Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
The Effect of Biotic and Abiotic Forces on Species Richness Peter J.T. White Faculty of Science, Department of Biology McGill University Montréal, Québec, Canada A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Doctor of Philosophy © Peter J.T. White, 2011 1 TABLE OF CONTENTS Table of Contents………………………………………………………………………………………………..….. 2 List of Tables………………………………………………………………………….….……………………………. 5 List of Figures……………………………………………………………………………………….…….…….…….. 8 List of Appendices…………………………………………………………………………………………………….13 Preface………………………………………………………………………………………………………………….… 14 Thesis Format and Style…………………………………………………………………………………… 14 Contribution of Co-Authors……………………………………………………………………………… 15 Original Contributions to Knowledge………………………………………………………………..17 References………………………………………………………………………………………………………. 21 Acknowledgements…………………………………………………………………………………………. 23 Thesis Abstract………………………………………………………………………………………………………… 26 Résumé…………………………………………………………………………………………………………………….28 General Introduction………………………………………………………………………………………………..30 References……………………………………………………………………………………………….……… 45 Chapter 1: Detecting Changes in Forest Floor Habitat after Canopy Disturbances…… 53 Abstract……………………..……………………………………………………………………………………. 54 Introduction…………………………………………………………………………………………………….. 55 Local Consequences of Damage………………………………………………………………..55 Landscape and Regional Investigations……………………………………………….…… 56 Habitat Implications of Ice Storms………………………………………………………….…56 Using Remote Sensing to Measure Habitat Change…………………….…………… 58 Objective………………………………………………………………………………….……………… 59 Method……………………………………………………………………………………………….…………… 59 Study Area………………………………………………………………………………………..……… 59 Satellite Imagery Processing……………………………………………………………………..60 VI Calculation…………………………………………………………………………………………… 61 Predicting the Spatial Pattern of CWD Influx……………………………………………. 63 Independent Validation of CWD Predictions…………………………………….……… 64 Methods Evaluation……………………………………………………………………………….… 65 Categorizing CWD Habitat…………………………………………………………………………66 Results………………………………………………………………………………………………………………67 Performance of VIs for Predicting CWD Influx……………………………………..…… 67 Independent Validation of CWD Predictions………………………………….………… 68 Categorization of CWD Influx into Different Habitat Types………………….…… 68 Methods Evaluation………………………………………………………………………….……… 69 Discussion……………………………………………………………………………………………………….. 69 Accuracy of VIs in Predicting Spatial Pattern of Damage…………………..……… 69 CWD Benefits for Mont St Hilaire Species…………………………………………..….… 71 The Benefits of Different Habitat Types for Insects……………………………..…… 72 CWD Influx and Species Movements……………………………………………………..… 74 Conclusion………………………………….……………………………………………………………. 76 Acknowledgements……………………………………………………………….………………………… 77 References………………………………………………………………………………………………………. 78 Figures………………………………………………………………………………………………………………89 Tables……………………………………………………………………………………………………………… 92 2 Linking Statement 1………………………………………………………………………………….………..…… 96 Chapter 2: Human-Disturbance and Caterpillars in Managed Forest Fragments.….…. 97 Abstract……………………………………………………………………………………………………….….. 98 Introduction……………………………………………………………………………………………….……. 99 Trailside Habitat in Forests………………………………………………………………………. 99 An Analogy of Forest Edge Habitat…………………………………………………………... 101 Hypotheses……………………………………………………………….….…….….….….………… 102 Method……………………………………………………………………………………………………………. 103 Study Area……………………………………………………………………………………………….. 103 Trail Index Calculation……………………………………………………………………………… 106 Caterpillar Surveying and Identification…………………………………………………… 107 Site Host Plant Availability……………………………………………………………………….. 108 Statistical Analysis……………………………………………………………………………………. 109 Results…………………………………………………………………………………………………………….. 110 Trail Index and Host Plant Availability……………………………………………………… 110 Random Versus Non-Random Distribution of Caterpillar Species…………….. 111 Discussion……………………………………………………………………………………………………….. 112 Possible Mechanisms of Negative Relationship……………………………………….. 112 Trails Versus Edges…………………………………………………………………………………… 114 Conclusion……………………………………………………………………………………………….. 116 Acknowledgements………………………………………………………………………………….……… 117 References………………………………………………………………………………………………………. 118 Figures……………………………………………………………………………………………………….……. 131 Tables……………………………………………………………………………………………………………… 134 Linking Statement 2………………………………………………………………………………………………… 136 Chapter 3: Testing Two Methods that Relate Herbivorous Insects to Host Plants….... 137 Abstract…………………………………………………………………………………………………………… 138 Introduction…………………………………………………………………………………………………….. 140 Overlooked in Conservation Planning…………………………………………………..…..140 The Relationship between Hosts and Lepidoptera………………………………..…. 141 Methods………………………………………………………………………………..….….….….…....…… 144 Study Area…………………………………………………………………………..…………………… 144 Caterpillar Survey and Identification………………………………………………………… 146 Controlling for Habitat Disturbance..…………………………………………………..…… 147 Analyses…………………………………………………………………………………………………… 147 Host Plant and Caterpillar Relationships..……………………………….…….….….….. 149 Testing Host Plant-Specific Preferences………………………………….…….….….….. 149 Results………………………………………………………………………………………………………….…. 150 Caterpillar Sampling…………………………………………………………………………………. 150 Host Plant and Caterpillar Relationships………………………………………………….. 151 CAPIr and CAPIa…………..………………………………………………………………………….. 151 Host Plant Importance Relative to Trail Index………………………………………….. 152 Discussion……………………………………………………………………………………………………..… 153 Biodiversity and Conservation….….….…………………………………………………….… 153 Low Richness and Abundance in Invasive Trees……………………………………….. 154 Mechanisms…..….….….….….….….……………………………………………….…..…..……..157 Conclusion……………………………………………………………………………………………..… 159 3 Acknowledgements……………………………………………………………………………………..….. 161 References………………………………………………………………………………………………….…… 162 Figures………………………………………………………………………………………………………..…… 172 Tables………………………………………………………………………………………………………….…… 177 Linking Statement 3………………………………………………………………………………………………… 181 Chapter 4: Intra-Seasonal Relationships between Insect Herbivores and their Hosts. 182 Abstract……………………………………………………………………………………………………….….. 183 Introduction………………………………………………………………………………………………….…. 184 Bottom-Up Effects: Foliar Quality.….….……………………….…….….…………………. 184 Bottom-Up Effects: Foliar Toxins.….….….….….….….….….….………………………… 186 Top-Down Effects.….….….….….….….….….….….….….….….….….….….…………….… 187 Current Gaps in Foliar Quality Research.….….….………….…….….…………………. 188 Tri-Trophic Relationships……………….….….….….….….….….…….….…………………. 191 Objective…………………………………………………………………………………………………. 192 Methods……………………………………………………………………………………………………….…. 193 Study Area…………………………………………………………………………………………….…. 193 Caterpillar Survey and Identification………………………………………………………...195 Measuring the Bottom-Up Effect…………………………..…..….……………….…..…... 195 Measuring the Top-Down Effect…………………………….….….……………….…..…... 196 Analysis.….….….….….….….….….….….….………………….……………..….……….…..…... 198 Methodological Assumptions…………………………………………………………………... 199 Results……………………………………………………………………………………………….……………. 200 Bottom-Up Foliar Quality……………………….………………………………………………… 201 Top-Down Pressure…………………………………………………………………………………. 201 Seasonality of Bottom-Up Effects.….….……………………………………………………..201 Tri-Trophic Relationships.….….….….….…………………….….….……………….…..…... 203 Discussion…………………………………………………………………………………………….….….….. 204 Bottom-Up: Foliar Quality………………….…………………………………….…….…………204 Bottom-Up: Toxins……………………………..….….……………………………….…………... 207 Top Down: Parasitoids…………………………………………………………………….…....… 208 Conclusion…………………………………………………………………………………….……….… 210 Acknowledgements……………………………………………………………………………………….… 211 References…………………………………………………………………………………………………….… 212 Figures…………………………………………………………………………………………………………….. 223 Tables……………………………………………………………………………………………………………… 228 General Conclusion……………………………………………………………………………………………….… 232 References……………………………………………………………………………………………………………... 237 Appendices……………………………………………………………………………………………………………… 239 4 LIST OF TABLES CHAPTER 1: Table 1 Landsat-5 satellite imagery used in calculation of vegetation indices for Mont St. Hilaire in 1996 and 1998. Table 2 Vegetation Indices (VIs) used to calculate forest damage after the 1998 ice storm at Mont St. Hilaire. The VIs incorporated four wavelengths of light: blue (450 – 520 nm), green (520 – 600 nm), red (630 – 690 nm) and near infrared (NIR, 760 – 900 nm). Table 3 The data for each regression model were split into random training and validation datasets (80/20 ratio, 66/16 plots) to create ten cross validation trials for each Vegetation Index. The R2 fit of each validation dataset is given respective of the regression coefficient and intercept calculated in its training dataset. NDVI provides the best for both training and validation datasets (R2 = 0.20 and 0.37, respectively). CHAPTER 2: Table 1 Trail Index is a consistent negative predictor of caterpillar abundance and caterpillar species richness both across and within study sites.Host plant availability is rarely significantly linked to caterpillar species abundance or richness, only explaining a large degree of caterpillar variance at Mont Shefford. Near-significant (*) and significant (**) p-values are marked. 5 CHAPTER 3: Table 1The relationship between host plants and caterpillars shows that (a) host plant (Shannon’s) diversity is a significant descriptor of caterpillar (Shannon’s) diversity when Trail disturbance is accounted for (total model adjusted R2 = 0.45F3,68 = 9.9) but (b)host plant richness is a non-significant descriptor of caterpillar richness when Trail disturbance is accounted for (total model adjusted R2 = 0.27, F3,68 = 20.1). Table 2CAPIa and CAPIr values for host plant trees (sorted in order of decreasing CAPIr values) are calculated as the difference between the observed and the average caterpillar abundances and richness in host plant tree species (see equations 1 and 2). Greater CAPI values indicate that a host plant is more preferred by the caterpillar assemblage. Table 3Host plant frequencies (combined) explained (a) 21.0% of caterpillar abundance, (b) 24.9% of caterpillar richness and (c) 27.9% of caterpillar Shannon’s diversity among quadrats. This was independent of trail index which, when combined with a F. americana interaction term, explained (a) 19.8%, (b) 22.3% and (c) 18.7% of the variances. CHAPTER 4: Table 1 Foliar nutrient properties in broadleaf trees in southern Quebec and Ontario in early June, early July and early August. Foliar nutrient data 6 are averaged across 34 different common deciduous tree species. This table is adapted from Ricklefs and Matthew’s (1982), Table 3. Table 2 A list of trees sampled (where n > 10) across the 72 study sites, and their associated caterpillar richness and abundance. Table 3Parasitoid pressure on host plants in August. Table 4 Emergence records for parasitoids in Ontario and Quebec documented by Natural Resources Canada, 1937-1949. 7 LIST OF FIGURES CHAPTER 1: Figure 1 Coarse woody debris influx at Mont St. Hilaire resulting from the 1998 ice storm. Mean coarse woody debris influx was 1.7 kg/m2 (standard deviation = 0.97 kg/m2), for a total estimated input of 16.8 metric tons per hectare. Coarse woody debris input values are spatially correlated up to 450 meters. Figure 2 The predicted biomass of coarse woody debris resulting from the 1998 ice storm is positively correlated to the amount of young (category 1 and 2) coarse woody debris that was measured in the summer of 2008 at 18 randomly located sites across Mont St. Hilaire. Figure 3 Concentrations of coarse woody debris at Mont St. Hilaire in (a) dry shaded habitat, (b), moist shaded habitat, (c) sun-exposed habitat and (d) wet habitat. CHAPTER 2: Figure 1 We sampled at four sites in the St. Lawrence River valley of southern Quebec, Canada (Figure adapted from Atlas of Canada 2010), each progressively farther from the center of Montreal (dashed bound). Urban development is highest in the western sites, giving way to farmlands eastward in the St-Lawrence Lowland, and then extensive forests in the Appalachian Highlands 8 Figure 2 Trail index calculation for a given pixel p at distance d from trail t that has a width w. Index shown untransformed (a) and logtransformed (b). Figure 3 Trail index across our four study sites: (A) Mont Royal, (B) Mont St. Bruno, (C) Mont St. Hilaire and (D1, D2, D3) Mont Shefford. The geospatial arrangement in this Figure does not reflect the regional geolocations of the sites (see Figure 1). CHAPTER 3: Figure 1 Caterpillars were collected from four sites in the St. Lawrence River valley of southern Quebec, Canada (Figure adapted from Atlas of Canada 2010). The matrix surrounding each site isdominated by agricultural lands and urban development with the exception of Mont Royal, which is a forest fragment in an exclusively urban setting. Figure2There was no relationship between the number of caterpillars reported to use a given host plant and either (a) CAPIr or (b) CAPIa scores. These relationships are expected to be positive as a host plant’s acceptability should be indicative of the caterpillar assemblage preference of that host plant relative to other host plants in the community. Figure3 A host plant replacement simulation for (a) caterpillar species richness and (b) caterpillar richness in the Mont Royal forest fragment. In these simulations F. grandifolia and A. pensylvanicum 9 were substituted for A. platanoides and R. cathartica (dashed line) and O. virginiana was substituted for F. americana (dotted line). The substituted species were chosen because they had high CAPIr and/or CAPIa scores and commonly share the same general canopy position as the species they replace. Replacement of invasive species with O. virginiana and A. pensylvanicum resulted in an increase of 30% in caterpillar species richness and 40% in caterpillar abundance. Replacement of F. americana with F. grandifolia resulted in an increase of 18% in caterpillar species richness and 37% in caterpillar abundance. CHAPTER 4: Figure 1 Caterpillars were collected from four sites in the St. Lawrence River valley of southern Quebec, Canada (Figure adapted from Atlas of Canada 2010) at the northern edge of the deciduous forest biome in eastern North America. The matrix surrounding each site isdominated by agricultural lands and urban development with the exception of Mont Royal, which is a forest fragment in an exclusively urban setting. Forest patches across the region are shown in dark gray, urban areas in light gray (including the City of Montreal at the left side of the pane). Figure 2 Average foliar qualities at quadrats in the months of June (Je), July (Jy) and August (Au) for (a) % water content (b) % nitrogen content, 10 (c) % phosphorus content, (d) toughness (grams), (e) % fiber content, (f) % lignin content and (g) % polyphenol content. Bars represent standard error. ANOVAs between months are significant for all foliar qualities at p < 0.001. Average foliar qualities for each quadrat were weighted based on the proportion of the total basal area occupied by each sampled host plant species in the quadrat. Host plant-specific foliar qualities were taken from Ricklefs and Matthew (1982). Figure 3Regression tree analyses of the determinants of caterpillar richness (left hand panels) and abundance (right panels) at the quadrat level. Analyses are shown for June (a, b), July (c, d) and August (e, f). The variables shown are % Phosphorus content (P), % Polyphenol content (Phenol),% Fiber content (Fiber) and % Lignin content (Lig). Water content, Nitrogen content and Toughness were included, but were not significant. The clause presented at each node is the condition corresponding to the left hand fork (the right hand fork would be the opposite condition). Each clause is paired with an R2 value associated with that node in brackets; this value is equal to the complexity parameter of the node. At each terminus the average caterpillar richness or abundance is given (depending on the tree) along with the number of quadrats that satisfy the conditions of the 11 fork (in brackets). Trees are pruned and show splits corresponding to R2> 0.05. Figure 4 Regression tree analyses for the determinants of caterpillar richness (a) and abundance (b) for the month of August. These trees were created using the same data as for Figure 3e and 3f (respectively) but with the addition of parasitoid pressure data for each quadrat. In (a) parasitoid pressure supersedes % fiber content as the most important determinant of caterpillar richness. In (b) parasitoid pressure is less important and supersedes % polyphenol content as the primary determinant of caterpillar abundance. 12 APPENDICES CHAPTER 2: Appendix A1 We surveyed 36 macrolepidopteran moth species across the four sites in our study region Mount Royal (R), Mont St. Bruno (B), Mont St. Hilaire (H) and Mont Shefford (S). All species IDs were based on 5th or 6th instar larvae identified using Wagner (2005). Appendix A2 Micromoth distribution. Appendix A3 Complete list of host plant species documented in vegetation surveys. Appendix A4Randomness was tested using randomization goodness-of-fit tests. P-values were calculated using 10,000 replicates of randomization. CHAPTER 3: AppendixB1 Caterpillar collections were made from 38 host plant trees in 72 quadrats across the four study sites. AppendixB2 A record of the macrolepidoptera and microlepidoptera morphospecies that were surveyed. Microlepidoptera morphospecies were given unique alphanumeric designations and subsequent individuals were verified with digital images. All macrolepidoptera were identified using Wagner (2005) and Handfield (1999). 13 PREFACE Thesis Format and Style This thesis is in a manuscript-based format, consisting of a set of four papers. The first chapter explores the geospatial pattern of damage and biological implications of the 1998 ice storm on an old-growth forest. The second, third and fourth chapters are focused on how disturbance and habitat quality impact caterpillar assemblage richness and abundance. All four chapters focus on forest fragments found in the same area of the St. Lawrence floodplain in southern Quebec (i.e. the Monteregié). This thesis was initially intended to have a remotesensing and GIS theme in all four chapters but after NSERC funding opportunities became available in the spring of 2008 the thesis evolved to centre on lepidopteran ecology and extensive field work was conducted in 2008 and 2009. Thus, the first chapter has a different focus than the final three chapters that comprise a cohesive unit. The first chapter, however, does link with the second chapter through the themes of natural disturbances (Chapter 1) and humanbased disturbances (Chapter 2) in forest fragment communities. 14 Contributions of Co-Authors Each of the four chapters was prepared as a manuscript for publication in a peer reviewed journal, generally focusing on biological reserve management for the conservation of biological diversity. I received valuable guidance and intellectual contributions from my supervisors B. McGill and M.J. Lechowicz. I was, however, the primary person designing the research questions, organizing and conducting the data collection, computing statistical analyses and writing the research for each chapter. I had two research assistants – R. MacKenzie and M. VonButtlar who helped me collect the woody debris, vegetation and Lepidoptera data used throughout this thesis. Chapter 1: M.J. Lechowicz and B. McGill were both valuable intellectual contributors. M.J. Lechowicz provided several rounds of editorial comments and suggestions as the manuscript was in varying stages of preparation. B. McGill provided very valuable insights and advice for the statistical procedures that were used. They both played key roles in sharpening my research focus and suggesting that the research focus on biotic implications. Prior coarse woody debris collections by M. Hooper in 1998were also instrumental in allowing a geospatial model to be computed. Chapter 2: M.J. Lechowicz and B. McGill played similar roles as in Chapter 1. 15 Chapter 3: This will be a single-authored manuscript. M.J. Lechowicz provided comments on the manuscript and encouraged me to think about the implications of my findings, particularly encouraging me to explore a post hoc invasive-species replacement exercise. B. McGill was an invaluable consultanthelping me construct the conceptual arguments I use in this chapter. He also provided several rounds of review before it was finalized. Chapter 4: This will also bea single-authored manuscript. B. McGill played a similar role as in chapter 3. He also encouraged me to use many of the statistical methods that were chosen. 16 Original Contributions to Knowledge Chapter 1: There has been much research on the impacts of the 1998 ice storm (and similar storms) on broadleaved forests. The bulk of this research has explored either the geophysical correlates to damage or the local biotic implications. For example, there is a wide array of studies that have mapped out the pattern of ice storm damage (King et al. 2005, Millward and Kraft 2004) or explored the topographic correlates to damage (Vandyke 1999) at a landscape orregional scale. These types of studies are valuable, but do not allow us to understand the large-scale biotic impacts resulting from amass coarse woody debris influx. Thus, the first original contribution to knowledge is the geospatial mapping of different types of coarse woody debris habitat (which support different biotic communities) across a landscape using freely available remote sensing imagery. Second, it has often been assumed that the Normalized Difference Vegetation Index (NDVI) is the best index to use in ecological studies to measure changes in forest canopy cover but tests of this assumption are rare. I provide empirical evidence to support this claim by showing that NDVI outperforms six other popular remotely sensed vegetation indices in a cross validation analyses predicting coarse woody debris influx resulting from canopy damage. 17 Chapter 2: The result of human disturbances on forest communities is an important conservation concern. One of the primary ways that humans impact natural communities is through the presence of recreational hiking trails. The impact of such trails has been explored for birds (Miller and Hobbs 2000), small mammals (Meaney et al. 2002) and ground-dwelling beetles (Grandchamp et al. 2000) (among other taxa), but it has never been considered for forest-dwelling caterpillar assemblages. In this chapter I show that there is a consistent and negative relationship between recreational trail presence and caterpillar assemblage richness and abundance. Furthermore, to accomplish this analysis I develop a simple index that quantifies the effect of trails on forest habitat as a function of trail width and trail proximity. This is a simple and effective tool that can be readily applied to other forest conservation areas to quantify a geospatial index of trail impact. The importance of this research extends to forest reserve management where trails are generally viewed as positive management tools because they direct and control the flow of pedestrian traffic. While the research in this chapter certainly does not suggest the elimination of trails, it suggests that trails need to be limited and managed to maximize the richness and abundance of forest-dwelling caterpillar assemblages. 18 Chapter 3: Insect herbivores have been linked to their host plants in two different ways depending on whether the unit of examination is an entire insect herbivore assemblageor an individualinsect herbivore species. Insect herbivore assemblage richness is often tied to host plant richness under the premise that a species-rich host plant assemblage provides a wider array of food and structural resources than a species-poor one (Haddad et al. 2001, Summerville and Crist 2004). This is contrasted by studies of individual insect herbivore species that often tie insect herbivores to host plants based on plant identity – some host plants are highly preferred while others are not (e.g.Delisle and Hardy 1997, Liebhold et al. 1995).Consumption of preferred host plant foliage often results in greater fecundity and developmental gains. It is unknown how host plant preferences at the individual species level scales up to an entire insect herbivore assemblage. In this chapter I test between these two relationships that posit: (1) insect herbivore assemblage richness is driven by host plant richness and (2) insect herbivore assemblage richness is driven by the presence of highly preferred host plants.I demonstrate that the latter is a much stronger descriptor of insect herbivore richness than the former. In addition, I showed thatinvasive host plants are strongly negatively linked to insect herbivore assemblage richness and abundance. A host plant replacement simulation showed that insect herbivore assemblage richness and abundance could increase by 30-40% if invasive host plants were replaced with functionally comparable native host plants. This 19 provided an empirical basis for continued and stronger forest management to eliminate and restrict the establishment of invasive plants. Chapter 4: Broadly speaking, there are two theories to explain the distribution of caterpillar species among host tree species. The first is that caterpillar species are distributed with respect to their foliar nutrient needs; the second is that caterpillar species are distributed so as to minimize their chances of coming into contact with their parasitoid enemies (Lill 2001). Certain caterpillar species have also been documented to feed on foliage high in polyphenol content to sequester toxins as a defense against parasitism (Turlings et al. 1995). Previous research has documented that a change in distribution strategies in certain caterpillar species may occur corresponding to the phenology of parasitoids (Lill et al. 2002). However these two pressures (bottom-up and top-down) have only rarely been considered in a seasonal context and almost never in the context of an entire caterpillar assemblage. My research in this chapter indicated that caterpillars are distributed amongst host plants with respect to foliar quality early in the season but that late in the season parasitoid pressure supersedes the importance of foliar qualityas an explanation for caterpillar richness and abundance patterns. In the caterpillar assemblage I examined, there was no evidence to suggest that polyphenol sequestration plays a large role in impacting caterpillar host plant choices. 20 References Delisle, J. and Hardy, M. 1997. Male larval nutrition influences the reproductive success of both sexes of the Spruce Budworm, Choristoneura fumiferana (Lepidoptera: Tortricidae). - Functional Ecology 11: 451-463. Grandchamp, A. C., Niemelä, J. and Kotze, J. 2000. The effects of trampling on assemblages of ground beetles (Coleoptera, Carabidae) in urban forests in Helsinki, Finland. - Urban Ecosystems 4: 321-332. Haddad, N. M., Tilman, D., Haarstad, J., Ritchie, M. and Knops, J. M. N. 2001. Contrasting effects of plant richness and composition on insect communities: a field experiment. – The American Naturalist 158: 17-35. King, D. J., Olthof, I., Pellikka, P. K. E., Seed, E. D. and Butson, C. 2005. Modelling and mapping damage to forests from an ice storm using remote sensing and environmental data. - Natural Hazards 35: 321-342. Liebhold, A. M., Gottschalk, K. W., Muzika, R., Montgomery, M. E., Young, R., O'Day, K. and Kelley, B. 1995. Suitability of North American Tree Species to the Gypsy Moth: A Summary of Field and Laboratory Tests. United States Department of Agriculture, Northeastern Forest Experiment Station, Radnor, PA Lill, J. T. 2001. Selection on herbivore life-history traits by the first and third trophic levels: the devil and the deep blue sea revisited. - Evolution 55: 2236-2247. 21 Lill, J. T., Marquis, R. J. and Ricklefs, R. E. 2002. Host plants influence parasitism of forest caterpillars. - Nature 417: 170-173. Meaney, C. A., Ruggles, A. K., Clippinger, N. W. and Lubow, B. C. 2002. The impact of recreational trails and grazing on small mammals in the Colorado Piedmont. - Prairie Naturalist 34: 115-136. Miller, J. R. and Hobbs, N. T. 2000. Recreational trails, human activity, and nest predation in lowland riparian areas. - Landscape and Urban Planning 50: 227-236. Millward, A. A. and Kraft, C. E. 2004. Physical influences of landscape on a largeextent ecological disturbance: the northeastern North American ice storm of 1998. - Landscape Ecology 19: 99-111. Summerville, K. S. and Crist, T.O. 2004. Forest moth taxa as indicators of lepidopteran richness and habitat disturbance: a preliminary assessment. - Biological Conservation 116: 9-18. Turlings, T. C., Loughrin, J. H., McCall, P. J., Röse, U. S., Lewis, W. J. and Tumlinson, J. H. 1995. How caterpillar-damaged plants protect themselves by attracting parasitic wasps. - Proceedings of the National Academy of Sciences 92: 4169-4174. Vandyke, O. 1999. A literature review of ice storm impacts on forests in Eastern North America. SCSS Technical Report #112. - Ontario Ministry of Natural Resources, Southcentral Sciences Section, pp. 1-29. 22 Acknowledgements My wife Blythe has stood by me throughout this journey and it would have been immeasurably more difficult for me to complete this thesis without her loving and unwavering support. She has brought such a fantastic richness to my life and has made our marriage a source of refreshment and rest. Over the past five years this thesis has been a major story that has been written across the pages of my life; Blythe has been the one to fill those pages with colour. MENTORS Throughout this thesis journey I have had had two invaluable mentors: Dr. Marty Lechowicz and Dr. Brian McGill. I am appreciative of all of the time and effort they have spent on me. I am very thankful that they chose to invest in me and I hope that I can live up to their expectations. EDUCATORS I would like to the two educators who inspired me to become a scientist. Robert Cassibo was my grade high school science teacher. He made science come alive in a way that forever changed my perception of the world in which we live. Without his influence I would not have become a scientist. Dr. Jeff Houlahan was my undergraduate ecology professor at the University of Ottawa. Through Jeff’s teaching, ecology became a relevant science and I developed a deep interest in the study of organisms and their 23 environments. Without his influence I certainly would not have pursued graduate-level ecology research. ENCOURAGERS There were many people who encouraged and motivated me over the course of my PhD. There are a few who merit special mention. (1) Richard Feldman and I started our PhDs in the McGill lab at the same time. He has been an academic brother to me encouraging me when I’ve been frustrated and challenging me to be a better student and scientist. (2) David Syncox motivates people to achieve great things. He has been a wonderful friend and has gone out of his way on innumerable occasions to be a listening ear for my frustrations and to encourage me towards reaching the goal of thesis completion. (3) I will also mention the McGill Lab Group who provided many excellent reviews and encouragement for my research: Sergio Estrada, Julie Messier and John Donoghue. FAMILY My family has been very supportive of my decision to pursue a PhD. My mother Linda often called asking “how things are going” and has been a constant reminder of how much I am loved by my family. My sister Mags has also very quick to read my papers and has been full of support. I am also thankful for the rest of my wonderful familyandthe encouragements they provided along the way – specifically my son Quentin, my dad Richard, my brother Andrew, my 24 sister-in-law Bethany. my brother-in-law Mike, my father-in-law Gary, my mother-in-law Beth, my brother-in-law Andy, my sister-in-law Steffani, my godmother Sharon, my godfather Bob and my best friend Todd and his wife Krista. FINANCIAL SUPPORT There were several sources of funding that allowed me to complete this degree. Thank you to Dr. Brian McGill, Dr. Martin Lechowicz and Dr. Richard Tomlinson via Dr. Brian Alters at the Tomlinson Project in University Level Science Education for your funding and support. Thanks also to the National Sciences and Engineering Research Council of Canada for two years of support. COLLEAGUES AND FRIENDS Finally, I have had many colleagues and friends at McGill University and at Emmaus Anglican Church to whom I owe a debt of gratitude for their friendship, encouragement and intellectual contributions. I dare not try to list them all in fear that I may leave some out. Thanks to all. 25 Thesis Abstract One central question in ecology is why some areas have many species and others have few. Many explanations have been proposed and often theforcesthat drive species richness are context-dependent. These forces aredivided into two general categories: biotic drivers and abiotic drivers. Biotic drivers are most commonly described in terms as top-down and bottom-up effects while abiotic drivers are commonly described in terms of climate and habitat disturbance. The objective of this thesis is to determine how these drivers affect species richness in terrestrial ecosystems. To test this I examine an insect herbivore assemblage in a disturbed forest fragment landscape in southern Canada.I use geographic information systems techniques to determine the impact of a natural episodic disturbance (i.e. an abiotic natural driver) and a chronic human disturbance (i.e. an abiotic human driver) on forest habitat quality (Chapter 1) and on a forestdwelling caterpillar assemblage (Chapter 2). I show that ice stormsresult in a heterogeneous pattern of spatial damage across a forest landscape, differing depending on the type of coarse woody debris examined. These different types of coarse woody debris provide habitat for a diversity of taxa. In contrast with natural disturbance, I found that human-based disturbance do not have a positive impact on caterpillar assemblages.Pursuant to this, I explore the concept of habitat quality from the perspective of host plant identity (Chapter 3) and host plant quality (Chapter 4). I found that caterpillar assemblages have strong host plant preferences and that these preferences may depend on quadrat-scale 26 foliar qualities (i.e. a biotic bottom-up driver)and parasitoid densities (i.e. a biotic top-down driver) at different times in the growing season. This thesis adds to a growing body of literature aimed to better understand the drivers of insect species richnessacross disturbed landscapes. In addition, this thesis develops several management-specific tools for measuring forest disturbance and provides valuable insight into how the selection of different tree species for planting initiatives can have important impacts on forest communities. Résumé 27 Les communautés forestières qui habitent les parcelles de forêts qui subsistent aujourd'hui sont affectées par les perturbations ainsi que par la qualité de l'habitat que leur procure les plantes-hôtes. Ces deux phénomènes ont un impact particulièrement important dans les paysages modifiés par l'activité humaine. Développer une meilleure compréhension de ces phénomènes va faciliter la prise de décision et les efforts de conservations visant à préserver et protéger la biodiversité des forêts. L’objectif global de cette thèse est d’étudier les divers aspects reliés à la qualité de l’habitat dans les parcelles forestières des collines montérégiennes du sud-est du Québec, Canada. J’utilise des techniques en système d’information géographique pour déterminer l’impact d’une perturbation naturelle épisodique (tempête de verglas) ainsi qu’une perturbation anthropogénique chronique (sentiers récréationnels) sur la qualité des habitats forestiers (Chapitre 1) et un assemblage de chenilles vivant en forêt (Chapitre 2). J’ai démontré que les dégâts engendrés par les tempêtes de verglas sont distribués de façon hétérogène à travers le paysage forestier, différant selon le type de débris ligneux grossiers examiné. Ces différents types de débris ligneux grossiers servent d’habitat à divers groupes taxonomiques. J’ai découvert que les perturbations anthropogéniques, au contraire des perturbations naturelles, n’ont pas eu d’impacts positifs sur les assemblages de chenilles vivants en forêt. J’ai également exploré le concept de qualité d’habitat en considérant l'identité de la plante hôte (Chapitre 3) et la qualité foliaire à l'échelle du quadrat (Chapitre 4). J'ai découvert que l'assemblage de chenille 28 démontre des préférences marquées pour certaines plantes hôtes et que ces préférences peuvent dépendre de différentes qualités foliaires à l'échelle du quadrat à différentes périodes durant la saison de croissance. Cette thèse contribue à la documentation croissante sur les facteurs qui affectent la richesse spécifique des communautés d'insectes vivant dans les forêts des paysages perturbés. De plus, cette thèse propose plusieurs outils spécifiques à la gestion pour évaluer les perturbations en milieu forestier et donne un aperçu de l'impact que peut avoir la sélection de différentes espèces d'arbres sur les communautés forestières lors de l'élaboration d'initiatives de plantation d'arbres. General Introduction 29 Since the beginning of modern ecology “species richness” has been one of the central fociin research. It is the subject of some of the most important ecological theories and has been associated with overall ecosystem health (Rapport et al 1999) and pristine habitat (Brooks et al. 2002). Species richness is also the primary measurement used in conservation biology where the primary goal is to protect species and their natural environments from extinction. Ehrlich and Ehrlich (1992) argue that there are four reasons why conserving richness should be a primary goal for humankind: (1) it is an ethical responsibility to preserve diverse forms of life on earth, (2) there are recreational and aesthetic valuesof high species richness, (3) there is a wide variety of medical, food and textile resources offered by high species richness, and (4) there is a wealth ofimportant ecological services offered by high species richness such as air filtration, water filtration, nutrient cycle maintenance and soil renewal. Each of these four factors may (arguably) be diminished with significantrichness losses. While similar arguments have been echoed by other authors and are rarely refuted (Ghilarov 2000, Randall 1991), we live in an era where species extinction rates are extraordinarily high, due in large part to human activities (Pimm et al., 1995). These extinctions are particularly high for insect species (Conrad et al 2006, Dunn 2005, Thomas et al 2004) and are often due to human-caused landscapescale habitat destruction for the purposes of timber exploitation, agriculture and landscape development (Brooks et al 2002). What Controls Richness? 