deciduous trees also was calculated at each site based on these data.

Vertical and horizontal distribution of foliage

Vertical and horizontal stratification of foliage was measured following Hubbell and

Foster (1987) as modified by Malcolm (1995). At every 2.5 m along 4 internal station transects spaced 25 m apart and 75 m long, percent foliage cover was measured using a 4 point scale (0-10,

10-50, 50-75 or 75-100%) at different height intervals using a 2.5-m-long pole for sighting purposes. Height intervals were 0-2.5, 2.5-5, 5-10, 10-15, 15-20, 20-25, and 25-30 m above ground and were identified using an optical rangefinder. Based on Principal Components

Analysis (PCA), scales in the intervals 0-2.5, 2.5-10 and 10-25 m were analyzed separately to represent shrub, understory and overstory strata, respectively (Sharkey 2008). Percent foliage cover was converted to foliage thickness scores as done in Malcolm (1995), and the mean, residual semi-variance and residual variance of foliage thickness were determined for each stratum. Because semi-variance and variance are naturally correlated with the mean, the mean

18 was partialed out (ie. the residual variance and semi-variance were used). From this point onwards, variance and semi-variance refer to residual variance and residual semi-variance, respectively. Variance provides an overall measure of heterogeneity (i.e., how variable was overstory foliage thickness in a stand?). Semi-variance is a way to examine the horizontal grain of the variability; a stand with a fine-grained overstory would have a close mix of gaps and no gaps, resulting in a high edge ratio (Sharkey 2008, Malcolm 1995). Mean, variance, and semi- variance of canopy height was also determined using the midpoint of the highest vertical stratum where foliage was recorded. Vertical foliage complexity was determined using the Shannon diversity index (H’) calculated from the across-site mean foliage thickness values of the seven height intervals.

At the 16 grid intersections, shrub openness was measured in four directions (N, S, E, W).

A 2 m long pole marked in 10 cm segments was held 5 m away from an observer and the number of segments completely unobstructed by foliage, twigs, snags or DWD was recorded (Kuttner

2002). The mean of the four measurements per intersection was calculated, and then a mean shrub openness value was calculated per site (Sharkey 2008).

Downed Woody Debris

Along the four 75-m-long transects measured for foliage distributions, the diameter (>7 cm) and decay class (using classes 1-5 by Hayden et al. [1995]) of downed woody debris was measured at the point where the DWD intersected the transect. Volume was calculated using Van

Wagner’s (1968) formula, and to reduce the number of decay classes, a PCA was conducted that identified two major decay classes: 1 and 2 (‘new’) and 3-5 (‘old’). DBH of snags >10 cm

19 diameter were measured at the 16 grid intersections within the 6-m radius plot and used to calculate basal area of snags.

Stand age, management history and productivity

Age since last disturbance and management history was obtained from forest resource inventory data (FRI) generated and ground-truthed by the Ontario Ministry of Natural Resources

(OMNR). Stands that were older than the commencement of logging in the area (>110 years) were considered un-logged (Sharkey 2008). All logged stands were clear-cut via horse-logging in the period 1935-1963, except two mixedwood stands mechanically logged in 1964 and 1971

(Sharkey 2008).

At the four corners of the grid and one station in the middle, vegetation type was determined using Ontario’s northeast Forest Ecosystem Classification (FEC). Eleven vegetation types were found in total and sites often had >1 vegetation types. Because vegetation types in

FEC have been suggested as a correlate of overall site productivity (Sharkey 2008: Arnup in lit.), vegetation types were ranked according to the number of herb species, and a weighted average of ranks for each site was calculated (Sharkey 2008).

Insect Sampling

At each site, I placed 1 pitfall trap at each of the 16 internal grid intersections spaced 25 m apart. Other studies have found this trap number sufficient for capturing most species in a site

(Werner and Raffa 2000). The traps were 25 m apart to minimize inter-trap interference and >50 m away from roads or water bodies (Digweed et al. 1995). The pitfall traps consisted of two

11-cm diameter containers (one within the other) with the rim set flush with the ground surface.

20

Only the inner container was removed when collecting samples to minimize substrate disturbance. Containers were filled partially with a 5% saline solution and a few drops of dish soap to break the surface tension. A plastic lid suspended with nails 3 cm above the trap covered the pitfall to minimize flooding and debris entry.

After one week (7 nights), the samples from a site were collected and consolidated.

Collection occurred over one week in June, July and August 2007 to cover a wide time range, with a second week per month collected in case of sample loss (Niemela et al. 1990). All

Carabidae were identified to species. Data were standardized per 1000 trap days to account for

18% of traps lost due to depredations, flooding, etc.

Statistical Analysis

Weibull live-tree distributions

I conducted a redundancy analysis (RDA) on the Carabidae correlation matrix (on log- transformed data) against live-tree distribution parameters (Weibull) for live stems >2.5cm dbh, and live trees >10cm dbh were conducted, and significance results were obtained using Monte

Carlo permutations (9999 permutations). To test whether Weibull shape or scale explained the greatest amount of variation (of the carabid matrix), I used forward selection (9999 Monte Carlo permutations).

Cohort class: multivariate analysis

RDA on the Carabidae correlation matrix (on log-transformed data) was also conducted against cohort class. Cohort class was used as a dummy constraining variable. Significance tests

21 were conducted with Monte Carlo permutations (9999 permutations). Associations with cohort classes were interpreted from ordination diagrams.

Cohort class: univariate analysis

Univariate tests examining species variation among cohort classes were also conducted.

For Carabidae species found in more than a third of sites, a one-way ANOVA was conducted on the log-transformed data (except for nitidicollis, for which transformation was not required). One common species, Synuchus impunctatus still violated assumptions of normality and homogeneity, hence a median test was used. All species found in less than a third of sites were tested using univariate median tests.

Unconstrained multivariate analysis of community structure

I examined the underlying gradients in Carabidae communities using Principal

Components Analysis (PCA) on the log-transformed correlation matrix. All structural variables were plotted passively and only those variables highly correlated to the ordination space (>0.40 on either axis) were shown in figures. Cohort class was tested as a dummy variable. RDAs of

PCA axis scores 1 and 2 as a function of the structural variables provided significance of variables via manual selection with 9999 Monte Carlo permutations.

Structural features: decomposition of variance

Partial Canonical Correspondence Analysis (PCCA) was used to test various sets of structural variables as predictors of the carabid community. Tests were conducted on the log- transformed correlation matrix. This test determines whether structural variables have a

22 significant effect alone and after other variables (eg. Weibull parameters) are considered, either jointly or in isolation. Downed woody debris (basal area of snags, volume of early-decay DWD and volume of late-decay DWD), canopy height (mean, variance and semi-variance) and understory foliage thickness (mean, variance and semi-variance of understory foliage thickness) were tested using PCCA. Other passively-plotted structural variables that appeared as key habitat features for the Carabidae community in the PCAs also were tested using PCCA. PCCA also was used to examine the predictive ability of age, productivity and percent deciduous composition in comparison with Weibull parameters.

Community Characteristics

Measures of total abundance, richness and diversity were calculated for the Carabidae communities. To obtain a measure of site-level richness corrected for abundance, I used a simple rarefaction curve technique (Sharkey 2008).The number of individuals captured was plotted against the number of species found in each site. A logarithmic curve was fitted to the data and the residuals from the curve were used as the corrected richness values. Diversity was measured using Shannon-Weiner index (H’).

Community metrics were explored using PCA on the correlation matrix, with all habitat variables plotted passively. Redundancy Analyses (RDA) using the axis 1 and 2 scores from the

PCA was used to test for significance of the structural variables via forward selection with 9999

Monte Carlo permutations. I also tested for differences in community metrics among cohort classes via Analysis of Variance (ANOVA). Examinations of residuals indicated that assumptions of normality and homogeneity were justified. Multivariate tests were undertaken with CANOCO for Windows version 4.5; univariate tests were undertaken with SAS v. 9.1.

23

RESULTS

A total of 2092 carabids from 25 species were captured (Table 1.2). Of these, 72% was from four species: Platynus decentis (Say) (27%), Pterostichus adstrictus Eschscholtz (23%), Pt. coracinus (Newman) (15%) and Pt. pensylvanicus LeConte (8%). Two non-native species were found: Pterostichus melanarius (Illiger) was found in five sites and Carabus nemoralis Müller in the two sites closest to the town of Kapuskasing (less than 10 km away). All sites containing non-native species had been previously logged, and all but one of these five sites contained annelid worms, another non-native resident of the upper soil and duff layer, which were noted in six sites.

24

Table 1.2- Abundances of carabid beetle species in 2007 in boreal mixedwoods of northeast Ontario, standardized to 1000 trap-days, in structural cohort classes. Organization of species is based on ecological synopses in Lindroth (1961-1969) and Larochelle & Lariviere (2003).

1 Overall mean abundance in cohorts Overall Univariate tests2 mean 1 2 3 4 (C/F, P) Species (n = 4) (n = 3) (n = 5) (n = 6) abundance Forest specialists Calosoma (Calosoma) frigidum Kirby 0 0 4.24 1.49 1.67 C=3.17, P=0.366 Sphaeroderus nitidicollis brevoorti LeConte 12.44 14.12 9.68 16.38 13.26 F=0.25, P=0.861 Sphaeroderus stenostomus lecontei Dejean 4.76 15.73 14.03 9.14 10.62 C=1.25, P=0.740 Agonum retractum LeConte 19.16 24.66 66.77 45.52 42.09 F=1.61, P=0.231 Pterostichus(Bothriopterus) pensylvanicus LeConte 82.05 (14.16) 41.55 18.00 31.95 40.81 F=0.34, P=0.798 Mean 23.68 (9.89) 19.21 22.54 20.89 21.69

Forest generalists Carabus (Archicarabus) nemoralis Müller 0 17.86 0 0 2.98 C=5.00, P=0.172 Scaphinotus (Nomaretus)bilobus (Say) 11.16 18.88 17.82 18.68 16.80 C=1.45, P=0.694 Harpalus (Euharpalops) fulvilabris Mannerheim 2.23 0 0 5.49 2.33 C=4.25, P=0.235 Calathus ingratus Dejean 49.75 13.26 9.74 14.95 20.95 F=0.88, P=0.477 Platynus (Platynus) decentis (Say) 113.37 285.71 168.05 139.36 165.95 F=1.28, P=0.318 Synuchus(Pristodactyla) impunctatus (Say) 28.09 75.65 77.06 20.04 46.94 C=5.17, P=0.160 Pterostichus(Bothriopterus) adstrictus Eschscholtz 133.65 115.46 98.78 48.90 92.68 F=1.39, P=0.287 Pterostichus (Euferonia)coracinus (Newman) 35.93 340.08 51.90 39.32 92.19 F=0.75, P=0.539 Pterostichus(Lenapterus) punctatissimus (Randall) 27.14 9.35 9.02 14.32 14.87 C=1.91, P=0.591 Pterostichus (Morphnosoma) melanarius (Illiger) 0.00 39.82 2.04 0.00 7.2 C=7.03, P=0.071 Trechus apicalis Motschulsky 0.00 0.00 2.60 0.00 0.72 C=2.60, P=0.457 Mean 36.28 83.28 39.73 30.25 42.15

25

Table 1.2- continued

Open-habitat species Harpalus somnulentus Dejean 0.00 0.00 4.08 0.00 1.13 C=2.60, P=0.457

Hygrophilous species Carabus granulatus granulatus Linné 0.00 2.98 0.00 0.00 0.5 C=5.00, P=0.172 C=3.50, P=0.321 Loricera pilicornis pilicornis (Fabricius) 3.25 0.00 0.00 0.00 0.72 Platynus (Batenus)mannerheimii (Dejean) 0.00 13.18 0.00 0.00 2.2 C=10.62, P=0.014 Mean 1.08 5.38 0.65 0 1.14

Harpalus spp. 2.75 0.00 0.00 0.00 0.61 C=3.50, P=0.321 Total mean abundance 24.93 48.97 26.37 19.38 27.49 1 Numbers in parenthesis are when one site with extremely high abundance was excluded. Site was not excluded for overall tests since only one species showed this trend. 2 Statistics and P values from univariate tests of cohort effects. F values (degrees of freedom =3) are from ANOVAs; Chi-square values (C) are from a median test.

26

Live-tree diameter distributions: multivariate analysis

Live-tree diameter distribution parameters (Weibull) were able to predict Carabidae communities: the live-tree diameter distribution of all stems ≥2.5 cm DBH was highly significant

(18.6% variance explained, P=0.007; Fig. 1.4), and the live tree-diameter distribution of all stems ≥10 cm DBH was almost significant (16.7% variance explained, P=0.059, Fig. 1.5).

Of the two Weibull parameters, Weibull scale of all stems was the most important parameter (Table 1.3); scale alone was significant for live-tree diameter distribution of all stems

≥2.5 cm DBH and ≥10 cm DBH (10.5%, P=0.023; 9.5%, P=0.049, respectively). As well, scale was significant when shape was already entered in the model for ≥2.5 cm DBH (but not ≥10 cm

DBH), whereas the converse was not true. Shape alone was not significant for either Weibull parameter sets. Weibull scale was associated with the first RDA axes in both RDAs (Fig. 1.4 &

1.5).

Cohort class: multivariate analysis

Cohort class, expressed as dummy variables, was a weaker predictor of communities than live-tree diameter distribution parameters as it was not quite significant in an RDA (22.2% variance explained, P=0.067; Fig. 1.6). In the ordination, Pterostichus punctissimus was found in cohort 1 space, Platynus decentis, Carabus nemoralis, Pterostichus melanarius, Synuchus impunctatus, Pterostichus adstrictus and Pterostichus coracinus were associated with cohort 2 space, Calosoma frigidum was associated with cohort 3, and Harpalus fulvilabris was associated with cohort 4 space. All of the above relationships are similar to those found in the unconstrained ordinations described in the next section (Fig. 1.7). Other less robust relationships were seen with Agonum retractum, which associated cohort 3-4 in the unconstrained ordination (see next

27 section), but with cohort 1 in the RDA. Sphaeroderus nittidicollis and Pterostichus pensylvanicus, which both associated with cohort 2 in the unconstrained ordination, were associated with cohort 3-4 in the RDA. Sphaeroderus lecontei and Sphaeroderus bilobus associated with cohort 3 in the unconstrained and cohort 2 in the RDA. Calathus ingratus associated with cohort 3-4 in the unconstrained and cohort 1 in the RDA.

The following rare species were also found: Loricella pilicornis (in cohort 1); Carabus granulatus (in cohort 2); Platynus mannerheimi (in cohort 1 and 2); and Trechus apicalis, and

Harpalus somnulentus (in cohort 3).

Cohort class: univariate tests

Univariate tests on cohort class were significant only for Platynus mannerheiimi (Table

1.2). This species was rare, found only in two sites, both of which were cohort 2. It is likely that this result is due to chance.

28

Table 1.3 - Redundancy analysis significance (forward selection) examining correlations between carabid communities of boreal mixedwoods of northeastern Ontario and Weibull shape and scale of all stems >2.5cm dbh (Weibull 2.5) and Weibull shape and scale of all stems >10 cm dbh (Weibull 10). Shown are percentage of variance explained and P values (significant values in bold).

Shape Scale (with scale (with shape Shape selected first) Scale selected first) Weibull 10 8.8%, P=0.224 7.1%, P=0.224 9.5%, P=0.049 7.9%, P=0.127 Weibull 2.5 8.0%, P=0.127 8.1%, P=0.091 10.5%, P=0.023 10.5%, P=0.020

29

Figure 1.4- Redundancy Analysis (RDA) for Carabidae in boreal mixedwood sites in northeastern Ontario in 2007, constrained by Weibull parameters of all stems >2.5cm dbh. ● cohort 1; □ cohort 2; ○ cohort 3; ■ cohort 4. Significance of each parameter is shown (**= P<0.01, *= P<0.05). Species codes are in Appendix 2.

30

Figure 1.5- As Figure 1.4 except that Weibull parameters are for stems >10 cm dbh.

31

Figure 1.6- As Figure 1.4 except that data are constrained by cohort class.

32

Unconstrained multivariate analysis of community structure

In the PCA on Carabidae abundance with structural variables plotted passively (Fig. 1.8), no clear separation exists among cohort classes (Fig. 1.7). However, Cohort 2 centroid is highly significant (24.2%, P=0.005), even in a forward selection (11.4%, P=0.039), and cohort 2 sites show some separation to the top right of the ordination diagram. Cohort 4 sites are also somewhat separated. Interestingly, both Weibull shape and scale of all stems >2.5 cm dbh were significant (15.7%, P=0.0497, 22.4%, P=0.011 respectively), although only Weibull scale remained significant in a forward selection following the selection.

Fine-grained understories (semi-variance of understory foliage thickness) seemed to be a slightly stronger predictor than Weibull parameters or cohort class (26.9%, P=0.007). Other habitat variables that were significant, although not in a forward selection, were variance of canopy height (18.4%, P=0.023), and shrub openness (16.8%, P=0.037).

Pterostichus coracinus, Carabus nemoralis, Carabus granulatus and Pterostichus melanarius were associated positively with Weibull scale 2.5 and semi-variance of understory foliage thickness. Shrub openness and Weibull shape 2.5 corresponded with high abundances of

Synuchus impunctatus and Playtnus decentis. The cohort 2 centroid was associated with all of the aforementioned species. Agonum retractum, in contrast, was directly opposite these environmental variables and species. Sphaeroderus lecontei, Scaphinotus bilobus, Sphaeroderus nittidicollis, and Pterostichus pensylvanicus associated with residual semi-variance of canopy height, along with vertical heterogeneity percent deciduous composition, and Weibull scale and shape of all stems. Calathus ingratus, Pterostichus punctissimus, Harpalus somnulentus, Trechus apicalis and Pterostichus adstrictus were negatively associated with these variables.

33

Figure 1.7 - Principal Components Analysis (PCA) for Carabidae (correlation matrix) from pitfall traps in 2007 in boreal mixedwoods of northeastern Ontario with environmental variables plotted passively (only those with axis scores >0.40 shown). ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Significance of each variable is shown (**= P<0.01, *= P<0.05).

34

Decomposition of variance

In Weibull comparisons, no variable set other than Weibull explained unique variance, although canopy height was nearly uniquely significant. Weibull did not explain unique variance, however, when understory foliage thickness, shrub openness, or vertical complexity was selected.

Live-tree diameter distributions (Weibull) shared the most variance with understory foliage thickness set (6%); while both were significant on their own, neither could significantly explain variation in the data when the other was selected as a covariate (Table 1.4). The weakness of shrub openness is further supported in this variance partitioning where it was not a significant predictor (7.4%, P=0.230; Table 1.4). It shared 3.7% of its variance with Weibull, and

Weibull could not explain unique variation once shrub openness was selected. Conversely, canopy height and understory foliage thickness both explained unique variation when shrub openness was a covariate. Vertical foliage complexity was a significant predictor of Carabidae communities (11.0, P=0.013), but did not explain any unique variance when Weibull was considered (7.6%, P=0.106) as it shared 3.2% of its variation with Weibull.