30 Why are some areas high in species richness while others are impoverished? The variance of richness within and among different habitats has long intrigued ecologists. There have been dozens of different mechanistic drivers of species richness proposed over the years and the strength of a proposed variable often seems to be context-dependent (Rosenzweig 1995, Adams 2009). Mechanistic drivers of richness identified at broad geographic scales tend to differ from those identified at small geographic scales (Currie 1991, Hawkins et al 2003); mechanistic drivers identified in disturbed habitat tend to differ from those identified in undisturbed habitat (Armesto and Pickett 1985).At the most general level these drivers fall into one of two categories: biotic mechanisms and abiotic mechanisms. Biotic mechanisms occur when living things affect the richness of a taxonomic group. These mechanisms can be divided into two sub-categories: top-down forces and bottom-up forces. Top-down forcesare observable when a species or taxonomic group at a higher trophic level affects a species or taxonomic group at a lower tropic level. Herbivores, for example, can suppress the species richness of their food plants (Crawley 1983, Hay 1985, Hunter and Price 1992) and predators/parasitoids can affect the species richness of their prey (Bruno and Cardinale 2008, Lill et al 2002). Bottom-up forcesare the opposite and are observable when a species or taxonomic group at a lower trophic level affects a species or taxonomic group at a higher trophic level. For example, grassland plots with many plant species can support more herbivorous 31 insect species than plots with few plant species (Siemann et al. 1998) and systems rich in prey can often support more predator species than those that are poor (Rosenzweig 1995). The direction of the effect (i.e. top-down vs. bottomup) often seems to be very system-dependent and opposing effects can be observed in analogous trophic structures– e.g. in some systems the bottom-up effect of plants on herbivores is dominant and in others the top-down effect of herbivores on plants is dominant. A third categoryof driving biotic forces exists where regulation within a trophic level occurs through competition or densitydependence. However this is more commonly observedat the species population level when measuring species abundance and performance rather than at the assemblage or community level when measuring species richness (e.g. Antonovics and Levin 1980, Ostfeld et al 2003). Abiotic mechanisms occur when non-living things affect the richness of a taxonomic group. Theseoften include climate variables like temperature, precipitation, solar radiation and habitat modification variables likefragmentation, isolation and habitat loss. At very large spatial scales precipitation and energy are strong correlates to species richness (Currie 1991, Hawkins et al. 2003). At smaller spatial scales disturbance events can play more important roles. Human disturbance through habitat modification and destruction generally results in species richness loss (e.g. Andren 1994, Helm et al. 2006, Ross et al. 2002) whereas natural disturbances like fire and severe weather events can result in species richness gains or losses depending on the 32 species group and habitat affected (e.g. species richness gain –Shafi and Yarranton1973, Facio 2003, Lafon 2004, Moretti et al 2004; species richness loss – Nekola 2002, Saint-Germain and Mauffette 2001, Swengel 2006). Thesis Objective The objective of this thesis is to examine how biotic forces (top-down and bottom-up effects) and abiotic forces (natural and human disturbances) drive terrestrial species richness. Understanding how these factors drive richness is particularly important in disturbed landscapes where native species richness is declining and extinctions are common. Since the direction and magnitude of these driving factors can be context-dependent, I aim to analyze them using a single taxonomic group in a specific ecological habitat type. The criteria I use for selecting a taxonomic group are: (1) it must be a group occupying mid-trophic levels and therefore subject to both bottom-up and top-down pressure, (2) it must be broad enough to be ecologically important to ecosystem function in disturbed landscapes, (3) it must be narrow enough to occupy one, roughly homogeneous, ecological guild, (4) there must be a plausible mechanism by which it can be affected by both natural and human disturbances. The criteria I use for selecting an ecological habitat are: (1) it must be subjected to periodic or chronic human disturbance, (2) it must be subjected to natural disturbance events, preferably where a recent large disturbance event has occurred, and (3) it must have high conservation value. 33 Taxonomic Group of Investigation One ecologically important mid-trophic level taxon that is often affected by bottom-up and top-down effects is insect herbivores. It isone of the most species-rich guilds in terrestrial ecosystems and is ecologically important for many reasons. First, many insect species are indicators thatreflect habitat quality and the impacts of environmental disturbances (Summerville et al. 2004, New 2004). Second, insect herbivores play critical roles in terrestrial ecosystems as food sources for bats (Goiti et al. 2009), birds (Murray et al. 1980), small mammals (Whitaker 1966), reptiles (Hamilton and Pollack 1956), amphibians (Moore and Stickland 1954) and other insects (Lill and Marquis 2001). Third, theyalso play a crucial role in ecosystem function by converting leaves into nutrient-rich frass that is readily consumed by detritivores and soil-dwelling organisms (Schowalter et al. 1986). Therefore, in many ways, the health of an ecosystem is tied to the species richness of the insect herbivores assemblage living therein. They are also commonly impacted by both bottom-up (Siemann et al. 2008) and top-down (Lill et al. 2002) effects. Conservation initiatives focusing on insects havetraditionallyusedone of two approaches: a single-species approach,or an assemblage approach. In the single-species approach, a threatened species is identified and steps are recommended to improve its survival chances. This could include increasing the abundance of important host plants or trying to control predator populations 34 (e.g. Schultz and Dlugosch 1999).Onedrawback of the single-species approach is that it is limited to species whose natural history is well known. This is particularly true for insect herbivores because, as a taxon, their natural history is very poorly known (New 2004). In fact, it is suggested that the majority of insect species in the world is currently undiscovered, let alone well-described (Gaston 1991). While most of these undiscovered species are in tropical ecosystems, the majority of well-studied species in temperate habitats tend to be defoliators capable of causing significant economic and aesthetic damage (e.g. the gypsy moth, Lymantria dispar) or species with bright colouration (e.g. the monarch butterfly, Danaus plexippus). Temperate non-defoliating species and cryptic species are often overlooked. Because of this narrow spectrum of research it becomes difficult to makeeffectiveconservation-oriented management decision based on one (or a few) species when the status of the rest of the assemblage isunknown.This approach has sometimes been justified by the “umbrella species” concept, but it is not always clear that thisachieves the desired goal (Andelman and Fagan 2000, Roberge and Angelstam 2004). Instead, the net result of this approach can be an overly large management investment for the benefit of few species. Conversely, an assemblage-based approach focuses on identifying, conserving and protectinghabitat that is high in species richness(New 2004). This requires less knowledge of the natural history of specific species because the variable of interest isspecies richnessin an assemblage rather than abundance in a species. It asks: “What characteristics of the habitat are 35 synonymous with high richness?” rather than “What characteristics of the habitat are synonymous with high abundance in one species?” The assemblagebased approach can be more effective on the whole but it requiresa contextspecific understanding of how assemblage richness responds to driving mechanisms. Geographic Extent of Study There are dozens of different ecosystems that host insect herbivores and many fit the desired criteria of being impacted by both human and natural disturbances. One such ecosystem that meets both criteria is the broadleaved forest of the Mixedwood Plains (MWP) in southern Canada. Historically this forest covered the entirety of the MWP but over the past century ithas been reduced to isolated fragments in a matrix of agricultural, industrial and urban development; less than 15% the area is currently forested(Allen 2001). The small amount of forest that remains is important habitat for threatened species yet continues to be at risk from human landscape modification(Environment Canada 2007, 2010). Foremost amongst these remnant fragments are those located in an areaof southern Quebec known as the Montérégiewhere less than 1% offorested land is protected(Ministère des Ressources Naturelles 2002).Many of these protected forest fragments tend to occupy the hilltops of the Monteregian Hills, a set of nine laccolith-based hills rising up from the floodplain. 36 Conservation on these hills has become increasingly challenging with the occurrence of episodic and chronic, human and natural disturbances across the MWP landscape. Although these hilltop forest fragments have escaped widespread deforestation, privately owned lots within the fragments have been subjected to residential development with associated vehicle access roads and hiking trails (CantonShefford 2010, Centre de la Nature Mont Saint Hilaire 2007, Les Amis De La Montagne 2008, Parcs Quebec 2010, Parcs Quebec 2010). Smaller-scale cutting (and subsequent replanting) and the periodical establishment of sugarbushes (resulting in the selective removal of all but sugar maple trees) has also occurred over the past century and has changed the forest tree species composition. Many of these disturbance events occurred before the forests on these hills were granted protected status. Additionally, periodic ice storms that damage canopy trees have left legacies of coarse woody debris on the forest floor, changing habitat quality for the forest insect community (Hooper et al. 2001). The persistence of forest habitat on these hills in a landscape of agricultural development is tied tohistoric events. Up until 10,000 ypb the Champlain Sea covered much of the St. Lawrence floodplain region; when it receded (c. 8,000 ybp) it left nutrient-rich sediment on the floodplain while the hilltops (which were islands in the Sea) remained comparatively nutrient-poor. Because of this sub-optimal soil quality and because of their challenging topography, the hills remained (in part or in whole) undeveloped when the rest 37 of the floodplain was converted for human-use throughout the 20th century. Forest fragments on three of these hills (Mont Royal, Mont St. Bruno and Mont St. Hilaire) are currently protected (to varying degrees) with mandates to preserve the diversity within their forests for conservation targets and forthe enjoyment of the general public. A fourth Hill (Mont Shefford) has designated conservation areas without explicit conservation mandates. Four Approaches to Determine Drivers of Richness The Monteregian forest fragments present an ideal ecosystem in which to study the effects ofbiotic and abiotic forcesoninsect species richness. I use four diverse approaches to examine this relationship. First, I approach it by using remote sensing and GIS tools to examine the consequences of an ice-storm disturbance (i.e. a natural abiotic driver) on saproxylic insects. Since there is a direct link between high quality habitat and guild species richness, determining the habitat consequences of a major natural disturbance is an important step for making conservation-oriented management decisions. In addition, I develop and test assessment tools to help quantify habitat quality and disturbance impact. The last major ice stormdisturbance in my study region occurred in 1998 when more than 100 mm of ice fell on Monteregian forest fragments. Various canopydamage models have been developed to gauge the impact of such events (e.g. Olthof et al. 2004, Proulx and Greene 2001, Stueve et al. 2007), but there is a need to better connect forest damage with ecologically meaningful habitat 38 changes.In Chapter 1 I usesatellite-derived remotely sensed vegetation indices to examine how this storm changed coarse woody debris habitat at Mont St. Hilaire, a hill in the MWP of Quebec. I develop geospatial data to document the influx of different types of coarse woody debris that are important saproxylic insect communities. The research goals of Chapter 1 are stated in the form of objectives rather than hypotheses. They are as follows: 1. To model the biotic changes resulting from coarse woody debris influx following a major canopy disturbance. 2. To use remotely sensed satellite images and geographic information systems to study the spatial heterogeneity, volume and connectivity of coarse woody debris habitat following a major canopy disturbance. 3. To test the performance of seven different vegetation indices in predicting habitat change after a major canopy disturbance. Second, I approach the question of how abiotic and biotic forces affect insect species by examining howman-made recreational trails (i.e. a human abiotic driver) impact forest-dwelling Lepidoptera assemblages. The biggest form of human impactwithin forest fragments is chronic disturbance fromrecreational hiking trails. Trails are particularly compelling to research because they are a type of abiotic disturbance that can be controlled by management decisions within forested parks and reserves. This sets them apart from other types of 39 disturbance that are either uncontrollable (e.g. ice storms) or extremely difficult to control without more powerful conservation mandates (e.g. to stop landscape-scale deforestation). There are often significant habitat modifications caused by the introduction of trails in forests. They are associated with high levels of vegetation trampling, a higher abundance of disturbance-tolerant plants, soil compaction, water drainage changes, light level changes, wind changes and temperature changes. As an analogy to trails, forest edge habitat has often been observed to have a higher than expected Lepidopterarichness due to a mixed assemblage – part forest-dwelling Lepidoptera species, part open-habitat Lepidoptera species. Trails have been positively, negatively, or neutrally associated with population abundances and species richness in certain taxa (e.g. birds: Miller 1998), but rarely linked to insect herbivore assemblages. A greater understanding of the impact of trails on Lepidoptera assemblages can be used to better protect and preserve richness in disturbed forest fragments. The hypotheses tested in chapter 2 are: 1. Trailside habitat will be beneficial for caterpillar richness and abundance. 2. The availability of acceptable host plants will be a determinant of caterpillar richness and abundance. Third, I approach the question of how abiotic and biotic forces affect insect species by examining how tree species richness and identity (i.e. biotic bottomup mechanisms) drive Lepidoptera richness and abundance within forest 40 fragments. In studies of single insect herbivore species, insect performance and distribution are strongly tied to host plant identity, indicating that some host plants are more preferred than others. This is contrasted by studies of insect herbivore assemblages that often equate high insect assemblage richness to high tree species richness. While it is well known that single insect herbivore species (even polyphages)can have strong host plant preferences (Liebhold et al. 1995, Maufette et al. 1983, Wint 1983) it is not known how well the host preferences of individual species scales to the assemblage level. The claim that higher host plant richness leads to higher insect herbivore richness is rooted in the assumption that richer communities have a greater diversity of foliar nutrients and a more structurally heterogeneous set of resources. This foliar and structural diversity would theoretically allow more insect herbivore species to coexist. On the other hand, comparisons of the hosting capabilities of trees show that some hosts are more highly preferred than others (Barbosa et al. 2000, Moran and Southwood 1982). Based on this, it is possible that the insect richness in a forest stand depends on having a high volume of preferred hosts. This assemblage-level “preferred-host” concept may be especially applicable for Lepidoptera in the MWP forests where a vast majority of species are polyphagic (Handfield1999). It is therefore possible for one or few hosts to be preferred by the majority of the assemblage.These two concepts are not necessarily mutually exclusive, but if the second outperforms the first it has important implications for forest management and suggests a revision in our understanding of the relationships 41 between insect herbivore assemblages and their environment. Does an abundance of preferred host plants correlate to high insect herbivore richness or is insect herbivore richness more closely related to host plant richness? In chapter 3 I test between these two possibilities. Hypotheses: 1. The diversity and richness of a Lepidoptera assemblage is locally driven by host plant richness. 2. Because host plant choice are made by individual Lepidoptera species, the relationships between the Lepidoptera assemblage and host plants will be accurately described as a function of host plant identity and abundance (i.e. host plant preferences are cumulative). Fourth, I approach the question of how abiotic and biotic forces affect insect species by examining how top-down and bottom-up pressures(i.e. two biotic drivers) on insect herbivore assemblages vary over the course of a growing season. The impacts of tri-trophic relationships on insect herbivores have been well studied (Mayhew 1997), but never for an entire assemblage in specific temporal windows over the course of a growing season. Individual insect herbivorespecies often discriminate between available host trees based on certain foliar nutrient qualities that change with leaf age (Feeny 1970, Ricklefs and Matthew 1982). A host plant might therefore be well usedearly in the season but be relatively unused late in the season due to a loss of nutritive value. 42 Top-down pressure applied by predators and parasitoids (Lill et al. 2002) complicates the seasonal relationship between insect herbivores and treeuse.This top-down pressure is often more intense later in the growing season. A full understanding of the drivers of insect herbivore richness needs to take into account the both bottom-up and top-down pressurein the context of a tritrophic relationship.In Chapter 4 I examine intra-seasonal tri-trophic relationships for a Lepidoptera assemblage in MWP forest fragments. The key foliar nutrient properties that have been linked as drivers of insect herbivore performance are nitrogen content, water content, phosphorus content, toxin content, fiber content, lignin content and leaf toughness. The nutrients among these that are positively linked to herbivore performance – nitrogen, phosphorus and water – tend to peak early in the season. The other nutrients peak in the middle or at the end of the season. Insect herbivore host preferences often correlate to host plant quality but this can be disrupted by high parasitoid abundances (Gratton and Denno 2003). Parasitoids are not typically evenly distributed among all host plants – some have high parasite loads (enemypacked space) while others have low parasite loads (enemy-free space). Host plant choices for insect herbivores thus become a delicate balance of minimizingexposure to parasitoids by avoiding enemy-packed space while maximizing the nutritive quality of the foliage consumed. Two important questions have arisen in trying to better understand the drivers of insect herbivore assemblage richness: (i) to what extent does bottom-up (foliar 43 nutrient quality) and top-down (parasitoid-presence) variables impact herbivore richness and (ii) how do the nature of bottom-up and top-down effects change from the beginning to the end of a growing season? Hypothesis: 1. The impact of top-down and bottom-up pressures will vary over the course of the growing season. Specifically: a. Foliar nitrogen content, water content and phosphorus content should be the primary drivers of Lepidoptera assemblage richness and abundance early in the season. b. Foliar toughness, fiber content and lignin content should be the primary drivers of Lepidoptera assemblage richness and abundance late in the season. c. Parasitoid pressure should have a significant impact on Lepidoptera assemblage richness and abundance late in the season. References Adams, J. 2009. Species Richness: Patterns in the Diversity of Life. - Springer Science, Praxis Publishing. Chichester, U.K. 44 Allen, E. 2001. Forest health assessment in Canada. - Ecosystem Health 7: 28-34. Andelman, S. J. and Fagan, W. F. 2000. Umbrellas and flagships: Efficient conservation surrogates or expensive mistakes? – Proceedings of the National Academy of Science97: 5954-5959. Andrén, H. 1994. Effects of habitat fragmentation on birds and mammals in landscapes with different proportions of suitable habitat: a review. – Oikos 71: 355-366. Antonovice, J. and Levin, D. A. 1980. The ecological and genetic consequences of density-dependent regulation in plants. – Annual Reviews of Ecology and Systematics 11: 411-452. Armesto, J. J. and Pickett, S. T. A. 1985. Experiments on distribution in old-field plant commuities: Impact on species richness and abundance. – Ecology 66: 230-240. Barbosa, P., Segarra, A. and Gross, P. 2000. Structure of two macrolepidopteran assemblages in Salix nigra (Marsh) and Acer negundo L.: abundance, diversity, richness, and persistence of scarce species. - Ecological Entomology 25: 374-379. Brooks, T. M., Mittermeier, R. A., Mittermeier, C. G., daFonseca, G. A. B., Rylands, A. B., Konstant, W. R., Flick, P., Pilgrim, J., Oldfield, S., Magin, G. and Hilton-Taylor, C. 2002. Habitat loss and extinction in the hotspots of biodiversity. – Conservation Biology 16: 909-923. 45 Bruno , J. F. and Cardinale, B. J. 2008. Cascading effects of predator richness. – Frontiers in Ecology and the Environment 6: 539-546. Canton Shefford. 2010. Carte Sentiers Parc Ecologique 2. - La Municipalité du Canton de Shefford, Shefford, Quebec. Centre de la Nature Mont Saint Hilaire. 2007. Gault Nature Reserve Forest Cover Type Map. - Centre de la Nature Mont Saint-Hilaire, McGill University. Conrad, K. F., Warren, M. S., Fox, R., Parsons, M. S. and Woiwod, I. P. 2006. Rapid declines of common, widespread British moths provide evidence of an insect biodiversity crisis. – Biological Conservation 132: 279-291. Crawley, M. J. 1983. Herbivory, the dynamics of animal-plant interactions. Blackwell Scientific, Oxford, England. Currie, D. J. 1991. Energy and large-scale patterns of animal- and plant-species richness. – TheAmerican Naturalist 137: 27-49. Dunn, R. R. 2005. Modern insect extinctions, the neglected majority. – Conservation Biology 19: 1003-1036. Eby, G. N. 1984. Geochronology of the Monteregian Hills alkaline igneous province, Quebec. - Geology 12: 468-470. Ehrlich, P. R. and Ehrlich, A. H. 1992. The value of biodiversity. – Ambio 21: 219226. Environment Canada. 2007. Species at Risk Database. - Government of Canada, Gatineau, Quebec. 46 Environment Canada. 2010. Protected Areas Data. Environmental Indicators. – Environment Canada, Gatineau, Quebec. Faccio, S. D. 2003. Effects of ice storm-created gaps on forest breeding bird communities in central Vermont. – Forest Ecology and Management 186: 133–145. Feeny, P. 1970. Seasonal changes in oak leaf tannins and nutrients as a cause of spring feeding by Winter Moth caterpillars. - Ecology 51: 565-581. Feininger, T. and Goodacre, A. K. 1995. The eight classical Monteregian hills at depth and the mechanism of their intrusion. - Canadian Journal of Earth Sciences 32: 1350-1364. Gaston, K. J. 1991. The magnitude of global insect species richness. – Conservation Biology 5: 283-296. Ghilarov, A. M. 2000. Ecosystem functioning and intrinsic value of biodiversity. – Oikos 90: 408-412. Goiti, U., Garin, I., Almenar, D., Salsamendi, E. and Aihartza, J. 2009. Foraging by Mediterranean Horseshoe Bats (Rhinolophus euryale) in relation to prey distribution and edge habitat. - Journal of Mammalogy 89: 493-502. Gratton, C. and Denno, R. F. 2003. Seasonal shift from bottom-up to top-down impact in phytophagous insect populations. - Oecologia 134: 487-495. Hamilton, W. J. and Pollack, J. A. 1956. The food of some colubrid snakes from Fort Benning, Georgia. - Ecology 37: 519-526. Handfield, L. 1999. Le Guide Des Papillons Du Quebec. - Broquet Inc. 47 Hawkins, B. A., Field, R., Cornell, H. V., Currie, D. J., ois Gue’gan, Kaufman, D. M., Kerr, J. T., Mittelbach, G. G., Oberdpff, T., O’Brien, E. M., Porter, E. E. and Turner J. R. G. 2003. Energy, water and broad-scale geographic patterns of species richness. – Ecology 84: 3105-3117 Hay, M. E. 1985. Spatial patterns of the herbivore impacts and their importance in maintaining algal species richness. – Pp. 29-34 in Proceedings of the 5th International Coral Reef Congress. Moorea, French Polynesia. Helm, A., Hanski, I. and Partel, M. 2006. Slow response of plant species richness to habitat loss and fragmentation. – Ecology Letters 9: 72-77. Hunter, M.D. and Price, P.W. 1992. Playing chutes and ladders: heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. – Ecology 73: 724-732. Hooper, M. C., Arii, K. and Lechowicz, M. J. 2001. Impact of a major ice storm on an old-growth hardwood forest. - Canadian Journal of Botany 79: 70-75. Lafon, C. W. 2004. Ice-storm disturbance and long-term forest dynamics in the Adirondack Mountains. – Journal of Vegetation Science 15: 267-276. Les Amis De La Montagne. 2008. Interactive Map of Mont Royal. - Les Amis de La Montagne, Montreal, Quebec. Liebhold, A. M., Gottschalk, K. W., Muzika, R., Montgomery, M. E., Young, R., O'Day, K. and Kelley, B. 1995. Suitability of North American Tree Species to the Gypsy Moth: A Summary of Field and Laboratory Tests. United 48 States Department of Agriculture, Northeastern Forest Experiment Station, Radnor, PA Lill, J. T. and Marquis, R. J. 2001. The effects of leaf quality on herbivore performance and attack from natural enemies. - Oecologia 126: 418-428. Maufette, Y., Lechowicz, M. J. and Jobin, L. 1983. Host preferences of the gypsy moth, Lymantria dispar(L.), in southern Quebec. – Canadian Journal of Forest Research 13: 53-60. Mayhew, P.J. 1997. Adaptive patterns of host-selection by phytophagous insects. – Oikos 79: 417-429. Miller, S. G., Knight, R. W. and Miller, C. K. 1998. Influence of recreational trails on breeding bird communities. - Ecological Applications 8: 162-169. Ministère des Ressources Naturelles. 2002. Rapport sur L'état des forêt québécoises 1995-1999,. - In: Ressources Naturelles et Faune,Gouvernement du Québec, Charlesbourg, Quebec Moran, V. C. and Southwood, T. R. E. 1982. The guild composition of arthropod communities in trees. – Journal of Animal Ecology 51: 289-306. Moretti, M., Obrist, M. K. and Duelli, P. 2004. Arthropod biodiversity after forest fires: winners and losers in the winter fire regime of the southern Alps. – Ecography 27: 173-186. Murray, N. D., Bishop, J. A. and Macnair, M. R. 1980. Melanism and predation by birds in the moths Biston betularia and Phigalia pilosaria. - Proceedings of the Royal Society of London. Series B, Biological Sciences 210: 277-283. 49 Nekola, J. C., 2002. Effects of fire management on the richness and abundance of central North American grassland land snail faunas. Animal Biodiversity and Conservation 25: 53–66. New, T. R. 2004. Moths (Insecta: Lepidoptera) and conservation: background and perspective. - Journal of Insect Conservation 8: 79-94. Olthof, I., King, D. J. and Lautenschlager, R. A. 2004. Mapping deciduous forest ice storm damage using Landsat and environmental data. - Remote Sensing of Environment 89: 484-496. Ostfeld, R. S., Canham, C. D. and Pugh, S. R. 2003. Intrinsic density-dependent regulation of vole populations. – Nature 366: 259-261. Parcs Quebec. 2010. Parc National de la Yamaska, 2010-2011. - Réseau Sépaq, Quebec, Quebec. Parcs Quebec. 2010. Parc National du Mont-Saint-Bruno, 2010-2011. - Réseau Sépaq, Quebec, Quebec. Pimm, S. L., Russell, G. J., Gittleman, J. L. and Brooks, T. M. 1995. The future of biodiversity. – Science 5222: 347-350. Proulx, O. J. and Greene, D. F. 2001. The relationship between ice thickness and northern hardwood tree damage during ice storms. - Canadian Journal of Forest Research 31: 1758-1767. Rapport, D. J., Costanza, R. and McMichael, A. J. 1999. Assessing ecosystem health. – Trends in Ecology and Evolution 13: 397-402. Randall, A. 1991. The value of biodiversity. – Ambio 20: 64-68. 50 Ricklefs, R. E. and Matthew, K. K. 1982. Chemical characteristics of the foliage of some deciduous trees in southeastern Ontario. - Canadian Journal of Botany 60: 2037-2045. Roberge, J. M. and Angelstam, P. 2004. Usefulness of the umbrella species concept as a conservation tool. – Conservation Biology 18: 76-85. Rosenzweig, M. L. 1995. Species Diversity in Space and Time. – Cambridge University Press, Cambridge, U.K. Ross, K. A., Fox, B. J. and Fox, M., D. 2002. Changes to plant species richness in forest fragments: fragment age, disturbance and fire history may be as important as area. – Journal of Biogeography 29: 749-765. Saint-Germain, M. and Mauffette, Y. 2001. Reduced ground beetle activity following ice damage in maple stands of southwestern Quebec. The Forestry Chronicle 77: 651-656. Schowalter, T. D., Hargrove, W. W. and Crossley, D. A. 1986. Herbivory in forest ecosystems. - Annual Review of Entomology 31: 177-196. Schultz, C. B. and Dlugosch, K. M. 1999. Nectar and hostplant scarcity limit populations of an endangered Oregon butterfly. – Oecologia 119: 231238. Shafi, M. I., and Yarranton, G. A. 1973. Diversity, floristic richness, and species evenness during a secondary (post-fire) succession. – Ecology 54: 897902. 51 Stueve, K. M., Lafon, C. W. and Isaacs, R. E. 2007. Spatial patterns of ice storm disturbance on a forested landscape in the Appalachian Mountains, Virginia. - Area 39: 20-30. Summerville, K. S., Ritter, L. M. and Crist, T. O. 2004. Forest moth taxa as indicators of lepidopteran richness and habitat disturbance: a preliminary assessment. - Biological Conservation 116: 9-18. Swengel, A. B. 1996. Effects of fire and hay management on abundance of prairie butterflies. – 76: 73-85. Thomas, J. A., Telfer, M. G., Roy, D. B., Preston, C. D., Greenwood, J. J. D., Asher, J., Fox, R., Clarke, R. T., and Lawton, J. H. 2004. Comparative losses of British butterflies, birds, plants and the global extinction crisis. – Science 303: 1879-1881. Whitaker. 1966. Food of Mus musculus, Peromyscus maniculatus bairdi and Peromyscus leucopus in Vigo County, Indiana. - Journal of Mammalogy 47: 473-486. 52 CHAPTER 1: Detecting Changes in Forest Floor Habitat after Disturbance Peter J.T. White1, Brian J. McGill2 and Martin J. Lechowicz1 1 McGill University, Department of Biology 1205 Dr. Penfield Ave., Montreal, Quebec H3A 1B1, Canada 2 University of Maine, School of Biology & Ecology Deering Hall 303, Orono, ME 04469, U.S.A. Abstract A massive ice storm hit northeastern North America in 1998, dropping more than 100 mm of freezing rain at its epicenter in southern Quebec, Canada. There has 53 been extensive study of which trees and areas received the most damage, but the biodiversity consequences of this damage at landscape scales have not received much attention. We assessed the effectiveness of seven remotelysensed vegetation indices - NDVI, EVI, DVI, RDVI, ARVI, NDVIgreen and VARI - for modeling the coarse woody debris (CWD) influx in an old growth forest reserve at the storm’s epicenter; NDVI was the best predictor of CWD influx. We categorized the geospatial CWD predictions from the NDVI-derived model to map the spatial distribution of sun-exposed, moist-shaded, dry-shaded and wet CWD microhabitats on the forest floor. Moist-shaded, dry-shaded and wet patches of CWD were large and well connected, but sun-exposed patches were small and sparse. Since these microhabitats affect the distribution and abundance of saproxylic insects, wood-rotting fungi, salamanders, birds, small burrowing mammals and plant species dependent on nurse-logs for establishment, the CWD influx from the 1998 ice storm may have revitalized local populations of these taxa through increased habitat availability as well as increased dispersal within the reserve. Introduction A catastrophic ice storm hit northeastern North America from January 4th to 10th, 1998. Southern Quebec was most heavily impacted with some areas receiving in 54 excess of 100 mm of freezing rain (Milton and Bourque 1999). The ice storm is considered the worst ever recorded in Canada (Milton and Bourque 1999, Vandyke 1999). The Quebec Ministry of Natural Resources estimated that 12% of trees in southern Quebec had more than 75% of their canopy destroyed and were unlikely to survive (Irland 1998). Local Consequences of Damage Much of the ecological research about the 1998 ice storm has focused on local correlates of forest damage and the structural consequences of damage. Local topography has been consistently linked to damage with specific elevations (Irland 2000, Weeks et al. 2009), slope angles (Bragg et al. 2003, Rhoads et al. 2002) and slope aspects (Bragg et al. 2003) all emerging as significant descriptors. The reported species-specific impacts of ice on trees have varied from study to study (Vandyke 1999) and from locality to locality (Brommit et al. 2004, Dugay et al. 2001, Hooper et al. 2001, Irland 1998, Rhoads et al. 2002). The damage caused by ice storms has been found to accelerate, slow, or have no impact on succession depending on the specific forest stand being examined (Boerner et al. 1988, Brommit et al. 2004, DeSteven et al. 1991, Takahashi et al. 2007). Landscape and Regional Investigations The landscape and regional scale impacts of ice storms have been best studied by blending satellite reflectance data, digital topography data, and ice accretion 55 data (King et al. 2005, McNab and Roof 2006, Millward and Kraft 2004, Olthof et al. 2004, Stueve et al. 2007). For the 1998 ice storm, Millward and Kraft (2004) used the Normalized Difference Vegetation Index (NDVI) to map canopy damage in a 2000 km2 forest in the eastern Adirondack region of New York State. They reported a complex and heterogeneous damage pattern, spatially autocorrelated at distances below 300 m. Olthof et al. (2004) used pre- and post- storm remotely sensed forest reflectance to create a regional damage map for eastern Ontario; this study is also documented in King et al. (2005). Their damage models discriminated between areas of low and high damage, identifying forest areas in need of management intervention. These few regional and landscape scale studies stand out in a literature dominated by studies at much smaller scales, often focusing on no more than several dozen plots. Habitat Implications of Ice Storms The biotic implications of a large coarse woody debris (CWD) influx that necessarily follows a major ice storm have rarely been considered on a landscape scale, even though a diverse and abundant CWD landscape can have profound positive impacts on forest diversity (Harmon et al. 1986). CWD created in different moisture, heat and sunlight environments can support significantly unique assemblages of saproxylic (CWD loving) insects. For example, it is common for many bark beetles to prefer sun-exposed CWD over shaded CWD. The warmer habitat is often a more inviting micro climate, and the sun-exposure 56 can lead to a quicker volatization of tree host compounds making the substrate more preferable for many species (seeBouget and Duelli 2004 for a review). In wet and moist habitats, the quick colonization of brown rot and white rot fungi can modify the lignin, cellulose and hemi-cellulose concentrations in decaying CWD making it favourable habitat for a different beetle assemblage (Kaila et al. 1994, Yee et al. 2004). Many saproxylic hoverflies also prefer to breed in very moist decaying wood (Rotheray and Stuke 1998). Yet, both of these assemblages are usually different from the saproxylic assemblage that uses waterlogged and floating CWD (Braccia and Batzer 2001). Aquatic insects (mainly in freshwater systems) use CWD as a feeding platform, food source, refuge from predation and substrate for oviposition (Cranston and McKie 2006, Harmon et al. 1986). The size, areal density and connectivity of CWD habitat are also important to consider. Large pieces of CWD decay at a significantly slower rate than small pieces (Vanderwel et al. 2006) and consequently provide a more stable habitat, which can be very important for taxa with limited dispersal abilities (Yee et al. 2004). Species richness and population abundance can be higher for small mammals, woodpeckers, herptofauna, fungi and plants in areas with a high volume of CWD (Keisker 2000). Furthermore, the connectivity of CWD patches can be vitally important as CWD can support more species when it is well connected compared to when it has patchy or clumped distribution (McMillan and Kaufman 1995, Schiegg 2000). 57 Using Remote Sensing to Measure Habitat Change NDVI is the most popular remotely sensed vegetation index (VI) used in ecological geospatial research, but there is often little rationale given for this choice (cf.Birky 2001, Millward and Kraft 2004, Newton 2007, Pettorelli et al. 2005 among many others). NDVI was originally proposed by Rouse (Rouse 1973, Rouse et al. 1974) and has a wide array of applications (Bannari et al. 1995); it is positively correlated to the Leaf Area Index (Fassnacht et al. 1997) and leaf biomass (Birky 2001) in temperate forests. In addition to NDVI, there have been more than 40 other VIs developed since the early 1970s (Bannari et al. 1995). Many of these indices were designed with specific purposes in mind, for example, to delineate low versus very low levels of vegetation (Richardson and Everitt 1992) or high versus very high levels of vegetation (Huete et al. 2002). Since red light is typically absorbed by chlorophyll and the near infrared light is reflected by foliage (Bannari et al. 1995), NDVI is a logical choice for measuring vegetation levels. But relying on these two wavelengths alone may omit important information: green light can be used to describe soil properties, blue light can be used to account for atmospheric abnormalities that may be present when using imagery from different dates. Because of these properties, it has been suggested that complete analyses should test and/or incorporate the use of more than one VI (Bannari et al. 1995, Liu et al. 2007, Payero et al. 2004). Objective 58 The goal of this paper is to use remotely sensed satellite images and geographic information systems to study the spatial heterogeneity, volume and connectivity of CWD habitat resulting from the 1998 ice storm at Mont St. Hilaire, a deciduous old-growth forest landscape; the secondary goal is to test the performance of seven different VIs in predicting habitat change after the ice storm. In this paper we use remotely sensed vegetation indices calculated from remotely sensed images, topography data and field measurements to create and validate a predictive landscape-level CWD influx model. Using hydrology data, topography data and canopy cover data we then categorize the influx model into four different CWD habitat types: sun-exposed, dry-shaded, moist-shaded and wet. Finally, we discuss the results and implications of this CWD influx for forest biodiversity. Methods Study Area Mont St. Hilaire is a laccolith-based hill situated on the St. Lawrence Floodplain, 30 km east of Montreal, Quebec (45°33'N; 75°09'W). Its annual average daily temperature is 6 oC fluctuating from a 21 oC monthly average in August to a -11 o C monthly average in January. On average, the region receives 836 mm of rain and 2140 mm of snow each year with 111 and 47 rainfall and snowfall days respectively (National Climate Data and Information Archive 2008). Mont St. Hilaire is topographically diverse in elevation, slope steepness and slope aspect 59 and covered in 10 km2 of old growth forest, primarily composed of Acer saccharum and Fagus grandifolia with significant numbers of Quercus rubra, Fraxinus americana,Tsuga canadensis spp. and Pinus strobus among other species (Arii 2004). Satellite Imagery Processing To model CWD input, weused multiband Landsat-5 Thematic Mapper (TM) imagery downloaded from the USGS EROS EarthExplorer (USGS 2007) for the years 1996 and 1998 (Table 1). Satellite imagery was selected from August to ensure that trees were at full leaf flush at the time of observation; scenes from 1997 could not be used due to high cloud cover in available images. Prior to the calculation of VIs, the Landsat imagery for each band was converted from digital numbers into radiometrically-corrected planetary reflectance (ρ) units using the following equation (adapted fromChandler and Markham 2003): [1] Gain and offset are band-specific rescaling values, Esunλ is the mean solar exoatmospheric irradiance, d is the earth-sun distance, and θs is the solar zenith angle. To account for atmospheric differences between the two images, we corrected satellite reflectance data using a relative radiometric normalization and atmospheric correction procedure (Hajj et al. 2008). Each band in the 1998 60 data was transformed to match the atmospheric conditions of the 1996 data using a linear regression equation derived by calculating the change in 150,497 invariant targets (i.e. pixels where no change occurred between years). All invariant targets were pixels between 60 and 160 km north of the ice storm epicenter where ice accretion was less than 5 mm and unlikely to cause canopydamage (Proulx and Greene 2001). VI Calculation We chose six additional VIs to compare to NDVI in ability to model change in CWD input from pre to post storm (Table 2). The additional VIs span the blue, green, red and infrared reflectance spectrum but were analogous to NDVI in the sense that (a) they are all simple ratio-based VIs and (b) they do not require any accessory parameters to compute. For all of these indices, large localized changes in VI values should correlate with canopy-level damage and gap formation that characteristically generate large amounts of CWD. The Enhanced Vegetation Index (EVI) was proposed as an improvement to the NDVI because it is less prone to saturate in high biomass regions (Huete et al. 2002) and is less affected by certain atmospheric and soil conditions that may degrade the NDVI signal (Liu and Huete 1995). The Difference Vegetation Index (DVI) offers more precision when measuring sparsely vegetated surfaces (Liu et al. 2007, Richardson and Everitt 1992). We included it in our model because heavy disturbance could uncover sparsely vegetated forest floor that may be better 61 measured with DVI. The Renormalized Difference Vegetation Index (RDVI) is an index that combines the benefits of NDVI and DVI; it performs well in both dense canopy and sparse vegetation (i.e. canopy gap) conditions (Roujean and Breon 1995). The Atmospherically Resistant Vegetation Index (ARVI) is a newer version of NDVI that takes into account and corrects for atmospheric light-scattering effects on the red band (Bannari et al. 1995, Kaufman and Tanre 1992). The ARVI accomplishes this correction by subtracting the difference in red and blue bands from the red band before final Index calculation. A parameter used in its calculation, γ, is an atmospheric self-correcting factor and is generally set to 1.0 for most remote sensing applications (Kaufman and Tanre 1992). The Green Normalized Difference Vegetation Index (NDVIgreen), has been found to be very sensitive to leaf chlorophyll levels (Gitelson and Merzlyak 1998) and can be less prone to atmospheric interference found in the red band of normal NDVI (Wolter et al. 1995). Finally, using red, green and blue bands, the Visible Atmospheric Resistant Index (VARI) was designed to be very resistant to atmospheric effects, employing the use of the blue-band similar to ARVI (Gitelson et al. 2002). Furthermore, this index omits near infrared light making it less affected by leaf orientation and soil moisture fluctuations than other indices (Zhan et al. 2007). We calculated the change in each index from 1996 to 1998 using the following equation: [2] 62 Predicting the Spatial Pattern of CWD Influx In the summer of 1998 Hooper et al. (2001) recorded the biomass of freshly fallen CWD at 103 plots across Mont St. Hilaire. We used these data as reference points to make our spatial predictions of CWD influx across the mountain. To reduce the effect of unwanted spatial autocorrelation in the Hooper et al. data, we omitted plots that were less than 30 m apart. Ice storms do not have highly spatially contagious damage because most tree damage comes from direct ice loading originating from storm-based precipitation (a contrasting example of this is tree damage caused by fire or pests/pathogens where damage on one tree results in damage on some or all of the surrounding trees). The exception to this is collateral damage occurring when an ice loaded tree falls, damaging trees in its immediate vicinity (Rhoads et al. 2002).However, given that the occurrence of large tree fall at Hooper’s sites was very low, we decided that a 30 m buffer would sufficiently minimize spatial autocorrelation of collateral damage. This filtering process left 82 of 103 plots. We used cross validation to test the ability of each VI to describe the influx of CWD. In each cross validation case, 20% of plots were used as holdouts to validate regression importance. This process was repeated ten times for each VI with a different set of holdouts used for each regression computation. The average regression coefficient, intercept and R2 were computed to describe the properties and predictive power of each VI. We tested the VIs as individual 63 regressions because of the high degree of collinearity between indices. We chose the highest performing VI (based on R2) to make our spatial map of CWD influx using the average relationship predicted by the 10 training datasets of that VI. Independent Validation of CWD Predictions In the summer of 2008, we chose eighteen new plots at random locations on Mont St. Hilaire to validate our mountain-wide CWD data predictions. At each plot we measured CWD greater than 75 mm in diameter in two perpendicular intersecting 60m x 60m transects. Each piece was scored into one of five decay classes as defined by Canada’s National Forest Inventory Ground Sampling Guidelines (Canadian ForestService 2008). CWD in the youngest of these classes (i.e. category 1 and 2) are most likely to represent wood dating back to the 1998 ice storm (Vanderwel et al. 2006, Zielonka 2006). As such, we estimated the total CWD volume at each site as the sum of diameters of CWD in both categories 1 and 2. Methods Evaluation In our method of computing CWD influx predictions we had to scale the CWD values measured by Hooper et al. (2001) in 113 m2 plots to a 30 m x 30 m (900 m2) scale to correspond to the resolution of Landsat-5 TM imagery. To do this, we superimposed 30 m x 30 m plots centered on the original Hooper et al. plots and assigned each one the CWD value corresponding to the smaller 113 m2 plot 64 within it. Although this rescaled the CWD data to the desired resolution, the original plots were not georeferenced with Landsat-5 imagery in mind and thus the superimposed 30 m x 30 m pixels corners did not align with the 30 m x 30 m VI data pixels. To adjust for this, we recalculated VI values for each superimposed pixel as a weighted mean of the four overlapping VI pixels based on the proportion of overlap. To calculate the impact of this upscaling on the statistics of our models, we mimicked the procedure using high resolution (1 m) multiple-return Light Detection and Ranging (LIDAR) data collected in 2003 (collection method described in below). With these data we created a high resolution binary map of canopy cover (tree height > 10 m) (Popescu et al. 2002, St-Onge and Achaichia 2001) and calculated the actual canopy cover at each of our 82 test plots in a 6 m radius (113 m2). The high resolution canopy cover data was then converted to 30 m x 30 m resolution with the same spatial referencing used in our VI data. Using weighted means based on the proportion of overlap, we then calculated the average canopy cover at each superimposed 30 m x 30 m site for a correlation comparison. This gave us an empirically-driven approximation of how the scaling and weighted scaling of canopy data and VI data (respectively) affected the statistical fit of our models. Categorizing CWD Habitat 65 We divided our mountain-wide CWD prediction data into four categories denoting areas likely to contain dry-shaded CWD, moist-shaded CWD, sunexposed CWD and wet CWD.First, we created a watershed map of Mont St. Hilaire by categorizing ponds and the area within 15 m buffer of stream line features as wet habitat most likely to contain wet CWD. We then used high resolution LIDAR data to create a canopy density map across Mont St. Hilaire. In May of 2003 multiple-return LIDAR data were collected for the reserve from a Piper Navajo C-GOVX aircraft from a height of 1300m (at a repetition rate of 50000 Hz, a scanning frequency of 37Hz and a scanning range +/- 15o).The total number of above-ground LIDAR hits was divided by the total number of LIDAR hits (above-ground plus last return) in 30 m x 30 m pixels to compute a canopy density map measuring light penetration to the ground layer. Sun-exposed habitat was categorized as non-wet habitat pixels with less than 75% canopy density. Dry-shadedhabitat was categorized as non sun-exposed and non-wet habitat pixels on slopes of a 20 degree incline or more (i.e. high drainage). Moistshaded habitat was categorized as non-wet habitat pixels with greater than 80% canopy density, on slopes with less than a 10 degree incline. Slopes with an inclination of between 10 and 20 degrees were filtered out, as were canopy densities between 75% and 80%.We calculated the number of patches, average patch size and average (Euclidian) interpatch distances for each CWD habitat type using Fragstats (McGarigal et al. 2002).Slope steepness and slope aspect for Mont St. Hilaire were derived from the Canadian Digital Elevation Data 66 (GovernmentofCanada 2000). Topography variables were scaled from a resolution of 22.5 m to a resolution of 30 m to match the TM satellite imagery. Results Performance of VIs for Predicting CWD Influx NDVI performed the best for both training and validation in our cross-validation trials (Table 3). It explained an average of 20% of the variance in training datasets and 37% of the variance in validation datasets. NDVIgreen was a close second, explaining 19% and 35% of the variance in training and validation datasets (respectively). The poorest performing VI was VARI, explaining 4% and 19% of the variance in training and validation sets (respectively). The average relationship between NDVI and CWD in our training datasets was: [3] Applying this relationship to hill-wide NDVI data, we created a CWD influx prediction map for Mont St. Hilaire (Figure 1). On average, the 1998 storm created an influx of 1.7 kg/m2 (standard deviation = 0.97 kg/m2) of CWD across Mont St. Hilaire (16.8 metric tons per hectare); 16057 tons of CWD were produced across the entire hill. Independent Validation of CWD Influx Predictions The 18 independent validation sites were predicted, based on fitting a regression to 1998 data, to gain between 0.75 kg/m2 and 3.2 kg/m2 of CWD from the ice 67 storm (mean = 1.6 kg/m2, standard deviation = 0.65 kg/m2). There was a significant positive relationship between the measured amount of young CWD and the predicted biomass of CWD across independent validation sites collected in 2008 (Figure 2, R2 = 0.32, p < 0.02). Categorization of CWD Influx into Different Habitat Types In total, 59.5% of the Mont St. Hilaire was categorized as likely to contain sunexposed, dry shaded, moist shaded, or wet (Figure 3). The remaining 40.5% was in pixels with intermediate topography or canopy cover and as such could not be assigned to one of the four habitat types. There were 334 tons of sun-exposed CWD across 22 hectares of sun-exposed habitat (average = 1.5 kg/m2). This habitat was arranged in 130 patches (average area = 1700 m2); the average interpatch distance was 128 m (standard deviation =95 m). There were 4559 tons of dry-shaded CWD across 310 hectares of dry-shaded habitat (average = 1.4 kg/m2). This habitat was arranged in 63 patches (average area = 48000 m2); the average interpatch distance was 77 m (standard deviation = 32 m). There were 2851 tons of moist-shaded CWD across 150 hectares of moist-shaded habitat (average = 1.9 kg/m2). This habitat was arranged in 163 patches (average area = 9100 m2); the average interpatch distance was 83 m (standard deviation = 34 m). There were 1057 tons of wet CWD across 70 hectares of wet habitat (average = 1.5 kg/m2). This habitat was arranged in 71 patches (average area = 9500 m2); the average interpatch distance was 74 m (standard deviation = 65 m). 68 Methods Evaluation According to the LIDAR canopy cover analysis, among the 82 training/validation sites the maximum canopy cover at a 113 m2 resolution was 91%; the minimum canopy cover was 2%; the mean percent cover was 32% (standard deviation = 20%). When canopy cover values were recalculated based on a geographically weighted mean at a 900 m2 resolution, the maximum canopy cover among the 82 training/validation sites was 68%; the minimum percent cover was 4%;the mean percent cover was 37% (standard deviation = 13%). These two datasets were significantly positively correlated with a slope of 1.0 (p < 0.001, R2 = 0.44). Discussion Accuracy of VIs in Predicting Spatial Pattern of CWD The use of VIs to map the CWD influx in a forest following an ice storm is largely unexplored. The CWD influx we calculated (16.8 tons / ha) is very similar to the Hooper et al. (2001) estimate of CWD increase at St. Hilaire following the ice storm (19.9 tons / ha). Our NDVI validation dataset produced a moderate R2 value (a mean R2 of 0.37) but the Methods Evaluationsection indicates that the correlation between data at 113 m2 and 900 m2 had an R2 = 0.44 (i.e. 44% of the variance explained). Thus, even if the prediction data at 113 m2 was perfect (as in the case of our LIDAR Methods Assessment), 56% of the explanatory power in our prediction models is likely to be lost by changing resolution from 113 m2 to 69 900 m2. In this context, our predictive NDVI-based model performed very well, accounting for 84% of the possible variance (i.e. 37% out of a projected possible 44%). The validity of this relationship was further supported by the relationship between the predicted and observed CWD at our 18 independent sample sites (R2 of 0.32). The unexplained variance in this relationship (68%) is in part attributed to the accuracy of our NDVI prediction model (as explained above), but it can also be attributed to the difficulty in determining the precise temporal origin of a piece of CWD found in on the forest floor. Site microclimate conditions like moisture, standing water, humidity and wind can slow or accelerate decomposition (Harmon et al. 1986), weakening the relationships apparent in our data. Damaged trees also can die but remain upright as snags which can decay at different rates than fallen CWD depending on the groundlayer microclimate conditions (particularly moisture) and the relative presence of saproxylic insects, microbes, birds and fungus contributing to decomposition and mechanical breakdown (Harmon et al. 1986). Pieces of CWD of different sizes and different wood densities also decay at different rates (Enrong et al. 2006, Harmon et al. 1986). CWD Benefits for Mont St. Hilaire Species The massive influx of CWD provides extra habitat for a diversity of taxa across the Mont St. Hilaire reserve. Perhaps the most obvious taxa to benefit from the habitat increase are those that directly contribute to the decomposition of CWD 70 – white rot fungi, brown rot fungi and wood boring saproxylic insects – which increase in abundance and diversity with increasing amounts of CWD (Jonsell et al. 1998, Kaila et al. 1997, Kaila et al. 1994, Yee et al. 2004). As CWD goes through escalating stages of decay large downed logs become important nursery habitat for seedlings, herbs and ferns that may otherwise be unable to colonize the forest floor (Clark et al. 1999). Throughout the decay process, CWD is also important habitat for many vertebrate species. Among Mont St. Hilaire mammals, the Woodland Jumping Mouse (Napeozapus insignis) and the Southern Red-backed Vole (Clethrionomys gapperi), are known to use CWD shelter habitat rather than relying primarily on tunnel-digging.Similarly, the Redbacked Salamander (Plethodon cinereus) is dependent on CWD, particularly in its juvenile amphibian stages (Richmond and Trombulak 2009). Insectivorous bird diversity is higher in heavily storm-damaged areas, but in many cases this is likely an indirect response of bird populations to an increased overall abundance of insects (Faccio 2003, Torgersen and Bull 1995). Because CWD directly benefits some forest-dwelling species more than others, we expect many of these biotic impacts to be spatially aligned with the areas of the reserve there the highest storm impact occurred. Our results show that the biggest concentrations of CWD influx are the eastern and southeastern slopes of the mountain and the low-lying area west of Lac Hertel (Figure 1). In addition, there are pockets of relatively high CWD influx scattered elsewhere throughout the reserve. 71 The creation of highly-connected CWD networks across the mountain can have additional impacts on some forest-dwelling species.Our results indicated that approximately 50% of the mountain received more than 1.7 kg/m2 of CWD influx and nearly 85% of the mountain received at least 0.7 kg/m2. These networks can act as transportation corridors for small mammals and saproxylic insects within and between habitat patches allowing an increase in both foraging territory size and dispersal distances (McMillan and Kaufman 1995, Schiegg 2000, Schiegg 2000). The Benefits of Different Habitat Types to Insects The high volume of CWD across a clear moisture gradient at Mont St. Hilaire may further benefit saproxylic habitat specialist species. In total, we documented 4893 tons of new CWD in dry habitat (sun-exposed and shaded) and 3908 tons of new CWD in moist and wet habitat. Ødegard (2006) showed that saproxylic insects were more diverse and abundant in wet forest stands than in dry forests stands. The very high volume of CWD in wet, moist and dry habitat at Mont St. Hilaire may encourage the persistence of this gradient. The baseline volume of CWD expected in old growth forests similar to Mont St. Hilaire is about 20 tons per hectare (Muller 2003). If this baseline were combined with our calculation of 8800 tons of new CWD in 552 hectares of moisture-defined habitat it yields ~20000 tons of CWD at St. Hilaire in habitats that can be defined by moisture content. It is unknown if this incredible volume of CWD results in a wider 72 moisture gradient or merely a taller distribution tightly clustered around the mean. Continued research on saproxylic distributions, diversity and CWD moisture conditions at Mont St. Hilaire is needed to shed more light on these patterns. The sparse sun-exposed habitat on the mountain (22 ha) and its comparatively low cumulative CWD volume (334 tons) suggests that the 1998 ice storm may not have had a profound long-term impact on saproxylic insects specific to these habitats. Generally speaking, the richness and abundance of sun-loving saproxylic beetles tends to vary with gap size. Larger gaps support more species and more individuals due to the increase in sun-exposure, wood volume, structural diversity and microhabitat diversity that they provide (Bouget 2005). In Sweden, 131 of 446 red-listed (i.e. endangered or threatened) saproxylic coleoptera species exclusively used sun-exposed CWD habitat rather than semi-shaded or shaded habitat (Jonsell et al. 1998). Bouget (2005) suggests that gaps need to be a minimum of 0.50 ha in size in order to support sustainable populations of habitat-specific saproxylic insect populations. Since the average sun-exposed patch size (i.e. gap size) we documented was 0.17 ha, the Mont St. Hilaire open habitat patches resultant from the storm may be too small to support these types of saproxylic populations. This small patch-size for sun-exposed CWD is radically different from what persisted in the summer immediately following the storm. Arii and Lechowicz (2007) showed that the opening of the canopy cover at Mont-St 73 Hilaire increased twofold compared to the pre-storm canopy cover measures (from an average of 11% to 22%). However, they and others also show that this canopy opening is short lived, and that the canopy recovers to pre-storm conditions within 3 to 7 years of disturbance (Arii and Lechowicz 2007, Beaudet et al. 2007, Darwin et al. 2004). It is unclear whether or not a short-term burst of sun-exposed CWD habitat like this can result in significant benefits for sun-loving saproxylic species. The LIDAR data from 2003 shows that the canopy at Mont St. Hilaire is very closed. In fact, at a 30m resolution, only two pixels (0.18 ha) had a canopy density of less than 50% and an additional nine pixels (0.81 ha) had a canopy density of less than 60%. CWD Influx and Species Movements In terms of dispersal, the increase in CWD at Mont St. Hilaire is more likely to benefit animal populations within the mountain than to encourage extramountain colonization from other forest fragments. Mont St. Hilaire is an isolated forest habitat situated within a matrix of agriculture, urban settlement and roads. The closest large forested areas are Mont Rougemont (15 km2 of forest, 7 km to the southeast) and Mont St. Bruno (4 km2 of forest, 9 km to the west). There are several smaller fragments within 2-3 kilometers of the reserve, but it is improbable that these small fragments would act as source populations for colonization to Mont St. Hilaire. Dispersal distances of a few to several kilometers are unrealistic for most vertebrates that use CWD (Bowman et al. 74 2002) – not to mention the extra difficulty they face of dispersing through suboptimal (matrix) habitat. This is almost certainly true for salamanders, which are not known as long-distance dispersers. For flightless saproxylic insects, colonization over distances greater than 500 m is uncommon (den Boer 1970). Even for full-winged beetles, depending on air currents, weather conditions and matrix properties, dispersal of more than a few kilometers is rare (Rainus 2006). Data from Burke and Goulet (1998) show that the diversity of native beetle species in deciduous forest fragments is linked forest fragment size, particularly for patches that are isolated at distances of more than 2 km from source populations. Among animals, the obvious exception to dispersal limitation in this case are avifauna where the distance between Mont St. Hilaire and surrounding habitat patches is within the natal and breeding dispersal ranges of many bird species, particularly migrants (Greenwood and Harvey 1982, Paradis et al. 1998). Within the Mont St. Hilaire, many species dependent on CWD as habitat won’t necessarily be constrained to the CWD habitat types as we have defined them; for species that are constrained to these specific habitats, our analyses suggests that within-mountain dispersal is a strong possibility. The average distance between moist-shaded debris patches (83 m) and dry-shaded CWD patches (77 m) are manageable dispersal distances for many forest-dwelling species. In addition, because the matrix-habitat in these cases is often very similar to preferred-habitat (i.e. old-growth temperate forest habitat) successful migration from one patch to another may occur more frequently. 75 Conclusion We have shown that the geospatial patterns of ice storm damage can be accurately predicted across forest landscapes using remotely sensed NDVI. The 1998 ice storm provided a large influx of CWD, which provided an array of CWD habitats across local environmental gradients. Although canopy cover has long since returned to pre-storm levels, the complex geospatial CWD pattern that was produced across the Mont St. Hilaire shows a change in forest floor habitat dynamics that favour deadwood-reliant fungi, insects, birds, small mammals and salamanders. Acknowledgements We are very grateful to M.C. Hooper and K. Arii who conducted the bulk of the data collection and conducted many of the initial Mont St. Hilaire 1998 ice storm studies. We would like to thank T. Work, M. Kalacska, R. Feldman, J. Messier and S. Estrada for their helpful reviews and suggestions. M. VonButtlar and R. MacKenzie helped with the deadwood collection data and L. Herzig helped with 76 the LIDAR data compilation and computations. This research was made possible by funds provided by the Natural Sciences and Engineering Research Council of Canada; PW would also like to thank Richard Tomlinson, T-PULSE and B. Alters for generously providing supplementary funding opportunities. References Arii, K. 2004. Ecology of American beech and sugar maple in an old-growth forest. Department of Biology. - McGill University. Arii, K. and Lechowicz, M. J. 2007. Changes in understory light regime in a beechmaple forest after a severe ice storm -Canadian Journal of Forest Research 37: 1770-1776. 77 Bannari, A., Morin, D., Bonn, F. and Huete, A. R. 1995. A review of vegetation indices. - Remote Sensing Reviews 13: 95 - 120. Beaudet, M., Brisson, J., Messier, C. and Gravel, D. 2007. Effect of a major ice storm on understory light conditions in an old-growth Acer-Fagus forest: Pattern of recovery over seven years. - Forest Ecology and Management 242: 553-557. Birky, A. K. 2001. NDVI and a simple model of deciduous forest seasonal dynamics. - Ecological Modelling 143: 43-58. Boerner, R. E. J., Runge, S. D., Cho, D.-S. and Kooser, J. G. 1988. Localized ice storm damage in an Appalachian Plateau watershed. - American Midland Naturalist 119: 199-208. Bouget, C. 2005. Short-term effect of windstorm disturbance on saproxylic beetles in broadleaved temperate forests: Part II. Effects of gap size and gap isolation. - Forest Ecology and Management 216: 15-27. Bouget, C. and Duelli, P. 2004. The effects of windthrow on forest insect communities: a literature review. - Biological Conservation 118: 281-299. Bowman, J., Jaeger, J. A. G. and Fahrig, L. 2002. Dispersal distance of mammals is proportional to home range size. - Ecology 83: 2049-2055. Braccia, A. and Batzer, D. P. 2001. Invertebrates associated with woody debris in a southeastern U.S. forested floodplain wetland. - Wetlands 21: 18-31. 78 Bragg, D. C., Shelton, M. G. and Zeide, B. 2003. Impacts and management implications of ice storms on forests in the southern United States. Forest and Ecology Management 186: 99-123. Brommit, A. G., Charbonneau, N., Contreras, T. A. and Fahrig, L. 2004. Crown loss and subsequent branch sprouting of forest trees in response to a major ice storm. - Journal of Torrey Botanical Society 131: 169-176. Burke, D. and Goulet, H. 1998. Landscape and area effects on beetle assemblages in Ontario. - Ecography 21: 472-479. Canadian Forest Service. 2008. Canada's National Forest Inventory Ground Sampling Guidelines. Natural Resources Canada, Ottawa, Ontario. Chandler, G. and Markham, B. 2003. Revised Landsat-5 TM radiometric calibration procedures and postcalibration dynamic ranges. - Ieee Transactions on Geoscience and Remote Sensing 41: 2674-2677. Clark, J. S., Beckage, B., Camill, P., Cleveland, B., HilleRisLambers, J., Lichter, J., McLachlan, J., Mohan, J. and Wyckoff, P. 1999. Interpreting recruitment limitation in forests. - Am. J. Bot. 86: 1-16. Cranston, P. S. and McKie, B. 2006. Aquatic wood - an insect perspective. - In: Grove, S. and Hanula, J. L. (eds.), Proceedings of a Symposium for the 22nd International Congress of Entomology. - United States Department of Agriculture. 79 Darwin, A. T., Ladd, D., Galdins, R., Contreras, T. A. and Fahrig, L. 2004. Response of forest understory vegetation to a major ice storm. - Journal of the Torrey Botanical Society 131: 45-52. den Boer, P. J. 1970. On the significance of dispersal power for populations of Carabid-beetles (Coleoptera, Carabidae). - Oecologia 4: 1-28. DeSteven, D. D., Kline, J. and Matthiae, P. E. 1991. Long-term changes in a Wisconsin Fagus-Acer forest in relation to glaze storm disturbance. Journal of Vegetation Science 2: 201-208. Dugay, S. M., Arii, K., Hooper, M. and Lechowicz, M. J. 2001. Ice storm damage and early recovery in an old-growth forest. - ENvironmental Monitoring and Assessment 67: 97-108. Enrong, Y., Xihua, W. and Jianjun, H. 2006. Concept and classification of coarse woody debris in forest ecosystems -Frontiers of Biology in China 1: 76-84. Faccio, S. D. 2003. Effects of ice storm-created gaps on forest breeding bird communities in central Vermont. - Forest Ecology and Management 186: 133-145. Fassnacht, K. S., Gower, S. T., MacKenzie, M. D., Nordheim, E. K. and Lillesand, T. M. 1997. Estimating the leaf area index of north central Wisconsin forests using the Landsat Thematic Mapper. - Remote Sensing of Environment 61: 229-245. 80 Gitelson, A. A., Kaufman, Y. J., Stark, R. and Rundquist, D. 2002. Novel algorithms for remote estimation of vegetation fraction. - Remote Sensing of Environment 80: 76-87. Gitelson, A. A. and Merzlyak, M. N. 1998. Remote sensing of chlorophyll concentration in higher plant leaves. - Advances in Space Research 22: 689-692. Greenwood, P. J. and Harvey, P. H. 1982. The natal and breeding dispersal of birds. - Annual Review of Ecology and Systematics 13: 1-21. Hajj, M. E., Agnès, d., Lafrance, B., Hagolle, O., Dedieu, G. and Rumeau, M. 2008. Relative radiometric normalization and atmospheric correction of a SPOT 5 time series. - Sensors 8: 2774-2791. Harmon, M. E., Franklin, F. E., Swanson, F. J., Sollins, P., Gregory, S. V., Lattin, J. D., Anderson, N. H., Cline, S. P., Aumen, N. G., Sedell, J. R., Lienkaemper, G. W., Cromack, K. and Cummins, K. W. 1986. Ecology of coarse woody debris in temperate ecosystems. - In: Advances in Ecological Research. Acedemic Press, Inc., pp. 168-188. Hooper, M. C., Arii, K. and Lechowicz, M. J. 2001. Impact of a major ice storm on an old-growth hardwood forest. - Canadian Journal of Botany 79: 70-75. Huete, A., Didan, K., Miura, T., Rodriguez, E. P., Gao, X. and Ferreira, L. G. 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. - Remote Sensing of Environment 83: 195-213. 81 Irland, L. C. 1998. Ice storm 1998 and the forests of the northeast: A preliminary assessment. - Journal of Forestry 96: 32-40. Irland, L. C. 2000. Ice storms and forest impacts. - The Science of the Total Environment 262: 231-242. Jonsell, M., Weslien, J. and Ehnstrom, B. 1998. Substrate requirements of redlisted saproxylic invertebrates in Sweden. - Biodiversity & Conservation 7: 749-764. Kaila, L., Martikainen, P. and Punttila, P. 1997. Dead trees left in clear-cuts benefit saproxylic Coleoptera adapted to natural disturbances in boreal forest -Biodiversity & Conservation 6: 1-18. Kaila, L., Martikainen, P., Punttila, P. and Yakovlev, E. 1994. Saproxylic beetles (Coleoptera) on dead birch trunks decayed by different polypore species. - Annales Zoologici Fennici 31: 97-107. Kaufman, Y. J. and Tanre, D. 1992. Atmospherically resistant vegetation index (ARVI) for EOS-MODIS. - Ieee Transactions on Geoscience and Remote Sensing 30: 261-270. Keisker, D. G. 2000. Types of wildlife trees and coarse woody debris required by wildlife of north-central British Columbia. –Research Branch of the Ministry of Forestry, Victoria, BC. King, D. J., Olthof, I., Pellikka, P. K. E., Seed, E. D. and Butson, C. 2005. Modelling and mapping damage to forests from an ice storm using remote sensing and environmental data. - Natural Hazards 35: 321-342. 82 Liu, H. Q. and Huete, A. 1995. A feedback based modification of the NDVI to minimize canopy background and atmospheric noise. - Ieee Transactions on Geoscience and Remote Sensing 33: 457-465. Liu, Z. Y., Huang, J. F., Wu, X. H. and Dong, Y.P. 2007. Comparison of vegetation indices and red-edge parameters for estimating grassland cover from canopy reflectance data. - Journal of Integrative Plant Biology 49: 299306. McGarigal, K., Cushman, S. A., Neel, M. C. and Ene, E. 2002. FRAGSTATS: Spatial Pattern Analysis Program for Categorical Maps. McMillan, B. R. and Kaufman, D. W. 1995. Travel path characteristics for freeliving white-footed mice (Peromyscus lecopus). - Canadian Journal of Zoology 73: 1474-1478. McNab, W. H. and Roof, T. 2006. Evaluation of Ikonos satellite imagery for detecting ice storm damage to oak forests in eastern Kentucky. - In: Prisley, S., Bettinger, P., Hunh, I.-K. and Kushla, J. (eds.), Proceedings of the 5th Southern Forestry and Natural Resources GIS Conference, Asheville, NS. - Warren School of Forestry and Natural Resources, University of Georgia, Athens, GA. Millward, A. A. and Kraft, C. E. 2004. Physical influences of landscape on a largeextent ecological disturbance: the northeastern North American ice storm of 1998. - Landscape Ecology 19: 99-111. 83 Milton, J. and Bourque, A. 1999. A climatological account of the January 1998 ice storm in Quebec. - Atmospheric Sciences and Climate Monitoring Division, Environment Canada, Ville Ste. Laurent, Quebec. Muller, R. N. 2003. Landscape patterns of change in coarse woody debris accumulation in an old-growth deciduous forest on the Cumberland Plateau, southeastern Kentucky. - Canadian Journal of Forest Research 33: 763-769. National Climate Data and Information Archive. 2008. Canadian Climate Normals or Averages 1961-1990 (Rougemont, Quebec). -Environment Canada, Fredricton, New Brunswick. Newton, A. C. 2007. Forest Ecology. - Oxford University Press. Ødegaard, F. 2006. Host specificity, alpha- and beta-diversity of phytophagous beetlesin two tropical forests in Panama. - Biodiversity & Conservation 15: 83-105. Olthof, I., King, D. J. and Lautenschlager, R. A. 2004. Mapping deciduous forest ice storm damage using Landsat and environmental data. - Remote Sensing of Environment 89: 484-496. Paradis, E., Baillie, S. R., Sutherland, W. J. and Gregory, R. D. 1998. Patterns of natal and breeding dispersal in birds. - Journal of Animal Ecology 67: 518536. 84 Payero, J. O., Neale, C. M. U. and Wright, J. L. 2004. Comparison of eleven vegetation indices for estimating plant height of alfalfa and grass. Applied Engineering in Agriculture 20: 385-393. Pettorelli, N., Vik, J. O., Mysterud, A., Gaillard, J.-M., Tucker, C. J. and Stenseth, N. C. 2005. Using the satellite-derived NDVI to assess ecological responses to environmental change. - Trends in Ecology & Evolution 20: 503-510. Popescu, S. C., Wynne, R. H. and Nelson, R. F. 2002. Estimating plot-level tree heights with lidar: local filtering with a canopy-height based variable window size. - Computers and Electronics in Agriculture 37: 71-95. Proulx, O. J. and Greene, D. F. 2001. The relationship between ice thickness and northern hardwood tree damage during ice storms. - Canadian Journal of Forest Research 31: 1758-1767. Rainus, T. 2006. Measuring the dispersal of saproxylic insects: a key characteristic for their conservation. - Population Ecology 48: 177-188. Rhoads, A. G., Hamburg, S. P., Fahey, T. J., Siccama, T. G., Hane, E. N., Battles, J., Cogbill, C., Randall, J. and Wilson, G. 2002. Effects of an intense ice storm on the structure of a northern hardwood forest. - Canadian Journal of Forest Research 32: 1763-1775. Richardson, A. J. and Everitt, J. H. 1992. Using spectral vegetation indices to estimate rangeland productivity. - Geocarto International 7: 63-69. 85 Richmond, L. S. and Trombulak, S. C. 2009. Distribution of Red-backed Salamander (Plethodon cinereus) with respect to cover-object characteristics in the Green Mountains of Vermont. - Northeastern Naturalist 16: 13-26. Rotheray, G. E. and Stuke, J.-H. 1998. Third stage larvae of four species of saproxylic Syrphidae (Diptera), with a key to the larvae of British Criorhina species. - Entomologist's Gazette 49: 209-217. Roujean, J. L. and Breon, F. M. 1995. Estimating PAR absorbed by vegetation from bidirectional reflectance measurements. - Remote Sensing of Environment 51: 375-384. Rouse, J. W. 1973. Monitoring the vernal advancement and retrogradation of natural vegetation. NASA/GSFCT Type II Report. Rouse, J. W., Haas, R. W., Schell, J. A., Deering, D. W. and Harlan, J. C. 1974. Monitoring the vernal advancement and retrogradation (Greenwave effect) of natural vegetation. NASA/GSFCT Type III Final Report. Schiegg, K. 2000. Are there saproxylic beetle species characteristic of high dead wood connectivity? - Ecography 23: 579-587. Schiegg, K. 2000. Effects of dead wood volume and connectivity on saproxylic insect species diversity-Ecoscience 7: 290-298. St-Onge, B. A. and Achaichia, N. 2001. Measuring forest canopy height using a combination of LIDAR and aerial photography data. - International Archives of Photogrammetry and Remote Sensing 34: 131-137. 86 Stueve, K. M., Lafon, C. W. and Isaacs, R. E. 2007. Spatial patterns of ice storm disturbance on a forested landscape in the Appalachian Mountains, Virginia. - Area 39: 20-30. Takahashi, K., Arii, K. and Lechowicz, M. J. 2007. Quantitative and qualitative effects of a severe ice storm on an old-growth beech-maple forest. Canadian Journal of Forest Research 37: 598-606. Torgersen, T. R. and Bull, E. L. 1995. Down logs as habitat for forest-dwelling ants - the primary prey of Pileated Woodpeckers in northeastern Oregon. Northwest Science 69: 294-303. USGS. 2007. EarthExplorer. Earth Resources Observation and Science (EROS) Center. U.S. Department of the Interior and the U.S. Geological Survey, Sioux Falls, South Dakota. Vanderwel, M. C., Malcolm, J. R. and Smith, S. M. 2006. An integrated model for snag and downed woody debris decay class transitions. - Forest Ecology and Management 234: 48-59. Vandyke, O. 1999. A literature review of ice storm impacts on forests in Eastern North America. SCSS Technical Report #112. - Ontario Ministry of Natural Resources, Southcentral Sciences Section, pp. 1-29. Weeks, B. C., Hamburg, S. P. and Vadeboncoeur, M. A. 2009. Ice storm effects on the canopy structure of a northern hardwood forest after 8 years. Canadian Journal of Forest Research 39: 1475-1483. 