In contrast, live-tree diameter distribution (Weibull) was uniquely significant when canopy height or DWD sets were selected as it shared little variance with these variables.

Weibull shared the least amount of variance with canopy height set and DWD set, and was able to explain unique variance when those variables were selected (while the contrary was not true).

The downed woody debris set (basal area of snags, decay class of early DWD, and decay class of late DWD) was not a significant predictor (Table 1.4). When the DWD and Weibull set were tested together, the DWD variables did not significantly explain the community variation. Using decomposition of variance, I found that only 1.2% of the variance was shared between the

Weibull and DWD set.

35

Table 1.4- Partial redundancy analysis examining the relative value of various habitat predictors in explaining carabid community composition from pitfall traps in boreal mixedwoods of northeastern Ontario in 2007 in comparison to Weibull parameters from the diameter distribution of stems >2.5 cm dbh.

Understory Vertical Canopy Shrub foliage DWD foliage height openness thickness complexity % P % P % P P % P % var. var. value var. value var. value value var. value Together 40.0 0.003 Together 37.5 0.042 Together 22.4 0.074 Together 33.7 0.157 Together 26.2 0.004 Weibull 18.6 0.007 Weibull 18.6 0.007 Weibull 18.6 0.007 Weibull 18.6 0.007 Weibull 18.6 0.007 Canopy Understory Shrub 22.4 0.087 24.9 0.042 7.4 0.230 DWD 16.3 0.622 Vert_H 11.0 0.013 height thickness openness Unique Unique Unique Unique Unique to 17.6 0.010 12.6 0.246 to 14.9 0.110 to 17.4 0.048 to 15.2 0.070 to Weibull Weibull Weibull Weibull Weibull Unique Unique Unique Unique Unique to to to 21.4 0.051 18.9 0.211 3.8 0.807 to 15.1 0.622 to 7.6 0.106 canopy understory shrub DWD Vert_H height thickness openness Shared 1.0 Shared 6.0 Shared 3.7 Shared 1.2 Shared 3.2

36

When constrained, productivity and percentage of deciduous composition, not age, were significant predictors of Carabidae communities (Table 1.5). Productivity and percentage of deciduous composition, however, were not significant once Weibull was considered. In variance partitioning, Weibull parameters continued to significantly explain unique variation when productivity was a covariate, but not for percentage of deciduous composition. This is likely due to Weibull sharing more variance with percentage of deciduous composition (4.1%) than productivity (1.6%). Only 0.6% of the variation was shared between age and Weibull, and when age was a covariate, Weibull continued to remain significant (37.5%, P=0.003).

37

Table 1.5- As Table 1.4 except for the stand level variables age, productivity (based on FEC herb richness) and percentage of deciduous composition.

Age Productivity Percentage of deciduous composition % var. P value % var. P value % var. P value Together 22.1 0.087 Together 24.1 0.041 Together 24.6 0.017 Weibull 18.6 0.007 Weibull 18.6 0.007 Weibull 18.6 0.007 Age 4.1 0.770 FEC 7.1 0.289 % deciduous comp. 10.1 0.025 Unique to Weibull 18.0 0.022 Unique to Weibull 17.0 0.035 Unique to Weibull 14.5 0.103 Unique to Age 3.5 0.846 Unique to FEC 5.5 0.406 Unique to % deciduous comp. 6.1 0.328 Shared 0.6 Shared 1.6 Shared 4.1

38

Community characteristics: univariate analysis

The logarithmic curve provided a relatively good fit to the richness vs. abundance plots, with richness increasing with abundance (Fig. 1.8). From this curve, I obtained a measure of richness standardized for abundance (hereafter termed simply "richness"). Abundance, richness and diversity did not differ significantly among cohort classes in one-way ANOVAs (F=3.09

P=0.063; F=0.51 P=0.679; F=0.89, P=0.471, respectively, df=3).

Community characteristics: multivariate analysis

The first axis of the community characteristics PCA represented a richness gradient and no visible cohort clustering was present (Fig. 1.9). Residual semi-variance of understory foliage thickness was significantly correlated to the community characteristics space (even in a forward selection with canopy height variance) and was positively correlated with species richness.

Diversity (H') increased with the variance of canopy height and overstory foliage thickness, both of which were significant although only variance of canopy height was significant in a forward selection (Fig. 1.9).

Abundance was positively associated with cohort 2 and Weibull shape of all stems, both of which were significant but not in a forward selection.

39

14

12

10

8 y = 2.469ln(x) - 0.2002

6 Species Species richness

4

2

0 0 20 40 60 80 100 120 140 160 Number of individuals

Figure 1.8. Carabid species richness plotted against abundance from pitfall traps in boreal mixedwoods of northeastern Ontario in 2007, with fitted regressions shown. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4.

40

Figure 1.9 - Principal Components Analysis (PCA) of carabid community metrics in of boreal mixedwood sites in northeastern Ontario, for all months 2007, with environmental variables (only those with axis scores >0.40 shown) plotted passively. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Significance of each parameter is shown (**= P<0.01, *= P<0.05. += P<0.05 after rsvund or rvarcanht was selected). Variable acronyms are identified in Appendix 1 and 2.

41

DISCUSSION

In agreement with Sharkey's (2008) findings in mixedwoods for small mammals, I found that characterization of stem diameter distributions was more effective at explaining carabid communities when relatively small stems (>2.5 cm DBH) were included as opposed to when trees of large diameter were included (≥10-cm DBH). Presumably, these smaller stems better reflect microclimate conditions relevant to small fauna (Sharkey 2008); they may also help to describe the overstory structure better in that they better characterize forest gaps. It would be important for forest managers to examine stands at a resolution higher than the conventional 10 cm dbh limit to properly manage habitat for Carabidae and small mammal communities.

Cohort class did not appear to be a particularly good coarse filter for Carabidae as only one species and no community characteristics differed significantly among cohort classes; furthermore, cohort class was not significant as a constraining variable in a redundancy analysis

(RDA). The cohort classes did not show distinct clustering in the PCA on the carabid matrix, however, there was some evidence that cohort classification could predict distinct communities;

In the PCA, cohort 4 sites appeared to be somewhat isolated from the other sites, and cohort 2 was a significant predictor, although performing worse than other habitat variables. Cohort 2 was also significantly associated with positive abundance, however, was not strong in a forward selection. My results agree with Sharkey (2008), who found that cohort class was nearly significant, however in her case it pertained to mammal communities. When examining forest communities, it may be more informative to look at stand development along a continuum, as is the case for Weibull parameters, rather than a (somewhat arbitrary) division of stands into classes. Weibull parameters were significant in examining constrained and unconstrained community data, and in several cases could explain unique variation once particular habitat

42 variables were selected. My results found Weibull scale of all stems to be the better predictor of

Carabidae assemblages which is contrary to Sharkey (2008), who found mammals communities differed more according to Weibull shape. Shape captures the number of stems in each size class, describing whether the forest is more even or uneven-aged. What seems to matter more for

Carabidae, however, is generally whether or not the stand is dominated by large trees as described by scale. Weibull shape showed some strength in the PCAs of community composition and characteristics, but not so in a forward selection. It seems that the cohort concept based on live-tree diameter distributions in general did not perform well as only one of the two parameters was needed to best characterize carabid communities.

In my study, fine-grained understories (i.e., high semi-variance), and the number of large stems (scale) were the most important correlates of carabid community structure; shrub openness and variance of the canopy height also performed relatively well. Interestingly, Weibull shared the highest amount of variation with understory foliage thickness parameters, suggesting that

MCM could capture this important habitat feature. Microclimate and ground-level structures such as percentage cover of herbaceous vegetation and amount of exposed ground, have been found to be the most important forest structures for Carabidae (Work et al. 2004, Lassau et al.

2005, Cobb et al. 2007). I did not directly measure those variables here, but presumably these variables (understory grain, shrub openness and variance of canopy height/overstory thickness) were correlated with fine-scale variation in conditions close to the ground. In a small area considered structurally homogeneous, it is possible to have high levels of compositional heterogeneity due to high microsite variation (Niemela et al. 1996). Shrub openness and understory foliage thickness variables may reflect variation due to gap phase dynamics (Koivula

& Niemela 2003, Klimaszewski et al. 2005). However, unlike my study, Koivula et al. (2002)

43 found no correlation between shrub cover and carabid assemblages. Larger-scale structural variables can also dictate microclimate conditions (Humphrey et al. 1999, Werner and Raffa

2000, Work et al. 2004). Canopy closure, for example, plays a key role in determining which species will thrive as it affects important microclimatic factors such as soil moisture and temperature (Niemela 1993, Spence et al.1996, Lassau et al. 2005, Buddle et al. 2006). In my study, Weibull scale did a better job at predicting carabid assemblages than the canopy height variables; however, I found some evidence that canopy height variables were, in addition, useful in understanding carabid community variation.

Downed woody debris (DWD) combined with snags was, perhaps surprisingly, was not a strong predictor nor did they predict carabid diversity or richness. Although DWD is often shown as important for ground-dwelling fauna (Martikainen et al. 2000, Hammond et al. 2001,

Cobb et al. 2007, Work et al. 2004), some Canadian studies have found little association between DWD and wildlife in mature mixedwoods (Sharkey 2008, Pearce et al. 2003). In one case, DWD was found to be more important for carabids in clear-cuts compared to closed forests, perhaps because DWD was not a limiting factor in mature forest sites (Pearce et. al. 2003).

Vanderwel (2009) similarly found evidence that late-decay DWD was important for small mammals, but only when in short supply. This may be the case for my stands, where we found no significant associations with DWD. It was not surprising that percentage of deciduous composition and productivity were not significant because we explicitly controlled for the former and implicitly controlled for the later (by studying only mixedwoods) (Sharkey 2008).

It has been suggested by many authors that multi-cohort stand development has a strong successional component. Using live-tree diameter distributions, however, some mature stands to

44 exhibit the structures associated with cohorts 1 and 2 (Sharkey 2008). This was contrary to the idea that age captures most of the structural variation in a stand, as a mature stand in theory might be expected to be in cohort 3 (Bergeron et al. 2007). In both mu study and Sharkey (2008), stand age did not appear to be a strong factor in explaining Carabidae or small mammals community variation, respectively, compared to stand development features themselves. For carabids, age was unable to explain unique variation. Most carabid habitat studies focus on time since disturbance as a variable of overwhelming importance; high diversity, abundance and richness is associated with young stands compared to mature stands, and species compositions change from open-habitat species to forest specialists over time (Beaudry et al. 1997, Koivula et al. 2002, Klimaszewski et al. 2005, Buddle et al. 2006, Niemela et al. 2007, Paquin 2008).

However, community changes along an age gradient are less drastic following canopy closure

(Paquin 2008), which may account for the low importance of age in my study. In Koivula et al.

(2002), Overall abundance did not change across a post-harvest successional gradient, although richness was higher in stands 5-10 years post-harvest. All stands in my study had closed canopies and ranged in age from 35-133 years. Buddle et al. (2006) found that aspen stands 28-29 years post disturbance, during a phase of succession described as canopy closure, had Carabidae communities that were similar to old-growth stands. Presumably, time since disturbance is a less useful variable for classifying cohorts and their wildlife communities because of inherent variability in site characteristics and variation in stand disturbances at the time of, and after, the stand replacing disturbance.

Percentage deciduous composition was able to explain significant variation in the assemblages (although not uniquely so once Weibull was selected), which has been found to be important in other studies (Niemela et al. 1992, Guillemain et al. 1997, Pearce et al. 2003).

45

Agonum retractum, Calathus ingratus, Pterostichus pensylvanicus and Platynus decentis are known to be associated with deciduous as opposed to coniferous leaf litter (Pearce et al. 2003), although it is not known why.

In my study, most species, which were those predominately associated with closed forests, were positively associated with the significant environmental variables. The few open habitat species that we captured, such as Harpalus somnulentus, are likely attributable to relic populations or dispersers seeking habitat (Koivula et al. 2002). In terms of species-level relationships, many common species were found in all sites, but nevertheless showed higher abundances as a function of particular structural features. Sharkey (2008) found associations between small mammals considered ‘old growth species’ and late cohort classes (high scale).

It is hard to determine conclusive results on the associations of species with cohort classes, however, as only cohort 2 was a significant predictor. For example, while we would expect the forest specialist Agonum retractum to associate with well-established, multi-age structure with evident gap dynamics, this species was associated with cohort 1 stands in the constrained ordination, but with cohort 3-4 in an unconstrained ordination (Larochelle and Lariviere 2003,

Klimaszewski et al. 2005). Another forest specialist, Calosoma frigidum, however, was associated with cohort 3 in constrained and non constrained ordinations; this species has been noted to be most abundant in undisturbed mature forest with thick leaf litter (Beaudry et al. 1997,

Larochelle and Lariviere 2003). Two forest specialists, Sphaeroderus nitidicollis and

Pterostichus pensylvanicus, associated with cohort 3-4 stands in the constrained analysis. Both, however, had only small vectors in the RDA of the Weibull parameters, indicating a general lack of association with cohort class. This result may have been skewed by the extreme abundance of

Pterostichus pensylvanicus in a cohort 1 stand. When that one site was removed, this species was

46 associated with cohort 3. Pterostichus pensylvanicus is considered a ubiquitous forest specialist associated with DWD used for over-wintering sites, but it also can use forest gaps, clearings and edges (Larochelle and Lariviere 2003, Work et al. 2004, Klimaszewski et al. 2005).

Synuchus impunctatus, Carabus nemoralis, and Pterostichus melanarius are forest generalists that were associated with high Weibull scale (cohorts 2 and 3) and have been associated with a range of forest habitat structures (Werner and Raffa 2000, Larochelle and Lariviere 2003,

Klimaszewski et al. 2005). Many of the forest generalists, such as Pterostichus punctissimus,

Pterostichus adstrictus and Calathus ingratus, were associated with low scale/shape (small stature, even-aged, homogeneous stands) (Niemelä et al. 1993, Klimaszewski et al. 2005). On the other hand, the generalists Platynus decentis, Pterostichus coracinus, and Scaphinotus bilobus were positively associated with stands beginning to show uneven-aged structure (cohort

2-3; high scale). Thus, generalists and specialists cannot be easily interpreted as favouring a particular cohort class.

One must be careful when concluding that a certain species has affinity for a certain habitat: structural features can vary in mutual dependence of each other, and such relationships can vary as a function of supply of one or another feature, hence determining whether or not a species is governed by a particular feature can be difficult (Thiele 1977). An additional caveat is the focus here on one year of sampling. Certainly, although sampling from one year can be informative (Niemela 1993), collections from multiple years are even more informative.

While other structural features (heterogeneity of canopy height and understory foliage thickness) were important for diversity and richness, cohort 2 and Weibull scale associated with abundance of Carabidae. However, I found no significant relationship between community

47 metrics such as richness, diversity and abundance and cohort class in the univariate analysis.

Previous studies have found that Carabidae species composition is more sensitive to changes in habitat and management than overall abundance, richness or diversity (Werner and Raffa 2000), in general agreement with my findings. Important changes in community composition can happen without a change in these community characteristics, suggesting that they are of limited utility for monitoring changes in biodiversity (Niemela 1997, Werner and Raffa 2000, Koivula et al. 2002, Work et al. 2004). Several studies have shown higher species richness and diversity of Carabidae in post-harvest (Beaudry et al. 1997, Koivula et al. 2002, Buddle et al. 2006) and old growth sites (Niemela 1997, Paquin 2008). However, few to no differences in density or diversity have been found between un-harvested stands and those that have undergone canopy closure (Werner and Raffa 2000, Koivula et al. 2002, Vance and Nol 2003, Buddle et al. 2006), or between cover types even when differences in faunal composition were present (Work et al.

2004).

Richness was strongly associated with fine-grained understories (semi-variance of understory foliage thickness) and diversity was associated with heterogeneous canopy height

(variance). The positive relationship with semi-variance and variance relates to the importance of heterogeneity in maintaining richness, abundance, and diversity. Heterogeneity promotes a diverse range of niches, which in turn promotes diverse, rich assemblages (Klimaszewski et al.

2005). While the majority of carabids can be found across a successional gradient, peak canopy closure may be the prime time for forest species; this may explain why cohort 2 and Weibull shape was associated with abundance, although not as strongly (Niemela 1993, Niemela et al.

1996). This emphasizes the need to maintain the range of cohort classes in the landscape in order

48 to effectively manage for biodiversity, although the most vulnerable species to the prevalent management regime of course are of greatest concern.

Two non-native species, Pterostichus melanarius and Carabus nemoralis, were associated with cohort 2 stands. Both species are known to associate with anthropogenic habitats

(Lindroth 1969, Werner and Raffa 2000, Larochelle and Lariviere 2003). The latter is a flightless species that can also inhabit mixedwood forests (Larochelle and Lariviere 2003) but has limited ability to disperse, perhaps explaining its occurrence in the two stands closest to Kapuskasing.

Pterostichus melanarius was found not just in the three cohort 2 sites, but also in cohort 3 and 4 sites further from town. This presumably reflects its proficient migratory abilities and it flexible habitat use of forest and anthropogenic habitats (Lindroth 1969, Niemela and Spence 1991,

Larochelle and Lariviere 2003). Although noted only anecdotally, invasive annelid earthworms were found in most sites occupied by the non-native carabids. Recently establishing themselves in northern areas, earthworms could potentially have an effect on the leaf litter and thus the species that depend on thick litter (for habitat or to prevent desiccation), such as Agonum retractum; they may also serve as food for non-native carabids, which readily feed upon them

(Larochelle and Lariviere 2003, Migg-Kleian et al. 2006). There are some native predators such as Carabus granulatus that have been found to use earthworms as prey; however, other predators that cannot eat earthworms and rely on litter-dependent prey may be negatively affected

(Lukasiewicz 1996, Migg-Kleian et al. 2006). Pterostichus melanarius has been found to have little impact on native carabid faunas in forested habitats, but a significant impact in urban and agriculture areas (Niemela & Spence 1991,1994, Carcamo & Spence 1994). These observations

49 suggest that even in these relatively remote boreal locations, changes in Carabidae communities are occurring as introduced earthworm and carabid species establish themselves.

In conclusion, weak support was found for unique Carabidae communities among cohort classes, aside from some separation of cohort 2 stands. The cohort concept used in this study may be too vague; it appears to be more effective to examine Carabidae communities along a continuum, using Weibull parameters, particularly scale, to classify the communities, and to incorporate other structural features along with live-tree diameter distributions to increase its predictive strength. Many structural features were important to Carabidae, with fine-grained understories and heterogeneous canopies positively associating with richness and diversity, and community composition. DWD was not important for Carabidae as it is likely not yet a limiting factor in my managed stands. Age since disturbance was not important to Carabidae, emphasizing the importance of examining habitat features rather than age in biodiversity management and the application of MCM.

50

CHAPTER 2- VERTICAL STRATIFICATION AND EFFECTS OF MULTI-COHORT MANAGEMENT IN THE ABUNDANCE OF DIPTERAN AND HYMENOPTERAN FAMILIES IN BOREAL MIXEDWOOD FORESTS

INTRODUCTION

In Ontario, forests are managed to emulate natural disturbance in an effort to maintain natural biodiversity. Currently, the boreal forest in northeastern Ontario is managed using clear-cut silviculture on a c. 100-year rotation, to approximate the stand-replacing disturbance of cyclical fire and insect outbreaks. However, current research suggests that the fire-return interval is much longer in the area, with some stands persisting 300 years without stand replacing disturbances (Bergeron et al. 2001). As a result, these mature and old growth stands, which experience small-scale disturbances, are not being managed in a manner consistent with actual natural processes under the current harvesting regime. Multi-Cohort Management (MCM) has been proposed as a way to manage the northeastern boreal forest such that mature stands will persist without substantial decreases in the annual allowable cut (Bergeron et al. 1999). MCM proposes the use of selection and partial cutting to approximate the proportions of multi-aged and even-aged stands naturally found in the landscape (Bergeron et al. 1999, Harvey et. al. 2001,

Bergeron et. al. 2001).

A major goal of natural disturbance emulation is to manage for biodiversity (Attiwill

1994, Bergeron and Harvey 1997, Hunter 1999), hence the responses of wildlife communities to

MCM, and indeed the importance of stand cohort management as a coarse filter strategy, is of critical importance. Insect communities are of particular interest because they are ubiquitous and occupy a wide range of habitats due to their feeding strategies and microhabitat preferences, and are an important component of food webs (Thiele 1977, Triplehorn & Johnson 2005). They are also are sensitive to short-rotation, intensive forestry, and there is a risk of species declining in

51 abundance or becoming extripated in managed landscapes (Haila et al. 1994, Niemela 1997).

Studies examining insect relationships with forest management have primarily focused on the ground-dwelling fauna (e.g., Spence et al. 1996, Heliola et al. 2001, Vance and Nol 2003, Work et al. 2004, Moore et al. 2004, Buddle et al. 2006). However, insects such as Hymenoptera and

Diptera can also be responsive to structural changes in forests due to management, warranting further study (Schowalter 1995, Su & Woods 2001, Deans et al. 2005, Reemer 2005, Akutsu et al. 2007). The response of forest insects, conversely, may not be consistent across the vertical profile; it is important to recognize that insects may specialize on one forest height stratum or another, with community responses to structure and forest management differing between understory and canopy (Murdoch et al. 1972, Basset et al. 2001b, Vance 2002, Akutsu et al.

2007, Grimbacher & Stork 2007). For instance, Akutsu et al. (2007) found a prominent decrease in abundance of flying insects in the understory in comparison to a more obscure response in the canopy following harvesting. These differences in response between strata are primarily due to manipulation of the structural characteristics and microclimates of the understory and canopy on which insect fauna depend. The understory, for example, contains most of the dead and decaying organic matter of a stand, which is important for the larval stage of many Diptera, while the canopy contains greater amounts of new living plant matter, a key food source for many herbivores (Basset 1991, Vance 2002, Atkusu et al. 2007, Grimbacher & Stork 2007). Some families in the understory, such as Syrphidae, can benefit from forest management practices that open the canopy (Humphrey et al. 1991, Reemer 2005, Deans et. al. 2007), while others such as

Diapriidae, which associate with un-harvested stand structures (such as high tree densities, older ages, greater canopy heights and diameters) may not (Deans et al. 2005). In comparison to the canopy, the understory (with the exception of gaps) is well protected by foliage cover from the

52 external physical environment, including high solar radiation, high wind speeds, and temperature extremes which are less favourable for desiccation-sensitive insects such as small dipterans

(Atkusu et al. 2007, Grimbacher & Stork 2007). Because certain forest features, such as canopy, height differ with the parameters that define cohort classes (Sharkey 2008), it is possible that

MCM could manage for particular insect taxa if their preferred structural features are predicted by cohort classification. It is therefore necessary to examine both understory and canopy communities in the context of forest management to avoid underestimating community interactions with structure, diversity and abundance (Lowman & Wittman 1996, Su & Woods

2001).

Little literature exists on the aerial insects of northern forest canopies (Vance 2002:

Winchester & Ring 1996). There are many reasons to expect a relatively impoverished fauna in the canopy of boreal forests compared to tropical forests. Boreal forests are more structurally simplified than tropical forests, lacking high canopies, phytotelmata, epiphytes and arboreal soil

(Ulyshen and Hanula 2007). Furthermore, harsh winters and deciduous leaf loss results is less food and fewer niches in the canopy over the year, negatively affecting the number of canopy- specialists (Schaefer 1991, Ulyshen and Hanula 2007). Tropical forest canopies, in general, can have much higher richness (Sutton et al. 1983, Basset et al. 1992, Basset 1992, Devries et al.

1997, Basset et al. 2001b) and abundance (Sutton and Hudson 1980, Kato et al. 1995, Basset et al. 2001b, Ribeiro & Basset 2007) than the understory due to more diverse and plentiful resources (Basset 1992, DeVries et al. 1997, Basset et al. 2001ab; see Grimbacher & Stork 2007 for contradicting results). In contrast, temperate forests can have little to no vertical stratification of insect richness (Fowler 1985, Ulyshen & Hanula 2007) and abundance (Le Corff and Marquis

1999, Vance 2002, Ulyshen & Hanula 2007). Trends in abundance between understory and

53 canopy in temperate forests can also be reversed from those found in tropical forests, with understories having greater insect diversity (Lowman et al. 1993, Ulyshen & Hanula 2007, Pucci

2008), richness (LeCorf and Marquis 1999) and abundances (Nielsen 1987, Preisser et al. 1998,

Lowman et al. 1993, LeCorff and Marquis 1999, Vance 2003; see Hollier & Belshaw 1993 and

Sobeka et al. 2009 for contradicting results). In most cases, studies in temperate forests found distinct assemblages in the canopy and understory, despite a difference or no difference in insect community characteristics (Hollier & Belshaw 1993, Winchester and Ring 1996, Vance 2007,

Ulyshen & Hanula 2007, Pucci 2008, see Fowler 1985 for contradicting results). These general trends hold true for Hymenoptera and Diptera in temperate forests with more individuals near the forest floor (Preisser et al. 1998) or similar numbers between strata (Vance 2002, Vance 2007,

Pucci 2008). To my knowledge, there has been no direct comparison of understory and canopy in northern boreal forests, however, it is likely that trends in the boreal will approximate those found in temperate forests more so than tropical forests.

In this chapter, my objectives were three-fold. First, as in chapter 1, I test the general utility of the multi-cohort concept as a coarse filter for boreal biodiversity conservation. In this case, I focus on family-level community structure, abundance, composition, richness and diversity of hymenopteran and dipteran families captured in aerial (Malaise) traps, and morphospecies of the Diapriidae. I expect hymenopteran and dipteran families may differ with cohort class and the live-tree diameter distributions used in the classification due to the affinity of particular families for particular structural features predicted by live-tree diameter distributions. Multi-cohort stands may favour species that use canopy openings such as

Syrphidae, Dolicopodidae, and Sarcophagidae, or species that associate with old-growth structures such as such the Mycetophilidae (Humphrey et al. 1999, Deans et al. 2007).

54

Diapriidae, which can be sensitive to tree height, may prefer multi-cohort stands as they have more large trees (Deans et al. 2005, Dennis unpubl.). Therefore, I predict that Weibull parameters will be a better indicator of insect communities than other classification parameters such as age, percentage of deciduous composition, and productivity.

A second objective is to examine whether dipteran and hymenopteran family communities show vertical stratification in boreal mixedwoods forest. I predict that

Hymenoptera and Diptera will show higher abundance, richness and diversity in the understory and fewer families in the canopy. For individual families, I expect to see similar trends in vertical stratification seen by Vance (2002) in Great Lakes - St. Lawrence forests: families that require ground-level resources (such as decaying material and soil) and shaded, damp microclimates will be at higher abundances in the understory stratum. An example might be

Diapriidae and Sciaridae, which associate with soil-level debris and fungi, being more abundant in the understory. Because the canopy has higher foliage biomass than the understory (Lowman

& Wittman 1996), I expect herbivores (eg. leaf miners such as Agromyzidae) would be more abundant in the canopy.

Finally, I wished to examine major habitat correlates of the understory and canopy faunas in boreal mixedwood forests. Based on previous literature, I expect late-decay DWD, understory foliage thickness, and canopy openness/thickness to be strong predictors of the understory communities (Okland 1994, Jokimaki et al. 1998, Humphrey et al. 1999, Basset et al. 2001b,

Jakel and Roth 2004, Dennis unpubl.). The canopy communities could be influenced by age, height, canopy thickness and proximity of trees (Stork 1988). Structural and compositional heterogeneity also plays a key role in maintaining diverse resources for canopy communities

(Lowman & Wittman 1996, Jakel & Roth 2004) and features that increase heterogeneity such as

55 deciduous composition (Lowman & Wittman 1996, Progar & Schowalter 2002, Jakel & Roth

2004, Schowalter & Zhang 2005), age of trees (Schowalter 1989, Simandl 1993, Progar &

Schowalter 2002, Jakel & Roth 2004), and percent deciduous composition may be significant predictors. Features that strongly influence microclimate such as canopy openness or percent cover, or features that promote movement or protection of certain taxa are likely important

(Akutsu et al. 2007, Grimbacher & Stork 2007).

METHODS

The study area and study design, including assignment of cohort classes and habitat sampling, were described in Chapter 1.

Insect Sampling

Flying insects were collected using aerial flight intercept Malaise traps as outlined by the

Environmental Monitoring and Assessment Network (EMAN) (Finnamore et al. 1998). At each site, two traps were set within a small gap, one at the understory level and the other within the canopy, directly above the understory trap using methods outlined by Vance et al. (2003)

(Fig.2.1). Traps were at least 50 meters away from roads or water, and the height of each trap was recorded (average canopy trap height was 12.4 m, range 9.7 m- 15.32 m). Traps were open for the same week in each of June, July and August in 2007. Samples from the three months were consolidated for each site with understory and canopy samples remaining separate. Some bottles could not be collected due to wind disturbance or bear predation. In anticipation of this, I also collected insects during a second week of sampling in each month. If a first week sample was missing for a trap, I used the second week sample for it instead. If both weeks were

56 unavailable, the grand mean of the samples from the same month was used, and for species found in less than 20% of samples, the grand mode was used (Jongman et al. 1995). Data were then standardized to 100 trap days. Traps had both upper and lower collection bottles.

Unfortunately, an unforeseen issue was flooding of the bottom bottles, which in some cases lead to sample dilution and improper preservation. As a result, the bottom samples were not processed and are not used.

All Diptera and Hymenoptera were identified to family level by the author and a contract taxonomist. Several studies have successfully used a higher-taxon approach, which I chose for my study due to its efficiency (Balmford et al. 1996, Akutsu et al. 2007, Vance et al. 2007). In addition, one family of wasps, Diapriidae, was identified to genus and morphospecies by a contract taxonomist using the same techniques applied in Dennis (unpubl.). Voucher specimens will be made available to the Canadian National Collection in Ottawa.

57

Figure 2.1- Malaise trap set-up for canopy and understory malaise in boreal mixedwood forests of northeastern Ontario.

58

Statistical Analysis

Diptera and Hymenoptera Families

Univariate tests of cohort class effects, height effects and their interaction were conducted using SAS v. 9.1. For dipteran and hymenopteran families that were present in more than a third of sites, split-plot tests were used to test for both cohort and height effects and their interaction. Cohort class was a "between-site" treatment and height was a "within-site" treatment.

For rarer families, median tests were used to test for cohort effects and binomial tests for height effects. In exploratory plotting of un-standardized abundance, ABI0217 for Diptera and

Hymenoptera and TEM0195 for Hymenoptera were determined as outliers using Chauvenet’s criterion and were removed prior to analysis. Data did not meet the assumptions of normality, and thus was rank-transformed prior to the test.

I also explored overall community characteristics (diversity, richness, standardized abundance) of Diptera and Hymenoptera at the family level to test if cohort class, trap height or interaction effects were present. The same approach used in Chapter 1 was used to correct family richness for abundance. Shannon Wiener diversity index and standardized abundance (i.e., corrected for trapping effort) were calculated for each order. After examining the residuals, it was determined that the assumptions of normality and homogeneity were satisfied, hence no transformation was used prior to a split-plot ANOVA.

For multivariate analyses, I tested Hymenoptera and Diptera families in the understory and canopy against cohort classes as dummy variables and Weibull parameters (of stems >2.5 in dbh) in separate constrained ordinations (square-root transformed, correlation matrix). In each ordination, I tested the significance of all canonical axes (9999 Monte Carlo Permutations).

59

Ordination diagrams were generated and manual selections of each Weibull parameter (9999

Monte Carlo Permutations) were conducted.

For the separate family matrices (square-root transformed data, correlation matrix), PCAs for canopy samples and understory samples were undertaken and the environmental variables entered passively. Constrained RDAs were conducted using PCA sample scores from the first two axes to determine significant correlations. Cohort class was tested as dummy variables. For understory Hymenoptera and Diptera, variance partitioning was conducted on Weibull parameters of cohort class to determine the amount of shared variance with canopy height (mean, variance, semivariance), understory foliage thickness (mean, variance, semi-variance), and woody debris (volume of late and early decay DWD and snag BA). The amount of shared variance between Weibull parameters and canopy height (mean, variance, and semivariance) and overstory foliage thickness (mean, variance and semivariance) was explored for canopy

Hymenoptera and Diptera. Age, productivity and percent deciduous composition was compared with Weibull parameters in variance partitioning for all data sets to determine the amount of shared variance.

For the separate family matrices, Principal Components Analysis (PCA) was used with samples classified by canopy and understory to show if communities are vertical stratified. The gradient length was <3 for both orders; thus PCA was the appropriate analysis. Data were square-root transformed and correlation matrices were used since there were several super- abundant families in both orders. All families were included in the ordinations. Scatter plots of the samples and species were generated. Scatter plot diagrams were created to show only families found in >10% of sites.

60

Diapriidae morphospecies

The male morphospecies data set for Diapriidae was used because it was less sparse than the female set. There was no need for a split-plot analysis as Diapriidae were essentially restricted to the understory. To identify a cohort effect, one-way analysis of variance tests were conducted on the standardized abundance of each morphospecies. Data did not meet the assumptions of normality, and thus was log-transformed. For rare families found in less than 3 sites (all morphospecies except MSP_3B1, and MSP_4B), median tests were used to test for a cohort effect. One-way analysis of variance tests were also used to explore overall community characteristics (diversity, richness, standardized abundance) of Diapriidae morphospecies.

Richness was standardized for abundance using the simple approach referred to previously.

Shannon Wiener diversity index and standardized abundance (corrected for trapping effort) were calculated. After examining the residuals, it was determined that the assumptions of normality and homogeneity were met.

The data showed a relatively large gradient length (>4) in a DCA; thus I used unimodal multivariate tests. Data was log-transformed prior to ordinations (log(x+1)). Morphospecies abundances were examined using DCA with environmental variables plotted passively.

Constrained CCAs on the DCA sample scores were used to determine which environmental variables were important for Diapriidae. Each variable was tested using 9999 Monte Carlo permutations in a manual selection. Cohort class was tested as a dummy variable. Variance partitioning was conducted on Weibull parameters of cohort class to determine the amount of shared variance with canopy height (mean, variance, semivariance), understory foliage thickness

(mean, variance, semi-variance), and woody debris (volume of late and early decay DWD and

61 snag BA). Age, productivity and percent deciduous composition was compared with Weibull parameters in variance partitioning for all data sets to determine the amount of shared variance.

To further explore possible cohort effect, I conducted constrained ordinations using the

Weibull parameters of all stems and cohort classes (dummy variables). The significance of variables was tested by use of manual selection (9999 Monte Carlo Permutations).

62

RESULTS

Diptera and Hymenoptera

A total of 22,569 individuals of 54 families from the order Diptera were collected (Table

2.1). While most families compromised less than one percent of the total catch, the most abundant families overall were Chironomidae (33.1%), Cecidomyiidae (19.4%), Sciaridae

(11.1%) and Ceratopogonidae (5.4%). Catches were higher in the understory traps (14,129 individuals) compared to canopy traps (8,440 individuals).

For Hymenoptera, 2,144 individuals from 34 families were captured (Table 2.1). Again, most families were less than 1% of the total catch, with Diapriidae (33.4%), Ichneumonidae

(25.6%), Braconidae (21.6%) and Mymaridae (4.7%) being the four most abundant overall.

In the univariate tests, cohort class effects were less prominent than height effects.

Among dipterans, Culicidae, Phoridae, Tachinidae, and Syrphidae significantly differed by cohort class (P<0.05, Table 2.1). Culicidae and Phoridae had distinctly higher abundances in cohort 4 (lowest abundance in cohort 2) (Table 2.1). Syrphidae also had distinctly lower abundances in cohort 2 and highest in cohort 1. Tachinidae was lowest in cohort 3 and highest in cohort 1. For hymenopterans, only Diapriidae and Ichneumonidae significantly differed with cohort class. Diapriidae and Ichneumonidae had much higher abundances in cohort 1 and lower abundances in cohort 2 (Table 2.1). Cohort class associations are interpreted in multivariate ordinations (next section).

Dipteran families that were significantly more abundant in the understory than the canopy were Anthomyiidae, Cecidomyiidae, Dryomyzidae, Mycetophilidae, Pipunculidae, Psychodidae,

Phoridae, Rhagionidae, Sciaridae, Syrphidae, Tipulidae, and Xylophagidae; families showing significantly higher abundances in the canopy than understory were the Agromyzidae,

63

Aulacigastridae, Chamaemyiidae, Lonchaeidae, Odiniidae, Periscelidae, Sarcophagidae and

Tachinidae (Table 2.1). The proportions of Anthomyiidae, Calliphoridae, Dolicopodidae,

Phoridae, and Tabanidae in the canopy versus the understory also differed with cohort class.

Hymenopteran families significantly more abundant in the understory were Ceraphronidae,

Diapriidae, Ichneumonidae, Tenthredinidae, and Vespidae were significantly different between trap heights (Table 2.1). Pompilidae and Chrysididae were the only hymenopteran families that were significantly more abundant in canopy. No interaction between trap height and cohort class was detected for hymenopterans.