87 Wolter, P. T., Mladenhoff, D. J., Host, G. E. and Crow, T. R. 1995. Improved forest classification in the Northern Lake States using multi-temporal Landsat imagery. - Photogrammetric Engineering and Remote Sensing 61: 1129. Yee, M., Grove, S., Richardson, A. and Mohammed, C. 2004. Brown rot in inner heartwood: why large logs support characteristic saproxylic beetle assemblages of conservation concern. - In: Grove, S. and Hanula, J. L. (eds.), Proceedings of a Symposium for the 22nd International Congress of Entomology. - United States Department of Agriculture, pp. 42-56. Zhan, Z. M., Qin, Q. M., Abduwasit, G. and Wang, D. D. 2007. NIR-red spectral space based new method for soil moisture monitoring. - Science in China Series B: Chemistry 50: 283. Zielonka, T. 2006. Quantity and decay stages of coarse woody debris in oldgrowth subalpine spruce forests of the western Carpathians, Poland. Canadian Journal of Forest Research 36: 2614-2622. 88 Figure 1Coarse woody debris influx at Mont St. Hilaire resulting from the 1998 ice storm. Mean coarse woody debris influx was 1.7 kg/m2 (standard deviation = 0.97 kg/m2), for a total estimated input of 16.8 metric tons per hectare. Coarse woody debris input values are spatially correlated up to 450 meters. 89 Figure 2The predicted biomass of coarse woody debris resulting from the 1998 ice storm is positively correlated to the amount of young (category 1 and 2) coarse woody debris that was measured in the summer of 2008 at 18 randomly located sites across Mont St. Hilaire. 90 Figure 3Concentrations of coarse woody debris at Mont St. Hilaire in (a) dry shaded habitat, (b), moist shaded habitat, (c) sun-exposed habitat and (d) wet habitat. 91 Table 1Landsat-5 satellite imagery used in calculation of vegetation indices for Mont St. Hilaire in 1996 and 1998. Scene Identifier Sun Elevation Sun Azimuth LT50140281996218XXX02 Acquisition Date 8/05/1996 51.0 o 128.4 o LT50140281998239XXX02 8/27/1998 48.5 o 142.6 o 92 Table 2Vegetation Indices (VIs) used to calculate forest damage after the 1998 ice storm at Mont St. Hilaire. The VIs incorporated four wavelengths of light: blue (450 – 520 nm), green (520 – 600 nm), red (630 – 690 nm) and near infrared (NIR, 760 – 900 nm). Name Equation Normalized Difference Vegetation Index (NDVI) Enhanced Vegetation Index (EVI) Difference Vegetation Index (DVI) Renormalized Difference Vegetation Index (RDVI) Atmospherically Resistant Vegetation Index (ARVI) Green Normalized Difference Vegetation Index (NDVIgreen) Visible Atmospheric Resistant Index (VARI) 93 Table 3The data for each regression model were split into random training and validation datasets (80/20 ratio, 66/16 plots) to create ten cross validation trials for each Vegetation Index. The R2 fit of each validation dataset is given respective of the regression coefficient and intercept calculated in its training dataset. NDVI provides the best for both training and validation datasets (R2 = 0.20 and 0.37, respectively). 94 Cross Validation Trial # 1 2 3 4 5 6 7 8 9 10 Average Training Dataset R2 Validation Dataset R2 NDVI EVI DVI RDVI ARVI NDVIgreen VARI NDVI EVI DVI RDVI ARVI NDVIgreen VARI 0.15 0.07 0.21 0.24 019 0.23 0.29 0.24 0.24 0.16 0.20 0.08 0.02 0.13 0.16 0.11 0.16 0.18 0.15 0.17 0.09 0.13 0.03 0.01 0.10 0.13 0.10 0.14 0.15 0.13 0.14 0.05 0.10 0.09 0.02 014 0.18 0.14 0.18 0.22 0.18 0.19 0.09 0.14 0.10 0.03 0.18 0.21 0.13 0.20 0.23 0.20 0.17 0.09 0.15 0.14 0.04 0.19 0.28 0.19 0.22 0.32 0.22 0.23 0.11 0.19 0.02 0.02 0.03 0.08 0.04 0.03 0.03 0.04 0.03 0.04 0.04 0.56 0.69 0.34 0.19 0.38 0.33 0.20 0.24 0.28 0.46 0.37 0.54 0.69 0.28 0.06 0.31 0.18 011 0.23 0.14 0.41 0.30 0.54 0.71 0.28 0.14 0.25 0.05 0.14 0.16 0.08 0.40 0.28 0.57 0.74 0.32 0.16 0.34 0.11 0.17 0.21 0.15 0.47 0.32 0.53 0.55 0.16 0.07 037 0.04 0.18 0.18 0.25 0.43 0.28 0.61 0.70 0.45 0.02 0.35 0.25 0.03 0.25 0.25 0.54 0.35 0.30 0.10 0.24 0.00 0.12 0.36 0.24 0.18 0.22 0.15 0.19 95 Linking Statement 1 In Chapter 1 I examined the geospatial and biotic impacts of an abiotic natural disturbance on a remnant forest. I generated detailed biotic impact data and it was evident that the patterns of influx for different types of coarse woody debris were highly heterogeneous. The long-term impact of this will be variable for different saproxylic guilds and other taxa reliant on specific forms of coarse woody debris. Remnant forest fragments are subjected to many forms of disturbance, both natural and human in origin. To best understand how disturbances shape forest communities the second category of abiotic disturbances, human-caused, needs to be examined. In Chapter 2 I explore this by analyzing the relationship between recreational trails and a forest insect assemblage. 96 CHAPTER2: Human-Disturbance and Caterpillars in Managed Forest Fragments Peter J.T. White1, Brian J. McGill2 and Martin J. Lechowicz1 1 McGill University, Department of Biology 1205 Dr. Penfield Ave., Montreal, Quebec H3A 1B1, Canada 2 University of Maine, School of Biology & Ecology Deering Hall 303, Orono, ME 04469, U.S.A. *This chapter was first published by the above authors in Biodiversity and Conservation (2011) 97 Abstract The impact of forest-edge habitat on caterpillar assemblages has been wellstudied, but the impact of trailside habitat has rarely been considered. We surveyed caterpillar populations in relation to recreational trails in forest fragments in southeastern Quebec, Canada. We found a consistent negative relationship between trails in the forest and both the abundance and species richness of caterpillars within and among forest fragments. Conversely, caterpillar presence was not related to the nearby presence of favourable host trees.We suggest that the negative effect of trails may be due to increased predation pressure in trailside habitat and to conditions that make trailside habitat less preferable for oviposition. These results underscore the importance of managing trails to limit the amount of intra-forest disturbance inforest fragment remnants. 98 Introduction The temperate deciduous forests of eastern North American have been subjected to widespread habitat destruction over the past two hundred years (Drushka 2003, Hannah et al. 1995). The forest fragments that remain have high levels of edge-related effects that can have a significant impact on many forestdwelling organisms(Alverson et al. 1988, Murica 1995, Wilcove 1985). Habitat destruction has been particularly widespread in the St. Lawrence Valley of southeastern Quebec, Canada where most of the historic mixedwood forests have been cleared in favour of agricultural development. The few forest fragments that remain are critical habitat for many forest-dwelling species (Warman et al. 2004).Since many of these remnant forest fragments are in close proximity to urban areas they experience high volumes of human traffic on both designated and unofficial hiking trails. Trailside Habitat in Forests Much has been made of the necessity to protect forest fragments from external threats (such as cutting and isolation), but there has been comparatively little focus on internal threats to forest fragments such as recreational hiking or walking trails.Trailside habitat is distinct from forest-edge habitat because the latter marks a transition to a different habitat type (e.g. agriculture, open field, urban area, etc.) whereas the former often does not. For many taxa, trailside communities differ significantly in species composition from forest interior 99 communities, the formeroften characterized by a high proportion of earlysuccession, disturbance-tolerant and invasive species (Dickens et al. 2005, Hall and Kuss 1989). Trails can facilitate soil-compaction via trampling that adversely affects root development and the growth of trailside woody plants (Bhuju and Ohsawa 1998). Trails can also lead to increased soil erosion, muddiness and vegetation trampling in trailside habitat (Bhuju and Ohsawa 1998, Dale and Weaver 1994, Farrell and Marion 2001). The few studies that have examined the impact of trailside habitat on forest-dwelling species often have uncovered a pattern similar to what is found associated with edge habitat. Recreational trails generally have a negative effect on small-mammal populations (Boyle and Samson 1985, Meaney et al. 2002, Sauvajot et al. 1998) – a relationship that is typically mirrored by small mammal populations in forest-edge habitat (Bayne and Hobson 1998, Miller and Hobbs 2000, Wolf and Batzli 2004); but see also (Anderson et al. 2003). Certain bat species are well known to prefer both forest-edge and trailside habitat over forest-interior habitat (Krusic et al. 1996, Patriquin and Barclay 2003). Many carabid beetles often favour heavily-trampled trailside habitat over forestinterior habitat (Grandchamp et al. 2000, Raymond et al. 2002), a similar pattern occurring in forest-edge habitat (Magura 2002, Magura and Tothmeresz 1997, Molnár et al. 2001; but see alsoDavies and Margules 1998). Among birds, many disturbance-tolerant species tend to favour trailside habitat over forest-interior habitat (Miller et al. 1998) in contrast to forest-edge habitat where the overall 100 species richness and abundance of the bird assemblage can be significantly higher than the forest-interior (Best et al. 1990, Strelke and Dickson 1980). An Analogy of Forest Edge Habitat The impact of forest-edge habitat on forest-dwelling moth populations has also been well-examined, but little research exists on the effects of trailside habitat. Insight from edge-effect literature on forest-dwelling moths can help inform an a priori hypothesis about the impact of trails. For example, in boreal forest habitat, Mönkkönen and Mutanen (2003) found no difference in moth species richness or abundance in edge versus non-edge (interior) habitat, which may suggest that trails could have little or no impact. Conversely, in tropical areas, Arctiidae moths have been found to be significantly more species rich in recovering secondary forests (edge-like habitat) compared to mature forest (non edge-like habitat)(Fiedler et al. 2007, Noske et al. 2009). Noske et al (2009) argue that this pattern may be because recovering forest stands tend to have more habitat niches than mature forest stands – a feature that could certainly be applicable to trailside habitat. In temperate regions, Summerville and Crist (2004) show that small forest fragments can have higher than expected moth richness in cases where there is a high host plant richness to offset the loss of forest area. A similar pattern could be predicted for trailside habitat especially if there is an increase in host plant richness driven by an increase in disturbance-tolerant plants. This higher-than-expected richness typically results from a greater 101 proportion of matrix-dwelling moths, sometimes at the expense of the forestdwellers (Summerville 2004). Fortin and Maufette (2001) found better caterpillar performance (bigger pupae, higher larval and pupal survivorship) and larger egg masses in edge habitat and connected their findings to an increased nutritional quality of edge trees. Roland (1993) uncovered higher caterpillar abundance at forest edges but suggested that the pattern may be due to an edge-associated decrease in top-down parasitoid or pathogen pressure. Although there are a plethora of proposed mechanisms (niche availability, host plant richness, host plant nutritional quality, parasitoid/pathogen pressure), the general trend shows either a neutral effect or a positive effect of edge or edge-like habitat on caterpillar species richness and abundance. Although many of these mechanisms could act as described in trailside habitat, no studies to our knowledge have examined the pattern of caterpillar richness or abundance associated with forest trails. Hypotheses In this study we therefore test two hypotheses about the drivers of caterpillar assemblages in relation to trailsides within forest fragments: 1. Trailside habitat is beneficial to caterpillar assemblages. We pose this hypothesis based on the set of forest-edge literature showing that there is a positive relationship between forest-edge habitat and moth species (richness or abundance). Presuming that trailside habitat is analogous to forest- 102 edge habitat, we predict caterpillar species richness and abundance to be higher in trailside habitat than in forest-interior habitat. 2. Host plant availability determines caterpillar species richness and abundance. Because the forest fragments we examine in this study are intensively managed we decided to put the impact of trails in a habitat context, to control for the impact of habitat suitability. We therefore examine whether caterpillar occurrence is related to host plant availability (i.e. the quantity of host plants at a given quadrat) and predict that quadrats with higher host plant availability will have more diverse and abundant caterpillar assemblages. Methods Study Area We studied caterpillar assemblages associated with the Monteregian Hills (Feininger and Goodacre, 1995) in the St. Lawrence River valley in southeastern Quebec, Canada (45°30'N, 73°30'W to 45°24'N, 72°35'W; Figure 1).During presettlement times all these Monteregian Hills would have had broadly similar forests embedded in a more or less continuously forested landscape (Richard and Grondin 2009), but now they exist as a series of large remnant forest fragments isolated in the developed landscape.We established eighteen 400 m2 (20 m x 20 m) quadrats in the forests at each of four sampling sites on Monteregian Hills. 103 Our first site was at Parc Mont Royal - an urban park in middle on Montreal, a city of 3.5 million inhabitants. A much-disturbed secondary broadleaf deciduous forest comprises about 100 ha of the 190 ha park on this Monteregian Hill in city centre. Many trees were cut in the 1950s and early 1960s (Brunel et al. 2005) and more than 70,000 trees were subsequently replanted up to the early 1990s in an attempt to restore the forest. Currently, the park has more than 110,000 trees and more than 80 species of woody plants (Brunel et al. 2005). Our second site was at Parc National du Mont St. Bruno – a protected provincial park in the eastern suburbs of greater Montreal. The forest at Mont St. Bruno is a broadleaf deciduous forest covering more than 500 ha of the 790 ha park. Aerial photography records indicate that 60 or more ha of forest were cut prior to the 1940s in the northern part of the park, and subsequently replanted or allowed to regrow (National Air Photo Library 1950). The forest has more than 85 species of woody plants. Our third site was at the Gault Nature Reserve on Mont St. Hilaire – a protected park and UNESCO Biosphere Reserve 38 km east of Montreal. The forest at Mont St. Hilaire is an old-growth broadleaf deciduous forest covering most of the 1000 ha reserve. This site has a long history of protection dating back to the 1600s (Arii 2004, Maycock 1961), and is recognized as a biodiversity hotspot not only for vascular plants (Holland 1980, Karst and Lechowicz 2007), but also moths and butterflies (Handfield 1999), mammals (Grant 1976, 104 Humphries et al. 2003), birds (Harris and Lemon 1974, Quellet 1967), and herpetofauna(Denman and Lapper 1964). Our fourth site was at Mont Shefford – one of the easternmost Monteregian Hills and well outside greater Montreal. At the time of this study different parts of the site nonetheless were in varying states of disturbance and urban development. Numerous small houses and some maple sugar bushes occur along several paved two-lane roads that run through the centre of Mont Shefford and there is a ski-hill on the northern slope as well. We conducted caterpillar sampling at three sub-sites around the hill. Six quadrats were in a 100 ha semi-disturbed patch of forest on the west side of Mont Shefford and six quadrats in a 25 ha patch of forest on the east side set aside as a forested community park. A quarter of the former site was a sugar bush and there were numerous trails used by deer-hunters throughout the area. The community park had several trails spread out within the park, but off-trail use was discouraged. Six quadrats were in a third sub-site, Parc de la Yamaska, a provincial park just north of Mont Shefford. Our quadrats were in the east-portion of the 450 ha semi-disturbed park that has several trails and a high volume of pedestrian traffic (ParcsQuebec 2010). 105 Trail Index Calculation We created a geospatial variable called trail index to measure the impact of trails at every 1 m2 pixel p across an entire site: [1] n trail index p = ∑ t =1 wt d pt where the width w (in meters) of trail tis divided by distance dpt , the distance between a given pixel p (at a 30 meter resolution) and trail t in meters. Trail index was calculated only for pixels that had at least one dpt less than 50 meters, the distance at which a given pixel p is considered to be forest interior habitat (Matlack 1993). Thus, for each pixel p, trail index is calculated as the cumulative impact of all ntrails within 50 meters of that pixel.Pixels where all dptdistances were greater than 50 meters were given a trail index value of zero. In our data trail index values scale from 0.02 (a pixel 50 meters away from a 1 meter trail) to 6.0 (a pixel 1 meter away from a 6 meter trail) for each trail (Figure2). We created a map of trail index for each of the four study sites based on documented trail data (Bossé 2005, CantonShefford 2010, CentredelaNatureMontSaintHilaire 2007, LesAmisDeLaMontagne 2008, ParcsQuebec 2010, ParcsQuebec 2010).There were no existing trail data for our first sub-site at Mont Shefford, so we georeferenced the trails that we observed adjacent to our six study quadrats in this area. Finally, we also calculated an index of overall trail impact for each 400 m2 quadratby averaging the trail index values of the pixels within the given 20 m x 20 m quadrat boundary. 106 Caterpillar Surveying and Identification Most studies examining moths in ecological settings use light traps to capture adult morphs, which yields large sample sizes from a wide area that can encompass multiple habitats. Beck and Linsenmair(2006)calculate that the attraction radius of light trapping is typically around 15 meters, although this may vary between an open and dense forest understory (but see alsoBaker and Sadovy 1978). For the purposes of surveying moth populations adjacent to recreational trails, caterpillar sampling via tree-beating(Futuyma and Gould 1979)is a more direct measure of the association between individuals and the trailside trees.We therefore focus our sampling on the caterpillar life-stage to ensure that the species we document are forest-dwelling rather than migrant from any adjacent matrix-habitat Caterpillars were collected at 18 quadrats at each of the four study sites. At each of the 72 quadrats, ten sugar maple (Acer saccharum) trees between 3 and 10 cm dbh were sampled by striking the bole and lower branches of each tree ten times with a 20 oz, 30" aluminum baseball bat and catching dislodged caterpillars on a sheet. Caterpillar collections were made three times at each quadrat (between June 1 and June 6, July 4 and July 9 and August 3 and August 6 in 2009), yielding a total of 21600 tree-strikes (10 strikes/tree x 10 trees/plot x 18 quadrats/site x 4 sites/study region x 3 survey windows). Macrolepidopteran moth caterpillars were identified to species with a dissecting microscope using Wagner’s Caterpillars of Eastern North America Field 107 Guide(2005). Microlepidopteran moths were counted and identified only to morphospecies for lack of an accurate identification guide. Macrolepidopteran moths collected in early instars were reared so that positive identifications could be made. Flightless moth species (with wingless or flight-limited adult female morphs) were retained in our dataset because of their ability to select host plants by tree-to-tree movement of late instars and aerial ballooning of first instars (Barbosa et al. 1989, Bell et al. 2005). In our data these included Lymantria dispar, Alsophila pometaria, Operophtera bruceata, Phigalia titea, Orgyia definita, and Orgyia leucostigma. Site Host Plant Availability All trees greater than 1 cm diameter at breast height (dbh) were counted and identified to species at each quadrat. Host plant preferences for macroLepidoptera were taken from Handfield’s Le Guide des Papillons du Quebec(Handfield 1999) and cross-referenced with Wagner’s account (2005). Host plant availability was calculated for each quadrat using the basal area measurement of each quadrat tree as an index of foliage biomass (Tucker et al. 1993). Specifically, the host plant availability of each quadrat was calculated as: m [2] hostplant availability = 108 n ∑∑ b i =1 j =i m ij where bij is the basal area in meters-squared for the jth of ni host plant species at a quadrat for the ith among m macromoth caterpillar species observed we identified in our study region. Statistical Analyses We used a general linear model (SPSS Inc. 2000) to examine how host plant availability and trail index relate to caterpillar richness and abundance across quadrats: [3] caterpillar species = trail index + hostplant availability Trail index, caterpillar species richness and caterpillar abundance were 1+ log10 transformed to improve normality. We also tested the probability of random distribution of caterpillars across our study quadrats for any caterpillar species found at 20% or more of the quadrats. For each of these caterpillar species we calculated the species-specific host plant availability at each quadrat. We then categorized quadrat host plant availability into one of five equal-interval categories: Very Low: 0-0.250 m2, Low: 0.251-0.500 m2, Medium: 0.501-0.750 m2, High: 0.751-1.00 m2 and Very High: > 1.00 m2. The expected occupancy of each species in each category was calculated as the total number of quadrats a given species was observed to occupy multiplied by the proportion of quadrats that fell into each of the above host plant availability categories.Because the distribution of a given species could incorporate as few as 15 quadrats (20% of 72), we used a randomization 109 test for goodness-of-fit (McDonald 2009) to see whether abundant caterpillar species were non-randomly distributed with respect to site host plant availability. P-values were calculated based on 10,000 replicates of distribution for each caterpillar across the five categories. Results We collected 75 species across our four sites, 36 macrolepidopteran moths (420 individuals, Appendix A1) and 39 microlepidopteran moths (253 individuals, Appendix A2). Of these, 21 species were collected on Mont Royal (11 macro, 10 micro), 46 on Mont St. Bruno (24 macro, 22 micro), 37 species on Mont St. Hilaire (20 macro, 17 micro) and 43 species on Mont Shefford (21 macro, 22 micro). Trail Index and Host Plant Availability Trails were most prevalent across Mont Royal where 95% of the forested area had non-zero trail index values (Figure 3); the average trail index was 0.164 across the entire site and 0.188 when averaged among quadrats. Trails were least prevalent at the Mont St. Hilaire site where only 29% of forested area had non-zero trail index values; the average trail index was 0.021 across the entire site and 0.015 when averaged among quadrats. The Mont St. Bruno and Mont Shefford sites were intermediate in terms of trail presence with 56% and 36% of forested areas having non-zero trail index values, respectively.Trail indices were 0.047 averaged across St. Bruno and 0.042 averaged across Shefford; trail index 110 values when averaged among quadrats were 0.028 for St. Bruno and 0.038 for Shefford. Average host plant availability for macromoths across all quadrats at each site ranged from 0.11 m2 to 1.34 m2 basal area (mean = 0.63 m2, standard deviation = 0.24) (see Appendix A3 for complete list of trees surveyed). Average host plant availability did not vary by site (ANOVA among sites, p = 0.80). Across all quadrats, trail index was a significant (negative) predictor of caterpillar richness and abundance; site host plant availability was a near-significant (positive) predictor of caterpillar richness and abundance (Table 1). Performing regression analyses on a site-by-site basis, trail index remained significant at all four sites whereas host plant availability was significant only at the Mont Shefford (Table 1). Random Versus Non-Random Distribution of Caterpillar Species Seven of the observed 36 macrolepidopteran moth species were present at more than 20% of the study quadrats: Cyclophora pendulinaria, Itame pustularia, Lambdina fiscellaria, Lithophane antennata, Lymantria dispar, Melanolophia canadaria and Morrisonia latex. Of these, only I. pustularia had a distribution across quadrats significantly different from random expectation (p = 0.0094, Appendix A4).I. pustularia occurred at high host plant availability quadrats more frequently than expected and low host plant availability quadrats less frequently than expected. 111 Discussion We show a consistent negative relationship between trail presence and caterpillar abundance and richness. This relationship was present both within and among the four forest fragments we examined. This relationship is contrary to our hypothesis that trailside habitat would be associated with an increase in caterpillar abundance and/or richness. We also did not find evidence that host plant availability was limiting caterpillar abundance or species richness across our sites. There was no consistent link between host plant availability and caterpillar abundance or richness across quadrats, and only one of the seven most abundant caterpillar species was distributed across quadrats nonrandomly in relation to host plant availability. Possible Mechanisms of Negative Relationship A few mechanisms may interact to explain the significant negative relationship between trails and caterpillars. First, changes in trailside microsite conditions may make potential host plants less attractive to gravid female moths (resulting in fewer caterpillars). Often, host selection by moths occurs via chemodetection of volatile organic compounds (VOC) emanating from candidate host plants (Mphosi 2007, Pophof et al. 2005, Shields and Hildebrand 2001). Changes in habitat temperature and moisture can result in changes to tree VOC emission levels (Tollsten and Müller 1996) potentially making them more or less attractive for oviposition. The assortment of unique habitat conditions at trailside habitat 112 (e.g. higher understory light levels, muddier terrain, more compact soil and more understory herbaceous vegetation) could therefore result in trailside trees having different VOC levels that attract fewer moths than interior-forest trees. This mechanism could also apply to species whose females are flightless or flightlimited as adults, resulting in redispersal by ballooning in early instars or intertree migration in later instars in Alsophila pometaria, Operophtera bruceata, Phigalia titea, Orgyia definita, Orgyia leucostigma and Lymantria dispar. Additionally, increased light levels in trailside habitat may lead to higher leaf toughness levels which can decrease host plant suitability for caterpillars (Choong 1996, Feeny 1970). Secondly, the aversion to trailside host trees may in part be driven by top-down predation or parasitism. In our system, insectivorous birds, bats, omnivorous rodents, coleopteran beetles and parasitic hymenoptera are major caterpillar predators that have been observed to play a role in regulating caterpillar populations (Barbosa et al. 2001, Grushecky et al. 1998, Lill et al. 2002, Medina and Barbosa 2002, Sanz 2001). While trailside habitat does not always benefit small mammal populations, caterpillar populations could experience high predation and parasitism rates from disturbance-tolerant (insectivorous) coleoptera, birds, bats and (parasitic) hymenoptera (Grandchamp et al. 2000, Holzschuh et al. 2009, Krusic et al. 1996, Miller et al. 1998). Thirdly, it is possible that overwintering egg and larval mortality is higher in trailside habitat due to lethal microclimatic conditions. High winter winds, 113 trampling and tracking for cross-country skiing all can remove insulating snow cover along trails (Leonard 1972) and the resulting exposure to extreme freezing conditions can lead to high rates of egg and larva death (Brokerhof et al. 1993, Cooke and Roland 2003, Layne and Peffer 2006, Waggoner 1985). This may seem like an unlikely explanation, particularly as edge habitat (which would be exposed to similar conditions) has been found to have high egg masses, larval and pupal densities in some caterpillar species (Bellinger et al. 1989, Fortin and Mauffette 2001, Roland 1993) – but survivorship in leeward versus windward edges has not been tested. If certain trailside habitats experience significantly harsher wintertime conditions then springtime caterpillar populations may be affected. Trails Versus Edges Many of the potential mechanisms driving reduced caterpillar presence in trailside habitat could arguably be applied to forest-edge habitat, yet forest-edge habitat often shows a positive rather than negative relationship with caterpillar abundance and richness. There are, however, two fundamental differences between forest-edge habitat and trailside habitat that could, in this case, account for the difference. In forest-edges moth populations transition from forest-interior species to edge-tolerant species as the habitat type transitions from undisturbed forest interior to disturbed successional edge-habitat (Fiedler et al. 2007, Noske et al. 2009). In many cases the abundance and richness of the 114 edge-dwelling moth assemblage is augmented by species from an adjacent soft matrix (e.g.Ricketts et al. 2001). For moths, a soft matrix would be agricultural habitat with sparse tree and shrub cover (Fischer et al. 2005). This type of habitat is not necessarily hostile to forest-dwelling moths, but generally lacks the host plants required to support stable populations. Similarly, for open-field moth species, a soft matrix is a forest-edge habitat that has a low abundance of suitable host- or food-plant species (Summerville and Crist 2003). Therefore, because forest-edge habitat is flanked by two distinct moth assemblages (forestdwelling and field-dwelling) that often occupy different habitat niches, abundance and richness tend to be higher. In contrast, trails that cut through forest interior habitat do not mark a transition to open field soft-matrix conditions. This was documented in a tropical forest system by Fiedler et al. (2007) who showed that Geometrid and Pyraloid moth richness decreased in edge habitat whereas Arctiid or Sphingid moth richness often increased. A second consideration is that thesurvey methods we used in this study are not the standard light-trapping methods most often adopted for Lepidoptera population studies in forest-edge habitat. Our tree beating methodology gives the advantage of surveying caterpillar populations at very specific locations in proximity to trails and allows us to control for microhabitat (host plant) characteristics. Gaston (1988) summarizes the many limitations of light trapping, including the inability to associate individual moths to specific microhabitats. This makes it very difficult to sample edge populations independent of matrix- 115 populations. Furthermore, the sampling effectiveness of light-traps deployed across a large region can be quite variable due to fluctuations in wind, cloud cover, moonlight and temperature from one site to another (McGeachie 1989, Yela and Holyoak 1997). Conclusion The strong negative association of trails with forest-dwelling moth richness and abundance suggests that management of forest fragments should focus on limiting proliferation of recreational-trails as a primary conservation activity. In our dataset, the host plant availability of a given site was rarely important in predicting moth richness or abundance, suggesting that tree abundance and richness maintenance could be designated a second-tier priority in Monteregian forest fragments. We don’t suggest that a healthy tree assemblage is not important for forest communities but rather that limiting intra-forest disturbance due to trail networks may be more important. Ours is not the first study to document the negative impact of recreational trails on diverse organisms in forest communities – significant and negative impacts have been widely reported (e.g.Bhuju and Ohsawa 1998, Dale and Weaver 1974, Dickens et al. 2005, Farrell and Marion 2001, Hall and Kuss 1989, Leung and Marion 1999, Miller et al. 1998, Wilson and Seney 1994). Ours is however one of the first to document a negative relationship specifically associated with moths, a popular indicator species of forest disturbance (New 1997, Summerville et al. 2004). 116 Leung and Marion (1999) note that closemonitoring of trail networks is needed in places where human disturbance is present and that pedestrian traffic should be limited to few well-defined trails. Both boardwalks and physical barriers can be effective for directing and minimizing the impact of pedestrian traffic on natural habitats (Doucette and Kimball 1990, Zhou and Tachibana 2004) but they can be costly to construct and maintain. However, forest fragments that have a high degree of human visitation – particularly those fragments that are managed with a mandate to protect wildlife and biodiversity for future generations – may require these types of extra measures to ensure that trails have a minimum impact on the surrounding habitat. Acknowledgements We would like to thank R. Feldman, J. Messier and S. Estrada for their helpful reviews and suggestions. M. VonButtlar and R. MacKenzie helped with the data collection. This research was made possible by funds provided by the Natural Sciences and Engineering Research Council of Canada. We would also like to thank D. Rodrigue and A. Mochon from Parcs Québec for their help at Parc National du Mont-Saint-Bruno and Parc National de la Yamaska. 117 References Alverson, W. S., Waller, D. M. and Solheim, S. L. 1988. Forests too deer: Edge effects in northern Wisconsin. - Conservation Biology 2: 348-358. Anderson, C. S., Cady, A. B. and Meikle, D. B. 2003. Effects of vegetation structure and edge habitat on the density and distribution of whitefooted mice (Peromyscus leucopus) in small and large forest patches. Canadian Journal of Zoology 81: 897-904. Arii, K. 2004. Ecology of American beech and sugar maple in an old-growth forest. Department of Biology. - McGill University. Atlas of Canada. 2010. Toporama – Online Topographic Maps. Natural Resources Canada, Ottawa, Ontario. Baker, R. R. and Sadovy, Y. 1978. The distance and nature of the light-trap response of moths. - Nature 276: 818-821. Barbosa, P., Krischik, V. and Lance, D. 1989. Life-history traits of forest-inhabiting flightless Lepidoptera. - American Midland Naturalist 122: 262-274. Barbosa, P., Segarra, A., Gross, P., Caldas, A., Ahlstrom, K., Carlson, R. W., Ferguson, D. C., Grissell, E. E., Hodges, R. W., Marsh, P. M., Poole, R. W., Schauff, M. E., Shaw, S. R., Whitfield, J. B. and Woodley, N. E. 2001. Differential parasitism of macrolepidopteran herbivores on two deciduous tree species. - Ecology 82: 698-704. Bayne, E. M. and Hobson, K. A. 1998. The effects of habitat fragmentation by forestry and agriculture on the abundance of small mammals in the 118 southern boreal mixedwood forest. - Canadian Journal of Zoology 76: 6269. Beck, J. and Linsenmair, K. E. 2006. Feasibility of light-trapping in community research of moths: Attraction radius of light, completeness of samples, nightly flight times and seasonality of Southeast-Asian hawkmoths (Lepidoptera: Sphingidae). - Journal of Research on Lepidoptera 39: 1836. Bell, J. R., Bohan, D. A., Shaw, E. M. and Weyman, G. S. 2005. Ballooning dispersal using silk: world fauna, phylogenies, genetics and models. Bulletin of Entomological Research 95: 69-114. Bellinger, R. G., Ravlin, F. W. and McManus, M. L. 1989. Forest edge effects and their influence on Gypsy Moth (Lepidoptera: Lymantriidae) egg mass distribution. - Environmental Entomology 18: 840-843. Best, L. B., Whitmore, R. C. and Booth, G. M. 1990. Use of cornfields by birds during the breeding season: The importance of edge habitat. - American Midland Naturalist 123: 84-99. Bhuju, D. R. and Ohsawa, M. 1998. Effects of nature trails on ground vegetation and understory colonization of a patchy remnant forest in an urban domain. - Biological Conservation 85: 123-135. Bossé, D. 2005. Municipalité du Canton de Shefford, Plan de Zonage. - Bonsant, D. (ed.). - Groupe Teknika, Shefford, Quebec. 119 Boyle, S. A. and Samson, F. B. 1985. Effects of noncomsumptive recreation on wildlife: A review. - Wildlife Society Bulletin 13: 110-116. Brokerhof, A. W., Morton, R. and Jonathan Banks, H. 1993. Time-mortality relationships for different species and developmental stages of clothes moths (Lepidoptera: Tineidae) exposed to cold. - Journal of Stored Products Research 29: 277-282. Brunel, S., Poitras, C., Burges, J., Dufour, M., Ghafouri, M., Chouinard, A., Mongrain, G., Dumais, H. and Turcotte, S. 2005. Étude de caractérisation de l’arrondissement historique et naturel du Mont-Royal. Commission des biens culturels du Québec, Québec, Québec. Canton Shefford. 2010. Carte Sentiers Parc Ecologique 2. - La Municipalité du Canton de Shefford, Shefford, Quebec. Centre de la Nature Mont Saint Hilaire. 2007. Gault Nature Reserve Forest Cover Type Map. - Centre de la Nature Mont Saint-Hilaire, McGill University. Choong, M. F. 1996. What Makes a Leaf Tough and How This Affects the Pattern of Castanopsis fissa Leaf Consumption by Caterpillars. - Functional Ecology 10: 668-674. Cooke, B. J. and Roland, J. 2003. The effect of winter temperature on forest Tent Caterpillar (Lepidoptera: Lasiocampidae) egg survival and population dynamics in northern climates. - Environmental Entomology 32: 299-311. 120 Dale, D. and Weaver, T. 1974. Trampling effects on vegetation of the trail corridors of north Rocky Mountain forests. - Journal of Applied Ecology 11: 767-772. Dale, D. and Weaver, T. 1994. Trampling effects on vegetation of the trail corridors of north Rocky Mountain forests. - Journal of Applied Ecology 11: 767-772. Davies, K. F. and Margules, C. R. 1998. Effects of habitat fragmentation on carabid beetles: experimental evidence. - Journal of Animal Ecology 67: 460-471. Denman, N. S. and Lapper, I. S. 1964. The herpetology of Mont St.