64

Table 2.1- Mean abundance (standardized to 100 trap-days) and univariate test results according to cohort classes (1-4) and trap heights (canopy versus understory) for families of Diptera and Hymenoptera collected from understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007

Mean abundance among sites Mean abundance Test results* per cohort class per trap type (F/Z/Chi-Square and P value) Overall 4 Mean 1 2 3 (Diptera n = Cohort Trap 5, Family abundance (n = 4) (n = 3) (n = 5) Hymenoptera Understory Canopy Class Height Interaction n=4) Diptera C=2.58 Z=1.00 2.12 0.00 1.59 0.00 5.56 2.12 0.00 Acartophthalmidae P=0.4604 P=0.3173 F=2.84 F=11.7 F=1.00 19.08 32.09 20.63 8.04 18.82 5.20 13.87 Agromyzidae P=0.0792 P=0.0045 P=0.422 F=1.36 F=1.88 F=1.37 10.88 10.85 3.17 13.33 12.70 6.12 4.76 Anisopodidae P=0.229 P=0.296 P=0.296 F=0.34 F=8.41 F=4.35 102.03 97.38 80.95 76.43 137.00 70.22 31.81 Anthomyiidae P=0.799 P=0.0124 P=0.0250 F=0.05 F=0.40 F=2.64 18.95 16.45 28.57 17.02 17.41 11.31 7.64 Asilidae P=0.984 P=0.539 P=0.093 C=2.83 Z=2.45 5.29 11.90 0.00 6.67 2.38 0.79 4.50 Aulacigastridae P=0.4178 P=0.0143 C=5.30 Z= -0.44 C=2.91 2.12 3.57 1.59 0.00 3.17 1.32 0.79 Bibionidae P=0.1510 P=0.6547 P=1.0000 C=3.4987 Z= 1.34 C=2.23 1.85 0.00 6.35 0.00 2.38 0.26 1.59 Bombyliidae P=0.3209 P=0.1797 P=1.0000 F=0.84 F=3.82 F=3.48 7.41 3.57 9.52 3.81 11.90 3.97 3.44 Calliphoridae P=0.496 P=0.072 P=0.0473 F=1.14 F=42.28 F=1.62 1160.74 1181.65 807.94 1396.19 1126.98 936.40 224.34 Cecidomyiidae P=0.3685 P=<0.0001 P=0.2311 F=1.00 F=0.46 F=2.16 320.68 406.07 182.54 180.36 449.75 160.79 159.89 Ceratopogonidae P=0.4220 P=0.5097 P=0.1419

65

Table 2.1- continued C=2.13 Z= 2.24 3.17 3.57 7.94 3.81 0.00 0.00 3.17 Chamaemyiidae P=0.5462 P=0.0253 C=3.37 Z=1.00 0.26 0.00 0.00 0.00 0.79 0.00 0.26 Chaoboridae P=0.3373 P=0.3173 F=0.56 F=0.42 F=0.66 1977.49 1434.96 1958.73 1574.64 2684.28 1013.37 964.12 Chironomidae P=0.6532 P=0.5294 P=0.5919 F=1.54 F=1.00 F=2.14 10.85 25.00 3.17 2.86 11.90 5.56 5.29 Chloropidae P=0.2505 P=0.3354 P=0.1444 F=1.22 F=0.14 F= 0.96 12.89 15.72 3.17 16.19 13.10 7.33 5.56 Clusiidae P=0.3417 P=0.7165 P=0.4404 F=3.75 F=4.02 F=2.39 32.44 28.27 6.35 10.71 66.37 27.35 5.09 Culicidae P=0.0387 P=0.0664 P=0.1155 F=0.35 F=0.49 F=6.79 71.94 61.54 53.97 50.00 106.15 44.30 27.65 Dolichopodidae P=0.7928 P=0.4960 P=0.0054 F=1.38 Z=-1.00 F=7.64 4.76 1.19 4.76 4.76 7.14 3.70 1.06 Drosophilidae P=0.7112 P=0.3173 P=0.1786 F=0.44 F=6.38 F=1.04 5.82 8.33 3.17 1.90 8.73 4.76 1.06 Dryomyzidae P=0.7272 P=0.0253 P=0.4082 F=3.57 F=4.38 F=1.09 77.79 88.58 28.57 45.48 122.12 55.10 22.69 Empididae P=0.0736 P=0.0565 P=0.3880 F=6.37 Z=1.34 3.70 4.76 3.17 0.00 6.35 1.59 2.12 Ephydridae P=0.0949 P=0.1797 F=0.57 F=1.55 F=1.70 8.99 2.38 1.59 7.62 18.25 7.41 1.59 Heliomyzidae P=0.6422 P=0.2351 P=0.2163 F=0.52 F=1.42 F= 0.19 56.90 89.25 52.38 56.25 38.12 23.55 33.35 Lauxanidae P=0.6751 P=0.2544 P=0.9024 F=2.94 F=18.81 F=1.75 48.21 88.67 15.87 29.05 53.37 11.83 36.38 Lonchaeidae P=0.0728 P=0.0008 P=0.2053 F=1.36 F=1.56 F=1.98 37.23 58.46 17.46 13.10 53.08 16.66 20.57 Milichiidae P=0.2998 P=0.2337 P=0.1667 F=2.69 F=1.18 F=0.95 136.14 153.87 58.73 132.92 165.70 94.72 41.42 Muscidae P=0.0894 P=0.2976 P=0.4432

66

Table 2.1- continued F=1.86 F=25.66 F=0.73 93.35 142.04 28.57 86.37 99.08 79.14 14.20 Mycetophilidae P=0.1870 P=0.0002 P=0.5521 F=0.71 F=15.45 F=2.05 40.26 32.65 77.78 55.71 13.69 3.09 37.17 Odiniidae P=0.5632 P=0.0017 P=0.1559 C=1.40 Z=1.13 C=2.63 3.17 4.76 4.76 2.86 1.59 0.53 2.65 Otitidae P=0.7059 P=0.2568 P=1.0000 C=4.25 Z=0.00 C=2.77 1.32 4.76 0.00 0.95 0.00 0.79 0.53 Pallopteridae P=0.2357 P=1.0000 P=1.00 C=1.52 Z=2.00 1.06 1.19 0.00 1.90 0.79 0.00 1.06 Periscelidae P=0.6768 P=0.0455 F=3.87 F=-3.15 C=6.77 230.28 211.58 103.17 174.11 353.10 185.62 44.66 Phoridae P=0.0352 P=0.0016 P=0.0662 F=0.72 F=20.48 F=2.36 46.32 42.31 31.75 69.88 36.66 39.88 6.45 Pipunculidae P=0.5563 P=0.0006 P=0.1156 C=1.38 Z=1.00 1.32 0.00 0.00 0.95 3.17 1.32 0.00 Platypezidae P=0.7099 P=0.3173 C=2.18 0.79 0.00 0.00 0.00 2.38 0.79 0.00 Psilidae P=0.5355 F=2.20 F=-2.00 C=1.76 97.30 46.53 14.29 119.82 153.89 78.83 18.47 Psychodidae P=0.1366 P=0.0455 P=0.8901 F=0.04 F=-3.05 C=7.05 15.11 11.45 9.52 15.24 20.24 12.20 2.91 Rhagionidae P=0.9887 P=0.0023 P=0.0769 C=2.18 Z=1.00 1.06 0.00 0.00 0.00 3.17 1.06 0.00 Rhinophoridae P=0.5355 P=0.3173 F=1.88 F=15.78 F=0.41 24.30 47.40 7.94 7.32 31.23 3.75 20.55 Sarcophagidae P=0.1834 P=0.0015 P=0.7509 C=1.61 Z=-0.71 C=3.31 5.03 2.38 1.59 5.71 7.94 3.17 1.85 Scathophagidae P=0.6578 P=0.4795 P=0.7857 C=2.56 Z=1.41 1.32 0.00 0.00 0.95 3.17 1.06 0.26 Scatopsidae P=0.4646 P=0.1573 F=1.21 F=30.48 F=0.87 662.06 779.96 388.89 541.55 820.49 492.32 169.74 Sciaridae P=0.3444 P=<0.0001 P=0.4805

67

Table 2.1- continued C=3.52 Z=-0.45 F=3.82 1.85 3.57 1.59 2.86 0.00 1.32 0.53 Sciomyzidae P=0.3186 P=0.6547 P=0.6000 F=2.14 F=0.03 F=3.26 264.32 551.03 26.98 356.67 114.88 125.56 138.76 Simuliidae P=0.1439 P=0.8703 P=0.0565 C=7.28 Z=1.00 2.12 5.95 0.00 0.00 2.38 1.85 0.26 Sphaeroceridae P=0.0634 P=0.3173 C=1.90 Z=0.00 C=5.54 2.91 10.71 1.59 0.00 0.79 2.38 0.53 Stratiomyidae P=0.5931 P=1.0000 P=0.3333 F=0.37 F=12.11 F=0.12 11.11 11.90 20.63 10.48 6.35 1.32 9.79 Strongylophthalmyiidae P=0.7760 P=0.0041 P=0.9496 F=4.88 F=8.33 F=2.19 25.85 37.94 6.35 17.68 34.35 20.94 4.91 Syrphidae P=0.0174 P=0.0127 P=0.1378 F=0.38 F=3.92 F=0.42 36.78 29.56 38.10 44.88 34.18 28.28 8.50 Tabanidae P=0.7717 P=0.0731 P=0.7406 F=4.22 F=12.47 F=2.00 178.37 411.49 114.29 79.11 137.72 59.57 118.80 Tachinidae P=0.0273 P=0.0037 P=0.1642 C=2.18 0.26 0.00 0.00 0.00 0.79 0.26 0.00 Tanypezidae P=0.5355 F=0.91 F=123.94 F=1.13 47.34 61.45 25.40 41.79 53.52 41.55 5.79 Tipulidae P=0.4644 P=<0.0001 P=0.3721 F=0.33 F=20.01 F=0.88 35.98 47.62 44.44 18.10 38.89 35.45 0.53 Xylophagidae P=0.8044 P=0.0006 P=0.4777 OVERALL MEAN 110.57 117.15 79.81 98.26 131.81 69.2 41.3

Hymenoptera C=2.20 Z=1.00 0.26 0.00 0.00 0.95 0.00 0.00 0.25 Ampulicidae P=0.5319 P=0.3173 C=1.65 Z=0.00 F=2.77 1.06 0.00 0.00 0.95 2.38 0.79 0.25 Aphelinidae P=0.6474 P=1.0000 P=1.000 C=3.00 Z=1.00 0.26 1.19 0.00 0.00 0.00 0.26 0.00 Argidae P=0.3916 P=0.3173

68

Table 2.1- continued F=0.47 F=1.74 F=1.38 122.54 132.53 58.73 80.08 183.16 58.23 62.68 Braconidae P=0.7072 P=0.2114 P=0.2958 C=2.71 Z=2.64 3.17 4.76 0.00 2.86 3.97 2.91 0.25 Ceraphronidae P=0.4381 P=0.0082 C=3.37 Z=2.33 C=1.78 6.88 10.71 1.59 1.90 11.11 2.12 4.51 Chrysididae P=0.3376 P=0.0196 P=1.0000 C=2.34 Z=1.41 0.53 0.00 1.59 0.95 0.00 0.53 0.00 Cimbicidae P=0.5045 P=0.1573 C=2.20 Z=1.00 0.53 0.00 0.00 0.95 0.79 0.53 0.00 Colletidae P=0.5319 P=0.3173 C=3.00 Z=1.00 0.53 1.19 0.00 0.00 0.79 0.53 0.00 Crabronidae P=0.3916 P=0.3173 C=0.00 0.53 0.00 0.00 0.00 1.59 0.26 0.25 Cynipidae P=1.00 C=1.70 Z=0.00 F=2.77 2.38 1.19 1.59 6.67 0.00 0.53 1.75 Cynipoidea P=0.6375 P=1.00 P=1.00 F=6.46 F=28.17 F=1.57 189.42 345.66 12.70 176.08 184.73 178.60 10.25 Diapriidae P=0.0075 P<0.0002 P=0.2480 C=1.33 Z=0.00 C=5.54 1.59 2.38 0.00 1.90 1.59 1.06 0.50 Dryinidae P=0.7224 P=1.00 P=0.3333 C=3.00 Z=1.00 0.26 0.00 0.00 0.00 0.79 0.26 0.00 Embolemidae P=0.3916 P=0.3173 C=1.54 Z=-1.00 C=2.46 12.43 9.52 3.17 11.43 19.84 10.05 2.26 Encyrtidae P=0.6727 P=0.3173 P=0.4286 C=2.80 Z=-1.63 C=2.63 2.38 1.19 3.17 1.90 3.17 1.85 0.50 Eulophidae P=0.4221 P=0.1025 P=1.000 C=2.34 Z=0.00 C=2.77 1.06 2.38 1.59 0.95 0.00 0.79 0.25 Figitidae P=0.5045 P=1.0000 P=1.00 C=6.64 Z=0.00 C=1.73 2.12 4.76 0.00 0.00 3.17 1.59 0.50 Formicidae P=0.0842 P=1.000 P=1.00 C=1.65 Z=1.41 3.44 0.00 0.00 10.48 1.59 3.17 0.25 Halictidae P=0.6474 P=0.1573

69

Table 2.1- continued C=2.20 Z=1.00 0.26 0.00 0.00 0.95 0.00 0.26 0.00 Heloridae P=0.5319 P=0.3173 F=8.54 F=11.48 F=0.80 145.26 187.85 44.44 154.53 159.55 122.70 22.62 Ichneumonidae P=0.0026 P=0.0054 P=0.5172 C=3.00 Z=1.00 0.53 0.00 0.00 0.00 1.59 0.00 0.50 Megachilidae P=0.3916 P=0.3173 F=0.58 F=1.52 F=0.03 26.76 18.35 33.33 29.34 26.94 17.82 9.73 Mymaridae P=0.6376 P=0.2411 P=0.9937 C=1.65 Z=1.41 0.53 0.00 0.00 0.95 0.79 0.00 0.50 Philanthidae P=0.6474 P=0.1573 C=0.86 Z=-1.89 F=2.97 3.97 4.76 3.17 1.90 5.56 3.17 0.75 Platygastridae P=0.8340 P=0.0588 P=0.5714 C=2.36 Z=1.41 1.06 2.38 1.59 0.95 0.00 0.79 0.25 Platygastroidea P=0.5008 P=0.1573 C=1.70 Z=2.00 3.44 3.57 3.17 2.86 3.97 0.79 2.76 Pompilidae P=0.6375 P=0.0455 C=1.65 Z=0.00 C=2.77 0.53 0.00 0.00 0.95 0.79 0.26 0.25 Proctotrupidae P=0.6474 P=1.00 P=1.00 C=5.66 Z=-0.630 C=2.00 10.85 22.62 1.59 3.81 13.49 7.14 3.51 Pteromalidae P=0.1291 P=0.5271 P=1.0000 C=1.62 Z=0.378 C=2.97 3.17 4.76 1.59 3.81 2.38 1.32 2.01 Scelionidae P=0.6553 P=0.7055 P=1.0000 C=1.20 Z=-1.89 C=5.74 7.14 9.52 3.17 5.71 8.73 6.08 1.00 Tenthredinidae P=0.7524 P=0.0588 P=0.2857 C=3.00 Z=1.00 0.53 0.00 0.00 0.00 1.59 0.53 0.00 Trichogrammatidae P=0.3916 P=0.3173 C=3.39 Z=-1.73 F=5.86 11.36 9.80 14.29 9.50 12.49 7.93 3.25 Vespidae P=0.3344 P=0.0833 P=0.5091 C=2.07 Z=1.41 0.53 1.19 0.00 0.00 0.79 0.00 0.50 Xiphydriidae P=0.5587 P=0.1573 TOTAL MEAN 16.69 22.38 5.50 15.28 18.90 12.7 4.0 Table 2.1- continued

70

*F = split-plot ANOVA cohort (df=3), height effects (df=1) and their interaction (df=3); C=median test (cohort effect, df=3); Z=exact likelihood ratio test (trap height effect, df=1), and F= binomial two-sided test (df=3).

71

For community metrics, cohort class had a significant effect on hymenopteran family richness only (F3=7.92, P=0.003). Dipteran family richness did not differ significantly among cohort classes (F3=0.87 P=0.4809), nor did the dipteran and hymenopteran family diversity

(F3=2.78 P=0.083 and F3=0.85 P=0.493, respectively) and family abundance (F3=0.44 P=0.728 and F3=1.16 P=0.367 respectively; Fig. 2.2, 2.3 & 2.4).

In contrast, for both Diptera and Hymenoptera, family abundance was significantly different with trap height, with more individuals occurring in the understory than in the canopy

(F1=9.00 P=0.010, and F1=9.61 P=0.009 respectively; Fig. 2.2, 2.3, & 2.4). Hymenopteran family richness was significantly different between the two trap heights, a trend which is clearly visible in the rarefaction curves (F1=6.17 P=0.029, Fig.3B). Differences in standardized dipteran family richness between trap heights were not significant (F1=0.49, P=0.4941). Family diversity did not differ with trap height for both Diptera and Hymenoptera (F1=0.36, P=0.561; F1=2.03

P=0.180, respectively). There was some interaction between cohort class and trap height detected for Hymenoptera family richness; the response of family abundance, richness and diversity did not significantly differ for Diptera (F3=1.39, P=0.289; F3=2.81, P=0.081 and F3=1.10, P=0.385; respectively) nor Hymenoptera except for family richness (F3=1.91 P=0.181; F3=3.63, P=0.045 and F3=0.95 P=0.447; respectively).

72

Figure 2.2- Abundance per 100 trap days of A. Diptera and B. Hymenoptera for cohort classes 1-4 and understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007. Standard error bars shown. Note contrasting scale.

73

Figure 2.3- Family diversity (Shannon Wiener indices) of Diptera and Hymenoptera for cohort classes 1-4 and understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007. Standard error bars shown.

74

40 A. 35

30

25

20

15

family richnessfamily 10 y = 4.7152Ln(x) - 2.3681 5

0 0 200 400 600 800 1000 1200 1400 1600 Number of individuals

14 B. 12

10

8 y = 1.0873ln(x) + 4.2943 6

4 Familyrichness

2

0 0 20 40 60 80 100 120 140 160 Number of individuals

Figure 2.4- Family richness against unstandardized abundance (Number of individuals) for A. Diptera and B. Hymenoptera from understory and canopy Malaise traps in boreal mixedwoods of northeastern Ontario in 2007, with fitted regressions shown. Note contrasting scale. ●=understory □=canopy.

75

In the multivariate analyses, there was a clear difference in dipteran and hymenopteran families caught in the understory versus the canopy (Fig. 2.5). In an unconstrained ordination, the first axis clearly represents trap height for both orders. More dipteran and hymenopteran families were associated with the understory. Associations according to height, as revealed by family vectors, agreed with the univariate tests (Fig. 2.5; Table 2.1).

76

Fig. 2.5– Principal Components Analysis (PCA) for families of A. Diptera and B. Hymenoptera found in boreal mixedwood sites in northeastern Ontario caught in understory and canopy Malaise traps. Only families found in >10% of sites shown. ABI0217 outlier removed (TEM01095 removed from Hymenoptera); □=canopy ●=understory. Variable acronyms are identified in Appendix 2.

77

Weibull parameters of all stems were not significant for either the Diptera or

Hymenoptera family matrices at either trap height (Table 2.2). When Weibull parameters were examined separately, however, Weibull scale was significant for understory Diptera, but not shape (9.8% of the variance, P=0.038 and 6.6% of the variance, P=0.356 respectively; Fig. 2.6).

Weibull scale and shape were not significant for canopy Diptera (scale: 6.2%, P=0.449 and shape:

5.7%, P=0.582). Similarly, they were not significant for Hymenoptera in the understory (scale:

7.8%, P=0.182; shape: 3.4%, P=0.982) or the canopy (scale: 6.7%, P=0.369; shape: 5.9%,

P=0.558; Fig. 2.7).

In constrained ordinations, cohort classes were not significant constraining variables for either the Diptera or Hymenoptera matrices (diagram not shown; Table 2.2). For Diptera that had a significant cohort effect in univariate tests, their vectors lined up with the following cohort centroids in a cohort-constrained ordinations: Culicidae (cohort 1), Phoridae (cohort 4),

Syrphidae (cohorts 1 and 4), and Tachinidae (cohort 1 & 4; results not shown). From the significant hymenopteran families, Ichneumonidae and Diapriidae in the understory were associated with cohort 1 space (results not shown).

In unconstrained ordinations (PCAs), cohort class separation is not apparent for Diptera and Hymenoptera at the understory and canopy level, apart from some separation of Dipteran families between cohort 1 and 2 centroids in the understory, with cohort 3 appearing to cluster near cohort 2 (Fig. 2.8 & 2.9). Cohort 1 was a significant predictor of understory Diptera, and cohort 2 was a significant predictor after mean understory foliage thickness was selected (Fig.

2.8, A). For canopy Diptera, cohort 1 was significant only in forward selection following the addition of variance of canopy height (Fig. 2.8, B).

78

I was interested to know what other structural variables may be important correlates of family abundances. Using RDA on PCA axis scores, I found that mean understory foliage thickness (30.3% P=0.011), cohort 1 centroid (19.8% P=0.023) and mean overstory foliage thickness (18.5% P=0.035) were important for the understory Dipterans (Fig. 2.8, A). In a forward selection, FEC productivity (20.5% var., P=0.016), cohort 2 centroid (11.7% var.,

P=0.032), and tree volume per hectare (10%, P=0.018) explained further variation once mean understory foliage thickness had been selected.

In contrast, dipteran families caught in the canopy were more sensitive to the residual variance of canopy height (18.2% var., P=0.030; Fig. 2.8, B). In a forward selection, cohort 1 explained further variation in canopy dipteran communities (18.9% var., P=0.032).

The Hymenoptera in the understory were sensitive to FEC productivity (28%, P=0.018), and nearly significant was the residual variance of shrub foliage thickness, a variable that was significant in a forward selection and almost significant on its own (16.8%, P=0.0255 and 16.9%,

P=0.051 respectively; Fig. 2.9, A). In contrast, the basal area of all stems was a significant predictor of hymenoptera families in the canopy (18.9%, P=0.033; Fig. 2.9, B). In a forward selection, the residual variance (15.9%, P=0.040) and semi-variance (15.9%, P=0.021) of shrub foliage thickness further explain further community variation after basal area was selected.

For understory Diptera and Hymenoptera, Weibull parameters, canopy height, and DWD were not significant predictors, but understory foliage thickness was significant only for understory Hymenoptera, sharing 5.5% of its variance with Weibull (Table 2.3). For canopy

Diptera and Hymenoptera, none of the selected habitat variables (Weibull, canopy height or overstory foliage thickness) were significant (Table 2.4).

79

Age and percentage of deciduous composition alone could not constrain dipteran nor hymenopteran families in either strata, however, productivity alone was significant for understory Hymenoptera (Table 2.5). Furthermore, productivity continued to significantly explain unique variance, while Weibull could not. Productivity and Weibull together became significant predictors of not only understory Hymenoptera, but also understory Diptera. Its highest shared variance with Weibull parameters was 3.8% for understory Hymenoptera.

80

Table 2.2 - Significance of all canonical axes for dipteran or hymenopteran family correlation matrix (square-root transformed) in understory and canopy traps in boreal mixedwoods of northeastern Ontario and their relationship to cohort class and Weibull parameters of all stems (>2.5cm dbh).

Cohort class Weibull 2.5 P value % var.* P value % var.* Diptera understory 0.094 22.0 0.059 16.0 Diptera canopy 0.323 19.9 0.506 12.3 Hymenoptera understory 0.821 16.4 0.573 11.8 Hymenoptera canopy 0.301 19.8 0.323 13.3

* Percent of total variance explained.

81

Figure 2.6 – Redundancy Analysis (RDA) of dipteran families in the A. understory and B. canopy constrained by Weibull parameters from boreal mixedwood sites in northeastern Ontario in 2007. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Only families found in >10% of sites shown. ABI0217 outlier removed. Variable acronyms are identified in appendix 1 and 2.

82

Figure 2.7– As Figure 2.6, except that ordinations are for hymenopteran families in the A. understory and B. canopy

83

Figure 2.8 – Principal Components Analysis (PCA) of dipteran families A. in understory traps and B. in canopy Malaise traps from boreal mixedwood sites in northeastern Ontario with environmental variables plotted passively (only those with axis scores >0.40 shown). ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Only families found in >10% of sites are shown. ABI0217 outlier removed. Significance for each variable are shown (**= P<0.01, *= P<0.05, +=0.05 in forward selection). Acronyms for variables are found in Appendix 1 and 2.

84

Figure 2.9- as Figure 2.8 except that ordinations are for hymenopteran families A. in understory traps and B. in canopy Malaise traps

85

Table 2.3- Partial redundancy analysis examining the relative value of various habitat predictors in explaining dipteran and hymenopteran communities from understory malaise traps in boreal mixedwoods of northeastern Ontario in 2007 in comparison to Weibull parameters from the diameter distribution of stems >2.5 cm dbh.

Canopy height Understory DWD foliage thickness % P % P % P value var. value var. value var. Understory Together 31.4 0.610 Together 32.8 0.513 Together 38.2 0.211 Diptera Weibull 17.6 0.157 Weibull 17.6 0.157 Weibull 17.6 0.157 Canopy 18.4 0.576 Understory 17.9 0.615 DWD 11.5 0.979 height foliage thickness Unique to 12.9 0.483 Unique to 12.9 0.483 Unique to 26.7 0.027 Weibull Weibull Weibull Unique to 13.8 0.864 Unique to 15.2 0.765 Unique to 20.6 0.338 Canopy understory DWD height foliage thickness Shared 4.7 Shared 4.7 Shared -9.1

Understory Together 32.5 0.414 Together Together 28.9 0.547 49.4 0.045 Hymenoptera Weibull 12.0 0.432 Weibull 12.0 0.432 Weibull 12.0 0.432 Canopy understory DWD height 17.3 0.491 foliage 39.4 0.021 13.3 0.699 thickness Unique to 15.2 0.289 Unique to 6.5 0.832 Unique to 15.6 0.305 Weibull Weibull Weibull Unique to 20.5 0.363 Unique to 37.4 0.028 Unique to 16.9 0.524 canopy understory DWD height foliage thickness Shared -3.2 Shared 5.5 Shared -3.6

86

Table 2.4- same as Table 2.3, except for canopy Malaise trap samples

Canopy height Overstory foliage thickness % var. P value % var. P value Canopy Together 26.2 0.896 Together 25.6 0.905 Diptera

Weibull 17.6 0.157 Weibull 17.6 0.157 Canopy Overstory height 16.4 0.740 foliage 17.5 0.649 thickness Unique to 9.8 0.787 Unique to 8.1 0.922 Weibull Weibull Unique to 18.2 0.681 Unique to 17.6 0.706 canopy overstory height foliage thickness Shared -1.8 Shared -0.1

Canopy Together 32.5 0.394 Together 31.2 0.454 Hymenoptera

Weibull 12.3 0.448 Weibull 12.3 0.448 Canopy 14.5 0.718 Overstory 19.4 0.372 height foliage thickness Unique to 18.0 0.185 Unique to 11.8 0.465 Weibull Weibull Unique to 20.2 0.356 Unique to 18.4 0.408 canopy understory height foliage thickness Shared -5.7 Shared 1.0

87

Table 2.5- same as Table 2.3, except examining age, productivity and percentage of deciduous composition for both understory and canopy Malaise trap samples

(%) FEC Age Deciduous productivity composition % P % P % P var. value var. value var. value Understory Together 23.9 0.216 Together 29.9 0.041 Together 23.6 0.224 Diptera Weibull 17.6 0.157 Weibull 17.6 0.157 Weibull 17.6 0.157 FEC Percent deciduous Age 4.6 0.723 10.9 0.103 8.2 0.248 productivity composition Unique to Unique to Unique to 19.3 0.114 19.0 0.070 15.4 0.254 Weibull Weibull Weibull Unique to percent Unique to Unique to FEC 6.3 0.408 12.3 0.056 deciduous 6.0 0.454 Age productivity composition Shared -1.7 Shared -1.4 Shared 2.2

Canopy Together 10.4 0.986 Together 21.2 0.366 Together 10.6 0.990 Diptera Weibull 17.6 0.157 Weibull 17.6 0.157 Weibull 17.6 0.157 FEC Percent deciduous Age 12.7 0.070 12.7 0.070 2.6 0.970 productivity composition Unique to Unique to Unique to 7.4 0.953 8.5 0.864 8.1 0.933 Weibull Weibull Weibull Unique to percent Unique to Unique to FEC 2.5 0.954 13.3 0.072 deciduous 2.7 0.947 Age productivity composition Shared 0.5 Shared -0.6 Shared -0.2

88

Table 2.5- continued

Understory Together 23.5 0.242 Together 36.4 0.036 Together 15.5 0.594 Hymenoptera Weibull 12.0 0.432 Weibull 12.0 0.432 Weibull 12.0 0.432 Percent FEC Age 6.2 0.331 28.2 0.010 deciduous 5.7 0.387 productivity composition Unique to Unique to Unique to 17.3 0.207 8.2 0.534 9.7 0.565 Weibull Weibull Weibull Unique to Unique to Unique to 11.5 percent 0.125 FEC 24.4 0.014 3.5 0.670 Age deciduous productivity composition Shared -5.3 Shared 3.8 Shared 2.3

Canopy Together 24.5 0.162 Together 15.3 0.654 Together 25.5 0.130 Hymenoptera Weibull 12.3 0.448 Weibull 12.3 0.448 Weibull 12.3 0.448 Percent FEC Age 6.9 0.298 4.9 0.472 deciduous 11.6 0.087 productivity composition Unique to Unique to Unique to 17.5 0.154 10.4 0.611 13.9 0.281 Weibull Weibull Weibull Unique to Unique to Unique to percent 12.2 0.091 FEC 3.0 0.786 13.2 0.076 Age deciduous productivity composition Shared -5.2 Shared 1.9 Shared -1.6

89

Diapriidae

In total, 567 males and 58 females were collected and identified. From the males, two subfamilies were caught: Diapriinae and Belytinae (Table 2.4). Diapriinae was represented by one morphospecies with one individual. Belytinae was much more abundant, with 42 morphospecies and 566 individuals. Female diapriids, which were not analyzed in this project, were represented by 1 individual from the subfamily Ismarinae, 9 individuals in 4 morphospecies within the subfamily Diapriinae, and 48 individuals in 35 morphospecies from the Belytinae.

Univariate tests revealed no significant relationships between any of the morphospecies and cohort class (Table 2.6). In CCA ordinations (results not shown), neither cohort class nor

Weibull parameters (all stems) were significant at predicting assemblages (18.0% P=0.502;

31.2% P=0.121; 21.8% P=0.141, respectively). When the Weibull parameters of all stems were considered separately, shape and scale were unable to significantly constrain morphospecies

(results not shown).This was also evident in the multivariate DCA, where there was only a minor separation between cohort classes (Fig. 2.10). The cohort 1 dummy variable, however, was able to significantly constrain the data (16.9%, P=0.022), but was not as strong of a predictor as the residual semi-variance of the overstory foliage thickness (17.4%, P=0.010) and residual variance of the understory foliage thickness (16.8%, P=0.017) which was able to significantly predict community variation even in a forward selection (17.7% P=0.007, with variance of overstory foliage thickness selected first).

In the variance partitioning, none of the variable sets (Weibull, canopy height, understory foliage thickness, DWD, age, productivity nor percent deciduous composition) were significant alone, uniquely or together with Weibull parameters (Table 2.7 & 2.8). Weibull shared the

90 highest amount of variance with understory foliage thickness and DWD (5.3% and 6.2%, respectively).

There was no cohort class effect present when examining community characteristics (Fig.

2.11, 2.12, & 2.13); morphospecies richness, abundance, and diversity were not significantly different according to cohort class (1.24, P=0.331; 2.34, P=0.118; 2.17, P=0.138 respectively).

91

Table 2.6- Mean abundance (standardized to 100 trap-days) and univariate test results of cohort class effects for male Diapriidae morphospecies collected from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007. Description of each morphospecies can be found in Appendix 3. Overall mean Overall Mean abundance among sites Test results* Morphospecies abundance (per 100 per cohort class (F/Chi-Square and P trap days) 1 2 3 4 value) MSP 1 6.97 10.71 0.00 0.95 12.82 C=4.54, P=0.208 MSP 1(1) 4.39 10.71 0.00 4.29 2.38 C=4.53, P=0.209 MSP 1A 4.30 7.14 0.00 0.95 7.14 C=1.93, P=0.586 MSP 1A1 1.06 1.19 0.00 0.95 1.63 C=1.24, P=0.742 MSP 1S 0.81 1.19 0.00 0.95 0.79 C=3.17, P=0.366 MSP 1S1 0.54 1.19 0.00 0.95 0.00 C=1.85, P=0.602 MSP 1B 0.54 0.00 0.00 0.00 1.59 C=1.38, P=0.701 MSP 1B1 0.54 1.19 0.00 0.95 0.00 C=1.86, P=0.602 MSP 1C 1.23 2.38 0.00 1.43 0.79 C=2.83, P=0.418 MSP 1C1 4.27 8.33 0.00 3.81 4.01 C=4.36, P=0.225 MSP 1C2 0.54 0.00 0.00 0.00 1.59 C=1.38, P=0.701 MSP 1D 5.07 11.90 0.00 2.86 4.80 C=2.64, P=0.450 MSP 1D1 1.60 5.95 0.00 0.95 0.00 C=2.83, P=0.418 MSP 2 0.81 1.19 0.00 0.95 0.79 C=0.79, P=0.851 MSP 2(1) 2.01 2.38 0.00 2.38 2.46 C=6.64, P=0.084 MSP 2A 3.99 0.00 0.00 7.62 5.56 C=2.61, P=0.456 MSP 2B 4.06 1.19 0.00 3.81 7.94 C=1.79, P=0.617 MSP 2D 2.28 1.19 0.00 5.24 1.59 C=1.98, P=0.576 MSP 2D1 1.80 3.57 0.00 1.59 1.59 C=4.71, P=0.195 MSP 2D2 1.85 0.00 0.00 5.71 0.79 C=5.67, P=0.129 MSP 3 18.39 40.48 1.67 5.52 22.22 C=1.45, P=0.694 MSP 3(2) 1.34 3.57 0.00 0.00 1.63 C=4.43, P=0.218 MSP 3B 12.04 21.43 0.00 8.89 14.10 C=3.65,P=0.302 MSP 3B1 15.66 26.19 0.00 15.14 16.63 F=2.12, P=0.144 MSP 3D 6.98 0.00 1.59 4.76 16.51 C=0.30, P=0.960 MSP 4 8.90 13.10 0.00 3.43 14.72 C=6.10, P=0.106 MSP 4(1) 1.34 1.19 0.00 2.86 0.79 C=1.79, P=0.617 MSP 4(2) 0.53 2.38 0.00 0.00 0.00 C=2.00, P=0.572 MSP 4A 8.65 15.48 0.00 3.81 11.90 C=4.53, P=0.209 MSP 4B 14.75 25.00 1.59 4.76 22.26 C=2.45, P=0.485 MSP 4B1 4.06 1.19 0.00 6.67 5.56 C=1.54, P=0.672 MSP 4E 4.93 3.57 0.00 8.89 4.76 C=2.30, P=0.513 MSP 4E1 0.81 1.19 0.00 0.95 0.79 C=0.79, P=0.851 MSP 4E2 0.54 0.00 0.00 0.00 1.59 C=1.38, P=0.701 MSP 5 0.53 0.00 0.00 0.95 0.79 C=2.18, P=0.536 MSP 11 3.24 11.90 0.00 0.95 0.79 C=5.07, P=0.166 MSP 17 0.28 1.19 0.00 0.00 0.00 C=2.00, P=0.572 MSP 18 0.54 0.00 0.00 0.00 1.59 C=1.38, P=0.701 DPN 5 0.26 1.19 0.00 0.00 0.00 C=2.00, P=0.572 Total_mean 3.50 5.59 0.11 2.65 4.53

*Test for abundant families, F= non-parametric one-way ANOVA (df=3); For rare families, C=median test (cohort effect, df=3).

92

Figure 2.10 – Detrended Correspondence Analysis (DCA) of diapriid morphospecies classified by cohort class from boreal mixedwood sites in northeastern Ontario in 2007 with environmental variables plotted passively (only those with axis scores >0.35 shown). ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4. Only families found in >10% of sites shown. *=P<0.05, **= P<0.01. ++= P<0.01 after rsvover was selected. Variable acronyms are identified in Appendix 1 and 2.

93

Table 2.7- Partial redundancy analysis examining the relative value of various habitat predictors in explaining diapriid communities from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007 in comparison to Weibull parameters from the diameter distribution of stems >2.5 cm dbh.

Understory Canopy foliage DWD height thickness % P % % P P value var. value var. var. value Together 26.4 0.616 Together 25.7 0.649 Together 32.7 0.3094 Weibull 11.6 0.417 Weibull 11.6 0.417 Weibull 11.6 0.417 Canopy Understory 14.1 0.666 19.4 0.331 DWD 14.9 0.606 height foliage thickness Unique to Unique to Unique to 12.3 0.412 6.3 0.893 17.8 0.154 Weibull Weibull Weibull Unique to Unique to Unique to canopy 14.8 0.635 understory 14.1 0.687 21.1 0.253 DWD height foliage thickness Shared 0.7 Shared 5.3 Shared 6.2

94

Table 2.8- same as Table 2.7 but examining age, productivity and percentage of deciduous composition in comparison to Weibull parameters

(%) Age FEC productivity deciduous composition % var. P value % var. P value % var. P value Together 15.6 0.566 Together 17.6 0.428 Together 17.6 0.434 Weibull 11.6 0.417 Weibull 11.6 0.417 Weibull 11.6 0.417 Percent FEC Age 6.9 0.264 5.5 0.403 deciduous 3.5 0.719 productivity composition Unique Unique to Unique to to 8.7 0.657 12.1 0.386 14.1 0.291 Weibull Weibull Weibull Unique to Unique to Unique percent 4.0 0.642 FEC 6.0 0.362 6.0 0.361 to Age deciduous productivity composition Shared 2.9 Shared 0.5 Shared 2.5

95

250

200

150

100

50 Mean Abundance (per 100 (perdays) trap Mean Abundance

0 1 2 3 4 Cohort Class

Figure 2.11- Mean morphospecies abundance per 100 trap days for Diapriidae from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007, shown in cohort classes 1-4. Standard error bars shown. No significant effects were detected.