-Hilaire, Rouville County, Quebec, Canada. - Herpetologica 20: 25-30. Dickens, S. J. M., Gerhardt, F. and Collinge, S. K. 2005. Recreational portage trails as corridors facilitating non-native plant invasions of the Boundary Waters Canoe Area Wilderness (U.S.A.). - Conservation Biology 19: 16531657. Doucette, J. E. and Kimball, K. D. 1990. Passive trail management in northeastern alpine zones: A case study. - In: Moore, T. A., Donnelly, M. P., Graefe, A. R. and Vaske, J. J. (eds.), Proceedings of the 1990 Northeastern Recreation Research Synposium. - United States Department of Agriculture, Forest Service, pp. 195-201. Drushka, K. 2003. Canada's forests : a history. - Forest History Society, McGillQueen's University Press. 121 Farrell, T. A. and Marion, J. L. 2001. Identifying and assessing ecotourism visitor impacts at eight protected areas in Costa Rica and Belize. - Environmental Conservation 28: 215-225. Feeny, P. 1970. Seasonal changes in oak leaf tannins and nutrients as a cause of spring feeding by Winter Moth caterpillars. - Ecology 51: 565-581. Feininger, T., Goodacre, A.K. 1995. The eight classical Monteregian hills at depth and their mechanism of intrusion.- Canadian Journal of Earth Sciences 32: 1350-1364. Fiedler, K., Hilt, N., Brehm, G. and Schulza, C. H. 2007. Moths at tropical forest margins - how mega-diverse insect assemblages respond to forest disturbance and recovery. - In: Tscharntke, T., Leuschner, C., Zeller, M., Guhardja, E. and Bidin, A. (eds.), The stability of tropical rainforest margins, linking ecological, economic and social constraints of land use and conservation. Springer Verlag, pp. 39-60. Fischer, J., Fazey, I., Briese, R. and Lindenmayer, D. 2005. Making the matrix matter: challenges in Australian grazing landscapes. - Biodiversity & Conservation 14: 561-578. Fortin, M. and Mauffette, Y. 2001. Forest edge effects on the biological performance of the forest tent caterpillar (Lepidoptera: Lasiocampidae) in sugar maple stands. - Ecoscience 8: 164-172. Futuyma, D. J. and Gould, F. 1979. Associations of plants and insects in deciduous forest. - Ecological Monographs 49: 33-50. 122 Gaston, K. J. 1988. Patterns in the local and regional dynamics of moth populations. - Oikos 53: 49-57. Grandchamp, A. C., Niemelä, J. and Kotze, J. 2000. The effects of trampling on assemblages of ground beetles (Coleoptera, Carabidae) in urban forests in Helsinki, Finland. - Urban Ecosystems 4: 321-332. Grant, P. R. 1976. An 11-year study of small mammal populations at Mont St. Hilaire, Quebec. -Canadian Journal of Zoology 54: 2156-2173. Grushecky, S. T., Liebhold, A. M., Greer, R. and Smith, R. L. 1998. Does forest thinning affect predation on Gypsy Moth (Lepidoptera: Lymantriidae) larva and pupae? - Environmental Entomology 27: 268-276. Hall, C. N. and Kuss, F. R. 1989. Vegetation alteration along trails in Shenandoah National Park, Virginia. - Biological Conservation 48: 211-227. Handfield, L. 1999. Le Guide Des Papillons Du Quebec. - Broquet Inc. Hannah, L., Carr, J. L. and Lankerani, A. 1995. Human disturbance and natural habitat: a biome level analysis of a global data set. - Biodiversity & Conservation 4: 128-155. Harris, M. A. and Lemon, R. E. 1974. Songs of song sparrows: reactions of males to songs of different localities. - The Condor 76: 33-44. Holland, P. G. 1980. Transplant experiments with Trout Lily at Mont St Hilaire, Quebec. - Journal of Biogeography 7: 261-267. 123 Holzschuh, A., Steffan-Dewenter, I. and Tscharntke, T. 2009. Grass strip corridors in agricultural landscapes enhance nest-site colonization by solitary wasps. - Ecological Applications 19: 123-132. Humphries, M. M., Kramer, D. L. and Thomas, D. W. 2003. The role of energy availability in mammalian hibernation: an experimental test in freeranging Eastern Chipmunks. - Physiological and Biochemical Zoology 76: 180-186. Karst, A. L. and Lechowicz, M. J. 2007. Are correlations among foliar traits in ferns consistent with those in the seed plants? - New Phytologist 173: 306-312. Krusic, R. A., Yamasaki, M., Neefus, C. D. and Pekins, P. J. 1996. Bat habitat use in White Mountain National Forest. - The Journal of Wildlife Management 60: 625-631. Layne, J. R. and Peffer, B. J. 2006. The influence of freeze duration on postfreeze recovery by caterpillars of Pyrrharctia isabella (Lepidoptera: Arctiidae): when is survival enough to qualify as recovery? - Journal of Experimental Zoology Part A: Comparative Experimental Biology 305A: 570-575. Leonard, D. E. 1972. Survival of Gypsy Moth populations exposed to low winter temperatures. - Environmental Entomology 1: 549-554. Les Amis de la Montagne. 2008. Interactive Map of Mont Royal. - Les Amis de La Montagne, Montreal, Quebec. 124 Leung, Y.F. and Marion, J. L. 1999. The influence of sampling interval on the accuracy of trail impact assessment. - Landscape and Urban Planning 43: 167-179. Lill, J. T., Marquis, R. J. and Ricklefs, R. E. 2002. Host plants influence parasitism of forest caterpillars. - Nature 417: 170-173. Magura, T. 2002. Carabids and forest edge: spatial pattern and edge effect. Forest Ecology and Management 157: 23-37. Magura, T. and Tothmeresz, B. 1997. Testing edge-effect on carabid assemblages in an oak-hornbeam forest. - Acta Zoologica Academiae Scientiarum Hungaricae 43: 303-312. Matlack, G. R. 1993. Microenvironment variation with and among forest edge sites in the eastern United States. - Biological Conservation 66: 185-194. Maycock, P. F. 1961. Botanical studies on Mont St. Hilaire, Rouville County, Quebec. - Canadian Journal of Botany 39: 1293-1325. McDonald, J. H. 2009. Handbook of Biological Statistics. - Sparky House Publishing, pp. 52-56. McGeachie, W. J. 1989. The effects of moonlight illuminance, temperature and wind speed on light-trap catches of moths. - Bulletin of Entomological Research 79: 185-192. Meaney, C. A., Ruggles, A. K., Clippinger, N. W. and Lubow, B. C. 2002. The Impact of Recreational Trails and Grazing on Small Mammals in the Colorado Piedmont. - Prairie Naturalist 34: 115-136. 125 Medina, R. F. and Barbosa, P. 2002. Predation of small and large Orgyia leucostigma (J. E. Smith) (Lepidoptera: Lymantriidae) larvae by vertebrate and invertebrate predators. - Environmental Entomology 31: 1097-1102. Miller, J. R. and Hobbs, N. T. 2000. Recreational trails, human activity, and nest predation in lowland riparian areas. - Landscape and Urban Planning 50: 227-236. Miller, S. G., Knight, R. W. and Miller, C. K. 1998. Influence of recreational trails on breeding bird communities. - Ecological Applications 8: 162-169. Molnár, T., Magura, T., Tóthmérész, B. and Elek, Z. 2001. Ground beetles (Carabidae) and edge effect in oak-hornbeam forest and grassland transects. - European Journal of Soil Biology 37: 297-300. Mönkkönen, M. and Mutanen, M. 2003. Occurrence of moths in boreal forest corridors. - Conservation Biology 17: 468-475. Mphosi, M. S. 2007. Comparative host selection and reproductive physiology for sunflower moth, Homoeosoma electellum (Hulst) (Lepidoptera: Pyralidae), and banded sunflower moth, Cochylis hospes Walsingham (Lepidoptera: Tortricidae) Entomology. - North Dakota State University, p. 198. Murica, C. 1995. Edge effects in fragmented forests: implications for conservation. - Trends in Ecology & Evolution 10: 58-62. National Air Photo Library. 1950. Roll Number A12956: photos 259, 260, 264. Natural Resources Canada, Ottawa, Ontario. 126 New, T. R. 1997. Are Lepidoptera an effective 'umbrella group' for biodiversity conservation? - Journal of Insect Conservation 1: 5-12. Noske, N. M., Hilt, N., Werner, F., Brehm, G., Fiedler, K., Sipman, H. J. M. and Gradstein, S. R. 2009. Disturbance effects on diversity of epiphytes and moths in a montane forest in Ecuador. - Basic and Applied Ecology 9: 412. Parcs Quebec. 2010. Parc National de la Yamaska, 2010-2011. - Réseau Sépaq, Quebec, Quebec. Parcs Quebec. 2010. Parc National du Mont-Saint-Bruno, 2010-2011. - Réseau Sépaq, Quebec, Quebec. Patriquin, K. J. and Barclay, R. M. R. 2003. Foraging by bats in cleared, thinned and unharvested boreal forest. - Journal of Applied Ecology 40: 646-657. Pophof, B., Stange, G. and Abrell, L. 2005. Volatile organic compounds as signals in a plant–herbivore system: Electrophysiological responses in olfactory sensilla of the moth Cactoblastis cactorum. - Chemical Senses 30: 51-68. Quellet, H. 1967. The distribution of the Cerulean Warbler in the province of Quebec, Canada. - The Auk 84: 272-274. Raymond, B., Vanbergen, A., Watt, A., Hartley, S. E., Corey, J. S. and Hails, R. S. 2002. Escape of pupal predation as a potential cause of outbreaks of the Winter Moth, Operophtera brumata. - Oikos 98: 219-228. 127 Richard, P.J.H. and Grondin, P. 2009. Histoire postglaciaire de la végétation, pp. 169-176, in Manuel de Foresterie, 2è édition, Ordre des ingénieurs forestiers du Québec, Éditions MultiMondes, Québec. Ricketts, T. H., Daily, G. C., Ehrlich, P. R. and Fay, J. P. 2001. Countryside biogeography of moths in a fragmented landscape: biodiversity in native and agricultural habitats. - Conservation Biology 15: 378-388. Roland, J. 1993. Large-scale forest fragmentation increases the duration of tent caterpillar outbreak. - Oecologia 93: 25-30. Sanz, J. J. 2001. Experimentally increased insectivorous bird density results in a reduction of caterpillar density and leaf damage to Pyrenean oak. Ecological Research 16: 387-394. Sauvajot, R. M., Buechner, M., Kamradt, D. A. and SChonewald, C. E. 1998. Patterns of human disturbance and response by small mammals and birds in chaparral near urban development. - Urban Ecosystems 2: 279297. Shields, V. D. C. and Hildebrand, J. G. 2001. Responses of a population of antennal olfactory receptor cells in the female moth Manduca sexta to plant-associated volatile organic compounds -Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology 186: 1135-1151. SPSS Inc. 2000. SYSTAT Version 10. - IBM Corporation. 128 Strelke, W. K. and Dickson, J. G. 1980. Effect of forest clear-cut edge on breeding birds in east Texas. - Journal of Wildlife Management 44: 559-567. Summerville, K. S. 2004. Do smaller forest fragments contain a greater abundance of Lepidopteran crop and forage consumers? - Population Ecology 33: 234-241. Summerville, K. S. and Crist, T. O. 2003. Determinants of lepidopteran community composition and species diversity in eastern deciduous forests: roles of season, eco-region and patch size. - Oikos 100: 134-148. Summerville, K. S. and Crist, T. O. 2004. Contrasting effects of habitat quantity and quality on moth communities in fragmented landscapes. - Ecography 27: 3-12. Summerville, K. S., Ritter, L. M. and Crist, T. O. 2004. Forest moth taxa as indicators of lepidopteran richness and habitat disturbance: a preliminary assessment. - Biological Conservation 116: 9-18. Tollsten, L. and Müller, P. M. 1996. Volatile organic compounds emitted from beech leaves. - Phytochemistry 43: 759-762. Tucker, G. F., Lassoie, J. P. and Fahey, T. J. 1993. Crown architecture of standgrown sugar maple (Acer saccharum Marsh.) in the Adirondack Mountains. - Tree Physiology 12: 297-310. Waggoner, P. E. 1985. How gypsy moth eggs freeze. - Agricultural and Forest Meteorology 36: 43-53. 129 Wagner, D. L. 2005. Caterpillars of Eastern North America. - Princeton University Press. Warman, L. D., Forsyth, M., Sinclair, A. R. E., Freemark, K., Moore, H. D., Barrett, T. W., Pressey, R. L. and White, D. 2004. Species distributions, surrogacy, and important conservation regions in Canada. - Ecology Letters 7: 374379. Wilcove, D. S. 1985. Nest predation in forest tracts and the decline of migratory songbirds. - Ecology 66: 1211-1214. Wilson, J. P. and Seney, J. P. 1994. Erosional impact of hikers, horses, motorcycles, and off-road bicycles on mountain trails in Montana. Mountain Research and Development 14: 77-88. Wolf, M. and Batzli, G. 2004. Forest edge - high or low quality habitat for WhiteFooted Mice (Peromyscus leucopus)? - Ecology 85: 756-769. Yela, J. L. and Holyoak, M. 1997. Effect of moonlight and meteorological factors on light and bait trap catches of Noctuid moths (Lepidoptera: Noctuidae). - Environmental Entomology 26: 1283-1290. Zhou, J. and Tachibana, H. 2004. Natural revegetation after elimination of disturbance of human treading in the Tennyogahara Mire, the Taisetsu Mountains, Japan. - Vegetation Science 21: 65-78. 130 Figure 1We sampled at four sites in the St. Lawrence River valley of southern Quebec, Canada (Figure adapted from Atlas of Canada 2010), each progressively farther from the center of Montreal. Urban development is highest in the centre of the Montreal Metropolitan Area decreasing towards the edges of the dashed boundary that marks the geographical limits of greater-Montreal. The urban development gives way to farmlands eastward in the St-Lawrence Lowland, and then extensive forests in the Appalachian Highlands. 131 Figure 2Trail index calculation for a given pixel p at distance d from trail t that has a width w. Index shown untransformed (a) and log-transformed (b). 132 Figure 3Trail index across our four study sites: (A) Mont Royal, (B) Mont St. Bruno, (C) Mont St. Hilaire and (D1, D2, D3) Mont Shefford. The geospatial arrangement in this Figure does not reflect the regional geolocations of the sites (see Figure 1). 133 Table 1Trail Index is a consistent negative predictor of caterpillar abundance and caterpillar species richness both across and within study sites.Hostplant availability is rarely significantly linked to caterpillar species abundance or richness, only explaining a large degree of caterpillar variance at Mont Shefford. Near-significant (*) and significant (**) p-values are marked. 134 Dependent Variable Independent Variable Caterpillar Abundance Caterpillar Species Richness Trail Index Site Host Plant Availability Trail Index Site Host Plant Availability Trail Index Caterpillar Abundance Site Host Plant Availability Trail Index Caterpillar Species Richness Site Host Plant Availability Site All All St. Bruno St. Hilaire Royal Shefford St. Bruno St. Hilaire Royal Shefford St. Bruno St. Hilaire Royal Shefford St. Bruno St. Hilaire Royal Shefford 135 Standardized Coefficient -0.66 0.17 -0.64 0.17 -0.592 -0.697 -0.473 -0.450 0.214 -0.356 -0.005 0.570 -0.498 -0.520 -0.479 -0.368 0.189 -0.222 0.059 0.453 Partial R2 0.434 0.030 0.416 0.030 0.315 0.465 0.223 0.196 0.042 0.122 0.001 0.322 0.223 0.259 0.228 0.131 0.032 0.047 0.004 0.199 p-value 1.8 x 10-10** 0.054* 5.7 x 10-10** 0.055* 0.019** 0.002** 0.054* 0.038** 0.357 0.073* 0.982 0.011** 0.056* 0.035** 0.052* 0.120* 0.442 0.340 0.798 0.060* Linking Statement 2 In Chapter 2 I showed that human disturbance associated with trails is negatively related to caterpillar richness and abundance in forest fragments. I developed a simple and effective index to measure the impact of trails. This indexcomplements the NSDVI-based remote-sensing ice storm impact framework developed in Chapter 1 to provide a management toolkit for holistically monitoring disturbances in remnant forest fragments.The next question that needs to be addressed relates to the biotic forces that affect species richness. Specifically, bottom-up effects from host plants. Both ice storms and recreational trails impact habitat quality, but for phytophagous insects “habitat quality” is highly dependent on the availability of suitable host plants. The methods used in Chapter 2 did not identify acceptable host plant abundance as a consistent determinant of caterpillar richness or abundance. In Chapter 3 I examine this in greater detail and expand my sampling regime to include a diversity of host plants at each quadrat (rather than just Acer saccharum). This will allow me to directly test the impact of different host plant species and overall host plant richness on caterpillar richness and abundance. 136 CHAPTER 3: Testing Two Methods that Relate Herbivorous Insects to Host Plants Peter J.T. White1 1 McGill University, Department of Biology 1205 Dr. Penfield Ave., Montreal, Quebec H3A 1B1, Canada 137 Abstract Insect herbivores are integral to terrestrial ecosystems. They provide essential food for higher trophic levels and aid in nutrient cycling. In general, research tends to relate individual insect herbivore species to host plant identity where a species will show preference one host over another. In contrast, insect herbivore assemblages are often related to host plant richness where an area with a higher richness of hosts will also have a higher richness of herbivores. In this study I test the ability of these two approaches (host plant identity/abundance vs. host plant richness) to describe the diversity, richness and abundance of an herbivorous Lepidoptera assemblage in temperate forest fragments in southern Canada. Many of these fragments are intensively managed in an effort to maximize the protection and preservation of biodiversity while simultaneously allowing for recreational use. Analyses indicated that caterpillar diversity, richness and abundance were better described by quadrat-scale host plant identity and abundance than by host plant richness. Most host plant-herbivore studies to date have only considered investigating host plant preferences at a species level; this type of assemblage level preference I show has been rarely considered. In addition, host plant replacement simulations indicate that increasing the abundance of preferred host plants could increase Lepidopterarichness and abundance by as much as 30% and 40% (respectively) in disturbed remnant forest fragments. This differs from traditional thinking that suggests higher levels of insect richness can be best obtained by maximizing plant richness. Host plant species that are highly preferred by the forest-dwelling caterpillar assemblage should be given special 138 management and conservation considerations to maximize biodiversity in forest communities. 139 Introduction Lepidoptera are very important in forest ecosystems. They are an intricate link between forest foliage and higher trophic levels. As larva (caterpillars) and as pupa, they are components of forest food webs providing an essential food source for birds, small mammals, snakes, amphibians and other insects(Hamilton and Pollack 1956, Lill et al. 2002, Moore and Strickland 1955, Murakami and Nakano 2000, Vasconcelos et al. 1996, Whitaker 1966). As adult moths, they are food for bats and birds (Goiti et al. 2009, Murray et al. 1980) and can be important flower pollinators (Pellmyr et al. 1996).Being herbivores in their larval life stage,Lepidoptera play a critical role in forest nutrient cycling, converting nutrient-rich leaves into nutrient-rich feces (either their own or those of a predator) that are easily digestible by soil organisms (Schowalter et al. 1986). Overlooked in Conservation Planning Even though forest-dwelling Lepidoptera play a central role in forest processes they are often overlooked in conservation planning (New 2004). This is in part because the natural history of most forest-dwelling Lepidoptera is very poorly known. In temperate regions research has focused on species capable of outbreak conditions like the spruce budworm (Choristoneura occidentalis), the gypsy moth (Lymantria dispar), the forest tent caterpillar (Malacosome dispar) or the fall webworm (Hyphantria cunea). More cryptic species and virtually all microlepidopteran species have been largely ignored. Acknowledging that individual species-targeted conservation management of Lepidoptera is often not possible, New (2004) suggests that an assemblage level approach could make it 140 easier to make conservation-oriented management decisions leading to the protection of entire lepidopteran assemblages and their natural habitats. The Relationship between Hosts and Lepidoptera The goal of this study is to determine how host plants richness, abundance and identity determine Lepidoptera assemblage richness in temperate deciduous forests. Across terrestrial ecosystems, theory has often focused on the richness of host plants driving the richness of Lepidoptera. This builds off the theory that richness at one trophic level determines the richness of the trophic level above (Andow 1991, Rosenzweig 1995). This means that in terrestrial ecosystems host plant richness would be a logical driver of insect herbivore richness. Indeed, this is what is often reported. In fields, plots with more forb, grass, legume and woody shrub species have been found to support higher insect herbivore richness and abundance than plots with less plant richness (Haddad et al. 2001, Siemann et al. 1998). A similar pattern has been observed in forest ecosystems where forest fragments with many tree species sustain higher insect herbivore richness than those with few (Summerville and Crist 2004). Along a successional gradient, insect herbivore richness can be tightly linked to plant species richness in young communities but more tightly linked to structural diversity in old communities (Southwood et al. 1979). The proposed mechanism to explain thesetypes of relationships is that a richer or more diverse host plant community provides more diverse foliar resources and more diverse structural resourcesthan a less diverse host plant community, allowing it to meet the physiological 141 and niche demands of more insect herbivore species (Lawton 1983, Murdoch 1972, Siemann 1998). The claim that host plant richness drives insect herbivore richness is problematic for two reasons. First, there are many exceptions to this relationship - especially when other factors are tested alongside host plant richness as competing explanatory variables. These factors include site-specific soil nutrient conditions (Hartley and Jones 2003), primary productivity and topography (Hawkins and Porter 2003), and habitat disturbance (Kruess and Tscharntke 2002). In addition, significant differences between insect herbivore richness have been observed among co-occurring host plants. For example,when the black willow tree (Salix nigra)and the box elder tree (Acer negundo) co-occur, the former tends to host a richer and more abundant Lepidoptera assemblage than the latter (Barbosa et al. 2000). Similarly, when the Norway maple (Acer platanoides) and sugar maple (Acer saccharum) co-occur, the former experiences significantly less insect herbivory than the latter (Cincotta et al 2009).These types of results seem to indicate that higher abundances of certain host plants in forest stands may be more important in facilitating a diverse and abundant insect herbivore assemblage. To my knowledge this assertion not been formally tested. Second, the claim that host plant richness drives insect herbivore richness is actuallydisconnected from the mechanisms that drive individual insect herbivore species abundance and distribution. For individual insect herbivore species the relationship with host plants is typically described in terms of host plant identity and host plant abundance rather than host plant richness (Thompson and Pellmyr 1991). For example, both the gypsy 142 moth (Lepidoptera: Lymantria dispar) and the winter moth (Lepidoptera: Operophtera brumata) are broad generalists but they tend to have faster developmental rates and higher population abundances when they feed on a select set of preferred host plants (Barbosa 1978, Liebhold et al. 1995, Maufette et al. 1983, Tikkanen et al 1999, Wint 1983). Many other similar examples exist (e.g. Busching and Turpin 1977, Capinera 1978, Deslile and Hardy 1997, Karban and English-Loeb 1997, Wiklund 1981). These types of preferences are usually driven by host plant-specific foliar nutrient qualities or natural enemy densities, which both have significant impacts on insect herbivore performance and survival (Hunter and Price 1992, Scriber and Slansky 1981, Thompson and Pellmyr 1991). Insect herbivore assemblage richness in a given locale is the culmination of these kinds of host plant choices made by individual species based on the host plants that are present. Since the individual choices are usually made based on host plant identity, insect herbivore richness may be best modeled by taking host plant identity into account. In this study I perform two tests to determine how host plants drive insect herbivore assemblage richness in temperate forests.First,I test whether caterpillar diversity and richness are related to host plant richness. Based on the aforementioned studies (and rationale), an increase in host plant species diversity or richnessis expected to be proportional to an increase in insect species diversity or richness. Second, I test whether caterpillar richness, diversity and abundance are more accurately described by the abundance of specific host plants. Both positive and negative host plant associations may be expected, synonymous to host plant choices made by individual insect herbivore species and indicative of the presence of preferred and non-preferredhosts. Given that 143 relationships between host plant identity and individual insect herbivore species are often evident, it is reasonable to test whether the cumulative selections made across all species results in detectable preference patterns at the assemblage level. I test these relationships with a Lepidoptera assemblage in forest fragments of the mixed wood plains in the St. Lawrence River valley of southeastern Canada. This area has historically experienced widespread forest habitat destruction; 85% of the original landscape has been cleared of old-growth forest in favour of agricultural, industrial and urban development (Allen 2001, Drushka 2003). This type of widespread habitat loss and associated habitat fragmentation can have significant detrimental effects on both generalist and specialist species (Bender et al. 1998). With this in mind, one of the biggest challenges is to manage forest fragments in a way that benefits forest-dwelling species assemblages, maximizing and preserving species richness in intensively developed landscapes. Methods Study Area The forest fragments I studied were associated with the Monteregian Hills (Feininger and Goodacre, 1995), in the St. Lawrence River valley in southeastern Quebec, Canada (45°30'N, 73°30'W to 45°24'N, 72°35'W; Figure 1). During pre-settlement times all these Monteregian Hills would have had broadly similar forests embedded in a more or less continuously forested landscape (Richard and Grondin 2009), but now exist as a series of large forest remnants isolated in the developed landscape. I established eighteen 400 m2 144 (20 m x 20 m) quadrats in the remnant forests at each of four sampling sites on Monteregian Hills. The first site was at Parc Mont Royal - an urban park in the middle of Montreal, a city of 3.5 million inhabitants. The forest at the park was cut in the 1950s and early 1960s (Brunel et al. 2005) and was subsequently reforested up until the early 1990s. The land area of the park is 190 ha, ~100 ha of which is forested. The second site was at Parc National du Mont St. Bruno – a protected provincial park in the eastern suburbs of greater Montreal. The forest at Mont St. Bruno is a broadleaf deciduous forest covering more than 500 ha of the 790 ha park. Aerial photography records indicate that 60 or more hectares of forest were cut prior to the 1940s in the northern part of the park, and subsequently replanted or allowed to regrow (National Air Photo Library 1950). The forest has more than 85 species of woody plants. The third site was at the Gault Nature Reserve on Mont St. Hilaire – a protected park and UNESCO Biosphere Reserve, 38 km east of Montreal. The forest at Mont St. Hilaire is an old-growth broadleaf deciduous forest covering most of the 1000 ha reserve. This site has a long history of protection dating back to the 1600s (Arii 2004, Maycock 1961). The fourth site was at Mont Shefford – one of the easternmost Monteregian Hills, 70 km east of Montreal. At the time of this study different parts of the site were in varying states of disturbance and urban development. I conducted caterpillar sampling in three sub-sites around the hill. Six quadrats were in a 100 ha semi-disturbed patch of forest on the west side of the Mont Shefford and six quadrats in a 25 ha patch of forest on the east 145 side of the hill set aside as a forested community park. A quarter of the former site was a sugar bush and there were numerous trails used by deer-hunters throughout the area. The community park had several recreational trails, but off-trail use was discouraged. Six quadrats were in a third sub-site, Parc de la Yamaska, a provincial park just north of Mont Shefford. Caterpillar Survey and Identification Caterpillars were collected at 18 quadrats at each of the four study sites. Prior to caterpillar surveying a vegetation analysis indicated that sugar maple (Acer saccharum) was the most abundant host plant across sites and was the only host plant that was present at every quadrat. Therefore, at each of the 72 quadrats, ten sugar maples and up to 10 of all other tree species between 3 and 10 cm dbh were sampled for caterpillars (Appendix B1). This sampling method was chosen to survey a representative proportion of host plants among quadrats. When a given host plant species was abundant in excess of 10 individuals, sample trees were chosen at random. Each sample tree was surveyed by striking the bole and lower branches ten times with a 20 oz, 30" aluminum baseball bat and catching dislodged caterpillars on a 1 m2 sheet. Caterpillar collections were made three times at each quadrat (between June 1 and June 6, July 4 and July 9, August 3 and August 6 in 2009). This resulted in a total of 2090 sampled trees and 62700 total tree-strikes (2090 trees x 10 tree-strikes per tree x three caterpillar collection periods). Macrolepidopteran moth caterpillars were identified to species with a dissecting microscope using Wagner’s Caterpillars of Eastern North America Field Guide(2005). 146 Microlepidopteran moths were counted and identified only to morphospecies for lack of an accurate identification guide. Macrolepidopteran moths collected in early instars were reared so that positive identifications could be made. Controlling for Habitat Disturbance Recent investigations have shown that forest-dwelling caterpillar assemblages are sensitive to intra-habitat disturbances. Any investigation of herbivore-host plant relationships should therefore take this into account. White et al.(2011) showed that there is a consistent negative relationship between recreational trail presence and caterpillar richness in forest fragments in southeastern Quebec, Canada. They suggested that this relationship may be due to increases in caterpillar parasitism/predation and/or changes in trail-side conditions that make trail-side habitat less suitable for caterpillars.Non-native tree species introductions are sometimes correlated to management and can have a negative impact on caterpillar species richness and abundance. To control for these effects I used a variable called Trail Index to measure the impact of trails at each quadrat. Trail index values for the quadrats in this study were described and calculated in White et al. (2011). Analyses I performed a two-run stepwise (forwards) multiple regression to determine the impacts of trail index and host plant frequencies on caterpillar abundance in the sampled quadrats. This type of analysis is very useful when a large number of independent variables 147 are used and the goal is to eliminate variables of marginal (or non-) significance. While it lacks the sophistication of other multivariate statistical methods (e.g. ordination or regression trees), one advantage it provides is that the final model it computes is independent of insignificant variables. It assumes a Gaussian distribution of the model residual values; collinearity in independent variables can be tested with interaction terms. The decision threshold to include a given independent variable in each step of each regression was based on p < 0.05. In the first run-through of the stepwise regression, I used 25 candidate independent variables to explain the variance in (log10 transformed) caterpillar abundances in the 72 study quadrats. This suite of 25 independent variable consisted of the host plant frequencies (24 species, Appendix A1) and the trail-index value of each quadrat. I only included host plants that were present at more than two quadrats, resulting in the exclusion of 13 of the original 38 host plant species. Acer saccharum was also excluded from the analysis because it is ubiquitous throughout the study region and I wanted to focus on the impact that additional host plant species had on quadrat caterpillar abundance. After the first run-through of the stepwise regression, I conducted a second run-through using the independent variables selected in the first run-through and the interactions between each of these variables and Trail Index. The standard coefficients and partial R2 values of the remaining independent variables were then calculated.I conducted two identical analyses using caterpillar species richness and caterpillar Shannon’s diversity (Shannon and Weaver 1949) as the dependent variables in place of caterpillar abundance. 148 Host Plant and Caterpillar Relationships I used two simple multiple regressions to test significance and strength of the relationship between host plants and caterpillars among quadrats. The first compared caterpillar richness to host plant richness and Trail Index (log10 + 1 transformed); the second compared caterpillar Shannon’s diversity to host plant Shannon’s diversity and Trail Index (log10 + 1 transformed). Interaction terms were included in both models to determine whether there was a relationship between independent variables. Testing Host Plant-Specific Preferences The preference of each tree species by the caterpillar assemblage was calculated as a Caterpillar Assemblage Preference Index (CAPIr) which measures the observed caterpillar richness in j trees of host plant species i minus the average caterpillar species richness in j trees drawn at random from the entire host plant-caterpillar dataset. It is calculated as: 1000 [1] CAPIri = Ri − ∑ {r} n =1 j 1000 where a given tree species i has a cumulative caterpillar species richness R summed across j trees that were surveyed. In this calculation, r is the caterpillar species richness in a subset of j individuals selected at random (with replacement) from the entire dataset of all 2090 host plant samples from all host plant species; 1000 subsets of rwere selected and averaged. CAPIr is essentially the actual caterpillar species richness in host plant species i with j individuals minus the average (i.e. expected) caterpillar species richness found in j trees. Thus, if a host plant species with j trees has a CAPIr value of x, it would be said to 149 support x more (or less if x is negative) caterpillar species than would be found if a random sample of j trees was sampled from the set of all trees. Similarly, the Caterpillar Assemblage Preference Index in terms of caterpillar abundance (CAPIa) computes the observed caterpillar abundance in j trees of host plant species i minus the average caterpillar abundance in j trees drawn at random from the entire host plant dataset. It is calculated as: 1000 [2] CAPIai = Ai − ∑{a} n =1 j 1000 where A and a are the abundances of host plant species i and of a random tree subset with j individuals respectively. For the purposes of this study, the acceptabilityof a host plant species is defined as the number of caterpillar species in an assemblage that are documented to use it. Host plant acceptability measures were taken from Wagner (2005) and Handfield (1999). Host plant acceptability was correlated to CAPIr and CAPIa to test whether caterpillars are distributed relative to the occurrence of acceptable host plants. Results Caterpillar Sampling I collected 1896 caterpillars including 53 macrolepidoptera species (1305 individuals) and 56 microlepidopteran morphospecies (591 individuals) (Appendix B2; Appendix A3 for botanical authorities) from 38 different host plant tree species (2090 total trees). The five 150 most common trees among quadrats (Acer saccharum, Fagus grandifolia, Fraxinus americana, Acer pensylvanicum and Ostrya virginiana) yielded 81% of caterpillar catches and high levels of caterpillar richness (Appendix B1). Host Plant and Caterpillar Relationships There was a significant positive relationship between host plant Shannon’s diversity and caterpillar Shannon’s diversity (Table 1a). Trail Index was also a significant descriptor; the total model 45% of the variance in caterpillar diversity. Host plant richness however was not significantly related to caterpillar richness (Table 1b). Trail Index remained a significant descriptor with the total model explaining 27% of the variance in caterpillar richness. CAPIr and CAPIa CAPIr values ranged between 7.9 (Prunus serotina) and -11.6 (Rhamnus cathartica); CAPIa values ranged between 121.5 (Fagus grandifolia) and -99.7 (Fraxinus americana). Increasingly positive CAPI scores indicate that a host plant is used by caterpillars more than average; increasingly negative CAPI scores indicate that a host plant is used by caterpillars less than average (Table 2). Host plant acceptability documented by Wagner (2005) and Handfield (1999) did a very poor job explaining the variance in CAPIr and CAPIa (Figure 2a,b). This indicates that there is no detectable connection between acceptable host plants and preferred host plants in my study system. 151 Host Plant Importance Relative to Trail Index For caterpillar abundance, Acer pensylvanicum, Ostrya virginiana and Fagus grandifolia had the highest partial R2 values in the step-wise regression. Host plant frequencies accounted for 21.0% of the variance in caterpillar abundance independent of Trail Index which accounted for 19.8% of the variance; this includes a significant interaction between Trail Index and F. americana frequency (Table 2a). The combination of independent variables accounted for an additional 20.5% of the variance (total R2 of model = 0.61). For caterpillar species richness, host plant frequencies of Acer pensylvanicum and Ostrya virginiana accounted for a combined 24.9% of the variance in caterpillar species richness among quadrats (Table 2b). Trail Index accounted for an additional 22.3% of the variance; this includes a significant interaction between Trail Index and F. americana frequency. The combination of independent variables accounted for an additional 11.5% of the variance (total R2 of model = 0.59). For caterpillar Shannon’s diversity, host plant frequencies of Acer pensylvanicum, Ostrya virginiana and Ulmus americana accounted for 27.9% of the variance in caterpillar diversity (Table 2c). Trail Index accounted for an additional 18.7% of the variance (this again includes a significant interaction between Trail Index and F. americana frequency); the combination of independent variables accounted for an additional 9.4% of the variance (total R2 of model = 0.56). 