96

3

2.5

2

1.5

1 (per 100 (perdays) trap

Mean Shannon-Weiner Diversity Mean Shannon-Weiner 0.5

0 1 2 3 4 Cohort Class

Figure 2.12- Mean Shannon-Wiener diversity of morphospecies per 100 trap days for Diapriidae from understory Malaise traps in boreal mixedwoods of northeastern Ontario, shown in cohort classes 1-4. Standard error bars shown. No significant effects were detected.

97

30

25

20

15 richness

10

5 y = 4.9743ln(x) - 9.7721

0 0 100 200 300 400 500 600 individuals

Figure 2.13- Morphospecies richness against unstandardized abundance (individuals) for diapriid morphospecies richness from understory Malaise traps in boreal mixedwoods of northeastern Ontario in 2007, with fitted regressions shown. ●=cohort 1 □=cohort 2 ○=cohort 3 ■=cohort 4.

98

DISCUSSION

The utility of cohort classification as a coarse filter approach towards understanding variation in abundances of dipteran and hymenopteran families was not well supported. Minimal support for differences among cohort classes were shown at the family level, with only 4 of 54 dipteran families (7.4%), and 2 of 35 hymenopteran families (5.7%) showing significant relationships with cohort class. Given the 5% level of significance, these results could easily be due to chance. Cohort class was not significant for dipteran or hymenopteran communities, and there was no clear separation of cohort class in dipteran or hymenopteran family ordinations, indicating a weak relationship with the community structure. This suggests that the classification scheme does encompass all features important for these communities. Family richness of understory Hymenoptera was sensitive to cohort class. For Diptera, cohort 1 had some predictive strength in an unconstrained analysis, but only one parameter defining cohort class, Weibull scale, was significant only for understory Diptera. Diapriidae showed no associations between cohort and morphospecies; however, cohort 1 had significant predictive ability in an ordination.

Thus, it appears that relying upon the distribution of tree diameters alone without other structural features is too simple an approach to predict insect communities in the boreal forest.

Most of the habitat variables important to aerial insects could be predicted by Weibull scale or shape as found by Sharkey (2008), except for productivity; thus, it may be that including productivity in the assignment of cohort classes would improve its strength for aerial fauna.

While some insect taxa can be associated with large-scale forest changes, others may be more sensitive to changes in smaller-scale forest structures (Jokimaki et al. 1998). The development of forest canopy structures (which increase habitat diversity, resource availability and moderate microclimate) naturally associated with succession fosters less variable community structure as a

99 stand progresses (Schowalter 1995). If there is no difference in canopy communities between old-growth, mature and partial harvest stands (Schowalter 1995), then it may be that cohort classification of stands post-canopy closure has little bearing on the insect communities compared to individual structures within stands.

Diapriid communities at the morphospecies level were not significantly different among cohort classes. Before MCM classification can be ruled out as a predictor of community assemblages, we need to consider more species level data, particularly species linked to structural features important in stand development. Certainly, the approach has shown some promise for the Carabidae (with respect to scale; see Chapter 1) and for small mammal communities (Sharkey et al. 2008).

Trap height had a much stronger effect on dipteran and hymenopteran assemblages than cohort class, with distinct subsets being found in each strata. Such findings agree with those from more temperate environments where the understory supports greater insect abundance than the canopy (Lowman et al. 1993, Winchester 1997, Preisser et al. 1999). This is in contrast however to the reverse trend found in tropical insect studies (Erwin 1982, Sutton et al. 1983, Stork 1988,

Basset et al. 1992, Basset et al. 2001ab). My results correspond with those of Preisser et al.

(1999), who found Diptera and Hymenoptera to be more abundant near the forest floor than the canopy in a temperate forest. This opposes what has been seen in the tropics, likely due to the understory having more niches and favourable shaded (cool) microclimates (Lowman et al.

1993).

In contrast, I found only Hymenoptera family richness, corrected for abundance, to be higher in the canopy than the understory, a finding which agrees with Vance et al. (2007) in

100

North American temperate forest canopies. This could be attributed to the number of canopy

‘tourists’ or rare species compared to the understory where wind currents are restricted by vegetation (Sutton & Hudson 1980, Grimbacher & Stork 2007).

Comparatively, life histories of insects in more temperate and boreal environments have a greater link to the forest floor than those in the tropics, which can have life histories completely restricted to the canopy (Erwin 1983, Preisser et al. 1999). The canopy environment is a less diverse environment in boreal mixedwoods than tropical forests due to greater seasonal leaf loss, a short synchronous leaf-out of deciduous trees, few epiphytes and aerial soils, all of which limit the habitat and food resources available for insects (Preisser et al. 1999). In particular, the understory supports a substantial amount of food in the way of soil detritus and understory herbs utilized by different insects in various life history stages (Preisser et al. 1999), and the structural openness of the understory in boreal forests(i.e., less dense foliage compared to canopy) may also favour insect flight (Preisser et al. 1999).

While some temperate forest studies found no or little difference in abundances between understory and canopy (Vance 2002, Pucci 2008, Ulyshen & Hanula 2007), all have found different species compositions between the two strata. In my study, 12 (22%) dipteran families were significantly associated with the understory. Vance (2002) found a similar proportion (20%) of dipteran families associated with the understory in the Great Lakes forest region, and all five of her understory families--Cecidomyiidae, Phoridae, Rhagionidae, Sciaridae and Xylophagidae-

- were also significant in my study. In Preisser et al. (1999), Mycetophilidae, Tipulidae,

Rhagionidae, and Sarcophagidae were all found in higher abundances in the understory than the canopy, again agreeing with my findings. It is possible that some of these associations are solely by chance (using a P value of 0.05) and any family-level association with forest structures must

101 be made cautiously as there can be a high amount of variability in life history strategies within a family. Because forest understories generally have the majority of decaying or downed wood material and interface with the soil environment in a stand, I expected insect families that mainly use these structures for larval development to be more abundant in the understory than the canopy. Furthermore, I expected to see associations with families that have an affinity for shaded and damp environments, and indeed, this was the case for a few dipteran families that were found in higher numbers in the understory. Moist, shaded environments are used by

Psychodidae, Rhagionidae, Sciaridae, Mycetophilidae, and Tipulidae adults, and decaying material or soil are the larval sites for several species within the families of Rhagionidae,

Sciaridae, Dryomyzidae, and Tipulidae (Peck and Russell 1976, McAlpine et al. 1981,1987.

Priesser et al. 1999, Vance 2002, Deans et al. 2005, Triplehorn & Johnson 2005, Marshall 2006).

Downed woody debris is also used by Xylophagidae larvae, and Phoridae are commonly seen near the forest floor being most abundant around decaying vegetation and matters

(McAlpine et al. 1981. 1987. Brown 1995, Triplehorn & Johnson 2005). Some families that primarily utilize plants were also found in the understory. Anthomyiidae larvae are primarily phytophagous or saprophagous, living on stems, roots, and foliage of decaying or living plants

(McAlpine et al. 1981. 1987, Triplehorn & Johnson 2005). For example, Cecidomyiidae could be found at all strata as they are abundant gall midges with some phytophagous, mycophagous, predaceous and parasitic species; however, Vance (2002) reasons that this family was restricted to the understory as it potentially could not stand the high canopy wind. It is also known that

Cecidomyiidae can use woody debris for larval development (McAlpine et al. 1981, Vance 2002,

Marshall 2006, Dennis unpubl.) which may further support this observation. Preisser et al. (1999)

102 suggests that the presence of canopy feeders in the understory could be due to species dropping down to more open areas to travel.

A smaller set of dipteran families (8, or 14.8%) were significantly associated with the canopy. Vance (2002) found 4 (16%) dipteran families associated with the canopy in the Great

Lakes forest region. Of the dipteran families found in canopies by Vance (2002) in eastern North

American temperate forests (Calliphoridae, Ceratopogonidae, Lonchaeidae, and Sarcophagidae), only the last two were significant in my study. Preisser et al. (1999) found no dipteran families more abundant in the canopy than in the understory of their forests. In my study, Agromyzidae,

Aulacigastridae, Chamaemyiidae, Lonchaeidae, Odiniidae, Periscelidae, Sarcophagidae and

Tachinidae were all found in significantly higher numbers in the canopy. I did expect to find more herbivorous insects in the canopy because of the high foliage biomass associated with boreal canopies, and new vegetation in the such upper strata can often be the preferred food

(Basset 1991, Lowman & Wittman 1996). Agromyzidae, which are primarily leaf miners, were more abundant in the canopy likely for the same reason (McAlpine et al. 1981. 1987. Triplehorn

& Johnson 2005). Several parasitoids and predators of herbivorous insects associated with the canopy. Certain families of Chamaemyiidae are predators of aphids and scales, Tachinidae larvae are parasitoids of a range of insect hosts but primarily Lepidoptera larva, while some

Sarcophagidae larvae are phytophagous or predaceous (McAlpine et al. 1981.1987, Triplehorn &

Johnson 2005, Marshall 2006). Interestingly, I found three rare families of which species can be associated with wounds in tree bark, Aulacigastridae, Periscelidae, and Odiniidae, to have an affinity for the canopy where they may be associated with higher surface areas of coarse and fine branches than the understory (McAlpine et al. 1981, 1987, Marshall 2006).

103

Among the Hymenoptera, 7 (20%) families (Ceraphronidae, Diapriidae, Ichneumonidae,

Platygastriidae, Tenthredinidae and Vespidae) were significantly associated with the understory.

Vance et al. (2007) found only 2 (5.7%) hymenopteran families associated with the understory

(Chrysididae and Diapriidae). It is not surprising that Diapriidae was the only hymenopteran family similar in both studies given that this family parasitizes primarily dipteran larvae in the soil (Masner 1993, Pucci 2008, Dennis unpubl.). Ichneumonidae, a ubiquitous parasitoid group, was also found by Preisser et al. (1999) and Pucci (2008) to be more abundant in forest understories than canopies. Similarly, Ceraphronidae are commonly found in the soil, and this could potentially explain their affinity for the understory (Masner 1993). Tenthredinidae are large, common sawflies that lay eggs in leaf edges or stems of a variety of herbs, shrubs, and trees and thus could be found throughout the forest profile, whereas some Vespidae families nest underground and could be more abundant in the understory than the canopy (Goulet & Huber

1993, Marshall 2006).

Pompilidae was the only hymenopteran family associated with the canopy. It is possible that these results are due to chance using a P value of 0.05. Vance et al. (2004) found Pompilidae to be associated with canopy space in ordinations, although in her study it was not significant.

Interestingly, Chrysididae was associated with the understory in Vance’s study, contrary to my findings.

Different structural features were important in predicting the insect communities in each stratum. The understory fauna appear to be driven by understory structures, productivity, and the thickness of overstory, while canopy fauna are more driven by forest stand parameters (eg. stand basal area) or heterogeneity of the canopy. An understanding of what structural features are

104 important to insect fauna may help develop a more robust MCM classification method, and this would allow us to better manage insect community diversity.

Productivity, although not a structural feature per se, was the only environmental variable that was significant for both dipteran and hymenopteran communities in the understory.

Productivity was much better than Weibull parameters at predicting insect communities. The understory fauna negatively responded to fine changes in productivity, as it was somewhat controlled for through site selection (Sharkey 2008). Because this variable was also significant for small mammal communities in an associated study on the same sites (Sharkey 2008), it may be an important candidate to examine further in its relationship to MCM.

Understory Diptera were also associated with higher understory foliage thickness and lower overstory foliage thickness. Lower overstory thickness would increase the amount of light penetration, and hence presumably is reflective of near-ground microclimates and also reflects characteristics and productivity (including leaf quality) of ground layer vegetation (increasing understory foliage growth); this is important habitat and food resource for herbivores (Basset et al. 2001). As a result, more gaps in forest canopies have often been seen to be related to higher species richness and abundance (Basset et al. 2001). Heterogeneity of understory foliage thickness (variance), which was important for Diapriidae, could reflect heterogeneity of soil and near-ground microhabitats, providing diverse niches for diapriid larvae that parasitize soil (Dennis unpubl.). Diapriidae were also sensitive to the grain size of variation overstory thickness, suggesting that they would be sensitive to large versus small gaps in the canopy.

In the canopy, most dipteran families were negatively associated with heterogeneous canopy height (variance), and this contradicts the idea that heterogeneity of tree crowns promotes

105 diverse biotic communities (Lowman & Wittman 1996). It could be that stands with highly variable canopy heights represent fragmented canopy habitats from a canopy dipteran perspective. Abundances of hymenopteran families in the canopy were generally negatively influenced by high basal tree area. Jokimaki et al. (1998) found that hymenopteran and dipteran communities can be sensitive to forest stand parameters such as the number of trees within a certain size class.

Neither downed woody debris, nor percentage of deciduous composition were significant predictors of aerial insect communities in my study. This finding agrees with those found for

Carabidae and small mammals which are said to associate with downed woody debris, and suggests that DWD has not yet a become a limiting factor in these stands (Chapter 1, Cobb et al.

2007, Sharkey 2008). Not all understory insects utilize DWD in their life cycle, and this could be masking the relationship with other insects that are known to be more sensitive to DWD, such as

Mycetophilidae (Okland 1994).

While canopy insect composition can be sensitive to tree age due to increasing structural complexity (Simandl 1993, Progar & Schowalter 2002), I did not see a significant relationship with age in my study. Neither Carabidae nor small mammals abundance changed significantly with stand age in the boreal mixedwood forests I sampled, which supports the idea that it is structure that is of overwhelming importance, not age, for insect communities (Chapter 1,

Sharkey 2008). Thus, the maintenance of insect and small mammal community structure on the landscape could be overlooked if stands are being managed according only to age.

106

CONCLUSIONS AND MANAGEMENT IMPLICATIONS

Natural disturbance plays a critical role in creating structural and compositional heterogeneity on the landscape; if forest silviculture can maintain the structures and processes via natural disturbance emulation, it will sustain the ecological resilience and biodiversity of a stand (Drever et al. 2006). In Ontario’s north-eastern boreal forest, Multi-cohort Management has been developed to better address the natural disturbance regime by managing for both even- age and multi-age stands, unlike current practices (Bergeron et al. 1999). Structural variation exists between stands of different cohort classes, and it is important to understand whether the ecological communities respond to this variation in the context of managing for biodiversity

(Sharkey 2008). Cohort class was not a strong predictor of Carabidae, Diptera, Hymenoptera nor

Diapriidae communities. The MCM concept showed greater strength, however, when applied along a continuum of forest structure used to classify cohorts (live-tree diameter distribution parameters), rather than a categorical designation of classes. Forest structure as measured by stem diameter distributions (ie. the parameters used to define cohort class) was an important correlate of Carabidae communities; Weibull scale of stems >2.5 cm DBH was the strongest correlate of Carabidae community composition in boreal mixedwoods. This finding reinforces

Sharkey’s (2008) finding that managers must manage beyond the 10 cm DBH limit to effectively describe the habitat of small mammal and carabid assemblages. Looking at stands along a structural continuum (Weibull parameters), however, was still not able to predict communities for epigaeic insects in the canopy and understory; only understory dipteran communities were sensitive to Weibull scale. While cohort classification as determined by Weibull parameters of live-tree diameter distribution may be an effective coarse-filter for Carabidae, multi-cohort management may not be effective for epigaeic insects; silvicultural practices that focus on

107 creating different classes of productivity, stand level heterogeneity of understory or overstory or live-tree basal area may be more appropriate.

While multi-cohort management was originally suggested to have an age-class element

(Bergeron et al. 1999), it was understood that stands of the same age will not always have the expected structures due to the natural variability in the landscape (Harvey et al. 2002). Age since last stand-replacing disturbance was not a strong correlate of Carabidae and aerial insect community composition in either stratum compared to other forest structures, justifying the importance of structure not age in determining insect communities. Furthermore, DWD quality and quantity was not a strong predictor of insect communities; however, based on other work, one would still expect it to be important for Carabidae and other associated taxa where stands are intensively managed. As a precautionary approach, managers need to ensure that silviculture techniques suggested for multi-cohort management will maintain natural volumes of DWD in the landscape.

The results here indicate that managing for carabid biodiversity should go beyond maintaining maximum diversity and richness. Forest specialists represent only a small percentage of carabid species, but are somewhat associated with uneven-aged, heterogeneous stands, which are expected to become rare under short rotation, clear-cut silviculture. If forests were managed only for peak diversity/richness of Carabidae, clear-cut silviculture would be the focus and forest specialists might be lost. By maintaining a mosaic of differently-structured stands as proposed by multi-cohort management, even though such managed stands might be younger than natural old-growth stands, it may be possible to maintain a full range of carabid species. For example, selection cutting after one rotation in deciduous forests had few long-term effects on Carabidae (Vance and Nol 2003) and small-gap harvesting could maintain forest

108 carabid communities since smaller harvested gaps had similar communities to those found in mature stands (Klimaszewski et al. 2005). More research is needed to examine effects of multiple rotations of selection and partial cutting under boreal forest MCM to determine whether they effectively emulate natural disturbances and maintain wildlife communities. Studies of other forest taxa will be important to fully understand cohort class effect and the importance of forest structures. Furthermore, the role of emerging threats such as invasive species combined with logging should be examined, which could directly influence carabid habitat and responses to

MCM.

While cohort class was not of great importance to Diptera and Hymenoptera, their communities were significantly different between understory and canopy. In general, most families that significantly associated with understory had an affinity for understory structures.

Families that associated with the canopy were mainly considered herbivores, parasitoids of herbivores or families that use tree wounds. The communities in each stratum responded differently to forest structures, which understory communities sensitive to understory structures such as productivity (Hymenoptera and Diptera), high understory foliage thickness (Diptera) and thus low overstory foliage thickness (Diptera) and canopy communities sensitive to forest stand parameters (live-tree basal area: Hymenoptera) and canopy height heterogeneity (Diptera).

Hymenoptera in both understory and canopy seem have a positive relationship with heterogeneity of shrub foliage thickness, as heterogeneity supports diverse communities.

Because hymenopteran and dipteran communities differ between strata and are sensitive to different structural variables at different heights, forest management in the northeastern boreal must consider both strata when assessing management impacts or else responses of these two groups could be underestimated (Lowman & Wittman 1996). Stratification of insect taxa and the

109 response to cohort class could vary if insects were assessed to a higher taxonomic resolution.

While a higher-taxa approach at the family level is often used in biodiversity monitoring projects due to a lack of taxonomic expertise or its efficiency, the inherent issues with this method should be considered.

In summary, while multi-cohort management as currently defined shows some strength in predicting carabid assemblages, it fails to fully predict the community composition which is sensitive to a wide range of structural variables not adequately described by live-tree diameter distribution. A cohort classification approach that includes important habitat features such as productivity, could be considered as a way to better assess forest structure and its relation to ecological communities via Multi-cohort Management in the future.

110

REFERENCES

Akutsu, K., C. V. Khen, and M. J. Toda. 2007. Assessment of Higher Insect Taxa as Bioindicators for Different Logging-Disturbance Regimes in Lowland Tropical Rain Forest in Sabah, Malaysia. Ecological Research 22:542- 550.

Arnup, R. W. 2008. Rob Arnup Consulting. Timmins, ON. In litt. (Sharkey 2008).

Attiwill, P. M. 1994. The Disturbance of Forest Ecosystems - the Ecological Basis for Conservative Management. Forest Ecology and Management 63:247-300.

Balmford, A., M. J. B. Green, and M. G. Murray. 1996. Using Higher-Taxon Richness as a Surrogate for Species Richness .1. Regional Tests. Proceedings of the Royal Society of London Series B-Biological Sciences 263:1267- 1274.

Basset, Y. 1991. Leaf Production of an Overstory Rain-Forest Tree and Its Effects on the Temporal Distribution of Associated Insect Herbivores. Oecologia 88:211-219.

Basset, Y. 1992. Influence of Leaf Traits on the Spatial-Distribution of Arboreal Arthropods Within an Overstory Rain-Forest Tree. Ecological Entomology 17:8-16.

Basset, Y. 2001a. Communities of Insect Herbivores Foraging on Saplings Versus Mature Trees of Pourouma Bicolor (Cecropiaceae) in Panama. Oecologia 129:253-260.

Basset, Y. 2001b. Invertebrates in the Canopy of Tropical Rain Forests - How Much Do We Really Know? Plant Ecology 153:87-107.

Basset, Y., H. P. Aberlenc, H. Barrios, G. Curletti, J. M. Berenger, J. P. Vesco, P. Causse, A. Haug, A. S. Hennion, L. Lesobre, F. Marques, and R. O'meara. 2001a. Stratification and Diel Activity of Arthropods in a Lowland Rainforest in Gabon. Biological Journal of the Linnean Society 72:585-607.

Basset, Y., H. P. Aberlenc, and G. Delvare. 1992. Abundance and Stratification of Foliage Arthropods in a Lowland Rain-Forest of Cameroon. Ecological Entomology 17:310-318.

Basset, Y., E. Charles, D. S. Hammond, and V. K. Brown. 2001b. Short-Term Effects of Canopy Openness on Insect Herbivores in a Rain Forest in Guyana. Journal of Applied Ecology 38:1045-1058.

Beaudry, S., L. C. Duchesne, and B. Cote. 1997. Short-Term Effects of Three Forestry Practices on Carabid Assemblages in a Jack Pine Forest. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 27:2065-2071.

Bergeron, Y. 2004. Is Regulated Even-Aged Management the Right Strategy for the Canadian Boreal Forest? Forestry Chronicle 80:458-462.

Bergeron, Y., P. Drapeau, S. Gauthier, and N. Lecomte. 2007. Using knowledge of natural disturbances to support sustainable forest management in the northern Clay Belt. Forestry Chronicle 83:326-337.

Bergeron, Y., S. Gauthier, V. Kafka, P. Lefort, and D. Lesieur. 2001. Natural Fire Frequency for the Eastern Canadian Boreal Forest: Consequences for Sustainable Forestry. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 31:384-391.

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.

111

Bergeron, Y., B. Harvey, A. Leduc, and S. Gauthier. 1999. Forest Management Guidelines Based on Natural Disturbance Dynamics: Stand- and Forest-Level Considerations. Forestry Chronicle 75:49-54.

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.

Boucher, D., L. De Grandpre, and S. Gauthier. 2003. Development of a Stand Structure Classification Systems and Comparison of Two Lichen-Spruce Woodlands in Quebec. Forestry Chronicle 79:318-328.

Bowman, J. C., D. Sleep, G. J. Forbes, and M. Edwards. 2000. The Association of Small Mammals with Coarse Woody Debris at Log and Stand Scales. Forest Ecology and Management 129:119-124.

Brown, B.V. 1995. Review of Scuttle Flies: the Phoridae. Bulletin of the Entomological Society of Canada 27: 41- 42.

Buddle, C. M., D. W. Langor, G. R. Pohl, and J. R. Spence. 2006. Responses to Harvesting and Wildfire: Implications for Emulation of Natural Disturbance in Forest Management. Biological Conservation 128:346-357.

Bultman, T. L. and G. W. Uetz. 1982. Abundance and Community Structure of Forest Floor Spiders Following Litter Manipulation. Oecologia 55:34-41.

Carcamo, H. A. and J. R. Spence. 1994. Crop Type Effects on the Activity and Distribution of Ground Beetles (Coleoptera, Carabidae). Environmental Entomology 23:684-692.

Cobb, T. P., D. W. Langor, and J. R. Spence. 2007. Biodiversity and Multiple Disturbances: Boreal Forest Ground Beetle (Coleoptera : Carabidae) Responses to Wildfire, Harvesting, and Herbicide. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 37:1310-1323.

Deans, A. M., J. R. Malcolm, S. M. Smith, and M. I. Bellocq. 2005. Edge Effects and the Responses of Aerial Insect Assemblages to Structural-Retention Harvesting in Canadian Boreal Peatland Forests. Forest Ecology and Management 204:249-266.

Deans, A. M., S. M. Smith, J. R. Malcolm, W. J. Crins, and M. I. Bellocq. 2007. Hoverfly (Syrphidae) Communities Respond to Varying Structural Retention After Harvesting in Canadian Peatland Black Spruce Forests. Environmental Entomology 36:308-318.

Dennis, R.W.J. 2009. Woody Debris and its Importance to Saproxylic Insect Communities in Boreal Mixedwoods of Northeastern Ontario. Master of Science in Forestry Thesis, Faculty of Forestry, University of Toronto.

Desender, K., A. Ervynck, and G. Tack. 1999. Beetle Diversity and Historical Ecology of Woodlands in Flanders. Societe Royale Zoologique De Belgique 129: 139-155.

Devries, P. J., D. Murray, and R. Lande. 1997. Species Diversity in Vertical, Horizontal, and Temporal Dimensions of a Fruit-Feeding Butterfly Community in an Ecuadorian Rainforest. Biological Journal of the Linnean Society 62:343-364.

Didham, R. K., J. Ghazoul, N. E. Stork, and A. J. Davis. 1996. Insects in Fragmented Forests: a Functional Approach. Trends in Ecology & Evolution 11:255-260.

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.

Drever, C. R., G. Peterson, C. Messier, Y. Bergeron, and M. Flannigan. 2006. Can Forest Management Based

112 on Natural Disturbances Maintain Ecological Resilience? Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 36:2285-2299.

Duchesne, L. C., R. A. Lautenschlager, and F. W. Bell. 1999. Effects of Clear-Cutting and Plant Competition Control Methods on Carabid (Coleoptera : Carabidae) Assemblages in Northwestern Ontario. Environmental Monitoring and Assessment 56:87-96.

Duchesne, L. C. and R. S. McAlpine. 1993. Using Carabid Beetles (Coleoptera: Carabidae) as a Means to Investigate the Effect of Forestry Practices on Soil Diversity’, Forestry Canada, Petawawa National Forestry Institute Technical Report 16.

Erwin, T. 1982. Tropical Forests: Their Richness in Coleoptera and Other Arthropod Species. The Coleopterists Bulletin 36:74-7.

Finnamore, A. T., N. N. Winchester, and V. M. Behan-Pelletier. 1998. Protocols for Measuring Biodiversity: Arthropod Monitoring in Terrestrial Ecosystems. Biodiversity Science Board of Canada, Ecological Monitoring and Assessment, Burlington, Ontario.

Fowler, S. V. 1985. Differences in Insect Species Richness and Faunal Composition of Birch Seedlings, Saplings and Trees - the Importance of Plant Architecture. Ecological Entomology 10:159-169.

Franklin, J. F. 1993. Preserving Biodiversity - Species, Ecosystems, or Landscapes. Ecological Applications 3:202- 205.

Goulet, H. 1974. Biology and Relationships of Pterostichus adstrictus Eschscholtz and Pterostichus pensylvanicus Leconte (Coleoptera: Carabidae). Quaestiones Entomologicae 10:3-33.

Grimbacher, P. S. and N. E. Stork. 2007. Vertical Stratification of Feeding Guilds and Body Size in Beetle Assemblages From an Australian Tropical Rainforest. Austral Ecology 32:77-85.

Guillemain, M., M. Loreau, and T. Daufresne. 1997. Relationships Between the Regional Distribution of Carabid Beetles (Coleoptera, Carabidae) and the Abundance of Their Potential Prey. Acta Oecologica-International Journal of Ecology 18:465-483.

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.

Hammond, H. E., D. W. Langor, and J. R. Spence. 2001. Early Colonization of Populus Wood by Saproxylic Beetles (Coleoptera). Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 31:1175- 1183.

Harvey, B. D., A. Leduc, S. Gauthier, and Y. Bergeron. 2002. Stand-Landscape Integration in Natural Disturbance-Based Management of the Southern Boreal Forest. Forest Ecology and Management 155:369-385.

Hayden, J., J. Kerley, D. Carr, T. Kenedi, and J. Hallarn. 1995. Ontario Forest Growth and Yield Program Field Manual for Establishing and Measuring Permanent Sample Plots. Ontario Ministry of Natural Resources. Ontario Forest Research Institute. Sault Ste. Marie, ON.

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.

Hill, J. K. 1999. Butterfly Spatial Distribution and Habitat Requirements in a Tropical Forest: Impacts of Selective Logging. Journal of Applied Ecology 36:564-572.

113

Hollier, J.A., and R.D. Belshaw. 1993. Stratification and phenology of a Woodland Neuroptera assemblage. Entomologist 112:169-175.

Holliday, N. J. 1991. Species Responses of Carabid Beetles (Coleoptera, Carabidae) During Postfire Regeneration of Boreal Forest. Canadian Entomologist 123:1369-1389.

Hubbell, S. P., and R. B. Foster. 1987. Large-Scale Spatial Structure of a Neotropical Forest. Revista De Biologia Tropical 35:7-22.

Humphrey, J. W., C. Hawes, A. J. Peace, R. Ferris-Kaan, and M. R. Jukes. 1999. Relationships Between Insect Diversity and Habitat Characteristics in Plantation Forests. Forest Ecology and Management 113:11-21.

Hunter, M. L. 1990. Wildlife, Forests, and Forestry: Principles of Managing for Biological Diversity. Prentice-Hall, Englewood Cliffs, NJ.

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.

Jakel, A. and M. Roth. 2004. Conversion of Single-Layered Scots Pine Monocultures Into Close-to-Nature Mixed Hardwood Forests: Effects on Parasitoid Wasps as Pest Antagonists. European Journal of Forest Research 123:203- 212.

Johnson, E.A. 1992. Fire and Vegetation Dynamics-Studies from the North American Boreal Forest. Cambridge Studies in Ecology, Cambridge University Press, Cambridge.

Johnson, E.A. and C.E. Van Wagner. 1985. The Theory and Use of Two Fire History Models. Canadian Journal of Forest Research 15: 214-220.

Jokimaki, J., E. Huhta, J. Itamies, and P. Rahko. 1998. Distribution of Arthropods in Relation to Forest Patch Size, Edge, and Stand Characteristics. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 28:1068-1072.

Jongman, R. H. G., C. J. F. ter Braak, and O. F. R. Van Tongeren. 1995. Data Analysis in Community and Landscape Ecology. Cambridge University Press, New York. 299 pp.

Kato, M., T. Inoue, A. A. Hamid, T. Nagamitsu, M. B. Merdek, A. R. Nona, T. Itino, S. Yamane, and T. Yumoto. 1995. Seasonality and Vertical Structure of Light-Attracted Insect Communities in a Dipterocarp Forest in Sarawak. Researches on Population Ecology 37:59-79.

Klimaszewski, J., D. W. Langor, T. T. Work, J. H. E. Hammond, and K. Savard. 2008. Smaller and More Numerous Harvesting Gaps Emulate Natural Forest Disturbances: a Biodiversity Test Case Using Rove Beetles (Coleoptera, Staphylinidae). Diversity and Distributions 14:969-982.

Klimaszewski, J., D. W. Langor, T. T. Work, G. Pelletier, H. E. J. Hammond, and C. Germain. 2005. The Effects of Patch Harvesting and Site Preparation on Ground Beetles (Coleoptera, Carabidae) in Yellow Birch Dominated Forests of Southeastern Quebec. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 35:2616-2628.

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. 2003. Gap Felling as a Forest Harvesting Method in Boreal Forests: Responses of Carabid Beetles (Coleoptera, Carabidae). Ecography 26:179-187.

Kotze, D. J., J. Niemela, R. B. O'hara, and H. Turin. 2003. Testing Abundance-Range Size Relationships in

114

European Carabid Beetles (Coleoptera, Carabidae). Ecography 26:553-566.

Kuttner, B. 2006. Description, Characterization, and Identification of Stand Structure Classes in Northeastern Ontario: the Application of Multi-cohort Concepts in the Classification of Stands from Four Forest Types to Cohorts. Lake Abitibi Model Forest Technical Report No. 3, Cochrane ON.

Larochelle, A. 1976. Manuel d’Identification des Carabidae du Québec. Cordulia (Suppl. 1).

Larochelle, A., and M.C. Larivière. 2003. A Natural History of the Ground-Beetles (Coleoptera:Carabidae) of America North of Mexico. Pensoft Publishers, Sophia, Bulgaria.

Lassau, S. A., D. F. Hochuli, G. Cassis, and C. A. M. Reid. 2005. Effects of Habitat Complexity on Forest Beetle Diversity: Do Functional Groups Respond Consistently? Diversity and Distributions 11:73-82.

Le Corff, J. and R. J. Marquis. 1999. Differences Between Understorey and Canopy in Herbivore Community Composition and Leaf Quality for Two Oak Species in Missouri. Ecological Entomology 24:46-58.

Lindroth, C. H. 1961–1969. The Ground Beetles (Carabidae, excl. Cicindelinae) of Canada and Alaska, parts 1–6. Opuscula Entomologica Supplementa 20, 24, 29, 33, 34, 35, Lund, Sweden.

Loucks, O.L. 1970. Evolution of Diversity, Efficiency, and Community Stability. American Zoologist 10: 17–25.

Lowman, M., P. Taylor, and N. Block. 1993. Vertical Stratification of Small Mammals and Insects in the Canopy of a Temperate Deciduous Forest: A Reversal of Tropical Forest Distribution? Selbyana 14: 25.

Lowman, M. D. and P. K. Wittman. 1996. Forest Canopies: Methods, Hypotheses, and Future Directions. Annual Review of Ecology and Systematics 27:55-81.

Lukasiewicz J. 1996. Predation by the beetle Carabus granulatus L (Coleoptera, Carabidae) on Soil Macrofauna in Grassland on Drained Peats. Pedobiologia 40:364–376.

Malcolm, J. R. 1995. Forest Structure and the Abundance and Diversity of Neotropical Small Mammals. Pages 179-197 in M. D. Lowman and N. M. Nadkarni, editors. Forest Canopies. Academic Press, San Diego.

Marshall , S.A. 2006. Insects. Their Natural History and Diversity. With a Photographic Guide to Insects of Eastern North America. Firefly Books. Richmond Hill, ON.

Martel, J., Y. Mauffette, and S. Tousignant. 1991. Secondary Effects of Canopy Dieback - the Epigeal Carabid Fauna in Quebec Appalachian Maple Forests. Canadian Entomologist 123:851-859.

Martikainen, P., J. Siitonen, P. Punttila, L. Kaila, and J. Rauh. 2000. Species Richness of Coleoptera in Mature Managed and Old-Growth Boreal Forests in Southern Finland. Biological Conservation 94:199-209.

Masner, L. 1993. Superfamily Proctotrupoidea. In H. Goulet and J. T. Huber.1993. Hymenoptera of the World: an Identification Guide to Families. 537-557. Research Branch, Agriculture Canada, Ottawa, Canada.

McAlpine, J. F., B. V. Peterson, G. E. Shewell, H. J. Teskey, J. R. Vockeroth, and D. M. Wood. 1981, 1987. Manual of Nearctic Diptera, Vol. 1 and 2. Research Branch, Agriculture Canada, Monographs 27 and 28.

McAlpine, J.F and D.M. Wood. 1989. Manual of Nearctic Diptera, Vol. 3. Research Branch, Agriculture Canada, Monograph 32.

Migge-Kleian, S., M. A. Mclean, J. C. Maerz, and L. Heneghan. 2006. The Influence of Invasive Earthworms on Indigenous Fauna in Ecosystems Previously Uninhabited by Earthworms. Biological Invasions 8:1275-1285.

115

Moore, J. D., R. Ouimet, D. Houle, and C. Camire. 2004. Effects of Two Silvicultural Practices on Ground Beetles (Coleoptera : Carabidae) in a Northern Hardwood Forest, Quebec, Canada. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 34:959-968.

Murdoch, W. W., C. H. Peterson, and F. C. Evans. 1972. Diversity and Pattern in Plants and Insects. Ecology 53:819.

Nielsen, B. 1987. Vertical Distribution of Insect Populations in the Fee Air Space of Beech Woodland. Entomologiske Meddelelser 54 :169-178.

Niemela, J., J. R. Spence, and D. H. Spence. 1992. Habitat Associations and Seasonal Activity of Ground-Beetles (Coleoptera, Carabidae) in Central Alberta. Canadian Entomologist 124:521-540.

Niemela, J. 1997. Invertebrates and Boreal Forest Management. Conservation Biology 11:601-610.

Niemela, J. 1993. Mystery of the Missing Species - Species-Abundance Distribution of Boreal Ground-Beetles. Annales Zoologici Fennici 30:169-172.

Niemela, J., Y. Haila, and P. Punttila. 1996. The Importance of Small-Scale Heterogeneity in Boreal Forests: Variation in Diversity in Forest-Floor Invertebrates across the Succession Gradient. Ecography 19:352-368.

Niemela, J., E. Halme, and Y. Haila. 1990. Balancing Sampling Effort in Pitfall Trapping of Carabid Beetles. Entomologica Fennica 1:232-238.

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. and J. R. Spence. 1991. Distribution and Abundance of an Exotic Ground-Beetle (Carabidae) - a Test of Community Impact. Oikos 62:351-359.

Niemela, J., J. R. Spence, and H. Carcamo. 1997. Establishment and Interactions of Carabid Populations: an Experiment With Native and Introduced Species. Ecography 20:643-652.

Nguyen, T. 2000. Classification of Northern Abitibi's Spruce-Moss Forest Stands According to their Internal Structure. Preliminary report to the Quebec Ministry of Natural Resources.

Okland, B. 1994. Mycetophilidae (Diptera), an Insect Group Vulnerable to Forestry Practices - a Comparison of Clear-Cut, Managed and Seminatural Spruce Forests in Southern Norway. Biodiversity and Conservation 3:68-85.