152 Discussion Host plant identity and abundance were statistically significant and strong predictors of caterpillar richness. This was contrasted by host plant diversity that had a statistically significant – but very weak – effect on caterpillar diversity. Tree richness and caterpillar richness were unrelated. These results are in sharp contrast with the idea that host plant richness drives insect herbivore richness (Lawton 1983, Southwood et al. 1979). Instead myresults indicate that host plant identity and abundance are more appropriate measures for explaining insect herbivore assemblage diversity, richness and abundance. This is a novel result that has not yet been described in analyses examining the relationship between host plants and insect herbivores. Biodiversity and Conservation My results suggest that reserve management should adopt an approach that identifies and promotes high biodiversity host plants. This is in contrast with other popular approaches such as maximizing stand structural complexity, maximizing floral biodiversity and using natural disturbance regimes (Battles et al 2001, Lindenmayer et al. 2006). Niemela and Neuvonen (1981, Neuvonen and Nimela 1983) suggested that the most important host plants in temperate forests for Lepidoptera biodiversity are those with the highest abundance. This is true in a static sense - in most northeastern broadleaved forests the sugar maple (Acer saccharum) is the most important for insect herbivores. By virtue of being the most abundant tree it hosts the highest insect herbivore species richness. But this is a narrow view that does not take into account low abundance host plants. In 153 situations where reserve management has a mandate to manage tree relative abundances to maximize overall forest health and biodiversity, a more nuanced approach is warranted. For the range of host plants in northeastern deciduous forest, this suggests that restoration and replanting efforts should include black cherry (Prunus serotina), serviceberry (Amelanchier spp.), mountain maple (Acer spicatum), striped maple (Acer pensylvanicum) and yellow birch (Betula alleghaniensis). In regions where different forest types persist, a system-specific analysis of tree hosting abilities should be conducted to identify high and low biodiversity host plants. Pair-wise host plant comparisons can be useful for this purpose and have been conducted for many common plants (Barbosa et al 2000, Cincotta et al 2009). That said, there is evidence to suggest that herbivore-hosting capabilities of trees can be conserved across large geographic scales. Moran and Southwood (1982) found that the relative species richness of insect herbivores and insect predators were very similar on five tree taxa present in both the United Kingdom and South Africa. This might suggest that the preference indices I calculated for broadleaved deciduous forests in southern Canada may be broadly applicable to maple-dominated broadleaved and mixedwood forests across northeastern North America. Low Caterpillar Richness and Abundance in Invasive Trees My results also indicate that invasive trees may be problematic in deciduous broadleaved forests. The impoverished caterpillar assemblages found on Acer platanoides (Norway maple) and Rhamnus cathartica (European Buckthorn) add to a growing body of evidence showing that non-native host plants are a detriment to forest insect assemblages. A. 154 platanoides was introduced in the late 1700s (Spongberg 1990) and has periodically been planted for forest restoration (Larson 1996, Webb and Kaunzinger 1993). However, it often outcompetes native tree species and is able to invade intact woodlands (Bertin et al. 2005, Wyckoff and Webb 1996). My results reinforce the trend identified by Cincotta et al (2009) showing that A. platanoides is not a favoured host plant of forest insect herbivores. Their results compared A. platanoides to A. saccharum, whereas I show it in relation to other common sub-dominant host plant trees where it ranked 21st among 24 tree species for both CAPIr and CAPIa (Table 3). Similarly, Rhamus cathartica ranked 24th and 22nd in CAPIr and CAPIa, respectively. Research has shown that Rhamnus cathartica is a detriment to forest communities as it modifies soil nitrogen conditions, reduces leaf litter levels, propagates the spread of invasive species, is not consumed by many native herbivores and has allopathic effects on native trees (Heneghan et al. 2004, Knight et al. 2007). While R. cathartica can be beneficial for sustaining insect populations in disturbed and urban settings (VanVeldhuizen et al. 2005) its negative association with forest-dwelling moth populations give further reason for its control in North American forest fragments (Gassmann 2005, Moriarty 2005). Curiously, Fraxinus americana also had markedly low CAPIr and CAPIa scores, ranking 22nd and 24th (respectively) among 24 tree species. Species in the genusFraxinustend to be higher than average in terms of leaf toughness (Ricklefs and Matthew 1982), support high caterpillar parasitoid loads(Lill et al. 2002) and can have prohibitively toxic phenolic compounds (i.e. in the case of the closely related Fraxinus pennsylvanica;Markovic et al. 1996). Despite these, they are however documented as widely used host plants by caterpillar species (Handfield 1999, Karban and Ricklefs 1984, 155 Wagner 2005). The step-wise regression analysis indicated that F.americana has a strong association with trails at my study sites which when combined with these other deterrents may have resulted in the low CAPI scores. From the point of view of forest management for biodiversity conservation, the full impact of A. platanoides and R. cathartica on forest dwelling moth assemblages can be enumerated with a host plant replacement simulation. At Mont Royal, caterpillar surveys included a total of 629 sampled host plants, the most abundant being A. saccharum (180 trees surveyed), F. americana (102 trees surveyed), R. cathartica (48 trees surveyed), T. americana (48 trees surveyed) and A. platanoides (39 trees surveyed); the remaining 212 surveyed trees were made up of 26 different host plant species. Caterpillar species richness and abundance collector curves can be created (using the second half of equations [1] and [2], respectively), first using all trees at Mont Royal and second by replacing the data collected from A. platanoides and R. cathartica on Mont Royal with data collected from O. virginiana and A. pensylvanicum from other locations (Figure 3). This results in an increase of 30% in caterpillar species richness and 40% in caterpillar abundance. In this exercise, O. virginiana was chosen as a replacement for A. platanoides because it has a similar abundance (averaged per site) and average height (5.6 meters versus 5.5 meters). A. pensylvanicum was chosen as a replacement for R. cathartica because these species are both often subdominant trees associated with disturbed areas (sun-loving) and are relatively similar in average height (5.4 meters versus 3.8 meters).A similar replacement simulation can be run for F. americana. While native, it has a CAPIr and CAPIa scores were amongst the lowest in the host plant data set. When the caterpillar 156 data collected from 102 F. americana surveyed at Mont Royal are replaced with data from F. grandifolia (both are common canopy-contributing species and have similar average heights; 7.4 meters versus 6.9 meters) caterpillar species richness increases 18%, abundance 37% (Figure 3). Mechanisms Given the important implications of these results, it is worthwhile considering the ecological mechanisms that may be driving the observed relationships. Since host plant choices are made by individual Lepidoptera species (and scale up to the assemblage level), any driving mechanism explaining the assemblage-wide pattern would need to be operational at the individual species level. The most well established factors used to explain host plant choices amongst individual species are host plant foliar quality (Feeny 1970, Matteson 1980, Buse et al. 1998), and the presence of natural enemies (Hunter and Price 1992, Siemann et al. 1998, Lill 2001). Nitrogen has been identified as one of the preeminent foliar nutrients associated with insect herbivore host plant selection and performance (Matteson 1980). This makes it a prime foliar candidate to explain the host plant preferences I observed. One of the most preferred host plant species in my study, Ostrya virginiana,is documented as having moderate to high levels of foliar nitrogen compared to other broad-leaved species (Mertzger 1990). However, the widely preferred host plant Acer pensylvanicum has a moderate to low level of foliar nitrogen (Zehnder et al. 2009) similar to that of the avoided host plant Fraxinus americana (Abrams and Mostoller 1995, Côté et al. 2002). One of the 157 most nitrogen-rich host plant species Betula papyrifera(Abrams 1998) was also has lower than average CAPIr and CAPIa values (Table 3). This would be consistent with the findings of Karban and Ricklefs (1984) who found no relationship between foliar nutritional quality and caterpillar species richness and abundance in broad-leaved deciduous caterpillar communities. Although a simple relationship between CAPI values and foliar nitrogen content does not seem apparent in this case, other foliar variables such as water content (Scriber and Slansky 1981) and foliar toxins (Gatehouse 2002) have also been commonly cited as determinants of insect herbivore performance and distribution. In an in-depth analysis of host plant-insect interactions, Futuyma and Gould (1979) concluded that the variation in insect populations among hosts is likely due to a multiplicity of plant leaf chemical variables. Top-down pressure from parasitoids can also have a significant impact on oviposition host choices of adult Lepidoptera (Karban and English-Loeb 1997, Thompson and Pellmyr 1991). Lill et al. (2002) documented parasitoid-host plant-caterpillar interactions in common host plant genera and showed that the genera Fraxinus and Acer were associated with higher than expected caterpillar parasitoid loads while the genus Ulmus with lower than expected caterpillar parasitoid loads. They did not include Ostrya or Fagus in their analyses. The positive association between Fraxinus and caterpillar parasitoids could explain the exceptionally low Fraxinus americana CAPIr and CAPIa values. However, Ulmus americana alsohad lower-than average CAPIr and CAPIa values, even though this genus is a documented predator-reduced space. This discrepancy could indicate that parasitism is not the primary driving mechanism of caterpillar assemblage 158 richness in Ulmus hosts in my study region. The wide range in CAPIr and CAPIa values for different Acer species in my data further suggest that a genus-level parasitoid-control of caterpillar assemblages may not be the dominant driver impacting caterpillar host-plant preferences. For example, A. pensylvanicum and A. spicatum had significantly positive CAPIr values (6.1 and 6.4 respectively) but A. rubrum and A. platanoideshad significantly negative CAPIr values (-6.1 and -9.7 respectively). Similarly, A. pensylvanicum and A. spicatum had significantly positive CAPIa values (44.5 and 14.6) but A. rubrum and A. platanoides had significantly negative CAPIa values (-10.3 and -22.3 respectively). If parasitoid regulation of caterpillars was occurring at the host plant genus level, then species within a given genera would be expected to have roughly similar CAPIr and CAPIa values. While it is still possible that host plant caterpillar parasitoid loads may play a role in driving caterpillar assemblage richness among the other host plant genera, it seems implausible that they are the sole driving mechanism determining caterpillar species richness and abundance among Acer or Ulmus host plants in my study. Conclusion Understanding the important determinants of insect assemblage richness and abundance in remnant forest fragments can improve management and conservation efforts. In a landscape where pristine forest habitat is rare, conservation-based management should attempt to maximize and maintain the richness in the forest fragments that remain.In this study, I have shown that the richness and abundance of an insect herbivore assemblage can be more effectively described in terms of host plant identity and host plant abundance. 159 Describing the insect herbivore assemblage in these terms is more consistent with singlespecies studies that show insect herbivore host choice is often a function of host plant identity rather than quadrat- or stand-scale host plant richness. In the system I examined, it does not appear as though top-down parasitoid control is the driving force behind host choice at the herbivore assemblage level (Lill et al. 2002). It also seems unlikely that host plant foliar nitrogen content is driving assemblage level host choice. Given that neither of these two mechanisms seem dominant, it is possible that the assemblage-level host plant selections are driven by a complex interaction of multiple foliar nutrient properties and top-down pressure (Mayhew 1997). It is also possible that there is a third factor (e.g. historical disturbance events) driving both the relative abundance of community host plant species and Lepidoptera assemblage host plant occupancy. In a direct conservation application of my results, host plant replacement simulations indicate that planting preferred host plants in the place of non-preferred host plants could result in a profound impact on insect herbivore assemblage richness and abundance. At a broader level, these results call for a shift in conservation management principles where some emphasis should be placed on identifying and protecting high value host plants that are synonymous with high levels of insect herbivore richness and abundance. 160 Acknowledgements I would like to thank P. Peres-Neto, C. Buddle, K. Summerville, J. Donoghue, R. Feldman and J. Messier for their important reviews. BothM.J. Lechowicz and B. McGill played integral roles commenting on drafts of this MS throughout its creation. I am grateful to M. VonButtlar and R. MacKenzie for help with data collection andam indebted to both D. Rodrigue (Supervisor, Conservation and Education Service at Parc du Mont St. Bruno) and A. Mochon (Supervisor, Conservation and Education Service at Parc de la Yamaska) for their help in selecting research locations within their respective parks. This research was made possible by funds provided by the Natural Sciences and Engineering Research Council of Canada. 161 References Abrams, M. D. 1998. The red maple paradox. - BioScience 48: 355-364. Abrams, M. D. and Mostoller, S. A. 1995. Gas exchange, leaf structure and nitrogen in contrasting successional tree species growing in open and understory sites during a drought. - Tree Physiology 15: 361-370. Allen, E. 2001. Forest health assessment in Canada. - Ecosystem Health 7: 28-34. Andow, D. A. 1991. Vegetational diversity and arthropod population response. – Annual Reviews of Entomology 36: 561-86. Arii, K. 2004. Ecology of American beech and sugar maple in an old-growth forest. Department of Biology. - McGill University. Barbosa, P., Segarra, A. and Gross, P. 2000. Structure of two macrolepidopteran assemblages in Salix nigra (Marsh) and Acer negundo L.: abundance, diversity, richness, and persistence of scarce species. - Ecological Entomology 25: 374-379. Battles, J. J., Shlisky, A. J., Barrett, R. H., Heald, R. C. and Allen-Diaz, B. H. 2001. The effects of forest management on plant species diversity in a Sierran conifer forest. – Forest Ecology and Management 146: 211-222. Bender, D. J., Contreras, T. A. and Fahrig, L. 1998. Habitat loss and population decline: a meta-analysis on the patch size effect. - Ecology 79: 517-533. Bertin, R. I., Manner, M. E., Larrow, B. F., Cantwell, T. W. and Berstene, E. M. 2005. Norway maple (Acer platanoides) and other non-native trees in 162 urban woodlands of central Massachusetts. - Journal of the Torrey Botanical Society 132: 225-235. Brunel, S., Poitras, C., Burges, J., Dufour, M., Ghafouri, M., Chouinard, A., Mongrain, G., Dumais, H. and Turcotte, S. 2005. Étude de caractérisation de l’arrondissement historique et naturel du Mont-Royal. Commission des biens culturels du Québec, Québec, Québec. Busching, M. K. and Turpin, F. T. 1977. Survival and Development of Black Cutworm (Agrotis ipsilon) Larvae on Various Species of Crop Plants and Weeds. - Environmental Entomology 6: 63-65. Buse, A., Good, J. E. G., Dury, S. and Perrins, C. M. 1998. Effects of elevated temperature and carbon dioxide on the nutritional quality of leaves of oak (Quercus robur L.) as food for the Winter Moth (Operophtera brumata L.). - Functional Ecology 12: 742-749. Capinera, J. L. 1978. Studies of host plant preference and suitability exhibited by early-instar Range Caterpillar larvae. - Environmental Entomology 7. Cincotta, C. L., Adams, J. M. and Holzapfel, C. 2009. Testing the enemy release hypothesis: a comparison of foliar insect herbivory of the exotic Norway maple (Acer platanoides L.) and the native sugar maple (A. saccharum L.). - Biological Invasions 11: 379-388. Côté, B., Fyles, J. W. and Djalilivand, H. 2002. Increasing N and P resorption efficiency and proficiency in northern deciduous hardwoods with 163 decreasing foliar N and P concentrations. - Annals of Forest Science 59: 275-281. Deslile, J. and Hardy, M. 1997. Male larval nutrition influence the reproductive success of both sexes of the Spruce budrowm, Choristoneura fumiferana (Lepidoptera: Tortricidae). – Functional Ecology 11: 451-463. Drushka, K. 2003. Canada's forests : a history. - Forest History Society, McGillQueen's University Press. Feeny, P. 1970. Seasonal changes in oak leaf tannins and nutrients as a cause of spring feeding by Winter Moth caterpillars. - Ecology 51: 565-581. Futuyma, D. J. and Gould, F. 1979. Associations of plants and insects in deciduous forest. - Ecological Monographs 49: 33-50. Gassmann, A. 2005. Developing biological control of buckthorns. - In: Skinner, L. C. (ed.) Symposium on the Biology, Ecology and Management of Garlic Mustard (Alliaria petiolata) and European Buckthorn (Rhamnus cathartica), pp. 58-61. Gatehouse, J. A. 2002. Plant resistance towards insect herbivores: A dynamic interaction. – New Phytologist 156: 145-169. Goiti, U., Garin, I., Almenar, D., Salsamendi, E. and Aihartza, J. 2009. Foraging by Mediterranean Horseshoe Bats (Rhinolophus euryale) in relation to prey distribution and edge habitat. - Journal of Mammalogy 89: 493-502. 164 Haddad, N.M., Tilman, D., Haarstad, J., Ritchie, M. and Knops, J.M.N. 2001. Contrasting effects of plant richness and composition on insect communities: a field experiment. – The American Naturalist 158: 17-35. Hamilton, W. J. and Pollack, J. A. 1956. The food of some colubrid snakes from Fort Benning, Georgia. - Ecology 37: 519-526. Handfield, L. 1999. Le Guide Des Papillons Du Quebec. - Broquet Inc. Hartley, S. E. and Jones, T. H. 2003. Plant diversity and insect herbivores: effects of environmental change in contrasting model systems. – Oikos 101: 617. Hawkins, B.A. and Porter, E.E. 2003. Does herbivore diversity depend on plant diversity? The case of Calilfornia butterflies. – The American Naturalist 161: 40-49. Heneghan, L., Rauschenberg, C., Fatemi, F. and Workman, M. 2004. European buckthorn (Rhamnus cathartica) and its effects on some ecosystem properties in an urban woodland. - Ecological Restoration 22: 275-280. Hunter, M.D. and Price, P.W. 1992. Playing chutes and ladders: heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. – Ecology 73: 724-732. Karban, R. and English-Loeb, G. 1997. Tachinid parasitoids affect host plant choice by caterpillars to increase caterpillar survival. - Ecology 78: 603611. 165 Karban, R. and Ricklefs, R. E. 1984. Leaf traits and the species richness and abundance of Lepidopteran larvae on deciduous trees in southern Ontario. - Oikos 43: 165-170. Knight, K. S., Kurylo, J. S., Endress, A. G., Stewart, J. R. and Reich, P. B. 2007. Ecology and ecosystem impacts of common buckthorn (Rhamnus cathartica): a review. - Biological Invasions 9: 925-937. Kruess, A. and Tscharntke, T. 2002. Constrasting responses of plant and insect diversity to variation in grazing intensity. – Biological Conservation 106: 293-302. Larson, D. W. 1996. Brown's Woods: an early gravel pit forest restoration project, Ontario, Canada. - Restoration Ecology 4: 11-18. Lawton, J.H. 1983. Plant architecture and the diversity of phytophagous insects. Annual Review of Entomology 28: 23-29. Lill, J. T. 2001. Selection on herbivore life-history traits by the first and third trophic levels: thedevil and the deep blue sea revisited. – Evolution 55: 2236-2247. Lill, J. T., Marquis, R. J. and Ricklefs, R. E. 2002. Host plants influence parasitism of forest caterpillars. - Nature 417: 170-173. Lindenmayer, B., Franklin, J. F. and Fischer, J. 2006. General management principles and a checklist of strategies to guide forest biodiversity conservation. – Biological Conservation 131: 433-445. 166 Markovic, I., Norris, D. M. and Cekic, M. 1996. Some chemical bases for Gypsy Moth, Lymantria dispar, larval rejection of Green Ash, Fraxinus pennsylvanica, foliage as food. - Journal of Chemical Ecology 22: 22832298. Matteson, W. J. J. 1980. Herbivory in relation to plant nitrogen content. - Annual Review of Ecology and Systematics 11: 119-161. Maufette, Y., Lechowicz, M. J. and Jobin, L. 1983. Host preferences of the gypsy moth, Lymantria dispar(L.), in southern Quebec. – Canadian Journal of Forest Research 13: 53-60. Maycock, P. F. 1961. Botanical studies on Mont St. Hilaire, Rouville County, Quebec. - Canadian Journal of Botany 39: 1293-1325. Mayhew, P.J. 1997. Adaptive patterns of host-selection by phytophagous insects. – Oikos 79: 417-429. Mertzger, F. T. 1990. Ostrya virginiana (Mill.) K. Koch eastern hophornbeam. - In: Burns, R. M. and Honkala, B. H. (eds.), Silvics of North America. Forest Service, United States Department of Agriculture. Moore, J. E. and Strickland E. H. 1955. Further notes on the food of Alberta Amphibians. – American Midland Naturalist 54: 253-256. Moran, V. C. and Southwood, T. R. E. 1982. The guild composition of arthropod communities in trees. – Journal of Animal Ecology 51: 289-306. Moriarty, J. 2005. Conventional management of buckthorn species. - In: Skinner, L. C. (ed.) Symposium on the Biology, Ecology and Management of Garlic 167 Mustard (Alliaria petiolata) and European Buckthorn (Rhamnus cathartica), pp. 58-61. Murakami, M. and Nakano, S. 2000. Species-specific bird functions in a forestcanopy food web. - Proceedings of the Royal Society of London 267: 1597-1601. Murdoc, W.W., Evans, F.D. and Peterson, C.H. 1972, Diversity and pattern in plants and insects. - Ecology 53: 819-829. Murray, N. D., Bishop, J. A. and Macnair, M. R. 1980. Melanism and predation by birds in the moths Biston betularia and Phigalia pilosaria. - Proceedings of the Royal Society of London. Series B, Biological Sciences 210: 277-283. National Air Photo Library. 1950. Roll Number A12956: photos 259, 260, 264. Natural Resources Canada, Ottawa, Ontario, Canada. Neuvonen, S. and Niemelä, P. 1983. Species richness and faunal similarity of arboreal insect herbivores. - Oikos 40: 452-459. New, T. R. 2004. Moths (Insecta: Lepidoptera) and conservation: background and perspective. - Journal of Insect Conservation 8: 79-94. Niemelä, P. and Neuvonen, S. 1981. Species richness of macrolepidoptera on Finnish deciduous trees and shrubs. - Oecologia 51: 364-370. Pellmyr, O., Thompson, J. N., Brown, M. I. and Harrison, R. G. 1996. Evolution of pollination and mutualism in the yucca moth lineage. - The American Naturalist 148: 827-847. 168 Richard, P.J.H. and Grondin, P. 2009. Histoire postglaciaire de la végétation, pp. 169-176, in Manuel de Foresterie, 2è édition, Ordre des ingénieurs forestiers du Québec, Éditions MultiMondes, Québec. Ricklefs, R. E. and Matthew, K. K. 1982. Chemical characteristics of the foliage of some deciduous trees in southeastern Ontario. - Canadian Journal of Botany 60: 2037-2045. Rosenzweig, M. L. 1995. Species Diversity in Space and Time. – Cambridge University Press, Cambridge, U.K. Schowalter, T. D., Hargrove, W. W. and Crossley, D. A. 1986. Herbivory in forest ecosystems. - Annual Review of Entomology 31: 177-196. Scriber, J. M. and Slansky, F. 1981. The nutritional ecology of immature insects. Annual Review of Entomology 26: 193-211 Shannon, C.E. and Weaver, W. 1949. The mathematical theory of communication. - University of Illinois Press, Urbana, Illinois. Siemann, E. 1998. Experimental tests of effects of plant productivity and diversity on grassland arthropod diversity. - Ecology 79: 2057-2070. Siemann, E., Tilman, D., Haarstad, J. and Ritchie, M. 1998. Experimental tests of the dependence of arthropod diversity on plant diversity. – The American Naturalist 152: 738-750. Southwood, T. R. E., Brown, V. K. and Reader, P. M. 1979. The relationship of plants and insect diversities in succession. – Biological Journal of the Linnean Society 12: 327-348. 169 Spongberg, S. A. 1990. A reunion of trees: the discovery of exotic plants and their introduction into North American and European landscapes. - Harvard University Press. Thompson, J.N. 1988. Evolutionary ecology of the relationship between ovipotision preference and perormance of offspring in phytophagous insects. – Entomologia Experimentalis et Applicata 47: 3-14. Tikkanen, O. P., Carr, T. G. and Roininen, H. 1999. Factors influencing the distribution of a generalist spring-feeding moth, Operophtera brumata (Lepidoptera: Geometridae), on host plants. – Environmental Entomology 28: 461-469. VanVeldhuizen, M., Ragsdale, D. W. and Skinner, L. 2005. Survey of insect fauna on common buckthorn, Rhamnus cathartica, in Minnesota. - In: Skinner, L. C. (ed.) Symposium on the Biology, Ecology and Management of Garlic Mustard (Alliaria petiolata) and European Buckthorn (Rhamnus cathartica), pp. 58-61. Vasconcelos, S. D., Williams, T., Hails, R. S. and Cory, J. S. 1996. Prey selection and baculovirus dissemination by carabid predators of Lepidoptera. Ecological Entomology 21: 98-104. Wagner, D. L. 2005. Caterpillars of Eastern North America. - Princeton University Press. 170 Webb, S. L. and Kaunzinger, C. K. 1993. Biological invasion of the Drew University (New Jersey) Forest Preserve by Norway maple (Acer platanoides L.). Bulletin of the Torrey Botanical Club 120: 343-349. Whitaker. 1966. Food of Mus musculus, Peromyscus maniculatus bairdi and Peromyscus leucopus in Vigo County, Indiana. - Journal of Mammalogy 47: 473-486. White, P. J. T., McGill, B. J. and Lechowicz, M. J. 2011. Human-disturbance and caterpillars in managed forest fragments. - Biodiversity and Conservation 20: 1745-1762. Wiklund, C. 1981. Generalist vs. specialist oviposition behaviour in Papilio machon (Lepidoptera) and functional aspects o the herarchy of oviposition preferences. – Oikos 36: 163-170. Wint, W. W. 1983. The role of alterntaive host-plant species in the lfe of a polyphagous moth, Operophtera brumata (Lepidoptera: Geometridae). – Journal of Animal Ecology 52: 439-450. Zehnder, C. B., Stodola, K. W., Joyce, B. L., Egetter, D., Cooper, R. J. and Hunter, M. D. 2009. Elevational and seasonal variation in the foliar quality and arthropod community of Acer pensylvanicum. - Environmental Entomology 38: 1161-1167. 171 Mont RoyalMont St. Bruno Downtown Montreal Mont St. Hilaire Mont Shefford CANADA 10 km U.S.A. U.S.A. Figure 1 Caterpillars were collected from four sites in the St. Lawrence River valley of southern Quebec, Canada (Figure adapted from Atlas of Canada 2010). The matrix surrounding each site isdominated by agricultural lands and urban development with the exception of Mont Royal, which is a forest fragment in an exclusively urban setting. The dashed boundary represents the Montreal Metropolitan Community (MMC). Urban development is very dense in downtown Montreal decreasing towards the MMC boundary. 172 Figure 2There was no relationship between the number of caterpillars reported to use a given host plant and either (a) CAPIr or (b) CAPIa scores. These relationships are expected to be positive as a host plant’s acceptability should be indicative of the caterpillar assemblage preference of that host plant relative to other host plants in the community. 173 174 Figure 3A host plant replacement simulation for (a) caterpillar species richness and (b) caterpillar richness in the Mont Royal forest fragment. In these simulations F. grandifolia and A. pensylvanicum were substituted for A. platanoides and R. cathartica (dashed line) and O. virginiana was substituted for F. americana (dotted line). The substituted species were chosen because they had high CAPIr and/or CAPIa scores and commonly share the same general canopy position as the species they replace. Replacement of invasive species with O. virginiana and A. pensylvanicum resulted in an increase of 30% in caterpillar species richness and 40% in caterpillar abundance. Replacement of F. americana with F. grandifolia resulted in an increase of 18% in caterpillar species richness and 37% in caterpillar abundance. 175 176 Table 1 The relationship between host plants and caterpillars shows that (a) host plant (Shannon’s) diversity is a significant descriptor of caterpillar (Shannon’s) diversity when Trail disturbance is accounted for (total model adjusted R2 = 0.45F3,68 = 9.9) but (b)host plant richness is a non-significant descriptor of caterpillar richness when Trail disturbance is accounted for (total model adjusted R2 = 0.27, F3,68 = 20.1). a) Caterpillar Shannon’s diversity = host Shannon’s Diversity + trail index + interactions Independent Variable Standardized t-value p-value Coefficient Host Shannon’s Diversity 0.32 2.2 0.035 Trail Index -2.7 -3.0 0.0037 Host x Trail Interaction 0.29 1.1 0.30 b) Caterpillar Richness = host plant richness + trail index + interactions Independent Variable Standardized t-value p-value Coefficient Host Richness 0.11 0.34 0.51 Trail Index -1.9 -2.2 0.032 Host x Trail Interaction 0.010 0.034 0.97 177 Table 2CAPIa and CAPIr values for host plant trees (sorted in order of decreasing CAPIr values) are calculated as the difference between the observed and the average caterpillar abundances and richness in host plant tree species (see equations 1 and 2). Greater CAPI values indicate that a host plant is more preferred by the caterpillar assemblage. 178 Host Plant Species # of Trees Surveyed Observed AB Average AB* CAPIa Observed SR Average SR* Prunus serotina 19 38 16.5 21.5† 17 9.1 † Amelanchier arborea 35 58 30.0 28.0 21 14.6 † Acer spicatum 20 32 17.4 14.6 16 9.6 † Acer pensylvanicum 206 221 176.4 44.6 52 45.9 † Betula alleghaniensis 10 17 8.5 8.5 11 5.1 Cratageus spp 19 15 16.5 -1.5 11 9.1 Carya cordiformis 15 9 12.9 -3.9 9 7.4 † Ostrya virginiana 145 239 124.0 115.0 39 38.0 Carpinus caroliniana 26 21 19.1 1.9 10 10.3 Malus pumila 11 4 9.4 -5.4 4 5.6 † Fagus grandifolia 229 317 195.5 121.5 46 48.4 Cornus alternifolia 13 3 11.2 -8.2 3 6.5 Ulmus americana 35 19 30.0 -11.0 11 14.6 Prunus nigra 11 3 9.4 -6.4 2 5.6 ‡ Betula papyrifera 18 5 15.6 -10.6 5 8.7 Tsuga canadensis 33 25 28.3 -3.3 10 14.0 ‡ Acer rubrum 12 0 10.3 -10.3 0 6.0 ‡ Rhus typhina 15 4 12.9 -8.9 1 7.4 ‡ Tilia americana 66 24 56.7 -32.7 16 23.1 ‡ Quercus rubra 34 11 29.1 -18.1 6 14.3 ‡ Acer platanoides 39 11 33.3 -22.3 6 15.7 ‡ Fraxinus americana 228 95 194.7 -99.7 38 48.3 ‡ Prunus virginiana 55 26 47.4 -21.4 10 20.5 ‡ Rhamnus cathartica 48 11 41.1 -30.1 7 18.6 * Average SR and AB values are based on the sample size (# of trees surveyed) for a given host plant. **Mostly composed of a mixture of Cratageus punctata Jacq. and Cratageus mollis Scheele and their hybrids. † CAPIa or CAPIr greater than 1 standard deviation above mean expected value. ‡ CAPIa or CAPIr greater than 1 standard deviation below mean expected value. 179 CAPIr 7.9† 6.4† 6.4† 6.1† 5.9† 1.9 1.6 1.0 -0.3 -1.6 -2.4 -3.5 -3.6 -3.6‡ -3.7 -4.0 -6.0‡ -6.4‡ -7.1‡ -8.3‡ -9.7‡ -10.3‡ -10.5‡ -11.6‡ Table 3Host plant frequencies (combined) explained (a) 21.0% of caterpillar abundance, (b) 24.9% of caterpillar richness and (c) 27.9% of caterpillar Shannon’s diversity among quadrats. This was independent of trail index which, when combined with a F. americana interaction term, explained (a) 19.8%, (b) 22.3% and (c) 18.7% of the variances. a) Caterpillar Abundance = host plant frequencies + trail index + interactions Independent Variable Standardized Coefficient Partial R2 Acer pensylvanicum 0.369 0.108 Ostrya virginiana 0.304 0.083 Fagus grandifolia 0.168 0.019 Log 10 (Trail Index) -0.610 0.198 Log10 (Trail Index) * F. americana 0.497 Total Model 0.613 b) Caterpillar Richness = host plant frequencies + trail index + interactions Independent Variable Standardized Coefficient Partial R2 Acer pensylvanicum 0.415 0.150 Ostrya virginiana 0.331 0.099 Log 10 (Trail Index) -0.636 0.223 Log10 (Trail Index) * F. americana 0.556 Total Model 0.590 c) Caterpillar Shannon’s Div = host plant frequencies + trail index + interactions Independent Variable Standardized Coefficient Partial R2 Acer pensylvanicum 0.453 0.175 Ostrya virginiana 0.321 0.094 Ulmus americana 0.189 0.010 Log 10 (Trail Index) -0.574 0.187 Log10 (Trail Index) * F. americana 0.527 Total Model 0.560 180 Linking Statement 3 In Chapter 3 I showed that there is a strong effect of bottom-up forces on insect herbivore richness. This mechanism was strongest when described as assemblage-level preference for specific host plants. Host plant species richness was a poor predictor of insect herbivore species richness. In addition I showed that the replacement of non-native host plants could have significant positive impacts on caterpillar richness and abundance. While this pattern is described in an exclusively “bottom-up” perspective, it is possible that both bottom-up and top-down forces are at work. Bottom-up forces can be described as the quality of foliage – often expressed as the concentration of foliar nutrients. Top-down forces can be described as the relative occupancy of enemies in prospective host plants (these enemies more often tend to be parasitoids in the case of Lepidoptera). Both of these forces may be highly seasonal in their nature. Foliar quality is usually highest in the beginning of the growing season when parasitoid abundance is low. Parasitoid abundance is often high in the latter half of the growing season when foliar quality is low. The question of how foliar quality impacts caterpillar assemblages thus has an intra-seasonal temporal component. In Chapter 4 I relate the intra-seasonal changes in quadrat-scale foliar quality (biotic bottom-up driver) to caterpillar richness and abundance, andI take parasitoids (biotic top-down driver) into consideration late in the growing season (August) to examine their impact. 181 CHAPTER4: Intra-Seasonal Relationships between Insect Herbivores and their Hosts Peter J.T. White1 1 McGill University, Department of Biology 1205 Dr. Penfield Ave., Montreal, Quebec H3A 1B1, Canada 182 Abstract Research has often linked host plant foliar qualities to insect herbivore distribution and performance but rarely have these factors been related to the richness or abundance of an entire caterpillar assemblage or examined in a seasonal context. This “bottom-up” effect can often be complicated by a “topdown” effect driven by the presence of natural enemies. The purpose of this research is to examine how a caterpillar assemblage in temperate broadleaved forest fragments is related to quadrat-scale foliar properties in three sampling windows within a growing season (early June, early July and early August). In addition, I test whether top-down parasitoid pressure supersedes foliar quality as the prime predictor of caterpillar assemblage richness and abundance in the third (August) sampling window. A regression tree analysis suggests that quadrat-scale foliar polyphenol content was important for caterpillar richness and abundance in June and August as a negative control. Foliar phosphorus was the primary predictor for the month of July. Parasitoid pressure superseded polyphenol content as the primary predictor of caterpillar richness in August. Since the foliar quality data, the parasitoid data and the caterpillar assemblage data I use come from different sources, my results should be interpreted with caution. However, there is good evidence to suggest that foliar quality plays an important role in affecting caterpillar assemblages early in the growing season, but that the nature of this effect differs throughout the growing season and can depend on parasitoid presence. 183 Introduction Determining the factors that affect insect herbivore distribution and performance has been an important topic in ecological research over the past four decades (Scriber 1978, Scriber 2010, Strong et al 1984, Wint 1983). These factors mostly fall into two categories: (i) bottom-up effects that pertain to the nutritional or structural quality of foliar resources and (ii) top-down effects that pertain to natural enemies finding and killing (or parasitizing) insect herbivores (Hunter and Price 1992, Mayhew 1997). In recent years, research has aimed at better understanding how these two forces shape insect herbivore communities given their high diversity (Wilson 1987) and the essential role they play in food web structure (Murakami and Nakano 2000, Vasconselos et al. 1996, Goiti et al. 2009) and ecosystem nutrient cycling (Schowalter et al. 1986). In this study, I hope to address two important gaps in the understanding of these tri-trophic relationships – namely (i) the effect of top-down and bottom-up forces on entire insect herbivore assemblages and (ii) the intra-seasonal variation in the nature and magnitude of top-down and bottom-up forces. Bottom-Up Effects: Foliar Quality A bottom-up effect occurs when the distribution or performance of insect herbivores is controlled or limited by aspects of foliar quality (Power 1992) – often defined by the concentration of specific nutrients present in edible foliage (Amwack and Leather 2000, Forkner and Hunter 2000). While different insect 184 herbivore species have different nutrient needs, foliar properties such as high nitrogen content and high water content are generally identified as positive controls of insect herbivore performance and distribution; high toughness and polyphenol (toxin) content are generally identified as negative controls (Amwack and Leather 2002, Choong 1996, Feeny 1970, Matteson 1980, Schowalter et al. 1986, Scriber 1977, Scriber 1878, Scriber and Slansky 1981). Feeny (1970) was among the first to describe this relationship for Lepidoptera larva (caterpillars) when he showed that Winter Moth (Operophtera brumata) early season herbivory was limited by available foliar nitrogen content but late season herbivory was limited by foliar tannin (toxin) content. Subsequent experiments also identified that foliar water content is an important driver of caterpillar performance (development times, weights, etc.) and caterpillar distribution (host plant choices) (Amwack and Leather 2002, Scriber and Slansky 1981). A third foliar quality, phosphorus, has more recently been identified as important to insect herbivore development (Perkins et al. 2004, Woods et al. 2002). While phosphorus is not generally considered of primary importance (like nitrogen or water content) its impact has a strong mechanistic basis for being a positive control as it is used in rRNA formation, which may be very important for fastgrowing organisms like caterpillars (Elser et al 1996). Of the negative control foliar nutrient controls, high foliar “toughness” can cause insect herbivores to eat foliage at a slower rate and interfere with gut processes (Clissold et al. 2009). Slower development times may also make them vulnerable to predators and 185 parasitoids for longer. Toughness is often described as a function of foliar lignin content and foliar fiber content; when these are present in high concentrations, they can result in host plant avoidance by gravid females (Rausher 1981 Slansky and Scriber 1981, Schowalter et al. 1986). Bottom-Up Effects: Foliar Toxins Another class of foliar nutrients, toxins (polyphenols, alkaloids and terpenes) have been both positively and negatively linked to host plant selection depending on the scenario. Feeny (1970) identified three mechanisms by which these types of toxic compounds can negatively affect caterpillars: (1) by acting as a repellent to discourage oviposition, (2) by interfering with protein synthesis and functioning and (3) by acting as a direct toxic compound. A fourth mechanism of impact has since been observed in cases where toxic compounds act as attractants to herbivore enemies, particularly parasitic wasps (DeMoraes et al 1998, Hoballah and Turlings 2001). Empirical tests have often shown diminished larval growth rates, lower larval weight, higher larval mortality and lower overall fitness associated with feeding on foliage high in toxin content (for reviews see: Amwack and Leather 2002, Pasteels et al. 1983, Schowalter et al 1986, Scriber and Slansky 1981). The ability of some insect herbivores to sequester plant toxins as a defense against predators and parasitoids is often juxtaposed to this (Blum 1983, Duffey 1980, Nishida 2002). For example, the wooly bear caterpillar (Grammia geneura) can increase its toxicity to natural 186 enemies by feeding on a diet highin pyrrolizidine alkaloids (Singer 2004a). This can be at the expense of feeding on host plants that result in faster development, but can be an advantageous behaviour where enemies are common. The eastern tent caterpillar (Malacosoma americanum) can sequester foliar hydrocyanic acid found in Prunus host plants (e.g. black cherry, P. serotina and choke cherry, P. virginiana), making their active defense against predators more effective (i.e. a more toxic regurgitation when attacked) (Peterson et al. 1987). There are many other such examples in the literature (e.g. Boppré 1990, Bowers 1992, del Campo et al. 2005, Marsh and Rothschild 1974, Moore et al. 1990) but Pasteels et al. (1983) contend that this phenomenon occurs far more frequently in monophagic and oligophagic insect herbivores than in polyphagic herbivores. To my knowledge this assertion has not been verified and since the ecology and life history of many insect herbivore species is poorly known (with the exception of well-documented defoliators), it is difficult to formally test. Top-Down Effects A top-down effect occurs when the distribution or performance of insect herbivores is controlled or limited by a natural enemy – often a predator or a parasitoid (Hunter and Price 1992). Top-down effects can play a direct role in regulating insect herbivore population size (Holmes et al. 1979, Hooks et al. 2003, Mols and Visser 2002, Sanz 2001) and their presence may force insect herbivores to make sub-optimal host plant choices in search of “enemy free” 187 space (Jeffries and Lawton 1984, Lill et al. 2002). This type of relationship has also been framed in the context of density-dependence – a high density of caterpillars (presumably on a high quality host plant) can become a magnet for would-be predators and parasitoids (Mayhew 1997). Because of this, some caterpillar species change their choice behaviour in favour of lower quality hosts. There is strong evidence to support this indirect negative effect of natural enemies on caterpillars. In northeastern North American forests tree genera with high parasitoid loads tend to have fewer caterpillar species than genera with low parasitoid loads (Lill et al. 2002). The pine beauty moth (Panolis flammea) has a higher survival rate on low quality host plants because of the prominence of parasitoids associated with high quality host plants (Leather and Walsh 1993).In field ecosystems, insect herbivore diversity is initially determined by host plant diversity (i.e. at initial community assembly), but it is regulated predator and parasitoid density thereafter (Siemann et al. 1998). Current Gaps in Foliar Quality Research There are two gaps in host plant host plant-caterpillar research that need to be addressed. First, the vast literature on foliar quality-caterpillar relationships has focused primarily on how foliar quality impacts single or only a few caterpillar species (e.g.Camara 1997, Clancy and King 1993, Feeny 1970, Hagen and Chabot 1986, Hough and Pimentel 1978, Hunter and Lechowicz 1992, Karban and English-Loeb 1997, Lill and Marquis 2001, Rausher 1981, Singer et al. 2004a, 188 Woods et al. 2002). These types of studies are often of limited utility in management or conservation contexts because they are difficult to apply to larger species groups. Important pests such as Alsophila pometaria, Lymantria dispar, Malacosoma disstria, Hyphantria cunea or H. textorare often studied but overall caterpillar assemblages are rarely considered (but see Karban and Ricklefs 1984). In other contexts, assemblage-based approaches have been very successful in describing factors that affect richness and abundance.For example, in a landscape-disturbance context, assemblage based approaches have demonstrated that species richness in moth assemblages is significantly higher in agricultural fields adjacent to forest habitat compared to those that were distant (Ricketts et al. 2001). In a forest fragmentation context, an assemblage-based approach was used to show how the host plant composition of a given fragment offset the species-loss expected due to small fragment size (Summerville 2004, Summerville and Crist 2003). Still other contexts exist (e.g. host plant preferences - White, Chapter 3), but an assemblage-based approach has not yet been used to examine bottom-up and top-down effects on Lepidoptera. An assemblage-based approach also has the advantage of including microlepidopteran species that are rarely studied because their taxonomy is very poorly known, even in well-studied locales. A second gap in our knowledge is that the impact of foliar quality on caterpillars is not often examined in a seasonal context. This is in part a function of the first gap (studies with one or few species); most species are very short 189 lived and it is only possible to study them in their limited seasonal timeframe. Yet, at the Lepidoptera assemblage level it is well known that significant composition changes occur over the course of the growing season - to such a degree that experimental designs have to take this factor into account. Because of these intra-seasonal compositional changes, it is possible that the balance between top-down and bottom-up regulators may change as well. It is not wellknown how early season versus late season caterpillars distribute themselves with respect to foliar qualities. In one rare example, Niemela and Haukioja (1982) showed that host plant use patterns among Finnish Lepidoptera change from the beginning to the end of the growing season. They attributed these changes to the timing of shoot growth patterns throughout the season, but did not have the data to implicate specific changes in foliar qualities. There are some examples of single insect herbivore species changing their feeding habits from early to late in their flight seasons – often tracking changes in host plant quality (e.g. Feeny 1970, Rausher 1981) – but this falls under the purview of the “first gap” above (i.e. most studies focus on one or few caterpillar species). Many tree species in temperate broadleaved forests have very significant seasonal trends in foliar qualities. Ricklefs and Matthew (1982) documented foliar qualities of 34 broad-leaved deciduous tree species in southern Canada and showed that average foliar nitrogen dropped by 19% and average leaf water content dropped by 14% from early June to August (Table 1). Since nitrogen and water content levels are often related to faster caterpillar development times 190 and larger population sizes (Matteson 1980, Scriber 1977, Scriber and Slansky 1981)a loss of nitrogen- or water-rich host plants could have significant consequences. Phosphorus, linked to caterpillar development, is 47% lower in early July than in early June and also 41% lower in early July than in early August (Table 1). Toughness, lignin content and fiber content also have strong intraseasonal variation. Toughness peaks late in the growing season whereas lignin content and fiber content peak mid-season (on average) (Table 1); all three are lowest early in the growing season. Like foliar quality, parasitoid (or predator) pressure can be highly seasonal, peaking multiple times or in the latter half of the growing season (Correa-Ferreira and Moscardi 1995, Damman 1987, Kato 1994, Liu et al. 2000, McAuslane et al. 1993, Okada 1989, Peña et al. 1996, Wong et al. 1984). It thus becomes quite difficult to analyze something like herbivorous caterpillar richness in a system where both foliar quality and parasitoid pressure are seasonally variable. For this type of analysis seasonally sensitive tri-trophic models may be warranted. Tri-Trophic Relationships Tri-trophic relationships occur when both bottom-up pressure (effects of foliar quality) and top-down pressure (effects of natural enemies) impact the performance or distribution of an herbivorous species (Leibold 1989). While the strength and significance of top-down and bottom-up forces vary depending on 191 the system in question (as discussed in the sections above), many authors argue for the primacy of bottom-up forces pointing out that all levels of the tri-trophic system will cease to exist in the absence of the lowest trophic level (i.e. no host plants = no herbivores = no natural enemies) butnot in the absence of the highest trophic level as herbivores can continue to graze on plants in the absence of natural enemies (Hunter and Price 1992, Power 1992, Siemann et al. 1998). Primacy aside, the seasonal nature of bottom-up foliar quality and topdown natural enemy abundance suggests that tri-trophic relationships should be studied in seasonal contexts. This is particularly appropriate when the unit of examination is a species assemblage and the period of examination is an entire foliar growth-season. In one of the few studies examining tri-trophic relationships in a seasonal context, Kato (1994) showed that a bi-voltine insect herbivore was distributed with respect to foliar quality in the first generation when parasitoids are absent, but distributed with respect to top-down forces (i.e. parasitoid presence) in the second generation when parasitoids are abundant. Further investigation is required to test if this type of pattern applies to an entire herbivorous assemblage. Objective The objective of this study is to use an assemblage-based approach to examine how caterpillar richness and abundance are related to foliar quality over the course of a growing season and, in addition, to determine if top-down parasitoid 192 pressure impacts this relationship. The base expectations are that caterpillar richness and abundance will be (i) positively related to foliar nitrogen content, water content and phosphorus content, (ii) negatively related to foliar toxin content, toughness, lignin content and fiber content, (iii) positively related to toxin content in the scenario where assemblage-level toxin sequestration occurs, and (iv) negatively related to parasitoid pressure. I hypothesize that the relative effects of these factors will change over the course of the growing season. Specifically, positive foliar properties (nitrogen, water content and phosphorus) should play a stronger role determining caterpillar distribution amongst trees early in the growing season when quality is high; negative foliar properties (toughness, lignin content, fiber content) should play a stronger role late in the growing season. Parasitoid pressure should also have a higher impact on caterpillar richness and abundance late in the growing season when parasitoids tend to be more common. Given that the impact of foliar polyphenol content could be either positive or negative it is difficult to predict how the magnitude of its impact may change over the growing season. Methods Study Area I surveyed caterpillars in four forest fragments located on the Monteregian Hills located in the St. Lawrence River valley in southeastern Quebec, Canada (45°30'N, 73°30'W to 45°24'N, 72°35'W; Figure 1). During pre-settlement times 193 all these Monteregian Hills would have had broadly similar forests embedded in a more or less continuously forested landscape (Richard and Grondin 2009). They now exist as a series of remnant forest fragments isolated in a developed landscape. Eighteen 400 m2 (20 m x 20 m) quadrats were placed randomly within each of the four sampling sites on Monteregian Hills. The first site was at Parc Mont Royal - an urban park in the middle of Montreal, a city of 3.5 million inhabitants. The park has a secondary-growth forest that covers approximately 100 of 190 total hectares. The second site was at Parc National du Mont St. Bruno – a protected provincial park in the eastern suburbs of greater Montreal. The forest at Mont St. Bruno is a secondary-growth forest that covers more than 500 ha of the 790 ha park. The forests are only minimally disturbed by a series of trails designated for public use. The third site was at the Gault Nature Reserve on Mont St. Hilaire – a protected park and UNESCO Biosphere Reserve, 38 km east of Montreal. The forest at Mont St. Hilaire is primarily an old-growth broadleaf deciduous forest covering most of the 1000 ha reserve. This site has a long history of protection dating back to the 17600s (Arii 2004, Maycock 1961). The fourth site was at Mont Shefford – one of the easternmost Monteregian Hills, 70 km east of Montreal. I conducted caterpillar sampling in three sub-sites around the hill. Six quadrats were in a 100 ha semi-disturbed patch of forest on the west side of the Mont Shefford and six quadrats in a 25 ha patch of forest on the east side of the hill set aside as a forested community park. A quarter of the former site was a sugar bush and 194 there were numerous trails used by deer-hunters throughout the area. The community park had several recreational trails, but off-trail use was discouraged. Six quadrats were in a third sub-site, Parc de la Yamaska, a provincial park just north of Mont Shefford. Caterpillar Surveys and Identification Caterpillars were surveyed at each quadrat in three time periods in the growing season of 2009: (i) June 1 to 6, (ii) July 4 to 9 and (iii) August 3 to 6. At each quadrat, ten sugar maples (Acer saccharum) and up to 10 of every other tree species between 3 and 10 cm dbh were sampled for caterpillars (Table 2). Each target tree was sampled by striking the bole and lower branches ten times with a 20 oz, 30" aluminum baseball bat; dislodged caterpillars were caught on a 1 m2 sheet held under the point of striking. All macrolepidopteran caterpillars collected were identified to species with a dissecting microscope using Wagner’s Caterpillars of Eastern North America Field Guide(2005). Microlepidopteran caterpillars were counted and identified to morphospecies for lack of an accurate identification guide. Any caterpillar that was collected in an early instar was reared to a late instar so that positive identifications could be made. Measuring the Bottom-Up Effect I computed the foliar quality of the trees surveyed at each quadrat using seasonal leaf properties reported by Ricklefs and Matthew (1982) for the study 195 species in similar forests in southern Canada. From their data, the foliar qualities I calculated for each quadrat were: percent water content, toughness (measured in grams), percent polyphenol content, percent nitrogen content, percent phosphorus content, percent lignin content and percent fiber content. In quadrats where more than one tree species was present (i.e. all but two quadrats), quadrat foliar qualities were averaged across host plant species weighted by their relative basal areas in a given quadrat. In other words, the caterpillar assemblage is related to an aggregated estimate of foliar quality in each quadrat; this is an approach routinely used in assessing the relationship between foliar traits and ecosystem functions such as decomposition (Kakazou et al 2006). Quadrats that contained host plants undocumented by Ricklefs and Matthew (1982) were included in the analysis only if three or fewer trees were undocumented. This resulted in foliar quality profiles being constructed for 52 of 72 possible quadrats. The tree species undocumented by Ricklefs and Matthews which resulted in the exclusion of quadrats were: Abies balsamea, Acer platanoides, Acer saccharinum, Acer spicatum, Amelanchier arborea, Crataegus spp., Dirca palustris, Picea glauca, Rhamnus cathartica, Prunus nigra, Robinia pseudoacacia and Tsuga canadensis. Measuring the Top-Down Effect An index of parasitoid pressure was created using the Forest Insect Survey data compiled by Natural Resources Canada. These data record the 196 occurrence of several common species of caterpillars sampled from coniferous and mixed wood forests across Ontario and Quebec between 1937 and 1949. The focal caterpillar species that Natural Resources Canada provided records for were: Acleris variana, Archips cerasivorana, Caripeta divisata, Hyphantria cunea, Lambdina fiscellaria, Nepytia canosaria and Semiothisa granitata (now Macaria granitata). Each of these species has a set of collection records documenting caterpillars collected on specific dates from specific host plants at specific locations. Each record also documented the number of parasitoids (and sometimes parasitoid taxa or species) that were observed emerging from collected caterpillars. The parasitoid emergence data associated with these records were divided into three subsets to match the three sampling periods that I used for my caterpillar collecting; a two week buffer was given around each sampling window for matching parasitoid emergence dates. Parasitoid emergence data were then pooled across the seven focal caterpillar species so that a parasitism likelihood score for each host plant could be created, simply calculated as the number of parasitoids observed divided by the number of caterpillars observed (both specific to a given host plant) multiplied by 100 to give the percentage of samples that were parasitized. There were not enough caterpillar collection records from June and July to construct a comprehensive set of host plant specific parasitoid pressure scores, so these months were omitted from the analysis; only 7 of my host plants were sampled by Natural Resources in the June sampling window and 11 in the July sampling window (see 197 Table 2 for a complete list). Finally, quadrat-specific parasitoid pressure values were calculated as the average parasitoid pressure of all trees present in each given quadrat. Analysis Quadrat-scale foliar qualities were used in a regression tree analysis (De’Ath and Fabricus 2000) to describe caterpillar richness and abundance across quadrats in June, July and August. Regression trees are good instruments to use when the dependent variable may not be linearly related to one or more independent variables homogenously throughout its numeric range. For example, when caterpillar richness is low it may have a strong relationship with nitrogen content and water content, but when it is high it may be limited by toughness and lignin. One of the advantages of regression trees is that they require no foreknowledge of the nature of the relationships between dependent and independent variables prior to analysis. In my case, a regression tree analysis is an appropriate choice given that the relationship between caterpillar richness and foliar nutrients is likely complex and non-linear (Futuyama et al 1984, Karban and Ricklefs 1984). Prior to regression tree computation, caterpillar abundance was log10 transformed to reduce the impact that outlier data points had on dataset variance. A second set of regression trees were calculated for August, including August parasitoid pressure values at each quadrat. Regression trees were pruned to omit nodes that had a complexity parameter of less than 0.05. 198 Methodological Assumptions There are two assumptions inherent to the methodology of calculating parasitoid pressure that should be taken into consideration before examining the regression tree results. First, I made the assumption that parasitoid host plant pressures in 2009 are similar to those observed between 1937 and 1949. In some areas, forests may have undergone significant compositional changes in the past six decades that may have caused certain parasitoids to adjust their preferred host plant targets for finding caterpillar hosts. For example, the caterpillar species Archips cerasivorana was often found on Prunus (cherry) host plants. If cherry trees have been significantly reduced in forest fragments over the past 60 years Archips may have adjusted its feeding habits to incorporate a much larger proportion of other host plant species. It is not known, were this the scenario, whether parasitoids that typically parasitize Archips and other Prunus feeders would be associated with Prunus in the same way they were from 1937 to 1949 or if they would become more closely associated with other host plant species. Even though most of the caterpillar species I analyze in this study are widely polyphagous, changes to caterpillar feeding preferences over the past 60 years could bring the parasitoid-association data into question because it is derived from parasitized caterpillar capture records. Second, the caterpillar assemblage I draw parasitoid data from consists of only seven species. I am assuming that the parasitoid-host plant associations 199 observed amongst these seven species are representative of the parasitoid-host plant associations present throughout the entire caterpillar assemblage. The accuracy of both these assumptions is not immediately testable, but is an important consideration. Extensive searches failed to find additional data with which to test the questions presented in this chapter. Collecting parasitoid data myself was not feasible given the timeframe and funding available for this project. As such, the choices that remained were to conduct the analyses as described with the understanding that any results may require further testing before definitive conclusions are drawn, or to leave parasitoids out of the analysis. I chose the former. Results In the June sampling period, surveys yielded 845 caterpillars from 26 macrolepidoptera species and 18 microlepidoptera morphospecies. In the July sampling period, surveys yielded 489 caterpillars from 25 macrolepidoptera species and 25 microlepidoptera morphospecies. In the August sampling period, surveys yielded 562 caterpillars from 32 macrolepidoptera species and 35 microlepidoptera morphospecies. May and June shared 22 of a combined 72 species; July and August shared 26 of 91 combined species; June and August shared 16 of 95 combined species. 200 Bottom-Up Foliar Quality Quadrat water content, nitrogen content and phosphorus content is highest in June (Figure 2a, b, c). Both water content and nitrogen content are lowest in August; phosphorus content is lowest in July. Quadrat toughness, fiber content and lignin content are lowest in June (Figure 2d, e, f). Both lignin and fiber content peak in July while toughness peaks in August. Average quadrat polyphenol content is lowest in August and highest in July (Figure 2g). Top-Down Pressure Parasitoid pressure is highest in Acer saccharum (13.6%), Ulmus americanum (13.3%) and Tsuga canadensis (11.9%). It is lowest in Acer pensylvanicum (0%), Cratagus spp (0%) and Fagus grandifolia (0%). These three host plant species also had a very low sample size (of caterpillars) within the Natural Resources Canada data (Table 3), suggesting that the results may be unreliable. Further testing shows no evidence a significant correlation between the number of caterpillars collected and parasitoid pressure across host plants (i.e. correlating column 2 and column 4 in Table 3: p = 0.19, R2 = 0.09, F1,19 = 1.85). Seasonality of Bottom-Up Effects Regression tree analyses of the June sampling period showed that foliar quality explained 0.41 of the variance in caterpillar species richness and 0.44 of the 201 variance in caterpillar abundance among quadrats (Figure 3a, b). The most important foliar quality in both regression trees was polyphenol content, which explained 0.33 of the variance in caterpillar richness and 0.44 of the variance in caterpillar abundance. Quadrats where the polyphenol content was greater than 10.9% (left hand fork of the regression tree 3a, b) had significantly fewer species and total caterpillars compared to quadrats where polyphenol content was lower. Phosphorus content explained 0.10 of the variance in the caterpillar richness regression tree (positive relationship, regression tree 3a). In the July sampling period, regression tree analyses showed that foliar quality explained a 0.42 of the variance in caterpillar species richness and 0.44 of the variance in caterpillar abundance among quadrats (Figure 3c, d). In both cases, phosphorus content was the most important foliar quality; sites with high phosphorus content (greater than 0.19%) had lower caterpillar richness and abundance than sites where phosphorus content was lower. The similarity between these two regression trees can be attributed to a strong correlation between caterpillar richness and abundance among quadrats in July (R = 0.82, F1,50 = 2.27, p< 0.0001). Lignin content was a significant variable in both trees, positively linked to caterpillar richness and abundance, but explaining only 0.06 and 0.10 of the variance respectively. Polyphenol content explained an additional 0.06 of the variance in the caterpillar richness tree (Figure 3c). In the August sampling period, regression tree analyses showed that foliar quality explained 0.30 of the variance in caterpillar species richness and 202 0.20 of the variance in caterpillar abundance among quadrats (Figure 3c, d). The most important foliar quality in the caterpillar richness tree was fiber content. Quadrats where fiber content was less than 28.5% (left hand fork of regression tree 3e) contained fewer caterpillar species than quadrats where fiber content was higher. For caterpillar abundance, quadrats that had high polyphenol content (greater than 10.3%) had significantly fewer species than quadrats where polyphenol content was lower. Polyphenol was additionally important in the caterpillar richness regression tree and provided a second node in the caterpillar abundance regression tree. Lignin content explained a small amount of variance in the caterpillar richness regression tree. Tri-Trophic Relationship When parasitoid pressure was added to the August analyses, there were notable changes to both regression trees. Quadrats where parasitoid pressure was greater than 6 % had significantly lower caterpillar richness and abundance than quadrats where parasitoid pressure was lower. In the caterpillar richness regression tree parasitoid pressure explained 0.28 of caterpillar variance (Figure 4a), superseding all of the foliar quality variables that were significant in Figure 3e. In the caterpillar abundance tree parasitoid pressure superseded polyphenol content for the primary regression tree node explaining 0.18 of the caterpillar variance (Figure 4b). Water content and phosphorus content explained smaller proportions of caterpillar abundance (0.10 and 0.06 respectively). 203 Discussion There appears to be a transition in the nature of bottom-up effects over the course of the growing season. Early in the season, low polyphenol content is the most important variable in regression tree models. In the middle of the summer, foliar phosphorus contentbecomes the most important variable, with low levels corresponding to high levels of caterpillar richness and abundance. By late summer, polyphenol and fiber content are both important, but when a tritrophic relationship is considered top-down parasitoid pressure supersedes both foliar variables to explain a higher proportion of the variance in caterpillar richness and abundance. There was no evidence to support the hypothesis that polyphenols are being selected for at the assemblage level, which would be expected if toxin sequestration for defense was a primary driver of caterpillarrichness or abundance among quadrats. Instead, the opposite was prevalent (a negative relationship between assemblage richness/abundance and polyphenol content). This set of results should however be interpreted with caution and as a preliminary finding rather than as a definitive test of these relationships across seasons. Bottom-Up: Foliar Quality Traditionally, nitrogen and water content have been identified as the more important foliar drivers of caterpillar distribution and performance; the negative effect of phosphorus I identified has received little attention. The insignificance 204 of nitrogen and water content in most intra-seasonal models could suggest that nitrogen and water content are not limiting foliar traits for caterpillars in northeastern temperate forests. These variables are often associated with feeding preferences in feeding trials, but the cost associated with redispersal in search of more nitrogen-rich and water rich foliage may outweigh the benefits. This non-effect of nitrogen and water content was also reported by Karban and Ricklefs (1984) who examined Lepidoptera assemblages in a similar forest type using the same foliar nutrient data. They concluded that top-down forces (i.e. parasitoids) or historical factors (i.e. that limit geographical distribution) might instead be responsible for Lepidoptera distribution amongst trees. My analysis agrees with this suggestion for the month of August (i.e. late in the growing season) but identified polyphenol content and phosphorus content as the driving nutrient foliar properties for June and July regression trees. The negative relationship between foliar phosphorus and caterpillars has a few possible explanations. The expectation was for a positive relationship, reflecting increased growth and shorter instar durations observed in some species (Apple et al. 2009, Perkins et al. 2004). One possibility is that foliar phosphorous is linked to other foliar qualities and the negative link described in the regression trees is merely coincidental. To test this, I ran a multiple regression analysis of quadrat foliar qualities (in July) where phosphorus content is a function of lignin content, fiber content and toughness. The results of this regression identify lignin content and toughness as significant positive predictors 205 of phosphorus content (p < 0.005, adj R2 = 0.24, F3,48 = 6.22) while fiber content is a negative predictor. Both toughness and lignin content make leaves less palatable to caterpillars (Choong 1996, Scriber 1977). Therefore, it is possible that quadrat-selection by Lepidoptera is driven by preference for low toughness and low lignin content and the negative relationship between caterpillars and phosphorus is merely a byproduct. This is exemplified by the foliar qualities of three very common host plants: striped maple (Acer pensylvanicum) hophornbeam (Ostrya virginiana) and white ash (Fraxinus americana). Striped maple and hophornbeam are highly preferred host plants for caterpillars whereas white ash is often avoided (White - Chapter 3); these three host plant species make up nearly 30% of the host plants I sampled. According to Ricklefs and Matthew (1982) the former two species have low toughness (453 g and 539 g respectively) and low phosphorus content (0.12 % and 0.16 % respectively) and the latter species has high toughness (672 g) and high phosphorus content (0.25 %) (all are July values). A similar pattern is evident with lignin content. Another possible explanation of the observed negative relationship between phosphorus and caterpillars is that the impact of dietary phosphorus is being mediated by another dietary element. Clancy and King (1993) suggest that this may be the case with the spruce budworm and identify dietary magnesium as the likely mediating element. In their research they showed that there was a negative relationship between dietary phosphorus intake and pupal mass and survival among caterpillars at high levels of dietary magnesium intake. At my 206 study quadrats the ratio of P to Mg in June is 1.64; in July it is 0.81 reflecting a 38% increase in magnesium and a 32% decrease in phosphorus from June to July (data not shown). It is not known whether or not (or to what extent) Clancy and King’s (1993) research on a conifer-feeding caterpillar species would be relevant to a deciduous-feeding caterpillar assemblage, but the dramatic change in relative foliar magnesium and potassium I observe among quadrats is consistent with a magnesium-mediating-phosphorus scenario. Bottom-Up: Toxins The negative relationship between the caterpillar assemblage and foliar polyphenol content is an example of where processes applicable to singlespecies are not scalable to entire assemblages. Foliar polyphenol content levels were derived from data published by Ricklefs and Matthew (1982),who state that the method they used to measure polyphenol content in leaves was “not specific for polyphenols but may also measure flavonoids and other easily oxidized compounds…”Thus, high polyphenol content levels in any two given host plant species may be representative of very different toxic compounds that may be tolerated (or sequestered) by different caterpillar species. While many examples of species-specific toxin sequestration exist (e.g. Barbehenn et al. 2005, Nishida 2002, Marsh and Rothschild 1974, Peterson et al. 1987, Singer 2004),the relationship between an entire insect herbivore assemblage has not yet formally been explored. Most of the positive relationships between 207 caterpillars and toxins are very species specific and unrelated Lepidoptera rarely tolerant to the same toxin (Nishida 2002). These types of species- and generaspecifictoxic compound tolerances may explainthe nature of the negative relationship between caterpillars and quadrat-scale polyphenols; caterpillartoxin relationships do not scale up to entire assemblages. An increase in quadratscale “polyphenols” would thus only truly be beneficial for caterpillars if the group of compounds in question could be sequestered by most of the caterpillars in the assemblage, which is not often the case (Pasteels et al. 1983). Top-Down: Parasitoids My analyses suggest that late-season top-down control of caterpillar assemblages may be occurring. It was not possible to examine parasitoid data in a seasonal context but the August analyses (Figure 4a, b) showed that top-down parasitoid pressure was negatively associated with caterpillar richnessand abundance late in the growing season. A negative impact of parasitoids or predatorson caterpillars is well established,(Gripenberg and Roslin 2007, Hunter et al. 1997, Preszler and Boecklen 1996, Scheirs and Bruyn 2002) but it has rarely been examined in relation to changes in bottom-up variables throughout a growing season. This raises questions about my June and July analyses– would the regression trees in Figure 3a-d look different if parasitoid data were taken into consideration? The reason that August parasitoids were the only ones included in my analyses was because of the limited parasitoid data within June 208 and July caterpillar records. I decided to try to address this question further with a simple post-hoc analysis examining an additional parasitoid dataset provided by Natural Resources Canada that detailed parasitoid records (without host plant associations) from the same period (1937 to 1949). Out of 487 emergence records in this dataset from mid-May to mid-August pooled across 13 parasitoid species, 293 were from the third sampling period (mid-July to mid-August - this includes a sampling window buffer as described in the methods), 173 were from the second sampling period (mid-June to mid-July) and only 21 were from the first sampling period (mid-May to mid-June) (Table 4). Based on the scarcity of parasitoids documented early in the growing season it seems unlikely that parasitoids would play a large role driving caterpillar assemblage richness or abundance in early June. The large increase in parasitoid records from June into July may be synonymous with an increasing significance of their impact on caterpillar assemblages. Gratton and Denno (2003) showed this type of trend with Prokelisia planthoppers that were distributed with respect to quadrat quality early in the season but with respect to predator (spider) densities late in the season. On an assemblage level, Morais et al (1999) documented a strong seasonal trend in tropical caterpillar abundance and postulated that the significant decline they observed towards the end of the season was likely due to an increase in predator and parasitoid activity. Both of these examples provide interesting parallels to my results and show that there is an empirical basis for an increased importance of late-season top-down effects. 209 Conclusion An assemblage-based approach has become more popular in studies of forest Lepidoptera and has been applied to landscape fragmentation questions (Ricketts et al. 2001, Summerville and Crist 2003), human-disturbance problems (White et al. 2011) and conservation problems (White - Chapter 3). In this study, I documented a change in the nature of the bottom-up relationship between host plant foliar quality and an herbivorous caterpillar assemblage over the course of a growing season. My results also suggested that late-season top-down effects might be greater than early-season top-down effects, at least in the case of caterpillar parasitoids. These results should be treated as preliminary, and a more direct measurement of bottom-up and top-down variables needs to be conducted before definitive conclusions can be made. That said, there is good reason to believe that many of the findings reported in this study are accurate. First, early season foliage tends to be more nutritious than late season foliage (Feeny 1970, Ricklefs and Matthew 1980). Host plant rejection and reselection (by larval ballooning or crawling)is more likely to happen when the likelihood of finding a high quality host plant is great (Mayhew 1997). Late in the season when average host quality is lower, caterpillars would be more likely to settle for a poor quality host because the chances of finding a better host are small. These observations support a hypothesis that bottom-up control is the predominant driver in early season tri-trophic relationships. Second, in my study system parasitoid abundance is very low in June and very high in August. While this in 210 itself is not evidence for the primacy of top-down effects in August, the scarcity of parasitoids in June supports a non-top-down (parasitoid) control scenario early in the growing season. While other top-down enemies may factor into the tri-trophic relationship, demand for prey can be considerably higher after predator young become independent, later in the growing season (Adams et al. 1991, Drent and Daan 1980). At the most basic level this study has shown that the nature of top-down and bottom-effects are probably variable across the length of a growing season and that tri-trophic analyses should take this intraseasonal variance into account. Acknowledgements I am very grateful to C. Buddle, P. Peres-Neto, K. Summerville, R. Feldman, J. Messier, S. Estrada B. McGill and M.J. Lechowicz for comments an early draft;B. McGill was also instrumental in helping to revise and refine the final draft. The vegetation and Lepidoptera field surveys were conducted with the help of R. MacKenzie and M. VonButtlar. The parasitoid data was assembled with the help of Isabelle Ochoa at the Canada Forest Service’s Great Lakes Forestry Centre. I would also like to thank A. Mochon (SEPAQ), D. Rodrigue (SEPAQ) and J. Lapalme (Les Amis de la Montagne) who were consultants for field work planning at Parc de la Yamaska, Parc Mont St Bruno and Parc Mont Royal (respectively). This research was funded by NSERC. 