OMNR 2001. Forest Management Guide for Natural Disturbance Pattern Emulation, Version 3.1. Ontario Ministry of Natural Resources, Queen's Printer for Ontario, Toronto.

OMNR 2009. List of Forest Management Units. OMNR website- http://www.mnr.gov.on.ca/ en/Business/Forests/ 1ColumnSubPage/STEL02_163535.html. January 15 2009.

Paquin, P. 2008. Carabid Beetle (Coleoptera : Carabidae) Diversity in the Black Spruce Succession of Eastern Canada. Biological Conservation 141:261-275.

Pearce, J. L., L. A. Venier, J. Mckee, J. Pedlar, and D. Mckenney. 2003. Influence of Habitat and Microhabitat on Carabid (Coleoptera : Carabidae) Assemblages in Four Stand Types. Canadian Entomologist 135:337-357.

Preisser, E., D.C. Smith, and M. D. Lowman. 1998. Canopy and Ground Level Insect Distribution in a Temperate

116

Forest. Selbyana 19:141–146.

Progar, R. A. and T. D. Schowalter. 2002. Canopy Arthropod Assemblages Along a Precipitation and Latitudinal Gradient Among Douglas-Fir Pseudotsuga Menziesii Forests in the Pacific Northwest of the United States. Ecography 25:129-138.

Pucci, T. 2008. A Comparison of the Parasitic Wasps (Hymenoptera) at Elevated Versus Ground Yellow Pan Traps in a Beech-Maple Forest. Journal of Hymenoptera Research 17:116-123.

Radforth, I. 1987. Bushworkers and Bosses: Logging in Northern Ontario, 1900-1980. University of Toronto Press, Toronto ; Buffalo. 336.

Rainio, J. and J. Niemela. 2003. Ground Beetles (Coleoptera : Carabidae) as Bioindicators. Biodiversity and Conservation 12:487-506.

Reemer, M. 2005. Saproxylic Hoverflies Benefit by Modern Forest Management (Diptera: Syrphidae). Journal of Insect Conservation 9:49–59.

Ribeiro, S. P. and Y. Basset. 2007. Gall-Forming and Free-Feeding Herbivory Along Vertical Gradients in a Lowland Tropical Rainforest: the Importance of Leaf Sclerophylly. Ecography 30:663-672.

Rowe, J. S. 1972. Forest Regions of Canada. Rev. edition. Dept. of the Environment. Canadian Forestry Service. Publication, no. 1300 Information Canada, Ottawa, ON. 172.

Saint-Germain, M. and Y. Mauffette. 2001. Reduced Ground Beetle Activity Following Ice Damage in Maple Stands of Southwestern Quebec. Forestry Chronicle 77:651-656.

Schaeffer, M. 1991. The animal community: Diversity and resources, In: E. Rohrig and B. Ulrich (eds.). Temperate deciduous forests. Ecosystems of the world . Elsevier, New York. .51–120.

Schowalter, T. D. 1995. Canopy Arthropod Communities in Relation to Forest Age and Alternative Harvest Practices in Western Oregon. Forest Ecology and Management 78:115-125.

Schowalter, T. D. 1989. Canopy Arthropod Community Structure and Herbivory in Old-Growth and Regenerating Forests in Western Oregon. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 19:318-322.

Schowalter, T. D. and Y. L. Zhang. 2005. Canopy Arthropod Assemblages in Four Overstory and Three Understory Plant Species in a Mixed-Conifer Old-Growth Forest in California. Forest Science 51:233-242.

Sharkey, C. A. 2008. Small Mammal Communities and Multicohort Stand Structure in Boreal Northeastern Ontario. Master of Science in Forestry Thesis, Faculty of Forestry, University of Toronto.

Simandl, J. 1993. Canopy Arthropods on Scots Pine - Influence of Season and Stand Age on Community Structure and the Position of Sawflies (Diprionidae) in the Community. Forest Ecology and Management 62:85-98.

Simila, M., J. Kouki, M. Monkkonen, and A. L. Sippola. 2002. Beetle Species Richness Along the Forest Productivity Gradient in Northern Finland. Ecography 25:42-52.

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 33:173-184.

Sobeka, S., T. Tscharntkea, C. Scherbera, S. Schielea and I. Steffan-Dewenter. 2009. Canopy vs. Understory: Does Tree Diversity Affect Bee and Wasp Communities and their Natural Enemies Across Forest Strata? Forest Ecology and Management 258:609-615.

117

Stork, N. E. 1988. Insect Diversity - Facts, Fiction and Speculation. Biological Journal of the Linnean Society 35:321-337.

Su, J. C. and S. A. Woods. 2001. Importance of Sampling Along a Vertical Gradient to Compare the Insect Fauna in Managed Forests. Environmental Entomology 30:400-408.

Sutton et al. 1983. The vertical distribution of flying insects in lowland rain-forests of Panama, Papua New Guinea and Brunei. Zoological Journal of the Linnean Society 78: 287-297.

Sutton, S. L. and P. J. Hudson. 1980. Vertical-Distribution of Small Flying Insects in the Lowland Rain-Forest of Zaire. Zoological Journal of the Linnean Society 68:111-123.

The Weather Network. 2009. www.weathernetwork.com . Historical Weather. April 9 2009.

Thiele, H. U. 1977. Carabid Beetles in their Environments: A Study on Habitat Selection by Adaptations in Physiology and Behaviour (Zoophysiology and ecology vol. 10). Springer Verlag, Berlin.

Triplehorn, C.A., and N. F. Johnson. 2005. Borror and Delong’s Introduction to the Study of Insects,. 7th Ed. Brooks/Cole, Toronto.

Ulyshen, M. D. and J. L. Hanula. 2007. A Comparison of the Beetle (Coleoptera) Fauna Captured at Two Heights Above the Ground in a North American Temperate Deciduous Forest. American Midland Naturalist 158:260-278.

Vance 2002. Canopy and Understorey Insect Communities of Sugar Maple and White Pine Forests of the South- Central Great Lakes-St. Lawrence region. Masters of Science in Forestry thesis, Faculty of Forestry, University of Toronto.

Vance, C. C., K. R. Kirby, J. R. Malcolm, and S. M. Smith. 2003. Community Composition of Longhorned Beetles (Coleoptera : Cerambycidae) in the Canopy and Understorey of Sugar Maple and White Pine Stands in South-Central Ontario. Environmental Entomology 32:1066-1074.

Vance, C. C. and E. Nol. 2003. Temporal Effects of Selection Logging on Ground Beetle Communities in Northern Hardwood Forests of Eastern Canada. Ecoscience 10:49-56.

Vance, C. C., S. M. Smith, J. R. Malcolm, J. Huber, and M. I. Bellocq. 2007. Differences Between Forest Type and Vertical Strata in the Diversity and Composition of Hymenopteran Families and Mymarid Genera in Northeastern Temperate Forests. Environmental Entomology 36:1073-1083.

Vanderwel, M. C., S. C. Mills, and J. R. Malcolm. 2009. Effects of Partial Harvesting on Vertebrate Species Associated With Late-Successional Forests in Ontario's Boreal Region. Forestry Chronicle 85:91-104.

Van Wagner, C. E. 1968. The Line Intercept Method in Forest Fuel Sampling. Forest Science 14:20-26.

Werner, S. M. and K. F. Raffa. 2000. Effects of Forest Management Practices on the Diversity of Ground- Occurring Beetles in Mixed Northern Hardwood Forests of the Great Lakes Region. Forest Ecology and Management 139:135-155.

Winchester, N. N. 1997. The Arboreal Superhighway: Arthropods and Landscape Dynamics. Canadian Entomologist 129:595-599.

Winchester, N. N. and R. A. Ring. 1996. Northern Temperate Coastal Sitka Spruce Forests With Special Emphasis on Canopies: Studying Arthropods in an Unexplored Frontier. Northwest Science 70: 94-103.

Work, T. T., C. M. Buddle, L. M. Korinus, and J. R. Spence. 2002. Pitfall Trap Size and Capture of Three Taxa of Litter-Dwelling Arthropods: Implications for Biodiversity Studies. Environmental Entomology 31:438-448.

118

Work, T. T., M. Koivula, J. Klimaszewski, D. Langor, J. Spence, J. Sweeney, and C. Hebert. 2008. Evaluation of carabid beetles as indicators of forest change in Canada. Canadian Entomologist 140: 393-414.

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. 34: 417-430.

119

APPENDICES

Appendix 1. Variable groupings, acronyms, and descriptions of habitat variables collected in boreal mixedwoods of northeastern Ontario in 2007.

Structure feature Acronym Description WEIBULL Weibull shape of all stems ≥2.5 cm w_shape_2.5 dbh Weibull scale of all stems ≥2.5 cm w_scale_2.5 dbh w_shape_10 Weibull shape of trees ≥10 cm dbh w_scale_10 Weibull scale of trees ≥10 cm dbh CANOPY HEIGHT Meancanht Mean canopy height Rvarcanht Residual variance of canopy height Residual semi-variance of Rsvcanht canopy height

OVERSTORY FOLIAGE Meanover Mean overstory foliage thickness THICKNESS Residual variance of overstory Rvarover (10-25 m level) foliage thickness Residual semi-variance of overstory Rsvover foliage thickness

UNDERSTORY Meanund Mean understory foliage thickness FOLIAGE THICKNESS Residualvariance of understory rvarund (2.5-10 m level) foliage thickness Residual semi-variance of rsvund understoryfoliage thickness

SHRUB STRATUM meanshr Mean shrub foliage thickness FOLIAGE THICKNESS Residual variance of the shrub rvarshr (0-2.5 m level) stratum foliage thickness Residual semi-variance of shrub rsvshr foliage thickness Shrub openness: # of unobstructed sections (each shrub_open 10cm high) out of 20 sections (0-2m stratum) VERTICAL FOLIAGE Shannon diversity of foliage COMPLEXITY (0-25m) vert_H' thickness in 7 height strata

120

Appendix 1- continued

DOWNED WOOD/ Volume per hectare of "new" SNAGS newDWD (decomposition clases 1 and 2) downed woody debris Volume per hectare of "old" oldDWD (decomposition classes 3, 4 and 5) downed woodydebris Basal area of snags snagBA

AGE Forest Resource Inventory (FRI) age age during 2006 survey year

COMPOSITION Measure of deciduous versus %decid coniferous tree species composition by basal area

121

Appendix 2- Species acronyms used for Carabidae species (pitfall traps), Diptera families and Hymenoptera families (understory and canopy malaise trapping) in boreal mixedwoods of northeastern Ontario in 2007.

Order Family or species name Acronym Carabidae Calosoma frigidum (Kirby) Cal_frig Sphaeroderus nitidicollis brevoorti (LeConte) Sph_niti Sphaeroderus stenostomus lecontei (Dejean) Sph_leco Agonum retractum (LeConte) Ago_retr Pterostichus pensylvanicus (LeConte) Pte_pens Carabus nemoralis (Müller) Car_nemo Scaphinotus bilobus (Say) Sca_bilo Harpalus (Euharpalops) fulvilabris Har_fulvi (Mannerheim) Calathus ingratus (Dejean) Cal_ingr Platynus decentis (Say) Pla_dece Synuchus impunctatus (Say) Syn_impu Pterostichus adstrictus (Eschscholtz) Pte_adst Pterostichus coracinus (Newman) Pte_cora Pterostichus punctatissimus(Randall) Pte_punc Pterostichus melanarius (Illiger) Pte_mela Trechus apicalis (Motschulsky) Tre_apic Harpalus somnulentus (Dejean) Har_somn Carabus granulatus granulatus (Linné) Car_gran Loricera pilicornis pilicornis (Fabricius) Lor_pili Platynus mannerheimii (Dejean) Pla_mann Harpalus sp. Har_spp. Diptera Acartophthalmidae Acar Agromyzidae Agro Anisopodidae Anis Anthomyiidae Anth Asilidae Asil Aulacigastridae Aula Bibionidae Bibi Bombyliidae Bomb Calliphoridae Call Cecidomyiidae Ceci Ceratopogonidae Cera Chamaemyiidae Cham Chaoboridae Chao Chironomidae Chir

122

Appendix 2- continued

Diptera Chloropidae Chlo (continued) Clusiidae Clus Culicidae Culi Dolichopodidae Doli Drosophilidae Dros Dryomyzidae Dryo Empididae Empi Ephydridae Ephy Heliomyzidae Heli Lauxanidae Laux Lonchaeidae Lonc Milichidae Mili Muscidae Musc Mycetophilidae Myce Odiniidae Odin Otitidae Otit Pallopteridae Pall Periscelidae Peri Phoridae Phor Pipunculidae Pipu Platypezidae Plat Psilidae Psil Psychodidae Psyc Rhagionidae Rhag Rhinophoridae Rhin Sarcophagidae Sarc Scathophagidae Scath Scatopsidae Scato Sciaridae Scia Sciomyzidae Scio Simuliidae Simu Sphaeroceridae Spha Stratiomyidae Stra Strongylophthalmyiidae Stro Syrphidae Syrp Tabanidae Taba Tachinidae Tach Tanypezidae Tanz Tipulidae Tipu

123

Appendix 2- continued

Diptera Xylophagidae Xylo Hymenoptera Ampulicidae Ampu Aphelinidae Aphe Argidae Argi Braconidae Brac Ceraphronidae Cera Chrysididae Chry Cimbicidae Cimb Colletidae Coll Crabronidae Crab Cynipidae Cyni Cynipoidea Cyni_ea Diapriidae Diap Dryinidae Dryi Embolemidae Embo Encyrtidae Ency Eulophidae Eulo Figitidae Figi Formicidae Form Halictidae Hali Heloridae Helo Ichneumonidae Ichn Megachilidae Mega Mymaridae Myma Philanthidae Phil Platygastridae Plat Platygastroidea Plat_ea Pompilidae Pomp Proctotrupidae Proc Pteromalidae Pter Scelionidae Scel Tenthredinidae Tent Trichogrammatidae Tric Vespidae Vesp Xiphydriidae Xiph

124

Appendix 3. Belytinae (Diapriidae) male morphospecies with their morphometric measurements from understory malaise trapping in Ontario’s northeastern boreal forest in 2007. Each unit is 1/40 mm. Morphospecies Morphometric measurements* MV SV BA RC CUB Pet. Length Pet. MD HW SC Ped F1 F2 Body Length Mandible M.PPD Keel RC MSP 1 11 2 7 8 12 10 4 22 10 2 10 7 90 SO NF CC MSP 1(1) 12 3 11 10 14 14 5 23 9 3 9 11 101 SO NF CC MSP 1A 10 3 7 10 12 10 6 21.5 10 2.5 10 7 94 SO NF CC MSP 1A1 8 3 11.5 8.5 8 10 8 24 10 2 10 9.5 114 SO F CC MSP 1S 8.5 2 6.5 6 9 9 4.5 20 7 2.5 7 5 79 SO NF CC MSP 1S1 3 3 9 9 4 9 4 17 9 2 6 4 64 LS NF CC MSP 1B 8 3 12 11 11 18 7.5 25 12 2 9 7 111 LS NF CC MSP 1B1 10 3 8 8 11 9 5.5 25 11 2.5 8 8 4 SO NF CC MSP 1B2 6 4 10 21 9 17 5 21 11 2 9 6 96 LS NF CC MSP 1C 7.5 4 11 16 5 9 6 21 9 2 6 5 76 SO NF CC MSP 1C1 6 3 13 16 10 12 7 22 10 2 8 6 85 LS NF CC MSP 1C2 6.5 4 9 13 4 8 5.5 24 9 2 7 8 91 SO NF CC MSP 1D 13 3 10 10 13 12 3 23 12 3 10 9 111 SO NF CC MSP 1D1 17.5 4.5 12.5 16 17 24 6 28 13 3 13 10 122 SO NF CC MSP 1D2 13 3 10 14 13 18 5 21 9 3 10 7 103 SO NF CC MSP 2 8 4 11 17 8 11 7 28 12 3 10 8 119 SO NF CC MSP 2(1) 14 4 9 12 13 13 6.5 25 15 3 10 8 119 SO NF CC MSP 2A 6 4 10 13 8 10 5 23 9 2.5 9 7 94 LS NF CC MSP 2B 6 4 11 13 4 13 6 25 12 3 8 6 96 LS NF CC MSP 2D 6 4.5 12.5 14.5 7 13 5.5 25 10 2 10 8 103 LS NF CC MSP 2D1 8.5 5 16 17 13 21 7 33 16 2 8 8 143 LS NF CC MSP 2D2 7.5 4.5 15 18 8 17 6 29 13 3 9 7 119 LS NF CC MSP 3 6 5 10 24 0 8 6 22 8 2.5 9 7 95 SO NF OP MSP 3(2) 8 5 9 23 0 8 6 21 8 3 8 6 88 SO NF OP MSP 3B 9 6 12 33 7 10 8 24 10 3 10 8 114 SO NF OP MSP 3B1 9.5 7 12 33 7 10 7 28.5 12 2 11 9 126 SO NF OP MSP 3B2 11.5 6 12 20 13 9 8 36 11 3 9 8 107 SO NF OP MSP 3D 10 7 16 42 10 13 9 28 9 2 12 12 134 SO NF OP MSP 4 9 5 13 17 10 11 7 29 9 2 9 7 116 LS NF CC MSP 4(1) 9 4 12 9 8 11 9 23 10 2 10 8 129 SO NF CC MSP 4(2) 9 4 12 16 8 13 8 29 11 3 10 7 133 SO D CC

125

MSP 4A 4 5 9 17 6 8 6 22 11 2.5 8 6 100 SO NF CC MSP 4B 6 8 12 22 7 9 7 26 12 3 8 7 118 SO NF CC MSP 4B1 3.5 4 9 15 2 7 5 17 8 2 7 5 75 SO NF CC MSP 4E 7 6 12 18 9 15 8 27 12 3 10 9 127 SO NF CC MSP 4E1 11 4 12 19 11 11 7 28 11 2 9 7 118 LS NF CC MSP 4E2 8 4 9 10 7 9 7 26 11 2 7 5 96 SO NF CC MSP 5 6 5 14 17 0 10 11 27 13 3 8 6 120 SO NF OP MSP 11 6 2 3 6 4 6 4 16 7 2 6 4 72 SO NF CC MSP 14 9 4 12 11 8 10 7 25 10 3 10 9 123 SO F CC MSP 17 4 4 11 13 5 10 9 25 12 3 9 7 115 SO NF CC MSP 18 7 3 10 20 6 9 7 22 12 3 8 7 109 SO F OP * MV: Medialis, SV: Stigmalis, BA: Basalis, RC: Redial Cell, CUB: Cubitalis, Pet: Petiole, MD: Median Diameter, HW: Head Width, SC: Scape, Ped: Pedicel, F1: Firsr Flagellomere, F2: Second Flagellomere, M.PPD K: Median Propodeal Keel, NF: Not Forked, F: Forked, D: Doubled, SO: Short Ordinary, LS: Long sickle shaped, OP: Open at apex, CC: Completely closed.