211 References Adams, N. J., Abrams, R. W., Siegfried, W. R., Nagy, K. A. and Kaplan, I. R. 1991. Energy expenditure and food consumption by breeding Cape gannets Morus capensis. – Marine Ecology Progres Series 70: 1-9. Apple, J. L., Wink, M., Wills, S. and Bishop, J. G. 2009. Successional change in phosphorus stoichiometry explains the inverse relationship between herbivory and lupin density on Mont St. Helens. - PLoS ONE 4: e7807. Arii, K. 2004. Ecology of American beech and sugar maple in an old-growth forest. Department of Biology. - McGill University. Awmack, C. S. and Leather, S. R. 2002. Host plant quality and fecundity in herbivorous insects. - Annual Review of Entomology 47: 817-844. Barbehenn, R., Cheek, S., Gasperut, A., Lister, E. and Maben, R. 2005. Phenolic compounds in red oak and sugar maple leaves have prooxidant activities in the midgut fluids of Malacosoma disstria and Orgyia leucostigma caterpillars. - Journal of Chemical Ecology 31: 969-988. Blum, M.S. 1983. Detoxication, Deactivation, and Utilization of Plant Compounds by Insects. In – Plant Resistance to Insects. Ed. Hedin, P.A. American Chemical Society, Las Vegas, Nevada. Boppré, M. 1990. Lepidoptera and pyrrolizidine alkaloids exemplification of complexity in chemical ecology. – Journal of Chemical Ecology 16: 165185. 212 Bowers, M. D. 1992. The evolution of unpalatability and the cost of chemical defense in insects. – In: Insect Chemical Ecology, An Evolutionary Approach. Eds: Roitberg, B. D. and Isman, M. B. Chapman and Hall, New York, NY. Camara, M. D. 1997. Predator responses to sequestered plant toxins in buckeye caterpillars: Are tritrophic interactions locally variable? - Journal of Chemical Ecology 23: 2093-2106. Choong, M. F. 1996. What Makes a Leaf Tough and How This Affects the Pattern of Castanopsis fissa Leaf Consumption by Caterpillars. - Functional Ecology 10: 668-674. Clancy, K. M. and King, R. M. 1993. Defining the Western Spruce Budworm's nutritional niche with response surface methodology. - Ecology 74: 442454. Clissold, F.J., Sanson, G.D., Read, J. and Simpson, S.J. 2009. Gross vs. net income: how plant toughness affects performance of an insect herbivore. – Ecology 90: 3393-3405. Correa-Ferreira, B.S. and Moscardi, F. 1995. Seasonal occurrence and host spectrum of egg parasitoids asociated with soybean stink bugs. – Biological Control 5: 196-202. Damman, H. 1987. Leaf quality and enemy avoidance by the larvae of a pyralid moth. – Ecology 68: 88-97. 213 De’Ath, G. and Fabricus, K. E. 2000. Classification and regression trees: a powerfu yet simple technique for ecological data analysis. – Ecology 81: 31783192. del Campo, M. L., Smedley, S. R. and Eisner, T. 2005. Reproductive benefits derived from defensive plant alkaloid possession in an arctiid moth (Utetheisa ornatrix). – Proceedings of the National Academy of Science 102: 13508-13512. DeMoraes, C. M., Lewis, W. J., Paré, P. W., Alborn, H. T. and Tumlinson, J. H. 1998. Herbivore-infested plants selectively attract parasitoids. - Nature 393: 570-573. Denno, R. F., Larsson, S. and Olmstead, K. L. 1990. Role of enemy-free space and plant quality in host-plant selection by Willow Beetles. - Ecology 71: 124137. Drent, R. H. and Daan, S. 1980. The prudent parent: energetic adjustments in avian breeding. - Ardea 68:225-52 Duffey, S.S. 1980. Sequestratin of plant natural products by insects.- Annual Reviews of Entomology 25: 447-477 Elser, J.J., Dobberful, D.R., Mackay, N.A. and Schampel J.H. 1996. Organism size, life history, and N:P stoichiometry. BioScience 46: 674-684. Feeny, P. 1970. Seasonal changes in oak leaf tannins and nutrients as a cause of spring feeding by Winter Moth caterpillars. - Ecology 51: 565-581. 214 Forkner, R. E. and Hunter, M. D. 2000 What goes up must come down? Nutrient addition and predation pressure on oak herbivores. – Ecology 81: 15881600. Gratton, C. and Denno, R. F. 2003. Seasonal shift from bottom-up to top-down impact in phytophagous insect populations. - Oecologia 134: 487-495. Gripenberg, S. and Roslin, T. 2007. Up or down in space? Uniting the bottom-up versus top-down paradigm and spatial ecology. - Oikos 116: 181-188. Hagen, R. H. and Chabot, J. F. 1986. Leaf anatomy of maples (Acer) and host use by Lepidoptera larvae. - Oikos 47: 335-345. Hoballah, M. E. F. and Turlings, T. C. J. 2001. Experimental evidence that plants under caterpillar attack may benefit from attracting parasitoids. Evolutionary Ecology Research 3: 553-565. Holmes, R. T., Schultz, J. C. and Nothnagle, P. 1979. Bird predation on forest insects: an exclosure experiment. – Science 206: 462-463. Hooks, C. R. R., Pandey, R. R. and Johnson, M. W. 2003. Impact of avian and arthropod predation on lepidopteran caterpillar densities and plant productivity in an ephemeral agroecosystem. – Ecological Entomology 28: 522-532. Hough, J. A. and Pimentel, D. 1978. Influence of host foliage on development, survival, and fecundity of the Gypsy Moth. - Environmental Entomology 7: 97-102. 215 Hunter, A. F. and Lechowicz, M. J. 1992. Foliage quality changes during canopy development of some northern hardwood trees. - Oecologia 89: 316-323. Hunter, M. D. and Price, P. W. 1992. Playing chutes and ladders: heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. – Ecology 73: 724-732. Hunter, M. D., Varley, G. C. and Gradwell, G. R. 1997. Estimating the relative roles of top-down and bottom-up forces on insect herbivore populations: A classic study revisited. - Proceedings of the National Academy of Sciences 94: 9176-9181. Jeffries, M. J. and Lawton, J. H. 1984. Enemy free space and the structure of ecological communities. – Biological Journal of the Linnean Society 23: 269-286. Kakazou, E., Vile, D., Shipley, B., Gallet, C. and Garnier, E. 2006. Co-variations in litter decomposition, leaf traits and plant growth in a Mediterranean oldfield succession.- Functional Ecology 20: 21-30. Karban, R. and English-Loeb, G. 1997. Tachinid parasitoids affect host plant choice by caterpillars to increase caterpillar survival. - Ecology 78: 603611. Karban, R. and Ricklefs, R. E. 1984. Leaf traits and the species richness and abundance of Lepidopteran larvae on deciduous trees in southern Ontario. - Oikos 43: 165-170. 216 Kato, M. 1994. Alternation of Bottom-up and Top-down Regulation in a Natural Population of an Agromyzid Leafminer, Chromatomyia suikazurae. Oecologia 97: 9-16. Leather, S.R. and Walsh, P.J. 1993. Sub-lethal plant defenses: the paradox remains. – Oecologia 93: 153-155. Leibold, M. A. 1989. Resource edibility and the effects of predators and productivity on the outcome of trophic interactrions. – The American Naturalist 134: 922-949. Liu, S., Wang, X., Guo, S., He, J. and Shi, Z. 2000. Seasonal abundance of the parasitoid complex associated with the diamondback moth, Plutella xylostella (Lepidoptera: Plutellidae) in Hangzhou, China. - Bulletin of Entomological Research 90: 221-231. Lill, J. T. and Marquis, R. J. 2001. The effects of leaf quality on herbivore performance and attack from natural enemies. - Oecologia 126: 418-428. Lill, J. T., Marquis, R. J. and Ricklefs, R. E. 2002. Host plants influence parasitism of forest caterpillars. - Nature 417: 170-173. Marsh, N. and Rothschild, M. 1974. Aposematic and cryptic Lepidoptera tested on the mouse. – Journal of Zoology 174: 89-122. Matteson, W. J. J. 1980. Herbivory in relation to plant nitrogen content. - Annual Review of Ecology and Systematics 11: 119-161. 217 Maycock, P. F. 1961. Botanical studies on Mont St. Hilaire, Rouville County, Quebec. - Canadian Journal of Botany 39: 1293-1325. Mayhew, P.J. 1997. Adaptive patterns of host-selection by phytophagous insects. – Oikos 79: 417-429. McAuslane, H., Johnson, F. A., Knauft, D. A. and Colvin, D. L. 1993. Seasonal abundance and within-plant distribution of parasitoids of Bemisia tabaci (Homoptera: Aleyrodidae) in peanuts. – Environmental Entomology 22: 1043-1050. Mols, C. M. M. and Visser, M. E. 2002. Great tits can reduce caterpillar damage in apple orchards. – Journal of Applied Ecology 39: 888-899. Moore, B. P., Brown, W. V. and Rothschild, M. 1990. Methylalkylpyrazines in aposematic insects, their hostplants and mimics. – Chemoecology 2: 4351. Morais, H. C., Diniz, I. R. and Silva, D. M. S. 1999. Caterpillar seasonality in a central Brazilian cerrado. - Revista de Biologia Tropical 47: 1025-1033. Niemela, P. and Haukioja, E. 1982. Seasonal patterns in species richness of herbivores:Macrolepidopteran larvae in Finnishdeciduous trees. – Ecological Entomology 7: 169-175. Nishida, R. 2002. Sequestration of defensive substances from plants by Lepidoptera. - Annual Review of Entomology 47:57-92. Okada, T. 1989. Parasitoids of the diamondback moth, Plutella xylostella (L.) (Lepidoptera Yponomeutidae): species and seasonal changes of 218 parasitism in cabbage fields. – Japanese Journal of Applied Entomology and Zoology 33: 17-23. Pasteels, J.M., Gregoire, J.C. and Rowell-Rahier, M. 1983. The chemical ecology of defense in arthropods. Annual Reviews of Entomology. – 28: 263-289. Peña, J. E., Duncan, R. and Browning, H. 1996. Seasonal abundane of Phyllocnistis citrella (Lepidoptera: Gracillariidae) and its parasitoids south Florida citrus. – Environmental Entomology 25: 698-702. Perkins, M. C., Woods, H. A., Harrison, J. F. and Esler, J. J. 2004. Dietary phosphorus affects the growth of larval Manduca sexta. - Archives of Insect Biochemistry & Physiology 55: 153-168. Peterson, S. C., Johnson , N. D. and LeGuyader, J. L. 1987. Defensive regurgitation of allelochemicals derived from host cyanogenesis by eastern tent caterpillars. - Ecology 68: 1268-1272. Power, M.E. 1992. Top-down and bottom-up forces in food webs: do plants have primacy? – Ecology 73: 733-746. Preszler, R. W. and Boecklen, W. J. 1996. The influence of elevation on tri-trophic interactions: Opposing gradients of top-down and bottom-up effects on a leaf-mining moth. - Ecoscience 3: 75-80. Price, P. W., Bouton, C. E., Gross, P., McPheron, B. A., Thompson, J. N. and Weis, A. E. 1980. Interactions among three trophic levels: Influence of plants on interactions between insect herbivores and natural enemies. - Annual Review of Ecology and Systematics 11: 41-65. 219 Rausher, M. D. 1981. Host plant selection by Battus philenor butterflies, the roles of predation, nutrition, and plant chemistry. - Ecological Monographs 51: 1-20. Richard, P.J.H. and Grondin, P. 2009. Histoire postglaciaire de la végétation, pp. 169-176, in Manuel de Foresterie, 2è édition, Ordre des ingénieurs forestiers du Québec, Éditions MultiMondes, Québec. Ricketts, T. H., Daily, G. C., Ehrlich, P. R. and Fay, J. P. 2001. Countryside biogeography of moths in a fragmented landscape: biodiversity in native and agricultural habitats. - Conservation Biology 15: 378-388. Ricklefs, R. E. and Matthew, K. K. 1982. Chemical characteristics of the foliage of some deciduous trees in southeastern Ontario. - Canadian Journal of Botany 60: 2037-2045. Sanz, J. J. 2001. Experimentally increased insectivorous bird density results in a reduction of caterpillar density and leaf damage to Pyrenean oak. – Ecological Research 3: 387-394. Schowalter, T. D., Hargrove, W. W. and Crossley, D. A. 1986. Herbivory in forested ecosystems. - Annual Review of Entomology 31: 177-196. Scriber, J. 1978. The effects of larval feeding specialization and plant growth form on the consumption and utilization of plant biomass and nitrogen: An ecological consideration. - Entomologia Experimentalis et Applicata 24: 694-710. 220 Scriber, J. 2010. Integrating ancient patterns and current dynamics of insect– plant interactions: Taxonomic and geographic variation in herbivore specialization. - Insect Science 17: 471-507. Scriber, J. M. 1977. Limiting effects of low leaf-water content on the nitrogen utilization, energy budget, and larval growth of Hyalophora cecropia; (Lepidoptera: Saturniidae). - Oecologia 28: 269-287. Scriber, J. M. and Slansky, F. 1981. The nutritional ecology of immature insects. Annual Review of Entomology 26: 193-211. Siemann, E., Tilman, D., Haarstad, J. and Ritchie, M. 1998. Experimental tests of the dependence of arthropod diversity on plant diversity. – The American Naturalist 152: 738-750. Singer, M. S., Carrière, Y., Theuring, C. and Hartmann, T. 2004a. Disentangling food quality from resistance against parasitoids: diet choice by a generalist caterpillar. - The American Naturalist 164: 423-429. Singer, M. S., Rodrigues, D., StiremanIII, J. O. and Carrière, Y. 2004b. Roles of food quality and enemy-free space in host use by a generalist insect herbivore. - Ecology 85: 2747-2753. Strong, D. R., Lawton, J. H. and Southwood (Sir), R. 1984. Insects on plants: community patterns and mechanisms. – Harvard University Press. Summerville, K. S. 2004. Do smaller forest fragments contain a greater abundance of Lepidopteran crop and forage consumers? - Population Ecology 33: 234-241. 221 Summerville, K. S. and Crist, T. O. 2003. Determinants of lepidopteran community composition and species diversity in eastern deciduous forests: roles of season, eco-region and patch size. - Oikos 100: 134-148. Summerville, K. S. and Crist, T. O. 2004. Contrasting effects of habitat quantity and quality on moth communities in fragmented landscapes. - Ecography 27: 3-12. White, P. J. T., McGill, B. J. and Lechowicz, M. J. 2011. Human-disturbance and caterpillars in managed forest fragments. - Biodiversity and Conservation 20: 1745-1762. Wilson, E. O. 1987. The little things that run the world (the importance and conservation of invertebrates). – Conservation Biology 1: 344-346. Wint, W. 1983. The role of alternative host-plant species in the life of a polyphagous moth, Operophtera brumata (Lepidoptera: Geometridae). Journal of Animal Ecology 52: 439-450. Wong, T. T. Y., Mochizuki, N. and Nishimoto, A. 1984. Seasonal Abundance of Parasitoids of the Mediterranean and Oriental Fruit Flies (Diptera: Tephritidae) in the Kula Area of Maui, Hawaii. – Environmental Entomology 13: 140-145. Woods, H. A., Perkins, M. C., Elser, J. J. and Harrison, J. F. 2002. Absorption and storage of phosphorus by larval Manduca sexta. - Journal of Insect Physiology 48: 555-564. 222 Mont Royal Mont St. Bruno Mont St. Hilaire Mont Shefford City of Montreal CANADA U.S.A Figure 1Caterpillars were collected from four sites in the St. Lawrence River valley of southern Quebec, Canada (Figure adapted from Atlas of Canada 2010) at the northern edge of the deciduous forest biome in eastern North America. The matrix surrounding each site isdominated by agricultural lands and urban development with the exception of Mont Royal, which is a forest fragment in an exclusively urban setting. Forest patches across the region are shown in dark gray, urban areas in light gray (including the City of Montreal at the left side of the pane). 223 b a %Water %Nitrogen %Phosphorus c Toughness (g) e d %Fiber %Lignin f g %Polyphenol Figure 2Average foliar qualities at quadrats in the months of June (Je), July (Jy) and August (Au) for (a) % water content , (b) % nitrogen content, (c) % phosphorus content, (d) toughness (grams), (e) % fiber content, (f) % lignin content and (g) % polyphenol content. Bars represent standard error. ANOVAs between months are significant for all foliar qualities at p < 0.001. Average foliar qualities for each quadrat were weighted based on the proportion of the total basal area occupied by each sampled host plant species in the quadrat. Host plant-specific foliar qualities were taken from Ricklefs and Matthew (1982). 224 Figure 3Regression tree analyses of the determinants of caterpillar richness (left hand panels) and abundance (right panels) at the quadrat level. Analyses are shown for June (a, b), July (c, d) and August (e, f). The variables shown are % Phosphorus content (P), % Polyphenol content (Phenol),% Fiber content (Fiber) and % Lignin content (Lig). Water content, Nitrogen content and Toughness were included, but were not significant. The clause presented at each node is the condition corresponding to the left hand fork (the right hand fork would be the opposite condition). Each clause is paired with an R2 value associated with that node in brackets; this value is equal to the complexity parameter of the node. At each terminus the average caterpillar richness or abundance is given (depending on the tree) along with the number of quadrats that satisfy the conditions of the fork (in brackets). Trees are pruned and show splits corresponding to R2> 0.05. 225 Caterpillar Richness b Phenol > 10.9 % (0.31) | Caterpillar Abundance (log10) Phenol > 10.9 % (0.44) | June a P < 0.24% (0.10) 2.1(9) 4.6 (10) P > 0.19| % (0.30) d 1.13 (43) P > 0.19| % (0.34) July c 0.54 (9) 7.0 (33) Lig < 10.5 % (0.06) 0.4 (10) 2.4 (11) 3.8(22) 5.7 (9) Fiber < 28.4 % (0.14) | 0.12 (10) 0.54 (11) f Phenol > 9.8 % (0.08) Lig > 9.2 % (0.08) 0.83 (31) 5.7 (22) Phenol > 10.3 % (0.13) | August e Lig < 10.5 % (0.10) Phenol > 8.2 % (0.06) Phenol < 6.2 % (0.07) 0.54 (10) 2.5 (10) 0.69 (19) 2.7 (9) 5.5 (11) 226 0.92 (36) Figure 4Regression tree analyses for the determinants of caterpillar richness (a) and abundance (b) for the month of August. These trees were created using the same data as for Figure 3e and 3f (respectively) but with the addition of parasitoid pressure data for each quadrat. In (a) parasitoid pressure supersedes % fiber content as the most important determinant of caterpillar richness. In (b) parasitoid pressure is less important and supersedes % polyphenol content as the primary determinant of caterpillar abundance. b a Parasitoids| > 6 % (0.18) Parasitoids >| 6 % (0.28) H2O > 51 % (0.10) 1.0 (19) P < 0.18 % (0.06) 0.93 (1259) 3.5 (35) 6.6 (17) 0.50 (9) 227 0.74 (15) Table 1Foliar nutrient properties in broadleaf trees in southern Quebec and Ontario in early June, early July and early August. Foliar nutrient data are averaged across 34 different common deciduous tree species. This table is adapted from Ricklefs and Matthew’s (1982), Table 3. Foliar Property June July August Water % Nitrogen % Phosphorus % Polyphenols (%) Lignin % Crude fiber % Toughness (g) 62.8 2.33 0.23 5.53 10.73 26.52 484 57.3 2.08 0.12 5.68 10.48 26.24 598 54.9 1.89 0.20 6.43 10.24 25.21 621 228 Relationship with Caterpillars + + + +/- Table 2A list of trees sampled (where n > 10) across the 72 study sites, and their associated caterpillar richness and abundance. Host Plant Tree Species Acer saccharum Fagus grandifolia Fraxinus americana Acer pensylvanicum Ostrya virginiana Tilia americana Prunus virginiana Rhamnus cathartica Acer platanoides Amelanchier arborea Ulmus rubra Quercus rubra Tsuga canadensis Carpinus caroliniana Acer spicatum Prunus serotina Cratageus spp Betula papyrifera Carya cordiformis Rhus typhina Cornus alternifolia Acer rubrum Malus pumila Prunus nigra Betula alleghaniensis # of Trees Surveyed 720 229 228 206 145 66 55 48 39 35 35 34 33 26 20 19 19 18 15 15 13 12 11 11 10 229 Total Caterpillar Richness 75 46 38 52 39 16 10 7 6 21 11 6 10 10 16 17 11 5 9 1 3 0 4 2 11 Total Caterpillar Abundance 673 317 95 221 239 24 26 11 11 58 19 11 25 21 32 38 15 5 9 4 3 0 4 3 17 Table 3 Parasitoid pressure on host plants in August. Host Plant Acer saccharum Ulmus americana Tsuga canadensis Carya cordiformis Prunus virginiana Fraxinus americana Salix nigra Quercus rubra Prunus serotina Malus pumila Acer negundo Alnus incana Acer rubra Ostrya virginiana Betula alleghaniensis Betula papyrifera Amelanchier arborea Tilia americana Acer pensylvanicum Cratagus spp. Fagus grandifolia # of Caterpillars Collected 127 919 590 225 1333 339 644 200 673 240 271 1137 96 68 81 1223 335 199 22 38 21 230 # of Parasitoids 20 141 80 22 106 20 35 9 27 8 9 34 2 1 1 15 4 1 0 0 0 Parasitoid Pressure 13.6 13.3 11.9 8.9 7.4 5.6 5.2 4.3 3.9 3.2 3.2 2.9 2.0 1.4 1.2 1.2 1.2 0.5 0 0 0 Table 4Emergence records for parasitoids in Ontario and Quebec documented by Natural Resources Canada, 1937-1949. Parasitoid Apanteleshyphantriae Apantelespolychrosidis Apecthisontario Campoplexvalidus Casinariaeupitheciae Casinariasemiothisae Hyposoterfugitivus Itoplectisconquisitor Meteorusbakeri Meteorushyphantriae Phobocampegeometrae Pimplapedalis Total mid-May to mid-June 0 16 0 0 0 0 0 2 0 1 0 2 21 231 mid-June to mid-Jul 1 21 41 1 0 1 18 77 0 7 3 3 173 mid-July to mid-August 11 9 63 0 3 2 22 168 3 9 0 3 293 General Conclusion One of the central aims of conservation biology is to better understand the forces that drive species richness, particularly across fragmented and disturbed landscapes (Andrén 1997, Fahrig 2003). Forest habitat represents one of the most modifiedhabitat types in North America (Ricketts and Imhoff 2003) and in Canada there is no ecozone more imperiled than the Mixedwood Plains in southern Ontario and Quebec (Gibbs et al. 2009). Because such a small proportion of the remnant forest fragments are protected,(Ministère des Ressources Naturelles 2002) there is little preventing the situation from getting worse in the coming decades. Understanding how biotic and abiotic forces affect species richness is important for the current and future conservation of species across this landscape. In this thesis I have focused on four facets of abiotic and biotic forces: (i) the impacts of natural disturbances, (ii) the impacts of human disturbances (iii) the effects of bottom-up forces, and (iv) the effects of topdown forces. Natural ice storm disturbances can have both positive and negative impacts on the richness in forest communities. There have been remote-sensing based tools used similar to the one I developed in Chapter 1 (e.g. King et al. 2005, Millward and Kraft 2004), but they have seldom been transformed into meaningful indices of habitat change relevant to different assemblages of coarse woody debris dependent guilds. While these previously-developed tools are useful on many levels, they are insufficient for making species-specific 232 predictions about ice storm impacts. Biomonitoring of rare or threatened species requires special knowledge about the geographical range of suitable habitat within a reserve. The ability to map the influx of specific types of coarse woody debris accomplishes this for the guilds in question. In addition, I have provided an important empirical example showing that NDVI outperforms other popular vegetation indices for making geospatial predictions of coarse woody debris influx following a major canopy-damaging disturbance. Beyond Mont St. Hilaire, there is now an empirical basis to use pre- and post-storm NDVI imagery to create a base map of damage in other protected forested areas using freely available archived Landsat 5 imagery (USGS 2007). There has been much research on the impacts of habitat destruction and fragmentation on forest communities (Fahrig 2003) while the impact of intraforest disturbances like trails are often overlooked. Biodiversity and abundance gains associated with trails are often due to the mixing of forest-dwelling and open-habitat species. This can result in an apparent positive impact of trailside habitat on overall biodiversity (somewhat analogous to edge-habitat) even though biodiversity may be declining amongst forest-dwelling species. Strong research documenting a negative relationship between trails and forest assemblages (Chapter 2) has been sorely needed to help make empiricallysupported management decisions to limit trails in forest reserves. Many forest reserve areas have a joint mandate to protect natural habitat while allowing access (through trails) for the enjoyment of the general public. In light of the 233 relationships I uncovered in Chapter 2, these two goals seem to be in direct conflict. One solution to this conflict is to manage trails in such a way so as to minimize impacts on trailside habitat. Recommendations to accomplish this goal are (i) construct boardwalks or physical barriers (i.e. rock walls or fences) to discourage pedestrians from walking off-trail (Doucette and Kimball 1990, Zhou and Tachibana 2004), (ii) limit the width of trails thereby increasing cross-trail canopy cover which will keep wind and temperature conditions in trailside habitat similar to that of forest-interior habitat, (iii) limit the prevalence of trails to reduce overall impact. In reserves that have some control over the amount public access (e.g. Mont St. Hilaire, or Mont St. Bruno and Parc de la Yamaska), trail traffic could be reduced by setting maximums for daily visitors. The bottom-up effect of host plants on insect herbivores has been a popular research area for pest-species capable of outbreak conditions (e.g.Maufette et al. 1983, Stuart and Polavarapu 1998) but has not been wellexamined across an entire species assemblage. For an assemblage, this question can be broken into two parts: (i) how are host plants related to insect herbivore assemblages (Chapter 3), and (ii) what is the mechanism behind host selection (Chapter 4). My analysis of caterpillars in the Monteregian showed that insect herbivore assemblages are more tightly linked to host plant identity and abundance; host plant richness was a poor predictor of caterpillars. Furthermore, caterpillars largely avoided non-native plants. The implications of these results are most applicable to Mont Royal, which has a high abundance of 234 non-native host plants (Acer platanoides and Rhamnus cathartica) and a low abundance of highly preferred host plants (A. pensylvanicum and O. virginiana). Although the eradication of A. platanoides would be impractical because they are often very large trees, it could be feasible to remove the vast majority of R. cathartica, replacing it with either A. pensylvanicum or O. virginiana. The mechanism behind these host plant preferences may be based on bottom-up foliar quality early in the season (June), but on top-down parasitoid pressure late in the season (August). These two mechanisms (bottom-up foliar nutrient pressure versus top-down parasitoid pressure) have been juxtaposed for a long time in ecology (Lill 2001). Mine is one of the first studies to draw links between these pressures and entire assemblages. The implication of this result is both research-oriented and conservation oriented. It underscores the need to examine tri-trophic relationships within assemblages in a seasonal context, largely because of temporal changes in foliar quality and parasitoidpressure. It also suggests that if there are significant declines in insect herbivore richness, late season parasitoidpressure may be a good factor to investigate as causal. Overall, the research presented in this thesis converges to single theme; conservation of forest fragment biodiversity in Mixedwood Plains forest fragments can be best achieved through the monitoring and understanding of both abiotic and biotic forces. Two very important management recommendations come from this. The first is to be aware of and manage human-presence in conservation areas. The second is to be aware of and 235 manage habitat quality (when necessary) through control of host plant species (planting or eradication) and coarse woody debris (protecting rare or important coarse woody debris habitat). 236 References Andrén, H. 1997. Habitat fragmentation and changes in biodiversity. - Ecological Bulletins 46: 171-181. Doucette, J. E. and Kimball, K. D. 1990. Passive trail management in northeastern alpine zones: A case study. - In: Moore, T. A., Donnelly, M. P., Graefe, A. R. and Vaske, J. J. (eds.), Proceedings of the 1990 Northeastern Recreation Research Synposium. - United States Department of Agriculture, Forest Service, pp. 195-201. Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. - Annual Review of Ecology and Systematics 34: 487-515. Gibbs, K. E., Mackey, R. L. and Currie, D. J. 2009. Human land use, agriculture, pesticides and losses of imperiled species. - Diversity and Distributions 15: 242-253. King, D. J., Olthof, I., Pellikka, P. K. E., Seed, E. D. and Butson, C. 2005. Modelling and mapping damage to forests from an ice storm using remote sensing and environmental data. - Natural Hazards 35: 321-342. Lill, J. T. 2001. Selection on herbivore life-history traits by the first and third trophic levels: the devil and the deep blue sea revisited. - Evolution 55: 2236-2247. Maufette, Y., Lechowicz, M. J. and Jobin, L. 1983. Host preferences of the gypsy moth, Lymantria dispar(L.), in southern Quebec. – Canadian Journal of Forest Research 13: 53-60. 237 Millward, A. A. and Kraft, C. E. 2004. Physical influences of landscape on a largeextent ecological disturbance: the northeastern North American ice storm of 1998. - Landscape Ecology 19: 99-111. Ministère des Ressources Naturelles. 2002. Rapport sur L'état des forêt québécoises 1995-1999. Ressources Naturelles et Faune, Gouvernement du Québec, Charlesbourg, Quebec. Ricketts, T. H. and Imhoff, M. 2003. Biodiversity, urban areas and agriculture: locating priority ecoregions for conservation. - Conservation Ecology 8: 115. Stuart, R. J. and Polavarapu, S. 1998. Oviposition preferences of the polyphagous moth Choristoneura parallela (Lepidoptera: Tortricidae): Effects of plant species, leaf size, and experimental design. - Environmental Entomology 27: 102-109. USGS. 2007. EarthExplorer. Earth Resources Observation and Science (EROS) Center. U.S. Department of the Interior and the U.S. Geological Survey, Sioux Falls, South Dakota. Zhou, J. and Tachibana, H. 2004. Natural revegetation after elimination of disturbance of human treading in the Tennyogahara Mire, the Taisetsu Mountains, Japan. - Vegetation Science 21: 65-78. 238 Appendix A1We surveyed 36 macrolepidopteran moth species across the four sites in our study region Mount Royal (R), Mont St. Bruno (B), Mont St. Hilaire (H) and Mont Shefford (S). All species IDs were based on 5th or 6th instar larvae identified using Wagner (2005). 239 Family Geometridae Lymantriidae Geometridae Noctuidae Geometridae Noctuidae Geometridae Geometridae Notodontidae Geometridae Geometridae Lymantriidae Noctuidae Noctuidae Noctuidae Noctuidae Noctuidae Arctiidae Noctuidae Notodontidae Geometridae Geometridae Geometridae Geometridae Noctuidae Noctuidae Geometridae Geometridae Lasiocampidae Lymantriidae Noctuidae Noctuidae Noctuidae Noctuidae Nolidae Notodontidae Genus and species Lambdina fervidaria Lymantria dispar** Melanolophia canadaria Lithophane antennata Itame pustularia Morrisonia latex Cyclophora pendulinaria Operophtera bruceata Symmerista leucitys Plagodis alcoolaria Hypagyrtis unipunctata Orgyia definite Crocigrapha normani Lithophane baileyi Zanclognatha cruralis Lithophane patefacta Morrisonia confuse Halysidota tessellaris Zale minerea Clostera spp Alsophila pometaria Phigalia titea Ennomos subsignaria Tetracis cachexiata Egira alternans Elaphria versicolor Lomographa vestaliata Pero ancetaria Malacosoma disstria Orgyia leucostigma Acronicta morula Bomolocha baltimoralis Eupsilia vinulenta Orthosia hibisci Baileya opthalmica Heterocampa guttivitta Distribution Among Sites BHS RBH RBHS BHS RBHS RBHS RBHS RBHS H BHS RBH BHS BHS HS BS BHS BH RH H S RB RB BS S BS BS B B B S B H S R H Y Number of Host Plants* 26 45 42 11 9 12 14 24 9 19 22 11 32 11 3 12 30 40 19 7 35 36 29 37 5 13 19 4 30 90 12 6 18 43 7 38 Total # of Individuals 70 59 52 52 44 20 15 14 11 7 6 6 6 6 6 5 5 4 4 4 3 3 2 2 2 2 1 1 1 1 1 1 1 1 1 1 * This is based on Handfield’s account (1999). He notes that some of these species have many more recorded host plants than he has documented. This is non-problematic for our study as all of the treespecies at our study quadrats are counted. ** Non-native species 240 Appendix A2 Micromoth distribution. Species ID A302 A309 A313 A315 A401 A402 A405 A406 A411 A501 A502 A503 A601 A604 A605 A606 A607 J1001 J1004 J1005 J1007 J1008 J1101 J1103 J1106 J800 J804b J805b J811 J901 J903 J905 Y1401 Y1405 Y1406 Y702 Y703 Y707 Y708 Distribution Among Sites RBHY R R H B BH B RBHY B HY H HY Y Y Y Y Y B RBHY B BHY RBHY BHY BY Y RBHY B BHY BR Y B BY Y Y Y BH RBH R BH Total # of Individuals 18 1 1 2 1 3 1 46 1 4 1 5 1 1 1 1 1 2 37 1 11 9 12 4 1 29 1 11 2 1 6 2 1 2 1 3 24 1 3 241 Appendix A3 Complete list of host plant species documented in vegetation surveys. Host Plant Species Abies balsamea (L.) Mill. Acer negundo L. Acer nigrum Michx. f. Acer pensylvanicum L. Acer platanoides L. Acer rubrum L. Acer saccharinum (L.) Small Acer saccharum Marsh. Acer spicatum Lam. Alnus incana (L.) Moench ssp. rugosa (Du Roi) R.T. Clausen* Amelanchier arborea (Michx. F.) Fernald Betula alleghaniensis Britt. Betula papyrifera Marsh Carpinus caroliniana Walter Carya cordiformis (Wangenh.) K. Koch Cornus alternifolia L.f. Crataegus spp** Dirca palustris L. Fagus grandifolia Ehrh. Fraxinus americana L. Malus pumila Mill. Ostrya virginiana(Mill.) K. Koch Picea glauca (Moench) Voss Pinus resinosa Aiton Pinus strobus L. Populus deltoides Bartram ex Marsh. Populus grandidentata Michx. Populus tremuloides Michx. Prunus nigra Aiton Prunus serotina Ehrh. Prunus virginiana L. Quercus rubra L. Rhamnus cathartica L. Rhus typhina L. Robinia pseudoacacia L. Salix nigra Marsh. Sorbus aucuparia L. Tilia americana L. 242 Tsuga canadensis (L.) Carrière Ulmus americana L. Viburnum lantanoides Michx. Viburnumopulus L. var. americanum Aiton † * Referred to as Alnus rugosa in this thesis. ** Mostly composed of a mixture of Cratageus punctata Jacq. and Cratageus mollis Scheele and theirhybrids. † Referred to as Viburnum trilobum in this thesis. 243 Appendix A4 Randomness was tested using randomization goodness-of-fit tests. P-values were calculated using 10,000 replicates of randomization. Caterpillar Species Cyclophora pendulinaria Itame pustularia Lambdina fervidaria Lithophane antennata Lymantria dispar Melanolophia canadaria Morrisonia latex <0.25 11.7 4.9 1.5 18.3 2.5 0.0 0.9 Expected Occurrence Actual Occurrence 0.26-0.50 0.51-0.75 0.76-1.0 >1.0 <0.25 0.26-0.50 0.51-0.75 0.76-1.0 >1.0 2.5 0.4 0.2 0.2 10 3 1 1 0 3.8 6.3 4.9 5.2 1 0 7 10 7 1.9 1.9 5.3 16.5 0 0 1 4 22 5.4 3.3 0.8 2.1 22 7 1 0 0 2.5 4.2 4.2 16.7 3 5 4 3 15 1.8 2.2 8.0 20.0 0 3 2 10 17 3.8 3.1 3.3 4.9 0 2 3 3 8 244 p-value 0.30 0.0094 0.19 0.20 0.55 0.63 0.43 Appendix B1Caterpillar collections were made from 38 host plant trees in 72 quadrats across the four study sites. 245 Host plant Acer saccharum Fagus grandifolia Fraxinus americana Acer pensylvanicum Ostrya virginiana Tilia americana Prunus virginiana Rhamnus cathartica Acer platanoides Amelanchier arborea Ulmus rubra Quercus rubra Tsuga canadensis Carpinus caroliniana Acer spicatum Cratageus spp Prunus serotina Betula papyrifera Carya cordiformis Rhus typhina Cornus alternifolia Acer rubrum Malus pumila Prunus nigra Betula alleghaniensis Viburnum lantanoides Sorbus aucuparia Alnus rugosa Viburnum trilobum Abies balsamea Acer negundo Acer saccharinum Dirca palustris Picea glauca Populus deltoides Robina pseudoacacia Salix sp Quadrat Occurrence of Host Plant 72 35 38 29 29 21 11 9 5 7 7 7 6 4 6 7 8 8 6 2 3 4 3 2 6 3 4 1 1 1 1 1 2 1 1 1 1 Total Trees Sampled Total Caterpillar Abundance Total Caterpillar Species 720 229 228 206 145 66 55 48 39 35 35 34 33 26 20 19 19 18 15 15 13 12 11 11 10 5 4 3 3 2 2 2 2 1 1 1 1 673 317 95 221 239 24 26 11 11 58 19 11 25 21 32 15 38 5 9 4 3 0 4 3 17 1 0 2 1 6 0 0 1 0 0 3 1 77 46 38 52 39 16 10 7 6 21 11 6 10 10 16 11 17 5 9 1 3 0 4 2 11 1 0 2 1 4 0 0 1 0 0 2 1 246 Appendix B2A record of the macrolepidoptera and microlepidoptera morphospecies that were surveyed. Microlepidoptera morphospecies were given unique alphanumeric designations and subsequent individuals were verified with digital images. All macrolepidoptera were identified using Wagner (2005) and Handfield (1999). 247 Family Genus Arctiidae Halysidota Arctiidae Lophocampa Drepanidae Euthyatira Drepanidae Pseudothyatira Geometridae Alsophila Geometridae Campaea Geometridae Cingilia Geometridae Cyclophora Geometridae Ectropis Geometridae Ennomos Geometridae Eufidonia Geometridae Eutrapela Geometridae Hypagyrtis Geometridae Itame Geometridae Lambdina Geometridae Lomographa Geometridae Melanolophia Geometridae Operophtera Geometridae Paleacrita Geometridae Pero Geometridae Phigalia Geometridae Phigalia Geometridae Plagodis Geometridae Prochoerodes Geometridae Tetracis Geometridae Xanthotype Lasiocampidae Malacosoma Lymantriidae Dasychira Lymantriidae Lymantria Lymantriidae Orgyia Lymantriidae Orgyia Noctuidae Achatia Noctuidae Acronicta Noctuidae Amphipyra Noctuidae Bomolocha Noctuidae Catocala Noctuidae Crocigrapha Noctuidae Egira Noctuidae Elaphria Noctuidae Eupsilia Noctuidae Lithophane Noctuidae Lithophane Species tessellaris caryae pudens cymatophoroides pometaria perlata catenaria pendulinaria crepuscularia subsignaria notataria clemataria unipunctata pustularia fervidaria vestaliata canadaria bruceata vernata ancetaria strigataria titea alcoolaria lineola cachexiata spp disstria plagiata dispar definita leucostigma distincta morula pyramidoides baltimoralis ultronia normani alternans versicolor vinulenta antennata patafacta 248 Total Microlepidoptera Total Abundance Morphospecies Abundance 5 A302 41 7 A307 1 1 A308 2 7 A309 1 6 A312 1 5 A313 1 1 A315 2 45 A401 1 1 A402 6 8 A404 7 7 A405 1 3 A406 52 15 A410 4 71 A411 2 183 A501 5 21 A502 1 112 A503 9 65 A504 3 7 A505 3 3 A601 1 1 A602 1 9 A604 6 10 A605 1 1 A606 3 2 A607 2 2 J1001 6 6 J1004 60 1 J1005 1 206 J1007 25 15 J1008 29 2 J1101 19 3 J1103 11 6 J1104 1 2 J1106 1 13 J1501 2 2 J1502 1 26 J703 9 6 J800 51 8 J804 5 7 J804b 4 197 J805b 21 9 J811 6 Noctuidae Noctuidae Noctuidae Noctuidae Noctuidae Noctuidae Noctuidae Noctuidae Nolidae Notodontidae Notodontidae Lithophane Morrisonia Morrisonia Orthosia Orthosia Zale Zale Zanclognatha Baileya Heterocampa Symmerista viridipallens var confusa latex hibisci rubescens minera unilineata cruralis ophthalmica guttivitta leucitys 249 10 12 54 11 12 9 1 20 20 5 44 J901 J904 J905 Y1401 Y1404 Y1405 Y1406 Y1407 Y1409 Y702 Y703 Y707 Y708 Y710 3 12 14 8 1 3 1 1 6 13 97 10 11 2